Opinion
Megaphylls, microphylls and the
evolution of leaf development
Alexandru M.F. Tomescu
Department of Biological Sciences, Humboldt State University, Arcata, CA 95521, USA
Originally coined to emphasize morphological differences, ‘microphyll’ and ‘megaphyll’ became synonymous with the idea that vascular plant leaves are not
homologous. Although it is now accepted that leaves
evolved independently in several euphyllophyte
lineages, ‘megaphyll’ has grown to reflect another type
of homology, that of euphyllophyte leaf precursor structures. However, evidence from the fossil record and
developmental pathways fails to indicate homology
and suggests homoplasy of precursor structures. Thus,
as I discuss here, ‘megaphyll’ should be abandoned
because it perpetuates an unsupported idea of
homology, leading to misconceptions that pervade plant
biology thinking and can bias hypothesis and inference
in developmental and phylogenetic studies. Alternative
definitions are needed that are based on development
and phylogeny for different independently evolved leaf
types.
The microphyll–megaphyll dichotomy
In vascular plant sporophytes, leaves are lateral appendages that share four defining features: vascularization,
determinate growth, bilateral symmetry (adaxial–abaxial
polarity; hereafter referred to as ad/abaxial polarity) and
definite arrangement (phyllotaxis). Despite these shared
characteristics, leaves are not homologous across all vascular plants. An early step toward the recognition of this
situation [1] was the defining of a major dichotomy between microphylls, small and simple leaves, and megaphylls, larger and more complex leaves. The dichotomy,
based on morphological criteria, was reinforced as subsequent progress in paleobotany led to the realization that
vascular plant phylogeny was itself divided by a major
dichotomy [2], dating back 415 million years to the Late
Silurian–Early Devonian, that paralleled, to some extent,
the taxonomic distribution of the two leaf types. The two
major lineages of that phylogenetic divide originated from
among two distinct grades of early vascular plants, the
zosterophylls and the trimerophytes. The descendants of
those two lineages form the two clades that comprise most
vascular plant phylogeny, the lycophytes and the euphyllophytes, and which were regarded as having evolved
microphylls and megaphylls, respectively. These ideas
provided the impetus and framework for an early wave
of thinking that would later become known as evo-devo,
which led to the formulation of seminal hypotheses of leaf
evolution.
Corresponding author: Tomescu, A.M.F. (mihai@humboldt.edu).
However, the microphyll–megaphyll divide is not as
clear cut, and morphological definitions that contrast
microphylls and megaphylls as mutually exclusive concepts of leaves are inconsistent. Moreover, current understanding of plant phylogeny and leaf development fails to
shed light on the origin of microphylls, and supports
several independent origins of megaphylls. Here, I review
plant phylogeny and developmental data from fossil and
extant trachephytes, as well as the current understanding
of genetic pathways controlling leaf development, to argue
that the megaphyll concept should be abandoned because
it perpetuates misconceptions and confusion based on
unsupported homology.
Morphological inconsistencies and overlap in the
microphyll–megaphyll dichotomy
Microphylls are defined as leaves of small size, with simple
venation (one vein) and associated with steles that lack leaf
gaps (protosteles). By contrast, megaphylls are defined as
leaves of generally larger size, with complex venation and
associated with leaf gaps in the stele [3]. However, each of
these criteria has inconsistencies [4] that highlight the
disconnection between phylogeny and current morphological definitions at the level of the microphyll–megaphyll
divide.
First, the presence or absence of leaf gaps does not
provide a basic distinction between the two types. In many
plants nested among the megaphyll-bearing euphyllophytes, leaf trace divergence is not associated with a leaf
gap (e.g. the non-protostelic Equisetum, cladoxylalean
pteridophytes and all extant seed plants [5]). Moreover,
several groups of euphyllophytes are similar to microphyllbearing lycophytes in that they have protosteles (e.g. the
extant filicalean ferns Lygodium and Gleichenia, fossil
Kaplanopteris and Botryopteris, some extinct coenopteridalean ferns and sphenophyllales, the aneurophytalean
progymnosperms, and even early seed plants such as
Elkinsia).
Second, leaf size and venation complexity are also
inconsistent as distinguishing criteria between microphylls
and megaphylls. Equisetum and some fossil sphenophyllales, as well as several extant and fossil gymnosperms, all of
which are megaphyllous euphyllophytes, have highly
reduced leaves supplied by one vein. Conversely, some
lycophyte microphylls are large (up to 1 m in some of the
extinct lepidodendrales or up to 0.5 m in extant Isoetes
species), whereas others have complex venation patterns
(some Selaginella species [6]) or morphologies (e.g. the dissected leaves of extinct protolepidodendraleans such as
1360-1385/$ – see front matter ß 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.tplants.2008.10.008 Available online 11 December 2008
5
Opinion
Colpodexylon, Protolepidodendron, Leclerqia and Estinnophyton [7]).
Phylogenetic and developmental perspectives
Definitions based on morphology thus seem to create more
confusion and overlap between the microphyll and megaphyll concepts instead of clarifying them. The validity of
the two concepts needs to be verified against comparative
developmental data, genetic developmental pathways controlling leaf development and plant phylogenies that illuminate the relationships between extant taxa and their
closest leafless relatives in the fossil record. Leaf evolution
was the earliest major plant biology question to be
addressed using a comprehensive evo-devo approach.
