Among vertebrates, only stegocephalians have an autopod. Thus, stegocephalians include all terrestrial and most amphibious vertebrates. The main invasion of land by vertebrates occurred in the Carboniferous. To understand this event, a survey of the biodiversity and phylogeny of early stegocephalians is useful.
Temnospondyls form a large group in which more than 150 genera are recognized. They appeared in the Early Carboniferous (about 340 Ma ago) and vanished in the Cretaceous, about 100 Ma ago. They include, among others, Eryops and Mastodonsaurus. Temnospondyls were midsized to large, generally between 30 cm and 3 m in total body length, but at least one species may have reached about 7 m (Steyer and Damiani, 2005). Throughout their history, temnospondyls have colonized various terrestrial and aquatic (salt- and freshwater) habitats. The oldest ones seem to have been amphibious, but in the Permian some species were terrestrial, at least as adults, whereas others were amphibious or aquatic (Fig. 5.1). They had four digits in the hand, and five in the foot.
Figure 5.1. Two Early Permian temnospondyls (280 Ma ago). These temnospondyls are shown in their presumed habitat. Eryops (the largest one) lunges to catch a chondrichthyan (a xenacanthid) while Trimerorhachis (smaller) escapes. The hand of Eryops actually had only four digits. Drawing by Douglas Henderson initially published by Czerkas and Czerkas (1990). Reproduced with permission.
The ribs of temnospondyls are often rather short, and when they are long, as in Eryops, they sometimes overlap. This strengthened the rib cage (as in birds) but prevented costal ventilation of the lungs. This suggests that other structures were responsible for bringing fresh air into the lungs. The large interpterygoid window in the palate of temnospondyls suggests the presence of a buccal pump analogous with that of anurans (Gans, 1970). Muscles may have enabled rhythmic increases in the mouth volume by raising the skin that closed the interpterygoid window, and by depressing the floor of the mouth. After closing the mouth and the nostrils, relaxing these muscles could have forced air into the lungs, through elastic recoil of the buccal skin. (See the box titled “The temnospondyl Iberospondylus schultzei.”)
THE TEMNOSPONDYL IBEROSPONDYLUS SCHULTZEI
This temnospondyl is the oldest known stegocephalian in the Iberian Peninsula. It dates from the Late Carboniferous (Stephanian C, between 302 and 304 Ma ago) and comes from near Puertollano, in the Ciudad Real province, in the center of Spain (Laurin and Soler-Gijón, 2006). It is a midsized Carboniferous temnospondyl (the skull length is 14 cm). It must have been amphibious because traces of the cephalic portion of the lateral line organ remain. Three well-preserved articulated individuals were found in the locality, and such a preservation mode suggests only a short postmortem transport of the carcasses. It seems to have tolerated salt or brackish water because the locality appears to have been coastal, and unmistakable evidence of marine influence has been found (Soler-Gijón and Moratalla, 2001). Iberospondylus is thus the oldest known temnospondyl in which saltwater tolerance is reasonably well established (the most famous and least disputed examples date mostly from the Triassic). The good preservation of the specimens may explain the presence of an otic lamella that occluded the otic notch; this indicates that this animal lacked a tympanum (ear drum). It must not have heard high-frequency airborne sounds well, but it probably heard well enough those of low frequency (less than 1000 Hz) and those transmitted in water or in the ground. Iberospondylus seems to occupy a fairly basal place among temnospondyls (Laurin and Soler-Gijón, 2006).
Skull of the temnospondyl Iberospondylus. A dermal ornamentation typical of temnospondyls is visible in dorsal view (left). The palatal view shows the large interpterygoid vacuity, also typical of this group. The scale is in centimeters. Modified from Laurin and Soler-Gijón (2001).
Temnospondyls experienced two extensive evolutionary radiations. The first one occurred from the Early Carboniferous to at least the end of the Early Permian. Temnospondyl diversity seems to have decreased in the Middle and Late Permian, but the low number of fossils from these times is insufficient to determine if this decrease was gradual, or if it occurred suddenly at the end of the Permian, when the biosphere experienced its greatest extinction event, at least in the oceans and seas. The Permo-Carboniferous temnospondyls have often been called “rhachitomous,” after their vertebral centrum (the part of the vertebra that located below the spinal cord; it surrounds the notochord and partly replaces it). That centrum is composed of a small, paired, dorsal pleurocentrum that articulates with the neural arches, and a small, crescentic intercentrum. Rhachitomous temnsopondyls form a paraphyletic group, since they include some close relatives of geologically more recent temnospondyls. The second evolutionary radiation of temnospondyls started in the Late Permian and gave rise to a great diversity of forms, most of which belong to the taxon Stereospondyli, also named after the morphology of its vertebral centrum. The stereospondyl centrum is composed of a large, relatively cylindrical intercentrum; the pleurocentrum is either absent, unossified, or very small. Starting with the Triassic, most temnospondyls were aquatic (Fig. 5.2) and fairly large (1.5 m to 3 m in total body length) and belong to the Stereospondyli.
Among all early stegocephalians, the ontogeny is best known in temnospondyls, because they are the most abundant Paleozoic stegocephalians in the fossil record. Aquatic larvae with external gills are known for a few temnospondyl species, and they were probably present in most species (Fig. 5.3). These larvae superficially resemble those of urodeles (salamanders and newts), but this does not imply close affinities, since seymouriamorphs also possessed such larvae; this developmental mode is probably primitive for stegocephalians.
