Assignment to Brachyopidae

We follow Watson (Reference Watson1956), Cosgriff (Reference Cosgriff1969), and Warren and Marsicano (Reference Warren and Marsicano1998, Reference Warren and Marsicano2000) in assigning Platycepsion wilkinsoni to the Brachyopidae. Warren and Marsicano (Reference Warren and Marsicano1998, Reference Warren and Marsicano2000) listed the following derived characters shared by P. wilkinsoni and the Brachyopidae: (1) deep skull with exoccipital condyles extending far posterior; (2) absence of tabular horns; (3) occipital portion of squamosal convex (although we cannot follow point [3] because we do not see a convex portion of the squamosal); (4) postglenoid area of mandible elongate; (5) narrow pectoral girdle. Point (5) deserves the following comment. In our view, a narrow pectoral girdle probably is not a characteristic of brachyopids or brachyopoids. Warren and Marsicano (Reference Warren and Marsicano1998) emphasized that the pectoral girdle of Platycepsion wilkinsoni is strikingly narrower than the width of the skull and pointed to similar proportions in the chigutisaurid Siderops kehli (Warren and Hutchinson, Reference Warren and Hutchinson1983). These authors thus hypothesized that a pectoral girdle much narrower than the skull could be a derived brachyopoid characteristic, but admitted that these proportions have been observed only in the two mentioned taxa. However, the condition in Siderops kehli is probably not comparable with that in P. wilkinsoni. Siderops kehli has a well-ossified skull, which is extremely broad also when three-dimensionally reconstructed (Warren and Hutchinson, Reference Warren and Hutchinson1983). In contrast, the wide skull of P. wilkinsoni is the result of complete flattening in one layer. In life, this skull was definitely deeper (Figs. 5, 6). More generally, a narrow pectoral girdle is a plesiomorphic character among temnospondyls, as exemplified by edopoids, Dendrysekos helogenes Steen, Reference Steen1934, zatracheids, and trematosaurids (Schoch and Milner, Reference Schoch, Milner and Wellnhofer2000, 2014; Witzmann and Schoch, Reference Witzmann and Schoch2006a), whereas the pectoral girdle is at least the width of the skull in plagiosaurids such as Gerrothorax pulcherrimus (Hellrung, Reference Hellrung2003) and probably Plagiosaurus depressus Jaekel, Reference Jaekel1914 (Nilsson, Reference Nilsson1937). An additional character—the wide and blunt posterior end of the interclavicle—is shared among chigutisaurids, brachyopids, and plagiosaurids. The interclavicle of P. wilkinsoni is consistent especially with that of Siderops kehli in outline, except for the concave posterior end.

Figure 5.

Platycepsion wilkinsoni (MMF12572), reconstruction of skeleton with body contour and external gills.

Figure 6.

Brachyopid skulls in dorsal view: (1) Brachyops laticeps Owen in Anonymous, 1854; (2) Xenobrachyops allos (Howie, Reference Howie1972); (3) Bothriceps australis; (4) Bathignathus watsoni; (5) Batrachosuchus browni; (6) Platycepsion wilkinsoni; (7) Notobrachyops picketti; (8) Vigilius wellesi; (9) Sinobrachyops placenticephalus Dong, Reference Dong1985. Platycepsion wilkinsoni is based on the present study, Notobrachyops picketti is based on Warren and Marsicano (Reference Warren and Marsicano1998), and all others are redrawn from Schoch and Milner (Reference Schoch, Milner and Seus2014) and references cited therein.

Derived characters (1) and (2) can also be found in tupilakosaurid dvinosaurians, but the elongate postglenoid process that is visible on the left mandibular ramus clearly precludes tupilakosaurid affinities. This is further supported by the fact that there is no evidence of a dorsally exposed palatine, which is a diagnostic character of tupilakosaurids (Schoch and Milner, Reference Schoch, Milner and Seus2014). The basioccipital, which makes a major contribution to the single occipital condyle in dvinosaurs, is absent and a wide gap remains between the exoccipitals. Finally, the short postorbital in P. wilkinsoni differs from the much expanded postorbital of tupilakosaurids. An additional and very distinctive brachyopid character of P. wilkinsoni is the unusually wide lateral line sulci (not present in dvinosaurs), which otherwise are known only from the genera Mastodonsaurus (Schoch, Reference Schoch1999, fig. 9) and Buettneria (Sawin, Reference Sawin1945, fig. 3), where outgroup comparison shows that they evolved in parallel. Despite the larval state of P. wilkinsoni, this feature is as pronounced as in Bathignathus watsoni (Haughton, Reference Haughton1925).

