A protoxylem pathway to evolution of pith? An hypothesis based on the Early Devonian euphyllophyte Leptocentroxyla

Abstract

Background and Aims

The Early Devonian (Emsian, 400–395 Ma) tracheophyte Leptocentroxyla tetrarcha Bickner et Tomescu emend. Tomescu et McQueen combines plesiomorphic Psilophyton-type tracheid thickenings with xylem architecture intermediate between the plesiomorphic basal euphyllophyte haplosteles and the complex actinosteles of Middle Devonian euphyllophytes. We document xylem development in Leptocentroxyla based on anatomy and explore its implications, which may provide a window into the evolution of pith.

Methods

Leptocentroxyla is preserved by permineralization in the Battery Point Formation (Quebec, Canada). Serial sections obtained using the cellulose acetate peel technique document branching pattern, anatomy of trace divergence to appendages, protoxylem architecture, and variations in tracheid size and wall thickening patterns.

Key Results

Leptocentroxyla has opposite decussate pseudo-whorled branching and mesarch protoxylem, and represents the earliest instance of central histological differentiation in a euphyllophyte actinostele. Tracheids at the centre of xylem exhibit simplified Psilophyton-type wall thickenings and are similar in size (at the axis centre) or smaller than the surrounding metaxylem tracheids (at the centre of appendage traces).

Conclusions

The position and developmental attributes of the simplified Psilophyton-type tracheids suggest they may have been generated by the protoxylem developmental pathway. This supports the delayed and shortened protoxylem differentiation hypothesis, which explains the evolution of pith by (1) delay in the onset of differentiation and lengthening of cell growth duration in a central protoxylem strand; and (2) shortening of the interval of differentiation of those tracheids, leading to progressive simplification (and eventual loss) of secondary wall thickenings, and replacement of tracheids with a central parenchymatous area. NAC domain transcription factors and their interactions with abscisic acid may have provided the regulatory substrate for the developmental changes that led to the evolution of pith. These could have been orchestrated by selective pressures associated with the expansion of early vascular plants into water-stresses upland environments.

INTRODUCTION

The stems of most extant euphyllophytes (i.e. non-lycophyte tracheophytes) possess eusteles or siphonosteles, i.e. vascular architectures characterized by the presence of a central parenchymatous core referred to as a pith. However, the fossil record demonstrates unequivocally that the earliest euphyllophytes had protosteles, which lacked a pith (Gensel and Andrews, 1984; Taylor et al., 2009). It is not surprising, therefore, that the evolutionary origin of the pith has elicited discussion and debate since the very beginnings of plant morphology as a direction of scientific inquiry.

Early discussions on the origin of the pith centred around two opposing views – the ‘extrastelar’ vs. ‘intrastelar’ hypotheses (Miller, 1971; Beck et al., 1982). The extrastelar hypothesis (Jeffrey, 1898) proposed that the pith of siphonosteles originated in protostelic axes by ‘invasion’ or ‘inrolling’ of parts of the cortex into the stele. In contrast, the intrastelar hypothesis – credited to either Van Tieghem and Douliot (1886) or Boodle (1901, 1903) (Miller, 1971; Beck et al., 1982) – posited that the pith originated by developmental change in the central tissues of the stele (i.e. xylem). Whereas the extrastelar hypothesis, finding no empirical support, remained a purely theoretical exercise, the intrastelar hypothesis gathered momentum with support from observations of fossil and living plants (e.g. Boodle, 1900, 1903; McLean Thompson, 1920; Posthumus, 1924; Bower, 1926; Miller, 1971).

The mechanisms proposed by the early authors for the intrastelar origin of the pith, although relatively vague, had obvious developmental foundations. Thus, Van Tieghem and Douliot (1886) invoked as an initial cause the increase in size of a protostele, as seen in the ontogenetic sequences of osmundalean stems. Boodle (1903) advanced that the pith evolved by progressive incomplete differentiation (implying parenchymatization) of the central region of the xylem of a protostele. McLean Thompson (1920, p. 715) explained the same ideas most clearly as a ‘change of destination of procambial elements [that] reached maturity as parenchymatous cells rather than as tracheides’.

Later, Miller (1971) discussed more specific mechanisms, bringing evidence from fossil zygopterid and osmundalean ferns in support of the derivation of pith parenchyma by developmental changes in the tracheids of the central xylem. The changes involved shortening and widening of central metaxylem tracheids and the appearance of a small number of intermixed parenchyma of same size and shape as these modified tracheids (with the implication that the latter were derived from tracheids), with continued change along these directions eventually leading to an entirely parenchymatous central area.

More recently, Stein (1993) used a modelling approach that relates properties of the auxin signalling system to the configuration of vascular tissues, with different auxin concentrations determining different cell fates. Without providing a unique explanation for the origin of the pith, Stein’s model suggests that histological differentiation at the centre of shoots is related to changes in the geometry of the shoot apex (enlargement of the apical dome) and increased density of lateral appendage primordia (shorter plastochrons), thus implying that the pith may be mostly a function of size and organization of the aerial shoot. This is consistent with the implications of scenarios proposed by Tomescu (2021) for the origin of stelar architectures that involve a pith (siphonosteles, eusteles), which involve radial reorganization of auxin concentration gradients as a result of an increase in axis size and the evolution of complex lateral appendages.

Considered in a deep time perspective, steles possessing a central parenchymatous zone are first seen among euphyllophytes in the Middle Devonian – in the moniliformopsid Arachnoxylon (Stein et al., 1983) and in representatives of several radiatopsid lineages [Actinoxylon (Matten, 1968), the progymnosperm Archaeopteris (Scheckler, 1978), and the younger (Mississippian) seed plant Diichnia (Beck et al., 1992]; see remarks in the Systematics section below for explanation of radiatopsids and moniliformopsids. Therefore, the evolution of steles possessing a pith may have begun in the Early Devonian, which has recently emerged as an interval wherein euphyllophytes underwent significant diversification, exploring broadly the morphospace of vascular anatomy (Bickner and Tomescu, 2019; Toledo et al., 2021; Durieux et al., 2021). There is good evidence that the type of central histological differentiation seen in Middle Devonian euphyllophytes opened the path toward the evolution of steles consisting of discrete bundles of vascular tissues that surround a central pith area, and that this evolutionary path was followed in parallel by several euphyllophyte lineages (Stein, 1993; Stewart and Rothwell, 1993; Tomescu, 2021). However, until now the connection between the Middle Devonian actinosteles exhibiting central histological differentiation and the typical actinosteles of the Early Devonian has remained elusive.

