Plant Sv.

stematiEs
and Eualutian
Entwiddung chichte
und SystemaTder Pflanzen Supplementum 1

Flawering Plants
Evolution and Classification
of Higher Categories
SvmpDsium, Hamburg, September 8-12, 187&

Editl!d bV H. Hubitzki

Springer-Verlag Wien new Yark

Prof. Dr. Klaus Kubitzki
Institut fur Allgemeine Botanik und Botanischer Garten,
Universitat Hamburg, Federal Republic of Germany

This work is subject to copyright.

All rights are reserved, whether the whole or part of the material is concerned,
specifically thoso of translation, reprinting, re-use of illustrations, broadcasting,
reproduction by photocopying machine or similar means, and storage in data
banks.

© 1977 by Springer-Verlag/Wien

With 118 Figures

Library of Congress Cataloging in Publication Data.

:lIain entry under title:
}'lowering plants. (Plant systematics and evolution - Entwicklungsgeschichte und S)"stematik
der Pflanzen: Supplementum; 1l. 1. Botany-Classification-Congresses. 2. Plants-Evolution-Con-
gresses. 3. Phanerogams-Congresses. 1. Kubitzki, Klaus, 1933-. II. Series: Plant systematics and
evolution: Supplementum; 1. QK95. F5H. 5~2'.OOl'2. 77-25200.

ISBN-13: 978-3-211-81434-5 e-ISBN-13: 978-3-7091-7076-2

DOl: 10.1007/978-3-7091-7076-2
Preface

The original suggestion to organize a symposium about the classi-

fication and evolution of the Flowering Plants was made at, the
International Botanical Congress at Leningrad in 1975, and the idea
was so well accepted by several colleagues that plans for such a
symposium quickly took shape. An organizing committee consisting
of Professor H. MERXMULLER, Miinchen, Professor V. H. HEYWOOD,
Reading, and Professor K. KUBITZKI, Hamburg, was set up. The
conference took place on 7-12 September 197tl in the Institut fiir
Allgemeine Botanik of the University of Hamburg under the auspices
of the International Association for Plant Taxonomy and was at-
tended by 80 participants from 14 countries.
There have been several meetings in recent years which have
dealt with the origin and evolution of the Flowering Plants so that
it might be questioned whether yet another symposium dealing with
more or less the same subject were really "justified. As the reader
will see from the contents of the book, this symposium differed
from similar ones held recently in two respects: 1. Emphasis was
given to methodological aspects of the classification of higher taxa,
and 2. much classificatory and evolutionary evidence relating to
the higher taxa of Flowering Plants was presented. It would have
been foolish in the extreme to expect from this meeting a general
agreement on one universally acceptable classificatory system; on
the contrary, it transpired very clearly that there is still a great
deal of dissent about the higher taxa of the Flowering Plants and,
for example, even the apparently well-established separation between
Monocotyledons and Dicotyledons is challenged.
If this meeting was successful it was because it had the important
function of permitting communication and promoting understanding
not only among taxonomists but also between different kinds of
practitioners, and there is no doubt that the impact of the newer
sub-disciplines such as phytochemistry, anthecology and ultrastructure
on classification and evolution is still increasing.
The conference was made possible through the financial aid ob-
tained from various organizations and institutions to which we are
most grateful: The Deutsche Forschungsgemeinschaft; the Freie und
Hansestadt Hamburg; the Hamburgische Universitatsstiftung; and the
VI Preface

Universitatsgesellschaft Hamburg. It should be pointed out here that

t.he two last-mentioned organizations depend wholly on private funds
and have taken over the role of private Maecenas whose import.ance
is only fully appreciated in these days of butgetary restrictions.
The President of the University of Hamburg offered a reception
for the participants on 7 September, and I am grateful for this
and for allowing the meeting to be held at the University. Also
the combined efforts of staff members and research students of this
institute are gratefully acknowledged.
Finally, I want to express my very grateful thanks to all con-
tributors for so extensively displaying their knowledge of the subject,
and to Professor V. H. HEYWOOD for revising the English of several
contributions.

Hamburg, October 25, 1977 K. KUBITZKI

Contents

General Principles and Methods

HEYWOOD, V. H.: Principles and Concepts in the Classification of
Higher Taxa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
MEEusE, A. D. J.: Delimitation of the Major Taxa of the Higher
Oycadophytina: Theoretical Criteria Versus Taxonomic Practice. . .. 13
KUBITZKI, K.: Some Aspects of the Classification and Evolution of
Higher Taxa. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 21
SPORNE, K. R.: Some Problems Associated With Character Correlations 33
CAGNIN, M. A. H., GOMES, C. M. R., GOTTLIEB, O. R., MARX, M. C.,
ROCHA, A.1. DA, G. F. DA SILVA, M. F. DAS, TEMPERINI, J. A.:
Biochemical Systematics: Methods and Principles ................ 53
CLIFFORD, H. T.: Quantitative Studies of Inter-relationships Amongst
the Liliatae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 77
BURTT, B. L.: Classification Above the Genus, as Exemplified by
Ge8neriaceae, With Parallels From Other Groups . . . . . . . . . . . . . . . .. 97

Evolutionary Aspects and Taxonomic Evidence

GOTTWALD, H.: The Anatomy of Secondary Xylem and the Classifica-
tion of Ancient Dicotyledons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 111
PHILIPSON, W. R.: Ovular Morphology and the Classification of Dico-
tyledons ..................................................... 123
KRACH, J. E.: Seed Characters in and Affinities Among the Saxi-
jragineae .................................................... 141
BEHNKE, H.-D.: Transmission Electron Microscopy and Systematics
of Flowering Plants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 155
CRONQUIST, A.: On the Taxont>mic Significance of Secondary Meta-
bolites in Angiosperms ........................................ 179
HEGNAUER, R.: Cyanogenic Compounds as Systematic Markers in
Tracheophyta ................................................ 191
GOTTSBERGER, G.: Some Aspects of Beetle Pollination in the Evolu-
tion of Flowering Plants ...................................... 211
EHRENDORFER, F.: New Ideas About the Early Differentiation of
Angiosperms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 227
VIII Contents

Evolution and Classification of Major Taxa

KENG, H.: Phyllocladus and Its Bearing on the Systematics of Coni-
fers ......................................................... 2:-l5
DAHLGREN, R.: A Commentary on a Diagrammatic Presentation of the
Angiosperms in Relation to the Distribution of Character States 253
HUBER, H.: The Treatment of the Monocotyledons in an Evolutionary
System of Classification ....................................... 285
THORNE, R. F.: Some Realignments in the Angiospermae ............ 299
ENDRESS, P. K.: Evolutionary Trends in the Hamamelidales-Fagales-
Group ...................................................... 321
BERG, C. C.: Urticales, Their Differentiation and Systematic Position 349
WAGENITZ, G.: New Aspects of the Systematics of Asteridae ......... 375

Summary Lecture
MERXMULLER, H.: Summary Lecture ............................. 397

Index to Plant Taxa .......................................... 406

SUbject Index ................................................. 414
Plant Syst. Evol., Suppl. 1, 1-12 (1977)
~ by Springer-Verlag 197'"

Department of Botany, Plant Science Laboratories,

University of Reading, England

Principles and Concepts in the Classification of Higher Taxa

By

V. H. Heywood, Reading

Abstract: Despite the very great interest shown by botanists during

the last 100 years in the "phylogenetic classification" of the Flowering
Plants, much more attention seems to have been paid to the phylogenetic
components of such schemes (despite the lack of adequate evidence) than
to their taxonomic/classificatory components (for which much evidence
is available). There is in fact an inbuilt conflict involved in constructing
a scheme that attempts to express both the dynamic/historical phylogenetic
component and a static horizontal present day classification, which makes
most of the conventional phylogenetic tree·like disgrams conceptually
suspect.
In the Flowering Plants, only the classes and families are widely accepted
whereas there are numerous plausible ordinal classifications. The kind of
questions we have to ask in the case of higher categories and taxa such as
the subclass and order are: are the criteria to be primarily phylogenetic,
or taxonomic in the sense of aiming at better circumscriptions of the taxa,
greater predictive value, 'Stability, etc.? Only when we have answered
these and similar questions can we decide whether we should be aiming at
cladogenetic schemes or anagenetic grades or an explicit combination of
both.

Introduction
In introducing a symposium on the evolution and classification of
higher taxa, I am conscious of the pitfalls that lie ahead if we proceed
without clarifying certain basic assumptions and establishing principles
to be followed. At the same time, I realise both the futility of attempting
this and my own inadequacy. Moreover, I am aware that I am out of
sympathy with much of the work on systems of classification that has
been done in the past and that is discussed later in this symposium,
largely on the grounds that it is un-scientific, misconceived and con-
ceptually unsound. If, however, my remarks have the effect of stimu-
lating a discussion that will lead to clear and logical thinking and action,
then I shall have fulfilled my role.
Plant By"t. Evo!., Supp!. 1 1
2 v. H. HEYWOOD:

Classification

I think it is generally agreed that one of the major roles of taxonomy

is to produce a classification (or system of classification) of organisms
that best reflects the totality of their similarities and differences. At
the level of the genus and above the units of classification are basically
a filing and retrieval device, or language of communication for use on a
single time-plane as HUGHES (1976) put it recently. The basic method
used is that of comparative morphology and is concerned with both
the content of the taxa recognised and their delimitation from others of
the same rank. This can be, and in the majority of cases is, undertaken
without using any phylogenetic information, although evolutionary
concepts may inform the taxonomist in his task. It is, as CRONQUIST
(1975) says, possible to perceive what we call natural groups without
thinking in terms of evolution, but such perception is facilitated by an
evolutionary frame of reference. This is not the same, however, as using
factual phylogenetic data, which is seldom possible. The evolutionary
frame of reference is, as regards extant plant groups, restricted to a loose
monophyletic requirement which, as we will see later, is determined from
comparative morphology and based on inference, so that it seldom, if
ever, affects the actual procedures of classification.
Before considering in detail the role of phylogeny in classification, I
think we should look at the current classificatory situation in the Flow-
ering Plants (Angiosperms), at the level of the family, order and above.
The number of Angiosperm families recognised varies from about
300-450, according to which system is followed. In purely practical
terms, the family is as basic a unit in taxonomy as is the species. In
other words, in the majority of cases a consensus is reached as to the
specific identity of a given plant, while there may be considerable debate
as to its generic position due to the vagaries of generic/subgeneric clas-
sification, but the moment any generic classification is accepted, there
is usually an automatic attribution to a family. Indeed in many, if not
most, cases one can place a plant in a family without knowing to which
genus or species it belongs-this is certainly true of the 200 or so families
recognised by BENTHAM & HOOKER and which formed part of our
basic training. The question that we must pose is how well we know the
families? As regards delimitation from other families there is still con-
siderable debate, especially in the case of many of the smaller families
recognised in recent years and there is still debate as to the familial or
subfamilial position of the major components of many of the larger
families-the Rosaceae, Leguminosae, Saxitragaceae, etc. To the non-
taxonomist, this may seem very strange-after many decades of apparent
intensive study, there is still debate as to which families should be re-
Principles and Concepts in the Classification of Higher Taxa 3

cognised, especially when we remember the relatively small number of

families involved!
We should perhaps enquire into the reaSOnS for this. Basically they
are that we have seldom made adequate comparative studies (a point I
shall return to later), which in turn depends on a detailed knowledge of
the content of families (again a point to be considered later), and also
because there is nO agreed principle for the delimitation of families other
than comparability of status in relation to allied families. We have to
remember that most families were recognised and described long before
the majority of the genera they contain today were described. They
have, therefore, been built up by a process of accretion. Originally they
were recognised, at least in the case of a natural family such as the
Compositae, Labiatae, Umbelliferae, Palmae, etc., by a process of mentalf
visual correlation of plants with similar characteristics. In a very non-
scientific way, they were recognised by a process of synthesis, i.e. from
below. Since then, thousands of genera have been added into families
when they appeared to be accommodated without too much difficulty
but the process of revision of families from below, On a synthetic basis,
has not been repeated, largely because it is too big a task to undertake
and we are not even sure how to proceed. A consequence of this is that
the family today is a somewhat vague abstraction. It is, in fact,
virtually impossible for us to envisage a family, let alone see or handle
one.
This does not, however, prevent us from talking about families being
advanced or primitive, or related, etc. when what we really mean is our
appreciation of certain component genera with which we are familiar
or about which some special piece of information has been available.
Even more basic is the fact that in very few cases do we know the actual
content of a family because, as noted above, so many additions and
alterations have been without any overall revision having been effected.
There is not even available an up-to-date account of families giving
their generic content, or even a series of detailed family descriptions.
The average family description is at such a level of abstraction as to be
useless except perhaps for diagnostic purposes (and even then of dubious
value). As a basis for information-processing or evolutionary interpreta-
tion it is of little value. When a phylogenist tells about the positioning
of a family or its status or origin, he is really talking in terms of his
personal appreciation of those members of the family with which he is
familiar. I need not elaborate On this point except to emphasize the need
for: (1) the accumulation at family level of a more scientifically organised
data-base; (2) at the same time, the need for detailed reviews (if not
revisions) of the Angiosperm families. How this is to be achieved is
something to which we should address ourselves.
1·
4 V. H. HEYWOOD:

Readers will doubtless be aware of a paper by Dr. JACOBS of Leiden

(1969) entitled "Large Families-not alone" in which he lamented the neglect
of the families from a revisionary point of view. Only by co·operative
effort can an adequate data base be achieved, possibly by using a computer-
based system recording the data or, if not, some centralised international
centre for the accumulation, listing, recording and storing of data. Without
an adequate data base, no major familial revision will be possible. In the case
of the Oompositae, we attempted last year (1975) at Reading, as part of
the work of a symposium, to revise the tribal classification of the family.
The method we adopted was for a specialist (or group of specialists) to
accept the responsibility of preparing for each tribe a review to generic
level, including an assessment of the content and sub-division of the tribe
and its status, relative to other tribes. Even the task of finding out which
genera existed and to which tribe they belonged proved to be quite a major
task, let alone the question of the validity of the genera and the tribe it-
self. There are still too many genera whose tribal position is uncertain and
quite major realignments of the tribes themselves were suggested. It would
be idle to pretend that we have solved most of the problems or that a
consistent tribal or generic ranking has been achieved, but at least we
are now in a position to assess what the content of the family is in terms
of genera in the first overall assessment since BENTHAM'S masterly mono-
graph over a century ago (1873). The results will be published as part of
the proceedings of the symposium (HEYWOOD et al. 1977). It is proposed
to hold a similar review of the family Leguminosae next year organised
by the Royal Botanic Gardens, Kew, and the Missouri Botanical Garden
in association with the University of Reading.

As regards the rank or taxon order, it has to be said that in practical

terms, it is of little value. There are at least ten current ordinal classifica-
tions available today and the order has no precise meaning unless one
adds sensu CRONQUIST, sensu ENGLER, sensu SOD, sensu HUTCHINSON,
sensu TAKHTAJAN or sensu THORNE, etc. I know of no principles or
norms or even conventions that apply to the circumscription and
ranking of orders (except, of course, their relative positional value in
the hierarchy) and I cannot find any coherent explanation in the litera-
ture. At least at the family level one can seek certain comfort in talking
about the family representing some conspicuous grouping, recognisable
in some cases even by the layman, or about it representing some evolu-
tionary peak or adaptive radiation. The order, on the other hand, seems
to be purely arbitrary as a taxonomic tank and apart from being supra-
familial, there is little to say about it. Is it even needed? It is certainly
not a useful information retrieval device unless qualified as noted above.
The diversity of content of the orders in the different classificatory
systems makes a nonsense of talking about their evolutionary rela-
tionships. How can we seriously discuss the phylogeny of units which in
view of the divergence of opinion as to their circumscription may not, in
the majority of cases, have an evolutionary coherence!
Certainly it may be useful to have some supra-familial groupings for
Principles and Concepts in the Classification of Higher Taxa 5

ease of reference, but I wonder whether an informal system might not

be the answer, or even an avowedly artificial one. The prevalent in-
sistence on the use of clades (the result of evolutionary divergence) in
classification is understandable but there is a case to be made for the
occasional use of artificial (and presumably polyphyletic) grades, a
practice previously adopted by ENGLER and (with hindsight) by BENT-
HAM & HOOKER and others.
Before leaving the classificatory aspects of higher taxa, perhaps a
word ought to be said about the two classes of the Angiosperms-the
Dicotyledons and the Monocotyledons. In a recent review EL-GAZZAR &
HAMZA (1975) have considered a number of characters in addition to the
conventional ones used to separate the two classes. These new characters
include pollen and stomatal morphology, distribution of calcium oxalate
crystals, lignin chemistry, duration of the mitotic cycle, etc. and they
find that the traditional distinction between the Monocotyledons and
the Dicotyledons is supported but that the Alismataceae, Butomaceae,
and Limnocharitaceae are better associated with the Nymphaeaceae and
N elumbonaceae in the Dicotyledons while, on the other hand, the Dios-
coreaceae fit better with the Aristolochiaceae.
Others, including MEEUSE, are of the opinion that the Monocotyledon-
Dicotyledon groups will have to be dismembered. So even at the highest
levels in the Angiosperms there is room for considerable debate about
content and circumscription.

Evolutionary or Phylogenetic Classification

It is probably misleading to talk of evolutionary or phylogenetic
classifications since one is forced to combine in such schemes both a
vertical dimension, representing the historical-evolutionary dynamic
elements and a horizontal dimension, representing present-day rela-
tionships, i.e. the actual classification to be used. Two-dimensional,
horizontal classifications, based largely on comparative morphology of
living Angiosperms, are perfectly valid and are an expression of the
traditional box-within-box methodology normally employed in biological
taxonomy.
On the other hand, trees are basically a false representation of the
pathways and pattern of evolutionary processes. There is seldom any
discernible basis for the branching pattern, not even for the major
branches which in the Angiosperms are in dispute. For the full represent-
ation of evolutionary relationships multi-dimensional diagrams would
be needed as remarked elsewhere (HEYWOOD 1966).
When a tree (or shrub) diagram is purported to represent both the
evolutionary and the classificatory components, the result is untenable.
6 v. H. HEYWOOD:

Unfortunately, the majority of recent phylogenetic/evolutionary

schemes of classification have been presented in the form of a branching
tree. While I agree with STEBBINS (1975) that the evolutionary tree is
better represented as a much-branched shrub, I disagree with him that
this form is characteristic of the CRONQUIST-TAKHTAJAN system. In
fact most of such tree or shrub systems can be either placed on their
sides or have the connecting lines removed and they become expressions
of present-day relationships. They are little more than a graphic exten-
sion of phenetics. In this connection the attitude of CONSTANCE (1964)
is a correct one, i.e. that such phylogenetic/evolutionary schemes or
diagrams are aids to teaching or serve as an intellectual stimulus. Harm
is caused, however, when they are regarded as more than this and as
statement of factual or even putative evolutionary (i.e. cladistic, patristic,
chronistic) relationships.
Systems such as those of DAHLGREN (1975) which are graphic repre-
sentations of phylogenetic trees in transection are more satisfactory,
although it must be noted that the tree itself is imaginary and the extent
to which the imagined branches determine the spacing of the present-day
groups clearly detracts from their usefulness. DAHLGREN writes, "To
present orders of families of Angiosperms as a two-dimensional model is
no innovation. Where this has been done, the relative position of the
groups has been determined by the degree of mutual similarity. One
disadvantage is that the reader, and sometimes even the constructor
of the system, has been inclined to look upon the system of now living
plants as an evolutionary tree, where some groups are regarded as
descendants of others in the diagram. This applies in particular to
HUTCHINSON 1969. Evidence for this type of evolutionary tree is usually
sparse or lacking. The present Magnoliales, in particular, is often regarded
as an ancestral group, other groups being frequently indicated as shooting
out of it like lateral buds. The introduction here of a third dimension,
time, is intended to prevent any such misinterpretation. It must be said
that practically nothing is known about the course of evolution in the
Angiosperms, so that the tree must be presented in such a generalised
form that no evolutionary details are shown. Even the two-dimensional
representation of the Angiosperms involves a tremendous number of
problems."
It seems to me quite clear that the vertical dimension which makes
the diagram a "tree" is derived from an interpretation of present-day
relationships between the groups and not vice versa. The question has
to be asked, therefore, if the time dimension adds anything of value or
simply misleads by giving apparent information. DAHLGREN does not
give a satisfactory answer to this point. His justification-that of avoid-
ing misinterpretations about the direct descent of contemporaneous
Principles and Concepts in the Classification of Higher Taxa 7

groups, one from another-seems gratuitous in that he has not made a

case for the use of a tree at all. He does, however, discuss the problems
of constructing a "reasonably functional two-dimensional diagram" for
the orders and families of the Angiosperms. It is worth quoting his
arguments in full since such two-dimensional "bubble" diagrams seem
likely to acquire popularity as methods of presenting Angiosperm
relationships and perhaps supplant the more customary tree dia-
grams.
DAHLGREN does not in fact believe that it is possible to construct
reasonably functional two-dimensional diagrams but continues, "To
place groups in exactly those positions that reflect their affinities becomes
increasingly difficult when consideration has to be taken to the number
of species in each group. For example, large 'bubbles' may prevent
other, related, groups from meeting in the model, and small groups
cannot be extended so as to approach sufficiently close to other groups
showing great similarity. In any system, it seems, some families or orders
apparently appropriately placed at the same time show several perhaps
phylogenetically important similarities to one or more remotely placed
group which in turn appears to occupy an appropriate position." He
concludes by saying that it is imperative that botanists should persevere
with the attempt to produce systems of this or similar types to survey
the many groups of Angiosperms. The need, he says, is pedagogic rather
than scientific. I cannot help observing, however, that his self-confessed
difficulties of juxtaposing "bubbles" as wished, due to the physical
constraints of the system, the failure to explain exactly why any given
set of juxtapositions is employed, plus the acute difficulties mentioned
above of determining the circumscription and content of orders, makes a
nonsense of the tree or time-dimension implied and I would suggest that
for pedagogic reasons, never mind scientific ones, it be ignored.
If the problem of presenting relationships in a two-dimensional
diagram can be overcome, then what is theoretically possible by way of
an evolutionary-phylogenetic classification is a series of horizontal
classifications in time, like a stratigraphy, culminating in the present-
day time-scale classification which would receive its explanation from
the past but not necessarily its form. It has to be stressed that even if
the full chronological fossil history of the Angiosperms were known, the
classificatory problems would not be automatically solved as these
depend on questions of ranking, of taxonomic judgement, criteria such
as relative size of gaps between taxa, taste, tradition, etc. Classifications
are made by man, although made possible by evolution.
Another point that is fundamental here, although often overlooked,
is that the rank of any taxonomic group followed through evolutionary
time will change. This has been clearly stated recently by HUGHES (1976)
8 v. H. HEYWOOD;

as follows, "If, however, a genus or higher taxon-concept is used to

include 'species' from a range of geologic time-planes, its significance in
terms of limits against other taxa (or even in terms of cluster centres)
will be different on each time-plane studied depending on the general
state of biological evolution at each of these times. This is not a measure
of the number of fundamental taxa in each higher taxon-concept which
may either increase or decrease with time, but a statement of the posi-
tion of the family (or order) in the context of all those other families (or
orders) existing at the time selected."
This leads to another aspect, what HUGHES calls "backwards sys-
tematics" (with perhaps a vague inference about systematists them-
selves!) i.e. the persistent use of living taxa of all ranks to accommodate
Palaeogene and Mesozoic fossils. In fact nearly all fossil Angiosperms
have been treated in this way, although it virtually precludes any
meaningful discussion of evolution within the families or groups con-
cerned. HUGHES asks, how can the details of descent of a living plant
taxon from some Cretaceous fossils be determined when the latter have
already been identified with the former? The lack of an adequate fossil
record, and the inadequate way in which fossils have been classified in
terms of extant taxa rather 'than in relation to members of their own
time-zone, has led to comparative morphology of extant plants being
used as a substitute for genuine fossil historical information by various
methods of extrapolation backwards.
It is, nowadays, standard practice to use information from the mor-
phology of living Angiosperms to attempt the reconstruction of the
"primitive" flower or even the ancestral Angiosperm in all its charac-
teristics. The difficulty about all these ingenious (or, in some cases,
ingenuous) theories is that there is no way they can be tested, short of
finding actual fossils for confirmation, and, what is more, they have fed
on each other to such an extent that the body of apparent evidence
(much of it derived in a circular manner) is so impressive that it is dif-
ficult at times to remember that there is no factual basis for it beyond
the present-day starting point and general trends in the fossil record.
If one adds to this certain other pieces of dogma such as the improbability
(= virtual impossibility) of the eight-celled female gametophyte and
associated double fertilisation having evolved more than once (although
this hypothesis can be challenged and has no factual basis-the eight-
celled gametophyte has not been seen in pre-Quaternary materia!!), we
are then more or less forced to accept another dogma, the monophyletic
origin of the Angiosperms, which in turn makes it easier to concentrate
our thinking on a single archetype or archetypal group. This is what one
might term canalisation of conceptual thinking.
The methods which may be employed for attempting to determine
Principles and Concepts in the Classification of Higher Taxa 9

ancestral or primitive characters by extrapolation from the features

possessed by contemporaneous organisms are several. While they have
been familiar to zoologists for several decades, few Angiosperm taxo-
nomists have applied them explicitly. There is a marked difference
between the botanical and zoological literature in this respect. This is
very evident if one consults the journal "Systematic Zoology", for
example, which has no botanical equivalent.
The main methods are correlation of characters used in conjunction
with what HENNIG (1966) calls "reciprocal illumination" or what
WALKER (1974, 1976) calls "spiral reasoning". As WALKER (1976) says,
in view of the wholly inadequate fossil record of Angiosperms, a combina-
tion of character correlation and spiral reasoning frequently represents
the only available way of determining the primitive character-state of
characters peculiar to Angiosperms.
Character correlation may be what WALKER calls random at one
particular taxonomic level or time-scale. This is the basis of the approach
of SPORNE in a series of papers (see p. 33). Character correlation may also
be directional, which consists of individually observing the direction of
character trends within a number of different lower taxa included in the
taxon for which it is hoped to determine which character-states are
primitive. This involves working out character phylogenies or semo-
phyleses. If there is a high cori'elation of many characters, some of
which are thought to be primitive on independent grounds, then, so the
argument goes, there is a strong probability that each character repre-
sents an ancestral or primitive state.
But characters are not taxa and while a primitive character is
usually ancestral, a primitive taxon is not. As WALKER says, a primitive
taxon is simply one that retains a large number of primitive characters
relative to some other taxon. Unfortunately, in the absence of fossils, it
is not easy to deduce the direction of character trends and it is impos-
sible to prove. STEBBINS (1975) has some very pertinent comments to
make on this question in relation to the developmental basis of the
individual characters concerned.
The definitive way of determining primitive characters (but not
primitive taxa) is the study of character-states sequentially in a time-
sequenced fossil record. This can seldom be done except, for example, in
the case of pollen (cf. WALKER 1976). Even when, as in pollen grains,
there is a reasonably complete fossil series, it tells us nothing about the
other characters of the plant produced the pollen. There is certainly no
guarantee that pollen similarity means taxonomic identity. Unfor-
tunately, there are virtually no cases where the "primitive" characters
found in fossil pollen are associated with other independent "primitive"
characters because of the lack of linked fossil material.
10 v. H. HEYWOOD:

WALKER (1976), by applying the methods of character correlation,

character trends and spiral reasoning, arrives at the conclusion that the
ancestral pollen possessed the following characteristics: anasulcate, with a
long sulcus; heteropolar; bilateral; boat.shaped-elongate; atectate and
primitively columella·less; more or less psilate; solitary; and large. In
the face of this, the most primitive pollen found in living Angiosperms
occurs in the Magnoliaceae, Degeneriaceae and Annonaceae.
When we look at the fossil record, however, we find that the earliest
fossil pollen accepted as Angiosperm is that of the genus Clavatipollenites
but this is too specialised in several features (such as the possession of
columellae) to have been ancestral and is probably most similar to that of
the Chloranthaceae amongst extant Angiosperms.
WALKER concludes that while a tremendous diversity has been found
in the pollen of living "primitive" Angiosperms, it has to be remembered
that most of it is at the level of monosulcate-derived pollen. And while
tricolpate or tricolpate.derived pollen is essentially restricted to the Angio-
sperms (dicotyledons) and can therefore be more readily identified in the
fossil record as being angiosperm rather than gymnosperm, it is too advanced
to inform us about the early evolution of the Angiosperms. At the critical
period in geological history, pollen of Angiosperms and Angiosperm ancestors
was at the monosulcate level of evolution and consequently difficult, if
not impossible, to distinguish as being angiosperm rather than gymnosperm
in the absence of other evidence.

Palaeobotanists interested in the orIgm of Flowering Plants must

look, WALKER says, for fossil Gymnosperms that produced large, more
or less psilate, boat-shaped-elongate pollen grains with a long anasulcus
and an atectate, primitively columella-less exine. "For it was probably
pollen of this type that some extinct groups of seed plants crossed the
line from gymnosperm to angiosperm."
Whether this line was crossed by one or several lines-whether the
Angiosperms were a monophyletic clade or a pleiophyletic grade-re-
main.s to be seen. To me it seems improbable that the whole syndrome
of angiospermy evolved neatly once in a single evolutionary line.
In summary, the fossil record even as it is available today, has much
to tell us if studied more thoroughly. So far it does little to confirm cur-
rent views as to the actual origins of the Angiosperms although it sup-
ports the relative primitive nature of the woody Ranalean families in
respect of some (but not all) features.
As STEBBINS (1975) comments, in the absence of significant fossils,
so many unknown factors exist that the method of extrapolation is
inadequate as a means of deducing what ancestral forms were like on
the basis of any living Angiosperms. He draws attention to the fact
that if one considers the fossil record of the conifers, the common an-
cestors of the main modern families, Pinaceae, Taxodiaceae, Cupressaceae,
Taxaceae, and Podocarpaceae, are so different that they could not pos-
sible be reconstructed except on the basis of fossils: this seems to be
Prin<;ljples and Concepts in the Classification of Higher Taxa 11

generally true of other groups for which there is an adequate fossil

record and there is no reason to believe that the situation in the Angio-
sperms will prove to be different. A recent report of a fossil dicotyledo-
nous Angiosperm flower from the Upper Cretaceous (TIFFNEY 1977)
tends to confirm this: its characteristics are distinctly angiospermous
yet their exact association does not appear to occur in any modern
group and attempts to place it in an extant angiosperm taxon have
failed.
Conclusion
While methods do exist for the extrapolation of phylogenetic trends
from extant forms, in the absence of adequate fossils, these seem to
have little bearing or effect on the so-called phylogenetic or evolutionary
systems of classification of the Angiosperms except to assist in working
out the relative degree of primitiveness or advancement of individual
higher taxa. They do, however, help give us an insight into the evolu-
tionary processes involved. Because of the size and complexity of the
problems involved, the methodologies of evolutionary extrapolation and
reconstruction have not been consistently applied in the construction of
systems of classification which owe much more to the traditional meth-
ods of taxonomy than to the principles of evolution. On the other hand,
we should press for the publication of the detailed reasons for the form
and content of the currently adopted systems of classification, such as
those of CRONQUIST, TAKHTAJAN, and THORNE. Admittedly, the latter
author has commenced publication of the detailed rationale for part of
his system, for example, his phylogenetic classification of the Annoni-
florae (THORNE 1974), a move greatly to be welcomed. Until this is
more commonly done, we must remain somewhat sceptical about their
systems, qua classifications, and modifications to them in the absence
of a data-base. The horizontal, basically phenetic approach of DAHL-
GREN is a welcome development, especially as there is now becoming
available an impressive body of comparative data (some of it presented
in this volume) which can be used to test it.
Botanists can scarcely complain if the currently available systems of
classification do not meet their needs unless they make these needs
clearly known.

References
BENTHAM, G., 1873: Notes on the classification, history and geographical
distribution of Compositae. J. Linn. Soc. (Bot.) 13, 335-577.
CONSTANCE, L., 1964: Systematic botany-an unending synthesis. Taxon 13,
257-273.
CRONQUIST, A., 1975: Some thoughts on angiosperm phylogeny and taxo-
nomy. Ann. Missouri Bot. Gard. 62, 517-520.
12 V. H. HEYWOOD: Principles and Concepts in the Classification

DAHLGREN, R., 1975: A system of classification of the angiosperms to be

used to demonstrate the distribution of characters. Bot. N otiser 128,
119-147.
EL·GAZzAR, A., and HAMZA, M. K., 1975: On the Monocots·Dicots distinc-
tion. Publ. Cairo Univ. Herb. 6, 15-28.
HEYWOOD, V. H., 1966: How many taxonomies? Rev. Roum. BioI. ser.
Bot. 11, 101-106.
- HARBORNE, J. B., and TURNER, B. L. (Eds.), 1977: The Biology and
Chemistry of the Compositae. London·New York: Academic Press.
HENNIG, W., 1966: Phylogenetic Systematics. Urbana, Chicago and London:
Univ. Illinois Press.
HUGHES, N. F., 1976: Palaeobiology of Angiosperm Origins. Cambridge:
University Press.
JACOBS, M., 1969: Large Families-not alone! Taxon 18, 253-262.
STEBBINS, G. L., 1975: Deductions about transspecific evolution through
extrapolation from processes at the population and species level. Ann.
Missouri Bot. Ga,rd. 62, 825-834.
THORNE, R. F., 1974: A phylogenetic classification of the Annoniflorae.
Aliso 8, 117-209.
TIFFNEY, B. H., 1977: Dicotyledonous angiosperm flowers from the Upper
Cretaceous of Martha's Vineyard, Massachusetts. Nature 265, 136-137.
WALKER,J. W., 1974: Aperture evolution in the pollen of primitive angio.
sperms. Amer. J. Bot. 61, 1112-1137.
- 1976: Comparative pollen morphology and phylogeny of the Ranalean
complex. In: Origin and early evolution of angiosperms (BECK, C. B.,
Ed.), 241-299. New York: Columbia Univ. Press.

Address of the author: Professor V. H. HEYWOOD, Department of Botany,

Plant Science Laboratories, The University, Whiteknights, Reading RG6
2AS, England.
Plant Syst. Evol., Suppl. 1, 13-19 (1977)
~ by Springer-Verlag 1977

Hugo de Vries-Laboratorium, University of Amsterdam, The Netherlands

Delimitation of the Major Taxa of the Higher

Cycadophytina: Theoretical Criteria
Versus Taxonomic Practice
By

A. D. J. Meeuse, Amsterdam

Abstract: The application of other than strictly morphological criteria

in attempts towards producing a "phylogenetic" classification of the Flowering
Plants is clearly unsound as long as the fundamental morphological inter-
pretation of certain magnoliophytic features remains unsettled. Such
"established" notions as gymnospermy versus angiospermy, the nature of
the functional reproductive units, the sporophyll concept, the assumed
primacy of phaneranthy and the entomophilous syndrome (including the
sex distribution), etc., which postulates primarily deciqe the basic framework
for the taxonomic arrangement in a system, may well be antiquated. A grea-
ter heterogeneity among the major taxa of the magnoliophytic assembly
is in better agreement with the available evidence and finds some support
when additional taxonomic criteria are applied. It is to be expected that
appreciable advances will result from such ancillary criteria if applied
to a framework based on a neological approach of the phylogeny and
morphology of the Magnoliophyta founded upon alternative assumptions
concerning their original anthomorphology and palaeoecology. The present
author is entirely in favour of the application of evidence from all sources
potentially providing useful taxonomic pointers, but does not believe
that such data can be decisive for the delimitation and placing of higher
taxa among the Angiosperms until some agreement concerning several
fundamental, morphological concepts has been attained. This inevitably
involves phylogenetic speculations concerning the nature and morphology
of the early magnoliophytes which already require the application of some
additional criteria such as palaeoecological evidence and certain palaeo-
botanic clues.

Although various kinds of criteria are gradually being given more

prominence in classificatory botany (cf., e.g., JOHRI 1963, MAHESHWARI
1964, KUBITZKI 1969, 1973, MEEUSE 1970, BA.TE-SMITH 1972, PHILIPSON
1974, BOUMA.N 1974, DAHLGREN 1975, DICKISON 1975, EYDE 1975,
JENSEN et al. 1975, SPORNE 1975), their application to the distinction
or rearrangement of higher taxa of the advanced cycadophytinous
14 A. D. J. MEEUSE:

spermatophytes has mostly remained restricted to the family level.

Recently the Thelygonaceae have, for instance, been transferred to the
Rubiales in the most recently published systems of classification, and the
inclusion of the Batidaceae and Tropaeolaceae in the Capparales has
been proposed by DAHLGREN (1975). These transfers were mostly based
on phytochemical evidence, but this is not the case in BEHNKE'S studies
of microplastids which clearly seem to provide an absolute criterion
(viz., the presence of a special type of plastid) for the segregation of
the Centrospermae (= Caryophyllidae) from all other A ngiospermae
(see BEHNKE 1972, and this issue). This is also the only example of
which one can truly say that it served to distinguish a major taxon at,
at least, ordinal level on the ground of a non-conventional differential
characteristic, although one must bear in mind that the separate status
of the centrospermous alliance had already been accepted much earlier
on the basis of traditional criteria. BEHNKE'S data confirm the caryo-
phyllalean affinity of several taxa whilst they point to the exclusion
of other families from the assembly.
Apart from this example, several other attempts have been made
to use criteria other than the customary macromorphological features,
e.g., characteristics of the ovule (PHILIPSON 1974, and this issue; see
also BOUMAN 1974) and the androecial ontogeny (LEINS 1971, MERX-
MULLER 1972), but if one takes stock of the systems of classification
proposed in the last decade (cf. for examplc those of EHRENDORFER
1971, TAKHTAJAN 1973, and DAHLGREN 1974), it is quite clear that al-
though the proposers of these systems aim at a taxonomic arrangement
reflecting the phylogenetic origin of the various groups, which suggests
a good deal of original thinking, it appears that the current systems
are not basically different in approach and scope from several proposed
by about the turn of the century or perhaps not even from some devel-
oped over a century ago (at least not in principle). It is the purpose of
the present paper to confront the methodological basis of those systems
of classification of the Magnoliophyta or Angiosperms which have at
present the greatest number of adherents with evidence which may
yield alternative criteria for the definition of primitive or basic taxo-
nomic categories within the Flowering Plants. Certain "classical"
typological fundamentals developed in phytomorphology and secondarily
employed in taxonomic classification inevitably dictate certain sequential
arrangements of taxonomic groups at the family or ordinal level and thus
lay down an overall foundation for the ultimate classification of all
Angiosperms. As we have seen, readjustments emanating from the
application of additional criteria mostly result in some minor "shifts"
and rearrangements without changing the existing major framework
of the systems in question. There is absolutely no point in discussing
Delimitation of the Major Taxa of the Higher Oyeadophytina 15

the application of unconventional criteria in the present context with-

out a previous analysis of the above-mentioned, essentially typological
tenets.
Systematists professing to aim (or claiming to have arrived) at a
phylogenetic system have subjectively made a principal, aprioristic
assumption regarding the more basic (= primitive) versus the more
secondary (= advanced) groups and the corresponding morphological
characteristics, and they started their classification from there. The
taxonomic features of the group accepted as the relatively most prim-
itive of all were in turn used in phylogenetic speculations regarding the
ancestry and the palaeoecology of the assumed basic taxon 1.
Gradually a number of primary and secondary postulates emerged
which were partly suffused with other doctrines such as the notions
of gymnospermy versus angiospermy, and the "sporophyll" concept.
The present author has repeatedly criticised this narrow point of view
(see MEEUSE 1974a, 1975b, 1976) and singled out the most relevant
basic postulates:
1. Flowering Plants are monophyletic.
2. All Flowering Plants bear "flowers" (or rather: functional re-
productive units) of one kind only; these conventional flowers are all
supposed to be uniaxial structures, basically corresponding (homologous)
with a leafy shoot.
3. Primitive magnoliophytic forms, both the extinct and the surviving
recent representatives, were (are) phaneranthous and entomophilous.
4. All angiospermous flowers are derived from the phaneranthous

1 This can be substantiated by two recent examples. In an introduction

to a symposium WALKER (1975: 515) explained that the foundations of
angiosperm phylogeny were to be discussed on the basis of the TAKHTAJAN
and CRONQUIST systems of classification which systems were chosen
"because they are the most thoroughly documented ... sys-
tems ... which have taken into consideration data from all
bases of angiosperm phylogeny".
The other example is also in a paper by WALKER (1976: 291) in which
he, when discussing early angiospermoid pollen types, rejects the con-
clusion drawn by KUPRIANOVA (1967) concerning the occurrence in the
Lower to Middle Cretaceous of plants closely allied to the recent Ohlorantha-
eeae on the ground that at that time such an already "advanced" family
could not yet have evolved ("if paleobotany is to tell us anything
about the origin of angiosperms pollen similar to that of
Magnolia, Degeneria, and Anaxagorea, not Asearina, must be
sought in the fossil record"). DOYLE et al. (1975: 451), on the other
hand, do not reject a relation between Barremian or Aptian Olavatipol-
lenites and pollen grains of recent taxa including Ohloranthaeeae (and
Myristieaeeae, also already advanced in the sense of WALKER), and thus
(implicitly) accept a possible taxonomic (and morphological) affinity be-
tween the latter and certain Cretaceous taxa.
16 A. D. J. MEEUSE:

archetype and this tenet in turn serves as a yardstick for the assessment
of the degree of evolutionary advancement of all other kinds of so-
called flowers as more or less modified derivatives of that preselected,
and supposedly archaic prototype of all "flowers"; the corresponding
taxa are subsequently arranged according to this assessment 8:s not,
or hardly, to extremely advanced descendants of the group with the
most basic floral morphology.
5. Flowering Plants have foliar carpels enclosing the ovules (this
condition is called angiospermy or angioody), and they differ in this
respect appreciably and fundamentally,from the gymnosperms.
The ensuing system of classification is rigid in that only one-direc-
tional sequences can be construed, all based on, and radiating from the
same primary group. It follows that taxa with simply constructed
functional reproductive units (FRUs = conventional "flowers"), such
as most of the so-called apetalous ones, are of necessity placed at the
end of such sequences as much derived (or depauperated, advanced,
or reduced but in any case secondarily originated) taxa, e.g., amenti-
ferous, cyperaceous, and salicaceous forms. (It must be pointed out
here that the opposite starting point forms the basis of the Englerian
and Wettsteinian systems in which priority of place is given to mono-
chlamydeous dicotyledons, and a more advanced status to phaneranthous
orders, but, mutatis mutandis, the same reasoning is followed although
in a more or less reverse order; however, the situation is more compli-
cated owing to a discrepancy in ENGLER'S views on Angiosperm phy-
logeny and his "system": see MEEUSE 1972.)
Each of the above-mentioned five postulates has been subjected
to severe criticism by the present author. Alternative assumptions
can be formulated as follows:
As opposed to (1): Flowering Plants are pleiophyletic in so far
that several evolutionary lineages leading to recent magnoliophytic
groups had already become segregated before their common progenitorial
taxon, from which such a group has descended, had completely attained
the level of advancement of a truly angiospermous plant form.
As opposed to (2): the FRUs of the Flowering Plants are all derived
from an ancestral, complex and pluriaxial structure (anthocorm), but
represent either a condensed, 'whole anthocorm, or only a subordinate
part of it, so that the conventional category of the flower is hetero-
geneous (MEEUSE 1975a); anthocorms and subordinate parts of antho-
corms are, moreover, Imi generis in respect of a leafy shoot and its
appendages.
As opposed to (3): primitive magnoliophytic groups (and their
immediate precursors) were aphananthous and anemophilous or at
best incipiently entolJlophilous; dicliny prevailed, incipient monocliny
Delimitation of the Major Taxa of the Higher Oycadophytina 17

only originating in those lineages which ultimately developed into

predominantly phaneranthous and zoophilous forms (MEEUSE 1976).
As opposed to (4): angiospermous FRUs reflect an early divergent
evolution and can be used as a yardstick to trace "impossible'" and
unacceptable derivations, in conjunction with the assessment of the
phylogenetic origin and the morphological status of the androecial
members (MEEUSE 1974)-as an example, the derivation of hamamelidid
taxa from a "ranalean" (= in practice a promagnolialean or even mag-
nolialean) progenitor, or vice versa, must be rejected as highly improbable
and rather be explained as the result of a parallel, independent evolution
of fairly long duration as accepted sub (1).
As opposed to (5): the "carpels" (monogyna in a more neutral
terminology) and most of the other floral appendages are considered
to be sui generis in respect of the functional leaves or trophophylls;
the genitalia (and most intrafloral semaphylls) are direct derivatives
of ovuliferous cupules and androsynangia-bearing organs of the pterido-
spermous and more advanced cycadophytinous gymnosperms and thus
link the Magnoliophyta with the latter, whilst still serving as a yard-
stick for the assessment of the relative degree of phylogenetic advance-
ment of the sex organs of recent angiospermous groups.
It is quite clear that, as was already pointed out before, there is
no reason to introduce other than morphological criteria before the
fundamental controversies are understood, and before one has agreed
upon the principles to be accepted as a starting point for the basic
framework of a classification.
According to the present author the Magnoliophyta are much more
heterogeneous than is generally assumed. Some recent summaries (see,
e.g., BATE-SMITH 1972, MERXMULLER 1972, PHILIPSON 1974), some
primarily based on morphological evidence, also suggest a more pro-
nounced intranscendence of the major groups of Angiosperms than is
generally accepted. This is in good agreement with certain indications
from phytochemical, palynological, anthecological, embryological, and
ontogenetic studies, which will not be discussed here in detail. Such
surveys also emphasize the great error repeated since at least a hundred
years ago of lumping (nearly) all sympetalous dicots in what is now
mostly called the "Asteridae". This group is so heterogeneous, and the
character of sympetaly has so frequently been mentioned as an example
of convergent evolution, that it is absurd to retain the artificial assembly
of the "Asteridae". This aggregate must be split up and referred to sev-
eral of the principal subordinate taxa of the dicots: Campanulales
and Asterales most probably belong, together with Araliales s.s. (Apiales
s.s., i.e., without Cornales and related families containing iridoid com-
pounds), to a lineage which also includes the ranalean and sapindalean-
Plant Srst. Evo\., Supp\. 1
18 A. D. J. MEEUSE:

rutalean orders; a rosoid-saxifragoid-cornoid affinity may be accepted

f~r tubiflorous (lamialean) groups; the Ericale8 may possibly have a
dilleniid origin, etc. Such a heterogeneity is also found among the
monocotyledons or Liliatae (compare, e.g., MEEUSE 1975c). Helobial
groups (Alismatales, etc.) appear to be rather isolated, and the lilialean
assembly is not necessarily basic in respect of Poales, Oyperales, Pan-
danales, Arecales, Zingiberales, and Arales. A long phylogenetic history
is more plausible than a monophyletic derivation from a single basic
group of protomonocots, which in its tum, as several workers have it,
is descended from a dicotyledonoid (nymphaeoid, ranalean-ranunculid,
or ranalean-piperid) progenitorial taxon.
It is quite clear that such views necessitate a reconsideration of
the starting point of a classification. One of the conclusions may be a
different dividing line between the principal subordinate major groups,
which segregates the non-ranalean orders from an aggregate of all
monocotyledonous and ranalean taxa, as strongly intimated by HUBER
in this symposium. For this reason the fundamental morphological
features and the associated evolutionary trends derivable from the
occurrence of such features ought to be re-studied and re-assessed more
thoroughly. It is possible that, especially as far as the phylogenetic
aspects are concerned, additional criteria may need to be used in this
connection. Certain fossil finds discussed by the present author (MEEUSlt
1976: 92-94) may, for instance, point to the more probable, early floral
architecture in protangiospermous forms.
References
BATE-SMITH, E. C., 1972: Chemistry and phylogeny of the Angiosperms.
Nature 236, 353-354.
BEHNKE, H.-D., 1972: Sieve-tube plastids in relation to Angiosperm sys-
tematics-an attempt towards a classification by ultrastructural ana-
lysis. Bot. Rev. 38, 155-197.
BOUMAN, F., 1974: Developmental studies of the ovule, integuments and
seed in some Angiosperms. Thesis, Univ. of Amsterdam. Naarden.
DAHLGREN, R., 1974: Angiospermernes taxonomi. Vol. 1. Copenhagen.
- 1975: Current Topics. The distribution of characters within an Angio-
sperm System. 1. Some embryological characters. Bot. Notiser 128,
181-197.
DICKISON, W. C" 1975: The bases of Angiosperm phylogeny: ve~etative
anatomy. Ann. Missouri Bot. Gdn. 62, 590-620.
DoYLE, J. A., VAN CAMPO, M., and LUGARDON, B., 1975: Observations
on exine structure of Eucommiidites and Lower Cretaceous Angiosperm
pollen. Pollen & Spores 17, 429~486.
EHRENDORFER, F., 1971: Systematik und Evolution. In: Lehrbuch der
Botanik fUr Hochschulen (STRASBURGER, E., et al.), 30. Aufl., p. 379-741.
Stuttgart: G. Fischer.
EYDE, R. H., 1975: The bases of Angiosperm phylogeny: floral anatomy.
Ann. Missouri Bot. Gard. 62, 521-537.
Delimitation of the Major Taxa of the Higher Cycadophytina 19

JENSEN, S. R., NIELSEN, B. J., and DAHLGREN, R., 1975: Iridoid compounds,
their occurrence and systematic importance in the Angiosperms. Bot.
Notiser 128, 148-180..
JOHRI, B. M., 1963: Embryology and taxonomy. In: Recent advances in
the embryology of Angiosperms (MAHESHWARI, P., Ed.), 395-444.
Delhi.
KUBITZKI, K., 1969: Chemosystematische Betrachtungen zur Gro13glie-
derung der Dicotylen. Taxon 18, 360-368.
- 1973: Probleme der Gro13systematik der Blutenpflanzen. Ber. dtsch.
bot. Ges. 85, 259-277 (1972).
KUPRIANOVA, L. A., 1967: Palynological data for the history of the Chlor-
anthaceae. Pollen & Spores 9, 95-10.0..
LEINS, P., 1971: Das Androeceum der Dikotylen. Ber. dtsch. bot. Ges.
84,191-193.
MAHESHWARI, P., 1964: Embryology in relation to taxonomy. In: Vistas
in Botany (TURRILL, W. B., Ed.), 4, 55-97. Oxford: Pergamon Press.
MEEUSE, A. D. J., 1970.: The descent of the Flowering Plants in the light
of new evidence from phytochemistry and from other sources. Acta
Bot. Neerl. 19, 61-72, 133-140..
1972: Sixty-five years of theories of the multiaxial flower. Acta Bio-
theor. 21, 167-202.
1974: Some fundamental principles in interpretative floral morphology.
In: Vistas in Plant Sciences (VARGHESE, T. M., Ed.), Vol. 1, 1-78.
Hissar.
1975 a: Changing floral concepts: Anthocorms, flowers, and anthoids.
Acta Bot. Neerl. 24, 25-36.
1975b (1974): Floral evolution and emended Anthocorm Theory. In:
Intern. Bioscience Monogr. (VARGHESE, T. M., Ed.), Vol. 1. Hissar.
- 1975c: Aspects of the evolution of the Monocotyledons. Acta Bot.
Neerl. 24, 421-436.
1976: Fundamental aspects of evolution of the Magnoliophyta. In:
Glimpses in Plant Research (NAIR, P. K. K., Ed.), 3,82-10.0.. New Delhi:
Vikas.
MERXMULLER, H., 1972: Systematic Botany: an unachieved synthesis.
BioI. J. Linn. Soc. 4, 311-322.
PHILIPSON, W. R., 1974: Ovular morphology and the major classification
of the dicotyledons. Bot. J. Linn. Soc. 68, 89-10.9.
SPORNE, K. R., 1975: A note on ellagitannins as indicators of evolutionary
status in Dicotyledons. New Phytol. 75, 613-618.
TAKHTAJAN, A. L., 1973: Evolution und Ausbreitung der Blutenpflanzen.
Jena: G. Fischer.
WAGENITZ, G., 1975: Blutenreduktion als ein zentrales Problem der Angio-
spermen-Systematik. Bot. Jahrb. Syst. 96, 448-470..
WALKER, J. W., 1975: The bases of angiosperm phylogeny: Introduction.
Ann. Missouri Bot. Gdn. 62, 515-516.
1976: Comparative pollen morphology and phylogeny of the Ranalean
complex. In: Origin and early evolution of the Angiosperms (BECK,
C. B., Ed.), 241-299. New York: Columbia Univ. Press.

Address of the author: Prof. Dr. A. D. J. MEEUSE, Plantage Middenlaan

2A, Amsterdam, The Netherlands.

2*
Plant Syst. Evol., Supp!. 1, 21-31 (1977)
© by Springer-Verlag 1977

Institut fur Allgemeine Botanik und Botanischer Garten,

Universitat Hamburg, Federal Republic of Germany

Some Aspects of the Classification and Evolution

of Higher Taxa
By

K. Kubitzki, Hamburg

Abstract: It is shown that a fully phylogenetic classification of the

flowering plants, though not impossible in principle, is hard to attain at
present because of the absence of any precise cladistic information about
this plant group. The unwarranted confusion of grades and clades imposes
a strongly typological element upon current systems intended to be phylo-
genetic. A plea is made for a striving after an understanding of the func-
tional significance of characters which is indispensable for aiming at a
true understanding of evolutionary processes and pathways.

In the study of the taxonomy and the evolution of the flowering

plants the higher levels of the taxonomic hierarchy have been relatively
neglected for a long time. The activity of most taxonomists has been
concentrated mainly around the production and revision of regional
Floras, and though much effort has been devoted in the last century
towards the production of phylogenetic systems of classification of
flowering plants, these have often been no more than attempted phylo-
genetic rearrangements of established groups, rather than a continued
reassessment of the taxa themselves in the light of new evidence
(YOUNG & WATSON 1970). Nevertheless, botanists have always been
ready to assimilate, and to integrate into their classifications, information
from new sources, though mostly in a somewhat selective manner,
with the main emphasis on information from-at any given time-more
fashionable fields, as more recently from population biology, natural
product chemistry, and ultrastructural research. It is an interesting
feature that there have always been certain newer developments
which, at the time of their appearance, have been overestimated, with
the danger of becoming discredited early, the rise and fall of serology
during the first half of this century being but one example. Today,
however, it is hardly likely that any of our highly valued ancillary
disciplines will suffer the same fate.
22 K. KUBITZKI:

As to evolutionary studies, these have focused during the last three

decades or so mainly around the study of populations, though in this
field also, important conclusions have been reached as to evolutionary
pathways and patterns at and above the level of the species. As a result
of these efforts, the process of evolution at the population level is now
so well understood that the leaders of this field, such as G. L. STEBBINS
and E. MAYR, have already turned to the question of the evolutionary
mechanisms involved in the differentiation of higher taxa, and the
renewed interest in these aspects of flowering plant evolution and
classification may well be due to this attitude. At any rate, we see
the factual basis of current classifications being analyzed by a growing
number of systematists, and these activities have recently led to sym-
posia like that on "The monocotyledons: Their evolution and compara-
tive biology" (in "Quarterly Review of Biology, Vol. 48, No. 3/4, 1973),
"Chemistry in botanical classification" (edited by BENDZ and SANTES-
SON), or "The bases of angiosperm phylogeny (in "Ann. Missouri Bot.
Gard. Vol. 62, No.3, 1976), and now also to this symposium which,
contrary to those just mentioned, has been planned by its organizers
to embrace also the methodological problems involved in the study
of higher taxa.
The latter is an important aspect since the consumers of classifica-
tory systems like teachers and students, plant physiologists, chemists,
etc., mostly remain ignorant about the processes of classification and
system-making, being primarily interested in their results. Thus it is hardly
appreciated outside taxonomic circles (and often not even there!)
that the systematist when being faced with problematical choices all
too often has to opt for one out of several equally tenable solutions
concerning circumscription and arrangement of taxa; moreover, few
people are really aware of just how vacillating the phylogenetic basis
of recent systems of classification in fact is!
If one looks into the classificatory basis of flowering plant syste-
matics, first of all the observer is impressed by the vast amount of
information which has been accumulated by generations of botanists
and other workers, without being readily available to, or at least not
commonly used by, the taxonomist. There are various reasons why
the data worked out by different disciplines often do not reach those
concerned with the task of building up classifications. One must agree
with HEYWOOD (1973a) that the main, but still unachieved task of
systematics, besides of its identificatory role, is "to collate, synthesize
and order these data in a systematic fashion so that conclusions of all
sorts may be drawn". It is probably true that some sorts of data are
easier to handle, in this respect, than others; this may apply, e.g., to
embryological characteristics which always have been well summarized
Classification and Evolution of Higher Taxa 23

(see, e.g., DAHLGREN 1975a), and, because of their innate unambiguity,

this seems to be especially true for chemical data, as shown by some
recent compilations of the systematic distribution and importance of
xanthones (REZENDE & GOTTLIEB 1973), of benzyltetrahydroisochi-
nolines (REZENDE et al. 1975) and of iridoid compounds (JENSEN et al.
1975). There are few examples, however, in which for a larger group
the obtainable evidence from all fields has been collated though numerical
methods which are perhaps the most important methodological achieve-
ment of systematic biology in recent times, are available for analyzing
the similarities and differences of organisms upon which most classifi-
cations are based. There is no doubt that numerical methods become
the more indispensable, the larger is the size of the taxa. It must be
mentioned, however, that in practice we are still very far from making
use of the wealth of information already accumulated.

This has led to the impression that taxonomists, in building their

classifications, proceed in a highly selective manner. With regard to the
lower taxa, this is only seemingly true and is a result of the common practice
among taxonomists, who regard the taxonomic decisions to be more im-
portant than the information on which the decisions have been based,
so that, in communicating their results, they justify their decisions in .the
form of a key or a description, but do not worry about listing the informa-
tion available about the organism (HEYWOOD 1973). As to the higher cate-
gories, taxonomists are often assumed to select deliberately those facts
which fit best in their preconceived ideas, but I think there are also inherent
difficulties which contribute to narrow the number of usable characters.
These may be seen in how to treat characteristics of scattered occurrence,
how to formulate an average picture of a taxonomic group, which, in it-
self, embraces different anagenetic grades, and, above all, how to evaluate
and integrate features which undoubtly have come into being as the result
of parallelism and convergence. It is because of this, as my colleague Dr.
HUBER informs me, that for the building up of an evolutionary classification
of the monocotyledons, e.g., only a surprizingly small number of characters
is suitable.

Mentioning of parallelism and convergence leads us to an inspection

of the evolutionary components of classifications since, if there were
divergence only, but no convergence, a purely phenetically based clas-
sification would by itself be a phylogenetic one. As PHILIPSON (1961)
has stressed the probabilities of the occurrence of convergence among
the flowering plants have been still underestimated so that many taxa
of any rank, at closer inspection, will be revealed as unnatural grade
groups. The question arises then if an evolutionary system of classifi-
cation is possible at all and if so, how to achieve it. If we are to construct
a system which intends to reflect the development of the plant kingdom
or a part of it in time and space, in other words if we are going to try
24 K. KUBITZKI:

to elucidate phylogeny (the origin and evolution of taxonomic groups),

and if such an attempt is based on knowledge of present-day forms alone,
then the answer must clearly be, that this is an virtually unachievable
goal. In view of the still meagre palaeontological evidence, a recon-
struction of the interrelationships between all kinds of flowering plants,
fossil and Recent, is for the present an unattainable task, though HUGHES
(1976), in his recent book, shows that it is possible, in principle, even if
such an attempt would enforce a profound change in methodology.
As STEBBINS (1974) in particular has pointed out, the common ancestors
of all modern flowering plants are completely extinct, and from all
modern groups of plants and animals that possess a reasonably good
fossil record it can be learned that the nature of the ancestor,; of ancient
and widespread modern classes can never be reconstructed with certainty
by putting together traits recognized as primitive which occur scattered
among modern forms 1. Nevertheless, continued efforts have been
made to deduce phylogeny in this strict sense by directly relying on
modern forms, though these have been condemned as "pseudophy-
logeny" by spokesmen of phylogenetic classification, as LAM and
ZIMMERMANN.
Widely prevalent, however, is a much more loose concept of phylo-
geny, as, for instance, is nicely expressed by JOHNSON (1972), who
feels phylogeny to imply primarily "retrospective predictivity; that
is, it predicts what we may hope to find out, in the future, about the
past-and thus 'explain' the present!" Nevertheless, we have avoided,
in the title of this symposium, the self-committing term "phylogeny"
and have preferred, to speak instead, of "evolutionary systems of classifi-
cation". I am aware that there are also strictly phenetic classificatory
approaches, like the notable one of YOUNG & WATSON (1970); yet
it is hard to see why at least an attempted evolutionary interpretation
of an otherwise perhaps more or less strictly phenetically based clas-
sification should not be allowed for.
I would now like to try to analyse the phylogenetic/evolutionary
features which current systems of classification do display. As to the circum-
scription of the families and, to some degree, also to the orders of the
flowering plants, and to their primary division in di- and monocotyledons,
a rather broad agreement has been reached (for a contrasting view, see
HUBER in this volume). The mutual relationships of the groups, however,

1 It should be noted here that reliance on primitive characters is strictly

rejected by the practitioners of phylogenetic systematics in zoology;
advanced characters are l\sed instead, though this method does not aim
so much for the recognition of the common· ancestor but for establishing
sister group relationships.
Classification and Evolution of Higher Taxa 25

are judged to be rather controversial, and their arrangement relative to

one another and their sequence have been dealt with quite differently
by various authors. If, however, the sequence and position of contem-
poraneous groups relative to each other (without practically any knowl-
edge of their cladistic relationship) really is the main phylogeneticj
evolutionary content of such classifications, then HEYWOOD (1973 b)
is right in considering such a "construction of phylogenetic schemes
of the angiosperm(s) ... a much overrated pastime which has been
pursued far too unscientifically for far too long". Furthermore, neither
the rank of the taxa, which of course is mainly determined by their
phenetic distance, nor the "box-in-box structure" of the system (formerly
often considered to be a proof a phylogeny) are very meaningful phylo-
genetic components. The main evolutionary element which influences
taxonomic arrangements may well be the significance of morphological
characters (in their broadest sense) in terms of primitiveness and advance-
ment, and the determination of their trends which lead to the establish-
ment of so-called "Merkmalsphylogenien" or "semophyleses". More-
over, there has been the recognition of genetic and ecological trends
(e.g., allogamy -,. autogamy) and of certain syndromes which also tend
to follow each other in regular successions (e.g., different modes of zoo-
gamy; zoogamy -+ anemogamy, etc.). In the last cases in which ec:ologi-
cally important traits are composed of several single features which
must cooperate harmoniously, their functional importance cannot be
overlooked; I doubt, however, if such always exists and can be indicated
with regard to isolated features. This is worth mentioning because
it is universally agreed that evolution is accompanied by a constant
improvement of adaptations at all levels, from which follows that the
primitive should be less well adapted than the advanced. One objec-
tion to this is the notion, put forward especially by STEBBINS (1974),
that it is normally not the single character but rather an adaptive
complex of several characters which is adaptive. In other words, we
mostly ignore completely what a character which may be important
to us (mainly for attempting to elucidate phylogeny) might signify
for the plant (see CROWSON 1970, KUBITZKI 1975). Because of this,
and because of the impossibility to test semophyleses along with palae-
ontological evidence, these character phylogenies tend to include a
strongly typological element. I would like to conclude that functional
improvement, as a concomitant of a certain evolutionary trend of a
single feature, can only be taken for granted if it is either self-evident
(as perhaps in the case of BAILEY'S sequence from tracheids to vessels
with simple perforation), if the anagenetic improvement has been
tested experimentally, or if this can be concluded on the basis of experi-
ments by analogy.
26 K. KUBITZKI:

These limitations do not at all minimize the value of the functional

point of view which has the advantage of leading away from typology
towards an understanding of adaptations. In this context, it is rather con·
fusing that flower characters, which have been considered for a long time
to be especially conservative and hence have been used from the very
beginnings of plant taxonomy as the major yardstick for tracing affinities,
have now become known to be extremely plastic in the face of the plant's
needs to maintain, or to improve, an effective pollination system. It is
because of this that biologists like CARLQUIST (1969) have claimed a mainly,
if not exclusively, functional interpretation of flower structures, and more
recently (CARLQUIST 1975) also of wood structure, claims which, however,
appear to me somewhat one·sided, because they overlook the fact that
every organism, past and present, besides the necessity of having been
or being adapted to its environment, has been or is also the carrier of a
phylogenetic load.

If we now turn more specifically to the higher taxa of the angiosperms,

it must be emphasized that delimiting and describing a taxon of higher
rank is equally an hypothesis as is the description of, say, a species. In
the latter case, the hypothesis is that there will be found in nature
thousands of individuals similar to those forming the limited sample
the description is based upon (HEYWOOD 1973). And while the testing
of this hypothesis in the case of the species is done by countless checking
and rechecking of other individuals in the field, herbarium and labora-
tory by later observers, in the case of higher categories the test will
be to find out how far the classification is compatible with evidence
which had not originally been used in the construction of the group.
In this context it would be interesting to look into the different
major subdivisions of the flowering plants which have been proposed
more recently. While TAKHTAJAN (1959) divided the dicotyledons
into 13 superorders, the monocotyledons into 5, in 1964 he proposed
a scheme with 6 subclasses in the dicotyledons and 5 in the mono-
cotyledons. This major classification has also been adopted, with slight
modifications, by CRONQUIST (1968), while THORNE (1968) continued
with 20 superorders in the dicotyledons and 5 in the monocotyledons,
and DAHLGREN (1975) with not less than 27 and 7 superorders, respec-
tively. To me there is no doubt that the lower number of taxa used for
the primary subdivision of dicotyledons and monocotyledons, while
not being beyond the grasp of the consumers, has, together with the
high didactic standard of TAKHTAJAN'S (1959, 1969, 1973) and CRON-
QUIST'S (1968) books, been one of the main reasons for the broad accep-
tance of these schemes.
Yet it is generally recognized that taxa of lower rank can
be defined much more precisely than those of higher rank; the
level of superorder is probably the last one in which a somewhat
meaningful diagnosis can be given; the su bc1asses, instead, can
Cfassification and Evolution of Higher Taxa 27

only be characterized by circumscribing them and by indicating some

general tendencies. While some of these taxa, like the Magnoliidae, and
to a lesser degree also the Asteridae, might represent fairly correctly
circumscribed taxonomic and evolutionary units, this may be doubted
for others. In another occasion (KUBITZKI 1969) I have pointed to the
somewhat vacillating value of the Rosidae and the Dilleniidae; and the
Hamamelididae, though constantly becoming smaller· and smaller
(cf. THORNE 1973), still give the impression of a vanishing grade group.
Things may become clearer by asking what is known or, what
can be inferred, about the mode of origin of the higher taxa of the flower-
ing plants. The idea of special mechanisms of macroevolution being
involved, as opposed to those of microevolution, has long been abandon~d.
More recently it has also become clear that every morphological feature
that distinguishes families and orders can in some groups vary at the
level of genera and species (STEBBINS 1974). In spite of this obstacle,
STEBBINS (1951), in a ingenious approach, has been able to demon-
strate that the combination of characteristics, found in the larger, more
successful families of the flowering plants must have evolved through
the guidance of natural selection because these families therewith
acquired particularly successful methods of solving the problems of
fertilization and seed dispersal. Less is known and can be expected to
having been operative as a guiding principle in the evolution at still
higher levels; so that the ideas presented about the respective importance
of secondary plant substances (see CRONQUIST and GOTTLIEB in this
volume) deserve especial attention.
A basic tenet of most of these considerations is the tacitly made assump-
tion of a reasonable monophyly of the higher taxa which is inferred from
the phenetic evidence. The future, however, will hold many surprises to
us, and with regard to several subclasses, we are already becoming aware
that monophylesis is questionable, at least. One example is the Asteridae, the
homogeneity of which has been challenged on the basis of leak-morpho-
logical studies (HICKEY & WOLFE 1976); another, where strong doubts as
to its evolutionary coherence seem to be justified is the Caryophyllidae.
While the core of this group, as characterized by its peculiar pigments,
sieve-tube leucoplasts, its deviating basic chromosome number x = 9
(RAVEN 1976) and other features must have a common, and possibly ancient,
origin, it may well be quite unrelated to Polygonales and Plumbaginales;
and in DAHLGREN'S (197 5) scheme the latter orders appear quite distant
from Caryophyllales. To give allowance for some pleiophyly is probably
inevitable, rather than making hasty adjustments of our classifications
to the stepwise progress of our knowledge, provided that resulting groupings
can clearly be defined phenetically.
A question that deserves special mentioning is that for the mono-
phyly of the angiosperms as a whole. The reasons for supporting this
hypothesis (summarized, e.g., by TAKHTAJAN 1959) are so well-known
28 K. KUBITZKI:

that they need not be repeated here. The classical argument of the uni-
formity of the embryo sac and of double fertilization has begun to loose
its persuasive power (MEEUSE 1964); and the constancy of the position
relative to each other of micro- and megasporangia in the flower may
well be related to its function in animal pollination, instead of signalizing
a common origin. On the other hand, it is clear that the older pleio-
phyletic hypotheses of GREGUSS (see, e.g., 1964) and EMBERGER (1960)
do not convince us any longer, because the former was based on rather
superficial traits, while the latter inferred pleiophyly from nothing
but the basic heterogeneity of present-day angiosperms. In more recent
times, several workers (see, e.g., MEEUSE 1971, PHILIPSON 1975) have
collated new evidence which supports the acceptance of rather different
evolutionary lines within the angiosperms. While PHILIPSON is cautious
about inferring pleiophyly from this, MEEUSE is convinced that this
heterogeneity clearly favours the acceptance of a pleiophyletic origin
of the flowering plants. Nobody, however, is at present able to indicate
where these different lines start, nor where they, if at all, converge,
and it is merely in the absence of a better alternative that the origin
of the flowering plants is postulated from one group of the seed ferns.
In the absence of any cladistic information, a decision between mono- and
pleiophyly seems to me at present virtually impossible.
The question of the mutual interrelationships between the major
groups of contemporary angiosperms is equally one of the more traumatic
matters in our field, and I shall only briefly comment on this here.
The question which subclass may be the key group of, and ancestral to,
the other subclasses of the flowering plants is becoming less important
as we are going to become aware of the complete disappearance of the
common ancestor(s) of that group. On the one hand, living groups of
higher rank can neither be ancestral to other living ones, nor can the
nature of this ancestor be inferred with certainty from comparisons
among Recent groups, as mentioned above. On the other hand, the differ-
ent subclasses of the angiosperms have attained different anagenetic
grades; the M agnoliidae representing the lowest, the Asteridae the
highest level of specialisation among the dicotyledons (with many
exceptions in both groups); and in monocotyledons, the situation is
similar, whereas WALKER'S (1975, Fig. 4) recent scheme is rather mis-
leading in this respect since all monocotyledons are confined in it to the
lowest grade. The question is then, are subsequent levels of specialisa-
tion by themselves indicative of a phylogenetic relationship between
major groups 'I If we put the question in this way the answer can only
be in the negative. I am aware that the Magnoliidae is the group with
the greatest concentration of known primitive characters, to which
constantly more cryptic new ones are added, as the archaic seed structure
Classification and Evolution of Higher Taxa 29

in Myristica and Annonaceae (CORNER 1976); the presumably primarily

columella-less exine in forms like Degeneria, Eupomatia, certain Nym-
pooeaceae and A nnonaceae (WALKER & SKVARLA 1975); and the' occur-
rence of the diarylpropanoid viranolol in V irola, one of the simplest
flavanoids so far discovered (BRAZ FrLHO et al. 1973). Primitiveness by
itself, however, cannot be the proof of the Magnoliidae being ancestral
to other subclasses, inasmuch as they occupy, as I have pointed out
earlier (KUBITZKI 1969, 1973), a phenetic ally isolated position 2. The
group for which a relationship with the Magnoliidae can be vizualized
is in my mind, the monocotyledons (cf. also HUBER, this volume), though
two of the most distinctive features of the Ranalean complex are even
absent from them: namely multistaminate flowers with a spiral phyllo-
taxis ("primary polyandry"), and the occurrence of benzyltetrahydro-
isochinoline alkaloids whose distribution and systematic importance
recently has been dealt with by REZENDE et al. (1975). If we are going
to postulate links between major extant groups we should be anxious
to ensure that they fulfill this role not only with regard to isolated
traits (as, e.g., Trochodendron with regard to supposed preadaptation
to anemogamy), but with regard to an overall similarity. Otherwise,
in our arrangements typological elements will largely continue to prevail.
To overcome these difficulties there is one way for elucidating
relationships between extant groups which, though it has been widely
used is far from being exhausted. This is, neither to use average para-
meters for major groups which in themselves often include a marked
gradation from primitive to advanced states, nor to consider evolutionary
tendencies (often leading to convergent, or at best homoiologous develop-
ments), but, as MARKGRAF (1955) has pointed out, to search in each
group for primitive characteristics often only present in one or a few
out of many genera which may be the clue for the understanding of
the whole complex. It is perhaps needless to say that this is possible
only in connection with monographic studies which have become so
unfashionable. By this way, however, the systematic arrangement
of major groups may still be improved largely. One other important
aspect of systematics is not so much concerned with classification itself
but with the processes and mechanisms of evolution which have led
to the present-day end products of evolution. If one asks which aspect
should have priority in future research the answer can only be that we

2 Though I have been rather misunderstood by several authors, since

I have neither "contested on chemical grounds ... the basal position of
the Ranalean complex" (HEYWOOD 1973b: 369), nor have I stated that
"modern Magnoliidae are advanced rather than primitive in chemical
features" (CRONQUIST 1976: 17)!
30 K. KUBITZKI:

have an obligation to attempt both approaches: it appears that not

much time is left for the study of diversity because the early destruction
of several of the earth's most fascinating biomes seems inevitable.

References
BENDZ, G., and SANTESSON, J. (Eds.), 1973: Chemistry in Botanical Classi-
fication. Nobel Symposium 25. New York-London: Academic Press.
BRAZ FILHO, R., FROTA LEITE, M. F., and GOTTLIEB, O. R., 1973: Con-
stitutions of diarylpropanoids from Virola multinervia. Phytochemistry
12,417-419.
CARLQUIST, S., 1969: Toward acceptable evolutionary interpretation of
floral anatomy. Phytomorphology 19, 332-362.
- 1975: Ecological Strategies of Xylem Evolution. Berkeley : University
of California Press.
CORNER, E. J. H., 1976: The Seeds of Dicotyledons. Vol. I, II. Cambridge:
Cambridge University Press. .
CRONQUIST, A., 1968: The Evolution and Classification of Flowering Plants.
London: Nelson.
- 1976: The taxonomic significance of the structure of plant proteins:
A classical taxonomist's view. Brittonia 28, 1-27.
CROWSON, R. A., 1970: Classification and Biology. London: Heinemann.
DAHLGREN, R., 1975: A system of classification of the angiosperms to be
used to demonstrate the distribution of characters. Bot. Notiser 128,
119-197.
1975a: The distribution of characters within an angiosperm system.
1. Some embryological characters. Bot. Notiser 128, 181-197.
DAVIS, P. H., and HEYWOOD, V. H., 1963: Principles of Angiosperm Taxon-
omy. Edinburgh and London: Oliver and Boyd.
EMBERGER, L., 1960: Les Vegeteaux Vasculaires. In: Traite de Botanique
(Systematique), Tome II, Fascicule I, II. (CHADEFAUD, M. et EMBERGER,
L.), Paris: Masson et Cie.
GREGUSS, P., 1964: The phylogeny of sexuality and triphyletic evolution
of the land plants. Acta BioI. Szeged. 10, 3-51.
HEYWOOD, V. H., 1973: Ecological data in practical taxonomy. In: Taxon-
omy and Ecology (HEYWOOD, V. H., Ed.), 329-347. London-New York:
Academic Press.
1973a: Taxonomy in crisis? Or taxonomy is the digestive system m
biology. Acta Bot. Acad. Sc. Hung. 19, 139-146.
1973 b: The role of chemistry in plant systematics. In: Chemistry in
Evolution and Systematics (SWAIN, T., Ed.), 355-375. London: Butter-
worths.
HICKEY, L. H., and WOLFE, J. A., 1976: The bases of angiosperm phylo-
geny: Vegetative morphology. Ann. Missouri Bot. Gard. 62, 538-589
( 1975).
HUGHES, N. F., 1976: Palaeobiology of Angiosperm Origins. Cambridge,
London, etc.: Cambridge University Press.
JENSEN, S. R., NIELSEN, B. J., and DAHLGREN, R., 1975: Iridoid compounds,
their occurrence, and systematic importance in angiosperms. Bot.
Notiser 128, 148-180.
•J OHNSON, L. A. S., 1972: Evolution and classification in Eucalyptus. Proc.
Liim. Soc. N.S. Wales 97, 11-29.
Classification and Evolution of Higher Taxa 31

KUBITZKI, K., 1969: Chemosystematische Betrachtungen zur GroB-

gliederung der Dicotylen. Taxon 18, 360-368.
1973: Probleme der GroBgliederung der Bliitenpflanzen. Ber. dtsch.
bot. Ges. 85, 259-277 (1972).
1975: Relationships between distribution and evolution in some hetero-
bathmic tropical groups. Bot. Jahrb. 96, 212-230.
MARKGRAF, F., 1955: Uber neuere Pflanzensysteme. Ber. dtsch. bot. Ges.
67, 20-22 (1954).
MEEUSE, A. D. J., 1964: Some phylogenetic aspects of the process of double
fertilization. Phytomorphology 13, 237-244.
1970: The descent of the flowering plants in the light of new evidence
from phytochemistry and from other sources. Acta Bot. Neerl. 19,
61-72, 133-140.
PHILIPSON, W. R., 1961: Relationship and convergence in angiosperms.
Phytomorphology 10, 367-376.
- 1975: Evolutionary lines within the dicotyledons. New Zealand J. Bot.
13, 73-91.
RAVEN, P. H., 1976: The bases of angiosperm phylogeny: Cytology. Ann.
Missouri Bot. Gard. 62, 724-764 (1975).
REZENDE, C. M. A. da M., and GOTTLIEB, O. R., 1973: Xanthones as sys-
tematic markers. Biochem. Syst. 1, 111-118.
- , - and MARX, M. C., 1975: Benzyltetrahydroisochinoline-derived al-
kaloids as systematic markers. Biochem. Syst. Ecol. 3, 63-70.
STEBBINS, G. L., 1951: Natural selection and the differentiation of angio-
sperm families. Evolution 5, 2,99-324.
- 1974: Flowering Plants. Evolution Above the Species Level. London:
E. Arnold.
TAKHTAJAN, A., 1959: Die Evolution der' Angiospermen. Jena: Fischer.
1964: The taxa of the higher plants above the rank of order. Taxon 13,
160-164.
1969: Flowering Plants. Origin and Dispersal. Edinburgh: Oliver and
Boyd.
1973: Evolution und Ausbreitung der Bliitenpflanzen. Stuttgart:
Fischer.
THORNE, R. F., 1968: Synopsis of a putatively phylogenetic classification
of the flowering plants. Aliso 6, 57-66.
- 1973: The "Amentiferae" or Hamamelidae as an artificial group: A
summary statement. Brittonia 25, 395-405.
WALKER, J. W., 1976: Evolutionary significance of the exine in the pollen
of primitive angiosperms. In': The Evolutionary Significance of the
Exine (FERGUSON, I. K., and MULLER, J., Eds.), 251-308. Linn. Soc.
Symp. Ser. No. 1.
and SKVARLA, J. J., 1975: Primitively columellaless pollen: a new concept
in the evolutionary morphology of angiosperms. Science 187, 445-447.
YOUNG, D. J., and WATSON, L., 1970: The classification of the dicotyledons:
A study of the upper levels of the hierarchy. Aust. J. Bot. 18, 387-433.

Address of the author: Prof. Dr. KLAUS KUBITZKI, JungiusstraJ3e 6-8,

D-2000 Hamburg 36, Federal Republic of Germany.
Plant Syst. Evol., Suppl. 1, 33-51 (1977)
© by Springer-Verlag 1977

The Botany School, University of Cambridge,

Cambridge, England

Some Problems Associated With Character Correlations

By

K. R. Sporne, Cambridge, England

Abstract: Among dicotyledons, 122 positive correlations at the 50: 1

level of significance (and 170 at the 20: 1 level) occur between twenty-six
characters, some of which are morphological, some anatomical and some
biochemical. Most of these characters are more abundant among families
known to have appeared early in the fossil record than they are among
those which appeared later. These facts have been used to discover which
families are primitive and which advanced, in a way which minimizes
subjective judgements. However, such judgements have not been com-
pletely eliminated; they have to be applied at all stages.
Decisions are necessary as to: which taxonomic scheme to use; which
level of taxon to take as the statistical unit; how to treat incomplete data
and "mixed taxa" ; whether correlated characters are functionally associated
and concerned in relative efficiency, or whether they are indicators of
relative advancement, having been involved in evolutionary trends; whether
all characters should be given equal weight.

At least 125 years have elapsed since arguments first started as to

the course of evolution among flowering plants; and they continue still.
The discussion has mainly concerned the nature of the most primitive
members of the group, one school of thought favouring the catkin-
bearing plants, while others favour ranalian types. Yet others, believing
that the angiosperms are polyphyletic, hold that several quite different
kinds of flowering plants could be equally primitive.
Until the early years of the present century, there seemed to be little
hope of settling such arguments, for the fossil record of flowering plants
was thought to give little or no help, and the various protagonists relied
almost entirely on subjective judgements and circular arguments.
However, BAILEY and his colleagues (whose work has been summarized
elsewhere, by SPORNE 1976a) showed that among dicotyledons certain
characters are associated (i.e. they occur together more frequently than
they would if they were merely randomly distributed within the group).
Plant Srs!. Evo!.. Snpp!. 1 3
34 K. R. SPORNE:

Furthermore, since some of these characters were more abundant in plants

from Cretaceous and Tertiary deposits than among present-day dicotyle-
dons, they concluded that the associated characters are primitive ones.
Further development of these ideas, and their extension to include
wood characters, led to the introduction by CHALK (1937) of simple
statistical tests of significance. By means of 2 X 2 contingency tests,
values of X2 were calculated, from which could be judged the probability
that the observed frequencies of association might have occurred by
chance. It was thus possible to discover which characters are statisti-
cally associated (i.e. "correlated").
This important step forward encouraged the hope that future dis-
cussions of angiosperm phylogeny would be based on objective reasoning
and that the subjective element might be considerably reduced. Starting
in 1939, I have been applying essentially the same techniques as those
of CHALK to the widest possi ble range of dicotyledon characters; and
the work is still in progress. Data have been analysed concerning some
forty characters, of which twenty-six are significant indicators of evolu-
tionary status.
These characters are numbered from 1 to 26 in Table 1, where the
degree of correlation between them is indicated by various symbols.
Positive correlations are shown by "plus" signs, and negative ones by
"minus" signs. Bold symbols indicate correlations that are significant
at the 50: 1 level, and faint symbols those that are significant at a level
between 20: 1 and 50: 1. "Crosses" indicate associations that are meaning-
less (e.g. to show a positive correlation between the woody habit and the
possession of particular wood characters would be absurd since, by
definition,wood characters can only be present in woody plants). A "dot"
implies absence of correlation. Lines 27 and 28 indicate the extent to
which the twenty-six characters occur more (or less) often than one
would expect among those families of flowering plants that have been
identified in the fossil record. Line 29 indicates the extent to which they
occur more (or less) frequently in rain-forest floras than in the rest of
the world.
The information contained in Table 1 includes, and also supplements,
that which has been published elsewhere (SPORNE 1974, 1975, 1976a,
CHENERY & SPORNE 1976). The purpose of this article is to examine the
procedures that have been adopted during the assembling, analysis and
interpretation of the data, the problems that have arisen and the extent
to which subjective judgements have had to be made in their solution.
Bearing in mind the lack of agreement among botanists as to the way
in which monocotyledons stand in relation to dicotyledons, it was decided
to treat the two groups separately, beginning with the dicotyledons.
The monocotyledons were subsequently studied by a colleague, LOWE
Some Problems Associated With Character Correlations 35

(1961). However, STEBBINS (1951) in a similar survey, but of a restricted

range of characters, chose to deal with the angiosperms as a whole.
One then had to decide whether to work with species, genera or
families as the statistical units. Theoretically, species should be the
obvious choice, for their delimitation is traditionally supposed to involve
much less subjective judgement than doe!;>, that of larger taxa. Further-
more, there is much less variation in the expression of characters within a
species than there is within a genus or a family. However, there are
probably some 200,000 species of dicotyledons in the world. To take
the species as the statistical unit would, therefore, be quite impracti-
cable; and this was even more true in 1939 than it is nowadays, when
computers are widely available. But, even today, the species is still
impracticable because of inadequate data. Thus, embryology, the study
of which is tedious and time-consuming, is known for only a minute
proportion of the total number of species. The same is true for biochemi-
cal characters. It is most improbable, therefore, that an embryologist
working in India, for example, would choose to investigate the very
same species as a plant biochemist working in Denmark. This being so,
it would be impossible to discover whether embryological and bio-
chemical characters are correlated or not, so long as the species was
taken as the statistical unit.
YOUNG & WATSON (1970), interested more in taxonomy than in phylo·
geny, carried out an analysis of the occurrence of some eighty-three at-
tributes among genera of dicotyledons but, in spite of using computerized
methods, they felt obliged to restrict their survey to 543 genera. Their
selection was made, first of all, by choosing those families with the greatest
number of genera and species. From these they selected the largest genera,
the number of genera chosen to represent each family depending on its
size. The resulting list of 600 genera was then reduced to 543 by removing
those for which the data proved to be either inadequate or of doubtful
accuracy. While such selection methods are, doubtless, justifiable in a
taxonomic investigation, they would be inappropriate in a phylogenetic
one, where small genera might be just as important as large ones for showing
the course of evolution.

I decided to use the family as the statistical unit, but not without
considerable misgivings, for this procedure demanded that. one should
treat as equal families as different in size as Compositae and Adoxaceae.
The former contains some 19,000 species, while the latter consists of a
single species. In deciding which taxonomic scheme to use, the main
criterion was availability of data. Differing views as to the relationships
between families were, at this stage of the procedure, irrelevant. Bearing
in mind the vast amount of factual information contained in the two
editions of ENGLER & PRANTL (1887-1915 and 1924-), ENGLER'S
system, as presented by DIELS (1936), was an obvious choice.
36 K. R. SPORNE:

The 259 families recognized by DIELS were, accordingly, listed in

alphabetical order on recording sheets, (alphabetical order being chosen
so as to avoid, as far as possible, any personal bias during the process of
recording data). On one sheet would be recorded presence or absence of
character "X" (for example) and, on another, presence or absence of
character "Y". Then, by superimposing the two sheets and viewing
them by transmitted light, those families with both characters X and Y
would be immediately obvious and could be recorded on a third sheet.
Suppose that the number of families with character X is "x" and that
the number with character Y is "y". Then, on a "null hypothesis",
X'y
the expected number (m') with both would be 259' If the actual occur-
rence (m) is greater than m', then the characters X and Yare said to
be positively correlated; and, if less, then negatively correlated. The
greater the difference between m and m', the more highly significant
is the correlation. By calculating the value of X2, the value of "p" can
be determined from tables, where p represents the probability that the
difference between m and m' has occurred purely by chance. In most
of my published work, the value of p = 0.02 was set as the criterion
of significance. This is more rigorous than the value of p = 0.05, which
is. normally set in biological work, but it was felt that greater caution
was necessary, because so many of the data are in an unsatisfactory
state. In Table 1, the "bold" symbols represent those correlations
which meet this criterion. However, many of my colleagues have asked
to see the effect of taking the less rigorous criterion of p = 0.05. The
result has been to introduce all the "faint" symbols in Table 1.
The procedure outlined above seems quite straightforward and
simple but, in fact, frequently proves otherwise. Perhaps the most
disturbing complication arises from the existence of "mixed" families,
i.e. those families in which some genera, or species, may exhibit a partic-
ular character, while others do not. The recording sheets for such a
character will include three symbols: "+" against those families in
which it is constantly present, "-" against those from which it is
constantly absent, and" ±" against those which are mixed. One pro-
cedure is to treat those marked "±" as if they had been marked "+",
thereby producing a list of families which possess the character non-
exclusively ("n.e."). Another procedure is to treat those marked "±"
as if they had been marked "-", since some of their members lack the
character; those marked "+" would then constitute a list of families
possessing the character exclusively ("e."). For any pair of characters,
X and Y, therefore, it might be necessary to carry out four estimates of
X2, as follows: X (e.) with Y (e.); X (n.e.) with Y (n.e.), X (e.) with
Y (n.e.), X (n.e.) with Y (e.).
Some Problems Associated With Character Correlations 37

With few exceptions, when this is done, the first gives the highest
value of '1..2 , and the last two give the lowest. Doubtless, this is a reflection
of the fact that mixed families tend to have something in common. Thus,
they could be mixed in respect of both characters X and Y (and also
in respect of many other characters) because they are large families that
ought to be split. A good example of such a family is provided. by the
Leguminosae, as defined by DIELS, which proves to be mixed in respect
of a large number of characters. Thus, even the number of carpels must
be recorded as variable, because of the few genera, like Archidendron,
with up to fifteen carpels. Most taxonomists now split the Leguminosae
into three families of which only one, M imosaceae, has to be recorded
as variable in respect of carpel number.
Even so, it would still be true to say of the flowers of Mimosaceae
that it is very rare for them to have more than one carpel. The temp-
tation is strong, when reading such a description to ignore the rare excep-
tions and to record a" +" for the character state "carpels one". However,
this temptation must be firmly resisted, for there are continuous series
of int.ermediat.es bet.ween "very rarely present." and "almost. always
present". These are variously described as "occasionally present",
"somet.imes present." , "often present", "usually present", etc. The
only way of avoiding the accusation of personal bias is to record a "±"
for all such statements. A single genus (or even species) is no less impor-
tant phylogenetically when it belongs to a large family than when it
belongs to a small one.
The meaning to be attached to the statement that character "X" is
present in a family, or that character "Y" is absent will depend, not only
on the kind of character, but also on the kind of publication in which
the statement appears. The most reliable are likely to be statements
about taxonomic characters, such as those used in ENGLER & PRANTL,
for they have been checked (presumably) in every species within the
family, and can usually be verified by reference to herbarium sheets.
Characters whose recognition requires long and tedious techniques may,
however, raise problems. It is common to find statements such as "The
following families exhibit character X", the families then being listed
without any clear indication as to the scope of the investigation. In this
form, the information is almost useless for statistical purposes, and the
only safe procedure is to record a "±" for each of the listed families,
recognizing that all one really knows is that the character is present
in at least some species belonging to those families. Families that are
not included in the list cannot, however, be assumed to lack character
"X", unless there is a clear statement that this is so. Omission of a family
might mean no more than that it has not been examined.
Ideally, one hopes to find the data presented in such a way as to
38 K. R. SPORNE:

show: (a) those families in which all the species examined exhibit
character "X", (b) those in which some species exhibit it but others do
not, (c) those in which none of the species examined exhibit the character,
and (d) those families which have not been examined. If there is a clear
statement as to which families have not been examined, then these
families must be excluded during any search for possible correlations
between character "X" and other characters. Thus, where information
is known to be lacking on two characters, the total "population" for
statistieal analysis may be much smaller than the 259 families recognized
by DIELS. However, it is better to work with small, and reliable, num-
bers than to pretend that the data are complete. There will also, of
course, be reductions in the total population by the exclusion of families
to which certain characters, by definition, cannot apply. Thus, when
analysing wood characters, herbaceous families must be excluded and,
when analysing fusion of petals, apetalous families must be excluded.
For one reason or another, therefore, the size of the statistical
population may be very much reduced. Thus, the correlation between
"apotracheal parenchyma" (character 4, in Table 1) and "leuco-antho-
cyanins" (character 9), the population is as low as 145, not only because
herbaceous families have to be excluded, but also those which BATE-
SMITH (1962) had been unable to examine biochemically and those whose
wood METCALFE & CHALK (1950) had not described.
Sometimes, the decision as to whether to exclude a particular family
or not is made more difficult by the problem of homology. This is clearly
illustrated by the Aizoaceae (or Ficoidaceae). LAWRENCE (1951), without
hesitation, states that the perianth is uniseriate, composed of a calyx of
five to eight sepals, and that there are no petals (the apparent petals being
petaloid stamens). In this he agrees with DIELS (1936). HUTCHINSON (1959),
however, is equally confident that both petals and sepals are present. The
decision is a vital one because, in the search for possible correlations with
petal number, petal fusion, corolla symmetry, etc., the Aizoaceae would
have to be excluded if they truly lack petals, but should be included if
they truly possess petals. Since my work has been based on DIELS, I ex-
cluded this family from such calculations. However, if I were now to repeat
all my investigations using HUTCHINSON'S scheme, the Aizoaceae ought
to be included.
Similar problems arise in connection with the aril which, strictly speak-
ing, is an extra envelope outside the integuments, which grows up around
the ovule from the point of attachment of the funicle. There are various
so-called "arillodes", which may resemble arils and may perform similar
functions, but whose homologies are obscure. They include "strophioles"
(swellings on the raphe), and "caruncles" (swellings near the micropyle).
Some morphologists have argued that such arillodes are merely analogous
structures. Subjective judgements cannot be avoided in such circumstances.

Some critics of the work summarized in Table 1 have argued that

subjective judgements must have played a part in the choice of charac-
Some Problems Associated With Character Correlations 39

Table 1. Correlations among twenty -six characters in dicotyledons; also,

correlations between these characters and their occurrence in pre-Oligocene
families, pre-Tertiary families and present-day rain-forest families. Bold
symbols indicate correlations of significance greater than 50: 1 and faint
symbols those of significance between 50: 1 and 20: 1. A "dot" indicates
that, the observed figure is not significantly different from that calculated
on a random basis. A "cross" indicates a meaningless correlation

'" "E'" . .e" . c

.
.2
.
u
Character '" u

.
~ (j ~ ~
. ~ E!
"~ ~. ~ . g .E ~ "
~ . g c 11" i! -g i " ~
c ;;
Correlations " 'c (j ~ c E
" .~ E S
f
0
"U
E ~

a. .- . ~ "
t 0

~ 'a 'g" " .=. "

0
"0
~
"0 C; ~ 0.
"0
c 0.
~
C 0

11 :@ f! ~ . '" . " ." .

0

:c " § ~ :::0. ~ '§

.~ U .0 c
.
in ~ u .~
d
'" " g "
(j "0

.
E u 0. _~
~ c O!
E
.~
8.'" ;; . .. :ir" -;:: ~ ;;" " ~
'" .
C
(j
:c ~
'".. ~
"
.~

]
Dicotyledons ~ E 0.
E
'! "is. c
'" a a
i~
C
~ ~ c
:
~ c
3 'c,
..
§ ~
"0
<50. (jE .!! .!!
>
a ~ 0
0
0
3
"8 "8
U1 U1 « V1 U1
0
u:
0
"
IT: ~ iii w u ~ ;E U1
0

~ :§ 8
"0
c "0
w a. ~ iU
0.2
<1
(1976) ~

_NM~~w~oomO_NM~~W~~mO_NM~~W

1 Woody habit +·xx+++++++++~++++~+~+~~i

2 Scalariform end - plates + +++ +++ + + + +
3 Scalariform side-walls + + + + + + ++
4 Apotracheal parenchyma x+ ++ +++ + + + -+ +
5 Un storeyed wood X+++ + .- + +
6 leaves alternate + + + +++++ .+++ ++
7 Stipules present + +++++++++++++++
8 Secretory ceUs + ++ +++++++ ++++++
9 leuco-anthocyanins ++ + ++ +++++ + ++++ ++
10 Flowers unisexual +++++++++ +++++ +++++++
11 Flowers Qctinomorphic ++ + +++++ +++ + + ++++
12 Petals free + + ++++++ ++ + ++++++++
13 Stamens pleiomerou5 1+ + +++++++ +++ +++++
14 Carpels pleiomerous + ++++++++ ++ ++++ + +
15 Carpels free ++ + ++ X +++ +
16 Axile placentation + + + ++++x + + +
17 Pollen binucleate ++ + ++++ + +
18 Seeds orillate + + +++ + +++++ ++++
t9 Two integuments + ++++++++++ + ++ +
20 Integument bundles ++ + +++++ +++++++ ++ +
21 Ovules crassinuceHate + - ++++++++
- ++++++++
+++ + ·
22 Endosperm nuclear +
++ +++ +++ +++
++++
. . ·· . .
23 Pollen pauciaperturate
.
24 Tapetum glandular
25 Ellagitannins present
+
+ + +
++
+ +
+
+ ..
26 Aluminium accumulation +++++ + + ·
27 Pre-Oligocene (Muller) ++++ +++ . ++ +- .+
28 Pre-Tertiary (C.G.H.l ++ +-+++++ + +++++ +
29 Rain-forest + +- + + +_+:...-_. ....:+'-..!.+_...:+_-'+'--_+'--_+-'--__.--=...J+

ters to be studied, but this is not so. In fact, far more characters have
been considered than the twenty. six listed in Table 1, but any which
gave non· significant results were merely not mentioned in any publica.
tions. Some characters, indeed, had to be abandoned at the outset,
40 K. R. SPORNE:

for lack of data. Among those with adequate data, but which proved
to be non-significant, were several involving facts provided by DAVIS
(1966), e.g. patterns of embryogenesis (Onagrad, Asterad, Solanad, etc.),
cytokinesis of pollen mother cells (i.e. successive or simultaneous),
pollen tetrad types (i.e. tetrahedral, isobilateral, decussate, etc.), number
of nuclei in the tapetal cells, pollen grains united when dispersed, and
endosperm ruminate. Data concerning girdling vascular bundles in
the flower receptacle, derived from SAUNDERS (1937, 1939) and supple-
mented by SPORNE (1976b), gave no significant correlations; neither
did data derived from YAKOVLEV & ZHUKOVA (1973) concerning the
colour of the embryo (i.e. whether green or not). Data derived from
SHUNJI IMAI et al. (1936) concerning the presence of phyto-ecdysones
gave no significant results; neither did those derived from BEHNKE
(1972), concerning the type of plastids in the sieve-tubes; nor did those
derived from YAMPOLSKY & YAMPOLSKY (1922) concerning dioecism.
The compound leaf, toothed leaf margin, inferior ovary, anemophilous
pollination, copious endosperm and starchy endosperm, likewise, gave
no significant correlations. Ovule shape was shown to be involved in
only one significant correlation, anatropous ovules being positively correl-
ated with unisexual flowers (SPORNE 1969).
Other critics apparently fail to understand the meaning of the
statement that two characters, A and B, (or their alternative states,
A' and B/) are correlated, and point to the existence of "exceptions"
in which A' occurs with B (or A with B'), as if this destroys the correla-
tion. It must be emphasized that the statement is a statistical generaliza-
tion. It is not a statement that A and B always occur together. If they
did, then statistical tests of significance would not be necessary.
It is perhaps surprising, however, that even when characters A and B
are correlated (and also, of course, the associated characters A' and B'),
the number of families exhibiting the combinations AB and A'B' can
be in a minority within the "population". For example, suppo"le that, in a
population of 100 families, 75 possess character A and 25 character A';
suppose, also, that 20 possess character Band 80 character B/. Further-
more, suppose that all those with character B also possess character A.
Clearly, the majority (55 families) possess the combination AB', yet
a 2 X 2 contingency test shows that A is positively correlated with B
(and A' with B'), for the expected number of families with both A and B,

on a random basis, is ----wo-

20 X 75
= 15. This is significantly less than the

observed number, 20. The value of X2 = 8.34 shows that the correlation
is highly significant, for it 'corresponds to a value of "p" which is less
than 0.01.
Some 'Problems Associated With Character Correlations 41

@ @ @ @

:/
@ (a)
:/
@ (b)

@@@ @@@
\:/
@ (e)
\:/
@(d)

@@@
\jl
~
'eV (e)

Fig. 1. Five model evolutionary processes, each of which, if repeated.

many times in different taxa, could have led to correlations today of A
with B and A' with B'. In each, the broken lines indicate the survival of
ancestral conditions, while the continuous lines indicate evolutionary
changes
42 K. R. SPORNE:

It has often been claimed, in discussions on evolution, that the

commonest combination of characters within a group represents the
primitive type for that group. However, the example just quoted shows
clearly that this is not necessarily true. Character A could be a primitive
one which has evolved so slowly that only 25% of the families have
achieved the advanced state, A', while character B could be a primitive
one which evolved so rapidly that only 20% have retained it. Alterna-
tively, the whole evolutionary scheme could be viewed upside down,
with A' as a primitive character evolving rapidly into A, and B' a primi-
tive character evolving slowly into B.
That one should expect primitive characters to show positive correla-
tions was suggested many years ago (SPORNE 1948). It follows from the
definition of a primitive taxon as one which has retained a relatively
large number of primitive (i.e. ancestral) characters. Primitive charac-
ters, instead of being randomly distributed among present-day families,
therefore, tend to be concentrated among those which are primitive.
This, in itself, is sufficient to cause positive correlations between primi-
tive characters (e.g. A and B) and, so long as the corresponding advanced
states (A' and B') are defined merely as the opposites of A and B, they
too must show positive correlation. However, if divergent evolution
has occurred, A diverging into C and D, while B has diverged into E
and F, then it does not follow automatically that any of the advanced
combinations, CE, CF, DE or DF, will be positively correlated.
Thus, although one should expect primitive characters to be posi-
tively correlated, advanced ones mayor may not be so, according to
the way in which they are described. But the very fact that advanced
characters, as well as primitive ones, may be correlated creates a problem
in that, without other evidence (e.g. from the fossil record), it is impos-
sible objectively to decide whether A and B are the primitive states,
or A' and B'. However, even in the absence of such other evidence,
one can at least suggest that characters have been involved in evolu-
tionary trends, and that if one pair is primitive the other is advanced.
STEBBINS (1951) questions this conclusion and puts forward an
alternative explanation for character correlations. He suggests that
characters which are correlated are those which, together, contribute to
outstanding biological success, and he describes organisms possessing
such characters as having attained "adaptive peaks". Thus, A + B +
C + D might represent one adaptive peak, while the opposite states,
A' + B' + C' + D' might represent another.
In certain circumstances, this explanation is completely compatible
with the idea that such characters represent the beginning and end of
evolutionary trends but, in other circumstances it is not. Fig. 1 illustrates
five model evolutionary processes, in all of which the broken lines
Some Problems Associated With Character Correlations 43

represent the survival of ancestral character· combinations, so as to

give the primitive condition today, while the unbroken lines represent
evolutionary changes, leading to advanced states. Each, if repeated
many times in different taxa, could have resulted in positive correIa.
tions, not only between A and B, but also between A' and B'. The first
two models, 1 a and 1 b, represent evolutionary trends. Thus, in 1 a, A
and B are primitive characters which have survived together from the
ancestral taxon, while A' and B' represent advanced characters. It is,
of course, unlikely that A' and B' would have appeared simultaneously,
as the model seems to imply but, for the sake of simplicity, the inter.
mediate combinations A'B and AB' have been omitted. Model 1 b
represents trends in the opposite direction, from A' to A, and from B'
to B, yet the statistical result is the same, viz. that A is positively correlated
with B, and A' with B'. In these models, the two possible explanations
of correlations are compatible with one another for, in order to have
survived at all, any character combinations that occur today must
be relatively efficient, whether they be primitive or advanced. Thus,
the combination AB may be specially efficient in those regions of the
present.day world whose climate resembles that in which fossil ancestors
lived, while A' B' may be specially efficient in other regions; and the
concept of adaptive peaks is, therefore, not incompatible with that of
evolutionary trends.
However, models 1 c, 1 d, and 1 e, represent evolutionary schemes in
which the correlations of A with B and A' with B' cannot be explained
on the basis of trends. Here, the concept of adaptive peaks provides the
only acceptable explanation. However, such schemes are applicable
only when the correlated characters are relatively few, and when they
are likely to be functionally associated with each other. Thus, they
might explain correlations between floral characters, which are intimately
concerned in pollination mechanisms, or between seed characters, which
are intimately concerned in dispersal and seedling establishment. They
are unlikely to explain all' the 122 correlations (at the 50: 1 level of
significance) between the twenty. six characters listed in Table 1.
Fig. 2 summarizes much ofthe information in Table 1, twenty.two of
the characters having been re.arranged in seven groups, as follows:
growth habit (one character); vessel characters (two); l~af characters
(three); flower characters (five); pollen characters (three); seed charac·
ters (five); biochemical characters (three). Characters 4, 5, and 16 have
been omitted because, by definition, they are subordinate ones. The
number of intra.group correlations (at the 50: 1 level of significance)
is shown in brackets, as a proportion of the total possible number.
Thus, among the five flower characters, there are seven correlations, out
of a possible ten. Such intra.group correlations could, indeed, be ex·
44 K. R. SPORNE:

plained on the grounds of functional association. The lines crossing

Fig. 2 indicate inter-group correlations. Thus, between the three leaf
characters and the five seed characters, there are twelve correlations,
out of a possible fifteen. It is unlikely that these can be explained

Woody
habit

"2 ~ (:;'\ V~ss~ls

0(+)
l.
3

p~r0---
Fig. 2. Correlations among twenty'two characters, summarized from Table 1.
The characters are re.arranged into seven groups, the number in each
being shown in the circles. Lines indicate inter-group correlations, whose
number in each case is shown as a fraction of the total possible number.
Alongside each group, the number of intra· group correlations is shown
in brackets

on the grounds of functional association; and it is much more likely that

they result from evolutionary trends. The same is true of most of the
inter-group correlations, for they are between characters which occur in
widely different regions of the plant, at widely different times in its
development, and even in different phases of its life-cycle. However,
if these correlations are, indeed, the result of evolutionary trends,
Some Problems Associated With Character Correlations 45

some knowledge of the fossil record is vital in establishing the direction

in which they have gone.

As explained elsewhere (SPORNE 1976a), the fossil record has been

used in two quite different ways, one of which is regarded by HUGHES
(1976) as improper, and the other less so. The first involves attempts to
assign fossil remains to modern taxa. Thus, fossil wood is compared with
known modern timbers, in the hope of finding an exact match. Similarly,
attempts have been made to match fossil leaves with those of living species,
but the results should be treated with extreme scepticism, not only because
of the improbability of locating the right species out of a total of some
200,000, but also because of the variability of leaf shape from plant to
plant, and even from branch to branch on one and the same plant. To
this should be added the remarkable variability shown by some hetero·
blastic species, as they pass through successive developmental phases. Fossil
pollen grains have been widely studied in recent years, in the confident
expectation that they would give more reliable identifications than either
fossil wood or leaves, but even so, they are not without problems. Thus,
although the pollen grains of many families can be assigned to a particular
genus, or even species, those of some others cannot easily be assigned to
their family.
The results of such attempts to identify fossil remains (be they of wood,
leaves or pollen grains) are often presented in the form of lists of families,
or genera, that are claimed to have existed in past ages. However, as HUGHES
points out, such claims involve the assumption that, because pollen grains
like those of (say) Ilex existed in Senonian times, the plant producing them
had the totality of features which characterize present·day Ilex. Such an
assumption is unwarranted, for it implicitly denies that evolution can
have occurred since the Senonian stage.

Nevertheless, if those families whose pollen has allegedly been

identified in pre-Oligocene deposits (MULLER 1970) are compared with
the whole present-day flora of the world, significant differences are
found. These are summarized in Line 27 of Table 1. Here, existence in
pre-Oligocene deposits has been treated, for statistical purposes, as if it
were a morphological character with which others are correlated; and
the symbols have the same meaning as in the main part of the Table.
Thus, a "+" sign indicates a positive "correlation", showing that the
relevant character occurs more frequently among those families that
have been identified by MULLER in pre-Pleistocene deposits than it
does in the present-day flora of the world. CHESTERS et al. (1967) give
details of the first recorded appearance of various families of dicotyledons,
identified on the basis of both pollen and other morpho~ogical charac-
ters. If those which were in existence by the end of the Cretaceous are
compared with the present-day world flora, significant differences are
shown to occur; and they are summarized in line 28 of Table 1.
It is important to realise that, in all these comparisons with "fossil"
families, we are concerned, not with the characters that they might
46 K. R. SPORNE:

have shown in the past, but with those that they exhibit now. In view
of the cautionary words of HUGHES, therefore, it is perhaps surprising
that any significant results are obtained at all. Even more surprising
is that the very characters which are most significant in such comparisons
are also those which appear in the body of Table 1. In view of this, it.
must be fairly certain that the twenty-six characters are primitive,
rather than advanced. Equally, it is fairly certain that those families
which appeared first in the fossil record have retained, to this day, a
large number of primitive characters and that their rate of evolutionary
progress must have been slower than that of the more advanced families
which appeared later.
The second way in which the fossil record can be used involves no
more than the accurate description of the fossils themselves (Le. no
assumptions need to be made about the missing parts of the plant).
Successive horizons are then compared, in order to discover any evolu-
tionary trends that may have occurred. This has been done, with great
success, by MULLER (1970), who has demonstrated a trend from few to
many in the number of apertures in the grain. Here, at last, are charac-
ters which the student of phylogeny can label with confidence as primi-
tive and advanced, respectively. "Pollen pauci-aperturate" represents
the primitive state (as opposed to "pollen multi-aperturate", the ad-
vanced state), and takes its place, as character 23 in Table 1. The fact
that it is correlated with more than ten of the other twenty-five charac-
ters leaves little doubt that they, too, are primitive. The negative cor-
relation in line 27 may seem surprising, but there is a simple reason, as
explained elsewhere (SPORNE 1975). It arises from the fact that it is
difficult to assign sulcate pollen grains to a particular family. As a result,
families with such pollen tend not to be listed as having occurred in
pre-Oligocene times; yet modern families with sulcate have pollen been
treated statistically as having pauci-aperturate pollen.
Recent work by HICKEY (1973) and by DOYLE & HICKEY (1976) holds
out great hope for the future use of leaf characters in a similar way. HICKEY
has devised a system of terminology for the various types of leaf architecture
in living dicotyledon~; DOYLE & HICKEY have now applied this system to
fossil leaves from successive mid-Cretaceous deposits in the U.S.A. They
have demonstrated clear trends, not only in overall shape and size, but
also in venation. As a result, yet more primitive characters can be added
to the list; and it will be interesting to discover how far they are correlated
with other primitive characters, among present-day dicotyledons. How-
ever, before this will become possible, much work still remains to be done
in assembling the necessary data.
When making comparisons that involve fossil floras, one is very
much aware that some plants were more likely than others to become
fossilized. Thus, many of the best localities for fossil angiosperms are
Some Problems Associated With Character Correlations 47

those which were estuarine, where macroscopic remains of leaves,

twigs or fruits were likely to have come from a circumscribed region,
or even from particular types of vegetation. For example, mangrove or
rain-forest plants might be over-represented in estuarine localities.
Such deposits, therefore, represent, not merely a minute sample of the
world's vegetation, but also one which is biassed.
Fossil pollen introduces less of this kind of bias, because pollen
grains can be transported much further than macroscopic remains, for
they can be carried in air currents, as well as in rivers and on the surface
of the sea. They can, therefore, have come from a much wider range of
habitats. However, the pollen record suffers from other kinds of bias.
Thus, pollen grains of some species, for a variety of reasons are unlikely
ever to have reached the site of deposition; others may be so delicate
that they never became fossilized or, if they did, then they fail to with-
stand the chemical treatment to which they are subjected by palyno-
logists. The effects of this kind of bias can be minimized if a particular
fossil flora is compared, not with the present-day flora, but with other
fossil floras. Such it comparison has been made (SPORNE 1973) between
the pre-Oligocene pollen record of MULLER (1970) and his complete list of
pollen identifications from all deposits up to, but excluding, the present
day; and a similar comparison was made between the pre-Tertiary fami-
lies recorded by CHESTERS et al. (1967) and their complete list of families
identified from all deposits up to, but excluding, the present day.
Although the number of significant differences is lower than for com-
parisons with the present-day world flora, nevertheless, the results
strongly support the conclusion that the characters listed in Table 1
are primitive.
The knowledge that certain characters are primitive and others
advanced can be used to make an assessment of the relative advance-
ment of the various families of dicotyledons for, by definition, primitive
families are those with a relatively large number of primitive characters,
and advanced families are those with relatively few primitive characters.
The first attempts (SPORNE 1949) to calculate an "advancement index"
were based on the twelve characters which, at that time, had been shown
to be correlated. These included '''the unisexual flower". The latest calcu-
lations (SPORNE 1969) are based on twenty-one characters, which do not
include "the unisexual flower". This character, more than any other,
is likely to have been involved in evolutionary reversals, the recognition
of which would have been a matter entirely of subjective judgement.
For this reason, if for no other, it was decided to omit thil controversial
character from the calculations. It should, however, he noted that its
inclusion would have made scarcely any difference to the final assess-
ments.
48 K. R. SPORNE:

The methods used for evaluating the advancement index of a family

are open to criticism for a number of reasons, of which the most serious
is that, yet again, subjective judgements are unavoidable. Basically,
the procedure is to assign "marks" to each family, according to the
number of advanced characters that it shows. However, the question
immediately arises as to whether all the twenty-one characters are
equally important as indicators of advancement. Because it is impossible
to discover how important they might be to the plant, it may be argued
that they should be treated as equivalent. Yet it is not possible to do
this where the expression of one character is dependent on the presence
of another. Thus, "apotracheal parenchyma" and "storeyed wood"
are both subsidiary to "woody habit". Illogical though it may be, the
decision was made to give one "mark" for the possession of "storeyed
wood" and one to "herbaceous habit", as if they were equally deserving.
Similar problems arise in relation to the corolla, where "petals fused"
and "corolla zygomorphic" are both subsidiary to "petals present";
and a similar decision was taken.
Another problem is that caused by "mixed" families, where it may be
argued that a family most of whose members possess a particular character
should receive a higher award than one in which the character is only
rarely present. Again, a decision has had to be made. All mixed families
have been treated alike, and have been awarded a "half-mark" for each
character in respect of which they are mixed. The effect of such a procedure,
when applied to a family like the Leguminosae, which is mixed in respect
of most characters, is to produce an advancement index with relatively
little meaning. The figure obtained is merely an average for a wide range
of component taxa which, in modern classifications, have been elevated
to the status of families, of which one would now have a low advancement
index, while the others would have relatively high ones.