Phylogenetic position within Brachyopidae

Platycepsion wilkinsoni is consistent with many brachyopids in the parabolic skull outline, the position of the orbits, and proportions of the preorbital and postorbital region (Fig. 6). The wide lateral line sulci are shared with Bathignathus watsoni and Batrachosuchus browni Broom, Reference Broom1903, but the highly apomorphic proportions of these taxa, such as the postfrontal (posteromedially extended and medially convex), slender parietal, and much wider supratemporal, are not shared by P. wilkinsoni, which instead has a more plesiomorphic condition similar to that of Bothriceps australis Huxley, Reference Huxley1859. We conclude that P. wilkinsoni was most likely related to a clade that was recovered by Warren and Marsicano (Reference Warren and Marsicano2000) and includes Vigilius wellesi Warren and Marsicano, Reference Warren and Marsicano2000, Bathignathus watsoni, and Batrachosuchus browni. Any further, more exact placement within the tree is hampered by the larval morphology.

Validity of the taxon Platycepsion wilkinsoni

Warren and Marsicano (Reference Warren and Marsicano1998, Reference Warren and Marsicano2000) pointed out that the contact between parietal and postorbital was the only autapomorphy of Platycepsion wilkinsoni. We do not see evidence for this “dvinosaurian” bone configuration and prefer the more conservative interpretation that postorbital and parietal are separated by the postfrontal. The poor preservation precludes a definitive statement and weakens this character as an autapomorphy of Platycepsion wilkinsoni.

A unique character among brachyopids (and other temnospondyls) forms the shape of the interclavicle. Cosgriff (Reference Cosgriff1969) stressed that in contrast to most temnospondyls, the posterior process of the interclavicle in P. wilkinsoni does not have a rounded posterior end but is rather straight. Indeed, as described above, the posterior end is even slightly concave. The only other brachyopid in which a complete interclavicle is known is Blinasaurus townrowi, whose interclavicle has a rounded posterior end (Cosgriff, Reference Cosgriff1974, fig. 26). Cosgriff (Reference Cosgriff1974) even consulted the different posterior ends of the interclavicles to justify that B. townrowi and Platycepsion wilkinsoni (referred by him to as Blinasaurus wilkinsoni) are different species. The interclavicles known in chigutisaurids have a similar diamond shape as in Platycepsion wilkinsoni and B. townrowi, but in contrast to P. wilkinsoni they have a rounded or pointed posterior end (Warren and Hutchinson, Reference Warren and Hutchinson1983). A possible exception is the chigutisaurid Koolasuchus cleelandi whose interclavicle might have a straight or faintly convex posterior end (Warren et al., Reference Warren, Rich and Rich1997, fig. 9). In plagiosaurids, the interclavicle is roughly triangular in shape, with the apex pointing anteriorly and the posterior border being straight (Nilsson, Reference Nilsson1937; Hellrung, Reference Hellrung2003). To our best knowledge, a concave posterior margin of the interclavicle appears to be a unique character among temnospondyls. We suggest this character as a possible autapomorphy of P. wilkinsoni, although we cannot rule out that the morphology of the interclavicle might have changed during further ontogeny. A further autapomorphic character of P. wilkinsoni might be the presence of five pairs of ceratobranchials. However, this character is weakened by the fact that the hyobranchial morphology is poorly known in other brachyopids.

Finally, the morphology of the skull as a whole differs from all other brachyopids and chigutisaurids in the large size of the orbits, the short postfrontal and postorbital, and the relatively small supratemporal, especially compared with the parietal. These features are probably larval traits and it is impossible to deduce the adult morphology of the taxon, but at present they give a distinctive cranial morphology that further distinguishes P. wilkinsoni, especially, from the taxa that have similar skull outlines—Bathignathus watsoni and Batrachosuchus browni. This said, we emphasize that comparison with such large specimens is necessarily limited because size differences are enormous. Only plagiosaurids have proportionally larger orbits, but their skull morphology is highly derived, with either having huge combined orbitotemporal fenestrae (Plagiosuchus pustuliferus [Fraas, Reference Fraas1896]) or skulls being more than twice wider than long (Plagiosterninae and Plagiosaurinae) (Schoch and Milner, Reference Schoch, Milner and Seus2014).

