The Briar Creek Bonebed (Case, 1915, spelled “Brier Creek”) was reported to yield partial skeletons at best, but mostly isolated bones (Case, 1915 and Romer, 1928). Teams from the University of Bonn under the leaderships of P.M. Sander and C.D. Shelton reopened the quarry in 2010 and 2011, collecting numerous bones, including those sampled for this study.

The bonebed is about 30 cm thick and consists of an accumulation of unsorted and disarticulated reptile and amphibian bones of variable preservation encased in a grey mudstone. Some bones have a heavy concretionary cover of ironstone. The Briar Creek Bonebed taphonomy is unique in that some of the bones are considered to have decomposed before preservation (Case, 1915). Case (1915) originally hypothesized that this area represented a pool or swamp that functioned as a macerating tank. The deposit could also have been an oxbow lake or swamp formed by the meandering of a river similar to the hypothesized paleoenvironment described by Sander (1989) for other Lower Permian bonebeds in Archer County. The isotope analyses indicate a non-marine origin of the localities from the Early to Middle Permian of the southwestern USA, including the Archer City Formation (Fischer et al., 2014).

Thirteen long bones, mostly complete, recovered during the 2010 and 2011 excavations pertaining to different anamniotic stegocephalians were chosen for consumptive sampling (Table 1). All bones belong to the histological collection of the University of Bonn, Steinmann Institute, Germany (IPBSH).

Taxonomic determination – Taxonomic determination of the sampled bones was made by comparison with the literature on Early Permian tetrapod morphology (Case, 1911, Cope, 1878, Kissel, 2010, Miner, 1925, Pawley and Warren, 2006 and Romer, 1957) and by personal studies of Lower Permian vertebrate collections curated in institutions in the United States undertaken by CDS. These include the Dallas Museum of Natural History (DMNH), the Museum of Comparative Zoology of Harvard University (MCZ), the Field Museum of Natural History in Chicago (FMNH), the Texas Memorial Museum (TMM) in Austin, TX, the Sam Noble Museum of Natural History at the University of Oklahoma in Norman, OK (OMNH), the American Museum of Natural History in New York City (AMNH), the University of Michigan Museum of Paleontology in Ann Arbor, MI (UMMP), and the Yale Peabody Museum of Natural History in New Haven, CT (YPM).

While Archeria was well described by Romer (1957) based on articulated skeletons from the Geraldine Bonebed, Archer County, Texas, regrettably there are no complete monographic descriptions of Eryops based on articulated specimens despite the exceptional preservation of several skeletons in the Geraldine Bonebed (Sander, 1987). Miner (1925) provided the description only of the pectoral limb of Eryops. Pawley and Warren (2006) redescribed the anatomy of Eryops, basing the descriptions on numerous isolated bones in various collections instead of on the articulated skeletons from the Geraldine Bonebed that are on exhibit in several North American museums (Sander, 1987).

Histological sections – All bones were sectioned transversely at or near the midshaft and reconstructed after cutting. The thin-sections were prepared according to standard petrographic methods (Lamm, 2014). Sections were ground and polished using wet SiC grinding powder (600 and 800 grit). The thin sections were then studied under a LEICA DMLP light microscope in normal and polarized transmitted light. Images of entire slides showing whole bone cross sections were produced with an Epson V740 PRO high resolution scanner.

The determination of histotypes – In order to test if the histological framework is consistent with taxonomy the sections were, independently of the taxonomic determination, sorted into histotypes based on the microanatomical and histological characters (character of primary tissue, vascular canals organization, degree of remodeling, growth marks). The histological terminology follows Francillon-Vieillot et al. (1990).

Humerus – All humeri (Table 1) represent the basal tetrapod tetrahedral form with a relatively low degree of torsion in the shaft. The short, thick humeral shaft is somewhat laterally compressed, and roughly quadrangular in cross section. Specimens IPBSH-23 and IPBSH-48 display an entepicondylar foramen that is unusual for temnospondyls. The most diagnostic bones are humeri IPBSH-48 and 23, having an expanded deltopectoral crest forming a highly developed flange which projects anteriorly. The anterior flange in the smallest specimen IPBSH-38 is not as distinctive as in IPBSH-23 and IPBSH-48. Based on this unique morphology of the deltopectoral crest all three bones were identified as Archeria (Romer, 1957).