The last and longest interval of the Early Devonian – the Emsian (407–393 Ma) – is especially interesting because it demonstrates the coexistence of (1) forms with plesiomorphic, terete centrarch protosteles (e.g. Psilophyton, Banks et al., 1975; Trant and Gensel, 1985) with (2) some of the earliest complex actinosteles, such as Gothanophyton (Remy and Hass, 1986), Gensel’s (1984) plant, Kenrickia (Toledo et al., 2021) and Adelocladoxis (Durieux et al., 2021), as well as (3) forms exhibiting intermediate xylem architecture with moderately ribbed actinosteles (e.g. Bickner and Tomescu, 2019; Pfeiler and Tomescu, 2021). One such transitional euphyllophyte described from the Battery Point Formation of Quebec (Canada), Leptocentroxyla tetrarchaBickner et Tomescu (2019), exhibits a unique anatomy that may represent the earliest stage of central histological differentiation in actinosteles and, thus, provide a connection between the Middle Devonian actinosteles with central parenchyma and the typical actinosteles of the Early Devonian. If so, it may offer clues to the processes involved in the evolution of steles with pith.

Here, we present new data on the anatomy of Leptocentroxyla that allow us to propose a new hypothesis for developmental changes that may have led to evolution of the pith. Specifically, data on protoxylem architecture, tracheid wall thickenings and patterns of variation in tracheid size suggest that delayed differentiation of central tracheids and lengthening of their growth duration, combined with progressive narrowing of the window of secondary wall synthesis, led to a parenchymatous central area. These data, combined with new information on trace divergence to lateral appendages, also lead to an emended species diagnosis for Leptocentroxyla tetrarcha.

MATERIALS AND METHODS

Leptocentroxyla was identified in the plant fossil assemblage preserved by calcareous cellular permineralization in a cobble collected in the 1960s by the late Dr Francis M. Hueber (Smithsonian Institution – U.S. National Museum of Natural History, Washington, DC; USNM) from the Battery Point Formation, on the southern shore of Gaspé Bay (Quebec, Canada) near Douglastown. This rock unit represents fluvial to coastal deposits, which host a broad diversity of plants. The age of the Gaspé flora is middle-to-late Emsian (Banks et al., 1975; McGregor, 1977; Hoffman and Tomescu, 2013), roughly 400–395 million years old [absolute age based on Cohen et al. (2013, updated 2017)]. The cobble along with all acetate peels and slides produced from it are stored at USNM under collection number USNM 557820-3 (specifically, slabs A and B).

The fossil was studied in serial anatomical sections produced with the cellulose acetate peeling technique (Joy et al., 1956). Slides of the acetate peels were mounted with Eukitt mounting medium (O. Kindler, Freiburg, Germany). The fossil was imaged using a Nikon Coolpix 8800VR digital camera mounted on a Nikon E400 compound microscope, and an Olympus DP73 digital camera mounted on an Olympus SZX16 microscope. Digital images were processed using Adobe Photoshop (San Jose, CA, USA). Material for scanning electron microscopy was obtained from cellulose acetate peels using the method detailed in Matsunaga et al. (2013). Scanning electron microscope images of gold-coated specimens were generated using an FEI Quanta 250 (Hillsboro, OR, USA) microscope.

The fossil is an axis fragment that traverses two adjacent cobble slabs and it is truncated at both ends by its intersection with the surface of the cobble (Fig. 1A). We calculated the length of the axis fragment by adding the remaining (unpeeled) thickness of the two cobble slabs, the thickness of rock already removed during the peeling process (estimated as the number of acetate peels multiplied by an average 25 μm of rock removed between successive peels) and 2.5 mm to account for the thickness of rock removed by sawing apart the two slabs (saw kerf). We inferred the branching pattern of the plant based on the anatomy of the divergence of vascular traces supplying the lateral appendages, as revealed by the study of serial sections.

Fig. 1.

Leptocentroxyla tetrarcha, axis preservation. (A) Position of the axis in two slabs that were sectioned using the cellulose acetate peel technique; letters indicate the approximate position of the sections illustrated in B–E. Note variation in the quality of preservation along the axis, from good at the proximal end (E) to more distorted at the distal end (B). (B) 557820-3 Atop 186f. (C) 557820-3 Abot 2d. (D) 557820-3 Btop 150b. (E) 557820-3 Bbot 90f. Scale bar = 500 µm.

To characterize patterns of tracheid type and size distribution, we recorded the number of thinner-walled metaxylem tracheids in the central region of the xylem, in multiple cross-sections through the axis. In the same cross-sections, we also measured metaxylem tracheid diameters in three regions of the xylem: the central region consisting of thinner-walled tracheids, the layer of thick-walled tracheids around the central region and the lobes of xylem of the actinostele. In each cross-section we only measured the diameter of the largest tracheid in each of the three regions.

RESULTS

Leptocentroxyla is represented by an axis fragment ~5.5 cm long documented in two adjacent rock slabs (Fig. 1). The quality of preservation varies along the axis, from good preservation at the proximal end to somewhat distorted, compressed tissues toward the distal end (Fig. 1). The primary xylem has four lobes in cross-section (Figs 1 and 3), with mesarch protoxylem strands at the lobe tips. The metaxylem exhibits Psilophyton-type (P-type) pitting, except for the central area, in which metaxylem tracheids are characterized by thinner walls with scalariform thickenings. Typical P-type pitting consists of horizontally elongated bordered pits that form an overall scalariform pattern and wherein each of the pits possesses multiple small apertures (e.g. Hartman and Banks, 1980). The inner cortex, incompletely preserved and presumed parenchymatous, includes multiple groups of 8–16 sclerenchyma cells that form a discontinuous layer between the xylem and outer cortex. The outer cortex has alternating sectors of sclerenchyma and parenchyma, the latter incompletely preserved. These features were described in detail by Bickner and Tomescu (2019). Below, we present new data that add details on four main aspects: the anatomy of trace divergence to lateral appendages and branching architecture; the architecture of protoxylem; secondary wall thickening patterns of the tracheids; and tracheid size variation.

Fig. 3.

