Psilophyton is a “trimerophyte” from the Early Devonian. Main cauline axes display both dichotomous and pseudomonopodial branching as lateral branching systems. First and second order branches bear dichotomous ultimate appendages (twice to thrice isotomous) that are non-planated (Fig. 1A). Lateral branching systems (LBS) are cauline and display radial symmetry. Thus, Psilophyton exhibits a primitive model of LBS and other kinds of LBS may have been derived from it.

Pseudosporochnus is a “cladoxylopsid”; this extinct group was very diverse and widespread during the Devonian. These tree-like plants displayed a cauline erect axis-trunk bearing photosynthetic branches (LBS) in an apical crown (Fig. 1B). The first-order branches were helically arranged and probably deciduous; both first and second order branches dichotomized (“digitate branching”) with helically arranged ultimate units that were tridimensional and several times dichotomous. The primary branches displayed a radial symmetry but the anatomy of other division orders is not well known. Ultimate appendages (UA) are distal ramifications of ultimate units (UU) and divide isotomously (Fig. 1B); they are more or less planated according to Stein and Hueber (1989) but are interpreted as definitely three-dimensional by Berry and Fairon-Demaret (1997) (p. 369). Webbing as well as the presence of a lamina are absent. We decided to treat the lamina as absent and planation as present in this case. Berry and Fairon-Demaret (1997) described Pseudosporochnus LBS but their terminology is different as from the one proposed here: their “LBS” correspond to our “ultimate units – UU” and their “ultimate branching units – UBU” correspond to our “ultimate appendage – UA”. We decided to follow our terminology in order to improve the homogeneity of vocabulary.

Equisetum is the only living genus belonging to Equisetidae (or Sphenophytes). Equisetum leaves are narrow, inserted in whorls around the cauline axis and fused into nodal sheaths. They are neither ramified nor circinate. They possess hydathodes and a single central vein. The latter character is confusing and may explain why the Equisetum leaf has often been described as a microphyll, but one that is not homologous of the microphyll of Lycopodiidae, also named lycophyll (Schneider et al., 2002). Sphenophyllum is an extinct sphenophyte genus belonging to Sphenophyllales. In contrast to that of Equisetum, the Sphenophyllum leaf is broad with a well developed lamina. Leaves are heterophyllous, spatulate and present wide ranges of variation in leaf size. Two veins are often present at the base of the leaf and bifurcate several times. Archaeocalamites is a late Devonian member of the Calamitaceae displaying leaves arranged in whorls on the distal branches and dichotomizing one to three times (Jennings, 1970).

The leaves of “zygopterid” ferns were helically borne on a creeping rhizome (Fig. 1C); they show a quadriseriate arrangement of pinnae (i.e. the pinnae were inserted as alternating pairs on either side of the primary rachis). The frond was up to four times divided and pinnules were small. Petiole anatomy (i.e. phyllophore type) is remarkable, the foliar trace presenting two perpendicular planes of bilateral symmetry. This anatomy is not present in any living fern group.

“Botryopterid” ferns, from the Early Carboniferous to the beginning of the Permian, exhibited an important foliar diversity. The leaves, helically borne on the stem, were at least six times pinnate. The petiole trace had bilateral symmetry and exhibited a unique anatomy, characteristic of the genus. The oldest species, Botryopteris antiqua, showed a quite narrow lamina (almost absent) whereas B. forensis displayed small pinnules.

Osmunda presents a short erect stem bearing roots and erect circinate leaves with persistent bases. The leaf is once to twice pinnate and is representative of the standard fern megaphyll, also called a frond. The frond is divided into opposite pinnae inserted on a primary central rachis, more or less in a single plane, and possesses bilateral symmetry. Each pinna has determinate growth and is composed of a number of pinnules, which are the laminate distal appendages. We propose here that the diversity of fronds observed in living ferns represents variation from this standard: extreme reduction in water ferns, such as Azolla spp. (Azollaceae), or more divided fronds in many species, such as the bracken fern, Pteridium aquilinum (Dennstaedtiaceae), or non divided leaves with an entire broad lamina as in Asplenium (Phyllitis) scolopendrium (Aspleniaceae), or with a peculiar indeterminate growth as in Lygodium (Lygodiaceae), or with a pseudodichotomous branching in Gleicheniaceae (Smith et al., 2006), etc.

