Postcranial anatomy and histology of Seymouria, and the terrestriality of seymouriamorphs

Neutron tomography

Neutron tomography measurements were performed at the DINGO thermal-neutron radiography/tomography/imaging station (Garbe et al., 2015) located at, and tangentially facing, the 20 MW Open-Pool Australian Lightwater (OPAL) reactor housed at the Australian Nuclear Science and Technology Organisation (ANSTO), Lucas Heights, New South Wales, Australia. The DINGO facility utilises a quasi-parallel collimated beam of thermal neutrons.

For the femur (ROMVP 80915) and the fibula (ROMVP 80917), the instrument was equipped with an Iris 15TM Large Field of View sCMOS camera (5,056 × 2,968 pixel, 16-bit) and Zeiss Ikon 100 mm f/2.0 Makro Planar lens. Based on a desired spatial resolution of ∼60 µm across the partially-embedded femur and fibula, a maximum sample width of 48.5 mm and minimum sample-to-detector distance of 28 mm, the DINGO instrument was configured with a 30 µm thick terbium-doped Gadox scintillator screen (Gd2O2S:Tb, RC Tritec AG) and 25.2 × 25.2 × 25.2 µm voxels for a Field-of-View of 100 × 74.5 mm.. To maximise counting statistics and minimise subsequent noise in the tomographic reconstruction, a collimation ratio (L/D) of 500 was used, where L is the neutron aperture-to-sample length and D is the neutron aperture diameter. This high-flux configuration traditionally illuminates a 200 mm ×200 mm area around the sample area with 4.75 × 107 neutrons cm-2s-1, leading to high background radiation and zingers on the detector. A newly installed slit system was implemented to restrict the neutron-irradiated area about the specimen to achieve optimum scan conditions and a divergence-limited spatial resolution of 56 µm. A total of 900 equally-spaced angle shadow-radiographs were obtained every 0.20° as the sample was rotated 180° about its vertical axis. Both dark (closed shutter) and beam profile (open shutter) images were obtained for calibration before initiating shadow-radiograph acquisition. To reduce anomalous noise, a total of three individual radiographs with an exposure length of 4.0 s were acquired at each angle (Mays, Bevitt & Stilwell, 2017) for a total scan time of 4.6 h.

For the sacral series (OMNH 79348), a collimation ratio (L/D) of 1,000 (Garbe et al., 2015) was used to ensure highest available spatial resolution. The field of view was set to 200 × 200 mm² with a voxel size of 73.3 × 73.3 × 73.3 µm and sample to detector distance of 70 mm. Neutrons were converted to photons with a 100 µm thick ZnS(Ag)/⁶LiF scintillation screen (RC Tritec AG); photons were then detected by an Iris 15 sCMOS camera (16-bit, 5,056 × 2,960 pixels) coupled with a Makro Planar 24 mm Carl Zeiss lens. The tomographic scan consisted of a total of 720 equally-spaced angle shadow-radiographs obtained every 0.25° as the sample was rotated 180° about its vertical axis. Both dark (closed shutter) and beam profile (open shutter) images were obtained for calibration before initiating shadow-radiograph acquisition. To reduce anomalous noise, a total of three individual radiographs with an exposure length of 8 s were acquired at each angle (Mays, Bevitt & Stilwell, 2017). These individual radiographs were summed in post-acquisition processing using the Grouped ZProjector function in ImageJ v.1.51 h. Total scan time was 5.7 h.

The individual radiographs were summed in post-acquisition processing using the ‘Grouped ZProjector’ plugin in ImageJ v.1.51 h in accordance with our previous measurements; this plugin was developed by Holly (2004). Tomographic reconstruction of the 16-bit raw data was performed using commercially available Octopus Reconstruction v.8.8 software package and the filtered back-projection algorithm to yield virtual slices perpendicular to the rotation axis. When these slices are stacked in a sequence, they form a three-dimensional volume image of the sample. The reconstructed volume data were downsampled by a factor of 2 in ImageJ to reduce computation time, then rendered and segmented with Avizo Lite 9.3.0.