Clearly, the time is ripe to change from the taxonomic scheme of

DIELS (1936) to a more recent one, especially now that computer facilities
are readily available. Even a superficial glance at recent classifications,
e.g. MELCHIOR (1964), TAKHTAJAN (1966, 1969) and CRONQUIST (1968),
shows that many of the old families have been split into new ones whose
characters are much less mixed.
In the meantime, the advancement index should be taken only as a
very rough guide to relative advancement. Nevertheless, it has already
proved its predictive value in relation to those characters which have
been investigated since the latest re-calculation of advancement indices.
Thus, the average advancement index for the fifteen families with
exclusively multi-aperturate pollen grains was found to be 77.2 (SPORNE
1972), whereas that for the 155 families with exclusively pauci-aperturate
pollen grains was 54.4. These figures gave a strong indication that further
investigation was justified; and the correlations listed in line 23 of
Some Problems Associated With Character Correlations 49

Table 1 were subsequently found to exist. Likewise, the fifty families

with ellagitannins in their tissues were found to have a low average
advancement index of 47.3, as compared with an average for the world
flora of 56.6 (SPORNE 1975), and the thirty-seven families in which
aluminium accumulators occur have an average of 49.0 (CHENERY &
SPORNE 1976). The correlations that were subsequently found are
listed in lines 25 and 26, respectively.
The most important conclusion to have come from calculating the
advancement indices of families has been to show that several families,
in quite different orders, are almost equally primitive. No single family
that is alive today can, therefore, be regarded as the ancestor of all
other dicotyledons. This should surprise no one who is aware of the
long time that has elapsed since angiosperms first appeared, but it may
help to silence the rather sterile arguments that still occur from time
to time.
Among the most advanced families, there are seen to be at least two
quite different kinds of flower. On the one hand, there is the zygomorphic,
gamopetalous type, found in Phrymaceae, Pedaliaceae, Martyniaceae,
etc.; on the other, there is the extremely reduced type found in Galli-
trichaceae and Hippuridaceae.
In the past, conclusions such as this would have involved a circular
argument which, when reduced to its essentials, would have run as
follows: "These families are primitive because they possess primitive
characters, and primitive characters are those which are possessed by
these primitive families." At last this circular argument has been
broken, for we have discovered which characters are primitive without
hav:ing had to ask the question: "Which families are primitive?" Indeed,
the asking of this question was postponed until an answer, which was
not entirely based on personal beliefs, could be expected with some
degree of confidence. However, it has not been possible to eliminate
subjective judgements completely. Indeed, subjective judgements.
have had to be made at every stage.

References
BATE-SMITH, K C., 1962: The phenolic constituents of plants and their
taxonomic significance. J. Linn. Soc. (Bot.) 58, 95-173.
BEHNKE, H.-D., 1972: Sieve-tube plastids in relation to angiosperm sys-
tematics-an attempt towards a classification by ultrastructural ana-
lysis. Bot. Rev. 38, 155-197.
CHALK, L., 1937: The phylogenetic value of certain anatomical features
of dicotyledonous woods. Ann. Bot. (London) 1, 409-427.
CHENERY, E. M., and SPORNE, K. n., 1976: A note on the evolutionary
status of aluminium-accumulation among dicotyledons. New Phytol.
76, 551-- 554.
Plant Sr_t.. E.-ul .. Suppl. 1
50 K. R. SPORNE:

CHESTERS, K. I. M., GNAUOK, F. R., and HUGHES, N. F., 1967: Angio-

spermae. In: The fossil record (HARLAND, W. B., et aI., Eds.), 269-288.
London: Geological Society.
CRONQUIST, A., 1968: The evolution and classification of flowering plants.
London: Nelson.
DAVIS, G. L., 1966: Systematic embryology of the angiosperms. New York:
Wiley.
DIELS, L., 1936: A. ENGLER'S Syllabus der Pflanzenfamilien, 11. Aufl.
Berlin: Borntraeger.
DOYLE, J. A., and HICKEY, L. J., 1976: Pollen and leaves from the mid-
Cretaceous Potomac Group and their bearing on early angiosperm
evolution. In: Origin and early evolution of angiosperms (BEOK, C. B.,
Ed.), 139-206. New York: Columbia University Press.
ENGLER, A., und PRANTL, K., 1887-1915: Die natiirlichen Pflanzenfami-
lien. (23 Vols.). Leipzig: Engelmann.
-, - (1924-): Die natiirlichen Pflanzenfamilien, 2. Aufl. Leipzig:
Engelmann.
HICKEY, L. J., 1973: Classification of the architecture of dicotyledonous
leaves. Amer. J. Bot. 60, 17-33.
HUGHES, N. F., 1976: Palaeobiology of angiosperm origins. Cambridge:
University Press.
HUTOHINSON, J., 1959: The families of flowering plants, Second edition.
Vol. 1: Dicotyledons. Oxford: Clarendon Press.
LAWRENOE, G. H. M., 1951: Taxonomy of vascular plants. New York:
MacMillan.
LOWE, J., 1961: The phylogeny of monocotyledons. New Phytol. 60,
355-387.
MELOHIOR, H., 1964: A. ENGLER'S Syllabus der Pflanzenfamilien, 12. Aufl.
2. Band: Angiospermen. Berlin: Borntraeger.
METOALFE, C. R., and CHALK, L., 1950: Anatomy of the dicotyledons.
(2 Vols.) Oxford: Clarendon Press.
MULLER, J., 1970: Palynological evidence on early differentiation of angio-
sperms. BioI. Rev. 45, 417-450.
SAUNDERS, E. R., 1937 and 1939: Floral morphology: a new outlook,
with special reference to the interpretation of the gynaeceum. (2 Vols.)
Cambridge; Heffer.
SHUNJI IMAI, TOMOYOSHI TOYOSATO, MICHIHIKO SAKAI, YASUO SATO,
SHOJI FUJIOKA, EIKO MURATA, and MINORU GOTO, 1969: Screening
results of plants for phytoecdysones. Chern. pharm. Bull., Tokyo 17,
335-339.
SPORNE, K. R., 1948: Correlation and classification in dicotyledons. Proc.
Linn. Soc., London 160, 40-47.
1949: A new approach to the problem of the primitive flower. New
Phytol. 48, 259-276.
1969: The ovule as an indicator of evolutionary status in angiosperms.
New Phytol. 68,555-566.
- 1972: Some observations on the evolution of pollen types in dicotyledons.
New Phytol. 71, 181-185.
- 1973: The survival of archaic dicotyledons in tropical rain-forests. New
Phytol. 72,1175-1184.
- 1974: The morphology of angiosperms. London: Hutchinson.
Some Problems Associated With Character Correlations 51

SPORNE, K. R., 1975: A note on ellagitannins as indicators of evolutionary

status in dicotyledons. New Phytol. 75, 613-618.
1976a: Character correlations among angiosperms and the i~portance
of fossil evidence in assessing their significance. In: Origin and early
evolution of angiosperms (BECK, C. B., Ed.), 312-329. New York:
Columbia University Press.
- 1976b: Girdling vascular bundles in dicotyledon flowers. Gdns' Bull.,
Singapore.
STEBBINS, G. L., 1951: Natural selection and the differentiation of angio.
sperm families. Evolution 5, 299-324.
TAKHTAJAN, A. L., 1966: Systema et phylogenia magnoliophytorum.
Moscow: U.S.S.R. Acad. Sci.
- 1969: Flowering plants-origin and dispersal (Transl. JEFFREY, C.).
Edinburgh: Oliver and Boyd.
YAKOVLEV, M., and ZHUKOVA, G., 1973: Angiosperms with green and
colourless embryo. (In Russian) Leningrad: V. L. Komarov Botanical
Institute, Academy of Sciences of the U.S.S.R.
YAMPOLSKY, C., and YAMPOLSKY, R., 1922: Distribution of sex forms in
the phanerogamic flora. In: Bibliotheca Genetica (BAUR, E., Ed.),
Vol. 3. Leipzig: Borntraeger.
YOUNG, D. J., and WATSON, L., 1970: The classification of dicotyledons:
a study of the upper levels of the hierarchy. Aust. J. Bot. 18, 387-433.

Address of the author: Dr. K. R. SPORNE, Botany School, Downing

Street, Cambridge CB2 3EA, England.
Plant Syst. Evo!., Supp!. 1, 53-76 (1977)
© by Springer-Verlag 1977

Instituto de Quimica, Universidade de Sao Paulo, Brazil

Biochemical Systematics: Methods and Principles

By

M. Aparecida H. Cagnin, Ceres M. R. Gomes, Otto R. Gottlieb,

M. Claudia Marx, A. Imbiriba da Rocha, M. Fatima das G. F. da Silva,
and 1. Aparicio Temperini*, Sao Paulo

Abstract: Chemosystematics will not evolve to become a scientific

discipline in absence of accepted chemotaxonomic procedures. Our approach
to this problem includes, initially, the mapping of metabolites by con-
sideration of their biosynthetic relationship, indicated by skeletons, substitu-
tion patterns and frequency of occurrence in nature. The maps form the
basis for three-dimensional graphic presentations of the chemical constitu-
tion (relative to a given metabolic class) of plant groups, which aid in the
deduction of basic chemosystematic principles, as well as for two-dimensional
graphic presentations of taxonomic distances among subunits of these
groups, which further the understanding of their evolutionary relations.
Progress of this type of work with respect to several phenylalanine (cou-
marins, benzylisoquinolines, Amaryllidaoeae alkaloids), phenylalanine-
acetate (isoflavonoids) and tryptophane-mevalonate (indole alkaloids)
derived biogenetic classes of compounds are described, and its potentiality
illustrated by reference to several taxa of flowering plants.

Introduction

The existence of a relation between chemical composition and

systematic affinity of plants was sensed ag~s ago (ROCHLEDER 1854).
Why are we, inspite of considerable effort (HEYWOOD 1973), still only
sensing this relationship, why are we still able to express it by nothing
better than this dangerous presence/absence criterion?
The answer: lack of taxonomic procedures applicable to chemical
data. In absence of scientific methodology, deduction of basic prin-
ciples is impossible and biochemical systematics will never attain status
of a scientific discipline.

* Read by OTTO R. GOTTLIEB to whom also enquiries should be ad-

dressed.
Work sponsored by FundaQao de Amparo it Pesquisa do Estado de Sao
Paulo.
54 M. APARECIDA H. CAGNIN et al.:

Chemotaxonomic Methods
Biogenetic Group Maps
Secondary metabolites can be classified into biogenetic groups,
BGs. At the head of each BG stand precursor units (P) which
generate all other representatives by a sequence of biosynthetic path-
ways. Such sequences can be mapped for skeletons, as shown for the
isoflavonoid BG (Fig. 1). Each skeleton is represented in nature by a
number of derivatives, through substitution by hydroxy, alkoxy and
other groups at specific C-atoms. In the :SG-maps diversity of derivatives
diminishes, generally from left to right, with increasing number of reac-
tion steps separating the particular skeleton from the precursor; and
from top to bottom, the reaction sequences being written in such a way
as to lead progressively to smaller collections of derivatives. The lower
the frequency of substitutional derivatives of a skeleton (or of a sequence
of skeletons), the less widespread, the more specialized, must be the
reaction step (or steps) leading to that skeleton (or sequence of skeletons).
The complete substitution pattern of a particular derivative probably
already exists in a particular P of the BG, and as many BG-maps can
be written as there are substitution patterns. Again, the lower the
frequency of skeletal variants endowed with a certain substitution
pattern, the less widespread, the more specialized, must be the reaction
step leading to that substitution pattern.

Diagrams of Chemical Constitution

The drawing of BG-maps is based on the critical revision of pub-
lished work and thus to a great extent independent of personal talent
or opinion. On such maps each compound is represented by a certain
position relative to each other. This information can be transferred
to tridimensional diagrams. The simplification involved in the transfer
entails a certain amount of judgement as to significance by the
operator who will be required to ask himself: what are the structurally
most relevant attributes of a natural compound with respect to its value
as systematic marker of the taxon under scrutiny, skeletons, oxygena-
tion patterns, or additional parameters such as secondary modifications
by oxidation, reduction, 0- or C-alkylation 1 A decision having been
reached concerning this matter, the attributes are marked on coordinate
axes, in order of decreasing frequency of occurrence.
One diagram is reserved for each subunit of a taxon. All compounds
of the chosen marker BG of the subunit are represented on this dia-
gram by points whose coordinates are the appropriate values, x, y and z,
of the chosen attributes. In the present case, Leguminosae genera,
c:oD
~ w{ , " _%
v~ ;,,: %I~ ,~;: ' -??:o
.'", '0 I
(l"" I S",
to
1':.21.) 1r~
-%~~~.2 . , 21.1.1.1.1.2
cc:n f
'I)
[

21 .1.1. 5
%(hi)
~
f
%-~~
(17."
'I.t, «h.,
~
o
g.
~
%
.( ,)
~
§.
i
~
(laJ) Ot
% Ot

Fig. 1. Biosynthetic relationship and code of the isoflavonoid biogenetic group of skeletons. The biosynthesis involves,
as PPs, one cinnamic acid and three acetic (malonic) acid units which form a chalcone (CHAL), the precursor of two
BGs, the flavonoids which derive from flavone (code 11), and the isoflavonoids (code 21). In 0: numbers of deriva-
tives X numbers of families in which they occur (REZENDE et aI. 1975)
56 M. APARECIDA H. CAGNIN et al.:

the attributes are oxygenation of the shikimate·derived B.rings without

(initial part of the x axis) or with (final.part of the x axis) a 2' ·oxy group,
5,7- or 7.oxygenation (y axis) and eventual additional oxygenation
of their acetate derived A rings (z axis) (Fig. 2). The skeletons of the
derivatives are marked by a code on the points of intersection. The
diagrams reveal similarities and dissimilarities of chemical constitu·
tion of the genera, and the examples featured in Fig. 3 give an idea

OXYGENATION OF ACETATE RING

OXYGENATION OF SHIKIMATE RING

5,7~--L_-'---'-_~J' ___L-....J..._L-....J...--.JL-..L.-...-l_ _
.-.:;t ..,. -10 ·0 .", ....-,,; .",
-.... -iD-,,;....
-ID -",
.",.. -.;- -.; =~ -.; -N
."j
-N -N· -,,; -N
-N -N
6

6,8 ADDITIONAL OXYGENATION OF ACETATE RING

Fig. 2. Key to diagrams of isoflavonoid constitution of Legnminosae genera

concerning development along an isoflavone line (1st line), a ptero-

carpanoid line (2nd line) and a rotenoid line (3rd line). Such similarities
and dissimilarities are visual guides in both aspects of the taxonomist's
activities (HEYWOOD 1973): (a) circumscription and description of
taxa; (b) formulation of phylogenetic relationships between taxa. Once
these relationships have been traced through lines (p. 56--62), the
diagrams also allow one to predict the chemical composition of extinct
ancestral forms and of unanalyzed existing groups.

Affinity Plots
The diagrams of chemical constitution of taxonomic units refer
to compounds and, as we shall see (p. 65--74), are useful in the deduction
of basic principles of chemical evolution. What, however, is also needed,
and the major goal of our effor·t, is the evolutionary mapping, not of
CLAORASTIS DIPTERYX PTERODON

11
~ 2( II I~ 1~/17 17
o·c:>t:O
::r'
~
8
0'
f.
PTEROCARPUS MACHAERIUM 00
21.1.1.11
21~~~_~21.1.1.1
J
21 21 21.1.1.1 '<
m
.,..
~·~~t::II·.tl 21.1 ~
2U.UJ
8
~.
c:>
I 1.1 I I I I. m
21
~ 2~y
21./ ~ 21 .21
a::
~
.,..
::r'
0
p..
m

~p..
MUNDULEA 21.2.1.1 MILLETTIA PISCIDIA
21.2.1 '"d
1.2 21 21.2
J , 21 21 ::l.
::l
c:>
1211 I • 121 I I ~I·'
-B'
(0
-~ '"
21

Q1
-.J
Fig. 3. Diagrams indicating isoflavonoid constitution of selected Leguminosae genera. .For expla-
nation of diagrams see key (Fig. 2) and text (p. 54/55 and p. 65/66)
58 M. APARECIDA H. CAGNIN et al. :

compounds, but of their sources, the organisms or groups of organisms

by which they are produced. The information contained in BG-maps
can also be used in ways which make this objective possible, in the
construction of phylogenetic trees.
The initial step in this direction involves quantification of the data
of BG-maps by the estimation, for each compound, of a relative
probability of occurrence (RPO). This probability diminishes
with increasing number of required reaction steps and increasing com-

Table 1. Relative probability of occurrence (RPO) of isoflavonoid skeleta

(cf. Table 2) given by distance from chalcone (in number of reaction
steps)/number of substitutional derivatives

Code Number of reaction steps

CHAL 2 3 4 5 6

1/135 = 2/20 = 3/11 = 4/42 = 5/22 = 6/3 =

21 0.0074 0.1000 0.2727 0.0952 0.2273 2.0000
4/19 = 5/4 = 6/1 =
0.2105 1.2500 6.0000
4/5 =
0.8000
2/17 = 3/7 = 4/5 =
21.2 0.1176 0.4286 0.8000
2/8 =
21.3 0.2500
2/1 =
21.4 2.0000
2/1 =
21.5 2.0000

plexity of reaction mechanism. The number of reaction steps can

clearly be estimated with relative ease. The specialization of a reaction,
however, can also be gauged considering its inverse proportionality
with the frequency of occurrence of the product (or sequence of related
products). The RPO for each compound can thus be estimated with
respect to skeleton and substitution.
Skeletons: An example of the calculations is given in Table 1
for the isoflavonoid BG, the pertinent figures being superimposed on the
position occupied by each isoflavonoid skeleton on the BG-map. P, in
column 1, is alloted 1 point and skeletons in successive colums, 2, 3, 4, ...
points. The point value of each skeleton is now divided by the frequency
of occurrence, as given by the number of its derivatives.
Su bstitutions: The precursor substitution pattern, here 5,7,4'-oxy-
genation (Table 2), is alloted 1 point, and successive patterns, formally
Table 2. Relative probability of occurrence (RPO) of isoflavonoid substitution given by distance from 5,7,4'-oxygenation
(each additional oxygenation or deoxygenation = 1 step)/number of skeletal derivatives

Position tc
of oxygenation o·
0
on rings ::r
A B 2',4',5' 4' 2',4' 3',4' 2',3',4' 3',4',5' 2',3',4',5' 2',4',6' 2',3',4',6' 2',5'
~.
0
~
4/55 = 2/22 = 3/38 = 3/11 = 4/10 = 4/2 = 5/3 = 4/3 = 5/3 = 3/1 = rJ.l
'<i
7 0.0727 0.0909 0.0789 0.2727 0.4000 2.0000 1.6666 1.3333 1.6666 3.0000
(!)
.,..'"
3/7 = 1/28 = 2/18 = 2/19 = 3/2 = S
1:1>
5,7 0.4286 0.0357 0.1111 0.1053 1.5000 ::to
0
5/10 = 3/9 = 4/1 = 4/8 = 5/3 = 5/2 = 6/1 = '"
6,7 0.5000 0.3333 4.0000 0.5000 1.6666 2.5000 6.0000 is:
(!)

5/10 = 3/3 = 4/3 = 4/2 = 5/3 = 6/1 = 4/1 = ~

0
7,8 0.5000 1.0000 1.3333 2.0000 1.6666 6.0000 4.0000
~
4/4 = 2/10 = 3/1 =
5,6,7 1.0000 0.2000 3.0000 ~p.
3/2 = I'd
5,7,8 1.5000 S·
S.
4/1 = ~
5,6,7,8 4.0000
'"
4/1 =
6,7,8 4.0000

01
<C
60 M. APARECIDA H. CAGNIN et al.:

derived by introductions or eliminations of substituents, receive addi-

tional points according to the number of operations involved. The point
value (REZENDE & GOTTLIEB 1973) for each substitution pattern is
now divided by the frequency of occurrence, as given by the number of
its representatives within the BG, irrespective of skeletons.

Table 3

Isoflavonoid constituent RPO of isoflavonoid

Structure Name skeleton substitution

R =H 13R)-laxifloran 0.2273 0.4000

R=H 13 R) -Ionchocarpan 0.2273 1.6666

R =H 16aR,11aR)-philenopteran 0.0952 1.6666

R = Me 16aR.11aR)-9-0Me.. 0.0952 1.6666

b~···O: OR
OMe
'>..
I
1. OMe
-- J::::::l R=H 16aS, 12aS)- rotenone 0.1176 0.0727.
~ R=H 16aS,12aS)-deguelin 0.1176 0.0727
~ R = OH I~)-tephrosin 0.4286 0.0727

Mean of skeletal R PO values

x = [10.2273 x 2) + 10.0952 x 2) + 10.1176 x 2) + 10.4286 x 1)J/16.5593
= 0.0790
Mean of substitutional RPO values
y = [10.4000 x 1) +11.6666 x 3) + 10.0727 x 3)J/58.4622
= 0.0961

For each compound a skeletal and a substitutional RPO-value can

thus be determined. As an example, Table 3 lists RPO-values for the
7 isoflavonoid constituents of the genus Lonchocarpus. Which of the
compounds represents the evolutionary state of the unit 1 None in isola-
tion, of course, but all together as represented by the weighted means
of their RPOs with respect to skeletons and substitutions.
These means, marked on the x (skeleton) and y (substitution) axes,
give a point which represents the biological unit. Once all units of a
taxon are plotted on the same graph, their taxonomic distances can
Y.10- 2

! Pt.rodon ,AOalborgla
... '"
a Bryo

rCordyla //,//-'
o·t:t!
2' ii ',/
... '" o
.,.,,/'"
f.e-
00
, ... ... -
1 .,,;/
. !! Baptilia ,."./""
t-h-D1 pferyx "' .... "
Neorautanlnia
1 !! D Swartzia ",..,."' ........
I~-j)-cladralti. /PMldlcaoo rMaChaarlum
_-.. -------..0 ~
I
; lJ ,/' ,/ -----
IO~ :; Lonchgcarpu~/ " --
. ", / I -----
~Mlldbra.diodendron
/
/
"
I
I
"
r
~
" I __ -
. fTrifolium / . . .- _--
TephrOllO " PIICldlO _--- _-oO.rrll ~
?Dlplotropil " Millattio / _--- _----- "'t:l
/ ,«,,\1-0 ,/ ...... --:--,::::---- -
.,J'G\,tc,'i " ___ ,.#~---;::'-- _ _ _ _ _ _ _ _ _ _ _ _ .0 Cyclolobium
s·s.
__ --' y Glycine _----::;:~~~.;c-;,;;..;----
".." .. ..bPu.rarLct Wi --:::. - - -cr"achyrrhizul i
--~.;.';:';:~-:';4ffi6~ral~ __ ----..opericOPlls
-------------- .... '1'1
20 45 B5 90

~.10-2

Fig. 4. Affinity plot of Leguminosae genera mapped along isoflavone ( .......... ), pterocarpanoid C __ ) and rotenoid
Q>
(___ ) lines. Near the origin of the plot, on the isoflavone lines (black dots): Ormosia, Amphimas, Wisteria, Ononis, ....
Atrormosia, Castanospermum, Myrocarpus, Myroxylon; Chromanthus, Calycotome, Cytisus, Ulex, Stauracanthus, Genista,
Erinacea, Laburnum, Adenocarpus, Chamaecitysus, Teline, Chamaespartium, Lygos, Lupinus, Thermopsis; Lathyrus, Cicer;
on the pterocarpanoid line (circles): Baphia, Maackia, Ougenia, Desmodia, Aldina, Sophora, Canavalia, Vigna; Lotus,
Flemingia, Platymiscium; Andira; on the rotenoid line (black squares): Mundulea, Amorpha
62 M. APARECIDA H. CAGNIN et al. :

be read directly. It is, additionally, possible to connect the plotted

points by reasonable lines, thus placing the units on to the branches
of a genealogical tree, as exemplified for Leguminosae genera in Fig. 4.
How can the viability of an affinity line be tested? By consideration
of the chemical composition of each unit. For this purpose, first, the
centres of the coordinate axes of the tridimensional diagrams (p. 56, 57)
are superimposed on the pertinent plotted points. Next, each diagram X
is considered with respect to its neighbours. If a neighbour, Y, farther
away from the origin of the plot, is characterized by compounds which
are identical or derivable from compounds of X, the units possess
chemical affinity and their connection, X ..... Y, is a reasonable postulate.
Only the BG precursors of X will frequently be missing from Y, possibly
because of a more efficient turnover in the more specialized unit.

Similarity Indexes for Biogenetic Groups

The assessment of the relative vigour with which a biological unit
develops certain pathways may also be based on BG-maps, comparing
number of derivatives (Du) and of skelet.ons (Su) produced by the unit

Table 4. Percent isoflavonoid similarity indexes for families of Embryobionta

Class Subclass Family 100 X S1/9261.54

Polypodiopsida Lepto8porangiatae Polypodiaceae 0.04

Pinopsida Podocarpaceae 0.04
Magnoliopsida M agnoliidae M yri8ticaceae 0.04
Hamamelididae Moraceae 0.04
Oaryophyllidae Ohenopodiaceae 0.02
Amaranthaceae 0.02
R08idae R08aceae 0.25
Legumino8ae 100.00
A8teridae Oompositae 0.06
Liliopsida Liliidae lridaceae 0.29
Stemonaceae 0.06

with number of derivatives (Dt) and of skeletons (St) produced by a super-

ordinate unit. Percentual similarity indexes (SI) relative to a given BG
may be tabulated for each unit under scrutiny, as exemplified for the
isoflavonoid BG (Table 4).
Biochemical Systematics: Methods and Principles 63

Chemosystematic Principles
Direction of Chemical Evolution
Any evolutionary series based on secondary chemical constituents
can theoretically be read in either direction (HARBORNE et al. 1976).
In other words, consider a sequence of metabolites, A -+ B -+ C ---+ D,
whose biosynthetic relationship we are able to predict. Now, if a taxon
contains C, one will not be able to say if it evolved from a taxon accu-
mulating B (by expansion of the reaction sequence) or from a taxon
accumulating D (by reduction of the reaction sequence). This, however,
is essential knowledge and it will never be possible to use chemical
characters as basis for phyletic classification until it becomes clear when
and where to expect expansion or reduction of a reaction sequence,
until some basic principles predict the direction in which an evolutionary
series has to be read.

Prior to examining natural reaction sequences, it is possible to distinguish

two types, according to their association with primary or secondary meta-
bolism. Most, if not all of the primary metabolites (PPs) are, of course,
synthesized by all plant species as precursors of biopolymers. In the present
context we are, however, interested only in their function as heads of bio-
genetic groups of secondary metabolites, which they do or do not produce
according to the individual plant, and it is thus permissible to discuss
expansion or retraction of a reaction sequence involving such primary
precursors. The second type of reaction sequence operates within each
biogenetic group of secondary metabolites. Having, thus, separated reaction
sequences into types, let us examine these in order.

First Principle
Results of SI-calculations are given in Table 5 for several biogenetic
groups in correlation with plant taxa, named according to CRONQUIST
(1961, 1968). In its present, far from complete version, the picture leads
to a working hypothesis for a first principle of biochemical systematics:
among taxa of Embryobionta of high hierarchical level, biochemical
evolution of secondary metabolites involves the gradual substitution of
biogenetic groups formed from primary precursors of the shikimate
pathway by biogenetic groups formed from primary precursors of the
acetate pathway.
Indeed, reduction in importance of the lignan BG, seems to have
led to accumulation of cinnamic acid, inhibitor of its synthesis from
phenylalanine (HASLAM 1974), which thus became available for the pro-
duction of the benzylisoquinoline BG. The introduction of nitrogen
into shikimate metabolites conferred probably an important compe-
titive advantage which primitive angiosperms had over most gymno-
~
~

Table 5. Percentual similarity indexes of several biogenetic groups for taxa of Embryobionta

Shikimate Shikimate + Mevalonate Mevalonate

Pathways
BOs Benzyliso. Amar. Alk. Cactus Anthran. Indole Steroid
Lignans Coumarins ~
quinolines Colchicins Alkaloids Ac. Deriv. Alkaloids Alkaloids
>
'1j

J>inophyta 53.35 0.00 0.00 0.00 ~

t>J
JJ agnoliophyta ....0
.l1agnoliopsida t:I
>
.11agnoliidae 33.05 100.00 2.29 0.04 tIl
H amamelididae 3.00 0.01 0.00 0.61
Caryophyllidae 0.00 0.00 0.00 tOO.OO 0.02 a
>
<;)
Dilleniidae 4.00 0.01 0.00 0.09
~
Rosidae i2.iS 5.42 0.00 100.00 100.00 Z
Asteridae 100.00 0.03 0.00 2.92 100.00 toO.OO cD
....
~
U/iopsida
Alismatidae 0.00 0.00 0.00 0.00
Arecidae (l.00 0.02 0.00 0.00
Oommelinidae 0.26 0.00 0.00 0.09
Uliidae 0.00 0.01 lOO.OO 0.04 toO.OO
Biochemical Systematics: Methods and Principles 65

sperms. Chemical evolution continued by further contraction of the

shikimate pathway and use of the consequently available chorismic
acid in the production of anthranilic acid and tryptophan. Coincidental
with this contraction was the introduction of a progressively higher
proportion of terpenoid chains into shikimate-derived compounds.
The ultimate consequence of this evolutionary trend is the suppression
of shikimates and the restoration of terpenoids as the dominant group
of biologically active compounds.

ASTERIOAE

/
/
/'
/
--
/
,,- /
ROSIDAE "-e

MAGNOLIIOAE

PINOPHYTA ---'-+-_ _----L_ _ _

1I::::_ _---1'---_ _ -'-~

o ~ ~ m 00%
Fig. 5. Percent similarity indexes for lignans (.) and benzylisoquino-
lines (.) in Pinophyta and Magnoliophyta

Partial reversal of this major trend is a concomitant possibility.

A plot of percentual SIs not only shows renewed vigour of diversifica-
tion of the lignan BG in connection with the Rosidae-Asteridae, but
also points to the competitive, substitutive development of the lignan vs.
the benzylisoquinoline BGs (Fig. 5).

The first principle cannot be viewed with unreserved confidence. This

is not due only to lack of completeness of our analysis, which we hope to
remedy in the near future, but chiefly to the fact that an element of cir-
cularity was involved in its deduction. Chemical evolution was correlated
with a morphological classification, itself based on inference of significance.

Secondary Metabolites as Systematic Markers

Isoflavonoid Lines in the Leguminosae. Isoflavonoids occur widely
scattered throughout the Polypodiophyta, Pinophyta and Magnoliophyta.
As shown by SIs (Table 4), however, only in the family Leguminosae
does structural diversity justify their use as systematic markers.
Plant first. Evol.. SIlPPI. 1 5
66 M. ApAREcIDA H. CAGNIN et a1. :

As may be gauged from theoisoflavonoid chemistry of the Legumi-

nosae available as at December 1975, several parallel evolutionary lines
can be distinguished: three isoflavone lines, a pterocarpanoid line and a
rotenoid line (Fig. 4). Although it seems clear that all three isoflavonoid
classes were already represented by biosynthetically simple derivatives
at an early stage of the evolutionary history of the family, the order
in which they are given indicates, nevertheless, progressive specializa-
tion in relation to their environment. Thus, whereas only the isoflavone
skeleton occurs in genera along lines 1 and 2, the peripheral genera of
the isoflavone line 3 show additionally the most widespread skeletal
variation: isoflavone -~ pterocarpanoid; and the peripheral species of
the more highly developed pterocarpanoid lines 4.2 and 4.4 contain
additionally rotenoids. Finally, the genera of the rotenoid line 5 seem
to be substantially lacking in the less specialized pterocarpanoids.

Th~ time is now ripe to ask how the tribal classification, based on chemical
characteristics, agrees with the morphological evidence. Although this is
an interesting topic, its importance must not be overestimated. The iso-
flavonoid data refer only to 232 species belonging to 70 genera. Worse
still, it is impossible to state how many of these species have been examined
with adequate thoroughness.
Given these cautionary arguments, let us compare the location of
tribes along the chemical lines based on isoflavonoid data (Fig. 6a from
top to bottom: 3 isoflavone one rotenoid and 3 pterocarpanoid lines) with
the relevant parts of a very recent reappraisal of morphological rela-
tionships within the Lotoideae (Fig. 6b, POLHILL 1976). If, indeed, chemical
evolution within the isoflavonoid BG proceeds by gradual diversification
of substitutions and skeletons, practically no serious discrepancy between
the two schemes is apparent. The slight differentiation of the primaeval
forms makes the determination of origins of lines frequently difficult.
Thus, e.g., although the lines leading to Trifolieae and Tephrosieae run
intriguingly close on the affinity plot, suggesting the existence of a common
ancestor, this is chemically unknown. The isoflavonoid data suggest,
furthermore, that the Dalbergieae, as well as the Phaseoleae are polyphyletic,
a fact which will hardly be contested on morphological grounds.

In conclusion, it is evident that considerable insight into the evolu-

tion of the Leguminosae results, if tribal divisions are extended back
along chemical lines of development until they become blurred in the
Sophoreae. It is intellectually pleasing to acknowledge that the con-
cluding part of this sentence paraphrases POLHILL (1976) who wrote
regional where we write chemical.

Indole Alkaloid Lines in the Apocynaceae. Monoterpene-indole alka-

loids occur in three related plant families. The present section is concerned,
nevertheless, only with their use as systematic markers in the Apocy-
naceae, subfamily Plumerioideae. Biosynthetically, tryptophan and
Oadieae Oadieae

l
Sophoreae Dalbergieae Sophoreae Dalbergieae
to
.....
o

I Genisteae _____ Thermopsideae \ -"h£rnw""""" G.. c:>

;.e,~ f.e-
rn
_ V icieae - Oicereae _ Galegeae - Trifolieae Galegeae _ Vicieae _ Oicereae _ _ Trilolieae

c:>
L------------------Tephrosieae
. I.
Tephros~eae
I
'"
~
o
1-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Dalbergieae 1E'
~
Po
I-d
1-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Phaseoleae ::l.
Phaseoleae
§.
].
'"
Phaseoleae
a b
~ Fig. 6. Evolutionary relationship of Lotoideae tribes, based on a) chemical evidence; b) morphological evidence
0>
-l
68 M. ApARECIDA H. CAGNIN et a1.:

mevalonate, the PPs of the indole alkaloids, produce such a large

number of skeletons, that it is impossible to present our BG map in
reasonable format for publication. Fortunately, however, this profusion
of skeleta can be grouped into so-called corynanthe, aspidosperma and
iboga types (GEISSMAN & CROUT 1969), IIcgeording to the disposition
of C-atoms in the mevalonate moiety, which may have preserved the

AspIdosperma
4,0
2,0
Vinca

1,3 ._Rauv olria

.................

1,0 Alstonia

rCobucala -"',; __ '" ____________ - - - - - - - --0 VoaconQo

fOchrosla ..........
."" TObernaemontano!/___,.L,.,..L"'---_ _ _ _ _ _ _ _ _ _ _ Cotharanthus
.......
I , .. , .....
.... Conopharyngia .... 1-- ...
.. "" callichilial ~~":'--='=--O-T~b.~n~n~h~----- -- - ---- ---.oPandaca
c::J •• .aIHktrio,P J",.......... o5chizozya io
§ .~1JHoPIOPhyton:...... Vallesio
~ ,,' ", ......~'--_ 'pabunlo

L··:~·~t...~-~-~/~~~c=ri=og=.r=··~c~r.~id~..4.nnu=m=PI:;;;.~IO~C~orf'pa~:==:;:==~:Hunt.rla
5 10
, ,M,tqdipup I ,
15 20 '3!,' 'Jo ' .L4' ,
!<~p~i~.,~~""""'~-'-''-'-;b.'-~ , I[]
25 --30
x
Fig. 7. Affinity plot of Apocynaceae genera mapped along corynanthe
( ) , aspidosperma ( _ _ ) and iboga (___ ) lines. Genera of un-
certain affinity are marked C1. Near the origin of the plot, on the corynanthe
line (black dots): Excavatia; on the aspidosperma line .(black squares):
Geissospermum, Diplorrhynchus, Gonioma, Rhazya, Picralima, Amsonia;
on the iboga line (circles): Peschiera, Stemmadenia, Rejoua, Ervatamia

original arrangement (corynanthe) or suffered two kinds of rearrange-

ment (aspidosperma, iboga).
The alkaloid chemistry, available as at December 1975, of 284 species
belonging to 36 genera of the Plumerioideae, suggests the existence of
three lines which, in order, indicate presumably progressive specializa-
tion (Fig. 7). (1) The corynanthe line which develops by oxygenation
of corynanthe type alkaloids. The peripheral genus Cabucala contains
already the biosynthetically simplest of the aspidosperma type alkaloids.
(2) The aspidosperma line, characterized, in addition to corynanthe
Biochemical Systematics: Methods and Principles 69

alkaloids, by aspidosperma type alkaloids. After an initial extension (2.1)

formed by genera showing little differentiation, the line branches into 2.2,
which develops by skeletal diversification, and 2.3 which develops by
skeletal and oxidative diversification. (3) The iboga line, characterized,
in addition to corynanthe and aspidosperma alkaloids, by iboga type
alkaloids. Oxidative diversification is restricted to iboga (and aspido-
sperma) constituents.
Now let us compare again chemical lines and morphological tribes.
The picture here is very clear. The genera of Rauvolfieae fall along
line 1, with the exception of Vallesia and Kopsia which seem closer to
line 2. The genera of Alstonieae form line 2 which sustains a branch
occupied by the genera of Carisseae (line 2.2). Although the peripheral
genera Alstonia and Catharanthus produce already inclusively iboga
type alkaloids, they should still belong to line 2 in view of the typical
oxidative diversification of their corynanthe alkaloids, absent from the
genera of line 3. The connection of Haplophyton to line 2 is unclear.
The genera of Tabernaemontaneae form line 3.

Benzylisoquinoline Lines in the Magnoliidae. Benzylisoquinoline

alkaloids occur widely scattered throughout the Magnoliophyta. As
shown by SIs, however, only in the orders of the subclass Magnoliidae
does structural diversity and natural distribution justify their use as
systematic markers. The biosynthesis of benzylisoquinolines involves,
as PPs, two tyrosine units which form a benzyltetrahydroisoquinoline
(BTIQ), the precursor of their BG.
The alkaloid chemistry of 132 genera, available as at December 1975,
suggests the evolution of the families of Magnoliidae along two inter-
connected lines (Fig. 8). (1) The aporphine line which develops by skeletal
and oxidative diversification of aporphine type alkaloids. The peripheral
pairs Magnoliaceae-Annonaceae and Monimiaceae-Lauraceae produce
already sporadic examples respectively of berberine and morphine
alkaloids of simple type and oxygenation. (2) The berberine line which
develops by skeletal and, in the case of subline 2.1 including the periph-
eral Papaveraceae and Fumariaceae, also by oxidative diversification
of berberine type alkaloids. These families are differentiated by the
capacity of diversifying skeletons and oxygenation of aporphine type
alkaloids, peculiarity of the Papaveraceae. Both produce only simple
morphines, in contrast to Menispermaceae on subline 2.2, characterized
by skeletal and oxidative variants of the aporphine and berberine, as
well as of morphine-type alkaloids.
The chemical lines of the affinity plot (Fig. 8) and the morphological
relations proposed by CRONQUIST (1968) for the orders Magnoliales·Ranun-
culales-Papaverales, agree down to details, such as the placement of Magnolia-
70 M. APARECIDA H. CAGNIN et al. :

ceae·Annonaceae and Monimiaceae-Lauraceae in separate clusters, and the

opinion that the Papaveraceae and the Fumariaceae are to be regarded as
parallel groups which show different individual specializations. Even the
position of Ranunculaceae on the affinity plot and CRONQUIST'S statement
that there is no obvious reason why the Ranunculaceae might not be ancestral
to all other families of the order are contradictory only in appearance.
Indeed, if Thalictrum, chemically as well as morphologically (TAMURA

Menispermoceae

120

9Papaveraceae

/
:_ Annanaeeae I
80
/
I
I
I
I
o Ranunculoceae I
I
I
·_Laurectae I
.lM·~~;~·;aceQe
I

40 I
I
, Moonolioceae I
I

1___ ___
I

o Nelumbonoceae ---------OFumOrioceae

Aristolochlacea8 Berberl!a~!...o!
~H.rnDndiac.a. ___ --- ----
_1 __ o.:-----Ranunculaceae line Tholictrum
20 40 60 80 100
X.10- 2

Fig. 8. Affinity plot of Magnoliidae families mapped along aporphine

(......), berberine (___ ) and berberine/morphine (___ ) lines

1962) the most advanced genus, is omitted from our calculation of average
chemical composition, the point representing Ranunculaceae falls precisely
on the line leading to Berberidaceae (Fig. 8).

Alkaloid Lines in the Amaryllidaceae. The Amaryllidaceae contain

a special type of alkaloids which has been isolated but very rarely from
other sources. Their biosynthesis involves, as PPs, phenylalanine and
tyrosin which form a benzylphenylethylamin, the precursor of their BG.
The alkaloid chemistry of 145 species belonging to 29 genera, avail-
able as at December 1975, suggests the existence of two distinct lines
within the Amaryllidaceae (Fig. 9). Development of the primitive stock,
Biochemical Systematics: Methods and Principles 71

Hobtaftthul
•
60,0

30.0

;."
.....,.: - • Crinum

Fig. 9. Affinity plot of Amaryllidaceae genera mapped along crinine ( )

(line 1) and lycoricidine lines (_ _ ) (line 2). Near the origin of the plot,
on the lycoricidine line (black squares): Eurycle8, Cyrtanthus, Calo8temma,
Eustephia, Eli8ena

which already contained representatives of both, was accompanied

either by modification predominantly of the phenylethyl moiety of
crinine type alkaloids, or by modification of the benzyl moiety of lyco-
ricidine type alkaloids.
72 M. APARECIDA H. CAGNIN et al.:

Now again, how does the chemical classification compare with

morphological evidence? The Euchareae appear as basic group which
leads either to genera pertaining to Orineae and to the genus Amaryllis
(line 1) or to a series of tribes (line 2). The most conspicuous fact here
concerns only the separation of Orineae and it seems worthwhile to
examine their relatives for possible morphological affinity. This sugges-
tion gains momentum through the observations of TRAUB (1957) con-
cerning the occurrence of hybrids Amaryllis X Orinum, Amaryllis X
Brunsvigia.
On presently available evidence, however, it must appear to the
morphologist that each tribe is equipped to perform its own parallel
chemical evolution, while it does appear to the chemist that each line
is equipped to perform its own parallel morphological evolution. The
analysis looks, nevertheless, at least partly meaningful.

Coumarin Lines in the U mbelliferae. Coumarins occur widely scattered

throughout the Magnoliophyta. The present section is concerned,
nevertheless, only with their use as systematic markers in Umbellilerae,
one of the two families in which they possess adequate structural diver-
sity. Biogenetically, all coumarins of Umbellilerae seem to be derived
from umbelliferone. Unsubstituted coumarin is synthesized by a different
route and hence does not belong to the same BG.
The coumarin chemistry, available as at December 1975, of 245 spe-
cies belonging to 47 genera of the Umbellilerae suggests the existence
of 3 lines (Fig. 10). (1) The coumarin line, which, as the other lines,
develops by progressive oxygenation of the coumarin skeleton. The
peripheral genus Bupleurum contains an angular furocoumarin. (2) The
linear furocoumarin line, divided into several sublines according to
co-occurring types of skeletons. The peripheral genera of subline 2.1,
Anethum and Apium, contain in addition two pyranocoumarins. (3) The
pyranocoumarin line, subdivided according to the angular or linear
nature of the skeletons.
Here, the points representing genera of particular tribes mostly do
not occupy particular lines. They are confined to circular sections,
which we nickname "annual rings" since they suggest evolutionary
development. Thus the genera of Peucedaneae occupy all sections,
Apieae all sections minus the outermost, Smyrnieae all sections minus
the two outermost, etc. The present situation, in which representatives of
several tribes contain compounds derived through several lines, is con-
sidered to be a sign of phylogenetic advancement for a family.
This recalls one of the recurring themes of our results: the whole
basic chemistry is imprinted on organisms at the beginning of the
taxon's evolutionary history. The possibilities of variation along one
Y.10-2

Angelica p /1
'/
10 //
1
I
/
120 1
1
/
/ b::l
I
Heracleum 1
/
1
/
/ Q

/
/
t
eo
80 /
/
/
/
/ Poucodanum f
~Bup'eurum
/ /'" ~
....
I ~/~ Q

// ~--~~/ '"
/
Pimpinella !
/ /
/
/
//
..... / /
..... / - '

40 / ........ / -
Prangos
Cnidium //
/
f
[
Llgu." cum
--- ~
40 60 80 100 150
i
X.10-2

Fig. 10. Affinity plot of Umbelliferae genera mapped along coumarin C ......... ..), furanocoumarin ~) and pyranocou,
marin (____) lines. Near the origin of the plot, on the coumarin line: A8trodisco8, DaUCU8, Scandix, Falcaria,
Thapsia, A8tydamia; on the furanocoumarin line: TrachY8permum, Petro8etinum, Foeniculum, Ooriandrum, Oicuta,
.....
w
Smyrniop8is, Levisticum, Laser, Phellopterus, Ferulago, Malabaila, Leptotaenia, Zozimia, Apium, Oymopterus, Oonium,
Anethum, Oachry8; on the pyranocoumarin line: Oapnophyllum, Pteryxia, Laserpitium. 10. Peucedaneae, 9. Apieae,
8. Smyrnieae, 11. LaBerpitieae, 7. Ooriandreae, 6. Scandiceae, 12. Dauceae
74 M. APARECIDA H. CAGNIN et al.:

particular biochemical pathway being exhausted, the line initiates

variation along another. In time, units along all lines will produce
compounds through several pathways of a BO. Thus, for morpho-
logically differentiated groups, clearcut chemical differentiation (see
Magnoliidae, Leguminosae, Apocynaceae) indicates primitiveness, where-
as blurred chemical differentiation (see Umbellilerae, Amaryllidaceae)
indicates advancement.

Second Principle
While the first principle was based, at least partly, on a correlation
with current morphological evolutionary thought, no such bias influenced
the deduction of a working hypothesis for a second principle: among
taxa of Embryobionta of low hierarchical level, biochemical evolution
of secondary metabolites involves gradual specialization by substitutional
or/and skeletal diversification of compounds within biogenetic groups.
Indeed, to summarize the proposed procedure, initially the evolu-
tionary sequences of compounds are mapped on biosynthetic (relative
closeness of structure to ubiquitous PPs indicates relative primitiveness
of a compound) and distributional (relative dispersion of occurrence
indicates relative primitiveness of a compound) grounds. Next, taxa
(presently we work with families or genera, but the method is, of course,
equally well suited for species or orders) are considered as average
conditions of the evolutionary stages of the compounds contained.
This results in the sequencing of taxa which are, finally, examined in
the light of existing morphological schemes. The operations are thus
performed in the order suggested by HEYWOOD (1973), to avoid elements
of circularity in the argument.

It is hoped, therefore, that the independence of our chemical method

and accepted morphological evidence is such as to warrant mutual compari-
son of information. The examples cited in which such a comparison was
performed show that, in a general way, within a biogenetic group, given a
sequence of metabolites, A -+ B -+ C ~ D (see p. 63), if a taxon contains C,
one is able to state that it may have evolved from a taxon accumulating
B. Neither A nor B need be accumulated by the more advanced taxon;
their transitory existence is assured by the biosynthetic mechanism.
Clearly, it is again proper to enquire about the possibility of reversal.
Although this may occur, we have come across so few cases which may
function as possible examples, that it must be considered a minor evo-
lutionary trend.

Third Principle
The precursors of all BOs are of course present in all plants where
they provide essential macromolecules. It is thus understandable,
or even plausible, that chemical characters are polyphyletic, i.e. that
Biochemical Systematics: Methods and Principles 75

compounds pertaining to any BG may appear widely scattered through-

out the plant kingdom. Clearly, however, only if secondary metabolites
are accumulated in certain plant groups will they serve a useful
purpose in biochemical systematics. It may appear, a priori, that the
distinction between haphazardly distributed metabolites and specially
accumulated metabolites would be difficult or even impossible. This
is not the case. Ubiquitous or widespread compounds usually remain
structurally close to the primary precursors, while useful markers
undergo structural variation through chemical evolution. As already
noted, evolutionary vigour of a BG may be ascertained by SIs.
One and the same species usually contains, against a background of
chemosystematically irrelevant metabolites, only a limited number of
relevant BGs. Different BGs with analogous functions may be produced
by closely related plant taxa, according to environmental conditions,
resulting in chemical vicariance. Enhancement of products along one
biosynthetic route seems to trigger a regulating mechanism which
suppresses the formation of compounds along another.

Morphological vicariance, due to differential environmental conditions,

is a well know phenomenon. We have ascertained that the chemical rela-
tionship of morphologically vicariant species can be closer than the chemical
relationship of congeneric species of the same habitat (GOTTLIEB & STE-
FANI 1970). Analogously, the morphological relationship of chemically
vicariant taxa may be closer than the morphological relationship of taxa
of similar chemical composition. An understanding of the interaction of
environment and secondary plant metabolism, incipient at the present
time, will have to be reached before the reasons for chemical vicariance
will become clear.

All this leads to a series of statements which we embody presently

in a third principle of biochemical systematics: in view of the ubiquity
of BG-precursors, presence of certain compounds in selected taxa
gives no assurance as to their affinity. Only if metabolites are considered
with respect to structural variation within their biogenetic route or
group may the analysis lead to a significant result. Even then, however,
due to chemical vicariance, difference in chemical composition of two
taxa says nothing about their lack of affinity.

Conclusion
Taxonomic methods (p. 54-62), applied to chemosystematic prob-
lems (p. 63-75), indicate that secondary metabolism is only moderately
useful, if employed as an auxiliary criterion in the 00nstruction of an
evolutionary classification. Indeed, at present, the interplay or con-
nection of form and chemistry of an organism is, at best, debatable.
76 M. APARECIDA H. CAGNIN et al.: Biochemical Systematics

Secondary metabolism, just as comparative morphology is an indepen-

dent, primary criterion. It is considered highly probable that phyletic
classifications of plants will be, before long, proposed by chemists, not
least because they promise to be more useful, than purely morphological
ones.

References
CRONQUIST, A., 1961: Basic Botany. New York: Harper & Row.
- 1968: The Evolution and Classification of Flowering Plants. London:
Nelson.
GEISSMAN, T. A., and CROUT, D. H. C., 1969: Organic Chemistry of Se-
condary Plant Metabolism, p. 533. San Francisco: Freeman, Cooper &
Co.
GOTTLIEB, O. R., and STEFANI, G. M., 1970: Xanthones from Kielmeyera
excelsa. Phytochemistry 9, 453-454.
HARBORNE, J. B., HEYWOOD, V. H., and KING, L., 1976: Evolution of
yellow flavonols in flowers of Anthemideae. Biochem. Syst. Ecol. 4,
1-4.
HASLAM, E., 1974: The Shikimate Pathway, p. 198. New York: Wiley.
HEYWOOD, V. H., 1973: The role of chemistry in plant systematics. Pure
Appl. Chern. 34,355-375.
POLHILL, R. M., 1976: personal communication.
REZENDE, C. M. A. da M., and GOTTLIEB, O. R., 1973: Xanthones as sys·
tematic markers. Biochem. Syst. 1, 111-118.
- , - and MARX, M. C., 1975: Benzyltetrahydroisoquinoline·derived al-
kaloids as systematic markers. Biochem. Syst. Ecol. 3, 63-70.
ROCHLEDER, F., 1854: Phytochemie, p. 260. In: Pflanzenchemie und Pflan-
zenverwandtschaft (MOLISCH, H., 1933), p. 3. Jena: G. Fischer.
TAMURA, M., 1962: Taxonomical and phylogenetical consideration of the
Ranunculaceae. Acta Phytotax. Geobot. 20, 71-81.
TRAUB, H. P., 1957: Classification of Amaryllidaceae. Subfamilies, tribes
and genera. Plant Life (Herbertia) 13, 76-83.

Address of the author: Prof. Dr. OTTO GOTTLIEB, Instituto de Qufmica,

Universidade de Sao Paulo, Caixa Postal 20780, 01000 Sao Paulo, Brazil.
Plant Syst. Evol., Suppl. 1, 77-95 (1977)
«:l by Springer-Verlag 1977

Botany Department, University of Queensland, Australia

Quantitative Studies of Inter-relationships Amongst

the Liliatae
By

H. T. Clifford, St. Lucia

Abstract: There is an immense amount of taxonomic data and new

information is accumulating rapidly, thereby making it increasingly dif·
ficult for taxonomists to comprehend more than a small amount of what
is available. Hence it is essential that advantage be taken of modern
computing facilities for storing and retrieving information but more
especially for generating' classifications. The introduction of such technology
should lead to a full exposure of the data and methodology thereby enabling
potential users of a classification to assess its value for the problem in hand.
As an example of the methodology a sample of eighty-eight liliate
families has been classified using data pertaining to fifty-one attributes
and using an intensely clustering sorting strategy. Of the four major groups
of families emerging from the analysis two, the Zingiberales and a group
of water plants corresponding closely with the Alismidae are well established
in traditional classifications. The two remaining groups did not agree closely
with any recognized grouping in established classifications but nevertheless
showed a strong internal homogeneity in that one is comprised largely of
wind·pollinated and the other mainly of insect-pollinated families.

Introduction
Taxonomy has long been investigated with the mystique of personality
playing a dominant role in decision making.· Such a viewpoint was clearly
expressed by HUTCHINSON (1960) when he stated "the delimitation of
families, of genera and of species is sometimes very much a matter of
taste and personal idiosyncracy but· I would also add of judgement and
experience". A similar sentiment is echoed by AIRy-SHAW (in WILLIS
1973) in the preface to his influential Dictionary where he states that
taxonomy is a "personal matter".
In like vein HEYWOOD (1974: 44) has written as follows in connec-
tion with what he sees as the failure to reconcile traditional and numerical
taxonomists: "Part of the answer, at least, lies in the extent to which
traditional methods of classification have involved such subtle processes
as Gestalt perception and a related phenoml'non, typology, which bring
78 H. T. CLIFFORD:

into play complex human mental processes which it is difficult to appre-

ciate and analyse, let alone machine copy." The suggestion that percep-
tion is at the basis of the taxonomies of all cultures has also been advo-
cated recently by DWYER (1976).
As perception is an intensely personal matter it is not surprising
that taxonomists sometimes expound their opinions with an almost
fanatical fervour. To avoid misunderstanding it is therefore essential
that any taxonomic work should include a statement as to the reason
for its production. The reasons may well differ from person to person
and from time to time with the same person. For example, it may be
intended that the classification reflect the probable phylogeny of the
group in question, the breeding relationship of its members, provide
well defined morphological groups useful for making subsequent identi-
fications or generate groups of maximum predictive value, to mention
but four possibilities. How the classification was generated should
also be recorded together with an account of the attributes used and
those neglected or rejected. The declaration of all three of these matters
will enable the users of taxonomic systems to assess their reliability
for the tasks to which they may be applied. For as MAcKAy (1969)
has observed, pattern is meaningful only to the agent who seeks it and
a classification of value in one situation may be valueless in another.
Because taxonomic studies are fundamental to so much biology
it is important that they should be as objective as possible. It would
seem more important that a monographer or reviser of a group should
discuss all the attributes studied rather than to list those regarded as
diagnostic, and to state how much data was available rather than to
list all the specimens examined. The attitude adopted to the formation
of taxonomic groups should also be stated. Were they defined in terms
of internal consistency- or in terms of discontinuities between groups
of equal rank? Those following after would then be able to build on
rather than repeat much of the work. To this end it is essential that
adequate records be kept and since many of these would be unacceptable
to journal editors means of establishing and maintaining data banks
must be sought.
Such data banks would also reveal the considerable extent to which
extrapolation inevitably creeps into classifications. For example, it
seems odd that cotyledons should serve as a basis for classification or
nomenclature when the seedlings of so few of the members of the class
have been described, especially as the attribute is neither constant
within nor confined to the class.
Most classifications are based on a wide range of attributes and then
one or few diagnostic attributes are chosen to circumscribe the groups
recognized. The tradition of extrapolating diagnostic attributes is of
Quantitative Studies of Inter-relationships Amongst the Liliatae 79

long standing, for example, JUSSIEU (1789) has algae, liliates and magno-
liates in his sixth order of the Acotyledons, the Naiades, indicating both
that he had seen few seedlings of the members of the order, and that
he based his grouping on other and undefined attributes. Likewise,
J?E CANDOLLE (1813) included the cycads amongst his Endogens or
Monocotyledoneae possibly because of their palm-like habit.
One means of implementing data banks would be to store the accu-
mulated information on magnetic tape and to use the computer as the
storage-retrieval system. Programs for these purposes are available
but before they can be used efficiently, considerable thought must be
given to procedures for updating the records, keeping track of their
reliabilities and for both adding to and deleting information with the
passage of time. Data recorded as appropriate at one time may be
inappropriate at a later date because the purpose for which the informa-
tion was required may change. Thus seeds possessed of an aril have
been described for many species but the nature of the aril is not the
same for all species.

For some users, including ecologists, the possession or otherwise of

an aril by a seed may alone be significant, but for a phylogenetically
interested taxonomist the ontogeny of the aril will be important. Hence to
a considerable extent the kinds of data accumulated will depend upon the
likely requirements of its users. The development of data banks has only
just begun and whether they are all to be computer based or published
more conventionally remains to be seen. To date the only large published
taxonomic data bank at generic level appears to be that prepared by
CLIFFORD & WATSON (1977) for the Poaceae of Australasia. It is clear
that patience and co-operation are required for the establishing of satis-
factory definitions and that the coding problem is still the "Achilles heel"
of numerical approaches to taxonomy (EL-GAzzAR &WATSON 1970).

In addition to its functioning as a data storage-retrieval system

the computer has an important role in the generating of classifications.
It is with respect to this function that there has been most apprehension
amongst practising taxonomists. Such distrust is difficult to appreciate
especially when considering higher level taxonomies where it seems
impossible for the human mind to assess all the available information.
To this end witness the few large families that have been monographed
this century.
Indeed JACOBS (1969) has maintained that in a single lifetime a
taxonomist is unlikely to be able to deal critically with more than about
1000 species. Hence, if his work is to be related to that of others some
assistance or co-operation is required. In this regard some chores are
more easily delegated than others. For example, it is not necessary for
monographers to write descriptions of thl'il" own taxa; nor, might it be
80 H. T. CLIFFORD:

added, their own keys, for such tasks could be readily undertaken by a
well trained plant morpholQgist. As was long ago pointed out by DIELS
(1924: 68), "Begrifflich und inhaltlich sind Phytographie und Systematik
zwei ganz verschiedene Dinge. Wer sie vermengt, kennt beide nicht."
However assistance in these ways leads only to amplification of the
descriptions of recognized taxa and doesn't really help with the general
problem of establishing classifications in which relationships may be
expressed. It is here that the computer has a central role to play in that
the user may specify the bases on which the taxa are to be grouped or
divided, the amount of similarity or dissimilarity to be allowed within
and between groups and the relative importance to be assigned to each
attribute. Such objectivity is to be welcomed especially as the right to
criticize and amend the results still rests in the hands of the taxonomist.
It is merely that the data and methodology are fully exposed.
Further advantages are that the process of classification may be
made interactive with the computer thereby enabling the taxonomist
to interrupt the analysis at any stage. For all but the largest data matri-
ces the cost of producing a classification or ordination is slight, and in
any event much of this cost involves the printing of the results. Hence
if these are displayed visually the consequences of changing the attributes
or taxa can be investigated almost instantaneously. Such a facility is
particularly useful for the study of higher level taxa where, for example,
some of the groups generated may possess only one state of a two- or
more-state attribute. To be able to so control the classificatory program
as to be able to experiment, in a matter of seconds, with different strategies
is an exciting prospcct.
A less exciting, but none-the-Iess important, aspect of computer aided
taxonomy stems from the ease with which descriptions of groups may be
generated and contrasted with one another. Such comparisons may
direct the attention of the user to errors in the data or to the discriminatory
value of attributes not previously appreciated as being taxonomically
significant.
In what follows, the results of subjecting a sample of liliates to a
number of numerical classificatory procedures are compared with their
relationships as expressed in two recently proposed classifications.
The higher level taxonomy of the Liliatae is by no means stable and
unfortunately it is impossible to decide with certainty which if any of
these classifications is correct. Furthermore, in the absence of a reliable
fossil record it is not possible to speculate with confidence on the phy-
logeny of the class. Finally, duc to a lack of sound theories of morpho-
genesis it is, for the majority of attributes, rarely possible to predict
with certainty which are their primitive and which their advanced
states.
Quantitative Studies of Inter-relationships Amongst the Liliatae 81

Hence, there are difficulties in deciding which structures are alike

because of their sharing a common ancestry and which are alike because
though differing in origin they are responding in a similar manner to the
same type of selection pressure. Lack of agreement as to which are ad-
vanced and which primitive characters is frequently the basis of the dif-
ferences observed between many of the proposed phylogenetic classifi-
cations. Such a lack of agreement is aggravated in that determination of
relationships depends largely on the framework within which the comparisons
are made and so reflects the interests of the individual taxonomist. Thus an
authority' on sedges might regard the carices as a family (Kobresiaceae)
quite separate from the Oyperaceae as did GILLY (1958) whereas a reviewer
of the flowering plants would probably regard them as different only at
subfamilial level. That is the nature of the classification cannot be divorced
from the viewpoint of the observer and depends upon whether the system
is being viewed globally or locally.

The choice of a sampling system for selecting both taxa and attrib-
utes poses problems. Here, in order to ensure that the total variability
within the class has been encompassed, most families recognized during
this century have been accepted as valid and each has been scored using
data from any or all of their species. Alternative schemes of sampling
are to choose a subset of species at random from the total set of liliates
or to choose of species in proportion to the numbers of species per family.
The former alternative is unrealistic in demanding a list of all species of
Liliatae and the latter is unrealistic in that small families are oversampled
or completely sampled in comparison with large families. The families
selected for study are listed in Table 2.
In circumscribing the class Liliatae a rather conservative viewpoint
has been adopted and disputing contenders for membership such as the
Nymphaeaceae and Petrosaviaceae have been neglected. Their considera-
tion would be more appropriate to the problem of classifying the Magno-
liophyta than to the problem in hand.
The choice of attributes has been determined by two principal
considerations, availability of information and the degree of agreement
or otherwise as to which structures and organs are homologous. Compar-
isons between taxa are most usefully made in terms of structures known
to be homologous, but, as has already been noted there are difficulties in
determining these homologies. Because there is dispute as to the nature
of the perianth in different families few such data have been recorded.
Furthermore, even when structures appear to be homologous they may
originate in quite different ways as has been discussed by TOMLINSON
(1969) for the paracytic stomates of the Bromeliaceae and Pandanaceae.
No conscious attempt has been made to select attributes on the
basis of their supposed taxonomic usefulness for such judgements cannot
be properly made until the classification has been established or accord-
ing to CRONQUIST (1968: 10) "perceived". The classification in turn
Plant Syst. Evol., Suppl. 1 6
82 H. T. CLIFFORD:

Table 1. The attributes and their states

Attribute States

1. -Endosperm nuclear, helobia~, 9(lllular

2. Embryo sac crassinucellar,tenuinucellar,
pseudocrassinucellar
3. Tapetum glandular, amoeboid
4. Embryogeny caryophyllad, asterad, onagrad, chenopodiad
5. Ovule form orthotropous, anatropous, hemianatropous,
campylotropous, anacampylotropuos
6. Origin of micropyle inner integ., both integ., neither integ.,
outer integ.
7. Secondary antipodal reported, not reported
nuclei
8. Sporangia per anther 2, 4, more than 4
9. Pollen nuclei 2,3
10. Anthers sessile or basifixed, dorsifixed
11. Pollen grains single, tetrads
12. Ovules per placenta 1, more than 1
13. Ovary Superior, subinferior, inferior
14. Placentation basal, axile, parietal, pendulous, marginal,
scattered
16. Anther dehiscence extrorse, lateral, introrse
17. Pollen aperturate, non ·aperturate
30. Pollen aperture distal, zonal, global
18. Pollen mother· cell simultaneous, successive
tetrad formation
19. Pollen shape thread·like, otherwise
20. Pollen aperture porate, colpate, spiroaperturate
21. Seeds arillate, nonarillate
22. Seeds operculate, non .operculate
23. Anthesis protandrous, protogynous, homogamous
15. Endosperm scarce, abundant
24. Endosperm ruminate, non·ruminate
28. Endosperm starchy, otherwise
25. Placentae per pistil 1,2,3,4, more than 4
26. Germination remotive epigeal, remotive hypogeal, admotive
27. Seedling axis bearing cataphylls, lacking cataphylls
29. Perisperm present, absent
31. Embryo macropodous, non.macropodous
32. Leaves 2·ranked, more than 2·ranked
33. Ligule present, absent
34. Leaf·base sheathing, not sheathing stem
35. Venation reticulate, convergent, pinnate
36. Guard ·cells present, absent
37. Guard·cell shape graminoid, otherwise
38. Subsidiary cells 0, 2, more than 2
39. Vessels (roots) present, absent
40. Vessels (stems) present, absent
41. Vessels (leaves) present, absent
42. Velamen present, absent
Quantitative Studies of Inter-relationships Amongst the Liliatae 83

Table 1 (continued)

Attribute States

43_ Hypodermis (leaf) present, absent

44_ Raphide bundles present, absent
45_ Embryo in seed basal, axile
46. Silica present in epidermis, absent from epidermis
47. Secondary growth occurring, not occurring
48. Squamulae present, absent
49. Root endodermis as in Iris, otherwise
50. Vernation conduplicate, convolute, plicate, involute, flat
51. Fruit achene or nut, berry, capsule, drupe, schizocarp

depends upon the sample of families investigated. For example, attrib-

utes constant within groups have little taxonomic value. Hence the
possession of two cotyledons is of little significance amongst a sample
of magnoliates but would be of considerable significance if the sample
included liliates as well. It ought to be noted that amongst a wider
sample of vascular plants two cotyledons would not necessarily be a
diagnostic character.
Aside from such theoretical considerations, comparisons between
families are also severely restricted by the lack of comparative family
descriptions. Unfortunately, these descriptions often contain much
extrapolated data and mistakes that have long passed unchalIenged.
To expose such misinformation demands recourse to the study of actual
plant material. Whilst this is in itself a pleasurable task it is at best
time consuming and at worst frustrating when the plant does not grow
locally. Of the fifty-one attributes scored for numerical analysis, thirty-
two refer to reproductive and nineteen to vegetative attributes. These
attributes and their states are defined in Table 1.
Of the total possible records for the chosen attributes about 75%
have been obtained. To quote alI sources from which the data were
acquired would result in an immense reference list and so only major
references will be cited. Wherever possible the data have been cross-
checked either with other references or with plant material. The prin-
cipal reference works consulted were as folIows: embryology (DAVis
1966); anatomy (TOMLINSON 1969, CUTLER 1969); seedlings (BOYD 1932);
vessels (CARLQUIST 1975); biochemistry (GIBBS 1974, HEGNAUER 1963);
seeds (MARTIN 1946); morphology (ARBER 1921).

Classificatory Procedures
As a first step in the analysis the dissimilarities between the families,
considered in pairs, were calculated using an information-gain statistic.
Clustering was then undertaken on the basi~ that at each stage in the
6·
84 H. T. CLIFFORD:

fusion cycle there was a mInImUm increase in diversity on fusion.

The procedure is highly group-sized dependent (CLIFFORD & WILLIAMS
1973) but has an advantage in that it results in sharply defined groups.
The relationships between the groups cannot be read directly from the
dendrogram and so the inter-family dissimilarities were subjected to a
principal co-ordinate analysis (GOWER 1967) which is an ordination
based on the inter-taxa dissimilarities. From the results of the two
analyses a set of four major sub-groups within the Liliatae was recognized.
Having established the groups their properties were investigated
in two ways. For each division of the dendrogram the frequencies of the
attribute states on either side of the division were determined; for each
of the first three principal co-ordinate axes the attributes most closely
correlated therewith were determined.
For most numerical taxonomic studies these analyses are adequate
in that they are the minimum and yet shed light not only on the classi-
fication. of the taxa but also indicate the role played by the attributes
in generating that classification (CLIFFORD, WILLIAMS & LANCE 1967).

It is unfortunate that the comparisons between families could not be

made with complete data as the classification may differ according to the
attribute set employed (CLIFFORD & LAVARACK 1974, CLIFFORD 1976).
However, to restrict the analysis to attributes known over all families
would have so reduced the attribute set as to make any discussion of results
of little value. For most families a wide variety of data was available so
the bias likely to be introduced through the use of special subsets of data
has been reduced.

Discussion of Results
The results arising from the application of the clustering strategy
are summarized at the ten-group level in Fig. 1, and the families con-
stituting each group are listed in Table 2. From the dendrogram it is
clear that four major groups may be recognized and that their members
differ markedly with respect to their habitat preferences and modes of
pollination. Those attributes by which the groups may be distinguished
vary from branch to branch of the dendrogram and since the classifica-
tion is polythetic the groups are defined mostly by their joint possession
or lack of several attribute states.
No detailed discussion will be undertaken of the family assemblages
within the ten groups. However, the inclusion of the Corsiaceae amongst
the water-plants calls for special comment. There were relatively few
data available for the family and these were mostly attributes common
to the class and so its presence in group 1 is subject to dispute. Early in
the analysis the Corsiaceae became associated with the Aponogetonaceae
and the somewhat variable Hydrocharitaceae with which it remained.
Quantitative Studies of Inter-relationships Amongst the Liliatae 85

That it is anomalous in the group is also supported by its position relative

to the other members of group 1 as shown in the ordination diagram
(Fig_ 2). Even so, it is interesting to note that DEYL (1955) included
the Corsiaceae in the same order as the Triuridaceae, a family CRONQUIST
(1968) considered as related to the water plants.

300-

Aquatic Terrestrial

250-

POLLINATION
200f-

~I
150 f-
1 Animal Wind

VEIOR

Bird Insect
10Of-

,l '--

--
50
r--

0
2 3 4 5 6 7 8 9 10

Fig. 1. Dendrogram resulting from clustering the families of Liliatae,

truncated at the to-group level with the principal habitats and pollination
mechanisms of the groups superimposed

The prime division within the Liliatae separates the majority of the
aquatic members (marine and freshwater) from the remainder. Four
attributes are strongly associated with this separation and these are
summarized in Table 3 where the numbers accompanying them indicate
in percentages the frequencies of the attribute states in each group. That
the percentages sometimes exceed 100 reflects the fact that some families
include individuals possessing more than one state of the attribute.
Within the aquatic plant families two groups may be recognized in
86 H. T. CLIFFORD:

Table 2. The family compositions of the 10 groups defined by the dendro-

gram in Fig. 1. Following each family name are two numbers, the first is
that indicating its position in the alphabetical list of family names, the
second the number of attributes for which the family has been scored

I. Mostly aquatic families (Alismidae)

1. Alismataceae 3, 46 2. Oymodociaceae 26,25
A ponogetonaceae 10,39 J uncaginaceae 44,38
Butomaceae 17,45 Lilaeaceae 46,26
Oorsiaceae 22,22 Naiadaceae 53,42
Halophilaceae 35,28 Posidoniaceae 62,24
H ydrocharitaceae 38,38 Potamogetonaceae 63,43
Limnocharitaceae 48,24 Ruppiaceae 67,39
Thalassiaceae 76,23 Scheuchzeriaceae 69,38
Zannichelliaceae 86,32
Zosteraceae 88,40

II. Mostly bird-pollinated, tropical, terrestrial families (Zingiberales)

3. Oannaceae 18, 47
Oostaceae 23, 42
Heliconiaceae 37,30
Lowiaceae 49, 28
Marantaceae 50, 39
Musaceae 52, 41
Strelitziaceae 73, 45
Zingiberaceae 87, 46

III. Mostly insect-pollinated, temperate, terrestrial families

4. Agavaceae 2,48 5. Asparagaceae 14,41
Alliaceae 4,35 Bromeliaceae 15,47
Aloeaceae 5,37 Burmanniaceae 16,35
Alstroemeriaceae 6,42 Oommelinaceae 21,49
A maryllidaceae 7,49 Dioscoreaceae 28,45
Aphyllanthaceae 9,33 Geosiridaceae 32, 15
A postasiaceae 11,29 H aemodoraceae 34,43
Gilliesiaceae 33,30 I sophysidaceae 41,21
Hypoxidaceae 39,40 Liliaceae 47,42
Iridaceae 40, 45 Ruscaceae 68,35
Orchidaceae 54, 43 Smilacaceae 70,32
Petrosaviaceae 57,29 Tecophilaceae 75,33
Philesiaceae 58,37 Trichopodaceae 79,35
Phylidraceae 59,38 Velloziaceae 73,37
Thismiaceae 77,21
)[anthorrhoeaceae 84,44

6. Araceae 12,49 7. Oroomiaceae 24,27

Lemnaceae 45,38 Roxburghiaceae 66,20
Petermanniaceae 56,3,3 Stemonaceae 72,33
Taccaceae 74,40
Trilliaceae 80,37
Quantitative Studies of Inter-relationships Amongst the Liliatae 87

Table It (continued)

IV. Mostly wind-pollinated terrestrial families

8. Abolbodaceae 1, 31 9. Anarthriaceae 8,33
Oartonemataceae 19,35 Oentrolepidaceae 20,41
Thurniaceae 78,30 Ecdeiocoleaceae 29,23
Eriocaulaceae 30,43 Flagellariaceae 31, 39
Xyridaceae 85,45 J oinvilleaceae 42,30
Hanguanaceae 36,31 Rapataceae 64,33
Juncaceae 43,46 Restionaceae 65,37
Mayacaceae 51,36
Pontederiaceae 61,47
Triuridaceae 81,28

10. Arecaceae 13,49

Oyclanthaceae 25,39
Oyperaceae 27,48
Pandanaceae 55,42
Poaceae 60,50
Sparganiaceae 71,43
Typhaceae 82,44

terms of the number of ovules per placenta in each group. All members
of group 2 have one and all members of group 1 have several ovules per
placenta.
Amongst the terrestrialliliates there are two major groups which are
behaviourly well separated in that one (8, 9, 10) has largely wind-pollin-
ated and the other (3, 4, 5, 6, 7) largely animal-pollinated members.
The four attributes by which these two groups may be distinguished
are summarized in Table 4.
The animal-pollinated liliates in turn comprise two distinct groups
of families one of which (3) contains a high proportion of bird-pollinated
and the other (4, 5, 6, 7) a high proportion of insect-pollinated species.
The former is also largely tropical in contrast to the latter being largely
temperate. The principal distinctions between the two groups are summa-
rized in Table 5.
Because the Zingiberales and Alismidae are clearly defined in the
cluster analysis and also stand out from the remaining families In the
principal co-ordinate analysis (Fig. 2) it was decided to reanalyse the
data excluding these two groups of families. A clustering analysis only
was undertaken and this confirmed the lack of well defined groups in
the "insect-pollinated" section of the liliates.
Within the largely insect pollinated families it is inappropriate to
recognize any further groupings above those resulting from truncating
the dendrogram at the ten-group level. That is because amongst the
88 H. T. CLIFFORD:

families of the four groups 4, 5, 6 and 7 there are no attributes which show
any marked correlations. For the same reason no attempt has been made to
recognize major groupings between the three groups (8, 9 and 10) of
largely wind-pollinated plants.

m------------------~~~~~~---------
e18 37 e24
eSO 66
,13
z
32 -72
e e 56
e57 80
23 e5
21e e4O Ii ~4 e84 ~ ,7
e75 47
e 14
e15 e6
e34 e79

e28

I
Fig. 2. The positions of the families of Liliatae relative to axes I and III
of a principal co-ordinate analysis. The numbers refer to the families of
Table 2; Z, Zingiberales; HI (Group 1) and H2 (Group 2), Alismidae

A consideration of the attributes most closely correlated with the three

principal co·ordinate axes revealed that the three attributes most closely
correlated with the first principal co-ordinate axis were presence or absence
of squamulae, the nature of the anther tapetum and the presence or absence
of endosperm in the seed. Of these three, squamulae and endosperm attri·
butes were earlier noted (Table 3) as being highly diagnostic of the primary
division within the Liliatae. The three attributes most closely correlated
with axis two were the presence or otherwise of vessels in the stems and
Quantitative Studies of Inter-relationships Amongst the Liliatae 89

Table 3. The four attributes most diagnostic of the primary subdivision of

the Liliatae into largely "Aquatic" and largely "Terrestrial" families

Percentage Frequency
Attribute
Aquatic Terrestrial

Endosperm
present 6 94
absent 94 8
Squamulae
present 87 2
absent 13 100
Stomates
present 44 100
absent 56 0
Embryo
macropodous 64 0
otherwise 36 100

Table 4. The four attributes most diagnostic for separating the "Wind·
pollinated" group of families from the "Animal-pollinated" groups

Percentage Frequency
Attribute
Animal Wind

Ovary
superior 42 100
subinferior 4 0
inferior 56 0
Endosperm
starchy 20 87
otherwise 80 13
Subsidiary cells
0 49 0
2 42 96
more than 2 12 4
Fruit
achene,nut 4 25
berry 22 0
capsule 78 58
drupe 0 21

leaves and the shape of the embryo, basal or axile, none of which was
diagnostic of any of the four groups originating from the classificatory
study. The three attributes most closely correlated with the third axis
were the presence or absence of perisperm, the type of leaf venation and
90 H. T. CLIFFORD:

the occurrence or otherwise of a leaf hypodermis, all three of which are

diagnostic for distinguishing between "bird.pollinated" Zingiberales and
the "insect-pollinated" Liliales.

Table 5. The five attributes that serve to separate the Zingiberales from
the remaining families of animal pollinated families ("Liliales")

Percentage Frequencies
Attribute
Zingiberales "Liliales"

Perisperm
present 100 0
absent 0 100
Venation
reticulate 0 22
convergent 0 78
pinnate 100 2
Leaf hypodermis
present 100 14
absent 0 89
Subsidiary cells
0 0 60
2 75 34
more than 2 38 5
Ovary
superior 0 51
subinferior 0 5
inferior 100 46

General Discussion
The emergence from the analysis of four major groups is of interest
in that, two recent classifications, namely those of CRONQUIST (1968)
and TAKHTAJAN (1969) recognize four subclasses in the Liliatae (Table 6).
Though there is an overall similarity between the two classifications
there are sufficient differences to justify comparing them with each
other and with the computer-generated groupings of families.
In all three systems the water families (Alismidae, Alismatidae)
stand out clearly from the remaining liliates and are widely recognized
as being a distinctive group of families. The inclusion of the Triuridaceae
in the subclass is problematical in that CRONQUIST includes them, but
TAKHTAJAN does not, placing the family instead amongst the "lilies".
The numerical classification has placed them in yet a further place
amongst the group of largely wind-pollinated families.
Next to the water-fa~ilies the wind-pollinated palms and their
allies are probably the most consistently recognized high level liliate
Quantitative Studies of Inter-relationships Amongst the Liliatae 91

Table 6_ The classifications of TAKHTAJAN (1969) and CRONQUIST (1968)

compared

TAKHTAJAN CRONQUIST

Alismidae Alismatidae
Alismanae
Alismales Alismatales
H ydrocharitales H ydrocharitales
Najadales Najadales
Triuridales
Liliidae Liliidae
Lilianae
Triuridales
Liliales Liliales (inc_ Iridales)
Iridales
Zingiberales
Orchidales Orchidales
Commelinidae Commelinidae
Juncanae
Juncales Juncales
Oyperales Oyperales (inc_ Poales)
Oommelinanae
Bromeliales Bromeliales
Oommelinales Oommelinales
Eriocaulales Eriocaulales
Restionales Restionales
Poales
Typhales
Zingiberales
Arecidae Arecidae
Arecanae
Arecales Arecales
Oyclanthales Oyclanthales
Arales Arales
Pandanales Pandanales
Typhales

assemblage_ However, there is less certainty as to the orders to be

included in the subclass (Arecidae)_ The Typhales are included by
TAKHTAJAN and excluded by CRONQUIST_ The numerical classification
again supports TAKHTAJAN but disagrees with them both in that t,hey
also include the Arales in the same major grouping as the palms_
Although the Arales (Lemnaceae, Araceae) appear amongst the
"lilies" in Fig_ 1 a further ordination analysis of the data excluding
the Alismidae and Zingiberales indicated that even amongst the "lilies"
92 H. T. CLIFFORD:

they stand out clearly from the remainder. Thus whilst the Arales are
not to be associated closely with the palms they sit uneasily amongst
the "lilies".
Whereas the grasses and the sedges are included in the Oommelinidae
by both TAKHTAJAN and CRONQUIST neither of these families is associated
with the Oommelinaceae in the numerical analysis. Instead they are
associated with the palms and their allies. The reality of this association
is supported by its recurrence for it also appeared in an earlier analysis
(CLIFFORD 1967) based upon much less data. Indeed, the Oommelinidae
has been largely disrupted by the n).lmerical analysis and the Oom-
melinaceae have become associated with the lilies, a relationship long
ago suggested by WARMING (1920).
Two differences of opinion between TAKHTAJAN and CRONQUIST with
respect to the familial compositions of the Oommelinidae and the Liliidae
have been discussed. A third difference of opinion concerns their treat-
ments of the Zingiberales. These TAKHTAJAN places with such confidence
in the Liliales that he relates them to the family Liliaceae, subfamily
Asphodeloideae. With equal confidence CRONQUIST places the gingers in
the Oommelinidae stating that in his opinion "The only respect in which
the Zingiberales are more like the Liliidae than the Oommeliniidae is
that the vessels are confined to the roots instead of occurring in all
vegetative organs".
Given such divergent opinions it would seem that different attributes
are being regarded as more or less significant by each taxonomist. In
terms of the numerical analyses which give no weight to any particular
attribute the Zingiberales emerge as a single homogeneous group with
no close links to any other group. The resolution of such disparate
viewpoints depends upon the solution of two quite separate problems.
The first of these is whether evolutionary and morphogenetic theories
can be devised to provide better guides to phylogenetic trends and the,
second is, given a series of groups, which criteria are to be employed to
determine their rank. At present means of distinguishing between
parallel and convergent evolution appea~ to pose irresolvable difficulties
for predicting phyletic relationships and so attention might be more
profitably directed towards the determination of rank.
Earlier taxonomists such as ENGLER (1883) regarded the Zingiberales
(Scitamineae) as a group equivalent in rank to the Alismidae. Such a rank
is supported in particular by the principal co-ordinate analysis (Fig. 2)
which, unlike the dendrogram, gives some indication of the separateness
of taxa independently of the way in which they cluster. Ordination
procedures are none-the-Iess not always a guide to rank. It may be
argued that rank is itself unimportant, and that like other aspects of
taxonomy "is a matter of judgement and not susceptible to precise and
Quantitative Studies of Inter-relationships Amongst the Liliatae 93

rigid decision" (HEYWOOD 1974: 280)_ However, this view would be

supported neither by all taxonomists nor all ecologists, some of whom
such as MAAREL (1972) have been using taxa of different rank~ as a
basis for the classification of vegetation.
The application of numerical methods to the determination of rank
is bedevilled by a basic principle of taxonomy, namely that taxa are
described in terms of their attributes. Consider the following common
situation. For a series of individuals or taxa a 2-way table is drawn up
in which the rows are taxa and the columns their attributes (or vice
versa). Given such a table with entries in the cells it is customary to
classify the taxa in terms of their attributes. It could equally be of
interest to classify the attributes in terms of the taxa.
Indeed host-parasite relationships are often investigated both ways.
Taxonomists concerned with the hosts often accept the parasites as
indicators of relationships amongst the hosts. Likewise parasite taxo-
nomists often look to the hosts for evidence of relationships amongst the
parasites. In so doing each accepts the classification of the other and so
the argument becomes somewhat circular.
To avoid this impasse methods are being sought for classifying two-
way tables using both the row and column information (CLIFFORD 1975).
To date little success has been achieved largely because most of the
methods available demand quantitative scores. One basis for fixing
such scores would be to define the states of attributes in terms of their
degree of phylogenetic advancement or otherwise. Regrettably too few
attributes lend themselves to being scored this way. Those interested in
pursuing this matter from a classificatory viewpoint should consult
MAcNAUGHTON-SMITH (1965) and those interested in ordination appro-
aches, HILL (1973).
The principal advantage of investigating rank from a combined
consideration of both taxa and attributes is that it avoids undue emphasis
being placed upon either a particular taxon or a particular attribute. It
would also eliminate the necessity of working with dissimilarity measures
for which the basis of dissimilarity was neglected once the measure had
been calculated. To lose sight of the attributes at any stage is unfortunate
for "any useful system of classification must concern itself both with
the objects being classified and the characters used to differentiate the
groups" (CRONQUIST 1968: 26).
Until joint consideration of attributes and taxa is possible by other
methods resort must be had to ordination procedures such as principal
co-ordinate analysis, the power of which has been illustrated with respect
to the manner in which the Zingiberales were segregated from the liliate
families in Fig. 2.
Finally it Hhould be noted that the numerical approaches to the
94 H. T. CLIFFORD:

classification of higher taxonomic categories differ in no ways from those

appropriate to lower categories. At all ranks it is essential that taxo-
nomic procedures be made as explicit as possible. It is here that quanti-
tative methods have a role to play in removing the mystique from taxo-
nomy which should be seen not be an esoteric art practised by the few
(KUBITZKI 1975) but as a cornerstone of biology practised by the ma-
jority.

The writer takes pleasure in acknowledging the receipt of much friendly

advice from Mr. DAVID HASSALL and Drs. M. B. DALE and V. H. BOUGHTON
during the preparation of this paper.

References
ARBER, A., 1925: Monocotyledons: a morphological study. Cambridge:
Cambridge University Press.
BOYD, L., 1932: Monocotyledonous seedlings. Trans. and Proc. Bot. Soc.
Edinburgh 31, 5-223.
CANDOLLE, A. P. DE, 1813: Theorie elementaire de la Botanique. Paris.
In: The families of flowering plants (HUTCHINSON, J., 1960), Vol. I.
Oxford: Clarendon Press.
CARLQUIST, S. J., 1975: Ecological Strategies of Xylem Evolution. Berkeley:
Univ. California Press.
CLIFFORD, H. T., 1975: Host-parasite relationships. In: Lecture Notes in
Mathematics (STREET, A. P., and WALLIS, W. D., Eds.), 452. Combi-
natorial Mathematics III. Berlin-Heidelberg-New York: Springer.
1976: The influence of attributes in the classification of the grasses
(Poaceae). In: Pattern Analysis in Agricultural Science (WILLIAMS,
W. T., Ed) .. Melbourne and Amsterdam: C.S.I.R.O. and Elsevier
Scientific Publishing Coy.
- and WATSON, L., 1977: Identifying Grasses. Brisbane: Queensland
University Press.
- WILLIAMS, W. T., and LANCE, G. N., 1969: A further numerical con-
tribution to the classification of the Poaceae. Aust. J. Bot. 17, 119-311.
LAVARACK, P. S., 1974: The- role of vegetative and reproductive at-
tributes in the classification of the Orchidaceae. BioI. J. Linn. Soc. 6,
97-110.
and WILLIAMS, W. T., 1973: Classificatory dendrograms and their
interpretation. Aust. J. Bot. 21, 151-162.
CRONQUIST, A., 1968: The evolution and classification of flowering plants.
London and Edinburgh:. Nelson.
CUTLER, D. F., 1969: Juncales. In: Anatomy of the Monocotyledons. Vol. IV
(METCALFE, C. R., Ed.). Oxford: Clarendon Press.
DAVIS, G., 1966: Systematic Embryology of the Angiosperms. New York:
John Wiley and Sons.
DEYL, M., 1955: The evolution of plants and the taxonomy of the mono-
cotyledons. Acta Musei Nationalis, Pragae XIB No.6, Botanica No.3,
1-143.
DIELS, L., 1924: Aufgaben der Phytographie und der Systematik. In:
Handbuch der biologischen Arbeitsmethoden. Abt. XI, Teil 1 (AB-
DERHALDEN, E., Ed.). Berlin: Urban nnd Schwarzenberg.
Quantitative Studies of Inter-relationships Amongst the Liliatae 95

DWYER, P. D., 1976: Systematics, Ecology and Biological Resources.

Search 7,294-298.
EL-GAzZAR, A., and WATSON, L., 1970: Taxonomy of Labiatae. New Phyt.
69, 451-486.
ENGLER, A., 1883: Syllabus der Pflanzenfamilien. Auf. 3. Berlin: Born-
traeger.
GIBBS, R. D., 1974: Chemotaxonomy of Flowering Plants. 4 Vols. Montreal
and London: McGill-Queen's University Press.
GILLY, C. L., 1952: Phylogenetic development of the inflorescence and
generic relationships in the Kobresiaceae. Iowa J. Science 26, 210-212.
GOWER, J. C., 1967: Multivariate analysis and multi-dimensional geometry.
Statistician 17, 13----':'28.
HEGNAUER, R., 1963: Chemotaxonomie der Pflanzen. Bd.2. Basel und
Stuttgart: Birkhauser.
HEYWOOD, V. H., 1974: Chemosystematics-an artificial discipline. In:
Chemistry in Botanical Classification, Nobel Symposium 25 (BENDZ, G.,
and SANTESSON, J., Eds.). New York and London: Academic Press.
HILL, M.O., 1973: Reciprocal averaging: an eigen-vector method for
ordination. J. Ecol. 61, 237-250.
HUTCHINSON, J., 1960: Families of Flowering Plants. 2 Vols. Oxford:
Clarendon Press.
JACOBS, M., 1969: Large families-not alone! Taxon 18, 253-262.
JUSSIEU, A. L. DE, 1789: Genera plantarum secundum ordines naturales
disposita. Paris. In: Historia Naturales Classica Tome 35 (STAFLEU,
E. A., Ed., 1964). Weinheim: J. Cramer.
KUBITZKI, K., 1975: Systematics and evolution of seed plants. In: Progress
in Botany 35 (ELLENBERG, H., et aI., Eds.). Berlin-Heidelberg-New York:
Springer.
MAAREL, E. VAN DER, 1972: Ordination of plant communities on the basis
of their plant genus family and order. In: Grundfragen und Methoden
in der Pflanzensoziologie (MAAREL, E., VAN DER, and TUXEN, R., Eds.).
The Hague: Junk.
MACKAY, D. M., 1969: Recognition and action. In: Methodologies of
Pattern Recognition (WATANABE, S., Ed.). London: Academic Press.
MACNAUGHTON -SMITH, P., 1965: Some statistical and other techniques for
classifying individuals. Home Office Res. Unit Rep. No.6. London:
H.M.S.O.
MARTIN, A., 1946: The comparative internal morphology of seeds. Amer.
MidI. Nat. 36, 513-560.
TAKHTAJAN, A., 1969: Flowering plants, origin and dispersal. Washington:
Smithsonian Institution Press.
TOMLINSON, P. B., 1969: Oommelinales-Zingiberales. In: Anatomy of the
Monocotyledons. Vol. III (METCALFE, C. R., Ed.). Oxford: Clarendon
Press.
WARMING, E., 1920: A Handbook of Systematic Botany. London: Allan
and Unwin Ltd.
WILLIS, J. C., 1973: A dictionary of the flowering plants and ferns, 8th ed.
Revised by H. K. AIRy-SHAW. Cambridge: Cambridge University
Press.

Address of the author: Dr. H. T. CLIFFORD, Botany Department, Uni-

versity of Queensland, St. Lucia, Brisbane 4067, Australia.
Plant Syst. Evo!., Supp!. 1, 97-109 (1977)
© by Springer-Verlag 1977

Royal Botanic Garden, Edinburgh, U.K.

Classification Above the Genus, as Exemplified

by Gesneriaceae, With Parallels From Other Groups
By

B. L. Burtt, Edinburgh

Abstract: Gesneriaceae is used as the starting point for a consideration

of some aspects of classification between family and genus. Gesneriaceae
may be split into two subfamilies, one with normal. cotyledons, the other
with one cotyledon becoming enlarged after germination: a parallel is
drawn with the use of the plane of foliar distichy in Zingiberaceae. In both
cases the resulting improvements are shown to be vital to the clear recogni-
tion and statement of evolutionary problems.
The status of Klugieae, one of the 4 tribes of Gesneriaceae-Oyrtandroideae
is re-examined and its elevation to subfamily is discussed in relation to
the system of the family. There is a rather general trend against simple
dichotomy at subfamily level, but its retention in Gesneriaceae is recom-
mended. The complexity of various classifications between family and
genus is assessed by use of the ratio between number of genera and the
rank immediately above. (terminal suprageneric taxa) and the number
of the latter that are monogeneric. When these ratios are too Iowa system
looses much of its practical value.
The treatment of possible linking genera between major groups is
discussed in relation to Jerdonia (Scrophulariaceae/Gesneriaceae) and Tri-
plostegia (Dipsacaceae / V alerianaceae) . The existence of such links does
not justify the union of the families. In contrast, new evidence confirms
the very close relationship of Selaginaceae to Scrophulariaceae tribe Manu-
leeae. This is not dependent on a small linking genus: the affinity is much
closer than is that of Manuleeae to most other tribes of Scrophulariaceae,
and the reduction of Selaginaceae to tribal rank must therefore be upheld.

In the last 140 years the classification of Gesneriaceae has come full
circle. The family first took on its present form when G. DON (1838) united
Gesneriaceae (New World genera having seeds with endosperm) and
Cyrtandraceae (Old World genera with seeds lacking endosperm), keeping
them as separate tribes· with three and four subtribes respectively.
Some 40 years later BEN'l'HAM (1876) shifted the primary emphasis from
endosperm to the position of the ovary, and used characters of fruit and
disc for his subtribes. As a result, some tribes and subtribes for the first
time contained a mixture of Old World and Npw World genera. FRITSCH
Plant Syst. 1'\'01.. Suppl. 1
98 B. L. BURTT:

(1894-1895) followed BENTHAM closely, but with a little more elabora-

tion.
These were the two systems available for the family when I came to
it in the 1930s, through the late Sir ARTHUR HILL'S interest in the
unequal cotyledons of the seedlings. Almost subconsciously, my mind
registered that all the plants with interesting seedlings were from the
Old World: the New World genera had equal cotyledons and were
something different. Returning to the family in the 1950s, I automati-
cally accepted the Old World genera as a taxonomic group. The justifica-
tion emerged gradually: all the Old World plants examined be camp
more or less anisocotylous after germination. Neither HILL, nor FRITSCH
before him, had realized this. HILL was looking for major differences
between the two cotyledons in size and persistence, such as that shown
in Streptocarpus, Ohirita or Monophyllaea. In some genera (Boea or
Oyrtandra for example) the size difference is small and persistence slight.
However, the tendency is pervasive throughout the Old World genera,
except the Australasian Ooronanthereae. In the New World anisocotyly
is seen only in the few Central American species of the Old World genus
Rhynchoglo88um. The subfamilies Oyrtandroideae and Gesnerioideae
therefore had to be remodelled so that these different patterns of growth
were separated. This was in effect a reversion to the position of 1838,
when the crucial character of seedling morphology was quite unknown.
It should be emphasized that there are other, less constant, features that
help to differentiate the two subfamilies, and also that -the use of the
seedling character is not a barrier to easy identification. An artificial key
to the genera can easily be made without mention of cotyledons.
There was one immediate gain from this re-alignment. Several Old
World genera had for long been hovering uncertainly on the borders
of Ge8neriaceae. They all had some anomalous features, though they
mostly showed the unilocular ovary alleged to separate Ge8neriaceae
from Scrophulariaceae. For most of these genera, study of the seedlings
has now been possible, but they are still unknown in the Formosan
Titanotrichum. Five (Brookea, Oharadrophila, Oyrtandromoea, J erdonia,
and Rehmannia) prove to have normal isocotylous seedlings, and they
thus fail this test for inclusion in Oyrtandroideae. Now it is quite possible
that we shall find some good species of Oyrtandroideae which have equal
cotyledons: there are many hundreds not yet examined and a taxonomist
never expects one single character to hold throughout a large group.
But in the genera just mentioned other features make their inclusion in
Oyrtandroideae doubtful: the equal cotyledons confirm these doubts, and
these plants must be excluded. Of course, they might belong to the
isocotylous subfamily Ge8nerioideae, but, without going into details,
they arp all best accomodated in our present concept of Scrophulariaceae.
Classification Above the Genus 99

Thus the value of the anisocotylous character is confirmed. It is not

of sporadic occurrence: it is a trend throughout one subfamily. Usually
the large cotyledon shows only slight post-germinal growth, more rarely
it attains the size of an ordinary foliage leaf, while in at least two groups,
in the quite distinct tribes Didymocarpeae and Klugieae, it reaches
maximum expression as the single, and sometimes very large, foliage
leaf of the adult plant.
Something of a parallel in remodelling with the help of a vegetative
character may be seen in Zingiberaceae. In this family a tall leaf-frond is
often developed and on it the individual leaves are distichously arranged.
The plane of distichy varies, however: it may be parallel or transverse to
the direction of the rhizome (WEISSE 1932, 1933). In the old classification
of K. SCHUMANN (1904), followed by LOESENER (1930), Zingiber itself is
associated with Alpinia and allied genera: HOLTTUM (1950) maintained
its greater resemblance to Hedychium, and transferred it to that tribe.
The general change of affinity was undoubtedly correct, though separate
tribal status for Zingiber can be justified (BURTT & OLATUNJI 1972). HOLT-
TUM apparently knew nothing about the leaf character, but in fact in the
Hedychieae and Zingi~ereae the plane of distichy of the leaves is parallel
to the long axis of the rhizome, while in Alpineae it is transverse to it.
In the old system both states of this character were to be found in a single
tribe: only after HOLTTUM'S revision is it possible to see that there is no
switch from one pattern to another in closely related genera: the difference
of pattern is in different tribes. For this we can put forward a hypothesis.
Alpineae are essentially plants of the tropical non-seasonal rain-forest:
they have leaf-fronds visible all the year round and their rhizomes are
very fibrous and behave rather as organs of spread than as food reservoirs.
Hedychieae and Zingibereae, in contrast, are more often plants of monsoon
climates: they have a rest period between the dying down of the old fronds
and growing up of the new, and their rhizomes are more fleshy (often with
associated root-tubers) and less far-spreading. However in Hedychieae
there are plants, such as some species of Kaempleria and Boesenb(3rgia,
which have no well-developed leaf-frond and no very marked horizontal
rhizome. To these plants the criterion of plane of distichy seems inappli-
cable. One hypothesis would be that the frondose tropical Alpineae gave
rise in seasonal climates to non-frondose plants, similar to these Hedychieae,
and that these in turn gave rise to plants with a leaf-frond. There is no
reason why the plane of distichy in the newly evolved frond system should
be the same as that in the old. The hypothesis could be investigated further
by careful studies in Kaempferia and allied genera. If it is disproved that
will be an indication that the two groups became separated very early
in the evolution of the family. Whatever the outcome, revision of the sys-
tem was clearly vital to a clear statement of the problem.

These two examples show the value of attention to patterns of growth

in defining groups above generic level. They also show that classifica-
tion between family and genus must provide a system within which the
evolutionary problems can be clearly recognized and formulated. Taxo-
nomy at this level is of the utmost importanc·p and this is one of its. major
100 B. L. BURTT:

functions. It is here, where speculation does not have to be quite so

awe-inspiring, nor quite so unprovable, as in the construction of a system
for the whole of the angiosperms, that we stand to gain most knowledge
of evolution.
In Gesneriaceae I have suggested (BURTT 1970) that continued
growth of one cotyledon was from the start an advantage, permitting
increased photosynthesis before, or during, the organization of the
plumular bud. The evolution of the unifoliate species of Streptocarpus,
or the genus Monophyllaea, shows that this feature had considerable
potential. The situation in Zingiberaceae is different. It may well be
that neither plane of distichy has an advantage in itself: it is the forma-
tion of a leaf-frond that is advantageous : there happen to be two ways
of doing it and both are found. Here the character is a morphological
end-point: not one with potential for further elaboration.
The two subfamilies of Gesneriaceae have followed very different
evolutionary paths, with some convergence but very little parallelism.
Within Oyrtandroideae (except in Klugieae, to which I shall return) there
is a marked tendency towards elongation of the ovary and fruit, as-
sociated with the restriction of the ovules to the tip of the lamelliform
placenta. A variety of fruit forms has developed: some are straight with
bilateral dehiscence, some straight with unilateral dehiscence, some
twisted in development and dehiscing when the spiral sutures gape on
drying; some are indehiscent and even these may be considerably
elongated. None of these long-fruited forms is paralleled in Gesnerioideae,
where fruit evolution has run from a short capsule to a berry and to an
inferior ovary.
Corolla-form remains fairly plastic throughout the family. As an
adaptation to bird-pollination a red arcuate corolla, associated with
similar patterns of pollen-presentation, is developed in Aeschynanthus
(Oyrtandroideae-Trichosporeae) , in Oolumnea (Gesnerioideae- Episcieae)
and Rechsteineria (Gesnerioideae-Gloxinieae). But the tribes to which
these belong are firmly committed to different forms of ovary. This
decides the method of ovular protection, whose importance was empha-
sized by GRANT (1950). In Aeschynanthus the elongate ovary becomes
even more slender and elongate, offering no catch-point to a probing
bill. In Oolumnea the roundish ovary is protected by a constriction near
the base of the corolla-tube; this acts as a capillary, presenting nectar
to the bird well above the level of the ovary. In Rechsteineria the ovary
is inferior. Thus the pattern of ovular protection depends on the starting
point, the evolutionary springboard, from which it developed.
Within Gesneriaceae-Oyrtandroideae the tribe Klugieae (incl. Loxo-
nieae) provides a good opportunity to examine the criteria of tribal status.
It is a small tribe of 6 well-marked genera, but there is a remarkable
Classification Above the Genus 101

range of morphological character. It stands apart from the rest of the

subfamily in the short ovoid ovary which is sharply contracted into the
style. The ovary may be completely unilocular with Y-shaped parietal
placentae similar to those of Gesnerioideae (rather than T-shaped as in
most Cyrtandroideae), or it may be bilocular with an axile placenta. The
inflorescence may be an open pair-flowered cyme, or condensed to a
helicoid pseudo-raceme. Fertile stamens may be four or two only, in
which case either the anterior pair (in Rhynchoglossum) or the posterior
pair (in Epithema) may persist. Other features are equally varied and
Klugieae has a wider geographical range than any other tribe: from
West Africa to the Solomon Islands, recurring in Central America. It
has all the appearance of a relict tribe comprising the surviving members
of a numerous and highly diversified group. This tribe is under intensive
study by WEBER (1975, 1976).
Meanwhile work on the other tribes, Cyrtandreae, Didymocarpeae, and
Trichosporeae has made it more and more evident that they are very
closely interrelated. Boeica (Didymocarpeae) has remarkable similarities
to Rhynchotechum (Cyrtandreae): there is the same short-tubed subregular
corolla with four stamens, and the same striking range of indumentum
from soft appressed silky hairs to harsh spreading bristles. The linear
capsule of Boeica contrasts with the round berry of Rhynchotechum to
give the tribal difference. The link between Didymocarpeae (seeds without
tails) and TricMsporeae (seeds with tails) is surely represented by
Loxostigma (BURTT 1975). It has one terrestrial species with tail-less
seeds and some epiphytic ones with tailed seeds.
Thus a better system than one with four equivalent tribes might
mark off Klugieae more sharply from the rest. There are two ways of
achieving this: either Klugieae can be raised to subfamily, or the other
three tribes united into one tribe with three subtribes. How are we to
decide? And what are the consequences of either step?
The system for Gesneriaceae with two subfamilies, Gesnerioideae and
Cyrtandroideae, has the advantage of representing what may have been
the fundamental cleavage in the family, caused by the development of
seedling anisocotyly. Klugieae, though sharing some characters with the
Gesnerioideae, are strongly anisocotylous and must therefore be regarded
as an offshoot of the cyrtandroid stock. Thus to rank it as a subfamily
equivalent to the other two might be considered unsatisfactory and it
would break the unity of the anisocotylous group. On the other hand,
there is in other families an increasing tendency amongst taxonomists to
avoid a simple dichotomy at subfamily level. This, of course, becomes
particularly obvious in a system such as THORNE'S (1968), where a broad
family concept results in numerous subfamilies-no less than 21 in
Liliaceae. Even without that extreme example, there is a notable
102 B. L. BURTT:

turning away from simple dichotomy; it has been abandoned, for example
in Commelinaceae (BRENAN 1966), Meliaceae (PENNINGTON & STYLES
1975; see also LEROY 1976), Proteaceae (JOHNSON & BRIGGS 1975).
There is thus a widespread readiness to compress the evolutionary
explosion, that must have been spread over a long period, into that
blank space on the paper between family and subfamily headings.
That may be justified when we have no clue as to how it took place. In
Gesneriaceae the seedling morphology supplies such a clue.
The alternative treatment of Klugieae is to rank it as a tribe co-
ordinate with a tribe Cyrtandreae, which would include the narrower
Cyrtandreae, Didymocarpeae, and Trichosporeae as subtribes. In 1962 I
abandoned subtribes under tribe Didymocarpeae (the only tribe large
enough to justify their use) because the existing ones were artificial; but
there are some 50 genera in the tribe, so that such groupings might well
be desirable, when we know how to make them. We lose that oppor-
tunity if Didymocarpeae is itself reduced to subtribal rank.
To reconstruct the evolution of Cyrtandroideae in the simplest way
that fits the available evidence, we need to imagine four major branch-
ings. First, that giving rise to the two subfamilies, Gesnerioideae and
Cyrtandroideae, from the basic family stock; second, a branch off Cyr-
tandroideae giving Klugieae; third, a branch giving Didymocarpeae;
fourth, a branch off Didymocarpeae giving rise to Trichosporeae. Even if
subtribes are not reserved for lower groupings, there are only three
ranks available for four levels of branching. Thus, as is well known,
even at this very simple level a system of classification cannot represent
the evolutionary situation.
Perhaps the best course at present is to make no change at all: to
retain two subfamilies and four co-ordinate tribes in Cyrtandroideae. If
desirable for a special discussion Klugieae could be designated group A;
the other three tribes group B. I do not object to such informal groupings,
but I am absolutely opposed to the addition of any new ranks which
would carry formal names. Subfamily, tribe and subtribe are quite
enough.
To return to the possible subdivision of Didymocarpeae and my
abandonment of subtribes there. FRITSCH had a tribe Streptocarpeae,
defined solely by the possession of a twisted fruit and this was retained
as a unit at subtribal rank by IVANINA (1967) in her carpological study
of the family. Such a fruit has probably developed at least 4 times in
Didymocarpeae. I recently found a letter, on a sheet belonging to the
Calcutta herbarium, from C. B. Clarke, the monographer of Cyrtandreae,
to Sir George King: it is dated 1882. Clarke wrote: "in Cyrtandraceae
we make rather too much of twisted capsule-valves: I have a section of
Didymocarpus which according to nature should be in Boea, but the
Classification Above the Genus 103

capsule valves are perfectly straight". Clarke was quite right. Now,
with more species to study, and with a far greater knowledge of other
characters, I have been able to remodel some of these genera "according
to nature", and three end up with either straight or twisted fruits: the
character is not a generic one. This, of course, is the death-knell to its
independent use in defining any higher category.
The decision to abandon subtribes in Didymocarpeae .was not taken
because no natural groupings could be made, but because of the number
of genera of uncertain affinity that were left. Most of the genera could
be grouped into about six subtribes, but there would be about 8 isolated
genera which could only form monogeneric subtribes. This I would
regard as unsatisfactory, for a system becomes useless when monotypic
groups outnumber the remainder.
How far then can classification between order and genus usefully go?
We need a measure of the subdivision of a group relative to the number
of genera it contains. Here I introduce the term "terminal suprageneric
taxon". It is self-explanatory; it refers to the lowest named group above
the generic level, of whatever rank it may be. The ratio of genera to
terminal suprageneric taxa, and the number of the latter that are
monogeneric, sum up a system of classification in a very practical way.
For example, for the tribe Campanulaceae-Lobelieae, WIMMER (1957)
gives us a system of 20 genera grouped in 14 terminal suprageneric taxa,
11 of them monogeneric. BRENAN (1966) has re-classified the 45 genera
of Commelinaceae into 15 equivalent groups, 6 of them monogeneric. In
Proteaceae JOHNSON & BRIGGS (1975) have 75 genera in 33 terminal
suprageneric taxa and 14 of these are monogeneric. This work on
Proteaceae is clearly of the highest quality, that is why it is an important
example. In Gesneriaceae BENTHAM had 71 genera in 9 terminal taxa;
FRITSCH had 84 in 23,5 monogeneric. New World genera are at present
heing re-studied by workers in America and I am uncertain about
groupings, but I estimate for the whole family 130 genera in 11 tribes:
for Cyrtandroideae 70 in 4: none monogeneric.
Gesneriaceae is a highly advanced, and presumably recent, family;
Proteaceae is more primitive and doubtless much older. They are very
different kinds of family. Can they be usefully compared in this way?
There are, of course, no universal definitions of subfamily, tribe or
subtribe: nevertheless we can strive to make these ranks of equivalent
value in different groups. The assessment of the resulting classification
in terms of the ratio of genera to terminal suprageneric taxa and the
number of these that were monogeneric would then give lj, measure of
comparison of the evolutionary state of two families. There would be,
however, no guarantee that the application of equivalent concepts in,
say, Magnoliaceae, Orchidaceae, and Compositae would give useful or
104 B. L. BURTT:

workable classifications in all those families. If the purpose of classifica-

tion within a family is to subdivide it into smaller groups that are useful
in the process of identification, useful in defining areas for more detailed
study, and useful as a whole in summarizing interrelations of the genera,
then the standards for the suprageneric groupings must come largely
from within.
The use of equal standards (if that were possible) throughout the
flowering plants, would result in a special classification useful for one
particular comparison: it could serve that purpose without any nomen-
clature being applied to the taxa between family and genus: one simply
needs totals at each level. For the more general needs of taxonomy served
by a classification of named categories, the standards most appropriate
to the internal structure of the family must be used.
A system between family and genus goes into far too much detail if
it reduces the number of names by only a third from the generic level to
the next one above it. It is quite wrong to think that all the recognizable
degrees of difference must be expressed in the classification. Taxonomy
has been called the science of diversity. Taxonomists must think in
terms of diversity, and not try to reduce every group to maximum uni-
formity. It may be necessary to make investigations in that way; but
just as we have to dissect the individual parts yet think in terms of the
whole organism, so we must analyse into components of maximum simi-
larity but then synthesize into groups of related diversity.
There is one further point about the isolated genera that come to
form monogeneric groups. Is there any real reason why these groups
should have formal latin names? Would it not be equally satisfactory to
use the generic name and simply preface it by the rank at which it was
being considered? For example, the recent reclassification of Meliaceae
(PENNINGTON & STYLES 1975) would read at subfamily level: M elioideae,
subfam. Quivisianthe, subfam. Capuronianthus, and Swietenioideae.
This device would avoid the creation of additional latin names, avoid the
farce· of trying to decide which characters of a single genus should
diagnose a subfamily, and give the immediate information that that
subfamily contains only one genus. It would be a device to reduce
nomenclatural innovations, and nomenclatural argument about rank
[for instance LEROY (1976) regards Capuronianthus as a tribe, not a
subfamily]. It would not make monogeneric groups more acceptable.
A relatively small number of genera are important in a system of
classification not merely because they raise the problem of a monogeneric
group, but because they are more or less intermediate between two
larger groups.
Jerdonia is a monotypic genus of the hills of S. W. India. WIGHT
(1848) placed it only tentatively in Gesneriaceae, saying he did not know
Classification Above the Genus 105

what else to do with it-but no one since has echoed his doubts. Yet it is
clearly out of 'place in Gesneriaceae. It does have a unilocular ovary, but
there are 4 parietal placentae, quite unlike the bifid lamellate placentae
of Gesneriaceae. Furthermore the seed is too large and the endosperm is
well-developed and alveolate. It has the characteristic pair-flowered
cyme, but this is found in several genera of Scrophulariaceae: Galceolaria,
Tetranema, and Pensterrwn at least. Jerdonia has an isocotylous seedling.
It is one thing to say that J erdonia is better placed in Scrophulariaceae;
another to decide on its affinity there.
The two subfamilies of Gesneriaceae must have had a common ances-
tral group. It is, as usual, unknown. The character-states that could
have given rise to the different expressions now found in these two
subfamilies are:

Inflorescence a pair-flowered cyme

Corolla zygomorphic
Aestivation ascending imbricate
Fertile stamens four
Ovary superior, unilocular
Seed with endosperm (alveolate?)
Seedling isocotylous
Stomata anisocytic (usually anomocytic in Scrophulariaceae)
lridoid glycosides absent (often present in Scrophulariaceae)

All those features are found in J erdonia. Of course J erdonia has some
special features of its own. Those four separate and very simple
placentae are not like those of either family; it has pollen in tetrads;
the haploid chromosome number is 14, whereas 8 seems probably basic
in Gesneriaceae. Nevertheless the agreement is remarkable. I think we
are justified in regarding J erdonia as the genus of Scrophulariaceae which
is closest to a missing group from which Gesneriaceae may have arisen.
There is no single clear-cut distinction, but the two families are main-
tainable, in fact must be maintained, as distinct entities. Unilocular
ovaries occur in Scrophulariaceae (quite apart from Jerqonia) and a
typical bilocular ovary with axile placenta is found in Gesneriaceae in
Monophyllaea, a genus whose enormously enlarged cotyledon makes it
an unmistakable mem!Jer of Gyrtandroideae. Speculation that Jerdonia
is close to the linking group gives no justification for merging the fam-
ilies.
One of the best examples of a small genus that links two families is
Triplostegia. This has the open cymose inflorescence and smell of
Valerianaceae; but whereas Valerianaceae has an ovary with one fertile
cell and two sterile ones, Triplostegia has a simple unilocular ovary as
found in Dipsacaceae. Furthermore, in Triplostegia therE is an involucel
surrounding each flower as in Dipsacaceae. This is not just a verbal
106 B. L. BURTT:

agreement due to applying the same term to rather different structures.