Radius – The single radius in this study, IPBSH-75, is a stout element with flared proximal and distal ends. The distal portion is flattened in the extensor - flexor plane and broader than the proximal portion. The extensor surface is convex in transverse section, whereas the flexor surface is gently concave. A sharp-edged ventromesial radial ridge passes down the midline of the anterior surface. On the flexor surface, the ventral radial crest is located slightly medial to the posterior edge. The proximal articular surface is sub-rectangular whereas the distal articular surface is more triangular. The morphology of radius IPBSH-75 is indicative of Eryops (Miner, 1925 and Pawley and Warren, 2006).

Ulna – The ulnae (Table 1) are relatively slender, and their shaft is anteroposteriorly flattened (Fig. 2A-C). The ulnar extensor keel has a rugosity on the proximal region, indicating the attachment to the triceps muscle (Fig. 2A). On the posterior surface, a low ulnar crest originates below the humeral articular surface and branches at the midpoint of the shaft into the posterolateral ulnar crest. In proximal view, the humeral articular surface is approximately quadrangular; only in ulna IPBSH-74 is it thinner. In specimens IPBSH-73 and IPBSH-74, the olecranon process is more ossified and raised like in IPBSH-76 (Fig. 2A–C). Ulnae IPBSH-76 and IPBSH-73 (Fig. 2A-B) both have a broad proximal head that is diagnostic of Eryops (Miner, 1925 and Pawley and Warren, 2006). Ulna IPBSH-74 (Fig. 2C) has a flat humeral articular surface suggestive of an affinity with Archeria (Romer, 1957).

Femur – The four femora (Table 1) represent the plesiomorphic condition of Permo-Carboniferous tetrapods, essentially a stout cylinder with expanded ends and ventral structures associated with muscular attachments (Fig. 2D–F). As is typical of Permo-Carboniferous tetrapods, the proximal articular surface is a ribbon-shaped oval that extends anteroposteriorly, being relatively narrow dorsoventrally, convex in outline dorsally and concave ventrally. The distal articular face of the femur is triangular. The posterior fibular condyle is rectangular, and the distal articular facet of the femur is bisected on the dorsal side by the prominent intercondylar fossa (fossa tendinalis) and tibial condyle. On the ventral side, proximal to the posterior edge of the articular surface, the distal end of the adductor crest constitutes the apex of a triangle. The fossa poplitea is located anteriorly to the apex. The ventral surface of the femur possesses an intertrochanteric fossa, internal and fourth trochanters, a longitudinal adductor crest, and a concave popliteal space (Fig. 2D–F). The internal trochanter lies close to the head of the bone at the proximal end of a raised and thickened ridge. As the ridge continues distally along the trochanter, it descends into the general contours of the shaft, which bears a prominent pit and associated rugosities, perhaps representing the fourth trochanter.

Femora were taxonomically assigned (Fig. 2D–F) mostly based on the different shape of the proximal head and different architecture of adductor crests.

In femora IPBSH-64 and IPBSH-78, the anterior part of the proximal articular surface is somewhat thicker than the posterior part. Both bones have a very prominent adductor crest, which arises near the fourth trochanter, essentially in line with the base of the internal trochanter. The adductor crest slants distally across the shaft to terminate laterally at the ventral tip of the fibular articular surface. In femur IPBSH-78, the adductor crest bifurcates distally, enclosing the popliteal space. Based on these characters femora IPBSH-64 and IPBSH-78 (Fig. 2D) can be identified as Eryops (Pawley and Warren, 2006).

The exquisite preservation of IPBSH-77 (Fig. 2E) allowed the identification of morphological characters (the articular face of the proximal end is nearly flat and descends almost perpendicularly along the posterior, low adductor crest) that are diagnostic of Diadectes (Case, 1911).