Histology

Histological preparation followed standard procedures (Padian & Lamm, 2013). All specimens were photographed prior to embedding in EP4101UV resin (Eager Polymers), which was allowed to cure for 24 h. ROMVP 80916 (partial femur), ROMVP 81198 (vertebra), and ROMVP 81199 (vertebra) were prepared at the Royal Ontario Museum (ROMVP), Toronto, Canada. Specimens were cut on the IsoMet 1000 precision saw (Buehler) and mounted to frosted plexiglass slides with cyanoacrylate adhesive. For the femora, the cut was made at the approximate region of the minimum diaphyseal circumference; for the vertebra, the first cut was made sagittally (anteroposteriorly) down the midline, and the second cut was made transversely through one of the two halves of the block. For ROMVP 80916 (larger, partial femur), the section is taken slightly proximal to the inferred minimum circumference due to the incomplete specimen’s nature.

Mounted blocks were trimmed using the IsoMet and ground on the Hillquist Thin Sectioning Machine lapidary wheel. Manual polishing using 1,000-mesh grit on glass plates and a combination of 1-µm and 5-µm grit on polishing cloths was used to remove scratches. ROMVP 81200 (partial femur) was prepared in a similar fashion but with a different equipment setup at the University of Toronto Mississauga. Cutting was performed on the Metcut-5 low speed saw (MetLab), initial grinding on the Metcut-10 Geo (MetLab), and manual grinding on a cutlery whetstone block. Imaging was done on two Nikon AZ-100 microscopes, both fitted with a DS-Fi1 camera and NIS Elements-Basic Research software registered to David C. Evans and to Robert R. Reisz.

Preparation of the specimens was performed by Diane Scott and Bryan M. Gee using pin vises and air scribes. Figures were prepared using Adobe Illustrator and Photoshop. The raw data are reposited online through MorphoBank (O’Leary & Kaufman, 2012; see Data Availability statement for additional details).

Histological interpretations and comparisons

Contextualizing the histological data of the specimens of Seymouria is complicated by the paucity of work on other stem amniotes, let alone seymouriamorphs specifically. Limb elements of Seymouria have never been histologically analyzed. The only other seymouriamorph femur to be histologically sampled is that of the European Discosauriscus, the femur of which is characterized by a parallel-fibered endosteal matrix with sparse vascularization comprised of radially and longitudinally arranged canals in small individuals that shifts to being dominated by radial vasculature in adults (Sanchez et al., 2008). Seymouria exhibits a lamellar matrix with a plexiform arrangement of the vasculature in the femur, and the degree of vascularization far exceeds that figured for Discosauriscus. The increased vascularization is indicative of relatively fast growth at the time of death in the sampled specimens of Seymouria, which also suggests relative immaturity. Skeletochronological markers also differ between the taxa. Discosauriscus possesses numerous, well-defined and evenly spaced LAGs, whereas Seymouria is characterized by distinctive growth zones and annuli bearing numerous closely spaced rest lines but without clear LAGs. Similar variation in skeletochronological markers has been reported in the Late Triassic metoposaurids Dutuitosaurus from Morocco and Metoposaurus from Poland; this disparity was hypothesized to be the result of differing seasonal activity patterns associated with climatic differences across paleolatitudinal gradients (Konietzko-Meier & Klein, 2013). Lastly, the medullary spongiosa is distinctly less developed in Discosauriscus (Sanchez et al., 2008: Fig. 2); the significance of this is unclear in the absence of additional data. It may relate to overall body size, as extant lissamphibians, relatively small in comparison to Seymouria, typically lack medullary spongiosa; however, the larger cryptobranchids possess a spongiosa (e.g., Laurin, Canoville & Germain, 2011). Similarly, small, semi-terrestrial to terrestrial temnospondyls also have little to no spongiosa (e.g., McHugh, 2015) compared to the larger Eryops with a much denser spongiosa (e.g., Konietzko-Meier, Shelton & Sander, 2016).

