Six axis orders are present among the studied material. It is impossible to reliably assess whether the most proximal and largest axis is the main stem or not. Considering this, we choose to name N the most proximal observed axis order, and N + 1, N + 2, etc. the subsequent axis orders. In transverse section, all the axes appear round to oval in outline. Lower-order axes include a four-lobed xylem strand, with phloem located between the xylem arms. Two planes of symmetry can be distinguished (Fig. 1A). The main plane of symmetry, referred to as the primary plane of symmetry (PPS, Fig. 1A and D) passes through the two main phloem groups. The secondary plane of symmetry (SPS) is perpendicular to the first; it determines the direction of departure of paired N + 1 axes, which are given off alternately.

The diameter of the basalmost axes (N-order) ranges from 3 to 5 mm along both planes of symmetry. They present a stele (about 2 mm in size, elongated in the PPS) composed of four lobes (Fig. 1A and D). The latter are arranged crosswise and separated by parenchyma in the central area. The xylem includes tracheids with a larger diameter toward the centre of the stele and tracheids with a smaller diameter in the outer region. The largest tracheids are interpreted as metaxylem, and the smallest as protoxylem. At least four protoxylem strands are present (arrows, Fig. 1D).

The protoxylem is slightly mesarch and located near the tip of each lobe. Phloem occurs in the bays between xylem lobes. The larger phloem clusters occur along the PPS and are characterized by 4 to 7 (30 to 50 μm in diameter) cells (PH, Fig. 1D) which may represent sieve cells. Smaller, and often less well preserved, phloem clusters occur along the SPS and consist of only 2 to 4 (30 to 20 μm in diameter) cells. The pattern of emission of traces to the N + 1 axes conforms to previous descriptions; departing traces of two axes N + 1 are first visible within the cortex (tN + 1 A and B, Fig. 1A). In addition, two small traces (tAP A and B) are also present; they arise very proximally from the traces of N + 1 axes. They belong to organs traditionally referred to as aphlebiae. These aphlebiae are profusely branched, more or less dichotomous, and nearly 1 cm long; they have been partially reconstructed in Fig. 3.

N + 1-order axes are significantly smaller than N-order axes. They range from 2.3 mm to 3.2 mm in diameter (Fig. 1B and C). They possess a massive tetrarch xylem strand without central parenchyma. Like in N axes, the protoxylem occurs at the tips of the metaxylem lobes and the phloem is located between the xylem arms. The departing traces of two N + 2 axes are visible within the cortex (Fig. 1B) as well as one aphlebia (AP). Fig. 1C corresponds to a more distal section of the same N + 1 axis; it illustrates the coalescent base of one pair of N + 2 axes as well as the departing traces of the subsequent/second pair of N + 2 axes, in an alternate arrangement.

The N + 2 axes measure 1.1 to 1.6 mm in diameter and are structurally almost the same as to those of the branches of previous orders. However, they present a subtriangular xylem strand but four protoxylem poles are still visible (Fig. 1E). Quadriseriate branching with the occurrence of basal aphlebiae is characteristic of these axes as documented by Bertrand (1909, plate VI, 32–33).

A more important change in xylem shape occurs in more distal branches. N + 3 axes (Fig. 1F–G) are much smaller in size, ranging from 1 × 0.8 mm to 0.5 × 0.3 mm, and they possess small triangular xylem strands. However, these small axes produce alternate pairs of N + 4 axes with a minute trace; this feature was documented by Bertrand (1909, plate VI, 37) and Chaphekar (1962, fig. 4).

Subsequent branching orders (N + 4 and N + 5 axes) show a general decrease in overall dimensions, as well as a strong diminution of xylem strand complexity that is progressively reduced to a small number of tracheids (Fig. 1F and G). Branching of these distal axes is dichotomous (iso- or anisotomous). We consider that the smallest axes (about 0.1 mm in diameter) correspond to the ultimate axis orders (Fig. 1H). Due to their similar size and anatomy, it is impossible to distinguish the distalmost axes from ultimate branches of aphlebiae.

The cortical tissues are very well preserved in all orders of axes. The inner cortex is a parenchymatous ground tissue consisting of cells decreasing in diameter toward the periphery (Fig. 1A, B, E and F). The outer cortex, darker in colour, is composed of smaller and, more sclerotic, thick walled cells. This cortical zonation is observed in axes of orders N to N + 3. By contrast, the more distal axes (N + 4 and N + 5) lack the sclerified zone (Fig. 1G–I).

