NovAMERICANitatesMUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2770, pp. 1-41, figs. 1-14 November 8, 1983

Colosteus scutellatus (Newberry), a Primitive Temnospondyl Amphibian from the Middle Pennsylvanian of Linton, Ohio

ROBERT W. HOOK' CONTENTS

Abstract ...... 2 Introduction ...... 2 Acknowledgments ...... 2 Abbreviations ...... 3 Methods ...... 3 Review of Previous Work ...... 4 Systematics ...... 6 Description ...... 7 Skull Roof...... 7 Palate ...... 12 Parasphenoid and Braincase ...... 16 Mandible ...... 16 Dentition ...... 18 Dermal Sculpture and Lateral Line Canals ...... 19 Branchial Arches ...... 20 Postcranial Remains ...... 21 Occurrences of Colosteid Remains ...... 29 Characteristics and Composition of the Colosteidae ...... 29 Shared Derived Characters of the Family ...... 29 Relationships Within the Family ...... 31

Relationships of the Colosteidae ...... 32 The Status of the Order Temnospondyli ...... 32 The Development of the Temnospondyl Otic Notch ...... 33 Colosteids, Edopoids, and Trimerorhachoids ...... 34 Systematic Conclusions ...... 35 Appendix ...... 35 Literature Cited ...... 37

' Coal Geology Research Group, Department of Geology, University of Kentucky, Lexington, Kentucky.

Copyright © American Museum of Natural History 1983 ISSN 0003-0082 / Price $3.40 2 AMERICAN MUSEUM NOVITATES NO. 2770

ABSTRACT Colosteus scutellatus (Newberry), known exclu- (Tatarinov, 1964) is recommended in the absence sively from the classic Coal Measure locality of of shared derived characters indicating a close re- Linton, Ohio, is the genotype and youngest rep- lationship between the colosteids and any other resentative ofthe Colosteidae, an archaic group of known temnospondyls. The structure ofthe stapes temnospondyls that also includes Pholidogaster and and skull roof suggests that the colosteoids may Greererpeton. The colosteids are a monophyletic represent the plesiomorphic sister group ofall oth- group distinguished from other temnospondyls by er temnospondyls. massive stapes and no otic notch; elongate pre- An overview ofthe status ofthe Temnospondyli frontal contacting narial border, premaxilla, and indicates that exclusion of the loxommatoids is maxilla; intertemporal reduced or absent with justified, but that a possible microsaur-temno- broad postorbital-parietal contact; single pair of spondyl relationship circumvents the possibility premaxillary tusks; and single, elongate Meckelian of temnospondyl monophyly. fenestra. The use of the superfamily Colosteoidea

INTRODUCTION The Temnospondyli are the largest and burkemorani Romer, a Colosteus-like form possibly most diverse order of fossil am- from the Mississippian of West Virginia, it phibians. Despite an impressive representa- became evident that the colosteids represent tion of Late Paleozoic and Triassic forms, an archaic lineage that might provide insights temnospondyls older than Pennsylvanian or into the early evolution of temnospondyls, if Westphalian age are virtually unknown. Con- not all Paleozoic amphibians. Subsequent sequently, the astonishing array of temno- study of additional colosteid material (Ro- spondyls that first appears in the Coal Mea- mer, 1972; Panchen, 1975; Smithson, 1982) sures of North America and Europe has has identified the need for a more detailed consistently frustrated paleontologists who understanding of the genotype of the family. wish to incorporate at least a modicum of The objective of the present paper is to phylogeny in their classifications ofthe order. provide a thorough description of the genus The variety and abundance of Pennsylva- Colosteus and an accurate characterization of nian temnospondyls is nowhere better doc- the Colosteidae. Although the affinities ofthe umented than in collections acquired from family are also considered, re-evaluation of the once-active coal mines of Linton, Ohio. other primitive temnospondyls is necessary In addition to a dozen species of fishes and before the problem of interrelationships can a score ofother tetrapods, this prolific locality be fully addressed. has produced seven temnospondyl genera, four of which are unique