Temnospondyls played a prominent role in theories about the origin of extant tetrapods because they were long considered to represent the stem group of anurans (Fig. 5.4A) or of all lissamphibians (Fig. 5.4B), and this remains a popular (but controversial) theory today. This may explain why the first reconstruction of the temnospondyl Mastodonsaurus looked like a huge frog, with a stubby body (we now know that it has a long, slender body with proportions more reminiscent of urodeles). These ideas hark back to the works of E. D. Cope, a pioneer of North American paleontology. They prevailed from the 1880s to the end of the 1990s. However, several recent studies that used computer-assisted phylogenetic analyses raise serious doubt about these theories (Laurin, 1998b; Vallin and Laurin, 2004; Marjanović and Laurin, 2008, 2009); temnospondyls seem to be stem tetrapods (Fig. 5.4C), and as such, are no closer to extant amphibians than to amniotes. This last hypothesis is accepted in this book, although alternatives are discussed. Thus, temnospondyls are not considered amphibians.
Figure 5.2. Metoposaurus, a large aquatic stereospondyl from the Triassic (210 Ma). This temnospondyl belongs to the Stereospondyli, which diversified intensively in the Triassic. Drawing by Douglas Henderson initially published in Long and Houk (1988). Reproduced with permission.
Figure 5.3. Larvae of the temnospondyl Tupilakosaurus wetlugensis. Larvae with external gills are known in a few temnospondyl species. Photo taken by the author in the Paleontological Institute (Moscow).
Figure 5.4. Phylogenetic hypotheses about the position of temnospondyls. The systematic position of temnospondyls is highly contentious. They have been considered to represent all or part of the stem group of anurans (A) or of all lissamphibians (B), or to be stem tetrapods (C). Temnospondyls are monophyletic only under this latter interpretation (C). Crosses designate extinct taxa. Extant taxa are emphasized in bold type. The name “Lissamphibia” is generally used only if the smallest clade that includes all extant amphibians excludes all currently known Paleozoic stegocephalians, such as temnospondyls (B, C). Amphibians are shown in dark gray, and reptiliomorphs, in light gray.
Embolomeres include only 16 genera and extend only from the Carboniferous (around 340 Ma ago) to the Triassic (around 220 Ma ago). Their geographic range comprised what is currently North America and European Russia, regions that were then part of Euramerica. Most measured between 1.5 m and 2.5 m in total length and were aquatic to amphibious, as shown by the frequent presence of grooves for the cephalic portion of the lateral-line organ and their long, deep tail, which was probably used for swimming. However, in the Triassic, some embolomeres found in Russia (the chroniosuchians) included terrestrial species (Fig. 5.5). The skull retains the intertemporal, a bone that had disappeared in most other stegocephalians by the Early Permian. The massive stapes suggests that the otic notch did not support a tympanum. These animals must have had fairly poor hearing on dry land. Palatal fangs suggest that they fed on relatively large prey. The relatively long ribs suggest that they had a good costal lung ventilation capability. Like those of many early stegocephalians, both hand and foot retained five digits.
They are characterized by a peculiar type of vertebral centrum composed of two cylinders (Fig. 5.6B). The primitive configuration for stegocephalians, as seen in temnospondyls, Ichthyostega, and other a few other Devonian sarcopterygians, consists of a rhachitomous centrum (Fig. 5.6A).
Since their discovery in the 19th century and until the 1990s, embolomeres have been considered to be among the basalmost reptiliomorphs (Fig. 5.7A). The character that initially justified this hypothesis in the 1880s (the presence of a single, median occipital condyle) has long been rejected because we now know that this character is primitive for stegocephalians, but other more convincing characters were proposed. For instance, two bones of the dermal skull roof (parietal and tabular) that are separated from each other in most other stegocephalians share a suture in embolomeres, amniotes, and other presumed reptiliomorphs (Fig. 5.8). Another potential synapomorphy of these taxa is the long posterior stem of the interclavicle (the stem is absent in temnospondyls and more basal taxa). However, most recent computer-assisted phylogenetic analyses (e.g., Vallin and Laurin, 2004; Marjanovié and Laurin, 2009) suggest that embolomeres are stem tetrapods (Fig. 5.7B). The characters that have been interpreted as synapomorphies of embolomeres and other presumed reptiliomorphs probably appeared in stem tetrapods. (See the box titled “The embolomere Proterogyrinus scheelei.”)
Figure 5.5. Chroniosuchian. This embolomere from the Late Permian and the Triassic lived in what is now Russia and was probably terrestrial. Large bony scutes on its back articulated with the neural spines and strengthened its axial skeleton, presumably to improve its ability to support the body weight outside the aquatic environment. Photo taken by the author in the Paleontological Institute (Moscow).
Figure 5.6. Stegocephalian vertebrae in right lateral view. Rhachitomous temnospondyl vertebra (A) showing the primitive morphology for stegocephalians (a ventral, crescentic intercentrum and a paired, dorsal pleurocentrum) and embolomerous vertebra of Eogyrinus, showing a centrum composed of two cylindrical elements (B). The intercentrum is in light gray, and the pleurocentrum, darker. The neural arch, in white, surrounds the spinal cord and is not fused to the vertebral centrum, contrary to most extant tetrapods. Cranial is to the right.
