PALEOZOIC LEPOSPONDYL AMPHIBIANS

AMER. ZOOLOGIST, 5:287-294 (1965). PALEOZOIC LEPOSPONDYL AMPHIBIANS DONALD BAIRD Dept. of Geology, Princeton University, Princeton, New Jersey SYNOPSIS. The lepospondyls are urodele-like amphibians of the Paleozoic Era, some o which have been considered ancestral or closely related to reptiles or modern amphibians. Their origin is enigmatic, and the earliest occurrences reveal a long-established division into three contrasting orders: the serpentine aistopods, the newt-like nectrideans and the deceptively reptilian microsaurs. The Subclasses Lepospondyli and Labyrinthodontia are not so disparate as to indicate a polyphyletic origin for the Amphibia. Which subclass gave rise to the present-day Lissamphibia is still disputed, but superficial similarities between lepospondyls and lissamphibians cannot mask the fundamental discrepancies in temporal range and morphology and (apparently) in life history as well. INTRODUCTION During most of the latter third of Paleozoic time, a span of some 90 million years, the swamps, streams, and ponds of the Northern Hemisphere were inhabited by a diverse assortment of amphibians which resembled the modern urodeles in their size range and adaptations. These amphibians are distinguished from their contemporaries, the labyrinthodonts, by vertebral centra which evidently ossified as cylinders around the notochord. By virtue of these "husk vertebrae" they are named Lepospondyli (or colloquially, lepospondyls), and set apart as a subclass of the Amphibia. Within the subclass three distinctive groups are recognized as orders. But classifying these ancient amphibians only emphasizes the fundamental enigma of their relationships. I may as well admit here that the origins and antecedents of the lepospondyls can only be guessed at, the affiliations within the group are obscure, and the connections (if any) between lepospondyls and present-day amphibians are disputed. This paper offers a skimming glimpse of the group in order to discuss (rather than answer) the question of its affinities. In a very real sense this survey is premature, for few of the known genera are adequately known, and several of the most crucial specimens have been lying in museum drawers, unprepared and unstudied or half-studied, for 80 to 100 years. Few paleontologists have showed much interest in lepospondyls as such. To put it bluntly, the saving grace of the lepospondyls has been that some of them look enough like reptiles to have attracted the attention of scientists interested in reptile origins: if they had not been described under a misapprehension, many of these curious amphibians might never have been described at all. For an excellent review of the Lepospondyli we are indebted to Dechaseaux (1955), who has judiciously sifted and weighed the published accounts, though of course much unpublished information was unavailable to her. Three component orders of the subclass can be distinguished with confidence: the limbless ATstopoda, the newt-like Nec- 287 tridea, and the sometimes reptile-like Microsauria. All three orders share a basic characteristic in the form of their vertebrae, which are composed of a single or twinned neural arch and a one-piece, hourglass-shaped centrum which is very probably homologous with that of the living lissamphibians (urodeles, frogs, and apodans), as Williams (1959) has argued. The centrum is thus a pleurocentrum, and the trunk vertebrae lack ossified intercentra, although chevron bones which have been observed in the tails of a few microsaurian genera may well be intercentral in homology. This characteristic lepospondylous vertebra (Fig. 1) occurs in fully developed form in the earliest known representatives of the group, and nothing is known of intermediate stages which might link it with the types of verte-

DONALD BAIRD FIG. 1. Trunk vertebrae of representative lcpospondyls. A, the aistopod Phlegethontia (from Gregory, retouched); B, the nectridean Diploceraspis (from Beerbower); C, the microsaur Cardiocephahis (from Gregory, Peabody and Price). brae found in labyrinthodonts or crossopterygian fishes. It simply appears de novo in the fossil record. "UNSEEN FEET": THE AISTOPODS Among lepospondyls, the most ancient pedigree is that of the aistopods which first appear in rocks of early Mississippian age, about 330 million years old. Indeed, aside from the late Devonian ichthyostegids the oldest known aistopod is the oldest known amphibian of any sort. Yet even at that remote date these serpentine amphibians were already highly specialized about as specialized, in fact, as any amphibians living or extinct. Only three genera have been discovered, the last of which became extinct in early Permian time some '260 million years ;igo. Having recently (1964) reviewed the group, I will merely outline its peculiarities here, using the best-known genus Phlegethontia (Figs. 1A, 2) as an example. The skull with its bulbous braincase and temporal fenestrae is amazingly sophisticated for so ancient a tetrapod; no comparable degree of elaboration was attained by labyrinthodonts or reptiles during the Carboniferous. Similar specializations can be seen in the skulls of apodans, lysorophid microsaurs, and amphisbaenid lizards; but Gregory (1948) has made it clear that these similarities are the result of convergent evolution. As comparisons between such highly evolved skulls are more hindrance than help in determining relationships, we must seek more reliable guides in the relatively conservative structures of the postcranial skeleton. The backbone is differentiated into neck, trunk, and tail, and comprises about 230 vertebrae in adults, the number apparently increasing with age. Neural spines are blade-like and the transverse processes are outgrowths of the centrum. The vertebrae FIG. 2. The aistopod Phlrgrthontia (skull from Gregory, skeleton from Dechaseaux). Abbreviations following Fig. 5.