In femur IPBSH-65 (Fig. 2F), difference in size between the anterior and posterior parts of the articular surface is slight. Femur IPBSH-65 with its massive proximal articular surface and low-reaching adductor crest belongs to Archeria (Romer, 1957).

Tibia – The single tibia in this study, IPBSH-68 (Table 1), is a slender bone with a gently expanded proximal head. The cnemial crest does not project sharply but rather it is a small rounded swelling at the extensor margin of the bone. The cnemial trough forms a shallow constriction on the extensor side of the femoral articular surface, marking the boundary between the anterior and posterior articular surfaces. The distal articular surface is more triangular, with a flattened flexor edge and strongly convex exterior side with continuous articulation facets to the fibula. The flexor surface bears a series of crests; however, the bone surface is poorly preserved so the morphological details are not obvious. Based on the morphology tibia IPBSH-68 belongs to Archeria (Romer, 1957).

Fibula – The single fibula in this study, IPBSH-81 (Table 1), has a slightly expanded and broadened head with a terminal articular surface for the femur, a flattened shaft, a broad proximal end, and a distal segment with a terminal surface for the articulation of the tarsus. In proximal view, the proximal surface twists at approximately 300 to the distal surface. The bone is strongly concave along its anterior (tibial) margin; the posterior edge is nearly straight. The articular surface for the femur is a crescentic oval, somewhat convex in outline along its extensor margin, concave on the opposite edge. The broad fibular extensor ridge bulges out from the femoral articular surface and distally passes directly down the proximal head on the extensor side. A short anterior fibular ridge is located on the anterior proximal surface. The flexor surface of the shaft is gently concave along its length. A deep fibular sulcus crosses the anterodistal corner. The shallow posterior fibular ridge runs along the posteromedial edge of the bone. Morphology of the fibula IPBSH-81 suggests an Eryops affinity (Pawley and Warren, 2006).

Five histotypes are recognized amongst the sampled long bones (Table 1).

Histotype I – This histotype is recognized in the smallest ulna (IPBSH-76; Table 1), based on the morphology assigned to Eryops. There are numerous vascular canals arranged in regular concentric rows (Fig. 3). The primary tissue is parallel-fibered and woven bone (Fig. 3C). There is a high number of osteocyte lacunae. Remodeling is concentrated at the medullary margin (Fig. 3B). No growth marks are present. This histological framework suggests an early ontogenetic stage, but it is not possible to narrow down the stage further. This is why we decided to treat the immature bone as a separate histotype.

Histotype II – Histotype II is only represented by femur IPBSH-77 (Fig. 4), which can be confidently assigned to Diadectes. The unremodeled cortex is relatively thin; the dorsal side consists of parallel-fibered bone, in the ventral region the tissue is more organized than the dorsal side. The remodeling of the cortex increases gradually towards the remodeled perimedullary region. There is no clear boundary between cortex and medullary region. The medullary region is large, with a few thickened trabeculae (Fig. 4A). This histotype has a diagnostic vascularization consisting of the slightly oblique, but clearly radially arranged vascular canals (Fig. 4B–C). The osteocyte lacunae are small, but numerous. Sharpey's fibers are only visible on the ventral side in the region of the adductor crest.

Histotype III – The primary tissue of this histotype, which is displayed in the femora that can confidently be assigned to Eryops, consists of lamellar or parallel-fibered bone. Longitudinal vascular canals are arranged in regular rows throughout the entire cortex, grading from simple vascular canals in the outer cortex to more primary osteons in the deeper cortex (Fig. 5A). The degree of remodeling increases from the bone periphery inwards from secondary osteons to erosion cavities (Fig. 5C). Between the rows of vascular canals are layers of a highly organized bone matrix with growth marks. In femur IPBSH-78, two to four growth lines in each annulus are visible (Fig. 5C). Throughout the entire section, the osteocyte lacunae are highly concentrated and circular in section (Fig. 5C). Sharpey's fibers are sparse, mostly present in the region of the adductor crest. The cortex transitions into the medullary region through the remodeled perimedullary region. However, the borders between the primary cortex, perimedullary region, and medullary space are relatively clear. The medullary region is occluded by a few thick trabeculae, but the central region remains empty.