Work on leaf evolution integrated all the data types available (morphology, anatomy, development, phylogenetic
hypotheses and the fossil record) to produce consistent
theories. Such pre-molecular era evo-devo approaches
led to several hypotheses for the evolution of leaves. Of
those, two prevailing hypotheses explained the origin of
microphylls from enations (small unvascularized flaps of
tissue that characterized early vascular plants), and that of
megaphylls from three-dimensional (3D) branching systems of undifferentiated photosynthetic axes [7].
Most modern phylogenies support the lycophyte–
euphyllophyte divide [8–12], but phylogenetic studies have
so far failed to provide unequivocal answers on the nature
of the ancestral structures and evolutionary processes that
generated the lycophyte leaf and have produced conflicting
hypotheses of euphyllophyte phylogeny [9,10,12]. A developmental feature potentially congruent with the lycophyte–euphyllophyte dichotomy could be the mode of
origination of leaf primordia on the flanks of the shoot
apical meristem (SAM). A study of Selaginella [13] led to
the idea that lycophyte leaves originate from a few initials
specified exclusively in the outermost cell layer of the SAM;
this is in contrast with euphyllophyte leaves, which
originate from larger populations of initials to which both
the outermost cell layer, as well as subjacent layers, contribute. The mode of origination of Selaginella leaves is
similar to the development of bryophytic gametophyte
leaves and fern scales, and could be the cause for a different
mode of ad/abaxial polarity determination that is
uncoupled from the layering of the leaf primordium [13].
Data available on other extant lycophytes seem to support
this proposed developmental dichotomy, but an exhaustive
survey of extant lycophytes, as well as data from fossils, are
needed to confirm the hypothesis.
Another direction of investigation is suggested by the
results of a study that found limited interrelationships
between phyllotaxis and cauline vascular architecture in
two Lycopodium species [14]. This uncoupling of two major
shoot morphological features contrasts the situation seen
in euphyllophytes, where phyllotaxis and cauline vasculature are well correlated. If confirmed by studies in other
lycophytes, this could represent another fundamental
developmental characteristic separating lycophyte leaves
from those of euphyllophytes.
Phylogenetic analyses could provide another set of
criteria for defining leaves in lycophytes and in the
different euphyllophyte groups by identifying the type of
6
Trends in Plant Science Vol.14 No.1
precursor structure at the origin of each, as reflected by
their respective leafless sister groups. That requires
inclusion of as many fossil taxa as possible in
morphology-based phylogenies. In lycophytes, three competing hypotheses propose evolution of the leaf (‘microphyll’): (i) by vascularization of enations; (ii) from
sporangia by sterilization; or (iii) from telome trusses by
reduction [7,8]. However, although monophylly of the
group is generally accepted, neither the scarce data on
genetic pathways controlling leaf development [15,16] nor
phylogenetic studies [8] lend unequivocal support to any of
the three hypotheses.
Our understanding of euphyllophyte phylogeny is hampered by major disparities between the results of studies
based exclusively on extant taxa and those of studies that
include fossils [8–12]. ‘Extant-only’ studies support monophylly of extant seed-free euphyllophytes (the ‘monilophyte’ clade), whereas studies including fossils support a
sequence of paraphyletic grades comprised of extant and
fossil seed-free euphyllophytes that led to the seed plants.
Although they have been gaining wide acceptance, the
results of extant-only studies are equivocal to the question
of euphyllophyte leaf origin because they lack fossil taxa
that: (i) could represent transitional stages bearing precursor structures in transformational series; and (ii) generate the amount of phylogenetic resolution needed to
answer such a question. Nevertheless, the fossil record
shows unequivocally that the common ancestor of ‘megaphyllous’ euphyllophytes was leafless. This indicates that,
among euphyllophytes, leaves are not homologous because
they evolved independently in at least two lineages or, as
indicated by phylogenies that include fossils, in as many as
nine different lineages. A close look at leaf development
and the genetic pathways that control it across euphyllophytes provides strong support for homoplasy of the leaf
across euphyllophytes.
How many different ‘megaphylls’ are there?
Multiple lines of evidence indicate that leaves that have
been categorized as megaphylls are fundamentally different from each other. Some are associated with eustelic
stems, whereas others are borne on siphonostelic or protostelic stems. In seed plants the adaxial domain of leaves
has a developmentally associated axillary meristem; such
structures are missing from seed-free plants. Cell differentiation and tissue maturation progress exclusively acropetally in fern leaves, but in seed-plant leaves,
development is a more complex process in which final
tissue maturation progresses largely basipetally. All of
these indicate that there are several different types of
‘megaphylls’ that have probably evolved independently.
Current understanding of plant phylogeny supports several independent origins of leaves in different euphyllophyte lineages.
The studies that used the most complete datasets pertinent to euphyllophyte phylogeny in general, and leaf
evolution among euphyllophytes in particular [9,12], support two contrasting topologies of the euphyllophyte tree.