The involucel of Triplostegia is just like that of Scabiosa (the involucel
of Morina, which is usually included in Dipsacaceae, is rather different
in detail). Valerianaceae was not separated from Dipsacaceae until 1802.
Perhaps if Triplostegia had then been known the family distinction
would never have been made. Now that it has been long accepted, what
is the best treatment of Triplostegia?
First, I absolutely reject the idea of making it a separate family.
Secondly, as a working taxonomist who is constantly called upon for
information, references etc., I am very conscious of what I call the
historical constraints on any major changes in classification. Were I
compiling a completely new system, knowledge of Triplostegia might
well cause me to unite Valerianaceae with Dipsacaceae: but that would
now be the wrong way to accommodate a genus of two species. The
importance of Triplostegia is that technically it belongs to Dipsacaceae,
but there is a strong resemblance to Valerianaceae. In Dipsacaceae it
looks superficially out of place: it is provocative, provides an obvious
link to Valerianaceae, and can be identified by the aid of numerous
existing family keys.· Placed in Valerianaceae it attracts little attention,
for it must be dissected for its special characters to be seen; but all keys
to families would have to be altered to accommodate it. I have no
hesitation in referring Triplostegia to Dipsacaceae. As already mentioned
it has the typical involucel (and it may be added calyx) of Scabiosa. It
does not require more than su btribal rank in tribe Dipsaceae.
Triplostegia is important from another point of view. It has been
suggested that the flower-head of Dipsacaceae is a compound structure:
a head of heads. This idea derives from the interpretation of the involueel
as the involuere of a primary capitulum now reduced to a single flower.
Triplostegia has a true involucel, but the flowers are solitary in an open
cyme. Condensation of this inflorescence would lead to the cymose
capitulum of Dipsacaceae. There is no question of an inflorescence of
capitula being reduced to a secondary head.
The relationship of Scrophulariaceae and Selaginaceae merits brief
attention. The family rank of Selaginaceae was upheld by BENTHAM &
HOOKER (1876), but it was reduced to a tribe of Scrophulariaceae by
WETTSTEIN (1891). Subsequently this position has been maintained by
most continental authors, but HUTCHINSON (1926) and following him
DYER (1975) have again given it family rank.
WETTSTEIN placed his tribe Selagineae next to the tribe Manuleeae,
with which it shares several features including synthecous anthers, and
the frequent adnation of bract to pedicel. JUNELL (1961) carried out a
critical comparative study of ovary structure, and made an interesting
discovery. While Selaginaceae has been diagnosed as having a single
Classification Above the Genus 107

ovule in a loculus, there are a few species of Selago that have two ovules
in each: and these ovules are placed medianly, one turned upwards one
downwards: neither occupy the apical position of the single ovule of
typical Selago. For these plants JUNELL made the new genus Tetraselago.
Among examples of Scrophulariaceae tribe Manuleeae, JUNELL
examined the monotypic genus Glumicalyx. This was important in his
final argument, for he found it had fewer ovules than the other genera
of Manuleeae and that the upper ones were turned upwards, the lower
downwards. This seemed to him a step in the direction of Tetraselago.
We can now go further (HILLIARD & BURTT 1977). Glumicalyx
is a genus of six species: three of these are to be added by transfer from
the neighbouring genus Zaluzianskya, one is new, but the last is to be
transferred from Walafrida, a genus of Selaginaceae, to which it bears
no little resemblance but differs sharply in having about 12 ovules in
each loculus. In Glumicalyx there is thus a range from about 80 down to
12 ovules in each loculus. The upper end of the scale retains contact
with the other genera of Manuleeae, while the lower looks towards the
condition in Tetraselago.
While Glumicalyx shows an approach to Tetraselago, another pos-
sible link has been discovered. This is a plant which has to be described
as a monotypic new genus: its interesting feature is that each ovary
loculus contains up to 6 ovules which are all located on the upper part
of the axile placenta, and only the uppermost 1-3 develop into seeds:
thus further sterilization of ovules in this ovary might lead to the single
apical ovule of Selago.
The evidence shows a close relationship between M anuleeae and
Selaginaceae. Without attempting a detailed study it can be said with
some confidence that the difference between Manuleeae and Selagi~eae
is no greater than that between Manuleeae and its neighbouring tribes
in Scrophulariaceae, and is much less than that between, say, Manuleeae
and Rhinantheae. Therefore Selaginaceae cannot be retained as a separate
family. The tribal rank accorded it by WETTSTEIN is the highest that
can be justified. The situation here contrasts with the comparisons made
between Scrophulariaceae and Gesneriaceae or between Dipsacaceae and
Valerianaceae. Ther.e a single small genus provided a possible link
between two well-separated groups. Here the importance of Glumicalyx
and Tetraselago is in showing that a small gap between the two groups in
one single character is indeed very small. It is the general close affinity
between Selaginaceae and M anuleeae that justifies ranking them as
collateral tribes.
When preparing this paper I jotted down a list of ways in which it
might be brought to a conclusion. There was an explanation of the ana-
lytic rather than synthetic approach (although analysis and synthesis
108 B. L. BURTT:

are but the left and right hand of the taxonomist, whose mind works
with both); there was a comparison of the position at family level and
below with that among orders and superorders, and of the role of infor-
mal systems at both levels. But when these, and other ideas, were
committed to paper, each proved to be not an ending but the begin-
ning of something new.
The best conclusion is to reiterate the fundamental importance of
sound classification (based on whole plants not on selected characters)
within each and every family. It is by working to get the best possible
system between family and genus, or in groups of families known to be
closely related, that we shall learn about the patterns of evolutionary
change. Without a better knowledge at this level we can scarcely hope
for a satisfactory superstructure.

I am grateful to Dr. P. KOOIMAN, Delft, for testing Jerdonia for iridoid

glycosides.

References
BENTHAM, G., and HOOKER, J. D., 1876: Genera Plantarum, 2. London:
Reeve, and Williams & N orgate.
BRENAN, J. P. M., 1966: The classification of Commelinaceae. Journ. Linn.
Soc. Bot. 59, 349-370.
BURTT, B. L., 1963: Studies in the Gesneriaceae of the Old World: XXIV:
Tentative keys to the tribes and genera. Notes Roy. Bot. Gard. Edinb.
24, 205-220.
- 1970: op. cit.: XXXI: Some aspects of functi<?fial evolution. Notes
Roy. Bot. Gard. Edinb. 30, 1-10.
- 1975: op. cit.: XL: The genus Loxostigma. Notes Roy. Bot. Gard.
Edinb.34, 101-105.
- and OLATUNJI, O. A., 1972: The limits of the tribe Zingibereae. Notes
Roy. Bot. Gard. Edinb. 31, 167-169.
DON, G., 1838: A general history of dichlamydeous plants, 4, 643-665.
London: Rivington et al.
DYER, R. A., 1975: Genera of South African flowering plants 1, 558. Pre·
toria: Dept. of Tech. Agric. Services.
FRITSCH, K., 1894-1895: Gesneriaceae. In: Die natiirlichen Pflanzenfamilien
(ENGLER und PRANTL), IV, 3 b, 133-185. Leipzig: Engelmann.
GRANT, V., 1950: The protection of the ovules in flowering plants. Evolution
4,179-201.
HILLIARD, O. M., and BURTT, B. L., 1977: Notes on some plants of
southern Africa, chiefly from Natal: VI. Notes Roy. Bot. Gard. Edinb.
35, 155-177.
HOLLTUM, R. E., 1950: The Zingiberaceae of the Malay Peninsula. Gard.
Bull. Singapore 13, 1-249.
HUTCHINSON, J., 1926: The families of flowering plants 1, 308. London:
Macmillan.
IVANINA, L. I., 1967: The family Gesneriaceae (the carpological review).
Leningrad: The Academy of Sciences of the USSR (In Russian).
Classification Above the Genus 109

JOHNSON, L. A. S., and BRIGGS, B., 1975: On the Proteaceae-the evolution

and classification of a southern family. Bot. Journ. Linn. Soc. 70,
83-182.
JUNELL, S., 1961: Ovarian morphology and taxonomical position. of Se-
lagineae_ Svensk Bot. Tidskr. 55, 168-192.
LEROY, J. F., 1976: Essais de taxonomie syncretique: I: Etude sur les
Meliaceae de Madagascar. Adansonia ser. 2, 16, 167-203_
LOESENER, T., 1930: Zingiberaceae. In: Die naturlichen Pflanzenfamilien
(ENGLER und PRANTL), 2. AufI., 15 A, 541-640. Leipzig: Engel-
mann.
PENNINGTON, T. D., and STYLES, B. T., 1975: A generic monograph of the
Meliaceae. Elumea 22,419-540.
SCHUMANN, K., 1904: Zingiberaceae. In: Das Pflanzenreich (ENGLER),
Heft 20. Leipzig: Engelmann.
THORNE, R. F., 1968: Synopsis of a putatively phylogenetic classification
of the flowering plants. Aliso 6, 57-66.
WEBER, A., 1975: Beitrage zur Morphologie und Systematik der Klugieae
und Loxonieae (Gesneriaceae): I: Die Spross- und Inflorescenzorganisa-
tion von Monophyllaea R. BR. Bot. Jahrb. 95, 174-207.
1976: Op. cit. II: Morphologie, Anatomie und Ontogenese der Elute
von Monophyllaea.R. BR. Bot. Jahrb. 95, 435-454.
WEISSE, A., 1932: Zur Kenntnis der Blattstellungsverhaltnisse bei den
Zingiberaceen. Ber. dtsch. bot. Ges_ 50 A, 327-366.
- 1933: Die Art der Distichie an den Achsensprossen von Zingiber. Ber.
dtsch. bot. Ges. 51, 13-20.
WETTSTEIN, R. VON, 1891-1893: Scrophulariaceae. In: Die naturlichen
Pflanzenfamilien (ENGLER und PRANTL), IV, 3 b, 39-107. Leipzig: Engel-
mann.
WIGHT, R., 1848: Icones plantarum Indiae orientalis 4 (2), 10, t. 1352.
Madras: The author.
WIMMER, F _, 1957: Oampanulaceae-Lobelioideae. In: Das Pflanzenreich
(ENGLER), Heft 107. Berlin: Akademie Verlag.

Address of the author: B. L. BURTT, Royal Botanic Garden, Inyerleith

Row. Edinburgh EH3 5LR, Scotland, U.K.
Plant Syst. Evol., Suppl. 1, 111-121 (1977)
© by Springer. Verlag 1977

Institut fiir Holzbiologie und Holzschutz

rler Bundesforschungsanstalt fiir Forst- und Holzw:irtschaft,
Hamburg, Federal Republic of Germany

The Anatomy of Secondary Xylem and the Classification

of Ancient Dicotyledons
By

H. Gottwald, Ham burg

Abstract: The stem wood of about 700 species belonging to 32 families

of the order Magnoliales s.1., plus further taxa exhibiting primitive wood
anatomical features, was investigated. On this basis, six structural
groups can be established each of which show a marked gradation from
primitive to advanced stages. Wood structure of Magnoliales sensu
TAKHTAJAN is only partially primitive, partially moderately derived,
while the most primitively structured heteroxylous taxa belong to the
"Dillenial-Hamamelidal" and "Theal" group, respectively. Accordingly,
there is no compelling evidence to support phylogenetic schemes
in which the Magnoliales is placed as the only common base for all recent
dicotyledons. For this reason, and since there is no structural mechanism
which justifies the derivation of the taxa, belonging to the "Dillenial-
Hamamelidal" and "Theal" groups which are very primitive in their wood ana-
to~y, from the types embracing M agnoliales sensu TAKHTAJ AN, a new phylo.
genetic model assuming an early separation of angiosperms into at least two
branches is discussed. Furthermore, possible affinities between primitive
and advanced taxa within the different structural groups are shown.

Introduction
The ancient dicotyledons ha ve a special attraction for wood anatomists
as, apart from the extraordinary position of this group, most of these
taxa are woody plants with a well developed secondary xylem and,
furthermore, all the vesselless taxa belong to them. Therefore, these
taxa, often regarded as ancestral, offer a suitable structural compara-
bility for tracing the limits and the position of the order M agnoliales
by means of wood anatomical features.
The secondary xylem of this ancient group was already investigated
by McLAUGHLIN (1933) and by LEMESLE (1953) but with very limited
material and with other purposes in mind. The present investigation is
based only on material of the secondary xylem of adult stem-wood
112 H. GOTTWALD:

samples belonging to the wood collection of the Bundesforschungsan-

stalt fur Forst- und Holzwirtschaft, which contains almost 45 000 micro-
scope slides covering about 16000 samples.

Materials and MethOlls

First, as ancient dicotyledons must be understood all the families which
comprise the order Magnoliales sensu TAKHTAJAN (1973). Second, all
additional taxa which are considered by MELCHIOR (1964) as belonging to
Magnoliales will be discussed. Third, this group of magnolian allies ~ill
be compared with 10 more families which likewise may be called ancient
due to their primitively structured secondary xylem. All in all the wood
of more than 700 species belonging to 165 genera of 32 families was in-
vestigated.
To demonstrate the anatomical details, a two-dimensional model is
Hsed in which the six vertical columns represent the various structural
principles of the families investigated. These six structural principles, or
"structural groups" are classified primarily according to the distribution
of the axial parenchyma as for instance as apotracheal diffuse or para-
tracheal. Further features used are the various distributions of the vessels
and specific features of high systematic value such as tannin-tubes in the
rays. The headings chosen for these six structural groups e.g. "Theal"
or "Laural" are based on the most typical family of each group. These
structural principles are only supposed to indicate the plan of architecture
or "topographic pattern" of the various taxa and insofar have no direct
link with the evolution of anatomical features (Figs. la, b). The second
dimension (horizontal) is an attempt to show the structural evolution of
the various families which in turn ,is classified into four grades subdividing
all structural groups, beginning with one understorey for the vesselless
taxa (A). The following three evolutionary grades (B, C, D) are based on
the evolution of the vessels from a purely multiple-perforated and scalari-
form-pitted type ascending to a s~mple-perforated and alternate-pitted
vessel; further, from a pronounced heterogeneous ray developing to a
homogeneous one and from a typical fibre tracheid with large bordered
pits to a small and scantily pitted cell of a more fibre-like type (Fig. la).
In making the classification all microscopic features were taken into ac-
count with exception of the longitudinal cell dimensions. Such measure-
ments were neglected since reasonable results can be obtained only with
large numbers of sampleR from many treeR of the Rame species and these
were not availahlp.

Wood Anatomical Descriptions

Magnoliales sensu TAKHTAJAN. Of the family Magnoliaceae
105 species representing all twelve genera, according to the last revision
of DANDY (cf. PRAGLOWSKI & DANDY 1974), were investigated including
the lesser known genera Alcimandra, Pachylarnax, and Tsoongiodendron,
discovered only in 1963_
The investigation clearly demonstrates that all twelve genera are
composed, without exception, according t.o the same principle of flt-ruc-
Anatomy of Secondary Xylem 113

ture which is characterized by radially grouped vessels and terminal

bands of axial parenchyma. This plan of architecture is sufficiently
stable and specific to make it a very simple task to identify any taxa
of M agnoliaceae as belonging to this family by means of wood anatomical
features. This structural type establishes one of the six structural
principles denominated "Magnolial". In contrast with this homogeneity
of structure, the Magnoliaceae embrace a wide range of structural evolu-
tion marked by a pronounced progression within the family. This evolu-
tionary drive is mainly visible in the vessels. Here the scalariform
perforations change to simple ones and the scalariform pits develop to
an opposite arrangement. Similar dynamics to the one linked to the
alterations in the vessels can be observed in the rays. Accordingly, the
wood structure of the genus Talauma is the most primitive and that
of the genus Liriodendron together with the subgenus Yulania of Ma-
gnolia are the most advanced in the family Magnoliaceae due to the non-
scalariform pitting of the vessels and the homogeneous rays.
This implies that the wood structure of Magnoliaceae is fairly primi-
tive to fairly advanced with a pronounced mosaic-like progression
reaching over three evolutionary steps (B, C, D).
The family of Winteraceae, the sole homoxylous taxon in M agnoliales
sensu TAKHTAJAN will be considered, for anatomical reasons, together
with the vesselless taxa Amborellaceae, Tetracentraceae, Trochodendraceae,
and Sarcandra of Chloranthaceae. On account of their vesselless state
the five taxa must be placed in the lowest of the four steps of evolution.
As obvious as this decision is, as uncertain must remain their assignment
to any of the six structural groups. Only the genus Sarcandra can be
correlated fairly well with the "Laural" principle. For the other four
homoxylous taxa agreement with the "Magnolial" principle is more
probable according to the distribution of the banded axial parenchyma,
than with the "Dillenial-Hamamelidal" or "Theal" principles. Further-
more it must be mentioned that Tetracentron has vestured vessel pits,
a common feature of only the Lauraceae and M yristicaceae within
Laurales sensu TAKHTAJAN.
Any possible relationship between these four homoxylous families can
only be hypothetical. There are weak indications for an affinity between
Trochodendraceae and Tetracentraceae which share the occurrence of
scalariform pitting in tracheids, thus representing the most primitive
state of all dicotyledons. On the other hand the families of Winteraceae
and Amborellaceae form a loose structural group with a few more ad-
vanced features. Both these statements, however, must remain very
vague as the structural heterogeneity within Winteraceae does not allow
its allocation to a fixed position.
From the pantropical family of Annonao~(1P the secondary xylem
Plant ::5)"8t. Evol., ::5uppl. 1 8
114 H. GOTTWALD:

of 40 genera was examined. This family forms a very homogeneous

structural group in spite of the large number of genera and species. The
family is to be placed in a specific "Annonal" structural principle near
to the "Magnolial" principle. The progression within the family is only
slight in terms of the variation of wood anatomical features, e.g. the
pit-diameter decreases from 10 fl-m in the genus Uvaria to the extremely
low diameter of less than 2 fl-m in Polyceratocarpu.~.
The evolutionary position in respect of the anatomical features of
all genera of Annonaceae is more advanced than that of the Magnoliaceae:
the vessels always show simple perforations, the vessel-pitting is alter-
nate, the rays are mostly homogeneous and the fibres have only minute
pits. In other words: in the Annonaceae the progression of the various
cell types is well balanced in contrast to those of the Magnoliaceae
which warrants its placement in the evolutionary grade D.
Of the small and mostly monotypic families the Himantandraceae
also possesses decidedly advanced features and is to be placed on a level
between the Annonaceae and the Magnoliaceae. As regards the struc-
tural principle the Himantandraceae tends more toward the "Magnolial"
than to the "Annonal" group, as had been suggested by McLAUGHLIN
(1933).
The families Canellaceae, Degeneriaceae, and Eupornafiaceae have
lesser pronounced structural principles. The ])egeneriaceae still show a
relationship to the "Magnolial" principle which is also indicated by
BARAN OVA (1972) with regard to the leaf epidermis, and has an evolll-
t:ionary state in the middle of Magnoliaceae in grade C.
The'family of Eupomatiaceae could be arranged within the "Laural"
as well as the "Dillenial-Hamamelidal" principle in a fairly advanced
position between grade C and D.
In the family Myristicaceae representatives of all 15 genera were
investigated. The secondary xylem exhibits a pronounced structural
uniformity and' requires a separate structural principlc between the
"Magnolial" and "Laural" groups which is in full accordance with
GARRAT (1933). The progression in Myristicaceae is indicated by multiple
vessel perforations in K nema developing to constantly simple ones in
Brochoneura; the evolutionary state of Myristicaceae can be compared
with the more advanced part of Magnoliaceae. A remarkable feature is
the constant occurrence of silica bodies in some genera which in tIlt'
Magnoliales s.l. occur only in the genus Talauma of Magnoliaceae and in
some genera of Lauraceae.
One can draw the conclusion that the seven heteroxylous families of
Magnoliales sensu TAKHTAJAN are spread out over four structural
principles of a more or less pronounced structural relationship. In con-
trast, the four homoxylous families cannot be allocated reliably to an~'
Anatomy of Secondary Xylem 115

structural principle unlike the homoxylous genus Sarcandra. Furthermore,

the genus Talauma shows the lowest structural evolutionary level of
any vessel-bearing taxon of Magnoliaceae sensu TAKHTAJAN without,
however, reaching the lowest possible position for heteroxylous taxa.
Moreover, Magnoliaceae is the family with the strongest intra-familial
progression and the most pronounced structural mosaic-evolution.
Finally it may be said that all features of the secondary xylem cor-
relate completely with the taxonomic concepts of DANDY (1974) and
TAKHTAJAN (1973).

Magnoliales sensu MELCHIOR. Of the additional families of

Magnoliales as circumscribed in "ENGLER'S Syllabus", the Lauraceae
will be considered first: its special anatomical features make it neces-
sary to form a separate "Laural" structural group. In comparison to
the previously mentioned taxa, Lauraceae possess the greatest structural
diversification of all investigated groups due to additional features as,
for instance, well-developed paratracheal parenchyma, oil-cells, septa-
tions in the fibres and storied rays. A remarkable characteristic of the
Lauraceae is the occurrence of tyloses in the fibres-besides those in
the vessels-which were found for the first time during this investiga-
tion and can be interpreted as indicating a slight relation to the M agno-
liaceae, the only family of all flowering plants where this unique
feature has previously been observed (GOTTWALD 1972). The progression
within Lauraceae is perceptible but mostly restricted to the vessels as
is also true of the Myristicaceae whose structural advancement is com-
parable in grade to tha,t of Lauraceae (C-D).
The families Hernandiaceae and Galycanthaceae are characterized by
the same "Laural" structural principle and can be placed next to Laura-
ceae according to their evolutionary state.
The monotypic family Gomortegaceae is also similar to the Lauraceae
but due to the pronounced apotracheal parenchyma it seems more
,'easonable to place Gomortega in the "Dillenial-Hamamelidal" group.
The advancement of Gomortega is comparable to the lesser-developed
part of M agnoliaceae.
With regard to the family M onimiaceae it was not possible to inves-
tigate more than about one third of the genera. Here, the Monimia-
ceae will be interpreted in its widest sense and the results of previous
investigations by GARRAT (1933) and MONEY (1950) are included. In
this sense the family extends from the "Theal" principle through the
"Dillenial-Hamamelidal" to the "Laur~l" principle. Within the Moni-
miaceae a pronounced structural progression, comparable to that of the
M agnoliaceae, can be noticed, ranging from primitive structures to
fairly advanced ones in the genus Peumus.
t\.
116 H. GOTTWALD:

) " THEAL " " OILLENIAl/HAMAMELIOAL"

Structural principle I

Structural i
evolution I
-+--------------~----------------~

•0
I
. I
-+ -1 SChjSandra~
I I
I
M0 N I •
• C M I A C.
• I
I
-r
0, rmmr-ll
lflw.illJ 1
o I
I I HAMAME-
ITRIMEN lAC J
O I lIDAC.
B
I I
- 0 I I EUCRYPHYAC . AEXTOXI-~
CAC . ILlI
01 THEAC . I ClAC.
10 ILL E N I A C.

----+----~­
Ve ssel less
I (

fam ili es A
(
(

Fig. 1 a
Fig. la-lb. Two-dimensional model of ancient taxa including all families
of the Magnoliale.~ (sensu TAKHTAJAN in thick boxes, sensu MELCHIOR in
thin boxes, others without boxes)
Anatomy of Secondary Xylem 117

~. ~.~ l.,,~~. . "~,,.~. . . .~. . ·. ·I~. . ~: ':'

" LA URAL " " MYRISTICAL " " MAGNOLIAL "

·1· :I .....
0: I .........."
•• •••• '" •• ,.. t.

.. ·on I •• I

'8
[RERNANj

I • t.
lA U
RAC .
T~T"':
~_
I
" TRI S
TICAC.
IHIMANTANORAC)I~
+---
:~
I
~~ L..-_""" MAGNO- I
L I A c.
'I CANfl-1
PIP(RAC. I lAC . I
I
LO£GENERIAC I

I
'--_. . .r-t -- --
, I
ACTINIOIAC . I
SAURU-
I I EUPTElEAC .1
I
II I
RAC .

CHL ORAN-
THAC. I
I
I I
_____ 1. ___ _ ____ .1. ___ _
Sorcondra
----------~I~=B-oR=E-Ll~Ac~.I~

=::-::-~(==:;--- I WIN T ERA C. )

--ITETRACENTRAC .1---.!:=i5r.;...;..,;;..;.;...;.;...;;.;......!
ITROCHOOENORACJ ;>

Fig. 1 b
118 H. GOTTWALD:

Of the remammg families of M agnoliales sensu MELCHIOR only

Trimeniaceae and Illiciaceae can be correlated reliably with a structural
principle. They are to be placed in the "Dillenial-Hamamelidal" group
in a low or very low position.

The horizontal position of the family Eupteleaceae remains uncertain

because structural arguments exist for its attribution to the "Dillenial-
Hamamelidal" group as well as to the "Magnolial" group, where it can be
placed in a relatively isolated position as ENDRESS (1 fl68) has already
mentioned. According to its low degree of advancement it must be placed
in a primitive position.
The same problem of an undecided structural arrangement exists for
the families of Austrobaileyaceae and Gercidiphyllaceae which, according
to their apotracheal parenchyma, have been placed in the "Theal" group
close to the "Dillenial-Hamamelidal" group. However, an unclearly de-
veloped terminal parenchyma also makes it possible to place Oercidiphyllurn
in the "Magnolial" group as already indicated by McLAUGHLIN (1933).
The placement of Schisandraceae within this model is difficult. The
genera Kadsura and Schisandra possess different specialized structures
due to their scandent habit which make a separate position for each of
them necessary. Schisandra, in which no terminal parenchyma can be
traced, must be put in the "Theal" group, and Kadsura, due to very scanty
apotracheal parenchyma in the "Dillenial-Hamamelidal" group in a fairly
low position. Further, a pronounced progression is found ill thc genus
Schi.~andra (Figs. 1 a, b).

A survey of these families which MELCHIOR, in con(rast to TAKHTA-

JAN, has incorporated in Magnoliales shows a distinct shift to the
"Theal" and the "Dillenial-Hamamelidal" structural principles to which
no family of the Magnoliales sensu TAKHTAJAN could be allocated with
certainty. It is also remarkable that the structural evolution within
these heteroxylous families included by MELCHIOR in M agnoliales is often
of a lower level than that of-Magnoliales sensu TAKHTAJAN.

Families Belonging to Orders Other Than the Magnoliales. In a

third series, a selection of ten further families belonging to nine orders
sensu TAKHTAJAN was investigated. These taxa can also be termed as
ancient because of the primitive structural features of their secondary
xylem.
Of these additional 10 families none belongs to the "Magnolia}",
"Myristical" or "Annonal" structural principle. The families of Chlo-
ranthaceae, Actinidiaceae, Saururaceae, Lardizabalaceae, and Piperaceae
are to be placed in the "Laural" group with an increasing degree of
advancement.
The greatest concentration occurred in the "Dillenial-Hamamelidal"
group where the families Dilleniaceae, Aextoxicaceae, and Eucryphiaceae
Anatomy of Secondary Xylem 119

demonstrate the lowest level of structural evolution of vessel-bearing

species, while the Hamamelidaceae must be aITanged in the same struc-
tural group in a slightly higher evolutionary position.
The last family to be considered, the Theaceae, has to be placed, due
to its specific plan of architecture, into a separate structural group.
The very primitive wood anatomical features described by KENG (1962)
warrant a low position in the "Theal" structural principle near the
Villeniaceae (Figs. 1 a, b).

Results

1. All eight families of Magnoliales sensu TAKHTAJAN are spread over.

four Htructural groups with a pronounced concentration in the "Ma-
gnolial" group.
2. Of these four structural groups the "Magnolial" principle and the
"Annonal" principle on one hand, as well as the "Myristical" and the
"Laural" principle on the other, can be seen as two pairs of closely
related types. Both can bc combined again in a wider structural block
in which, however, the Canellaceae and Eupteleaceae hold an isolated
position.
3. The vessellm;s families callnot be clearly linked to one of the six
structural principles although some features allow an association with
the above mentioned "Magnolial-Annonal-Myristical-Laural" block.
4. Five of the six structural groups have a worldwide distribution;
only the "Magnolial" one, consisting of four families, does not occur in
Africa including Ma<!.agascar.
5. About two thirds of the investigated heteroxyl families-outside
the Magnoliales sensu TAKHTAJAN-are concentrated in the "Dillenial-
Hamamelidal" and the "Theal" structural groups, while only one third
belongs to tho "Laural" group. That means that all families which
MELCHIOR considers as Magnoliales, besides those of TAKHTAJAN, are
represented in the "Laural-Myristical-Magnolial-Annonal" block.
6. The "Dillenial-Hamamelidal" and the "Theal" structural groups
form a second block of differing basic anatomical structure.
7. The evolutionary state of the seven heteroxyl families of M a-
l/1wliales sensu TAKHTAJAN ranges, in comparison to other ancient families,
f!"Om an only fairly primitive to a relatively highly advanced level.
8. The most primitive structures of vessel-bearing taxa can be found
in the "Dillenial-Hamamelidal" and the "Theal" structural groups,
outside the Magnoliales sensu TAKHTAJAN. Hence, according to the
features of the secondary xylem, the order Magnoliales sensu TAKHTA-
.J AN is not at all in accord with his statement of being "the most
primitive order of flowering plants".
120 H. GOTTWALD:

Conclusions
It seems evident that, within the span of evolution that can be de-
duced from living plants, no family of the "Theal-Dillenial-Hamamelidal"
block has proceeded from the "Magnolial-Annonal-Myristical-Laural"
block or vice versa. The M onimiaceae can be seen as a possible bridge
or a "branching-off family" originally connecting the two main struc-
tural blocks at a very early undetermined state of phylogeny.
From the structural point of view these conclusions can lead to the
following three interpretations:

Concept I
A complex of various families with the most primitive structures,
belonging to the "Dillenial-Hamamelidal" and the "Theal" block, can
be regarded as the base for all living dicotyledons. Apart from this base
a separate "dead end" has to be considered in which all vesselless fam-
ilies are combined in the lowest possible position.
However, this model of structural evolution would assume that the
various families of the other block with the "Magnolial-Annonal-
Myristical-Laural" structures (including all heteroxylous taxa of Ma-
gnoliales sensu TAKHTAJAN) have been derived from the "Dillenial-
Hamamelidal" and the "Theal" block. This, however, does not seem
possible as such a fundamental structural metamorphosis is not con-
ceivable without a pronounced development of intermediate evolutionary
characters for which we have no evidence.

Concept II
Another model would place the taxa with the lowest state of evolu-
tion in an appropriately low position with a base consisting of the
four exclusively vesselless families from which the other ancient taxa
are derived. This also does not seem practicable, since it has not been
possible up to now to find evidence for the necessary structural links
between the vesselless and the other living heteroxylous families.

Concept III
According to the evidence derived from the anatomical structures of
the secondary xylem, a model for the recent dicotyledons is proposed
consisting of two main branches representing the two structural blocks
and with a third separate dead-end branch comprising the vesselless
families. In other words, the anatomical evidence does not support any
concept in which all being angiosperms are derived from a single living
order or its forerunners (e.g. the Magnoliales of TAKHTAJAN). Further-
more, no structural alteration is imaginable which would generate a
very primitive structural group from a less primitive one,-ll both
Anatomy of Secondary Xylem 121

groups are also distinguished by fundamentally different structural

principles. Due to this last mentioned fact a derivation of the Dillenia-
ceae, Theaceae and others from Magnoliales cannot be explained by a
"heterobathmic mechanism".
In terms of further evolutionary development the following lines
based on anatomical evidence may be suggested: from the "Annonal"
group towards Ebenaceae; from the "Magnolial" group towards Rutaceae;
from the "Dillenial-Hamamelidal" group towards Rosaceae (p.p.) and
from the "Laural" group towards Leguminosae (p.p.) and Boraginaceae.
Finally it should be strongly emphasized that all the statements
made here are based only on the features of the secondary xylem.
However, wood anatomy cannot on its own be used to devise and estab-
lish phylogenetic models and classificatory systems. Therefore, the wood
anatomist must make use of available systems where he either may find
confirmation of his own results or else may put forward additional
evidence for further discussion and improvements.

References
BARANOVA, M., 1972: Systematic anatomy of the leaf epidermis in the
Magnoliaceae and some related families. Taxon 21, 447-469.
ENDRESS, P., 1968: Gesichtspunkte zur systematischen Stellung der Eupte-
leaceen. Ber. Schweiz. Bot. Ges. 79, 229-278.
ENGLER,' A., 1964: Syllabus der Pflanzenfamilien (MELCHIOR, R., Rrsg.),
12. Auf!. Berlin: Borntraeger.
GARRATT, G. A., 1933: Bearing of wood anatomy on the relationships of
the Myristicaceae. Tropical Woods 36, 20-44.
- 1933: Systematic ~natomy of the woods of the M onimiacaee. Tropical
Woods 39, 18-44.
GOTTWALD, R., 1972: Tyloses in fibre tracheids. Wood Sci. Technol. 6,
121-127.
KENG, R., 1962: Comparative morphological studies in Theaceae. Univ.
Calif. Publ. Bot. 33, 269-384.
LEMESLE, R., 1953: Les caracteres histologiques du bois secondaire des
Magnoliales. Phytomorphology 3,430-445.
McLAUGHLIN, R. P., 1933: Systematic anatomy of the woods of the Magno-
liales. Tropical Woods 34, 3-39.
MONEY, L., BAILEY, I. W., and SWAMY, B. C. L., 1950: The morphology
and relationships of the Monimiaceae. J. Arnold Arb. 31, 372-404.
PRAGLOWSKI, J., and" DANDY, E., 1974: Magnoliaceae. World Pollen and
Spore Flora 3, 1-45.
TAKHTAJAN, A., 1973: Evolution und Ausbreitung der Blutenpflanzen.
Stuttgart: Gustav Fischer.

Address of the author: RELMu'r P. J. GOTTWALD, Wiss. Direktor, Insti-

tilt fUr Rolzbiologie und Rolzschutz der Bundesforschullgsanstalt fUr Forst-
und Rolzwirtschaft, Leuschnerstr. 91, D-2000 Hamburg 80, Federal Repub-
lic of Germany.
Plant Syst. Evo!., Supp!. 1, 123-140 (1977)
© by Springer.Verlag 1977

Department of Botany, University of Canterbury, New Zealand

Ovular Morphology and the Classification of Dicotyledons

By

W. R. Philipson, Christchurch

Abstract: Four types of ovule are defined by integument number and

nucellar thickness. Their distribution throughout the dicotyledons is
summarized and their importance in defining major subclasses is advocated.
In particular, those families with ovules having a single integument are
considered to form a complex of orders and families which includes most
sympetalous and several polypetalous groups (including Oornales, Ara·
liales, Escalloniaceae and Pittosporaceae). This complex (provisionally
referred to as the Unitegminae) is believed to comprise the modern repre·
sentati\"() of a distinct evolutionary line within the dicotyledons.

Introduction
Features of ovular morphology have long been used to define taxa
at all level;;. The Danish botanist WARMING (1878) first clearly ;;tated
the distinction between the crassinucellate- bitegmentary ovule and
the tenuinucellate-unitegmentary ovule. He recognized the taxonomic
importance of these ovular types and named them type dichlamyde
and type monochlamyde, these terms being derived from the sub-
divisiont; of the dicotyledons of which they were characteristic.
After Ilome decades WARMING (1914) drew together more complete
information about these and other types of ovule and wall able to
relate these to recognized taxa. OSSIAN DAHLGREN (1927) published
his careful studies of the nucellus and reassessed the significance of
ovular characters in classification. Neither WARMING nor DAHLGREN
t;et out their views as formal systems. This had been done by VAN TIEG-
HEM (1901) who gave great prominence to ovular (or related) characterll
in a system which embraced the whole plant kingdom. His complete
reliance on ovular types when drawing up his major subdivisions
(Orders) resulted in a system that was too rigid and which contained
inconsistent and unnatural groupings. It is likely that VAN TIEGHEM'1l
uncritical proposals lead to the taxonomic value of the ovule coming
into disrepute. Nevertheless, as I hope to ;;how, several of his principal
124 w. R. PHILIPSON:

proposals are helpful. A major obstacle to their general acceptance

has been the occurrence of the "sympetalous type" ovule (as the
unitegmic-tenuinucellate ovule has frequently been called) within
many polypetalous families. It was a review by MAURITZON (1939)
which convinced most comparative morphologists that the sympetalous
ovule had been evolved in parallel repeatedly. Consequently it has
come to have a less dominant role in systematics.
Recently I have suggested (PHILIPSON 1974) that the difficulties
seen by MAURITZON would be largely overcome if a distinction were
made between polypetalous families in which the so-called sympetalous
ovule is constant as opposed to those in which it (or one or another
of its features) occurs only sporadically. If this is done, correlations
between ovular types and major family complexes become evident.
At the same time, and again in the following year (PHILIPSON 1975), I
proposed the view that dicotyledonous families in which the sympetalous
ovule is .dominant form a long-established and independent line of
evolution. It should be emphasized that these two proposals were
distinct. The first was in the field of classification-the recognition
and definition of taxa-even though this was not formalized by applying
names to the groups proposed. The second was an interpretation of
that classification. (In claiming that the ovule is of value in taxonomy
at the higher level, I am in effect saying that it largely conforms with
several other independent characters, that it therefore confirms existing
boundaries, that it clarifies some doubtful areas, and I!l0reover that
it enlarges previous concepts.) In what follows the types of ovule being
considered will be described, their distribution among higher categories
summarized, and the bearing of this on the framing of major subdi-
visions of the dicotyledons will be discussed.

Definition of Ovule 'l'ypes

Consideration of ovule morphology will be limited to two features,
namely, the number of integuments and the presence or absence of
additional layers within the epidermis at the apex of the nucellus.
The combinations of these two characters allow only four types to be
defined and in fact all of these occur, though with widely differing
frequencies. It would clearly be an advantage to consider a wholc
syndrome of ovular characters but sufficiently consistent information
over a wide sample is not available. No doubt additional characters
would reveal other types of ovule, such as that characteristic of the
Centrospermae or the orthotropous ovule' of several apetalous families.
The present application of ovular morphology to taxonomy, there-
fore, must be regarded only as a beginning.
Ovular Morphology and the Classification of Dicotyledons 125

Figs. 1- 4. Examples of the four types of ovules. Fig. 1. Hernandia nymphaei-

jolifl, (PRESL) KUBITZKI, type I; 2 integuments, crassinucellate. Ovule
X 40; nucellus X 250. Fig. 2. Halesia carolina L., type II; 2 integuments,
tenuinucellate. Ovule X 75; nucellus X 250. Fig. 3. Griselinia littoralis
RAOUL, type Ill; 1 integument, crassinucellate. Ovule X 40; nucellus X 250.
Fig. 4. Forstera tenella HOOK. f., type IV; 1 integument, tenuinucellate.
Ovule X 50; nucellus X 250
126 W. R. PHILIPSON:

The four ovule types defined by combinations of the two characters

being considered will subsequently be referred to by numbers as fol-
lows:
hitegmentary-crassinucellate Type I
hitegmentary-tenuinucellate Type II
unitegmentary-crassinucellate Type III
unitegmentary -tenuinucellate Type IY
Representative examples of these four types are shown in Figs. 1-4.
It is unlikely that any of these types are strictly homogeneous,
with their various parts truly homologous. For example, BHANDARI
et al. (1976) suggest that the origin of the integument primordium
may prove to be wholly dermal in the sympetalae in general, whereas
in other groups with tubular corollas (e.g. Cucurbitaceae, Convolvulaceae)
it is subdermal and vascularized. Even if much more were known
about the morphogenesis of the integuments and the nucellus it is
unlikely that many convincing differences could be established between
large sub-groups within these types.· At least two considerations ma~e
such distinctions difficult to prove. In the first place, considerable
differences in morphogenesis have· been described in ovules whose
close relationship cannot be doubted. It is doubtful, therefore, if such
differences would ever provide convincing evidence that two forms
were unrelated. BOUMAN (1975) has found considerable variation in
development within a natural family, the Crucijerae. In Brassica and
Sinapis initiation of the outer integument is subdermal; in Lunaria
it is partly subdermal and partly dermal; while in CapseUa it is com-
pletely dermal. Similarly initial divisions in the formation of the
nucellar beak in Euphorbia were found by BOR & BOUMAN (1974) to
be subdermal, though epidermal cells of the flank later contribute
to the beak. In another genus of the Euphorbiaceae, Codiaeum, they
found the initiation of the beak to be by divisions of both the parietal
cells -and of cells of the nucellar dermatogen. If such variations occur
in related plants, it may prove difficult to demonstrate convincing
differences between ovules in relation to these particular characters.
Secondly, it must be borne in mind that each ovular type, as defined
above, comprises ovules of. very different general appearance. As
extreme examples, the ovules of Aralidium and Rhododendron may
be compared (Figs. 5, 6). Both are unitegmentary but the integument
of Aralidium is very thick and is traversed by many vascular bundles
(personal observation), whereas that of Rhododendron is thin and no
vascular strand is present even in the funicle (PALSER et al. 1971).
In spite of this great range in form there are good grounds for believing
that these two examples represent extremes of a single series. In the
Ovular Morphology and the Classification of Dicotyledons 127

Fig. 5. Ovule of Aralidium pinnati/idum MIQ. in sagittal section, X 50.

and in transverse section, X 40. Integument massive and with many
vascular bundles
Fig. 6. Ovule of Rhododendron leptanthum F. v. MULL. X 75. The thin
integument is prolonged beyond the micropyle and the chalazal end of
the ovule also forms a "tail" in the seed

Araliaceae (in which the integument is thick and the nucellus may be
of more than one layer) the nucellus usually degenerates so that the
embryo sac comes into direct contact with the integument, the inner
layer of which may form a well differentiated endothelium. These
are characteristics of the ovule typical of many sympetalous families.

DAVIS (1966) distinguishes between strictly crassinucellate ovules,

in which the archesporial cell cuts off a primary parietal cell (which usually
divides further) and pseudo-crassinucellate ovules, in which no primary
parietal cell is formed, but in which apical cells of the nucellar epidermis
128 w. R. PHILIPSON:

divide to form a nucellar cap. In each case the megaspore-mother-cell

is capped by more than one layer of cells, though the origin of these cells
is different. In the original presentation of this classification (PHILIPSON
1974) this distinction was referred to, but not taken into account in making
comparisons. It was felt that insufficient attention had been given to this
feature, so that conclusions could not be based on the inadequate informa-
tion available. DAVIS (1966) recorded pseudo-crassinucellate ovules in few
dicotyledonous families, namely Frankeniaceae, Oalycanthaceae, Podo-
phyllaceae and Oobaeaceae and also in some members of Ranunculaceae,
Boraginaceae, Ehretiaceae, Labiatae and Asclepiadaceae (though this last
may have been in error). In all the sympetalous families in this list, the
pseudo-crassinucellate condition occurs in one or a few members of other-
wise uniformly tenuinucellate relationships (regarding Oobaeaceae as a
satellite family of the Polemoniaceae). The situation in the Ranunculaceae
(where several genera are pseudo-crassinucellate) is of greater interest,
but it scarcely affects the comparisons made here. For these reasons no
subdivision of the four ovular types is attempted on the origin of the
nucellar tissue.

Distributiou of Ovular Types

The distribution among the orders and families of dicotyledons

of these two characters, both separately and in association, has been
set out by O. DAHLGREN (1927), WUNDERLICH (1959) and PHILIPSON
(1974) and was further discussed by R. DAHLGREN (1975b). The last
author stresses that the characters are known from too few examples
for broad generalizations to be firmly established. Undoubtedly a
much larger sample is desirable, but it is felt that the main conclusions
reached in this presentation rest on features sufficiently constant to
encourage some confidence in their acceptance.
It is not necessary to restate here the detailed information already
published by the above authors and also by DAVIS (1966). All that is
required is a summary. This can be provided most conveniently by
describing the distribution of the combinations of the two characters,
that is to say, of the four ovular types (Table 1). The names of taxa
are used as in CRONQUIST (1968) unless otherwise stated.
At the risk of over-simplification, the distribution of the ovular
types can be described very briefly. Type I is found in nearly all members
of CRONQUIST'S subclasses Magnoliidae, Hamamelididae and Caryophyll-
idae, and also in most members of the greater parts of his subclasses
Dilleniidae and Rosidae (Table 1, column 1). However, a substantial
number of families of the Dilleniidae are characterized by Type II
ovules and a few by Type IV (Table 1, columns 2 and 4), and an even
larger section of families of the Rosidae have ovules of Types III and IV
(Table 1, columns 3 and 4). In the remaining subclass, the Asteridae,
virtually all members have Type IV ovules (Table 1, column 4).
Ovular Morphology and the Classification of Dicotyledons 129

Table 1. The distribution of ovular types among the sub-classes and families
of Dicotyledons (minor exceptions omitted)

(1) Type I (2) Type II (3) Type III (4) Type"IV

Magnoliidae Theaceae Cornaceae Asteridae

(except. Circeastaceae Guttiferae Davidiaceae Ericales
Sabiaceae Ochnaceae Nyssaceae Diapensiaceae
Peperomiaceae) M arcgraviaceae Alangiaceae Escalloniaceae
Lecythidaceae Garryaceae Hydrangeaceae
Primulaceae Araliaceae Philadelphaceae
Caryophyllididae M yrsinaceae Aquifoliaceae Pittosporaceae
Theophrastaceae I cacinaceae Alseuosmiaceae
Coridaceae Umbelliferae
Hamamelidae 8tyracaceae Grubbiaceae
(except Myricaceae (except Halesia)
Eucommiaceae Ebenaceae
.Tuglandaceae (etc., etc.)

Dilleniididae Rafflesiaceae Hippuridaceae Sarraceniaceae

(except families in Byblidaceae Theligonaceae Loasaceae
upper part of co\. 2 Podostemaceae Bruniaceae Limnanthaceae
and Parnassiaceae J uglandaceae Santalaceae
Ericales l' ahliaceae Myricaceae Opiliaceae
Diapensiales 8tackhousiaceae Sabiaceae Hydnoraceae
Sarraceniaceae Oxalidaceae Eucommiaceae Circaeastaceae
Loasaceae) Tropaeolaceae Peperomiaceae
Balsaminaceae
Rosidae
(except families of
this sub-class listed
in other 3 columns)

The outline of the classification proposed here is that those families

which consistently possess ovules of Type II form a subclass (the
Thealean-Primulalean complex), and that those with ovules of Types in
and IV should be united into another subclass. Since most of what
follows relates to this second subclass, it will be useful to give this
taxon an informal name. Accordingly, the convenient term Uniteg-
minae is adopted from the system of VAN TIEGHEM (1901).
Before advancing further arguments in favour of this classification,
some additional details of the distribution of ovular types need to be
considered. In each of the last three columns of Table 1, the families
above the dotted line form what appear to be natural and coherent
groups. The families listed below the dotted line represent small, often
highly specialized families, frequently of uncertain relationship, and
Plant Syst. Eyol., Suppl. 1 9
130 W. R. PHILIPSON:

for the most part without evident relationships among themselves.

Possibly some should be associated with the group above the dotted
line, especially as a number are reported to contain iridoid compounds
(JENSEN et al. 1975), substances which occur rather consistently within
many unitegmic families, as will be discussed later. Even so these
aberrant families indicate that morphological modification of the
ovule must have occurred repeatedly during the evolution of the di-
cotyledons. This is shown even more forcibly by the several polypetalous
and apetalous families in which aberrant ovular characters occur in
only a few of their members. These were referred to by the authors
cited above but are omitted here. It is to be expected that their true
systematic relationships will become clearer on more intensive study.
Families characterized by Type I ovules will not be considered
further. They represent about half of the dicotyledons and clearly
include several major taxa. One proposal advanced here is that the
families above the dotted line in Table 1, column 2 constitute a major
subdivision which spans thc boundary between the polypetalous and
the sympetalous condition. The affinity of the Theales and Prim'ulales
is acknowledged by several authors and the case for the adoption of
this major taxon will not be further advanced here, as the case has
already been stated (PHILIPSON 1974). The main part of the proposal
is that the groups of families above the dotted line in both columns 3
and 4 of Table 1 together form a single subclass, the Unitegminae.
The evidence for regarding such a large and diverse group as a single
taxon will now be reviewed.

The Unity of the Unitegminae

It is significant that the association of polypetalous families here
included in the Unitegminae had been recognized previously as a com-
plex of related families without any reference to their ovular morphology.
When considering the relationships of the trans-Pacific genus Griselinia
PHILIPSON (1967) concluded that this genus could not be referred to
any existing family, but showed affinity with the Escalloniaceae, the
Cornaceae and the Araliaceae. These families were considered to be
part of a complex which included the satellite families around the
Cornaceae as well as families often referred to as the "woody Saxi-
Iragaceae". Of polypetalous families omitted at that time but now
included, the Pittosporaceae and Alseuosmiaceae are now considered to
be close to the complex on other grounds besides ovular morphology,
The status of the Aquiloliaceae and Icacinaceae will be discussed pres-
ently as will that of the Actinidiaceae and Grubbiaceae,
RODRIGUEZ (1971), in discussing the relationships of the Umbel-
Ovular Morphology and the Classification of Dicotyledons 131

lijerae, considered a rather wider range of families, notably including

some sympetalous groups, a topic to which we shall revert presently.
Nevertheless, at the base of this flabellate array of families he clearly
recognized the same core-complex of Escalloniaceae-Cornaceae-Ara-
liaceae, each with their satellite families. RODRIGUEZ also reached
these conclusions with only minor reference to ovular morphology.
The fact that the ovules of all the members of this complex are unitegmic
(and most also tenuinucellate) would be a remarkable coincidence if
it were not due to their natural relationship. Only two small families
that might be regarded as immediate associates of the complex have
Type I ovules. These are Brexiaceae and Grossulariaceae (in the narrow
sense). However, BENSEL & PALSER (1975b) conclude from evidence
of floral anatomy that Ribes is similar to other members of the Saxi-
jragoideae. It may, therefore, have no close relationship with the Uni-
tegminae. The association together of Brexia and Ixerba in a small
family may be artificial, but at least Ixerba appears correctly placed
near the Escalloniaceae (BENSEL & PALSER 1975a) and therefore should
be accepted as a member of the Unitegminae in which the ovule is Type I.
On the basis of the stereo structure of the exine, HIDEUX & FER-
GUSON (1976) also suggest an evolutionary link between the Escal-
loniaceae, Cornaceae, and Araliaceae, which together form the core of
the Unitegminae. They include in their consideration the Cunoniaceae
and Saxijragaceae (sensu stricto), the whole group in their opinion
occupying a somewhat central position palynologically.
One of the most telling points in favour of the unity of the Uni-
tegminae is the existence of many anomalous genera which cannot
be placed with certainty in any of the larger families of the complex
but which appear to combine the characters of several of them. TAKHTA-
JAN (1969) explains the motives behind his close association of Ara-
liaceaeand Cornaceae. The existence of Mastixia, Helwingia and Tori-
cellia. nullified his attempts to separate these families. The discussion
by RODRIGUEZ (1971) brings out the complexities of the distribution
of characters among these genera. RODRIGUEZ also accepted the im"-
portance of Griselinia as a type in which features of Araliaceae, Cor-
naceae and Escalloniaceae were blended. He supports the view, already
advanced by CRONQUIST (1968) and TAKHTAJAN (1969), that Corokia
forms a bridge between Cornaceae and Escalloniaceae. I wish to call
attention to another genus which has been variously placed in the
Cornaceae and Araliaceae. The mono-typic Malayan genus Aralidium
shares a wide range of characters with Griselinia and, having no resin
canals and a dorsal raphe to the ovule, also approaches the Cornaceae
more closely than the Araliaceae. However, the pits of the fibres of
its secondary xylem are un bordered (unpublished observation), a
132 w. R. PHILIPSON:
feature characteristic of Araliale8 and contrasting with the bordered
pits found in Griaelinia and other Oornales.
The distinction between the fibre-tracheids of the Oornaceae and
the libriform fibres of the Araliaceae is one of several characters which
show that the complex of families under consideration is by no means
uniform. A general discussion of this heterogeneity will be given later,
but the condition in Aralidium illustrates that these distinctions can
be bridged and do not necessarily prevent the acceptance of the whole
complex as a single coherent major taxon. At this level, taxa are not
defined by the possession of anyone character, or even by a group
of characters, but rather by a majority of a whole syndrome of characters.
In the particular case of this complex the syndrome is not easily iden-
tified because it consists in large part of generalized non-specific
characters. RODRIGUEZ (1971) characterized the well-established con-
cept of the UmbeUillorae (including Araliaceae, Umbellilerae and Oor-
naceae "in the widest sense) "by mostly pentamerous or tetramerOUB
flowers with a reduced calyx, free petals and stamens, inferior ovaries
mostly with a pulviniform disk or a stylopodium supporting one or
more styles, tending to 5, 4, 2, or 1 locuies with one pendent, anatropous
functional ovule per locule; by drupaceous or schizocarpous fruits (a
few baccate), and seed with abundant endosperm and small, straight
embryos; by occasional trends to unisexuality and apetaly; and by
usually involucrate, cymose, umbellate or capitate inflorescences".
To include the EscaUoniaceae (and its allies) this definition requires
only slight modification in respect to ovary position and placentation.
When discussing Griselinia (PHILIPSON 1967) I summed up the charac-
teristics of these families as "being generally similar in wood anatomy,
pollen morphology, and floral and fruit morphology". In now seeking
to define this complex by its ovular morphology I am not proposing a
single-character classification. Rather, my intention is to use the ovule
as the character which serves best to define a group already identified.
It serves best because it is precise and also because it coincides more
consistently than any other character with the boundaries under
consideration.
Even if the coherence of the polypetalous members of the Uni-
tegminae were to be accepted, the question of their relationship with
the sympetalous families (excluding Primulales and Ebenales) remains.
These families make up the subclass Asteridae of CRONQUIST (1968).
It is considered that various groups of sympetalous families relate to
different polypetalous groups, but that all of these fall within the
Unitegminae. That is to say, the Asteridae are derived by several routes
from a common ancestral complex.
Chemical evidence now substantiates the subdivision of the Asleridae
Ovular Morphology and the Classification of Dicotyledons 133

into two major groups, each with distinct relationships among poly-
petalous families. HEGNAUER (1969) has published a scheme illustrating
the close links between the Compositae and the Araliales on the one
hand, and between the Caprifoliaceae-Rubiaceae and the Cornaceae-
Hydrangeaceae on the other. A very similar scheme is also illustrated
by FROHNE & JENSEN (1973: Fig. 98) and JENSEN et al. (1975) con-
sider that the iridoid-containing polypetalous and sympetalous orders
to "make up a relatively homogeneous and probably monophyletic
group". Another chemical character of this group is the absence of
polyines which are frequent in the Compositae and the Araliales.
Three further links between the unitegmic polypetalous orders
and the sympetalous orders will be mentioned. Firstly, the Empetraceae
and Clethraceae are generally associated with the Ericales, and a similar
affinity is frequently assigned to the Cyrillaceae. FAGERLIND (1947)
has brought forward evidence, especially embryological, supporting
a similar link between the Grubbiaceae and the Ericaceae. Much earlier
LINDLEY (1846) had proposed a relationship between the Actinidiaceae
and the Ericaceae, and a succession of authors until the present day
(TAKHTAJAN 1969) have supported this. Secondly, the suggested
relationship between the Aquifoliaceae and the Oleaceae (e.g. WETT-
STEIN 1930-1935) was upheld by MAURITZ ON (1936) mainly on
embryological evidence. Thirdly, the two small segregate families
Columelliaceae and Alseuosmiaceae both have sympetalous corollas,
but are thought to be most closely related to the polypetalous families
Escalloniaceae (STERN et al. 1969) and Pittosporaceae (CRONQUIST
1968) respectively, aQd indeed the corolla of the Pittosporaceae itself
is sometimes incipiently tubular.
Having briefly referred to the ramifications of the complex among
sympetalous orders, we may now look for indications of its origins
among bitegmic polypetalous alliances. That such relationships exist
is suggested by the presence of bitegmic ovules in Ixerba and Quintinia.
The relevance of this type of ovule in Brexia and Ribes has already
been discussed. The transition from ovules of Type I to those of Type IV
probably took place in two steps, since a few families of the Unitegminae
have crassinucellate (perhaps only pseudo-crassinucellate) ovules. It
is particularly interesting that the Araliaceae stand in this relationship
to the Umbelliferae, and even within the latter family one example
of a pseudo-crassinucellate ovule has been reported (GUPTA 1964).
Similarly a gradation from Type I to Type III, wi~,h some few examples
of Type IV, exists within the Celastrales. The inclusion of the Aquifoliaceae
and the lcacinaceae within the Unitegminae is debatable, as the acquisi-
tion of the single integument may have occurred independently within
the Celastrales. However, the inclusion of these two families together
134 W. R. PHILIPSON:

in the same order appears questionable as iridoid compounds are known

in the Celastrales only from members of the lcacinaceae. This suggests
a close tie between this family and the iridoid-containing Unitegminae.
This still leaves the position of the Aquifoliaceae problematical.

The existence of these transitional taxa documents what would in any

event be suspected, namely that the unitegmic ovule was derived from the
bitegmic ovule. To search for the stock from which the Unitegminae arose
among extant polypetalous plants is not likely to be productive. A group
with simple actinomorphic flowers is probable. One such derivation was
suggested by HEGNAUER (1969) when he drew attention to striking chemical
resemblances between the Araliaceae and Piuosporaceae on the one hand
and the Rutaceae on the other. He had earlier (HEGNAUER 1963) discussed
similarities between the alkaloids of the Rutaceae and the Polycarpicae
(Magnoliales) , and KUBITZKI (1969) developed this suggestion. It might
be possible, therefore, to construct on chemical grounds an evolutionary
line extending from the Magnolian stock through .ij.utalean-like plants
to polypetalous Unitegminae and eventually to the'-sytnpetalous families.
A diagram of such a scheme has in fact been published by FISH & WATER-
MAN (1973).

Variability of Some Characters Within the Unitegminae

One of the strongest arguments in favour of the reality of the
Unitegminae is the almost complete restriction of iridoid compounds
to groups with a single integument. Only two reports of iridoid com-
pounds in plants with two integuments are given by PLOUVIER &
FAVRE-BoNVIN (1972) and JENSEN et al. (1975) in their reviews of the
distribution of these substances. Both these exceptional plants (Li-
quidambar and Daphniphyllum) are usually placed in the Hamameli-
dales, and both have bitegmic ovules, but Daphniphyllum is of very
doubtful affinity: it is sometimes placed in the Euphorbiales and it
should be noted that THORNE (1968) places it next to the Pittosporaceae.
Although iridoid compounds are so closely associated with unitegmic
families, the converse is not necessarily true, for there are many uni-
tegmic taxa from which no iridoid substances have been reported. The
distribution of iridoid compounds is not random, Jm: they occur in
assemblages of related orders or families. They therefore indicate a
subdivision within the Unitegminae, defined by the presence or absence
of these compounds. It is to be expected that any large section of
dicotyledonous orders will be heterogeneous in respect to several
characters. Within the Unitegminae this is illustrated. by various views
on the relationship of the Araliales to the Cornales. EYDE (1967),
CRONQUIST (1968) and R. DAHLGREN (1975) diverge from the formerly
more generally held opinion that they are closely related. While the
separation of these two orders can be based on morphological grounds
Ovular Morphology and the Classification of Dicotyledons 135

(e;g. resin ducts, position of raphe), chemical data are more compelling
(HEGNAUER 1969, FRORNE & JENSEN 1973, JENSEN et al. 1975). The
implications of the distribution of iridoid compounds will first be con-
sidered, followed by comparisons with the distribution of a few other
eharacters.
The following summary of the distribution of iridoid compounds
relies on the information given by JENSEN et al. (1975) which should
be consulted for much detail here omitted (the names of plant groups
are used in the sense of CRONQUIST 1968). Iridoid compounds occur
in the Cornales and in the Hydrangeaceae, Grossulariaceae and Icacinaceae
among polypetalous families, and in the following sympetalous alliances:
(ientianales, Ericales, Scrophulariales, Dipsacales, Rubiales, and La-
miales (except Boraginaceae). Conversely, iridoid compounds are not
known to occur among unitegmic polypetalous families in the Araliaceae,
Umbelli/erae, Pittosporaceae and Aqui/oliaceae, nor among the fol-
lowing sympetalous families: Solanaceae, Convolulaceae, Polemoniaceae,
Boraginaceae, Campanulaceae and Compositae.
Acceptance of this important chemical cleavage within unitegmic
groups does not detract from the validity or usefulness of the Uniteg-
minae. It is usual for a major group to embrace subdivisions of which
the common ancestral stock is unknown and probably extinct. The
monocotyledons and dicotyledons within the angiosperms provide a
parallel. Nor does the recognition of thc larger inclusive group detract
from the importance of the evolutionary line marked by the presence
of iridoid compounds. Probably we shall never know conclusively
whether the acquisition of iridoid substances preceded or followed
the reduction in integument number in this evolutionary line. For
thc resolution of phylogenetic problems events need to be placed in
sequence. If the acquisition of iridoid compounds came first, then the
group here called Unitegminae must be polyphyletic. But I b~lieve
the most probable sequence is that the unitegmic ovule arose in a
core group from which the whole Unitegminae has been derived. Within
this core, at an early stage, two divergent lines arose by the acquisition
of iridoid compounds. This belief is based on the general similarities
between polypetalous members of the complex and also on a preference
for the most simple explanation (the principle of parsimony in evolution).
In any event, even if the cladistic relations of the members of the
U nitegminae cannot be clarified, the concept of this grollp appears
a logical major taxon within the dicotyledons. FRORNE & JENSEN
(1973: Fig. 98) depict the orders which are included here in the U ni-
tegminae as two evolutionary lines linked by a common origin. This
stock which gave rise to both the Araliales and the Cornales must
have been unitegmic and thus the whole complex is drawn together.
136 W. R. PHILIPSON:

A contrary OpInIOn has been expressed by .JENSEN ot al. (1975) who

suggest that iridoid compounds developed along an evolutionary line ill
which the ovules were just about to switch from the bitegmic t9 the unitegmic
condition. The principal factor in support of this view is the presence of
iridoid compounds in Liquidambar and Daphniphyllum, which are bitegmic
and probably members of the Hamamelidales. If the iridoid compounds
are considered to have arisen subsequent to the reduction in integument
number, then it would be necessary to postulate an independent origin for
these compounds in the Hamamelidales, though .JENSEN et al. (1975) con·
sider the possibility that these two genera represent relicts of primitive
iridoid.bearing groups with bitegmentary ovules. However, a double
origin of the iridoids appears more likely than the alternative, which is
the independent evolution of the Type IV ovule in the Aralialean and t,}\C
Cornalean groups-groups which show affinity on a number of other grounds.
It is of some significance that a number of other characters which
vary within the Unitegminae show no correlation with the distribution
of iridoid compounds. Among several such characters the distribution
of three will be considered.
Nodal anatomy. Variability of the leaf-trace pattern within
families is now known to be much more variable than at the time
SINNOTT (1914) first drew attention to the taxonomic value of this
character (HOWARD 1974). Nevertheless, a particular nodal type often
predominates in a family or large group of families. This is true within
the Unitegminae, most of which (in both iridoid and non-iridoid contain-
ing groups) are basically unilacunar. However, most of the polypetalous
members are tri-Iacunar (or in the Araliales, multi-lacunar). Of the
sympetalous groups in which tri-Iacunar nodes predominate, one,
the Oompositae is non-iridoid containing, and another, the lJipsacales,
is iridoid-containing.
Bi- and tri-nucleate pollen grains. Many of the familie!:!
or orders within the Unitegminae include examples of both hi- and tri-
nucleate pollen (BREWBAKER 1967), indicating that the step from two
or three nuclei has occurred repeatedly within this group. This applies
equally to both iridoid-containing and non-iridoid-containing families.
Of the principal polypetalous orders of the Unitegminae it is only the
Oornales that have bi-nucleate pollen.
Endos perm types. Most non-iridoid containing families have
nuclear endosperm (R. DAHLGREN 1975b) though there are important
exceptions, and both ab-initio cellular and nuclear endosperm occur
in the Solanaceae and particularly in the Oompositae. Conversely,
most iridoid-containing groups have ab-initio cellular endosperm,
but again with notable exceptions (e.g. Rubiales, Gentianales).
The lack of correlations between the distributions of these variou!:!
characters serves to remind 11s that we are dealing with the end products
of a long and intricate evolutionary history throughout which the
Ovular Morphology and the Classification of Dicotyledons 137

Angiosperms have displayed great versatility. It is against such a

confused background that the ovule and the iridoid substances appear
as enduring markers of a main stream of evolution (Fig. 7). Evidently
the Type IV ovule, once acquired remained very stable throughout

BITEGMINAE UNITEGMINAE

Fig. 7. Schematic representation of some major sub·divisions of the di-

cotyledons. Living groups are labelled. The families and orders making
up t.he Unitegminae are listed more fully in Table 1. Names of taxa as in
CRONQUIST (1968) except that Boraginaceae are excluded from the Lamiales

massive radiating evolution that produced about half of the existing

dicotyledons, including many most successful families. It is to the
establishment of this successful dynasty that the sympetalous ovule
owes its importance in systematics.
In contrast, the single integument or the thin nucellus evolved
repeatedly, as is shown by their sporadic presence in several polypetalous
families most of whose members have Type I ovules. The amount of
138 W. R. PHILIPSON:

radiating evolution following these events appears to have been slight,

and the effect on major classification is, therefore, negligible. The
repeated acquisition of these characters is not surprising if, as seems
probable, the reduction in integument number has adaptive significance.
It is normal within angiosperms for any opportunity to be taken up
by several alliances advancing over a broad front. STEBBINS (1974)
brings to notice an interesting hypothesis deriving the angiospermous
ovule from a structure like the cupule of some Oaytoniales by reduction
in the number of ovules per cupule. In the Oaytoniaceae each ovule was
enclosed within its own single integument and several occurred within
each cupule. In the Oorystospermataceae the ovules were reduced to
one. The cupule and its contents then become equivalent to a bitegmic
ovule. It may then be supposed that one of the integuments is super-
fluous and consequently would be lost. Moreover, this loss would most
likely have occurred not once but repeatedly.

References
BENSEL, C. R., and PALSER, B. F., 1975a: Floral anatomy in the Saxi-
fragaceae sensu lato. 1. Introduction, Parnussioideae and Brexioideae.
Amer. J. Bot. 62, 176--185.
- 1975b: Floral anatomy in the Saxifragaceae sensu lato. II. Saxi-
fragoideae and lteoideae. Amer. J. Bot. 62, 661-675.
BHANDARI, N. N., BOUMANN, F., and NATESH, S., 1976: Ovule ontogeny
and seed coat structure of Scrophularia himalensis Royle. Bot. J ahrb.
Syst .. 95, 535-548.
BOR, J., and BOUMAN, F., 1974: Development of ovule and integuments
in Euphorbia milii and Oodiaeum variegetum. Phytomorphology 24,
280-296.
BOUMAN, F., 1975: Integument initiation and testa development in some
Oruciferae. Bot. J. Linn. Soc. 70, 213-229.
BREWBAKER, J. L., 1967: The distribution an phylogenetic significance
of binucleate and trinucleate pollen grains in the Angiosperms. Amer.
J. Bot. 54, 1069-1083.
CRONQUIST, A., 1968: The Evolut·ion and Classification of Flowering Plants.
Boston: Houghton Mifflin.
DAHLGREN, K. V. 0., 1927: Die Morphologie des Nuzellus mit besonderel'
Beriicksichtigung der deckzellosen Typen. Jb. wiss. Bot. 67, 347-426.
DAHLGREN, R., 1975a: A System of classification of the Angiosperms to
be used to demonstrate the distribution of characters. Bot. Notiser.
128, 119-147.
1975 b: The distribution of characters within an Angiosperm System 1.
Some Embryological Characters. Bot. Notiser 128, 181-197.
DAVIS, G. L., 1966: Systematic embryology of the Angiosperms. New York-
London-Sydney: John Wiley.
EYDE, R. H., 1967: The peculiar gynoecial vasculature of Oornaceae and
its systematic significance. Phytomorphology 17, 172-182.
FAGERLIND, F., 1947: Die systematische Stellung del' Familie Grubbiaceae.
Svensk bot. Tidsskr. 41, 315--320.
Ovular Morphology and the Classification of Dicotyledons 139

FISH, F., and WATERMAN, P. G., 1973: Chemosystematics in the RutaceaeII.

The Chemosystematics of the ZanthoxylumJFagara complex. Taxon 22,
177-203.
FROHNE, D., and JENSEN, U., 1973: Systematik des Pflanzenreichs unter
besonderer Beriicksichtigung chemischer Merkmale und pflanzlicher
Drogen. Stuttgart: Gustav Fischer.
GUPTA, S. C., 1964: The embryology of Coriandrum sativum L. and Foeni-
culum vulgare MILL. Phytomorphology 14, 530-547.
HEGNAUER, R., 1963: The taxonomic significance of alkaloids. In: Chemical
Plant Taxonomy (SWAIN, T., Ed.), 389-427. London-New York:
Academic Press.
1969: Chemical evidence for the classification of some plant taxa. In:
Perspectives in Phytochemistry (HARBOURNE, J. B., and SWAIN, T.,
eds.), 121-138. London-New York: Academic Press.
1971: Chemical patterns and relationships of Umbelliferae. In: The
biology and chemistry of the Umbelliferae (HEYWOOD, V. H., ed.),
267-277. London-New York: Academic Press.
HIDEUX, M. J., and FERGUSON, 1. K., 1976: The stereostructure of the
exine and its evolutionary significance in Saxifragaceae sensu lato. In:
The evolutionary significance of the exine (FERGUSON, 1. K., and MUL-
LER, J., eds.), 327-377. London: Academic Press.
HOWARD, R. A., 1974: The stem-node-leaf continuum of the Dicotyledoneae.
J. Arnold Arb. 55,125-173.
;JENSEN, S. R., NIELSEN, B. J., and DAHLGREN, R., 11)75: Iridoid Com-
pounds, their occurrence and systematic importance in the Angiosperms.
Bot. Notiser 128, 148-180.
K UBITZKI, K., 1969: Chemosystematische Betrachtungen zur Gro13gliede-
rung der Dicotylen. Taxon 18, 360-368.
LINDLEY, J., 1847: The Vegetable Kingdom, edT 2. London: Bradbury
and Evans.
MAURITZON, J., 1936: Zur Embryologie und systematischen Abgrenzung
der Reihen 7'erebinthales und Celastrales. Bot. Notiser 1936, 161-212.
1939: Die Bedeutung del' embryologischen Forschung fiir das natiir-
liche System der Pflanzen. Acta Univ. Lund. ser. 2, 35, 1-70.
FALSER, B. F., PHILIPSON, W. R., and PHILIPSON, M. N., 1971: Embryology
of Rhododendron. Introduction and ovule, megagametophyte, and early
endosperm development in R. yunnanense. J. Ind. Bot. Soc. 50 A,
172-188.
PHILIPSON, W. R., 1967: Gri,~elinia FORST. fil.-Anomaly or link. New Zea-
land J. Bot. 5, 134-165.
1974': Ovular morphology and the major classification of the dicotyle-
dons. Bot. J. Linn. Soc. 68, 8!l--108.
1!l75: Evolutionary lines within the dicotyledons. New Zealand J. Bot.
13, 73-91.
PLOUVIER, V., and FAVRE-BoNVIN, J., 1!l72: Les iridoides et seco-iridoides:
repartition, structure, proprietes, biosynthese. Phytochemistry, 10,
1967-1722.
RODRIGUEZ, R. L., 1971: The relationship of the Umbellales. In: The Biology
and Chemistry of the Umbelliferae (HEYWOOD, V. H., ed.), 63-91.
London: Academic Press.
140 W. R. PHILIPSON: Ovular Morphology and the Classification ...

SINNOTT, E. W., 1914: Investigations on the phylogeny of the Angio-

sperms, I. The anatomy of the node as an aid in the classification of
Angiosperms. ArneI'. J. Bot. 1, 303-322.
STEBBINS, G. L., 1974: Flowering Plants, Evolution above the species
level. Cambridge, Mass.: Belknap Press.
STERN, W. L., BRIZICKY, G. K., and EYDE, R. R., 1969: Comparative
anatomy and relationships of Oolumelliaceae. J. Arnold Arb. 50, 36-75.
- SWEITZER, E. M., and PHIPPS, R. E., 1970: Comparative anatomy and
systematics of woody Saxi/ragaceae, Ribes. In: New Research in Plant
Anatomy. Suppl. 1 (ROBSON, N. K. B., CUTTER, D. F., and GREGORY,
M., eds.), Bot. J. Linn. Soc. 63, 215-237. London: Academic Press.
TAKHTAJAN, A., 1969: Flowering Plants, Origin and Dispersal. Trans.
Jeffrey, C. Edinburgh: Oliver and Boyd.
THORNE, R. F., 1968: Synopsis of a putatively phylogenetic classification
of the flowering plants. Aliso 6, 57-66.
TIEGHEM, P. VAN, 1901: L'oeuf des plantes considere comme base de leur
classification. Ann. Sci. nat. (Bot.) ser 8, 14, 213-390.
WARMING, E., 1878: De l'ovule. Ann. Sci. nat. (Bot.) ers. 6, 5, 177-266.
- 1914: Observations sur Ie valeur systematique de l'ovule. Mindeskr.
Japetus Steenstrups F0ds. 24, 1-45.
WETTSTEIN, R. VON, 1930-1935: Randbuch del' systematischen Botanik,
4th ed. Leipzig-Wien: Deuticke.
WUNDERLICH, R., 1959: Zur Frage del' Phylogenie del' Endospermtypen
bei den Angiospermen. Osterreich. Bot. Zeitschr. 106, 203-293.

Address of the author: Professor W. R. PHILIPSON, Department of

Botany, University of Canterbury, Christchurch, New Zealand.
Plant Syst. Evo!., Supp!. 1, 141-153 (1977)
@ by Springer-Verlag 1977

Institut fur Systematische Botanik der Universitat Munchen, Federal

Republic of Germany

Seed Characters in and Affinities Among the Saxifragineae

By

J. E. Krach, Miinchen

Abstract: Neither the Rosales nor the Saxifragales, as defined by the

majority of recent authors, can be regarded as being monophyletic. Also
the term "slightly pleiophyletic" does nothing to elucidate the deep gaps
separating the several evolutionary lines usually combined to form the
order (or even suborder) Saxifragales.
The clear-cut concepts of an order Orassulales leave room for only the
herbaceous families Orassulaceae (including Penthorum and most of the
Saxifrageae), Astilbaceae, and Peltiphyllaceae. Some further genera (Vahlia,
Eremosyne, and Oephalotus) seem to be distantly related to these families.
The Hydrangeaceae and the Escalloniaceae proper share a remarkable
number of characteristics, which allows their combination as the Escal-
loniales.
The two otherwise very isolated genera Ribes and Phyllonoma may
be related.
The Ounoniales (inc. Brunellia, Bauera, PoUingeria and perhaps the
herbaceous Francoaceae) are another closely related group of families.
Only very doubtfully the woody orders Escalloniales, Grossulariales
and Cunoniales possess any closer affinity to the herbaceous families
Saxifragaceae and Crassulaceae.
Most of the taxa referred to as saxifragaceous have been so poorly
investigated that it is at present premature to make any far-reaching con-
clusions regarding their true relationship. Until more knowledge has been
obtained only their evolutionary trends can be suggested without a full under-
standing of their meaning for the taxonomic treatment of the whole group.

The Rosiflorae are still considered by many botanists as a natural

and even as a monophyletic order the broad limits of which correspond
to the "Reihe Rosales" of ENGLER (1930), although there is a multi-
plicity of publications for decades which point to the untenability of such
a procedure. In some of the more recent systematic proposals the three
Englerean Unterreihen Hamamelidineae, Rosineae, and Saxifragineae
are treated with otherwise unaltered boundaries as orders.
142 .1. E. KRACH:

About the current classification of the genera related to Hamamelis,

ENDRESS in his contribution to this symposium has said much more
than I can. As far as the Rosineae in the Englerean sense are concerned
I would like to restrict myself to one comment. The lumping together
not 'only of the Rosaceae and Fabaceae, but also N euradaceae, Conna-
raceae, Chrysobalanaceae, and Krameriaceae, is less and less convincing
the more we know not only about the morphology of the flowers, but
about the pollen, the anatomy of the wood and the biochemistry of
representatives of these groups.
With the following arguments I hope to show also that the Saxi-
Iragineae of ENGLER (1930). corresponding to the Saxilragales of many
more recent authors, consist instead of several evolutionary lines which
are best not combined into one single order.
If one is in search of the evolutionary lines of a higher taxon it is
certainly useful to find a starting-point, common to each of these lines;
that is to determine certain basic characteristics from which the devel-
opment of the recent taxa might have begun. For such a heterogenous
taxon as the Saxilragineae classified by ENGLER, which envelops Cras-
sulaceae and Cunoniaceae, Pittosporaceaeand Cephalotaceae as well, this
becomes .a rather challenging task. It is best to limit oneself first to
subcategories in order to achieve any result at all.
But presumably, merely by being forced to accept narrower horizons
one is at first inclined to doubt in the monophyletic structure of the
entire group. However, the search for unifying factors in a large group, in
itself implies a certain homogeneity within this group. So it seems to me
that the process used has itself some prejudicial connotations which
cannot entirely be avoided.
A rather broad consensus of basic and essential characteristics
within the Englerean Saxilragineae can be demonstrated, which applies
to more than half of the genera of this taxon:
seeds anatropous, raphe clearly visible in the ripe seed:
two double-layered integuments;
outer epidermis of the seed coat thickened following the Reseda-type:
inner epidermis of the seed coat compressed to form a tannini-
ferous pigment-layer;
seed coat with tannins, but without lignin: cell-walls in the seed coat
of the ripe seeds without visible pits:
in the ripe seeds scarcely any nucellar-but at least some endo-
spermous tissue remains;
embryo straight, in the direction of the longitudinal axis of the seed;
cotyledons shorter than hypocotyl. the plumule, if present at all,
being only a slight bump;
accumulation of sedoheptulose. no hamamelose:
Seed Charact.ers in and Affinities Among the Saxifragineae 143

root tips containing anthocyanin;

largely herbaceous, rarely having secondary xylem and phloem;
tendency to form five-whorled flowers with five parts in each 1.
By this correlation of characteristics the Crassulaceae, Penthorum
and most of ENGLER'S Saxilrageae are bound together. Vahlia, Eremo-
syne and Cephalotus might also be assigned to this otherwise very
homogeneous group.
For the purposes of this discussion-the division of the Saxilragales
in the broadest sense into monophyletic taxa-it is relatively unimpor-
tant whether one calls this grouping an order or a superfamily. For
clarity I will henceforth refer to this group as the Crassulales.
Within this order Crassulales one can distinguish between several
diverging evolutionary pathways. They can be differentiated not only
by the presence of characteristics unique to each line but also by the
lack of certain characters which are common to the other lines.
The seeds of Vahlia, depicted in ENGLER & PRANTL and HUTCHIN-
SON (1964) as anatropic, do not show any traces of a raphe when ripe.
This genus is clearly further distinguished by the mode of placentation
-two (or three) large placentas hanging from the apex of the loculus-
and the tenuinucellate ovules. The inclusion of Eremosyne in this
group is uncertain. Many of the essential characters allowing a proper
systematic placing are still unknown. I have never seen ripe seeds
of Cephalotus. According to the present descriptions, this genus ob-
viously is no less independent than Vahlia.
It is also possible to separate two small families, the Astilbaceae and
the Peltiphyllaceae from the proper core of the order. The independence
of these two families seems to have come about through an early depar-
ture from the main evolutionary lines.
The great similarity of traits exhibited by each of these families 2
might of course be due to the small number of genera and species in
eaeh, and their consequently lower variability.
The members of both these families still live in the (postulated)
primary habitat (mesophytic forests) of the ancestors of the whole order
Crassulales. Thus their development must have been separate from the

1 The number of the parts between 2 and 20 is due to reduction and,

also, to secondary multiplication.
2 A further, more accurate investigation would, in my opinion, settle
the question of whether Peltiphyllum has one or two integuments. In ripe
seeds I have seen four clearly distinguishable cell-layers. In my opinion
this is of greater importance than finding (or overlooking) a cuticule be·
tween them. It is generally less useful to discuss the number of integuments
than to give the number of cell.layers in the integuments, a procedure I
consider to yield more information.
144 J. E. KRACH:

evolutionary mainstream of this group, which adapted soon to xeric or

montane-alpine biotopes.
A closely related group of genera (united by KLOPFER (1973) into the
subfamily Heucheroideae) also remained mainly in mesophytic biotopes.
They are reminiscent of the Astilbaceae and Peltiphyllaceae in their
biochemistry and morphology (stipules, pinnate laminas).
If one only looks a little deeper than the gross-morphological details,
there emerge webs of interrelationships between characters within the
taxa concerned that tempt one to conclude that the real Saxilragaceae
{corresponding to the Saxilrageae of ENGLER) and the Crassulaceae had
a very long common development. I think this hypothesis is better
than the two other possibilities (at the usual risk of overestimating the
worth of the own findings) :
1. a multitude of parallel developments of analogous Rtructures in
the Saxilragaceae and the Crassulaceae,
2. repeated turnoffs from the evolutionary mainstream which lead
to the Crassulaceae of today by genera we now ascribe to the Saxi-
Iragaceae.
The contrary-a separate derivation of each of the several subfamilies
or tribes of the Crassulaceae-is not a possibility at all. According to my
knowledge nobody has yet furnished convincing proof that there are
subfamilies or tribes within the Crassulaceae which are separated from
each other by clear and mutually exclusive sets of characters.
Cross-connections of the type which exist between nearly all genera of
the Crassulaceae (as shown by agreements in single, or sets of, characters)
extend to genera of the Saxilrageae also. The development of the several
evolutionary lines is, in my opinion, best described as a radiation out-
ward from near one point (somewhat like the rays of an umbel).
To combine Crassulaceae _and Saxilragaceae into one single family
perhaps carries the process of unification too far. However, I am sure
it is closer to the truth (real relationships) than to assign these two taxa
to two completely different lines of evolution, and therefore to two
separate orders (Rosales for the Crassulaceae; Hamamelidales for the
Saxilragaceae), as for example GUNDERSEN (1950) did in his modification
of the BESSEY -SKOTTSBERG system.
A far more difficult task than the tracing of the evolutionary hran-
chings-in principle already long-known-of the Crassulales as accepted
here is the identification of a common ancestor of this taxon and any
possibly related taxa.
HUTCHINSON'S subdivision of the dicots into Herbaceae and Lignosae
is certainly interesting but has gained no general acceptance. According
t,o students of systematics and taxonomy since BAILLON, there remain
only a Rmall and unimportant number of herbaceous species related to the
Seed Characters in and Affinities Among the Saxifragineae 145

Crussulales 1 • And of this initially small number, some genera have sub-
sequently proved to be not saxifragaceous at all, but to belong instead to
several other orders or even subclasses. Therefore it seems necessary to
search next·related taxa among the woody plants, particularly because
of the rudimentary secondary thickening present here and there in several
groups of the Crassulales.

Because, sometime after 1872, ENGLER included Rib((s into his Saxi-
/rageae, this genus is generally considered to be the next-of-kin to the
Grassulales. Only in 1973 did TAKHTAJAN explicitly point to the closer
relationship of Escalloniaceae to the Grassulales. Seed-anatomy can
only provide very few indications for such a relationship: thus a unifica-
tion of Grassulales and Escalloniaceae into a single order has to be re-
garded as very problematical, at least, particularly in view of the many
divergent characteristics and tendencies.
If one attributes only the correctly placed genera to the Escal-
loniaceae 2 one finds that they can be characterized by the following
characteristics:
lamina entire;
no stipules;
formation of an intrastaminal disc;
presence of hamamelose, accumulation of aluminium;
fruit a septicidal capsule (berries are very rare) composed of between
five to two, more or less fused carpels;
transition from hypogynous through perigynous to epigynous flowers;
placentas axial or lateral with numerous seeds;
tenuinucellate;
tendency towards forming a single double-layered integument;
tendency to reduction of the embryo-size in ripe seeds;
tendency towards lengthening of the walls of the outer layer of the
seed coat, which is formed according to the Erica-type.
The seven genera of the Escalloniaceae can be easily separated into
two tribes: the round-seeded Guttsieae (Guttsia, Abrophyllum, Argo-
phyllum) and the Escallonieae (Quintinia, Forgesia, Escallonia, Valdivia)
with scobiform seeds.
It is possible to expand the family Escalloniaceae by adding Anop-
terus in a subfamily of its own (reduction of the number of the carpels
to two, fused to a unilocular perigynous ovary, seeds with large wings,

1 Unfortunately the fairy tale of a great similarity between "he

Spiraeoideae and the Astilbaceae is still perpetuated, and is used as a basis
for establishing a much closer taxonomic relationship between the rather
distantly related Rosales and Crassulales than is in fact justified.
2 In some current systems this family has been turned into a catch·all
for free.floating taxa with no precise taxonomic attachments.
Plant Syst. Evol., Suppl. 1 10
146 J. E. KRACH:

presence of a third layer in the seed coat, extreme reduction of the seed-
core, minute embryo).
Theoretically one could add a further subfamily, the Oarpodetoideae
(including Oarpodetus and Argyrocalymna), but these genera more
probably belong to the Ericales as evidenced by pollen tetrads and
hypertrophic cuticule of the nucellus. The round seeds, anatomically
somewhat similar to those of the Outtsiae, and the reserve cellulose
indicate a special position for these two genera among the Ericales, too.
The assembly of characters which typify the Escalloniaceae, and
which are regarded, not only by myself, as progressive, also fully char-
acterize another family, the Hydrangeaceae, from the Englerean Saxi-
jragineae. Beneath the cloak of the Saxijragales s.l. these two families
quite certainly form a genuinely monophyletic taxon, to be referred to
here as Escalloniales. Within the Escalloniaceae the Escallonieae have
achieved the developmental stratum which is already broadly distributed
among the Hydrangeaceae. Naturally, the Escalloniaceae and the Hydran-
geacae are characterized by their own phylogenetic trends.

H ydrangeaceae Escalloniaceae

the number of cell-layers in the integument only one two-layered

varies according to genus (and investigator); integument
normally only 2 cell-layers remain traceable (except A 'lJOpteru8)
in the ripe seeds
petals free tendency to sympetaly
tendency to reduction of the number of the many seeds per placenta
seeds
endosperm cellular endosperm nuclear
tendency towards formation of palmately veined leaves pinnately veined
leaves
stamens 5 to many stamens (4-) 5 (-6)

The uniform and well-definable family Hydrangeaceae can, in my

opinion, be divided into the two subfamilies centred on Philadelphus
and Hydrangea respectively only if one is willing to restrict oneself to
the use of one single characteristic. To do so ignores important evo-
lutionary ties. Thus I cannot support, e.g., HUTCHINSON'S splitting of
the Hydrangeaceae into two families, but I consider the basic place he
gave the Oarpentiereae and their connection by Oarpentiera to the
Philadelpheae correct. Anyhow the division into tribes by HUTCHINSON
seems to be more natural than that of ENGLER from the view-point of
seed-anatomy. On the basis of seed anatomy-and therefore in need of
Seed Characters in and Affinities Among the Saxifragineae 147

further support-a subdivision of the Hydrangeaceae is easily estab-

lished into five tribes: Carpentiereae, Decumarieae, Philadelpheae,
Hydrangeae, and Dichroeae.
Kirengeshoma proves not to be a member of this family. I even
wonder whether this monotypic japonic genus is a member of the
Saxi/ragales in a very broad sense at all. Seed anatomical investigation
provides no support for leaving this, often cultivated, garden plant in its
attributed place in the system, where it has been an oddity also because
of the inconsistency of its appearance with the rest of the group.
In this connection I should like to point to the fact that uncritical
acceptance of results and findings in the literature, especially regarding
the morphology of Hydrangeaceae, may lead to false conclusions. Even
the drawings in ENGLER & PRANTL are not always as precise as could be
desired, as BUXBAUM (1951) already has explicitly mentioned. These
inaccuracies cannot always be attributed to the rather woodcut-like
printing methods used in those texts.
Instead of the supposition that there is a closer kinship between
Escalloniaceae and Hydrangeaeceae, one might be inclined to think of
them as two evolutionary streams which developed parallel to one another
in the southern and northern hemispheres respectively. But even then
the development had to begin with woody plants, which bore a large
number of crassinucellate, double-integumented seeds in hypogynous
capsules fused from 5 carpels with axial placentas. Besides containing
a medium-sized embryo, the ripe seeds also had plenty of endospermous
-but no nucellar-tissue. Thus they were also likely to be the ances-
tors of the Crassulales.
Naturally, bearers of those characters are widespread throughout the
plant kingdom. Thus the possession, or the possible derivation from
possessors of those characters, is not necessarily a sign for an actual
relationship to the Crassulales or Escalloniales either. Among those taxa
which have been held to be related with the Escalloniales, Ribes and the
Cunoniaceae share a considerable amount of these characters.
The Pittosporaceae, whose relationships might be found also
within this group, have characteristics intermediate between the
primitive ones mentioned above and the progressive ones. The theore-
tical possibility that this family had shared part of an evolutionary
pathway with the Escalloniaceae, which are also inhabitants of the
southern hemisphere, cannot be fully denied.
Despite having the basic characters already described, the genus
Ribes is still rather isolated within the Rosi/lorae, not only by the
morphology of its seeds, but also by a host of other differing characters.
The graving suspicion I first got from this, that is that Ribes might
not belong to this group at all, is based on the fact that many similarities
10"
148 J. E. KRACH:

consist of rather primitive characteristics, which are broadly distributed

within the Rosidae-Dilleniidae, at least among the more primitive taxa.
Meanwhile, however, a number of progressive characteristics that
Ribf'-s shares with other members of the Englerean Saxi/ragaceae have
come to light, making the probability of a parallel development, con-
eeiveable for a single character, in this case rather small. Widespread
progressions are: the palmate venation of the laminae, the lack of
stipules, the sometimes monosexual flowers, or the development of a
berry. Even the shape of the receptacle in Ribes, which ENGLER used
as a basis to unite this genus with the Saxi/rageae, might be found
among other taxa of the Saxi/ragales in a broad sense with further
careful investigation. This hypanthium is otherwise not occurring very
widespread, if one excludes the Onagraceae-Myrtales which are certainly
only very distantly related.
The small subsidiary cells, which also occur in Escalloniaceae (which
genus?) and Cunoniaceae can certainly not be called a primitive character.
The indumentum, composed either of long villi with multicellular feet and
rounded heads, or-even further developed-composed of hairs with
feet recessed in epidermal depressions and with shield- or bowl-shaped
heads as occur similarly in Escalloniaceae and the as yet poorly known
Pterostemon, is an equally progressive character. The dwarfish form of
the embryo in comparison to the rather large seeds and the massive
endosperm, and even the occurrence of rests of nucellar tissue in the
ripe seeds, can be ascribed to neoteny.
There is still no consensus as to whether the crystal-layer, so signi-
ficant for Ribes, develops from the external epidermis of the inner
integument or from the internal epidermis of the outer integument.
The answer to this question may be vital to clarify the relationship of
Ribes to the Cunoniaceae. The development of the inner epidermis of
the inner integument into a tannin-layer, however, is a rather widely
distributed feature within the Rosi/lorae (and of course in other sub-
classes also).
Independent progressions which do not necessarily appear in all
subgenera are the development of spines (from epidermal tissue as in
Rosa!) and of glands (which are sometimes even secretory) on leaves,
stems and fruits. Ribes also demonstrates notably unique developments
in its seed structures: The development of an aril-like tissue from
the funiculus occurs nowhere else among the Rosi/lorae as far as I know,
but the name "aril" is unfortunately used for several tissues of
heterogeneous origin.
The many-layered outer integument may still be an archaic
character. It is equally possible that the development of the highly
specialized sarcotesta could be attributed to a secondary proliferation
Seed Characters in and Affinities Among the Saxitragineae 149

of the cell-layers. The occurrence of still functional chloroplasts even in

cells of the ripe seed-coat, can hardly be designated as a new develop-
ment, but at best as an example of partial neoteny. I am not quite
certain, whether the storage of reserve-cellulose in the cell-walls of the
endosperm can be similarly explained. This storage of reserve-cellulose
occurs in similar structures within the M ontiniaceae as well, which
exhibit comparable structures also in other areas of seed development.
A convincing proof of a closer relationship between Ribes and the
Montiniaceae however, is still lacking.
Further, very independent and progressive characteristic~ are found
in the embryogeny which one as yet has not been able to place
satisfactorily in any standard category.

I cannot agree with W. H. CAMP (in GUNDERSEN 1950) who states that
Ribes has been derived from the Escalloniaceae as this view is based only on
superficial, gross·morphological similarities. Not only are the two families
Escalloniaceae and Grossulariaceae too dissimilar, but the Escalloniaceae have
also achieved a much higher degree of development in many important
aspects. The thought that Ribes developed at the time of the uplifting
of the Andes as a specialized, low·temperature adapted mountain shrub
which then crossed to North America via the Andean bridge and after-
wards spread throughout the holarctis is, nonetheless, interesting. The
occurrence in South America of only the two subgenera Parilla and Berisia,
which are, according to JANCZEWSKI (1913), highly derived, speaks against
CAMP'S theory, which also fails to explain the occurrence of one single
member of Parilla (P. sardourn) in another continenti. The development
of Escallonia-like (or, as I would rather have it, Fuchsia·like) and, more·
over, unisexual flowers in these two subgenera can be easily explained as
a convergent development (adaptive evolution with the same pollinators:
hummingbirds and Sphingidae).

Besides the far less striking correspondences with the Escalloniaceae,

Cunoniaceae, and, as has now been emphasized, with the M ontiniaceae
too, Ribes exhibits an actual relationship concerning seed structure only
to one genus within the Rosiflorae. This is the small (only 4 species)
genus Phyllonoma, which is restricted to the Andes between Mexico and
Bolivia. The inflorescences of Phyllonoma, which are reminiscent of those
of Helwingia, project from the middle or almost from the top of their
leaves.
The seeds of the species (P. ruscifolia) , which I have investigated,
completely lack the crystal-layer so significant for Ribes, but this

1 I suppose that the classification of Ribes sardo'urn is also indication

of the deeply interwoven relationships of characteristics within the one
and single genus Ribes. Due to this web, any subdivision of this genus
can be based only on the use of a very small number of characteristics,
rather than by regarding different character-matrices simultaneously.
150 J. E. KRACH:

evolutionary step-if it is one-is anticipated by the occurrence of

"windows" in the crystal-layer even in some of the investigated seeds
of Ribes uva-crispa (not only the cultivars), and Ribes diacanthum.
Both these genera share-in addition to characteristics of the flower
and the fruitl-the following traits:
seed-coat composed of a multi-layered outer and a two- (or three-?)
layered inner integument;
outer epidermis of the seeds fully packed with mucilage (myxo-
testa!) ;
continual reduction in cell-volume of the outer integument prece-
ding from the outside inwards;
metamorphosis of the cells of the epidermis of the inner integument
to a thick tannin-layer which retains its cellular structure;
storage of reserve-cellulose in unpitted cell-walls along with fats and
aleurone;
tendency towards development of seeds with sma1reIDbryos.
Except for the first and fourth, these characteristics have to be
viewed as a common progression away from the original characteristics
of the ancestral taxon. I am quite certain, that further investigations
will reveal further common traits between Ribes and Phyllonoma.
Therefore I find it proper to unite these two genera-and only these
two-into the order Grossulariales. Both these genera command such a
great number of independent characters, however, that they are better
each treated as separate monogeneric families.
Potentially, the Cunoniaceae possess a seed-coat which has far more
layers than that of Ribes. This is demonstrated by the seeds of Schizo-
phragma ilicina, which are merely four millimeters long, but nonetheless
sit inside a stone which is of the size of a peach pit and equally ruminated.
Is it conceivable that there is a selective mechanism which in this case
could induce such an already well-protected seed to increase the number
of layers in its seed-coat?
Thus, for the present, this wealth of layers (approximately 25 cell-
layers organized into seven distinct strata) is regarded as a primordial
characteristic. Such multilayered seed-coats, however, occur only in
species in which the seed itseH is not the diaspore. By means of gradual
reduction, the seed structure of all the other Cunoniaceae in a broad
sense (including Bauera and BruneUia = hereafter referred to as Cuno-
niales) can be derived from the Schizophragma model.
ENGLER'S classification of the Cunoniaceae into tribes is partially
based on insufficient knowledge of the individual members. HUTCHINSON

1 Phyllonoma had not yet been investigated for other characters. The
recent results of MORI and KALLUNKI could no more be included.
Seed Characters in and Affinities Among the Saxifragineae 151

completely dispensed with a subdivision, as he often did when uniting a

decidedly incongruent conglol!lerate into a family.
By considering the different degrees of reduction of seed-coat struc-
ture (sometimes combined with transformations in single tissues) we
can differentiate between the following groups: Schizophragma; Brunel-
lia; A istopetalum, Bauera; Ounoniaceae S.S.l; Oeratopetalum. The Ouno-
niaceae s.s. could be further divided into three groups2, which partially
correspond with the tribes of ENGLER. By this grouping I deliberately
avoid defining the single groups as (sub)tribes or (sub)families.
Through a further such reduction one can arrive at the seeds of
Francoa, although I am not at all certain whether this derivation
is entirely hypothetical or not. Yet I cannot deny that it is possible to
link this genus to the Ounoniales, perhaps as a suborder of its own, but
sufficient positive evidence is still lacking.

It, is far less difficult to seek out the relatively clear-cut evolutionary
trends in the development of the seeds of the Ounoniales than to correlate
these structures with those of the fruits, flowers and inflorescences. Al-
though the seed of Schizophragma within its stone retains primitive characters,
I do not believe that the one-seeded drupe of this genus should be regarded
in any way as primordial to the Ounoniales as a whole. Dehiscent capsules
composed of five to two locules containing many seeds are more likely to
be primitive. One can regard as progressive the reduction in the number
of the seeds partially associated with a reduction in size of the flowers
and fruits, and their concentration into a capitulum. Another trend is
towards the production of indehiscent fruits, gradually gaining size, which
were initially dry (Aistopetalum) but then became berries (Opocunonia)
or drupes (Schizomeria).

Based on the previous results it is possible to conclude that from a

hypothetical prototype (the so-called prosaxifrageous basis) several
main-lines of evolution yielded four currently tangible orders: Oras-
sulales; Escalloniales; Grossulariales; Ounoniales. These mainstreams
are disparate in the number of their genera and of course in the number
of their species. Any attempt to amalgamate these four taxa as sub-
orders 3 into one order called Saxilragales 4 is hindered by the suspicion
that not these four taxa only evolved from this common ancestor. If
one were to imagine an organism from which Orassulales, Escalloniaks,
Grossulariales, and Ounoniales might all have evolved, one can, without

1 Including PoUingeria.
2 I did not get. ripe seeds from ENGLER'S tribe IV or V.
3 Despite t.his I would not object a lowering in rank of each of these
categories.
4 Moreover I do not like the rather vague term Saxifragales which,
chameleon-like, means something different to every user.
152 J. E. KRACH:

excessive mental gymnastics, imagine that a majority of the taxa of the

Rosidae could easily have had that organi~m as an ancestor as well.
From this prototype, so far defined by seed· and flower·charac·
teristics and a few biochemical traits, one can also conceivably derive
several certainly distant taxa, if one stipulates the same number of
progressions, although not in the same directions. Thus, the doubts
whether the four orders treated really had a common ancestor sub·
sequent to the basic branching of the Rosidae or even earlier could
perhaps be diminished though surely not obviated.
By this I certainly do not mean to imply that all that which has,
up to now, comfortably hidden under the blanket name of Saxi/ragales,
should remain there. Most of the current systems include under the
category of Saxi/ragales, or even Rosales, one or another family-not
to mention the mostly not· mentioned genera-which have absolutely
nothing to do with these taxa. There are other families whose affiliation
to these taxa appears to be questionable, because the investigation of
other complexes of traits besides the gross· morphology of leaf and flower
is still lacking. Thus any broadly supported conclusions are premature.
Perhaps in this case it is best to take to heart a command I learnt from
H. HUBER: It is far better to admit to an incomplete system of (at least
sufficiently) completely known plants than to boast of a complete
system of incompletely known plants.

I am very greatly obliged to WILLIAM COLMER;;, M.A. for his patient

help with the English translation.

References
(For further literature see KRACH 1976)
BUXBAUM, F., 1951: Erneuerung der Systematik der h6heren Pflanzen.
Wien: Springer.
ENGLER, A., und PRANTL, H., eds., 1930: Nat. Pflanzenfamilien, 2. Auf!.,
Bd. 18 a: Saxifragineae. Leipzig: W. Engelmann.
GIBBS, R. D., 1974: Chemotaxonomy of flowering plants. Vo!' I and IV.
Montreal and London: McGill·Queen's Univ. Press.
GUNDERSEN, A., 1950: Families of Dicotyledons. Waltham, Mass: Chronica
BotanicaCo.
HIDEUX, M. J., and FERGUSON, 1. K., 1976: The stereostructure of the
exine and its evolutionary significance in Saxifragaceae sensu lato.
In: The evolutionary significance of the exine (FERGUSON, 1. K., and
MULLER, J., eds.). Linnean Society Symposium series no. 1.
HOOGLAND, H. G., 1961: Studies in the Cunoniaceae 1. Austr. Journal
Botany 8, 318-341.
HUBER, H., 1963: Die Verwandtschaftsverhaltnisse der Rosifloren. Mitt.
Bot. Staatssamm!. Miinchen 5, 1-48.
HUTCHINSON, J., 1964: The genera of flowering plants. Vo!' 1. Oxford.
- 1967: The genera of flowering plants. Vol. 2. Oxford: Clarendon Press.
Seed Characters in and Affinities Among the Saxifragineae 153

JANCZEWSKI, E. DE, 1913: Monographie des Grosseillers V. Bull. Internat.

Acad. Sc. Cracovie B 10, 714-741.
KLOPFER, K., 1973: Florale Morphogenese und Taxonomie der Saxifra-
gaceae sensu lato. Feddes Rep. 84, 475-516.
KRACH, J. E., 1976: Samenanatomie der Rosifloren 1. Die Samen der
Saxifragaceen. Bot. Jahrb. Syst. 97, 1-60.
MORI, S. A., and KALLUNKI, J. A., 1977: A revision of the genus Phyllonoma
(Gro88ulariaceae). Brittonia 29, 69-84.
TAKHTAJAN, A., 1973: Evolution und Ausbreitung der Blutenpflanzen.
J ena: G. Fischer.

Address of the author: Dr. J. ERNST KRACH, Iustitut fUr Systematische

Botanik, MenzingerstraBe 67, D-8000 Miinchen 19, Federal Republic of
Germany.
Plant Syst. Evo!., Supp!. 1, 155-178 (1977)
© by Springer-Verlag 1977

Lehrstuhl fur Zellenlehre der Universitat Heidelberg,

Federal Republic of Germany

Transmission Electron Microscopy and Systematics

of Flowering Plants
By

H.-Dietmar Behnke, Heidelberg

Abstract: From the large number of TEM investigations of plant cells

only two characters have so far proved useful for the classification of higher
taxa of flowering plants.
1. Dilated cisternae of endoplasmic reticulum in root cap cells and
bundle parenchyma cells are a characteristic feature of members of Cap-
parales and will be discussed with respect to their systematic reliability.
2. The micromorphological differences identified within sieve-element
plastids represent the only character presently available which is generally
applicable to the higher taxa of seed plants: The major distinction be-
tween types of sieve-element plastids is based on the presence (P-type)
or absence (S-type) of protein accumulations, while their micromorphology
and combinations define subtypes of sieve-element plastids. Specific sub-
types characterize the Monocotyledoneae (P II) and the Centrospermae
(P III). Magnoliales/Laurales (P I) and Fabales (P IV) contain distinct
subtypes in some of their families and genera, while others have only S-type.
Aristolochiales, Vitineae, Eucryphiaceae, Gunneraceae contain subtype
P I-plastids.
Classification and delimitation of higher taxa in flowering plants can
be aided by utilizing the different types of sieve-element plastids, as has
been best demonstrated in the Centrospermae but is also applicab\e to
monocotyledons and a number of other taxa. An analysis of investigated
species and projection into the total number of extant angiosperms results
in a distribution of 35% P-type and 65% S-type.
In discussing phylogenetic relationships the plastid data are super-
imposed on CRONQUIST'S, DAHLGREN'S, EHRENDORFER'S and TAKHTAJAN'S
angiosperm systems. Finally, arguments for the P-type being the ancestral
sieve-element plastid .are presented and it is suggested that on multiple
occasions the S-type (by loss of protein) has been derived.

Introduction
During the last two decades transmission electron microscopy
(TEM) aided by ultramicrotome techniques has proven a powerful
tool in plant morphology. Ultrastructural details increased our know-
ledge of cell organisation and stimulated research in various fields,
156 H.·D. BEHNKE:

including, for example, physiology, biochemistry, and genetics. More-

over, these data greatly improved our understanding of the way cells
function. Despite the large number of TEM investigations limited
information is relevant for systematics; in addition, most of the data
are species-specific and not applicable to higher taxonomic categories.
With regard to seed plants a recent survey on "Electron Microscopy
and Plant Systematics" (COLE & BEHNKE 1975) lists only a few characters
which are useful in the classification of families and higher taxa and
most of these were determined with the scanning electron microscope
(SEM).
Comparative pollen morphology includes by far the most electron
microscopic investigations concerned with plant systematics. The
ultrastructure of the pollen wall has been studied since the introduction
of TEM into botany, but its systematic implications are interpretable
only in combination with results derived from light microscopy (LM),
SEM, and TEM studies, and therefore are excluded from the present
discussions. For a comprehensive survey of palynological research
concerned with the phylogeny of flowering plants see WALKER & DOYLE
(1975).
Thus, at the present time there are only two TEM-determined
characters available, both recently developed, which fit the require-
ments of our survey: they are applicable to higher categories of flowering
plants and are entirely based on TEM methods: 1. Specific dilated
cisternae of endosplasmic reticulum occurring in a variety of ccll types
in at least two families of Capparales, and 2. specific plastids present
as two main types in the sieve elements of the food-conducting phloem
tissue in all taxa of flowering plants (as well as gymnosperms).
Both these characters will be considered in detail and their reliability
as systematic markers tested against several current systems for flower-
ing plants.

Material and TEM Methods

TEM studies using the ultrathin sectioning techniques require fresh,
living material for the satisfactory preservation of cytoplasmic organelles.
For comparative investigations, material from different sources is
best fixed in aldehydes followed by osmic acid, dehydrated by acetone
and embedded in epoxy resins, all by standard procedures (e.g. for sieve-
element plastids described by BEHNKE 1975a). Small pieces (about 1-5 mm 3 )
of the plant material which is penetrated and surrounded by polymerized
epoxy resins, after recutting into yet smaller parts (0.1 to 0.2 mm side
length) containing the appropriate tissue (e.g. phloem with sieve elements),
are sliced into 50 to 100 nm (= rmn- 6 ) ultrathin sections. These when
viewed with a TEM, in general, give pictures of dissected organelles, not
always showing all their specific structures or inclusions. For example,
in the investigations of sieve-element plastids (the diameter of which is
Transmission Electron Microscopy and Systematics of Flowering Plants 157

about 2 f1m) 20 to 40 contiguous serial ultrathin sections need to be screened

in order to be aware of all their contents. Complete series, however, are
difficult to cut in routine preparations and are not needed in most samples,
such as those containing either S-type or P II-P IV subtype plastids (see
Fig. 25). But if those P I-forms, which contain only tiny protein crystalloids
(e.g. in Laurus and Magnolia), are under investigation, complete series
would prove helpful, unless the protein is detected early among the single
sections.
Since fresh material is a prerequisite for TEM studies, it is often difficult
to obtain the necessary samples for a systematic survey on the distribution
of a particular TEM-determined character. Although some plant collections
from remote areas such as Australia, New Zealand, and Africa which were
sent by Air Mail did survive, in general, our investigations have utilized
the limited taxa grown in European Botanical Gardens. The difficulties
in obtaining fresh material from Oceania, South America and Africa leave
incomplete the investigation of many interesting groups of Magnoliidae
as well as some of the Oaryophyllidae (e.g. Barbeuiaceae, Microtea, Mol-
luginaceae) and others.
Compared to the problem of obtaining fresh plant material all the other
problems associated with the material are insignificant:
Like many other biologists not basically trained in taxonomy I often
must rely on the specification of Botanical Gardens, which are not always
correct (Myristica, in my experience, is a good example: I examined four
collections before I obtained one which I now believe is Myristica frag-
rans l ). There are, of course, errors in collecting species (see, for example,
BEHNKE 1976, p. 34 concerning Phytolacca acinosa).

TEM-Characters in Higher Taxa of Flowering Plants

1. Dilated Cisternae
;Dilated sections of the endoplasmic reticulum with a filamentous
content in root cells of Raphanus sativus L. were first described and
termed "dilated cisternae" (DC) in 1965 by BONNETT & NEWCOMB.
Similar structures containing a tubular protein were found in the phloem
parenchyma of leaf veins of Brassica chinensis L. by FAVALI & GEROLA
(1968). IVERSEN (1970b) initiated the first systematic overlook on the
occurrence of DC in Brassicaceae; JORGENSEN et al. (in prep.) detected
DC in Capparis cynophallophora L.
A compilation of all data concerning DC (cf. Table 1) suggests
that these organelles are typical of Brussicaceae and Capparaceae.
Their internal structures (protein filaments or protein tubules) appear
to depend on the location of the cells inspected: While in root cap and
root epidermal cells the DC content is filamentous, in bundle parenchyma

1 The investigated M. dactyloides GAERTN. (Bot. Garten, Miinchen)

and M. fragrans ROUTT. (Garten der Landw. Botanik, Bonn) also differ
from the dubious Myristica species (which were Annona) by the presence
of "extruded nucleoli" in their sieve elements.
158 H.-D. BEHNKE:

cells of leaves and stem the DC contain protein tubules. The latter are
closely-packed and parallel to the long axis of the cisternae, which at
the same time determine their length, while their diameter is about
25 nm (Figs. 1-3).
IVERSEN (1970a) performed a histochemical test for myrosinase
and discussed the localization of this glucosinolate-hydrolyzing enzyme
in the DC. Since, however, DC were not found in other glucosinolate
containing families like Resedaceae (IVERSEN 1970b), Bataceae, Gyro-
stemonaceae, Salvadoraceae (J0RGENSEN et al. in prep.), their involve-
ment in the hydrolysis of glucosinolates is not cogent. At least the
cited families can do well without DC. So far, DC represent a micro-
morphological character for Brassicaceae and Oapparaceae, only.

Table 1. Genera of Capparales which contain DC.

Rrassicaceae
Allyssoide8
Alyssum } IVERSEN (1970b)
Arabis
Armoracia JORGENSEN et al. (in prep.)
Brassica FA VALl & GEROLA (1968), IVERSEN (1970b)
Cardamine
Cardaminop8i.~

1
IVERSEN (1970b)
Cochlearia
De8curainia
Diplotaxi8 CRESTI et al. (1974)
Draba IVERSEN (1970b)
Eruca CRESTI et al. (1974)
Iberis IVERSEN (1970b)
Isatis BEHNKE & ESCHLBECK (unpubl.)
Lepidium IVERSEN & FLOOD (1969)
L-unaria BEHNKE & ESCHLBECK (unpubl.)
Raphanus BONNETT & NEWCOMB (1965)
Sinapi8 IVERSEN (1970b), HAVE LANGE & COURTOY (1974)
Si8ymbrium IVERSEN (1970b)
Thlaspi IVERSEN (1970b), HOEFERT (1975)

Capparaceae
Cappari8 .10RGENSEN et al. (in prep.)
Cleome BEHNKE (unpubJ.)

l<'igs. 1-3. Dilated cisternae (DC) in Capparale8. Figs. 1 and 2. DC in

bundle parenchyma cells of Raphanus 8ativu8, Bras8icaceae (hypocotyledons).
Inside the DC are parallel arranged protein tubules (0 about 300 nm;
length: same as DC). 1: x 10,000; 2: x 30,000. Fig. 3. DC in phloem-
parenchyma cells of Cappari8 cynophallophora, Capparaceae (stem).
Contents of the DC not dissolved into single protein tubules. x 10,000.
ER = endoplasmic reticulum, SE = sieve element; marker = 1 fJ.m
Transmission Electron Microscopy and Systematics of Flowering Plants 159

Figs. 1-3
160 H.-D. BEHNKE: Transmission Electron Microscopy

2. Sieve-Element Plastids
Only a decade has passed since sieve-element plastids were first
demonstrated to incorporate ultrastructural details which can be
used to characterize different plant taxa. The Dio8coreaceae, in many
morphological and anatomical respects uncommon among monocoty-
ledons, was the first family reported to contain sieve-element plastids
whose micromorphology differed from the then known pattern (BEHNKE
1965, 1967). A subsequent test of 21 families including representatives
of nearly all respective orders of M onocotyledoneae revealed sieve-
element plastids with the same characteristics (BEHNKE 1969a). Con-
tinued research on all dicotyledon subclasses stepwise (follow e.g.
BEHNKE 1969b, 1972, 1975a, 1975b) led to the present knowledge
on the distribution of plastid types and subtypes among the flowering
plants, based now on 690 species from about 190 families!.