to Linton (Hook, ACKNOWLEDGMENTS 1981). The most common of these indige- nous temnospondyls is Colosteus scutellatus, I am greatly indebted to Dr. Donald Baird, a heavily scaled amphibian with an elongate whose commitment over the past 30 years body and diminutive limbs. Because similar to the collection, preparation, and identifi- forms are unknown from younger deposits, cation ofLinton fossils made the present study such as the Texas redbeds (Romer, 1935), possible. I am also grateful to Dr. Robert R. early workers generally considered Colosteus Reisz for support and thoughtful criticism. as a novel adaptation to the coal swamp en- My thanks are due to Dr. Andrew R. Mil- vironment. Thus, after a somewhat check- ner, Messrs. John Chom, Stephen Godfrey, ered taxonomic history, Colosteus and the Timothy R. Smithson, and Carl F. Wellstead family to which it gives its name, the Colos- for free exchange of information on unpub- teidae, were removed from the main line lished material; to Dr. Robert L. Carroll for of temnospondyl evolution (Romer, 1947, initially directing my attention to the need 1966). for a review of the Linton colosteids and for With the 1969 description of Greererpeton making available unpublished drawings by 1983 HOOK: COLOSTEUS SCUTELLA TUS NEWBERRY 3

Mrs. Pamela Gaskill; to Drs. Richard Lund YPM, Peabody Museum of Natural History, Yale and Kathy V. Bossy for loans of peels of Lin- University ton specimens; to Dr. Malcolm J. Heaton for interest in my work and advice on the prep- ANATOMICAL aration of specimens and figures; to Ms. Di- a, angular ane Scott for helpful suggestions in the pro- a.p.f., anterior palatal fenestra duction of illustrations; and to Mr. Steven art, articular Jaunzems for expert photographic services. c, clavicle I also thank Drs. Anne E. Fry and Dennis cb, ceratobranchial ossifications Radabaugh of Ohio Wesleyan University for ch, choana their in c.r., conical recess assistance recovering Dr. Baird's peels ct, cultriform process from the research materials of the late Dr. cth, cleithrum John N. Chase. Dr. Robert W. Morris of Wit- co, coronoid tenberg University and Mr. Michael C. Han- d, dentary sen of the Ohio Geological Survey provided ect, ectopterygoid valuable guidance in the early stages of this eo, exoccipital work. f, frontal It is similarly a pleasure to acknowledge h, humerus the assistance of individuals at institutions hb, hypobranchial cartilage where I have studied Linton material. These i, interclavicle ic, intercentrum include Drs. Eugene S. Gaffney and John G. im.f., intermandibular fenestra Maisey, American Museum of Natural His- j, jugal tory; Dr. Angela C. Milner, British Museum 1, lacrimal (Natural History); Dr. David S. Berman, Car- m, maxilla negie Museum of Natural History; Dr. Mi- n, nasal chael E. Williams, Cleveland Museum of p, parietal Natural History; Dr. John R. Bolt, Field Mu- pal, palatine seum of Natural History; Dr. Donald Brink- pa.f., para-articular foramen man and Mr. Charles R. Schaff, Museum of pra, prearticular Comparative Zoology, Harvard University; pc, pleurocentrum pf, postfrontal Dr. Stig M. Bergstrom, Orton Museum, Ohio pl, pleuracanth shark tooth State University; and Ms. Ruth L. Elder, Mu- pm, premaxilla seum of Paleontology, University of Michi- po, postorbital gan. pp, postparietal Drs. C. S. Churcher, Thomas S. Parsons, prf, prefrontal and P. J. Pointing read an early version of pt, pterygoid the manuscript. The suggestions of Drs. qj, quadratojugal Baird and Milner improved the final draft. r, radius sa, surangular ABBREVIATIONS sp, splenial spp, postsplenial INSTITUTIONAL st, supratemporal t, tabular AMNH, American Museum of Natural History u, ulna BM[NH], British Museum (Natural History) u.p., uncinate process CM, Carnegie Museum of Natural History v, vomer FMNH, Field Museum of Natural History of manus MCZ, Museum ofComparative Zoology, Harvard I-IV, digits University OSU, Orton Museum, Ohio State University METHODS PU, Museum of Natural History, Princeton Uni- versity Vertebrate remains from Linton have been UMMP, Museum of Paleontology, University of reduced to virtually two dimensions