Archaeopteris is an extinct tree that lived during the Late Devonian. It was the main species of the first widespread Devonian forests. It has been interpreted as one of the earliest known modern trees with trunks showing massive wood and secondary phloem. The trunk produced deciduous LBS consisting of two orders of branches. Laminate or dichotomous non-laminate photosynthetic ultimate appendages – traditionally interpreted as small leaves (Beck, 1971) – were helically borne on ultimate (B2) and penultimate branches (B1) (Fig. 1D). B1 and B2 display radial symmetry whereas laminate or non-laminate ultimate appendages exhibit bilateral symmetry and a complex dichotomous venation.

Elkinsia is an early Spermatophyte from the Upper Devonian; the slender erect stem bore helically arranged LBS interpreted as dimorphic leaves branching dichotomously to pinnately and presenting a diversified morphology (Fig. 1E). Fertile leaves were tridimensional whereas the vegetative ones displayed very small laminate pinnules at their distal ends. Fertile leaves seemed to have an apical location on the main axis. We have to specify that the reconstruction of Elkinsia is based on dispersed elements but we assume the presence of pinnules on vegetative leaves.

Foliar characters have already been used several times in phylogenetic analyses, especially in ferns (Kenrick and Crane, 1997, Pryer et al., 1995, Rothwell, 1999, Schneider, 2007, Schneider et al., 2009 and Stevenson and Loconte, 1996). However, because of an unequal distribution of taxonomic sampling (including or not fossil taxa and/or modern lineages), the treatment of foliar organs (i.e., selection of descriptors and coding of states) is also quite biased or contrasts among studies.

The term megaphyll was described by Kenrick and Crane (1997) as leafy appendages of seed plants, ferns and basal Equisetiidae. True megaphylls are characterized by a developed lamina and complex venation patterns. Although they are typically of large size with a pinnate organization, megaphylls show an extreme diversity in terms of form and size (Kenrick and Crane, 1997).

Some foliar characters traditionally used in the literature are not used in the present study. For example, circinate vernation is a character considered as the main criterion defining extant ferns (Polypodiidae, Marattiidae and Ophioglossidae). However, circination is not a character common to the whole of ferns: for instance there is no standard circinate vernation in Ophioglossidae (Smith et al., 2006). Furthermore, the rarity of young crosiers in fossils makes it difficult to confirm circinate development in extinct taxa (Schneider et al., 2009).

To discuss the foliar organ concept, we propose that all lateral photosynthetic organs (including leaves) be considered to be lateral branching systems (LBS, Fig. 1). The comparison of their characters will allow a better discrimination between different kinds of foliar organs (from unwebbed lateral branches to megaphylls). However, the anatomical description of some of these lateral branching systems remains unclear. The variability of foliar organs thus must be described in detail in order to collect as much information as possible. In this respect, we propose to focus principally on four anatomical characters of importance regarding foliar organs (versus cauline axis). Some have been already discussed in previous studies (Galtier, 2010 and Sanders et al., 2009). We have to point out that phyllotaxy is also a major criterion characterizing leaves. We have chosen, however, not to use it in our consideration of the LBS concept. Indeed, this character describes developmental reiteration of leaves on the cauline axis; because it can often be difficult to distinguish foliar from cauline organs on fossils, we thought that the use of this character would have been misleading. To be rigorous, we should have considered the insertion of ramifications on each order of the LBS (B1, B2, … Fig. 1) for which the distinction between cladotaxy and phyllotaxy is unclear.