Leptocentroxyla tetrarcha, branching architecture as reflected by the pattern of trace emission to lateral appendages. A series of three cross-sections (C proximal to A distal) through a branching region demonstrates pseudo-whorled branching. Whereas one pair of opposite traces that are pinching off in C have already separated from the stele in B (orange arrowheads; bottom right part of the xylem in C is distorted taphonomically), the other pair of opposite traces (black arrowheads) are just pinching off throughout the sequence (C–B–A) and are separating at a level distal to the topmost section (A). Note missing trace at bottom right in A (asterisk), which has exited the axis and entered the base of a branch at this level. The inset diagram shows the position of the three planes of section (orange lines corresponding, from top to bottom, to A, B and C) with respect to the branching of the xylem (grey central band) and the axis; the darker dots in the xylem represent traces of appendages diverging perpendicular to the plane of the diagram. The divergence of the two pairs of opposite traces is only slightly offset vertically, which renders the branching pattern similar to a whorled arrangement. (A) 557820-3 Bbot 159f. (B) 557820-3 Bbot 119f. (C) 557820-3 Bbot 87f. Scale bar = 500 µm.

Anatomy of trace divergence

Serial cross-sections reveal the anatomical pattern of divergence of vascular traces that supply lateral appendages from the four lobes of the stele (Fig. 2). Proximal to a level of trace divergence, a pinching of the xylem lobe is conspicuous close to the tip of the lobe (Fig. 2A). Distal to this level, the xylem trace to the appendage begins to separate in a radial direction (Fig. 2B) and crosses upwards through the inner cortex as it diverges away from the xylem lobe (Fig. 2C). The traces to lateral appendages are elongated radially throughout their trajectory (Figs 2 and 3).

Fig. 2.

Leptocentroxyla tetrarcha, anatomy of trace divergence to lateral appendages, protoxylem architecture and tracheid pitting. Serial sections (A–C, basal to distal) of a xylem trace pinching off from the tip of a xylem lobe (A, B) and diverging in radial direction through the inner cortex (B, C); note radially elongate shape of trace to lateral appendage and groups of sclerenchyma in the cortex (on right side of xylem lobe and appendage trace). Groups of three or four smaller metaxylem tracheids close to the tips of the xylem lobe indicate the position of inconspicuous protoxylem strands (arrowheads in A–E); note that although protoxylem tracheids are not always recognizable, the patterning of tracheid sizes indicates the position of protoxylem strands. Distally to the trace divergence point, a second protoxylem strand is present (arrowheads in D and E), interpreted as resulting from a divergence in the single protoxylem strand at the lobe tip in C. The metaxylem of the main axis (F, G) features Psilophyton-type thickenings (F and G top and bottom), except for the central tracheids (G, centre and H), which are thinner-walled and characterized by scalariform wall thickenings that separate horizontally elongated oval bordered pits; note remnants of rare partitions separating the large apertures of these bordered pits into smaller apertures (arrowheads in H). (A) 557820-3 Bbot 87f. (B) 557820-3 Bbot 90f. (C) 557820-3 Bbot 190f. (D) 557820-3 Bbot 119f. (E) 557820-3 Bbot 182f. (F-H). 557820-3 Bbot 110. Scale bars = 150 µm (A–E); 50 µm (F, G); 35 µm (H).

Branching architecture

Considered at the level of the entire axis, with its four lobes of primary xylem, the pattern of trace divergence to lateral appendages reveals the branching architecture of Leptocentroxyla. The divergence pattern consists of pairs of opposite traces branching off at different levels (orange vs black arrowheads in Fig. 3). This indicates a branching architecture that would look roughly whorled from the outside, but in which the two decussate pairs of traces are slightly offset vertically (Fig. 3). Thus, one pair of opposite traces diverge at a level slightly higher above the divergence of the other pair of opposite traces, within the general area of the node.

Protoxylem architecture

The protoxylem forms strands in the vicinity of the tips of primary xylem lobes (Fig. 2), consistent with a mesarch pattern of primary xylem maturation. The strands are inconspicuous (i.e. shallow protoxylem strands, sensuStein, 1993), sometimes consisting of one or perhaps two very small and barely identifiable tracheids (e.g. Fig. 2A, C) and sometimes featuring no clearly recognizable tracheids (e.g. Fig. 2B). Nevertheless, whether protoxylem tracheids are recognizable or not, the position of the protoxylem strands is clearly indicated by tracheid shape and size patterns in the metaxylem, which features groups of two to four smaller, thinner-walled tracheids in the locations of protoxylem strands (Fig. 2).

Proximal to the divergence of the xylem trace that supplies a lateral appendage, a second protoxylem strand is present along the midline of the corresponding xylem lobe (Fig. 2D, E). This is probably the result of a divergence in the single protoxylem strand of the lobe and generates the protoxylem strand that will supply the trace of the next lateral appendage. However, the overall inconspicuousness of protoxylem strands makes it difficult to identify unequivocally such divergences in the protoxylem strands of the xylem lobes. For the same reason, the protoxylem strands of Leptocentroxyla are hard to trace continuously, in general, and diagnostic annular or helical thickenings cannot be demonstrated in any of their tracheids. Intriguingly, a corresponding tracheid size–shape pattern indicative of the position of the protoxylem is not present in the xylem traces to the lateral appendages, even at levels proximal to their physical separation from the stele (Fig. 2A, B). Instead, the central regions of these traces are occupied by a core of smaller metaxylem tracheids with thinner walls (Figs 2A–C and 4) and simpler secondary wall thickenings that are closely comparable to the tracheids that occupy the central area of the primary xylem (see below).

Fig. 4.

Leptocentroxyla tetrarcha, secondary wall thickenings of tracheids in lateral appendages. The tracheids in the traces to lateral appendages feature Psilophyton-type thickenings (A, B), except for the central region of the trace, which consists of smaller tracheids with thinner walls characterized by scalariform thickenings that separate oval horizontal bordered pits (A, centre; B, centre; C); note pit in C (centre, arrowhead) with aperture separated by a median vertical partition and remnant of such a partition in a pit in B (arrowhead). (A) 557820-3 Bbot 90f. Scale bar = 50 µm. (B) 557820-3 Bbot 110. Scale bar = 30 µm. (C) 557820-3 Bbot 90f. Scale bar = 25 µm.

Secondary wall thickening patterns of tracheids

The metaxylem of Leptocentroxyla typically consists of tracheids with Psilophyton-type secondary wall thickenings, the type that characterizes basal euphyllophytes (Fig. 2). However, the central area of the xylem strand is occupied by thinner-walled tracheids (Figs 2 and 5). These tracheids are as large as the typical metaxylem tracheids that surround them (Fig. 5B), but they lack the P-type thickenings of the latter. Instead, tracheids in the central area are characterized by a pattern of scalariform secondary wall thickenings that separate horizontally elongated oval bordered pits (Fig. 2). Locally, the oval pits retain remnants of vertical partitions that divided them into several smaller apertures (Fig. 2). The resulting geometry is similar to that of typical P-type thickenings, but with larger, often fused perforations of the pit membrane. This same type of thickening is present in the tracheids that form the central areas of traces that supply the lateral appendages (Fig. 4). As in the central tracheids of the main axis, the central tracheids of lateral traces, smaller than the peripheral tracheids, feature a scalariform thickening pattern with oval horizontal pits that preserve, occasionally, partitions separating them into multiple apertures (Fig. 4).