by post- Michigan mortem crushing. Early descriptive efforts 4 AMERICAN MUSEUM NOVITATES NO. 2770 were further vexed by the partial or complete corded with the description; further dimen- pyritization ofthe original bone material (see sions may be taken directly from the speci- Moodie, 1915). It was not until the 1950s, men drawings if necessary. when the technique of acid-etching was fol- The present study considers the morphol- lowed by the use of latex as a microcasting ogy of an animal whose remains present no compound (Baird, 1955), that detailed study perceivable characteristics from which sexual of Linton specimens became feasible. In this maturity or absolute age can be inferred. innovative manner, Baird has prepared most Consequently, the terms subadult and adult ofthe important vertebrate material from this are based arbitrarily on absolute size, not de- locality. Using the spray technique described velopmental stage: specimens with an esti- by Heaton (1 980b), I have cast the remaining mated skull length (measured along the dor- unprepared temnospondyl specimens of Lin- sal midline) less than 15 cm. are considered ton collections ofNorth American museums. to be subadults. Although semantically in- These latex "peels" provide detailed casts congruent, this usage provides a means by from which study of Colosteus and other Lin- which large, diagnostic materials can be dif- ton amphibians can proceed. ferentiated from smaller specimens that may The drawing of significant Linton speci- lack the apomorphic states produced by al- mens is necessitated by the generally inade- lometric growth (A. R. Milner, 1 980a). With- quate representation achieved by photogra- out this discrimination, subadults may form phy alone. Rather than using standard camera the basis of erroneous phylogenetic state- lucida or grid ocular techniques (Carroll and ments-the classic example being the infa- Gaskill, 1978), drawings were traced from mous "order" Phyllospondyli (Romer, 1939). twofold photographic enlargements onto my- lar drafting film while the corresponding peel REVIEW OF PREVIOUS WORK was examined under high magnification. This method is not only faster than those cited The taxonomic histories of most speci- above, but also more accurate because it re- mens presently referred to Colosteus are tor- duces the amount ofperipheral distortion and tuously complex. Since such details are of allows free movement ofthe specimen during little general interest, their treatment has been drawing. relegated to the Appendix. It is appropriate, Reconstructions were drafted on paper and however, to summarize previous work by way based largely on specimen drawings. Resto- oftracing the publication history ofthe genus ration of the skull in dorsal view (fig. 1A) is Colosteus and the family Colosteidae. derived from several nearly complete sub- The first report of fossil vertebrates from adult skulls ofintermediate size (average skull the Ohio Diamond Coal Company Mines at length of 9 cm.). The reconstruction of the Linton was given by J. S. Newberry in 1856; palate (fig. 1 B) is based on four subadult spec- it included a brief description of Pygopterus imens whose estimated skull lengths range scutellatus, a supposedly new species of pa- from 5.5 to 12 cm. Palatal areas not repre- leoniscoid fish. Although no figures were sup- sented by actual remains (about 35% of the plied or specific specimens adequately de- composite) were restored with slight modi- scribed, a general picture of a heavily scaled, fications after Greererpeton, a colosteid whose elongate form with flattened head and point- palatal architecture is reasonably well docu- ed snout was established. mented (Smithson, 1982). In the absence of In 1869, E. D. Cope erected the batrachian adequate information on the occiput of Co- genus Colosteus on the basis of Linton ma- losteus, no reconstruction of this region has terial lent him by Newberry. Cope cited three been attempted. species, C. crassiscutatus, C. marshii and C. Accurate measurement of Linton fossils is foveatus, and provided measurements by made difficult by the crushed and incomplete