Comparisons with other Paleozoic tetrapods are also limited by a paucity of comparative work. Of the major Paleozoic clades (e.g., pelycosaurian synapsids, ‘lepospondyls’), temnospondyls are the best-sampled (Sanchez et al., 2010a; Sanchez et al., 2010b; McHugh, 2014; Konietzko-Meier, Shelton & Sander, 2016). The relative thickness of the cortex and the development of the medullary spongiosa are most comparable to that of the co-occurring trematopid Acheloma dunni, a terrestrial taxon (Sanchez et al., 2010b; Quemeneur, de Buffrénil & Laurin, 2013). The spongiosa is less developed than in either definitively aquatic taxa such as the late Permian rhinesuchid Rhinesuchus (McHugh, 2014) or in controversially aquatic taxa such as the early Permian eryopid Eryops (Sanchez et al., 2010b; Quemeneur, de Buffrénil & Laurin, 2013; Konietzko-Meier, Shelton & Sander, 2016), and the cortex is not extensively thickened as in the definitively aquatic dvinosaur Trimerorhachis (Sanchez et al., 2010b; Quemeneur, de Buffrénil & Laurin, 2013). A large number of Mesozoic temnospondyls, which are predominantly aquatic, have also been sampled (Steyer et al., 2004; Konietzko-Meier & Sander, 2013; Sanchez & Schoch, 2013). Many of these taxa exhibit similar structure to that of Trimerorhachis, often with a high degree of pachyostotic development and with greatly reduced or nearly absent medullary cavities. McHugh (2015) sampled the small-bodied Early Triassic lydekkerinid Lydekkerina, a semi-aquatic taxon with terrestrial capabilities (e.g., Pawley & Warren, 2005; Canoville & Chinsamy, 2015) and the amphibamiform Micropholis, a fully terrestrial taxon (e.g., Schoch & Rubidge, 2005). Both taxa exhibit a similar histological and microanatomical organization to that of terrestrial Paleozoic temnospondyls and to that of Seymouria. Collectively, the temnospondyl comparisons support an inferred terrestrial lifestyle of Seymouria. However, it is important to note that the spongiosa of Seymouria is more developed than in any of the co-occurring terrestrial temnospondyls at Richards Spur (Castanet et al., 2003; Quemeneur, de Buffrénil & Laurin, 2013; Richards, 2016; Gee, Haridy & Reisz, 2020) in which the spongiosa is either weakly developed (Trematopidae) or virtually non-existent (Dissorophidae, Amphibamiformes). However, it is comparable in the developed medullary spongiosa to that of Eryops, the degree of terrestriality of which has long been debated (Konietzko-Meier, Shelton & Sander, 2016, and references therein). The significance of the spongiosa in Seymouria is uncertain at present and warrants further work to compare with other co-occurring taxa and with more closely related stem amniotes.

The vertebral histology is also difficult to compare with coeval tetrapods, let alone with closely related taxa. Vertebrae are uncommon in histological studies compared to limb elements, and most studies that that have examined the vertebrae of Paleozoic tetrapods have focused on the inter- and pleurocentra (Konietzko-Meier, Danto & Gadek, 2014; Danto et al., 2017; Danto et al., 2019). However, both the centra and the neural arches contribute valuable information regarding the lifestyle of Seymouria. Previous workers have often suggested that the neural arch would have been subject to strong biomechanical constraints during locomotion in early tetrapods (Rockwell, Evans & Pheasant, 1938; Olson, 1976; Holmes, 1989). The prominent expansion of the neural arch and the development of the zygapophyses in Seymouria lends support to this hypothesis. Discosauriscus is the only other seymouriamorph to have its internal vertebral anatomy examined (Danto et al., 2016). The main difference is in the construction of the neural arch, which is comprised of thick, compact lamellar bone in Discosauriscus; in contrast, the neural arch of Seymouria is largely hollow. Based on the size of the sampled Discosauriscus material, the individual was likely premetamorphic and still aquatic, which would explain the higher degree of ossification. Whether this might have changed in later stages of ontogeny if or when individuals metamorphosed into a terrestrial adult form remains unknown. Beyond seymouriamorphs, neural arches have not been sampled in many clades, which may be because most Paleozoic tetrapod clades have multipartite vertebrae in which the arch readily detaches from the centra during preservation. Furthermore, isolated neural arches have not traditionally been utilized as an ideal case study for exploring histological questions compared to either the centra or to other postcranial elements. Danto et al. (2016) sampled a number of Paleozoic lepospondyl taxa in which neural arches were preserved. Some of the aquatic taxa (e.g., an indeterminate nectridean) exhibit a similar spongy bone composition of the arch, but the interior of the arch is relatively well-ossified with little empty space.