PALEOZOIC LEPOSPONDYL AMPHIBIANS 289 are pierced by the spinal nerves, a condition found also in urodeles, but not in nectrideans or microsaurs. Ai'stopod ribs lie with their straight shafts nearly parallel to the body axis, their strange K-shaped proximal ends articulating by the capitulum alone. As in other lepospondyls, the belly is armored with a herringbone pattern of slender bony scales; the dorsal scales (when present) are distinctively pebble- or oatshaped. But the chief peculiarity of the ai'stopods is their total lack of limbs and girdles. Not even a rudiment can be found in any of the well-preserved skeletons from coalswamps of Pennsylvanian age, and the unique early Mississippian skeleton seems to be equally limbless. It might be tempting to speculate that limbs were never present in this lineage, and that ai'stopods originated directly from a crossopterygian fish through the suppression of the fins. Common sense, however, suggests that the obvious dividing-points between neck, trunk, and tail were once marked by girdles. This consideration and the basic similarities between aistopods and the four-legged nectrideans and microsaurs convince me that the forerunner of the aistopods had limbs. If these serpent-amphibians had already lost their appendicular skeletons by early Mississippian time, we must search for their four-legged ancestors among the mists of the Devonian, a period from which very few amphibians (and no lepospondyls) are known at present. Until such an ancestor is unearthed, the origins of the group must remain a mystery. NECTRIDEANS: NEWTS AND ARROWHEADS Probably the lepospondyls most familiar to non-specialists are the horned genera Diplocaidus and Diploceraspis from the early Permian. With their massive sculptured skulls shaped like broad arrows, these bizarre forms readily catch the eye of the museum visitor and student. As extreme evolutionary end-products, however, they offer little help toward an understanding of the Order Xectridea and its affinities. More representative of the order are the FIG. 3. Skulls of the nectrideans Keraterpeton (left) and Sauropleura (from Steen). Pennsylvanian genera such as the jjrimitive horn-bearer Keraterpeton (Figs. 3, 4B), and the urocordylids Ptyonius and Urocordylns. Little about the latter genera will be found in the literature, for Urocordyhis (sensu stricto) is represented by a single skeleton which has never been prepared, while Ptyonius was mistakenly sunk into synonymy in 1930, and has not yet struggled back to the surface. So until paleontologists can catch up with their descriptive chores, the common but somewhat specialized genus Sauropleura (Figs. 3, 4A) must serve as an example. Keraterpeton and Sauropleura, the most newt-like and elegant of the lepospondyls, seem to deserve an ordinal name meaning "girl swimmers." Both clearly show the basic nectridean trademark: a high-sided swimming tail with fan-shaped neural and haemal spines. Unlike those of most ancient tetrapods, the haemal arches are not intercentral in origin, but form an integral part of the pleurocentrum. Accessory articulations between vertebrae are another nectridean characteristic. The relatively short trunk consists of a fixed number of vertebrae bearing two-headed ribs. Limbs and girdles are diminutive but fully formed, with four toes in the forefoot and five in the hindfoot; the sculptured trio of thoracic plates clavicles and interclavicle look very much like those of labyrinthodonts. Belly scales range from needle-like to rhomboidal in different genera, and dorsal scales are unknown. Family variants in the nectridean skull pattern can be seen in Figure 3. Both

290 DONALD BAIRD FIG. 4. Reconstructions of lepospondyl skeletons. A, the urocordylid nectridean Sauropleura; B, the horned nectridean Keraterpeton; C, the microsaur Microbrachis (from Steen). families are typically short-snouted and have the usual amphibian pattern of bones in the skull roof. Urocordylids retain a small intertemporal bone, while keraterpetontids have traded it for a horn-shaped enlargement of the tabular which, in primitive genera, apparently connected the skull to the T-shaped upper bone of the shoulder girdle. In the earliest keraterpetontids the palate is closed, with only slit-like openings between the pterygoids and the parasphenoid, but these palatal vacuities widen in the course of time and become broad windows in the terminal genera Diplocaulus and Diploceraspis. Nectrideans make their first appearance, already full-blown and diversified, in lower Pennsylvanian deposits about 300 million years old; like the other lepospondyl orders, they disappear from the scene in early Permian time. Some similarities in vertebral structure which appear to ally nectrideans more closely to ai'stopods than to microsaurs have been pointed out by Gregory, Peabody, and Price (1956), but these resemblances are counterbalanced by the fact that the spinal nerves perforate the vertebrae in aistopods but pass between the vertebrae in nectrideans. Thus the relationship between these orders must be remote at best. To the extent that they are the least specialized and the most labuinthodont-like of the lepospondyls, the Carboniferous nectrideans may be thought to represent the basic stem of the subclass. In all but the most basic features, however, they are so far removed from the other orders that any attempt to reconstruct a composite common ancestor, an eo-lepospondyl, would be largely an exercise in science fiction. MICROSAURS: LITTLE BOCUS REPTILES As the Microsauria form the subject of Dr. Gregory's contribution to this symposium, a brief review for the sake of comparison is sufficient here. Microsaurs, which range from late Mississippian to early Permian time, form a natural but diversified group which has yet to be fully sorted into families (Romer, 1950). They are longtrunked and (for the most part) feeblelimbed, with either three or four toes in the forefoot (Fig. 4C). Their vertebrae, a trademark of the order, are three-piece affairs with one suture dividing the neural arch and another separating arch from centrum. Since ossification may obliterate either suture or both, however, too much importance should not be attached to their condition. Separate caudal chevrons which are presumably intercentral in origin have been reported in Microbrachis and Pantylus. The characteristic dorsal scales are o\al with radial striations superimposed on

PALEOZOIC LEPOSPONDVL AMPHIBIANS 291 pasph _JM ^_ FIG. 5. Microsaur skulls. A, Dolichopareias; B, Euryoclus; C, Ostodolepis; D, Microbrachis; E, Hyloplesion (with palate); V, Tuditanus; G. Lysoroplws. Right postparietal stippled, supratemporal hatched. (A-D from Romer; E modified from Dechaseaux; F original, from the newly prepared type specimen; G original, based on Sollas.) a angular d dentary eo exoccipital ep epipterygoid t frontal 1, la lacrimal in, mx maxilla n nasal ABBREVIATIONS FOR FIG. 2, 5, AND 6. p parietal pal palatine pasph parasphenoid pc posttrontal pro, pmx premaxilla po- postorbital pp postparietal prf prefrontal pt pterygoid a concentric growth pattern; the belly scales are rod-like or rhomboidal. Microsaur skulls (Fig. 5) are characterized by the absence of an otic notch and by the large size of the supratemporal bone. The palate is closed in gymnarthrids such as Euryoclus but conspicuous palatal vacuities occur in members of other groups, for example Microbrachis, Hyloplesion and Tuditanus. The genus Tuditanus from the middle of Pennsylvanian of Ohio requires special and apologetic mention because it has erroneously been called a reptile by Baird (1958) and "the oldest known reptile, Eosauravus" by Peabody (1959). Since those papers were q quadrate qj quadra tojugal sa surangular so supraoccipital sp splenial sq squamosal st supratemporal v vomer published, however, preparation of the type specimens has put matters in a different light: Eosauravus is a synonym of Tuditanus, trunk intercentra are not present, the scales are typically microsaurian, the forefoot has only four toes, and the supposedly reptilian tarsus is not irreconcilably different from that of the microsaur Hyloplesion. The deceptive nature of these little tetrapods is doubly illustrated in Fig. 6. Tuditanus, once thought to be a reptile, is now known to be a microsaur; while Cephalerpeton, once thought to be a microsaur, is now known to be a stem-reptile of the family Romeriidae, a structural ancestor

292 DONALD BAIRD FIG. 6. Deceptive similarity in skulls of the microsaur Tuditanus (above) and the true reptile Cephalerpeton (original reconstructions from the type specimens). Abbreviations following Fig. 5. for most of the Eureptilia and their offspring the birds and mammals. Besides looking superficially alike, Tuditanus and Cephalerpeton inhabited the same coalswamp and shared the trait of elusiveness, for only two skeletons of each have ever been found. Two problematical groups are mentioned last because not everyone is willing to include them with the microsaurs. Noteworthy for their antiquity are the adelospondyls (Fig. 5A) from the late Mississippian of Scotland, very rare and insufficiently studied. Their skulls are essentially microsaurian, although exceptionally long behind the orbits; the elongate body has microsaur-like vertebrae and a shoulder girdle of sculptured dermal bones; minute but well-developed limbs are present, and three toes can be seen in the forefoot; the belly scales are slender and rhomboidal. Watson (1929), the only previous commentator on adelospondyl taxonomy who has actually examined an adelospondyl, points out the strong similarities between these Mississippian forms and the Pennsylvanian and Permian