Types, Subtypes, and Forms of Sieve-Element Plastids

The ultrastructural features of sieve-element plastids which are of
systematic value are associated with accumulations of protein and starch.
Starch is present as grains differing in number, size and shape
(Figs. 4-6). Single or even all grains of a sieve-element plastid may
consist of many darkly stained, small particles (Fig. 6) thus reflecting
their glycogene-like chemical composition, while others appear to be
solid and evenly formed spherules.
Protein is accumulated as crystalloids and filaments, and the amount,
shape and arrangement of both may differ among sieve-element plastids
of different taxa.
The protein accumulations are used as the main discriminating
features in the classification of types, subtypes, and forms of sieve-
element plastids:
Types. Presence of a protein accumulation defines the P-type
plastids, whereas a b sen c e of any protein accumulation defines the
S-type plastids.
In this definition, the starch accumulation is of no primary impor-
I As of TAKHTAJAN (1973).

Fig. 4. Longitudinal section through stem phloem of Lindera praecox,

Lauraceae: Sieve elements (SE) with many plastids. x 6,000
Figs. 5-7. S-type sieve-element plastids. Fig. 5. Stephania glabra, Meni-
spermaceae: starch grains (S) of different size. X 20,000. Fig. 6. Byblis
gigantea, Byblidaceae: starch grains (S) composed of globular particles
with glycogen-like appearance. X 20,000. Fig. 7. So-subtype plastids
without any starch grains in Celtis orientalis, Moraceae. x 20,000.
M = mitochondrium, PC ~ phloem-parenchyma cell, marker = 1 fJ.m
Figs. 4-7
162 H.·D. BEHNKE:

tance for the classification of the plastids since it may be· present or
absent in both types. If in S-type plastids starch is absent (see e.g.
Fig. 7 and BEHNKE 1973) the plastids are classified as belonging to
subtype So.
Subtypes. For the division of P-type sieve-element plastids into
subtypes some conspicuous protein accumulations and their combina-

FORMS OF P-TYPE PLASTIDS

Crystalloids po ly gon al cun~a le

''' 'OJ cryst.IIDldsJ

~~
PIa PIc

pmb pma pmc

Plb PI" P Ua

PUb . .
, \~ .ddrt.... lh..,cryst
compoot<l.r ",bonus
,.\ d.If.,,,\ '~'''' '

Fig. 8

tions, as typically present in some taxa, were chosen: whereas sub-

type P II which is thought to be primitive in flowering plants basically
incorporated single crystalloids of different sizes and shapes and/or
irregularly arranged filaments, subtypes P II-P IV circumscribe mor-
phologically very distinct and probably more advanced P-type plastids.
Characteristic features are: many cuneate crystalloids in P II (as
of all Monocotyledons), one ring-shaped bundle of filaments in PIlI
(as of all Centrospermae) , many, prevalent polygonal crystalloids in
P IV (as of 60% of Fabales).
Forms. Most of the subtypes can be subdivided into different forms
of P-type sieve-element plastids which are typical of some families or
1 For clarification· a consecutive numbering system of all subtypes
is introduced; however, this simplified system does not render the previous
detailed formulas (BEHNKE 1975a) invalid.
Transmission Electron Microscopy and Systematics of Flowering Plants 163

family groups within the respective higher taxon. Forms are indicated
by lower case letters. Out of the number of possible forms resulting
from a free combination of crystalloids and filaments, Fig. 8, in a
two-dimensional alignment, lists those realized (cf. also Fig. 25). Starch
was found to be present at least in some species belonging to forms
P la-c, P IIa, Pilla, P IV.

The Distribution of P- and S-type Sieve-Element Plastids. Their

Subtypes and Forms Among Flowering Plants
P-Type. A brief account for all families of flowering plants which con-
tain P-type sieve-element plastids, also mentioning the appropriate
subtypes and forms, is given in Table 2. A description of the subtypes
and their main representatives follows here:
P I-Subtype. The P I-subtype is held the most primitive among
the P-subtypes of flowering plants. Of all subtypes P I shows the

Table 2. Families of ,flowering plants which contain P·type sieve-element

plastids

1. Dicotyledons
Atherospermataceae Pia Gunneraceae P Ib
Achatocarpaceae Pilla Gyrocarpaceae Plb
Agdestidaceae Pilla H alophytaceae Pilla
Aizoaceae Pilla H ectorellaceae Pilla
Amaranthaceae P IIlc Hernandiaceae Plb
Annonaceae Pia Lauraceae* Plb
A ristolochiaceae P laJP lIa Leeaceae P Ib
Basellaceae Pilla M agnoliaceae * Plb
BrasBicaceae * PIb Mimosaceae PIV
Oactaceae Pilla M olluginaceae P IlIaJb
Oaesalpiniaceae PIV M onimiaceae P laJb
Oalycanthaceae Pia N yctaginaceae PIlla
Oanellaceae Pic Petiveriaceae Pilla
Oapparaceae * P Ib Phytolaccaceae Pilla
Oaryophyllaceae Plllb Portulacaceae Pilla
Ohenopodiaceae P IIlc Putranjivaceae Plb
Didiereaceae Pilla Rhabdodendraceae Plb
Dysphaniaceae P Illc Stegnospermaceae P IIlb
Eucryphiaceae P Ib Tetragoniaceae Pilla
Eupomatiaceae P Ib Ulmaceae * Plb
Fabaceae* PIV Vitaceae P Ib
Gisekiaceae Pilla

2. Monocotyledons P II,
all of the families (36, as of TAKHTAJAN 1973) so far investigated, with
Poaceae belonging to form P lIb.
* Families which contain P-type and S-type genera.
11·
164 H.-D. BEHNKE: Transmission Electron Microscopy

highest variability within the limits classified earlier: e.g., the sizes
and shapes of the single crystalloids within forms PIa and P Ib show
considerable variation among the taxa to be concerned.
The primary distribution of P I is within Magnoliales, Laurales
and Aristolochiales.
Roughly two thirds of the so far investigated families of Magnoliales
and Laurales contain P I sieve-element plastids: Magnoliaceae (Fig. ta),
Eupomatiaceae, Annonaceae, Oanellaceae (Fig. 12), Monimiaceae (Fig. 10),
Atherospermataceae, Hernandiaceae, Gyrocarpaceae, Oalycanthaceae, Lau-
raceae (Figs. 14, 15) (see Table 2 for distribution of forms among these
families); while the other contain S-type plastids, only: M yristicaceae 1,
Winteraceae, A ustrobaileyaceae, Trimeniaceae, Ohloranthaceae.
Members of the two families from which the orders Magnoliales
and Laurales derive their names contain both S-type and PI-subtype
species and are considered representative for the plastid distribution
within these orders. The P-type species of their genera Magnolia,
Michelia, Laurus, Lindera, Ocotea, Sassafras, Umbellularia a~l contain
form P Ib plastids. However, their crystalloids tend to be the tiniest
of all of those recorded in P-type plastids. In addition, some crystalloids
are so inconspicuous that they are easily overlooked in a routine
examination (Figs. 13-15). This faint appearance of protein (as of
Ocotea and Sassafras in Figs. 14, 15) is suggested to be an indication
that here, in the M agnoliales and Laurales, is one of the multiple transi-
tions from P-type to S-type plastids (see also discussion on p. 172).
Aristolochiales (and its family Aristolochiaceae) is heterogeneous
with respect to P-subtypes: Aristolochia (Fig.9) has PIa plastids
whereas Asarum forms P II plastids, typical of monocotyledons.
Other higher taxa which contain PI-subtype plastids (form
P Ib) are: 1. Eucryphiaceae, treated either in Rosales (CRONQUIST
1968, THORNE 1968), Saxifragales (TAKHTAJAN 1973) or Ounoniales
(DAHLGREN 1975); 2. Gunneraceae, included in Haloragales (CRON-
QUIST 1968) or Hippuridales (TAKHTAJAN 1973), but raised into the
unigeneric Gunnerales by DAHLGREN (1975); 3. Vitaceae and Leeaceae
1 Formerly recorded as P-type (BEHNKE 1971, 1972), see comments
under "Material and TEM Methods".

Figs. 9-12. P I-subtype sieve-element plastids. Fig. 9. Aristolochia ar-

gyroneura, Aristolochiaceae: form P I a plastid with polygonal crystalloid
(C), irregular filaments (1<') and starch grains (8). X 50,000. Fig. 10. Peumus
boldus, Monimiaceae: form P Ib plastids with polygonal crystalloids (C)
and starch grains (8). X 30,000. Fig. 11.0is8Us antarctica, Vitaceae: form
P Ib plastids. X 30,000. Fig. 12. Oanella alba, Oanellaceae: form PIc
plastids with irregular to ring-like arranged filaments (F) and starch grains
(8). X 30.000. marker = 1 fJ-m
PIa

Figs. 9-12
166 H.-D. BEHNKE: Transmission Electron Microscopy

Figs. 13-15. P I-subtype sieve-element plastids with small and incon-

spicuous crystalloids (arrows): Fig. 13. Magnolia anguatifolia, Magno-
liaceae; Fig. 14. Sassafras albidum, Lauraceae; Fig. 15. Ocotea foetens,
Lauraceae. Figs. 13 and 14 X 20,000; Fig. 15 X 30,000. marker = 1 (.10m

usually together with S-type family Rhamnaceae aligned within Rham-

nales, however, placed into Cornriles by THORNE (1968) and suggested
by JOHNSTON (1973) to be next to Saxitragales.
In addition, a few isolated genera scattered throughout dicotyledons
also belong to the group of plants which form P I b sieve-element
plastids: Ulmus (in Ulmaceae), Rhabdoderuiron (usually in Rutaceae) ,
Brassica (of Brassicaceae), and Capparis (of Capparaceae).
It is remarkable that this P I-subtype, except for perhaps P IV
(see p. 170) is the only subtype which occurs in genera or families which
are systematically treated near S-type containing taxa.
P II~subtype. Cuneate crystalloids, oriented towards the centre of
the plastids, are the characteristics of the P II-subtype (Fig. 16). The
sieve-element plastids of all of the so far investigated members of Mono-
cotyledoneae fall, without exception, within the range of this subtype

Figs. 16-18. P II-subtype sieve-element plastids with cuneate crystalloids

(C): Fig. 16. Vafiilla planifolia, Orchidaceae; X 40,000. Fig. 17. Oostua
montanua, Oostaceae (with additional starch grains, S); X 15,000. Fig. 18. Sac-
charum officinarum, Poaceae {with additional crystalloid composed of sub-
structures with wider spacing than in cunes; see arrow); X 25,000. marker
in Fig. 16 and 17 = 1 (.10m
Figs. 16-18
168 H.·D. BEHNKE:

and there is only one P II-subtype genus which does not belong to the
monocotyledons: Asarum. Some species, or even entire orders (e.g.
Zingiberales) , have starch accumulations in addition to the cuneate
crystalloids of the P II-subtype (Fig. 17).
The family Poaceae besides the cuneate crystalloids in their sieve-
element plastids contains one or more small crystalloids composed
of substructures (Fig. 18) with a wider spacing than in the cunes;
these plastids therefore being defined as formP lib.
P Ill-subtype. The P III-subtype is confined to the Gentrospermae
and in my view can be used for the delimitation of this order (see
MABRY & BEHNKE 1976). A ring-shaped bundle of protein filaments
apposed at the periphery of the plastids is the subtype-characteristic
(Figs. 19-22). The three presently recognized forms of this subtype
differ by the presence or absence and the shape of an additional central
crystalloid: form P IlIa with a globular crystalloid (Figs. 19, 20) in
A chatocarpaceae , A gdestidaceae , A izoaceae, Basellaceae, Gactaceae, Di-
diereaceae, Gisekiaceae, Halophytaceae, Hectorellaceae, Molluginaceae
(in part), Nyctaginaceae, Petiveriaceae, Phytolaccaceae, Portulacaceae,
Tetragoniaceae; form P Illb with a hexagonal crystalloid (Fig. 21) in
Stegnospermaceae, Garyophyllaceae and Limeum (of Molluginaceae);
form P I1Ic (Fig. 22) without any crystalloid in Amaranthaceae, Gheno-
podiaceae and Dysphaniaceae (for detailed description see BEHNKE
1976).
Absence of the P III-subtype has proven to be of considerable
help for excluding from the Gentrospermae the families Bataceae, Gy-
rostemonaceae, Theligonaceae, Vivianiaceae (BEHNKE 1976) all of which
were doubtfully aligned with the Gentrospermae. With a few exceptions
(e.g. Barbeuiaceae, Microtea of Phytolaccaceae, and Molluginaceae)
the plastid investigation of the Gentrospermae with about 160 species
recorded is regarded as representative.
The division of the Gentrospermae into three family-groups as done
by BEHNKE (1976), each classified by a distinct form of PIlI-subtype
plastids, almost exactly corresponds to an earlier proposal by FRIED-
RICH (1956). In his dendrographic figure, based on flower diagrams,
FRIEDRICH (1956: Fig.5) derives the three suborders Portulacineae,
Garyophyllineae, Ghenopodiineae from the basic Phytolaccineae. Except
for the Gyrostemonaceae which he keeps in his family list but separates
in his diagram and for the Achatocarpaceae which are separated in the
diagram, but kept within the Phytolaccineae, FRIEDRICH'S Phytolac-
cineae and Portulacineae fit into the form P III a group of BEHNKE
(1976). It is of special interest to note that FRIEDRICH (1956) derives
Garyophyllineae from Phytolaccineae through Stegnosperma and part
of Molluginaceae, where BEHNKE (1976) found the specific form P I1Ib
Transmission Electron Microscopy and Systematics of Flowering Plants 169

pma

0,5 urn 19

10,5 urn

Figs. 19- 22. P III-subtype sieve-element plastids with ring-like bundle

of filaments (F): Fig. 19. Decaryia madagascariensis, Didiereaceae: form
P III a with globular crystalloid (C); X 80,000. Fig. 20. Gibbaeum angulipes,
Aizoaceae: form P IIIa, but bundle cross-sectioned (F); X 30,000. Fig. 21. Si-
lene cucubalus, Caryophyllaceae: form P IIIb plastid with polygonal crys-
talloid (C); X 60,000. Fig. 22. Alternanthera sessilis, Amaranthaceae: form
P IIIc plastid without crystalloid; X 30,000. marker in Figs. 20, 22 = 1 !-Lm
170 H.·D. BEHNKE:

plastids. A minor disagreement is in placing Paronychiaceae within

Chenopodiineae (FRIEDRICH 1956) instead of lumping them with Caryo-
phyllaceae s. str. because of identical plastid subtypes (BEHNKE 1976).
P IV -subtype. Of all the P-subtypes this is the least detected, pro-
bably because it is the most restricted. P IV-subtype sieve-element
plastids, which contain a few polygonal crystalloids of variable size

Figs. 23 and 24. Sieve-element plastid of Fabales. Fig. 23. P IV-subtype

plastid with several polygonal crystalloids (C) and starch grains (S) in
Astragalus cieer, Fabaceae. Fig. 24. S-type plastid of Hippoerepis comosa,
Fabaceae. x 25,000; marker = 1 !-Lm

(Fig. 23), are found in about 60% of the investigated Fabales, the
others having S-type plastids' (Fig. 24). In Mimosaceae and Caesal-
piniaceae only P IV-subtype plastids were found, however, the total
number of species (35) studied so far, is insufficient to reach a final
decision on the distribution of the S- and P-types.
S-Type • . S-type sieve-element plastids lacking protein deposits
but with starch accumulating in grains of different sizes and forms
(e.g. Figs. 5, 6, 24) are the most common in flowering plants. A rough
calculation based on projections from investigated species indicates
that 65% of all flowering plants have S-type plastids.
Among the subclasses listed by TAKHTAJAN (1973) and their orders
investigated the following are almost without exception S-type taxa:
Ranunculidae: Illiciales, Nelumbonales, Ranunculales, Papaverales,
Sarraceniales.
Transmission Electron Microscopy and Systematics of Flowering Plants 171

Hamamelididae: Trochodendrales, Cercidiphyllales, Eupteleales, Ha-

mamelidales, Eucommiales, Urticales (except for Ulmus), Barbeyales,
Casuarinales, Fagales, Betulales, Myricales, Juglandales, Leitneriales
(for detailed list see BEHNKE 1973).
Dilleniidae: Dilleniales, Paeoniales, Capparales (exceptions see
under PI-subtype), Tamaricales, Salicales, Ericales, Euphorbiales.
Asteridae: Gentianales, Scrophulariales, Asterales.
In the other dicotyledon subclasses the S-type is:
1. Prevailing in Rosidae: Saxi/ragales (except for Eucryphiaceae),
Rosales, Nepenthales, Myrtales, Rutales, Sapindales, Geraniales, Cor-
nales, Celastrales, Oleales (but with priority of P-type in Fabales and
Rhamnales, P-type also in Gunneraceae).
2. Equal with P-type in Magnoliidae: S-type orders: Piperales,
Ral/lesiales, Nymphaeales; some S-type families in: Magnoliales, Lau-
rales.
3. In a minority in Caryophyllidae: Polygonales, Plumbaginales,
and Gyrostemonaceae, Bataceae, Vivianiaceae of Caryophyllales (sensu
TAKHTAJAN).
So-subtype. Sieve-element plastids which show neither starch nor
protein accumulations are recorded as So-subtype (see definition on
p. 162); these include all investigated species of Ral/lesiaceae, Crassu-
laceae and some species of Ulmaceae, Moraceae, Urticaceae, Rosaceae,
Coriariaceae.
As for the other types and subtypes of sieve-element plastids a
reason for the presence or absence of both protein and starch is not
evident; different concentrations and qualities of carbohydrates present
in the sieve-elements (as listed by KLUGE 1967) also offer no obvious
answer.

Interrelationships Between Types and Subtypes of Sieve-Element Plastids

Subtypes and forms of P- and S-type plastids and their suggested
interrelationships are presented diagramatically in Fig. 25. Based on
the comparatively small number of taxa investigated (about 690 species
and 550 genera of 190 families 1) this compilation might appear to
be· incomplete. However, the detection within flowering plants of
P-type plastids, nevertheless, is probably relatively representative:
a comparison with the diagram given in 1975 (Fig. 18 in BEHNKE
1975b) which included only 500 species from flowering plants easily
demonstrates that data for an additional 190 species (with the
emphasis on Rosidae) did not significantly change the concept.

1 As of TAKHTAJAN (1973).
172 H .-D. BEHNKE:

Therefore, with regard to the origin of plastid types and subtypes

in flowering plants we still consider the P-type as primitive and the
S-tne as derived.
P I-subtype is by far the most ancestral type of protein-accumulat-
ing plastids demonstrating the maximum variability of forms; it is
also most often in taxa which are closely associated with others con-

INTERRELATIONSHIPS BETWEEN PROTEIN AND STARCH ACCUMULATING SIEVE-ELEMENT PLASTIDS

Fig. 25

taining S-type plastids (possIbly indicating points of transitions from

the P- to S-type, e.g. in Magnoliaceae and Laurw;eae) . This subtype is
also most similar to the sieve-element plastids of Pinaceae (only record
of P-type in Gymnospermae, BEHNKE 1974).
Besides P I, subtype P IV also occurs in taxa which are allied
with others containing S-type plastids. It is interesting to note that its
protein crystalloids could possibly be derived by a splitting and/or
multiplication of the single crystalloid in P I.
Both P II- and P III-subtypes are more uniform in their basic
pattern, and are therefore especially characteristic for the taxa in
which they occur. They are considered to be derived comparatively
early from the P I-subtype. There are, in the extant taxa investigated
Transmission Electron Microscopy and Systematics of Flowering Plants 173

so far, no transitions from P II- and P III-subtypes to the S-type.

Indeed, starch has been lost almost completely in these subtypes.
Complete loss of enzymes for the protein accumulations and con-
sequently the formation of the more advanced S-type plastids is sug-
gested to have occurred several times (as indicated in Fig. 25). It is
not to be excluded, of course that S-type plastids could also directly
be derived from the original angiosperm ancestors.

51EVE ELEMENT PLASTIDS AND THE SUBCLASSES DF FLOWERING PLANTS

(~rr."9od .tt~, CRONQUI$T 1968 )

Fig. 26

The Incorporation of the Plastid Data Into Existing Systems of

Flowering Plants
CRONQUIST'S (1968), EHRENDORFER'S (1971) and TAKHTAJAN'S(1973)
more or less similar systems for flowering plants initially appeared to
be adequate for superimposition of the distribution of P-type plastids as
a discriminating feature.
At the class level the uniform P-type Liliopsida (P II-subtype)
are clearly separated from the Magnoliopsida with P- and S-type taxa.
174 H .·D. BEHNKE:

As far as subclasses are concerned, there is a restriction of P-type

plastids to only three of the six subclasses as e.g. recognized by CRON-
QUIST (1968): Magnoliidae, OaryophyUidae and Rosidae (Fig. 26). The
same graph (Fig. 26), however, demonstrates that unlike in Liliopsida

SIEVE-ELEMENT PLASTIDS AND THE SUPERORDERS OF DICOTYLEDONS

(arranged after EHRENDORFER 1971)

A~'erana e

lamianae

......_ ............... Asteridae - _...............-:1

t
I
... . ..
GJ

... ..
Q>
c c to
'"
. .. ~'.."
.~ IV C :2
c ~ ~ :2
'" '" c IV « ~ q;
C
'"
~
C
IV U ll. E

-0 w
~ 3 to
o '"
~ 0:: E
/
to
/ J:

0;; ; ; ; ====,-\ Rosidae ='E~~'",,4=::::!J

Ranuncu lanae

Magnolianae ~""--=;~~~~~
MAGNOU AT AE Magnoliidae

; P - TYPE, 5 - TVPE

these three subclasses also contain S-type plastids (varying from 10%
to 85%).
A subdivision of the subclasses into superorders (e.g. demonstrated
by EHRENDORFER 1971) confines P-type plastidsin Magnoliidae to
part of the M agnolianae but in Rosidae still allocates · ihem to two
superorders, Rosanae, and Oelastranae (Fig. 27) .
Descending to the order level ~ .g. of TAKHTAJAN ' S (1973)dendro-
gram we are still confronted with the same problem: in addition to
Transmission Electron Microscopy and Systematics of Flowering Plants 175

the basic Magrwliales and Laurales there are at least five orders which
contain both types of plastids (see Fig. 28): Oaryophyllales, Saxifra-
gales, FalJales, Hippuridales, Rhamnales; probably also Oapparales.
DAHLGREN'S recently developed system of flowering plants (1975)
reduces the number of heterogeneous orders by two. His separating

SIEVE-ELEMENT PLASTIDS AND THE ORDERS OF DICOTYLEDONS

(arranged aile' TAKHTAJA 1973 )

~ ___ . ' I.,p'

~S · I,,,,

Fig. 28

the P-type family Gunneraceae (as order Gunnerales) from the S-type
families Hippuridaceae and Haloragaceae and excluding the S-type
families Bataceae and Gyrostemonaceae from Oaryophyllales closely
corresponds to the conclusions derived from sieve-element plastids
(see Fig. 1 in BEHNKE & DAHLGREN 1976). The dismemberment of the
OaryophyUidae (and placing the S-type orders Polygonales and Plum-
baginales next to Primulales) also accentuate the specifity of the P 111-
subtype found in OaryophyUales. For a closer connection of those taxa
exhibiting the P I-subtype DAHLGREN'S (1975) system also does not
provide a. ready solution.
176 H.-D. BEHNKE:

Considering the difficulties in incorporating sieve-element plastid

data into existing systems, a modified alternative is offered (Fig. 29),
which at least is in accordance with the distribution of the P- and
S-type plastids as presently known.
In this alternative proposal TAKHTAJAN'S (1973) division of Magnoli-
opsida into subclasses is maintained since at present S-type plastids
cannot be used for classification. Changes in the circumscrip-
tion of subclasses are undertaken with the Garyophyllidae and the

DISTRIBUTION OF SIEVE-ELEMENT PLASTIDS WITHIN FLOWERING PLANTS

(BEHNKE 1976)

P -TYPE S - TYPE
(35 ./.) (65 "10)
.. m
ASTERIDAE

w
«
Cl
::::;
..J
>- F.abanail!"
:x: I
D.. ROSIDAE
o _1._ •. __ ._ . . -_ • . . . . •
>- SaA .fra anae
Il::
«
u

DILlENllDAE

Fig. 29

Rosidae. The Garyophyllidae are strictly circumscribed by the P III-

subtype containing families of the order Garyophyllales see BEHNKE
1976), while the orders Polygonales and Plumbaginales are included
in Rosidae (DAHLGREN 1975). In addition, within the Rosidae two new
superorders are proposed:
A superorder Saxi/rdganae is proposed to include besides all S-type
families of Saxifragales (sensu TAKHTAJAN) the P I-subtype families
Eucryphiaceae, Gunneraceae, V itaceae and Leeaceae.
Eucryphiaceae was placed into Saxifragales by TAKHTAJAN (1973).
WAGENITZ (1975) list.s some differential characters which deny a
close relationship between Gunneraceae and Haloragaceae (and Myrtales
as a whole); DAHLGREN (1975) alignes his Gunnerales within Saxi-
/raganae. The families Vitaceae and Leeaceae were proposed by JOHN-
STON (1974) to be refered to Saxifragales.
Another new superorder Fabanae is split from Rosanae to include
Transmission Electron Microscopy and Systematics of Flowering Plants 177

P IV-subtype containing Caesalpiniaceae, Mimosaceae and the Fabaceae

which contain both P IV-subtype and S-type species.
Except for a few scattered genera these proposals meet the require-
ments of the distribution of the total of 65% S-type and 35% P-type
within flowering plants.

The author is grateful to Prof. T. J. MABRY (Austin, Texas) for many

helpful discussions and for linguistic corrections of the English manuscript;
to Prof. F. EHRENDORFER (Wien) for comments on the hierarchy and sub-
classification of sieve-element plastids; and to Prof. H. MERXMULLER
(Miinchen) for literature comments. Skillful technical assistance by Miss
B. SCHMIDT and Mrs. D. LAUPP during the unpublished part of this investiga-
tion is acknowledged. Supported by grants from "Deutsche Forschungs-
gemeinschaft" .

References
BEHNKE, H.-D., 1965: Uber das' Phloem der Dioscoreaceen unter beson-
derer Beriicksichtigung ihrer Phloembecken. II. Mitteilung: Elektronen-
optische Untersuchungen zur Feinstruktur des Phloembeckens. Z.
Pflanzenphysiol. 53, 214-244.
1967: Uber den Aufbau der Siebelement-Plastiden einiger Dioscoreaceen.
Z. Pflanzenphysiol. 57, 243-254.
1969a: Die Siebri:ihren-Plastiden der Monocotyledonen. Vergleichende
Untersuchungen iiber Feinbau und Verbreitung eines charakteristischen
Plastidentyps. Planta (Berl.) 84, 174-184.
1969 b: Ultrastructure of angiosperm sieve-tube plastids in relation to
systematics. Abstr. XIth Intern. Bot. Congress (Seattle), p. 12.
1971: Sieve-tube plastids of Magnoliidae and Ranunculidae in relation
to systematics. Taxon 20, 723-730.
1972: Sieve-element plastids in relation to angiosperm systematics.-An
attempt towards a classification by ultrastructural analysis. Bot. Rev.
38, 155-197.
- 1973: Sieve-tube plastids of Hamamelididae. Electron microscopic
investigations with special reference to Urticales. Taxon 22, 205-210.
1974: Sieve-element plastids of Gymnospermae: Their ultrastructure in
relation to systematics. Plant Syst. Evol. 123, 1-12.
1975a: P-type sieve-element plastids: A correlative ultrastructural
and ultrahistochemical study on the diversity and uniformity of a
new reliable character in seed plant systematics. Protoplasma 83,
91-101.
1975 b: The bases. of angiosperm phylogeny: ultrastructure. Ann.
Miss. Bot. Gard. 62, 647-663.
1976: Ultrastructure of sieve-element plastids in Garyophyllales (Gentro-
spermae), evidence for the delimitation and classification of the order.
Plant Syst. Evol. 126, 31-54.
and DAHLGREN, R., 1976: The distribution of characters within an angio-
sperm system. II. Types of sieve-element plastids. Bot. Notiser 129,
287-295.
BONNETT, H. T., and NEWCOMB, E. H., 1965: Polyribosomes and cisternal
accumt'liations in root cells of radish. J. Cell BioI. 27, 423--432.
Plant Syst. Evol.. Suppl. 1 12
178 H.-D. BEHNKE: Transmission Electron Microscopy

COLE, G. T., and BEHNKE, H.-D., 1975: Electron microscopy and plant
systematics. Taxon 24, 3-15.
CRESTI, M., PACINI, E., and SIMONCIOLI, C., 1974: Uncommon paracrys-
talline structure formed in the endoplasmic reticulum of the integu-
mentary cells of Diplotaxis erucoides ovules. J. Ultrastruct. Res. 49,
218-223.
DAHLGREN, R., 1975: A system of classification of the angiosperms to be
used to demonstrate the distribution of characters. Bot. Notiser 128,
119-147.
EHRENDORFER, F., 1971: Spermatophyta. In: Lehrbuch der Botanik fUr
Hochschulen, 30th ed., 584-745~ Stuttgart: Fischer.
FAVALI, M. A., and GEROLA, F. M., 1968: Tubular and fibrillar compo-
nents in the phloem of Brassica chinensis L. leaves. Giorn. Bot. Ital.
102,447-467.
FRIEDRICH, H. C., 1956: Studien iiber die natiirliche Verwandtschaft der
Plumbaginales und Oentrospermae. Phyton (Austria) 6, 220-263.
HAVELANGE, A., and COURTOY, R., 1974: Description et essais de caracteri-
sation cytochimique d'un composant inconnu dans les cellules meriste-
matique de Sinapis alba L. (Cruciferes). C. R. A~. Sc. Paris 278,
1191-1193.
HOEFERT, L. L., 1975: Tubules in dilated cisternae of endoplasmic reti-
culum of Thlaspi arvense (Orucilerae). Amer. J. Bot. 62, 756-760.
IVERSEN, T.-H., 1970a: Cytochemical localization of Myrosinase (~-thio­
glucosidase) in root tips of Sinapis alba. Protoplasma 71, 451-466.
1970b: The morphology, occurrence, and distribution of dilated cisternae
of the endoplasmic reticulum in tissues of plants of the Oruci/erae.
Protoplasma 71,467-477.
and FLOOD, P., 1969: Rod-shaped accumulations in cisternae of the
endoplasmic reticulum in root cells of Lepidium sativum seedlings.
Planta 86, 295-298.
JOHNSTON, M. C., 1974: Rhamnales. Encyclopaedia Britannica. Macro-
paedia Vol. 15, 793-796.
,T0RGENSEN, L. B., BEHNKE, H.-D., and MABRY, T. J., in prep.: Protein-
accumulating cells and dilated cisternae of the endoplasmic reticulum in
three glucosinolate-containing genera: Armoracia, Oapparis, Drypetes.
Planta.
KLUGE, H., 1967: Untersuchungen iiber Kohlenhydrate und myo-Inosit
in Siebr6hrensiiJten von Holzgewachsen. Dissertation Darmstadt.
MA-BRY, T. J., and BEHNKE, H.-D. (eds.), 1976: Evolution of Centrospermous
Families. Plant Syst. Evol. 126 (1).
TAKHTAJAN, A., 1973: Evolution und Ausbreitung der Bliitenpflanzen.
Stuttgart: Fischer.
THORNE, R., 1968: Synopsis of a putatively phylogenetic classification of
the flowering plants. Aliso 6, 57-66.
WAGENITZ, G., 1975: Bliitenreduktion als ein zentrales Problem der Angio-
spermen-Systematik. Bot. Jb. Syst. 96, 448-470.
WALKER, J. W., and DOYLE, J. A., 1975: The bases of angiosperm phylo-
geny: palynology. Ann. Miss. Bot. Gard. 62, 664-723.

Address of the author: Prof. Dr. H.-D. BEHNKE, Lehrstuhl fiir Zellen-
lehre, 1m Neuenheimer «'eId 230, D-6900 Heidelberg 1, Federal Republic
of Germany.
Plant Syst.. Evol.. Suppl. t. 179-189 (1977)
(<;) by Springer-Verlag 1977

The New York Botanical Garden. Bronx. New York. U.S.A.

On the Taxonomic Significance of Secondary Metabolites

in Angiosperms
By

A. Cronquist, New York

Abstract: I here put forward the speculative interpretation that changes

ill the major groups of chemical repellents had an important role in the rise
of major new groups of dicotyledons. Each set of repellents tends to lose
its effectiveness as insects and other predators become resistant to it.
A new kind of repellent gives the plants a competitive advantage and
permits the evolutionary expansion of a new group. Thus the isoquinoline
alkaloids of the Magnoliidae gave way to the tannins of the Hamamelidae,
Rosidae and Dilleniidae, and these in turn gave way to the iridoid compounds
that were most effectively exploited by the Asteridae. Within the Asteridae,
the rise of the relatively recent family Asteraceae may relate to a shift to
polyacetylenes and sesquiterpene lactones in place of the already less
effective iridoids. The pattern is blurred b"y the continued evolutionary
experimentation with new repellents by each major taxonomic group,
and by the exploitation of the same new group of repellents by different
taxa at about the same time.

Chemical data have until now played only a small role, or virtually
none, in the formulation of widely recognized systems of classification of
flowering plants. Yet the accumulation of information over the past two
or three decades has shown that the chemistry correlates well enough
with morphology and other feat.ures so t.hat. it should be t.aken into
account.
Unfortunately, effort.s to incorporate the chemical dat.a into t.he
syst.em of classification have so far met with only limited success. Belief
in the overriding, a priori importance of chemistry has led some botanists
to taxonomic conclusions completely unacceptable to other botanists
who rely more on morphological features.
The benzyl-isoquinoline alkaloids furnish an example of both t.he
usefulness and the problems of the chemical data. It is now well recog-
nized that these alkaloids are widespread in the M agnoliidae, but rare
elsewhere. They furnish one of the most. useful markers of the subclass,
12"
180 A. CRONQUIST:

and their presence in the Papaverales is one of the important features

which has led to the general agreement that the Papaverales are related
to the Ranunculales rather than to the Capparales. (I here use the name
Magnoliidae in the broad sense, to include the Ranunculidae of TAKH-
TAJAN.)
KUBITZKI (1969, 1973) has argued that the distinctive chemistry
of the Magnoliidae makes this group an unlikely ancestor for other
angiosperms. Yet on the basis of comparative morphology the M agno-
liidae do appear to be phyletically basal within their division; the
palaeobotanical data provide stronger support for this point of view year
by year, almost day by day. On the basis of the fossil record it is now
difficult to challenge the concept that the earliest angiosperms had
monosulcate pollen, and it is becoming clear that the mono cots began
to diverge from the dicots before the origin (or at least before the evo-
lutionary diversification) of triaperturate pollen (DOYLE 1973). The
only modern dicots with monosulcate pollen belong to the Magnoliidae
(WALKER & DOYLE 1975; WALKER 1976). The evidence from leaf
architecture is not yet so conclusive, but it is now apparent that early
fossil angiosperm leaves find their closest modern counterparts among
the Magnoliidae (HICKEY & WOLFE 1975). Thus KUBITZKI'S conclusion,
based largely on chemical evidence, is at odds with the conclusion based
on other evidence.

The betalains have been a notorious cause of dispute between chemically

oriented and morphologically oriented taxonomists. MABRY & TURNER
have adamantly maintained in various publications (e.g., 1964) that the
presence of betalains is a sine qua non for admission to the order that has
often been called Oentrospermae. Other botanists (ECKARDT 1976, CRON-
QUIST 1973) have been equally adamant that it makes no sense to exclude
the Oaryophyllaceae from that order merely because they have anthocyanins
instead of betalains. Only very recently MABRY (1976) has concluded that
the Oaryophyllaceae and Molluginaceae can be restored, as a suborder, to
the same order that includes the other centrospermous families.
Chemically oriented botanists have frequently pointed out the chemical
similarity of the Asteraceae to the Apiaceae (Umbelliferae) (e. g., HEGNAUER,
1971). Morphologically oriented botanists have not even deigned to reply
in print, because on morphological grounds is seems impoSsible to associate
these two families.
DAHLGREN (1975) has recently assigned major importance to the presence
of iridoids and some other compounds in his novel system of classification,
but here again the chemically defined groups are morphologically hetero-
geneous.

So we have a dilemma. The chemical data obviously must be taxo-

nomically important, yet reliance on them to establish evolutionary
relationships frequently gives results that are out of harmony with
Taxonomic Significance of Secondary Metabolites in Angiosperms 181

morphologically based systems. We need a new rationale for under-

standing the significance of chemical data to the general system. I here
propose such a new rationale.
It is so difficult to provide a plausible Darwinian interpretation for
much of the morphological evolution at the level of families and orders of
angiosperms that most phylogenists have concentrated on the establish-
ment of trends and relationships, with little more than lip service to
their Darwinian significance. Chemical differences have also often been
discussed almost in a vacuum, as if they had no real importance to the
plants. Although a Darwinian interpretation of many of the morpho-
logical differences still escapes me, I suggest that survival value does
provide the key to the evolutionary and taxonomic interpretation of
chemical differences among the major groups of angiosperms.
FRAENKEL (1959) and JANZEN (1975) are among those who have
emphasized the importance of secondary metabolites in plants as repel-
lents. In accordance with their views, I believe that most of the chemical
compounds which have. been considered of taxonomic importance at
the level of families and above are repellents. They discourage predators
or pathogens. The known chemical differences among the major taxa
are largely differences in defensive weapons. Even the betalains are
reported to be toxic to certain fungi (KIMLER 1975).
Every weapons system has its costs-in the case of plants, metabolic
costs. No one plant can afford to maintain a complete, functional array
of all possible defensive weapons; the metabolic tax would be insup-
portable. Adoption of a new weapons system is therefore likely to mean
abandonment of an old one. Among the metabolic costs of weaponry I
here include not only the direct use of energy to produce the weapons,
but also the adjustments and compromises necessary to prevent the
plant from being hoist on its own petard. The plant must be able to
withstand its own repellents.
It should be no surprise to any student of history that any defensive
weapon sooner or later loses its effectiveness. The same may be true of
offensive weapons, but we are here concerned with defense. For several
hundred years the Great Wall of China served its intended purpose of
keeping out the barbarians, but eventually it was breached. No con-
ceivable wall could turn back an airplane. Anti-aircraft guns may dis-
courage trespassing airplanes, but they are useless against ballistic
missiles. And so it goes.
We are all acquainted with the decline in effectiveness of DDT over
the course of two or three decades, and the appearance of penicillin-
resistant strains of pathogenic bacteria over the same time span. Here
the massive human disturbance caused rapid evolution of resistance to
poisons. Although the rate is slower, the chemical weapons of plants also
182 A. CRONQtTIST:

lose much of their effectiveness as times goes on. The predators and
pathogens develop resistance. Compounds that originally served as
repellents may even become attractants for specialized predators.
The secondary metabolites of flowers often differ, at least in detail,
from those of vegetative organs. Even the different parts of a flower may
differ in this way (GIANNASI 1975). The possible biological and taxo-
nomic significance of chemical differences between vegetative and floral
parts has scarcely begun to be evaluated, although it seems obvious that
flowers have more use for attractants than do vegetative parts. It is the
compounds in the vegetative parts that have attracted attention as
possible markers of major taxonomic groups. Some of these, such as
the betalains, occur in the flowers as well, but others may not. It is
clear enough that differences in the flavonoids and other secondary
metabolites of flowers are taxonomically significant at the level of genera
and species, but it does not now appear that such primarily floral chemical
differences are useful at the higher taxonomic levels. More information
may of course modify or reverse this view. RTO npoaumeT YBIIAeT.
EHRLICH and RA VEN (1965) recognized an intimate correlation between
the evolution of new repellents by plants and the evolution of butterflies
capable of feeding (in the larval stage) on plants with such defenses. They
further suggested that plants with new repellents "would in a sense
have entered a new adaptive zone. Evolutionary radiation of the plants
might follow, and eventually what began as a chance mutation or
recom bination might characterize an entire family or group of related
families." The remainder of the present paper largely follows the direc-
tion forecast by these authors.
As a speculative interpretation, put forward for consideration by
other botanists, I suggest that the evolution of chemical repellents plays
an important but complex role in the rise and diversification of new
families, orders, and subclasses of angiosperms. When one set of repel-
lents loses some of its effectiveness, the time is ripe for another set. A
suitable new set of repellents gives its possessors a competitive advantage
and permits their evolutionary expansion. The increased abundance of
the new group fosters the evolution of resistance to their repellents by
predators. Then another new group of repellents may be exploited, in
the continuing struggle between predator and prey. A long period of
relative disuse may even permit the revival of an old set of repellents,
after the predators have lost resistance.
Evolutionary experimentation with new repellents is constantly
going on. Every member of a taxonomic group has the potential op-
portunit.y to try something a little different, by way of a single mutation
ncar the end of a biosynthetic chain. Thus the chemical possibilities of a
given set of repellents are explored and exploited.
Taxonomic Significance of Secondary Metabolites in Angiosperms 183

Since every set of repellents gradually loses its effectiveness, op-

portunities arise for the selective development of new sets. It is not to
be expected that the first members of a new set of repellents will be. the
most useful ones. Therefore the effectiveness of existing repellents must
have declined to a level that can be surpassed by one of the less effective
members of a potentially more effective new set, or nothing much is
likely to happen.
The same set of repellents may be developed and exploited by two
or more major taxa at about the same time. Competitive exclusion here
takes an oddly delayed form. The evolutionary opportunity is initially
open to anything that can produce the right mutations. The opportunity
declines in value as the predators become resistant.
Under special local conditions, or with the aid of happenstance
fixation of mutations, individual species or even larger groups may turn
to repellents already widely used by other taxa. Such re-adopted repel-
lents can not confer any broad-scale advantage on their possessors, and
the evolutionary expansion of the taxonomic group that produces them
is likely to be blocked by preadapted predators (LEVIN 1976).
The hypothesis provides for all degrees of usefulness of repellents.
Some may initially be so much better than their predecessors that they
foster tremendous evolutionary expansion of the groups that produce
them. The tannins and the iridoids may represent this sort. Others may
give their possessors a degree of advantage, but not enough to foster the
evolution of a major new group. Such repellents might evolve several
times in different major taxa, in each case allowing some expansion of
the group before a competitive balance is restored. The triterpenoid
saponins might possibly belong to this class. Still others may be very
effective, but impose such a heavy metabolic tax that their net value is
severely limited. They might be compared to medicines with virtually
no margin between the therapeutic and the toxic dose. Repellents of
this class might be tried repeatedly, but never lead to the evolution of a
major new group. Hydrogen cyanide may belong to this class.
Under this hypothesis, the basic pattern in evolution of repellents is
one of successive shifts from one major set to another, in response to
progressive increase in resistance by the predators. The pattern is blur-
red by the constant evolutionary experimentation with new repellents
and refurbished old ones. All the major taxa have individual species or
genera that produce one or· another unusual repellent in addition to or
instead of the common ones. When one of these unusual repellents
confers a sufficient competitive advantage over a long enough period of
time, the small taxonomic group becomes a large one, and the uncommon
repellent becomes a common one. Doubtless it is easier for the predators
to evolve resistance to some repellents than to others. Therefore the
184 A. CRONQUIST:

evolutionary expansion of a group of plants with a particular set of

repellents will be governed in part by how long it takes the predators to
become resistant.
The hypothesis implies that the development of a new set of repel-
lents is neither very hard nor very easy. If it were very hard, then any
particular set of repellents would probably evolve only once, and the
correlation of chemistry and morphology would be much stronger than
it is. If it were very easy, then the old repellents would not persist, but
would be abandoned with the speed of mutual fund managers abandoning
a falling stock.

I claim no expertise in chemistry, but it seems to me that the genetic

difficulties in the development of a new set of repellents are rather modest.
Anthocyanins are widespread in the Magnoliidae as well as in other angio-
sperms. Given the machinery to produce anthocyanins, not many extra
parts are needed to produce tannins. MABRY (personal communication)
has recEmtly concluded that probably only two or three enzymes need be
added to the common angiosperm supply to permit the formation of
betalains. Likewise it appears that the production of iridoid compounds
does not require any great number of enzymes beyond those common to
angiosperms in general.

For each of the principal sets of repellents, it appears that most of

the necessary equipment is a part of the common heritage of angiosperms,
and only one or a few parts need be added to produce a working machine.
It is easy to be impressed with the number of links in the biosynthetic
chain leading to any metabolite, and equally easy to overlook the fact
that most of these links are used in more than one chain. Assembly of a
new chain may require the formation of only one or two new links. Yet
these new links are not just floating around waiting to be picked up.
The proper mutations must occur, and the frequency of these mutations
differs from one taxonomic group to another, in accordance with well
known principles governing partial genetic control of mutability
(DOBZHANSKY 1941). The number of mutations necessary to permit
the plants to withstand their own new set of repellents is unknown, but
may be presumed to be more than just one or two.
Thus the formation of new,kinds of repellents may be about at the
level of difficulty that the hypothesis requires.
Now let us see how these theoretical considerations might apply
to the problems at hand. Here again I propose a speculative interpreta-
tion, which mayor may not prove to be basically correct. It must with-
stand critical analysis before it can become anything more than a working
hypothesis.
I suggest that the Magnoliidae are indeed the phyletic ally basal
group of angiosperms. For whatever reasons, they have used benzyl-
Taxonomic Significance of Secondary Metabolites in Angiosperms 185

isoquinoline alkaloids as their most important defensive weapons. By

the time the M agnoliidae became common and widespread, in the late
Cretaceous, their predators had probably developed a consiQ-erable
resistance to this set of repellents. The time was therefore ripe to develop
and explore a new set. The most immediately available new repellents
were the tannins, notably including ellagic acid and the proanthocy-
anidins. The M agnoliidae characteristically produce anthocyanins, and the
same basic shikimic acid pathway leads to both anthocyanins and tan-
nins. The Hamamelidae, Rosidae, and Dilleniidae, which were differ-
entiating from the Magnoliidae in the late Cretaceous, all exploited
tannins as repellents and discarded the outmoded benzyl-isoquinoline
alkaloids.
We may note that STEBBINS (1965), and DOYLE & HICKEY (1976)
think that the earliest angiosperms may have been shrubs, and that
angiosperm trees came in a little thereafter. If their view is correct,
then the change to an arborescent habit may have changed the balance
of predators (more insects, fewer vertebrates) and set the stage for
insect-oriented rather than vertebrate-oriented repellents. I don't know
the relative effectiveness of tannins and benzyl-isoquinoline alkaloids
on modern insects and vertebrates, let alone the Cretaceous ones. My
suggestion here is merely an example of the sort of approach that might
be useful.
In the late Cretaceous, tannins were a refurbished rather than an
absolutely new repellent. Some gymnosperms were producing tannins as
far back as the late Palaeozoic (NIKLAS & GENSEL 1976), and many
conifers continue to do so to this day. Yet the major evolution of insects
during much of the Cretaceous period had occurred with angiosperms
as the principal hosts, and it can reasonably be assumed that the insects
preying on angiosperms in the late Cretaceous were not highly resistant
to tannin.
-Once the tanniferous groups of angiosperms began to dominate thc
landscape, the insects and other plant predators had to develop resistance
to tannin. In so doing, they probably lost some of their resistance to
benzyl-isoquinoline alkaloids. Thus the diminishing abundance of the
Magnoliidae may have caused a reciprocal increase in the effectiveness
of their repellents, which have subsequently been tried out with modest
success by a few more advanced groups. Perhaps at some time in the
long-distant future the benzyl-isoquinoline alkaloids can again be
effectively exploited by a new group of angiosperms.
By about the beginning of the Tertiary period, the insects ,md other
predators may have become fairly resistant to tannins, so that the time
was ripe for a new set of repellents. Some members of the Rosidae, Dil-
leniidae, and even a few of the Hamamelidae then began to exploit
186 A. CRONQUIST:

iridoids in place of tannins. In some few of the Rosidae the use of iridoids
was combined, perhaps fortuitously, with sympetaly, leading to the
formation and rapid expansion of the Asteridae. The similar combina-
tion of sympetaly and iridoid repellents in the Ericales (Dilleniidae) did
not lead to a comparable evolutionary expansion, because the Ericales
were by then already committed to mycotrophy, limiting their evolu-
tionary opportunities.
Like the tannins before them, the iridoid compounds eventually
began to lose some of their effectiveness, and other groups of repellents
proliferated. At the end of the Oliogocene, the Asteraceae began to
exploit polyacetylenes and sesquiterpene lactones together. This very
successful combination may have been the most important factor in thc
explosive rise and diversification of the family.
The principal arsenal of the Asteraceae may now already have passed
its peak of effectiveness. The evolutionary expansion of the group is so
recent that extinction has not yet produced good generic lines, but
already the family is experimenting successfully with new defenses. The
Senecioneae, which are certainly not a primitive tribe, have produced
the nearly unique Senecio alkaloids. Senecio is the largest genus in the
Asteraceae, by any reasonable specific standards, and it has become
cosmopolitan and ubiquitous. The most recently evolved tribe appears
to be the Lactuceae, which does not appear until late Miocene, about
15 million years ago (RAVEN & AXELROD 1974). The Lactuceae have
largely abandoned polyacetylenes and sesquiterpene lactones, and
produce latex instead.
Very recently (BOWERS et al. 1976) it has been shown that Ageratum
houstonianum produces an effective insect anti-juvenile hormone. This
substance, a simple chromene, is exciting interest as a possible source of
insect-specific pesticides. It remains to be determined how widespread
such substances are in the Asteraceae or in angiosperms in general. They
mayor may not have anything to do with the success of the Asteraceae
in comparison to other families.
Polyacetylenes are exploited by the Apiales (especially the Apiaceae)
and Campanulaceae, as well as the A steraceae , but these other groups
apparently lack sesquiterpene lactones. It is reasonable to suppose that
the absence of the iridoids from the Campanulaceae is secondary, because
these are present in more primitive families of Asteridae and even in the
Goodeniaceae, which are usually referred to the Campanulales. On the
other hand, there is no reason to suppose that the precursors of the
Apiales ever had iridoid compounds. Their most likely ancestors, the
Sapindales, do not have them. By the time the Apiales began to produce
polyacetylenes, the opportunity for effective exploitation of iridoids may
already have passed.
Taxonomic Significance of Secondary Metabolites in Angiosperms 187

The Apiales have a fairly long fossil history. Fossils considered to

represent the Araliaceae occur throughout the Tertiary and even into
the Upper Cretaceous. It is not at all certain, however, that the e.arliest
members of the group had polyacetylenes. There is a considerable
overlap in chemical repellents between the Araliales and the presumably
ancestral Sapindales (HEGNAUER 1971), but the polyacetylenes are not
in the arsenal of the Sapindales. Thus it is possible that polyacetylenes
were not invented in the Araliales until long after the origin of the order.
The Caryophyllales are chemically noteworthy for the substitution of
betalains for anthocyanins in most families. The partial absence of proan-
thocyanidins and other tannins from the group may be biologically more
important than the absence of true anthocyanins. Most of the mechanism
for the production of anthocyanins (and proanthocyanidins) is obviously
present at least in some members of the order, because some of them
produce yellow flavonoid pigments as well as betalains (ALSTON 1967).
Although MABRY & TURNER (1964) have proposed that the betalain
families are a very apcient group, their known fossil record is less than
60 million years long (MULLER 1970). I suggest that a search for the
biological significance of the betalains should concentrate on their
repellent (and fungicidal) properties, rather than on their function as
flower pigments.
The principles here expounded should be applicable to monocots as
well as dicots, but I am not yet ready to propose a comprehensive inter-
pretation. We may note that the intercalary meristem of grass leaves
may be a more important factor in the success of the family than any
repellent that may be present, but the siliceous deposits in the cells do
discourage attacks by chewing insects (FRAENKEL 1959).
I hope that the approach here suggested will permit the harmonious
integration of the data on secondary metabolites into the general
taxonomic system. The thesis and antithesis of morphological and
chemical char'acters may thus be resolved into a general synthesis.

1 thallk Dr. DAVID GIANNASI and Dr. KARL NIKLAS for advice alld
counsel during the preparation of this paper. Responsibility for the contents
if', as always, my own.

References
ALSTON, R., 11167: Biochemical systematics. In: Evolutionary biology
(DOBZHANSKY, T. S., HECHT, M. K., and STEERE, W. C., eds.), Vol. 1,
1!17-305. New York: Appleton-Century-Crofts.
BOWERS, W. S., OHTA, T., CLEERE, J. S., and MARSELLA, P. A., 1976:
Discovery of insect anti-juvenile hormones in plants. ;:;;cience 193,
542-547.
(;RONQUIST, A., 1973: Chemical plant taxonomy: A generalist's view of
a promising specialty. Nobel Symposium 25, 29-- 39.
188 A. CRONQUIST:

DAHLGREN, R., 1975: A system of classification of the angiosperms to be

used to demonstrate the distribution of characters. Bot. N otis. 128,
119-147.
DOBZHANSKY, T. S., 1941: Genetics and the origin of species. 2nd. ed.
New York: Columbia University Press.
DOYLE, J. A., 1973: Fossil evidence on early evolution of the monocoty-
ledons. Quart. Rev. BioI. 48, 399-413.
and HICKEY, L. J., 1976: Pollen and leaves from the Mid-Cretaceous
Potomac Group and their bearing on early angiosperm evolution. In:
Origin and early evolution of angiosperms (BECK, C. B., ed.), 139-206.
New York: Columbia University Press.
ECKARDT, T., 1976: Classical morphological features of centrospermous
families. Plant Syst_ Evo!. 126, 5-25.
EHRLICH, P. R., and" RAVEN, P. H., 1965: Butterflies and plants: a study
in coevolution. Evolution 18, 586-608.
FRAENKEL, G. S., 1959: The raison d'etre of secondary plant substances.
Science 129, 1466-1470.
GIANNASI, D. E., 1975: The flavonoid systematics of the genus Dahlia
(Composltae). Mem. N.Y_ Bot. Gard. 26 (2), 1-125.
HEGNAUER, R., 1971: Chemical patterns and relationships of the Umbel-
liferae. In : The biology and chemistry of the Umbelliferae (HEYWOOD,
V. H., ed.), 267-277. Suppl. 1 to Vol. 64; Bot. J. Linnean Soc.
HICKEY, L_ J., and WOLFE, J. A., 1975: The bases of angiosperm phy-
logeny: vegetative morphology. Ann. Missouri Bot. Gard. 62, 538-589.
JANZEN, D. H., 1975: Ecology of plants in the tropics. London: Edward
Arnold.
JENSEN, S_ R., NIELSEN, B. J., and DAHLGREN, R., 1975: Iridoid com-
pounds, their occurrence and systematic importance in the angiosperms.
Bot. Notis. 128, 148-180.
KIMLER, L., 1975: Betanin, the red beet pigment, as an antifungal agent.
Bot. Soc. Amer. Abstr. 36.
KUBITZKI, K., 1969: Chemosystematische Betrachtungen Zl1r Grol3gliede-
rung der Dicotylen. Taxon 18, 360-368.
- 1973: Probleme der Grol3systematik del' Blittenpflanzen. Bel'. dtsch.
bot. Ges. 85, 259~277 (1972).
LEVIN, D. A., 1976: The chemical defenses of plants to pathogens and herbi-
vores. Ann. Rev. Eco!. Syst. 7, 121- -159.
MABRY, T. J., 1973: Is the ordElr Centrospermae monophyletic ~ Chemistry
in Botanical Classification. Nobel Symposium 25, 275-285.
1976: Pigment dichotomy and DNA-RNA hybridization data for
centrospermous families. Plant Syst. Evol. 126, 79-94.
and TURNER, B. L., 1964: Chemical investigations of the Batidaceae.
Betaxanthins and their systematic implications. Taxon 13, 197-200.
MULLER, J., 1970: Palynological evidence on early differentiation of angio-
sperms. BioI. Rev. Cambridge Philos. Soc. 45, 417-450.
NIKLAS, K., and GENSEL, P. G., 1976: Chemotaxonomy of some paleozoic
vascular plants. Part 1. Chemical compositions and preliminary cluster
analyses. Brittonia 28, 353---378.
RAVEN, P. H., and AXELROD, D.1., 1974: Angiosperm biogeography and
past continental movements. Ann. Missouri Bot. Gard. 61, 539-673.
Taxonomic Significance of Secondary Metabolites in Angiosperms 189

. STEBBINS, G. L., 1965: The probable growth habit of the earliest flowering
plants. Ann. Missouri Bot. Gard. 52, 457-468.
WALKER, J. W., 1976: Comparative pollen morphology and phylogeny of
the Ranalean complex. In: Origin and early evolution of angiosperms
(BECK, C. B., ed.), 241-299. New York: Columbia University Press.
and DOYLE, J. A., 1975: The bases of angiosperm phylogeny: palynology.
Ann. Missouri Bot. Gard. 62, 664-723.

Address of the author: Dr. A. CRONQUIST, New York 'Botanical Garden,

Bronx Park, Bronx, NY 10458. U.S.A.
Plant Syst. Evol., Suppl. 1, lIH-209 (1977)
© by Springer-Verlag 1977

Laboratorium voor Experimentele Plantensystematiek,

University of Leiden, The Netherlands

Cyanogenic Compounds as Systematic Markers in Tracheophyta

By
Robert Hegnauer, Leiden

Abstract: A short chronological review of cyanogenesis (Fig. 1) in

vascular plants is given (Table 1). Subsequently the chemistry (Tables 2-4;
Fig. 2) and biochemistry (Figs. 3-9) of cyanogenic plant constituents are
summarized. Five biogenetica;l groups (A-E) of cyanogenetic compounds
are presently known from vascular plants. The occurrence and function
(Fig. 10) of cyanogenesis is sketched and causes of conflicting statements
in phytochemical literature are briefly discussed. Finally the presently
known distribution of the different pathways, and the resulting constituents,
as well as of still unidentified cyanogenic compounds, is given. The paper
ends with a short systematic appreciation of cyanogenesis as a systematic
marker at higher categories. The differences between Pteridophyta, Gymno-
spermae and Angiospermae and the similarities between Liliopsida and the
magnoliid and ranunculid part of Magnoliopsida are stressed and the uni-
formity of Passiflorales, including Flacourtiaceae as well as the hetero-
geneity of Rosaceae are outlined and discussed from the taxonomic point
of view.

Introduction
Since FLlrCKIGER (1883: 722-728,950-955) reviewed the distribution
of hydrocyanic acid in nature, several aspects of the phenomenon
cyanogenesis were periodically summarized in literature. In the follow-
ing discussion of the chemistry and biochemistry of cyanogenesis and
cyanogenic constituents and in their evaluation as systematic characters
no references are given as a rule. The reader is advised to consult the
following recent reviews, when more information or original papers
are needed: DILLE.MANN (1958), TSCHIERSCH (1967), BUTLER (1969),
CONN (1969, 1973), EYJOLFSSON (1970), HEGNAUER (1962-1973a,
1973b), NAHRSTEDT (1973), TAPPER and REAY (1973), GIBBS (1974),
SEIGLER (1975).
Cyanogenesis in the Vegetable Kingdom
Cyanogenesis may be defined as the ability of certain plants to
release hydrocyanic acid (prussic acid) after injury of cells. Usually
cyanophoric plants contain one or several cyanogenic glycosides and
192 R. HEGNAUER:

--
R. OH
I ,,/
C + Sugar(s)
R2/ "CN
0( - Hydroxynitrile
Cyanogenic glycoside (= Cyanohydrin)
n I

R.
'C=O + HCN
R{'
Aldehyde Hydrocyanic
or Ketone acid
Fig. 1. The process of cyanogenesis operating in most cyanophoric plants.
Reaction I: catalyzed by ~.glycosidases. Reaction II: the rate of spon-
taneous release of HCN depends upon temperature and pH; the reaction
may also be catalyzed by oxynitrilases

Table 1. Early history of cyanogenesis in the plant ~ugdom

0;;;;

Year Species Plant parts Family Authors

1801 Prunus amygdalus Seed Rosaceae BOHM

1802 Prunus laurocerasus Leaf Rosaceae SCHAUB
1803 Prunus persica and Leaf; flower Rosaceae SCHRADER
P.spinosa
1803 Prunus armeniaca Seed Rosaceae VAUQUELIN
1812 Prunus padus Bark Rosaceae BERGEMANN
1812 Many Prunoideae, Seeds; leaves; Rosaceae Several
to Maloideae and bark; roots authors
1885 Spiraeoideae
1836 Manihot esculentuR Tuberous Euphorbiaceae HENRY and
roots BOUTRON-
CHARLARD
1840 Lucuma mammosa Seed Sapotaceae GAYTON
18431 Phaseolus lunatus Seed Leguminosae MARCADIEU
1862 Ohardinia orientalis Fruit Asteraceae EICHLER
1862 Lepidium sativum Seeds; seed- Brassicaceae SCHULZ
lings
1867 X imenia americana Unripe fruits Olacaceae ERNST
1870 Vicia sativa Seeds Leguminosae RITTHAUSEN-
KREUSLER
1871 M arasmius oreades BasidiocarpR Agaricaceae LosEcKE
1874 Ipomoea dissecta Leaves Oonvolvulaceae FLUCKIGER-
HANBURY
Linum usitatissimum Seeds,seed- Linaceae
lings, stems

I
A quuegia V'U 19ari,s Flowering Ranunculaceae
plant
1884 Saxifragaceae .JORIssEN
Ribes aureum Young shoots
Arum maculatum Young leaves Araceae
Glyceria maxima Flowering Poaceae
plant
Cyanogenic Compounds as Systematic Markers in Tracheophyta 193

Table 2. Phytochemical history of cyanogenic plant constituents: First

two periods (1830-1840: ROBIQUET and BOUTRON-CHARLARD; HENRY and
BOUTRON-CHARLARD; WOHLER and LIEBIG. 1890-1910: The plant physio-
logy or TREUB-GUIGNARD period)

First Products of enzymatic

isolation Name of hydrolysis
Source
of pure constituent (A.= acetone;
compound B = benzaldehyde)

1830 Amygdalin Seeds of Prunus HCN + +

B 2 glucose
amygdalus (bitter (1837); sugar is gentio-
almonds); Rosaceae biose (1923)
1834/36 Manihotoxin Tuberous roots of Identical with linama-
M anihot esculentus; rin (1906)
Euphorbiaceae
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 0 _ _ _ _ _ _ _ _ _ _ _ _ _ _, _

1891 Linamarin Seedlings of Linum HCN + A + 1 glucose

usitatissimum ;
Linaceae
1902 Dhurrin Young plants of HCN +
p-OH-B +1
Sorghum vulgare; glucose
Poaceae
1903 Phaseolunatin Seeds of Phaseolus Identical with linama-
lunatus; Leguminosae rin (1906)
1904 Gynocardin Seeds of Gynocardia HCN +dihydroxycy-
odorata; Flacour- clopentenone +
1 glu-
tiaceae cose (1966, 1970)
1905 Sambunigrin Leaves of Sambucus HCN + + B 1 glucose
nigra; Caprifoliaceae
1905 Prulaurasin Leaves of Prunus HCN + B + 1 glucose;
laurocerasus ; is racemized prunasin
Rosaceae (1941)
1906 Vicianin Seeds of V icia HCN +
B +
1 arabi-
angustifolia (sativa); nose +1 glucose; di-
Leguminosae saccharide = vicia-
nose (1910)
1907 Prunasin Twigs of Prunus HCN +
B +
1 glucose
padus; Rosaceae

the corresponding more or less substrate-specific enzymes which catalyze

the release of hydrocyanic acid (Fig. 1). Sometimes, however, these
enzymes are inactive or totally lacking; such plants may contain large
amounts of cyanogenic glycosides without being cyanophoric after
damage.
Hydrocyanic acid was first detected in living nature by BOHM in
1801, and amygdalin, the first plant glycoside obtained as a pure and
crystalline material, was described in 1830 by the French scientists
ROBIQUET and BOUTRON-CHARLARD. Since 1801 seeds, flowers, leaves

Plant Syst. Evol.. Suppl. 1 13

194 R. HEGNAUER: Cyanogenic Compounds 88 Systematic Marker

Table 3. Phytochemical history of cyanogenic plant constituents: Third

period (The toxicological or RDIINGTON and STEYN-F'INNEMORE period)

First Products of enzymatic

isolation Name of hydrolysis
of pure constituent Source (OH-B = hydroxy-
compound benzaldehyde)

1935 Acacipetalin Leaves of South- HCN + isobutyric acid

african Acacia taxa; (= dimethylketene
Leguminosae + H20) + 1 glucose;
revised structure:
HCN + methyl-
acrolein + 1 glucose
(1975)
1936 Goodia-glu- Leaves of Goodia HCN + p-OH-B + 1
coside loti/olia; Legu- glucose (= p-gluco-
minosae syloxybenzaldehyde-
cyanohydrin)
1936 Phyllanthin Leaves of Phyl- HCN + p-OH-B + 1
(renamed lanthus gaaBtroemii ; glucose
taxiphyllin Euphcwbiaceae
in 1964)
1936 Zierin Leaves and twigs of HCN + m-OH--B + 1
Zi8ria laevigata; glucose
Rutaceae
1937 Acalyphin Leaves of Acalypha Structure still unknown
indica; Euphcwbiaceae
1938 Lotaustralin Herb of Lotus auBtra- HCN + methylethyl-
lis; Leguminosae ket~ne + 1 glucose
1948 Lucumin Seeds of Oalocarpum HCN + benzaldehyde
sapota (= Lucuma + disaccharide; sugar
mammoaa) ; Sapota- is primverose (1971)
ewe

and (or) barks of many rosaceous plants were shown to be strongly

cyanophoric. Two tropical alimentary crop plants, Cassava and Lima
bean, belong to the first non-rosaceous taxa which were demonstrated
to be cyanophoric. The early history of research on cyanogenesis is
summarized in Table 1.
After JORISSEN had shown in 1884 that several non-rosaceous
European plant species release hydrocyanic acid (Table 1), plant
physiologists, toxicologists and phytochemists became much interested
in cyanogenesis and its causes. Between 1885 and 1910 Dutch (e.g.
VAN ROMBURGH, GRESHOFF, TREUB) and French (e.g. GUIGNARD,
HEBERT) scientists dete!lted many new cyanophoric plants. In 1906
GRESHOFF published a list of cyanogenic taxa containing 179 species
representing 34 families of Angiosperms and one family of Fungi.
Table 4. Phytochemical history of cyanogenic plant constituents: Fourth
period (The period of biogenetical and ecological investigations)

First Products of enzymatic

isolation Name of hydrolysis
of pure constituent Source (OH-B = hydroxy-
compound benzaldehyde)

1966 Nandina-glu- Leaves of Nandina Identical or enantiome-

coside domestica; ric -with Goodia-gluco-
Berberidaceae side
1966 Proteacin Seeds of Macadamia HCN + p-OH-B + 2
integrifolia ; glucose; one glucose
Proteaceae linked to the phenolic
OH
1969 Triglochinin Herb of Triglochin HCN + triglochinic
maritima; Jun- acid (ketene + H20)
caginaceae + 1 glucose; seco-
dhurrin structure
(1972)
1969 Barterin Rootbark of Barteria HCN + hydroxycy-
(= Bar- fistulosa; Flacour- clopentenone + 1 glu-
terioside) tiaceae cose (1970)
1969 Deidamin (re- Leaves of Deidamia HCN + cyclopente-
named deida- clematoides; Passi- none + 1 glucose
clin in 1970) floraceae (1970)
1969 Cyanolipid Seedoil of Oordia HCN + hydroxy-
verbenacea ; methylacrolein + 2
Boraginaceae fatty acids (C16-C22)
1970 Cyanolipids Seedoils of many HCN + a C5 -aldehyde
Sapindaceae + one or two fatty
acids
1971 Tetraphyllin-A1. Fruits of Tetrapathaea Identical with deidaclin
1971 Tetraphyllin-B J tetrandra; Passi- Identical with barterin
floraceae
1972 Thalictrum- Herb of Thalictrum A monomethylester of
glucoside aquilegifolia; triglochinin ; methyla-
Ranunculaceae tion probably during
isolation! (1976)
1973 Holocalin Seeds of H olocalyx HCN + m-OH-B + 1
balansae; Legumi- glucose
nosae
1974 Cardio- Herb of Oardio- HCN + hydroxyme-
spermin spermum hirsutum; thylacrolein + 1 glu-
Sapindaceae cose
1975 Dihydroaca- Leaves of South- HCN + isobutyralde-
cipetalin african Acacia-taxa; hyde (= dihydroac-
Leguminosae and rolein) + 1 glucose
leaves of Hetero-
dendron oleaefolium ;
Sapindaceae
1976 Sorbaria-CN - Sorbaria arborea ; HCN + hydroxyme-
glucoside Rosaceae thylacrolein + glu-
cose + poOH-benzoic
acid
196 R. HEGNAUER:

To-day we know about 25 cyanophoric species of Fungi, about

50 species of cyanogenic ferns, 6 species of cyanogenic Gymnosperms
and about 2000 species of cyanogenic Angiosperms.

Chemistry and Biochemistry of Cyanogenic Plant Constituents

Structural studies with cyanogenic plant constituents started in
1837, when WOHLER and LIEBIG showed that amygdalin is composed
OH
I
H-'C-CN

¢
O-Glyc.
I
R1- C-CN
I
R2 O-Gluc.

I n

Gluc.-O "CN
'c
II
C
;' ....
Rl R2

m
Fig. 2. Presently known chemical types of cyanogenic plant constituents.
I: cyanohydrin glycosides (= glycosides of oc-hydroxynitriles = oc-hydr-
oxycyanides); .the usual type of cyanophoric plant constituents (compare
Figs. 1, 3,4,6, 7, 8, 9). II: unstable cyanogenic compounds (oc-hydroxy
group free; compare Figs. 5 and 6; sugar, if present, combined with a
phenolic hydroxyl). III: glycosides of oc-unsaturated oc-hydroxycyanides
(i.e. glycosides of ketene cyanohydrins; cf. Fig. 7; hydrolysis generates
highly unstable ketenes which by addition of water generate stable organic
acids). IV: cyanolipids of seed oils (n may be 14, 16, 18 or 20; cf. Fig. 8)

of HON, benzaldehyde and glucose. These German chemists simul-

taneously detected a protein-like compound in almonds which catalyzes
hydrolysis of amygdalin; this proteinaceous matter was called emulsine.
It was in fact the second enzyme to be described in biochemical litera-
ture. Since 1837 the chemical investigation of cyanogenic plant con-
stituents advanced slowly and rather irregularly. In fact, the four
periods sketched in Tables 2-4 can be discerned.
As a result of the endeavour of many scientists working during
the four periods outlined in Tables 2-4 the chemistry, biochemistry
and biology of cyanophoric .plant constituents became known to some
extent.
Cyanogenic Compounds as Systematic Markers in Tracheophyta 197

Chemistry of Cyanogenic Compounds

Formally the four chemical types of cyanophoric compounds re-
presented in Fig. 2 may be distinguished.
For the plant taxonomist, however, biogenetic pathways and their
resulting products a.re more reliable characters than mere structural
details and chemical reactivities of the individual plant metabolites,
which allow their chemical typification.

R, [J '" 0
R,
)CH-~H-COOH _ )CH - ~H - COOH
R2 NH2 R2 NHOH

Amino acid N - Hydroxyamino acid

Aldoxim Cyanidtt
n

R, /O-Glyc.
m 'C
/'
R2
"- CN
0( - H ydrox ycyanidtt Cyonogttnic glyco.r.idtt
(=Cyonohydrin)

Fig. 3. The biosynthesis of cyanogenic glycosides (chemical type I; cf.

Fig. 2).- I: this reaction is probably catalyzed by highly substrate-specific
enzymes which, however, usually do not discern between valine and iso-
leucine; the enzymes catalyzing the reactions have not yet been isolated.
II: enzymes catalyzing this reaction have not yet been characterized.
III: more or less substrate·specific ghicosyltransferases have been isolated
from several cyanophoric plants

Biochemistry of Cyanogenic Compounds

ROSENTHALER (1923) suggested that the amino acids phenylalanine,
tyrosine and valine might be the precursors of prunasin, dhurrin and
linamarin. In the meantime this hypothesis was extended and refined
by the experimental work of many biochemists. Today several steps
of cyanoglycoside synthesis by plants are rather well known (Fig. 3).
198 R. HEGNAUER:

R1, /CN
/C
R2 'O-Glyc.
Biosynthetic Rr R2 Sugar
Group (0 Glyc.)

0 - - -- ..
Phenylalanine Glucose Prunasin uod Sambunigrin
-H
A Gentiobiose Amygdal in
Vicinose Vicianin
-- - --
- - - - -- - -- - - - - - ---
HoO
Primverose J.ucumin
(see fig. 5)
-
-H Holocalin and Zierin

Tyrosine R) = H: Taxiphyllin and

B
(see fig. ~) ~O-O- -H Glucose

R) =
Dhurrin
Glucosyl: Proteacin

-- -- - --- -- - -_.
Valine and

- -- -----
CH3 - -C") Glucose Linamarin
Isoleucine 1----- --
C C2H S - -C"3 Glucose Lotaustralin

-II Glucose Dihydroacaci pe tal i n

~" -

- - fo-- - - - - - - - - - - - - - -
CH) -
Leucine _C!!;l_
~---
D CH 2 - C- -II Glucose Acacipetalin

.. -- - ":"l. - - - - -- ---- ----------

-
I
(see fig. 7)
CH 2 c- -H Glucose R4 = H: Cardiospermin
I
C"2 DR 4 R4 = .p,-Hydroxybenzoyl:
~-CN-glucoside

Cyclo-
pentenyl-
glycine?
-w -
- C" _
~-'"'-'",-
CH-CHOH-CH 2 -
}
Rr +R2
Glucose
Glucose
Deidaclin
Barterin
E - CH _ CH-CHOH-CHOH - Glucose Gynocardin
. (see fig. 8)

Fig. 4. The 5 biosynthetic main groups (A-E) of cyanophoric constituents

of vascular plants. N. B.: References to other figures in column 1 should
be corrected as follows: group A and B (see Fig. 6); group D (see Fig. 8);
group E (see Fig. 9). In column 4 read vicianose instead of vicinose

According to the amino acids involved five main biosynthetic

groups of cyanophoric compounds have been demonstrated to occur
in Tracheophyta (Fig. 4) and an additional group is present in :Fungi
(Fig. 5).
The greatest number of presently known cyanogenic glycosides
belongs to the aromatic groups A and B (Fig. 4). The differences be-
Cyanogenic Compounds as Systematic Markers in TracheQphyta 199

COOH COOH
I I
HO-C-CN HO-C-CN
I I
CH3 H

cyanohydrin of cyanohydrin of
pyruvic acid glyoxylic acid

Fig. 5. Cyanogenic constituents of Fungi (TAPPER & MACDONALD 1974)

11\ CN CN
w I I
c
0 R,-O-C-H H-C-O-R,
u
>-
c;,,....,
<:1:
u II R2¢ OR2
'i: rf RJ
w '-" ~
e0
';; C~anOhYdrin with Cyonohydrin with
c {S -Configuration (R) - Configuration
g
c
UJ (L- series of glycosides) (D-series of glycosides)

Sambunigrin : R,=Clucosyl ; Prunasin : R,=Glucosyl ;

'"
." R:z=~=H R2=~=H
·f Zierin : R,=Glucosyl ; =
Holocalin : R, Glucosyl;
w R2 =OH; R3=H R;r=OH; R3-H
w Amygdalin: R,= Gentiobiosyl ;
~ R:z .. R3 =H
< Vicianin : R, =Vicianosyl ;
Q.
:> ~=~=H
...0 Lucumin : R," Primverosyl ;
~ R:z= ~=H

Dhurrin : R,= Glucosyl ; Taxiphyllin : R,= Glucosyl ;

R2 =H'; R3 =OH R2=H: R3 =OH
~ Proteacin : R, = Glucosyl ;
tw R2 =H; R3=Glucosyloxy
-- ----- - - - - - - - - ---

o
--------------- --:
w
.....,
ro '0"'"
'"
CD
CN
I
Q.
:>
0
CHOH
L. Nandina - Glucoside unstable
~ Goodia - Glucoside cyan,o-'
, phorlc
0- Glucosyl compounds

Fig. 6. The presently known aromatic cyanogenic glycosides (groups A

and B of Fig. 4)

tween individual compounds relate to the configuration of the agly-

cones, the nature of the sugars, and in group B to the positions of the
attachment of glucose (Fig. 6). Another variant of group B is repre-
200 R. HEGNAUER:

sented by triglochinin (Fig. 7) which was shown to be an isomerized

seco-dhurrin (or seco-taxiphyllin).
Only two glucosides, linamarin and lotaustralin, are known from
group C (Fig. 4). They often occur together suggesting the already

C)= C,. . . CN
H

HO-O--~-CN HOOC
HOOC O-Glue.
o -Glue.
Dhurrin or (and) Taxi- Trigloehinin
phyllin

Fig. 7. Triglochinin belongs to the tyrosine-derived cyanogenic glucosides

(group B of Fig. 4). 1: the sequence of reactions leading to triglochinin
involves ring fission; therefore triglochinin is a so-called seco-compound

- I
Cyonophorie R, R2
I
compounds

Aeoeipetolin H Glueosyl

r"
Cordiosprzrmin OH Glueosyl

NC
X
Dihydro -
O-Glue. NC OR2
Sorborio-
CN-Glucosidrz O=C

(]
0, Glueosyl

ocoeiprztalin OH
Cyonolipid s O-CO-[CH2J-CH
,e 3
CO-[CH,J;;,CH,

Fig. 8. The recently di!lcovered and rapidly growing group of leucine-

derived cyanogenic compounds (group D, Fig. 4). Concerning cyanolipids
it should be noted that they may be spontaneously cyanophoric (cyano-
hydrin hydroxyl not esterified) or non-cyanogenic (no hydroxyl in IX-posi-
tion). Sorbaria-CN -glucoside according to N AHRSTEDT (1976)

mentioned (Fig. 3) lack of specificity of the ammo acid aminohydr-

oxylase concerned.
Group D (Fig. 4) comprises momentally 4 glucosides and several
cyanolipids (Fig. 8). Most probably the number of leucine-derived
cyanophoric plant constituents will increase considerably in future.
The strange cyclopentenoid representatives (Fig. 9) of group E
of Fig. 4 form a puzzling cluster of cyanophoric glucosides. Theoretically
Cyanogenic Compounds as Systematic Markers in Tracheophyta 201

they could be derivatives of the non-proteinogenous amino acid cyclo-

pentenyl glycine. The latter once was detected in seeds of a species
of Hydnocarpus, a genus known for the presence of gynocardin-type
glucosides. At present, however, experimental research abou:t the
biosynthesis of these compounds is still lacking.

GlucosidlZS R, R,

DlZidaclin (= H H
~O-GIUC. TlZtraphyllin A)
R, eN Barterin (= Tetra- H OH
Ro phyllin B)
Gynocardin OH OH

Fig. 9. The presently known representatives of the gynocardin group

(E in Fig. 4) of cyanogenic glucosides

Occurrence, Function, Distribution and Systematic Meaning

of Cyanogenesis and Cyanogenic Compounds in Vascular Plants
On p. 196 the number of presently known cyanophoric species
of vascular plants is given as 2056. This is a very rough estimate. There
are several reasons why more precise statements are impossible, some
of which deserve to be mentioned shortly in this context.
(a) Formerly most phytochemists paid little attention to serious
plant identification; a considerable number of taxa is probably er-
roneously reported as cyanophoric.
(b) Because of the enormous amount of nomenclatural and taxo-
nomic synonymy many taxa are reported twice or several times under
different names as cyanophoric in original literature and reviews.
(c) Research workers interested in cyanophoric plants applied
different methods to demonstrate cyanogenesis. Some of the proce-
dures used lack specificity for HCN. Consequently the literature about
cyanophoric plants contains a number of false positive reports.
(d) The sensibility of analytical methods for the detection of HCN
varies widely and depends upon many factors. In our context, .a plant
:;hould release at least 10-20 mg HCN per kg fresh weight. This implies
that one gram of fresh plant material usually reacts positively when
the GUIGNARD-MIRANDE- or (and) the FEIGL-ANGER-MIRANDE-test
(FIKENSCHER & HEGNAUER 1977) are applied. If more sensitive proce-
dures are used, most or possibly all plants can be shown to release
trace amounts of HCN (RosENTHALER 1923, GEWITZ et al. 1974). From
the systematic and ecological point of view, however, only accumula-
tion of cyanophoric constituents represents a character worthwhile
of con:;ideration.
Ecological Aspects of plant cyanogenesis

Plant Predators

II II
(Phytopathogenic microorganisms; Plants:
.Phytophagous animals):

A: Species sensitive Examples: .

to HCN Trifolium hybridum
Lotus uliginosus
Sambucus ebulus
Campanula scheuchzeri

,-------
I Evolution of
I resistance, e.g.
, by production
I of HCN metabolizing
_ _ _ _ _ _ _ _ .J'
, enzymes
L..

Intermediate situation
(some cyanogenic phenotypes originate
in populations by mutations)

B: Species not sensi- ~ (t) ~7:

~ ) Ir-----------,
Fixation of cyano- , ~
tive to HCN ~ I genesi~ (or tart of i~s I~
'-------~----®-I1---- ~s:)~ b.: ~e~e~t:o~J
Cyanogenic Facultatively
Species cyanoaenic species
(i.e. species poly-
e morphous and (or)
polytypic with regard
to cyanogenesis

Examples: Examples:
Helica (many species) Trifolium rep ens
G'i'Y'CHi a maxima Lotus corniculatus
Cystopteris fragilis Pteridium aquilinum
Sambucus nigra Taxus baccata
Campanula coch- Juniperus oxycedrus .

<i>and e leariifolia Sambucus racemosa s.l.

Campanula rotundifolia

Fil!. to. Possible role of cyan0l!enesis in vascular plants (modified after

JONES 1972, 1973). (+) Plant potentially suitable as host (food). (-) Plant
not suitable as host (food). 1. e.l!. plants which produce cyanogenic gly-
cosides, but no splitting enzymes. 2. Evolution of very efficient mechanisms
preventing autointoxification by HCN (e.g. rapid conversion of HCN to
asparagine via ~.cyanoalanine). This mechanism is fixed within the taxon
because it seems to be efficient in all habitats occupied by the taxon.
3. Evolution of similar detuxification. mechanisms which, however, are
still subject to selection, because in sume biotopes part of the mechanism
may be inhibited, and the system, therefore, becomes deleterious to the
taxon concerned
R. HEGNAUER: Cyanogenic Compounds as Systematic Markers 203

(e) Reports in literature are inconsistent for a number of taxa.

Besides factors such as mentioned sub (a) to (d) other possibilities
should always be considered as eventual explanations for contradictions.
Many plants accumulate cyanogenic compounds only in one or a 'few
organs; if different plant parts are tested, the results may be different.

Table 5. Variation of cyanogenesis within one species: Field and herbarium

observations with Centaurea scabiosa L. (sensu lato)

Collection Cyanogenesis-tests performed

Localityl
year in the field 2 in the laboratory 3

1946 Kt. Tessin, CH 1 very strong

1947 Kt. Tessin, CH 1 negative
1947 Kt. Graub., CH 1 negative
1962 Tolkamer, NL 1 negative
1966 Kt. Glarus, CH 1 negative
1967 Kt. Glarus, CH 1 negative
1967 Kt. Graub., CH 1 negative 1 negative
1968 Kt. Tessin, CH 1 negative
1969 Alpes Maritimes, F 1 negative
1969 Kt. Glarus, CH 2 negative 1 negative; 1 very
strong
1972 Kt. Glarus, CH 3 negative, 3 negative; 1 very
1 positive strong
1972 Kt. Graub., CH 1 positive 1 very strong
1973 Kt. Graub., CH 1 positive 1 very strong
1974 Kt. Glarus, CH 6 negative, 5 negative; 6 very
5 positive strong
19.74 Kt. Graub., CH 6 positive, 6 very strong;
1 negative 1 negative

1 Most collections and tests performed in Switzerland; "Kt. Glarus"

(includee lake Walensee); "Kt. Graub." means valley of the Rhine between
Chur ~nd Sargans.
2 GUIGNARD-MIRANDE-Test; mostly without adding emulsin.
3 Leaves of herbarium material; some droplets of water and some mg
of emulsin added; observation after 4 hours; GUIGNARD-Test.

Moreover, accumulation of cyanogenic plant constituents is often

conditioned in a taxon-specifk manner by ecological factors and the
age of the plant investigated and its different parts. Therefore the same
plant may react positively or negatively depending on the timing of
an investigation. Finally, genetical variation within species with
regard to cyanogenesis is much more frequent than is generally assumed
(FIKENSCHER & HEGNAUER 1977). Genetical polymorphism and poly-
typism enable population geneticists and ecologists to deal with cyano-
204 R. HEGNAUER:

genesis and its function (JONES 1972, 1973; Fig. to). At the same time
variability of cyanogenesis within species may be a significant feature
in experimental plant systematics and a nuisance when used as a
character in (X-taxonomy (e.g. Table 5).

Distribution and Systematic Implications

Without any doubt the number of cyanogenic taxa and cyanogenetic
compounds known will increase considerably in the future. Nevertheless
the facts available at present and summarized in the following suggest
that the biogenetical groups A to E represent characters worth to be
considered by plant taxonomists.
Pteridophyta. Cyanogenesis has been reported in literature for re-
presentatives of the genera Asplenium, Blechnum (Lomaria), Cystopteris,
Dryopteris, Lindsaea, Polypodium, Pteridium, Schizaea and Thelypteris,
and observed by the author in Microgramma lycopodioides (L.) COPELlo
Prunasin was isolated from species of Cystopteris and Pteridium, and
vicianin from several species of Davallia. Within the range of Pterido-
phyta only some Ferns seem to produce and accumulate cyanophoric
compounds and phenylalanine seems to be the only precursor used
for this purpose.
Gymnospermae. Cyanogenesis occurs in needles and young twigs of
Juniperus oxycedrus L., Metasequoia glyptostroooides Hu et CHENG
and 4 species of Taxus. In the taxa mentioned tyrosine is the precursor
of the cyanogenic glucoside present.
Angiospermae. In the main group of modern vascular plants cyano-
phoric constituents occur very erratically and all compounds known
nowadays are represented. One of its main groups, Liliopsida (formerly
Liliatae), was already analyzed by HEGNAUER (1973b). Cyanogenic
taxa are relatively frequent in Araceae, Juncaceae, Juncaginaceae
(incl. Lilaea), Poaceae (= Gr:amineae) and Scheuchzeriaceae. Outside
these families cyanophoric species are very rare. From Commelinaceae,
e.g., one species, Tinantia erecta SCHEIDEW., was demonstrated to be
cyanogenic by several authors. Tyrosine-derived cyanophoric glucosides
(dhurrin or its enantiomer taxiphyllin: triglochinin) were observed
in all species of Liliopsida which were investigated in some detail.
Obviously Liliopsida use tyrosine only for the synthesis of cyanogenic
compounds.
Most of the presently known cyanogenic plant constituents are
restricted to the second major group of Angiosperms, the Magnoliopsida
(formerly Magnoliatae). A detailed analysis of the 7 subclasses (TAKH-

1 See Addenda.
Cyanogenic Compounds as Systematic Markers in Tracheophyta 205

TAJAN 1973) of Magnoliopsida reveals many gaps in the chemical

identification of cyanophoric plant constituents, and some biochemical
tendencies which the present author esteems significant to plant taxo-
nomy. The presently known facts are summarized next.
The compilation makes use of the system of TAKHTAJAN
(1973) and is based on HEGNAUER (1960 and 1962-1973a and unpublished
own observations) and GIBBS (1974; only positive results of GIBBS
himself utilized). In the compilation subclasses, superorders, orders,
families and, in brackets, the number of genera are given. The
following abbreviations are used: n.i. = cyanogenic compounds
not yet characterized; p.c. = cyanogenesis should be confirmed;
P = phenylalanine-derived compounds (Fig. 6); D = tyrosine-derived
glucosides (Fig. 6); T = triglochinin variant of D (Fig. 7); Li = bio-
genetical group C (Fig. 4); L = leucine-derived glucosides (Fig. 8);
CN-Lip = cyanolipids (Fig. 8); G = gynocardin group of glucosides
(Fig. 9); ? after group of comp'aunds (e.g. G?) means that the identifi-
cation should be confirmed. Statements concerning groups of cyano-
phoric constituents within families do not mean that their presence
was established for all species and genera known to be cyanophoric.
Magnoliaceae (2) T, e.g., refers to the presence of triglochinin in Lirio-
dendron tulipifera L.; the cyanophoric species L. chinensis SARG.
and Magnolia sprengeri PAMPAN have not yet been thoroughly in-
vestigated.
Magnoliidae-Magnolianae: Magnoliales: Magnoliaceae (2) T, De-
generiaceae (1) n.i., Annonaceae (1) n.i., Canellaceae (1) n.i., Winteraceae
(1-2) n.i.; Laurales: Calycanthaceae (2) n.i., Lauraceae (1) p.c.
Ranunculidae-Ranunculanae: Ranunculales: Menispermaceae (1)
n.i., Ranunculaceae (6-11) D and T, Nandinaceae (1) D, Berberidaceae (1)
n.i.; Papaverales: Papaveraceae (4) T (Eschscholtzia) and Li? (Papaver
nudicaule), Fumariaceae (1) n.i.
Hamamelidae-Hamamelidanae: Trochodendrales: Trochoden-
draceae (1) n.i.; Hamamelidales: Platanaceae (1) n.i. 1 ; U rticales: Ulmaceae
(1) n.i., Moraceae (1) n.i.; Juglandales: Juglandaceae (1?; possibly false
positive GUIGNARD-test reaction). .
Caryophyllidae-C!aryophyllanae: Caryophyllales: Gyrostemonaceae
(1) n.i., Molluginaceae (1) n.i., Cactaceae (1) n.i., Amaranthaceae (3) n.i.,
Chenopodiaceae (5) n.i.
Dilleniidae-Dillenianae: Theales: Clusiaceae (1) n.i.; Violales:
Flacourtiaceae (15) G; Passiflorales: Passifloraceae (incl. Paropsieae)
(6-8) G, Turneraceae (3) G?, Malesherbiaceae (1) G?; Cucurbitales: Cu-
curbitaceae (1) n.i.; Capparales: Capparidaceae (3) n.i., Tovariaceae (1)

1 See Addenda.
206 R. HEGNAUER:

n.i., Brassicaceae (10) n.i., Resedaceae (2) n.i. (in Oapparales cyanogenic
glycosides and [or] hydrocyanic acid are probably by-products of
glucosinolate synthesis or decomposition).-Ericanae: Ericales: Erica-
ceae (3) n.i., Epacridaceae (2) n.i., Pyrolaceae (21; possibly false positive
GUIGNARD-test reactions); Ebenales: Sapotaceae (4) P.-Malvanae: Mal-
vales: Elaeocarpaceae (2) P1, Sterculiaceae (1-2) n.i., Malvaceae (1) n.i.;
Euphorbiales: Euphorbiaceae (11) Li, D and possibly P.
Rosidae-Rosanae: Saxi/ragales: Ounoniaceae (1) p.c., Davidsoniaceae
(1) p.c., Escalloniaceae (1) n.i., Iteaceae (1) D 1, Grossulariaceae (1) D1,
Hydrangeaceae (2) P and possibly D, Orassulaceae (7-8) n.i., Saxi-
/ragaceae (2) n.i. ; Rosales: Rosaceae (many genera and species of Spiraeoi-
deae, Prunoideae and Maloideae; from Rosoideae only Kerrieae, Adeno-
stomeae and Oercocarpinae are cyanogenic) P in Prunus, Maloideae
and a few species of Spiraeoideae and L in Sorbaria (compare Fig. 8)
and possibly in other taxa of Spiraeoideae and tp!l ~ew cyanogenic
taxa of Rosoideae mentioned above; Fabales: Mimosaceae (4) P and L
and possibly Li, Oaesalpiniaceae (3) P, Fabaceae (approximately 20) P,
D and Li; Nepenthales: Droseraceae (31; probably false positive GUIG-
NARD-test reactions).-Myrtanae: Myrtales: Myrtaceae (1) P, Melasto-
mataceae (3) P1, Oliniaceae (1) Il, Onagraceae (3) n.i., Lecythidaceae (1)
n.i.; Hippuridales: Haloragaceae (3) n.i.-Rutanae: Rutales: Rutaceae
(2) P; Sapindales: Sapindaceae (14) L and CN-Lip ; Geraniales: Linaceae (1)
Li; Polygalales: Tremandraceae (2) n.i.-Aralianae: Oornales: Apicaeae (1)
D (needs to be confirmed).-Celastranae: Oelastrales: Aqui/oliaceae (1)
p.c., lcacinaceae (1) P1, Salvadoraceae (11; see remarks sub Oapparales),
Oelastraceae (1) n.i., Oorynocarpaceae (1) n.i.; Rhamruiles: Vitaceae (1)
n.i.; Santalales: Olacaceae (2) P.-Proteanae: Proteales: Proteaceae
(10D).
A steridae-Lamianae : Dipsacales: Oapri/oliaceae (1-2) P; Gentiana-
les: Gentianaceae (3) n.i., Apocynaceae (1) n.i., Asclepiadaceae (3) P'I,
Rubiaceae (5) P1; Polemoniales: Oonvolvulaceae (2) P1 and Li?, Bo-
raginaceae (3) CN-Lipl; Scrophulariales: Solanaceae (1) n.i., Scrophu-
lariaceae (1-2) P, Myoporaceae (1) P, Acanthaceae (3) n.i.; Lamiales:
Verbenaceae (2) n.i., Lamiaceae (3) n.i.-Asteranae: Oampanulales:
Oampanulaceae (1) n.i., Goodeniaceae (1-2) n.i.; Asterales: Asteraceae
(15-20) P and Li.
Consultation of the foregoing compilation of cyanogenic vascular
plant taxa reveals a number of systematically interesting fact.s. These
are summarized in the following conjectures.
(a) Each of the three major taxa of vascular plants, Pteridophyta
(phenylalanine only), Gymnospermae (tyrosine only) and Angiospermae

1 See Addenda.
Cyanogenic Compounds as Systematic Markers in Tracheophyta 207

(all pathways illustrated by Fig. 4), has its own biochemical characters
with regard to production of cyanophoric compounds.
(b) The aromatic glycosides (Fig. 6) occur very erratically in vascular
plants, but are present in all their divisions, classes and subclasses.
Most probably they represent the phylogenetically oldest cyanophoric
plant constituents among the compounds treated in this paper.
(c) Within Angiosperms Liliopsida and the first two subclasses
of Magnoliopsida, which represent together WETTSTEIN'S Polycarpicae,
seem to use exclusively the tyrosine-pathway (Figs. 4, 6, 7) for the
synthesis and accumulation of cyanogenic constituents. This feature
might be systematically significant.
(d) Within Magnoliopsida phenylalanine-derived compounds are
the more frequent and the more wide-spread aromatic cyanogenic
constituents, if Polycarpicae are excluded.
(e) Group C (Fig. 4) seems to parallel group A with regard to erratic
occurrence and wide distribution within Magnoliopsida. The fact
that linamarin and lotaustralin generally occur together suggests that
one or more enzymes engaged in their biosynthesis do not discriminate
between valine and isoleucine. This might be caused by a structural
similarity of the two amino acids, i.e. branching of the chain at C-3.
(f) Leucine-derived compounds (Figs. 4 and 8) occur in leaves of
Rosaceae (Sorbaria: Spiraeoideae), Mimosaceae and Sapindaceae as
glucosides, and in seeds of many Sapindaceae as cyanolipids. Perhap's
all taxa mentioned belong to one and the same evolutionary
line of Magnoliopsida as suggested long ago by HALLIER f.
(g) The gynocardin group of constituents (Fig. 9) is highly charac-
teristic of Passiflorales and of two tribes of Flacourtiaceae (Oncobeae
and Pangieae). This character represents an extremely va-
luable systematic marker within Dilleniidae.
(h) Contrary to the statement of HEGNAUER (1960) cyanogenesis-
if adequately and carefully used as a character in plant systematics-
may prove to be of considerable value to a natural classi-
fication of plants, even at the higher systematic levels.
Seemingly the cyanophoric constituents show what was called by the
present author (HEGNAUER 1971) an asystematic distribution of
secondary plant metabolites. Nevertheless it is my conviction that a
thorough analysis of such characters can reveal tendencies of bio-
chemical evolution and, therefore, ultimately might furnish argu-
ments which can he used in efforts to improve presently accepted
classifications.
(i) The facts known to-day suggest that more than one biosynthetic
group (Fig.4) of cyanophoric compounds occur only in very large
genera or families belonging to Dilleniidae (Eupkorbiaceae) , Rosidae
208 R. HEGNAUER:

(Rosaceae, Acacia, Fabaceae) and Asteridae (Asteraceae). In such taxa

the chemical character accumulation of cyanogenic compounds
is taxonomically significant at infrafamiliar levels. Linamarin, e.g.,
is a character of many Loteae, Trilolieae and Phaseoleae in Fabaceae
and of Oalenduleae in Asteraceae. In Rosaceae most cyanophoric taxa
with the basic chromosome number x = 9 do accumulate neither
amygdalin nor prunasin (e.g., species of Exochorda, Gillenia, Sorbaria,
Kerria, Neviusia, Rhodotypos, Ooleogyne, Adenostoma and Oercocarpus).
Whether most of them contain compounds similar to the Sorbaria-CN-
glucoside (NAHRSTEDT 1976) has still to be demonstrated.
(k) The erratic occurrence of cyanogenesis in Boraginaceae (cyano-
lipids in the seed oil of Oordia verbenacea 1 ; not yet identified compounds
in seedlings of Borago ollicinalis) deserves the attention of phytochemists
and taxonomists. The systematic position of this family of Asteridae
is still uncertain.

I thank Dr. A. NAHRSTEDT, Institut fur Pharmazeutische Biologie,

Freiburg i. Br., for information and a manuscript copy on Sorbaria-CN-
glucoside, and Dr. M. ETTLINGER, Chemical Laboratory II, University of
Copenhagen, for a stimulating discussion of likely biosynthetic routes to
gynocardin-type glycosides. I am obliged to Dr. B. HENNIPMAN, Rijks-
herbarium, Leiden, for the identification of the Brazilian fern Microgramma
Iycopodioides (L.) COPEL.

References

BUTLER, G. W., 1969: Metabolism of cyanoglucosides, mustard-oil glyco-

sides and selenium-containing compounds in plants. Proc. Roy. Austra-
lian Chern. Inst. 36, 65-70.
CONN, E. E., 1969: Cyanogenic glycosides. Agric. Food Chern. 17,519-526.
-- 1973: Biosynthesis of cyanogenic glycosides. Biochem. Soc. Symp. 38,
277-302.
DILLEMANN, G., 1958: Composes cyanogenetiques. In: Handbuch der
Pflanzenphysiologie (RUHLAND, W., et aI., eds.), Band VIII (redigiert
von MOTHES, K.), 1050-1075. Berlin-Gottingen-Heidelberg: Springer.
EYJOLFSSON, R., 1970: Recent advances in the chemistry of cyanogenic
glycosides. Fortschr. Chern. Org. Naturstoffe 28, 74-108.
FIKENSCHER, LUCIE H., and HEGNAUER, R., 1977: Die Verbreitung der
Blausaure bei den Cormophyten. 11. Mitteilung. Ober die cyanogenen
Verbindungen bei einigen Compositae, bei den Oliniaceae und in der
Rutaceen-Gattung Zieria. Pharm. Weekblad 112, 11-20.
FLUCKIGER, F. A., 1883: Pharmakognosie des Pflanzenreiches, 2. AufI.
Berlin: R. Gaertner's Verlagsbuchhandlung.
GEWITZ, H.-S., et aI., 1974: Presence of HCN in Chlorella vulgaris and its
possible role in controlling the reduction of nitrate. Naturo 249, 79-81.

1 See Addenda.
Cyanogenic Compounds as Systematic Markers in Tracheophyta 209

GIBBS, R. D., 1974: Chemotaxonomy of Flowering Plants, 4 volumes.

Montreal: McGill-Queen's University Press.
HEGNAUER, R., 1960: Die systematische Bedeutung des Blausiiuremerk-
males. Pharm. Zentralhalle 99, 322-329.
1962-1973a: Chemotaxonomie der Pflanzen, Bande 1-6. Basel-
Stuttgart: Birkhauser Verlag (Systematic distribution of cyanogenesis
and cyanogenic constituents; consult the heading Cyanogenese in
indices).
1971: Pflanzenstoffe und Pflanzensystematik. Naturwissenschaften 58,
585-598.
1973b: Die Verbreitung der Blausaure bei den Cormophyten. 10. Mit-
teilung. Die cyanogenen Verbindungen der Liliatae und Magnoliatae-
Magnoliidae: Zur systematischen Bedeutung des Merkmals der Cyano-
genese. Biochemical Systematics 1, 191-197.
JONES, D. A., 1972: Cyanogenic glucosides and their function. In: Phyto-
chemical Ecology (HARBORNE, J. B., ed.), 103-124. London-New York:
Academic Press.
1973: Co-evolution and cyanogenesis. In: Taxonomy and Ecology
(HEYWOOD, V. H., ed.), 213-242_ London-New York: Academic Press_
N AHRSTEDT, A., 1973: Cyanogene Glykoside in hoheren Pflanzen. Pharmazie
in unserer Zeit 2, 147-155.
- Ul76: Ein neues Cyanogenes Glykosid aus Sorbaria arborea (Rosaceae)_
Z. Naturforsch. 31 C, 397-400_
ROSENTHALER, L., 1923: Zur Priifung der Treubschen Hypothese. Bio-
chern. Z. 134, 215-224.
SEIGLER, D. S., 1975: Isolation and characterization of naturally occurring
cyanogenic compounds. Phytochemistry 14, 9-29.
SCHUTTE, H_ R., 1973: Biosynthese von cyanogenen Glykosiden und Senf-
olglucosiden. Fortschr. der Botanik 35, 103-119.
TAKHTAJAN, A., 1973: Evolution und Ausbreitung der Bliitenpflanzen.
Jena: VEB Gustav Fischer Verlag.
TAPPER, B. A., and MACDONALD, M_ A., 1974: Cyanogenic compounds in
cultures of a psychrophilic basidiomycete (snow mold). Canad_ J. Micro-
bioI. 20, 563-566.
- and REAY, P_ F_, 1973: Cyanogenic glucosides and glucosinolates
(mustard oil glycosides)_ In: Chemistry and Biochemistry of Herbage
(BUTLER, G_ W., and BAILEY, R. W., eds.), Vol. 1, 447-476. London-
New York: Academic Press.
TSCHIERSCH, B., 1967: Blausaure und Blausiiureglykoside, eine tTbersicht.
Pharmazie 22, 76-82.

Addenda
(1) Cyanogenesis in ferns: The following genera should be added
to the list given on p. 204: A ctinopteris , Gheilanthes, Pteridanetium,
Microlepia, Pteridrys, Stenochlaena, Gampyloneurum, and Phlebodium,
and the number of presently known cyanogenic species approaches 50
(HARPER, N. L., et. al., Phytochemistry 15,1764-1767 [1976]; HARPER,
N_ L., SEIGLER, D. S., Econ. Botany 30, 395-407· [1976]; Proc. Okla.
Acad. Sci. 56, 95-100 [1976]).
Plant Syst. Evol.. Suppl. 1 14
210 R. HEGNAUER: Cyanogenic Compounds as Systematic Markers

(2) Platanus contains triglochinin and dhurrin (FIKENSCHER, L. H.,

RUIJGROK, H. W. L., Planta Medica 31, 290-293 [1977]).
(3) Cyanolipids do not occur in the seed oil of Cordia verbenacea
(SEIGLER, D. S., Biochem. Systematics Ecology 4, 235-236 [1976]).

Address of the author: Prof. Dr. R. HEGNAUER, Laboratorium voor Ex·

perimentelePlantensystematiek, Schelpenkade 148, Leiden, The Netherlands.
Plant Syst. Evo!., Supp!. 1,211-226 (1977)
© by Springer-Verlag 1977

Departamento de Botanica,
Faculdade de Ci€mcias Medicas e Biol6gicas de Botucatu, Brazil

Some Aspects of Beetle Pollination in the Evolution

of Flowering Plants
By

Gerhard Gottsberger, Botucatu

Abstract: Pollination by beetles seems to have strongly influenced

the evolution of angiosperm flowers. Beetles are a predominant group of
potential visitors and pollinators of flowers since earliest times, so can-
tharophily is apparent within groups of flowering plants of the most diverse
evolutionary levels. In the subclass Magnoliidae, cantharophily is a do-
minant feature of many families, but even in these archaic groups a more
open, unspecialized type of beetle pollination and a more specialized one
can be distinguished. Specialization here was probably connected with
increase in flower size, numerical increase and grouping and flattening of
the sexual organs, etc. The strongly protogynous flower attracts beetles
through imitative odours, to which the insects are already conditioned in
their other activities. Secondary polyandry, such as occurs in the more
basic groups of Rosidae, Dilleniidae and Oaryophyllidae, is in many cases
related to cantharophily and may find its functional explanation in this
mode of pollination. This probably somewhat more recent radiation into
beetle-pollination might have caused the stamens to increase in number
in order to save some of them from the crude visitors. Cantharophily in
secondarily polyandrous groups is also frequently connected with the proto-
gynous condition of flowers and with odours which act directly on the
instincts of the visitors. During these new waves of flower-biological radia-
tion other insects besides beetles must have already been in existence. The
pollination of primitive Rosidae, Dilleniidae and Oaryophyllidae is there-
fore much less exclusive compared with the Magnoliidae,· viz., beetles
and other insects often frequent flowers jointly. Since beetles have con-
tinued as the predominant insect group until today, cantharophily can be
observed also in advanced groups of angiosperms. In this case, cantharo-
phily is no longer a sign of primitiveness as it is in the Magnoliidae and to
some extent probably also in secondarily polyandrous groups, but a rela-
tively recent adaptation into a still existing ecological niche.

Introduction
In a meeting such as the present one in which the eV0lution and
classification of higher categories is to be illuminated from different points
of view, it seems appropriate to speak about the functional aspects of
flowers.
212 G. GOTTSBERGER:

It has been shown many times and by many examples that morpholo-
gical, anatomical, chemical and physiological characteristics of flowers
largely correspond to the necessities of pollination and that many
of these features must have evolved in response to the agents
involved.
From the time that the idea of co-evolution of plants and of pol-
lination agents gained probability, it was considered significant to study
these correlations for at least two reasons. First, functional aspects
underline structural ones and so may show us the ecological background
for flower characteristics. Second, reproductive organs of plants and
their function can be seen in a historical context. The different pol-
lination agents (excluding here the relatively timeless wind and water)
evolved in different ages. For example, within insects, beetles are
especially old, but butterflies and social bees are more recent. The
appearance of new groups of potential agents of pollination certainly
had very strong effects upon flowering plants. Waves of radiation into
new ecological conditions must have occurred during the historical
development of angiosperms.
It was certainly tempting to use flower-biological concepts for
phyloge.netic reasoning. Thus, we often say without further consideration
.that because beetles are an old group, plants with cantharophily are
always derived from phylogenetically old stocks, or, since birds and
mammals are more modem animals, therefore we should at least suspect
that plants with pollination by birds and mammals should be considered
as derived phylogenetically from more modem groups, and so on.
This kind of reasoning was used to a large extent and probably
worked in many cases. However, we largely neglected to distinguish
between organisms which probably remained more or less in the original
condition, and others which perhaps because of a changing environment
had to adapt to new conditions. Today, flower-biological concepts are
changing to a form that better fits into the whole framework.

I shall now attempt to demonstrate some of these Hew concepts as

I personally see them at the moment: The main example will be cantharo-
phily, which is a feature of many primitive angiosperms, but which also
accompanies the differentiation of flowering plants up to its most evolved
members. The main part of t.his paper will be to show you the supposed
evolution of flowers as far as cantharophily is concerned. The second
subject of this symposium, classification, will not really be considered,
although cantharophily would be a good example to use in dealing with
this question. Secondary polyandry, a distinctive feature in the more pri-
mitive members of some subclasses and used even as a differential-diagnostic
character is, in my opinion, also largely connected with beetle pollination
and may find its functional explanation through t.his mode of pollina-
tion.
Some Aspects of Beetle Pollination 213

Beetles and Beetle Flowers

The trouble with talking about beetles is that we know so little
about them. Our knowledge of the sense-physiology of beetles in general
and of flower beetles in particular is very rudimentary and not comparable
with what we know about honeybees, bumblebees, butterflies, moths
and even flies.
The orthognathic position (perpendicular to body axis) of mouth
parts, which is the original condition and common in the majority of
beetles, limits the length of the mouth parts (FAEGRI & VAN DER PIJL
1971: 114). Consequently such beetles are able to visit only relatively
flat flowers which are somewhat dish-, bowl- or brush-shaped. The
famous Nemognathus and others, because of their extremely prolonged
mouth parts, which are in a prognathic position (parallel to body axis),
are also able to explore deep-shaped flowers. But these are rare and
extremely specialized cases. On their flower-visits beetles may feed on
pollen, on floral organs like petals, tepals, stamens and carpels, or on
nectar as far as it is available. Very often they damage the flowers
considerably.
How are such beetles attracted by flowers? Data are accumulating
which show that most beetle flowers emit characteristic odours, which
are usually somewhat similar to those of fresh or rotten fruits, or aminoid,
similar to carrion, decayed fish, sperm, or the beetles themselves. Both
fruity and aminoid odours of flowers seem to stimulate strongly the
sexual instincts of beetles (PORSCH 1950: 290; GOTTSBERGER 1970; etc.);
copulation is induced and very often even egg deposition occurs within
the flowers.
It is very often said that the general syndrome of beetle pollination
and beetle flowers is rather unspecialized due to the lack of specialization
of the visitors. Some authors (VAN DER PIJL 1969: 91; FAEGRI & VAN
DER PIJL 1971) divide beetle pollinated flowers into two groups. The
more primitive one has larger flowers, frequently amorphic or haplo-
morphic sensu LEPPIK, and the second and more advanced one has small
flowers condensed in inflorescences. Protogyny is a common feature in
beetle flowers.