Symmetry of lateral branching systems: The presence of one or two planes of symmetry characterizes bilateral symmetry (Fig. 2). Bilateral versus radial symmetry is defined from the cross section of vegetative organs depending of the position of protoxylem strands and of the global shape of the vascular supply of the LBS or the leaf (Galtier, 2010). Comparing symmetry of stem, LBS and leaf permits identification – or not – of an anatomical distinction between organs. It is difficult to distinguish between a foliar or cauline nature of lateral ramifications, especially in putative basal groups like “cladoxylopsids”. We also must note that the bilateral symmetry of lateral branching systems is not homologous with the dorsiventrality of rhizomes, which results in their prostrate subterranean habit. Symmetry of LBS is defined here for the first-order branches (B1 in Fig. 1).

Abdaxity: Fig. 2 brings to light differences between bilateral symmetry and abdaxity with the examples of fern and seed plant petioles. In the ferns Osmunda and Psalixochlaena and in the seed plant Elkinsia, the leaf trace possesses a single plane of symmetry with a clear distinction between abaxial (inferior) and adaxial (superior) faces. This differentiation is supported by an adaxial concavity on the rachis. In contrast, there are two perpendicular planes of symmetry in the petiole or phyllophore of Zygopteris preventing the differentiation of abaxial from adaxial face; in this case there is no abdaxity. Recognition of this aspect of lateral-appendage organization also depends on determination of the position of the protoxylem strand, which is adaxial (abdaxity S) in ferns versus abaxial (abdaxity I) in seed plants and Archaeopteris. Abdaxity is a new term, introduced here, for a character similar but more complete than the “abaxial/adaxial identity” of Sanders et al. (2009), it is defined conjointly by: •

the presence of a single plane of symmetry;

Furthermore, bilateral symmetry and abdaxity are defined for the first-order branch of the leaf in this study, but they can be different in the more distal branching orders. That is why these characters generally vary along leaf ramification divisions. Notice that the “leaves” of Tmesipteris lack a petiole and thus the character is not applicable.

In this study, we described the lamina through two characters: planation and webbing (Zimmermann, 1938). However, caution is required in the observation of these descriptors on fossils because taphonomic processes attendant fossilization can confer the appearance of flattened and webbed morphology (Chaloner, 1999). For the present study, the absence or presence of planation and webbing is coded for the whole LBS: it can as well be borne on first-order branches as on ultimate appendages.

Planation: planation is the transition from three-dimensional to bi-dimensional terete segments. Planate organs are thus flattened branching axes organized into a single plane (Beerling and Fleming, 2007).

Webbing: webbing consists of filling in flattened branches with laminar tissue. In developmental terms, the webbing of telomes to produce a laminate leaf blade involves, first, the production of lateral outgrows and, second, fusion of adjacent branches (Beerling and Fleming, 2007). The acquisition of planation then webbing induces the development of the lamina.

It is necessary to rely on relevant phylogenetic analysis to understand megaphyll evolution and especially to define one or several kinds of foliar organs among Euphyllophytes. Consequently, we propose to use one phylogenetic tree illustrating the main consensus hypotheses published by Schneider (2007). The tree presented and used here (Fig. 3 and Fig. 4) is synthetic, meaning it is not exactly the same as any of the published ones, in order to provide a larger and summarized view of recent phylogenetic knowledge and hypotheses on the subject (see below). The characters tested (symmetry, abdaxity, planation and webbing) were added after the fact by the authors in order to permit discussion of their evolution.