lysorophids (discussed below). Further research, I predict, will only strengthen the existing evidence that the adelospondyls are an early-divergent family of microsaurs. The lysorophids, or molgophids, range from middle Pennsylvanian to early Permian, 1 and are best known from a Permian species assigned to Lysorophus (Fig. 5G). Like the microsaurs, they have extremely long bodies with relatively tiny limbs, a stemmed interclavicle, and either three or four toes in the forefoot. The skulls are highly specialized with a bulbous braincase (in most genera), a closed palate, and freefloating maxillae and premaxillae. Despite their atypical features, the skulls share the microsaur characteristics of a long postorbital region and a large supratemporal bone; the supraoccipital bone is usually wedged in between the postparietals as in Ostodolepis (Fig. 5C). Molgophid vertebrae are clearly microsaurian in pattern, with or without a divided neural arch, with or without a persistent neurocentral suture. Obscure structures which appear to be paired haemal arches are present in the tail of Molgophis. The scales are feebly ossified, but seem to be of microsaurian type. These peculiar amphibians form a closely knit family which, like so many others mentioned in this paper, is much in need of study. Lysorophids have been assigned hither and yon by various classifiers, but are best interpreted, I believe, as specialized microsaurs. LEPOSPONDYL AFFINITIES The three orders just described are sufficiently clear-cut that their status as natural groups seems beyond question. Despite their peculiarities, these orders share enough basic features to justify brigading them as a subclass of the Amphibia. Their relationships to other amphibian groups, however, are by no means clear. l A supposed lysorophid from the late Triassic has been described by Huene and Bock (1954) under the name Lysoroceplialus. Their specimen is not a lysorophid and probably not an amphibian at all, but appears instead to be the incomplete and ihimaged skull roof of a palaeoniscoid fish.

PALEOZOIC LEPOSPONDYL AMPHIBIANS 293 Except in the most specialized forms, the skulls, girdles, and limbs of lepospondyls are essentially similar to those of labyrinthodont amphibians and primitive reptiles. Though the lepospondyl skull and the labyrinthodont skull might conceivably have originated independently of each other at the crossopterygian fish level, a duplicate origin of the tetrapod appendicular skeleton is simply incredible. Thus common sense assures us that the Amphibia must be a monophyletic class, Jarvik (1960) to the contrary notwithstanding. The available evidence which is only sketchily summarized in this paper indicates that the Lepospondyli, Labyrinthodontia, and Eureptilia had a tetrapod ancestor in common, and that this ancestor must have lived well back in Devonian time, some 380 million years ago. Until more ancient tetrapod fossils have been discovered, and until those already discovered have been fully described, there will be little point in further speculation on amphibian origins. Devonian ancestors are to be found not in hypothetical reconstructions but somewhere under a rock. Evidence for and against the lepospondyls as ancestors of the present-day amphibians is presented in the accompanying paper by Dr. Estes, who views the ancient amphibians in the light of the modern groups. Looking backward in time, he concludes that the living Lissamphibia are more plausibly derived from the Labyrinthodontia than from the Lepospondyli. As a student of Paleozoic amphibians I look forward in time to seek their descendants among the Cenozoic forms and come to the same conclusion. The following evidence impresses me most: Lepospondyls disappear from the fossil record in the early Permian, but lissamphibians do not appear until the early Triassic; and no connecting forms are known. The vestigial dorsal scales of apodans resemble those of microsaurs, but are equally similar to the dorsal scales of labyrinthodonts, as Noble (1931) has pointed out. Lissamphibians (aside from sirenids) have pedicillate teeth with a distinct separation between pedicel and crown; lepospondyls do not. Urodeles have spinal nerves which pierce the vertebrae; lepospondyls (aside from aistopods) do not and the totally limbless aistopods can hardly be made into urodele ancestors on the basis of their intravertebral spinal nerves. The open palate of modern amphibians may be derived as easily from that of labyrinthodonts as from that of some microsaurs or nectrideans. Similarities between highly specialized skulls of lepospondyls and lissamphibians should be discounted, for structures which are so obviously subject to convergent evolution make most unreliable guides to kinship. In summary, then, the lissamphibians are remote from the lepospondyls both in time and in