Interpretation of the ontogeny of Platycepsion wilkinsoni

The present study demonstrates for the first time soft-tissue preservation of external gills in P. wilkinsoni. External gills are a larval specialization in temnospondyls and other tetrapods (see discussions in Schoch and Witzmann, Reference Schoch and Witzmann2011; Witzmann, Reference Witzmann2013). Therefore, P. wilkinsoni is not only immature, but indeed a larval form. The alternative interpretation, that P. wilkinsoni is a paedomorphic adult that retained its larval external gills (see Watson, Reference Watson1956), is unlikely for the following reasons. The extremely short snout and the large orbits and pineal opening stand in contrast to an interpretation as a small adult or juvenile. Only adult plagiosaurids have proportionally shorter snouts, but their skulls are highly aberrant with huge orbits (Schoch and Milner, Reference Schoch, Milner and Seus2014). What is even more important is the extremely low degree of ossification and differentiation of the postcranial skeleton. Adult temnospondyls with paedomorphic skull proportions (e.g., dvinosaurians or lapillopsids and lydekkerinids among stereospondyls) have well-ossified postcranial skeletons throughout (Schoch and Milner, Reference Schoch, Milner and Wellnhofer2000, Reference Schoch, Milner and Seus2014). Thus, the interpretation of the specimen as a larva appears the most plausible for us.

Platycepsion wilkinsoni is the only unambiguously known larva of a stereospondyl, despite the fact that growth series are known from certain Early Triassic stereospondyls such as Warrenisuchus aliciae (Warren and Hutchinson, Reference Warren and Hutchinson1988) (see, however, the comment of Steyer [Reference Steyer2003] that this series is not represented by a single species), Edingerella (Watsonisuchus) madagascariensis (Lehman, Reference Lehman1961) (Steyer, Reference Steyer2003), and Benthosuchus sushkini (Efremov, Reference Efremov1929) (Bystrow and Efremov, Reference Bystrow and Efremov1940). Even their smallest specimens lack unambiguous larval characters.

It is worth mentioning here that the ~40 mm long, poorly preserved skull of Notobrachyops picketti Cosgriff, Reference Cosgriff1973, has been assigned to brachyopids with reservation (Warren and Marsicano, Reference Warren and Marsicano1998, Reference Warren and Marsicano2000). If this assignment is correct, then N. picketti is the second smallest brachyopid after P. wilkinsoni. However, the proportionally long preorbital region, the rather small orbits, pineal opening and, to a lesser degree, the absence of lateral line sulci support the interpretation of Warren and Marsicano (Reference Warren and Marsicano1998) that the specimen is an adult. It thus provides no ontogenetic data and cannot be compared with P. wilkinsoni.

In the following, a short summary of the general larval ontogeny in temnospondyls is given. It is then compared with the degree of ossification and differentiation of the skeleton in P. wilkinsoni to derive information about its developmental stage and, as far as possible, the sequence of ossification.

General features of Paleozoic temnospondyl larvae.—A great number of well-preserved specimens enabled detailed study of the formation of skull and postcranium in Permo-Carboniferous larval temnospondyls (e.g., Boy, Reference Boy1974; Milner, Reference Milner1982; Schoch, Reference Schoch1992, Reference Schoch2002, Reference Schoch2003, Reference Schoch2004; Werneburg, Reference Werneburg1992, Reference Werneburg2002, Reference Werneburg2021; Witzmann and Pfretzschner, Reference Witzmann and Pfretzschner2003; Witzmann, Reference Witzmann2005, Reference Witzmann2006; Fröbisch et al., Reference Fröbisch, Carroll and Schoch2007, Reference Fröbisch, Olori, Schoch and Witzmann2010). These studies indicate that rapid ossification of the dermal skull roof and palate early in ontogeny, contrasted with a much slower pace in postcranial development, is plesiomorphic for temnospondyls. A more derived early ontogeny is found in the Branchiosauridae, a group of larval or paedomorphic dissorophoids in which formation of the dermal skull bones was delayed, whereas the postcranial elements, especially the limbs, ossified and differentiated earlier (Boy, Reference Boy1974; Schoch, Reference Schoch1992, Reference Schoch2004).

The ossification sequence in skull and postcranium of Paleozoic temnospondyls generally proceeded in the following pattern. Endocranial bones started to ossify after formation of the skull roof, the first elements to ossify being usually the exoccipitals and quadrates. Postcranial ossification commenced with undifferentiated, paired neural arches in the anterior trunk region and ossified in an anterior to posterior direction, followed by successive formation of the ribs, the humerus, and the dermal bones of the pectoral girdle. The endochondral scapulocoracoid ossified much later (first the scapula and, much later, the coracoid), and the ilium was the first element of the pelvic girdle that started ossification already in small larvae, followed by the ischium and, much later, the pubis. The latter remained cartilaginous in many aquatic forms. Initial ossification of the femur occurred after the humerus and the dermal pectoral girdle had started to ossify (Schoch, Reference Schoch1992; Witzmann and Pfretzschner, Reference Witzmann and Pfretzschner2003).