Fig. 5.

Leptocentroxyla tetrarcha, tracheid size variation. (A) Location of the central (purple), peripheral (green) and lobe areas (blue) of the axis xylem where maximum tracheid diameters were measured in each section. 557820-3 Bbot 90f. Scale bar = 200 µm. (B) Maximum metaxylem tracheid diameters in the central, peripheral and lobe tracheids (N = 40 for each tracheid type); bottom whisker = 25^th percentile − 1.5 × inter-quartile range; top whisker = 75^th percentile + 1.5 × inter-quartile range; dots = outliers; horizontal line = median; dashed horizontal line = mean. (C) Maximum metaxylem tracheid size is (weakly) positively correlated with the number of thin-walled tracheids in the central area of the xylem (r = 0.296, central tracheids; r = 0.376, peripheral tracheids; r = 0.290, lobe tracheids).

Tracheid size variation

The unusual anatomy of the xylem, with the central core of thin-walled tracheids, prompted a closer look at patterns of tracheid sizes, which we evaluated as a way to identify potential trends. However, despite the conspicuous differences in cell wall thickness and their different positioning in the xylem, the sizes of metaxylem tracheids in the three regions of the xylem – central thin-walled tracheids, peripheral thick-walled tracheids and thick-walled tracheids in the xylem lobes (Fig. 5A) – are closely comparable (Fig. 5B). All three tracheid types measure just under 40 µm in maximum diameter, on average (Fig. 5B) (N = 40 tracheids for each xylem region; see Supplementary Data Table S1 for measurements). Additionally, all three regions of the xylem show positive correlation between the size of metaxylem tracheids and the number of thin-walled tracheids in the central area as recorded in individual sections (Fig. 5C). Although the correlation is weak (r = 0.296 for central tracheids, 0.376 for peripheral tracheids and 0.290 for lobe tracheids), the relationship is nevertheless interesting, given that the number of central thin-walled tracheids varies widely (between 13 and 33) and without following a linear pattern along the axis.

DISCUSSION

Histological differentiation at the centre of protosteles

Among protostelic axes of living plants, histological differentiation at the centre of the protostele is known only in some monocot roots (e.g. Zea), in psilotophyte axes and in the stems on some ferns (e.g. Gleichenia); in the latter two cases, the differentiation is size-dependent (Ford, 1904; Bower, 1935; Bierhorst, 1971): an increase in axis size is accompanied by a transition from a typical protostele to a stele wherein the central area is occupied by cells other than tracheids. In most cases these cells are parenchyma and the steles are referred to as either medullated protosteles or siphonosteles, depending on whether there is evidence for the central tissues differentiating from procambium or not (Tomescu, 2021). In Psilotum, the central area consists of fibre-like sclerenchyma (Ford, 1904; Bierhorst, 1971). Importantly, none of these instances shows a central stele area occupied by a different type of tracheid, as seen in Leptocentroxyla. In the fossil record, actinosteles with central parenchymatous areas are first seen among euphyllophytes in the Middle Devonian (Matten, 1968; Scheckler, 1978; Stein et al., 1983; Beck et al., 1992). Similar to the examples from living plants, most of these fossil plants exhibit size-dependent transitions from smaller bona fide actinosteles, with tracheids at the centre, to larger steles with central parenchymatous areas, but none has a central stele area occupied by a different type of tracheid, as seen in Leptocentroxyla.

The protostele is the plesiomorphic organization of vascular tissues in tracheophytes. All Early Devonian and older euphyllophytes possess protosteles with a central core of xylem. The oldest protosteles are haplostelic – the xylem forms a terete strand – a reflection of the small axis size, lack of morphologically differentiated lateral appendages and simple pattern of hormonal regulation of vascular architecture in the plants that possess them. In the euphyllophyte clade, more complex, actinostelic protosteles appear in the Emsian (Gensel, 1984; Remy and Hass, 1986; Bickner and Tomescu, 2019; Pfeiler and Tomescu, 2021). However, up to the end of the Early Devonian, none of these protostelic euphyllophytes shows histological differentiation at the centre of the stele, other than the presence of a central protoxylem strand in those possessing centrarch xylem maturation. In this context, the actinostele of Leptocentroxyla, with its core of metaxylem tracheids that are markedly different from those at the periphery of the primary xylem strand, stands out as the oldest occurrence of central histological differentiation in protosteles. This anatomical feature is important as it may provide clues to the evolution of more complex steles, such as siphonosteles and eusteles that possess a central parenchymatous area – the pith.

Elements of a hypothesis for the evolution of the pith

Metaxylem tracheids

with different ontogenetic trajectories: paedomorphism. The thin-walled tracheids at the centre of the axis stele (and lateral appendage traces) have a conspicuous scalariform pattern of thickenings, which would not have allowed for their longitudinal stretching post-differentiation, without rupturing of the secondary wall. Thus, they differentiated beyond the zone of elongation of axes, i.e. as metaxylem (Esau, 1965, 1977).

The presence of vertical partitions that occasionally cross the apertures of the horizontally elongated bordered pits that form the scalariform pattern indicates that these secondary wall thickenings represent simplified versions of the P-type wall thickenings of the typical metaxylem tracheids. This difference in secondary wall thickenings between the two types of metaxylem tracheids reflects a difference in ontogenetic trajectory. Current understanding of the development of secondary wall thickenings in tracheary elements (e.g. Oda and Fukuda, 2013; Sugiyama et al., 2017; Oda, 2018; Buttò et al., 2019) indicates that a narrowing of the window (interval) of differentiation leads to thinner secondary wall buildup with simpler, less complete patterns of thickening. Consequently, the most parsimonious explanation for the thin-walled tracheids of the axis and lateral traces is that their differentiation followed the same pathway as typical P-type tracheids, but did not proceed all the way to completion. Instead, the secondary wall building process of these tracheids stopped earlier along the pathway, probably due to a shorter duration of the cell differentiation (maturation) phase. Thus, these ‘incompletely’ developed secondary wall thickenings are best regarded as simplified P-type thickenings and their ontogenetic trajectory can be regarded as paedomorphic (in a broad sense), in comparison to the typical metaxylem tracheids.