Figure 5.7. Phylogenetic position of embolomeres. Only the main phylogenetic hypotheses are shown. In A, Embolomeres as reptiliomorphs. This hypothesis prevailed from the 1880s to the 1990s. In B, Embolomeres have been placed among stem tetrapods by most computer-assisted analyses from the 1990s onward.
Figure 5.8. Stegocephalian skulls. These skulls show the primitive condition for the relative position of the parietal (P) and the tabular (T), in which these bones are widely separated from each other by intervening bones, as in Ichthyostega (A), and the condition, often considered characteristic of reptiliomorphs (but more likely a synapomorphy of a larger clade), of the presence of a contact between both bones, as seen in embolomeres, seymouriamorphs (B) and amniotes. Parietal and tabular in dark gray; postparietal and supratemporal in light gray. Scale: 1 cm.
Seymouriamorphs have so far been found only in Permian localities, which is surprising given that all plausible phylogenies suggest that this group must have appeared in the Carboniferous (Fig. 5.9). This great gap in their fossil record may result from an erroneous dating of some localities where they occur. Indeed, seymouriamorph-bearing localities in Kazakhstan, Tadjikistan, and Xinjiang (China), which were all on the Kazakhstan plate in the Carboniferous, have been dated as Permian, but this rests largely on the seymouriamorphs. Phylogenetic analyses suggest that the seymouriamorphs found on this continent were among the oldest, most basal ones. Furthermore, this plate collided with the Euramerican plate (which included European Russia and North America, among other regions) early in the Permian. If seymouriamorphs had first appeared on the Kazakhastan plate in the Carboniferous, the collision between these plates could explain the sudden appearance of seymouriamorphs in Euramerica in the Early Permian.
Figure 5.9. Geographical distribution of seymouriamorphs. This map shows the position of continental plates in the Middle Permian (265 Ma ago) and the areas that have yielded seymouriamorph fossils (large black dots). The Kazakhstan continent is framed into a black polygon. Modified from Gradstein et al. (2004).
THE EMBOLOMERE PROTEROGYRINUS SCHEELEI
This embolomere, among the oldest ones, was found in West Virginia and dates from the Early Carboniferous (Serpukhovian, about 318 to 326 Ma ago). With a skull length of about 12 cm, it was smaller than most later embolomeres, whose skull usually measured between 15 and 35 cm. Grooves for the lateral-line organ on the skull, the weak ossification of the appendicular skeleton, the absence of neurocentral fusion (between neural arch and vertebral centrum), and the long, deep tail, in which the haemal and neural arches may have supported a fin, suggest a mostly aquatic lifestyle. The otic notch may have housed a spiracle, the first gill slit, which oxygenates blood going to the head. The intercentrum of Proterogyrinus retained the primitive crescentic shape that was lost in later embolomeres, in which this bone became cylindrical. Proterogyrinus is probably a fairly basal embolomere (Holmes, 1984).
Skull in lateral (A), dorsal (B), and palatal (C) view, and skeleton (D). In dorsal view (B), note the intertemporal (It) and the contact between tabular (t) and parietal (p). The palate (C) shows a very small interpterygoid vacuity (Fi), contrary to temnospondyls. The palatal (C) and lateral (A) views also shows the large palatal fangs. A groove for the lateral-line organ is visible behind and below the orbit (A, B). Reproduced from Holmes (1984, figs. 1, 3) with permission from the Royal Society of London.
Figure 5.10. Skull of Seymouria baylorensis. Cranial reconstruction of Seymouria baylorensis, the first seymouriamorph ever described (Broili, 1904).
Figure 5.11. Skeleton of Seymouria. Skeleton of Seymouria sanjuanensis, from the Early Permian of Germany. Photo provided by D. Berman. Reproduced with permission of the Carnegie Museum of Natural History and the Museum der Natur, Gotha, Germany.
Figure 5.12. Seymouriamorph vertebra. The pleurocentrum (dark gray) of the vertebral centrum is firmly fused to the neural arch (white). The intercentrum is small and crescentic (light gray).
Fewer than 15 genera are recognized (Laurin, 2000; Bulanov, 2003), and most seem to have been terrestrial as adults (Fig. 5.10). This is suggested by the high degree of ossification of their endoskeleton (Fig. 5.11), by the neurocentral fusion (Fig. 5.12), and by the presence of seymouriamorphs in some fossiliferous localities, such as Fort Sill in Oklahoma, in which only fairly terrestrial taxa are encountered. Furthermore, seymouriamorphs may have been among the first stegocephalians with a tympanic middle ear (adapted to hear high-frequency airborne sounds, such as most animal vocalizations). The gracile stapes (the main ear ossicle) was capable of transmitting high-frequency sounds. The otic notch, at the back of the skull (Fig. 5.10), may have supported the tympanum.
They were initially considered amniotes (Broili, 1904), because of their presumed terrestrial lifestyle and because of presumed synapomorphies with amniotes (see “Embolomeres” earlier in this chapter), but the discovery of aquatic larvae with external gills in the 1940s (see the box titled “The seymouriamorph Discosauriscus austriacus”) showed that this hypothesis was false (amniotes lack larvae). Nevertheless, most paleontologists thought that seymouriamorphs were reptiliomorphs at least until the 1990s (Fig. 5.13A). As for embolomeres and temnospondyls, most recent phylogenetic analyses suggest that seymouriamorphs are stem tetrapods (Fig. 5.13B).