One of these topologies, not emphasized here, proposes an
unorthodox placement of the zosterophylls–lycophyte clade
as the sister group to the lignophytes (sensu Ref. [17]) and,
Opinion
Trends in Plant Science
Vol.14 No.1
Figure 1. Microphyll and megaphyll origins. ‘Extant + extinct’ polysporangiophyte phylogeny (based on Refs [9,33]) supports a single origin of microphylls (blue line) and
several independent origins of megaphylls (red lines); red dots represent alternative megaphyll origins. Paraphyletic grades within which relationships are not detailed are
collapsed and represented by the broader branches of the tree; names of lineages with extant representatives are in green; names of extinct lineages are in black. Fern
clades 1 and 2 are as defined in Ref [9]: fern clade 1 – Stauropteridales; fern clade 2 – Zygopteridales + Cladoxylales; fern clade 3 includes living and extinct Filicales and
Hydropteridales. Aneurophytales and Archaeopteridales are progymnosperms. The pteridosperms are seed ferns: HMC (hydrasperman, medullosan and callistophytalean
seed ferns); P (peltaspermalean and corystospermalean seed ferns); and GPBC (glossopteridalean, pentoxylalean, bennettitalean and caytonialean seed ferns).
hence, paraphylly of the euphyllophytes. The other topology (Figure 1) supports the more generally accepted
hypothesis of euphyllophyte monophylly [8]. However,
independent of euphyllophyte monophylly, because of
the inclusion of a high number of fossil representatives
of all lineages (some of which are entirely extinct), both of
the proposed topologies bring considerable resolution to
the question of euphyllophyte leaf evolution – both support
up to nine independent origins of euphyllophyte leaves
(Figure 1). The morphology of fossil taxa that occupy basal
positions in the different lineages that have evolved leaves
independently indicates that in all cases the precursor
structures that evolved into leaves were systems of 3Dbranching undifferentiated axes.
‘Megaphyll’ precursor structures
The origin of euphyllophyte leaves from systems of 3Dbranching axes, long advocated by students of the fossil
record, received support from cladistic studies that proposed that synapomorphies of the euphyllophyte clade
include: (i) monopodial or pseudomonopodial branching;
and (ii) small, pinnulelike vegetative branches [8]. The
corollaries of these synapomorphies are the presence of a
main axis with subordinated lateral branching systems
and determinate growth of the lateral branching systems. These proposed synapomorphies led to a hypothesis of partial homology, namely that megaphylls are
homologous at the level of precursor structures (‘megaphyll precursors’ [8]): 3D dichotomous lateral branches
that shared a fundamental leaf characteristic, determinacy [8].
However, currently there is no support for homology of
lateral appendage determinacy across euphyllophytes.
Structural homology means sameness as a result of inheritance from a common ancestor that had the structure in
discussion, yet evidence that the common ancestor of
euphyllophytes had determinate lateral branching systems is equivocal. In fact, a recent study of 380–340million-year-old fossils (Heather L. Sanders, PhD thesis,
Ohio University, 2007) shows that ferns and seed plants
7
Opinion
acquired determinacy and ad/abaxial symmetry of lateral
branching systems in a different sequence: whereas evolution of ad/abaxial symmetry in ferns preceded that of
determinacy, ad/abaxial symmetry in seed plants is a
more-derived character, preceded by the evolution of determinacy. This finding has signficiant implications because it
demonstrates that neither determinacy nor ad/abaxial
symmetry were present in the common ancestor of the
lineages from which ferns and seed plants evolved. Therefore, homology of ‘megaphyll’ precursor structures could be
invoked only at the most basic level of body plan organization, that of 3D branching systems of undifferentiated
axes. However, at that level everything would be homologous to everything else in the organography of polysporangiophyte sporophytes, because it all evolved from the
same fundamental precursor structure, the branched system of undifferentiated axes.
Furthermore, if megaphyll precursor structures were
homologous as determinate lateral branching systems
across all basal euphyllophytes, then we should be able
to find evidence of process homology [18] for leaf determinacy across all extant ‘megaphyllous’ lineages. In other
words, we would expect the gene pathways that control
determinacy in ‘megaphyllous’ leaves to be shared among
all euphyllophytes. However, although they are still short
of addressing the whole breadth and depth of polysporangiophyte phylogeny, genetic studies of leaf development
indicate that this is not the case. Interactions between
class 1 KNOX (KNOTTED-LIKE HOMEOBOX) genes and
ARP (ASYMMETRIC LEAVES, ROUGH SHEATH and
PHANTASTICA) genes, a main candidate regulator of
determinacy in leaf development, show a breadth of diversity that eliminates the possibility of process homology,
and so do the genetic pathways involved in another major
leaf feature, ad/abaxial polarity.
Process homology and homoplasy
Process homology refers to common inheritance of developmental genetic pathways [18]. In terms of gene pathways, three major determinants of leaf development have
been commonly considered in discussions of the evolution
of shoot systems: (i) interactions between class 1 KNOX
genes and ARP genes; (ii) class III homeodomain-leucine
zipper (HD-Zip) genes and their interactions with KANADI
genes and microRNAs 165 and 166; and (iii) YABBY genes
and their interactions with class 1 KNOX genes and leaf
ad/abaxial polarity pathways.
KNOX–ARP interactions
Class 1 KNOX genes (hereafter referred to as KNOX) are
thought to be responsible for maintaining the meristematic
status of tissues characterized by cell totipotency and thus
for indeterminacy of growth. Consistent with this role, they
are expressed at the shoot apex in all extant lineages of
tracheophytes (Box 1). Conversely, genes of the ARP group
are expressed in the leaf primordia of all living tracheophytes (Box 1), where they are thought to induce determinacy of growth by promoting cell fate determination,
starting with the specification of leaf primordium founder
cells at the periphery of the apical meristem. Thus, expression of KNOX genes at the shoot apex and of ARP genes in
8
Trends in Plant Science Vol.14 No.1
leaf primordia seem to be common denominators of leaf
development in all extant tracheophytes. Therefore, it is
tempting to suggest that some form of KNOX–ARP
antagonistic interaction is the fundamental shared mechanism responsible for determinacy of lateral appendages,
which would then represent a ‘process synapomorphy’ that
characterized all leaf precursors in tracheophytes. However, a closer look at KNOX phylogeny and KNOX–ARP
patterns of expression in the different tracheophyte
lineages indicates complex evolutionary patterns. These
demonstrate that KNOX–ARP interactions are modulated
differently in the different lineages (Box 1) and weaken the
case for KNOX–ARP interactions as a major regulator of
determinacy.