which the type specimens can be identified nature of specimens. As such, it is not prac- today. However, Cope later admitted (187 1 a, tical to attempt a conventional tabulation of p. 41) that in describing the type species, C. dimensions. Incidental measurements are re- crassiscutatus, he had "overlooked" New- 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 5 berry's original account of P. scutellatus, ropleura, and the familial name lapsed to the thereby implying that both taxa were inad- Sauropleuridae (Hay, 1902). This oversight vertently based on the same specimen. Since was amplified by R. L. Moodie, who, in a Pygopterus was (and is) a valid fish genus, series ofpapers culminating in a 1916 mono- and the material in question was certainly not graph, compounded existing taxonomic piscine (Cope, 1873), Cope recognized the problems with additional errors of his own proper combination of Colosteus scutellatus invention (see Appendix). (Newberry, 1856) for the type species. With In an ambitious, if not commendable, at- this synonymy, Colosteus predates Amphi- tempt to obtain "a reduction in the number bamus ("Raniceps," "Pelion") lyelli (Wy- of species" at Linton, A. S. Romer (1930, p. man, 1857) as the first tetrapod recorded- 79) resurrected Colosteus but retained only albeit as a fish-from the Linton deposit. the type species in his redescription of the Newberry also described (1856) and later genus. Although he corrected many existing figured (1873, pl. 39, fig. 6) Rhizodus angus- mistakes, Romer "committed some new tus, a presumed crossopterygian fish from ones" himself (Romer, 1968, p. 67). Partic- Linton. Romer (1930) noted that the type of ularly unfortunate was his attempt to aug- this species was unknown and, on the basis ment the postcranially impoverished sample ofNewberry's figure, recorded R. angustus as of Colosteus with a sizable quantity of head- a synonym of C. scutellatus since the latter less "lepospondyl" material, described orig- species has page priority. However, because inally by Cope as Molgophis and Pleuroptyx. the description and figure of R. angustus are The improbable nature ofthis graft was elab- nondiagnostic (there is nothing to bar it from orated upon by Steen (1931) and merits no being any one of the larger Linton labyrin- further consideration. thodonts) and renewed search for the speci- Romer (1930) also reinstated the family men has been unsuccessful, it may be thank- Colosteidae in which he associated a second fully regarded as a nomen dubium and nomen Linton genus, Erpetosaurus, with the geno- oblitum. type while transferring Sauropleura to the In an 187 lb summary of the "Batrachian Urocordylidae and Amphibamus (as "Pe- Fauna" of Linton, Cope erected a new form, lion") to the Peliontidae (subsequently as- Oestocephalus marshii, without description. signed to the Dissorophidae by Carroll, 1964). He later referred this name to Ptyonius Because most of his material consisted of marshii (Cope, 1874) and revealed that both poorly ossified, branchiosaur-like subadults, taxa were based on his 1869 description of Romer placed the family among the Phyl- C. marshii; recent review of the urocordylid lospondyli; similarly influenced by the work nectrideans has upheld this synonymy (Bos- of Watson (1921), Steen (1931) followed this sy, Ms). In the same 187 lb report, Cope mistaken interpretation. Upon dissolution of erected another nomen nudum, Colosteus the phyllospondyls, Romer deemed Colos- pauciradiatus, only to supply description and teus and Erpetosaurus to be primitive rha- figures in subsequent papers (1874, 1875). chitomes (1939) that had descended from The holotype (AMNH 6920, fig. 1 lA) ofthis ichthyostegids (1945), and separated, there- taxon is an interclavicle that was originally fore, from the main line of temnospondyl included in the C. crassiscutatus paratype se- evolution (1947). ries (Cope, 1869, p. 23; contra 1874, p. 275 Jarvik (1948, 1952) rejected the possibility and 1875, p. 408); however, it does not merit of a close relationship between Ichthyostega specific separation because it cannot be dif- and colosteids, but proposed no