One implication of the same unique type of tracheids with paedomorphic ontogeny at the centre of both the main axis and its lateral appendages is that the two may be structurally identical and that the smaller size of the appendages and lack of lobation of their xylem represent just the starting point for epidogenetic growth (developmental pattern that results in a progressively larger and generally more complex primary body apically; Eggert, 1961). Completion of the latter would then lead to the same actinostelic architecture with central paedomorphic tracheids, distally in the appendages. Such epidogenetic changes from smaller, simpler stelar organization to larger and more complex steles are known in several plant lineages – e.g. lepidodendralean lycopsids (Eggert, 1961), ferns (Bower, 1926) or progymnosperms (Carluccio et al., 1966; Scheckler, 1976, 1978). If the two orders of branching recognized in Leptocentroxyla are structurally identical, this implies that there are no structural distinctions between different orders in the hierarchy of branching, and therefore the plant had iterative branching architecture (Berry and Stein, 2000).

Paedomorphic metaxylem tracheids positionally homologous to protoxylem.

Leptocentroxyla is certainly a euphyllophyte, as demonstrated by its mesarch protoxylem. All actinostelic euphyllophyte axes/stems, without exception, have centrarch or mesarch xylem maturation and in all of them the traces to lateral appendages have one or several protoxylem strands located more-or-less centrally. Therefore, the central paedomorphic tracheids in the appendage traces of Leptocentroxyla occupy the position of a protoxylem strand. However, their wall thickening pattern indicates that they are metaxylem. Nevertheless, like protoxylem tracheids, they are smaller than the surrounding tracheids that have typical P-type wall thickenings.

Metaxylem tracheids with the same paedomorphic ontogeny occupy the centre of the xylem of the main axis. Unlike protoxylem tracheids and unlike the paedomorphic tracheids of the appendage traces, these tracheids occupy a much larger area of the centre of the xylem and they are equal in size (and not smaller than) to the surrounding tracheids. Nevertheless, they have the same type of secondary wall thickenings as the tracheids that are positionally homologous to protoxylem in the appendage traces. Thus, their positional homology to protoxylem, although more tenuous, cannot be rejected, especially in light of the hypothesis discussed below. Besides, the centrarch xylem maturation of Psilophyton (Banks et al., 1975; Gensel, 1979; Trant and Gensel, 1985), the basal-most known euphyllophyte, indicates that a central protoxylem strand is plesiomorphic in the group.

Paedomorphic tracheids

with the same growth duration as the typical metaxylem tracheids. As part of their maturation, cells undergo growth in size and differentiation into a specific cell type. In sclerenchyma cells, including tracheary elements, the growth and differentiation phases are typically sequential – differentiation, involving the building of secondary wall thickenings, proceeds after growth to mature size has been completed. In a group of cells that start maturing roughly at the same time, such as those of a xylem strand maturing from procambium cells, those that start differentiating earlier (e.g. protoxylem or the paedomorphic tracheids at the centre of Leptocentroxyla appendage traces) are smaller because they have a shorter duration of growth. Conversely, cells of similar sizes undergo growth over equivalent time intervals. Therefore, the similar sizes of paedomorphic and typical metaxylem tracheids in the main axis of Leptocentroxyla indicate that they had the same duration of growth.

The delayed and shortened protoxylem differentiation hypothesis

A summary of statements.

The data and interpretations presented above have several implications:

• (1) The tracheids with thinner cell walls at the centre of the xylem in the main axis and lateral appendages are produced by the same developmental processes and are, therefore, developmentally equivalent.

• (2) The tracheids with thinner cell walls at the centre of the axis and lateral appendages are positionally equivalent. Those of the lateral appendages are positionally homologous to protoxylem, and therefore the central tracheids with thinner walls at the centre of the axis are (at least in some aspect) also homologous to protoxylem.

• (3) The simpler secondary wall thickenings of these tracheids imply a paedomorphic ontogenetic trajectory characterized by a shorter interval of differentiation as compared to the typical metaxylem tracheids.

• (4) The size similarity between the paedomorphic tracheids at the centre of the axis and the typical metaxylem tracheids of the axis implies equivalent intervals of cell growth preceding differentiation.

In a nutshell, the paedomorphic tracheids at the centre of the Leptocentroxyla axis are protoxylem homologues (at least in some aspects) that have undergone a prolonged period of cell growth (delayed differentiation) – reaching the same sizes as the surrounding metaxylem tracheids – and a briefer period of differentiation and secondary wall construction (shortened differentiation) – which resulted in thinner secondary walls and simpler secondary wall thickening patterns. If Leptocentroxyla illustrates the earliest stage in the evolution of central histological differentiation in the stele, a hypothesis for the evolution of the pith (Fig. 6) – in terms both of mechanism and of histological origin – arises at the intersection of the statements above.

Fig. 6.

A possible path to the evolution of pith as explained by the delayed and shortened protoxylem differentiation hypothesis – by central histological differentiation in a protostele; grey vertical lines and cylinder tops/bases denote immature and differentiating cells; coloured vertical lines and cylinder tops/bases denote mature tissues: red = tracheids, green = parenchyma. The starting point is a euphyllophyte with centrarch primary xylem maturation (i.e. a central protoxylem strand), like the Early Silurian Psilophyton, wherein tracheids of the central protoxylem strand mature (depicted by red secondary wall thickenings) in the zone of cell elongation, whereas the rest of the primary xylem (metaxylem) matures beyond the zone of elongation with production of P-type secondary wall thickenings. From this starting point, delayed differentiation (maturation) of the central tracheids allows them to grow in size before they differentiate; if the delay in differentiation lasts until these cells exit the zone of elongation, then these cells will differentiate in the same way as the metaxylem tracheids, with P-type thickenings (theoretical stage). If the interval of maturation is shortened in the central cells with delayed maturation, these cells will have less time to build secondary wall thickenings, which will result in simpler thickenings, like the central simplified P-type tracheids of Leptocentroxyla. Further shortening of the maturation interval of the central cells of the stele would eventually lead to complete elimination of the secondary wall building phase in the differentiation process of these cells, which will mature into parenchyma cells (as seen in some Devonian moniliformopsids – e.g. Arachnoxylon – and radiatopsids – e.g. Actinoxylon).

Constructing the hypothesis.