Beetle Pollination in Different Subclasses of Angiosperms

1 now would like to describe cantharophily as it appears in the
different subclasses of the Magnoliatae. In order to keep the discussion
short, I shall restrict myself to dicotyledons.
Having studied the phylogenetically old Magnoliidae, which ha,ve so
many characters in common with the assumed ancestors of the angio-
sperms, it has been learned that beetle pollination is a dominant and
very often exclusive feature within this group. But unfortunately,
214 G. GOTTSBERGER:

studies have not started at the right end and the knowledge now repre-
sented in textbooks refers to highly specialized cases of cantharophily.
Until recently we were aware of pollination methods only in the Eupo-
matiaceae, Calycanthaceae, Magnoliaceae, A nnonaceae , and Nymphaea-
ceae, which in my opinion are all cases of a more or less advanced type
of cantharophily.
But what is about pollination in Winteraceae? Winteraceae are
considered as being especially primitive because of their wood, their
unifacial stamens and carpels, their placentation, etc. (see GOTTSBERGER
1974). The high number of primitive characters in the Winteraceae is
perhaps explained partly by the fact that they are high palaeopolyploids
with the highest polyploid chromosome numbers yet known in angio-
sperms. Palaeopolyploids tend to preserve more primitive characters
than diploids (or low polyploids) which have diverged more actively
(EHRENDORFER et ai. 1968: 349).
For a primitive group we might expect also a primitive mode of
pollination. The only thoroughly investigated case of pollination in the
Winteraceae is that of Drimys brasiliensis (GOTTSBERGER et aI., in
progress). This species belongs to the New World section, Drimys, which
because of its morphology, anatomy, cytology and chemistry is con-
sidered more primitive than the Old World section Tasmannia (see
KUBITZKI & VINK 1967: 14; EHRENDORFER et al. 1968: 338), nowadays
treated by some authors as a separate genus, Tasmannia (see EHREN-
DORFER 1976: 229). The hermaphrodite flowers of Drimys brasiliensis
show a beetle pollination syndrome which, however, is different from
that of other cantharophilous M agnoliidae. The sexual organs are
exposed and unprotected during the whole anthesis in the permanently
open flower. The flower odour is not fruit-like or aminoid but sweet and
recalls a "typical" blossom odour. Such an absolute odour (FAEGRI &
VAN DER PIJL 1971: 87-89) probably has no "meaning" outside the
flower, and unless an insect has "learnt" this odour from the flower, it
will not start any reaction. As we will see later on, the more specialized
flowers in respect to beetle pollination, such as these of the Annonaceae,
Magnoliaceae, Eupomatiaceae, Calycanthaceae, and Nymphaeaceae, emit,
as far as it is known, so-called imitative fruit-like or aminoid odours
which "imitate" odours to which the insects are already conditioned by
instincts and by "experience" in their other activities, and which would
normally start an instinctive "meaningful" reaction (FAEGRI & VAN
DER PIJL 1971: 87). Flowers with such imitative odours frequently
attract a crowd of beetles. The flowers of Drimys brasiliensis, with
their sweet absolute odour attract only a few species of beetles, which
do not further harm the flowers but eat only the pollen.
Being acquainted with the pollination syndrome in other families
Some Aspects of Beetle Pollination 215

of the Magnoliales, Laurales, and Nymphaeales, I want to emphasize the

difference between their specialized type of cantharophily and the open,
obviously less specialized one manifested in Drimys brasiliensis.
A very fine case of specialized cantharophily is to be seen in the
Annonaceae. In species of Guatteria, for example, the open but still
developing flowers remain dish-shaped and greenish during several
weeks. At anthesis the petals turn yellowish and fold together to form
a pollination chamber. It is from this moment on or a little later that
the flowers start to emit a strong fruit-like odour which particularly
attracts beetles that normally inhabit fruits and feed on them. This
kind of odour can be considered an imitative one. The fruit- beetles act
and react in the flower as they do on their normal substrate, feeding on
the soft odoriferous perianth or stamens, copulating inside the dark
pollination chamber and depositing their eggs. The closing of the flower
has a somewhat protective effect, as larger and even more destructive
beetles have difficulties in entering the flower interior with the sexual
organs and thus very often only gnaw the soft perianth. Further, the
dark, fragrant floral chamber possibly has similar properties to the
dark, fragrant interior of fruits where the beetles normally live, thus
attracting these fruit beetles quite precisely by deceit (GOTTSBERGER
1970). Further examples of an advanced mode of pollination in which
imitative fruit- or also carrion-like odours act upon the instincts of fruit
or carrion beetles can be found in most other Annonaceae species (see
WESTER 1910; UPHOF 1933; CORNER 1940; ZIMMERMAN 1941 ; VAN DER
PIJL 1953; KRAL 1960; GOTTSEERGER 1970). The closed flowers of
Annonaceae, contrary to the open, unprotected flowers of Drimys,
receive frequent visits of mostly gnawing and thus destructive pol-
linators; it is noticeable that their many stamens and carpels with hard
heads are very densely aggregated on the floral axis before disintegration.
Species of Magnolia have been recently investigated in respect to
their pollination (THIEN 1974). The beetles which enter the flowers to
feed on secretions of the gynoecium, stigmatic papillae, petal secretions
or pollen, may also cause considerable damage (KNUTH 1904: 303;
HEISER 1962: 262). I have already suggested before (GOTTSBERGER
1974: 463) that most Magnolia species may function by deceit-attracting
fruit-eating or otherwise specialized beetles. This is now partly confirmed
(THIEN et al. 1975); not only are some of the pollinators beetles that
visit decaying fruits, but the floral odours of Magnolia contain many
methyl esters which are present also in the odours of many fruits.
A very similar mode of pollination characterizes the Brazilian
Magnoliacea Talauma ovata (GOTTSBERGER, in progress). The large
cream-white tepals open at the apex to form a small entrance through
which the beetles crawl into the interior of the flower. Just as in Magnolia
216 G. GOTTSBERGER:

species, beetles seem attracted by strong fruit-like odours. In an early

stage the odour is clearly apple-like, sometimes mixed with the odour of
terpentine, sometimes like an indefinable fruit ester, and at a later time
of anthesis, because of accumulation of the beetle's excrements, the
odour changes to a pungent one, somewhat similar to a mixture of fish
and the beetles themselves. The flowers of Talauma, like most Magnolias,
Annonaceae or other cantharophilous flowers of the Magnoliidae, are
protogynous. The beetles trapped by the strong imitative odours may
deposit pollen on stigmata in the earlier stage of flower development and
receive new pollen shortly before they leave when the stamens have
dehisced. At the later phase, their bodies usually have become sticky from
crawling over the non-sweet stigmatic secretions, so that pollen easily
adheres to them. Mter anthesis and opening of the flowers, the heavy
damage done by the beetles becomes evident. Tepals are eaten and
stamens and carpels are gnawed. For the time being, and until there is
proof, one may at least suspect that the beetles involved in the pollina-
tion of Talauma are also such species as live and breed on fruits. Just
as in flowers of Annonaceae we discovered beetle larvae in shed floral
axes and in undeveloped carpels. The beetles not only cause crossing
of flowers in Talauma but also selfing, which was proved experimentally.
Self-compatibility is another widespread feature of cantharophilous
Magnoliidae (see HEISER 1962; GOTTSBERGER 1970; THIEN 1974; etc.).
With CARLQUIST (1969: 340) I share the idea that the large number
of sexual organs, the flattening of the stamens, their dense aggregation
along the floral axis and consequently the large dimensions df M agno-
liaceae flowers might well be interpreted as an adaptation better to
escape from damage by specialized, crude pollinators (see GOTTSBERGER
1974: 463).
Beetle pollination syndromes in Eupomatia (Eupomatiaceae), Caly-
canthus (Calycanthaceae), Victoria and other Nymphaeaceae are com-
parable insofar as their flowers all offer special nutritious tissues to the
visitors. In Eupomatia laurina beetles eat the flat inner staminodes
(HAMILTON 1897), in Calycanthus occidentalis the beetles feed on special
food bodies on the staminodes, stamen tips and innermost petals (GRANT
1950 a), and in Victoria amazonica, carpellary appendages in the flower
cavity provide food for the beetles (PRANCE & ARIAS 1975). The proto-
gynous flowers of the above-mentioned species emit strong fruit-like
odours and usually attract a mass of beetles which are kept for some time
in the flower cavity. Perigyny may be a mode of ovule protection in
these species (see GRANT 1950 b).
There is certainly still very much to be done ill pollination studies
on the more tropical Mag1wliidae, especially in Magnoliales and Laurales,
of which only a few have actually been investigated. Therefore, I do not
Some Aspects of Beetle Pollination 217

wish to leave behind this uniform picture of exclusiveness of beetle pollina-

tion in primitive angiosperms, and will mention also some exceptions_ For
example, species of Lauraceae are visited chiefly by beetles and flies
(KNUTH 1904, CORNER 1940). On the other hand, in flowers of Monimiaceae
and Siparunaceae I have never observed beetles at all. In the species of
Mollinedia (Monimiaceae) with urceolate receptacles, pollination of uni-
sexual flowers is accomplished by ThY8anoptera. The female insects bore
holes into male and female buds and deposit their eggs in the interior of
the receptacles. Their larvae start development in buds' and become adult
when flowers open. Afterwards they fly out, go to other flowers and are
responsible for the pollen transfer (GOTTSBERGER, in progress). This strange
mode of pollination reminds one of the much more sophisticated mode
in Ficw~, where the urceolate syconia are used as breeding places by wasps.

But let us return to the main point. After having discussed some
pollination syndromes of entomophilous members of the Magnoliales,
Laurales and Nymphaeales we may summarize and try to interpret our
results. Drimys, a member of Winteraceae, the presumably most primitive
present-day angiosperm family, has an open type of beetle pollination
without further specialization. I believe that the unifacial stamens and
carpels have remained as an archaic feature in its flowers because of its
unspecialized cantharophily. The beetles attracted casually by a flowery
odour, and not by deceit, are mostly small and do not further harm the
sexual organs of t,he flowers. Because of this less destructive mode of
pollination I think that there was and is no selection pressure to enlarge,
to flatten or to increase the number of the sporophylls or to aggregate
them densely along the floral axis.
On the other hand, flattened structures, and/or a high number of
sporophylls, are characteristic for flowers of the other families mentioned
above, which all show a specialized cantharophily. Their flowers
utilize deceit, attracting fruit-, or carrion- or dung-beetles. The side
effects of these more crude visitors seem to be compensated by larger
flowers, by the production of food bodies or thick petals, by a semi-
inferior ovary, or by connective shields, or by mechanisms like closing
which help to exclude the larger beetles. The flowers offer protection,
alimentation, breeding places and imitation of the normal substrate
of the beetles.
Returning once more to the Winteraceae 1 wish to draw attention
to another feature which seems to underline the primitiveness of flower
structures in this family. The dioecious species of Tasmannia, contrary
to Drimys, have anemophilous, unshowy and unisexual flowers (PER-
VUKHINA 1967). The fact of manifestation of anemophily alr~ady in
Winteraceae was used, however, in a sense to show that primitive angio-
sperms were not quite fixed as to one or the other mode of pollination
even in their earliest times. Doubts were brought up as to what extent
218 G. GOTTSBERGER:

entomophily is indeed the original and basic condition in angiosperms.

I believe that another way of interpretation can be given. If the W intera-
ceae are really such a primitive and unspecialized group as we believe
them to be, then, and although the present-day members are now fixed
in their archaism, their ancestors probably had plasticity enough to
differentiate quite actively. It is a principle in evolution that usually
the more basic and unspecialized groups or members of a group are
able to diverge actively, whereas the advanced and more specialized
ones are often more fixed. The only possible way of flower-biological
radiation in the earliest times of angiosperms was probably from can-
tharophily to anemophily. At that time there probably did not exist
other pollination agents besides beetles and the wind. Some of the
ancestors of the Winteraceae radiated into anemophily with a subsequent
change-over to dicliny, as seen in Tasmannia. From a certain point of
specialization on, such a radiation towards anemophily does not seem to
have occurred anymore. I am not aware of any example of pure ane-
mophily in the Magnoliaceae, Annonaceae, Calcyanthaceae, Eupornatiaceae
or Nymphaeaceae, although there exist tendencies towards dicliny (for
Annonaceae see FRIES 1959), supposedly here more an attempt to escape
autogamy. I shall come back to the question of an early radiation into
anemophily when I present the conclusions at the end of this paper.
Finishing the account of beetle pollination syndromes and presum-
able implications in the flower structures of some Magnoliidae, I admit
that my reasoning is somewhat oversimplified, but I think it necessary
to show some basic trends likely to have occurred during the evolution
of the ancestors of these groups. In reality I am aware that reversals of
trends may also have frequently occurred, which finally brought out the
tremendous diversity in this subclass. Unless we have a more complete
knowledge of how so called primitive flowers function, we will not he
able to explain their characters in a more satisfying way. For the time
being my contribution is meant to open discussion and also to show where
more observation is imperative. An understanding of cantharophily in
the Magnoliidae may be essential also for an understaading of what this
mode of pollination may have caused in the flower structures of other
subclasses.
The theory to be presented now is that the secondary numerical
increase of stamens in other subclasses may represent an adaptation to
pollination by beetles and other pollen-eaters. Such an interpretation
certainly needs substantiation to be reliable. Therefore, let me first
present the facts.
I was brought to this iciea when 1 studied the pollination mechanisms
of some of the polyandrous members of the Dilleniidae. In the Dil-
leniaceae I have observations for Davilla and Doliocarpus. Davilla
Some Aspects of Beetle Pollination 219

elliptica, a shrub in the Brazilian cerrado vegetation opens its yellow-

petaled flowers at dawn, which was at 6: 20 in the morning in June.
The two innermost sepals, enlarged and scoop-like, separate and the
petals and stamens unfold. Immediately there was flower visiting by
the int~oduced Apis melli/era and later on also by the native Meli-
poninae-bees, which all collected pollen. During the whole morning
there was also frequent approach by many beetle species, which cut
the stamens and fed on the filaments, anthers and pollen. :From about
9 : 00 in the morning on, individuals of another species of a curculionid
beetle arrived, usually crawling directly to the ovary. This beetle was
seen to bite holes into the ovary's interior, afterwards depositing an egg
in one of the locules and than changing to another flower. At about
11 A.M. the petals fall, and the two large sepals start to fold together,
during which time there are still visits by the last mentioned beetles. At
about 2 P.M. and 3 P.M. of the same day the ovary of flowers is completely
enclosed by the two inner sepals. Thus, the flowers are pollinated by bees
and beetles, of which the last are quite destructive. Some of them gnaw on
the stamens and others use the ovary as a breeding place. The folding
of the sepals after flowering seems to be an adaptation to save some of
the ovaries from destruction. .From three different counts it was seen
that about 25% to 30% of ovaries are destroyed by the beetle's larvae.
From another count in Davilla rugosa it became evident that the destruc-
tion sometimes attains nearly 100%. Unfortunately, when I made these
observations some years ago, I did not pay special attention to the flower
odour of Davilla. In Doliocarpus schottianus the flower odour has a heavy
fruit-like component; its stamens and petals are also cut or gnawed by
beetles.
in the Clusiaceae (Theales) beetles are involved in the pollination of
several species of K ielmeyera. K. variabilis has large white-petaled,
slightly protogynous flowers, which emit a strong fruit like odour. Besides
some bees, numerous beetles are regularly attracted by the flowers and
are co-pollinators. They crawl over all flower parts and do much destruc-
tion in gnawing petals and cutting stamens. Older flowers are often
without any stamens because the beetles have rasped them all at the
base. .For some beetles there is already proof that they are true fruit
eaters; they were found also on ripe, odoriferous fruits of Campomanesia
(M yrtaceae ).
In species of Clusia, beetle pollination is almost an exclusive feature.
Pollination in these dioecious species is attained quite precisely because
the strong odour, with a fruit-like component, is localized only on the
massive stamens and the large stigmata. Here also there occurs some
gnawing of flower parts.
Species of Paeonia (Paeoniaceae) have long been known as having
220 G. GOTTSBERGER:

beetle flowers with cantharophilous odours (DELPINO 1873; DIELS 1916;

PORSCH 1950; GRANT 1950 b; VAN DER PIJL 1960; LEPPIK 1964; etc.).
There is further suspicion of regular participation of beetles in the pol-
lination of some Cistaceae (e.g. KNOLL 1914) and Tiliaceae (e.g. species of
Luehea, BAKER et al. 1973; GOTTSBERGER, unpublished). AI1!o here,
destruction of stamens commonly occurs.
For the basic polyandrous group of the Rosidae data are most
instructive also. A very large number of species of Rosaceae belonging
to the genera Spiraea, Aruncus, Physocarpus, Sorbus, Pyrus, Amelanchier,
Crataegus, Rubus, Filipendula, Rosa, Sorbaria, Prunus, Chaeno'fneles,
M espilu.y, Potentilla, Geum, etc. are known to be pollinated by beetles.
Usually beetles are co-pollinators together with other insects; flower
destruction may vary from population to population but occurs fre-
quently (KNUTH 1904; PORSCH 1950; GRANT 1950b; KNOLL 1956;
PROCTOR & YEO 1973; etc.). It is remarkable that a large number of
these beetle pollinated flowers emit cantharophilous odours and are
protogynous. Some beetles on Rosaceae even belong to species found
otherwise in Magnolia flowers (KNUTH 1904: 334; DAUMANN 1930: 111).
For the polyandrous Caryophyllidae similar observations exist.
Mesembryanthemum of the Aizoaceae frequently has beetles as co-pol-
linators of the flowers (KNUTH 1904: 278-279) and in the Cactaceae
beetles are nearly omnipresent flower visitors (KNUTH 1904: 518-519;
PORSCH 1939 ; VAN DER PIJL 1961: 45). One only needs to check the list
of flower odours given by PORSCH (1939) to see how commonly fluit-like
or aminoid odours occur in this family.
These are the flower-biological data, some of which have been
known for decades. Analyzing the -facts, my interpretation tends in the
direction of seeing in the polyandry of the basic groups in Rosidae,
Dilleniidae and Caryophyllidae an adaptive character that provides an
excess of stamens and pollen for foraging beetles and other pollen-eaters.
This adaptive character of many ·stamens is manifested also in some more
advanced members with more advanced pollination syndromes.
One argument against my interpretation might be. the belief that
beetles are better attracted to these polyandrous flowers, because
they have many stamens and therefore provide larger amounts of food
than do flowers with few stamens. This argument is, however, easy to
invalidate by observations. In the polyandrous members of more ad-
vanced families of these subclasses, as for example, Caryocaraceae,
Cochlospermaceae, Begoniaceae, Capparaceae, Chrysobalanaceae and
others, I never saw the flowers visited regularly by beetles and also did
not often see damaged flowers. They have advanced pollination systems,
being visited by bats, bees or moths, and there is no regular utilization
of their polyandrous flowers by beetles. There may be casual visits hy
Some Aspects of Beetle Pollination 221

beetles, as occur on every plant, and there may be also a casual damage,
but there is not usually any frequent visiting. Another good argument
for beetles as the real co-pollinators of Dilleniaceae, Rosaceae, Clusiaceae,
Cactaceae, etc. is the widespread occurrence of the imitative fruit-like or
aminoid odours, which deceive beetles precisely; this character is known
already from the beetle-pollinated Magnoliidae. Protogyny concludes
the picture of this syndrome.
In the last few years it has become evident that the primary polyandry
of Magnoliidae cannot be directly compared with the so called secondary
one in other subclasses (LEINS 1964,1971; KUBITZKI 1973; etc.). However,
it is certainly right to say that if we assume that there existed an evolu-
tionary trend within the Magnoliidae towards a numerical increase of
stamens as a response to selection pressure, than this advanced condition
might be called secondary polyandry also (STEBBINS 1974: 226). But as
the numerical increase of stamens in Magnoliidae follows the spiral ar-
rangement and in other subclasses it usually does not, the problem remains
the same. It is still difficult to imagine how any form of secondary poly-
andry in Rosidae, Dilleniidae and Oaryophyllidae may have derived directly
from the spiral one of the Magnoliidae. There must have been first oligo-
merisation with a cyclic arrangement of stamens and from this condition
on a new initiation of stamen primordia.

As for the Caryophyllidae, it is commonly assumed that they derived

via the Phytolaccaceae from some Ranunculanae ancestors. In the
M agnoliidae, besides the initially mentioned two trends of radiation
into anemophily and specialized cantharophily, obviously at least a
third trend has existed towards oligomerisation and fixation of numbers
of flower parts. The last cited trend might have started from a time when,
besides the beetles, other, probably nectar-sucking, visitors had attained
enough frequency to exert selection pressure on flowers. The derivation
of the Caryopyllidae via the M agnoliidae with oligomerised stamens is
therefore at least theoretically possible. For the polyandrous Rosidae
and Dilleniidae an origin via the woody Saxifragales is often as-
sumed, in which group also flowers with oligomerised stamens predomi-
nate.
One reason why such an adaptation to beetle pollination has oc-
curred may have been the presence of cantharophilous odours in flowers
of the assumed ancestors. Cantharophilous odours together with the
protogynous condition are indeed quite frequent in present-day Magno-
liidae (even in those with a reduced number of stamens) and in Saxifrri-
gales. Such an imitative flower odour could have been a kind of pre-
adaptational character, based on which new waves of radiation towardb
beetle pollination have occurred.
Is there any way to explain why there is a centrifugal stamen develop-
ment in the polyandrous Caryophyllidae and Dilleniidae but a centri-
222 G. GOTTSBERGER:

petal one in the Rosidae? Are these characters evolved by chance or are
these opposite directions of stamen development explainable? Let us
try to use the immensely diversified Saxilragales as a model for deriving
Rosidae and Dilleniidae. A large number of Saxilragales are charac-
terized by a nectariferous disc in an intrastaminal position, thus separat-
ing the stamens and the ovary. The stamens are inserted at the periphery
of the flowers close to the perianth and there is space between them and
the gynoecium. If we assume that the polyandrous Rosidae are derived
via the Saxilragales with intrastaminal nectariferous discs, then in
response to a selection pressure for numerical increase of stamens a
centripetal development in direction of the flower-centre may have
occurred, filling the space of the former nectariferous disc, now partly or
totally functionless. That this idea may have some truth is shown by the
fact that the whole Rosidae are characterized by an intra- or also inter-
stamina I disc which appears everywhere within this group, especially in
the more derived members (with nectar sucking pollinators!) but also in
more primitive ones. Other present-day Saxilragales are without such
a disc and their stamens and ovary are close together. The Dilleniidae
may have been derived from Saxilragales-like (or other) ancestors
without an intrastaminal disc. In response to selection for numerical
increase' of stamens meristems are more likely t.o have increased towards
the periphery of flowers, because development towards the centre was
spatially limited by the gynoecium. The tendency to form a disc is
indeed much less pronounced in the Dilleniidae. Polyandrous Caryophyll-
idae may be comparable with the Dilleniidae, in being derived from
ancestors also without an intrastaminal disc, therefore appearing now
also with centrifugal stamen development. The evolution of secondary
polyandry, based on these reflections, may well have followed the
direction of least resistance.
To complete the picture of evolution of dicotyledons influenced by
beetle pollination, I wish finally to present some selected examples in
which another trend, that towards aggregation of small flowers into
inflorescences becomes evident. There are virtually hundreds of cases
but I shall restrict myself to those appearing in Polemoniaceae, Proteaceae
and Asteraceae. These three families have been well studied so that we
are aware of their major flower-biological radiation.
The Polemoniaceae, a basically bee-pollinated group, show occasional
radiation into beetle pollination (GRANT and GRANT 1965). In cantharo-
philous Linanthus parryae the flowers are large and cup shaped, but in
I pomop8i8 congesta small flowers are aggregated into terminal capitate
heads.
In the Proteaceae there is "a 'retrograde' development of pollination
syndromes from tilt' brush blossom t.ype back to the more primitive
Some Aspects of Beetle Pollination 223

bowl-shaped one concomitant with the progressive morphological

development of the blossom. This development is apparently accompanied
by a similar 'retrograde' development amongst the pollinators, from
pure or predominating ornithophily back to the assumed most primitive
stage of cantharophily" (FAEGRI 1965). In Leucadendron discolor the
inflorescence is even condensed into a structure which forms a perfect
parallel with Magnolia (V AN DER PIJL 1969: 92). Cantharophily is
evidently an end line in both families. There is no doubt that this is to
be considered a secondary specialization for beetle pollination.
In the Asteraceae, the aggregation of small flowers into flower heads
seems to be an original adaptation to beetles (see KNUTH 1905; GRANT
1950 b; VAN DER PIJL 1960; LEPPIK 1960, 1970; etc.), from which a
radiation towards more specialized pollinators, like Hymenoptera,
Lepidoptera, birds, etc. seems to have occurred. The fact that Compo-
sitae and other groups presumably follow an evolutionary sequence from
the more primitive pollination syndrome towards a more advanced one,
has led LEPPIK (1970: 327) to put forward a theory which says that
there is a general trend of floral evolution in various groups with a
recapitulation of the historical appearance of pollinators and pollination
syndromes. This theory, however, does not work for all cases. In the
Polemoniaceae and Proteaceae an opposite development has in fact
occurred.
The basic condition of cantharophily in the Compositae led several
authors to think that they must be a very old group having originated in
the early beetle-era. This assumption is not likely and in my opinion is
not even necessary. From the origin of angiosperms up to the present,
be{ltles were and are potential flower visitors and pollinators. There has
been a continuous trend of adaptation towards cantharophily in the
most diverse groups of most diverse evolutionary level. Therefore, the
basic adaptation to beetle pollination in Asteraceae might well have
occurred more recently. Also, the palaeontological data plead for a
relatively late appearance of this family. Only in mid-Tertiary does the
Compositae began to play a major role in the vegetation (W AGENITZ
1976).
Conclusion
Beetles probably influenced the origin of the angiosperms which
led to the formation of a basically entomophilous, hermaphrodite flower.
The original flowers must have had very primitive structures and
cantharophily supposedly was of an open unspecialized type. Present-
day members which still seem to show some of these original feat-ues
are some Winteraceae, like Drimys. The presence of anemophily in this
family probably indicates that there already existed a strong tendency
224 G. GOTTSBERGER:

to radiate into anemophily in the ancestral complex now extinct. It is

likely that this line or these lines led to entomophilous-anemophilous
and/or pure anemophilous groups of the Magnoliidae and some Hama-
melididae. From the ancestral complex there might have been several
waves of radiation towards specialized cantharophily concomitant with
a development of specialized cantharophilous flower characters, as for
example a numerical increase of sexual organs and their flattening
(e.g. Magnoliaceae, Annonaceae, etc.). Another, but probably somewhat
later trend, presumably connected with the increased influence of suck-
ing insects, led to groups with oligomerized flower parts. From such
groups with a reduced number of stamens, several waves of radiation
towards cantharophily seem to have occurred again, concomitant with a
secondary numerical increase of stamens. These radiation waves may
have occurred at a time when other flower visitors besides the beetles
were already in existence. Flowers of primitive members of secondarily
polyandrous groups show a mixed pollination syndrome with beetles
and other insects as co-pollinators. A nearly exclusive beetlQPollination
was reached by some Magnoliidae in earlier times when beetles probably
were the predominant visitors. Beetles continued as potential pollinators
up to the present, so that there was a continuous tendency of radiation
into beetle pollination. In more modern groups, however, instead of an
increase in size and number of flower parts, there is a new trend toWards
condensation of flowers into inflorescences.
This is my actual picture of a presum(jld evolution of dicotyledons
regarding, however, only cantharophily. We now only need to fill the
space with wind and water, bees, flies, moths, butterflies, birds and bats
to get a more complete account.

My thanks go to Prof. Dr. L. VAN DER PIJL, The Hague, who gave
valuable suggestions on the content of the paper and to Dr. G. EITEN,
Brasilia, and Mr. B. L. BURTT, Edinburgh, who corrected the language
and the style.

References
BAKER, H. G., BAKER, ,I., and OPLER, P. A., 1973: Stigmatic exudates
and pollination. In: Pollination and Dispersal (BRANTJES, N. B. M.,
and LINSKENS, H. F., eds.), 47-60. Nijmegen: Dept. Bot., Kath.
Universiteit. .
CARLQUIST, S., 1969: Toward acceptable evolutionary interpretations of
floral anatomy. Phytomorphology 19, 332-362.
CORNER, E. J. H., 1940: Wayside trees of Malaya. Vol. 1. Singapore: Govern-
ment Printing Office.
DAUMANN, E., 1930: Das Bliitennektarium von Magnolia und die Futter-
korper von Calycanthus. Planta 11, 108-116.
DELPINO, F., 1873: Ulteriori osservazioni sulla dicogamia nel regno vege·
tale. Milano: G. Bernardoni.
Some Aspects of Beetle Pollination 225

DIELS, L., 1916: Kaferblumen bei den Ranales und ihre Bedeutung fur die
Phylogenie der Angiospermen. Ber. deutsch. bot. Ges. 34, 758-774.
EHRENDORFER, F., 1976: Evolutionary significance of chromosomal dif-
ferentiation patterns in gymnosperms and primitive angiosperms. In:
Origin and early evolution of angiosperms (BECK, C. B., ed.), 220-240.
New York: Columbia University Press.
KRENDL, F., HABELER, E., and SAUER, W., 1968: Chromosome numbers
and evolution in primitive angiosperms. Taxon 17, 337-353.
F AEGRI, K., 1965: Reflections on the development of pollination systems
in African Proteaceae. J. South Afr. Bot. 31, 133-136.
- and PIJL, L. VAN DER, 1971: The principles of pollination ecology.
Ed. 2. Oxford: Pergamon Press.
FRIES, R. E., 1959: Annonaceae. In: Die naturlichen Pflanzenfamilien
(ENGLER, A., and PRANTL, K., eds.), Ed. 2, 17a II, 1-171. Berlin:
Duncker & Humblot.
GOTTSBERGER, G., 1970: Beitrage zur Biologie von Annonaceen-Bluten.
6sterr. bot. Z. 118, 237-279.
- 1974: The structure and function of the primitive angiosperm flower-a
discussion. Acta bot. Neerl. 23, 461-471.
GRANT, V., 1950a: The pollination of Calycanthus occidentalis. Amer. J.
Bot. 37, 294-297.
1950 b: The protection of the ovules in flowering plants. Evolution 4,
179-201.
and GRANT, K., 1965: Flower pollination in the Phlox family. New York-
London: Columbia Univ. Press.
HAMILTON, A. G., 1897: On the fertilisation of Eupomatialaurina R. BR.
Proc. Linn. Soc. New South Wales 22,48-55.
HEISER, CH. B., 1962: Some observations on pollination and compatibility
in Magnolia. Proc. Indiana Acad. Sci. 72, 259-266.
KNOJ.L, F., 1914: Zur 6kologie und Reizphysiologie des Androceums von
Cistus salvifolius L. Jahrb. wiss. Bot. 54, 498-527.
- 1956: Die Biologie der Bliite. Berlin-Gottingen-Heidelberg: Springer.
KNUTH, P., 1904-1905: Handbuch der Bliitenbiologie. 3. Band, 1. und
2. Teil (LOEW, E., ed.). Leipzig: Verlag W. Engelmann.
KRAL, R., 1960: A revision of Asimina and Deeringothamnus (Annonaceae).
Brittonia 12, 233-278.
KUBITZKI, K., 1973: Probleme der Gro/3systematik der Blutenpflanzen.
Ber. dtsch. bot. Ges. 85, 259-277.
- and VINK, W., 1967: Flavonoid-Muster der Polycarpicae als systemati-
sches Merkmal. II. Untersuchungen an der Gattung Drimys. Bot. Jb.
87, 1-16.
LEINS, P., 1964: Das zentripetale und zentrifugale Androecium. Ber. dtsch.
bot. Ges. 77, 22-26.
- 1971: Das Androecium der Dikotylen. Ber. dtsch. bot. Ges. 84,
191-193.
LEPPIK, E. E., 1960: Early evolution of flower types. Lloydia 23, 72-92.
1964: Floral evolution in the Ranunculaceae. Iowa State J. Sci. 39,
1-101.
1970: Evolutionary differentiation of the flower head of the Compositae.
II. Ann. Bot. Fennici 7, 325-352.
Plant Syat. Evol., Suppl. 1 15
226 G. GOTTSBERGER: Some Aspects of Beetle Pollination

PERVUKHINA, N. V., 1967: The pollination of the primary angiosperms

and the evolution of modes of pollination (Russ.). Bot. Zh. (Moskow)
52, 157-188.
PIJL, L. VAN DER, 1953: On the flower biology of some plants from Java.
Ann. Bogor. t, 77-99.
- 1960: EcologJcal aspects of flower evolution. I. Phyletic evolution.
Evolution 14, 403-416.
- 1961: Ecological aspects of flower evolution. II. Zoophilous flower
classes. Evolution 15, 44-59.
- 1969: Evolutionary action of tropical animals on the reproduction of
plants. BioI. J. Linn. Soc. 1, 85-96.
PORSCH, 0., 1939: Das Bestaubungsleben der Kakteenbliite. Oactaceae,
Jahrb. deutsch. Kakteen-Ges., 1-142.
- 1950: Geschichtliche Lebenswertung der Kastanienbliite. Osterr. bot.
Z. 97, 269-321.
PRANCE, G. T., and ARIAS, J. R., 1975: A study of the floral biology of
Victoria amazonica (POEPP.) SOWERBY (Nymphaeaceae). Acta Amaz. 5,
109-139.
PROCTOR, M., and YEO, P., 1973: The pollination of flowers. Glasgow:
W. Collins Sons and Co. Ltd.
STEBBINS, G. L., 1974: Flowering plants. Evolution above the species level.
Cambridge, Mass.: Belknap Press.
THIEN, L. B., 1974: Floral biology of Magnolia. Amer. J. Bot. 61,1037-1045.
- HEIMERMANN, W. H., and HOLMAN, R. T., 1975: Floral odors and
quantitative taxonomy of Magnolia and Liriodendron. Taxon 24,
557-568.
UPHOF, J. C. TH., 1933: Die nordamerikanischen Arten der Gattung ABi-
mina. Mitt. deutsch. dendrol. Ges. 45, 61-76.
WAGENITZ, G., 1976: Systematics and phylogeny of the Oompositae (A8ter-
aceae). Plant Syst. Evol. 125, 29-46.
WESTER, P. J., 1910: Pollination experiments with Anonas. Bull. Torrey
Bot. Club 37, 529-539.
ZIMMERMAN, G. A., 1941: Hybrids of the American papaw. J. Hered. 32,
83-91.

Address of the author: Univ.-Doz. Dr. G. GOTTSBERGER, Departamento

de Botanica, Faculdade de Ciencias M6dicas e Biol6gicas de Botucatu,
Universidade Estadual PauIista "Julio de Mesquita Filho", BR-18600
Botucatu, S. P. Brazil.
Plant Syst. Evo!., Supp!. 1, 227-234 (1977)
© hy Springer. Verlag 1977

Botanisches lnstitut der Universitat Wien, Austria

New Ideas About the Early Differentiation of Angiosperms

By

Friedrich Ehrendorfer, Wien

Abstract: Three premises are introduced: Origin of angiosperms from

one common ancestral group, stamens and carpels as appendicular (phyl-
lomic) sphorangiophores, and adaptive nature of basic angiosperm characters.
The widely accepted assumptions that Magnoliidae link up directly to the
Rosidae-Dilleniidae alliance, and that Hamamelididae are heterogeneous
and have originated from several Rosidae·Dilleniidae groups are hardly
tenable: There are drastic differences between the Magnoliidae and Rosidae-
Dilleniidae, in the morphology of their functionally similar entomophilous
flower types, and in many other respects; the H amamelididae demonstrate
considerable internal coherence together with ambivalent affinities to both,
Magnoliidae and Rosidae-Dilleniidae. To solve this dilemma, the hypo-
thesis is discussed that the extant Hamamelididae mark the remnants and
descendants of an ancient "transitional field" from the Magnoliidae to the
Rosidae-Dilleniidae (Fig. 1). The Hamamelididae would thereby correspond
to an early phase of flower reduction and trends towards anemophily, the
Rosidae·Dilleniidae to a subsequent phase of floral elaboration and inten·
sified adaption towards zoophily.

There are three premises (1-3) which are essential to the following
discussion; let me start with them without going into too many details:
1. The angiosperms have originated from one common ancestral
group. There is growing evidence from many fields to back such an
assumption in spite of continued scepticism (e.g. MEEUSE in this sym-
posium): Paleobotany demonstrates the simultaneous appearance of
typical monosulcate and later of tricolpate angiosperm pollen grains,
leaves, etc. during the Lower Cretaceous (Barremian to Albian) on a
world wide scale (cf. BRENNER 1976, DOYLE and HICKEY 1976, HUGHES
1976, DOYLE 1977). These data clearly suggest an origin of the group in
the tropical zones and a spread towards the poles, paralleled by increas-
ing diversification. Among the very large array of morphological,
anatomical, palynological, biochemical, and other characters connecting
all angiosperms, but lacking in other spermatophytes, the embryological
similarities always have been particularly impressive. They are not
15·
228 F. EHRENDORFER:

limited to the well known and unique structure of the embryo sack, the
peculiar double fertilization, and the subsequent secondary endosperm
formation, but include unexpected details concerning the entrance of
the pollen tube through the degenerating synergide and the remarkablp-
ultrastructure -of the "filiform apparatus". As a taxon the Angiospermae
are particularly coherent. This is most evident from the fact that none
of the very numerous taxonomic subdivisions proposed has been generally
accepted (not even in regard to Monocotyledoneae and Dicotyledoneae;
cf. HUBER in this symposium). On the other hand, the separation from
all other Spermatophyta (including the Gnetatae-obviously the most
closely related extant Gymnospermae), never has been a problem.
2. The stamens and carpels of angiosperms ar~ sporangiophores
of an appendicular (phyllomic) nature, homologous throughout the
group. The increasing knowledge of pollen sac- and ovule-bearing
organs in pteridosperms and -their descendants hints to their morpho-
logical interpretation along modified classical concepts without taking
refuge to unnecessarily complex auxiliary hypotheses (cf. EHRENDORFER
1971). From such a view-point the flowers of all angiosperms appear
comparable (as would have to be expected from their common origin).
3. The origin and successful diversification of angiosperms must
have been linked, just as in all other groups of organisms, to their basic
characters, giving them adaptive superiority over their predecessors
and allowing them to enter a new ecological niche. Their primarily
hermaphrodite flowers with enclosed ovules evidently have made
possible wind-independent, more precise, and more economic insect
pollination. (Size, ornamentation, and ultrastructure of earliest known
angiosperm monosulcate pollen suggest entomophily and germination
on the stigma of closed carpels! DOYLE 1977, J. MULLER, pers. comm.)
Neotenic gametophytes, double fertilization and secondary endosperm
must have allowed angiosperms faster and more economic reproduction
(STEBBINS 1974), their improved conducting system and increased
flexibility of vegetative growth clearly have stimulated ecological
expansion. Consequently, early angiosperms may have grown as small,
subordinate and scattered woody plants in the understory and along
the margins of unstable successional gymnosperm forests in ± humid
seasonal tropical climates (cf. the thin pollen walls of earliest angio-
sperm pollen; DOYLE and HICKEY 1976).
If these three premises are accepted, we can suggest on a comparative
basis a likely basic flower model for the angiosperms (Fig. 1-1): genera-
lized, entomophilous, rather large, with spiral arrangement of numerous
flower elements, undifferentiated perianth (perigon), primary polyandry,
medium-sized, monosulcate, psilate to reticulate pollen, carpels with
stigmatic margins but without styles, and a trend to staminal nectaries.
New Ideas About the Early Differentiation of Angiosperms 229

Among extant angiosperms certain Magnoliidae seem to have preserved

such a flower type rather closely.
If we want to interpret other angiosperm flower types with the
help of our Magnoliidae model, and if we try to speculate about pos-
sible phylogenetic links between extant Magnoliidae (s. lat.) and the
Rosidae-Dilleniidae alliance, we encounter two major obstacles, to be
outlined in the following paragraphs (A and B).

Fig. 1. Proposed relationships between several major dicotyledonous angio-

sperm' groups and their flower types. I basic Magnoliidae (polymerous,
perigon, primary polyandry); II basic Hamamelididae (and some basic
Rosidae-Dilleniidae) (oligomerous, hermaphrodite); III advanced Hamameli-
didae (strongly oligomerous, unisexual); IV basic Rosidae-Dilleniidae
(calyx and corolla, androecium 1-2 cyclic); V some Dilleniidae (secondary
polyandry: centrifugal); VI some Rosidae (secondary polyandry: centri-
petal), VII advanced Asteridae (synsepaly, sympetaly, dorsiventrality,
oligomery, axial nectary)

A. Among the Rosidae-Dilleniidae alliance we also find entomophi-

lous and rather large, polyandrous flower types but they differ from our
Magnoliidae model (Fig. 1-1) in some important aspects: (spiro)cyclic
arrangement of fewer flower elements, differentiated perianth (calyx
related to perigon, corolla related to stamens), secondary polyandry
(usually through centripetal or centrifugal elaboration: LEINS 1975),
rather large, tricolpate and often more strongly sculptured pollen,
carpels with stigmas on styles, and a trend towards axial nectaries
(Fig. 1-V, VI). A direct derivation of the entomophilous flower models
230 F. EHRENDORFER:

I ~ V, VI is hardly possible, if functional aspects are considered: Why

are there new semaphylls (i.e. petals) in V + VI, if a showy pergion
was already present in I? Why is there secondary polyandry in V + VI,
as primary polyandry had already been established in I? Why the
different types of nectaries in I and V + VI? It is these questions, so
far not answered, together with the obvious differences between Magno-
liidae and Rosidae-Dilleniidae in chemistry (KUBITZKI 1969, 1973) and
other characters which have supported ideas about a pleiophyletic
origin of Dicotyledoneae.
B. The Hamamelididae (in the sense of TAKHTAJAN 1969 and CRON-
QUIST 1968) are nowadays usually regarded as secondarily si~plified
in flower structure as a consequence of adaptation towards wind pollina-
tion (e.g. ENDRESS in this symposium; Fig. 1-II, III). This and the
recognition of affinities of Hamamelididae to several groups of the
Rosidae-Dilleniidae alliance has recently led to a total demolition of the
group (cf. THORNE 1974 or DAHLGREN in this symposium: his Hamame-
lidanae include Cunoniales hesides Fagales, Casuarinales, Hamameli-
dales, Trochodendrales, etc.; Juglandales and Myricales are placed in
Rutanae, Didymelaceae (= Didymelales) and Barbeyaceae (= Barbey-
ales) in Celastranae, Eucommiales in Cornanae, Urticales in Dillenianae).
Such a procedure is based on the opinion that the Hamamelididae are
a polyphyletic assemblage, that their orders are ± independent ane-
mophilous derivatives from several zoophilous Rosidae-Dilleniidae
groups, and that therefore they should be regarded as "dead ends"
in a morphological and evolutionary sense. This interpretation obscures
important relationships and fails to explain several relevant pheno-
mena: Why are the various orders of Hamamelididae connected by
palynological, anatomical (wood, leaf venation, etc.), biochemical, and
other affinities which can not be explained as accidental or as a result
of parallel adaption to secondary anemophily'l How is it possible that
there are not only the unquestionable affinities of Hamamelididae to
various groups of Rosidae-Dilleniidae but at the same time to the
Magnoliidae, if they are to be derived separately from the former only'!
Let us remember in connection with the last question: The wide-spread
occurrence of 3-merous flowers in Magnoliidae and in Hamamelididae
(e.g. Fagaceae, Platanaceae, Daphniphyllaceae, Barbeyaceae etc.); the
remarkable similarities in the pattern of chromosomal differentiation
within Hamamelididae and between Hamamelididae and woody Magno-
liidae (s. str.) (EHRENDORFER 1976a), the obviously ambivalent affinities
of Trochodendraies, Cercidiphyllales and Eupteleales in regard to partly
homoxylous wood, chloranthoid and platanoid leaf dentation (HICKEY
and WOL1<'E 1975), ultrastructure of 3-colpoidate or 3-colpate pollen
(WALKER 1976a, b), and chemistry: lack of benzyl-isoquinoline alkaloids
New Ideas About the Early Differentation of Angiosperms 231

(as in Magnoliidae s. str.), but presence of polyphenolics (like leu co-

delphinidine) and ellagic acid (as in Rosidae-Dilleniidae); the broadly
documented biochemical, anatomical, chromosomal and other affinities
(benzyl-isoquinoline alkaloids, bitter substances, idiobasts, chromosomal
patterns, choricarpy, etc.) which connect Magnoliidae and the primitive
Rutaceae-Zanthoxyleae and -Toddalioideae (FISH and WATERMAN 1973,
EHRENDORFER 1976a), which in turn are clearly linked through the
Rhoipteleaceae to the Juglandaceae and Myricaceae. Why do evidently
wind-transported and Hamamelididae-like pollen appear so remarkably
early in the fossil record (Albian-Cenomanian), and why are there well
documented records of several Hamamelididae (reminiscent of Cercidi-
phyllaceae, Platanaceae, Fagaceae, Juglandaceae, etc.) (MULLER 1970,
BRENNER 1976, DOYLE and HICKEY 1976) which antedate their pre-
sumed zoophilous Rosidae-Dilleniidae predecessors? The widely accepted
but evidently over-simplified reductional interpretation of the Hamame-
lididae has no answer to these questions. This deficiency again has pro-
vided arguments for iq.eas about primarily simple flowers in Hamame-
lididae and their direct gymnospermic origin, independent from Magno-
liidae and the Rosidae-Dilleniidae alliance.
Is there a solution to the problems concerning the two basic zoo-
philous flower types of dicotyledons, i.e. Magnoliidae and Rosidae-
Dilleniidae (A), and the ambivalent affinities of the evidently very an-
cient Hamamelididae (B)? To this end I would like first to discuss the
flower types illustrated in Fig. 1. This diagram indicates that the pri-
marily polyandrous Magnoliidae type (I) and the secondarily poly-
androus Rosidae-Dilleniidae types (II) can be linked through types II
and "IV. Among less advanced Hamamelididae we encounter the flower
types II which tend towards anemophily (but often still attract animal
pollinators): small size, (spiro)-cyclic arrangement of a reduced number
of flower elements, simple and inconspicuous perigon, oligandry, carpels
± free, with stigmas exposed on styles, small and smooth tricolpate to
triporate pollen,. and lack of nectaries. The development from II to III
illustrates extreme reduction and unisexual differentiation of flowers
typical for many exclusively an~mophilous Hamamelididae, often
reaching a stage which appears hardly reversible. On the other hand,
flower type IV indicates an adaptive reversal to a new level of zoophily;
it is typical for many Rosidae-Dilleniidae, and can be characterized by
increasing flower size, origin of a new type of semaphylls, i.e. petals
from stamens, larger pollen grains with variously sculptured exine and
elaborated apertures. There are trends towards secondary polyandry,
particularly in beetle pollinated groups (cf. GOTTSBERGER in this sym-
posium), but also trends towards the formation of axial nectaries for
the attraction of more advanced flower visitors with sucking mouth
232 F. EHRENDORFER:

parts. Finally, flower model VII depicts well-known further steps in

the specialization of sympetalous dicotyledons.-If we accept the se-
quence of changes in flower structure and function as outlined in this
paragraph and in Fig. 1, the question listed under A) can be answered.
In terms of phylogenetic relationships the data presented suggest
that the entomophilous Magnoliidae and Rosidae-Dilleniidae are not
of obscure parallel origin, but that the extant Hamamelididae mark the
remnants and descendants of an ancient (Lower to Upper Cretaceous
"transitional field" from the Magnoliidae to the Rosidae-Dilleniidae.
This has already been suggested, particularly on the basis of palyno-
logical evidence, by WALKER (1976: Fig. 4 and pers. comm.). We there-
fore come to the unorthodox view that the Hamamelididae are not alto-
gether "dead ends" but indicators for an early phase of flower reduction
and adaptation towards anemophily in the evolution and differentiation
of the angiosperms. It is obvious that such an interpretation solves the
problems raised under B), particularly those concerning the ambivalent
affinities of Hamamelididae to Magnoliidae and Rosidae-Dilleniidae.
As an ecological and functional corollary to the hypothesis presented,
one has to assume that early entomophilous angiosperms, after a phase
of establishment and coexistence with anemophilous gymnosperms
expanded into more northern and more xeric habitats, not yet colonized
by seed plants. A tendency towards anemophily must have been an
adaptive advantage under such circumstances, with more uniform
pioneer popUlations in an open environment, and with the obligate lag
of animals (including potential pollinators) following the plants in the
conquest of new ecological niches. It could have been that these develop-
ments have contributed to the increase in pollen wall thickness and the
origin of 3-colpate pollen (for better accomodation to volume changes
due to varying humidity and/or better germination chances?) in the
early angiosperms (DOYLE, VAN CAMPO and LUGARDON 1975).
From the considerable amount of evidence available to support the
hypothesis discussed, only little can be added at this, occasion to the
paleobotanical, palynological, anatomical, morphological and bio-
chemical data already briefly mentioned. The suggestion that petals,
particularly in the Rosidae, Dilleniidae and A steridae , are of androecial
origin, dates back to C. DE CANDOLLE and GELAKOVSKY, and has been
well substantiated (BAUM: 1950). Crucial instances include families,
where apetalous and petal iferous taxa coexist, and where the origin of
diplochlamydeous from monochlamideous flowers can be inferred, as in
Hamamelidaceae (J. WALKER, pers. comm., but cf. ENDRESS in this
symposium), Cunoniaceae (also with trends towards numerical increase
of stamens), Rutaceae-Zanthoxyleae/-Toddalioideae, Buxaceae-Euphor-
biaceae, etc. The ancient Proteaceae have remained on a monochlamydeous
New Ideas About the Early Differentiation of Angiosperms 233

level but have compensated this in their return to elaborate zoophily

through the formation of ± efficient pseudanthia, just as numerous
Hamamelidaceae and Moraceae (cf. ENDRESS and BERG in. this sympo-
sium), Euphorbiaceae, etc. Finally, a note on remarkable cases of paral-
lelism :n the Oaryophyllidae and the Monocotyledoneae, groups which
appear directly linked to the Magnoliidae: For both groups one can
suggest-just as for the Hamamelididae and Rosidae-Dilleniidae-an
early phase of flower reduction. and trends towards anemophily (e.g. in
primitive members of Molluginaceae, A izoaceae , Phytolaccaceae or of
Dioscoreaceae, Arecaceae, etc.), followed by a phase of intensified adapta-
tion towards zoophily, with development of showy tepals or petals,
nectaries (axial in Oaryophyllidae, gynoecial-septal in Monocotyledoneae),
and occasional increase of stamen numbers (cf. EHRENDORFER 1976).
Let me finish by saying that I hope that the facts presented, and the
suggestions and hypotheses based upon them will stimulate further
research on the fascinating and wide-open problem of the early differen-
tiation of the angiosperms; to the benefit of the facts, and at the expense
of the hypotheses.
The author is vel y grateful to Dr. J. WALKER (Amherst) for stimulating
discussions and personal communications during his stay as Visiting Pro-
fessor at. the Universit,y of Massachusetts.

References
BAUM, H., 1950: Unifaziale und subunifaziale Strukturen im Bereich der
Bliitenhiille und ihre Verwendbarkeit fiir die Homologisierung del'
Kelch- und Kronblatter. OsterI'. Bot. Z. 97, 1-43.
aRENNER, G. J., 1976: Middle cretaceous floral provinces and early migra-
tions of angiosperms. In: Origin and early evolution of angiosperms
(BECK, C. B., ed.), 23-47. New York and London: Columbia Univ.
Press.
CRONQUIST, A., 1968: The evolution and classification of flowering plants.
London: Nelson & Sons Ltd.
DOYLE, J. A., 1977: Patterns of evolution in early angiosperms. In: Pat-
terns of evolution (HALLAM, A., ed.), 501-546. Amsterdam: Elsevier.
- HICKEY, L. J., 1976: Pollen and leaves from the mid·cretaceous Potomac
group and their bearing on early angiosperm evolution. In: Origin and
f'arly evolution of angiosperms '(BECK, C. B., ed.), 139--206. New York
and London: Columbia Univ. Press .
.. _- VAN CAMPO, M., LUGARDON, B., 1975: Observations on exine structure
of Eucommiidites and lower cretaceous angiosperm pollen. Pollen et
Spores 17, 429--486.
EHRENDORFER, F., 1971: Spermatophyta. In: Lehrbuch del' Botanik fiir
Hochschulen, 30. Auf!., 586-741. Stuttgart: G. Fischer.
-- 1976a: Evolutionary significance of chromosomal differentiation pat-
t.erns in gymnosperms and primitive angiosperms. In: Origin and early
evolution of angiosperms (BECK, C. B., ed.), 220-·240. New York and
London: Columbia Univ. Press.
234 F. EHRENDORFER: New Ideas About the Early Differentiation

EHRENDORFER, F., 1976b: Closing remarks: Systematics and evolution of

centr03permous families. Plant Syst. Evol. 126, 99-105.
FISH, F., WATERMAN, G., 1973: Chemosystematics in the Rutaceae II.
The chemosystematics of the Zanthoxylum/Fagara complex. Taxon 22,
177-203.
HICKEY, L. J., WOLFE, J. A., 1975: The base of angiosperm phylogeny:
Vegetative morphology. Ann. Missouri Bot. Gard. 62, 538-589.
HUGHES, N. F., 1976: Palaeobiology of angiosperm origins. Cambridge:
Univ. Press.
KUBITZKI, K., 1969: Chemosystematische Betrachtungen zur Gro/3gliederung
der Dicotylen. Taxon 18, 360-386.
- 1972: Probleme der Gro/3systematik der Bliit.enpflanzen. Ber. dtsch.
bot. Ges. 85, 259-277.
LEINS, P., 1975: Die Beziehungen zwischeri multistaminaten und einfacholl
Androeceen. Bot. Jahrb. Syst. 96, 231---237.
MULLER, J., 1970: Palynological evidence on early differentiation of angio-
sperms. BioI. Rev. 45, 417-450.
STEBBINS, G. L., 1974: Flowering plants. Evolution above the species level.
Cambridge: Harvard Univ. Press.
TAKHTAJAN, A., .1969: Flowering plants. Origin and dispersal. Edinburgh:
Oliver & Boyd.
THORNE, R. F., 1974: The "Amentiferae" or Hamamelidae as an artificial
group: A summary statement. Brittonia 25, 395~-405.
WALKER, J. W., 1976a: Comparative pollen morphology and phylogeny of
the Ranalean complex. In: Origin and early evolution of angiospermH
(BECK, C. B., ed.), 241--299. New York and London: Columbia UIlI\,.
Press.
- 1976 b: Evolutionary significance of the exine in the pollen of primit.ive
angiosperms. Linn. Soc. Symp. Ser. 1, 251--308.

Address of the author: Prof. Dr. FRIEDRICH EHRENDORFER, Botanischcs

Inshtut der Universitat Wien, Hennweg 14, A-1030 Wien, Austria.
Plant Syst. Evo!., Supp!. 1,235-251 (1977)
© by Springer-Verlag 1977

Department of Botany, University of Singapore

Phyllocladus and Its Bearing on the Systematics of Conifers

By

Hsuan Keng, Singapore

Abstract: The so-called phylloclade--the flattened, simple or compound,

leaf-like organ which emerges from the axil of an acicular leaf-of Phyl-
locladu8 is a complicated lateral branch system. It possibly represents a
relic structure resembling the photosynthetic organ of the advanced mem-
bers of Progymnosperms. It is therefore suggested to uphold the erection of
a monogeneric family Phyllocladaceae as proposed by CORE, and to accept
the basic scheme of' classification of Coniferales as proposed by BUCH-
HOLZ.

Introduction
My previous studies on the genus Phyllocladus, one of the most
curious members of conifers, were based on a single Malesian species,
Ph. hypophyllus HOOK. f. (KENG 1963a, 1963b). From April to August,
1976, I visited Tasmania and New Zealand and had a good opportunity
to examine living and preserved material and herbarium specimens
of the other four known species, to have discussion with many specia-
lists in different fields, and to have access to much literature. It is
therefore necessary to further elaborate the theoretical conclusions
reached earlier (KENG 1974, 1975) and bring them into a more definite
form. This is presented in the following three sections: 1. On Phyllo-
cladus; 2. Comparison between Phyllocladus and the Archaeopteris-
like progymnosperms, and 3. Phyllocladus and the systematics of
conifers.

On Phyllocladus
The unique feature of the genus Phyllocladus L. C. & A. RICHARD ex
MIRBEL, as indicated in the generic name, is the presence of the so-
called phylloclade or cladode because it is emerged from the axil of
an acicular structure which is generally regarded as 'he true leaf (GOEBEL
1933). Phylloclades are extremely variable in shape and size, as they
are highly subjected to environmental changes. Bro~dly speaking
236 H. KENG:

there are two types of phylloclades, simple and pinnately compound.

Simple phylloclades are found in Ph. alpinus HOOK. f. (Fig. 21) and
Ph. aspleniifolius (LABILL.) HOOK. f. (Fig. 22), while pinnately com-
pound ones are characteristic of Ph. glaucus CARR. (Fig. 19), Ph. hypo-
phyllus (Fig. 20) and Ph. trichomanoides D. DON (Fig. 18).
Simple phylloclades are 1-4 cm long, mostly rhomboid in outline.
They have finely toothed or nearly entire margin, but sometimes
shallowly or deeply 2~3-lobed on one or both sides below. In still other
cases, they are pinnatifid but very rarely also found to be pinnately
compound. In general, only some, not all, of the pinnatifid phyllo-
clades bear a terminal bud which is covered with awl-shaped scales.
These buds either remain dormant, or give rise to a short shoot of
usually 1-4 cm long, coated spirally over its length by lanceolate scale
leaves which arc deciduous. Thc short shoot is normally ended up in
a new terminal bud which is surrounded by a crop of 3-5 simple or
sometimes pinnatifid phylloclades which are usually arranged III a
false whorl.
The pinnately compound phylloclades can reach 10-15 cm long,
consist of 5-10 (-12) segments. These segments, in taxonomic literature
(e.g. ALLAN 1961, LAUBENFELS 1969), are often conveniently designated
as phylloclades or cladodes. The pinnate phylloclades, like the simple
ones, mayor may not possess a terminal bud. In the latter case, there
is always a large prominent terminal segment which is often partly
lobed or partly fused with one or both lateral segments beneath. Fur-
ther development of the terminal bud into a short shoot is quite similar
to that of the simple phylloclades just described, except the short
shoot of the pinnate phylloclades is usually much longer and stouter,
and the new crop of phylloclades (3-5, up to 10) thus produced are
normally pinnately compound, very rarely simple.

A complete and comprehensive taxonomic account of the genus Phy/.

locladus was prepared by PILGER (1903). The latest taxonomic descriptions
of the species from various geographical regions are scattered in local floras
and other sources (e.g. ALLAN 1961, CURTIS & MORRIS 1975, LAUBENFELS
1969, etc.).
Morphological studies of some Phyllocladus species, principally 011
Ph. alpinus, and sometimes supplemented by Ph. glaucus or Ph. tricho·
manoides, were carried out by ROBERTSON (1906), KILDAHL (190811,), YOUNG
(1910), SINNOT (1913), HOLLOWAY (1937), BUCHHOLZ (1941) and others.
SINNOT (1913), for example, discussed the morphological significance
of the strobilar anatomy and compared the development of the male and
female gametophytes and of the embryo with other members of the Podo·
carpaceae. He stressed that the structure of the female gametophyte and
embryo of Phyllvctadus agrees more closely with that of Dacrydium, espe-
cially in the reduced Humber .of archegonia and the structure of the neck,
than with that of any other members of the family Podocarpaceae.
Phyllocladus and Its Bearing on the Systematics of Conifers 237

E0J:.. e •• • 00 ;Iuu 0
9 8

~~?@J
6

O·
4
't .. •.•
1

••

3
Fig. 1. An old pinnate phylloclade of Ph. glaUCtt8 giving rise to a short
shoot with a crown of new pinnate phylloclades. (Scale = 4 ern)
Figs. 2-9. Diagramatic transverse sections made from different parts as
indicated in Fig. 1

Vasculature of the segments of a pinnate phylloclade of Ph. hypo-

phyllus has been reported in a previous paper (KENO 1963 b). Fresh
material of Ph. glaucus (Fig. 1) and Ph. trichomanoides were sectioned
with the aid of a Hooker Plant microtome and a freezing microtome.
Their vasculatures are very similar and are briefly described below.
238 H. KENG:

Transverse sections (Fig. 2) of the stalk of last year's pinnate

phylloclade indicate that 1. in the centre a cylinder of secondary
xylem envelops the primary wood, and 2. in the cortex a leaf trace
enters the leaf subtending the lowermost segment of the pinnate
phylloclade.
As a result of activation of the terminal bud of the pinnate phyllo-
clade, a new shoot is formed. This new shoot is covered with spirally
arranged, single-veined, deciduous, lanceolate leaves and is crowned
with a crop of young pinnate phylloclades (Fig. 1). Transverse sections
from the lower level of the shoot (Fig. 3) show an undulating central
cylinder of primary growth with leaf-traces in the cortex. Sections
from a higher level indicate that the undulation becomes more pro-
nounced, finally five large V-shaped ribs alternating with four or five
smaller ones gradually become evident. Further upwards (Fig. 4),
a single bundle, associated with a resin canal, diverges from the pole
of one of the ribs to supply the leaf which is subtending the entire
pinnate phylloclade. This is followed by the departure of two blocks
of vascular tissues which constitute the rest of the rib. These two
blocks of vascular tissues form the basic vasculature of the lowermost
pinnate phylloclade. Subsequently the departure of another single
hundle, followed by two blocks of vascular tissues which form the
hasic vasculature of a second pinnate phylloclade (Fig. 5). Eventually,
a crown of 3-5, up to 8-10 new pinnate phylloclades can be formed.
Sections of the stalk of a young pinnate phylloclade show a central
vascular system consisting of two closely associated blocks of vascular
tissues. The divergence of a single bundle from one side followed by
two small blocks of vascular tissue from the same side constitutes
the vasculature of the lowermost segment and its subtending leaves
(Figs. 6, 7). As soon as these two small blocks of vascular tissues enter
the base of a segment, divergence begins (Fig. 8). A single trace enters
into the leaf-like structure of the segment, and two smaller patches
of vascular tissues follow to form the basis of the vasculature of one
of the several small branches within a segment. Eventually, these smal-
ler branches result in five to seven leaf-like structures with their
xylem portions oriented towards the centres (Fig. 9).
The wood structure of Phyllocladus species has been described
by GREGUSS (1955), PATEL (1967), and others. The latter author re-
marked that the wood of Phyllocladus species can be separated from
that of other New Zealand members of the Podocarpaceae by its distinct
rings, large cross-field pits, abundant tangential pits and lack of axial
parenchyma. He further emphasized that the presencc of abundant
tangential pits in PhyllocladU.9 can he regarded as a useful diagnostic
feature.
PhyUocladus and Its Bearing on the Systematics of Conifers 239

Haploid chromosome numbers of three New Zealand species of

Phyllocladus were reported by HAIR & BEUZENBERG (1958) which
are uniformly 9, the lowest among the living conifers. Those of the
Tasmanian and Malesian species remain unknown.

At the present time the genus PhyllocladU8 is confined to the Australasian

and Malesian regions (Fig. 10). Fossil records, however, reveal its former

Fig. 10. Present (in solid lines) and past (in broken lines) distribution of
Phyllocladu8. Five living species are found in 1. Tasmania (with one species),
2. New Zealand (with three spp.) and 3. Malesia (with one sp.). (Base map
copyright by the University of Chicago Press)

much wider range of distributiOIi in geological time. After enumerating

and evaluating the palaeo-botanical data, COUPER (1960) reaches the fol-
lowing conclusions (Fig. 10):1. Macrofossils identified as PhyllocladU8
on cuticular structure and morphology of the cladodes are recorded from
the Oligocene of continental Australia and the Oligocene, later Tertiary
and Quarternary of New Zealand. 2. Microfossils (pollens) are· found spo-
radically in Tertiary sediments in Australia, Tasmania, and New Zealand.
They are also found in the Lower Tertiary of Seymour Island in western
Antarctica. 3. No fossils of Phyllocladu8 are known from the Cretaceous
and Tertiary of northern South America, New Guinea, Borneo or India.
A phyllocladean affinity has been suggested for some fossil woods from
southern South America, but these identifications are inconclusive.
The absence of fossils in the tropics and in the northern Hemisphere
deserves special attention. It supports the view that the present extension
of Phyllocladus into eastern Malesia was a relatively recent development
(HAIR 1963, LAUBENFELS 1969).
240 H. KENG:

Comparison Between Phyllocladus and the Archaeopteris-like

Progymnosperms
The photosynthetic organs of Phyllocladus are generally known
as phylloclades or cladodes. The prevailing view to-day in interpreting
the origin and nature of these organs is expressed in the following
statement made by ROBERTSON (1906: 259) over half a century ago:
"Phyllocladus is ... characterized by the reduction of its true leaves
to pointed scales, and the expansion of certain of its stem-branches
into flattened leaf-like structures". A simple photosynthetic organ
as commonly found in Ph. alpinus and Ph. aspleniifolius (the type
species of the genus) (Figs. 21, 22) could be summarily described as a
phylloclade, without paying much attention to its detailed internal
structure. Superficially it is a flattened green organ arising from the
axil of a true leaf, the latter often being reduced. In the case of the
pinnately compound photosynthetic organs as found in Ph. glaucus,
Ph. hypophyllus and Ph. trichomanoides (Figs. 18, 19, 20), however,
such terms as phylloclade or cladode are hardly applicable. Observa-
tions made on the Malesian species, Ph. hypophyllus (KENG 1963b)
reveal that the whole pinnate photosynthetic organ is subtended by
a leaf-like structure, and that it consists of usually 6-8 segments. Each
segment is also situated in the axil of a leaf-like structure and it is
composed of a number of smaller branches. These smaller branches,
moreover, are individually subtended by a leaf-like structure but they
are fused together laterally. Examinations of the pinnate photosynthetic
organs of the mature plants of Ph. glaucus and Ph. trichomarwides
(Figs. 14, 15) show basically the same configuration as that of the
Malesian species, but the bipinnate nature becomes much more pro-
nounced in the photosynthetic organs of the seedlings of Ph. tricho-
manoides (Figs. 16, 17 and 28).

Transverse sections of a mature segment of the pinnate photosynthetic

organs also indicate some unusual features (KENG 1963b). The upper and
lower epidermis are well defined. Mesophyll is more or less differentiated
into two parts comparable to palisade and spongy tissues. The deeply
sunken stomata possess guard cells almost indistinguishable from those
found in the juvenile leaves. Furthermore, many of these photosynthetic
organs, simple or pinnate, possess a terminal bud, which upon activation,
gives risc to a short shoot with a numbcr of spirally arranged leaves on the
axis together with a crown of new photosynthetic organs (Fig. 1).

The photosynthetic organ of Phyllocladus is thus much too com-

plicated to be designated as a phylloclade in its original sense as re-
ferring to the specialized branch of an angiospermous plant. Or in other
words, phylloclade is a misnomer when applied to the photosynthetic
Phyllocladus and Its Bearing on the Systematics of Conifers 241

organ of Phyllocladus. It becomes apparent that the photosynthetic

organ of Phyllocladus represents a unique structure among gymno-
sperms. Gymnosperms were plausibly derived from the Psilophyton-
like earlier vascular plants through progymnosperms (BECK 1962,
1966; BANKS 1968; and others). It is therefore probably not too far-
fetched to attempt to compare the photosynthetic organs of Phyllo-
cladus with those of Archaeopteris, one of the most advanced pro-
gymnosperms.
The photosynthetic organs of Archaeopteris (Fig. 11) was thought at
one time to be the ancestral form of ferns. They were thus described as
a bipinnate "frond" (e.g. BECK 1962). Each "frond" is composed of pinnae
and pinnules arranged on the rachis and its branches respectively. This
plant has been specially noted for the possession of a large stipule.like
structure at the base of the "frond" and for the presence of the so-called
rachial pinnules. CARLUCCIO et al. (1966) found the so-called "rachis"
has a radially symmetrical vascular system like a stem and that the axes
of the lateral pinnas are characterized by a similar type of vasculature.
Based on these facts, they concluded that the foliar organ of Archaeopteris
is not a bipinnate "fro.nd" but rather should be regarded as a planated
lateral branch system. According to this interpretation, the actual leaves
are helically arranged and comprise 1. the pairs of "rachial pinnules",
and 2. the fertile and sterile appendages (pinnules) borne on the lateral
axes which had previously been regarded as pinnae. BECK (1970) later
accepted this view and further noted that the entire lateral branch system
appears to be subtended by a leaf which he had earlier (1962) designated
as a "stipule".

Photosynthetic organs of Archaeopteris macilenta LESQ. and Phyllo-

cladus trichomanoides are presented in Figs. 11 and 14, and for con-
venience 'of comparison and discussion, a set of new, simple, lucid
terms are proposed in brackets along with the terms used in the previous
sections of this paper on Phyllocladus on the one hand, and the terms
used by BECK in his 1971 paper on Archaeopteris on the other.
The lateral branch system (primary branch) of Archaeopteris is
composed of a number of distichously disposed lateral branches (se-
condary branches). The large notched foliar structure (primary laminar
appendage) has only recently been recognized as a leaf (BECK 1970).
The leaves, or formerly the rachial pinnules (secondary laminar ap-
pendages), along the main axis are probably totally different from
those on the lateral branches. The latter are possibly referable both
to the tertiary branches and the tertiary laminar appendages (Fig. 12):
those distichously disposed along the secondary branches might re-
present the tertiary branches, while those inserted between the distichous
leaves represent the tertiary laminar appendages. Another possibility
would be that these tertiary laminar appendages (like the secondary
laminar appendages of Phyllocladus), might be of deciduous nature,
Plant 8yst. Evol., 8uppl, 1 16
~
..•. ITon. brandl l
IT.... I ""'-Q~I

...... "11"-. from _1,,'9

ISocondoty brQod\ l
16 " i'Soc1Om_ Q!'P0'"4 I

IT.... Iotn QPP I

Figs. 11-16. Comparison of a lateral branch system of Archaeopteris ma-

cilenta (Fig. 11) with a young pinnate phylloclade of Ph. trichomanoides
(Fig. 14) . Figs. 12 and 13, sterile and fertile lateral branches of Archaeop-
teris. Figs: 15 and 16, cleared segments showing the vasculatures from an
adult plant and a seedling of Phyllocladus, respectively. (Fig. 11 from BEOK
1962, Figs. 12 and 13, from BEOK 1971)

{\r 18

~ 1l:1'7 ~20
, 2/J ~2' V
Figs. 17-22. Photosynthetic organs of Phyllocladus, (a) from the seedling
(Fig. 17) and adult plant (Fig. 18) of Ph. trichomanoides, and (b) from the
adult plants of Ph. glaucus (Fig. 19), Ph. hypophyllus (Fig. 20), Ph. alpinus
(Fig. 21) and Ph. aspleniifolius (Fig. 22). (All schematic)
H. KENO: Phyllocladus and Its Bearing on the Systematics of Conifers 243

thus the persistent ones are all tertiary branches. These tertiary bran-
ches and their laminar appendages become extremely clearly differentia-
ted in the fertile lateral branches (Fig. 13, redrawn from BECK 1971,
or as reconstructed by PHILLIPS et al 1972).

The putative ancestor of Phyllocladus probably possessed a large photo-

synthetic organ resembling closely that of Archaeopteris. It probably

Figs. 23- 27. Photosynthetic organs (± reduced) of Phyllocladus bearing

oyulate strobili: Ph . trichomanoides (Fig. 23), Ph. glaucus (Fig. 24), Ph.
hypophyllus (Fig. 25), Ph. alpinus (Fig. 26) and Ph. aspleniifolius (Fig. 27).
(All schematic)

consisted of a twice· divided primary branch, and several of these branches

were probably arranged on the main axis forming a false whorl (cf. the
reconstruction of Tetraxylopteris by BECK 1957, in his Fig. 32). Each
primary branch probably bore a large terminal bud. At present the photo-
synthetic organs found in the seedlings of Ph. trichomanoides (which has
conspicuous tertiary branches with subtending tertiary laminar appendages
as well as a large terminal bud) is probably the nearest to the putative
ancestral form (Figs. 17 and 28).
The simple photosynthetic organs (of which further ramification has
been arrested) as found in Ph. aspleniifolius (Fig. 22) and Ph. alpinus
(Fig. 21). probably represent a more speciali?ed form.
16·
244 H. KENO:

With regard to the reproductive structures, among Phyllocladus

species, Ph. glaucus (Fig. 24) possesses the most well-organized ovulate
cones which consist of a large number of cone-scales. Ph. hypophyUus
(Fig. 25), has less well-organized cones and the latter are generally
seated on the notch of a secondary branch (segment). The combination
of vegetative and reproductive functions performed by a single photo-
synthetic organ is highly remarkable among the living seed plants.
The ovulate reproductive structure of a putative ancestor of Phyllo-
cladus probably possessed a highly well-organized cone seated on the
upper part of a secondary branch.
The above comparison indicates some similarity between the photo-
synthetic and reproductive organs of progymnospermous Archaeopteris
and those of coniferous Phyllocladus, yet the geological and morpho-
logical gaps between them are enormous. For instance, Archaeopteris
is found in Upper Devonian to the beginning of Carboniferous, over
340 million years ago (BANKS 1968), whereas Phyllocladus fossils are
known only from the Oligocene onwards, less than 40 million years
ago (COUPER 1960). Morphological gaps include the arrangement of
pits in the radial walls of the tracheids which in Archaeopteris are in
groups separated by spaces free of pits (ARNOLD 1930) and the hetero-
sporous sporangia which stand erect in one or more rows on adaxial
surface of fertile ultimate appendages (BECK 1976). In Phyllocladus
pits occur on radial walls of tracheids in one or rarely in two rows
(PATEL 1967), the staminate and ovulate strobili are well organized,
and the ovule and seeds are of highly evolved form.
Therefore any attempt to postulate a connection between them
needs to be extremely cautious. Without further evidence, it can only
be said that the photosynthetic and reproductive organs of Phyllocladus
might have been derived from a putative ancestor which had general
features resembling Archaeopteris or its related genera.

Phylloc:ladus and the Systematics of Conifers

It is not intended to give an exhaustive historical sketch of classi-
fication systems of conifers here, as this has been adequately done
by PILGER (1926), FLORIN (1955) and others. For the convenience of
discussion, the following brief account will be sufficient.
In the first edition of ENGLER & PRANTL's "Die natlirlichen Pflan-
zenfamilien", EICHLER (1887) divided Conilerae into two subfamilies:
Pinoideae (with two tribes, Abietineae and Cupressineae) and Taxoide~
(also with two tribes, Podocarpeae and Taxeae-the genus Phyllocladus
was included in the latter). In the 9th and 10th edition of ENGLER'S
Syllabus (ENGLER & Grw 1924). the two subfamilies were elevated
PhyUodadus and Its Bearing on the Systematics of Conifers 245

Fig. 28. A seedling of Phyllocladus trichomanoides showing the bipinnate

photosynthetic organs. Inset, a bipinnate photosynthetic organ. (Photo-
graphed by Mr. C. J. MILES; material supplied by Dr. B. P. J. MOLLOY,
Botany Division, DSIR, Christchurch, New Zealand)
246 H. KENG:

to two families, Taxaceae and Pinaceae, with the former further divided
into three subfamilies (namely, Podocarpoideae, Phyllocladoideae and
Taxoideae) and the latter, four tribes (A raucarieae , Abieteae, Taxodieae
and Cupresseae). All the subfamilies and tribes, except the Phyllo-
cladoideae (which were retained as a subfamily in Podocarpaceae) were
promoted to family rank by PILGER (1926), plus a new family Cephalo-
taxaceae, formerly under the Taxaceae. Thus the Coniferales constitutes
the following seven families: Taxaceae, Podocarpaceae, Cephalotaxaceae,
Araucariaceae, Pinaceae, Taxodiaceae, and Cupressaceae.
The composition and classification of the Conifers were significantly
altered in the 12th edition of ENGLER'S Syllabus (MELCHIOR & WER-
DERMANN 1954), in which PILGER and MELCHIOR proposed the fol-
lowing scheme:

Class Coniferopsida
Order Cordaitales
Order Coniferales (Lebachiaceae, Voltziaceae, Cheirolepidaceae, Proto·
pinaceae, Pinaceae, Taxodiaceae, Cupressaceae, Podocarpaceae, Ce-
phalotaxaceae, A raucariaceae)
Class Taxopsida
Order Taxales (Taxaceae)

Following this brief account of the historical background, discussion

will be concentrated on three essential problems, namely, 1. the in-
clusion of the Taxaceae in conifers, 2. the establishment of the Phyllo-
cladaceae, and 3. the systematics of conifers.

The Inclusion of the Taxaceae in Conifers

SAHNI (1920) first noted that the taxads (including Cephalotaxus)
were so different from the other conifers that they deserved the rank
of a separate order, Taxales. FLORIN (1938-1945, summarized in 1948,
1951) studied the reproductive organs of fossil taxads-Palaeotaxus
of Triassic and Taxus (or Marskea, cf. HARRIS 1976) of Jurassic-and
found that they were clearly different from those of true conifers.
Therefore he accepted the separation of the taxads-including the
following five genera: Taxus, Amentotaxus, Torreya, Austrotaxus and
Pseudotaxus, but not Cephalotaxus-from the Coni/erae into a separate
order, Taxales, and even a new class, Taxopsida. Many authors, such
as CHAMBERLAIN (1935: 439), PULLE (1937) and others, expressed a
different view, namely that the single ovulate strobilus of taxads is
most probably derived from the multiovulate cone. The present writer
(KENG 1963b, 1969) pointed out that the development and evolution
of the ovulate strobili in Phyllocladus might indicate the possible way
in which the single, seemingly terminal ovule of taxads could have been
Phyllocladus and Its Bearing on the Systematics of Conifers 247

achieved. He also drew attention to the fact that PhyUocladus is closely

related to the Taxaceae and Podocarpaceae on the one hand, and Amento-
laxus, to the Taxaceae and Cephalotaxaceae on the other. Therefore
he suggested that the Taxaceae should be retained in the Coniferales.

From palaeobotanical studies, HARRIS (1976) recently observed that

"there is something in common between the ovuliferous. short shoot of
Taxus and the fertile shoot of Lebachia", and suggested that early fossil
conifers like W alchia could "shift" their ovules from a lateral to a terminal
position. He therefore implied that the class Taxop8ida is unnecessary.
(Both Lebachia and Walchia belong to the family Lebachiaceae which is
the earliest member of the Coni/erale8, cf. MELCHIOR & WERDERMANN
1954.)

The Establishment of the Phyllocladaceae

The somewhat intermediate morphological features of Phyllocladus
between the Taxaceae and Podocarpaceae lead to quite diverse taxo-
nomic treatment of the genus, depending on the relative weight as-
signed to these features. For example, such characters as the leafy
microsporophyll bearing two microsporangia at the base and the winged
microspores resemble those of the Podocarpaceae; the erect ovules and
the arillate seed appear to be similar to those of the Taxaceae. For
this reason, three different schemes of classification of this genus have
been proposed. The first includes it in the Podocarpaceae (or formerly
Podocarpoideae or Podocarpineae) (e.g. PILGER 1926); the second in-
cludes it in the Taxaceae (or formerly Taxoideae, Taxineae) (e.g. EICHLER
1887); and the third is to consider Phyllocladus as representing a
separate taxon coordinate with and intermediate between the Podo-
carpaceae and Taxaceae (PILGER 1903, ENGLER & GILG 1924, CORE
1955).

Morphological studies of Phyllocladus by many authors (e.g. KILDAHL

1908b) led to more or less the same conclusions as ROBERTSON'S (1906:
264): "Phyllocladus occupies an intermediate position between Podocar-
poideae and Taxoideae, but with greater affinities for the former". YOUNG'S
view (1910: 91) slightly deviates from this, she observed that PhyllocladU8
has primitive characters of the Taxineae which become eliminated in the
Podocarpineae, and it also has primitive characters of the Podocarpineae
which have been eliminated in the Taxineae.

The present writer (KENG 1973, 1975) emphasized the fact that
the photosynthetic organs of Phyllocladus commonly known as phyllo-
clades are probably a very ancient structure, and that they could have
been misinterpreted morphologically. It may represent a relic structure
of which the distinction between leaf and branch has not been sharply
differentiated. Therefore the present writer supported the proposition
248 H. KENG:

made by CORE (1955) to establish the family Phyllocladaceae. He further

considered this monogeneric family to be probably the most primitive
members of the Coniferae. The fact that this family has external and
internal morphological characters in common with the Taxaceae on
the one hand, and with the Podocarpaceae on the other could be ex-
plained by assuming that this family inherited these traits directly
from the ancestral stock from which both Taxaceae and Podocarpaceae
were subsequently evolved.

The Systematics of Conifers

With the re-instatement of the Taxaceae in the Coniferales, and
with the establishment of the family Phyllocladaceae, the living members
of this order thus include the following eight families: Phyllocladaceae.
Taxaceae, Podocarpaceae, Cephalotaxaceae, Pinaceae, Taxodiaceae, Cu-
pressaceae, and Araucariaceae. This treatment resembles closely the
basic scheme proposed by PILGER (1926) with the addition of the Phyl-
locladaceae which was formerly a sub-family of the Podocarpaceae.

Attemps have been made, from time to time, to subdivide this order
into two to several smaller groups. For examples, GAUSSEN (1944-1952,
cited in FLORIN 1955) recognized three sub-orders: (a) Taxineae, with
the Taxaceae, (b) Podocarpineae, with the Podocarpaceae, and (c) Pinoi-
dineae, with the remaining families. PULLE (1937) created five orders:
Araucariales, Podocarpales, Pinales, Oupressales (with Taxodiaceae and
Oupressaceae) and Taxales (with Oephalotaxaceae and Taxaceae). The
fragmentation of this order culminated in a system of FLORIN (in ERDTMAN
1952) in which two classes and six orders were proposed. They are (a) the
Taxad Class or Taxopsida, with the order Taxales, and (b) the Conifer
Class, or Ooniferopsida, with five orders: Araucariales, Podocarpales, Pinales,
Oupressales, and Oephalotaxales.
BUCHHOLZ'S (1934) proposal to divide the order Ooniferales into two
sub-orders, which he called Phanerostrobilares and Aphanostrobilares,
appears to be the most reasonable one and thus deserves special attention.
It adheres in principle, to the original schemes of EICHLER (1887) and ENG-
LER & GILG (1924). This general scheme was later adopted by CHAMBERLAIN
(1935, as Taxares and Pinares), JANCHEN (1949, as Taxales and Pinales)
and others. The main difference between these two sub-groups, in the
words of CHAMBERLAIN (1935: 229, 230) is the one sub-group (called Pha-
nerostrobilares, Pinares, Pinineae or Pinales by different authors) "with
an obvious (ovulate) cone", including Pinaceae, Araucariaceae, Taxodiaceae
and Oupressaceae; and the other sub-group (called Aphanerostrobilares,
Taxares, Taxineae or Taxales) "without such an obvious cone", including
Podocarpaceae, Taxaceae and Oephalotaxaceae, and to be added here, the
Phyllocladaceae.

At the first glance, the validity of this distinction seems to be

dubious, as it is ambiguous. Members of the first sub-group do possess
Phyllocladus and Its Bearing on the Systematics of Conifers 249

very obvious cones, although most of the members of the second sub-
group lack an obvious cone, yet in some cases such as Cephalotaxus
(Cephalotaxaceae) , Microcachrys (Podocarpaceae) and Phyllocladus,
especially Ph. glaucus (Fig. 24) (Phyllocladaceae) , "obvious" ovuliferous
cones are in fact present.
FLORIN'S monumental work (1938-1945, summarized in 1951) has
almost conclusively demonstrated that the seemingly simple ovuli-
ferous cones of Pinus and possibly other members of one sub-group
(Pinineae) have evolved from a much more complicated structure
such as those found in the palaeozoic fossil groups Lebachia, Ernestio-
dendron, Walchia, and Pseudovoltzia. But whether the same conclusion
could be extended to explain the cones of the members of the other
sub-group (Taxineae) is highly debatable. Without further evidence,
it is thought better to keep these two subgroups separate.
For this reason, the following scheme of classification which is a
modification of BUCHHOLZ'S (1934), is presented below.

Order Coniferales
Sub-order Taxineae
Family 1. Phylloeladaceae
Family 2. Podocarpaceae
Family 3. Taxaceae
Family 4. Cephalotaxaceae
Sub-order Pinineae
Family 5. Pinaceae
Family 6. Araucariaceae
Family 7. Taxodiaceae
Family 8. Cupressaceae

This study was carried out during- a Sabbatical Leave from the Uni-
versity of Singapore. An Inter·University Exchange Scheme grant was
awarded by the British Council. Professors L. CONSTANCE and W. R.
PHILIPSON and Dr. K. R. SPORNE kindly criticized the manuscript. Pro-
fessors W. R. PHILIPSON (University of Canterbury, Christchurch, New Zea-
land) and W. D. JACKSON (University of Tasmania, Hobart, Australia)
kindly provided laboratory and other facilities, while many individuals,
including B. C. ARNOLD, G. BROWNLIE, R. K. CROWDEN, W. M. CURTIS,
J. W. DAWSON, H. K. MAHANTY, B. J. MOLLOY, L. G. MOORE, A. E. OR-
CHARD, P. WARDLE and many others provided helpful discussions. My
wife, Mrs. R. S. KENO prepared several drawings for illustration.

References
ALLAN, H. H., 1961: Flora of New Zealand, Vol. 1. Wellington: R. E. Owen.
ARNOLD, C. A., 1930: The genus Callixylon from the U.'lper Devonian of
Central and Western New York. Pap. Mich. Acad. Sc. 11, 1-50.
BANKS, H. P., 1968: The early history of land plants. In: Evolution and
Environment (DRAKE, E. T., ed.), 73-107. New Haven: Yale Univ.
Press.
250 H. KENG:

BECK, C. B., 1957: Tetmxylopteris schmidtii gen. et sp. nov., a probably

pteridosperm precursor from the Devonian of New York. Amer. J. Bot.
44, 350-367.
1962: Reconstructions of Archaeopteris and further consideration of its
phylogenetic position. Amer. J. Bot. 49, 373-382.
1966: On the origin of gymnosperms. Taxon 15, 337--339.
1970: The appearance of gymnospermous structure. BioI. Rev. 43,
379-400.
1971: On the anatomy and morphology of lateral branch systems of
Archaeopteris. Amer. J. Bot. 58, 758-784.
1976: Current status of the Progymnospermopsida. Rev. Palaeobot.
Palyn. 21, 5-23.
BUCHHOLZ, J. T., 1934: The classification of Conilemles. Trans. Illinois
Acad. Sci. 25, 112-113.
- 1941: Embryogeny of the Podocarpaceae. Bot. Gaz. 103, 1--37.
CARLUCCIO, L. W., HUEBER, F. M., and BANKS, H. P., 1966: Archaeopteris
macilenta, anatomy and morphology of its frond. Amer. J. Bot. 53,
719-730.
CHAMBERLAIN, C. J., 1935: Gymnosperms, structure and evolution. Chicago:
Univ. Press.
CORE, E. L., 1955: Plant taxonomy. Englewood: Prentice-Hall.
COUPER, R. A., 1960: Southern Hemisphere Mesozoic and Tertiary Podo-
carpaceae and Fagaceae and their palaeogeographic significance. Proc.
Roy. Soc. B. 152, 491-500.
CURTIS, W. M., and MORRIS, D.!., 1975: The Student's Flora of Tasmania.
2nd ed. pt. 1. Hobart: Government Printer.
EICHLER, A. W., 1887: Conilerae. In: Die nat. Pflanzenfamilien, Teil II, 1
(ENGLER, A., and PRANTL, K.), 28-116. Leipzig: W. Engelmann.
ENGLER,. A., and GILG, E., 1924: Syllabus der Pflanzenfamilien. Ed. !I &10.
Berlin: Gebr. Borntraeger.
ERDTMAN, H., 1952: Chemistry of some heartwood constituents of conifers
and their physiological and taxonomic significance. In: Progress ill
Organic Chemistry, 22-63. London: Butterworths Sc. PubIs.
FLORIN, R., 1948: On the morphology and relationships of the 7'axaceae.
Bot. Gaz. 110, 31-39.
1951: Evolution in Cordaites and Conifers. Acta Hort. Berg. 15, 285-388.
1955: The systematics of the Gymnosperms, In: A century of progress
in the natural sciences 1853-1953, 233--403. San Francisco: Calif.
Acad. Sc.
GOEBEL, K., 1933: Organographie der Pflanzen. 3. Auf!. Jena: G. Fischer.
GREGUSS, P., 1955: Identification of living Gymnosperms on the basis of
Xylotomy. Budapest: Akad. Kiado.
HAIR, J. B., 1963: Cytogeographical relationships of the southern podo-
carps. In: Pacific Basin Biogeography (GRESSI'I'T, J. L., ed.), 401---414.
Honolulu: Bishop Museum Press.
and BEUZENBERG, E. J., 1958: Chromosomal evolution in the Pudu-
carpaceae. Nature (London) 181, 1584-1586.
HARRIS, T. M., 1976: The Mesozoic Gymnosperms. Rev. Palaeobot. Palyn.
21, 119-134.
HOLLOWAY, J. T., 1937: Ovule anatomy and development and embryogeny
in PhyllocladU8 alpinU8 HOOK. and in P. glaucus CARR., Trans. R. Soc.
N. Z. 82, 329-410.
Phyllocladu8 and Its Bearing on the Systematics of Conifers 251

JANCHEN, E., 1949: Das System der Koniferen, Sitzungsber. Oest. Akad.
Wiss. math.-naturw. Kl. (div. 1) 158, 155-162.
KENG, H., 1963a: Phyllocladu8 hypophyllus HOOK. f. Gard. Bull. Singaporc
20, 123-126.
1963 b: Aspects of morphology of PhyllocladU8 hypophyllu8. Ann. Bot.
n.s. 27, 69-78.
1969: Aspects of morphology of A mentotaxu8 formo8ana with a note on
the taxonomic position of the genus. J. Arnold Arb. 50, 432-446.
1973: On the family Phyllocladaceae. Taiwania 18, 142~145.
1974: The phylloclade of PhyllocladU8 and its possible bearing on thc
branch systems of progymnosperms. Ann. Bot. n.s. 38, 757-764.
1975: A new scheme of classification of the Ooniferale8. Taxon 24,
289-292.
KILDAHL, N. J., 1908a: The morphology of Phylloclad118 alpinus. Bot.
Gaz. 46, 339-348.
- 1908 b: Affinities of Phyllocladu8. Ibid. 46, 464-465.
LA UBENFELS, D. J. DE, 1969: A revision of the Malesian and Pacific Rain
Forest Conifers. I. Podocarpaceae, in part. J. Arnold Arb. 50, 274-369.
MELCHIOR, H., and WERDERMANN, E., 1954: A. ENGLER'S Syllabus del'
Pflanzenfamilien. Ed. 12, I. Berlin: Gebriider Borntraeger.
PATEL, R. N., 1967: Woody anatomy of Podocarpaceae indigenous to New
Zealand. 3. Phyllocladu8. N. Z. J. Bot. 6, 3-8.
PHILLIPS, T. L., ANDREWS, H. N., and GENSEL, P. G., 1972: Two hetero·
sporous species of Archaeopteris from the Upper Devonian of West
Virginia. Palaeontographica 139 B, 47-71.
PILtlER, R., 1903: Taxaceae. In: Das Pflanzenreich, IV. (ENGLER, A.), 5.
Leipzig: W. Engelmann .
..- 1926: Ooniferae. In: Die nat. Pflanzenfamilien, 2. Auf!. (ENGLER, A.,
and PRANTL, K.), 13, 121--407. Leipzig: W. Engelmann.
PULLE, A., 1937: Remarks 011 the system of the Spermatophytes. Med.
Bot. Mus. Utrecht 43, 1--17.
ROBERTSON, A., 1\)06: Some points ill the lllorphology of Phyllocladll8
alpimts HOOK. Ann. Bot. 20, 259-265.
SAHNI, B., 1920: On certain archaic features in the seed of Taxu8 baccata,
with rcmarks on the antiquity of the 'l'axineae. Ann. Bot. 34, 117-133.
SINNOT, E. W., 1913: The morphology of the reproductive structures in
the Podocarpinae. Ann. Bot. 27, 39-82.
YOUNG, M. S., 1910: The morphology of l'odocarpineae. Bot. Gaz. 50,
82-100.

Address of the author: Prof. Dr. H. KENG, Department of Botany,

Bu\,it Timah Road, Singapore 10.
Plant Syst. Evol., Supp!. 1, 253-283 (1977)
© by Springer-Verlag 1977

Botanical Museum of the University of Copenhagen, Denmark

A Commentary on a Diagrammatic Presentation of the

Angiospel'ms in Relation to the Distribution of Character States
By

Rolf Dahlgren, Copenhagen

Abstract: The difficulty, not to say impossibility, of presenting groups

of orders as variously shaped figures in a two-dimensional diagram to
indicate relationships is first emphasized. However, the method can be
used to demonstrate the distribution of character states. The distributional
patterns of a number of character states are presented with brief comments,
viz. sympetaly, epigyny and perigyny, centrifugal succession in multi-
staminate androecia, pollen grains with one aperture, the successive type
of microsporogenesis, apocarpy, three-nucleate pollen grains (the "dry"
stigmas and the C4 pathways being largely correlated with this attribute),
unitegmic and tenuinucellate ovules and cellular and helobial types of
endosperm formation. In addition, the distribution in dicotyledons is out-
lined for the synthesis of eUagic acid, iridoids, benzylisoquinoline alkaloids,
glucosinolates and polyacetylenes derived from fatty acids. In mono-
cotyledons other characters prove useful in gross taxonomy, for instance
the presence of vessels in the stem, the presence of starch in the endosperm
and the presence of oxalate raphides and silica bodies, as well as more
conventional characters such as floral symmetry, epigyny versus hypogyny,
stamen number and fruit type. Comments on the possible taxonomic
significance of the characters shown are included. An apparent connection
between certain constellations of orders (superorders) is pointed out.
Finally, an appeal is made for further cooperation in presenting the distribu-
tion of other character states which may be of taxonomic importance at
the higher levels.

Introduction
The method of illustrating an angiosperm system as an imaginary
phylogenetic tree in transection was first used by the author and his
colleagues in a. textbook of angiosperm taxonomy (DAHLGREN et al.
1974, 1975a and b, 1976a). Some 25 different characters, most of which
were <?Onsidered to be of importance in gross taxonomy, yrere shown
in this diagram. The details of the system and the method of presenta-
tion needed further elaboration, however. A revised system with a
diagram was presented in 1975 (DAHLGREN 1975a). Diagrams showing
254 R. DAHLGREN:

the distribution of some embryological features were presented in

DAHLGREN (1975b).
As in DAHLGREN (1975a) reservations must be made when presenting
a -two-dimensional diagram to illustrate phylogenetic relationships.
The attempt has been made to place the variously shaped figures

Fig. 1. The diagrams lIsed in this article presented as the transection of an

imaginary phylogenetic "shrub"

representing the orders close to one another when members of the

orders are similar in several characters that reflect presumably phylo-
genetic relationships. The size of these figures is as far as possible an
expression of the number of species in the orders. Where members of
an order closely resemhle members of more than one other order the
figure representing the order has been extended in two or more direc-
t.ions. Other figures have been rounded.
Differentiation within the ancestral forms of angiosperm orders
naturally was extensive and affected the appearance of different parts
Commentary on a Diagrammatic Presentation of the Angiosperms 255

of the plant. While some characters underwent considerable evolu-

tionary change others remained at a level that most of us would regard
as primitive. Greater importance is often attached to the primitive
state of pollen grains (viz. the monosulcate type) or to that of the gy-
noecium (viz. the apocarpous) than to the primitivity of other characters.
Earlier evolutionary trends may have continued to develop in the
same direction, but in many cases evolution may have been reversed
or at least taken a different course. Whereas there may be a conspicuous
similarity between major taxa of plants with regard to certain characters
this may be counteracted by equally conspicuous differences in other
characters. Thus a diagram showing the variation pattern of one
character will seldom or never coincide with the pattern of variation
of another independent character. Mfinities as judged by the sum of char-
acter similarities will therefore always be relative. Moreover, no such
diagram can be completely accurate, since the position of a group as
determined by similarities to two or three other groups in a two-dimen-
sional diagram will inevitably correspond poorly with its position in
relation to a fourth of fifth group whose positions are determined by
similarities to still other groups.
The probability of a polyphyletic origin for the angiosperms has
been stressed, for example, in several papers by MEEUSE. But I do
not consider it likely that the angiosperms have evolved from widely
different groups of gymnosperms. The combination of several very
characteristic attributes (secondary endosperm, 8-nucleate embryo
sac, companion cells in the phloem, etc.) would hardly have evolved
independently in different gymnosperm groups.
Is there anything to be gained from a diagrammatic presentation
of the angiosperms? Firstly, I am convinced that for numerous features
their distributional patterns can be shown in the diagram presented
here. Secondly, the evaluation of distributional patterns of different
categories of characters will doubtless cast light on phylogenetic rela-
tionships between groups of orders.
The distributional patterns of a number of characters will be shown
together with brief comments on the following pages. Some major
features of the system presented will also be pointed out together with
a discussion on the subclasses of CRONQUIST (1968) and TAKHTAJAN
(1969).

Distributional Patterns of Some Characters Within the Angiosperms

Some of the following characters will either be dealt with in greater
detail in forthcoming papers or have already been presented (DAHL-
GREN 1975b, JENSEN et al. 1975, BEHNKE and DAHLGREN 1976).
~
01
C>

........,..
• ~. "L""""'f,"-U "(l.UMM)N
. Am
r~"':I~~~.I.I ~ ~ NY...u.(AUI
MVtM!I()C;HAIII'''LIJ
1l1\MID."'1

::0

z
I
PoII.n graiM
~ ",th _ _ tur.
_ .111~ ... ·t •

.. l~urOl. Monolu lc •••• nd mGftOl.u lc:.,.-Clethted

W~ ..... 2 .... """. op.rtu ...

l"ig. 2. Distribution of pollen grains with one aperture or of pollen grains without or with more than one aperture considered
t.o be derived from types with one aperture. In MagnoZianae the apertures do not lie in the equatorial plane except in Illiciales
and BOrne Chloranthaceae. Inaperturate pollen grains are found in many Laurales, AriBtolochiales and Piperales and some
Nymphaealesin the dicotyledons and in Zosterales and Najadales, many Zingiberales, Smilacaceae, etc., in the monocotyledons.
Grains with 2 t.o many apertures occur sporadically, the most conspicuous group being perhaps AliBmatoles. Greater detail is
found in works of WALKER and collaborators
"d

!
rJ)
(1
o
~
l.>J ~
<
~
~
I'IIAJAOALLI
rJ) • ~ .AI."""" A"U N1U.....IIU,LlI

I~~'::::ll~ ", 8J ~"UI

!
o
i 11Y0000000IUf.L" ::l
II>
tj

i·
:3
~.
(";)

'"d

g.
I
s..
~
.. CD

~
0,9.
~
SUCC_UNe type 0' ~- ......

~ Fig. 3. Approximate distribution of the successive type of microsporogenesis. The diagram must be interpreted with reserva-
~
tion, as the often very few records have been taken as representative of the whole families. In Juncales and Oyperales pollen N>
01
grains are dispersed in tetrads (3 microspores being aborted in Oyperales); their microsporogenesis is very dubiously succes- """
sive, which is indicated by lighter shading. In dicotyledons successive microsporogenesis is known in certain Magnolianae
and Nymphaeanae, superorders which in other characters also show affinities to the monocotyledons. Further, this type is
found in Rafllesiaceae, sometimes considered to be allied to Magnoliales and Aristolochiales, in Podostemaoeae (which might
well be placed with the Alismatanae if it were not for the two cotyledons) and some Apocynaceae and Asclepiadaceae
258 R. DAHLGREN:

1. The distribution of monoaperturate pollen grains or of in-

aperturate or bi- to polyaperturate grains derived presumably from
monoaperturate types is hown in Fig. 2. It can be seen that they are
found in (a) the monocotyledons, (b) the superorder Magnolianae
and (c) in Nymphaeanae (= Nymphaeales). Apertures in these groups
are rarely in the equatorial plane (except in Illiciales and a few Ohloran-
thaceae). For details of Fig. 2, see for example WALKER & DOYLE
(1976) and WALKER (1974, 1976).
2. A connection between monocotyledons and certain Magnolianae
is also shown in the distribution of protein-accumulating (P-type)
sieve-tube plastids (BEHNKE & DAHLGREN 1976). All groups of
monocotyledons have P-type plastids of a rather distinctive type
with cuneate protein bodies matched interestingly enough by those
in Asarum, Aristolochiaceae (BEHNKE 1971, 1975: 93). The sieve-tube
plastids in Oaryophyllanae differ from those in other angiosperms
except some Magnoliales but resemble those of certain Pinatae and
contribute to separate out Oaryophyllanae as a very distinct group.
3. The successive type of microsporogenesis (Fig. 3) indicates
a connection between the majority of the monocotyledons and certain
Magnolianae and Nymphaeanae, for example in members of Myristica-
ceae, Lauraceae, Aristolochiaceae, Oabombaceae and Oeratophyllaceae.
The successive type of microsporogenesis is also found in many Apo-
cynaceae and Asclepiadaceae, where it has doubtless arisen independently,
and in Rattlesiaceae and Podostemaceae, where the character has not
yet been decisively evaluated. The basic data on microsporogenesis is
very limited and the diagram is thus provisional.
4. Apocarpy (Fig. 4) is regarded as a primitive state compared
with syncarpy (incl. paracarpy). It has a wider distribution in di-
cotyledons than the character states previously shown. Thus it is
found not only in M agnolianae but is dominant in the Ranunculanae
and Rosanae. Its distribution in the monocotyledons is relatively
restricted, however, occurring in three or four independent groups,
viz. in a number of palm genera, in most Alismatanae and in the Triuri-
dales and a few Melanthiaceous genera. A "disjunctive" distribution
can be excpeted for certain so-called primitive states. They may have
been conserved in some disparate groups, whereas in most groups
the characters are advanced, the gynoecium in this case being variously
syncarpous. When the carpel is solitary syncarpy is naturally an impos-
sibility, as in Amygdalaceae (in Rosales), Fabales, Proteales and Elae-
agnales. This condition is not to be equated with apocarpy. Nor must
the parietal placentation in monomerous gynoecia be compared to
that of the syncarpous gynoecia. A transition from apocarpous to syn-
carpous gynoecia is found in several orders (Magnoliales, Ranunculales,
o

gI
o
::s
\Il
tj

i'
~.
~

I-d

I
g'
o
....
-" ~
.' 5-
qg ,
o
~ ..,oearpy (Il10<. I""" one capel)
;::::::::::::: one Ir.. clrpel
....
~
1'"
Nl
. Fig. 4. Distribution in angiosp!lrms of apocarpous gynoecia or of gynoecia consisting of (apparently) one carpel only. In the Ql
monocotyledons apocarpy is found in most families of Alismatanae, in Triuridales, in some Melanthiaceae (Liliales) and in co
certain palms. In the dicotyledons apocarpy is concentrated to Magnolianae , Nymphaeanae, Rosanae and Saxilraganae (Saxi·
Iragales). Outside these groups free carpels occur occasionally in Hamamelidanae, Rutanae, Celastranae and Caryophyllanae.
Doubts are often connected with some groups having so·called monomerous gynoecia. They are found in approximately the
same parts of the diagram as the apocarpous groups, culminating in the Fabales, Proteales and Elaeagnales
260 R. DAHLGREN:

Rosales, Rutales, Sapindales, Dilleniales, Saxitragales). A case of pre-

sumably secondary apocarpy is met with in Resedaceae (e.g. Caylusea).
5. Endosperm formation (DAHLGREN 1975b: 189-193) is cellular
in at least some family in each of all the orders of Magnolianae and
Nymphaeanae. Cellular endosperm is also found in members of Ne-
lumbonaceae, Lardizabalaceae and Circaeasteraceae in Ranunculanae
and in Mitrastemon (Rafflesiaceae) in Rafflesianae. This type also
occurs in TrochoderUlraceae and Cercidiphyllaceae here placed in Ha-
mamelidanae. All these groups are usually classified as primitive. In
addition, the ab initio cellular endosperm is found in many Saxitra-
ganae and Gentiananae (though Gentianales generally has the nuclear
type), in most Cornanae, in Lamianae, Loasanae, Solananae and Cam-
panulanae and in many Asteranae and Celastranae. This feature is
thus common to some of the most "primitive" and some of the most
"advanced" groups. That the cellular type is the original state was
probably first stated by WUNDERLICH (1959: 274). In the "advanced"
order~, according to her, it can have been retained as a consequence
of the restricted nucellussize. However, the endosperm is nuclear
ab initio in for example the small nucelli of Theales.
6. The distribution of benzylisoquinoline alkaloids (Fig. 5)
covers some of the often apocarpous groups (see also KUBITZKI 1969).
They are found in many taxa of M agnolianae, though so far there is
only a couple of records in Piperales (see, e.g., HANSEL et al. 1975)
and none as yet in Illiciales. Further, the benzylisoquinoline alkaloids
are known in many Ranunculanae, reaching a maximum of complexity
and variation in Papaverales. Complex benzylisoquinoline alkaloids
are also found in certain genera of Rutaceae (FISH & WATERMAN 1973)
and a new find was recently reported in Heracleum (Apiaceae) in
Aralianae by GUPTA et al. (1976). The presence of benzylisoquinoline
alkaloids in some genera of Rhamnaceae (Celastranae) indicates that
this family may have some affinity to Ranunculanae and Magno-
lianae, although this is dubious. Benzylisoquinoline alkaloids are also
found in Croton (Euphorbiaceae).
7. The distribution of sesquiterpene lactones and of polyacety-
lenes derived from fatty acids (Fig. 6) supports a connection between
the A steranae , Campanulanae, A ralianae , Celastranae (Santalales),
Rutanae (Rutales) and possibly also Ranunculanae and Magnolianae.
Thus polyacetylenes are chiefly found in Asteranae and Campanulanae but
also in Apiaceae, Araliaceae and Pittosporaceae (Aralianae) , in Ola-
caceae, Opiliaceae, Santalaceae, V iscaceae and Loranthaceae (Santalales,
Celastranae), in Simarubaceae (Rutanae) and in Lauraceae and An-
nonaceae (Magnolianae). Further, it is also known in isolated members
of Sterculiaceae (Dillenianae) and Fabaceae (Rosanae). Reports of their
§
I
.tJ
i·
3
~.
o

i
~.o
::s
_u o
...
~
OfIl,lbullon ot
>
::s
os.
~-. bOonl,Uooqu lnoUne .lkalO'" ~
1,-. ~
Fig. 5. Distribution of benzylisoquinoline alkaloids and of iridoids (hatching and dots respectively). The former group of
compounds has its main distribution in Magnolianae (except Illiciale8 and most Piperale8) and Ranunculanae, but is absent t-:>
0>
in Nymphaeanae . Occurrences in some Rutaceae (Rutales), and Heracleum (Araliale8) are of great interest. The presence of -
these alkaloids in Rhamnaceae may be of importance in splitting the present Rhamnales. Iridoids occur in the OOf"nanae,
Gentiananae, Loasanae and Lamianae. Secoiridoids occur chiefly in Gentiananae but also in some Oornales and Loasale8. In
groups with bitegmic ovules, iridoids are recorded in Liquidambar and DaphniphyUum (here in Hamamelidales) and in Fou-
quieriaceae, which is probably better placed. near Ericales than in Saxifragale8 (indicated by an arrow)
I>:>
0>
I>:>

;0
t1
:..
=:
t"'

~
Z

_us

Dlatrlbutlon Of poly.e.ty~n ••

Fig. 6. Distribution of polyacetylenes derived from fatty acids. The greatest concentration is found in Asteranae, Aralianae
(incl. Pittosporales), Oampanulanae and Santalales (Oelastranae). Scattered occurrences are reported in Simaroubaceae (Rutales),
Lauraceae (Laurales), Annonaceae (Magnoliales), Fabaceae (Fabales) and Sterculiaceae (Malvales). The records in E'uphor.
biaceae (Euphorbiales) and Valerianaceae (Dipsacales) are very dubious, and a find in Scaevola in Goodeniaceae deserves parti-
",1110 .... r n o .... -t~ ........ no ~ .. ~a Co rlo"' .. ~n .. ; .......... +...."' .......... \,." .... "' .. 4- ........." ...,
Cl
o
3
3
§
§
o
::l
II'

;.
1
'"d

i
~
~
o·
::::
o
..,
fOA\..U
~
>
::l
<l9.
o
CenuUug.llucc ....lon In multilt.mln.te .ndro.c:l.
. wllh ,tlmen •• "~ ... Ind/OI' It.men trunk

::::::::::::: wll",,"1 any 01 I ..... .

1t.:)
Fig. 7. Distribution of centrifugal succession in multistaminate androecia. Hatching represents groups with stamen fascicles a>
~
and/or stamen trunk bundels, the dots groups with none of these structures; sometimes as in Geraniaceae, with centrifugal
obdiplostemony. Most data from TUCKER (1972)
264 R. DAHLGREN:

presence in Ricinus and Valeriana are dubious. One case, Scaevola

(Goodeniaceae, Gentiananae) , an iridoid-containing genus, does not
conform to the expected pattern.
8. Androecial characters, particularly centrifugal v. centri-
petal succession in multistaminate androecia (Fig. 7), have
been dealt with by MERXMULLER (1972), KUBITZKI (1973), TUCKER
(1972) and LEINS (1972, 1975) and are of importance in distinguishing
the subclasses Rosidae and Dilleniidae in TAKHTAJAN'S and CRON-
QUIST'S systems. Taxa with multistaminate androecia displaying
centrifugal succession and usually fascicular stamens occur in a number
of groups, most of which are not closely related:

Certain Caryophyllanae (in Aizoaceae and Cactaceae).

A great many Dillenianae (Dilleniales, Cistales and Malvales).
Certain Violanae (Violales, Capparales), some of which may be closely
related to certain Dillenianae.
Many Theanae: mainly Theales (incl. .Lecythidaceae).
Certain Ericales (Cornanae), e.g. Actinidiaceae.
Genera in the apparently isolated Loasaceae, here treated as a separate
superorder, Loasanae.
Some Butomales and Hydrocharitales, Alismatanae.
Some genera of palms (Arecanae).

The obdiplostemonous taxa have not been included here.

Facts seem to indicate that in many (or all?) cases the multista-
minate-fascicular androecia with centrifugal succession are derived
from androecia with one or two whorls of stamens. The character is
of taxonomic value but should not be over-estimated as it has arisen
within several lines of evolution. There are also obvious inconsistencies
as in Begoniaceae (Violales), where the stamens develop centripetally
in contrast to other multistaminate V iolales; and as in Punicaceae
(Myrtales) which has centrifugal succession in contrast to most other
Myrtales. The consistently spiral organization of the stamens in Magno-
liales does not hold either: stamen trace bundles are found for example
in Degeneria (STEBBINS 1974: 221).
9. The presence of glucosinolates (ETTLINGER, unpubl. ms.) or
similar isothiocyanate precursors characterizes Tovariaceae, Reseda-
ceae, Capparaceae and Brassicaceae, which make up to the nucleus of
the order Capparales, but also a number of small families: Tropae-
oiaceae, LimnantJt.aceae, Bretschneideraceae, M oringaceae, Pentadiplan-
draceae, Caricaceae, Salvadoraceae, Gyrostemonaceae. and Bataceae,
most of which are of uncertain taxonomic position. In addition, glu-
cosinolates are found in Drypetes (incL Putranjiva) , usually placed
in Euphorbiaceae, but possibly meriting distinction as a separate family,
Putranjivaceae. The question of whether all or most of the glucosinolate
Commentary on a Diagrammatic Presentation of the Angiosperms 265

fa.milies are at all closely related or whether they form several inde-
pendent aggregates has not yet been settled. Research is being done
on several of the families. Typical so-called "myrosin cells" are found
in most but not all of the glucosinolate families. Glucosinolates and
cyanohydrins are formed from amino acids by the same initial reaction.
Several of the glucosinolate families also seem to be allied to Flacour-
tiaceae and Passijloraceae, which contain cyanogenic compounds (ETT-
LINGER, op. cit.).
10. Ellagic acid and ellagitannins are of taxonomic importance.
Much has been written on this by BATE-SMITH (e.g. 1973), who has
kindly supplied me with the data presented in Fig. 8. The absence
of ellagitannins in minor groups is of restricted taxonomic value. For
example they tend to be lacking in herbaceous plants, whereas they
may be present in related woody plants. There is a noticeable absence
of ellagitannins in all monocotyledons and in the sequence of groups
Magnolianae-Ranunculanae-Rutales (of Rutanae)-Aralianae-Aster-
anae-Campanulanae-Solananae, many orders of which contain either
benzylisoquinoline alkaloids or polyacetylenes or both (see above).
Ellagitannins are also (generally) lacking in another sequence of super-
orders, Cornanae-Gentiananae-Lamianae-Loasanae, i.e. those in which
iridoids are produced, although there are some exceptions where both kinds
of compounds occur together, viz. in some Ericales, some Cornales and
the monogeneric Fouquieriaceae. Other superorders where ellagitan-
nins are of very restricted occurrence are Violanae (except Tamari-
cales), Dillenianae (except Dilleniales and certain Malvales-Euphor-
biales) , Thymelaeanae and Rutanae (except some Geraniales and Ju-
glandales). Individual orders where they are usually or consistently
lacking are Fabales and Polygonales, a fact that should be taken into
consideration in the future.
The presence of ellagic acid in some members of Nymphaeanae,
viz. in Nymphaeaceae and Cabombaceae, is unique in taxa with mono-
aperturate pollen grains. Nelumbo is lacking in ellagic acid and con-
tains benzylisoquinoline alkaloids. This together with its tricolpate
pollen grains suggests a position near Ranunculales and at some distance
from N ymphaeales.
11. Epigynous and perigynous flowers (Fig. 9) characterize
taxa at different levels. Epigyny can with certainty be classified as a
derived condition not likely to revert to hypogyny. Where epigyny
has appeared early it is likely to characterize larger categories: Ara-
lianae; Asteranae; most M yrtanae; Gentiananae: Rubiaceae; Orchidales;
and some other Lilianae. Concrescence between a well-developed
floral axis or perianth and the carpel walls is likely to arise within
different groups and thus to be scattered in the system. The distribution
~
0>
0>

~
t1
>-
Il:
S
~
z

POAI,.U

Ollt,lbullon 01 ollagle Kid

Fig. 8. Preliminary survey of the distribution of ellagic acid and/or ellagitannins in the angiosperms. Families with records
of ellagic acid are shaded; partially shade only in certain larger families where their presence is known to be sporadic (as in
Oaesalpiniaceae, Fabales). Ellagitannins are concentrated mainly to the Hamamelidanae-Rosanae (Rosales)-Theanae-Dil-
lenianae (in part)-Myrtanae- Saxifraganae-Oornanae (Ericales)-Primulanae (in part) complexes, being conspicuously
absent or rare in most other parts of the system. Striking occurrences outside the complexes mentioned are the orders Nym-
phaeales, Tamaricales, Santalales and Ju.glandales and, e.g., the families Geraniaceae, Aceraceae, Simarubaceae and Limnan-
thaceae. The absence of ellagitannins in all monocotyledons in the Magnolianae-Ranunculanae-Rutanae- Violanae-Ara-
lianae-Asteranae-Oampanulanae-Solananae complexes and in Gentiananae-Loasanae-Lamianae is conspicuous (see
also BATE-SMITH 1973). The data, kindly supplied by E. C. BATE-SMITH, are preliminary
Cl
o
3
3
§
g
§
II>
t::1

i'
3
~
o·
I-d

i
::l
~

~
0'
::l
o
...,
~
::r
(l)
.'
~
::l
()q
::: : ::: per igynous 11__..