Skeletochronological interpretations. In the absence of comparative histological data, most inferences regarding the life history of seymouriamorphs have been based on external anatomy of different skeletal regions. For example, most individuals of Discosauriscus retain lateral line canals on the skull, indicating an aquatic lifestyle, but this may also reflect a biased relative abundance of premetamorphic individuals in the fossil record (e.g., Klembara et al., 2006). Although definitive adults of this taxon, terrestrial or otherwise, are unknown (Klembara, Martens & Bartík, 2001), previous authors have inferred that Discosauriscus underwent metamorphosis (Klembara, 1995), or that if some species were paedomorphic, they were derived from an ancestor that did metamorphose into a terrestrial adult (e.g., Boy & Sues, 2000). For Discosauriscus austriacus, Sanchez et al. (2008) reported that metamorphosis occurred around the sixth year of life. Although not explicitly stated as such, determination of metamorphosis in that study was rendered feasible through the sampling of limb material of Discosauriscus from articulated skeletons that would permit correlation of the skeletochronological data from the histological analysis with external osteological features traditionally used for relative age determination.

Characterizing ontogeny in Seymouria is complicated by a general paucity of associated postcrania compared to Discosauriscus such that most characterizations are solely based on the cranium (e.g., Klembara et al., 2006; Klembara et al., 2007). Indeed, it is not known for certain that Seymouria underwent metamorphosis, as no larval forms have been recovered. The smallest reported specimen is one from the Bromacker quarry with a skull measuring 2.1 cm (Berman & Martens, 1993), but the preservation is too poor to be described for comparative anatomical purposes. Most of the other well-described cranial specimens exceed 8.0 cm (e.g., Berman, Reisz & Eberth, 1987), and small specimens remain poorly represented (e.g., Klembara et al., 2006). Simultaneously, the histological signals of metamorphosis in early tetrapods, if they exist, remain poorly understood. In extant lissamphibians, which may be the most appropriate extant analogue for seymouriamorphs, there is often a ‘transformation mark’ (Schroeder & Baskett, 1968) or a ‘metamorphosis line’ (Rozenblut & Ogielska, 2005) that demarcates overwintering following metamorphosis. Unfortunately, the criteria for identifying this line are somewhat ambiguous in the literature beyond it being the closest circumferential line to the medullary cavity and likely being closely spaced to the first LAG. Identifying this feature may thus rely mainly on an absence of endosteal resorption (present in our sectioned specimens) that would obliterate this mark. The line has a similar appearance to LAGs (e.g., Tsiora & Kyriakopoulou-Sklavounou, 2002), and there is some debate over whether there is a correlation of this line with deposition of woven-fibered bone (e.g., see Castanet & Smirina, 1990; Guarino et al., 2003). To the best of our knowledge, this feature has never been identified in a Paleozoic tetrapod, and the applicability of crown lissamphibian life history to seymouriamorphs on the amniote stem is unknown.