We arbitrarily added some fossil representatives not present in Schneider (2007) (such as “cladoxylopsids”, “zygopterids”, “botryopterids” and “anachoropterids”) in order to increase the diversity of fossil foliar forms. We also removed some living taxa to reduce the sampling and to focus our study on basal lineages. Our synthetic tree proposes the monophyly of Monilophytes, but with paraphyletic “cladoxylopsids” (Pseudosporochnus and Ibyka) as sister to a clade including Equisetidae and ferns. In Schneider's supertree (fig. 2 in Schneider, 2007), Ibyka appears as closely related to Equisetidae, confirming a traditional view (Stewart and Rothwell, 1993). We prefer the hypothesis of Taylor et al. (2009) who proposed that Iridopteridales (including Ibyka) share more characters with Pseudosporochnales (“cladoxylopsids”) than with Equisetidae. In the study of Schneider (2007), other “cladoxylopsids” are lacking, and relationships of Ibyka with traditional “cladoxylopsids” were thus not tested. In the maximum parsimony analysis combining fossils and living taxa (Fig. 1A in Schneider, 2007), Equisetidae display an unresolved position within the Monilophytes. We propose to regroup Equisetidae with ferns in order to keep the hypothesis that horsetails would be embedded within ferns in accordance with other phylogenies focused on living taxa and including molecular data (Qui et al., 2006 and Schneider et al., 2009). Furthermore, a clade that groups together some Paleozoic extinct ferns (Botryopteris and Psalixochlaena) is proposed to be related to Osmunda and Polypodium, and represents extinct Polypodiidae or leptosporangiate ferns (Taylor et al., 2009). In addition, the Euphyllophytes are rooted by extinct Aglaophyton (non vascular Polysporangiophyte), extinct Rhynia (“rhyniopsids”) and living Lycopodium (Lycopodiidae), which together represent the basal Tracheophytes. We also added Psilophyton to the Euphyllophyte sample because this genus is a key taxon when considering early LBS diversity. Lignophytes are represented in Schneider's tree by Archaeopteris, Pinus and Cycas; we propose to insert the additional Aneurophyton, Elkinsia and Medullosa to increase fossil representatives. The positions of extinct Spermatophytes are in accordance with Hilton and Bateman (2006).

In order to study the evolution of the selected supposed foliar organ characters within Euphyllophytes we used Mesquite version 2.74 software (Maddison and Maddison, 2010), which allows the inferred history of each selected character to be traced on the phylogenetic hypothesis by applying the maximum parsimony (MP) criterion and by treating characters as unordered and unweighted. If inference for a character on the tree provides several character distributions (resulting in an ambiguous state for some nodes), all optimizations (including ACCTRAN and DELTRAN) are tested. The four characters of interest were coded as follows: •

symmetry of LBS: non-applicable, radial, bilateral;

Symmetry and abdaxity are defined for differentiated lateral branches (related to pseudomonopodial growth). Outside Euphyllophytes, both characters are coded as non-applicable because taxa do not have pseudomonopodial growth. Coding of the selected characters for all the taxa is reported in Table 1, determined by personal observations on Paris, Liège and Montpellier collections and from the published data mentioned above.

Radial symmetry is inferred as ancestral in Euphyllophytes.

The Maximum Parsimony (MP) proposes a single scenario (Fig. 3A): one appearance of bilateral symmetry in Spermatophytes and one appearance in ferns within Monilophytes. The non-applicable status for Psilotum suggests a secondary loss of the LBS rather than illustrating a plesiomorphic state preceding the acquisition of pseudomonopodial growth.

Absence of abdaxity is inferred as ancestral in Euphyllophytes.

A single MP scenario proposes two appearances of abdaxity (Fig. 3B): once in Spermatophytes and once in ferns with reversal in Rhacophyton and Zygopteris. The inclusion of Tmesipteris and Psilotum in ferns suggests that abdaxity was present in ancestors but can no longer be observed. Peculiar foliar features or absence of leaves in, respectively, Tmesipteris and its close relative Psilotum are proposed as the result of reduction (minimization), which can thus be interpreted as an evolutionary reversion.

Absence of planation is inferred as ancestral in Polysporangiophytes (Fig. 4A).

Planation would have appeared in Lycopodium (with the appearance of the lycophyll), in a clade grouping Archaeopteris and Spermatophytes and in Monilophytes with two equally parsimonious scenarios: one appearance in Monilophytes with one reversal in Ibyka (ACCTRAN optimization) or one appearance in Pseudosporochnus and one appearance in ferns (DELTRAN optimization). The absence of planation in Psilotum is explained by a regressive loss of the leaf.

Absence of webbing is inferred as ancestral in Polysporangiophytes.