Developmental stage of P. wilkinsoni and comparison with other larval temnospondyls.—The dermal skull roof of Platycepsion wilkinsoni is well ossified and bears distinct ornament with broad lateral lines, and the exoccipitals not only have started to ossify, but are already well differentiated with conspicuous occipital condyles and paroccipital processes. Similarly, the viscerocranium is in an advanced stage of development with massive branchial arches and a well-differentiated stapes. The ossified branchial arches are no larval character. Their presence in adult brachyopids (Warren et al., Reference Warren, Rozefelds and Bull2011) indicates that they also were retained in P. wilkinsoni in the adult stage.

The stapes is of special interest because this element has been rarely reported in brachyopids. In the brachyopid Batrachosuchus watsoni and the plagiosaurids Gerrothorax pulcherrimus and Megalophthalma ockerti Schoch, Milner, and Witzmann, Reference Schoch, Milner and Seus2014 (Watson, Reference Watson1956; Witzmann et al., Reference Witzmann, Schoch, Hilger and Kardjilov2012; Schoch et al., Reference Schoch, Milner and Witzmann2014), it forms a highly aberrant, blade-like element. In contrast, the probably closely related chigutisaurids had a rather plesiomorphic stapes, with a long rod-like shaft (Warren and Hutchinson, Reference Warren and Hutchinson1983). In Platycepsion wilkinsoni, the stapes was still in an immature stage (as indicated by its consistency with the stapes of Micromelerpeton credneri), but clearly represents the rod-like type found in chigutisaurids and most other stereospondyls, which is also widespread in Paleozoic taxa (Schoch, Reference Schoch2017). These data suggest that the blade-like stapes of brachyopids and plagiosaurids was derived, and not paedomorphic as suggested by Schoch (Reference Schoch2017), but it remains unclear what the stapes of adult P. wilkinsoni looked like.

In contrast to the skull and viscerocranium, the postcranial skeleton was in a much less advanced stage. Still completely unossified are the scapulocoracoid, ischium and pubis, and the vertebral centra. This corresponds to an early larva of an eryopiform (Sclerocephalus haeuseri or Onchiodon labyrinthicus Geinitz, Reference Geinitz1861). The femur is short and undifferentiated. The neural arches are poorly developed and still paired, and the centra are completely unossified. Vertebral development, starting with paired neural arches and subsequent ossification of centra, is plesiomorphic for temnospondyls and probably for early tetrapods in general (Danto et al., Reference Danto, Witzmann and Fröbisch2016). Because brachyopids have disc- or spool-shaped centra as adults (Warren et al., Reference Warren, Rozefelds and Bull2011), P. wilkinsoni shows that the plesiomorphic pattern of development was conserved also in Mesozoic stereospondyls, irrespective of whether the adult vertebral morphology was rhachitomous (horseshoe-shaped intercentra), stereospondylous (disc-like intercentra), or diplospondylous (disc-like intercentra and pleurocentra). The ribs are moderately short and straight, similar to those of larval dvinosaurians, eryopids, and Micromelerpeton credneri, but much longer than in branchiosaurids. It is not clear whether adult brachyopids developed uncinate processes. The only other described brachyopid with ribs is an adult specimen of Bothriceps australis, in which slender, almost straight ribs without uncinate processes have been described (Warren et al., Reference Warren, Rozefelds and Bull2011). However, these ribs are derived from the posterior trunk region, whereas ribs with uncinate processes are usually present in the “thoracic” region (e.g., Witzmann and Schoch, Reference Witzmann and Schoch2006b). At least the closely related chigutisaurids developed distinct uncinate processes as adults, as described by Warren and Hutchinson (Reference Warren and Hutchinson1983) and Marsicano (Reference Marsicano1993), whereas uncinate processes are unknown in plagiosaurids (Nilsson, Reference Nilsson1937, Reference Nilsson1946; Hellrung, Reference Hellrung2003).