As a result, determining whether a specimen of Seymouria is ‘postmetamorphic’ is based largely on an established precedent of using external features from which admittedly arbitrary and gradational terms such as ‘juvenile,’ ‘sub-adult,’ and ‘adult’ are derived. It bears noting that these terms may refer to different biological attributes (e.g., sexual maturation vs. the process of metamorphosis) and may be used differently by various workers. With respect to ROMVP 80916, the large partial femur that was sectioned, there is a well-established precedent for identifying the element as belonging to a probable ‘adult’ that had completed metamorphosis. ROMVP 80916 is incomplete, but the preserved portion is the same size as the complete ROMVP 80915, which measures 5.5 cm in length and which is on the larger end of previously reported specimens (e.g., 6.4 cm; White, 1939). Based on comparisons with articulated specimens of Seymouria (Berman, Reisz & Eberth, 1987; Berman et al., 2000), this femur would belong to an individual with a skull length exceeding 10 cm, which falls well within a range for which specimens have been previously described as ‘adults.’ Femora of the previously described articulated skeletons are smaller than ROMVP 80916 yet the skulls possess numerous features accepted as evidence for both somatic maturity and terrestriality, such as the absence of lateral line grooves, ossified carpals and tarsals, firmly interdigitated sutures, and pronounced ornamentation (Boy & Sues, 2000). ROMVP 81200 is smaller, measuring only 2.6 cm as preserved. By comparison with ROMVP 80915, assuming isometric scaling of the element, ROMVP 81200 would have been around 4.2 cm in length when preserved. Using this estimate and comparisons with articulated specimens, the skull of this individual would have been around 8 cm, which is only slightly below the lower size bound reported for most specimens of Seymouria (e.g., Berman, Reisz & Eberth, 1987) and which would at least represent a ‘sub-adult’ based on previous designations. Further evidence for a postmetamorphic determination may be found in the nature of the preservational environment of Richards Spur. Beyond the enigmatic and extremely rare aïstopod Sillerpeton permianum, there is no evidence of aquatic tetrapods, either larval forms of metamorphosing adults or obligately aquatic adult forms, even though material of very small-bodied tetrapods is captured (MacDougall et al., 2017). Regardless of whether this represents a biased sample, it is clear that the fissure fills were not conducive to the capture of aquatic tetrapods.

Based on a postmetamorphic interpretation of the sectioned material, the onset of metamorphosis in Seymouria may thus be constrained to probably occurring by the second year of life. It may be inferred that the larger ROMVP 80916 was probably older than two years, although this cannot be proven nor can the amount of growth cycles lost to remodeling be determined at present. Metamorphosis in Seymouria occurred much earlier than in Discosauriscus, which was suggested to undergo a delayed metamorphosis around the sixth year of life (Sanchez et al., 2008). The more compact and sparsely vascularized lamellar bone of Discosauriscus also support interpretations of slower growth and a protracted aquatic larval stage in this taxon (Sanchez et al., 2008). This disparity among closely related taxa may reflect the different environments that these taxa inhabited (correspondent with their differing lifestyles), as Seymouria is primarily found in fluvial environments of North America, and Discosauriscus is primarily found in lacustrine settings in Europe. Both taxa likely experienced distinct seasonality, but the presence of distinct LAGs in Discosauriscus suggests a stronger relative influence of environmental fluctuations on this taxon. Some seasonality at Richards Spur, previously demonstrated through other lines of evidence (e.g., Woodhead et al, 2010; MacDougall et al., 2017) is evident in our samples from the clearly defined annuli and growth zones (Figs. 11–12), and it seems probable that these environmental conditions (particularly the ephemerality of bodies of water) would have favored an early onset of metamorphosis in Seymouria. Taxa living in the same habitat and experiencing similar conditions can exhibit different responses to those conditions—amniotes are typically less constrained by water stress relative to non-amniotes, for example. A hypothesis of variable responses to similar climate at Richards Spur is supported by the differing skeletochronological records of co-occurring tetrapods from the site. The dissorophoid temnospondyls exhibit clear LAG patterns (Richards, 2016; Gee, Haridy & Reisz, 2020), whereas the varanopid synapsids exhibit a mixture of LAGs and annuli with growth zones (Huttenlocker & Shelton, 2020), and the captorhinid eureptiles exhibit only annuli with growth zones (Peabody, 1961; Richards, 2016; Huttenlocker & Shelton, 2020). Other hypotheses to explain the differing patterns include secondary environmental factors that may themselves be influenced by climate but in more nuanced and asymmetrical fashions, such as prey availability or limited spatial occupation within the habitat.