Webbing would have appeared three times in Tracheophytes (Fig. 4B): in Lycopodium (resulting in the lycophyll), in a clade grouping Archaeopteris and Spermatophytes, and in ferns (the absence of webbing in Psilotum is again explained by a regressive loss of the leaf).

Cauline lateral branches result from anisotomous branching of the main axis. They are not plagiotropic and are not necessarily specialized for the same functions as extant leaves (photosynthesis, respiration, evapotranspiration), and are thus distinct from the next trait, lateral branching systems (LBS). For instance, lateral systems of Psilophyton (Fig. 1A) are branched cauline axes that end in ultimate appendages. They correspond to the cauline ramified system sensu Gerrienne (1997).

Although this term has been widely used in the literature, we propose here a new meaning in order to permit comparison of the different kinds of foliar organs within the Euphyllophytes. “LBS” encompasses every organ involved in the principal functions of leaves within extant organisms (photosynthesis, respiration, evapotranspiration) regardless the morpho-anatomy, except for the case of organs with a radially symmetrical stele. Plagiotropic ramifications, possibly deciduous, as for instance, the penultimate branches of Archaeopteris, typically belong to this latter category. The lateral branching system is a ramification borne lateral to a main axis, with cauline and/or foliar parts, and possibly bearing ultimate appendages on distal extremities. This LBS definition encompasses the different kinds of foliar organs and facilitates the comparison between homologous structures. LBS can encompass true stems bearing leaves. For example, B1 of Pseudosporochnus are true stems/branches and ultimate appendages of Archaeopteris are true megaphylls (with or without lamina). The organs described below (quadriseriate leaf, pinnate leaf, …) are subsets of LBS.

Ultimate appendages are small, dichotomous, ultimate branches, mostly non-planated and unwebbed (Fig. 1B, UA). They are borne on ultimate units (Fig. 1B, UU) that are helically arranged along and up to the distal extremities of proximal appendages (= LBS) in “cladoxylopsids” and “progymnosperms”. Anatomy of ultimate appendages is distinct from the stem anatomy and their morphology can be variable in “cladoxylospids”. Some of them are similar to terete cylindrical cauline (telomic) axes (Ibyka) whereas others can be flattened (Pseudosporochnus) or even possess bilateral symmetry and a lamina varying from small to broad, as in Archaeopteris; in this latter case they are currently interpreted as leaves or “small megaphylls” (Galtier, 2010). The term ultimate appendages also encompasses dichotomously branched, “megaphyll precursors” (sensu Kenrick and Crane, 1997) present in basal Euphyllophytes.

This term defines the case where secondary axes are inserted in pairs alternately on the main rachis revealing the existence of two planes of bilateral symmetry but no abdaxity (Fig. 1C). A lamina may be present on distal divisions of the organ in the form of generally narrow pinnules, as in Zygopteris (Galtier, 2010). Such organization has not yet been observed in Lignophytes.

We define this as a large branched plagiotropic lateral organ with a developed lamina (not only restricted to distal appendages). This corresponds to the lateral, webbed appendages found in all living Euphyllophytes. Applied to living organisms, the megaphyll is a vascular organ with determinate growth, bilateral symmetry (and abdaxity) definitely organized on the stem in some kind of phyllotaxy (Sanders et al., 2009 and Tomescu, 2008). Taking into account fossil taxa and additional anatomical characters, we will be able to check the validity of this megaphyll concept within the whole Euphyllophyte clade, or show that this concept hides a diversity that is rather explained by various character combinations illustrating contrasting histories and processes. Characters discussed in the present study are insufficient to distinguish Lignophyte leaves and Monilophyte leaves on the basis of major anatomical differences (Fig. 2) between abdaxity S (Monilophytes) and abdaxity I (Lignophytes). However, the literature suggests some differences, which can be summarized as follows: •

megaphyll: Lignophyte leaves in which the adaxial face is associated with a lateral cauline meristem and in which both cellular differentiation and tissue maturation are basipetal (Tomescu, 2008). This pattern differs from that found in other Euphyllophyte groups;

The lycophyll (Kaplan, 2001 and Schneider et al., 2009) is a Lycophyte apomorphy, not observed in Euphyllophytes. It is a lateral organ with simple venation and limited development of laminar tissue, straight or forked, with one vein or, in some instances, two running in parallel. The so-called “microphyllous leaf” of Sphenophytes is not homologous with the lycophyll and must be studied in the context of frond evolution in Monilophytes. It is generally accepted that the “unbranched microphyllous leaf” of Equisetum and advanced calamitean fossils evolved by reduction from the dichotomous leaf of archaeocalamiteans (Stanich et al., 2009).