The thin-walled tracheids at the centre of the Leptocentroxyla axis had the same duration of cell growth, prior to differentiation, as the typical metaxylem tracheids, but they differentiated over a shorter time interval. Under progressively shorter differentiation intervals, these central paedomorphic tracheids would eschew entirely the formation of a secondary cell wall (energy- and time-demanding) and would end up differentiating as parenchyma. Thus, in terms of mechanism, one pathway toward evolving a pith could involve delayed differentiation of central tracheids and lengthening of their growth duration, combined with progressive narrowing of the window of secondary wall synthesis – i.e. delayed and shortened differentiation. Developmental changes along these lines could have led to configurations such as the medullated protosteles seen in Actinoxylon, Archaeopteris and Arachnoxylon (Matten, 1968; Scheckler, 1978; Stein et al., 1983).

In terms of histological origin, the anatomy of Leptocentroxyla suggests that the precursor tissue whose development changed to form the pith was protoxylem. The paedomorphic tracheids at the centre of the appendage xylem are positionally homologus to protoxylem and are smaller than the typical xylem tracheids. Thus, these tracheids can be regarded as a different expression of protoxylem. Then, the paedomorphic tracheids of the main axis, characterized by the same ontogenetic trajectory and located centrally, therefore both positionally and developmentally equivalent to those of the lateral appendages, may also be regarded as a different expression of protoxylem. Thus, these tracheids with delayed and shortened differentiation, which could represent an incipient stage in the evolution of the pith, may be derived from protoxylem.

Together, the considerations above support the delayed and shortened protoxylem differentiation (DSPD) hypothesis for the evolution of the pith (Fig. 6). This hypothesis involves as a first step a delay in the onset of differentiation and the lengthening of the duration of cell growth in cells of a pre-existing central protoxylem strand, until this duration equates to that of the surrounding metaxylem tracheids, so the central tracheids have diameters comparable to those of the latter. This step is also associated with an increase in the number of central protoxylem cells that may precede, follow or parallel the increase in the duration of cell growth (not pictured in Fig. 6). The second step brings about a shortening of the interval of differentiation for the tracheids at the centre of the stele, which leads to simplification of their secondary wall thickenings. If continued, the progressive simplification of cell wall thickenings would eventually lead to complete loss of secondary wall thickening and replacement of the tracheids at the centre of the xylem by a parenchymatous area.

The DSPD hypothesis and previously published data and hypotheses.

The DSPD hypothesis implies that pith and protoxylem are akin developmentally at a fundamental regulatory level. Here we discuss data that support a pith–protoxylem association and speculate on its bearing on previously advanced hypotheses for the evolution of pith. While a close association between pith and protoxylem may appear intriguing, when considered within a broader context it is not as far-fetched as it seems. Atypical protoxylem, differentiated as parenchyma, is known in several Devonian euphyllophytes (e.g. the Middle Devonian Langoxylon; Scheckler et al., 2006 – see also discussion therein). Data from extant ferns also demonstrate a connection between pith and protoxylem. Documenting stem anatomy in the Hymenophyllaceae, Boodle (1900, 1903) refrained from referring to the central parenchyma as pith, because of the protoxylem strand present at the very centre of the parenchymatous area in two species, Hymenophyllum scabrum and Vandenboschia (Trichomanes) radicans. These are the only instances of protoxylem strands present at the centre of a shoot among all extant and extinct plants, outside of the centrarch and centrarch–mesarch Devonian trimerophytes and radiatopsid euphyllophytes. The presence within the central parenchymatous areas of protoxylem strands provides further support for the strong connection between pith and protoxylem.

In his study modelling differentiation in the stele tissues, Stein (1993) attributed the origin of pith at the centre of the stele to changes in the geometry of the shoot apex and in the density of lateral appendages, leading to a larger apical dome and compact arrangement of lateral hormone sources. Stein proposed several hypotheses to explain the differentiation of the pith under radially polarized hormone (auxin) gradients. In Stein’s hypoinduction hypothesis, assuming a continuum of auxin concentration-dependent cell fates (highest for primary xylem and lower for phloem and then parenchyma), the pith is the result of auxin concentrations that are insufficient to induce vascular tissue differentiation. Bringing this into an evolutionary perspective, Tomescu (2021) proposed that the lowering of central auxin concentrations could have resulted from a trend of peripheral redistribution of auxin concentrations under a limited supply of hormone. While hypoinduction may explain the origin of the pith in some plant lineages, because it proposes that pith differentiates under lowest auxin concentrations, it is inconsistent with the DSPD hypothesis that supports shared developmental regulation between pith and protoxylem, which is thought to reflect highest auxin concentrations.

The DSPD hypothesis is more consistent with Stein’s (1993) hyperinduction scenario, which posits that pith parenchyma is the result of auxin concentrations high enough to inhibit differentiation of specialized conducting cells. In the hyperinduction hypothesis, protoxylem is the type of vascular tissue determined by the highest non-inhibitory auxin concentrations, and it, along with protoxylem parenchyma and pith parenchyma are considered distinct terms of a developmental continuum; thus, pith is ‘protoxylem in (auxin) overdrive’. Under this scenario, the paedomorphic metaxylem tracheids at the centre of the Leptocentroxyla axis differentiate under auxin concentrations that are high enough to inhibit protoxylem differentiation and yet still too low to completely inhibit the differentiation of xylem altogether. In turn, this would imply that metaxylem differentiation is determined not only by auxin concentrations lower than those responsible for protoxylem differentiation, but also by concentrations within a narrow range above those determining protoxylem differentiation and below those inhibiting xylem differentiation entirely.

Regulatory mechanisms potentially involved in the evolution of pith.

Developmental regulation studies have revealed two pathways that determine the identity of cells in the central tissues of axial plant organs in ways that are relevant to the hypotheses discussed here. Both pathways involve the regulation of NAC domain transcription factors, a large group of transcription factors, many of which control various aspects of secondary cell wall biosynthesis (Yao et al., 2012). One of these pathways controls switches between parenchyma and sclerenchyma identity by regulating the activity of NST class NAC transcription factors, whereas in the other pathway interactions between abscisic acid (ABA) signalling and VND class NAC transcription factors control xylem cell developmental trajectories.

With respect to the first of the two pathways, Sanchez et al. (2012) have proposed that the default identity of cells at the centre is one resembling that of sclerenchyma cells (like those of the xylem), and that the pith differentiates because regulatory pathways determining this default identity are actively repressed, systemically. They based this inference on the results of Mitsuda et al. (2007), Zhong et al. (2010) and Wang et al. (2010), which indicate that WRKY genes (WRKY12 in Arabidopsis and STP in Medicago) repress secondary cell wall formation (and lignification) in cells of the pith region, which differentiate into parenchyma, as a result. Specifically, WRKYs repress the promoter of the NST2 transcription factor, an NAC transcription factor belonging to a regulatory cascade that promotes secondary cell wall formation.