~ Ol>igynou, l
Fig. 9. Distribution of epigyny and perigyny in angiosperms. Epigyny has developed independently in many groups. The ~
largest of these are the Lilianae ~especially Orchidales) , the Myrtales, the Araliales-Asterales(-Campanulales) complexes and
the Rubiaceae in Gentianales. In some of these it has doubtless appeared in more than one evolutionary branch. Epigyny N>
0>
is less frequellt in most of the so·called "primitive" groups such as Magnolianae, Nymphaeanae and Ranunculanae, and is ...
likewise rather uncommon in Rutanae, Dillenianae, Lamianae, Primulanae, Solananae, etc. In monocotyledons epigyny is
absent in Commelinanae, Arecanae, Aranae and Typhanae, all groups where the tepals are usually reduced or glume.like.
Perigyny is a somewhat ill·defined concept, and in a slightly broader sense has a more extensive distribution than is shown
here
268 R. DAHLGREN:

of epigyny, perigyny and hypogyny should be considered in relation to

position of nectaries, function of fruit dehiscence and other biological
phenomena of importance. In monocotyledons the total absence of
epigyny in Commelinanae, Aranae and Arecanae can be explained by
the early reduction of the perianth to hyaline scales, bristles or hairs
(or lack of perianth) in most of the families.
12. Sympetaly (Fig. 10) may well be compared with epigyny
and can be expected to appear in widely different groups as a con-
sequence of intercalary concrescence (STEBBINS 1974: 111). A distinc-
tion can be made between the "main sympetalous groups" or the
Sympetalae, (most of) which are often considered to be closely related,
and the other sympetalous groups. The "Sympetalae" include Asteranae,
Campanulanae, Solananae, Lamianae, Primulanae, Gentiananae, many
Cornanae, and Plumbaginales (Plumbaginanae) in the present system.
It is generally agreed that sympetaly in other (minor) groups has evolved
along a number of separate lines and in many cases perhaps at a late
stage of evolution. The problem as to whether the "main sympetalous
groups" form one, two or several natural complexes is often discussed.
As indicated above (under polyacetylenes, sesquiterpene lactones) as
well as below (under iridoids) it seems that the Sympetalae (excl. Pri-
mulales and Plumbaginales) form at least two main branches: the iridoid-
bearing branch (lacking polyacetylenes, sesquiterpene lactones and
usually inulin) and the iridoid-Iacking sympetalous branch. The for-
mer, comprising Cornanae (often choripetalous; including Ericales)-
Gentiananae-Lamianae seem8 to approach The,anae and/or Saxi-
fraganae-Hamamelidanae, while the latter, comprising Asteranae-Camp-
panulanae-Solananae, seems to approach Aralianae(-Rutanae) and
Santalales of the Celastranae. Primulanae and Plumbaginales (Plumbagin-
anae) probably fall outside both ofthese groups and are perhaps best placed
near Theanae. Moreover the ovules of Primulanae and Plumbaginanae
are usually bitegmic, whereas in the other large sympetalous groups
they are unitegmic.
13. Iridoids (Fig. 5) are mostly confined to sympetalous plants
(JENSEN et al. 1975), but only in the superorders Cornanae, Gentiananae
and Lamianae. As far as is known they are absent in Solananae, Cam-
panulanae and Asteranae. With the exception of a few genera, Daphni-
phyUum, Fouquieria (DAHLGREN et al. 1976b) and Liquid,ambar all
the iridoid-bearing taxa have unitegmic, generally tenuinucellate
ovules. With the chief exception of most Gentianales their endosperm
is usually cellular ab initio. Terminal endosperm haustoria are common
in the iridoid groups. An unexpected occurrence of this combination
of features is found in Loasaceae which is possibly best placed near
Cornales or the orders of Gentiananae.
(')
o

j
g
tj
'"
!iI'
~
~
8
~.

[
~.
o
::l
~u o....

~
~
1l,9.
DltllitJu\1oft 01 .ympol.,y (dicotyledOnt only)

Fig. 10. Distribution of sympetalous corollas in dicotyledons (syntepaly in monocotyledons is not included). Where fusion of
III
pairs or of three of the petals has occurred (such as in Balsaminaceae or Polygalaceae) or where two petals of five I,tre united
(such as in Fabaceae) these are not included but are indicated by asterisks. Some problems arise where there is doubt as to
i~
Q>
whether calyx or corolla is represented in monochlamydeous flowers: in Santalales Loranthaceae is shaded but not Viscaceae. ~

The perianth in Thymelaeaceae and Proteaceae is interpreted as a calyx. Sometimes the petals are connate apically to form a
calyp;ra (Marcgraviaceae, some Vitaceae). The polyphyletic origin of sympetaly is obvious. However, the assemblage of super-
orders in the right-hand part of the diagram (Primulanae-Plumbaginanae: Plumbaginales-Cornanae-Gentiananae-La-
mianae-Solananae-Campanulanae-Asteranae) comprising the "Sympetalae" in the broad sense is worthy of note although
it probably does not represent a natural group
270 R. DAHLGREN:

The importance of iridoids as an indication of relationships was

not accepted by STEBBINS (1974: 135). Nor is it acknowledged by
THORNE in his system of angiosperms (1968; pers. comm.), where
iridoid-containing families are widely scattered. The concentration
of seco-iridoids and indole alkaloids chiefly to the orders of Gentiananae
as well as the high correlation with embryological characters support
their taxonomic value, however.
14-15. The distribution of unitegmic and of tenuinucellate
ovules and its implications have been stressed in particular by PHILIP-
SON (1974), and have also been discussed by DAHLGREN (1975b) largely
on the basis of an extensive study by WUNDERLICH (1959). Unitegmic
ovules characterize Cornanae, Gentiananae, Lamianae, Loasanae,
Solananae, Campanulanae, Asteranae and Aralianae, but also occur in a
considerable number of other groups. The presence of both bitegmic
and unitegmic ovules in, for example, Ranunculaceae, Rosaceae, Pi-
peraceae . and Salicaceae prove that the unitegmic condition is highly
polyphyletic. The tenuinucellate and unitegmic conditions usually
occur together (especially in sympetalous groups), but there are notable
exceptions. In a number of usually restricted, isolated groups with
unitegm~c ovules (such as in the families just mentioned) these are
crassinucellate. In most Theales and Primulales the ovules are bitegmic
and tenuinucellate.
16. The distribution of 2- and 3-nucleate pollen grains has
been presented by BREWBAKER (1967) and outlined in the present
system by DAHLGREN (1975b). Certain major groups of dicotyledons
are consistently 3-nucleate viz. Caryophyllanae, A ralianae , Asteranae.
Plumbaginanae, Thymelaeanae (Dichapetalaceae excluded) and Bras-
sicaceae (Violanae). BREWBAKER pointed out the coincidence in di-
cotyledons between 3-nucleate pollen grains and sporophytic self-
incompatibility and between 2-nucleate pollen grains and gameto-
phytic self-incompatibility. Recent studies show further correlation.
Thus almost all groups with sporophytic self-incompatibility (HESLOP-
HARRISON 1975: 413) and thus also 3-nucleate pollen grains (but also
other groups) have the "dry" type of papillate stigma. The fact that
3-nucleate pollen grains and "dry" (smooth in SEM micrographs)
stigmas and 2-nucleate and "wet" (secretory) stigmas occur together
even down to specific level is shown by HANSEN (1976: 343) for
members of Balanophoraceae. The basic studies of dry/wet stigma
distribution by K. R. SHIVANNA at Kew have not yet been published
(HESLOP-HARRISON 1976: 143). In the case of the 3-nucleate pollen
grains it seems that the active materials for pollen tube growth are
conveyed in the exine of the pollen grain wall and are originally derived
from the tapetum (thus being of sporophytic origin); further. there
Commentary on a Diagrammatic Presentation of the Angiosperms 271

appears to be no synthesis in them at all (HESLOP-HA~RISON 1968: 90).

Thus we have here a syndrome of interdependent characters of pro-
found biological importance. The taxa photosynthesizing over the C4
pathway are also mainly restricted to groups with 3-nucleate pollen
grains.
Some Distributional Patterns in Monocotyledons
The taxonomy of the monocotyledons does not present problems
of the same magnitude as in the dicotyledons. The superorders Alis-
tanae, Lilianae, Commelinanae, Zingiberanae, Typhanae, Arecanae and
Aranae are those which most taxonomists tend to agree upon, Aranae
and Zingiberanae perhaps being placed with Arecanae and Lilianae
respectively. Whether Bromeliales and Haemodorales are best placed
with Lilianae or Commelinanae, and whether Triuridales is best placed
with Lilianae or Alismatanae depends on the evaluation of characters.
Typhales (~ Typhanae) may be closest to Juncales. The fact that they
can be attacked by the same species of Uromyces as Acorus (Araceae)
(NANNFELDT 1968: 9'4) supports another affinity, but Typhales differs
from Arales in having helobial endosperm formation. It may on the
other hand be closest to families in Lilianae, as supported by the
serological investigations of LEE and ~'AIRBROTHERS (1972).
Some criteria that contribute to distinguish between Lilianae and
Commelinanae are shown here. They have been presented with com-
ments in DAHLGREN et al. (1976a).
The presence or absence of vessels (Fig. 11) in different
parts of the plant presents an interesting pattern of variation (WAGNER,
in· prep.). Commelinanae, Arecanae and Typhanae consistently have
vessels in the stem, Aranae and Alismatanae lacking them in the stems
(Alismatanae usually also in the roots). Zingiberanae and Lilianae
are variable in this respect. Lilianae generally lacks vessels in the stems,
exceptions being, for example, Dioscoreales, many Bromeliales and
certain Asparagales, particularly some of those that are bacciferous
and/or have secondary thickening growth.
Epigyny is common in Lilianae and particularly so in Zingiberanae.
It is also found in Hydrocharitaceae (Alismatanae). All these groups
have a well-developed perianth. Reduction of the perianth is extensive
in Commelinanae, which has consistently hypogynous flowers. Like-
wise, zygomorphic and asymmetric flowers are concentrated to Lilianae
and Zingiberanae. In the monocotyledons tri-nucleate pollen grains
are concentrated to Commelinanae (except most Commelinales, Erio-
caulales and some Cyperales) , Alismatanae, some saprophytic Lilianae
and certain Aranae (BREWBAKER 1967).
The endosperm where present (it is absent in the seeds of Or-
272 R. DAHLGREN:

Fig. 11. Top: Approximate distribution of vessels with scalariform perfora.

tion or with entire perforation (sometimes also with scalariform perforation)
in monocotyledons. Bottom: Approximate distributioncrif seeds with
starchy endosperm or starchy perisperm.-·--Both characters serve to some
extent. t.o illustrate the distinct.ness of Commelinanae and Lilianae. The
diagram corresponds to the monocotyledonous sector of the main
diagram although the orders have been separated from one another and
thei,· positions slight.ly alt.erered (moreover Pontederiaceae is treated as a
>lepaJ"ate order)
Commentary on a Diagrammatic Presentation of the Angiosperms 273

Onl.l. r.phldes

a ~

'1il1OCAUU.lU
"""'''''''''u

SlUe. bodle.

Fig. 12. Top: Approximate distribution of oxalate raphides in monoco-

tyledons. Alismatanae and Commelinanae except most Commelinaceae are
characterized by the absence of oxalate raphides. In Lilianae they are
much commoner in Asparagales than in Liliales, where they are scattered
in Alstroemeriaceae and Melanthiaceae. Bottom: Approximate distribu-
tion of silica bodies in monocotyledons. The shape of the silica bodies
characterizes families or minor taxa. They are typical of most Commelinanae
and Arecanae but are lacking in most Lilianae

Plant Syst. Evol.. Suppl. 1 18

274 R. DAHLGREN:

Table 1. Classification of the families of angiosperms according to DAHL-

GREN (1975a), where further information is obtainable. An asterisk (*)
marks the families with weakly based or dubious position. Alternative
positions are given for many of these in the publication mentioned

Magnolianae
M agnoliales: Winteraceae, Degeneriaceae, Himantandraceae, Magno-
liaceae, Annonaceae, Oanellaceae, Myristicaceae, Eupomatiaceae
Laurales: Monimiaceae, Trimeniaceae, Lauraceae, Idiospermaceae,
Austrobaileyaceae, Gomortegaceae, Amborellaceae, Oalycantha.ceae, Hernan-
diaceae, Lactoridaceae *, Ohloranthaceae *
A ristolochiales: Aristolochiaceae
Piperales: Saururaceae, Piperaceae
Illiciales: Illiciaceae, Schisandraceae

RaJJlesianae
Ralflesiales: Rajjlesiaceae, Hydnoraceae

Ranunculanae
N elumbonales: Nelumbonaceae
Ranunculales: Lardizabalaceae, Menispermaceae, Sargentodoxaceae,
Kingdoniaceae, Ranunculaceae, Oircaeasteraceae, Hydrastidaceae, Glauci-
diaceae, Podophyllaceae, Nandinaceae, Berberidaceae
Papaverales: Papaveraceae, Hypecoaceae, F-umariaceae

Nymphaeanae
N ymphaeales: Oabombaceae, Nymphaeaceae, Barclayaceae, Cerato-
. phyllaceae

Rutanae
Rutales: Rutaceae, Oneoraceae, Surianaceae, Simaroubaceae, Kirkiaceae,
Burseraceae, Meliaceae
Polygalales: Malpighiaceae, Trigoniaceae, Vochysiaceae, Xanthophyl-
laceae, Polygalaceae, Krameriaceae, Emblingiaceae *
Sapindales: Ooriariaceae, Anacardiaceae, Podoaceae, Julianaceae,
Akaniaceae, Uapacaceae*, Sapindaceae, Aitoniaceae, Aceraceae, Hippo-
castanaceae, Sabiaceae*, Meliosmaceae, Koeberliniaceae
J uglandales: Rhoipteleaceae, Jugla,ndaceae
M yricales: Myricaceae
Leitneriales: Leitneriaceae
Geraniales: Zygophyllaceae, Nitrariaceae, Peganaceae, Balanitaceae,
Erythroxylaceae, Dirachmaceae*, Geranialfeae, Ledocarpaceae, Vivianiaceae,
Biebersteiniaceae, Ixonanthaceae, Humiriaceae, Httgoniaceae, Linaceae, Le-
pidobotryaceae, Averrhoaceae, Oxalidaceae, Hypseocharitaceae
Basaminales: Balsaminaceae

Aralianae
A ral-iales: Araliaceae, Torricelliaceae, Apiaceae
l'ittosporal es: l'ittosporaceae
Commentary on a Diagrammatic Presentation of the Angiosperms 275

Table 1 (continued)

Asteranae
A sterales: Asteraceae

Dillenianae
Dilleniales: Paeoniaceae, Dilleniaceae
Oistales: Oistaceae, Bixaceae
Malvales: Spaerosepalaceae, Oochlospermaceae*, Elaeocarpaceae*, Ster-
culiaceae, Huaceae*, Tiliaceae, Dipterocarpaceae, Bombacaceae, Malvaceae,
N euradaceae *
U rticale8: Ulmaceae, Hymenocardiaceae, Moraceae, Oannabaceae, Ur-
ticaceae
Euphorbiales: Euphorbiaceae, Pandaceae*, Aextoxicaceae*, Picro-
dendraceae

Thymelaeanae
Thymelaeales: Dichapetalaceae*, Thymelaeaceae

Violanae
V iolales: Flacourtiaceae, Passifloraceae, Dipentodontaceae, Scyphoste-
giaceae, Violaceae, Turneraceae, Malesherbiaceae, Achariaceae, Oucurbi-
taceae, Begoniaceae, Datiscaceae*, Oaricaceae
Tamaricales: Tamaricaceae, Frankeniaceae
Salicales: Salicaceae
Oapparales: Limnanthaceae, Tropaeolaceae, Bretschneideraceae*, Sal-
t.adoraceae*, Moringaceae, Resedaceae, Tovariaceae, Oapparaceae, Penta-
diplandraceae, Brassicaceae, Gyrostemonaceae *, Bataceae *

Celastranae
Oelastrales: Buxaceae*, Simmondsiaceae*, Stylocerataceae*, Didy-
melaceae *, Barbeyaceae *, Geissolomataceae *, Avicenniaceae *, Staphyleaceae,
Sphenostemonaceae, Aquifoliaceae, Oelastraceae, Stackhousiaceae, Siphono-
dontaceae *, Goupiaceae *, Lophopyxidaceae *, M ontiniaceae *
San tales: Olacaceae, Opiliaceae, Loranthaceae, Misodendraceae, San-
lalaceae, Eremolepidaceae, Viscaceae
Rhamnales: Rhamnaceae, Vitaceae, Leeaceae

Solananae
Solanales: Solanaceae, Goetzeaceae, Nolanaceae, Oonvolvulaceae, Ouscu-
taceae, Oardiopterygiaceae, Oobaeaceae, Polemoniaceae, Hydrophyllaceae,
Ehretiaceae, Boraginaceae, Wellstediaceae, Lennoaceae*, Hoplestigmataceae*

Campanulanae
Oampanulales: Oampanulaceae, Pentaphragmataceae, Lobeliaceae, Sphe-
nocleaceae

Hamamelidanae
Trochodendrales *: Trochodendraceae, Tetracentraceae, Eupteleaceae,
OercidiphyUaceae
IS·
276 R. DAHLGREN:

Table 1 (rontinufd)

H amamelidales: Myrothamnaceae, Hamamelidaceae, Platanaceae, AI-

tingiaceae, Daphn'iphyllaceae, Rhodoleiaceae *
Oasuarinales: Oasuarinaceae
Fagales: Fagaceae, Oorylaceae, Betulaceae
Balanopales *: Balanopaceae
Ounoniales: Ounoniaceae, lteaceae, Brunelliaceae, Eucryphiaceae,
Baueraceae, BruniacefJe

Rosanae
Rosales: Orossosomataceae*, Rosaceae, Malaceae, Amygdalaceae, Oon-
naraceae *, M elianthaceae *, Ohrysobalanaceae *
Fabales: Mimosaceae, Oaesalpiniaceae, Fabaceae

Proteanae
Proteales: Proteaceae

Myrtanae
M yrtales: Lythraceae, Punicaceae, Rhizophoraceae, Dialypetalantha-
ceae*, Orypteroniaceae, Oombretaceae, Oliniaceae, Melastomataceae, Penae-
aceae, Myrtaceae, Onagraceae
Elaeagnales *: Elaeagnaceae
Trap'ales: Trapaceae
H aloragales: Haloragaceae

Saxifraganae
Saxifragales: Orassulaceae, Penthoraceae, Saxifragaceae, Fouquieri-
aceae*, Francoaceae*, Brexiaceae*, Oephalotaceae, Tremandraceae, Vahli-
aceae, Ribesiaceae *, Greyiaceae *
Podostemales*: Tristichaceae, Podostemaceae
Gunnerales: Gunneraceae

Balanophoranae
Balanophorales: Balanophoraceae, Oynomoriaceae

Plumbaginanae
Plumbaginales: Plumbaginaceae, Limoniaceae
Polygonales: Polygonaceae

Primulanae
Primulales: Myrsinaceae, Aegicerataceae, Theophrastaceae, Primula-
ceae, Oordidaceae
Ebenales: Ebenaceae, Sapotaceae, Lissocarpaceae, Styracaceae

Theanae
T heales: Stachyuraceae, Ochnaceae, Quiinaceae, Medusagynaceae, Scy-
topetalaceae *, Sarcolaenaceae, Strasburgeriaceae, Oncothecaceae, Theaceae,
Pentaphylacaceae, Marcgraviaceae, Oaryocaraceae, Pelliceriaceae, Napo-
Commentary on a Diagrammatic Presentation of the Angiosperms 277

Tabl. 1 (continued)

leonaceae, Bonnetiaceae, Foetidiaceae, Lecythidaceae, Symplocaceae, Olu-

siaceae, Ancistrocladaceae *, Elatinaceae *
N epenthales: Nepenthaceae, Dioncophyllaceae *
Dr 0 s era 1e s: Droseraceae, Lepuropetalaceae, Parnassiaceae

Cornanae
Ericales: Actinidiaceae, Olethraceae, Oyrillaceae, Roridulaceae, Erica-
ceae, Monotropaceae, Pyrolaceae, Epacridaceae, Diapensiaceae, Byblidaceae*,
Empetraceae, Grubbiaceae *
Sarraceniales: Sarraceniaceae
Eucommiales: Eucommiaceae
Oornales: Garryaceae, Alangiaceae, Oornaceae, Davidiaceae, Nyssaceae,
lcacinaceae, Escalloniaceae, Oolumelliaceae, Stylidiaceae, Hydrangeaceae,
Alseuosmiaceae, Sambucaceae, Adoxaceae

Gentiananae
Dipsacales: Oaprifoliaceae, Valerianaceae, Triplostegiaceae, Dipsaca-
ceae, Morinaceae*, Oalyceraceae*
Oleales: Oleaceae
Goodeniales: Goodeniaceae.
Gentianales: Loganiaceae, Buddlejaceae, Retziaceae, Rubiaceae, Menyan-
thaceae, Gentianaceae, Apocynaceae, Asclepiadaceae

Loasanae
Loasales: TJOasaceae

Lamianae
Scrophulariales: Scrophulariaceae, Selaginaceae, Globulariaceae, Len-
tibulariaceae, Plantaginaceae, Pedaliaceae, Trapellaceae, Martyniaceae, Oro-
banchaceae, Gesneriaceae, Bignoniaceae, Henriqueziaceae, Myoporaceae,
Acanthaceae
H ippuridales: Hippuridaceae
H ydrostachyales *: Hydrostachyaceae
Lam i ale s: V erbenaceae, Oallitrichaceae,. Lamiaceae.

Caryophyllanae
Oaryophyllales: Phytolaccaceae, Agdestidaceae, Stegnospermataceae,
Achatocarpaceae, Nyctaginaceae, Aizoaceae, Molluginaceae, Didiereaceae,
Oactaceae, Portulacaceae, Hectorellaceae, Basellaceae, Ohenopodiaceae, Dys-
phaniaceae, Halophytaceae, Amaranthaceae, Oaryophyllaceae

Alismatanae
A lismatales: Alismataceae, Limnocharitaceae
H ydrocharitales: Butomaceae, Hydrocharitaceae, Aponogetonaceae
Zosterales: Scheuchzeriaceae, Juncaginaceae, Potamogetonaceae, Zoste-
raceae, Posidoniaceae, Zannichelliaceae, Oymodoceaceae
Najadales: Najadaceae
278 R. DAHLGREN:

Table 1 (continued)

Lilianae
Dioscoreales: Dioscoreaceae
Stemonales:.Stemonaceae, Trilliaceae
Asparagales: Smilacaceae, Philesiaceae, Ruscaceae, Convallariaceae,
Asparagaceae, Dracaenaceae, Hypoxidaceae, Tecophi.teaceae, Phormiaceae,
Xanthorrhoeaceae, Aphyllanthaceae, Asphodelaceae, Anthericaceae, Ixoli-
riaceae, Agavaceae, Hemerocallidaceae, Hyacinthaceae, Alliaceae, Amaryl-
lidaceae.
Taccales: Taccaceae
H aemodorales: Haemodoraceae, Pontederiaceae, Philydraceae
Liliales: Colchicaceae, Iridaceae, Alstroemeriaceae, Liliaceae, Melan-
thiaceae
Triuridales: Triuridaceae
Burmanniales: Burmanniaceae, Corsiaceae, Thismiaceae
Orchidales: Apostasiaceae, Cypripediaceae, Orchidaceae
Bromeliales: Bromeliaceae, Velloziaceae*

Typhanae
Typhales: Sparganiaceae, Typhaceae

Zingiberanae
Zingiberales: Lowiaceae, Heliconiaceae, Musaceae, Strelitziaceae, Zingi-
beraceae, Costaceae, Cannaceae, Marantaceae

Commelinanae
Commelinales: Commelinaceae, Cartonemataceae, Mayacaceae, Xyri-
daceae, Abolbodaceae, Rapateaceae
E riocau lales: Eriocaulaceae
J uncales: Juncaceae, Thurniaceae
Cyperales: Cyperaceae
Centrolepidales: Centrolepidaceae
Poales: Restionaceae, Ecdeiocoleaceae, Flagellariaceae, Joinvilleaceae,
Poaceae

Arecanae
A recales: Arecaceae
Pandanales: Pandanaceae
Cyclanthales: Cyclanthaceae

Aranae
Arales: Araceae, Lemnaceae

ckidales, Alismatanae and many Aranae) contains starch (Fig. 11)

in all Commelinanae but also in certain Lilianae such as Bromeliales,
Haemodorales and some Asparagales and Liliales.
Oxalate raphides (Fig. 12) are common in monocotyledons
except in Commelinanae where they seem to be confined to the (ento-
Commentary on a Diagrammatic Presentation of the Angiosperms 279

lI!-0philous) Commelinaceae; they are also absent in the Alismatanae and

rare in Zingiberanae. Within Lilianae they are commoner in Aspara-
gales than in Liliales.
Silica bodies (Fig. 12) in cells of the vegetative parts show an
occurrence that is partly "vicarious" in relation to that of the oxalate
raphides. They are present in most Commelinanae but also in Arecanae,
while they are absent in Alismatanae and most Lilianae (except Bro-
meliaceae). Zingiberanae is variable in this respect. Finally, steroidal
saponins, which are commoner in monocotyledons than in dicotyledons,
are concentrated chiefly to Lilianae (incl. Bromeliales) and outside
this superorder are of restricted occurrence.

Important Major F~atures Within This Angiosperm System in Relation

to the Subelasses in TAKHTAJAN'S and CRONQUIST'S Systems

The main features of the angiosperm system as outlined by TAKHTA-

JAN (1969) and CRONQUIST (1968) have received wide support. Like
THORNE (1968) I have found it necessary, however, to refrain from
using their subclasses. My reasons are as follows:
1. In the dicotyledons in particular certain orders placed in different
subclasses display obvious similarities indicating close affinities.-
Example: The families of Ericales (placed in Dilleniidae) display obvious
similarities to those of Cornales (placed in Rosidae). These in their
t.urn resemble families in Dipsacales (placed in Asteridae).
2. Relevant data justify excluding major groups from certain sub-
classes thus restricting their circumscription markedly.-Example:
Caryophyllales is clearly distinct from Polygonales and Plumbaginales,
which should both be placed in another part of the system. Further
t.axa must also be removed from the vicinity of Caryopkyllales, for
example the monogeneric Theligonales which is closely connected with
Rubiaceae, and Bataceae and Gyrostemonaceae with other affinities.
All these groups comprise the subclass CaryophyUidae which must
thus be changed radically.
3. Entirely new constellations of orders can be discerned which
do not correspond to the subclasses previously distinguished.-Example:
The constellation comprising Cornanae (Ericales, Cornales, Sarraceniales,
'!Eucommiales)-Gentiananae (Dipsacales, Goodeniales, Oleales, Gen-
tianales)-Lamianae (Scropkulariales, Hippuridales, ?Hydrostackyales,
Lamiales)-Loasanae (Loasales). This group is characterized by the
presence of iridoids; unitegmic nearly always tenuinucellate ovules;
usually cellular endosperm formation (exception most Gentianales);
usual absence of ellagitannins and absence of polyacetylenes, sesquiter-
280 R. DAHLGREN:

pene lactones and inulin. Another (similar) natural group is Asteranae-

Aralianae with other characteristics.
4. The presence of orders that can only with difficulty be referred
to any of the subclasses.-Example: Nympkaeales, which by virtue
of its monoaperturate (or inapertur~te) pollen grains agrees with the
orders of M agnolianae but lacks their essential oils and benzylisoquino-
line alkaloids. Ellagic acid is sometimes present in Nympkaelaes whereas
it is totally lacking in Magnolianae and Ranunculanae. Nympkaeales
exhibits conspicuous similarities to taxa in the monocotyledonous
Alismatanae, especially Hydrockaritales. It is thus difficult to place
in any subclass and it is best treated as a separate group between the
Magnolianae and Alismatanae. Even more problematic is the posi-
tion of Ratllesiales, Balanopkorales, Proteales, Elaeagnales and Podo-
stemales. Eucommia, placed in the subclass Hamamelidae, has important
embryological and phytochemical features in common with the Cornanae.
Thus I consider it justified to split up the subclasses to facilitate
the redistribution of their constituents. This will make it easier to
take advantage of recent advances in, for example, phytochemistry
and ultrastructural research.

Aims of the System and Diagram Used

A diagram such as that presented here can be used to survey the
range of distribution, consistency and concentration of a particular
character state. Where different attributes tend to occur in the same
groups useful patterns of correlation and interdependence can thus be
more easily discerned and assessed.
After the distributional patterns of a considerable number of char-
acters have been illustrated and compared it will presumably be possible
to deduce certain taxonomic conclusions. In the diagram the families
have fixed positions within the orders; if they repeatedly prove to
deviate in important characters from other families of the order or
superorder where they have been placed they should probably be moved
to another position in a revised version. There are several cases where
it may also be justified to radically rearrange orders or superorders.
It is hoped that in the course of time it will be possible to revise the
system and diagram on the basis of a comparison of the diagrams in
combination with new achievements presented in the literature.
Various suggestions for improvements have already begun to ac-
cumulate. The following can be mentioned as examples:

Fouquieriaceae should be removed from Saxifragales to the vicinity of

Ericales (DAHLGREN et al. 1976b).
Oochlospermaceae should be included in Bixaceae and placed in Oistales.
Commentary on a Diagrammatic Presentation of the Angiosperms 281

Elaeocarpaceae should probably be removed from Malvales.

Dichapetalaceae is probably best placed in Euphorbiales.
Gyrostemonaceae and Bataceae should probably be removed from Cap-
parales, into a separate order.
The position of several of the smaller families in Celastrales should be re-
considered; M ontiniaceae, for example, should be placed in Cornales.
Rhamnales should be divided, Vitaceae and Leeaceae being placed in a
separate order.
Rosales and Fabales are probably better treated in sepaTate superorders,
Fabales perhaps closer to Sapindales.
Connaraceae seems to be better placed in Sapindales than in Rosales.
A position for Elaeagnaceae closer to Proteales should be considered.
The superorders Plumbaginanae and Primulanae could perhaps be closer
to Theanae.
Centrolepidaceae shobld be di vided into two families.

Finally I should like to appeal to specialists in all relevant fields

for contributions presenting the distribution of any attribute. Contribut-
ing information on affinity between groups of angiosperms at family
level or higher will also be welcomed.

Several colleagues have contributed considerably to this article. The

chemical data have been supplied by S. ROSENDAL JENSEN and B. JUHL
NIELHEN, Technical University of Denmark, with the exception of those
on ellagic acid which I have received from E. C. BATE-SMITH, Agricultural
Research Council, Cambridge, England. P. WAGNER, Central Botanical
Library, Copenhagen, has kindly supplied the data on vessels in mono-
cotyledons. Mrs. MARGARET GREENWOOD·PETERSSON, Lund, Sweden, has
revised the English text.

References
BATE-SMITH, E. C., 1973: Systematic distribution of ellagitannins in relation
to the phylogeny and classification of the angiosperms. Nobel Symposium
25,93-102.
BEHNKE, H.-D., 1971: Zum Feinban der Siebrohren·Plastiden von Ari-
stolochia und Asarum (Aristolochiaceae). Planta 97, 62-69.
1975: P.type sieve element plastids: a correlative ultrastructural and
ultrahistological study on the diversity and uniformity of a new reliable
character in seed plant systematics. Protoplasma 83, 91-101.
and DAHLGREN, R., 1976: The distribution of characters within an angio-
sperm system. 2. Sieve-element plastids. Bot. Notiser (Lund) 129,
287-295.
BREWBAKER, J. L., 1967: The distribution and phylogenetic significance
of binucleate and trinucleate pollen grains in the angiosperms. Amer. J.
Bot. 54, 1069-1083.
CRONQUIST, A., 1968: The evolution and classification of flowering plants.
London-Edinburgh: Nelson.
DAHLGREN, R., 1975a: A system of classification of the angiosperms tD be
used to demonstrate the distribution of characters. Bot. N otiser (Lund)
128, 119-147.
282 R. DAHLGREN:

DAHLGREN, R., 1975 b: Current topics. The distribution of characters

within an angiosperm system. I. Some embryological characters. Bot.
Notiser (Lund) 128, 181-197.
in cooperation with HANSEN, B., JAKOBSEN, K., and LARSEN, K., 1974:
Angiospermernes taxonomi, 1. (In Danish.) K0benhavn: Akademisk
Forlag.
1975a: Ibid. 2.
1975b: Ibid. 3.
1976a: Ibid. 4.
JENSEN, S. R., and NIELSEN, B .•J., 1976b: Iridoid compounds in l!'ou-
quieriaceae and notes on its possible affinities. Bot. N otiser (Lund)
129, 207-212.
ETTLINGER, M., (unpubl. ms.): Plant analysis by butterfly; the occurrence
of benzylglucosinolate in Batis maritima and its context in systematics
and chemical evolution.
Fum, F., and WATERMAN, P. G., 1973: Chemosystematics in the Rutaceae
II. The chemosystematics of the Zanthoxylum/Fagara complex. Taxon
22, 177-203.
GUPTA, B. D., BANERJEE, S. K., and HANDA, K. L., 1976: Alkaloids and
coumarins of Heracleum wallichii. Phytochemistry 15, 576.
HANSEN, B., 1976: Pollen and stigma conditions in Balanophoraceae. Bot.
Notiser (Lund) 129,341-345.
HESLOP-HARRISON, J., 1968: Ribosome sites and 8 gene action. Nature
218,90-91.
1975: Incompatibility and the pollen-stigma interaction. Ann. Rev.
Plant Physiol. 26, 403-425.
1976: A new look at pollination. Rep. K MaIling Res. 8tn. for H175,
141-157.
HANSEL, R., LEUSCHKE, A., and GOMEZ-POMPA, A., 1975: Aporphine-type
alkaloids from Piper auritum. Lloydia 38, 529-530.
JENSEN, S. R., NIEI.SEN, B. J., and DAHLGREN, R., 1975: Iridoid com-
pounds, their occurrence and systematic importance in the angiosperms.
Bot. Notiser (Lund) 128, 148-180.
l{uBITZKI, K., 1969: Chemosystematische Betrachtullgen zur Uro/3gliede-
rung del' Bliitenpflanzen. Taxon 18, 360--368.
- 1973: Probleme del' Gro/3systematik del' B1iitenpflanzen. Bel'. dtsch.
bot. Ges. 85, 259-277 (1972).
LEE, D. W., and FAIRBROTHERS, D. E., 1972: Taxonomic placement of
the Typhales within the monocotyledons: preliminary serological in-
yestigation. Taxon 21, 39-44.
LEINS, P., 1972: Das Androeceum del' Dikotylen. Bel'. Deutsch. Bot. Ges.
84, 191-193.
--- 1975: Die Beziehungen zwischen multistaminaten und einfachen An-
droeceen. Bot. Jahrb. Syst. 96, 231--237.
MERXMULLER, H., 1972: Systematic botany---an unachieved synthesis.
BioI. J. Linn. Soc. 4, 311--321.
NANNFELDT, J. A., 1968: Fungi as plant taxonomistt:!. Festskrift till T. SE-
GERSTEDT. Acta Univ. Uppsala, 85-95.
PHILIPSON, W. R., 1974: Ovular morphology and major classification of
the dicotyledons. Bot. J. Linn. Soc. 68, 89-108.
STEBBINS, G. L., 1974: Flowering plants. Evolution above the species
level. Cambridge, Mass.: Harvard Univ. Press.
Commentary on a Diagrammatic Presentation of the Angiosperms 283

TAKHTAJAN, A., 1969: Flowering plants. Origin and dispersal. Edinburgh:

Oliver & Boyd.
THORNE, R. F., 196B: Synopsis of a putative phylogenetic classification of
the flowering plants. Aliso 6, 57-66.
TUCKER, S. C., 1972: The role of ontogenetic evidence in floral morphology.
Adv. PI. Morph., 359-369.
WAGNER, P., in prep.: Distribution of vessels in roots, stems and leaves of
the monocotyledons. Bot. N otiser 130.
WALKER, J. W., 1974: Aperture evolution in the pollen of primitive angio.
sperms. Amer. J. Bot. 61,1112-1137.
1976: Comparative pollen morphology and phylogeny of the Ranalean
complex. In: Origin and early evolution of angiosperms (BECK, C. B.,
ed.), 241-299. Columbia Univ. Press.
and DOYLE, J. A., 1976: The bases of angiosperm phylogeny: Palynology.
Ann. Missouri Bot. Gard. 62, 664--723 (1975).
WUNDERLICH, R., 1959: Zur Frage der Phylogenie del' Endospermtypen
bei den Angiospermen. Osterreich. Bot. Zeitschr. 106, 203-293.

Address of the author: Prof!)ssor ROLl!' DAHLGREN, Botanical Museum

of the University of Copenhagen, Gothersgade 130, DK·1123 Copenhagen,
Denmark.
Plant Syst. Evol., Suppl. 1, 285-298 (1977)
@ by Springer-Verlag 1977

Institut fur Allgemeine Botanik und Botanischer Garten,

Universitat Hamburg, Federal Republic of Germany

The Treatment of the Monocotyledons in an Evolutionary

System of Classification
By

Herbert Huber, Hamburg

Abstract: The more characters to distinguish monocotyledons and

dicotyledons that are taken into consideration, the more numerous become
the families which exhibit both mono- and dicotyledonous characters. In
particular, monocotyle~ons and ranalean dicotyledons appear intimately
related and the gaps between them do not allow a distinction into classes
to be made. The author, therefore, considers the monocotyledons and the
ranalean dicotyledons as two oxtreme wings of a single natural unit, with the
Annonaceae, Aristolochiaceae, Nymphaeaceae, and Piperaceae as connecting
links. Within the monocotyledonous wing, 12 natural units of higher than
ordinal level (superorders) may be tentatively recognized and arranged
according to the number of dicotyledonous features they possess and to the
degree of their systematic isolation. Whereas the groups which presE'nt at
least a few clearly dicotyledonous characters (like Arales and Helobiae)
occupy a more isolated position, the reduction or absence of dicotyledonous
characters (as in the anemophilous monocotyledons) is accompanied by
decreasing isolation. This indicates that among monocotyledons which
have attained a high evolutionary level, convergence largely camouflages
relationship. .

For more than half a century the fact has been acknowledged that
certain characters, widely distributed among, and generally attributed
to, the monocotyledons, occur sporadically in the dicotyledons as weD,
particularly within the ranalean orders. The occurrence of these mono-
cotyledon-features outside the monocotyledons present a considerable
problem in angiosperm classification, especially if one takes into ac-
count how few characters form the basis of our current definition of
the monocotyledons. This overlap of characters is reflected in most
modern treatments of angiosperm classification by the definition of
monocotyledons as an early and strongly specialized offshoot of the
ranalean group, an opinion against which there is little objection,
provided it is admitted that differentiation took place before the recent
orders and families had emerged.
286 H.HuBER:

A fundamental bipartition of angiosperms into monocotyledons

and dicotyledons has nevertheless been unanimously maintained by
practically all recent writers. This may serve convenience but, as I see
the situation, it is not an appropriate way to demonstrate existing
relationships. I should, therefore, like firstly to point out the problem
of the delimitation of the monocotyledons from the ranalean dicotyle-
dons, and secondly to illustrate what I consider to be the natural units
above the rank of family within the monocotyledonous group.
Characters that obscure the delimitation between the monocotyle-
dons and the dicotyledons may be roughly divided into two categories:
the first which are centred within the dicotyledons, only extending
casually or in very few groups into the monocotyledons, and the second,
on the contrary, represented by typical monocotyledon-characters
which are uncommon or rare within the dicotyledons. There is a striking
difference in the overall distribution of these two sets of characters:
Among the numerous features which belong to the first category there
is not a single one which is exclusive to monocotyledons and the ranalean
dicotyledons, however broadly one may construe the latter. All charac-
ters are commonly found in the non-ranalean dicotyledons as well. On
the other hand, the typical monocotyledon characteristics which invade
the dicotyledons are almost totally restricted to the ranalean group
and many of them occur within the central orders distinguished by
their synthesis of benzyltetrahydroisoquinoline alkaloids. I am con-
sequently inclined to consider that the differentiation of the angio-
sperms into non-ranalean dicotyledons and the ranalean complex,
including the monocotyledons, took place prior to that into the ranalean
dicotyledons and the monocotyledons.
The characters of the second category are particularly important
because they indicate how imperfect the differentiation into ranalean
and monocotyledonous orders is still today. When present in dicotyle-
dons, these characteristics tend to group into syndromes, as is well
shown by Annonaceae and Aristolochiaceae which share several char-
acters with monocotyledons. These characters are: (a) sieve tube
plastids with protein crystalloids; (b) median position of prophyll;
(c) primarily trimerous perianth; (d) nectar secretion from floral phyl-
lomes instead from a disc; (e) unisulcate or inaperturate pollen grains.
Among these characters, prophyll-position is perhaps the most reo
markable, because this can hardly be considered primitive. According
to FRIES (1959) the Annonaceae uniformly present single median pro-
phylls within the floral region, and predominantly so but with numerous
exceptions within their vegetative shoot, but normally with a dico-
tyledonous emhryo. Apparently the tendency to produce single median
prophylh'l arost' pt'ripherically in thE' plant body and extended proxi-
Treatment of the Monocotyledons 287

mally until it finally took over on the embryo whose. cotyledons are
equivalent to prophylls, thus giving rise to a monocotyledonous embryo.
Another example of concentrl}tion of monocotyledon-characters
outside of the monocotyledons proper is in the Piperaceae. Again we
find a primarily trimerous perianth, and pollen grains unisulcate or
derived from the unisulcate type. In addition to this, the Piperaceae pre-
sent medullary stem bundles; the stomata frequently surrounded by
a rosette of subsidiary cells; and a tendency towards a completely
closed reticulation of veinlets.
Benzyltetrahydroisoquinoline alkaloids have recently (HANSEL & al.
1975) been discovered in one species of Piper, but on the whole, argu-
ments to include the Piperaceae and Saururaceae, in the ranalean
group are weaker than for the Annonaceae and Aristolochiaceae. In parti-
cular, the Piperaceae have perisperm, otherwise known in the ranalean
orders only from Nymphawles (without Nelumbo).
This, however, is the third example of a major assemblage of mono-
cotyledonous characters occurring outside monocotyledons proper
and perhaps the most 'perplexing one. Apart from the medullary stem
bundles and unisulcate pollen grains (at least in Brasenia, Cabomba
and Nuphar) , the order exhibits one most unexpected feature, not,
otherwise known outside typical monocotyledons, i.e., as HAINES & LYE
(1975) have pointed out, an embryo possessing not only a single cotyle-
don but forming a coleoptile. Finally, the helobial formation of endo-
sperm in Cabomba is worth mentioning. This clearly monocotyledonous
assemblage of characters is counterbalanced by an apparently primarily
polymerous, spirally constructed flower, which is quite out of place
witllin the monocotyledons. The same is true of the ellagitannins,
reported from Nymphaeaceae but absent from both monocotyledons
and the ranalean dicotyledons. As to sieve tube plastids, the Nym-
phaeales and the Piperales form part of the ordinary dicotyledonous
pattern and correspond neither with typical monocotyledons nor with
the Annonaceae.

The Nymphaeale8 and Piperale8 are not elosely related to eaeh other nor
to any particular group within or without the monocotyledons. Nevertheless,
their general affinity strongly tends in the latter direction, far more than
t,owards the ranalean orders. Apart from the characters already referred to,
this fact is illustrated (Fig. 1) by the presence of laticiferous tubes in Nym-
phaeaceae, Helobiae (Ali8mataceae), Araceae, and Scitamineae (Musa); the
occurrence of perispermous seeds in Nymphaeales, Piperales (Chloranthaceae
excluded), and Scitamineae; and the cellular formation of endosperm in
most. Nymphaeaceae, many Piperaceae (not always in P?:per) , and in th('
Arales altogether. Generally speaking, the Piperale8, Nymphaeales, Arales,
Helobiae, and Scitamineae have in common a tendency not to employ endo-
sperm as a principal storage tissue in their seeds but substitute it paItly
288 H. HUBER:

or completely, either by a storage nuceJlus commonly called perisperm, or

by storage embryos. In the Arales, the latter is merely a tendency; in the
Helobiae no trace of a storage tissue other than the embryo is left in the ripe
seeds. I cannot pretend that perispermous seeds constitute a particularly
derived character but storage embryos certainly do; in the monocotyledons,
however, they do not indicate a very successful evolutionary tendency.

The five orders mentioned above are equally distinct from each
other as from the rest of the monocotyledons. Accordingly I do not
suggest any particularly close relationship of the Piperales with the
Arales as suggested by EMBERGER (1960), although one must admit
that Arales present certain piperalean features, which can not be re-
jected as casual.
The A rales agree well with the Piperales among other characters
in leaf articulation and venation; stomata frequently with a rosette
of subsidiary cells; cellular formation of endosperm; atropous ovules
(in part of Araceae); and pseudomonomerous ovaries in Peperomia
and a few Araceae. They differ fundamentally in the loss of vessels
in the shoot, thus representing an outstanding monocotyledonous
progression. Another highly evolved feature which the Arales have
in common with various specialized but unrelated monocotyledons
is an ameboid periplasmodium. The two last mentioned characters
and to a certain degree the possession of subsidiary cells arranged in
a rosette are shared by Arales, Helobiae, and Scitamineae. In these
three groups the pollen is frequently inaperturate and trinucleate
when shed. As to pollen apertures, unisulcate grains, supposed to be
primitive in angiosperms, are common in the Piperales and Nym-
phaeales; they occur in few members of the Arales and Helobiae, but
not in the Scitamineae. Inaperturate pollen, considered derived from
unisulcate, occurs in the Piperales (Peperomia), Nymphaeales (Oerato-
phyllum), Arales (particularly abundant in the geophytic genera),
. and is frequent in the Helobiae. This may suggest an adaptation to
aquatic pollination, but the predominance of inaperturate pollen in
Scitamineae contradicts this explanation.

Abbreviations (Fig. t):

Ara Arales Gram Graminales Phi Philydrales
Arec Arecales = Poales Pip Piperales
AspaJ' Asparagales Haem H aemodorales Pont. Pontederiales
Bro Bromeliales Hel Helobiae Scit Scitamineae
Cyp Oyperales June J1~ncales Ste Stemonales
Dio Dioscoreales Lil Liliales Tac Taccales
Far Far'inosae Nym N ymphaeales Typh Typhales
Orch Orchidales
Pand Pandanales
Treatment of the Monocotyledons 289

Fig. 1. Diagram showing specific superposition of characters, indicative of

ordinal and higher level in the monocotyledons. Orders, or groups of orders,
such as the Farinosae and Helobiae, surrounded by a signature, commonly
possess the character indicated in the key, those placed outside do not.
Names crossing a line indicate that the respective feature is present in a part
of the order or orders only. Characters restricted to one order have been
neglected, for which reason the Taccales, e.g., appear closer to the Liliales
than they are; nor have primitive characters, such as apocarpy or presence
of proanthocyanidins, been indicated as they tend to occur in conservative
representatives of several evolutionary lines, nor have derived features
obviously polyphyletic in origin, such as anemophilous and epigynous
flowers (for abbreviations see p. 288)

S·type Sieve tube plastids pre· perianth differentiated into

sent calyx and corolla
vessels only in the rootR septal nectaries present
()-o-o silicia accumulation in partic- seeds with storage nucellus
ular cells (perisperm)
)( ...... ......
)( )(
endosperm forma.tion cellular
l<-X-X laticiferous tubes present.
'-1-1- stomata with 2 or more sub· , .. ... ~ .....I .. .. · endosperm with hemicellulose
sidiary cells
II........ ........ endosperm with starch
perianth petaloid not differ-
entiated into calyx and corolla 0 .... ·0 .....0 embryo lateral
Plant Syst. Evo!. , Supp!. 1 19
290 H. HUBER:

Apart from the Arales, the Helobiae represent another example

of extreme heterobathmy. In addition to the characters already men-
tioned, tHe Helobiae possess, insofar as they are zoophilous, heterochla-
mydeous flowers as found in the monocotyledons, e.g. in the Scitamineae,
Bromeliaceae, and Enantioblastae. Secondarily polyandrous flowers
mark another progression. On the other hand, apocarpy, laticiferous
tubes, intrapetiolar stipules and the terminal plumule of Ottelia can be
considered primitive or, more correctly, reminders of the dicotyledons.
If one disregards the reduced aquatic plants, such as the Lemnaceae
and Z08tera, the most derived group among the isolated evolutionary
lines of monocotyledon affinity is the Scitamineae. This is evident
from the following vegetative and reproductive characters: the flowers
tend to become monosymmetric or asymmetric; the number of stamens
becomes reduced; septal nectaries are universally present; and silica
bodies or silica idioblasts are frequent in leaves and stems. The latter
character is often correlated with the arborescent habit. Rosette trees
indeed occur in Scitamineae. There is now a tendency to consider this
habit rather primitive among monocotyledons, but at least as far as
the Scitamineae are concerned I agree with STEBBINS (1974) in rejecting
this opinion. As it seems, the arborescent genera Ravenala from Mada-
gascar, Strelitzia from South Mrica, and Phenako8permum from the
Guayanas are pollinated either by bats or by birds (Strelitzia only).
Disjunction in distribution indicates age, but pollinators do not: bats
as well as Pas8eri/ormes date back, at most, into the lower Tertiary.
Not much is known about the forest floor herb Orchidantha, but I
think this would make a much better model of a primitive Scitaminea.
With respect to the somewhat marginal position occupied by Pi-
perales, Nymphaeales, Arales, Helobiae, and Scitamineae, the "central"
group of monocotyledons shows at least one character common to
most of its members: this is, the presence of a well developed storage
endosperm, whereas perisperm is scarce, or mostly absent, and storage
embryos are almost lacking. Orchids, with mycotrophic germination,
make the only considerable exception, but their seeds, which lack
storage tissue throughout, have no equivalent in the previously mentio-
ned groups either. Perhaps less important is the observation that,
contrary to the marginal monocotyledonous orders, the tendency
towards inaperturate pollen is much more weakly developed within
the "central" group, a few A8paragale8, Iridaceae, Hanguana, and
part of Xyridaceae excepted.
Distinction of "marginal" and "central" groups does not imply that the
latter are derived from the former nor does it imply the absence of primitive
and dicot.yledonous features from the central group. Within the latter,
Liliiflorae can be called reasonably distinct, at least if one limits them to
Treatment of the Monocotyledons 291

·the orders Dioscoreales, Stemonales, Asparagales, and Liiiales, as I have

defined them elsewhere (HUBER 1969).

From an anatomical point of view, the Liliiflorae are almost suf-

ficiently described by the possession of anomocytic stomata, and the
absence of silica bodies or silica idioblasts. Vessels are mostly confined
to the roots, but this is not so in Dioscoreales and S'f!I,ilacaceae, which
I consider, as to this character, to be the most primitive members
of this assemblage. The outer integument consists commonly of more
than two cell-layers and the inner integument is of two layers only.
The endosperm accumulates fat oil, protein, and mostly hemicellulose
deposited on the cell walls. The Liliiflorae primarily possess nectari-
ferous tepals, but many of them replace this original condition by
septal nectaries. Apart from this, the Asparagales produce rosette
trees, with or without secondary thickening of the stem. An unusual
character is the steroid saponins which abound in the Dioscoreales,
Stemonales, and numerous Asparagales, particularly in their phanero-
phytic members. They have been found in smaller quantities in the
Liliales and Orchidales, two orders presumably related to each other,
although the Orchidales certainly did not descend from any recent
member of Liliales.
The liliiflorous monocotyledons have some important characters
in common with the Scitamineae (suppression of vessels in the shoot,
septal nectaries, and presence of chelidonic acid being the most note-
worthy) and the Arales (presence of steroid saponins and again sup-
pression of vessels), but more pronounced is affinity with what I have
previously called the palmaceous evolutionary line (Arecanae of DAHL-
GREN 1976), that is palms including the Cyclanthaceae and screw pines.
The relationship of these three orders seems fairly well established.
As to anatomical characters, the palmaceous monocotyledons
are distinguished by (a) stomata mostly surrounded by a rosette of
subsidiary cells; (b) Arecales, but not Cyclanthales and Pandanales, by
accumulation of silica; (c) Arecales and Cyclanthales, but not Pand-
anales, by postgenital partition of the leaves; (d) Arecales and Pand-
anales by presence of vessels throughout their vegetative organs;
(e) Arecales and Pandanales, but not Oyclanthales (which are zoo-
philous), by a tendency towards reduction in number of ovules, a
characteristic frequently correlated with anemophily; (f) all three orders
by occurrence of polyandrous flowers. Pollen in the Oyclanthaceae and
Pandanaceae often exhibits ulcoidate apertures which are almost ab-
sent from the Liliiflorae. Chelidonic acid, widely distributed in the
Liliiflorae, has never been reported from palmaceous monocotyledons.
Most noteworthy is the arborescent habit of most palms and many
111*
292 H. HUBER:

Pandanales. With respect to the association of this character with evi-

dently derived features (silica accumulation in Arecales, septal nectaries
in A sparagales , silica accumulation and. septal nectaries in the Sci-
tamineae and Bromeliales), I consider the pachycaul habit in the mono-
cotyledons not as an archaic but a derived feature, and regard a plant
like Freycinetia or a climbing Gyclanthaceae as a much better choice
for a primitive palmaceous representative.
The affinity of the palmaceous monocotyledons with the Lilii-
florae is stressed by presence of abundant endosperm storing hemi-
cellulose, apart from fat oils and aleurone, whereas starch is mostly
absent from the mature endosperm. Another seed-character in favour
of an affinity of palmaceous with liliiflorous monocotyledons is rumi-
nate endosperm as present in the Trichopodaceae, species of Yucca
and many palms. This character, almost invariably correlated with
woody habit throughout angiosperms (the Trichopodaceae are the
only noteworthy exception) may indeed indicate primitiveness, as
recently expressed by CORNER (1976). Steroid saponins are not common
in palms, but occur in a few genera in large quantities (HEGNAUER
1963).

Faced with this list of characters that are shared by Liliiflorae and
palmaceous monocotyledons, why not., one may wonder, attribute the
Dioscoreaceae to the latter. The presence of vessels in axis and petioles, the
tendency to reduce flowers (which are unisexual and anemophilous in
Dioscoreaceae proper) and ruminate seeds might indicate this. Nevertheless,
the structure, venation and vernation of the leaves, stomata (anomocytic
in Dioscoreales), the tendency towards a geophytic habit, epigynous flowers,
presence of chelidonic acid and absence of silica permit the Dioscoreales to
be seen as undeniable members of the Liliiflorae.
Araceae, conventionally but wrongly associated with the palmaceous
group, differ markedly by an opposite tendency in the evolution of the
androecium, as the number of stamens tends to become reduced and not
increased as in many palmaceous monocotyledons, quite apart from the
vessel and endosperm characters already described.

The crucial areas in monocotyledon systematics are the families

whose seeds present two commonly superposed specializations, i.e.
both integuments consisting of two cell layers; and accumulation of
starch instead of hemicellulose as the principal storage carbohydrate in
the endosperm. The great diversity of families which exhibit these
characteristics indicates that this syndrome must have arisen several
times. There are examples of a positive correlation of this syndrome
with almost every character reported above both for the liliiflorous
and palmaceous monocotyledons, save three exceptions: anomocytic
stomata; postgenital partition of leaf or compound or lobed leaves
Treatment of the Monocotyledons 293

throughout; and ruminate seeds (there is, however, an indication of

ruminate seeds in AngiozantMs, Haemodoraceae).
It is logical to compare the flj,milies with the petaloid perianth
not differentiated into calyx and corolla with Liliillorae. Additional
liliiflorous characters found in the families which present seeds and
perianth as described are: septal nectaries in Haemodoraceae and Pon-
tederiaceae; and suppression of vessels in the aerial organs of Phily-
draceae and Pontederiaceae. As in the Liliillorae, neither is silica ac-
cumulated nor has ulcerate or ulcoidate pollen been reported. Apart
from this, the Haemodoraceae resemble the Liliillorae in presence of
chelidonic acid, although they differ from the majority by having
vessels distributed throughout the vegetative organs. Information
available at present does not allow to decide if the similarities between
the Haemodoraceae, Philydraceae, and Pontederiaceae are due to af-
finity, as suggested by DAHLGREN (1974), or to convergence.
In my opinion, the Haemodoraceae are distantly related to the
Taccaceae, a family which combines dicotyledonous features, one of
them not otherwise encountered in monocotyledons (DAVIS 1966),
with suppression of vessels in the shoot, both integuments two-layered
and an endosperm, free from starch, storing hemicellulose although
in modest quantities. A remote affinity may be deduced from the
presence of a peculiar pigment in the seed coat both in the Haemo-
doraceae and the Taccaceae and other characteristics as well (HUBER
1969).
In contrast to the zoophilous and primarily homoiochlamydeous
families, all but one of the families which combine a perianth either
differentiated into calyx and corolla, or glumaceous, reduced and
anemophilous with the above mentioned superposition of specialized
seed characters, develop vessels in all their vegetative organs and they
mostly accumulate silica in peculiar cells (except the Sparganiaceae,
Typhaceae, and the greater part of the Juncaceae). In these families,
silica accumulation is frequently associated with ulcerate or ulcoidate
pollen (exceptions: Commelinaceae, Eriocaulaceae, Mayacaceae, Ra-
pateaceae, Typhaceae). Endosperm formation, helobial throughout
Haemodoraceae, Philydraceae, and Pontederiaceae, is nuclear in the
calyciflorous and anemophilous families with exception of the J uncaceae
(and possibly the Sparganiaceae and Typhaceae as well). These characters
viewed together indicate affinity with the palmaceous evolutionary
line and I would not hesitate to conEider the familes in question en bloc
as reduced offshoots of palmaceous ancestry, if these families were
indeed uniformly anemophilous or achlamydeous and there were not
conspicuous examples of highly evolved entomophily, presenting a
perianth differentiated into calyx and corolla, as in the Commelinaceae
294 H.HuBER:

and Xyridaceae. As a matter of fact there is no way in which one can

imagine these calyciflorous families derived from, or in any definable
way related to, the palmaceous orders. However, there seems to exist
at least a chance of understanding the calyciflorous families better.
It has long been known that both calyciflorous and certain anemo-
philous families (Centrolepidaceae, Eriocaulaceae, Plagellariaceae, Gra-
mineae, Rapateaceae, and Restionaceae) possess an almost unique feature
not otherwise observed in the angiosperms: that is the embryo removed
from the axial position, which it originally helds, and attached laterally
to the endosperm. This group of families corresponds with the Pari-
nosae as defined by HAMANN (1961), including the Gramineae.
There is a strange resemblance of the Enantioblastae--the name I prefer
for this assemblage-with the Scitamineae, with which they share, apart
from heterochlamydeous flowers, silica accumulation, stomata with sub-
sidiary cells arranged in a rosette (in Enantiobla.stae due to their mostly
narrow leaves particularly in broad-leaved Commelinaceae) and, if one has
the perfume grasses in mind, accumulation of essential oils in one-celled
idioblasts. More stringent, however, is the affinity of the Enantioblastae with
the Bromeliaceae. Among the monocotyledons with both two·layered
integuments and farinose endosperm, the Bromeliaceae are the only family
which combines differentiation of perianth into calyx and corolla wit.h
suppression of vessels in the shoot. The bromeIiads indeed exhibit a most
puzzling constellation of characters of different evolutionary lines. They
agree with the Enantioblastae in their mostly two-layered outer integument"
farinose endosperm, and commonly lateral position of t,he embryo; with
the Enantioblastae and Scitamineae, in silica accumulation, stomata provided
with subsidiary cells arranged in a rosette, and heterochlamydeous flowers;
with the Liliiflorae in unisulcate or two- or multiaperturate but not ulcerate
nor ulcoidate pollen grains; with the Liliiflorae and Scitamineae, in suppression
of vessels in leaves and axis, occasionally arborescent habit (Puya), septal
nectaries. and frequently helobial formation of the endosperm (in Scitami-
neae, Zingiberaceae only) ; with the Scitamineae, finally, in a decided tendency
towards ornithophily combined with persist,ent petaloid bracts.
Leaves with a thorny toothed margin which are characteristic for the
terrestrial bromeliads do not occur in either the Enantioblastae or the
Scitamineae but are not infrequent in the Liliiflorae (particularly among
Asparagales), in the Juncales (Prionium and Thumia), and in the Velloziaceae,
which seem to be the closest relatives of the Bromeliaceae. Steroid saponins,
widely distributed in liliiflorous monocotyledons, occur in the Bromeliaceae
in but small quantitieR, whereas chelidonic acid seems to be absent.

Generally speaking, the relationship of the Bromeliaceae to the Enantio-

blastae, Juncales, lI-nd Scitamineae is more prominent than to the Lilii-
florae (Velloziaceae excepted), but they do not fit well into either evolu-
tionary line.
Having now eliminated the families with an embryo laterally at-
tached to the endosperm, there remains a complex of families that
uniformly possesses vessels throughout the vegetative body, anemo-
Treatment of the Monocotyledons 295

philous flowers, mostly ulcerate or ulcoidate pollen grains frequently

coherent to form a tetrad and, as far as the families embrace more
than one genus, with at least one instance of a rosette tree. This group
comprises the Cyperaceae, Juncaceae, Sparganiaceae,. Thurniaceae,
and Typhaceae, most of which have been attributed by various autho-
rities and to some degree, to the palmaceous monocotyledons-cor-
rectly, as I should say, at least as far as the Cyperales (with one family
only) and Typhales (two families) are concerned, although arguments
for this attribution admittedly consist rather in absence of serious
obstacles than in positive evidence.
As to the Juncales (Juncaceae and Thurniaceae), I am less confident
to take their palmaceous affinity for granted. Apart from helobial
formation of the endosperm, this order rather rigidly maintains pluri-
ovulate ovaries, which is an unexpected feature in an natural unit which
should be long adapted to anemophily. The only alternative place
for the Juncales, if not in the palmaceous group, would be in such a
superficially dissimilar assemblage as the Haemodorales and Philydrales,
provided the two eventually prove to be related with each other, which
is still uncertain.
To conclude this paper, I shall put forward a few ideas as to the
delimitation of dicotyledons from the monocotyledons and the ar-
rangement of the orders and superorders within the latter. The dif-
ferentiation of the ranalean complex in its broadest sense (i. e., including
the monocotyledons) into the ranalean and the monocotyledonous wing
has not yet reached such a degree of perfection that intermediates have
been eliminated. Characteristic specializations in the central groups, both
of the ranalean dicotyledons and the monocotyledons, frequently allow
attribution of intergrades to either extreme, e.g. the Annonaceae and
Aristolochiaceae to the ranalean wing; the Dioscoreales and Taccales to the
monocotyledonous wing; the Piperales and N ymphaeales are best placed
halfway in between the two extremes. It would not appear logical to con-
sider the break between the Piperales and Nymphaeales on the one side
and the Arales, Dioscoreales, Helobiae, and Scitamineae on the other,
deeper or more indicative as to isolation than discrepancies between
the Piperales and Nymphaeales, or between the A rales , Dioscoreales,
and Scitamineae.
Provided systematics maintains its claim to express natural and
not administrative units, it is suggested that monocotyledons should
not be considered a class of their own, but rather a bundle of loosely
coherent evolutionary lines, some of them more isolated than others
and almost deserving the rank of a subclass, others intricately reti-
culate and, in this case, only with hesitation being ranked as super-
orders.
296 H.HuBER:

Table 1. Conspectus of the monocotyledonous superorders. (It should be

held in mind that the Juncales and Velloziales can not be attributed defi·
nitely to any superorder)

Dicotyledonous features Intermediate evolution· Dicotyledonous features rare or

prevailing ary state ab1'ent

a) isolated evolutionary
lines:
Piperiflorae (one or- Ariflorae (one order: Scitamineae (one order: Zingi-
der: Piperales) Arales) berales)
Nymphaeiflorae (one Helobiae or Alismati-
order: Nymphaeales) florae (2-4 orders)

b) reticulate evolution-
ary lines:
Lilliflorae (probably 5 or- Haemodoriflorae (one family:
ders: Dioscoreales, Ste- H aemodoraceae)
monales, Asparagales, Pontederiiflorae (2 orders: Pon-
Liliales, Orchidales) tederiales and Philydrales)
Tacciflorae (one family: Bromeliiflorae (possibly 2 or-
Taccaceae) ders: apart from Bromeliales
Palmiflorae (with 5 or 6 also V elloziales)
orders if construed Enantioblastae or Commelini-
broadly: Arecales, OycZ- florae (superorder embracing
anthales, Pandanales, all families with mealy endo-
Oyperales, TyphaZes, sperm and lateral embryo
and doubtfully Jun- except Bromeliaceae ; these
cales) families have been arranged
in one [THORNE 1968] to 4
[DAHLGREN 1976] orders)

The angiosperms, in my OpInIOn, represent one single and in spite

of their diversification remarkably homogeneous class which hardly
exhibits discontinuities deep enough to justify the recognition of sub-
classes. I therefore tend to arrange the angiosperm superorders in an
informal way, uniting the ranalean and monocotyledonous super-
orders so as to contrast them against with the non-ranalean dicotyle-
dons, and within this enlarged ranalean complex, the already mentioned
two extreme wings become obvious, with a few ambiguities in between.
For didactic purposes, I would propose an arrangement of the mono-
cotyledonous superorders as indicated in Table 1. All of these mono-
cotyledonous superorders, the bromeliads and Enantioblastae perhaps
excepted, exhibit a certain proportion of primitive, that is to say di-
cotyledonous characters, although there is a tendency for them to be
lost as a higher evolutionary level is attained.
Treatment of the Monocotyledons 297

With respect to their origin, an arrangement according to their

dicotyledonous similarities would appear feasible but as such a proce-
dure would lead towards a linear' sequence this would' necessarily
obscure the reticulation of affinities which is so prominent in this as-
semblage. A clearer method of illustrating relationships seems to con-
sist in plotting essential character constellations (Fig. 1) instead of
presumptive natural units as is usually done. Thus at least partly
the procedure on which systematic decisions are based, can be visualized,
what may well contribute towards a more objective system of classi-
fication.

References
BEHNKE, H.-D., 1975: The bases of angiosperm phylogeny: Ultrastructure.
Annals Missouri Bot. Gard. 62, 647-663.
CARLQUIST, S., 1975: Ecological strategies of xylem evolution. Berkeley-
Los Angeles-London: University of California Press.
CHEADLE, V. 1., 1953: Independent origin of vessels in the monocotyledons
and dicotyledons. Phytomorphology 3, 23-44.
and TUCKER, I. M., 1961: Vessels and phylogeny of monocotyledons. In:
Recent Advances in Botany (IX. International Bot. Congress 1959) 1,
161-165. Toronto.
CORNER, E. J. H., 1976: The Seeds of the Dicotyledons. Vol. 1,2. Cambridge:
University Press.
DAHLGREN, R., 1974-1976: Angiospermernes taxonomi, bind 1~~4. K0ben-
havn: Akademisk Forlag.
DAVIS, G. L., 1966: Systematic Embryology of the Angiosperms. New York-
London-Sidney: John Wiley & Sons.
EMBERGER, L., 1960: Les vegetaux vasculaires. In: Traite de botanique
(S;vstematique), (M. CHADEFAUD et L. EMBERGER), tome 2. Paris: Mas-
son & Cie.
ENGI,ER, A., 1964: Syllabus der Pflanzenfamilien, 12. Aufl. (MELCHIOR, H.,
Hrsg.). Berlin: Borntraeger.
ERDTMAN, G., 1952: Pollen Morphology and Plant Taxonomy: Angiosperms.
Stockholm: Almquist & Wiksell.
FRIES, R. E., 1959: Annonaceae. In: Die natiirlichen Pflanzenfamilien,
2. Aufl., Band 17 a II. Berlin: Duncker & Humblot. .
HAINES, R. W., and LYE, K. A., 1975: Seedlings of Nymphaeaceae. Bot.
Journ. Linn. Soc. 70, 255-265.
HAMANN, U., 1961: Merkmalsbestand und Verwandtschaftsbeziehungen
der Farinosae. Willdenowia 2, 639-768.
1962: Wei teres tiber Merkmalsbestand und Verwandtschaftsbeziehungell
del' "Farinosae". Willdenowia 3, 169-207.
HANSEL, R., LEUSCHKE, A., and GOMEZ-POMPA, A., 1975: Aporphine-type
alkaloids from Piper auritum. Lloydia 38, 529--530.
HEGNAUER, R., 1963: Chemotaxonomie der Pflanzen, Band 2: MOllo-
cotyledoneae. Basel und Stuttgart: Birkhauser.
HUBER, H., 1969: Die Samenmerkmale und Verwandtschaftsverhaltnisse
der Liliifloren. Mitt. Bot. St.aatss. Mtinchen 8, 219~-538.
298 H. HUBER: Treatment of the Monocotyledons

MEEUSE, A. D. J., 1975: Aspects of the evolution of the monocotyledons.

Acta Bot. Neerl. 24, 421-436.
MOORE, H. E., Jr., and UHL, N. W., 1973: Palms and the origin and evolu-
tion of the monocotyledons. Quart. Rev. BioI. 48, 414--436.
NETOLITZKY, F., 1926: Anatomie der Angiospermen-Samen. Handbuch der
Pflanzenanatomie, Band 10. Berlin: Borntraeger.
PORSCH, 0., 1914: Die Abstammung der Monokotylen uud die Bliiten-
nektarien. Ber. dtsch. bot. Ges. 31, 580-590.
RAMSTAD, E., 1953: JIber das Vorkommen und die Verbreitung von Cheli-
donsaure in einigen Pflanzenfamilien. Pharm. Act. Helv. 28, 45-57.
STEBBINS, G. L., 1974: Flowering Plants. Evolution Above the Species
Level. London: E. Arnold.
- and KHUSH, G. S., 1961: Variation in the organisation of the stomatal
complex in the leaf epidermis of monocotyledons and its bearing 011
their phylogeny. Am. Journ. Bot. 48, 51-59.
STONE, B. C., 1972: A reconsideration of the evolutionary status of the family
Pandanaceae and its significance in monocotyledon phylogeny. Quart.
Rev. BioI. 47,34-45.
SUESSENGUTH, K., 1920: Beitrage zur l<'l'age des systematischen Anschlusses
der Monocotylen. Beih. Bot. Zbl. 38 (II), 1--79.
THORNE, R. T., 1968: Synopsis of a putatively phylogenetic classification
of the flowering plants. Aliso 6, 57-66.

Address of the author: Prof. Dr. HERBERT HUBER, Institut fiir Allgemeine
Botanik und Botanischer Garten der Universitat Hamburg, Jungiusstr. 6~8,
D-2000 Hamburg 36, Federal Republic of Germany.
Plant Syst. Evol., Suppl. 1, 299-319 (1977)
© by Springer-Verlag 1977

Rancho Santa Ana Botanic Garden, Claremont, California, U.S.A.

Some Realignments in the Angiospermae

By

Robert F. Thorne, Claremont, Calif.

Abstract: The system of classification that I have developed since 1950

to indicate the phylogenetic relationships of the higher taxa of the flowering
plants has for the most part been published only in synoptical form. Since
some of my more iconoclastic alignments have not been fully explained in
print, this paper attempts to explain my reasons for some of the more
innovative of these positionings.
My classification of 'the Angwspermae deviates considerably from others
now widely accepted in several major ways. Because I attempt to stress
relationships more than differences, my taxa tend to be more inclusive,
while the differences within the major categories are recognized through
the use of subcategories like suborders and subfamilies. Family names are
in accord with the International Code of Botanical Nomenclature though
the nine "exceptional" names are dropped as obsolete. Also for the sake of
uniformity in phylogeny I have extended the principle of priority to the
names of orders and other higher categories up to the class, anticipating a
future rule or recommendation of the Code.
Taxa that are treated in some detail with explanation of the alignments
are ~ the Aquifoliaceae, Sarraceniaceae, and N epenthaceae of the Theales and
Plumbaginaceae of the Primulales in the Theiflorae; Fouquieriaceae with
Polemoniineae and Solanineae of the Solanales in the Malviflorae; Gyroste-
monaceae, Stylobasium DESF., and Emblingia F. MUELL. of Sapindineae and
Batis L. of Batineae, Rutales, in the Rutiflorae; Orossosomataceae of Rosineae,
Rosales, and Buxaceae (excluding Simmondsia NUTT.), Buxineae, and Bata-
nopaceae, Daphniphyllineae, in the Pittosporales of the Rosiflorae; Asteraceae
of the Asteriflorae near the Oorniflorae and Lamiiflorae and all three less
closely with Saxifragineae of the Rosiflorae; Alismatiflorae as less primitive
in the Monocotyledoneae than the Li~iillorae; and Juncineae and Poineae in
the Oommeliniflorae.

Introduction
Since 1950 I have been developing a system of classification of the
Angiospermae that presents as accurately as possible the relationships
of the higher taxa in that class. As a teacher of taxonomy I learned
very early how resentful students can be when expected to learn a system
of classification that they and their instructor realize is highly artificial,
300 R. F. THORNE:

grouping together taxa of dubious or distant relationship. In the early

1950's there were available no classifications that could be accepted
as even approximately phylogenetic. Therefore, I started to construct
my own system on a tentative and synthetic basis, stripping orders of
obviously or probably unrelated families and transferring them to other
orders where they did seem to belong or to a temporary fluctuating pool
of taxa incertae sedis.
Elsewhere I have published (THORNE 1958, 1963, 1974 c, 1976) the
principles which I have followed, my modus operandi, in this attempt
to devise a phylogenetic classification. I mention here only that basically
I have tried to find and use information that I regard as pertinent from
all parts of the flowering plants at all stages of their development. 1
have never hesitated to use information made available by morphologists,
cytologists, biochemists, physiologists, geologists, herbivores, predators,
parasites, or any other group deeply interested in the flowering plants
as objects of study, food, shelter, or hosts. It is, of course, presumptious
of me and ·of my fellow phylogenists, to attempt to gather, evaluate, and
apply this mass of data, but someone has to do it. There are always
available some presumptious fools who will march in where angels are
too prudent to tread.
Fortunately today there are a number, some would :;ay a plethora, of
classifications that purport to be phylogenetic, and that are indeed far more
natural than the classifications earlier in the century. None, including my
own, is yet truly phylogenetic, nor probably will anyone of them at.tain
perfection for botanical generations to come, if ever. What bothers students
Of the angiosperms about these various systems is that they differ both in
minor details and sometimes in major groupings. This is to be expected,
considering the different emphasis that each phylogenist places upon the
data available to him. None of us is completely right nor completely wrong,
and each makes important suggestions that. specialists can examine and
accept, 01' discard if found inadequate. Surely all good phylogenists examin{'
carefully the systems and the justifications given by their fellow botanists.
Where these phylogenists continue to differ, each regards his own interpreta.
tion of the data as more realistic than that of his rivals.

My system of classification differs considerably from other I:lystems

now widely accepted, and much of it I have presented only in synoptical
form without thorough discussion of my reasons for various realign-
ments. Because my system attempts to stress relationships more than
differences, my taxa tend to be more inclusive than those of my fellows
with more disintegrative tendencies. As a teacher I consider it pedagog-
ically unsound to expand the number of taxa unnecessarily through
the narrow definition of categories. I realize that in this period of taxo-
nomic and economic inflation I am fighting a rearguard action, but there
is a quixotic excitement about tilting with authoritative windmills and
Some Realignments in the Angiospermae 301

dodging juggernaut bandwagons. I think the students, and therefore

time, are on my side in this fight against debasement of categories.
In another departure from current practice I have extended the
principle of priority to orders and other higher categories up to the class.
Because LINDLEY first consistently applied the ending "-ales" to generic
roots, I have used his "Nixus Plantarum" (1833) as the starting work
for ordinal names. Family names are in accord with the "International
Code of Botanical Nomenclature" (STAFLEU 1972) though the nine
exceptional names are here disregarded as obsolete. In these matters I
think I am merely anticipating ultimate decisions for the "International
Code".
The divergences, however, of my system from other recent systems
that probably interest the reader most are those rather startling realign-
ments that are new or at least different from standard treatments. I
shall try to explain my reasons for accepting some of these more icono-
clastic alignments, selecting primarily those groupings that I have not
discussed at some length elsewhere (THORNE 1973, 1974 a, 1974 b,
1974 c, 1975, 1976).
In a long paper (THORNE 1976: Fig. 2) I have attempted a phylogenetic
shrub that presents visually but somewhat imperfectly how I relate the
major superorders, orders, and suborders to one another in the Angiospermae.
Placed toward the center are the superorders that I think retain the most
primitive original features of the protoangiosperms, as the Annonillorae,
Hamamelidillorae, and Rosillorae. Radiating from the center are the other
taxa that have achieved higher evolutionary plateaux in their characteristics,
and farthest from the center are the taxa that deviate most from their
presumed protoangiospermolls ancestry. I do accept both the class and the
two subclasses as monophyletic taxa. In reference to this phyletic shrub I
shall here work outward from t,he center in discussing those realignments
that may seem most novel to the reader.

Theiflorae
My treatment of the Annoniflorae (THORNE 1974) has heen available
for some time. A similar treatment of the Theiflorae, a related but
somewhat more specialized superorder, is not yet completed. I shall
discuss, therefore, some of the members of the group that are not gen-
erally placed in or near the Theales by other taxonomists, particularly
the Aquifoliaceae, Sarraceniaceae, Nepenthaceae, and Plumbaginaceae.
Aquifoliaceae. A prime target for phyletic demolition
is the extraordinary melange of largely unrelated families with small
haplostemonous flowers, witl\ or without glandular disks, that many
taxonomists continue to recognize as the Celastrales. The Aquifoliaceae
seemed to me especially out of place near the Celastraceae. The unspe-
cialized stem anatomy, theaceous appearance of the foliage, and gener-
302 R. F. THORNE:

ally circumtropical distribution of !lex L. early suggested the Theaceae

as probable relatives. To students of tropical botany this should not be
a startling suggestion. Sterile specimens of !lex are commonly confused
with sterile specimens of Eurya THUNB., Adinandra JACK., Oleyera
THUNB., Ternstroemia MUTIS ex L. f., ,and other theaceous genera.
Indeed, !lex species approach members of the Theaceae so closely in
stem, leaf, and flower characteristics, that I regard them as haplostem-
onous, functionally dioecious, small-flowered cousins of the camellias.
Among the features of the more specialized members of the Theaceae
shared by species of !lex are: basically tropical distribution; ancient
fossil record; similar growth habit and foliage; sexuality bisexual to
polygamous or functionally dioecious; sepals and petals imbricate;
stamens reduced to 5 or 4 and often attached basally to the separate or
basally connate petals; syncarpous ovary with mostly fewer than 1010-
cules; ovules 2 or mostly 1 on each of the axile placentae; single style
with lobed or toothed stigma; and fruit a baccate drupe. In xylem
anatomy the Aquiloliaceae are about as primitive as the Theoideae and
Ternstroemioideae but considerably less specialized than members of the
other theaceous subfamilies.
Presumably the close relationship between the two families, which
must have stemmed from common Cretaceous ancestry, has been ob-
scured for most temperate taxonomists by their incomplete knowledge
of the Theaceae, usually restricted to the commonly cultivated species
with showy, many-stamened, bisexual flowers and capsular fruits.
Actually definitive differences between members of the two families
are relatively few, most evident in ovule and seed characteristics. The
unitegmic ovules of the A quiloliaceae , for example, develop into
rather unspecialized seeds with a rudimentary embryo; whereas, the
bitegmic ovules of the Theaceae develop into rather specialized seeds
with well developed embryo and little or no endosperm. Examined
species of !lex lack the sclerenchymatous idioblasts so characteristic
of theaceous species though species of both families have mucilaginous
epidermal cells. Most hollies have leaves supplied with minute, mostly
caducous stipules, thus differing from the exstipulate leaves of the
Theaceae.Most Theaceae have bisexual flowers; most !lex species are
functionally dioecious or polygamo-dioecious. Most Theaceae have
numerous stamens, fascicled or basally monadelphous; most hollies have
stamens equal in number to the sepals and separate from one another.
Hollies have ovules reduced to 1, less commonly 2, in each carpel;
theaceous species mostly have more than 2 in each carpel. The two
families thus are now quite distinct from each other but surely closely
related within the same suborder. BAAS (1975), however, interprets the
relationships of the Aquiloliaceae quite differently.
Some Realignments in the Angwspermae 303

Sarraceniaceae. In view of DEBuHR's recent paper (1975)

on' the phylogenetic relationships of the Sarraceniaceae, no de-
tailed discussion is needed here. I think he haR rather definitively
placed the Sarraceniaceae in its own suborder Sarraceniineae adjacent
to the Theineae of the Theales. Although the evidence he presented for
this relationship is rather overwhelming, I might add to it that there
seems a well-marked tendency within the Theales toward. leaf dimorphism
and ascidiform structures, as displayed also in the saccate floral bracts
of the Marcgraviaceae and well-known pitchers on many leaves of
Nepenthes L., discussed below. It may be worth mentioning here too
that the Marcgraviaceae and Theaceae, especially members of the Bon-
netioideae, Tetrameristoideae, and Kielmeyeroideae (MAGUIRE 1972,
MAGUIRE et al. 1972) are well represented on the Guayana Highlands
with Heliamphora BENTH. of the Sarraceniaceae. It seems quite possible
that this ancient area is the original cradle of the Sarraceniaceae and
many, if not most, of the fam"iIies of the closely related Theineae.

Nepenthaceae. Although I do not regard the Sarraceniaceae

and the Nepenthaceae, the paleotropical analogs of the New World
pitcher-plants, as families with direct close relationship, I do now include
both in the Theales (THORNE 1976), the former as the Sarraceniineae
adjacent to the Theineae (now including the Olethrineae) and the latter
as the Nepenthineae between the Scytopetalineae and Hypericineae.
Nepenthes species have some striking stimilarities to the Dioncophyl-
laceae, three monospecific genera of West African lianas, and somewhat
fewer resemblances to the Ancistrocladaceae and Hypericaceae (including
the often segregated Olusiaceae).
Among the likenesses between Nepenthes and the Dioncophyllaceae
(AIRY SHAW 1952, METCALFE 1952, SCHMID 1964) are: the leaf-climbing
lianous habit; alternate, exstipulate, entire, petiolate, often dimorphic
leaves, some at least with midrib excurrent into two recurved hooks
(Dioncophyllaceae) or prolonged into an ascidium-bearing "tendril"
(Nepenthes); abundant development of glands; stratified phloem;
peripheral rings of vascular bundles em bedded in cortical fibers; fiber-
tracheids; numerous ovules maturing into endospermous (but very
different) seeds borne in loculicidally dehiscent capsules. Although some
of these similarities may well represent convergences, there appear to be
enough basic resemblances to indicate distant common origin despite
the quite divergent flowers, pollen, placentation, and seeds.
Ancistrocladus WAJ,L.. sole genus of the Ancistrotadaceae, shows
fewer similarities to Nepenthes but more with the Dim/'cophyllaceae
(VAN STEENIS 1948, METCALl<'E 1952, ERD;J;MAN 1958, KENO 1967,
1970, GOTTWALD & PARAMESWARAN 1968). As1de from the overlapping
304 R. F. THORNE:

range'! of the two groups in tropical West Africa, they retain in common:
woody, climbing habit by non-homologous hooks; largely exstipulate,
alternate, simple, petiola,te, evergreen leaves with actinocytic stomata
and sunken peltate glands; fiber-tracheids and other similar stem
anatomical characteristics, as vascular strands embedded in cortical
fibers; contorted petals; stamens mostly 10 in 2 whorls; similar pollen
grains; and erect, turbinate embryo with thin leafy cotyledons embedded
in starchy endosperm. The differences in calyx, gynoecium, placentation,
and fruit indicate rather distant common ancestry, however, within the
Scytopetalineae of the Thea,les. The suggested relationship of Ancistro-
Cladu8 to the Hugonieae of the Linaceae, which I had previously accepted
(THORNE 1968) seems less likely and probably the result of convergence.

Plumbaginaceae. A review of the similarities between the

Plumbaginaceae and Primulaceae has convinced me that HUTCHINSON'S
treatment (1973) of the two families in one order, Primu:lales, most nearly
fits the phylogenetic facts of any recent treatments. Hence, I have
reduced the Plumbaginales to subordinal rank, Plumbaginineae, in my
revised synopsis (THORNE 1976). Among the more convincing charac-
teristics possessed in common by the two families are: similar habit,
mostly perennial herbs or subshrubs with leaves mostly exstipulate,
simple, and often rosulate; flowers actinomorphic, bisexual, penta-
merous, bracteate, and commonly heterostylous; calyx gamosepalous
and commonly persistent; corolla sympetalous, though rarely nearly or
quite polypetalous, imbricate or convolute in vernation; stamens in a
single whorl, mostly epipetalous and oppositipetalous, with the outer,
oppositisepalous, whorl obsolete or reduced to staminodia; gynoecium
largely superior, unilocular with free-central or free-basal placentation
and ovules many to five (Primulaceae) or only one (Plumbaginaceae);
ovules bitegmic, anatropous or semianatropous developing into seeds
with straight embryo in copious endosperm; fruits various though
sometimes a circumscissily dehiscent capsule (pyxidium) in both families;
and similar fla vonoids (HARBORNE 1967) and amyloid hemicelluloses
in the seeds (BENTVELZEN 1962).
The differences between the two families are sufficent for the recogni-
tion of two suborders. The Primulineae, with the closely related Myrsi-
naceae, Theophrastaceae, and Primulaceae, are distinguished by their
single style and one or usually more, mostly tenuinucellate ovules on a
free-central or basal placenta. The monofamilial Plumbaginineae differ
by their membranous calyx; wholly or partly distinct styles; solitary,
crassinucellate ovule on a basal placenta; large embryo; embryological
features; and distinctive pollen grains.
The recent attempts to link the Primulaceae, Plumbaginaceae, or
SomE' Realignments in the Angiospermae 305

both to the Polygonaceae and Caryophyllaceae of the Chenopodiflorae

over-emphasize the convergences due to common possession of a unilo-
cular ovary with free-central to basal placentation. The anatropous
rather than campylotropous ovules, endosperm rather than perisperm,
central rather than peripheral embryo, and non-centrospermous pollen
grains, seeds, and chemistry of the Primulales negate any close common
ancestry between the Primulales and Chenopodiales. The position of
the Polygonales anent, both the Primulales and Chenopodiales needs
further, thorough study.
The relatively close common ancestry of the Primulales with the
Ebenales, Ericales, and more primitive Theales within the superorder
Theiflorae would seem to be too obvious to dwell upon here.

Malviflorae
There is much difference of opinion among phylogenists as to whether
the Sympetalae, as traditionally treated by the Englerians, are a mono-
phyletic group of related orders arising from immediate common ances-
tors or are a polyphyletic group of non- or only distantly-related orders
that have converged in achieving an evolutionary plateau represented
by a successful syndrome of evolutionary characteristics. I believe the
8ympetalae, like the Polypetalae, Apetalae, Amentiferae, Herbaceae, and
similar artificial groups, are polyphyletic. This is evident already from
my treatment in the Theiflorae of the Ericales, Ebenales, and Primulales.
Another group that I think has little to do with the Gentianiflorae,
Lamiiflorae, and Asteriflorae, all indeed more or less related to one another
through the more primitive Rosiflorae, is the order Solanales, which I
place in the Malviflorae with the Malvales, Urticales, Rhamnales, Euphor-
biales, and Campanulales. Although I have discussed this group briefly
elsewhere (THORNE 1974 a), I should like here to consider the North
American Fouquieriineae of the Solanales.
Fouquieriaceae. The largely Mexican Fouquieriineae, with the
sole family Fouquieriaceae and sole genus Fouquieria KUNTH,
have been ably monographed by HENRICKSON (1967, 1969, 1972, 1973).
Because of its distinctive features and rather isolated phyletic position
I have treated the family (THORNE 1968, 1976) in its own suborder,
equivalent to the Solanineae and Polemonineae, in the Solanales. Despite
its specialized habit, it retains from the common protosolanalean
ancestors rather more primitive features than members of the other two
suborders, among them: relatively unspecialized xylem characteiistics;
corolla lobes imbricate in bud; separate stamens, non-epipetalous (but
filaments connected slightly to petal bases and falling with spent corolla),
basically ten in one whorl (bllt up to 23 in some species); placentation
Plant :;yst. EYol., :;uppl. 1 20
306 R. F. THORNE:

basally axile but imperfectly so above with three septiform placentae

nearly meeting but not united by their margins; and bitegmic ovules.
The closeness of relationship of the Fouquieriineae to the three
families of the other two suborders varies from feature to feature, as
should be expected from three distinct evolutionary lines diverging from
common ancestors. In general, however, the Fouquieriaceae seem to
retain more characteristics in common with the more primitive genera,
like Cantua Juss. ex LAM., Cobaea CAV., and Huthia BRAND., of the
Polemoniaceae (GRANT 1959) than with members of the Solanaceae or
Convolvulaceae. Among these more prominent features are: the shrubby
or small tree habit; mostly alternate, simple, exstipulate, deciduous
leaves with anomocytic stomata; showy, bisexual, actinomorphic,
bracteose flowers with 5 imbricate, persistent sepals or calyx-lobes and
5-lobed, tubular corolla; 3-carpellate and basically 3-loculate gynoecium
with 3- branched, single style; fruit a 3-valved, loculicidal capsule; and
seeds flat, broadly winged, with firm-fleshy endosperm surrounding a
well-developed embryo with spatulate or elliptic cotyledons and a
small radicle. An undescribed shrubby genus of Polemoniaceae from
Baja California under study by REID MORAN (personal communication)
has somewhat the habit of Fouquieria with substantial woody stem,
primary leaves persisting and becoming spinescent, and fascicled secon-
dary leaves arising in the axils of the spines.
My colleague R. SCOGIN has made a thorough study of the compara-
tive biochemistry of the Fouquieriaceae as compared with such putative
relatives as the Polemoniaceae, Solanaceae, Convolmllaceae, Tamaricaceae,
Frankeniaceae, and Ericaceae. His results (personal communication)
show a striking correlation chemically between Fouquieria and the first
three families but little correlation with members of the Tamaricales or
Ericale8. Among the chemical characteristics that he reports linking
Fouquieria closely with other families in the Solanale8 are: presence of
coumarins, as scopoletin and scopolin, a pelargonidin glucoside, an
acylated anthocyanin (hyacinthin), and triterpene and steroidal saponins;
and absence of proanthocyanidins, negatively charged flavonoids, the
cyclical polyalcohol pinitol, and tannins. The primary chemical dif-
ferences between Fouquieria and other members of the Solanales,
according to SCOGIN, are the presence of ellagitannins and iridoids in
and absence of alkaloids and methyl ethers of flavonols from Fouquieria.
DAHLGREN et al. (1976), emphasizing the iridoids in Fouquieria, interpret
the chemistry of the genus quite differently. It is noteworthy that all
families of the Solanales are heavily developed in Mexico or the south-
western United States or both, especially Fouquieria, the arborescent
species of Ipomoea L., and many Polemoniaceae and Solanaceae. Ericaceae
usually are not found in association with Fouquieria.
Some Realignments in the Angiospermae 307

I am somewhat puzzled by the supposed relationship claimed by some

phylogenists between Fouquieria and the Tamaricales because I can see few
resemblances between the Fouqu1".eriaceae and the Tamaricaceae And Fran-
keniaceae. The Tamaricaceae are indigenous only to the Old World and the
Frankeniaceae are rare within the New World range of Fouquieria. I presume
those botanists seeing mutual relationships there must have been influenced
by the incompletely axile placentation of the upper portion of the fouquieri-
aceous ovary, often described as parietal, a very widespread placental
condition. SCOGIN'S study, while dismissing the liklihood of relationship of
Fouquieria with the Tamaricales, does indicate considerable relationship
between the Tamaricaceae and Frankeniaceae. CAMPBELL & THOMSON
(1976 and personal communication) have noted the remarkably similar
structure of the salt glands of Frankenia L. and Tamarix L., whereas
WALIA & KAPIL (1965) found the embryology of Frankenia to be quite
different from that of members of the Tamaricaceae.

Rutiflorae
Although I have discussed the superorder Rutiflorae or some of its
component taxa elsewhere (THORNE 1974 a, 1974 b, 1976), my recent
transfer of the Gyrostemonaceae from the Ohenopodiflorae to the Ruti-
florae warrants some discussion here. The Rutiflorae are a distinct
evolutionary line parallel to but not closely related to the Theiflorae
and perhaps with closer common ancestry with the Annoniflorae than
that between the Theiflorae and Annoniflorae. This is evidenced es-
pecially by the striking chemical similarities between the rather primi-
tive Rutaceae and the Annonales and Berberidales. The essentially tropi-
cal, large order Rutales, including the Sapindineae, is especially note-
worthy because of its inclusion of the anemophilous Juglandineae and
probable close common origin with the also anemophilous Myricales and
Leitneriales. The very strong tendency within this superorder toward
anemophily has largely been overlooked by many phylogenists, who
prefer to trace most of the old amentiferous families to the Hmnamelidi-
florae, which do indeed also display strong anemophilous tendencies.
Gyrostemonaceae. The recent purification of the Oheno-
podiales (Oentrosper1nae) by the removal of extraneous unrelated taxa
formerly forced unnaturally into dubious alliance with the centrosperms
has led to some unexpected alignments. Theligonum L. has found a
good home in the Rubiaceae (WUNDERLICH 1971); the Primulaceae and
Plumbaginaceae are discussed above; Batis L. is discussed below; and
Simmondsia NUTT. still defies all efforts to place it properly. A small
Australian family, Gyrostemonaceae, earlier removed from the Phytolac-
caceae, has lately been excluded, quite correctly, from the Ohenopodiales
(GOLDBLATT et al. 1976). Since we have had considerable field experience
in Australia with this family and have rather good representation of the
five genera in our herbarium, L. DEBUHR, recently returned from Western
20·
308 R. F. THORNE:

Australia, and I have studied the group to see if we could place it with
any of the other Australian families. We found that the Gyrostemona,ceae
possess many features in common with the largely Australian Dodo-
naeeae of the Sapindaceae and the West Australian Stylobasium DESF.
(PRANCE 1965) and some features with the West Australian Emblingia
F. MUELL. and the widely disjunct Batis L. We suspect that this group
of taxa, mostly anemophilous shrubs or subshrubs of the more arid
areas of Australia, form a related complex within or near the Sapindaceae.
I have elsewhere (THORNE 1976) treated the Gyrostemonaceae as a family
immediately adjacent to the Sapindaceae, including the Dodonaeoideae
and Stylobasioideae, and suggested that LEINS (ERDTMAN et al. 1969)
was probably correct in assigning Emblingia to this grouping. Once
again anemophily in the Rutiflorae seems to be far more prevalent than
previously realized.
Most of the highly distinctive combination of features that char-
acterize·the Gyrostemonaceae can be found among the Sapindaceae. In
addition to the overlapping Australian ranges are: the primarily shrubby
or subshrubby habit and stems with normal growth and moderately
specialized stem anatomy; alternate, simple, entire, spatulate to com-
monly linear leaves with very small or no stipules and anomocytic
stomata; a petalous, unisexual flowers (species monoecious or dioecious)
with connate calyx 8-4-2-lobed or entire; stamens 6 to numerous on the
rim or top of a discoid or convex receptacle, producing mostly isopolar,
tricolpate, prolate-spheroidal to prolate pollen grains of medium size
(PRIJANTO 1970 a); gynoecium of 1 or 2 to numerous (not more than
6 in Sapindaceae), connate carpels with more or less free styles; locules
of ovary each with 1 axile, campylotropous, bitegmic ovule; fruit dry, an
indehiscent nutlet or a schizocarp in which each membranous carpel
dehisces and separates from the others and the discoid axis; and a curved
seed with small, basal arillode. This distinctive combination of features,
not duplicated in anyone group of the Sapindaceae, plus the often nu-
merous carpels, rather peculiar central, disk-like axis, and central,
fleshy or oily endosperm does clearly separate the Gyrostemonaceae from
the apparently closely related but exendospermous Sapindaceae.

Bataceae. This monogeneric family consists of two species, the

dioecious American species Batis'maritima L. and the Australian-Papuan
monoecious species B. argillicola VAN ROYEN. Because of its usual as-
sociation in coastal habitats with members of the Chenopodiaceae and
because of its obvious succulence, Batis was long assumed to be
centrospermous despite the absence of the usual centrospermous char-
acteristics. The strikingly similar pollen grains of Batis and the Gyroste-
monaceae have been noted hy several palynologists (ERDTMAN 1952,
Some Realignments in the Angiospermae 309

PRIJANTO 1970 a, 1970 b, NOWICKE in GOLDBLATT et al. 1976). Because

of these palynological similarities and the reported presence of gluco-
sinolates, DAHLGREN (1975) and GOLDBLATT et al. (1976) have placed
the Bataceae and Gyrostemonaceae in the Capparales. Although I can see
no more natural relationship of these families to the Capparales than to
the Chenopodiales, there does seem to be great merit in the suggested
relationship of the two families to each other.
Batis species resemble plants of the Gyrostemonaceae in: their uni-
sexuality and presumed anemophily; shrubby habit with normal,
relatively specialized stem anatomy; linear leaves with minute stipules;
apetalous, unisexual flowers, some (B. argillicola) borne singly in the
leafaxils ; isopolar, tricolporoidate, subprolate to prolate, pertectate,
non-baculate pollen grains of medium size (PRIJANTO 1970 b); pistil
with one ovule in each locule; fruit a "septicidal berry" (fleshy capsule 1)
in B. argillicola. It is significant that the Australiall-Papuan species
(VAN ROYEN 1956, 1957) is considerably less specialized than the American
species in its less tidal habitat, less sprawling habit of growth, monoecism,
mostly solitary flowers, and free, septicidally-dehiscent fruits. The
ultimate antecedents of Batis maritima, therefore, like those of the often
associated Dodonaea viscosa L. (Sapindaceae) , Scaevola plumieri (L.)
VAHL (Goodeniaceae), and Cassytha filiformis L. (Lauraceae) of maritime
tropical America, may well be Australasian.
Batis is separated from the Gyrostemonaceae by a number of distinc-
tive features, such as: leaves with paracytic stomata; membranous
spathella (or calyx) and spatulate, valvate tepals (or staminodia); 4 or
5 alternitepalous stamens; pollen grains with compound apertures with-
out opercules; bicarpellate but tetralocular pistil; and anatropous, basal
ovules ripening into exalbuminous seeds with a large, straight embryo
with spatulate cotyledons. I now prefer tentatively to treat Batis as a
separate family Bataceae and suborder Batineae following the Gyroste-
monaceae and Sapindaceae in the Sapindineae of the Rutales. The
absence of glucosinolates from most of the Sapindineae might be bother-
some to some botanists, but myrosin cells, often associated with these
sulphur compounds, have been reported in the sapindaceous Bret-
schneidera HEMSL. (SCHOLZ 1964). If one regards the differences between
the Bataceae and Sapindineae as of the same approximate magnitude
as those differences that separate the Juglandineae and Myricales or the
Anacardiaceae and Leitneriales, the Bataceae then should be treated as
the Batales following the Rutales with the M yricales and Leitneriales in
the Rutiflorae. Certainly, this whole Australasian anemophilous complex
(Oyrostemonaceae, Stylobasium, Emblingia, and Batis) should be exam-
ined carefully by experts to see if the apparept relationship with the
Sapindaceae is truly the result of common anceMry.
R. F. THORNE:

Rosiflorae
The Rosiflorae are, like the Annoniflorae and Hamamelidiflorae, a
phyletically important superorder which retains a large number of
primitive features. It is central to the understanding of the relationships
and origins of many other superorders of flowering plants. The Rosiflorae
appear to be more closely related to the Hamamelidiflorae than to the
Annoniflorae, but surely all three had at least distant common ancestors
in early Cretaceous time. Space does not permit extended discussion of
the Rosiflorae, but I wish to consider here one representative family, the
Orossosomataceae, which I include in the Rosineae of the Rosales within
this superorder, along with the Pittosporales and Proteales.
Crossomataceae. This small family of southwestern North
America, containing probably only three species in two genera, has
been variously placed by different botanists, usually in the Rosales near
the Rosaceae, Ohrysobalanaceae, Oonnaraceae, and Fabaceae or in the
Dilleniales (Dilliniineae of my Theales) with the Dilleniaceae and Paeonia-
ceae. To resolve the question of relationship, TATSUNO (1976) has made a
thorough study of the Orossosomataceae and its reputed relatives,
emphasizing biochemistry. Her research did reach two important
conclusions: a) that the new Arizona genus Apacheria MASON fits well
chemically with Orossosoma NUTT.; and b) that stamen initiation and
maturation in Orossosoma is centripetal, not centrifugal as reported
previously. My conclusion from TATSUNO'S evidence is that the Oros-
sosomataceae must be retained in "lhe Rosineae since they show over-
whelming morphological and other similarities to the other families of
that suborder. Among the more striking, non-chemical similarities found
in the Orossosomataceae and various members of the Rosaceae, Fabaceae,
Oonnaraceae, and Ohrysobalanaceae are: woody habit, from small tree to
low shrub, with relatively specialized wood (such as vessel elements with
simple perforation plates and alternate intervascular pitting); leaves
alternate, simple, and exstipulate with anomocytic stomata; flowers
bisexual, perigynous, and basically pentamerous; androecium with
many stamens (15-50), centripetally initiated; gynoecium of 5-3 distinct,
stipitate carpels; ovules campylotropous, bitegmic, and crassinucellate
with a multicellular archesporium; and fruits follicular producing hlack,
reniform seeds with a fimbriate aril and a large, curved embryo in
abundant endosperm. The relatively specialized stem anatomy, absence
of fasciculate grouping of stamen traces, and the centripetal development
of the stamens, among many other features, surely rule out close rela-
tionship with members of the Dillenineae of the Theales. Although one
must view popular names with considerable skepticism, the only bona
fide common name I have heard for a member of the Orossosomataceae
Some Realignments in the Angiospermae 311

is "wild-apple" for Crossosoma cali/ornicum NUTT. on Santa Catalina

Island, where it is a frequent shrub and indigenous.

Buxaceae. I have long been fascinated by t.he disjunct, re-

lictual, world distribution and primitive morphology of the Buxaceae
and much puzzled by its probable relationships. It surely is an ancient
family without close relatives. Overly impressed by their reported cro-
tonoid pollen grains, I first placed the Buxaceae and the Thymelaeaceae
(THORNE 1968) in the Euphorbiales. I have more recently (THORNE
1976) removed both families from the Euphorbiales, the Thymelaeaceae
to the Myrtales and the Buxaceae to the Pittosporales. The former is still
under study, but I wish to discuss the Buxaceae here.
The extremely primitive xylem anatomy of the Buxaceae combined
with the somewhat specialized features of the flower is reminiscent of
the rather similar combination in some other pittosporalean refugees from
the Euphorbiaceae, as Daphniphyllum BLUME and Balanops BAILL. In
comparing the Buxaceae, excluding the highly aberrant Simmondsia
NUTT., with more primitive members of my Pittosporales, I found the
following features to be retained in common: woody habit, small trees
or shrubs, to perennial herbs; wood primitive with very small, medium
to very long vessel elements, exclusively solitary and with 15-20-bar-
red scalariform perforation plates, and very small, scalariform to op-
posite or transitional intervascular pitting; xylem parenchyma diffuse
and apotracheal; rays markedly heterocellular, to 2 or 4 cells wide with
numerous uniseriates; fiber- tracheids with large to small bordered pits
and very short to moderately long; leaves evergreen, exstipulate, simple,
mostly glabrous and entire; flowers mostly unisexual (species then
usually monoecious), apetalous and sometimes naked, in tight bracteate
racemes or spikes; sepals imbricate, usually 4, or absent; stamens 6-4
(rarely numerou!3) with large anther, dehiscing longitudinally; gynoecium
syncarpous (usually abortive in or absent from male flowers) with 3 or 2
carpels and locules each with 2 anatropous ovules, bitegmic and cras-
sinucellar, pendulous from the apex of each locule; fruit a loculicidally
dehiscent capsule or a drupe; and seeds with straight, linear embryo in
copious fleshy endosperm. Although tempted to combine the Buxineae
and Daphniphyllineae into one suborder, I have tentatively retained the
two as distinct because of the rather notable buxaceous features such
as stomata surrounded by a rosette of subsidiary cells, predominant
monoecism, axile placentation, and relatively well-developed embryo
in the seed.

Simmondsia chinensis (LINK) SCHNEID., the popular jojoba of south-

western North America, has recently been studied by BROWN (1976), to try
to ascertain biochemically and otherwise what its relationships to the
312 R. F. THORNE:

Buxaceae or other putative relatives might be. Alt.hough het· biochemical

st.udies strongly support the removal of Simmond8ia from the Buxaceae,
they did not place the genus definitively near any other putative relatives.
Among the more striking features in which Simmond8ia differs from t,he
Buxaceae are: the relatively specialized vessel elements with simple perfora-
tion plates and alternate intervascular pitting; tracheids with large, distinctly
bordered pits; anomalous growth from successive cambia; anomocytic
stomata on both surfaces of the leaf; petiole with arcshaped vascular strand;
anemophily, with no trace of nectaries nor odor; dioecism; tricolpate, oblate-
spheroidal pollen grains; deciduous styles; absence of obturator and other
embryological differences; large seeds with investing embryo, cotyledons
with abundant wax, and little or no endosperm; and numerous biochemical
differences as presence of leucocyanidin, leucodelphinidin, and tannins, and
la.ck of alkalOids, and sinapic and ferulic acids. I have accordingly placed
Simmond8ia in my list of taxa incertae sedis.

Balanopaceae. Balanops has commonly been placed near

the Fagales presumably due to the superficial resemblance of the drupe
subtended by an involucre to the fagaceous acorn in an acorn-cup.
Some time ago I removed Balarwps from near the Fagales to my taxa
incertae sedis (THORNE 1974 a). Later in examining our herbarium
material in an effort to place the genus, I was impressed by its resem-
blance in clustered, pseudoverticillate, leathery leaves to species of
Daphniphyllum (HUANG 1965). In drawing up a chart of likenesses
between the two genera, I was astonished by the remarkable "fit"
between them: Dioecious trees or shrubs; primitive xylem anatomy with
vessel elements solitary, with many-barred (more than 20) scalariform
perforation plates and rare scalariform intervascular pitting; narrow
rays with numerous uniseriates composed of upright cells; cork sub-
epidermal; leaves simple, exstipulate, alternate to subverticillate, mostly
entire, with stomata only on the lower surface; flowers unisexual,
apetalous or naked; staminate flowers with 6-3 or no sepals, stamens
14-2, usually 5 or 6, on short filaments, sometimes with apiculate COll-
nectives and sometimes with rudimentary ovary; pollen grains commonly
tricolpate, suboblate, the colpi with granular membranes; female
flowers with 6-3 or no sepals, staminodia small or absent, ovary imper-
fectly 4-2-loculate (placentation thus parietal) with 2 anatropous ovules
in each locule; and fruit a drupe with 3-1 pyrenes. Surely the two
genera must be treated in one order or preferably, I think, in one suborder,
Daphniphyllineae of the Pittosporales, with rather close common origin
with the Buxineae discussed above and less close common origin with the
Pittosporineae. With the Pittosporaceae there is close resemblance often
in habit, foliage, and parietal type of placentation. I do think Balanops
and Daphniphyllum deserve their traditional recognition as monogeneric
families because of notable differences in absence of sepals and staminodia
from Balanops, deeply bipartite styles in Balanops, ovules (unitegmic
Some Realignment,s in the Angiospermae 313

and erect in Balanops but bitegmic and pendulous in Daphniphyllum),

and seeds (rudimentary apical embryo in thick, fleshy endosperm in
Daphniphyllum but large, straight embryo in sparse endosperm in
Balanops). These are differences unexpected in a single family.
Daphniphyllum overlaps in range with members of the Buxaceae and
Pittosporaceae and Balanops with the Pittosporaceae, but the Asiatic-
Papuan Daphniphyllum is apparently separated from the Australasian-
Fijian Balanops by at least the Torres Straight. One must admire the
early phyletic perceptivity of BENTHAM & HOOKER, who apparently
considered the Buxaceae and Daphniphyllum (their Family 153) and
Balanops (Family 154) as related taxa (also treated by METCALFE &
CHALK (1950) as adjacent families 244-246).

Asteriflorae
Because of our dissatisfaction with the failure of synantherological
specialists to come up with a realistic classification of the Asteraceae, my
colleague S. CARLQUIST and I have proceeded to publish (CARLQUIST
1976, THORNE 1976) a revised classification of the family with a redistri-
bution of the tribes, placing the Mutisieae, Vernonieae, Eupatorieae,
Cardueae, and Arctoteae into the Cichorioideae with the Cichorieae. The
remaining tribes, clustered about the Heliantheae, are retained in the
Asteroideae. I shall leave the justification for this reclassification to
CARLQUIST (1976). In a recent paper WAGENITZ (1976) has independently
divided the composite tribes into two groups closely resembling ours,
though he places the Eupatorieae with the Asteroideae.
1 wish here to give some of my reasons for insisting that the re-
lationships of the Asteraceae are not be sougbt in the Campanulaceae
nor Rubiaceae, as suggested by some syngenesiologists, but in the
Cornale.~ (see also WAGENITZ 1976), Dipsacales, and Lamiales, and
ultimately for all these in the rosalean Saxijragineae. With the Cornales,
Dipsacales, and Lamiales the Asterales share: primitively woody habit;
exstipulate leaves; largely bisexual, tetracyclic flowers; locules mostly
reduced to two or one with each commonly bearing a single, anatropous,
unitegmic ovule; and marked tendencies toward the herbaceous habit,
much dissected leaves, flowers crowded into bracteate capitula, floral
dimorphy and zygomorphy, and maturation of indehiscent, dry, one-
seeded fruits. With the Cornales and Dipsacales the Asterales share a
floral tube adnate to the inferior ovary, calyx lobes greatly reduced or
modified, and strong tendencies, a'l in some Saxitragineae, toward the
neutralization, enlargement, or other mo~ification of lateral flowers of
the crowded inflorescence into a pseudanthium. The Asteraceae have
advanced considerably over the Cornales in their sympetalom. corolla,
314 R. F. THORNE:

greater dimorphy and zygomorphy of the flowers, syngenesious stamens,

and one-seeded cypselar fruits. They differ from the Lamiales in their
consistent epigyny, cypselas, stamen isomery, and less consistent floral
zygomorphy. The basal attachment of the single ovule in the ovary
readily di'ltinguishes the Asteraceae from members of the Cornales and
Dipsacales.
Liliiflorae vs. Alismatiflorae
By selecting the generally accepted primitive features retained
among the least specialized members of all the extant monocot super-
orders one can derive a probable description of the common protomonocot
ancestor of these groups: terrestrial, rhizomatous perennial with cambium
lacking from stem; vessel elements, tracheid-like with scalariform end-
plates, evolved from tracheids only in the roots; leaves alternate, simple,
entire, broad-Ianceolate, parallel-veined, with broad, sheathing base;
flowers bisexual, actinomorphic, hypogynous, spiro cyclic or pentacyclic,
and entomophiloll'l with parts separate and trimerous; perianth with
two series of little differen~iated, petaloid tepals; stamens six in two
whorls, producing anasulcate pollen grains; nectaries probably lacking
and flowers pollinated by beetles or other insects with chewing mouth-
parts; three separate carpels with unsealed or lightly sealed stigmatic
margins bearing bitegmic, crassinucellate, anatropous ovules; and
fruits a whorl of three follicles maturing seeds with a rudimentary
embryo in abundant non-starchy endosperm. Such a protomonocot
would presumably be growing under warm, moist, shaded, forest condi-
tions, possibly in riparian swamps but surely not in open water. This
description fits more closely some extant forest liliads, aroids, or arecads,
rather than Sagittaria L. (Alismataceae) , Aponogeton L. f. (Aponogeto-
naceae), or Potamogeton L. (Potamogetonaceae).
Some members of the Alismati/lorae, as Butomus L. (Butomaceae),
Aponogeton, Stratiotes L. (Hydrocharitaceae) , and the All:smataceae,
retain a val'iable number of primitive features from their common
ancestors. Among these retained characteristic'l are bisexual, hypogynous,
trimerow-l flowers with separate parts; monosukate pollen grains;
carpels with many anatropous ovules on submarginal placentae; and
follicular fruit (Butomus). Yet these same plants combine these primi-
tive with rather many specialized features: palustrine or aquatic habitats
and habits of growth; rf'latively specialized vessels, often with porous
end-plates, in the roots; separate tepals well-differentiated into three
outer greenish sepals and three inner white or colored petals; increased
number of stamens and carpel'l, often mono-ovulate, on an elongated
receptacle; pollen grains trinucleate at release; gynoecium paracarpous,
epigynous, and with laminar placentation; ovules orthotropous or
Some Realignments in the Angio8permae 315

campylotropous; fruit usually one-seeded achenes or many-seeded

berries; and mature seeds with enlarged, curved embryo and no endo-
sperm. These are hardly characteristics of a "primitive" monocot.
Their more completely aquatic relatives, like Halophila THOU. and
Thalassia BANKS (Hydrocharitaceae) , Halodule ENDL. and Zanichellia
L. (Zanichelliaceae), Ruppia L. (Potamogetonaceae), and Na.ias L. (Na.ia-
daceae), are among the most specialized of all monocots, often devoid of
vascular tissue, cuticle, and stomata; with flowers naked, unisexual, and
reduced sometimes to a single stamen or pseudomonomerous pistil; and
pollen grains nonaperturate and convervoid and pollination and dispersal
hydrophilous. The mycophytic Triuridales, often considered related
to the Ali8matale8, may have closer common ancestry with the Lilii/lorae
than with the Ali8mati/lorae.
Because of the apocarpy and other primitive features in the few
alismatiflorean taxa mentioned above, many botanists have regarded
this relatively specialized superorder as the most primitive group of
Monocotyledoneae. Based upon this assumption, the Alismati/lorae
have often been suggested as transitional between the monocots and the
apocarpous Nymphaeales or Berberidales, possibly because they often
share an aquatic or near-aquatic habitat. The similar modifications for
successful adaptation to aquatic niches (laminar placentation or increased
number of mono-ovulate carpels among them) and possibly the retention
of some primitive features (including vessels restricted to the roots and
apocarpy) are expected evolutionary convergences in these quite un-
related groups. To obviate any more attempts to use these aquatics as
links between the dicots and monocots, I have in my recent synopsis
(THORNE 1976) delayed consideration of the Ali8mati/lorae by placing in
the basal position in the Monocotyledoneae (Liliidae) the Lilii/lorae with
the single order Liliales and the three suborders Liliineae, Iridineae, and
Orchidineae. Some of the apocarpous liliads have many more primitive
features than even the least specialized helobiads listed above. Further-
more, the liliads appear to represent the "main-stream" of monocot
evolution; whereas, the alismatads (helobiads) are surely a palustrine-
aquatic offshoot from the basic, terrestrial, protomonocot stock. Placing
the helobiads first in the Monocotyledoneae has been as misleading as the
Englerian placement of the specialized anemophilous orders in lead-off
position for the Dicotyledoneae.

Commeliniflorae
In this context it probably is unnecessary to stress the commelinaleall
affinities of the Juncaceae, Cyperaceae, and Poaceae. In recent years,
however, I have several times heard knowledgable botanists, who
316 R. F. THORNE:

should have known better, refer to the lilialean ancestry of each of these
families. The most conservative members of the Commelinales do share
some of the characteristics listed above for the protomonocot, but
even the least specialized are somewhat more advanced in the di- to
hexapel'igenous stomata (FRYNS-CLAESSENS and VAN COTTHEM 1973);
biseriate, well-differentiated perianth; triloculate gynoecium with axile
placentation; mostly orthotropous ovules; and seeds with a mostly
basal or lateral embryo in starchy, often mealy, endosperm or perisperm.