Box 1. Genetic pathways in leaf development
Broadening of the taxonomic spectrum for which genetic developmental data are available, along with innovative developmental
approaches to the plant fossil record, will help answer specific
questions that will lead to a deeper understanding of the evolution
of leaf development in all tracheophyte lineages.
Is KNOX expression in the shoot apex (Figure Ia, dark green) and
ARP expression in leaf primordia (Figure Ia, red) shared by all
vascular plants, living and extinct? With respect to extinct
lineages, this might seem a strictly rhetorical question. However,
several studies [25,35,36] have already used morpho-anatomical
fingerprints to demonstrate developmental processes and regulatory mechanisms in plant fossils, showing that documentation
of such characteristics in extinct lineages has become reality.
Discovery of morpho-anatomical fingerprints for regulatory
mechanisms of leaf development ([37], Heather L. Sanders, PhD
thesis, Ohio University, 2007) could enable the inclusion of many
extinct lineages in studies addressing the evolution of leaf
development.
Is KNOX expression in complex leaf primordia (Figure Ib, light
green) exclusively characteristic of euphyllophytes, or were KNOX
genes also expressed in the leaf primordia of lycophytes with
complex leaf morphology (now extinct), such as Leclerqia?
KNOX downregulation in leaf primordium initials (Figure Ib,
yellow) has been documented in Selaginella and seed plants but
is absent in filicalean ferns and some angiosperms (Medicago). Is
this pattern due to loss of a shared characteristic of lycophytes
and euphyllophytes or to parallel evolution in ferns and Medicago?
Was ARP expression in the shoot apical meristem (SAM; Figure Ib,
pink) present in the common ancestor of lycophytes and
euphyllophytes and subsequently lost in seed plants, only to
evolve independently in some angiosperms (tomato)? Or was it
acquired independently in lycophytes, ferns and flowering plants?
What are the functions of HD-ZipIII genes in ferns? Are they
responsible only for SAM functioning and vascular tissue
patterning, as they are in all other extant vascular plants (Figure
Ic, purple), or are they also involved in leaf ad/abaxial symmetry
patterning, as is the case in seed plants (Figure Ic, light blue)? Is
the latter function a process synapomorphy of seed plants or of all
euphyllophytes?
Are HD-ZipIII–KANADI interactions conserved across all embryophytes, as suggested by the presence of KANADI homologs in
Selaginella and Physcomitrella, and is their canalization into
regulation of (i) vascular tissue radial patterning and (ii) leaf ad/
abaxial polarity regulation homologous or homoplasic in different
vascular plant lineages? Are any of the other regulatory pathways
of leaf ad/abaxial polarity documented in angiosperms active and
interacting in other lineages?
What gene pathways control leaf ad/abaxial polarity patterning in
lycophytes? Are any of the pathways that are involved in leaf ad/
abaxial polarity patterning in seed plants also active in lycophytes?
Opinion
Figure I. Expression patterns of genes associated with leaf development.
Polysporangiophyte phylogeny is simplified based on the tree in Figure 1 (main
text) and shows only extant lineages for which gene expression data are available
(data from Refs [15,16,18,21–24,26,29,34]); thicker branches are placeholders for
paraphyletic grades within which relationships are not detailed: 1,
protracheophytes and rhyniophytes; 2, Isoetales, Protolepidodendrales,
Lycopodiales, Drepanophycales and zosterophylls; 3, trimerophytes, Psilotum
and Stauropteridales; 4, Marattiales and Ophioglossales (fern clade 3 comprises
extinct and living Filicales + Hydropteridales); 5, Zygopteridales, Cladoxylales,
sphenopsids, progymnosperms plus the hydrasperman, medullosan and
callistophytalean seed ferns; 6, peltaspermalean and corystospermalean seed
ferns; 7, glossopteridalean, pentoxylalean, bennettitalean and caytonialean seed
ferns. (a) KNOX and ARP expression patterns potentially shared by all living
tracheophytes: KNOX genes expressed in the SAM (dark green) and ARP genes
expressed in leaf primordial (red). (b) KNOX and ARP expression patterns that
demonstrate widely divergent types of interaction: KNOX genes expressed in leaf
primordia (plants with complex leaf primordia) (light green); KNOX genes
downregulated, even if only transiently, in leaf primordia (Medicago shows no
downregulation, in contrast to the main angiosperm pattern) (yellow); ARP genes
expressed in the SAM (Lycopersicon is the only angiosperm known with this
pattern) (pink). (c) HD-ZipIII and YABBY functions: HD-ZipIIIs involved in SAM
functioning and vascular tissue patterning that are potentially repressed by
microRNAs 165 and 166 (purple); HD-ZipIIIs involved in leaf ad/abaxial polarity
patterning (light blue); YABBYs involved in leaf ad/abaxial polarity patterning
(brown). Question marks refer to questions detailed in the box text.