Figure 5.13. Systematic position of seymouriamorphs. Only the main phylogenetic hypotheses about seymouriamorphs are shown. A, Among reptiliomorphs. This hypothesis prevailed from their discovery till the 1990s. B, Among stem tetrapods. This hypothesis is supported by most computer-assisted analyses published from the late 1990s onwards.
THE SEYMOURIAMORPH DISCOSAURISCUS AUSTRIACUS
This seymouriamorph found in the Czech Republich dates from the Early Permian (Sakmarian, about 285 to 295 Ma ago). It is represented by growth series documenting its ontogeny, from aquatic larvae with external gills with a weakly ossified endoskeleton to postmetamorphic individuals that had lost their external gills. The good ossification of the latter suggests a fairly terrestrial lifestyle. It was long thought that even the largest known individuals were relatively young (less than two years old), but recent skeletochronological study showed that these specimens were sexually mature and about ten years old (Sanchez et al., 2008). For a long time, it was considered a temnospondyl, because it had been confused with branchiosaurs, a group of temnospondyls also represented by growth series including many larvae. The growth series of Discosauriscus show that the relative size of the eye decreases throughout ontogeny, as in most vertebrates. Discosauriscus is one of the few known stegocephalians that seems to have possessed electrosensory organs (Klembara, 1994). The gracile stapes of the closely related taxon Seymouria suggests that the otic notch of Discosauriscus supported a tympanum enabling it to hear high-frequency airborne sounds.
The seymouriamorph Discosauriscus austriacus. This skull (A, specimen K 13) represents a late larval or metamorphic stage. To the right, the sclerotic ring (that protects the eye) is visible in the orbit. Grooves for the lateral-line organ are visible between the nares and orbits; the temporal ramus of this organ extends to the upper edge of the otic notch. The skeleton (B) is from a young postmetamorphic individual. Reproduced with permission (from the Royal Society of London [A] and from the Royal Society of Edinburgh [B]) from Klembara (1997, fig. 2) and Klembara and Bartík (2000, fig. 1).
Amphibians include the lissamphibians (anurans, urodeles, and gymnophionans) and extinct taxa that are more closely related to lissamphibians than to amniotes (see “Are Animals Still Conquering the Land Today?” in Chapter Two). According to this definition, the stegocephalian groups described above are not amphibians, even though they have been (and often still are) considered as such by many authors. Genuine amphibians appeared in the Early Carboniferous (about 340 Ma ago); Paleozoic amphibians include all or most taxa that have collectively been called “lepospondyls” (Fig. 5.14), but since this name probably designates a paraphyletic group, it will not be used below. The term “amphibian” is preferable.
Paleozoic amphibians were generally small (from 10 cm to 30 cm in total body length). Several synapomorphies suggest that these taxa are closely related to each other, with the possible exception of adelogyrinids (Ruta and Coates, 2007). For instance, amphibians have lost the complex labyrinthine infolding of the dentine that characterized the teeth of the first stegocephalians (Fig. 5.15) and that gave them the name “labyrinthodonts” (this designates a paraphyletic group, so this name is not used here). The neural arches fused firmly to the vertebral centrum early in ontogeny (Fig. 5.16), contrary to other stegocephalians, whose neural arches fused to the centrum only in adults, if at all.
Figure 5.14. Amphibian phylogeny. The “lepospondyls,” only known from the Paleozoic, have nearly always been considered amphibians, but their relationships to extant amphibians and temnospondyls (often considered amphibians) is fairly controversial. The hypothesis presented here is from Laurin (1998b).
Paleozoic amphibians include several taxa (Fig. 5.17), such as the limbless adelogyrinids and aïstopods; the lysorophians, with diminutive limbs, and a long, slender body that could reach a length of 1 m; and nectrideans, characterized by neural and haemal arches with broadened, crenulated distal ends (Fig. 5.16). Many amphibians are usually classified among “microsaurs,” but this is a paraphyletic group that includes Paleozoic amphibians that do not fit into the other taxa mentioned above (adelogyrinids, aïstopods, etc.). The name “Microsauria” will not be used further in this book.
Figure 5.15. Labyrinthine infolding of the tooth dentine. This type of tooth characterizes most early stegocephalians (temnospondyls, embolomeres, seymouriamorphs, etc.). Amphibians lack such dentine infolding. Reproduced from Owen (1860).
Figure 5.16. Nectridean vertebra. The shape of the neural (white) and haemal (light gray) arch is unique to nectrideans, but the fusion between the neural arch and the vertebral centrum (dark gray) is a synapomorphy of nearly all amphibians.
Figure 5.17. Paleozoic amphibians. A lysorophian (with a slender body and tiny limbs) is surrounded by several individuals of the nectridean Diplocaulus of various growth stages. The adults of the latter have a very broad head shaped like a triangle. Drawing by Douglas Henderson initially published by Czerkas and Czerkas (1990). Reproduced with permission.
We know almost nothing about the ontogeny of Paleozoic amphibians because their larvae have almost never been fossilized. It has been suggested that Microbrachis (see the box titled “The amphibian Microbrachis pelikani”) had external gills (Carroll and Gaskill, 1978), but more recent study of this taxon has failed to support this claim (Vallin and Laurin, 2004; Milner, 2008).