For the second pathway, working on the Arabidopsis root, Ramachandran et al. (2021) showed that ABA has a positive effect on the expression of VND (and a number of other xylem differentiation genes). Specifically, their experiments modulating water and ABA availability demonstrate that the ­effects of ABA signalling on the rate of differentiation and the fate of xylem cells are controlled independently via the activation of distinct VND genes. Specifically, VND2 and VND3 promote differentiation of metaxylem cells, especially those at the centre of the xylem, whereas VND7 promotes xylem cell fate change from metaxylem to protoxylem. Because four other eudicot genera (Brassica, Nicotiana, Solanum, Phtheirospermum) showed the same ABA–VND effects in the roots, Ramachandran et al. (2021) further suggested that interactions between ABA and VND genes may represent a universal molecular toolkit regulating xylem cell developmental adjustments.

Together, these studies indicate that in all likelihood interactions between NAC domain transcription factors and ABA were one of the major pathways whose modulation led to evolution of pith along a path consistent with the DSPD hypothesis presented here. Along this path, modulation of NAC–ABA interactions would have effected changes in the developmental trajectories of cells in the central tissues of plant axes, such as protoxylem to metaxylem cell fate transitions (VND genes) and sclerenchyma (including tracheary elements) to parenchyma transitions (NST and VND genes). Since NAC–ABA interactions are modulated by water relations (Ramachandran et al., 2021), water stress could have provided the selective pressures that led to the evolution of pith. For instance, the expansion of early vascular plants from water-rich lowland areas into more water-stressed, upland environments during the Early Devonian could have been a major driver in the evolution of pith.

While the NAC–ABA modulation hypothesis is an appealing explanation for the evolution of pith, it is worth noting that cell fate in the central tissues of plant axes is determined by multiple regulatory interactions. Ramachandran et al. (2021) have suggested based on transcriptomic data that MYB46 and MYB83 may also govern metaxylem differentiation in interaction with ABA in the Arabidopsis root. The same authors also noted that ABA is involved in another pathway important for cell identity in the stele, by activating miR165 in the endodermis, which induces a reduction in the levels of HD-ZIP III transcription factors that are important for determination of xylem cell fate. On the other hand, miR165 is also under control of SHR (Carlsbecker et al., 2010), and levels of HD-ZIP III expression in provascular tissue are also determined by auxin (e.g. Mattsson et al., 2003).

Thus, mechanisms other than the NAC–ABA interaction could have also been involved in developmental changes, consistent with the DSPD hypothesis, that led to the evolution of pith. In this context, it is important to remember that all the available information on the regulatory mechanisms that determine cell and tissue identity at the centre of plant axes comes from studies of a few angiosperms, and that data from seed-free plants are sorely needed to understand the broader taxonomic and phylogenetic scope of these regulatory mechanisms. Especially relevant to the issues discussed here would be, for example, data on the molecular regulation of the size-dependent transition from a typical prostele configuration to a medulated protostele or a siphonostele with parenchyma at the centre, in the pteridophytes Gleichenia and Psilotum (Bower, 1935; Bierhorst, 1971).

SYSTEMATICS

Division:

Tracheophyta

Subdivision:

Euphyllophytina

Leptocentroxyla tetrarcha Bickner et Tomescu 2019 emend. Tomescu et McQueen 2022

Emended specific diagnosis (emendations in bold tyoe):

Small axes (~2 mm diameter) circular-subcircular in cross-section, with offset opposite decussate pseudo-whorled branching. Architecture probably iterative (sensuBerry and Stein, 2000). Primary xylem ~1.1 mm across, a four-ribbed actinostele, with thick diamond-shaped central area and slender ribs. Inconspicuous mesarch, shallow protoxylem strands at rib tips. Metaxylem tracheids up to 60 µm wide, with angular, polygonal outlines and Psilophyton-type secondary wall thickenings, except for conspicuous central area of thinner-walled tracheids with scalariform thickenings derived from simplified Psilophyton-type thickenings. Xylem traces to lateral appendages radially elongated, lacking protoxylem, with central area occupied by well-defined core of thinner-walled tracheids derived from simplified Psilophyton-type thickenings. Inner cortex wide, parenchymatous, with groups of large sclerenchyma forming a discontinuous layer around the xylem. Outer cortex thin, primarily sclerenchymatous. Epidermis of radially flattened parenchyma.

Remarks

Comparison with coeval actinostelic euphyllophytes. The only Early Devonian plant that approaches the branching pattern of Leptocentroxyla is Pertica quadrifaria (Kasper and Andrews, 1972). However, the anatomy of Pertica is currently not known, which precludes further comparison. Irrespective of anatomy, the pseudo-whorled branching of Leptocentroxyla follows a decussate pattern that is different from that or Pertica quadrifaria, which approaches a helical pattern with occasionally longer internodes.

The mesarch protoxylem strands of the xylem lobes place Leptocentroxyla firmly among the euphyllophytes. Euphyllophytes with a lobed xylem are diverse in the Middle and Upper Devonian (Scheckler, 1974, 1976; Beck, 1976), but only a handful are known from the Lower Devonian (Gensel, 1984; Remy and Hass, 1986; Chu and Tomescu, 2015; Bickner and Tomescu, 2019; Toledo et al., 2021; Pfeiler and Tomescu, 2021; Durieux et al., 2021). A comparison between Leptocentroxyla and some of these Early Devonian euphyllophytes has been provided by Bickner and Tomescu (2019). Our new data demonstrate additional differences between Leptocentroxyla, Gothanophyton (Remy and Hass, 1986) and Gensel’s (1984) plant (Table 1). While all three plants have conspicuously lobed primary xylem with mesarch maturation, Leptocentroxyla is significantly smaller and has P-type tracheids, whereas Gothanophyton and Gensel’s plant (which was described from the Battery Point Formation, like Leptocentroxyla) have tracheids with scalariform to oval pitting. Additionally, Leptocentroxyla has pseudo-whorled branching and lacks the central protoxylem strand that characterizes both Gothanophyton and Gensel’s plant. In Leptocentroxyla, traces to laterals arise singly and diverge radially, similar to those of Gensel’s plant but unlike those of Gothanophyton, which has lateral traces consisting of paired bundles. In contrast to the traces of Gensel’s plant, which are tangentially elongated, those of Leptocentroxyla are elongated radially. All these differences emphasize the distinctiveness of each of these three plants.