Trends in Plant Science
Vol.14 No.1
KNOX phylogenies show that members of the gene
family expressed in the SAMs of different living lycophyte
and euphyllophyte lineages are not orthologs of each other
[15,19] because some of them are more closely related to
KNOX genes with different patterns of expression than to
those with similar expression patterns. This has been
taken to indicate independent evolution of the SAM
expression pattern (and presumably function) of KNOX
genes in the different lineages [15]. Alternatively, it has
been proposed [19] that the different patterns of expression
could be the result of subfunctionalization after duplication
of a single KNOX gene inherited from the tracheophyte
common ancestor, where it covered all of the expression
domains of the different derived members of the family,
including the SAM. This seems plausible in light of the
finding that in mosses (Physcomitrella) KNOX genes
promote production of sporogenous tissue by preventing
prematures sporogenesis [20], a function similar to the
maintenance of meristematic status, and thus indeterminacy, at the shoot apices of higher embryophytes.
KNOX and ARP genes have been identified in all the
tracheophytes studied to date. It has been proposed that
KNOX–ARP interactions have evolved independently in
lycophytes and euphyllophytes [15], but an alternative
hypothesis is that they are shared across all vascular
plants and, in lycophytes, they could have contributed to
the acquisition of determinacy of sterilized-sporangia
appendages [19]. Irrespective of that discussion, KNOX–
ARP interactions are modulated differently between
extant plant lineages. A reflection of this situation is that
KNOX expression is not limited to the SAM; neither is ARP
expression limited to leaf primordia (Box 1). Additionally,
the degree of antagonism in KNOX–ARP interactions varies between lineages:
KNOX genes are downregulated in the leaf primordia of
lycophytes and seed plants (cycads, conifers, Welwitschia and angiosperms; no information available
for Ginkgo). Although this pattern has been attributed
to ARP gene activity, unequivocal evidence is not
available for all lineages that exhibit KNOX downregulation. In fact, in some angiosperms, ARP genes are
not needed for downregulation of KNOX genes (which
are still downregulated in the leaf primordial of some
Arabidopsis and maize ARP loss-of-function mutants
[21]), and they do not repress all KNOX genes in the leaf
primordium (a feat achieved only by YABBY genes [16]).
KNOX genes are expressed in the leaf primordia of ferns
and seed plants that have complex leaves that start off
as complex primordia [22]. In most of those seed plants,
KNOX expression characterizes later stages of leaf
primordium development and follows a transient stage
of KNOX downregulation that is thought to be linked to
leaf primordium specification. By contrast, ferns do not
exhibit KNOX downregulation associated with any
stage of leaf primordium development; additionally,
KNOX proteins do not seem to have the same role in leaf
development in Ceratopteris as they do in angiosperms
(Heather L. Sanders, PhD thesis, Ohio University,
2007). Recently, the angiosperm Medicago has been
shown to lack KNOX downregulation during leaf
primordium development [23].
9
Opinion
Ferns are also peculiar in that both KNOX and ARP
genes are expressed in both the SAM and leaf primordia,
suggesting that interactions between the two gene
groups are anything but antagonistic.
ARP expression in the SAM has been documented not
only in ferns but also in Selaginella and has been
hypothesized to be responsible for apical branching in
these plants [15]. However, the same pattern of
expression has been documented in an angiosperm
(Lycopersicon [24]) that is characterized by axillary and
not apical branching.
Genes of the ASYMMETRIC LEAVES2/LATERAL
ORGAN BOUNDARIES (AS2/LOB) family act in
concert with AS1 to downregulate KNOX genes in
angiosperm leaf primordia. Putative AS2/LOB homologs have been identified in Selaginella and Physcomitrella [19], but their patterns of expression and
phylogeny at the level of all living tracheophytes are
not resolved.
In conclusion, across the lycophyte–euphyllophyte
divide, gene expression patterns do not indicate unequivocally whether particular types of KNOX–ARP interactions are homologous or homoplasic; neither do they
differentiate unequivocally lycophyte leaves from those
of euphyllophytes. Within euphyllophytes, the existence
of a shared fundamental mechanism underpinning leaf
development has been proposed based on studies of the
evolution of leaf morphology [25], and KNOX–ARP interactions have been suggested as a candidate for that role
[15]. However, the multitude of interaction types suggested
by the diversity of KNOX and ARP expression patterns
indicates that this is not the case and does not translate
into any clear patterns of process homology or homoplasy
in leaf development, even just across euphyllophytes.
Class III HD-Zip genes
Class III HD-Zip genes (hereafter referred to as HD-Zip)
have been identified in representatives of most extant
streptophytes [26]. Based on HD-Zip expression patterns
and inferred functionality, as well as on gene phylogeny, it
has been proposed [26] that the common ancestor of vascular plants had a single HD-Zip gene that was involved in
regulation of apical meristem functioning and radial patterning of vascular tissues (Box 1). Evolutionary duplication of the ancestral HD-Zip and subsequent lineagespecific functional diversification led to the acquisition of
new roles by the resulting paralogs in lycophyte and
euphyllophyte lineages. Consequently, HD-Zip functions
mark a fundamental difference in developmental regulation between lycophyte and seed-plant leaves [16].