Lissamphibians appear only in the Early Triassic, with the stem anurans (salientians) Triadobatrachus from Madagascar and Czatkobatrachus from Poland. Triadobatrachus retains a short tail, and the bones of its zeugopod (radius and ulna in the arm; tibia and fibula in the leg) have not fused into a radioulna and a tibiofibula, contrary to extant anurans. Urodeles and gymnophionans appear only in the Jurassic, and the oldest representatives of these taxa belong to their stem group, as shown by the retention of several primitive characters. For instance, Eocaecilia, the oldest known gymnophionan, retains diminutive limbs, which are lost in all crown gymnophionans.
THE AMPHIBIAN MICROBRACHIS PELIKANI
This amphibian was found in the famous locality of Nyrany, in the Czech Republic, and dates from the Late Carboniferous (Moscovian, from 306 to 312 Ma). It is represented by several individuals that represent a growth series, and it was the first taxon previously considered a “microsaur” to be described in detail (in 1883). This amphibian was perhaps neotenic, because it retained lateral-line grooves on its skull into adulthood, its limbs were small, and its endoskeleton was poorly ossified, although these characters may also reflect its aquatic lifestyle. These phenomena are difficult to tease apart because in extant amphibians, neoteny is often associated with an aquatic lifestyle. In any case, Microbrachis was probably among the most strictly aquatic, and the most neotenic, of all known Paleozoic amphibians. A few other genera share some or all of these features (lysorophians, adelogyrinids, and some nectrideans also appear to have been aquatic, neotenic forms), but most Permo-Carboniferous amphibians were probably amphibious to terrestrial. Microbrachis is the type genus of the monotypic (and redundant) family Microbrachidae (it includes only this genus), and of a larger amphibian taxon called “Microbrachomorpha” (Carroll and Gaskill, 1978), but this group is probably paraphyletic (Anderson, 2001; Vallin and Laurin, 2004). As in the vast majority of other Paleozoic amphibians, its skull lacked an otic notch and its stapes was massive, indicating that the tympanum was absent, as in urodeles and gymnophionans.
The amphibian Microbrachis pelikani. Reconstruction of the skull in dorsal (A), palatal (B), occipital (C), and lateral (D) views. The parietal (p) contacts the postorbital (po); this character is claimed to unite “microbrachomorphs.” The interpterygoid fenestra (fi), visible in palatal view (B), is relatively narrow. The stapes (s) is fairly massive. This animal, like all Paleozoic tetrapods, lacks palatal fangs. Modified from Vallin and Laurin (2004).
As we saw above, some taxa previously interpreted as reptiliomorphs, such as embolomeres and seymouriamorphs, are probably stem tetrapods. On the contrary, diadectomorphs are genuine reptiliomorphs, as studies form the early 20th century already suggested. They were so similar to amniotes (Fig. 5.18) that some authors have even considered them amniotes (Berman et al., 1992) and have suggested that they laid amniotic eggs (Lee and Spencer, 1997). However, both suggestions represented a minority opinion, and the type of egg laid by diadectomorphs is conjectural (Laurin and Reisz, 1999). Diadectomorphs were initially part of the taxon Cotylosauria, named after the shape of their occipital condyle (the bony structure to which the skull articulates with the vertebral column), but later, the meaning of that name was misinterpreted and cotylosaurs were considered the ancestral group of amniotes. In fact, diadectomorphs are the closest known relatives of amniotes, but they are not their ancestors. This close relationship is supported by the presence of two sacral vertebrae (whose ribs articulate with the hip), instead of a single one, as in most other stegocephalians. This character may reflect the presumably fairly terrestrial lifestyle of diadectomorphs and amniotes.
Figure 5.18. Diadectomorphs, amniotes, and seymouriamorphs. The carnivorous amniote Dimetrodon tries to catch a diadectid in the water while two other diadectids and a Seymouria (to the right) rest on a log. Drawing by Douglas Henderson initially published by Czerkas and Czerkas (1990). Reproduced with permission.
Figure 5.19. Diadectomorph skeleton. Skeleton of the small diadectid Orobates as preserved. It was found in Germany and dates from the Early Permian (Berman et al., 2004). Picture provided by D. Berman. Reproduced with permission of the Carnegie Museum of Natural History and the Museum der Natur, Gotha, Germany.
Diadectomorphs were relatively large (1.5 m to 2 m in total body length). Only eight genera are currently recognized, and they date from the Late Carboniferous and the Early Permian. At least some of them (limnoscelids) were amphibious to aquatic and perhaps piscivorous, (see the box titled “The diadectomorph Limnoscelis paludis”) whereas others (diadectids and Tseajaids) were probably more terrestrial (Fig. 5.19). Diadectids were herbivorous, which is exceptional for Paleozoic stegocephalians and is the only case known outside Amniota. Diadectids had very broad molars that seem well suited to chew plants, and their stubby body also suggests a herbivorous diet. No diadectomorph seems to have had a membranous tympanum, but diadectids had a bony plate linked with the ear that may have been used to transmit sounds (either low-frequency airborne sounds or sounds transmitted through water). Their neural arches were strongly convex, as in some early amniotes. The functional significance of this character remains enigmatic.