In its overall degree of ossification, P. wilkinsoni resembles early larvae of Permo-Carboniferous eryopiforms such as the eryopid Onchiodon labyrinthicus (at a skull length between 16–20 mm; Witzmann, Reference Witzmann2005), the basal stereospondylomorph Sclerocephalus haeuseri (at a skull length of ~20 mm; Schoch, Reference Schoch2003; Witzmann, Reference Witzmann2006), and the more derived stereospondylomorph Archegosaurus decheni (at a skull length between 30–40 mm; Witzmann, Reference Witzmann2006). However, P. wilkinsoni differs from Paleozoic temnospondyl larvae in a more pronounced dermal ornament and distinct lateral lines, and in the earlier ossification and differentiation of the occiput. This earlier development of the skull might be a stereospondyl characteristic, as indicated by early juveniles of Warrenisuchus aliciae (Warren and Hutchinson, Reference Warren and Hutchinson1988) and Benthosuchus sushkini (Bystrow and Efremov, Reference Bystrow and Efremov1940), which likewise already possess well-ossified exoccipitals and well-developed dermal ornament.

Degree of ossification and implications for ossification sequence.—Based on the degree of ossification of the different skeletal elements in Platycepsion wilkinsoni, it is possible to detect similarities and differences in a small part of the ossification sequence between P. wilkinsoni and Paleozoic temnospondyls. In the dissorophoid Micromelerpeton credneri, the ilium started to ossify prior to the scapulocoracoid, which in turn ossified earlier than the stapes, and which was followed later by ossification of the exoccipitals (Boy, Reference Boy1995; Witzmann and Pfretzschner, Reference Witzmann and Pfretzschner2003). The same sequence was probably present in the eryopid Onchiodon labyrinthicus (although ilium and scapula are already ossified in the smallest available specimen; Witzmann, Reference Witzmann2005) and was documented in the basal stereospondylomorph Sclerocephalus haeuseri (Witzmann, Reference Witzmann2006; Schoch and Witzmann, Reference Schoch and Witzmann2009a). In the more derived and more aquatic stereospondylomorph Archegosaurus decheni, the ossification sequence had partially changed: the scapulocoracoid started to ossify after an ossified stapes had formed. In P. wilkinsoni, the ossification of the scapulocoracoid is not only delayed with respect to the formation of the stapes, but also with respect to the exoccipitals, which ossified prior to the scapulocoracoid. This heterochronic shift might be correlated with the strict aquatic lifestyle of both adult Archegosaurus decheni and brachyopids throughout their life histories, in which the limbs were hardly used for terrestrial locomotion.

Paleobiological implications

The striking contrast between the strongly ossified skull and hyobranchium on the one hand and the feebly ossified and poorly developed postcranium, including the tiny limbs, on the other hand might give some interesting hints concerning the lifestyle of P. wilkinsoni. The branchial arches (which are more strongly developed than the ribs) and the well-differentiated exoccipitals indicate that Platycepsion wilkinsoni relied on strong suction feeding similar to plagiosaurid temnospondyls (Witzmann and Schoch, Reference Witzmann and Schoch2013). Posteroventral rotation of the robust hyobranchial apparatus expanded the buccal cavity and generated powerful suction of prey into the mouth. The morphologically already advanced exoccipital condyles probably facilitated rapid mouth opening by skull elevation, which has been shown to occur in plagiosaurid feeding (Jenkins et al., Reference Jenkins, Shubin, Gatesy and Warren2008). In contrast to most actinopterygians and Latimeria chalumnae Smith, Reference Smith1939, suction feeding was akinetic in P. wilkinsoni because all skull roofing bones were already tightly sutured. This mode of feeding was probably retained in P. wilkinsoni throughout its life history. It suggests that brachyopids had a uniphasic trophic life history, even though a change in prey items from invertebrates and small fishes to larger aquatic vertebrates was likely.

The poor degree of postcranial ossification of P. wilkinsoni might indicate that it was a lurking predator, at least in this growth stage, waiting for bypassing prey. The external gills were probably the main site of gas exchange in the larva, conceivably supplemented by the developing lungs when air breathing was necessary (e.g., depletion of oxygen in the water) and possibly by skin breathing. Ventral grooves for branchial arteries are not preserved on the gill arches in Platycepsion wilkinsoni, whereas adult brachyopids have ventrally grooved gill arches (Warren et al., Reference Warren, Rozefelds and Bull2011), indicating the presence of internal gills (Schoch and Witzmann, Reference Schoch and Witzmann2011). We conclude that internal gills were not yet developed in Platycepsion wilkinsoni, but developed later at the end of the larval stage or thereafter. This is analogous to the situation in extant lepidosirenid lungfishes, which possess larval external and adult internal gills plus lungs (Graham, Reference Graham1997).