Whereas HD-Zips are involved in procambium specification and vascular tissue patterning in lycophyte leaves
(Selaginella), in seed-plant leaves, they are involved in
procambium specification and patterning, as well as in
primordium specification and ad/abaxial polarity (Box 1).
Particularly, in seed-plant leaves, HD-Zips determine
adaxial polarity as a result of antagonistic interactions
with KANADI genes (promoters of abaxial identity in leaf
tissues) and microRNAs 165 and 166 [27,28]. A conserved
target sequence for microRNAs 165 and 166 suggests that
10
Trends in Plant Science Vol.14 No.1
negative regulation of HD-Zip by those microRNAs might
be shared across all embryophytes [29]. By contrast,
absence of HD-Zip expression data for seed-free euphyllophytes (in which HD-Zips have been identified [16]) and
scarcity of data on non-angiosperm KANADI homologs
(which were identified, nevertheless, in Selaginella and
Physcomitrella [19]) preclude understanding of the role of
HD-Zip–KANADI interactions in leaf evolution.
In conclusion, phylogenetic patterns of HD-Zip functions parallel the lycophyte–euphyllophyte divide and concur with homoplasy of leaves between the two clades.
However, the functions and interactions of these genes
are insufficiently documented to bring resolution to questions of homology and homoplasy in leaf evolution among
euphyllophytes.
YABBY genes
The YABBY gene family is believed to be specific to seed
plants because members have been identified in both
angiosperms and gymnosperms but none were detected
in seed-free plants [19] (Box 1). To date, YABBY functions
have been investigated only in angiosperms. There, they
are responsible for repression of all KNOX genes in the leaf
primordium [16] and regulation of leaf ad/abaxial polarity
and laminar growth, and they might also influence phyllotaxis [30]. However, in terms of leaf ad/abaxial polarity
regulation, YABBY genes act downstream of all other
known polarity determinants. These include three pathways (ETTIN–AUXIN RESPONSE FACTOR4, AS1–AS2
and HD-ZipIII–KANADI) with complex interactions that
differ significantly among studied angiosperm species, and
none of these pathways are concerned exclusively with leaf
ad/abaxial polarity [28].
Genetic pathways – overview
Studies of genetic pathways of leaf development that
address gene phylogeny, expression patterns and functionality are opening new perspectives in the way we think
about leaf evolution. KNOX–ARP interactions have been
proposed as a fundamental mechanism responsible for
determinacy of leaf precursor structures. However, patterns of KNOX and ARP expression at the shoot apex and
in leaf primordia exhibit a multitude of combinations that
indicate widely divergent types of KNOX–ARP interaction
in different major lineages and raise the question whether
that is an important interaction for leaf determinacy. If the
latter is true, no independent evidence is available to
indicate whether it represents homology or homoplasy,
in terms of process, between lycophytes and euphyllophytes. The diversity of KNOX and ARP expression patterns rather supports many independent origins of
determinacy of leaf precursor structures (even among
euphyllophytes), possibly by different modulation of
KNOX–ARP interactions, but it is unclear whether those
interactions existed before the evolution of determinacy or
evolved initially as a determinacy mechanism.
The expression patterns and known or inferred functions of HD-Zip genes mark a fundamental difference in
developmental regulation between the seed-plant clade,
where HD-Zips participate in leaf ad/abaxial polarity regulation in interaction with KANADI genes and microRNAs
Opinion
165 and 166, and lycophytes with no HD-Zip involvement
in leaf ad/abaxial polarity. Data on HD-Zips from seed-free
euphyllophytes are needed to understand whether that
fundamental difference corroborates the lycophyte–
euphyllophyte divide as a ‘process synapomorphy’ of
euphyllophytes or reflects a seed plant synapomorphy.
The latter is probably the case for YABBY genes,
which are involved in seed plant leaf ad/abaxial polarity
regulation (among other functions, such as laminar growth
and phyllotaxis) and which seem to be absent in seed-free
plants.
Conclusions and future perspectives
The classic concepts of microphyll and megaphyll pervade
thinking on the evolution of leaf development. As such,
they influence significantly the process of science in plant
biology by contributing to the shaping of evolutionary
hypotheses and to inferences of developmental studies.
However, the two concepts, as currently defined, are equivocal, partially overlapping and inconsistent with current
understanding of plant phylogeny. Most workers are aware
of this situation when they agree that ‘microphylls’ (as
referring to lycophyte leaves) probably have a single common origin and that ‘megaphylls’ (as referring to euphyllophyte leaves) evolved independently in several lineages.
Under this scenario, the ‘megaphyll’ (and its more recent
version, the ‘euphyll’) has been retained to account for a
hypothesis of partial homology whereby the leaves of all
euphyllophytes can be traced back to a common ancestor
that had determinate lateral branching systems. However,
the fossil record and genetic pathways controlling leaf
development (as documented in different vascular plant
lineages) indicate that euphyllophyte leaves are neither
homologous at the level of their precursor structures nor at
the level of the genetic pathways that control their development. Thus, by grouping together non-homologous
structures, the megaphyll concept perpetuates an unsupported evolutionary scenario. The centrality of the concept
to many hypotheses on the evolution of leaf development
makes this issue more than just a matter of naming things,
and I argue that the ‘megaphyll’ should be abandoned
before it introduces more bias in plant science. ‘Leaf’,
accompanied by a specifier (e.g. ‘filicalean leaf’) works just
as well, and it is neutral in terms of any implication of
homology. Ultimately, we need unequivocal definitions of
the different leaf types based on development (including
genetic pathways) and phylogeny.