structure. Still another line of evidence, however, remains to be considered. THE PROBLEM OF METAMORPHOSIS Juvenile labyrinthodont amphibians, the "branchiosaurs" or "phyllospondyls" of classic authors, are well known for their highly cartilaginous skeletons and filamentous external gills. The smallest individuals differ so much from adult labyrinthodonts, and show so few diagnostic characters, that despairing taxonomists have lumped most of them into the catch-all genus Branchiosaurus (where few of them properly belong, certainly none of the Permian forms). Although limbless tadpoles of labyrinthodonts have not been found, the gill-bearing and semi-ossified "branchiosaurs" are generally accepted as being unmetamorphosed larvae. Both of the main labyrinthodont subdivisions, the rhachitome-stereospondyl group and the anthracosaur group, had larvae of this sort. Furthermore, large, gilled labyrinthodonts which are obviously neotenous are also known. Thus there can be no doubt that metamorphosis (or its alternative, neoteny) was characteristic of the labyrinthodont life cycle. Quite a different situation is found in the Paleozoic lepospondyls. Here, examples from each of the orders show that juvenile individuals are essentially miniature adults, readily assignable to adult-based species

294 DONALD BAIRD wherever the specimens are well-enough preserved to be identified at all, and differing from adults only in proportions which are subject to allometric growth. No juvenile lepospondyl shows traces of external gills. Although the matter needs further investigation before positive statements can be made, at present I know of no evidence for metamorphosis in Paleozoic lepospondyls. If the lepospondylous amphibians were characterized by direct development, as seems to be the case, it is difficult to conceive of them as the ancestors of metamorphosing amphibians such as frogs and salamanders. For this additional reason I concur with Estes and other authors in believing that lissamphibian origins are to be sought among the Labyrinthodontia rather than among the Lepospondyli. Parenthetically, let us hope that this observation on the apparent absence of metamorphosis in lepospondyls will not provoke still another attempt to link the microsaurs with the reptiles! Now that the Lissamphibia are accepted as constituting a natural group, the taxonomic status of that group needs to be put into perspective. Under Romer's 1945 classification, the Class Amphibia had been bisected into the Subclass Lepospondyli (the lepospondyls as defined here plus the urodeles and apodans) and the Subclass Apsidospondyli (the labyrinthodonts plus the frogs). When the lissamphibian orders have been subtracted from this system and combined, three alternative placements are possible: to include them as a superorder in the Lepospondyli (where I feel confident they do not belong), to place them in the Apsidospondyli as a superorder equivalent in rank to the Labyrinthodontia (which is more plausible), or to recognize their distinctness in morphology and temporal range by treating them as a third major component of the Amphibia, the Subclass Lissamphibia. The last alternative seems best to me. In that case the categories Apsidospondyli and Labyrinthodontia become conterminous, and most students will doubtless prefer to designate this taxon by the more familiar name. We may thus consider the Class Amphibia as comprising the Subclasses Labyrinthodontia, Lepospondyli, and Lissamphibia. REFERENCES Baird, D. 1958. The oldest known reptile fauna. Anat. Rec. 132:407-408.. 1964. The a'istopod amphibians surveyed. Mus. Comp. Zool. Harvard, Breviora 206:1-17. Dechaseaux, C. 1955. Lepospondyli, p. 275-305. In J. Piveteau, ed., Traite de paleontologie, Vol. 5. Masson et Cie., Paris. Gregory, J. T. 1948. A new limbless vertebrate rom the l'ennsylvanian of Mazon Creek, Illinois. Am. Jour. Sci. 246:636-663. Gregory, J. T., F. E. Peabody, and L. I. Price. 1956. Revision of the Gymnarthridae, American Permian microsaurs. Peabody Mus. Nat. Hist. Bull. (Yale) 10, 77 p. Huene, F., and W. Bock. 1954. A small amphibian skull from the Upper Triassic of Pennsylvania. Wagner Free Inst. Sci. Bull. 29:27-35. Jarvik, E. I960. Theories de 1'evolution des vertebres. Masson et Cie., Paris. 104 p. Noble, G. K. 1931. The biology of the Amphibia. McGraw-Hill Book Co., New York. 577 p. Peabody, F. E. 1959. The oldest known reptile, Eosaurainis copei Williston. Smithsonian Misc. Coll. 139:1-15. Romer, A. S. 1945. Vertebrate paleontology. Univ. Chicago Press. 687 p. -. 1950. The nature and relationships of the Paleo7oic microsaurs. Am. Jour. Sci. 248:628-654. Watson, I). M. S. 1929. The Carboniferous Amphibia of Scotland. Palaeontologia Hungarica 1:219-252. Williams, E. E. 1959. Gadow's arcualia and the de- \elopment of tetrapod vertebrae. Quart. Rev. Biol. 34:1-32.