The different types of organs defined above exhibit the combination of anatomical and morphological traits as summarized in Table 2, with corresponding taxa noted. All of the characters proposed are observable on both fossil and/or living plants. The presence of a megaphyll or a megafrond is often supported by the presence of a lamina, and consequently by the presence of planation and webbing. We propose here to overcome this viewpoint, defining the presence of a leaf by, at least, the presence of planation (such as the ultimate appendages of Pseudosporochnus). Conversely, bilateral symmetry is a necessary criterion but not sufficient to define megaphyll or megafrond. Abdaxity is an important anatomical character to identify modern leaves but we showed that some flattened and laminate leaves (such as quadriseriate leaves of “zygopterids”) do not display abdaxity. As with abdaxity, the presence of a webbed lamina is not sufficient to define a foliar organ in a broad sense even if abdaxity and webbing are required to identify megaphyll and megafrond. Finally, bilateral symmetry and planation are necessary to define a leaf sensu lato; abdaxity and webbing are necessary to characterize megaphyll and megafrond (even if they can be reduced in some derived taxa) (Table 2). The four characters discussed are essential to define the leaf concept but we must not lose sight that to be considered leaves, LBS have to be repeated over and over in an orderly arrangement on the stem, initiated by apical meristematic activity (phyllotaxy).

Considering the origins of the leaf sensu lato, the question is: would those characters have appeared simultaneously or not during the land plant evolution? By inferring evolution of anatomical and morphological traits, independently of the organs, we thus have the possibility to examine evolutionary relationships and/or potential pathways between the diverse observed organs.

The appearance of bilateral symmetry of lateral branching systems is a first important stage regarding the differentiation of cauline and foliar organs. The presence of bilateral symmetry reveals a modification of LBS anatomy in Monilophytes. Indeed from the radial symmetry of cauline axes, lateral ramifications developed one or two bilateral symmetry planes (as observed in extinct “zygopterids” and Rhacophytales for the second case; Phillips and Galtier, 2005). In addition to the bilateral state, the number of symmetry planes should also be taken into account. In the present state of knowledge, the “zygopterids” and Rhacophytales appear to be the earliest ferns, and likely sisters to other ferns (Galtier, 2010). This could suggest that the symmetry in one plane was derived from the model with two planes, likely by the reduction of the phyllophore. In this hypothesis, the “true” megafrond in ferns would thus be derived from quadriseriate lateral organs. Such an interpretation implies that relationships of “zygopterid” ferns and Rhacophytales to the other ferns (particularly early botryopterids) are better known than they actually are. Phylogeny in Monilophytes, and especially in ferns, including fossil taxa needs to be improved. The genetic background for morpho-anatomical characters already may have been present in some groups even if the morphological expression of that genetic background had not yet been expressed or was not yet clearly observable (Endress, 2010). Consequently, the previous hypothesis can be tempered because the first appearance of bilateral state can occur earlier than the base of the fern clade.

If we define the megaphyll by its symmetry, then the megaphyll sensu lato would have appeared twice: once (the megaphyll) in Spermatophytes and once (the megafrond) in ferns. However, the study of this character needs to be improved because it was, up to now, mainly defined for first-order branches. Indeed, some taxa (e.g. Pseudosporochnus) have no anatomical distinction between first-order branches and the cauline axis. Rather, a bilateral symmetry is observed in second order branches (Stein and Hueber, 1989). Thereby, it is necessary to distinguish branch orders in order to discuss the evolution of this character and avoid misinterpretation.