Histotype IV – The tissue type is recognized by a thick, lamellar bone layer near the surface and is seen in bones (humeri, a femur and a tibia) that can confidently be assigned to Archeria (Fig. 6C–H); moreover, in the ulna (IPBSH-73) and fibula (IPBSH-81), both have a morphology diagnostic of Eryops (Table 1, Fig. 6A and B). There is a sharp contrast between the borders of the primary lamellar layer and the highly vascularized remodeled cortex (Fig. 6B). The degree of remodeling here depends on the size of the bone. In humerus IPBSH-38, dense vascularization is mostly simple canals with the deposition of primary lamellar bone in some vascular canals; small patches of incipient fibrolamellar bone (FLB) are also present (Fig. 6C–D). Primary osteons dominate the cortex of humerus IPBSH-48 (Fig. 6E–F), whereas humerus IPBSH-23, ulna IPBSH-73, tibia IPBSH-68, and fibula IPBSH-81 mostly show secondary osteons (Fig. 6G–H). In femur IPBSH-65, two vascularized layers separated by an avascular layer are present. The outer layer is composed of primary osteons and the inner layer is composed of secondary osteons deeper in the cortex. This region of the cortex gradually transitions to the medullary region with thick trabeculae (Fig. 6A). The perimedullary region was created by the extensive remodeling of the medullary region, which is open in the center (Fig. 6A). The osteocyte lacunae in the avascular layers are dense and flattened; they are more circular in the vascularized regions. The presence of Sharpey's fibers is extensive throughout the section.

Histotype V – A highly porous cortex is present in two bones, ulna IPBSH-74 and radius IPBSH-75, which morphologically belong to Archeria and Eryops, respectively. The outermost layer of these bones consists of thin lamellar bone (Fig. 7A). The vascular canals are visible only next to the lamellar layer. The larger part of the cortex is highly porous, resembling spongy bone with regular thin trabeculae consisting of primary and secondary tissue (Fig. 7B–D). No growth marks are preserved. In the central part of the sections, there is an open medullary cavity. Osteocyte lacunae are numerous and spindle-shaped. Extensive Sharpey's fibers are visible through the primary cortex.

The previous description of the faunal content of the Briar Creek Bonebed provided by Romer (1957) and the morphological characters of these bones limit the probable identity of the potential taxa to three well known genera: Eryops, Archeria, and Diadectes (Table 1). However the histological variation visible among the studied bones is higher. The bones based on the morphology assigned to Archeria represent two histotypes (IV and V). The Eryops bones show higher variability and represent the histotypes I, III, IV and V (Table 1). For humeri and femora, the histotype seems to be directly related to the taxon. Histotype II is diagnostic of Diadectes, histotype III of Eryops, and histotype IV of Archeria (Fig. 4, Fig. 5 and Fig. 6, Table 1). That pattern is not observed in the ulnae, fibula and radius. These bones do not repeat the histological framework visible in the propodials representing the new histotypes (I and V) or the histotypes known from femora, but assigned to different taxa. Both large ulnae are similar in length, but show a discrepancy between morphology and histology (Table 1). Ulna IPBSH-73 (Fig. 2B), which is assigned to Eryops based on morphology, shows histotype IV (Fig. 6A-B, Table 1), diagnostic of Archeria in propodials. The flat humeral articular surface of ulna IPBSH-74 suggests that this bone belongs to Archeria (Fig. 2C), but it shows histotype V (Fig. 7), not seen in any propodials. Moreover, the second bone with histotype V (radius IPBSH-75) morphologically resembled Eryops. Fibula IPBSH-81 represents histotype IV, but morphologically it is slender and not as broad distally as described by Romer (1957) for Archeria.

Specimens morphologically identified as Eryops show four different histotypes, and morphologically Archeria bones represent two histotypes (Table 1).