Phylogeny and the fossil record show that the ancestors
of the two crown clades of vascular plants, lycophytes and
euphyllophytes, were leafless, which is consistent with
homoplasy of leaves between the two clades. Within the
lycophyte clade, homology of leaves is widely accepted,
thus the microphyll concept could be retained if it is
redefined to designate lycophyte leaves. Within euphyllophytes, fossil-based phylogenies support as many as nine
independent origins of leaves. Justification for the use of
‘megaphyll’ to designate euphyllophyte leaves stemmed
from the idea that they all evolved from precursor structures that were homologous as branching systems with
determinate growth. However, genetic pathways suggest
homoplasy between euphyllophyte lineages in terms of leaf
Trends in Plant Science
Vol.14 No.1
developmental processes, and studies of fossil ferns and
seed plants indicate that the branching axes of the common
ancestor of the two lineages had not evolved determinacy
nor ad/abaxial symmetry. Thus, use of a specialized term
has no logical justification because it would group together
structures that are not homologous.
Despite spotty taxonomic coverage, data on developmental genetic pathways add another layer of complexity
to the understanding of leaf evolution. The diversity and
phylogenies of genes in major gene families involved in leaf
development are well documented in several angiosperm
taxa. Putative homologs of some of those genes have been
identified in gymnosperms, filicalean ferns, Psilotum, the
lycophyte Selaginella, the moss Physcomitrella, liverworts
and hornworts, and phylogenies have been constructed for
some of those genes. By contrast, while attempts are being
made to study developmental genes in non-flowering
plants, the difficulties have thus far prevented much from
being learned. Studies of expression patterns have been
performed only on a limited subset of genes and taxa, and
studies of gene functions are rare outside the angiosperm
clade. Mechanisms controlling ad/abaxial leaf polarity,
characterized in significant detail in flowering plants,
are undocumented in seed-free plants. Nevertheless, available data indicate that KNOX–ARP interactions are too
diverse among (and sometimes within) major lineages to be
unequivocally regarded as the fundamental mechanism of
leaf determinacy in tracheophytes. It is unclear which
aspects of the KNOX–ARP interactions are homologous
and which are homoplasic between lycophytes and euphyllophytes.
Two major unanswered questions are – what were the
precursor structures of lycophyte leaves, and exactly how
many independent origins of euphyllophyte leaves there
were? Answers to these questions and understanding of
leaf evolution could come from addressing several more
specific questions (Box 1). These will require broadening of
the taxonomic spectrum for which genetic developmental
data are available [31]. Whether or not the monilophyte
clade stands the test of time [10,12,32], the seed-free
euphyllophytes include a considerable amount of diversity.
Of these, only one lineage, the filicalean ferns, has been
sampled to date. Thus, even considering extant euphyllophytes only, several lineages that are highly divergent
morphologically require detailed study: eusporangiate
ferns (Marattiales and Ophioglossales), psilotophytes,
sphenopsids and heterosporous leptosporangiate ferns
(Hydropteridales). Getting the full picture of leaf development will also require data on lycophytes other than
Selaginella (the only taxon of the clade studied to date),
such as homosporous lycophytes and the more highly
derived, heterosporous Isoetales. Inclusion of data from
all of these taxa, ideally supplemented with data on fossil
lineages, will illuminate patterns of gene evolution, expression and functionality that could ultimately be used as
characters in phylogenetic studies.
Finally, one could also wonder what good is the microphyll if the megaphyll is to be abandoned, because the two
concepts were defined as mutually exclusive. However,
irrespective of whether we decide to call lycophyte leaves
microphylls, defining their synapomorphies represents a
11
Opinion
valid task with significant implications. That task is rendered difficult by the lack of understanding of the evolutionary history of lycophyte leaves. Although some fossil
evidence points to origin by vascularization of enations,
two other proposed evolutionary mechanisms (sporangial
sterilization and reduction of branching systems) cannot be
rejected, and neither phylogeny nor genetic developmental
data provide unequivocal resolution to the enigma of the
precursor structures of the lycophyte leaf. One potential
lycophyte synapomorphy and microphyll-defining character that needs confirmation is the origination from only a
few initials, all of which are specified in the outermost cell
layer of the shoot apex. Another potential shared feature of
lycophyte leaves is a partial uncoupling of phyllotaxis and
cauline vascular architecture. These emphasize the need
for basic studies of leaf developmental anatomy and
morphology, a need which is by no means limited to the
lycophytes – it is humbling to realize, in the post-genomics
era, how much we do not know about development and
morphology in many plant species.
Acknowledgements
I thank Gar Rothwell, Heather Sanders and two anonymous reviewers for
their constructive critique, helpful comments and suggestions.