In all branching orders, the periphery of the cortex is occupied by a narrow zone of spongy parenchyma composed of elongate and thin-walled cells separated by large intercellular spaces; it directly underlies a prominent epidermis (Fig. 1E, J–K). We interpret this tissue as a hypodermal aerenchyma. The development of this tissue causes folding and wrinkles of the axis surface (Fig. 1A, E, J). The aerenchyma, though reduced to a single cell layer, is also visible in the distalmost axes as confirmed by the longitudinal section, (Fig. 1I). The epidermal cells are square or rectangular in transverse section, measuring about 20 × 20 μm (Fig. 1E, H–K). Such epidermal cells are present even in the smallest axes. The stomata, similar to those illustrated by Bertrand (1909), have been rarely observed and only in ultimate axis order and aphlebiae.

We were able to follow and reconstruct an N + 1 axis from its departure from an N axis to its tip (Fig. 2 and Fig. 3). It measures slightly more than 3 cm long and bears alternately two successive pairs of n + 2 laterals with aphlebiae (Fig. 2). Proximally, this axis measures about 1 mm in diameter. The first branching occurs at about three centimetres from the N axis. The attachment of one pair of N + 2-order axes is illustrated in Fig. 2A, where the section is slightly oblique. The two vascular strands of the N + 2 axes are visible within the cortex, together with the corresponding pair of aphlebia traces (tAP). The transverse section of the N + 1 axis becomes circular just after emission (Fig. 2B). Unexpectedly, the backward oriented apical region of the same N + 1-order axis can be seen on the right (arrow, Fig. 2B) near one branched aphlebia. A series of longitudinal oblique sections of the apical region of the N + 1 axis can be seen on Fig. 2C–E. These peels demonstrate the connection between the two regions of the curved N + 1 axis. Vascularization is visible nearly up to the apex. However, the poor preservation prevents any detailed description of the topmost apical region. We notice the presence of an irregular surface at the lateral left side of this backward oriented apex (arrows, Fig. 2D). This may represent poorly preserved primordia of N + 2 axes to be borne distally. The two (proximal and distal) regions of the vascular strand are first visible side by side (vN + 1, Fig. 2E), and then connected (Fig. 2F). The tracheids are cut more longitudinally indicating that this section is located within the fold of the curved axis. The Fig. 2G illustrates a section situated higher within the fold and a single vascular strand is visible. This figure shows also the common base of a second pair of N + 2 axes. These downward oriented axes are free in Fig. 2F; they are very small, as expected in immature axes; however they are occurring on the expected side. This is an unconventional position considering that the preceding emission of the first pair of N + 2 axes occurred on the same side (Fig. 2A). Indeed, the typical quadriseriate branching of Stauropteris corresponds to the alternate emission of paired axes as illustrated in Fig. 1C. The apparent attachment of the two successive pairs of N + 2 axes on the same side results from the strong curvature of the top of the N + 1 axis and of the displacement of the second pair of N + 2 axes. This is confirmed by the ultimate sections of the series within the recurved left side of cortex. On the Fig. 2H, the two vascular traces of the N + 2 axes are the only ones present in the cortex; they are cut longitudinally through this curved cortical side. Beyond this level, only smaller and smaller tangential sections of the cortical region are visible and the top of the curve is reached at Fig. 2I. We provide on Fig. 3 the detailed reconstruction of a portion of S. oldhamia where one large N axis is bearing a pair of N + 1 axes with their basal aphlebiae. One N + 1 axis was fully developed (about 3 cm long) up to its first ramification, i.e. the level of attachment of the first pair of N + 2 axes with corresponding aphlebiae. Beyond this point, the N + 1 axis measures only about 0.5 cm long up to its apical region; it shows the attachment of a second pair of typically immature and displaced N + 2 axes, then it terminates as an attenuated conical apex. Information about the morphology and eventual branching of the first pair of N + 2 axes is missing.

In this paper, we confirm the distinctive organization of the plant comprised of 6 successive orders of axes, ranging in diameter from 5 mm to 0.1 mm. The vascular system also decreases in size and complexity from four-lobed xylem strands to very small terete strands in the terminal axes. S. oldhamia is here interpreted as an orthotropic photosynthetic axis bearing determinate lateral branches, the whole exhibiting a bilateral symmetry. Due to the lack of information on the basalmost/proximal region (including roots), it is suggested that the monotypic Stauropteris assemblage contained in the studied coal ball was transported, and represented only the aerial part of the original plant. Actually, we failed finding evidence of attachment of the largest (N-order) axes either to any rooting organ, or to a plagiotropic shoot and we are reduced to speculation about general morphology of the plant. In the last hypothesis, S. oldhamia with its orthotropic photosynthetic branching system, could be compared to the living Psilotum which is sometimes viewed as a “living rhyniophyte” (Bierhorst, 1971), an interpretation challenged by Kaplan (2001) who considered that the simplified nature of the leaf of Psilotum is suggestive of an evolutionary reduction rather than an enation homology. S. oldhamia is however clearly distinct from Psilotum in its repetitive quadriseriate branching system of bilaterally symmetrical axes and in the absence of any small foliar organ. Nevertheless, due to the lack of information concerning a number of characters, the stauropterids appear as a basal clade of ferns, near the “trimerophytes” and the Psilotophytes, in some recent phylogenies (e.g., Rothwell, 1999).