There are several potential explanations for this histotype variability: it may be the result of methodological error such as erroneous morphological interpretation or non-midshaft plane of section. Variability might also have an intraspecific origin, or an interspecific origin, with bones of similar morphologies belonging to different species.

Methodological origin – The low specialization, postulated paedomorphic character of postcrania and lack of an ontogenetic series of large Temnospondyli may lead to ambiguous taxonomic determination of unassociated postcranial elements (Pawley and Warren, 2004). Erroneous taxonomic assignment of these bones thus may be the origin of the problematic interpretations applied to these specimens. However, in the bones described here, the morphological features characteristic for each taxon are well preserved and their taxonomic affinities based on the morphology seems to be unequivocal (Case, 1911, Cope, 1878, Kissel, 2010, Miner, 1925, Pawley and Warren, 2006 and Romer, 1957). Nevertheless, it is possible that more taxa than currently recognized are present in the locality.

The plane of section also might have an influence on the apparent histotype. Konietzko-Meier and Klein (2013) showed that the bone microstructure is highly variable within millimeters of the sampled region. The histological midshaft with the neutral zone (the area where growth started) is more locally restricted in plesiomorphic tetrapod long bones than the morphological midshaft. Different sampling planes thus may intersect a variable microstructure of the medullary region; the density of trabeculae increases proximally and distally together with decreasing cortical bone thickness. Laurin and Buffrénil (2015) documented the misinterpretation of lifestyle based on the microstructural features of Ophiacodon because the section plane was not precisely mid-diaphyseal. However, the specimens from Briar Creek Bonebed varied not only on the microstructural level, but also represent the different histology (growth pattern, tissue type, vascularization pattern), suggesting that the plane of section cannot explain all of the observed variability.

Intraspecific variability – There are a number of reasons for a high intraspecific variability of bone microstructure; environmental fluctuations, developmental plasticity and different individual age, intraskeleton variability, various growth strategies, genetic variability, sexual dimorphism or other drastic changes in the life cycle (Woodward et al., 2014). Previously sampled anamniotic tetrapods that best demonstrate intraspecific histological variability are Dutuitosaurus ouazzoui (Steyer et al., 2004), Metoposaurus diagnosticus from Poland (Konietzko-Meier and Klein, 2013 and Konietzko-Meier and Sander, 2013) and Gerrothorax pulcherrimus (Sanchez and Schoch, 2013).

The histological variant of the smallest Eryops ulna (IPBSH-76) may result from developmental immaturity. The cortex in the smallest ulna is histologically immature, representing histotype I (Fig. 3). However, because it is immature, it is possible that the vascular pattern may change during ontogeny. Even if histotype I, represented by the smallest ulna, only represents juvenile ontogenetic variation of Eryops, we still observe a discrepancy between the histology and morphology among other bones, which based on the morphology, can be assigned to a taxon and lack clear immature characters.

One of the potential explanations is environmental plasticity. Environmental plasticity has been documented in the bones of Gerrothorax pulcherrimus, where it is expressed as a variable growth rate, duration of juvenile period, age at maturity, and life span (Sanchez and Schoch, 2013). This variability is seen in sympatric and allopatric populations. Regardless, the bones of this taxon always demonstrate common histological features in tissue organization and general growth pattern (Sanchez and Schoch, 2013), in contrast to the intrageneric histological variability observed in the bones from the Briar Creek Bonebed.

One of the main problems with the histological sections studied here is the origin of histotype V, visible in two taxa and bones: Eryops radius and Archeria ulna. A possible scenario for the origin of this histotype is provided by the extensive cortical erosion that occurs during the development of Gerrothorax (Sanchez and Schoch, 2013). In the earliest stage, the cortex is very compact and composed of fibrous or parallel-fibered bone, occasionally crossed by some radial extrinsic fibers. Later in development, this cortex is deeply pierced by numerous bays of erosion and becomes lamellar in the periphery (Sanchez and Schoch, 2013). This suggests that our bones with a spongy structure (histotype V) may represent the late ontogenetic forms of one of the other histotypes (III or IV) visible in the femora and humeri (Fig. 7). However, it is not possible to determine which one with the data at hand.