THE DIADECTOMORPH LIMNOSCELIS PALUDIS
Fossils of this species were found in various sites in the southwestern United States; they date from the Early Permian (Asselian, 295 to 299 Ma). The long, slender body and the weak ossification of its body suggest an amphibious to aquatic lifestyle, as the specific epithet implies. Its sharp fangs suggest that it fed on relatively large prey. This species played a central role in hypotheses on the origin of amniotes. Romer (1957) considered, like many contemporary paleontologists, that Limnoscelis was perhaps the ancestor of amniotes, or at least fairly similar to it. Furthermore, given its relatively old geological age, he suggested that its relatively aquatic lifestyle was more likely to be primitive than to represent a return to the aquatic environment. Thus, he suggested that the first amniotes were primitively aquatic and that they ventured onto land mostly to lay their eggs. The amniotic egg, which is adapted to being laid on dry ground, would then have preceded the appearance of terrestrial adults (at least among amniotes). On land, these eggs were sheltered from the many aquatic predators, whereas relatively few large predators roamed the land. This hypothesis is no longer largely accepted today, because the oldest known amniotes seem to have been terrestrial, and there is no evidence that the other diadectomorphs (such as Diadectes and Tseajaia) were aquatic.
The diadectomorph Limnoscelis paludis. Cranial reconstruction in dorsal (A), palatal (B), occipital (C), and lateral (D) views. Large fangs are present, but contrary to those of temnospondyls and embolomeres, they do not grow out of the palate; instead, they are located on the premaxilla and maxilla, which also support the upper jaw teeth.
The oldest known amniotes date from the middle of the Late Carboniferous (315 Ma). They were found near Joggins, in Nova Scotia (Canada), and were preserved in fossilized giant lycopsid tree stumps. Various older fossils (from the Early Carboniferous or the early Late Carboniferous), generally poorly preserved or fragmentary, had been interpreted as amniotes (Baird and Carroll, 1967; Smithson and Rohlfe, 1990), but serious doubts have been raised about these interpretations (Laurin and Reisz, 1992, 1999), or their proponents have themselves provided new interpretations (Smithson et al., 1994). The oldest known fossils are of small body size, but this may result from their preservation in the giant tree stumps that may have acted as passive traps. Given their diameter (60 cm or less), they could contain only fairly small vertebrates.
When they first appear, amniotes are represented by their two main clades, sauropsids (reptiles and related extinct taxa) and synapsids (mammals and their extinct relatives). Synapsids dominated the terrestrial vertebrate fauna throughout the remainder of the Paleozoic and until the great end-Permian crisis (250 Ma ago) eliminated most of them. Sauropsids then went on to dominate for the entire Mesozoic, until the Cretaceous/Paleogene crisis (65 Ma ago) eliminated many large reptiles, including most (but not all) dinosaurs (birds are dinosaurs, and several lineages existed before the end of the Cretaceous). Since then, synapsids (represented exclusively by mammals) have again dominated the terrestrial vertebrate fauna. The oldest synapsids (those from the Carboniferous and the Early Permian) have long been called “pelycosaurs,” but this group is paraphyletic (therapsids are nested within it), and this name will not be used further in this book. (See the box titled “The amniote Haptodus garnettensis.”) Synapsids included the largest terrestrial (or amphibious) herbivores and carnivores in the Paleozoic. One of the most familiar carnivores is the strange sailback Dimetrodon (Fig. 5.18). Herbivorous synapsids include caseids, which reached a fairly large size (Cotylorhynchus reached nearly 3 m in total length), and edaphosaurids, the oldest known herbivorous amniotes, such as Edaphosaurus (Modesto and Reisz, 1992).
The oldest known fossil amniotic egg dates from the Triassic, but the amniotic egg is produced in mammals (it is laid only in monotremes) and in reptiles; therefore, parsimony suggests that it must have existed in their last common ancestor (by definition, that was the first amniote). The long gap in the fossil record of amniotic eggs was initially intriguing, because in the Mesozoic, fossilized amniotic eggs are relatively abundant. This gap probably results from the absence of mineralization of the external membrane of the egg of early amniotes (Laurin, Reisz, and Girondot, 2000). Such eggs still exist in monotremes, the platypus and the echidna. A flexible, weakly mineralized outer egg membrane is a synapomorphy of reptiles, but such an egg has a very low fossilization potential. The strongly mineralized egg shell that is so familiar to us (because we eat so many hen eggs) is an archosaur synapomorphy, and it also appeared at least three times within turtles and within squamates. The oldest taxon with a strongly mineralized egg shell (Archosauria) appeared in the Triassic (their oldest fossils date from the Middle Triassic), which explains why no amniotic egg has been found in Permian or Carboniferous localities, despite the relative abundance of skeletal remains of amniotes in these same deposits.
Figure 5.20. Diadectomorph and amniote skulls. The skulls of the diadectomorph Limnoscelis (A) and of the amniote Captorhinus (B) show the relationship between the frontal bone (gray) and the orbit (lighter gray) in both groups. The position of the prefrontal and postfrontal, between the frontal and the orbit, as seen in diadectomorphs (A), is primitive and very widespread in stegocephalians, whereas the presence of a frontal contribution to the orbit, as seen in most amniotes (B), is derived and rare.
ABBREVIATIONS: F, frontal. Pf, postfrontal. Prf, prefrontal.
Since the amniotic egg has left no Paleozoic fossils, and since it is difficult to identify the species that have laid the fossilized eggs we have, paleontologists use other characters to identify amniotes in the fossil record. For instance, the frontal bone borders the orbit dorsally in amniotes (Fig. 5.20B), whereas in most other stegocephalians, two other bones (pre- and postfrontal) occupy this position (Fig. 5.20A).