Abdaxity is an essential character to understand the evolution of leaf anatomy. Evolution of abdaxity and symmetry is fully congruent in Spermatophytes. Besides, we can notice that abdaxity is absent from first-order branches in Archaeopteris but present in its ultimate appendages (abdaxity I) as shown in Fig. 2, suggesting that anatomical changes first occurred in the last-order branch systems. For this reason, we can assert that Archaeopteris leaves are not homologous with Spermatophyte megaphylls. Interpretation of the evolution of abdaxity in ferns depends on the phylogenetic position of the “zygopterids” and the Rhacophytales. Because of phylogenetic irresolution, the absence of abdaxity in “zygopterids” and Rhacophytales is interpreted as two independent reversals. However, if such taxa are sisters to other ferns, as already suggested here above, abdaxity would have appeared in a clade grouping all ferns but excluding “zygopterids” and Rhacophytales, and thus after the appearance of bilateral symmetry. This suggests that the megaphyll could be defined by the strict combination of both these anatomical features only in Lignophytes but not in Monilophytes. Here again, new phylogenetic hypotheses including fossils are needed to clarify the evolution of abdaxity and its relation to symmetry in ferns.

Lamina is traditionally defined as the result of planation and webbing processes (Zimmermann, 1938). It is a planated two-dimensional structure likely to be present on the whole leaf or restricted to distal portions. Phyllods and cladods of cauline origins but with foliar function are essentially defined by their planate form converging on being a lamina (Bell, 2008). Thus a lamina or a lamina-like structure does not necessarily mean the presence of a leaf in absence of other evidence. Present data (Stein and Boyer, 2006) suggest that the development of veins and laminae and the transition from axes to veins are progressive. There is no word to designate a structure displaying planation without webbing as in Pseudosporochnus. Furthermore, where is the limit between a broad lamina and slightly webbed appendages? To address these issues, further phylogenetic analysis will be required in order to better define the concept of “lamina” and understand its evolution. Leaves are usually defined by the presence of a lamina, an interpretation based on observations that can be quite subjective (Sattler and Rutishauser, 1997). A standard megafrond is a compound leaf with numerous branching orders and a narrow distal lamina. In this case, it is often difficult to distinguish between a real absence of a lamina (i.e. a structure that does not yet exist) and an absence of a lamina that has secondarily disappeared. There are numerous examples of living ferns with fronds that lack laminae, the result of regressive loss, interpreted as hygrophilous adaptive strategies (e.g. in Hymenophyllaceae and the genus Abrodictyum; Ebihara et al., 2006). It is thus important to compare in detail the organisms without laminae with their putative relatives in order to examine the possibility that the character has been lost.

Bilateral symmetry of first-order branches would have appeared after planation within Lignophytes and Monilophytes (with the ACCTRAN opitimization): the progymnosperm Archaeopteris displays planation and webbing only for its ultimate appendages (= vegetative leaves) and Pseudosporochnus exhibits planation only at distal ends of ramifications. Therefore, in these examples, the appearance of planation and bilateral symmetry was concomitant only in last-order branching (ultimate appendages). However, planation would have appeared together with webbing in a clade grouping Archeopteris and Spermatophytes. The evolution of these characters is different in Monilophytes: both are present for the whole of ferns and horsetails but are not present in “cladoxylopsids” (Ibyka and Pseudosporochnus). Indeed, planate ultimate appendages are present in Pseudosporochnus but not in Ibyka. Webbing is absent in “cladoxylopsids” and seems to appear once in the clade grouping horsetails and ferns. With the ACCTRAN optimization, planation would have been acquired before webbing (with reversal in Ibyka). With the DELTRAN optimization, the planation observed in some “cladoxylopsids” would be not homologous with planation observed in ferns and horsetails (it would be an example of a convergence). The “cladoxylopsid” planation is restricted to ultimate appendages, whereas planation is observed on the whole lateral organ in most ferns (except Rhacophytales and “zygopterids”) and horsetails, supporting the hypothesis of absence of homology. In this case, the combination of webbing and planation defines the megafrond only in the ferns and horsetails and not in the “cladoxylopsids”. The form of the leaves in “cladoxylopsid” makes this group key for understanding the evolution of the lamina, at least in Monilophytes. In available phylogenies, these taxa, whose monophyly is still questioned, are under-represented. Further studies should integrate additional “cladoxylopsids” in order to refine our understanding of their relationships, and to determine more accurately the phylogenetic location of planated structures in this lineage.