The published histological studies of an Eryops femur (Sanchez et al., 2010 and Quémeneur et al., 2013) describe the microanatomy of this bone as highly spongy. This suggests that histotype V is characteristic of Eryops, especially if the radius (IPBSH-75) representing this histotypes morphologically belongs to this genus. Also, in the largest sampled Eryops femur (IPBSH-78) that displays histotype III, the shape of the erosion cavities is similar to those observed in spongy bones, in which osteoclast activity increases in the medullary and perimedullary regions. However, even in the largest femora (IPBSH-78), a clearly developed lamellar-zonal cortex is always present, contrary to histotype V.

Based on the histological framework, bones with histotype V may also belong to Archeria. The remnants of the primary tissue are still visible in the spongy cortex interstices between secondary osteons (Fig. 7C) and the lamellar bone. The outermost layer visible in histotypes V is similar to those in histotype IV. This suggests that histotype V represents the late developmental stage of histotype IV, and all bones with histotypes IV and V, despite the strong morphological variations, belong to one taxon, which in this case is Archeria.

However, histotypes IV and V may represent adaptations to very different modes of life. Osteosclerosis observed in the histotypes IV may be produced by bone mass increase, which results in heavy, dense bone typical of slow swimmers living in shallow water (Canoville and Laurin, 2009, Houssaye, 2009, Laurin et al., 2009, Laurin et al., 2011 and Ricqlès and Buffrénil, 2001). The spongy condition present in the histotypes V is characteristic of open marine active swimmers (Houssaye et al., 2013, Houssaye et al., 2014 and Ricqlès and Buffrénil, 2001). Based on these observations, Archeria may present two opposite life strategies; pelagic, slow swimmer and active swimmer. Histotype IV confirms the traditional interpretation of Archeria as an aquatic, slow animal. Histotype V is closer to the hypothesis proposed by Laurin and Soler-Gijón (2010), which suggests a possible marine lifestyle in Archeria, though not necessarily pelagic.

This dual signal visible in the bones assigned to one taxon is plausible if the animal's mode of life changed during its life cycle. The ulnae, with histotypes IV and V, are very similar in size, and based on the morphology (ossification through the olecranon process), they seem to be at the same ontogenetic stage. The similarity in morphological stage excludes the slow, developmental origin of different histotypes IV and V and suggests a parallel development of described structures during the same ontogenetic stage.

The most drastic change in the mode of life during the life cycle of extant amphibians is associated with metamorphosis. The earliest known records of metamorphosis are based on morphological characters of Permo-Carboniferous dissorophoids (Schoch, 2009, Schoch, 2014 and Schoch and Fröbisch, 2006) and seymouriamorphs (Laurin, 2000). In Lissamphibia, metamorphosis is sometimes visible histologically as a metamorphosis line, which is a thick line close to the medullary region, but a transition from one histotype to another has not been observed (i.e. Ento and Matsui, 2002, Esteban et al., 1999, Esteban et al., 2002, Hemelaar, 1988, Jakob et al., 2002, Khonsue et al., 2001, Khonsue et al., 2002, Montori, 1990 and Olgun et al., 2001). However, as of yet the histological variation between the pre-metamorphic and post-metamorphic forms, not only for early stegocephalians, but also for lissamphibians, is unknown.

The remaining possibility is the parallel development of different microstructures throughout life history representing i.e. the sexual dimorphism. However, it is not possible to assess this cause of variability here because we lack data on the sex of the sampled individuals.

To understand the source of this variability requires further sampling of more ontogenetic series of other taxa.

Interspecific variability – Our initial histological studies of various temnospondyl intercentra suggested that temnospondyl bone histology may contain a taxonomic signal (Konietzko-Meier et al., 2014). We cannot exclude the possibility that several tetrapod taxa with similar bone morphologies, but different histotypes, are preserved in the Briar Creek Bonebed. This suggests either that the taxonomic diversity of the Briar Creek locality has been underestimated in the past, or that infraspecific histological variability in early limbed vertebrates was greater than currently recognized.