An alternative interpretation is to consider S. oldhamia and the other stauropterids as demonstrating initial stages in the evolution of the leaf (of which the parent stem remains unknown). Emberger (1968) distinguished “Phyllophorées” ferns characterized by fronds, either entirely cauline (Holophyllophorées = stauropterids) or partially cauline (Hétérophyllophorées = e.g. zygopterids and Rhacophyton). In stauropterids, all successive divisions of the frond possessed at least two planes of symmetry (bilaterality) but they were not dorsiventrally symmetrical; furthermore, as stated above, information on an eventual stem bearing the Stauropteris frond is missing. In contrast, zygopterids showed a clear-cut distinction between the parent stem and helically arranged fronds with a regular phyllotaxis. This is supported by anatomical data: stems had a radially symmetrical protostele while the petiole (= phyllophore) possessed two planes of symmetry perpendicular to each other; therefore this organ was not dorsiventrally symmetrical, in contrast to all the higher order rachides. It is now accepted that zygopterids provided some of the best-documented series of evolutionary processes leading to the megaphyll, even if they did not evolve a fully dorsiventral leaf (Galtier, 2010 and Phillips and Galtier, 2005). The zygopterid leaf showed a fundamental step toward the acquisition of the dorsiventrality (cf. Kaplan, 2001) or adaxial/abaxial identity (= “abdaxity” character of Corvez et al., 2012), which is yet absent in the phyllophore but present in the next-order rachises of zygopterids. Abdaxity, as defined from leaf traces with abaxial protoxylem in seed plants but adaxial protoxylem in ferns (Galtier, 2010), allows characterizing the megaphyll of most Euphyllophytes. By analogy with the situation in zygopterids, it is our opinion that the aerial branching system of Stauropteris may represent a leaf precursor in which the proximal (N to N + 3) axes show a bilateral symmetry; however, the generalized absence of abdaxity is indicative of a less advanced stage in the evolution of the frond than in zygopterids and even in rhacophytaleans where abdaxity is only absent in the two most proximal orders of axes of the frond.

Due to their anatomical and morphological particularities, the affinities of Stauropteridales have been and remain matter of debate. In earliest discussions, Stauropteris was classified within the zygopterid ferns (Bertrand, 1909) and later included within the Coenopteridales (Andrews and Boureau, 1970 and Eggert, 1964). The latter group is now known to be paraphyletic. Stauropteridales are more accurately treated (e.g., Taylor et al., 2009) as a distinct group, near the Rhacophytales and the Zygopteridales, demonstrating a different early stage in the evolution of the frond. This is confirmed by a recent phylogenetic analysis (Corvez, 2012) whilst stauropterids appear as a basal clade near the Psilotophytes in other analyses (e.g., Rothwell, 1999).

Of particular interest is the occurrence, in S. oldhamia, of an aerenchymatous hypodermis and of a prominent epidermis in all branching orders. We confirm the presence of stomata as previously mentioned by Bertrand (1909) and Chaphekar (1962). In the absence of any laminate organ, we interpret the hypodermis as necessary to the assimilatory function, which suggests that photosynthesis was occurring in all orders of axes.

It is certainly significant that the aerenchymatous hypodermis is absent from the older S. burntislandica found in volcanic environments of Pettycur (Bertrand, 1909 and Surange, 1952) and of the Roannais (Galtier, 1971). Such a hypodermis has been occasionally mentioned in other representatives of the genus. In S. berwickensis, from both volcanic and fluviatile environments (Scott and Galtier, 1996) the occurrence of this tissue was suggested but not illustrated by Long (1966). Cichan and Taylor (1982) mentioned a narrow hypodermis in the main axis of S. biseriata. In both cases, the hypodermis seems to be restricted to the proximal N-order axis only, and its aerenchymatous nature was not mentioned.