Phylogenetic inferences show that the definition of a “true” leaf, as a combination of anatomical changes and planation and webbing processes, cannot be easily applied in Lignophytes because the laminar surface would have appeared before the anatomical changes necessary to support it functionally. Actually, this observation is due to the character coding of Archaeopteris: bilateral symmetry and abdaxity have been coded as absent because these characters were defined for first-order branches of LBS. However, if we had considered only the vegetative leaves of Archaeopteris (Fig. 1D), bilateral symmetry and abdaxity would have been coded as present. Finally, the combination of anatomical changes and planation and webbing processes were concomitant in vegetative leaves of ultimate branches of LBS, at least in Archaeopteris.

In Monilophytes, bilateral symmetry could have appeared concomitantly with webbing and planation processes in ferns (including horsetails), if we suppose that planation observed in some “cladoxylopsids” is not homologous with that observed in ferns. This evolution suggests that, among the four discussed characters, planation is the first necessary step through the emergence of the leaf at the Euphyllophyte scale. Furthermore, bilateral symmetry combined with appearance of the lamina (planation and webbing) was present before the appearance of abdaxity, at least in “zygopterids” and Rhacophytales, if we assume that both of these taxa are sisters to other ferns (as we discussed here above).

Concerning Tmesipteris, the symmetry of the lateral organ was a priori coded bilateral because of the planate shape and the single central vein, but we cannot exclude here that this organ could be, in fact, given its absence of clear abdaxity, a cladod (i.e. a planate stem with a foliar function). Tmesipteris and its relative Psilotum are now well recognized as true ferns, but exhibiting particular morphology illustrating numerous regressive processes (loss of roots and reduction of leaves) (Smith et al., 2006).

In this review we emphasized that Lignophytes and Monilophytes do not display homologous leaves even if both are traditionally called megaphylls (Tomescu, 2008). In order to remove ambiguities about foliar differences between Monilophytes and Lignophytes, we proposed to distinguish the two kinds of leaves by use of the terms megafrond, the leaf observed in Monilophytes, and megaphyll, now restricted to the Lignophytes, (as suggested here above). Besides, in the absence of megaphyll or megafrond, some ultimate appendages of extinct “progymnosperms” and “cladoxylopsids” display distinct and various forms, also suggesting analogy rather than homology. We note that the term megafrond could be a priori not appropriate for narrow leaves of Sphenopsids even though phylogenetic analysis shows that leaves of Sphenophytes and ferns are homologous. Sphenophyte leaves are thus interpreted as regressed megafronds.

Galtier (2010) argued that leaves evolved from two complex and concomitant processes. On one hand there is a transition from radial to bilateral symmetry, linked to abdaxity. On the other hand, the blade resulted from planation and webbing. The independent evolution of the two processes would be illustrated by the morpho-anatomy of “zygopterids” and some “botryopterids”, and also of Elkinsia and early seed plants. This interpretation is based mainly on paleobotanical studies without any phylogenetic considerations. Our inferences do not contradict this hypothesis by suggesting the scenario in which lamina (defined sensu lato) did not necessarily evolve concomitantly with anatomical changes in Euphyllophytes. Hypotheses proposed by our phylogenetic inferences encourage additional evolutionary studies that could benefit from more complete phylogenies and the new vocabulary proposed here. For example, it would allow examination of the independence between planation processes and anatomical changes in distinct clades, especially in Monilophytes. It would also emphasize the necessity to treat phyllotaxy as a combination of branching patterns of different ramification orders. Indeed, this character should be one of the first of importance involved in the origin of leaves.