Juncaceae and Cyperaceae.Some of the primitive members

of the Juncaceae (like Thurnia HOOK. f.) and Cyperaceae, the
two families constituting the Juncineae, like the primitive genera of the
closely related Bromeliineae, Pontederiaceae, Commeliniineae, and
Eriocaulineae, appear to be centered in tropical South America with
special concentrations on the Guayana Highlands and Brazilian Planalto,
usually most abundant in moist, open habitats. The Juncaceae and
Cyperaceae are closely linked by their chromosomes with diffuse centro-
meres; pollen grains dispersed as tetrads or pseudomonads; typical
commelinalean seeds with basal, lentiform or derived capitate embryo
in abundant starchy endosperm; and overall specialization for anemophily.

Poaceae. Like the Juncineae and heavily Australasian Flagellari-

ineae, the monofamilial Poineae are anemophilous. Which features are
part of the anemophilous syndrome and which are retained from proto-
commelinalean ancestors are not always easy to determine. Presumably
characteristics such as the growth\habit, stomatal apparatus, primitive
grass flower, pollen grains, seeds, and other embryological features,
which are markedly commelinalean, are retained characteristics little
tainted by the prevailing anemophily. The poaceous seed with marginal
embryo lateral to the abundant horny and mealy, starchy endosperm in
itself epitomizes the position of the Poaceae both as a member of the
Commelinales and of the independent suborder Poineae.
I regret that space does not permit the presentation of arguments
for the alignment of the other three hundred or so angiosperm families.
This small sampling, however, may give some indications of the kinds of
data and the philosophy upon which I have based my alignment of the
,mperorders and their constituent orders, suborders, families, and sub-
families.
A number of my associates at the Rancho Santa Ana Botanic Garden
have supplied data, specimens, other materials, and suggestions. I am es-
pecially grateful to RON SCOGIN, SHERWIN CARLQUIST, LARRY DEBUHR,
JAMES HENRICKSON, ALICIA TATSUNO, and SHANNON BROWN. Botanists at
other institutions, too numerous to mention most of them, havc over the
years generously supplied information, separates, specimens, slides, preserved
Some Realignment.s in t.he Angiospermae 317

materials, and guidance in the field. Particularly helpful in this project have
been REINO ALAVA, University of Turku; B. L. BURTT, Royal Botanic
Garden, Edinburgh; ROLF DAHLGREN, Botanical Museum, University of
Copenhagen; TH. ECKARDT, Berlin-Dahlem; ARTHUR GIBSON, University of
Arizona; B. P. M. HYLAND, Atherton, Australia; HSUAN KENG, University
of Singapore; JOHN PARHAM, Queensland State Herbarium, Brisban~;
REID MORAN, San Diego Museum of Natural History; G. R. PROCTOR,
Institute of Jamaica, Kingston; and RUDOLF SCHMID, University of Califor-
nia, Berkeley.

References
AIRY SHAW, H. K., 1952: On the Dioncophyllaceae, a remarkable new family
of flowering plants. Kew Bull. 1951, 327-347.
BAAS, P., 1975: Vegetative anatomy and the affinities of Aq1~ifoliaceae,
8phenostemon, Phelline, and Oncotheca. Blumea 22,311-407.
BENTVELZEN, P. A. J., 1962: Primulaceae. Flora Males., ser. I, 6 (2), 173-192.
BROWN, S. C., 1976: Biochemistry of 8immondsia chinensis. M. A. Thesis,
Claremont Grad. School, Claremont, Calif. 72 p.
CAMPBELL, N., and THOMSON, W. W., 1976: The ultrastructure of Frankenia
salt glands. Ann. Bot. 40, 681-686.
CARLQUIST, S., 1976: Tribal interrelationships and phylogeny of the Astera-
ceae. Aliso 8, 465-492.
DAHLGREN, R., 1975: A system of classification of the angiosperms to be
used to demonstrate the distribution of characters. Bot. Notiser 128,
119-147.
JENSEN, S. R., and NIELSEN, B. J., 1976: Iridoid compounds in Fou-
quieriaceae and notes on its possible affinities. Bot. Notiser 129, 207-212.
DEBUHR, L. E., 1975: Phylogenetic relationships of the 8arraceniaceae.
Taxon 24, 297-306.
ERDTMAN, G., 1952: Pollen Morphology and Plant Taxonomy. An Introduc-
tion to Palynology. I. Angiosperms. Stockholm: Almqvist & WIKSELL.
- -- .1958: A note. on the pollen morphology in the Ancistrocladaceae and
Dioncophyllaceae. Veroffentlich. geobot. Inst. Rubel Ziirich 33, 47-49.
--- LEINS, P., MELVILLE, R., and METCALFE, C. R., 1969: On the relationships
of Emblingia. Bot. J. Linn. Soc. 62, 169-186.
FRYNS-CLAESSENS, E., and VAN COTTHEM, W., 1973: A new classification
of the ontogenetic types of stomata. Bot. Rev. 39, 71-138.
GOLDBLATT, P., NOWICKE, J. W., MABRY, T. J., and BEHNKE, H.-D., 1976:
Gyrostemonaceae: status and affinity. Bot. Notiser 129, 201-206.
GOTTWALD, H., und PARAMESWARAN, N., 1968: Das sekundare Xylem unci
die systematische Stellung der Ancistrocladaceae und Dioncophyllaceae.
Bot. Jb. 88, 49-69.
GRANT, V., 1959: Natural History of the Phlox Family. I. Systematic
Botany. The Hague: Martinus Nijhoff.
HARBORNE, J. B., 1967: Comparative biochemistry of the flavonoids-IV.
Correlations between chemistry, pollen morphology and systematics in
the family Plumbaginaceae. Phytochem. 6, 1415--1428.
HENRICKSON, J. S., 1967: Pollen morphology of the Fouquieriaceae. Aliso 6,
137-160.
1969: The succulent Fouquierias. Cactus and Succulent J. 41, 178-184.
1972: A taxonomic revision of the Fouquieriaceae. Aliso 7, 439---537.
1973: F01tquieriaceae DC. World Pollen and Spore Flora 1, 1-12.
318 R. F. THORNE:

HUANG, T. C., 1965: Monograph of Daphniphyllum (1). Taiwania 11, 57-98.

KENG, H., 1967: Observations on Ancistrocladus tectorius. Gard. Bull.
Singapore 22, 113-121.
- 1970: Further observations on Ancistrocladus tectorius (A ncistrocladaceae).
Gard. Bull. Singapore 25, 235-:237.
LINDLEY, J., 1833:' Nixus Plantarum. London.
MAGUIRE, B., 1972: The botany of the Guayana Highland-Part IX. Bon·
netiaceae. Mem. N. Y. Bot. Gard. 23, 131-165.
DE ZEEUW, C., HUANG, Y.-C., and CLARE, C. C., jr., 1972: The botany
of the Guayana Highland-Part IX. Tetrameristaceae. Mem. N. Y. Bot.
Gard. 23, 165-192.
MELCHIOR, R. (Ed.), 1964: A. ENGLER'S Syllabus der Pflanzenfamilien,
12. Auf!., Bd. II. Angiospermen. Berlin: Gebr. Borntraeger.
METCALFE, C. R., 1952: The anatomical structure of the Dioncophyllaceae
in relation to the taxonomic affinities of the family. Kew Bull. 1951,
351-368.
and CHALK, L., 1950: Anatomy of the Dicotyledons: Leaves, Stem, and
Wood in Relation to Taxonomy, with Notes on Economic Uses. 2 vols.
Oxford: Clarendon Press.
PRANCE, G. T., 1965: The systematic position of Stylobasium DESF. Bull.
Jardin Bot. Etat Bruxelles 35, 435-448.
PRIJANTO, B., 1970,11,. Gyrostemonaceae. In: World Pollen Flora (ERDTMAN,
G., Ed.), 2, 5-13. Copenhagen: Munksgaard.
-- 1970 b: Batidaceae. In: World Pollen Flora (ERDTMAN, G., Ed.),3,
5-11. Copenhagen: Munksgaard.
ROYEN, P. VAN, 1956: A new Batidacea, Batis argillicola. Nova Guinea, n.
ser. 7, 186-196.
- 1957: Batidaceae. F!. Males., ser. I, 5, 414--415.
SCHMID, R., 1964: Die systematische Stellung der Dioncophyllaceen. Bot.
Jb. 83, 1-56.
SCHOLZ; R., 1964: Sapindales. In: A. ENGLER'S Syllabus der Pflanzen-
familien, 12: Auf!., Bd, II. Angiospermen (MELCHIOR, R., Ed.), 277-288.
Berlin: Gebr. Borntraeger.
STAFLEU, F. A. (Chairman, Ed. Comm.), 1972: International Code of Bota-
nical Nomenclature. Reg. Veget. 82, 1-426.
STEENIS, C. G. G. J. VAN, 1948: Ancistrocladaceae. F!. Males., ser. I, 4 (1),
8--10.
TATSUNO, A. M., 1976: A biochemical profile of Crossosomataceae. M. A.
Thesis, Claremont Grad. School, Claremont, Calif. 125 p.
THORNE, R. F., 1958: Some guiding principles of angiosperm phylogeny.
Brittonia 10, 72-"l7.
- 1963: Some problems and guiding principles of angiosperm phylogeny.
Amer. Natural. 97, 287-305.
1968: Synopsis of a putatively phylogenetic classification of the flowering
plants. Aliso 6 (4), 57-66.
1973: Inclusion of the Apiaceae (Umbelliferae) in the Araliaceae. Notes
Roy. Bot. Gard. Edinburgh 32, 161-165.
1974 a: The "Amentiferae" or Hamamelidae as an artificial group: a
summary statement. Brittonia 25, 395-405 (for Oct.-Dec., 1973).
1974 b: Sapindales. Encyc!. Brit., Ed. 15, 16, 239-244.
1974 c: A phylogenetic classification of the Annoniflorae. Aliso 8, 147-
209.
Some Realignments in the Angiospermae 319

THORNE, R. F., 1975: Angiosperm phylogeny and geography. Ann. Missouri

Bot. Gard. 62, 362-367.
- 1976: A phylogenetic classification of the Angiospermae. Evo!. Bio!. 9,
35-106.
WAGENITZ, G., 1976: Systematics and phylogeny of the Oompositae (Astera-
eeae). Plant Syst. Evo!. 125, 29-46.
WALIA, K., and KAPIL, R. N., 1965: Embryology of Frankenia LINN. with
some comments on the systematic position of the Frankeniaeeae. Bot.
Notiser 118, 412-429.
WUNDERLICH, R., 1971: Die systematische Stellung von Theligonum.
(Zugleich eine kritische Zusammenstellung einiger embryologischer,
anatomischer und morphologischer Merkmale del' Rubiaeeae.) Osten.
Bot. Z. 119, 329-394.

Address of the author: Dr. ROBERT F. THORNE, Rancho Santa Ana

Botanic Garden, Claremont, CA 91711, U.S.A.
Plant Syst. Evo!. Suppl. 1, 321-347 (1977)
@ by Springer-Verlag 1977

Mitteilungen aus dem Botanischen Museum der Universitii.t Ziirich Nr. 288

Evolutionary Trends in the Hamamelidales-Fagales-Group*

By

Peter K. Endress, Zurich

Abstract: Within the Hamamelidales-Fagales-complex putative evolu-

tionary trends are outlined. Structure of inflorescences, flowers, pollen,
fruits, and seeds, and peculiarities of the life history, particularly the general
delay of ovule development and syngamy in relation to anthesis are demon-
strated in the context of pollination and dispersal biology. The most out-
standing trend with many correlated characteristics is exhibited by the
adaptation to wind pollination in some groups of the Hamamelidales and
especially in almost the whole of the Fagales. This seems to be one of the
reasons for the dominant. position of the Fagales in the vegetation of tem-
perate regions.

Introduction
Ten years ago I have demonstrated similarities between the Ha-
mamelidales and Fagales (ENDRESS 1967). They have turned out to
be a surprisingly systematically homogeneous group. At the same
time, it was a contribution to the complex problem of a monophyletic
or pleiophyletic origin of the angiosperms, which had been discussed
earlier with controversial interpretations. In the meantime, the view
of the homogeneity of the Hamamelidales-Fagales-complex together
with the idea of a monophyletic origin of the angiosperms has widely
been accepted. The last decade is marked out by an intensive stimula-
tion and increase of knowledge in thc field of origin and macrosystematics
of the angiosperms, mainly initiated by the works of CRONQUIST (1968)
and TAKHTAJAN (1969). The increasing interest has led to the organiza-
tion of several mutually stimulating symposia devoted to these pro-
blems, and also to further surveys and classification proposals by
various authors.
The present paper gives a brief review of the actual knowledge of the
group with main emphasis on putative evolutionary trends. It should be
understood as a sketch with many hypothetical interpretations, which

* Dedicated to Professor Dr. FRIEDRICH MARKGRAF, in occasion of

his 80th birthday.
Plant Syst. Evo\., Suppl. 1 21
322 P. K. ENDRESS:

may be corrected in the future by further investigations. Besides the already

published data some unpublished results of the present author are included.

The group is conceived as containing the following families and

subfamilies: Hamamelidales: Hamamelidaceae (Disanthoideae, Hama-
melidoideae, Exbucklandioideae, Liquidambaroideae, Rhodoleioideae), Pla-
tanaceae, M yrothamnaceae; Fagales: Betulaceae (Ooryloideae, Betu-
loideae) , Fagaceae (Fagoideae, Oastaneoideae, Quercoideae). In general,
only literature of the last decade is cited. The older literature has
been reviewed in the following publications: Hamamelidaceae (ENDRESS
1967), Platanaceae (HUTCHINSON 1967), Myrothamnaceae (JAGER-
ZURN 1966), Betulaceae (ENDRESS 1967, ABBE 1974), Fagaceae (ELIAS
1971, SOEPADMO 1972, ABBE 1974). A short review of the Hamamelidales
is given by ENDRESS (1974 b), of the Fagales by ABBE (1974 b) and
HJELMQVIST (1974).

General Similarities Between Hamamelidales and Fagales

Some of the most outstanding similarities will be listed (Fig. 1).
The group consists of woody plants with relatively primitive secondary
xylem (diffuse porous, vessel elements often with scalariform perfora-
tions, vessels thin-walled, angular and narrow), which on the whole
is somewhat more advanced in the Fagales than in the Hamamelidales
(see further SHIMAJI 1962, BAAS 1969, MOSELEY 1974, CARLQUIST
1976). The phyllotaxis is often distichous. The leaves are stipulate.
In the deciduous taxa the venation is mostly craspedodromous (see
further ENDRESS 1968, SKVORTSOVA 1975). In bud the position of the
leaf blade with plicate vernation and the arrangement of the stipules
may be identical (Fig. 1). The midrib may have the same configuration
of vasculature (Fig. 1). Massive, multicellular glandular hairs may
be present on the petiole (Fig. 1). Expanding buds often show epinastic
movements. Sieve-tube plastids are uniformly of the S-type (BEHNKE
1973). Chemically, the common accumulation of polyphenols (quercetin,

Fig. 1. Some of the most striking structural similarities between Hamameli-

dales and Fagales. Stipulate leaf with craspedodromous venation. Vascula-
ture pattern in the midrib. Glandular hairs on the petiole. Folding of leaf
blade in bud. Diagram of l)ud with distichous phyllotaxis, orientation of
leaf blade and of adaxial and abaxial stipules. Secondary xylem with
scalariform perforation plates in vessels and alternate intervascular pitting.
Transverse section of young ovary with identical position of the ovules.
Inflorescences: epinastic spikes, terminating a vegetative shoot (from
ENDRESS 1967, except SEM photographs, leaf of Oorylm ferox, and shoot
diagram, which are original)
Evolutiona.ry Trends in the Hamamelidales-Fagales-Group 323

Hlrnamelis vlrglrOana I
_Ollrrn
Os! rya carp;n;toIia

:I<'ig_ 1
21·
324 P. K. ENDRESS:

Rhodo - "isan- Hamamelidoideae - - . - - - - - - - - - - - -

leioideae ihoideae
r-PoI-Ii-nat-o-rs-+-Bi-rds----+--FI-ies------.:----~~.---'-Va-,,-'ou-Sin-sect-S-(few-spec--ia-IiZ-ed-)--'-r
i'riSi.;dect~s

Pollination
unit

5exdistribution
Visual
I attractants

.
large
large small large
:!: 10 5 4 5 T 12-32 15-20 9-14/6-10
FJo.vers/inflor 5-10 2 3 3 -30 20-30 10-20 4-8

Fig. 2. Structural base of divergence

myricetin, ellagic acid, leucoanthocyans) (HEGNAUER, cited in ENDRESS

1967, JAY 1968, MEARS 1973, LEBRETON 1976) is noteworthy. The
inflorescences are basically racemous, often spikes or pendulous cat-
kins, which may be enriched by further dichasial ramification (also
in Hamamelidaceae: ENDRESS 1967). The ontogeny of unisexual in-
florescences may be markedly different for males and females (Fig. 10).
The floral structure, especially the gynoecium, shows an astonishing
congruence in representatives of both orders, when the whole ontogeny
from floral initiation up to fruiting is considered (Figs. 8, 9). Further
similarities and various evolutionary trends of the reproductive organs
are presented in the following sections.

Pollination Biology - Inflorescence, Flower and Pollen Structurc

In contrast to the relatively uniform ground plan (Figs. 3, 4) the
external appearance of the inflorescences and flowers varies considerably
within the group, especially in the Hamarnelidales (ENDRESS 1967,
1970, 1974 a, BOGLE 1970). Here, obviously, an adaptive radiation has
led to mainly bee, fly, bird, and ",ind pollinated flowers.
The basic inflorescence type in Hamamelidaceae is a spike or a
system of several spikes. Only in Distylium, Distyliopsis, and Matudaea
this system of spikes is reduced to a system of single flowers, arranged
in a terminated raceme or a panicle. The possibility of such an evo-
Evolutionary Trends in the Hamamelidales-Fagales-Group 325

Fagoideae ~ Casta -
neoideae
Wind - .. Insects
"nowIlIOnia ~~

<\ I ~ I <. I 0" t o"

,'- inflonlscMlc:e
fU Ild'/ o"Cl ' °fl ---
Mass of
• filaments

- --. Nectar- -
(Castanea)

sma.
medium large I mad..,., small

&~
1- 4 8 - 90 112
0" >100 0" 1- 3 0">100

in pollination biology (Original)

lutionary direction has largely been neglected so far in the literature

(ENDRESS 1969, 1970, 1971).
Largely unspecialized insect (often bee) pollinated flowers as in
Corylopsis are medium sized, bisexual, pentamerous, diplo- or haploste-
monous, nectariferous (on scales surrounding the ovary base), and bear
a double perianth with petals of a conventional shape. The general
appearance resembles flowers of for example, the genus Saxi/raga.
From such forms other types can be derived. Hamamelis and Disan-
thus are adapted to fly pollination. The flowers are relatively smalL
The long, ribbon-like petals are greenish, yellow or red. Shiny surfaces
have evolved: in Hamamelis on the exposed inside of the sepals and
nectariferous staminodia, in Disanthus on the nectariferous base of
the petals. An unpleasant floral smell is common in both genera. The
flowers of other genera closely related to Hamamelis, such as Loro-
petalum and Trichocladus are apparently not adapted to fly pollination.
A completely other type of insect (mainly bee and bumble-bee)
pollinated flowers has evolved in the genus Fothergilla with its bottle-
brush-like inflorescences. The insects feed on pollen. No nectar is pro-
duced. The corolla is totally lacking. The calyx is minute. At the same
time, the stamen filaments are thickened beneath the anthers, they.
are white and showy. Furthermore, the stamen number is increased
(Fig. 5). Thereby, two different processes are apparently involved:
an increase of the floral sectors to mostly 7, and an increase of the
326 P. K.ENDRESS:

DisIytium
Myrotharnnus

~~ /
I Disanthus

I
I
Some
HamamelidOideae \
(e.g. Corylopsis) \
Exbucklandia \
Liquidambaroideae" Platanus
"- .......

CorylopsiS
(some species)

Sinowilsonia

-------!-1~-------------
~ : ~~-~
I
I
I
I
I
I
I
I
I
Castaneoideae Nothofagus
Betula I Quen:us
Alnus c1'
I
j4'ig. 3. Inflorescences (Original, partly after ENDRESS 1967, 1970)
/~, ® ~@J'~ ~@
/~.::~~:,,\~
II iill 0,\
((yo~~
0 tOO 0
~//(/. 000. '~~'\\'\\®
\
\" IIllB ,)/ \\ 1 0 A (to I ~1,
0 .;) t;Ij
~-="~/ Corylopsis
~:~o f/ !J ~nli
\\: @(~J@)
Fagoideae Nothotagus Castaneoideae
', ...... ~_I Quercoideae (partly)
I i
,,~ - \ ~'
/00;0
j
100 .,
~ "\@( o. 0°0°0 0 0 ~
0"0 ~ ~ ~ II rg;.------....::~®-
(. ~® ~ ®
U}l 0 • 0 0 0
8~0~:\'....~. \@I ~ 0 °° 0° ~
\~O ~ O~J ... = V ~ ~ ~ ~OOOO/ (0;; ~~o~~ ~
00 / S'llOW,lson,a -........ \.." / ,(0 0
07
,\1 ~
'- ° 0 Fothergila -........ '. ~ ~ S·
~ II cFagUs. Quercoi~ Nothotagus
0 0Y
, 0 @± I \ astaneo,deae g:
°o~oo ~\~------ :--'- --~@) - - -,@,
Distylium /' /' / ' \ ~I \ (,(ffi\) r@
/ @±: \
h --=~~
~v,1ZJ2 I
I "'....\ \ \@l\.W
"----"
[
!t
Betuloideae
/ 0 \ ~~(!f':~:)y I Carpinus

o I
/00
\/' .*.
0
I
!
~
Platanus @.
I
I
@ ...-....®
o co
@
~
t
CY 8 8
o~o / Myrothamnus ' /-==..."'" I 0 0 00 f
o ~&§ 0 /;0,J~ @- (0 0 \ I 0 -S-
Betula Coryloideae
00 '0 @~ / L"'iSItLij \) I Alnus ~
....... , OO~/ . / ~~ ,§
-==- ~ I Wl'U' 0 I
--- - Rhodoleia / ~

~
Fig. 4. Floral diagrams. Encircled the basic number of floral parts ~ sepal number. ± means: deviations occur frequently Nl
..;r
(Original, a few diagrams modified after ENDRESS 1967, 1976)
328 P. K. ENDRESS:

stamens per sector to 3 or 4. The following considerations favour the

hypothesis of a secondary polyandry (ENDRESS 1976): 1. The onto·
genetical sequence of the stamen initiation is centrifugal. 2. The initia·
tion pattern in the other polyandrous genus of the family, Matudaea,
is completely different. Therefore, no "common polyandrous ground
plan" of the hamamelidaceous flower has to be supposed. 3. The stamen
groups are opposite to the sepals, which suggest a loBS of a whorl of
petals. 4. A secondary increase in stamen number favoured the evolu-

Fothergilla major 100 pm" Matudaea trinervia ~

Fig. 5. Divergent androecium initiation patterns in polyandrous Hamame·

lidaceae (from ENDRESS 1976)

tion of the special pollination syndrome by an improvement of the

visual attraction apparatus and an increased food supply for pollinators.
Parrotiopsis, closely related to Fothergilla, has less showy stamens,
somewhat fewer in number, without thickened filaments. Visual at·
tractants are, in addition, enlarged white bracts subtending the yellow
inflorescences and thus forming a pseudanthium.
Another type of pseudanthium, still more elaborated, has evolved
in Rhodoleia. Here, the flowers are arranged in heads. Petals are only
formed at the periphery of the inflorescence. This is the only case of
an extreme floral zygomorphy in the Hamamelidales.Fagales.complex.
The inflorescence is subtended by brown scales. It resembles a large
red flower and is comparable to the capitulum of Compositae. Nectar
is produced abundantly. Birds have been observed as visitors of t.hese
flowers.
Many genera are adapted to wind pollination. As a rule, the flowers
are unisexual, apetalous, inconspicuous. The stigmatic surface is large
(Fig. 8). The pollen production is high, effected by an enlargement of
Evolutionary Trends in the Hamamelidales.Fagales·Group 329

the anthers (Fig. 6), by an increase in the stamen number per flower
(e.g. Pa"otia, Matudaea) or by an increase in the flower number per
inflorescence (e.g. Sinowilsonia).As already discussed, in the highly
polyandrous genus Matudaea the stamen initiation patterais comp)etely
different from Fotkergilla. In Pa"otia, Sycopsis, Distylium, and
Matudaea with incomplete sexual separation, pollination may be ef·
fected by both wind or insects. In Distylium and Distyliopsis the flowers
are completely devoid of a perianth, but they are clearly definable
(ENDRESS 1970, in contrast to BOGLE 1970, MEEUSE 1975). Parrotia
and Sycopsis have pendent flowering heads with long, but only slightly
flexible filaments. Sinowilsonia shows unisexual pendulous catkins,
which are only slightly flexible. The closely related genus Fortunearia
is somewhat less specialized, the sexual separation less marked, the
inflorescences erect, the flowers with rudimentary petals, inconspicuous,
and at least partly wind pollinated.
In the Liquidambaroideae the flowers are densely arranged in uni·
sexual, globose heads. A perianth is almost totally lacking.
The inflorescences of the Platanaceae are very similar. However,
the flowers show a dou ble, although inconspicuous perianth, and 3 to 9
free instead of 2 united carpels (see following section).
The Myrothamnaceae closely resembles the genus Distylium in
general floral appearance. The flowers are without a perianth, the
number of floral parts is highly variable, even the inflorescence structure
is similar (JAGER.ZURN 1966).
The Fagales exhibit a simplification of flowers and specialization
of inflorescences as general trends connected with their extreme adap.
tation to wind pollination. In the Betulaceae (ENDRESS 1967, MACDO·
NALD 1971, SATTLER 1973, ABBE 1974 a, KORCHAGINA 1974) the male
inflorescences are lax, pendulous, many.flowered catkins, the female
ones pendent or erect small spikes. The flowers are minute, but increased
in number, especially in the male inflorescences, firstly, by the occurrence
of dichasia of threes or twos instead of single flowers, and secondly,
by an increased number of primary flowers as well. It is interesting
to note that also Corylopsis of the Hamamelidaceae sometimes shows
trimerous dichasia, moreover with the same position of the 2·merous
gynoecium in respect to the bract as in Carpinus (ENDRESS 1967).
Corylus exhibits the most extreme specialization and divergent evolu·
tion of male and female inflorescences. At anthesis, the female ones
do not differ superficially from purely vegetative buds, with the excep·
tion of the exserted stigmas of the 6 to 10 flowers, wheref.s the male
catkins in Corylusavellana bear more tha.n 100flowers. In the betulaceous
flower, which is always unisexual, the simple perianth is 4· or 2·merous
or even lacking. The tepals correspond to the sepals of the Hamame·
330 P. K; 'ENDRESS:

lidaceae. The androecium contains 4 to 1 stamens, opposite the sepals.

The gynoecium is always 2-merous.
In the Fagaceae (SOEPADMO 1972, ABBE 1974 a), the Fagoideae and
Quercoideae are generally wind pollinated, the Castaneoideae at least
partly insect pollinated. In the wind pollinated genera, the male in-
florescences, catkins or heads (Fagus) or at least the filaments (Notlw-
fagus) are lax and pendulous; the anthers and the stigma tend to be
large. In the insect pollinated taxa, the inflorescences are stiff and erect;
nectaries may be present (Castanea); the anthers tend to be very small,
dorsifixed and versatile, the stigma seems to be extremely small. The
flowers are basically 6-merous, not 3-merous as is usually believed,
because, firstly, male flowers mostly have 12 stamens in a diploste-
monous arrangement, secondly, the perianth is uniform, all parts behave
ontogenetically like the sepals in Hamamelidaceae, thirdly, also in a
5-merous or 4-merous flower the sepals usually are not arranged in
one whorl, but in a spiral or they are decussate (see e.g. Di8anthus,
Hamameli8, Fig. 4) .. I suppose that the 6-merous condition is derived
from a 5-merous one. The increase in stamen number by this process
(see also Fothergilla of the Hamamelidaceae) may again be favourable
in pollination biology in different ways. In Notlwfagus with only few-
flowered inflorescences the stamen number may be increased up to
90 per flower in exceptional cases. In contrary, the catkins of the Quer-
coideae have only 6-staminate flowers, but instead of single flowers
there are often many dichasial clusters composing the catkin, which
increase the stamen number considerably, up to 70 per cluster.
Unexpectedly, in the insect pollinated Castaneoideae the stamen
number per cluster may even be higher. But this seems to be compatible
with the absence of petals. Again, bottlebrush-like inflorescences have
evolved, in which the showyness is effected by the dense mass of stamen
filaments in many hundreds of flowers. In compensation, the anthers
are extremely small. These extreme features: the many-flowered
dichasial clusters, the minute anthers, the unisexual flowers (present
in all Fagale8) and the pluriovulate but one-seeded fruits (see following
section) could favour the hypothesis of a secondary reversal to
entomophily in the Castaneoideae.
Bracts, sepals and cupule. In many Hamamelidaceae and
Fagaceae the young flowers are protected until anthesis by their own
sepals, in the beginning also by their bracts. In other Hamamelidaceae
(Corylop8i8, Fothergilla, Parrotiop8i8, Parrotia, Sycop8i8, Distylium,
etc.), in the Myrothamnaceae, and in most Betulaceae the sepals are
much reduced or even lacking. Here, the floral bracts ("Tragblatter")
have taken over the protective function up to the last stages before
anthesis (ENDRESS 1975). In some Betulaceae (Alnus, Betula, Corylus)
Evolutionary Trends in the Hamamelidales-Fagales-Group 331

they are highly specialized to "hypopeltate" organs. All these genera

are characterized by dense inflorescences, often catkins. In Betulaceae,
these peltate scales have taken over an additional function as an ar-
resting mechanism for the released pollen grains (FAEGRI and VAN DER
PIJL 1971).
Another specialization is found in the Liquidambaroideae and Pla-
tanaceae. Here, both sepals and bracts, of the single flowers are reduced.
But the sterile apices of the anthers are broad and massive. They are
contiguous and provide some kind of protection in the last develop-
mental stages before anthesis for the fertile, lower parts of the anthers
in the male inflorescences (ENDRESS 1975).
In the female flowers of the Fagaceae, the cupule is an additional
protective organ, which may also have certain functions in fruit dis-
persal. The diversity in its evolution has rendered the cupule an im-
portant organ in the classification of the family (see e.g. FORMAN 1966).
Petals. In the Hamamelidaceae various stages of reduction can
be observed. Corylop,sis and Rhodoleia show petals of a normal shape.
In Disanthus, Hamamelis, Loropetalum, etc., they are narrow and rib-
bon-like. Forlunearia and Sinowilsonia have tiny, filamentous petals,
often lacking in Sinowilsonia. Many other Hamamelidaceae, the Myro-
thamnace'J,e, and the Fagales have apparently lost their petals.
Stamens. The Hamamelidales uniformly show an apical connective
protrusion of various length (Fig. 6) and mostly basifixed anthers.
The same features are realized in Notho/agus. In the insect pollinated
Hamamelidaceae and in the Platanaceae the anthers open by means
of valves-a very unusual character in angiosperms. Most of the wind
pollinated Hamamelidaceae, the Myrothamnaceae, and the Fagales
have simple slits. In the Coryloideae, Notho/agus species, and certain
Hamamelidaceae with a reduced calyx (Matudaea, Distylium), the anthers
are hairy. Generally, large anthers have evolved in wind-pollinated
genera, whereas small anthers are present in insect pollinated groups.
Pollen grains exhibit a wide range of exine structure and aperture
patterns, especially in the Hamamelidaceae (CHANG 1964, WALKER and
DOYLE 1976, preliminary personal SEM observations, see Fig. 7).
Surveys on the Fagales are given by TAKEOKA and STIX 1963, KUPRIANOVA
1965, BURRICHTER et al. 1968, STRAKA 1975, SUROVA 1975, HANKS and
FAIRBROTHERS 1976}. The distribution of the types shows correlations
with the general degree of advancement ot the genera, in particular
with the pollination type (general discussion in WHITEHEAD 1969).
In the Hamamelidaceae the insect pollinated groups have tricolpate
pollen with a reticulate surface pattern and with long and often granulate
colpi. The meshes of the reticulum are especially large in Disanthus,
Exbucklandia, and Dicoryphe. In the wind pollinated genera, there
332 P. K. ENDRESS:

fJQ( rt\\\
QII
ifl
)~
rL
,j=:"
~
lv _
Corytopsis / ' "
wi_iae / "
I \
~
) \ Liquodambar
) styraciflua /
/ \ \
'-
( Matudaea
trinervia

/ • lmm
I ---- -
I
1/
! ...-/1
,.-
,.-
,..,-----.
/ ~\ ) castanea
saliva

-- I
I
(~I
Quercus
rebut

Ii I
j
I
) j

L J
Platanus
orientalis ! ~
Fig. 6. Stamens. Fuml, sizl', and dehiscencl' patterns. Dehil:>cing line,;
marked wit.h thick lines (Original, Myrothamntt8 after JAGER-ZURN 1966)
Evolutionary Trends in the Hamamelidales-Fagalea-Group 333

Fig. 7. Pollen (Original). Voucher numbers indicated (E = ENDRESS)

334 P. K. ENDRESS:

is a clear tendency towards a decrease in the diameter of the surfacE>

meshes (Distylium, Sycopsis, Matudaea, etc.) or even to a nearly imper-
forate tectum forming a smooth surface (Liquidambar) , which is also
realized predominantly in the Fagales. A second trend is shown in
the shortening of the colpi (e.g. Distylium) and at the same time in
an increase in number of the apertures, as e.g. in Sycopsis, Matudaea,
and Liquidambar with polyrugate or polyporate pollen grains. Rho-
doleia superficially resembles the 3-colporate and smooth-surfaced
Fagaceae. The Platanaceae are similar to the tricolpate and reticulate
Hamamelidaceae. The Myrotkamnaceae differ by having tetrad pollen.
The strongest adaptation to wind pollination has apparently occurred
in the pollen grains of the Betulaceae. They are 3- (or more) -pororate
and vestibulate (anguloaperturate) which makes them more buoyant.
(Also some Hamamelidaceae show a tendency towards anguloaper-
turate grains, e.g. Distylium and Sycopsis.) The psilate, smooth surface
in Betulaceae is accompanied by a granulate, and not columellate,
infratectal exine structure, a very rare and scattered feature in angio-
sperms (VAN CAMPO and LUGARDON 1973, DOYLE, VAN CAMPO and
LUGARDON 1975, WALKER 1976).

Dispersal Biology - Gynoecium and Fruit Structure

In the whole group the gynoecium structure is rather uniform
(Figs. 1, 8, 9). In the Hamamelidaceae the gynoecium is 2-merous, thE>
ovary syncarpous and 2-locular, the styles and stigmas free. The posi-
tion of the ovary varies between superior and inferior. Each carpel
is markedly ascidiform. In contrast to certain other groups of the
angiosperms, the ascidiform part begins to form only after the ovule
initiation (ENDRESS 1967, 1974a). It widens as the ovule(s) enlarge(s).
It produces a downward prolongation of the ovarial cavity giving
space to the developing pendulous ovule(s). It is relatively less marked
in the many-ovulate gynoecia. The ovules, if one per carpel, are pen-
dulous, diagonally positioned in the two carpels (Figs. 1, 9), emerging
immediately above the upper end of the ascidiform region. The
placentation is very uniform. However, there is some confusion be-
tween axile and parietal (ENDRESS 1967, WEAVER 1969, BOGLE 1970).
In some genera the ovary development is markedly delayed at anthesis
(see following section), and therefore the ventral slit of the carpels
still open. But up to the stage of syngamy the ventral slit baR
generally closed, and hence, the placentation is clearly axile.
Variations in gynoecium structure are exhibited by the stigmatic
surface area (see third section). It is restricted to the stylar tip in ento-
mophitous genera, but covers the whole ventral side of the styles in
Evolutionary Trends in the Hamamelidales-Fagales-Group 335

anemophilous taxa (Fig. 8). Another kind of variation is represented

by the ovular number. An intermediate number is shown by Disan-
thus, Ohunia, and Exbucklandia (6), Neostrearia (3), and Oorylopsis
(3, 2 of them reduced). Rhodoleia and the Liquidambaroideae develop
many ovules per carpel in an acropetal succession. In most or all other
genera, there is only one ovule in a carpel.

MyrothamnUS lTlO5Cha\a P\aIanus orienta/is carpn..e betuLIs

CasUInea saUve

Fig. 8. Gynoecium structure. Stigmas stippled (Original, Myrothamnus

after JAGER-ZURN 1966)

Also in the multiovulate genera only one or few ovules ripen in each
carpel. But the seeds are relatively small, the testa thin (MELIKIAN
1973, MOHANA RAO 1974), and often more or less winged. Here, wind
dispersal may play a role. In the genera with uniovulate carpels a
uniformly, very elaborated and peculiar ejection mechanism for a
short range dispersal has evolved. The ripe thick-coated, smooth-
surfaced, and very hard seeds are shot out of the loculicidal and ventri-
cidal capsules for a distance of several meters. As many representatives
of the family occur on river banks, flowing water is presumably an
additional dispersal agent for longer distances. The seed surface is
water repellent and the seeds float freely on the water surface for several
1mm

Fig. 9. Gynoecium development. Median longitudinal sections. The ascidi·

form part of the ovarial cavity is marked by horizontal lines (amplified
after ENDRESS 1967, Fagus after FEY)
P. K. ENDRESS: Evolutionary Trends in Hamamelidales-Fagales-Group 337

hours or days. Other additional agents, e.g. animals, have not been
reported so far.
In the Myrothamnaceae the 3- to 4-merous gynoecium is syncarpous,
the carpels ascidiform only at the lowest part of the ovary, the numerous
ovules have an axile placentation above the ascidiform portion. The
fruit is a ventricidal capsule. Dispersal agents are not known.
The Platanaceae have evolved a much more effective dispersal
mechanism than the probably closely related Liquidambaroideae. It
is likely that the most important differences between these two groups
are just represented by the features of their two divergent dispersal
syndromes. The carpels bear a single orthotropous ovule in their ascidi-
form region. They are indehiscent and wind dispersed individually
by means of a pappus-like hairy tuft at their base. A nearly apocar-
pous gynoecium with an increased number of carpels per flower (3 up
to 9) is correlated with this dispersal syndrome.
In the Fagales the fruits are one-seeded and indehiscent through-
out as a consequence of their strong adaptation to wind pollination.
Therefore, many features of fruit differentiation diverge from the
Hamamelidaceae. On the other hand, it is most important for the under-
standing of the uniformity of the group, that the early stages until
fertilization are strikingly similar in both orders. This is especially true
for the Betulaceae when compared with the Hamamelidaceae (Figs. 1, 9)
(degree of congenital and postgenital "fusion" of the 2 carpels, placenta-
tion of the single ovule per carpel).
In the Fagaceae the gynoecium is often 3-carpellate and 6-ovulate
(constantly 2 ovules per carpel). But in both families only one ovule
per gynoecium gives rise to a seed. The other ones degenerate after
fertilization. Therefore, the proportions of the ripe ovary differ from
the Hamamelidaceae. The single seed may become much larger than
the two seeds of the Hamamelidaceae in relation to the fruit size. The
"fertile" one of the two or more locules finally occupies the whole
space within the hard outer coat of the fruit wall. This originates in
2 different ways. In many Fagaceae (perhaps excluding Quercus) the
ascidiform part of the carpels becomes extremely elongate (Fig.9).
At least in some genera the cavity is filled with unicellular hairs below
or around the ovules. Finally, the hairs are pressed against the walls
by the large seed together with the remains of the "sterile" carpels.
In Betulaceae, in contrast, the massive base below the locules
enlarges. In consequence, a large amount of the ovarial tissue around
the developing seed degenerates (Fig.9). Carpinus takes an inter-
mediate position. Therefore, in both families, the ripe fruit has func-
tionally only one large ovarial cavity. This has led to misinterpretations
of the gynoecium morphology of the Fagales.
Plant Syst. Evo!., Supp!. 1 22
338 P. K. ENDRESS:

Inspite of the uniform indehiscence of the fruits in the Fagales,

the dispersal mechanisms show even more variations than in the Hama-
melidaceae. In Betulaceae and Fagaceae smaller, wind dispersed, and
larger, animal and water dispersed fruits have evolved. Alnus, Betula,
and partly Nothotagus (WARDLE 1970) show wings on the outer side
of the fruit itself as attributes of wind dispersal. In Carpinus and
Ostrya wings are formed by the floral bracts. Corylus, Ostryopsis and
most Fagaceae form larger nuts, which are predominantly dispersed
by animals. These nuts have evolved a hard coat, which can, but only
with difficulty, be opened by such animals (STEBBINS 1974).

Further Ecological Features-Peculiarities of the Life History

The most striking trend in the life history of the group is the delay
of the ovule development in relation to pollination (Figs. 9, 10). Gen-
erally, syngamy takes place weeks or even months after pollination.
This is possibly true for all Fagales and for most of the Hamamelidales.
Preliminary studies in the Hamamelidaceae have shown that from
16 species investigated out of 16 genera only 3 have fully developed
embryosacsat anthesis, in 6 cases not even the meiocyte has differentia-
ted. In this respect Hamamelis virginiana and Corylus avellana are the
most extreme species known so far of the two orders with a gap of up
to respectively 7 and 4 months between pollination and syngamy.
In Corylus avellana not even the ovules are initiated at anthesis. It
would be intereJting to know, if they develop only in pollinated
'flowers.
At least the extratropical members of both orders are predominantly
spring or even late winter flowering. Here, a wide gap has evolved
between male and female meiosis. Male meiosis often takes place in
autumn, possibly in order to facilitate the disposal of ripe pollen grains
in time, whereas female meiosis occurs long after pollination. In Hama-
rnelis and in Corylus avellana this gap is up to 8 months.
The general adaptation of the life history to spring flowering in
the Hamamelidales, or ancestors of the group, may have been an im-
portant basis (preadaptation) for the evolution of the wind pollinated
Fagales and for their successful radiation and spreading in temperate
regions, in contrast to the less adapted Hamamelidales, which, today,
are in a stage of isolated survival or even of extinction. They have
predominantly a deciduous habit. They grow in communities with
low number of tree species, but large number of individuals of the
same species, or even in pure stands. And they flower at the end of the
deciduous season. Pollination is favoured by the absence of leaves
(WHITEHEAD 1969), and at the same time the following development
of the fruits is favoured by the beginning of the vegetation season.
Month 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9

t.".l
Rowe r! Fruit <:
fo"~ !on ~ o
Inltiat_", r--!OSII "- ~~ ~ ri~ a-
"- I ./' ot"
\ 1 "- .'c g.
Corylopsis willrnottiae / ,
/ / \
/ \
, / ,
/ / \ \
/ /
/ / I \ \
/ /
/ \
/ /
,I \
~
/ / \
Hamamelis japonica I
I I - I i
I I
I
I I ----
--- I
::l
I I --- I ~
I I
- I

HamameUs virginiana ,,
----
-- - -\ S·
I , ~
,I (D
----- ---
\
\ ----
t! 1----
, --- 1\. \
Sinowilsonia henry;
~ I-- I I
----- I I
--- - ~- +-- ---- --r--- \ "
9 I / \
/ I / \

~--------------.- 1---
_.- ~T - ~- ~+- ~- -- r---1--- -- ~- 1 - - ---"i - ~-1-- -- ~-
f~
I \ \
/ I ~
--- \ \
cf -::. -, \ \
\
Corylus avell_ ---- ----- " -, ,
1'( I \
'. ~
~ ,r -- -'- - 1 I
- - - -- -- - -- - -- - I-- - - - - I L_ 1 - - - j- -- - - -I - - .... -==:::-::: --- I--~ - - - r- --/. -L-_ 1 - - - - - f----
I I
I I
cf -- --- f I
\ f
Fagus silvatlca
~
\ I I
f
Fig. 10. Timing of major events in flower and fruit development, showing the tendency towards a remarkable dela.y of ovule
t
~ development and syngamy in relation to pollination (marked with thick line). Approximate times for wild or cultivated plants ~
in Central Europe or North America (Original, after data from SHOEMAKER 1905, FEY, and personal observations) ~

'"
340 P. K. ENDRESS:

As a much less important parallel the fly pollinated syndrome

in Disanthus and Hamamelis (third section) may have evolved with a
shift of the flowering time to winter or even autumn. Diptera are the
first insects to be active in the cold season during any short period of

V=
(AF=
/r'IIr ......·\lUl.t/iI

Fig. 11. Ovules. Diverging integument structure (at the tetrad stage).
Ovules with a single sporogenous cell and with a massive sporogenous tissue
(Original, except indicated figures). Voucher numbers indicated (E = END-
RESS)

relatively high temperature. Therefore, they are the most important

pollinators among the insects during the winter.
The structure and ontogeny of the ovules and seeds is a particularly
interesting aspect of the life history of this group. The ovules are crassi-
nucellar and anat.ropous (orthot.ropous in Platanus). In the Hama-
melidale8 and Fagaceae (except. Nothofauus) t.hey are bitegmic, in t.he
Betulaceae and in Nothofauus unit.egmic. A tendency towards an ex-
Evolutionary Trends in the Hamamelidales.Fagales.Group 341

tremely long outer integument is apparent in some groups: Exbuck·

landia of the Hamamelidaceae (KAUL and KAPIL 1975), Myrothamnaceae
(JAGER.ZURN 1966) and Fagaceae (Fig. 11). It may partly function
as an obturator. In Exbucklandia it contributes the wing of the seed.
The differentiation of a massive sporogenous tissue in the ovule has
been observed in Corylopsis species of the Hamamelidaceae and in the
Coryloideae (Fig. 11). In the other genera investigated so far only a
single meiocyte has been observed. The frequent occurrence of more
than one young embryosac in one ovule is common in Hamamelidaceae
and Coryloideae. The embryosac development conforms to the Poly-
gonum type in all species investigated. Quercus gambelii is the only
species which has been studied extensively in its ultrastructure (Mo-
GENSEN 1972, 1973, 1975a, b, BROWN and MOGENSEN 1972, SINGH and
MOGENSEN 1975). The knowledge of embryo formation is still too
scanty for a comparative survey.
Certain inhomogeneous "embryological" characters in the group
may be correlated with diversity in seed and embryo size. Firstly, in
the medium- and large-seeded Fagales, but not in Hamamelidales nor
in the small-seeded Fagale.~ (Betuloideae) , is the outer or the single
integument vascularized. Secondly, the endosperm formation is nuclear
or cellular (KAPIL and KAUL 1974) in Hamamelidales, and extremely
nuclear in Fagales. In general, small embryos (as in Hamamelidales)
may possibly rather be accompanied by an endosperm which becomes
cellular after only a short nuclear phase or even from initiation; in
contrast, large embryos (as in Fagales) have a faster growing endo-
sperm with a long nuclear phase.
EHRENDORFER (1973) and STONE (1974) point to further ecological
functions of the differing seed size. The small fruits and seeds show
an epigeal germination (Hamamelidales, Betulaceae except Corylus,
Fagoideae) , and large fruits are hypogeal (Corylus, Fagaceae except
Fagoideae). Large fruits with high storage capacity are further adapted
to shady climax forest environments, whereas small diaspores with
less nutrient supply and epigeal germination are better adapted to
open woodlands.
Cytology
Recent surveys of the chromosome numbers of the whole group
are given by EHRENDORFER (1976) and RAVEN (1976), of the Hama-
melidaceae by GOLDBLATT and ENDRESS (1977). The chromosome num-
bers are relatively constant for genera or even subfamilies with the
exception of polyploid series occurring in several genera. Base numbers
seem to be x = 6, 7, and 8. The numbers x = 6 and x = 8 probably
have evolved several times from x = 7. Hamamelidoideae, Rhodo-
342 P. K. ENDRESS:

leioideae, and Fagaceae (except Nothotagus and Trigonobalanus) show

n = 12 (or higher ploidy levels). Platanaceae, Betuloideae and partly
Corylus with at least n = 14 are based on x = 7. Disanthoideae, Ex-
bucklandioideae, Liquidambaroideae, Carpinus, and Ostrya exhibit
the x = 8 assemblage. Disanthus of the Hamamelidaceae (GOLDBLATT &
ENDRESS 1977), Carpinus and Ostrya, all with n = 8, seem to be the
only paleodiploids known so far from the whole complex.

Distribution and Phylogeny

Early fossils are still too scanty for reconstructing the phylogeny
of the group in some detail. However, some extrapolations are possible.
Earliest probable remains of the Hamamelidales are known from the
Upper Cretaceous (KNAPPE and RUFFLE 1975, DOYLE and HICKEY 1976),
of the Fagales from the latest Cretaceous (WOLFE 1974, RAVEN and
AXELROD 1974). A time sequence of the pollen types represented in
the group is shown by WOLFE, DOYLE and PAGE (1976): 1. reticulate
tricolpates (as in primitive Hamamelidaceae): Lower Cretaceous, 2. tri-
colporates (as in Fagaceae): Upper Cretaceous, 3. triporates (as in
Betulaceae): Late Cretaceous.

The main distribution centre of the group is Eastern Asia. Paleobotani-

cally speaking, the distribution is predominantly Laurasian (RAVEN and
AXELROD 1974) except the African respectively Madagascan Dicoryphe,
Trichocladus (Hamamelidaceae) , and Myrothamnaceae. The way of migra-
tion of Nothofagus to the southern hemisphere and of its radiation is still
controversial (VAN STEENIS 1971, SCHUSTER 1976).

Features of living genera may contribute knowledge to the problem

as well. Some evidence has been discussed in the preceding sections.
In Hamamelidaceae, Disanthus is especially interesting with an unusual
combination of primitive features: the only known paleodiploid of
the family (n = 8), ovary superior, medium number of ovules per
carpel, embryo sac ripe at anthesis, insect pollinated, flowers penta-
merous, bisexual, perianth double and well developed, calyx imbricate,
pollen reticulate and tricolpate. Corylopsis, also with relatively many
primitive characters, is of particular interest as well, because of the
remarkable similarity with certain BetuZaceae. Both genera are tem-
perate and deciduous. In addition, it is important, that polyandry
in Hamamelidaceae has turned out to be advanced. In BetuZaceae,
primitive and advanced features are more mixed, however, the Cory-
loideae, on the whole, show more resemblance to the Hamamelidaceae
than the Betuloideae. In the Fagaceae, Nothotagus probably shows
the closest affinity to Hamamelidaceae and Betulaceae: gynoecium
Evolutionary Trends in the Hamamelidales-Fagales-Group 343

-partly 2-merous with a 2-winged fruit, a few species completely lacking

a cupule, only a single integument (as Betulaceae), anthers long and
with connective protrusion (as Hamamelidales); some vegetative char-
acters (mostly common with Fagus and partly other genera): phyl-
Iotaxis distichous, wood diffuse porous, vessels thin and angular, vessel
elements with scalariform perforations, ray cells of the type 3 of KRIBS,
germination epigeal.
The insect pollinated genera of the Hamamelidaceae are partly
tropical, partly temperate. The wind pollinated representatives of
both orders are predominantly temperate. However, some probably
wind pollinated genera have a more tropical or not typically temperate
distribution. But, at least, they often occur in open habitats such as
river banks (Distylium, Matudaea) or semideserts (MYTothamnaceae).
The Myrothamnaceae are adapted to dry regions by their extremely
poikilohydrous behaviour (VIEWEG & ZIEGLER 1969).
From various grounds, there is some evidence, that the Fagales,
particularly the Betulaceae, may have originated from temperate de-
ciduous Hamamelt:daceae or Hamamelidaceae-like ancestors. On the
whole, the elaborate adaptation to wind pollination seems to have
been one of the most important prerequisites for the successful adaptive
radiation and distribution of the Fagales and for the conquest of their
dominant position in wide parts of the temperate regions.

I thank Professor Dr. C. D. K. COOK for reading the manuscript and

Professor Dr. H. R. HOHL for use of the scanning electron microscope.
DORIS METZGER and J. J. PITTET have made the SEM photographs, which
is also grat.!'fully acknowledged.

References
ABBE, E. C., 1974 a: Flowers and inflorescences of the ·'Amentiferae". Bot.
Rev. 40, 159-261.
- 1974 b: Betulales. Encylopaedia Britannica, Macropaedia (ed. 15), 2,
872-875. Chicago: Helen Hemingway Benton Pub\.
BAAS, P., 1969: Comparative anatomy of Platan1~S kerrii GAGNEP. Bot.
J. Linn. Soc. (London) 62, 413-421.
BEHNKE, H. D., 1973: Sieve-tube plastids of Hamamelididae. Electron
microscopic investigations with special reference to Urticales. Taxon 22,
205---210.
BOGLE, A. L., 1970: Floral morphology and vascular anatomy of the Hama-
melidaceae: The apetalous genera of Hamamelidoideae. J. Arn. Arb. 51,
310-366.
BROWN, R. C., and MOGENSEN, H. L., 1972: Late ovule and early embryo
development in Quercus gambelii. ArneI'. J. Bot. 59, 311-316.
BURRICHTER, E., AMELUNXEN, F., VAHL, J., and GIELE, T., 1968: Pollen-
und Sporenuntersuchungen mit dem Oberflachen-Rasterelektronen-
mikroskop. Z. Pflanzenphysiol. 59, 226- 237.
344 P. K. ENDRESS:

CARLQUIST, S., 1976: Wood anatomy of Myrothamnus flabellifolia (Myro·

thamnaceae) and the problem of multiperforate perforation plat,es. J.
Arn. Arb. 57, 119-126.
CHANG, K.-T., 1964: Pollen morphology in the families Hamamelidaceae and
Altingiaceae (in russ.). Flora and systematics of the higher plants 13,
173-232. Moscow-Leningrad: Nauka.
CRONQUIST, A., 1968: The evolution and classification of flowering plants.
London-Edinburgh: Nelson.
DOYLE, J. A., and HICKEY, L. J., 1976: Pollen and leaves from the Mid-
Cretaceous Potomac group and their bearing on early angiosperm
evolution. In: Origin and early evolution of angiosperms (BECK, C. B.,
Ed.), 139-206. New York-London: Columbia Univ. Press.
VAN CAMPO, MADELEINE, and LUGARDON, B., 1975: Observations on
exine structure of Euoommiidites and Lower Cretaceous angiosperm
pollen. Pollen Spores 17, 429-486.
EHRENDORFER, F., 1973: Adaptive significance of major taxonomic char-
acters and morphological trends in angiosperms. In: Taxonomy and
ecology (HEYWOOD, V. H., Ed.), 317-327. London-New York: Academic
l~ress.
- 1976: Evolutionary significance of chromosomal differentiation patterns
in gymnosperms and primitive angiosperms. In: Origin and early evolu-
tion of angiosperms (BECK; C. B., Ed.), 220--240. New York-London:
Columbia Univ. Press.
ELIAS, T. S., 1971: The genera of Fagaceae in the southeastern United
States. J. Arn. Arb. 52, 159-195.
ENDRESS, P. K., 1967: Systematische Studie iiber die verwandtschaftlichen
Beziehungen zwischen den Hamamelidaceen und Betulaceen. Bot.
Jahrb. Syst. 87, 431-525.
1968: Zur systematischen Stellung von Parrotia C. A. MEY. und SycopBis
OLIvo und zum neuen Bastard X Sycoparrotia semidecidua P. ENDRESS
et J. ANLIKER. Schweiz. Beitr. Dendro!. 16-18, 5-22.
- 1969: Molinadendron, eine neue Hamamelidaceen-Gattung aus Zentral-
amerika. Bot. Jahrb. Syst. 89, 353-358.
1970: Die Infloreszenzen der apetalen Hamamelidaceen, ihre grund-
satzliche morphologische und systematische Bedeutung. Bot. Jahrb.
Syst. 90, 1-54.
1971: Bliitenstande und morphologische Interpretation der Bliiten bei
apetalen Hamamelidaceen. Ber. dtsch. bot. Ges. 84, 183-185.
1974a: Unbekannte BIiitenpflanzen - Probleme der GroJ3systematik.
Vierteljschr. naturf. Ges. Ziirich 119, 1-21.
1974 b: Hamamelidales. Encylopaedia Britannica, Macropaedia (ed. 15),
8, 578-580. Chicago: Helen Hemingway Benton Pub!.
1975: N achbarliche Formbeziehungen mit Hiillfunktion im Infloreszenz-
und Bliitenbereich. Bot. Jahrb. Syst. 96, 1-44.
1976: Die Androeciumanlage bei polyandrischen Hamamelidaceen und
ihre systematische Bedeutung. Bot. Jahrb. Syst. 97, 436--457.
FAEGRI, K., and VAN DER PIJL, L., 1971: The principles of pollination
ecology, 2nd Ed. Oxford-New York-Toronto-Sydney-Braunschweig:
Pergamon Press.
FEY, B., --In preparation (Doctoral dissertat,ion): University of Ziirich,
Switzerland.
Evolutionary Trends in the Hamamelidales-Fagale8-Group 345

FORMAN, L. L., 1966: On the evolution of cupules in the Fagaceae. Kew

Bull. 18, 385-419.
GOJ.DBLATT, P., and ENDRESS, P. K., 1977: Cytology and evolution in
Hamamelidaceae. J. Arn. Arb. 58, 67-71.
GUSEJNOVA, N. A., 1976: On cytoembryology in Platanaceae (russ.). Bjl1ll.
Glav. Bot. Sada (Moscow) 102, 67--71.
HANKS, S. L., and FAIRBROTHERS, D. E., 1976: Palynotaxonomic investiga-
tion of Fagu8 L. and N othofagus BL.: light microscopy, scanning electron
microscopy, and computer analyses. In: Botanical Systematics 1,
1--141 + II (HEYWOOD, V. H., Ed.). London-New York-San Francisco:
Academic Press_
HJELMVUIST, H., 1974: Fagales. Encyclopaedia Britannica, Macropaedia
(ed. 15),7, 139-142. Chicago: Helen Hemingway Benton Publ.
HUTCHINSON, J., 1967: The genera of flowering plants. Dicotyledones,
Vol. 2. Oxford: Clarendon Press.
J AGER-ZURN, IRMGARD, 1966: Infloreszenz- und bliitenmorphologische
sowie embryologische Untersuchungen an Myrothamnus WELW_ Beit,r.
BioI. Pfl. 42,241--271.
JAY, M., 1968: Distribution des f1avonoides chez les Hamamelidacees et
families affines. Taxon 17, 136--147.
KAPIL, R. N., and KAUL, USHA, 1974: Embryologically little known taxon--
Parrotiop8is jacqttemontiana. Phytomorphol. 22, 234-245.
KAUL, USHA, and KAPIL, R. N., 1975: Exbtteklandia pop?dnea--from
flower to fruit. Phytomorphol. 24; 217-228.
KNAPPE, H., and RUFFLE, L., 1975: Beitrage zu den Platanaceen-Funden
und einigen Hamamelidales der Oberkreide. Wiss. Z. Humboldt-Univ.
Berlin, math.-nat. Reihe 4, 487--492.
KORCHAGINA, 1. A., 1974: On the nature of the flower of Betulaceae (russ.).
Trans. Moscow Soc. Naturalists, BioI. Sel'., Sect. Bot. 51, 50--74.
KUPRIANOVA, L. A., 1965: The palynology of the Amentiferae (russ.).
Moscow-Leningrad: Nauka.
LEBRETON, PH., 1976: Quelques donnees chimiotaxinomiques relatives aux
Fagacees. Bull. Soc. bot. France 123, 293-· 298.
MACDONALD, A. D., 1971: Floral development in the "Amentiferae". Ph.D.
Thesis, McGill University, Montreal, Quebec.
MEARS, J. A., 1974: Chemical constituents and systematics of Amentiferae.
Brittonia 25, 385--394.
MEEUSE, A_ D. J., 1975: Floral evolut,ioll ill the Hamamelididae. Acta bot..
Neerl. 24, 155--191.
MELIKIAN, A. P., 1973: Seed coat types of Hamamelidaceae and allied
families in relation to their systematics (russ.). Bot. Zhurn. U.S.S.R. 58,
350--359.
MOGENSEN, H. L., 1972: Fine structure and composition of the egg apparatus
before and after fertilization in Qttercus gambelii: the functional ovule.
Amer. J. Bot. 59, 931-941.
1973: Some histochemical, ultrastructural, and llutritiona,1 aspects of
the ovule of Quercus gambelii. Amer. J. Bot. 60, 48-54.
1975 a: Ovule abortion in Quercus (Fagaceae). ArneI'. J. Bot. 62, 160-165.
1975 b: Fine structure of the unfertilized, abortive egg apparatus in
Quercus gambelii. Phytomorphol. 25, 19-30.
MOHANA RAO, P. R., 1974: Seed anatomy in some Hamamelidaceae and
phylogeny. Phytomorphol. 24, 11:1- 1:19.
346 P. K. ENDRESS:

MOSELEY, M. F., Jr., 1974: Vegetative anatomy and morphology of Amenti-

ferae. Brittonia 25, 356-370.
\{AVEN, P. H., 1976: The bases of angiosperm phylogeny: cytology. Ann.
Missouri bot. Gard. 62, 724--764.
- and AXELROD, D. I., 1974: Angiosperm biogeography and past con-
tinent,al movements. Ann. Missouri bot. Gard. 61, 539-673.
SATTLER, R., 1973: Organogenesis of flowers. A photographic text-atlas.
Toronto-Buffalo: Univ. of Toronto Press.
SCHUSTER, R. M., 1976: Plate tectonics and its bearing on the geographical
origin and dispersal of angiosperms. In: Origin and early evolution of
angiosperms (BECK, C. B., Ed.), 48-138. New York-London: Columbia
Univ. Press.
SHIMAJI, K., 1962: Anatomical studies on the phylogenetic interrelationship
of the genera in the Fagaceae. Bull. Tokyo Univ. Forests 57, 1-64.
SHOEMAKER, D. N., 1905: On the development of Hamamelis virginiana.
Bot. Gaz. 39, 248-266.
SINGH, A. P., and MOGENSEN, H. L., 1975: Fine structure of the zygote
and early embryo in Quercus gambelii. Amer. J. Bot. 62, 105-115.
SKVORTSOVA, N. T., 1975: Comparative morphological investigations in
representatives of the family Hamamelidaceae and their phylogenetic
relationships (russ.). In: Problems of comparative morphology of the
seed plants (BUDANTSJEV, L. J., Ed.), 7-24. Leningrad: Nauka.
SOEPADMO, E., 1972: Fagaceae. In: Flora Malesiana (VAN STEENIS, C. G. G.J.,
Ed.), 7,2,265-403. Groningen: Wolters-Noordhoff Publ.
STEBBINS, G. L., 1974: Flowering plants. Evolution above the species level.
Cambridge, Mass.: Harvard Univ. Press.
STONE, D. E., 1974: Patterns in the evolution of amentiferous fruit,s. Brit--
tonia 25, 371-384.
STRAKA, R., 1975: Pollen- und Sporenkunde. Einc Einfiihrung in die Paly-
nologie. Stuttgart: Fischer.
SUROVA, T. G., 1975: Electronmicroscopical studies on pollen and spores
of plants (russ.). Moscow: Nauka.
TAKEOKA, M., and STIX, ERIKA, 1963: On the fine structure of the pollen
walls in some Scandinavian Betulaceae. Grana Palynol. 4, 161-188.
TAKHTAJAN, A., 1969: Flowering plants. Origin and dispersal. Edinburgh:
Oliver & Boyd.
VAN CAMPO, MADELEINE, and LUGARDON, B., 1973: Structure gremle
infratectale de I'extexine des pollens de quelques gymnospermes et
angiospermes. Pollen Spores 15, 171-187.
VAN STEENIS, C. G. G. J., 1971: Nothofagus, key genus of plant geography,
in time and space, living and fossil, ecology and phylogeny. Blumea 19,
65-98.
VIEWEG, G. H., and ZIEGLER, R., 1969: Zur Physiologie von Myrothamn'us
flabellifolia. Ber. dtsch. bot. Ges. 82, 29- 36.
WALKER, J. W., 1976: Evolutionary significance of the exine in the pollen
of primitive angiosperms. In: The evolutionary significance of the
exine (FERGUSON, I. K., and MULLER, J., Eds.), 251---308. Linn. Soc.
Symp. Ser. 1. London: Academic Press.
and DOYLE, J. A., 1976: The bases of angiosperm phylogeny: palynology.
Ann. Missouri bot. Gard. 62, 664-723.
WEAVER, R. E., Jr., 1969: Studies in the North American genus Fothergilla
(Hamamelidaceae). J. Arn. Arh. 50, 599- 619.
Evolutionary Trends in the Hamamelidales-Fagales-Group 347

WARDLE, J., 1970: The ecology of Nothojag'us solandri. 3. Regeneration. New

Zeal. J. Bot. 8, 571-60S.
WHITEHEAD, D. R., 1969: Wind pollination in the angiosperms: evolutionary
and environmental considerations. Evolution 23, 2S-35.
WOLFE, J. A., 1974: Fossil forms of Amentijerae. Brittonia 25, 334-355.
DOYLE, J. A., and PAGE, VIRGINIA M., 1976~ The bases of angiosperm
phylogeny: paleobotany. Ann. Missouri Bot. Gard. 62, S01-S24.

Address of the author: Prof. Dr. PETER K. ENDRESS, Institut fur Syste-
matische Botanik der Universitiit, Zollikerstr. 107, CR-SOOS Zurich, Switzer-
land.
Plant Syst. Evol., Suppl. 1, 349-374 (1977)
© by Springer. Verlag t 977

Institute for Systematic Botany, State University Utrecht., The Netherlands

Urticales, Their Differentiation and Systematic Position

By

C. C. Berg, Utrecht

Abstract: The Urticales constitute a group of plants with surprlsmg

diversity and confusing complexity, especially in the Moraceae round which
family the present paper is centred. Main trends in the differentiation of
the growth habit, flower, fruit, and inflorescence are sketched and as far as
possible connected to distribution, pollination, and dispersal. The reduced
state of the urticalean flower is a main factor in the differentiation of the
reproductive structures. Two main complexes may be distinguished in the
variation patterns: one centred round adaptations to wind pollination, the
other round adaptations to insect pollination, the latter resulting in several
types of pseudanthous or pseudocarpous inflorescences. In both complexeA
there occur well adapted large genera (Ficu8, Cecropia) and apparently less
well adapted, as well large genera (Dor8tenia, Elato8tema). The Urticale8
appear to belong to a central complex in the Angiosperms. The affinity of
the order with the Malvale8 is reasonably clear, while its connections with
the Hamamelidale8 and their allies are much more vague.

Introduetion
The Urticales constitute a distinct order, the circumscription of
which has hardly changed since about 1850. It comprises the Ulmaceae,
Moraceae, Urticaceae, and Cannabaceae. A few minor unigeneric families
have been inserted in the Urticales: Theligonaceae, Barbeyaceae, and
Eucommiaceae, but either they have been removed subsequently or
their inclusion in the Urticales is still questionalble. The Hymenocar-
diaceae yet another unigeneric family, has recently been placed in the
Urticales (cf. DAHLGREN 1975); according to WEBSTER (1967, 1975)
it is euphorbiaceous. Morphological, anatomical, and cytological char-
acters place the Cannabaceae in a rather isolated position within the
Urticales and might even be considered to make its position in the order
questionable (cf. BECHTEL '921, TIPPO 1938, RAVEN 1975). This family
will not be discussed further in the present paper. Within the Urticales
the Ulmaceae can be clearly delimited. In contrast the limits between
the Moraceae and the Urticaceae are confused, and transferring the
350 C. C. BERG:

Oonocephaloideae from the Moraceae to the Urticaceae (CORNER 1962)

does nothing to clarify the distinction between the two families. It is
preferable to regard the Moraceae s. str., Urticaceae s. str., and Oono-
cephaloideae as separate groups, perhaps as subfamilies of a single
family (cf. BERG 1973). The Moraceae s. str. can be conveniently sub-
divided into 4 tribes (BERG 1973):
M oreae: with among others Antiaropsis, A rtocarpus , Bagassa,
Batocarpu8, Bleekrodea, Broussonetia, Oardiogyne, Olarisia, Oudrania,
Fatoua, Maclura, Malaisia, Morus, Olmedia, Phyllochlamys, Pleco-
spermum, Sloetiopsis, Sorocea, Sparattosyce, Streblus, Treculia, Trophis.
Dorstenieae: with among others Bosqueiopsi8, Brosimum, Dorstenia,
Helianthostylis, Scyphosyce, Trilepisium (= Bosqueia), Utsetela.
Oastilleae C. C. BERG (ined.): with Antiaris, Oastilla, Helicostylis,
Maquira, Mesogyne, Naucleopsis, Perebea, Pseudolmedia.
Ficeae: with Ficus only.
The.Oonocephaloideae comprise: Oecropia, Oous8apoa, Myrian-
thus, Musanga, Poikilospermum, Pourouma.
On the basis of flower (cf. BECHTEL 1921) and wood (cf. TIPPO
1938) characters the Ulmaceae are regarded as the most primitive
group in the order and the Urticaceae as the most advanced. The Oono-
cephaloideae seem to he intermediate between the Moraceae and Urti-
caceae in this respect. While the Ulmaceae and Urticaceae are rather
uniform, the Oonocephaloideae show some diversity and the M oraceae
exhibit an even greater diversity. This diversity is found in the growth
habit, flower, fruit, seed and particularly in the inflorescence and
infructescence.
Habit
There is a wide variatioll in habit among the Urticales: trees, from
huge to small (Ulmaceae, Moraceae, and Oonocephaloideae) , and shrubs
(mainly Urticaceae and Moraceae); woody climbers, climbing with
the aid of roots (in Ficus, Urera) 01' spines (Maclura s.I.), 01' without
special organs [e.g. Malaisia scandens (LOUR.) PLANCR., Myrianthus
scandens LOUIS ex HAUMAN and Mrican species of Urera); (hemi-)epi-
phytes, strangling or not (in Ficus, Ooussapoa and Poikilospermum):
undershrubs (mainly Urticaceae, but also in some moraceous genera
such as Dorstenia, Fatoua, and Perebea); herbs, perennial (mainly
Urticaceae, but also Dorstenia and Fatoua) or annual (Urticaceae):
succulents (Dorstenia), and tuber-geophytes (Dorstenia).
Knowledge of the characters differentiating growth habit (life
form, tree form, gl'Owth-rhythm,differences between juvenile and
adult trees, shoot-morphology, deciduous habit, protection of apical
meristem, shedding of branches, shedding and abortion of shoot apices,
Urticales, Their Differentiation and Systematic Position 351

bud characters, and bud fo,rmation) in tropical trees is still fragmentary

and superficial. This is a field which is still in the early stages of being
explored (cf. TOMLINSON and GILL 1973) and is a work for which HALLE
and OLDEMAN'S monograph on the habit and growth of tropical trees
(1970) can serve as a basis.
The importance of a thorough knowledge of growth-habit dif-
ferentiation for tropical forestry and for the cultivation of tropical
crops is evident, but its significance for classification, evolution, and
plant geography has still to be investigated.
An attempt to arrange and interpret the available data relating
to the Urticales is presented below. From HALLE and OLDEMAN'S study,
work on moraceous taxa by JARRETT (1959, 1960) and CORNER (1967),
and my own observations it can be concluded that in arborescent
Urticales three, roughly definable growth habits are dominant. These
types of growth habits match or comprise categories recognized and
described by HALLE and OLDEMAN, who called them the "modele de
CORNER", "modele de RAUH", "modele de TROLL", etc.
1. Trees with a monopodial trunk, having monopodial subverticillate
to more or less diffusely arranged branches. The leaves are borne in
spirals or if the branches are thin they tend to be distichous. This growth
habit comprises the related models of RAUH and ATTIMS. It is common
in Ficus and the Conocephaloideae and it appears to characterize Arto-
carpus subg. Artocarpus (cf. JARRETT 1959, 1960).
2. Trees with a monopodial trunk and a sparse to dense system of
more or less vertically growing monopodial branches bearing the leaves
in spirals. In the axils of the leaves more or less distinctly phyllomorphic
horizontally growing branches with distichous, often more or less
asymmetrical leaves. These phyllomorphic branches arise in succession
from almost every node. In the Castilleae they are shed, a phenomenon
which sets this tribe apart. This growth habit, which matches the
related models of COOK and Roux, is characteristic for the Castilleae
and occurs in the Ulmaceae (Trema, Celtis) and in some Moreae (Chloro-
phora regia A. CHEV. and Artocarpus lakoocha RoxB. which belongs
to the subg. Pseudojaca). The genera Celtis, Chlorophora GAUD., and
Artocarpus (subg. Pseudojaca) also' have species which show the next
type of growth habit, which matches the model of TROLL. The model
of Roux can be converted into that of TROLL experimentally (Roux
1968, HALLE and OLDEMAN 1970) and this change probably occurs
i.n C. regia with age.
3. Trees in which the trunk and main branches are sympodial.
The often asymmetrical leaves are borne in two rows, but spirally
arranged leaves can be found on initially more or less vertical growing
parts of the trunk (of the sapling) or branches. The leaves may already
352 C. C. BERG:

be distichous in the seedling. This growth h(l.bit occurs in many Ulma-

ceae (Ulmus, Holoptelea, OeUis) , in many Dorstenieae, and in many
Moreae (e.g. in Oklorophora excelsa (WELW.) BENTH. and HOOK. and in
Artocarpus nitidus TREC. which is also a member of the subg. Pseudo-
jaca).
Besides these three predominant tree forms, there is a less common
form found in the M oraceae: a monocaul (= unbranched) or sparingly
branched (small) "tree with relatively short internodes and relatively
large, shortly petiolate, entire leaves arranged in spirals. This form
is known for several Ficus species (cf. CORNER 1967), NaucleopsiB
stipulariB DUCKE, and Dorstenia oligogyna (PELLEGRIN) C. C. BERG
(ined.). In spite ofits occurrence in quite different groups of the Moraceae,
the plants are strikingly similar, even in the leaf shape, the leaf venation,
and the presence of large (sub)persistent stipules. CORNER (1967)
assigned them to the group of pachycaul trees. In Ficus he found
that pachycauly is correlated with the presence of primitive characters
in the inflorescence and flowers. The same is found in Naucleopsis
stipul,a.ris (cf. BERG 1972) and in Dorstenia, in which the West African
D. d.iettii GUILLAUMET, a very close relative of D. oligogyna, can be
regarded as the most primitive species in the genus.
In Ficus, which exhibits an extremely wide range of growth habits
indeed, relatives of species with the simple tree form show the features of
the first type of growth habit described above, the type which is the
dominant tree form in the genus.
In Naucleopsis there is a gradation in the tree forms observed.
The series starts with N. stipularis and is followed by a group of species
with rather small pachycaul to sub-pachycaul trees having few rather
thick and long phyllomorphic self-pruning branches with relatively
large leaves [this group matches the features of the model of COOK
and includes, e.g. N. ulei (WARB.) DUCKE and N. macropkylla MIQ.]
and goes on to a group of species comprising leptocaul trees, up to 30 m
tall, with many relatively slender permanent branches and small phyllo-
morphic branches wit.h small leaves [this group matches the model
of Roux and includes, e.g. N. mello-barretoi (STANDL.) C. C. BERG].
These tree forms can be found in other taxa belonging to the Oastilleae
as well.
The species closest to Dorstenia oligogyna and D. djettii, like D.
a/ricana (BAILL.) C. C. BERG (ined.) and D. kameruniana ENGL. show
the features of the third type of growth habit described above, the
type which is dominant in the tribe Dorstenieae. On the other hand,
D. oligogyna and D. djettii are related to D. elliptica BUR., which mostly
forms un branched or sparingly branched shoots with rhizomes and
spirally arranged leaves.
Urticales, Their Differentiation and Systematic Position 353

Growth-habit differentiation in Ficus, NaucleopsiB (and related genera,

see BERG 1977), and Dorstenia appears to fit the picture of tree-form evolu-
tion drawn by CORNER (1949). Although the presence of simply structured
trees in these three genera might be the result of convergent tree-form
reduction, in agreement with CORNER (1967) I am inclined to regard the simple
tree-form (without relating it to the primitive Angiosperm habit or monocaul
non-angiospermous seed plants) as representing one of the first stages of
tree differentiation. Not only this seems to be the case in the Moraceae,
but also in other groups, as this tree form is rather widely distributed
among tropical arborescent families. On the other hand, the inflorescences
of these three moraceous genera belong to the most advanced ones in the
family.

Monocauly also occurs in Cecropia ulei SNETHLAGE. The genus

Cecropia and several species of Artocarpus (e.g. A. communis J. R. and
G. FORSTER) have also been counted among pachycaul trees (CORNER
1949). The pachycaul, and sometimes monocaul, trees of these taxa
(as well as those of, e.g., the euphorbiaceous genus Macaranga) are
components of secondary vegetation and have thick, rapidly growing
and soft woody stems With long internodes and large and often incised
leaves having long petioles. I am not sure whether it is correct to group
this type of tree with the pachycaul trees occuring in Ficus, Naucleopsis,
and Dorstenia, which are components of the lower storeys of evergreen
tropical forests. Monocauly in Cecropia is probably due to reduction.
Most of the species exhibiting the features of the second type of
growth habit described above are usually genuine components of ever-
green tropical lowland forests with relatively small distribution areas.
Many of the species showing the first, but more having features of the
third growth habit described above occur in drier tropical habitats
like (semi- )deciduous forests and savannas.
Ulmaceae (species of Ulmus and Celtis) and Maraceae (species of
Marus and Broussonetia), which are trees with sympodial growth,
even occur in extra-tropical regions. These species show abortion
and shedding of shoot apices at the end of the growth season, a pheno-
menon common in trees of northern temperate region (cf. MILLINGTON and
CHANEY 1973) but apparently rare in tropical trees (cf. KWAN KORIBA
1958, HALLE and OLDEMAN 1970). This phenomenon also occurs in
Holoptelea grandis (HUTCH.) MILDBR., which is related to Ulmus and
occurs in the lowlands of tropical West Africa. Shedding of shoot
tips appears to be a distinctive feature of Morus and Broussonetia.
It is also found in M orus mt,sozygia STAPF, occurring in African tropical
lowlands, and in the Madagascan Br0U8sonetia greveana (BAILL.) C. C.
BERG (ined.); in these taxa it is correlated with deciduous habit and the
formation of more or less well-developed scaled axillary resting buds.
Abortion of shoot tips is also reported for Maclura pomifera (SMITH
Plant Syst. Evol., Suppl. 1 23
354 C. C. BERG: Urticales, Their Differentiation and Systematic Position

1963}, a species of the third moraceous genus which has members in

northern temperate regions. In Maclura s.l. which includes Chlorophora
tinctoria (L.) GAUD. ex BENTH. & HOOK., Plecospermum, Cardiogyne,
and Cudmnia (cf. CORNER 1962), short-shoots end in spines. Shedding
of shoot apices and the formation of spinous shoot tips (because of
abortion of the apical meristem) may be regarded as related pheno-
mena (cf. MILLINGTON and CHANEY 1973).
The group of moraceous genera comprising Morus, Broussonetia, and
Maclura s.l. shows in its distribution similarities with families and genera
like the Magnoliaceae and Hamamelidaceae and Berberis, Oelastrus, and
Euonymus (cf. GOOD 1974), viz. centred more or less distinctly in northern or
altitudinally subtropical to temperate regions; the areas range in the Old
World from Japan through Sri Lanka to Madagascar and East Africa and
through the Philippines and Indonesia to New Caledonia and in the New
World from S.E. North America through Central America and the Andes
southwards. The distribution of this group of moraceous taxa as well as
that of !,!ome ulmaceous groups suggests that their adaptations to temperate
conditions did not arise in (seasonal) tropical lowland conditions. The presence
of such species under these conditions appears to be secondary.

Aerial and stilt roots contribute to the diversity of arborescent

taxa in the Urticales. Such roots are common in Ficus, and they also
seem to be a characteristic feature of the Conocephaloideae, a group
in which they are functional not only in epiphytic taxa (Coussapoa
and Poikilospermum) but also in those occurring in inundated places
(Cecropia species). On the other hand, whether they are functional
or not in other species (e.g. of Pourouma and Myrianthus) still has to
be established.
Variation ranging from an almost arborescent habit through a
frutescent, suffrutescent, and herbaceous to a succulent habit is found
in Dorstenia, which, after Ficus, is the most diverse and interesting
genus of the Momceae. An outline of this variation is shown in Fig. 1.
Most non-succulent frutescent to herbaceous African species of
Dorstenia have branched underground creeping rhizomes, which can be
regarded as important for vegetative reproduction (of. RICHARDS
1952). These species, as well as the only African species with creeping
leafy stems (D. picta BUR.), are components of the undergrowth of
evergreen forest and most of them have rather small areas of distribu-
tion. The phanerophytic forms with thin, often more or less woody
rhizomes gradually pass via forms with swollen rhizomes (D. preussii
ENGL.) and those with more or less irregularly shaped discoid tubers
(D. cuspidata HOCHST.) into truly geophytic forms with regularly
shaped placentiform tubers (D. benguellensis WELW.). The acaulescent
tuberiferous D. barnimiana SCHWEINF. is probably derived from a
caulescent tuberiferous form. The tuber-forming taxa occur in the
---
/
/
/
/
/
/
/

•
(

/
-~.- ---
~'. @

Fig. 1. Schematic drawings of the habits of various Dorstenia species,

showing the main trends in differentiation. 1 D. djettii; 2 D. africana;
3 D. elliptica; 4 D. psiluTUS; 5 D. letestui; 6 D. preussii; 7 D. cuspidata;
8 D. barnimiana; 9 D. benguellensis; 10 D. picta; 11 D . elata; 12 D. ramosa;
13 D. cayapia; 14 D . gigas; 15 D. crispa
356 C. C. BERG:

drier regions of Central and East Mrica and have a (rather) wide distri-
bution. The stem succulent'! like D. gigas SCHWEINF., which forms
up to 2 m tall branched shrubs, and D. crispa ENGL., which forms
low monocaul stems, cannot be clearly connected with the rhizome-
and tuber-forming group of species.
D01"stenia species from South and Central America can be morpho-
logically connected with D. picta. Most of them are components of the
undergrowth of evergreen tropical forests and form short monocaul
stems. In species occurring in relatively dry habitats the stems are
subterranean and somewhat tuberous (cf. CARAUTA, VALENTE and SUCRE
1974). The absence of long creeping rhizomes or tubers derived from
them causes differences in habit and life form between Mrican and
American species, notwithstanding their occurrence in similar habitats.
The above considerations outline the variations in growth-habit
(pachycaul to leptocaul, arborescent to herbaceous) recognizable on
the basis of the available evidence. It is undoubtedly worth-while
investigating this promising field much more closely in dealing with
the distribution, origin, and evolution of plant groups. As is apparent
from the above discussion, growth-habit variation may well be of
importance in taxonomy in spite of statements to the contrary (cf.
WHITMORE 1975).
Flower
The small flowers of the Urticales consist basically of 4-5 tepals,
4-5 epitepalous stamens, and a bicarpellate pistil with a single ovule.
There is not the space here to give a detailed survey and the reader is
referred to CORNER'S work on the classification of the Moraceae (1962),
WEDDELL'S monograph of the Urticaceae (1856-1857), and PLANCHON'S
study on the Ulmaceae (1848).
The extremes in the variation of the flower in the Urticales are the
bisexual flower with 5 tepals and 10 or sometimes even more stamens, as
in Ampelocera (Ulmaceae), and the flower consisting of a pistil or a
single stamen only, as occurs in Treculia and Brosimum, respectively.
Although reduction of the number of floral parts is the ma:in trend,
multiplication sometimes occurs as in the ulmaceous genera Holop-
telea and Ulmus and the moraceous genus Naucleopsis.
In the Ulmaceae the flowers are bisexual or unisexual by abortion.
Differences between staminate and pistillate flowers are not pronounced
and specialized stamens and strongly reduced flowers do not occur
in this family.
In the other groups of the Urticales the flowers are unisexual. Pistil-
late and staminate flowers can be quite different in size and form of
the perianth. More or less strongly reduced flowers often occur mainly
Urticales, Their Differentiation and Systematic Position 357

in groups in which the stamens are straight before anthesis or if inflexed

gradually straightening at anthesis, as in Bosqueiopsis.
Inflexed stamens which bend outwards suddenly and elastically
at anthesis, thus shedding their pollen simultaneously, are distinctive
for the Urticaceae s. str. and occur in many genera of the Moreae.
Flowers with such stamens normally have a well-developed perianth
with 4-5 tepals, 4-5 stamens, and a more or less conspicuous pistillode.
They still show the basic structure of the Urticales flower, with the
exception of the reduced pistil. This completeness of structure appears
to be connected with the presence of the explosive stamens. A more
or less strong expression of the bisexual condition is correlated with a
less reduced state of the flower, at least as regards the number of tepals
and stamens.
In groups of the Urticales with straight stamens (Moraceae and Cono-
cephaloideae) the number of tepals and stamen'! is often reduced to 3
or 2. Further reduction can result in disappearance of the perianth,
leaving a single stamen. Brosimum lactescens (S. MOORE) C. C. BERG
shows reduction from a 4-merous staminate flower to a 2-merous one
and B. utile (H. B. K.) PITTIER from a 2-merous one to a perianthless
flower with a single stamen (cf. BERG 1972).
Loss of and diminished protection of young stamens by the perianth
means that the stamens must be protected in other ways. In Brosimum
the necessary protection is given by bracts which form a pseudo-
perianth, by firm peltate bracts which cover the young flowers, and/or
by a foveate receptacle which bears the staminate flowers in pits.
In Castilla the stamens are protected by the cup-shaped or bivalvate
receptacle which is closed until anthesis, in Pseudolmedia and N au-
cleopsis by the bracts of the involucre, and in the peculiar inflorescence
of Trilepisium by an expanded margin of the receptacle. Even if a
delicate perianth is present in staminate or pistillate flowers (as in
Treculia and Broussonetia, respectively), bracts appear to have a pro-
tective function. In the Moraceae and Dorstenieae these bracts are
usually (hypo }peltate, a form which often arises in connection with
the protection of flowers and floral parts (cf. ENDRESS 1975).
In some moraceous genera with straight stamens, as in Clarisia
and Antiaris, the staminate flowers can become more or less disor-
ganized, by losing the distinct relationship between tepals and stamens.
The distribution of straight and inflexed (explosive) stamens sug-
gests that the latter type is derived. The inflexed stamens probably
appeared early in the evolution of the Urticales and separated its mem-
bers into two groups. This separation put its stamp on the differentia-
tion of the Urticales, especially in relation to pollination. In some
genera, such as Cudrania and possibly also Sorocea and Antiaropsis,
358 c. C. BERG:

which taxonomically are more or less closely related to genera with

inflexed stamens, the straight stamens are probably derived; this is
suggested by their shape and insertion and/or dehiscence and/or the
direction of the filament at different flowering stages.
Wind pollination probably occurs in all taxa with inflexed stamens.
In the group of Urticales without such stamens, more or less distinct
adaptations to wind pollination occur. A remarkable case is found in
Cecropia. In many of its species the anther is detached from the straight
filament by abscission, but it remains loosely attached to the flower
(by the sticky ends of the appendices of the anther or by a bundle of
stretched spiral thickenings of the tracheary elements of the staminal
vascular bundle) and is movable; clearly, these species are anemophilous.
In other taxa, e.g. in several representatives of the Ulmaceae, but
also in Maquira coriacea (KARSTEN) C. C. BERG and Brosimum lactescens
(S. MOORE) C. C. BERG which both belong to groups not generally
adapted to wind pollination, adaptation to wind pollination followed
a more common way by producing slender filaments and relatively
large anthers. In the pendulous catkins one can find another ada,pta-
tion to wind pollination, which is important especially for taxa with
non-specialized stamens (e.g. Bagassa, Batocarpus, Sorocea).
The deciduous habit connected with flowering may also be regarded
as an adaptation to wind pollination (cf. WHITEHEAD 1969).
An important consequence of wind pollination is that plants de-
pending on it require more or less open habitats, like (semi- )deciduous
forests, secondary vegetations, savannas, forest margins, riversides,
etc., i.e. places with sufficient air movement. Light demand is cor-
related with their presence in such habitats. Some arborescent, light-
demanding, and more or less distinctly deciduous taxa which depend
on wind pollination, like Chlorophora excelsa and Bagassa (Juianensis
AUBL., can penetrate evergreen forest through accidental clearings.
They become tall enough to make wind pollination possible. A few
anemophilous taxa, like species of Elatostema and Sloetiopsis usam-
barensis ENGL., occur in the undergrowth of evergreen forests and can
stand heavy shade, the former as herbs and the latter as small trees.
But they chiefly occur near small streams where air movement can be
expected. S. usambarensis also occurs as shrubs on exposed rocks in
river beds and coastal shrub.
Pistillate flowers lacking a perianth or having a reduced one occur
in several moraceous genera in which the flowers are enclosed in
cavities formed by the receptacle of the infloresecnce (Ficus) or by
bracts (Treculia, Broussonetia). If the pistillate flower is completely
immersed in and fused with. the receptacle of the inflorescence a perianth
can seldom be distinguished. The tepals of the pistillate flower are
Urticale8, Their Differentiation and Systematic Position 359

often fused, forming a tubular perianth which in turn is often fused

with the pistil or fruit in megaspermous taxa, and enlarged at fruit ..
Fusion of the tepals often precedes fusion of the flowers and occurs
in many members of the Oastilleae.
Anatomical studies on the flowers of Urticales (BECHTEL 1921)
and of Artocarpus (SHARMA 1965) indicate that the pistil of the Urticales
is derived from one with at least 3 carpels forming an ovary which has
3 or more locules with 2 or more axillary ovules per carpel. Ovaries
with 2 or 3 locules and/or 3 styles occur rarely in the Moraceae and
Ulmaceae. Further reduction of the gynoecium leads to the pseudo-
monomerous pistil with an anatropous basal ovule, as occur in the Urtica-
ceae and Oonocephaloideae. Reduction of one of the two stigmas is
observed in several Moraceae and can be regarded as a first step in
this reduction which has no apparent adaptive significance.
The shape of the stigma varies considerably. In the anemophilous
taxa it is usually filiform (with a moderate length and often pilose)
or comose.
Fruit
In the Ulmaceae the fruits are either samaras (in Ulmus and the
related genus Holoptelea) or drupe (let)s. In the Urticaceae the fruits
are achenes or drupelets. The achenes can sometimes be enclosed by
a winged fruiting perianth (in Memoralis) enveloped by a succulent
fruiting perianth, forming pseudo-drupelets which are often aggregated
(in many taxa of the Boehmerieae) or ejected by staminodes (e.g. in
Elatostema) .
The basic type of fruit in the Moraceae seems to be a dehiscent
drupe and is found in several genera belonging to different tribes:
A ntiaropsis , Bleekrodea, Broussonetia, Dorstenia, Fatoua, Ficus (cf.
CORNER 1962), Scyphosyce, Sparattosyce, Streblus, Sloetiopsis, and
Utsetela. This fruit is more or less stipitate and the white fleshy exo-
carp splits into two valves that squeeze out or eject the endocarp body
which is often smooth and blackish in megaspermous taxa, or greyish
and tuberculate in microspermous taxa (Dorstenia species, Fatoua). The
endocarp bodies drop near the plants. How they are dispersed sub-
sequently and whether they are eaten by animals is unknown. Common
gregarious occurrence of several taxa with these fruits suggests short-
distance dispersal. In most cases the fruits or infructescences containg
them are inconspicuous (often greenish) and often hidden by the foliage
of the plants. In Antiaropsis and probably also in Bleekrodea the presence
of a red fruiting perianth seems to promote dispersal because of the
attractive red-black colour contrast.
The stipes and/or the white colour of the indehiscent drupelets
360 c. C. BERG:
such as occur in Chlorophora excelsa, Baga8sa guianensis AUBL. and
several species of Br0U8sonetia, the way in which the exocarp is thickened
in the indehiscent drupelets of Morus and related genera, and the
occurrence of a transition from the dehiscent to the indehiscent drupelet
in Broussonetia suggest that these two types of indehiscent drupes
are clo'.le to the dehiscent type and that they may well be derived
from the latter (see Fig. 2). These indehiscent drupelets are enclosed
in succulent and often coloured spicate or globose infructescences
which are attractive to animals.

2 3

Fig. 2. Three types of moraceous fruits. 1 dehiscent drupelet (Dorstenia);

2 stipitate indehiscent drupelet (Broussonetia); 3 broadly based indehiscent
drupelet (Morus)

In many Moraceae, especially in megaspermous taxa, the fruit

is fused with the tubular enlarged fleshy and coloured perianth or im-
mersed in and fused with an enlarged fleshy and coloured receptacle
of the inflorescence, thus forming solitary or aggregated pseudo-drupes
or drupaceous infructescences. In most of these cases it is not possible
to distinguish a pericarp. The succulent perianths fused with the fruits
are often yellow- to red-coloured, but sometimes they are black, e.g.
in Sorocea. In this genus the pseudo-drupes are bome by contrasting
red-coloured (often swollen) pedicels and the rachis of the racemose
inflorescence. The enlarged receptacles fused with the fruits are usually
yellow- to red-coloured as well, but are black in Trilepisium.
If the dehiscent drupe (not being adapted to long-distance dis-
persal) is the original type of fruit in the Moraceae, the coloured perianth
in Antiaropsis, the formation of indehiscent drupes, enclosure of the
fruits in coloured perianths or receptacles, and aggregation into syn-
carpous infructescences may be regarded as improvements in bringing
about dispersal.
Urticalea, Their Differentiation and Systematic Position 361

In the Conocephaloideae there are various kinds of fruit. In Pou-

rouma and Myrianthus the megaspermous fruits together with the
enlarged perianths form pseudo-drupes. The microspermous dry fruits
of Cecropia and Musanga are enclosed by greenish succulent perianths.
In the epiphytic genera Coussapoa and Poikilospermum the pericarp
consists of an endocarp with small protuberances and a mucilaginous
mesocarp and the endocarp body, surrounded by the mucilaginous
layer, is (often) squeezed out by the basally thickened fruiting perianth.
The seeds vary from large, when they are usually without endosperm
but with a more or less complicated embryo having thick and often
unequal cotyledons and a short radicle, to smdll, when they are usually
with endosperm and a rather simple embryo, rather flat and equal
cotyledons, and a long radicle. As pointed out by CORNER (1962)
microspermy is connected with a herbaceous habit, occurrence in
secondary vegetations, or epiphytism. Megaspermy is more or less
clearly linked with an arborescent habit and occurrence in tropical
forests. Connections between seed characters and primitiveness appear
to be lacking in the Urticales. Large seeds occur in many ~oraceae
(Castilleae, most Dorstenieae, and several Moreae like Artocarpus and
Treculia) and in several Ulmaceae. Small seeds occar in the Urticaceae,
in several Moraceae (Ficus, Dorstenia, and several members of the
Moreae). Dorstenia shows a distinct morphological series in its range
of large to small seeds: large seeds occur in the frutescent species,
medium-sized ones mainly in the suffrutescent species, and small
ones in the herbaceous and succulent species.
The testa is usually thin and its protective function has been trans-
ferred to the endocarp, which is usually crustaceous but may even
be woody.
Inflorescence
An explanation of the morphological structure of the inflorescence
can be given in general terms, although several details are still pro-
blematical (cf. GOLENKIN 1894, BERNBECK 1932). The basic structure
of the Urticales inflorescence appears to be a compound dichasium
(or cymose panicle, cf. CORNER 1967). The basal (= central) flowers
in the bisexual inflorescence are pistillate but in the Urticaceae often
staminate (cf. BERNBECK 1932).
In the aggregation of small flowers, the Urticales follow a common
trend in the Angiosperms.
The Ulmaceae mostly have loose or condensed cymose or secondarily
racemose inflorescences with bisexual, bisexual and staminate, or
unisexual flowers. The number of flowers can be reduced to one in
the pistillate inflorescence.
362 C. C. BERG:

The main variations which are observed In the inflorescences of

the Urticaceae s.l. are tendencies towards:
1. shortening of the axes;
2. reductiou in the number of the axes;
3. fusion of the axes;
4. dorsi-ventral orientation (= adaxial orientation of the flowers);
5. change from dichasial to monochasial growth;
6. reduction in the number of flowers;
7. change from bisexual to unisexual.

Repeatedly branched inflorescences are common in the Urticaceae

s. str. and the Oonocephaloideae. In the former group the variation
mostly results in paniculate, racemose, spicate, capitate, or sessile
glomerate inflorescences, but sometimes in discoid (and involucrate)
inflorescences (as in Elatostema in which genus they are transformed
into a fig-like inflorescence in Elatostema ficoides WEDD.). The inflores-
cences of the Urticeae s. str. are either unisexual or bisexual. In most
genera of the Oonocephaloideae the flowers usually become clustered
in heads at the ends of the branches of the unisexual inflorescence.
Reduction in the ramification of the inflorescence can result in a single
head, spike, or umbel in the pistillate inflorescence. In Oecropia the
flowers are arranged in digitate clusters of spikes which are enveloped
by a caducous spathe, thus forming an unique type of inflorescence.
In the Urticeae s. str. and Oonocephaloideae, as well as in the Moraceae,
pistillate and staminate infloreseences are often (quite) different in
size and shape; the pistillate inflorescences are usually the most derived
ones.
Branched inflorescences are rare in the Moraceae. They are found
in Fatoua and Bleekrodea and in both genera are mostly bisexual.
The other members of the Moreae have mostly unisexual spikes, ra-
cemes, heads, or the more derived discoid-involucrate inflorescences

Fig. 3. Schematic drawings of the inflorescences of Moracooe, showing

some of the main trends in differentiation. A. Castilleae, pistillate inflor-
escences. 1-3 hypothetical; 4a Pereboo, Helicostylis, etc.; 4b,Pseudolmedia.,
Perebea, etc.; S Pereboo, Oastilla, etc.; 6a Oastilla; 6b Antiaris, Mesogyne;7
N aucleopsis. B. Castilleae, staminate inflorescences. 1-3 hypothetical; 4aPere-
boo, Maquira, etc.; 4b Pereboo humilis; S Helicostylis; 6 Naucleopsis;
7 Pseudolmedia; 8 Oastilla. C. Moreae, pistillate inflorescences. 1 Trophis,
Sorocoo, etc.; 2 Trophis, Sorocoo, etc.; 3a Trophis involucrata; 3b Olmedia
aspera; 4a Sorocoo sprucei; 4b Sorocoo duckei; Sa Olarisia ilicifolia; Sb Ola-
risia racemosa; 6a Antiaropsis; 6b Antiaropsis, Phyllochlamys; 7 Sparat-
tosyce
Urticales, Their Differentiation and Systematic Position 363

@ ~
~
3~
I
~ - ftii
4b

\~ .~
SS? S~?
I ~

\ 6~
~-8~
364 C. C. BERG:

(as in AntiarOp8is, Olmedia, and PhyllochlamY8) or the urceolate· in-

florescences (in Sparatto8yce). Further reduction can result in un i-
florous pistillate inflorescences.
The Oastilleae have unisexual, discoid to globose or to urceolate,
involucrate inflorescences, which are often uniflorous in the pistillate
ones but rarely so in the staminate ones. The Dor8tenieae have discoid
to globose or to urceolate, basically bisexual inflorescences, which
mayor may not be more or less distinctly involucrate. They can be-
come unisexual. Pistillate inflorescences, on the other hand, can be-
come uniflorous. The Ficeae have bisexual (or functionally unisexual)
urceolate inflorescences.
Schematic drawings of the variation in the inflorescences of some
groups of the Moraceae are given in Figs. 3 and 4.
The more or less distinctly involucrate inflorescences, f>uch as
occur in the Oastilleae, in several Dor8tenieae, and in a few Moreae,
are more or less distinctly pseudanthous.
A remarkable aspect in the variation of the more or less pseu-
danthous inflorescences is that some trends in the differentiation of
the angiospermous flower are repeated: many to few parts (= flowers),
free to fused parts (= flowers), superior to inferior parts (= flowers),
free or adnate to a hypanthium-like of the (cup-shaped) receptacle
of the inflorescence. For the genus Dor8tenia can be added the change
from an actinomorphic to a zygomorphic inflorescence.
Distribution of the sexes is very variable. In the predominantly
anemophilous families of the Ulmaceae and Urticaceae bisexual inflores-
cences are common and dioecism is not a dominant feature. How-
ever, dioecism is characteristic for the Oonocephaloideae. In the Oas-
tilleae, with their strictly unisexual inflorescences and flowers which
seldom show vestiges of the other sex, monoecious taxa are rather
common. Furthermore, in some of these taxa (Antiari8 and Oastilla)
both monoecoius and dioecious specimens can be found. In the Dor8te-
nieae, with their basically bisexual inflorescences, staminate inflorescen-
ces can occur alongside the bisexual ones (in B08quieop8i8 and He-
liantho8tyli8) .
The inflorescences are solitary, or in the Urticaceae s.l. often in
pairs, in the axils of the leaves. Secondary inflorescences can be borne

Fig. 4. Schematic drawings of the inflorescences of Moraceae, showing some

of the main trends in differentiation. D. Mostly Dorstenieae. 1 hypothetical;
2 Bleekrodea (Moreae); 3a and 3b, Bleekrodea madagascariensis; 4 Utsetela;
5a and 5b, Helianthostylis; 6 Trymatococcus; 7a, 7b, and 7c Bosqueiopsis
gilletii; 8 Brosimum; 9a and 9b, Brosimum lactescens; lOa and lOb, Brosi-
mum utile; 11 Dorstenia djetii; 12 Dorstenia africana; 13 Dorstenia barteri;
14 Scyphosyce: 15 Trilepisium; 16 Ficus
Urlicales, Their Differentiation and Systematic Position 365

~
:
I .'
II
+
~

o
+
-
@
* b
Af
~ 1\
"
'J

y" , '

.~..
@ \,.'0

b @J c

Fig. 4
366 C. C. BERG:

on more or less condensed leafless shoots as in cauliflorous taxa or in

taxa with ulliflorous inflorescences like Clarisia racemosa R. & P.
which has its uniflorous pistillate inflorescences in secondary racemes.
In most Castilleae the inflorescences are borne in clusters in the leaf
axils. In some of them (e.g. Antiaris, Mesogyne, and Castilla) staminate
and pistillate inflorescences can occur in the same cluster; the staminate
ones surround the pistillate one, thus simulating to some extent the
arrangement of "male" and "female" parts in the bisexual flower.
The variation in the inflorescence of the Urticales is related to and
partly governed by the structure and variation in the flower and fruit
and the dimensions of the seed. Consequently, it is related to pollination
and dispersal as well.
The small monochlamydeous flowers of the Urticales do not produce
nectar. A few taxa are known to produce odour; mostly it is only
staminate flowers including those of some anemophilae, which do this.
The perianth is uncoloured, at least at anthesis. These disadvantages
with regard to insect pollination appear to have been more or less
successfully overcome in two ways. First, by adaptation to wind pol-
lination, largely achieved by modification of the stamens (see p. 357-358).
In groups having such stamens the shape of the inflorescence does
not seem to be so important although arrangement of the staminate
flowers in catkins may promote shedding of the pollen. In groups
whose stamens are not adapted to wind pollination, pendulous spicate
. or racemose staminate inflorescences are essential for pollination.
The presence of such inflorescences is correlated with an arborescent
or frutescent habit and if arborescent often with deciduousness as well.
In these groups of the Urticales variation in the pistillate inflorescence
is chiefly related to dispersal.
The arrangement of flowers (which do not have stamens adapted
to wind pollination) in secondary functional units more suited to in-
sect pollination is the other way be which disadvantages due to the
reduced state of the flower have been overcome. In many cases these
units simulate more or less closely the structure and appearance of
flowers. In some Moraceae pseudanthy is very pronounced. The
staminate inflorescences of most Naucleopsis species resemble the flowers
of Tovomita (Guttiferae) and Ternstroemia (Theaceae) and the bisexual
inflorescences of Trilepisium resemble flowers of several Myrtaceae.
The significance of the differences between staminate and pistillate
inflorescences in groups which otherwise have more or less distinctly
pseudanthous inflorescences is puzzling.
Pseudanthy is connected with occurrence in tropical forests. Un-
fortunately, data about pollination in groups with pseudanthous in-
florescences are (almost) lacking. As a rule, the fructification rate is
Urticales, Their Differentiation and Systematic Position 367

normal, which, assuming that reproduction is sexual, suggests suc-

cessful pollination. However, some facts about Dorstenia suggest
that adaptation to insect pollination is less satisfactory. In many
Dorstenia species some features (colour, shape of the bract-anns and
appendages) of the more or less distinctly pseudanthous bisexual in-
florescences with their minute flowers show a tendency towards adap-
tation to pollination by flies. A study of several populations of Dorstenia
africana (BAILL.) C. C. BERG (ined.), which has yellow, faintly smelling
inflorescences, showed that a very low percentage of the inflorescences
set fruit which indicates only occasional pollination. The same is found
in Scyphosyce manniana BAILL. which in many characters resembles
Dorstenia. The occurrence of apomixis (cf. GUSTAFSON 1946), as well
as the presenct; of rhizomes to ensure vegetative reproduction (cf.
RICHARDS 1952) and possibly the production of a wide range of inflores-
cence forms, often bizarre and apparently without significance, could
well be related to unsuccessful adaptations to insect pollination.

It is remarkable that the urticaceous genus Elatostema which resembles

Dorstenia species in so many features also contains apomictic species (cf.
FAGERLIND 1944). As in Dorstenia, many species of Elatostema have discoid,
more or less distinctly involucrate, mostly bisexual inflorescences with short
stamens in the staminate flowers. They usually form creeping rhizomes and
often occur in the undergrowth of evergreen forests, a habitat which does not
promote wind pollination. Elatostema even resembles Dorstenia in regard
to the ejection of the fruits (containing seeds without endosperm in contrast
with other Urticeae s. str.) and also in its taxonomical complexity.

Successful adaptation to insect pollination is only known for Ficus,

with its pseudocarpous inflorescences. More or less pseudocarpous
(although quite different from Ficus in structure and appearance)
are the pistillate inflorescences which occur in Artocarpus and Treculia.
In spite of some records on pollination in Artocarpus (JARRETT 1959 to
1960, FAEGRI & VAN DER PIJL 1972) how it takes place in these genera
is uncertain.
Cauliflory in the Moraceae is connected with the presence of these
pseudocarpous inflorescences. In Ficus one can find almost the whole
range of types of cauliflory (cf. CORNER 1933, RICHARDS 1952) and it
is probably related to dispersal rather than to pollination (cf. VAN DER
PIJL 1972). The same may possibly be true for the other moraceous
taxa with pseudocarpous inflorescences.
The more or less pseudanthous and, of course, the pseudocarpous
pistillate and bisexual inflorescences form at the fruiting stage one-
to many-seeded structures which often resemble true fruits. In Treculia
africana DECAISNE and several species of Artocarpus these structures
resemble the fruits of the bombacaceous genus Durio. The spinous
368 c. C. BERG:

infructescences of Treculia acuminata BAILL., T. obovoidea N. E. BROWN,

and several species of Naucleopsis resemble the fruits of Oncoba (Fla-
courtiaceae) and Nephelium (Sapindaceae). The infructescences formed
by several multiflorous Oastilleae are reminiscent of the compound
fruits which occur in the Annonaceae. Other inflorescences, like those
in species of Brosimum, form drupaceous infructescences.

General Remarks
Surveying the differentiation patterns of the Urticales I am inclined
to regard them for the greater part and especially with reference to
reproductive structures as the results of two, not always clearly dis-
tinguishable processes. In the first of these processes ~daptive aspects
are not apparent or may even be lacking, while in the second one de-
velopments seem to be adaptive responses. to changes caused by the
first process. Examples of the first process can be found in the trends
towards the reduction of flowers and inflorescences - trends which can
also be found in other groups of the Angiosperms. The process with
distinct adaptive aspects can be found in developments re-establishing
successful -pollination and dispersal.
The reduced state of the flower can be regarded as a leading factor
concerning the variation in the Urticales, especially with regard to
their reproductive structures. Two main complexes can be distinguished.
In one complex, the adaptations centre round insect pollination; in
the other, the adaptations relate to wind pollination. The former com-
plex is clearly linked to the conditions of evergreen (to semi-deciduous)
tropical lowland forests. The group of plants involved can be roughly
characterized by an arborescent habit (comprising rosette trees), by
pseudanthous (to pseudocarpous) inflorescences with further reductions
in the flowers and unspecialized stamens, and by megaspermous (ag-
gregated) pseudo-drupes.
The number of Moraceae which can be regarded as characteristic
representatives of this group of plants is quite small, but the concerned
species show considerable diversity, notwithstanding the rather constant
conditions under which they probably arose. The large and diverse
genera Ficus and Dorstenia do not fully match the features of this
group of M oraceae, in view of the occurrence of microspermy in both
genera, the herbaceous habit in Dorstenia, and the extension of their
areas of distribution far outside the rain forest. This last seems in
Dorstenia to be due mainly to the herbaceous haht. and- in Ficus to
the special way of pollination, which makes it independent of its pol-
linators in growth habit, habitat, and arrangement of the inflorescences
(cf. VAN DER Pm, 1972).
Urticales, Their Differentiation and Systematic Position 369

The other complex is linked to the occurrence in more or less open

(mesic to semi-arid) habitats, chiefly tropical but also subtropical to
temperate (see p. 353). This group of plants comprises the majority
of the Urticales (Ulmaceae, Urticaceae, and many taxa of the Moraceae)
and shows a wide variation in habit (arborescent to herbaceous) and
inflorescences (showing adaptation to wind pollination or not, seldom
pseudanthous), the least reduced staminate flowers and mostly special-
ized stamens. The fruits are dry or succulent, microspermous or mega~
spermous. The plants are essentially light-demanders and are often
pioneers. In several of its characters, especially in those of the inflores-
cences (see p.367), the urticaceous genus Elatostema tends towards
the group of Moraceae belonging to the first complex.
The Conocephaloideae, a diver£le group of light-demanding plants
with aerial or stilt roots, cannot 'be satisfactorily placed in either of
the two complexes. It comprises three groups of morphologically and
ecologically vicarious genera: Coussapoa and Poikilospermum, Pou-
rouma and Myrianthus, and Musanga and Cecropia. Cecropia alone
can be fitted into the second complex, as it exhibits clear adaptations
to wind pollination, which are lacking in the other genera.
The four, more or less exceptional genera mentioned above (Ficus,
Dorstenia, Elatostema, and Cecropia) demonstrate that abundant
speciation, occurrence in a wide range of habitats, and considerable
morphological diversity do not depend solely on structures well-adapted
to pollination and dispersal. These four genera are the largest or second
largest ones of the subdivisions to which they belong.
Ficus and Cecropia can be regarded as well-adapted to insect and
wiIld pollination, respectively. They are well adapted for dispersal
and they occupy a wide or relatively wide range of habitats. Not only
successful adaptation, but the fact of having evolved rather complicated
and unique structures to which they owe successful pollination and
dispersal, seem to promote speciation and diversification. In contrast
with Ficus and Cecropia, developments in Dorstenia have not led to
successful adaptations either to (insect) pollination or to dispersal.
Nevertheless, this genus comprises numerous species, occupies a wide
range of habitats, and shows considerable diversity, expecially in the
inflorescence and growth habit. The strange patterns in the differentia-
tion of the inflorescence, the switching over to asexual (apomictic
and/or vegetative) reproduction, and the considerable cytological
differentiation (cf. LE COQ 1963, FEDOROV 1969) can be regarded as
responses to adaptive failures in the reproductive structures.
Elatostema is the urticaceous parallel of Dorstenia (see p.367).
The combination of the discoid-involucrate inflorescence and reflexed
stamens seems to be an important disadvantage with regard. to pol-
Plant SYilt. Evol .• Suppl. 1 24
370 c. C. BERG:

lination. While the trends in differentiation led Dorstenia outside the

rain forest, the reverse apparently occurred with Elatostema.
For the M oraceae STEBBINS (1974) assumed a reversal from wind
pollination to insect pollination. This assumption is not supported
by the differentiation patterns found in the Urticales. STEBBIN'S as-
sumption appears to be based on the consideration that (most 1) angio-
spermous groups arose under semi-arid conditions and that their
presence in tropical rain forest is secondary.
Interpretation of morphological (and anatomical) differentiations
and reading of morphological series depend largely on the nature of
the conditions (humid tropical, humid subtropical, montane, semi-
arid) which are accepted as being those under which the Angiosperms
or angiospermous groups originate.d. Postulating these conditions
seems to be easier for those groups which have retained primitive
characters than for old groups with derived characters.
In the Urlicales, the most derived and advanced forms of flowers
and inflorescences and, on the other hand, the presumed primitive
and less derived growth habits are bound up with humid tropical low-
land conditions.