Two Long-Established Phylogenies
The first phylogenies of early stegocephalians were produced in the 1880s (see “Temnospondyls” and “Embolomeres” earlier in this chapter), well before the advent of cladistics and phylogenetic analysis software. It is thus not surprising that they were overturned by some (but not all) recent analyses that benefited from the tremendous technological advances of phylogenetics in the last decades. Nevertheless, they deeply influenced generations of paleontologists and are still widely accepted, in a slightly modified form, by several practicing systematists, and as such, they deserve to be presented here.
THE AMNIOTE HAPTODUS GARNETTENSIS
This species comes from near Garnett, Kansas, and dates from the Late Carboniferous (Kasimovian, between 304 and 307 Ma ago). It is one of the oldest known amniotes represented by well-preserved, partly articulated skeletons. Like most early amniotes, it seems to have been fairly terrestrial, and was probably carnivorous, as suggested by its dentition. It is closely related to the sphenacodontids and to therapsids (that group that includes most Middle-Permian and more recent synapsids, such as dicynodonts and mammals). Contrary to at least some therapsids, the back of its mandible was not modified to facilitate hearing airborne sounds. Its massive stapes was still part of the jaw suspension (it braced the upper jaw against the braincase). The transversal row of large teeth on the transverse flange of the pterygoid (Pt, visible in palatal and lateral views) characterizes most early amniotes. The large tooth in the anterior portion of the maxilla (M) is a caniniform tooth, probably homologous with the canine tooth found in mammals. However, contrary to the canine, two caniniform teeth were sometimes functional on the same maxilla, whereas a single canine is functional at any given time. Furthermore, the caniniform tooth was replaced throughout life, like all teeth, in Haptodus. Haptodus thus had a polyphyodont dentition, like most early vertebrates, whereas mammals are diphyodont (they possess two successive dentions, the milk dentition in children and the adult dentition). Diphyodonty must have appeared in synapsids in the Triassic or early in the Jurassic, and this was perhaps linked with the complex molars of mammals, whose occlusal surfaces must interlock closely to ensure efficient chewing. The high metabolism of mammals requires a fast digestion of food (especially of plant matter, which is not very nutritious), and the complex molars allow mammals to break up food into tiny particles, thus increasing the surface-to-volume ratio and speeding up digestion. This complex interlocking would presumably have been difficult to achieve in continuously replacing teeth. Like all Permo-Carboniferous tetrapods, Haptodus must have retained a slow metabolism and an ectothermic body temperature regulation system (which relies on behavior and the ambient climate to maintain body temperature near its optimal value). Its skin was probably devoid of fur and it probably laid eggs.
The amniote Haptodus garnettensis. Reconstruction of the skull in dorsal (A), palatal (B), and lateral (C) views, and of the mandible in lateral (D) and medial (E) views. Reproduced from Laurin (1993).
Most phylogenies published before 1997 implied that all known stegocephalians were either amphibians or reptiliomorphs (Fig. 5.21); in most cases, no limbed vertebrate was considered a stem tetrapod, at least until the discovery of Devonian stegocephalians. Thus, embolomeres, seymouriamorphs, diadectomorphs, and amniotes were considered to be reptiliomorphs, whereas “lepospondyls,” temnospondyls, and lissamphibians were considered amphibians. The “lepospondyls” were sometimes considered polyphyletic because some were classified among reptiliomorphs by a few authors (“lepospondyls” are paraphyletic in Figure 5.21). The position of the Devonian stegocephalians such as Acanthostega and Ichthyostega was more controversial. Most previous studies considered them amphibians, but at least since the 1980s, some authors considered them stem tetrapods. This phylogeny suggested that the invasion of land by vertebrates occurred in parallel in amphibians and in reptiliomorphs (Fig. 5.21).
These phylogenies were initially produced by E. D. Cope, who suggested, as early as 1880, that embolomeres were reptiliomorphs and, in 1882, that temnospondyls were amphibians. These hypotheses were initially supported by a few characters. For instance, reptiliomorphs were thought to share the presence of a single occipital condyle, and amphibians displayed a vertebral centrum formed (or dominated) by an intercentrum (this is clearly correct only among temnospondyls). This phylogeny was nevertheless accepted by the vast majority of paleontologists and remains popular today. When the characters used by Cope to justify his hypotheses were refuted (because they are primitive, like the single occipital condyle, or based on an error of interpretation, like the composition of the amphibian vertebral centrum, which is a pleurocentrum in lissamphibians and “lepospondyls”), other characters were proposed, but strangely, the phylogeny was not seriously questioned. This can perhaps be attributed to the weight of tradition, which plays an important, if seldom acknowledged, role in science (Laurin, 2008a). The weight of tradition may seem inappropriate in science, but it is not always harmful, because it prevents the scientific community from prematurely rejecting good established theories when new alternatives are proposed. Recent stegocephalian phylogenies produced without computers retain a topology compatible with Cope’s suggestions (Panchen and Smithson, 1988).
Figure 5.21. Classical stegocephalian phylogeny (simplified). The position of the Devonian stegocephalians, such as Ichthyostega and Acanthostega, was controversial for most of the 20th century. They were generally considered to be amphibians, and less often, stem tetrapods. The horizontal gray lines indicate the independent acquisitions of an amphibious or terrestrial lifestyle. In some cases (nectrideans, “microsaurs,” temnospondyls, embolomeres), only some species of these groups are amphibious or terrestrial.