The presumably generalized assimilatory function of all the axes of S. oldhamia is a very important feature. It contrasts with the situation in all known earliest ferns where the assimilatory function is restricted to the ultimate branching orders, in the form of either unwebbed cylindrical segments or of small laminated pinnules (Galtier, 2010). This character also contrasts with the occurrence of laminated pinnules in several coeval ferns from Bouxharmont: Ankyropteris, Corynepteris and Psalixochlaena (Holmes and Fairon-Demaret, 1984, Phillips and Galtier, 2005 and Phillips and Galtier, 2011).

An assimilatory function generalized to all axis orders is considered plesiomorphic to all plant groups (Kenrick and Crane, 1997). The presence of this character in the Carboniferous Stauropteris plant is puzzling. Two contrasting hypotheses can be considered. First, it is a plesiomorphic character that survived among Stauropteridales. Second, it is the result of a character reversion. We consider the second hypothesis as unlikely because the earliest representatives of the Stauropteridales, currently represented by the Devonian genus Gillespiea (Erwin and Rothwell, 1989), were similarly lacking specialized assimilatory organs. Furthermore, the assimilatory hypodermis seems to be more developed in the youngest S. oldhamia than in the earlier stauropterids; this may represent an adaptation to a humid coal swamp environment. There is however no clear evidence suggesting that S. oldhamia lived in semi-aquatic environments. Another hypothesis is suggested by the study of isoetalean lycopsids. In a reinterpretation of the parichnos system of Pennsylvanian lepidodendrales, Green (2010) suggested that it could represent an unusual metabolism similar to that of modern Isoetaceae. It consists of a particular photosynthetic pathway related to aquatic CAM photosynthesis (Keeley, 1998). In this model, the parichnos system represents internal gas spaces used as a carbon-concentrating mechanism. The parichnos system is thus similar to aerenchyma and used to provide upward CO₂ transport and downward O₂ transport. Such mechanism of CO₂ enrichment is particularly interesting within the context of the high O₂, low CO₂ late Palaeozoic atmosphere. A similar process could be proposed for Stauropteris. Indeed, the occurrence of aerenchyma in all branches order as well as the lack of stomata is consistent with this hypothesis.

The Stauropteridales appear to be a very specialized group occupying a particular position among earliest fern evolution. They are devoid of any lamina, like the very different, and supposedly leafless, cladoxylaleans. They are also quite distinct anatomically from the rhacophytaleans, zygopterids and other fern groups which exhibit, from the Late Devonian, the slow process of acquisition of laminate pinnules.

Several reconstructions of S. oldhamia have been published (e.g., Chaphekar, 1962, Eggert, 1964 and Mägdefrau, 1967) and reproduced in textbooks. The study of this coal ball containing several branched axes allows confirmation of the overall validity of these reconstructions. All these reconstructions, however, emphasize a very regular organization of the plant. They show organs emitted in opposite pairs giving a repetitive quadriseriate aspect to all branching orders. For example, Chaphekar (1962) and Mägdefrau (1967) illustrated N + 1 axes bearing at least four successive pairs of N + 2 axes and, in turn, N + 2 axes bearing at least four successive pairs of N + 3 axes at regular intervals. Here we describe the distalmost parts of the plant. They highlight an organization contrasting with previously published reconstructions. Indeed, we notice in the ultimate branching orders (after N + 3) a progressive loss of the four-lobed organization prevailing in other orders. Additionally, we provide the detailed reconstruction of a portion of this plant where one N axis is bearing immature N + 1 to N + 2 axes. The N + 1 axis was fully developed up to its first ramification. In contrast, the remaining part of this N + 1 axis, up to its apical region, measures only about 0.5 cm and includes a second level of ramification. The very short distance between the levels of attachment of the two successive pairs of N + 2 axes is suggestive of an arrest of growth. This is confirmed by the rapid decrease in diameter of the N + 1 axis after its second branching level as well as by the conical shape of its apical region. This also suggests, for S. oldhamia, some kind of determinate growth comparable with the development found in the leaves of some modern ferns. This may be considered as an argument to interpret the whole aerial branching system as a primitive fern frond (see discussion above). If there is no doubt that N + 1 to N + 3 axes were arranged as paired branches, evidence is missing of a quadriseriate arrangement of all branching orders. In addition, it is probable that the interval between successive branches was quickly decreasing and that the number of the successive paired branches was smaller than 4 or 5, as suggested in previous reconstructions. Despite the curvature observed in the distal-most region of this N + 1 axis, it is impossible to assess whether it is resulting from taphonomic processes or corresponding to an original circination. In spite of the exquisite preservation of the epidermal tissue, the absence of any ramentum, including in the apical region, appears characteristic of this plant.