alternative ferentiated from comparably sized interclav- placement for either Colosteus or Erpetosau- icles of the type species. rus. This dissent was followed by Lehman The family Colosteidae was erected by Cope (1955), who recognized the "?Colosteidea" in 1875 to accommodate Colosteus, Sauro- as a superfamily of the suborder Rachitomi pleura, and Amphibamus. However, in one [sic]. Through four volumes of the Bibliog- ofhis last papers, Cope (1897) made the mis- raphy ofFossil Vertebrates (Camp and Alli- take of synonymizing Colosteus with Sau- son, 1961; Camp et al., 1964, 1968, 1972), 6 AMERICAN MUSEUM NOVITATES NO. 2770 the Colosteidae was included with the Elpis- affinities of Erpetosaurus lie closer to saur- tostegidae and Otocratiidae in the non-rha- erpetontid trimerorhachoids than to colos- chitomous suborder Elpistostegalia. More teids (Hook, 1980). appropriately, Tatarinov (1964) erected the superfamily Colosteoidea under the Rha- SYSTEMATICS chitomi and recognized two families, the Co- losteidae and Otocratiidae. Romer, in the last CLASS AMPHIBIA edition of Vertebrate Paleontology (1966), ORDER TEMNOSPONDYLI questioned his own assignment of the colos- SUPERFAMILY COLOSTEOIDEA teids to the superfamily Edopoidea, and, at TATARINOV, 1964, SENSUNOVUM a later date, indicated further apprehension EMENDED DIAGNOSIS: Primitive tem- in stating that the two Linton forms "ap- nospondyl amphibians characterized by mas- proach the metoposaurs more closely than sive stapes, extensive tabular-squamosal the trimerorhachids" (1968, p. 78). contact, and no otic notch; elongate prefron- The discovery of a Mississippian-age co- tal reaches external naris, contacting pre- losteid did surprisingly little to clarify the maxilla and maxilla, excluding lacrimal and phylogenetic position of the family. As the nasal from narial opening; intertemporal mi- first describable non-loxommatoid temno- nute or absent, broad postorbital-parietal spondyl from the Lower Carboniferous, contact; single pair of premaxillary tusks on Greererpeton burkemorani Romer (1969) ap- posterolateral palatal flange; single, elongate peared to be identical in many respects to the Meckelian fenestra; approximately forty pre- Pennsylvanian-age Colosteus. Panchen (1975) sacral vertebrae; extensive dorsal and ventral subsequently concluded that two problemat- scalation, ventral scales rhomboidal with ical labyrinthodonts from the British Lower crenulated posterior margins. Carboniferous, Otocratia modesta Watson Differentiated from edopoids by well- (1929) and Pholidogaster pisciformis Huxley developed lateral line system, marginally sit- (1862), represented a single taxon that was uated external nares, no exposure of septo- very similar to Greererpeton. Although these maxilla on skull roof, orbital lacrimal, ac- important discoveries led to the realization cessory tusks on palatine and ectopterygoid, of the great antiquity of the colosteid stock separate sphenethmoid and otico-occipital (Panchen, 1972, 1973), no attempt was made braincase regions, and retention of anterior to reconcile such with existing classifications. flange and entepicondylar foramen on the hu- Thus, in the most recent comprehensive merus. scheme of amphibian classification (Carroll Distinguished from trimerorhachoids by and Winer, 1977), the Colosteidae consisted laterally directed external nares, narrow in- ofPholidogaster, Greererpeton, Colosteus, and terpterygoid vacuities completely enclosed by Erpetosaurus, and was assigned to the Tri- pterygoids, thin cultriform process not su- merorhachoidea. tured to vomers on palatal surface, premax- As reported by various workers (Romer, illa borders choana, no vomerine tusks, and 1968; Panchen, 1975), the late Dr. John N. dentary teeth markedly larger than maxillary. Chase, in collaboration with Dr. Baird, had begun a redescription of Colosteus and Er- FAMILY COLOSTEIDAE COPE, 1875 petosaurus. Before his death in 1977, Dr. DIAGNOSIS: As for the superfamily, which Chase had photographed peels ofappropriate is monotypic. specimens and had initiated drawing efforts as- of important material. In 1979, with the