Position of the Urticales

In recent classifications of the Angiosperms the Urticales can be
found in a position either near the Hamamelidales (TAKHTAJAN 1969,
CRONQUIST 1968) or near the Malvales (THORNE 1968, STEBBINS 1974,
DAHLGREN 1975).
The reduced monochlamydeous flower, the common occurrence of
anemophily, and the presence of catkins have tied the Urticales to the
"Amentiferae", especially to the Fagales. Many of these amentiferous
orders have been connected with the Hamamelidales and of these rela-
tionships, that of the Fagales with the Hamamelidales is established
and generally accepted.
Most arborescent Urlicales inhabiting subtropical to temperate
regions resemble representatives of the Fagales in growth habit and
morphology of the shoots and leaves. TIPPO (1938) concluded from the
results of a comparative anatomical study of the wood from members
of the Urticales and their presumed allies that the Urticales can be
considered to be,derivatives of the Hamamelidales.
On the other hand, differences in the structure of and variation in
tha inflorescence (ENDRESS 1970), fruit, and seed do not support the
presumed relationship between the Urticales and Hamamelidales and
their allies. Furthermore, a 4-(5-)merous flower with a single whorl
of small sepaloid perianth parts, bicarpellate pistils, and adaptations
Urticales, Their Differentiation and Systematic Position 371

to wind pollination appear in reduction series in separate groups of the

Angiosperms. Similarities in growth habit and morphology of the
shoots and leaves can be ascribed to similarities in environmental
conditions. Phytochemical differences between the Urticales and Ha-
mamelidales shed doubt on the presumed relationship of those groups
as well (cf. MEEUSE 1975).
In his monograph of the Urticaceae WEDDELL (1856-1857) discussed
several characters indicating affinity of this family with the Malvales,
especially with the Tiliaceae. In commenting on the ENGLER system
of the Angiosperms HALLIER (1903) found in the characters of the
Urticales more indications for a connection with the Malvales and
Euphorbiaceae than with the Hamamelidaceae, Fagaceae, and Betulaceae.
In summary, the reasons most of which have already been used
by WEDDELL and HALLIER for considering the Urticales to be related
to the Malvales are: the presence of bast fibres and mucilaginous cells
and ducts in both orders, the common occurrence of connate stamens
(revealed in the Ulmaceae by the anatomical study of BECHTEL 1921)
and valvate tepals in the Urticales and the occurrence of unilocular
anthers with two pollen-sacs in Ficus subg. Urostigma sect. Malvanthera
(CORNER 1960); similarities between the orders with regard to the
variation in growth habit and leaves (simple or, if deeply incised or
compound, then palmate), and the patterns in variation of the inflores-
cence, fruit, and seed of the Urticales which matches that of the Malvales
more clearly than that of the Hamamelidales and their allies.
WEDDELL regarded the differences in the androecium of the Urticaceae
and the Malvales and the persistent perianth in the Urticaceae as con-
tradicting the affinities of the orders. The latter point does not seem
to be very important and the differences in the androecium are less
acute if the androecium of the Ulmaceae and the results of VAN HEEL'S
study on the androecium of the Malvales (1966) are taken into account.
Neither from the anatomical characters of the wood (pers. comm.
Dr. A. M. W. MENNEGA) nor from the chromosome numbers arguments
against the supposed relationship of the orders can be obtained.
Most facts indicate that the Urticales belong to a central complex
in the Angiosperms, comprising at least the Urticales, Malvales, Eu-
phorbiales, and the Violales sensu CRONQUIST (1968). This is largely
in agreement with DAHLGREN'S view (1975). The affinities of the Urti-
cales with the M alvales are closest between the Ulmaceae and Tiliaceae,
the least derived families of the orders. Both orders and the Euphorbia-
ceae provide many tropical pioneer tree species (cf. WHITMORE 1975).
This is connected with differentiation in growth habit and other char-
acters and in this respect the Euphorbiaceae show similarities to the
Urticales. The relation between these groups seems to be collateral.
24·
372 C. C. BERG:

The relationships of the Urticale8 with the Violale8 are rather vague
and point to the Flacourtiaceae, as do many taxa of the central complex
(cf. TAKHTAJAN 1969), and this indicates a key position for that family.

The author is much indebted to Dr. N. G. BISSET for the correction of

the English text and for his valuable critical remarks, to Dr. A. M. W. MEN-
NEGA for her critical reading of the text, to the Netherlands Foundation for
Tropical Research (WOTRO) for the grants which enabled the author to
carry out field studies on MO'I'aceae that form part of the basis of the present
paper. Mr. H. RYPKEMA prepared the drawings.

References
BECHTEL, A. R., 1921: The floral anatomy of the Urticales. Amer. Jour.
Bot. 8, 386-410.
BERG, C. C., 1972: Flora Neotropica Monograph No.7. Olmedieae and
Brosimeae (Moraceae). New York: Hafner.
--- 1973:.Some remarks on the classification and differentiation of Moraceae.
Meded. Bot. Mus. Herb. Utrecht No. 386.
- 1977: The Oastilleae, a tribe of the Moraceae, renamed and redefined due
to the exclusion of the genus Olmedia from the "Olmedieae". Acta Bot.
Need. 26, 73-82.
BERNBECK, F., 1932: Vergleichende Morphologie der Urticaceen- und
Moraceen-Infloreszenzen. Bot. Abhandl. Heft 19.
CARAUTA, J. P. P., VALENTE, M. DA C., and SUCRE, B. D., 1974: Dorstenia L.
(MO'I'aceae) dos Estados da Guanabara e do Rio de Janeiro. Rodriguesia
27, 225-278.
CORNER, E. J. H., 1933: A revision of the Malayan species of Ficus: Oovellia
and Neomorphe. Jour. Malay. Br. Asiat. Soc. 11, 1-65.
1949: The durian theory or the origin of the modern tree. Ann. Bot.
N. S. 13, 367-414.
1960: Taxonomic notes on Ficus LINN., Asia and Australasia. Gard.
Bull. Singapore 17, 368-485.
1962: The classification of MO'I'aceae. Gard. Bull. Singapore 19,187-252.
1967: Ficus in the Solomon Islands and its bearing on the Post-Jurassic
history of Melanesia. Philos. Trans. Roy. Soc. London B. 252, 23-159.
CRONQUIST, A., 1968: The evolution and classification of flowering plants.
Boston.
DAHLGREN, R., 1975: A system of classification of the Angiosperms to be
used to demonstrate the distribution of characters. Bot. Notiser. 128,
119-197.
ENDRESS, P. K., 1970: Die Infloreszenzen der apetalen Hamamelidaceen,
ihre grundsatzliche morphologische und systematische Bedeutung. Bot.
Jahrb. 90, 1-54.
- 1975: Nachbarliche Formbeziehungen mit Hiillfunktion im Infloreszenz-
und Bliitenbereich. Bot. Jahrb. 96, 1-44.
FAEGRI, K., and PIJL, L. VAN DER, 1971: The principles of pollination
ecology. Oxford etc.: Pergamon Press.
FAGERLIND, F., 1944: Die Samenbildung und die Zytology bei agamosper-
mischen und sexuellen Arten von Elatostema und einigen nahestehenden
Gattungen nebst Beleuchtung einiger damit zusammenhangender
Probleme. Kung. Svensk Vet. Akad. Hand!. 21, 1-130.
Urticales, Their Differentiation and Systematic Position 373

FEDOROV, A. A., 1969: Chromosome numbers of flowering plants. Leningrad:

V. L. Komarov Botan. Institute.
GOLENKIN, M., 1894: Beitrag zur Entwicklungsgeschichte der Inflorescenzen
der Urticaceen und Moraceen. Flora 78, 97-132.
GOOD, R., 1974: The geography of flowering plants. London: Longman.
GUSTAFSON, A., 1946: Apomixis in higher plants. Lunds Univ. Arsskr. N. F.
Avd. 2,42, 1-66.
HALLE, F., and OLDEMAN, R. A. A., 1970: Essai sur l'architecture et la
dynamique de croissance des arbres tropicaux. Paris: Masson & Cie.
HALLIER, H., 1903: Uber die Verwandtschaftsverhaltnisse bei ENGLER'S
Rosalen, Parietalen, Myrtifloren und in anderen Ordnungen der Diko-
tylen. Abhandl. Naturw. Hamburg 18,3-98.
HEEL, W. A. VAN, 1966: Morphology of the androecium in Malvales. Blumea
13, 174-394.
JARRETT, F. M., 1959-1960: Studies in Artocarpus and allied genera. Jour.
Arnold Arb. 40, 1-37, 113-155,298-368; 41, 73-140, 320-340.
KWAN KORIBA, 1958: On the periodicity of tree growth in the tropics. Gard.
Bull. Singapore 17, 11-81.
LE COQ, C., 1963: Contribution!,! a l'etude cyto-taxinomique des Moracees et
des Urticacees. Revue Gen. Bot. 70, 385-426.
MEEusE, A. D. J., 1975: Floral evolution in the Hamamelididae. III. Hama-
melidales and associated groups including Urticales, and final conclusions.
Acta. Bot. Neer!. 24, 181-191.
MILLINGTON, W. F., and CHANEY, W. R., 1973: Shedding of shoots and
branches. In: Shedding of plant parts (KOZLOWSKI, S. T.), 149-204.
New York-London: Academic Press.
PLANCHON, J. E., 1848: Sur les Ulmacees et la famille des Urticees. Ann.
Sci. Nat. Paris Bot. III. 10, 244-341.
PIJL, L. VAN DER, 1972: Principles of dispersal in higher plants. Berlin-
Heidelberg-New York: Springer.
RAVEN, P. H., 1975: The bases of Angiosperm phylogeny: Cytology. Ann.
Missouri Bot. Gard. 62, 724-764.
RICHARDS, P.jW., 1972: The tropical rainforest. Cambridge: University Press.
Roux, J., 1968: Sur Ie comportement des axes aeriens chez quelques plantes
a rameaux vegetatifs polymorphes; Ie concept de rameaux plagiotrope.
Ann. Sci. Nat. Paris. Bot. XII. 9, 109-256.
SHARMA, M. R., 1965: Morphological and anatomical investigations on
Artocarpus FORST. III. The flower. Phytomorphology 15, 185-201.
SMITH, C. A., 1963: Shoot apices in the family Moraceae with a seasonal study
of Madura pomi/era (RAF.) SCHNEID. Bull. Torrey Bot. Club 90,237-258.
STEBBINS, G. L., 1974: Flowering plant. Evolution above the species level.
Cambridge, Mass.: Belknap Press.
TAKHTAJ AN, A., 1969: Flowering plants. Origin and dispersal. Edinburgh:
Oliver & Boyd.
THORNE, R. F., 1968: Synopsis of putatively phylogenetic classification of
the flowering plants. Aliso 6, 57-66.
TIPPO, 0., 1938: Comparative anatomy of the Moraceae and their presumed
allies. Bot. Gaz. 100, 1-99.
TOMLINSON, P. B., and GILL, A. M., 1973: Growth habits of tropical trees:
Some guiding principles. In: Tropical forest ecosystems in Africa and South
America: A comparative review (MEGGERS, B.J.,AYENSU,E. S., and DUCK-
WORTH, W. D.), 129-143. Washington: Smithsonian Institution Press.
374 C. C. BERG: Urticales, Their Differentiation and Systematic Position

WEBSTER, G. L., 1967: The genera of Euphorbiaceae in the Southeastern

United States. Jour. Arnold Arb. 48,303-430.
- 1975: Conspectus of a new classification of the Euphorbiaceae. Taxon 24,
593-601.
WEDDELL, H. A., 1856-1857: Monographie de la familIe des Urticees. Arch.
Mus. Hist. Nat. Paris 19, 1-591.
WHITEHEAD, R., 1969: Wind pollination in the Angiosperms: Evolutionary
and environmental considerations. Evolution 23, 28-35.
WHITMORE, T. C., 1975: Tropical rain forests of the Far East. Oxford:
Clarendon Press.

Address of the author: Dr. C. C. BERG, Institute for Systematic Botany,

State University Utrecht, Transitorium II, Heidelberglaan 2, Utrecht,
The Netherlands.
Plant Syst. Evol., Suppl. 1, 375-395 (1977)
«:l by Springer-Verlag 1977

Systematisch-Geobotanisches Institut, Universitat Gtittingen,

Federal Republic of Germany

New Aspects of the Systemat.ics of Asteridae

By

Gerhard Wagenitz, Gottingen

Abstract: Five main groups can be distinguished in the Asteridae:

1. Gentianales-Rubiales (and OleQles?); 2. Polemoniales (including Solanaceae
and Boraginaceae); 3. Scrophulariales-Lamiales; 4. Dipsacales; 5. Campa-
nulales-Asterales. The Loasaceae, Fouquieriaceae, and Columelliaceae have
several characteristics of the Asteridae but can scarcely find their place in
this group as circumscribed here. Several plants with much reduced flowers
have to be included in the Asteridae: Theligonum in the Rubiales; Callitriche,
Hippuris and perhaps also Hydrostachys in the Scrophulariales-Lamiales
group. For the Scrophulariales-Lamiales and the Dipsacales, affinities
with the woody Saxifragales and Cornales are most probable and several
genera have been shifted between these groups. The Asterales show remark-
able phytochemical agreements with the Araliales. Several authors have
proposed a breaking up of the Asteridae but the evidence for this does not
seem conclusive.

Introduction
The Asteridae comprise the "core" of the old Sympetalae or Garno-
petalae. The name and the status as a subclass Asteridae are only 12 years
old (TAKHTAJAN 1964) but the group is not a new one, being equivalent
to a great extent to the "Sympetalae, Reihe Haplosternones" (EICHLER
1875), "Anisokarpe Gamopetalen" (GOEBEL 1882) and the "Sympetalae-
Tetracyclicae" (WARMING & MOBIUS 1929). The Asteridae have been
accepted as a taxonomic and phylogenetic unit in the systems of CRON-
QUIST (1957, without name; 1968), TAKHTAJAN (1959, "Oberordnung
Tubillorae"; 1966-1973), and STEBBINS (1974). In several other recent
systems families of the Asteridae may be found dispersed in two or more
places implying a multiple origin of the group (HUTCHINSON 1926-1973,
EMBERGER 1960, S06 1961-1975, THORNE 1968, DAHLGREN 1975).
I shall discuss here the characters of the Asteridae, their main groups,
the position of some aberrant families, and lastly the question of af-
finities with one or more groups outside the Asteridae.
376 G. WAGENITZ:

Characters and Tendencies Common to the Asteridae

According to CRONQUIST (1968) about a third of the species of
dicotyledons belong to this subclass. A calculation on the basis of the
species number!;! given in the "Syllabus der PfIanzen-Familien" (MEL-
cmOR 1964), with corrections for some families, amounts to about
58,700 species or 34% of the dicotyledons (WAGENITZ 1967). Never-
theless this subclass is more uniform in terms of the floral diagram and
embryological characters than the others.
With very few exceptions the flowers are basically pentamerous
with a connate or much reduced calyx, a sympetalous corolla and one
whorl of stamens alternating with the corolla-lobes and adnate to the
tube. The number of stamens is often reduced to 4 or 2 (more rarely 3
or only 1) in zygomorphic flowers. The gynoecium is syncarpous with
the notable exception of the Apocynaceae and Asclepiadaceae. In most
cases there are only two carpels, but exceptions to this rule are numerous.
Some of the carpels may be sterile, eventually leading to a pseudomono-
merous gynoecium. The most constant character is the structure of the
ovules which are unitegmic, tenuinucellate and anatropous (or hemi-
anatropous). Very few exceptions are known: campylotropous ovules
are said to occur in the Acanthaceae and almost orthotropous ones in
Avicennia of the Verbenaceae (DAVIS 1966), while the ovule of Phryma
reported earlier as orthotropous (COOPER 1941) is hemianatropous
(WmPPLE 1972). The ovules of the Oonvolvulaceae are said to be bitegmic
and tenui- or crassinucellate by DAVIS (1966). The structure of the
nucellus is apparently variable in thTs family but the ovules are unitegmic
(see SCHNARF 1931, GOVIL 1970), the statement to the contrary in DAVIS
must be due to a mistake. Not constant for the group and not restricted
to it, but more common here than in any other subclass is the cellular
endosperm (in the following "cellular endosperm" is used as an ab-
breviation for"ab initio cellular endosperm formation").
The morphological trends (progressions) of the flowers in the Aste-
ridae are known from other subclasses too. The following are of para-
mount importance:

1. Flowers (corolla) regular ~ irregular

2. Reduction in the number of stamens
3. Reduction in the number of ovules
4. Ovary superior ~ inferior

If flower morphology is surveyed in conneotion with pollination

ecology three diverging tendencies are especially noteworthy. One is
found in many members of Scrophulariaceae, Gesneriaceae, Acanthaceae,
and Labiatae (Lamiaceae) exhibiting rather large showy zygomorphic
New Aspects of the Systematics of Asteridae 377

flowers which are adapted to more or less restricted groups of pollinators.

The structure of the flower guarantees that the pollen of the four or
two stamens is deposited on the insects or birds in such a manner that
pollination is effective. Very specialized floral mechanisms have evolved
in these families, e.g. in Salvia (HILDEBRAND 1866, CORRENS 1891,
GRANT & GRANT 1964), Pedicularis (SPRAGUE 1962, MACIOR 1968, 1970)
or Oalceolaria and Diascia (VOGEL 1974). A second trend leads to the
development of long and narrow floral tubes restricting effective visits
to animals with long and thin mouth-parts. Good examples may be
found in the Polerrwniaceae (GRANT & GRANT 1965), Solanaceae, Apocyna-
ceae, and Rubiaceae (SILBERBAUER-GOTTSBERGER & GOTTSBERGER 1975),
but even in some Scrophulariaceae (VOGEL 1954). A third principle has
been also very successful: numerous small flowers are aggregated into
a dense inflorescence which may finally form a pseudanthium in the
sense of TROLL (1928) as in the Oompositae or Dipsacaceae. Besides of
these two families a similar tyPe of inflorescence can be found e.g. in
the Polerrwniaceae (Jporrwpsis sect. Microgilia), Rubiaceae (Oephalan-
thus), Valerianaceae, Globulariaceae, Oalyceraceae, and Oampanulaceae
(Jasione, Phyteuma). Finally one group should be mentioned which in
the dicotyledons is unparalleled in the complication of flower structure:
the Asclepiadaceae with their elaborate system for the transfer of pollen
or pollinia by "translators".

Although I am well aware of all the difficulties involved in the calcula-

tion of an "advancement index", I think that in the case of the Asteridae it
gives a good idea of the relatively advanced position of this group. According
to SPORNE (1969, 1974) 38 of the 61 families of dicotyledons with an advance-
ment index of 70% or more belong to the Asteridae and the families of this
subclass have a total range of this index from 57% (Apocynaceae) to 100%
(Oallitrichaceae, Hippuridaceae, Hydrostachyaceae, Phrymaceae), with only
5 families below 70%. The same is shown in the diagram of STEBBINS (1974)
where no order of Asteridae approaches the center.

The Main Groups of the Asteridae

Five main groups (evolutionary lines) seem rather evident in the
Asteridae, taking into account a combination of morphological, embryo-
logical and phytochemical characters (Table 1).

Gentianales-Rubiales (and Oleales)

The families of the Gentianales (Oontortae) have already been brought
together in one order by BARTLING (1830). The close affinity of the
Rubiaceae-distinguished mainly by the inferior ovary and the lack of
intraxylary phloem-has been noted earlier but was formally accepted
W
'I
00

Table 1. Characters distinguishing main groups in the A8teridae

Inflores o Symmetry Iridoid

Leaves Stipules of Flowers Stamens Ovary
cence Endosperm Compounds

1 a. Gentianales opposite + ..... 0 monotelic * isomerous superior nuclear +

(cellular)
b. Rubicile8 opposite + monotelic * isomerous inferior nuclear + ~
(cellular)
c.Oleales opposite 0 monotelic * oligomerous superior cellular + ~
~
2. Polemoniale8 alternate 0 monotelic * (.p isomerous superior nuclear 0
(opposite) (polytelic) _ cellular ~
N
3. Serophularicile8 o opposite or 0 polytelic (isomerous) superior cellular
t +
Lamiales alternate ..... oligomerous (inferior)
4. Dip8aeales opposite 0 monotelic (*) t isomerous inferior cellular +
..... polytelic ..... oligomerous
5 a. Campanulale8 alternate 0 monotelic * t isomerous inferior cellular +0
..... polytelic (oligomerous)
b. A8terale8 opposite or 0 polytelic * t isomerous inferior nuclear or 0
alternate cellular
New Aspects of the Systematics of Asteridae 379

only in recent times (WAGENITZ 1959, CRONQUIST 1968, TAKHTAJAN

1966-1973).
The most important common characters of this group are: leaves
opposite, entire, with stipules (sometimes rudimentary, lacking in the
Gentianaceae), intraxylary phloem present (except in Rubiaceae). Flow-
ers actinomorphic with convolute, valvate or rarely imbricate aestivation,
stamens isomerous. Endosperm nuclear (cellular in few Gentianaceae aud
Rubiaceae). Essential chemical characters are the presence of indole
alkaloids (Loganiaceae, Apocynaceae, Rubiaceae) and cardenolides
(Apocynaceae, especially Echitoideae, Asclepiadaceae). The indole
alkaloids belong to the group of complex seco-iridoids; simple seco-
iridoids and a few iridoids also occur (JENSEN et al. 1975).
The Gentianales-Rubiales are the only group in the Asteridae where
true stipules occur (see WEBERLING 1957 on stipule-like organs in Capri-
foliaceae) and they have the· rare combination of nuclear endosperm
and presence of iridoids. The position of two families each with only
few genera is controversial: the M enyanthaceae agree with the Gentiana-
ceae (in which they have been included for a long time) in chemical
characters (HEGNAUER 1969) but they differ in several anatomical
features (LINDSEY 1938) and in having cellular endosperm (VIJAYA-
RAGHAVAN & PADMANABAN 1969). CRONQUIST (1968) has consequently
transferred this family to the Polemoniales, but they have surely no
close relative here and I still think it is possible that they are an aberrant
member of Gentianales. The Buddlejaceae (at least the best known
genus Buddleja) have morphological (lack of true stipules, HASSELBERG
1937), anatomical, embryological and phytochemical characters which
differ from those of Gentianales and agree with the Scrophulariales
(WAGENITZ 1959). It must be admitted, however, that the borderline
between the Loganiaceae and Buddlejaceae is not as clearcut as it may
appear. This has prompted the monographers of the Loganiaceae not
to accept the Buddlejaceae (LEENHOUTS 1962, LEEUWENBERG 1967),
although they do not doubt the affinities of Buddleja with the Scrophu-
lariaceae, and LEENHOUTS had to admit that for Buddleja "palynology
reveals a close relationship to the Scrophulariaceae and hardly any to
the Loganiaceae" (PUNT & LEENHOUTS 1967).
The family Oleaceae is singular in several respects and this has led
to the fact that it is nearly the only one of the "old" members of Asteridae
whose place in this subclass has recently been debated. TAKHTAJAN
(1969, 1973) has placed the Oleales near the Celastrales (compare MAu-
RITZON 1939). But its embryological characters are those of Asteridae
and so is the floral diagram, although there are a few members with
free petals or without petals. Oleaceae have often been included in the
Gentianales, but they differ in the following characters: leaves sometimes
380 G. WAGENITZ:

divided (true pinnate leaves are very rare in Asteridae! HICKEY &
WOLFE 1975, using leaf characters, include Oleales in the~osidae near
Rhamnales), without stipules, intraxylary phloem lacking, usually
4 petals and 2 stamens (extramely rare in flowers with a strictly regular
corolla), cellular endosperm. CRONQUIST (1968) has Oleaceae in his
Scrophulariales "for a lack of a better alternative", but they differ e.g.
by tbe monotelic (determinate) inflorescence (TROLL 1969) and the regular
corolla, and no family which is near to Oleaceae can be found in the
Scrophulariales. Oleaceae have quite an array of iridoid compounds,
especially seco-iridoids (JENSEN et al. 1975), with one group (Oleuropein
group) apparently restricted to this family. The spectrum of these
iridoids is more consonant with an affinity to the Gentianales. On the
other hand, planteose bas recently been found in seeds of Fraxinus; it is
an oligosaccharide so far only known from several families of the Tubi-
florae s.l. (JUKES & LEWIS 1974). On the whole the erection of a separate
order seems fully justified and it is best treated here as an appendix to
the Gentianales-Rubiales line.

Polemoniales and Scrophulariales-Lamiales

Tbese two groups together correspond to the "Tubiflorae" (+ Plan-
taginales) in the system of ENGLER (1897; see also MELCHIOR 1964 and
WETTSTEIN 1935). In most other systems the families are grouped into

Table 2. Characters of main groups in the Tubiflorae

Polemoniales Scrophulariales-Lamiales

FlowerS actinomorphic ± zygomorphic

(few exceptions)
Stamens isomerous mostly oligomerous
Inflorescence monotelic 1 polytelic 1
(polytelic in most
Oonvolvulaceae)
Endosperm nuclear or cellular cellular
lridoid compounds o + in most families
Alkaloids + in several families o
Stored carbohydrates often starch mainly stachyose and
other oligosaccharides

1 The two main-types of inflorescences distinguished by TROLL (1964,

cf. WEBERLING 1965) correspond to a certain degree to the determinate and
indeterminate types of the older morphologists. Data on the occurrence of
these types in the families of the Tubiflorae may be found scattered in the
publications of EICHLER 1875, DANERT 1958, TROLL 1964, 1969, WEBERLING
1957b (for additional information I thank Prof. WEBERLING).
New Aspects of the Systematics of Asteridae 381

two or more different orders but there is no consensus about their

delimitation. If morphological and phytochemical data are combined,
two main groups emerge (DAHLGREN 1975): Polemonialea with ;Fole-
moniaceae, Hydrophyllaceae, Boraginaceae s.l., Oonvolvulaceae, Solana-
ceae, and Nolanaceae (presumably also the Lennoaceae), Scrophularialea
with the bulk of the TUbi/lorae, and the allied Lamialea with Verbenaceae
(incl. Phrymaceae, WHIPPLE 1972), Oallitrichaceae and Labiatae (Lamia-
ceae). The differences are shown in the following table (Table 2).
The delimitation of the Polemoniales adopted here differs from the
system of CRONQUIST by the inclusion of the Boraginaceae, and from
that of TAKHTAJAN by the inclusion of the Solanaceae. Solanaceae have
often been put near the Scrophulariaceae and WETTSTEIN (1891) wrote
that the border between these families could only be drawn artificially.
But in fact differences in the inflorescence-type (DANERT 1958, TROLL
1964: 177-179), the anatomy (lack of intraxylary phloem in Scrophu-
lariaceae) and in chemical characters make it easy to distinguish between
these families 1. On tl).e other hand, Oonvolvulaceae and Solanaceae both
have intraxylary phloem, chemically allied types of alkaloids (HEG-
NAUER 1973, FROHNE & JENSEN 1973) and they can store starch (sweet
potatoes and potatoes are well known exemples). These two families
together with the Nolanaceae seem to form one group in the Polemonialea,
Polemoniaceae and Hydrophyllaceae a second one, while the Boraginaceae
is rather isolated (this is also in line with anatomical evidence: PATEL &
INAMDAR 1971, INAMDAR & PATEL 1973, and corresponds well to the
division in small orders by HUTCHINSON 1973). The Scrophularialea
with a greater number of families are more homogeneous. The close
coimections between most families of this order have been emphasized
by many authors (e.g. TAKHTAJAN 1959, MELCHIOR 1964, CRONQUIST
1968, IHLENFELDT 1967) and cannot be discussed here. In embryogeny
Scrophularialea and Lamiales seem to be very uniform as compared to
the Polemoniales (YAMAZAKI 1974).

Dipsacales
The Dipsacalea include the Oapri/oli'J,ceae, Valerianaceae, Morinaceae
(VIJAYARAGHAVAN & SARVESHWARI 1968) and Dipsacaceae. Thisrather
homogeneous group was left when the Rubiaceae had been removed from
the old Rubialea (WAGENITZ 1959). Like the Scrophulariales they usually
have zygomorphic flowers, cellular endosperm and iridoid compounds

1 FAIRBROTHERS et al. (1975) interpreted the serological data of HAW-

KES & TURNER (1968) as being in better accordance with the indusion of
Solanaceae in the Scrophulariales. But as only two genera of Solanaceae
(Salpiglossis and Schizanthus) were tested and all the reactions were faint,
it seems hazardous to use this evidence.
382 G. WAGENITZ:

(although mostly of a different type). The ovary is inferior and there is

a strong tendency for the reduction of carpels and of stamens although
the corolla is never as distinctly zygomorphic as in most Scrophulariales.
DAHLGREN (1975) has united the Dipsacales with the Oleales, Goodeniales,
and Gentianales to form his Gentiananae. I should like to stress that in
my opinion the Dipsacales are nearer to the Scrophulariales than to the
Gentianales.

The position of the genera Sambucus (Oaprifoliaceae or Sambucaceae)

and Adoxa (Adoxaceae) is doubtful. The isolated position of the genus Sambucus,
especially as regards morphological and anatomical characters, has been
known since long (HOCK 1892, SCHWERIN 1920, FUKUOKA 1972), it is less
evident in chemical characters (HEGNAUER 1964, BOHM & GLENNIE 1971),
but this may change if we had more information. In embryology the occur-
rence of the rare Adoxa-type of the development of the embryo sac in
Sambucus and Adoxa is remarkable. DAHLGREN has placed Sambucus and
Adoxa in his Oornales, I shall return to this idea later on.

Oampanulales-A sterales
Oampa,nulales are a group recognized as early as 1830 (by BARTLING)
including Goodeniaceae (incl. Brunoniaceae, CAROLIN 1959, DUIGAN 1961),
Stylidiaceae and Oampanulaceae and Lobeliaceae. Only recently this
order has been broken up, e.g. by THORNE (1968) and DAHLGREN (1975).
Mainly on account of the presence of iridoid compounds DAHLGREN has
excluded the Goodeniaceae (incl. Brunoniaceae) and the Stylidiaceae. For
'the Goodeniaceae embryological differences between this family and the
Oampanulaceae are also mentioned (JENSEN et al. 1975), but of these
only the absence of endosperm haustoria seems to me important and
VIJAYARAGHAVAN & MALIK (1972), while favouring the formation of a
separate order Goodeniales, would place it "very close to the Oampanu-
lales". In my opinion the evidence for keeping these families together
(morphological characters, presence of inulin) still prevails.
In a recent article (WAGENITZ 1976) I have made some comments
on the affinities of the Oampanulales and Asterales. The similar pollen
presentation mechanisms and similar morphological tendencies are
noteworthy but might be due to parallelism, but they are supported by
phytochemical agreements : the common occurrence of inulin is well
known, but recently the acetylenes, very characteristic for the Oompositae
(mainly in the Asteroideae), have found to be widespread in the Oam-
panulaceae (BOHLMANN et al. 1973). The Oalyceraceae have been of interest
in this connection as they approach the Oompositae quite closely in some
characters. But there has always been a debate as to their proper place.
TAKHTAJAN (1969) has an order Oalycerales between Oampanulales and
Asterales. With most Oampanulales they have in common the binucleate
New Aspects of the Systematics of Aateridae 383

pollen (only trinucleate pollen is known from the Dipsacales, BREW-

BAKER 1967). Quite recently the detection of secologanin, a simple
seco-iridoid (JENSEN et a1. 1975), has been added to those argum(lnts in
favour of a position in or near the Dipsacales (CRONQUIST 1968, THORNE
1968, DAHLGREN 1975). But surely this is not a well known family and
the same is true for the Pentaphragmataceae and Sphenocleaceae. Ac-
cording to embryological studies (KAPIL & VIJAYARAGHAVAN 1965,
MAHESHWARI & KAPIL 1967) these two monogeneric families are cor-
rectly placed in the Oampanulales, but phytochemical data are totally
lacking.
Aberrant Genera Recently Added to the Asteridae
As stated above the main groups of Asteridae have a good set of
common characters and have been put into this subclass (or its older
equivalents) by numerous authors. But there are several genera or small
families which agree with the Asteridae in most of the "key characters"
but are quite aberrant in several others. According to floral morphology,
they can be put into'two different categories.

Families of Plants With Well-developed Perianth

a) Loasaceae. The Loasaceae have a combination of embryological
characters which is rarely (if ever) found outside the Asteridae: unitegmic,
tenuinucellate ovules, cellular endosperm with the development of ter-
minal haustoria. This makes them highly anomalous in the Parietales
(or Oistales, Violales, Passi/lorales), their traditional place (SCHNARF
1933). In line with the embryological characters is the occurrence of
iridoids (KOOIMAN 1974, JENSEN et a1. 1975). TAKHTAJAN (1959, 1973)
included Loaaaceae in the Polemoniales, but this order has no iridoids
according to present knowledge, and of course there are several con-
spicuous differences in flower morphology. Aberrant in comparison with
all groups of the Asteridae is the tendency to increase the number of
stamens: there are 4-5 stamens, alternating with the free (!) petals
(DANDY 1967) only in the small subfamily Gronovioideae. The variability
of, the androecium is astonishing: recently it has been shown that
Mentzelia differs strikingly in the development of the groups of stamens
from three other investigated genera of Loasaceae (LEINS & WnmARD
1973). I find it impossible to include this family in any of the existing
orders of Asteridae, DAHLGREN had apparently the same impression
when erecting a separate superorder.
b) Fouquieriaceae. Members of this small family with two genera and
eleven species have the following characters which might point to an
affinity with the Asteridae: sympetalous corolla, tenuinucellate ovules
(although with two integuments I), cellular endosperm, presence of several
384 G. WAGENITZ:

iridoid substances (DAHLGREN et al. 1976). In many systems the Fou-

quieriaceae have been placed in the Parietales or a smaller group of this
affinity (Violales, Tamaricales) , in modern systems e.g. by TAKHTAJAN
(1959-1973), CRONQUIST (1968) and HUTCHINSON (1973), but the char-
acters mentioned above would be very aberrant here. The bitegmic ovule,
number of stamens, iridoid substances make the genus equally anomalous
in the Polemoniales (MELCHIOR 1964 has it near the Polemoniaceae,
THORNE 1968 in hls Solanales). DAHLGREN et ai. (1976) have thus
proposed placing the Fouquieriaceae in a monotypic order near the Eri-
cales and Oornales.
c) Oolumelliaceae. This monogeneric family has been referred to the
Tubillorae by many authors since the end of the last century (FRITSCH
1894, WETTSTEIN 1935, MELCHIOR 1964, HUTCHINSON 1926-1973,
TAKHTAJAN 1959-1969). THORNE (1968) has included it in his Saxi-
Iragaceae s.1., CRONQUIST (1968) in the Rosales. CRONQUIST was already
using information from the work of STERN et aI., published 1969, on an
exhaustive investigation of the anatomy. No close relation could be
detected, but the anatomical evidence was in favour of a connection
with the woody Saxilragales. No detailed embryological or phytochem-
ical investigations are available, only GIBBS (1974) seems to have made
a few simple tests, the most remarkable being a strong positive reaction
to the tannin-test (unusual for the Tubillorae).
d) Hoplestigmataceae. This farr..ily has been placed near the Bora-
ginaceae in several recent systems. The basis for this seems to be rather
shaky. Hoplestigma has many petals and 20-30 stamens free from the
corolla to mention only the most conspicuous differences from the
Asteridae. Other places to which it has been assigned are: Bixales
(HUTCHINSON 1973), Violmes (CRONQUIST 1968), Ebenales (GILG 1908,
WAGENITZ in MELCHIOR 1964). As no recent investigations and no data
at all on the embryology and phytochemistry are known to me, a discus-
sion seems rather pointless.

Families of Plants With Reduced Flowers

One of the more conspicuous differences between the modern concept
of Asteridae and equivalent older groupings is the inclusion of several
monogeneric families of plants without a corolla: Theligonaceae, Hip-
puridaceae, Hydrostachyaceae, and Oallitrichaceae. This has been discus-
sed in some detail elsewhere (WAGENITZ 1975), and it may suffice to
repeat here the main arguments for the new systematic position of these
four groups in a tabulated form (Table 3).
A leading role in the assignment of these families to the Asteridae
has been played by embryological characters, especially the unitegmic-
>cI
[ Table 3. Plants with reduced flowers recently added to the Asteridae
U>
';;l
r"
Main evidence Main differences to allied
:;; Family Old position New position for new position 1 families (in new position)
¥-
U>

~
" Theligonaceae Caryophyllales Rubiales Endosperm nuclear Unisexual flowers ~
Myrtales iridoid substances simple perianth
opposite leaves with stipules male flowers with 6-30 stamens
~
CD
a
raphides present ovary with 1 ovule &l"
....o
H ippuridaceae Myrtales Scrophulariales Endosperm cellular Apetalous
(H aloragales) iridoids: aucubin, catalpol ovary inferior ~
stachyose 1 carpel (?) with one ovule

H ydrostachyaceae near Rosales Scrophulariales Endosperm cellular Unisexual flowers

micropylar haustorium perianth 0
5·
type of embryo development two styles
f
rn
o
median stipules (?) ....
~
Callitrichaceae Euphorbiales Lamiales Endosperm cellular Unisexual flowers
Myrtales haustoria present perianth 0
~
(Haloragales) iridoids: aucubin, catalpol carpels transverse t
opposite leaves two styles
type of glands

1 For all groups must be added here: unitegmic-tenuinucellate ovules!

.,...
~
00
at
386 G. WAGENITZ:

tenuinucellate ovules (PHILIPSON 1974 has Hippuridaceae and Theli-

gonaceae in his group with unitegmic-crassinucellate ovules, but compare
JUEL 1911 and WUNDERLICH 1971). Only in the Theligonaceae have
phytochemical researches, testifying the presence of iridoids (KOOIMAN
1971), given rise to an investigation and evaluation of morphological,
anatomical and embryological characters. For Hydrostachys phyto-
chemical data are still lacking; they would be most welcome to corrob-
orate the evidence from the thorough investigations of morphology and
embryology by RAUH & JAGER-ZURN (1966).

Affinities of the Asteridae

The idea that the Sympetalae are not a natural group is an old one
and for several decades taxonomists have looked for affinities between
the orders of Sympetalae and non-sympetalous groups. If only floral
morphology is taken into account this is scarcely more than a guess and
some of the older suggestions as to the origin of groups of the Asteridae
are clearly obsolete. I can scarcely see any foundation for the idea that
the Geraniales (BESSEY 1915, WERNHAM 1913) or more specifically the
Linaceae (HALLIER 1912) occupy a key position as a starting point for
the Tubiflorae. For the Polemoniaceae recently KAPIL et al. (1969) have
again brought forward this proposition, but in fact they advance more
embryological differences than morphological similarities! I think we
'should also dismiss the long cherished idea of a direct connection between
the Umbelliflorae and Rubiales. This seems to have been suggested by a
superficial resemblance of the ovary in derived members of these groups:
Umbelliferae (Apiaceae) and the Galieae of the Rubiaceae.
Recently chemical data together with embryological and morpho-
logical characters help to give a better foundation to the ideas on the
affinities of some groups of the Asteridae. One complex of families gets
more and more interest in this respect: the woody Saxifragales (Ouno-
niales, sometimes included in Rosales) and Oornales. If the broad concept
of Oornales proposed by HUBER (1963) and DAHLGREN (1975) is accepted,
we have to deal with these Oornales alone. In these families several
characters can be found which are familiar to us as typical for the
Asteridae: unitegmic-tenuinucellate ovules occur in most Escalloniaceae
and Hydrangeaceae, while unitegmic-crassinucellate ovules are widespread
in the Oornales s. str. (PHILIPSON 1974), cellular endosperm has been
found in the Hydrangeaceae, Davidiaceae, and Oornaceae (and in several
herbaceous groups of the Saxifragales); finally iridoid compounds are
present in the Escalloniaceae, Hydrangeaceae and many Oornales. The
flowers are not very specialized; they show a strong tendency towards
New Aspects of the Systematics of Asteridae 387

Table 4. Groups of uncertain systematic position shifted between the

Saxifragales·Oornales and Asteridae

Proposed position in:

Saxifragales or Oornales Asteridae

Oolumellia Saxifragaceae Tubiflorae

(Oolumelliaceae) (THORNE 1968) (FRITSCH 1894;
Rosales s. 1. MELCHIOR 1964)
(CRONQUIST 1968) Scrophulariales
Saxifragales (TAKHTAJAN 1966)
(TAKHTAJAN 1973)

Sambucus Oornales Dipsacales-Oaprifoliaceae

(Sambucaceae) (DAHLGREN 1975) (most authors)

Adoxa Saxifragaceae Rubiales

(Adoxaceae) (WARMING & MOBIUS 1929; (WETTSTEIN 1935;
GUNDERSEN 1950) ENGLER & DIELS1936)
Saxifragales Dipsacales
(HUTCHINSON 1973) (WAGENITZ 1959;
Oornales TAKHTAJAN 1966-1973;
(DAHLGREN 1975) CRONQUIST 1968;
THORNE 1968)

Alseuosmiaceae Rosales s. 1. Oaprifoliaceae

(CRONQUIST 1968) (TAKHTAJAN 1969;
Saxifragales HUTCHINSON 1973)
(TAKHTAJAN 1973)
Oornales
(DAHLGREN 1975)

Berenice Saxifragaceae-Escallonioideae Oampanulaceae

(ENGLER 1930) (ERDTMAN &
METCALFE 1963)

Donatia Saxifragaceae Oampanulales-Stylidiaceae

(ENGLER 1890/91) (WAGENITZ in
Saxifragales MELCHIOR 1964;
(HUTCHINSON 1973) CRONQUIST 1968)
Rosales s. 1. . Oampanulales-Donatiaceae
(THORNE 1968) (TAKHTAJAN 1959-1973)
Oornales
(DAHLGREN 1975)

an inferior ovary; and a sympetalous corolla may occur (e.g. Polyosma,

more often in the Crassulaceae belonging to the herbaceous families of
Saxifragales). The comparison of Asteridae with this complex is rendered
difficult by the fact that the Saxifragales and Cornales are very diverse
25'
388 G. WAGENITZ:

groups. The "old" Cornales have been chosen by GIBBS (1974) as one of
his examples for "chaos in taxonomy", and the situation is still more
complicated if the broad concept of Cornales proposed by DAHLGREN is
accepted. It is remarkable that several genera have been shifted between
the Saxifragales-Cornales and the Asteridae. Examples for this have
been combined in table 4. The most intimate connection can be seen
between Cornaceae and Caprifoliaceae with Sambucus (and Viburnum 1)
as a link, this has been confirmed by serological investigations (HILDE-
BRAND & FAIRBROTHERS 1970 a, b). Hydrostachys could have been
added to the families in table 4. When discussing the systematic position
of this genus, RAuH & JAGER-ZURN (1966) had to choose between the
Saxifragales and Scrophulariales (a very valuable conspectus of characters
of these groups can be found in this publication).
On a phytochemical basis another connecting line has recently been
drawn between the Araliales and the Asterales, the main point being
the common occurrence of polyacetylenes and sesquiterpene lactones
(HEGNAUER 1971, 1973, BOHLMANN et al. 1973, WAGENITZ 1976).
These substances are rare in the angiosperms, and if confronted with
this evidence most systematists with some inclination to chemotaxonomy
feel that there "is something in it". In embryology unitegmic-cras-
sinucellate ovules occur in the A raliaceae , unitegmic-tenuinucellate in
Umbelliferae (Apiaceae).
It should be noted here that the two groups to which these "phyto-
chemical connecting lines" go, the Cornales and the Araliales, have been
united in one order in most older systems (and recently re-united in the
system of TAKHTAJAN 1973). And RODRiGUEZ (1971) in a very well
documented discussion of the relationships of the U mbelliflorae speaks
about "a remarkable network of interrelations and similarities across the
gap between the two main lines". In this connection RODRIGUEZ uses
the picture of a puzzle which does not fit together and that is exactly
the feeling I have.
We shall now ask if the evidence discussed above makes it possible to
divide the Asteridae into two (or more) distinct phyletic lines with dif-
ferent affinities. The authors advocating such a division have drawn
the line in different ways and it will not be possible to discuss here all
the proposals which have been made.
One possibility was brought forward by HEGNAUER (1964: 544),
FROHNE & JENSEN (1973: 176) and JENSEN et al. (1975: 169). Using
mainly the presence of iridoid compounds as a guiding principle JENSEN
et al. build up two groups of orders, one (Gentianales, Rubiales, Oleales,
Scrophulariales, Lamiales, Dipsacales) with iridoids is brought in close
connection to the Cornales s.l. (and Ericales); the other group (Pole-
moniales, Campanulales s. str., Asterales), especially the A sterales , show
New Aspects of the Systematics of Asteridae 389

the phytochemical agreements with the Araliales mentioned above. If

the iridoids are thus taken as a key-character the old Tubiflorae have to
be put irito quite different lines and so have the families of the Cam-
panulales. Too much stress is laid now on this very fashionable group of
substances. Their presence, which shows such a surprisingly good cor-
relation with cellular endosperm and unitegmic ovules, is surely an
important character. But what about their absence? As far as is known
the Asclepiadaceae differ from all other families of the Gentianales by the
absence of iridoid compounds, and in the Scrophulariaceae there are
several tribes without these compounds too (KOOIMAN 1970). In these
cases connection with related iridoid-containing groups by common
morphological characters has been so strong, that no one has thought of
excluding these groups. I am not sure that the differences between the
families of Campanulales and the orders of the old Tubiflorae are so
far-reaching as to exclude a common origin. GIBBS (1974) on account of
a broad survey of chemical characters thinks the Tubiflorae s.l. are "an
order of surprising homogeneity". As far as palynology goes it gives
apparently more arguments for than against a connection between the
different groups of the Tubiflorae, and the evidence is at least not
incompatible with an origin in the Rosidae (WALKER & DOYLE 1975).
On the basis of their studies on leaf-characters (venation, tooth-
type, etc.), HICKEY & WOLFE (1975) have quite recently proposed a
division of Asteridae into two groups:
a) Dilleniid-Leafed Asterids (Gentianales, Rubiales, Polemoniales
incl. Solanaceae, Campanulales, Asterales);
b) Rosid-Leafed Asterids (Dipsacales, Scrophulariales, Lamiales).
The Tubiflorae are here divided in the same manner as according
to the chemical characters, but the Gentianales-Rubiales are found "on
the other side of the line". As expressed by their names, the authors
think that the first of these groups is based in the Dilleniidae, the other
in the Rosidae. Taxonomists must surely be thankful,to the authors for
bringing into the discussion a new set of characters, but I feel that it is
going too far to propose such a division on this evidence, especially as it
is evident from the description of the leaf characteristics that the sample
investigated of the Asteridae is still rather small.
In the first attempt to apply numerical methods to the classification
of dicotyledons (YOUNG & WATSON 1970) only the larger families are
treated. For this and other reasons not too much weight should be
attributed to this analysis. But it is interesting to note that in the main
group "Tenuinucelli" there are four secondary groups: 1. "Asclepioids"
corresponding to the Gentianales-Rubiales (and Sapotaceae and several
families of doubtful position), 2. "Acanthoids" which are families of the
Scrophulariales-Lamiales, 3. Compositae and 4. Umbelliferae. Several
390 G. WAGENITZ:

other families of the Asteridae come out as intermediate between 1 and 2.

Unfortunately, the woody families of the Saxifragales have not been
included in this study.
Conclusion
We cannot prove with certainty that the Asteridae are a mono-
phyletic group, and doubts about this viewpoint are possible, but to me
the evidence seems to be equally inconclusive for a distinction of several
phyletic lines of different origin. The pattern of the distribution of
characters cannot be explained without parallel evolution in morpho-
logical, embryological but surely also phytochemical characters. And
we often simply do not know which characters we can rely on as indicating
phyletic affinity or only a certain level of evolution. This is a major
difficulty, and at the same time we have two other problems: still very
inadequate knowledge of the distribiftion of many characters, and the
problem of handling the data. This is too well known to dwell on it
any more. I feel strongly after preparing this lecture that the traditional
method of looking into the literature and compiling the data is unsatis-
factory and that data-banks are necessary. Three titles of articles written
by systematists with a wide experience come to my mind. What is
Systematic Botany? An unending synthesis (CONSTANCE 1964), an
unachieved synthesis (MERXMULLER 1972) or even the Stone of Sisyphus
(HEYWOOD 1974)?
References
BARTLING, F. G., 1830: Ordines nati\H'ales plantarum eorumqll,e characteres
et affinitates. Gottingen: Dieterich.
BESSEY, C. E., 1915: The phylogenetic taxonomy of flowering plants. Ann.
Missouri Bot. Gard. 2, 109-164.
BOHLMANN, F., BURKHARDT, T., and ZDERO, C., 1973: Naturally Occurring
Acetylenes. London-New York: Academic Press.
BOHM, B. A., and GLENNIE, C. W., 1971: A chemosystematic study of the
Caprifoliaeeae. Canad. J. Bot. 49, 1799-1807.
BREWBAKER, J. L., 1967: The distribution and phylogenetic significance
of binucleate and trinucleate pollen grains in the angiosperms. Amer. J.
Bot. 54, 1069-1083.
CAROLIN, R. C., 1959: Floral structure and anatomy in the family Goodenia-
eeae DUMORT. Proc. Linn. Soc. New South Wales 84,242-255.
CONSTANCE, L., 1964: Systematic botany, an unending synthesis. Taxon 13,
257-273.
COOPER, D. C., 1941: Macrosporogenesis and the development of the seed
of Phryma leptostachya. Amer. J. Bot. 28, 755-761.
CORRENS, C., 1891: Zur Biologie und Anatomie der Salvienbliithe. Jahrb.
Wiss. Bot. 22, 190-240.
CRONQUIST, A., 1957: Outline of a new system of families and orders of
Dicotyledons. Bull. Jard. Bot. Bruxelles 27, 13-40.
- 1968: The Evolution and Classification of Flowering Plants. London:
Nelson.
New Aspects of the Systematics of Asteridae 391

·DAHLGREN, R., 1975: A system of classification of the angiosperms to be

used to demonstrate the distribution of characters. Bot. Not. 128,
119-147.
- , JENSEN, S. R., and NIELSEN, B. J., 1976: Iridoid compounds in Fou-
quieriaceae and notes on its possible affinities. Bot. Not. 129, 207-212.
DANDY, J. E., 1967: Loasaceae. In: The Genera of Flowering Plants (Angio-
spermae) (HUTCHINSON, J.). Dicotyledones 2, 353-362. Oxford: Cla-
rendon.
DANERT, S., 1958: Die Verzweigung der Solanaceen im reproduktiven
Bereich. Abh. Deutsch. Akad. Wiss. Berlin, Kl. Chern. 1957; Nr. 6,
1-183.
DAVIS, G. L., 1966: Systematic Embryology of the Angiosperms. New York
etc.: J. Wiley.
DUIGAN, S. L., 1961: Studies of the pollen grains of plants native to Victoria,
Australia. 1: Goodeniaceae (including Brunoniaceae). Proc. Roy. Soc.
Victoria N. S. 74, 87-109.
EICHLER, A. W., 1875: Bliithendiagramme. Band 1. Leipzig: Engelmann.
EMBERGER, L., 1960: Les vegetaux vasculaires. In: Traite de Botanique
(Systematique). Tome II. (CHAUDEFAUD, M. et EMBERGER, L.). Paris:
Masson.
ENGLER, A., 1890-1891: Saxifragaceae. In: Natiirl. Pflanzenfam. (ENGLER,
A., und PRANTL, K., Hrsg.) III 2 a, 41-93. Leipzig: Engelmann.
- 1897: Ubersicht iiber die Unterabteilungen, Klassen, Reihen, Unterreihen
und Familien der Embryophyta siphonogama. In: N atiirl. Pflanzenfam.
(ENGLER, A., und PRANTL, K., Hrsg.), Nachtrage zum n.-IV. Teil,
341-357. Leipzig: Engelmann.
1930: Saxifragaceae. In: Natiirl. Pflanzenfam. 2. Aufl. 18a, (ENGLER, A.,
und PRANTL, K., Hrsg.), 74--226.
- und DIELS, L., 1936: Syllabus der Pflanzenfamilien. 11. Aufl. Berlin:
Borntraeger.
ERDTMAN, G., and METCALFE, C. R., 1963: Affinities of certain genera
incertae sedis suggested by pollen morphology and vegetative anatomy.
Kew Bull. 17, 249-256.
FAIRBROTHERS, D. E., MABRY, T. J., SCOGIN, R. L., and TURNER, B. L.,
1975: The bases of angiosperm phylogeny: chemotaxonomy. Ann. Mis-
souri Bot. Gard. 62, 765-800.
FRITSCH, K., 1894: Oolumelliaceae. In: Natiirl. Pflanzenfam. IV. 3b (ENG-
LER, A., und PRANTL, K., Hrsg.), 186--188.
FROHNE, D., und JENSEN, U., 1973: Systematik des Pflanzenreichs unter
besonderer Beriicksichtigung chemischer Merkmale und pflanzlicher
Drogen. Stuttgart: G. Fischer.
FUKuoKA, N., 1972: Taxonomic study of the Oaprifoliaceae. Mem. Fac. Sci.
Kyoto Univ. Ser. BioI. 6, 15-58.
GIBBS, R. DARNLEY, 1974: Chemotaxonomy of Flowering Plants. 4 Vol.
Montreal-London: McGill-Queen's Univ. Press.
GILG, E., 1908: Die systematische Stellung der Gattung Hoplestigma und
einiger anderer zweifelhafter Gattungen. Bot. Jahrb. Syst. 40, Beibl. 93,
76--84.
GOEBEL, K., 1882: Grundziige der Systematik und Speciellen Pflanzen-
morphologie. Leipzig: Engelmann.
GOVIL, C. M., 1970: Oonvolvulaceae. In: Symposium on Comparative Embryo-
logy of Angiosperms. Bull. Indian National Sci. Acad. 41, 246-249.
392 G. WAGENITZ:

GRANT, K. A., and GRANT, V., 1964: Mechanical isolation of Salvia apiana
and Salvia mellifera (Labiatae). Evolution 18, 196-212.
GRANT, V., and GRANT, K. A., 1965: Pollination in the Phlox Family.
New York-London: Columbia University Press.
GUNDERSEN, A., 1950: Families of Dicotyledons. Waltham, Mass.: Chronica
Botanica.
HALLIER, H., 1912: L'origine et Ie systeme phyletique des Angiospermes
exposes it l'aide de leur arbre genealogique. Arch. Neerl. Sci. Exact. Nat.
ser. III. B. 1, 146-234.
HASSELBERG, G. B. E., 1937: Zur Morphologie des vegetativen Sprosses der
Loganiaceen. Symb. Bot. Upsal. 2 (no. 3),1-170.
HAWKES, J. G., and TUCKER, W. G., 1969: Serological assessment of rela-
tionships in a flowering plant family (Solanaceae). In: Chemotaxonomy
and Serotaxonomy, Syst. Ass. Special Vol. 2 (HAWKES, J. G., Ed.),
77-88. London-New York: Academic Press.
HEGNAUER, R., 1964, 1969, 1973: Chemotaxonomie der Pflanzen. Band 3,
5, 6. Basel-Stuttgart: Birkhauser.
1971 : Chemical patterns and relationship of Umbelliferae. In: The Biology
and Chemistry of the Umbelliferae (HEYWOOD, V. H., Ed.), 267-277.
London: Academic Press.
HEYWOOD, V. H., 1974: Systematics-the Stone of Sisyphus. BioI. J. Linn.
Soc. 6, 169-178.
HICKEY, L. J., and WOLFE, J. A., 1975: The bal;les of angiosperm morphology:
vegetative morphology. Ann. Missouri Bot. Gard. 62, 538-589.
HILDEBRAND, F., 1866: Uber die Befruchtung der Salvia-Arten mit Hiilfe
von Insekten. Jahrb. Wiss. Bot. 4, 451-478.
HILLEBRAND, G. R., and FAIRBROTHERS, D. E., 1970a: Serological investiga-
tion of the systematic position of the Oaprifoliaceae, I. Correspondence
with selected Rubiaceae and Oornaceae. Amer. J. Bot. 57, 810-815.
- - 1970 b: Phytoserological systematic survey of the Oaprifoliaceae.
Brittonia 22, 125-133.
HOCK, F., 1892: Zur systematischen StellUng von Sambucus. Bot. Centralbl.
51, 233-234.
HUBER, H., 1963: Die Verwandtschaftsverhaltnisse der Rosifloren. Mitt.
Bot. Staatssamml. Miinchen 5, 1-48.
HUTCHINSON, J., 1926: The Families of Flowering Plants. I. Dicotyledons.
London: Macmillan.
- 1973: The Families of Flowering Plants. Ed. 3. London: Oxford Univer-
sity Press.
IHLENFELDT, H.-D., 1967: Uber die Abgrenzung und die natiirliche Glie-
derung der Pedaliaceae R. BR. Mitt. Staatsinst. Allg. Bot. Hamburg 12,
43-128.
INAMDAR, J. A., and PATEL, R. C., 1973: Structure, ontogeny, and classifica-
tion of trichomes in some Polemoniales. Feddes Repert. 83, 473-488.
JENSEN, S. R., NIELSEN, B. J., and DAHLGREN, R., 1975: Iridoid compounds,
their occurrence and systematic importance in the angiosperms.
Bot. Not. 128, 148-180.
JUEL, H.O., 1911: Studien liber die Entwicklungsgeschichte von Hippnris
vnlgaris. Nova Acta Regiae Soc. Sci. Upsal., ser. IV. 2, nr. 11.
JUKES, C., and LEWIS, D. H., 1974: Planteose, the major soluble carbo-
hydrate of seeds of Fraxinns excelsior. Phytochemistry 13, 1519-1521.
New Aspects of the Systematics of Asteridae 393

KAPIL, R. N., RUSTAGI, P. N., and VENKATARAMAN, R., 1969: A contribu-

tion to the embryology of Polemoniaceae. Phytomorphology 18, 403-412.
and VANI, R. S., 1967: Nyctanthes arbor-tristis L.: embryology and
relationships. Phytomorphology 16, 553-563.
- and VIJAYARAGHAVAN, M. R., 1965: Embryology of Pentaphragma
horsfieldii (MIQ.) AIRY SHAW with a discussion on the systematic position
of the genus. Phytomorphology 15, 93-102.
KOOIMAN, P., 1970: The occurrence of iridoid glycosides in the Scrophula-
riaceae. Acta Bot. Neer!. 19, 329-340.
- 1974: Iridoid glycosides in the Loasaceae and the taxonomic position
of the family. Acta Bot. Neer!. 23, 677-679.
LEENHouTs, P. W., 1962: Loganiaceae. In: Flora Malesiana I. 6 (2), 293-387.
LEEuWENBERG, A. J. M., 1967: Notes on American Loganiaceae II. Revision
of Peltanthera BENTH. Acta Bot. Neerl. 16, 143-146.
LEINS, P., und WINHARD, W., 1973: Entwicklungsgeschichtliche Studien
an Loasaceen-Bliiten. Osterr. Bot. Ztschr. 122, 145-165.
LINDSEY, A. A., 1938: Anatomical evidence for the Menyanthaceae. Amer.
J. Bot. 25, 480-485.
MACIOR, L. W., 1968: Pollination adaptation in Pedicularis groenlandica.
Amer. J. Bot. 55, 927-932.
- 1970: The pollination ecology of Pedicularis in Colorado. Amer. J. Bot. 57,
716-728.
MAHESHWARI, P., and KAPIL, R. N., 1967: Some Indian contributions to
the embryology of angiosperms. Phytomorphology 16, 239-291.
MAURITZON, J., 1939: Die Bedeutung der embryologischen Forschung fiir
das natiirliche System der Pflanzen. Acta Univ. Lund. N. F. Avd. 2.35.
Nr. 15, 1-70.
MELCHIOR, H. (Hrsg.), 1964: A. ENGLER'S Syllabus der Pflanzenfamilien.
12. Auf!. Band 2. Angiospermen. Berlin-Nikolassee: Gebr. Borntraeger.
MERXMULLER, H., 1972: Systematic Botany-an unachieved synthesis.
BioI. J. Linn. Soc. 4, 311-321.
PALIWAL, G. S., and SRIVASTAVA, L. M., 1969: The cambium of Alseuosmia.
Phytomorphology 19, 5-8.
PATEL, R. C., and INAMDAR, J. A., 1971: Structure and ontogeny of stomata
in some Polemoniales. Ann. Bot. (London) 35, 389-409.
PHILIPSON, W. R., 1974: Ovular morphology and the major classification of
the dicotyledons. Bot. J. Linn. Soc. 68, 89-108.
PUNT, W., and LEENHouTs, P. W., 1967: Pollen morphology and taxonomy
in the Loganiaceae. Grana Palyn. 7, 469-516.
RAUH, W., und JAGER-ZURN, I., 1966: Zur Kenntnis der Hydrostachyaceae.
1. Teil, Bliitenmorphologische und embryologische Untersuchungen an
Hydrostachyaceen unter besonderer Beriicksichtigung ihrer systema-
tischen Stellung. Sitzungsber. Heidelb. Akad. Wiss. Math.-Nat. K!.
Jg.1966, 1. Abh., 1-117.
RODRIGUEZ, R. L., 1971: The relationships of the Umbellales. In: The Biology
and Chemistry of the Umbelliferae (HEYWOOD, V. H., Ed.), 63-91.
London: Academic Press.
SCHNARF, K., 1931: Vergleichende Embryologie der Angiospermen. Berlin:
Gebr. Borntraeger.
1933: Die Bedeutung der embryologischen Forschung fiir das natiirliche
System der Pflanzen. BioI. Gen. 9, 271----288.
394 G. WAGENITZ:

SCHWERIN, F. VON, 1920: Revisio generis Sambucus. Mitt. Deutsch. DendroI.

Ges. 29, 194-231.
SILBERBAUER-GOTTSBERGER, r., und GOTTSBERGER, G., 1975: Uber sphingo-
phile Angiospermen Brasiliens. Plant Syst. EvoI. 123, 157-184.
S06, R., 1961: The present aspect of the evolutionary history of Telomo-
phyta. Ann. Univ. Sci. Budapest Sect. BioI. 4,167-178.
- 1967: Die modernen Systeme der Angiospermen. Acta Bot. Acad. Sci.
Hung. 13, 201-233.
- 1975: A review of the new classification systems of flowering plants
(Angiospermatophyta, Magnoliophytina). Taxon 24, 585-592.
SPORNE, K. R., 1969: The ovule as an indicator of evolutionary status in
angiosperms. New Phytol. 68, 55&---566.
- 1974: The Morphology of Angiosperms. London: Hutchinson.
SPRAGUE, E. F., 1962: Pollination and evolution in Pedicularis (Scrophu-
lariaceae). Aliso 5, 181-209.
SPRAGUE, T. A., 1927: The morphology and taxonomic position of the
Adoxaceae. J. Linn. Soc. Bot. 47, 471-487.
STEBBINS, G. L., 1974: Flowering Plants. Evolution above the Species
Level. Cambridge, Mass.: Belknap Press.
STERN, W. L., BRIZICKY, G. K., and EYDE, R. H., 1969: Comparative
anatomy and relationships of Golumelliaceae. J. Arnold Arbor. 50, 36-75.
TAKHTAJAN, A., 1959: Die Evolution der Angiospermen. Jena: VEB G. Fi-
scher.
1964: The taxa of the higher plants above the rank of order. Taxon 13,
160-164.
1966: Sistema i filogenija cvetkovych rastenij (Systema et phylogenia
Magnoliophytorum). Moskva-Leningrad: Nauka.
1969: Flowering Plants. Origin and Dispersal. City of Washington:
Smithsonian Inst. Press.
1973: Evolution und Ausbre,itung der Bliitenpflanzen. Stuttgart:
G. Fischer.
THORNE, R. F., 1968: Synopsis of a putatively phylogenetic classification of
the flowering plants. Aliso 6, 57-66.
TROLL, W., 1928: Organisation und Gestalt im Bereiche der Bliite. Berlin:
J. Springer.
- 1964, 1969: Die Infloreszenzen. Typologie und Stellung im Aufbau des
Vegetationskorpers. 1. Band; 2. Band, 1. Teil. Stuttgart: G. Fischer.
VIJAYARAGHAVAN, M. R., and MALIK, U., 1972: Morphology and embryology
of Scaevola frutescens K. and affinities of the family Goodeniaceae. Bot.
Not. 125, 241-254.
- and PADMANABAN, U., 1969: Morphology and embryology of Gentau-
rium ramosissimum DRUCE and affinities of the family Gentianaceae.
Beitr. BioI. Pfl. 46, 15-37.
- and SARVESHWARI, G. S., 1968: Embryology and systematic position of
Morina longifolia WALL.Bot. Not. 121, 383-;-402.
VOGEL, S., 1954: Bliitenbiologische Typen als Elemente der Sippenglie-
derung dargestellt anhand der Flora Siidafrikas. Bot. Studien 1. J ena:
VEB G. Fischer.
1974: Olblumen und olsammelnde Bienen. Tropische und subtropische
Pflanzenwelt 7. Akad. Wiss. Lit. (Mainz), Math.-Naturw. Kl. Wies-
baden: F. Steiner.
New Aspects of the Systematics of Asteridae 395

WAGENITZ, G., 1959: Die systematische Stellung der Rubiaceae. Ein Beitrag
. zum System der Sympetalen. Bot. Jahrb. Syst. 79,17-35.
1967: Betrachtungen liber die Artenzahlen der Pflanzen und Tiere.
Sitzungsber. Ges. Naturf. Freunde Berlin N. F. 7, 79-93.
1975: Bllitenreduktion als ein zentrales Problem der Angiospermen-
Systematik. Bot. Jahrb. Syst. 96, 448-470.
1976: Systematics and phylogeny of the Compositae (Asteraceae). Plant
Syst. Evol. 125, 29-46.
WALKER, J. W., and DOYLE, J. A., 1975: The bases of a~giosperm phylo-
geny: palynology. Ann. Missouri Bot. Gard. 62, 664-723.
WARMING, E., and MOBIUS, M., 1929: Handbuch der systematischen Botanik.
4. Aufl. Berlin: Gebr. Borntraeger.
WEBERLING, F., 1957a: Morphologische Untersuchungen zur Systematik
der Caprifoliaceen. Akad. Wiss. Lit. Mainz, Abh. Math.-Naturw. Kl.
Jg. 1957, Nr. 1, 1-50.
- 1957b: Die Infloreszenzen von Bonplandia CAV. und Polemonium
micranthum BENTH. und ihre vermittelnde Sonderstellung unter den
Bliitenstanden der Polemoniaceae. Beitr. BioI. Pfl. 34, 195-211.
- 1965: Typology of inflorescences. J. Linn. Soc. Bot. 59, 215-221.
WERNHAM, H. F., 1913: Floral evolution: with particular reference to the
sympetalous dicotyledons. New Phytol. Reprint no. 5. Cambridge.
WETTSTEIN, R. VON, 1891: Solanaceae. In: Natiirl. Pflanzenfam. IV, 3b
(ENGLER, A., und PRANTL, K., Hrsg.), 4-38.
- 1935: Handbuch der Systematischen Botanik. 4. Aufl. Leipzig-Wien:
F. Deuticke.
WHIPPLE, H. L., 1972: Structure and systematics of Phryma leptostachya L.
J. Elisha Mitchell Sci. Soc. 88,1-17.
WUNDERLICH, R., 1971: Die systematische Stellung von Theligonum.
Osterr. Bot. Ztschr. 119, 329-394.
YAMAZAKI, T., 1974: A system of Gamopetalae based on the embryology.
J. Fac. Sci. Sect. 3. Bot. (Tokyo) 11, 263-281.
YOUNG, D. J., and WATSON, L., 1970: The classification of dicotyledons:
a study of the upper levels of the hierarchy. Aust. J. Bot. 18, 387-433.

Address of the author: Prof. Dr. GERHARD WAGENITZ, Lehrstuhl fiir

Pflanzensystematik, Systematisch-Geobotanisches Institut der Universitat,
Untere Karspiile 2, D-3400 Gottingen, Federal Republic of Germany.
Plant Syst. Evol., Suppl. 1, 397-405 (1977)
© by Springer-Verlag 1977

Summary Lecture
By

Hermann Merxmiiller, Botanische Staatssammlung Miinchen

In the past decades we have all attended many symposia on micro-

evolution, biosystematics, the species problem and so on, whereas
meetings concerned with the higher categories have been rare events.
This may at least partly be due to the conviction of many botanists
that most problems here are solved already whereas others may con-
sider the creation of a, truly evolutionary system or a generally acceptable
classification to be an unattainable aim in our time. As much as I
belong to this second group of people I thoroughly enjoyed this Con-
ference-especially as it revealed so many signs of concern.
In the late fifties it was the numerical taxonomists who worried
us with the idea that an equally good or even better taxonomy could
be made without the preoccupation of phylogeny-an idea which in
its basic principle ,greatly influenced many of us. Nevertheless there
was no manifest break with tradition as regards higher taxa and their
classification. This was due, I believe, to the fact that at the same
time the CRONQUIST and TAKHTAJAN systems were published, the over-
whelming success of which may have depended partly on this early
sign of detente-so many years before the so-called political one. There
was a kind of truce, just slightly disturbed by the steady grumbling
voices of MEEUSE, MELVILLE and a few others. But the new seriou·s
worry has come from the chemists who as exact scientists could not
accept this terrible mixture of "Science and Art" which taxonomists
are used to tolerate.
We should admit that we ought to be worried and as a matter of
fact we all are-at least in our subconsciousness. HEYWOOD did not
simply play his well-known r6le of enfant terrible, castigating us by
calling our aims a much overrated pastime, charging us with spiral
reasoning, deploring our canalization of conceptual thinking. The
same uneasiness has been expressed at this symposium, perhaps less
forcefully, by KUBITZKI and MEEUSE as well as by CLIFFORD and
SPORNE. Trying to avoid in this Summary Lecture at least some of the
inbuiIt conflicts resulting from unclear definitions, confusion of aims
398 H. MERXMULLER:

and so on, I shall simply use the Conference key-words, i.e. Evolution,
Classification, Higher Categories, and add a few words about some
Higher Taxa, Characters and Methods .
. With regard to Evolution I may begin with one of the so-called
main problems, the old question of Polyphyly versus Monophyly.
My impression was that there is hardly anyone amongst the participants
who'still believes in an extreme polyphyly in the sense as e.g. GREGUSS
has proposed or as LAM has advocated at least at the zenith of his
theory. It may be somewhat symbolic that even the term polyphyly
meanwhile has been slightly changed to "polyrheithry". But on the
other hand, and more surprising, nobody seems to defend also a strictly
monophyletic theory, believing in a common ancestor as having been
one or two single plants. There is an ever increasing amount of consent
about smaller or larger ancestral groups, an idea expressed by ENGLER
already in 1926, formulated as a theory by SUESSENGUTH (1938), called
a "moderate pleiophyly" by SUESSENGUTH & MERXMULLER (1954)
and strongly stressed by CRONQUIST who likes to speak in the last
years about his "rather loose concept of monophyly". Such concepts
fit rather well with examples such as those shown by HUBER with
his groups of monocot lines; they would be compatible with many
facts presented here by EHRENDORFER; and I begin to wonder whether
MEEusE's polyrheithric groups really must have had entirely unrelated
ancestors in his opinion. We should not forget SIMPSON'S remark that
once in the past the orders of today may have been genera only. In
this context the old question of poly- versus monophyly could prove
in the future to have been an overrated, if not a pseudo-problem.
A much more serious problem can be expressed by the statement
that science has made an enormous progress in the understanding
of evolutionary processes but that we understand rather little as yet
about the evolution of real taxa. There are rather few, if any,
cases where we truly succeeded in establishing the real genealogy of a
pair of recent species-and this is naturally even more true for the
higher taxa. We never know whether the last lines leading towards
an extinct base are really converging to one point; BURTT'S paper on
the Gesneriads offered one more splendid example. In such cases one
mostly tries to reconstruct the presumed ancestor by adding together
the so-called primitive traits. .
Both these words "basic" and "primitive", so often used during
this conference, are possibly somewhat dangerous-I already stressed
the fact in an earlier paper how much we are sometimes preoccupied
by the choice of our terms. There are hardly any "basic groups" living
in our time (with the possible exception of some groups leading to
future not yet existing ones which we certainly would not recognize
Summary Lecture 399

as such). On the other hand it is highly improbable that plants pos-

sessing "primitive" characters only live to-day or even have lived in
the past. We probably should reconsider, in a more generalized way,
the term of Hetero bathmy which may be an overall characteristic
of all living things. In such a context it is not at all surprising when
SPORNE'S Theales/Violales are lower in his correlation scheme than
the Magnoliads or when most of these Magnoliads in GOTTWALD'S
wood anatomical diagrams rank higher than some Dilleniids or Rosids.
Even the incredible predominance of unisexuality at the "base" of
SPORNE'S scheme could be understood more easily with the acceptance
of such premises.
As the word "primitive" usually carries with it the connotation
of anthropomorphic terms such as "worse", or at least "imperfect",
"less progressive", etc. it should be replaced by "conserv a ti ve"
(at least for people who do not interpret this term in the same way
as "primitive"). But whatever term we use, we always should keep
in our mind that each plant, past or present, is or has been "perfect",
that it always lives or 'has lived in perfect agreement with its environ-
ment, that it is or has been fully adapted. STEBBINS has emphasized
that it is not the characters but complexes of characters which may
be adaptive-and I would add: complexes of conservative and new
characters together. That is the reason why I consider it useless to
try to reconstruct an extinct plant or group by summing up conserva-
tive traits only. Concerning the term "progress" I may add that one
of the very few general progressive trends in the evolution of Angio-
sperms which I can see, is specialization, interdependence of parts
and -eventually collectivism (as was shown here once more in the Urti-
cales by BERG), developments similar to those of ants, bees and man-and
it is doubtful if everybody would call this a "progress".
A highlight of this Conference have certainly been the new ideas
on evolutionary strategies, based on the interaction of plants
with pollinators and predators. The enormous importance of the pol-
lination syndromes is generally agreed on; but the possible sequence
of the different flower adaptations to different pollen vectors and the
following steps of specialization have rarely been considered in such
great detail as was done here by EHRENDORFER and GOTTSBERGER.
Likewise CRONQUIST'S consideration of the chemotaxonomically fashion-
able metabolic compounds as a sequence of repellants is a fascinating
new idea. These are three concepts which might reasonably and equally
be applied to the great traits of the Angiosperm development-but
the authors do not always work in the same direction.
CRONQUIST explicitly uses his theory for an explanation of what
he thinks to be at least partly asystema tic distributions of the metabo-
400 H. MERXMULLER:

lites concerned: "The same group of repellants may have been explored
and exploited by different taxonomic groups at the same time". GOTTS-
BERGER, besides his nice demonstration of early adaptation and speciali-
zation in the Magnoliids, produced an interesting interpretation of
the secondary polyandry as a secondary adaptation to cantharophily
as well as of oligomerization as an adaptation to sucking pollinators.
But he admits that this theory does not work for all cases, and we
know that at least in the later phases of insect evolution all these factors
worked side by side, often even in very small groups.
It is only EHRENDORFER who believes that his theory (the evolu-
tionary basis of which seems to be perfectly sound and convincing)
is applicable to an overall explanation of the presumed sequence of
the four subclasses concerned (if not for Oaryophyllidae and Liliopsida,
too). I somewhat doubt this absolute and exclusive validity of his
theory. Nature is so intrinsicly interwoven, each small group in such
a close contact with its environment, that such events must have oc-
curred not once but many times. There ma~e the danger of producing
once more grades, not clades. The same may be feared in the case of
CLIFFORD'S scheme of the monocots, even if it is, of course, a phenetic,
not an evolutionary one. Further considerations would lead us, here
too, to the well-known problems of homology, convergence, reversal
and so on, which have been treated by PHILIPSON, SPORNE and others
also-but I now wish to make some remarks on classification.
Classification means "order". In one of his latest papers, HEY-
WOOD sees here the general role of systematics: "It has, besides its
identificatory role, to collate, synthesize and order the data in a syste-
matic fashion so that conclusions of all sort may be drawn".
But in the context of order there remains always GILMOUR'S question:
order-to which purpose? If I want a general reference system, a filing
and retrieval device, I can live comfortably with the alphabetical one
of WILLIS. For teaching I do not know of any better than our CRON-
QUIST/TAKHTAJAN system whose high didactical standard has contri-
buted as much to its great success. as did its few "balloons" which are
so easily to handle in the very reduced teaching standards of today.
The only phenetic system I know is the certainly still imperfect one of
YOUNG & WATSON which nevertheless would merit much more considera-
tion than it has had so far. MEEUSE is the only one who wants to make
a truly phylogenetical system (a term the Organizing Committee deli-
berately excluded from the program) which reflects the phylogenetic
origin and therefore needs the definition of "primitive, basic" features
free from typological views. For this postulate we have no answer,
I fuM. .
The rest of us enjoy "evolutionary systems" which means nothing
Summary Lecture 401

more than the old romantic natural system, with certain arrangements
and sequences, perhaps enriched with hyphens and arrows or with
boxes and boxes-in-boxes. With regard to the two systematic sche-
mes presented at this Conference (DAHLGREN, THORNE) it was interesting
to see that both prefer arrangements where the evolutionary sequences
go from the center of the scheme towards its periphery. Compared with
TAKHTAJAN'S scheme they therefore look somewhat phenetic but
both are of course as supposedly evolutionary as TAKHTAJAN'S one is.
In both schemes I miss a large central hole which ought to represent
the extinct ancestral groups; and in both I see some dangers with
regard to the very differently shaped groups (one may compare the
figures of DAHLGREN'S Saxi/ragales and Gampanulales or of THORNE'S
Rosi/lorae and Lamii/lorae) which may lead to conceptions quite other
than those intended by the authors. Nevertheless, there seemed to
be a rather unanimous agreement on the utility, especially of DAHL-
GREN'S scheme.
But there is another point I want to stress here, namely the infla-
tion of categories between order and class. For their 50-70 orders
of dicots CRONQUIST and TAKHTAJAN need 6 or 7 subclasses and some
13 superorders; THORNE enumerates 20, DAHLGREN even 27 super-
orders. This reminds me a little of the old Englerian times in which
so many infraspecific categories like proles, subvariety, subforma
and so on have been in vogue. We should deliberate whether this super-
ordinal inflation is not another fruitless trial of expressing quite doubt-
ful phylogenetical contexts by box-in-box schemes.
At all events this inflated classification is a perfect means for con-
fusihg the non-taxonomists by the fashionable application of non-
existing nomenclatural rules to these higher categories. I really wonder
how we dare to offer our students such abominable sources of con-
fusion like Magnoliophyta, Magnoliophytina, Magnoliopsida, Magnolianae
and so on. But really intolerable (as has been mentioned several times
in the discussions) is the strict adherence to the typification rule in
the case of orders and superorders with a quite different content. At
this meeting we have been simply ,amused by the necessity of defining
the order "Gornales" every time by the addition of "sensu CRONQUIST"
or "sensu HUBER" or "sensu DAHLGREN"-but I may draw attention
to the fact (citing HEYWOOD once more) that we are working mainly
for non-taxonomists which simply cannot understand that each of
these Gornales-authors means quite a different group of families and
that it is only the Gornaceae that is included by all authors in this
ordcr.
Here we should try to make some steps towards a solution-and
maybe such a solution for both problems could be the increasing use
Plant Syst. Evol., Suppl. 1 26
402 H. MERXMULLER:

of informal ranks as has been proposed several times during the

symposium. HUBER does not like a formal rank to be given to his
monocot groups; BURTT wants no formal taxa between his family
Gesneriaceae and its genera and, if necessary, informal ones only. Could
we not try to make informal "order groups", when necessary of course,
comparable to our species aggregates and similarly freed from rules
and authorships ~ Such procedures could eliminate also some unne-
cessary monsters like the monotypic Rattlesiitlorae or Balanophoranae.
A final word may be said concerning the shape of our higher taxa,
including the families, one of the "basic units" of HEYWOOD. CRON-
QUIST as well as THORNE are in favour of larger entities, mainly for
pedagogical reasons as far as I can see. Here I disagree entirely. I am
in favour of families (and orders) as clean as possible, well circumscribed,
with predictive value, and this is possible only when heterogeneous
material has been removed from them. As an example I may recall
the Wellstediaceae which does not share many characters with the
Boraginaceae which it nevertheless has been attributed to by MELCHIOR;
therefore it did not reappear in any newer scheme. I sometimes wonder
what surprises we might expect if numerical taxonomists were to
start not with the families of the Syllabus but with its so-called sub-
families or even tribes. SPORNE has made some remarks on the same
problem.
This does not mean that I want to recommend an inflation of
.families and orders-why should we not have "free floating" genera
and families or even something like the "Incognitales" proposed by
BAILEY some thirty years ago ~ By the way, such procedures would
at least recall to our minds more often such incredible cases like
Ernblingia which, in our years and in the same paper, has been attri-
buted to the Capparidaceae, Polygalaceae, Sapindales and Goodeniaceae,
r.espectively. This is a rather bad illustration of what some people
like to call Science and Art.
Let me now come to the higher categories, including some
special higher taxa. We have agreed, I believe, that the Order is
one of the few better and more natural categories in use. As nobody
dared to give a definition of this category-as a matter of fact, the
definition of categories other than the so much treated species seems
to be an important goal of future symposia-one may be inclined to
imitate HESLOP-HARRISON defining the order as a group of families
which a renowned author has studied thoroughly and declared to be
coherent. This is not said for fun only but for adding that the next
highlights of this Conference for me have been the presentations of
the Hamamelidales by ENDRESS and of the Urticales by BERG (as well
as of the Gesneriads by BURTT). This in my opinion is systematics
Sunnrnary Lecture 403

at its best-and the authors concerned should forgive me for not

attempting to sum up in an imperfect manner what they have presented
in such a lucid form.
The other (presumably) undisputed category was the Class, which
here means the DicotfMonocot distinction-and which nevertheless
offered one of the great problems of this Conference. One may have
got the impression that the time is near for re-installing the Monocots
in the first position, before the true Dicots-as the older ones amongst
us learned it in their early days. WALKER in his latest palynological
paper has already inserted the Monocots after his Magnoliidae sensu
strictissimo and before the Ranunculidae and the other subclasses.
HUBER has admitted, in the discussion of his present paper, that he
also would prefer to put the borderline between the Monocots and the
Magnoliids on one side, and the true Dicots on the other side. BEHNKE'S
scheme of the distribution of his different types and subtypes of sieve-
element plastids once more was strongly in favour of such a solution
and even the bioElynthetic pathways of the cyanogenic compounds
(HEGNAUER) lead in the same direction, the tyrosin pathway char-
acterizing once more the Monocots and Magnoliids together. The
"undisputed" classes need, I fear, a question mark.
A short remark to the Monocots: we have been offered two different
new systems, HUBER'S one with the cluster of 1~ partly converging
partly diverging strains, and CLIFFORD'S machine-made one with
10 equally well understandable groups. Both groupings look quite
acceptable although both differ greatly from our familiar ones. Any
judgement or decision seems to be premature if not impossible today;
and this is exactly the situation we are faced with in many other cases.
The situation of the Su bclasses has become somewhat deplorable.
It certainly will be useful to retain terms like "dilleniid" and so on
as a kind of key-words, as hints-but a formal use seems to be scarcely
justified. The Magnoliids in the strict sense are just a group of left
overs which as we saw should perhaps be treated together with. the
(informal) Monocots. The Ranunculids are scarcely more than an iso-
lated order or order group, hardly worthy of an higher esteem than
the purified Oaryophyllales which themselves are one of the nicest
and most clear-cut orders we have.
With regard to the Rosidae, Dilleniidae and even the Asteridae
the available information makes a clear-cut separation increasingly
impossible. I may recall DAHLGREN'S remark about the difficulty
of shaping subclasses in the case of Oornales, Ericales and Dipsacales
(considered by him and others as rather nearly related albeit treated
as members of three different subclasses by others), recall WAGENITZ'S
and THORNE'S presentations treating different evolutionary lines
26'
404 H. MERXMULLER:

leading from the Rosidae/Dilleniidae complex to different Asteridae

groups; recall also PIrrLIPSON'S convincing special groups combining
Rosidae and Dilleniidae elements or GOTTWALD'S similarly aberrant
anatomical schemes.
KUBITZKI cited STEBBINS'S remark that, in contrast to animals,
in plants subsequently to the initial differentiation of higher taxa,
extensive adaptive radiation must have occurred, having led to ample
parallelism and convergence. Anthropologists have been using for a
long time the term "Tier-Mensch-"Obergangsfeld" (zone of transition
between animal and man): I begin to wonder whether the less advanced
Rosidae and Dilleniidae and even the early Asteridae may also form
such a zone of transition-and later on only are diverging into rosid,
dilleniid and asterid specializations. Possibly this would fit PHILIPSON'S
scheme, at least to some extent.
There remains only the so-called Hamamelididae. When restricted
to ENDRESS'S group, it is a perfectly good order or order group of early
Rosidae, i.e. a further member of the just mentioned zone of transition.
When, however, this "subclass" is admitted in the sense of EHREN-
DORFER-tlien it would form a typical other (i.e. prior) zone of transition.
Concerning the systematics of special "superorders" and orders
I shall make a few petulant remarks, only as they did not play a pro-
minent role in this Qonference. Certainly, the stars seem to be in favour
of a definite transfer of the Urticales to the Malvales/ Euphorbiales
order group; PHILIPSON has shown us very persuasive groupings like
his Theales/Primulales order group; BEHNKE has produced a con-
vincing scheme of the Caryophyllales relationship. But may I be al-
lowed to point out gently that none of the problems nor of the solu-
tions was entirely new. All have been discussed already fifty or even
hundred years ago, e.g. by HALLIER to name only one-and have
been rejected in more recent systems and therefore forgotten.
But let me turn again to enjoyable matters-and as always, I find
it difficult to express my appreciation adequately. I refer to the papers
concerning Methods and Characters. One may mention the lucid
presentation of numerical methods and their difficulties by CLIFFORD,
the likewise clear and pertinent presentation of the advancement
index and its implications by SPORNE. The same applies to the fascina-
ting lectures of our chemist friends, HEGNAUER'S remarkable intro-
duction to the cyanogenic compounds as possible systematic markers
and GOTTLIEB'S promising approach to the mapping of metabolites
and their graphic presentation. It should be stressed that the latter
contribution is one of the few attempts really to collate and to order,
in a systematic fashion, the obtainable evidence; its importance also rests
upon the fact that it is not based on taxonomic correlations but on
Summary Lecture 405

. chemical and biosynthetical evidence and hence represents what is

vitally needed: an independent criterion.
It was likewise impressive to listen to those lectures which presented
us with the results of yearlong investigations of single characters,
foremost to PHILIPSO~'S extremely remarkable studies on ovules, to
GOTTWALD'S overwhelming amount of wood anatomical data, to
BEHNKE'S convincing resume on his ultrastructural results. DAHLGREN
has shown us a surprisingly high number of new character distribution
maps on the base of his useful scheme. For me, these all have been
great events and I am sure that the publication of these papers will
allow us the welcome opportunity of a more intrinsic study and use
of all the data offered.
I began with some words of concern. After a. week of papers and
discussions I cannot say that this has ceased. But are we in confusion
and despair? I certainly do not believe so. The cry for a data bank
was loud as always-but it was rightly stressed in the discussion that
here also the question "to which purpose" has to be answered first.
What I believe we should do is first to keep in mind some of HEY-
WOOD'S words, to give clear distinctions and definitions of what we
want to do as well as to indicate clearly and distinctly what we have
done. I still consider it extremely important that each of us continues
collecting data and searching for further characters, making revisions
and monographs-all this being somewhat tedious for great spirits
but the only way we can proceed. We certainly shall not refuse further
speculation, we gladly shall accept new ideas, but possibly we should
not try to insert each new fact into the system as fast as possible. If
we do not want to have many different systems we ought to treat
carefully our systems of today. As EHRENDORFER said in one of his
latest papers, "taxonomy is-and hopefully will remain-a compro-
mise between knowledge, tradition and convenience. This convenience
. . . necessarily entails a certain stability ... Our system need not
reflect immediately every new idea about possible evolutionary af-
finities." It is not in spite but because of all these worries that I am
sure you all share my conviction that this allegedly old-fashioned
systematy today is even fuller of fascination than it ever was.

Address of the author: Prof. Dr. H. MERxMuLLER, Botan. Staatssamm-

lung, Menzinger Str. 67, D-8000 Munchen 19, Federal Republic of Germany.
Index to Plant Taxa
Abrophyllum 145 Aquifoliaceae 133, 135, 206, 301
Acacia 208 Araceae 204, 287
Acanthaceae 206, 376 Arales 18, 287, 288, 295
Achatocarpaceae 168 Araliaceae 130-132, 134, 135, 187,
Actinidiaceae 118, 133 260
Adoxa 382 Araliales 17, 132
Aeschynanthus 100 Aralidium 126 seq.
Aextoxicaceae 118 Araucariaceae 246
Agdestidaceae 168 Archaeopteris 241, 242
Ageratum 186 Archidendron 37
Aistopetalum 151 A rctoteae 313
Aizoaceae 38, 168, 220 Arecales 18, 292
Alismataceae 5, 287 Arecidae 91
Alismatales 18 Argophyllum 145
Alismatidae 91 Argyrocalymna 146
Alismidae 87 seq. Aristolochiaceae 5, 164, 258, 286,295
Alnus 330, 338 A ristolochiales 164
Alpineae 99 Artocarpus 351-353, 359, 361, 367
Alpinia 99 Asarum 164, 258
Alseuosmiaceae 130, 133 Ascarina 15
A maranthaceae 168,205 Asclepiadaceae 128, 206, 258, 376,
A maryllidaceae 70 377, 379, 389
Amaryllis 72 Asparagales 279, 290
A mborellaceae 113 Asteraceae 180, 186, 206, 208, 223,
A mentotaxus 247 313
Ampelocera 356 Asterales 17,171,382
Amygdalaceae 258 Asteranae 268
A nacardiaceae 309 Asteridae 17, 27, 28, 128, 132, 186,
A naxagorea 15 375
A ncistrocladaceae 303 Asteroideae 313
Anethum 72 Astilbaceae 144
Angiozanthos 293 A therospermataceae 164
Annonaceae 10, 29, 69, 113, 114, 164, A ustrobaileyaceae 118, 164
205, 214, 215,260, 286, 295 Avicennia 376
Anopterus 145
Antiaris 357, 366 Bagassa 358
Antiaropsis 357, 359, 364 Balanopaceae 312
A pacheria 310 Balanophoraceae 270
Apiaceae 180, 186, 260 Balanophorales 280.
Apiales 17 Balanops 311, 312
Apium 72 Barbeyales 171
Apocynaceae 66, 206, 258, 376, 377, Basellaceae 168
379 Bataceae 158, 168, 171, 175, 264,
A ponogeton 314 279, 281, 308, 309
Index to Plant Taxa 407

Batidaceae 14 Capparaceae 157, 205, 220, 264

Batis 307-309 Capparales 14, 171, 180,309
Batocarpus 358 Capparis 157, 166
Bauera 150, 151 Caprifoliaceae 133, 206, 379, 381,
Begoniaceae 220, 264 388
Berberidaceae 205 Cardiogyne 354
Berisia 149 Cardueae 313
Betula 330, 338 Caricaceae 264
Betulaceae 329, 330, 340 Carpentiera 146
Betulales 171 Carpinus 329, 337, 338, 342
Bleekrodea 359, 362 Carpodetus 146
Boea 98 Caryocaraceae 220
Boeica 101 Caryophyllaceae 168, 180, 305
Boesenbergia 99 Caryophyllales 176, 187, 279
Boraginaceae 121, 128, 135, 206, 208, Caryophyllidae 14, 27, 128, 176, 220,
381 221, 233
Bosqueiopsis 357 Cassytha 309
Brasenia 287 Castanea 330
Brassica 157, 166 Castilla 357, 366
Brassicaceae 157, 206, 264, 270 Casuarinales 171
Bretschneideraceae 264 Caylusea 260
Brexiaceae 131 Cecropia 353 seq.
Brochoneura 114 Celastraceae 206
Bromeliaceae 290, 294 Celastrales 133, 171
Brookea 98 Celtis 351-353
Brosimum 356-358, 368 Centrolepidaceae 281, 294
Broussonetia 353, 354, 359 Centrospermae 168, 180
Brunellia 150, 151 Cephalotaceae 142
Brunoniaceae 382 Cephalotaxaceae 246
Buddlejaceae 379 Cephalotus 143
Bupleurum 72 Ceratopetalum 151
Butomaceae 5 Ceratophyllaceae 258
Butomus 314 Ceratophyllum 288
Buxaceae 311 Cercidiphyllaceae 118, 260
Cercidiphyllales 171, 230
Charadrophila 98
Cactaceae 168, 205, 220 Chenopodiaceae 168, 205
Caesalpiniaceae 170, 177, 206 Chirita 98
Calceolaria 377 Chloranthaceae 10, 15, 113, 118, 164,
Callitrichaceae 49, 381 258, 287
Calycanthaceae 115, 128, 164, 205, Chlorophora 351, 352, 354, 358
214 Chrysobalanaceae 220, 310
Calycanthus 216 Chunia 335
Calyceraceae 377, 382 Cichorieae 313
Cambomba 287 Cichorioideae 313
Cambombaceae 258, 265 Circaeasteraceae 260
Campanulaceae 135, 186, 206, 313, Cistaceae 220
377
Clarisia 357, 366
Campanulales 17, 305, 382, 389
Campanulanae 268 Clavatipollenites 10, 15
Canellaceae 114, 119, 164, 205 Clethraceae 133
Cannabaceae 349 Clusia 219
Cantua 306 Clusiaceae 205
408 Index to Plant Taxa

Cobaea 306 Degeneriaceae 10, 114, 205

Cobaeaceae 128 Diascia 377
Cochlospermaceae 220, 280 Dichapetalaceae 270, 281
Columelliaceae 133, 384 Dicoryphe 331, 342
Columnea 100 Didiereaceae 168
Commelinaceae 92, 102, 103, 204, Didymocarpeae 99, 101, 102
279, 293, 294 Dilleniaceae 118, 119, 121,218,310
Commelinales 316 Dilleniales 171, 260, 265, 310
Commelinidae 91, 92 Dilleniidae 27, 128, 185, 218, 221,
Compositae 4, 133, 135, 136, 377 230
Connaraceae 281, 310 Dioncophyllaceae 303
Convolvulaceae 135, 206, 306, 376, Dioscoreaceae 5, 292
381 Dioscoreales 291, 295
Cordia 210 Dipsacaceae 105, 106, 377, 381
Coriariaceae 171 Dipsacales 135, 279, 313, 381
Cornaceae 130, 131, 133 Disanthus 325, 331, 335, 340, 342
Cornales 132,135,136,171,265,279, Distyliopsis 324, 329
313, 386, 388 Distylium 324, 329-331, 334
Corokia 131 Dodonaea 309
Corsiaceae 84 Doliocarpus 219
Corylopsis 329-331, 335, 341, 342 Dorstenia 350, 352, 359, 361, 367,
Corylus 329, 330, 338 369, 370
Corynocarpaceae 206 Drimys 214,217
Coussapoa 350, 354, 361, 369 Droseraceae 206
Crassulaceae 142-144,171,206,387 Drypetes 264
Crassulales 147, 151 Dysphaniaceae 168
Crineae 72
Crossosoma 310
Crossosomataceae 310 Ebenaceae 121
Ehretiaceae 128
Croton 260
Elaeagnaceae 281
Cruciferae 126
Elaeagnales 258, 280
Cucurbitaceae 205
Elaeocarpaceae 206, 281
Cudrania 354, 357
Elatostema 358, 359, 362, 367, 369,
Cunoniaceae 131, 142, 148-151, 206
370
Cunoniales 151, 386
Emblingia 308, 402
Cupressaceae 246
Empetraceae 133
Cuttsia 145
Enantioblastae 290, 294
Cyclanthaceae 291, 292
Epacridaceae 206
Cyperaceae 295, 315, 316
Epithema 101
Cyperales 18
Eremosyne 143
Cyrillaceae 133
Ericaceae 133, 206
Cyrtandreae 101, 102
Ericales 18, 133, 135, 146, 171, 186,
Cyrtandroideae 98, 100-102
265, 268, 279
Cyrtandromoea 98
Eriocaulaceae 293, 294
Ernestiodendron 249
Dacrydium 236 Escallonia 145
Daphniphyllum 134, 136, 268, 311, Escalloniaceae 130-133, 145-149,
312 206, 386
Davidiaceae 386 Escalloniales 151
Davidsoniaceae 206 Euchareae 72
Davilla 218,219 Eucommia 280
Degeneria 15, 29, 264 Eucommiales 171, 279
Index to Plant Taxa 409

Eucryphiaceae 118, 164, 176 Halophila 315

Eupatorieae 313 Halophytaceae 168
Euphorbiaceae 126, 206, 207 Haloragaceae 175, 176, 206
Euphorbiales 171, 265, 305, 311 Hamamelidaceae 119, 324, 331, 337
Eupomatia 29, 216 Hamamelidales 171, 322
Eupomatiaceae 114, 164, 214 Hamamelididae 27, 128, 185,230
Eupteleaceae 118, 119 Hamamelis 325, 331, 338, 340
Eupteleales 171, 230 Hanguana 290
Exbucklandia 331, 335, 341 H ectorellaceae 168
Hedychieae 99
Fabaceae 177,206,208,260,310 Hedychium 99
Fabales 171, 258, 265, 281 Heliamphora 303
Fagaceae 330, 340, 341 Heliantheae 313
Fagales 171, 322, 329 Helobiae 287,290, 295
Fagus 330 Helwingia 131, 149
Fatoua 350, 359, 362 Heracleum 260
Ficus 350 seq. Hernandiaceae 115, 164
Flacourtiaceae 205, 207, 265 Himantandraceae 114
Flagellariaceae 294 Hippuridaceae 49, 175, 386
Forgesia 145 Hippuridales 279
Fortunearia 331 Holoptelea 352 seq.
Fothergilla 328-330 Hoplestigmataceae 384
Fouquieria 268, 305, 306 Huthia 306
Fouquieriaceae 265, 280, 305, 383 Hydnocarpus 201
Francoa 151 Hydrangea 146
Frankeniaceae 128, 307 Hydrangeaceae 133 seq., 206, 386
Freycinetia 292 Hydrocharitales 280
Fumariaceae 69, 205 Hydrophyllaceae 381
Hydrostachyales 279
Gentianaceae 206 Hydrostachys 388
Gentianales 135, 171,260,279,377 Hymenocardiaceae 349
Geraniales 171, 265
Hypericaceae 303
Gesneriaceae 97, 376
Gesnerioideae 98, 100, 101 Icacinaceae 133-135, 206
Gisekiaceae 168 Ilex 302
Globulariaceae 377 Illiciaceae 118
Glumicalyx 107 Illiciales 170, 258, 260
Gomortegaceae 115 Ipomopsis 222
Goodeniaceae 186, 206, 382 Iridaceae 290
Goodeniale8 279 Iteaceae 206
Gramineae 294 Ixerba 131, 133
Griselinia 130-132
Grossulariaceae 131, 135, 206 Jerdonia 98, 104, 105
Grossulariales 150, 151 Juglandaceae 205, 231
Grubbiaceae 133 Juglandales 171, 265
Guatteria 215 Juncaceae 204, 293, 295, 315, 136
Gunneraceae 164,171, J75, 176 J uncaginaceae 204
Gyrocarpaceae 164 Juncales 296
Gyrostemonaceae 158, 168, 171, 175, Juniperus 204
205, 26~ 279, 281, 307, 309
Kadsura 118
Haemodoraceae 293 Kaempferia 99
Halodule 315 K ielmeyera 219
410 Index to Plant Taxa

K irengeshoma 147 M alesherbiaceae 205

Klugieae 99 seq. Malvaceae 206
Knema 114 Malvales 265, 305
Maquira 358
Marcgraviaceae· 303
Labiatae 128, 376, 381 Martyniaceae 49
Lactuceae 186 M astixia 131
Lamiaceae 206 Matudaea 324 seq.
Lamiales 135, 279, 313, 380, 381 Mayacaceae 293
Lardizabalaceae 118,260 M elastomataceae 206
Lauraceae 69, 113-115, 164, 172, Meliaceae 102
205, 217, 258, 260 Memoralis 359
Laurales 113, 164, 171 Menispermaceae 69, 205
Lebachia 247, 249 M entzelia 383
Lecythidaceae 206 M enyanthaceae 379
Leeaceae 164, 176, 281 Mesogyne 366
Leguminosae 4, 37, 62, 66, 121 Metasequoia 204
Leitneriales 171, 307, 309 M ichelia 164
Lemnaceae 290 Mimosaceae 37,170,177,206,207
Lennoaceae 381 Mitrastemon 260
Liliaceae 92 Mollinedia 217
Liliales 90, 279, 315 Molluginaceae 168, 180, 205
Liliatae 77 Monimiaceae 69, 115, 120, 164, 217
Liliidae 91, 92 Monophyllaea 98, 100, 105
Liliiflorae 290, 292 Montiniaceae 149, 281
Limeum 168 Moraceae 171, 205, 233, 349, 353
Limnanthaceae 264 Morina 106
Limnocharitaceae 5 Morinaceae 381
Linaceae 206, 304 Moringaceae 264
Linanthus 222 Morus 353,354
Lindera 164 Musa 287
Liquidambar 134, 136, 268, 334 Musanga 361, 369
Liriodendron 113 M utisieae 313
Loasaceae 268, 383 Myoporaceae 206
Loasales 279 Myrianthus 350, 361, 369
Lobeliaceae 382 Myricaceae 231
Loganiaceae 379 Myricales 171,307,309
Lonchocarpus 60 M yristica 29
Loranthaceae 260 Myristicaceae 15, 113-115, 164, 258
Loropetalum 331 Myrothamnaceae 329 seq.
Lotoideae 66 Myrsinaceae 304
Loxonieae 100 M yrtaceae 206
Loxostigma 101 M yrtales 171

Maclura 350, 354 Najas 315

Magnolia 15, 113, 164, 215, 220 N andinaceae 205
Magnoliaceae 10, 69, 112, 114, 115, Naucleopsis 352 seq.
164, 172, 205, 214 Nelumbo 265, 287
Magnoliales 111, 112, 164, 171,258 N elumbonaceae 5, 260
Magnoliidae 27-29, 69, 128, 179, Nelumbonales 170
180, 184, 18~ 213, 214, 229 Neostrearia 335
M alaisia 350 N epenthaceae 303
Index to Plant Taxa 411

N epenthales 171 Phyllocladus 235 seq.

Nepenthes 303 Phyllonoma 149, 150
Nephelium 368 Phytolaccaceae 168
N olanaceae 381 Pinaceae 246
Nothofagus 330 seq. Piper 287
Nuphar 287 Piperaceae 118, 270, 287
Nyctaginaceae 168 Piperales 171, 260, 295
Nymphaeaceae 5, 29, 214, 265 Pittosporaceae 130, 133-135, 142,
Nymphaeales 171,280,287,295 260
Pittosporales 310-312
Ocotea 164 Platanaceae 205, 331, 337
Olacaceae 206, 260 Platanus 210
Oleaceae 133, 379 Plecospermum 354
Oleales 171, 279 Plumbaginaceae 304
Oliniaceae 206 Plumbaginales 27,171,176,279
Olmedia 364 Poaceae 168, 204, 315, 316
Onagraceae 206 Poales 18
Oncoba 368 Podocarpaceae 236, 246, 248
Opiliaceae 260 Podophyllaceae 128
Orchidales 291 Podostemaceae 258
Orchidantha 290 Podostemales 280
Ostrya 338, 342 Poikilospermum 350, 354, 361, 369
Ostryopsis 338 Polemoniaceae 135,222,306,377,381
Polemoniales 379-381
Polyceratocarpus 114
Paeonia 219 Polygonaceae 305
Paeoniaceae 310 Polygonales 27, 171, 176, 265, 279
Paeoniales 171 Polyosma 387
Pandanales 18, 292 Pontederiaceae 293, 316
Papaveraceae 69, 205 Portulacaceae 168
Papaverales 170, 180, 260 Potamogeton 314
Parietales 383, 384 Pourouma 361, 369
Parilla 149 Primulaceae 304
Parrotia 329, 330 Primulales 270
Parrotiopsis 328, 330 Proteaceae 102, 103, 206, 222, 232
Passifloraceae 205, 265 Proteales 258, 280, 310
Pedaliaceae 49 Pseudovoltzia 249
Pedicularis 377 Pteridophyta 204
Peltiphyllaceae 144 Pterostemon 148
Pentadiplandraceae 264 Punicaceae 264
Pentaphragmataceae 383 Pyrolaceae 206
Penthorum 143
Peperomia 288
Perebea 350 Quintinia 133, 145
Petiveriaceae 168
Peucedaneae 72 Rafflesiaceae 171, 258, 260
Peumus 115 Rafflesiales 171, 280
Phenakospermum 290 Ranunculaceae 128, 205, 270
Philadelphus 146 Ranunculales 69, 170, 258
Phryma 376 Ranunculidae 180
Phrymaceae 49, 381 Rapateaceae 293, 294
Phyllochlamys 364 Ravenala 290
Phyllocladaceae 247 Rechsteineria 100
412 Index to Plant Taxa

Rehmannia 98 Scyphosyce 359, 367

Resedaceae 158, 206, 260, 264 Selaginaceae 106
Restionaceae 294 Senecioneae 186
Rhabdodendron 166 Simarubaceae 260
Rhamnaceae 1.66, 260 Simmondsia 307,311
Rhamnales 171, 305 Sinowilsonia 329, 331
Rhododendron 126 Siparunaceae 217
Rhodoleia 328, 331, 335 Sloetiopsis 358, 359
Rhoipteleaceae 231 Smyrnieae 72
Rhynchoglossum 98, 101 Solanaceae 135, 136, 206, 306, 377,
Rhynchotechum 101 381
Ribes 131, 133, 145 seq. Solanales 305, 306
Rosaceae 121,171,206-208,220, Solananae 268
270 Sorocea 357, 358, 360
Rosales 171,260,281,310 Sparattosyce 359, 364
Rosidae 27, 128, 176, 185, 186, 220, Sparganiaceae 293, 295
221, 230 Sphenocleaceae 383
Rosiflorae 141 Spiraeoideae 145
Rubiaceae 133, 206, 279, 313, 377 Stegnospermaceae 168
Rubiales 135 Stemonalc8 291
Ruppia 315 Sterculiaceae 206, 260
Rutaceae 121, 134, 206, 231, 260, Stratiotes 314
307 Streblus 359
Rutales 171,260 Strelitzia 290
Streptocarpeae 102
Sagittaria 314 Streptocarpus 98, 100
Salicaceae 270 Stylidiaceae 382
Salicales 171 Stylobasium 308
Salvadoraceae 158, 206, 264 SYCOpSi8 329, 330
Salvia 377
Sambucus 382, 388
Santalaceae 260 Taccaceae 293
Sapindaceae 206, 207, 308 Taccales 295
Sapindales 171,186,187,260,281 Talauma 113 seq., 215, 216
Sapotaceae 206, 389 Tamaricaceae 307
Sarcandra 113, 115 Tamaricale8 171,265
Sarraceniaceae 303 Taxaceae 246, 248
Sarraceniales 170, 279 Taxodiaceae 246
Sassafras 164 Tetracentraceae 113
Saururaceae 118, 287 Tetragoniaceae 168
Saxifragaceae 130, 144, 148, 206, 384 Tetraxylopteris 243
Saxifragales 171, 221, 260, 386, 388 Thalassia 315
Saxifragineae 141, 142 Theaceae 119,121,302
Scabiosa 106 Theales 260, 270
Scaevola 264, 309 Theligonaceae 14, 168, 386
Scheuchzeriaceae 204 Theligonales 279
Schisandra 118 Theligonum 307
Schizophragma 150, 151 Theophrastaceae 304
Scitamineae 287, 290, 294, 295 Thurniaceae 295
Scrophulariaceae 98, 105, 206, 376, Thymelaeaceae 311
377 Tiliaceae 220, 371
Scrophulariales 135, 171, 279, 380, Tinantia 204
381 Triuridales 258
Index to Plant Taxa 413

Toricellia 131 Vahlia 143

Tovariaceae 205, 264 Valdivia 145
Treculia 356,361, 367, 368 Valerianaceae 105,106,377,381
Trema 351 Velloziaceae 294
Tremandraceae 206 Velloziales 296
Trichocladus 342 Verbenaceae 206, 381
Trichopodaceae 292 Vernonieae 313
Trichosporeae 101, 102 Viburnum 388
Trigonobalanus 342 Victoria 216
Trilepisium 357, 360, 366 V iolales 264
Trimeniaceae 118, 164 Viscaceae 260
Triplostegia 105, 106 Vitaceae 164, 176, 206, 281
Triuridaceae 85 Vivianiaceae 168, 171
Triuridales 315
Trochodendraceae 113, 205, 260
Trochodendrales 171, 230 Walafrida 107
Trochodendron 29 Walchia 247, 249
Tropaeolaceae 14, 264 Wellstediaceae 402
Turneraceae 205 Winteraceae 113, 164, 205, 214
Typhaceae 293, 295
Xyridaceae 290, 294
Ulmaceae 171,205,349 seq.
Ulmus 166, 352 seq.
Umbelliferae 72, 133, 135, 180 Yucca 292
Umbelliflorae 132
Umbellularia 164
Urera 350 Zaluzianskya 107
Urticaceae 171, 349, 371 Zanichellia 315
Urticales 171, 305, 349 Zingiberaceae 99, 100, 294
Utsetela 359 Zingiberales 18,87,90,92,168
Uvaria 114 Zostera 290
Subject Index

acetylenes 382 dilated cisternae 157

adaptive peaks 42 dispersal syndromes 337
amygdalin 193, 196
anemophily 218 seq., 291, 295 ellagic acid 185, 265, 280
angiosperms, earliest 185 ellagitannins 49, 265, 306
anisocotyly 98, 99 endosperm, cellular 287, 376, 381
anthocorm theory 16 endosperm haustoria 382
apocarpy 258, 315 endosperm, helobial 287, 293, 294
aril 38 endosperm, nuclear versus cellular
attractants 182 136
endosperm, ruminate 292
backwards systematics. 8 entomophily 228, 293, 330
beetle pollination 213 epigyny 265, 268, 271
benzyltetrahydroisochinolin alkaloids . evolution, chemical 56, 63
23, 29, 69, 185, 230, 260, 265, evolutionary trends 25
280 seq. exine, columella-less 29
betalains 180 seq.
butterflies 212 family, category of 2, 3, 402
flower characters 26
cantharophily 212 seq. fossil floras 46, 47
centrifugal stamen development
221 seq., 264 genera linking families 105, 106
glucosinolates 158, 264
centripetal stamen development 264
growth habit in Urticales 350 seq.
characters, primitive 8, 9, 25, 42, 47,
81, 399 heterobathmy 121, 290, 399
character phylogenies 25 heterochlamydeous flowers 294
characters, advanced 25, 81 homoxylous angiosperms 113
chelidonic acid 291 seq. hypogyny 268
chemical data 23
class, category of 5, 403 indole alkaloids 66, 270, 379
classification, evolutionary 5 inflation of categories 300, 401
classification, phenetic 24 inflorescences, monotelic versus poly-
coleoptile in Nymphaeaceae 287 telic 380
convergence 23 integuments, initiation 126
coumarins 72 inulin 382
cauliflory 367 iridoid compounds 23, 130 seq.,
cupule 17 180 seq., 268 seq., 306, 379 seq.
cyanogenic compounds 191 seq., 265 isoflavones 56, 65
cyanogenic taxa 201, 204
laticiferous tubes 287
data banks 78,.79 leaf characters 46
delay of syngamy 338
diagrammatic representation of angio. medullary stem bundles 287
sperms 253 seq. megaspermy 361
Subject Index 415

metabolites, primary versus secondary pollination by Diptera 340

63,65 pollination, by insects 331, 366 seq.
microspe~y 361 pollination in Hamamelidales
microsporogenesis, successive type of 324 seq.
258 polyacetylenes 186, 260, 265, 388
monocauly 353 polyandry, primary 29, 221, 228,
monocliny 16 230
monocotyledonous characters in dicot- polyandry, secondary 212, 221 seq.,
yledons 286 290, 328, 400
monocotyledonous embryo 287 polyphyly 255, 398
monogeneric groups 103 seq. polyreithry 398
monophyly 27, 28, 321 predators 183, 184
mosaic-evolution 115 priority in names of higher taxa 301,
myrosin cells 265 401
proanthocyanidins 185, 187
progymnosperms 241
nectariferous tepals 291
prophyll, median 286
nodes, uni- versus trilacunar 136
protogyny 213, 221
numerical methods 23
protomonocot, hypothetical 314
pseudanthia 233, 364, 366, 377
odour, absolute versus imitative 214 pseudocarpy 367
order, category of 4, 402 pterocarpanoids 56
origin of higher taxa 27
ovular morphology 123 rank, determination of 93
ovule types 124, 126 repellents 181 seq.
ovules, bitegmic 270 rotenoids 56
ovules, tenuinucellate 270
ovules, unitegmic 270 saponins 183
sarcotesta 148
ovuliferous cones 249
oxalate raphides 278, 279 seco-iridoids 270, 379, 380, 383
self-incompatibility 216, 270
senecio alkaloids 186
pachycauly 292, 352 septal nectaries 291 seq.
perigyny 265, 268 sesquiterpene lactones 186, 260, 388
perisperm 287, 288 sex distribution in Urticales 364
phaneranthy 16, 17 sieve element plastids 14, 27, 158,
photosynthesis, C4-pathway of 271 258,287
phylloclade 235 silica bodies 279, 291
phylogenetic classification 21, 24 social bees 212
phylogenetic trees 6 sporophyll theory 15
phylogenetic trends 11 starch endosperm 278
pleiophily 16, 27, 28 . steroid saponins 279, 291 seq.
pollen, bi- versus trinucleate 136, stigma, dry versus wet 270
270, 383 stomata, anomocytic 291
pollen, inaperturate 258, 288 storage embryos 288
pollen, monoaperturate 258 subclass, category 26 seq., 279, 403
pollen, monosulcate 10, 46, 227, 287 subfamily, category 2
pollen, multi-aperturate 48 suborder, category 404
pollen, pauci-aperturate 46, 48 superorder, category 26
pollen, tricolpate 227, 229 sympetaly 17, 268
pollen, ulcoidate 295
pollination, aquatic 288 tannins 183 seq.
pollination by beetles 213 taxa incertae sedis 312
416 Subject Index

terminal suprageneric taxon 103 wind pollination 328, 334, 338, 358,
366,368
vesselless angiosperms 113, 119 wood anatomy 111 seq.
vessels 271, 191 xanthones 23
vicariance, morphological versus
chemical 75 zoophily 233