Future work will be necessary to identify histological markers of pre- versus postmetamorphic life stages. The present study cannot inform further on this because of the absence of definitive larval specimens of Seymouria, which in turn creates more uncertainty regarding the nature of metamorphosis, if it occurred in this taxon, as well as the external osteological changes associated with this process if it did. Even if our interpretation of the sampled specimens as postmetamorphic is accepted, the timing of either the onset or completion of metamorphosis relative to the time of death remains unclear. In order to more precisely constrain histological markers of metamorphosis will require sampling of taxa, such as Discosauriscus, for which metamorphosis is definitively known and for which osteological changes associated with this process are well-documented.

Lifestyle interpretations. In early tetrapods, interpretations of lifestyle (e.g., aquatic vs. terrestrial) are often based on the presence or absence of features such as lateral line grooves and the degree of development of external features of the limbs (e.g., Moodie, 1908; Schoch, 2002; Witzmann, 2016). Histology has more recently been utilized as a means to further test these hypotheses by means of comparisons with extant taxa in which mode of life can be definitively observed and with the classically utilized external anatomical features (Germain & Laurin, 2005; Kriloff et al., 2008; Sanchez et al., 2010a; Quemeneur, de Buffrénil & Laurin, 2013; Konietzko-Meier, Shelton & Sander, 2016). Seymouria is widely accepted to lack lateral line grooves, although they have been suggested by some to have been present in juveniles (Berman & Martens, 1993, but see Klembara et al., 2006), suggesting a transition in lifestyle throughout ontogeny. Additionally, the limb bones are well-developed, with prominent attachment sites for musculature and distinct processes (Figs. 4–5), and the neural arches are greatly expanded compared to other Paleozoic tetrapods with prominent zygapophyses inferred to have supported the axial column in a terrestrial animal (Sullivan & Reisz, 1999). Our data further corroborate the interpretation of the Richards Spur locality as an assemblage dominated by terrestrial fauna (MacDougall et al., 2017).

The centra also contribute information through inferences on the skeletal mass of the element(s). The two traditionally utilized criteria are the thickness of the periosteal domain and the presence or absence of calcified cartilage. Greatly thickened domains (pachyostosis) and retention of calcified cartilage throughout ontogeny are frequently seen in large-bodied aquatic temnospondyls and probably served to increase the skeletal mass for buoyancy control (Danto et al., 2016). In both Discosauriscus and Seymouria, the periosteal domain is relatively thin, and calcified cartilage is primarily found around the notochordal canal (Danto et al., 2016). In Seymouria, this is the only location of this tissue, whereas calcified cartilage occurs sporadically in the endochondral domain of at least immature individuals of Discosauriscus.

What then can be concluded regarding the histological data from Seymouria postcrania and the lifestyle of the taxon? The femoral microanatomy, specifically the relatively thin cortex and the modest development of the medullary spongiosa, is more compatible with that of a terrestrial animal by comparison with other Paleozoic tetrapods (primarily temnospondyls) that have been inferred to be terrestrial (e.g., Sanchez et al., 2010b; Quemeneur, de Buffrénil & Laurin, 2013). Based on studies of femoral and tibial microanatomy in extant tetrapods (Kriloff et al., 2008; Quemeneur, de Buffrénil & Laurin, 2013), these features also support a primarily terrestrial lifestyle. Collectively, this corroborates the conclusions of previous authors that Seymouria was most likely a terrestrial animal (White, 1939; Berman & Martens, 1993; Sullivan & Reisz, 1999; Marchetti, Mujal & Bernardi, 2017). The vertebral histology also confers support for a terrestrial lifestyle. The periosteal domain is thin, calcified cartilage is sparse and confined to the margin of the notochordal canal, and the neural arch is largely hollow. These data correspond favorably with the broad expansion of the arch and the zygapophyses, which Sullivan & Reisz (1999) interpreted to be for the stiffening of the axial column following White (1939).