Table 1.

Comparison of Leptocentroxyla with coeval actinostelic euphyllophytes.

	Leptocentroxyla	Gensel’s (1984) plant	Gothanophyton
Axis size (mm)	2–5	7–10	7–15
Actinostele type	4-lobed	3-lobed	4-6-lobed
Xylem maturation	Mesarch	Mesarch	Mesarch
Xylem centre	Thin-walled metaxylem	Protoxylem strand	Protoxylem strand
Appendage taxis	Opposite decussate (pseudo-whorled)	Alternate	Alternate (pseudo-whorled)
Metaxylem pitting	Psilophyton-type	Scalariform to oval pits	Scalariform to oval pits
Trace divergence	Single, radial	Single, radial	Paired, tangential
Trace shape	Radially elogated	Tangentially elongated	Lobed

Considered in a broader taxonomic perspective, the mesarch xylem maturation and the central protoxylem strands of Gensel’s plant and of Gothanophyton place these plants among the radiatopsid euphyllophytes. In contrast to these, the xylem anatomy of Leptocentroxyla makes its position among the euphyllophytes, in terms of radiatopsid vs. moniliformopsid affinities, uncertain (as discussed below).

The position of Leptocentroxyla among euphyllophytes. Early euphyllophytes with a lobed central xylem strand (actinostele) have been divided by Beck and Stein (1993) into two groups based on their protoxylem architecture. Those in the ‘radiate protoxylem’ group have a central protoxylem strand from which additional protoxylem strands diverge outwards, running upwards and obliquely through the xylem lobes to enter the xylem traces that diverge from the lattter to supply lateral appendages. Only later-diverging members of the radiate protoxylem group, notably some of the basal-most actinostelic seed plants, lack the typical central protoxylem strand (Rothwell and Erwin, 1987). In contrast to the radiate protoxylem architecture, euphyllophytes included in the ‘permanent protoxylem’ group have protoxylem strands that run vertically in the plane of the xylem lobes, without diverging from a central protoxylem strand, which is lacking in most members of the group. The two groups are currently referred to as radiatopsids (Radiatopses) and moniliformopsids (Moniliformopses), respectively (Kenrick and Crane, 1997).

Leptocentroxyla lacks the central protoxylem strand characteristic of basal radiatopsid euphyllophytes. Instead, the central area is occupied by a core of thin-walled metaxylem tracheids with simplified P-type pitting. With its single protoxylem strands that run vertically at the tip of each xylem lobe and seem to diverge only to produce strands corresponding to the divergence of traces that supply laterals, Leptocentroxyla is more similar to moniliformopsids, or to younger radiatopsids. However, the relationship between these diverging protoxylem strands and the strands of simplified P-type tracheids seen in the xylem traces could not be ascertained in the specimen available. On the other hand, neither younger radiatopsids nor moniliformopsids are known to possess a central xylem area with thin-walled metaxylem. Additionally, the vascular traces that supply the laterals of Leptocentroxyla do not have typical protoxylem at the centre – they have smaller simplified P-type tracheids, again, unlike any other known euphyllophytes, be they radiatopsids or moniliformopsids. Yet, if the thin-walled P-type tracheids with simplified P-type thickenings in the axis xylem and in the lateral appendage traces are equivalent to protoxylem, as hypothesized above, then the anatomy of Leptocentroxyla is more consistent with radiatopsid affinities. A test of all these considerations that could illuminate the taxonomic affinities of Leptocentroxyla will be facilitated by the discovery of additional specimens, which could reveal unequivocally whether the mesarch protoxylem strands observed in the stele ribs are connected to the areas of simplified P-type tracheids at the axis centre and in the lateral appendage traces. Until then, based on available data, the type of xylem organization documented in Leptocentroxyla has no known equivalent among extinct or living plants. Along with the P-type tracheids that are not known in younger Devonian euphyllophytes, these set apart Leptocentroxyla from all radiatopsids and moniliformopsids currently known.

CONCLUSIONS

The Early Devonian (Emsian; ~400–395 Ma) euphyllophyte Leptocentroxyla tetrarcha is characterized in more detail, with an emended diagnosis. This plant combines plesiomorphic Psilophyton-type tracheid thickenings with xylem architecture intermediate between the plesiomorphic haplosteles of basal euphyllophytes and the complex actinosteles of Middle Devonian euphyllophytes. Newly developed data on Leptocentroxyla broaden the diversity of morphology and anatomy documented among Early Devonian plants. We demonstrate opposite decussate branching forming a whorled appearance (pseudo-whorled), marking the earliest known occurrence of this branching pattern in euphyllophytes. We also demonstrate that tracheids at the centre of the xylem exhibit simplified Psilophyton-type wall thickenings resulting from a shorter interval of secondary wall building. This marks the earliest instance of central histological differentiation in a euphyllophyte actinostele. Developmental attributes reflected by the anatomy of the simplified Psilophyton-type tracheids – duration of the cell expansion and cell differentiation phases – combined with their position at the centre of the xylem, suggest that these tracheids may represent an expression of the protoxylem developmental pathway. This interpretation supports a hypothesis – the delayed and shortened protoxylem differentiation hypothesis – for the evolution for central histological differentiation in the stele that could explain the advent of the pith as a central tissue in euphyllophyte axes. If Leptocentroxyla does indeed illustrate the earliest known stages in the evolution of a pith, then this hypothesis posits that evolving a pith involves (1) a delay in the onset of differentiation and lengthening of the duration of cell growth in a pre-existing central protoxylem strand, leading to an increase in cell size (diameter) prior to differentiation; and (2) a shortening of the interval of differentiation of the tracheids, leading to progressive simplification of wall thickenings and, eventually, to loss of secondary wall synthesis and replacement of the tracheids with a central parenchymatous area. Data from living angiosperms suggest that modulation of interactions between NAC domain transcription factors and ABA have underpinned the changes in developmental trajectories of central xylem cells that led to the evolution of pith. The fact that these interactions are modulated by water stress provides a potential link between the evolution of pith and the expansion of vascular plants from humid lowland environments into drier upland areas.

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Table S1: Tracheid measurements.

FUNDING

Initial assessment of the NMNH collection of Gaspé specimens was made possible by an American Philosophical Society grant to A.M.F.T.

CONFLICT OF INTEREST

The authors declare they have no conflict of interest.