References
1 Bower, F.O. (1908) The Origin of a Land Flora, a Theory Based Upon
the Facts of Alternation. Macmillan
2 Banks, H.P. (1968) The early history of land plants. In Evolution and
Environment (Drake, E.T., ed.), pp. 73–107, Yale University Press
3 Gifford, E.M. and Foster, A.S. (1989) Morphology and Evolution of
Vascular Plants. (3rd edn), Freeman
4 Kaplan, D.R. (2001) The science of plant morphology: definition,
history, and role in modern biology. Am. J. Bot. 88, 1711–1741
5 Beck, C.B. et al. (1982) Stelar morphology and the primary vascular
system of seed plants. Bot. Rev. 48, 691–815
6 Wagner, W.H. et al. (1982) Complex venation patterns in the leaves of
Selaginella: megaphyll-like leaves in lycophytes. Science 218, 793–794
7 Stewart, W.N. and Rothwell, G.W. (1993) Paleobotany and the
Evolution of Plants. (2nd edn), Cambridge University Press
8 Kenrick, P. and Crane, P.R. (1997) The Origin and Early Diversification
of Land Plants. Smithsonian Institution Press
9 Rothwell, G.W. (1999) Fossils and ferns in the resolution of land plant
phylogeny. Bot. Rev. 65, 188–218
10 Pryer, K.M. et al. (2001) Horsetails and ferns are a monophyletic group
and the closest living relatives to seed plants. Nature 409, 618–622
11 Friedman, W.E. et al. (2004) The evolution of plant development. Am.
J. Bot. 91, 1726–1741
12 Rothwell, G.W. and Nixon, K.C. (2006) How does the inclusion of fossil
data change our conclusions about the phylogenetic history of
euphyllophytes? Int. J. Plant Sci. 167, 737–749
13 Harrison, C.J. et al. (2007) Growth from two transient apical initials in
the meristem of Selaginella kraussiana. Development 134, 881–889
12
Trends in Plant Science Vol.14 No.1
14 Gola, E.M. et al. (2007) Vascular architecture in shoots of early
divergent vascular plants, Lycopodium clavatum and Lycopodium
annotinum. New Phytol. 174, 774–786
15 Harrison, C.J. et al. (2005) Independent recruitment of a conserved
developmental mechanism during leaf evolution. Nature 434, 509–514
16 Floyd, S.K. and Bowman, J.L. (2006) Distinct developmental
mechanisms reflect the independent origins of leaves in vascular
plants. Curr. Biol. 16, 1911–1917
17 Rothwell, G.W. and Serbet, R. (1994) Lignophyte phylogeny and the
evolution of spermatophytes: a numerical cladistic analysis. Syst. Bot.
19, 443–482
18 Jaramillo, M.A. and Kramer, E.M. (2007) The role of developmental
genetics in understanding homology and morphological evolution in
plants. Int. J. Plant Sci. 168, 61–72
19 Floyd, S.K. and Bowman, J.L. (2007) The ancestral developmental tool
kit of land plants. Int. J. Plant Sci. 168, 1–35
20 Singer, S.D. and Ashton, N.W. (2007) Revelation of ancestral roles of
KNOX genes by a functional analysis of Physcomitrella homologues.
Plant Cell Rep. 26, 2039–2054
21 Piazza, P. et al. (2005) Evolution of leaf developmental mechanisms.
New Phytol. 167, 693–710
22 Bharathan, G. et al. (2002) Homologies in leaf form inferred from
KNOXI gene expression during development. Science 296, 1858–
1860
23 Di Giacomo, E. et al. (2008) Characterization of KNOX genes in
Medicago truncatula. Plant Mol. Biol. 67, 135–150
24 David-Schwartz, R. and Sinha, N. (2007) Evolution and development in
plants: bridging the gap. Int. J. Plant Sci. 168, 49–59
25 Boyce, C.K. and Knoll, A.H. (2002) Evolution of developmental
potential and the multiple independent origins of leaves in
Paleozoic vascular plants. Paleobiology 28, 70–100
26 Floyd, S.K. et al. (2006) Evolution of class III Homeodomain-leucine
zipper genes in streptophytes. Genetics 173, 373–388
27 Barkoulas, M. et al. (2007) From genes to shape: regulatory
interactions in leaf development. Curr. Opin. Plant Biol. 10, 660–666
28 Kidner, C.A. and Timmermans, M.C.P. (2007) Mixing and matching
pathways in leaf polarity. Curr. Opin. Plant Biol. 10, 13–20
29 Floyd, S.K. and Bowman, J.L. (2004) Ancient microRNA target
sequences in plants. Nature 428, 485–486
30 Goldshmidt, A. et al. (2008) Signals derived from YABBY gene
activities in organ primordia regulate growth and partitioning of
Arabidopsis shoot apical meristem. Plant Cell 20, 1217–1230
31 Nardmann, J. and Werr, W. (2007) The evolution of plant regulatory
networks: what Arabidopsis cannot say for itself. Curr. Opin. Plant
Biol. 10, 653–659
32 Cantino, P.D. et al. (2007) Towards a phylogenetic nomenclature of
Tracheophyta. Taxon 56, 822–846
33 Hilton, J. and Bateman, R.M. (2006) Pteridosperms are the backbone of
seed-plant phylogeny. J. Torrey Bot. Soc. 133, 119–168
34 Sano, R. et al. (2005) KNOX homeobox genes potentially have similar
function in both diploid unicellular and multicellular meristems, but
not in haploid meristems. Evol. Dev. 7, 69–78
35 Rothwell, G.W. and Lev-Yadun, S. (2005) Evidence of polar auxin flow
in 375 million-year-old fossil wood. Am. J. Bot. 92, 903–906
36 Rothwell, G.W. et al. (2008) A fossil record for growth regulation: the
role of auxin in wood evolution. Ann. Mo. Bot. Gard. 95, 121–134
37 Sanders, H. et al. (2007) Paleontological context for the developmental
mechanisms of evolution. Int. J. Plant Sci. 168, 719–728