Terrestriality in seymouriamorphs. Assessing the range of ecologies among seymouriamorphs from a macroevolutionary standpoint is important because the group has historically been regarded as being well-situated for understanding the skeletal modifications associated with terrestriality. Seymouria is one of the best seymouriamorphs for examining such modifications because complete, articulated skeletons are known (e.g., Berman, Reisz & Eberth, 1987), but it then becomes important to assess whether a terrestrial or aquatic lifestyle is the plesiomorphic state among seymouriamorphs. Given that seymouriamorphs, and reptiliomorphs more broadly, are frequently used as exemplars for the skeletal changes associated with terrestrial adaptation, clarifying the primitive condition of this group is critical for informing accurate comparisons. A conceptual phylogeny, adapted from Klembara (2011), is presented in Fig. 14, illustrating the distribution of lifestyles among seymouriamorphs.

Our data provide strong evidence at the histological and microanatomical scales to support the longstanding hypothesis of terrestriality in Seymouria. This is not a particularly controversial idea; numerous aspects of the external morphology, such as the well-ossified limbs and the massively expanded vertebrae have long been cited as evidence for this lifestyle (Romer, 1956). Although Berman & Martens (1993) described a possible indication of a lateral line system in juvenile specimens of S. sanjuanensis from Germany, subsequent work (Klembara et al., 2006) on an early juvenile did not find any evidence for a lateral line system in other S. sanjuanensis from the same locality. As such, while it is often inferred that Seymouria underwent metamorphosis as with other seymouriamorphs and a number of other terrestrial tetrapods (e.g., some temnospondyls), definitive aquatic larval forms and morphological transitions associated with the presumed metamorphosis are unknown.

At least one other seymouriamorph, Karpinskiosaurus, is also represented only by specimens that lack lateral line grooves (Klembara, 2011). Kotlassia has also been historically regarded as lacking lateral line grooves (e.g., Bystrow, 1944), but the Kotlassia of most previous authors is actually a combination of material referable to the type species, Kotlassia prima, and material properly referable to Karpinskiosaurus (see Bulanov, 2002; Klembara, 2011, for discussion). Whether these grooves are definitively absent in the holotype of Ko. prima is not apparent from previous works that accounted for this historical discrepancy. For Karpinskiosaurus and Seymouria, it has been proposed that these taxa underwent metamorphosis relatively early in their development and lived on land for the majority of their lives (Klembara, 2011).

In contrast, most other seymouriamorphs are known from individuals with lateral line grooves, including Ariekanerpeton (Klembara, 2005), Discosauriscus (Klembara, 1996), Spinarerpeton (Klembara, 2009), and Utegenia (Malakhov, 2000). The most recent phylogenetic analysis that focused on seymouriamorph phylogeny is that of Klembara (2011), which followed a series of anatomical work that re-described virtually all known seymouriamorphs. Mapping the distribution of ecologies onto this topology suggests that seymouriamorphs are primitively aquatic (Utegenia being the earliest diverging taxon) with two separate shifts to terrestriality, one in Karpinskiosaurus and one in Seymouria (Fig. 14). However, caution must be exercised in inferring the phylogeny of a clade in which metamorphosis is known to occur, because biases in the record of premetamorphic larval forms versus that of postmetamorphic terrestrial adults (if such determinations can be made to begin with) can produce misleading data. As with Seymouria (Berman et al., 2000), it has been proposed that Discosauriscus transitioned from an aquatic to terrestrial lifestyle throughout its ontogeny, but even the largest known specimens of Discosauriscus are believed to be juveniles, and none have been recovered from the terrestrial environments that the adult individuals may have inhabited (Klembara, Martens & Bartík, 2001). This may relate to a relatively protracted larval stage that was recovered through the histological work of Sanchez et al. (2008) in which metamorphosis may not have begun until year six of an individual’s life. The latest phylogenetic analyses (Klembara, 2011) do not bear out the slippage that is predicted when coding taxa based on immature specimens (i.e., Discosauriscus is a highly nested seymouriamorph), but this does not negate the potential for this disparity to affect the phylogeny. Ariekanerpeton, Spinarerpeton, and Utegenia are also likely represented only by juveniles (Klembara & Ruta, 2003; Klembara & Ruta, 2005; Klembara, 2009), which warrants consideration.