An alternative hypothesis, less widely accepted, implies that extant amphibians are polyphyletic, at least compared with Paleozoic stegocephalians. According to its proponents, anurans are nested within temnospondyls, whereas gymnophionans are nested within “lepospondyls.” The position of urodeles has been more variable; an origin among “lepospondyls” was favored by early studies (Carroll and Currie, 1975; Carroll and Holmes, 1980), but an origin among temnospondyls (not the same as for anurans) is favored by more recent studies (Schoch and Carroll, 2003; Carroll, 2007; Fröbisch et al., 2007; Anderson, 2008; Anderson et al., 2008). These phylogenies are incompatible with most recently published phylogenies produced by computer-assisted analyses (Laurin, 2002). Furthermore, the developmental data that have provided much of the support for these hypotheses actually do not support them (Schoch, 2006; Germain and Laurin, 2009). Finally, two of the few data matrices that supported lissamphibian polyphyly actually support a monophyletic origin of extant amphibians among “lepospondyls” when recoded to better reflect the descriptive literature (Marjanović and Laurin, 2008, 2009). Globally, there is very little objective support for extant amphibian polyphyly.
A Recent Alternative
The first computer-assisted phylogenetic analyses of Paleozoic stegocephalians (Gauthier et al., 1988; Trueb and Cloutier, 1991) seemed to support Cope’s venerable ideas, but this resulted from a rather restricted choice of taxa. One of these studies included only temnospondyls and lissamphibians, and the other included only embolomeres, seymouriamorphs, diadectomorphs, and amniotes. Given these taxonomic samplings, it was impossible to test which Paleozoic taxa were amphibians and which ones were reptiliomorphs (because to test these hypotheses, lissamphibians and amniotes had to be present in the same analysis). The first analysis that included most relevant higher taxa (Laurin and Reisz, 1997) suggested that lissamphibians derive from “lepospondyls.” Temnospondyls are stem tetrapods, rather than amphibians (Fig. 5.22). According to the same phylogeny, embolomeres and seymouriamorphs are not reptiliomorphs; instead, they are stem tetrapods. All Devonian stegocephalians are also stem tetrapods. This phylogeny requires reevaluating the evolution of most characters in early stegocephalians. For instance, it suggests that the first terrestrial stegocephalians were stem tetrapods (Fig. 5.22).
Figure 5.22. Recent stegocephalian phylogeny (simplified). Reptiliomorphs and amphibians include far fewer taxa than in older phylogenies because many of their presumed members (temnospondyls, embolomeres, and seymouriamorphs) have been transferred to the tetrapod stem. The horizontal gray lines indicate the independent presumed appearances of amphibious or terrestrial lifestyles. In some cases (temnospondyls, embolomeres), only some species of these taxa have become terrestrial, and some species reverted to an aquatic lifestyle (adelogyrinids, and some species among nectrideans, “microsaurs,” amniotes, and lissamphibiens).
SYNAPOMORPHIES OF THE CLADES: 1, temnospondyls and tetrapods; pentadactyl limb. 2, embolomeres and tetrapods; presence of a contact between parietal and tabular. 3, seymouriamorphs and tetrapods; fusion of the vertebral centrum to the neural arch in the vertebra. 4, tetrapods; loss of the intertemporal in the dermal skull roof. 5, cotylosaurs; presence of two sacral vertebrae. 6, amphibians; atlantal neural arch fused in the sagittal plane (it is paired in most other taxa). 7, lysorophians and lissamphibians; loss of two of the three coronoids that were present in the mandible of the first stegocephalians and loss of the jugal bone.
Such a radical change in stegocephalian phylogeny could be accepted neither rapidly nor without much resistance. A few studies produced large data matrices that supported a monophyletic origin of lissamphibians among the temnospondyls (Ruta et al., 2003; Ruta and Coates, 2007). However, three theses, including one in progress, have reworked this matrix, and all conclude that Lissamphibia is monophyletic but nested within “lepospondyls” (Marjanović and Laurin, 2009). Furthermore, “lepospondyls” are more closely related to amniotes than to temnospondyls in all phylogenies supported by data matrices published after 1996, whereas old phylogenies placed both temnospondyls and “lepospondyls” in an amphibian clade that excludes amniotes. Thus, the old phylogeny is rejected by practically all recent relevant studies, but the debate about lissamphibians continues and will probably not be settled for five or ten more years. This long period of debate is necessary because much time is required to compare the various data matrices, find the characters or taxa responsible for the incompatibilities, and determine the correct coding. This work requires exhaustive literature searches, an examination of many fossils, and new descriptions and analyses, but this painstaking research is the price to pay for progress in paleontology.
Ontogeny is best known only in a few temnospondyls and seymouriamorphs, among Paleozoic stegocephalians. At least a few species of both groups had larvae with external gills. We do not know if embolomeres, diadectomorphs, and Paleozoic amphibians had such larvae.
Our ideas about phylogenetic relationships among the main stegocephalian taxa have changed drastically following the first computer-assisted phylogenetic analyses that included all the main clades. From the 1880s till the mid-1990s, we thought that most stegocephalians were either amphibians (temnospondyls and “lepospondyls”) or reptiliomorphs (embolomeres, seymouriamorphs, diadectomorphs, and amniotes). Most recent analyses suggest that temnospondyls, embolomeres, and seymouriamorphs are stem tetrapods. Only “lepospondyls” are amphibians, and reptiliomorphs include mostly diadectomorphs and amniotes.