COLOSTE US 1869 sistance of Dr. Baird, I was able to recover GENUS COPE, the peels formerly in Dr. Chase's possession TYPE SPECIES: Colosteus scutellatus (New- and to critically evaluate the association of berry, 1856). Erpetosaurus with the Colosteidae. On the DIAGNOSIS: Colosteoid temnospondyl basis of this work, which will be reported in characterized by only three bones, jugal, lac- full at a later date, it is apparent that the rimal, and postfrontal, forming orbital mar- 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 7 gin; open lateral line system; single row of Leptophractus lineolatus Cope, 1877, p. 576; as C. minute coronoid teeth. Dorsal scalation re- scutellatus, Romer, 1930, pp. 100-101, 106. sembles gastralia. - Megalocephalus lineolatus (Cope, 1877); Beaumont, 1977, pp. 73-76. Colosteus scutellatus (Newberry, 1856) Anisodexis enchodus Cope, 1885, p. 406; as C. scutellatus, Romer, 1930, pp. 100-101, 106. SYNONYMY: - M. lineolatus (Cope, 1877); NEW SYNON- YMY. Pygopterus scutellatus Newberry, 1856, p. 98. Diceratosaurus robustus Moodie, 1909b, pp. 67- Colosteus crassiscutatus Cope, 1869, p. 23. 69, fig. 5; as C. scutellatus, Romer, 1930, pp. Colosteus scutellatus, Cope 1871a, p. 41; 1874, p. 101, 106, fig. 10. 275; 1875, pp. 407-408, pl. 33, fig. 1, pl. 34, = Erpetosaurus radiatus (Cope, 1874); NEW fig. 2; 1877, p. 578. Romer, 1930, pp. 100- SYNONYMY. 108, figs.8 (part),9, 11. Steen, 1931, pp. 858- 860, figs. 6, 7, p1. 2, fig. 1. SPECIFIC DIAGNOSIS: As for genus. Colosteus pauciradiatus Cope 187 lb, p. 177 (no- HOLOTYPE: AMNH 6916, formerly 8584G men nudum); 1874, p. 275; 1875, p. 408, pl. and 8666G of Columbia University, collect- 40, figs. 1, 2. ed by Prof. John S. Newberry (ca. 1855): lat- Sauropleura scutellata, Cope, 1897, pp. 86, 88. erally compressed anterior portion of a sub- Moodie, 1909a, p. 26; 1916, pp. 156-157, pl. adult individual, including skull, mandibles, 14, fig. 3, pl. 21, fig. 5. Sauropleura pauciradiata Cope, 1897, p. 88. partial ribs, incomplete pectoral girdle and Moodie, 1916, pp. 158-160, fig. 34. scalation; preserved in counterpart blocks. Sauropleura longidentata Moodie, 1909b, pp. 74- HORIZON: Cannel coal below coal seam 76, figs. 18, 19; 1916, pp. 160-16 1, pl. 16, identified as the Upper Freeport Coal (New- figs. 2, 3. berry, 1871, 1874), Allegheny Group, Middle Macrerpeton deani Moodie, 1916, pp. 184-185, Pennsylvanian; equivalent to late Westpha- fig. 40, pl. 21, figs. 1, 2. lian D of Europe (Baird, 1964). LOCALITY: Coal mines originally owned by NON: the Ohio Diamond Coal Company (Murphy, 1980), Linton, Saline Township, Jefferson Rhizodus angustus Newberry, 1856, p. 99; as C. County, Ohio; approximately 2.5 km. south scutellatus, Romer, 1930, pp. 100-101, 106. of Wellsville, Ohio. = nomen dubium, nomen oblitum. HYPODIGM: See Appendix. Molgophis macrurus Cope, 1868, pp. 220-221; questionably assigned to C. scutellatus, Ro- mer, 1930, pp. 101, 106. DESCRIPTION = M. macrurus, Romer, 1952, p. 76. Colosteusfoveatus Cope, 1869, p. 24. SKULL ROOF = Saurerpeton obtusum (Cope, 1868); NEW The skull is a flat, posteriorly broad struc- SYNONYMY. In Colosteus marshii Cope, 1869, p. 24. ture with gently sloping cheeks. dorsal view = Ptyonius marshii Cope, 1874, p. 265. (fig. 1 A), the outline of the skull and position Molgophis brevicostatus Cope, 1875, p. 369, pl. of the orbits bear resemblance to the long- 44, fig. 1; questionably assigned to C. scutel- snouted trimerorhachoid Neldasaurus (Chase, latus, Romer, 1930, pp. 101, 106. 1965); smaller skulls show proportionally = M. macrurus Cope, 1868; NEW SYNONY- larger orbits and shorter snout lengths. MY (Wellstead, personal commun.). Whether or not the antorbital length of the Pleuroptyx clavatus Cope, 1875, pp. 370-371, pl. largest skulls would have equaled that of 44, fig. 2; questionably assigned to C. scutel- Greererpeton is not known. The skull width latus, Romer, 1930, pp. 101, 106. is greatest just anterior to the jaw articulation = P. clavatus; Steen, 1931, p. 860. the skull Sauropleura newberryi Cope, 1875, pp. 404-405, and is roughly 75 percent of length. pl. 37, fig. 2; as C. scutellatus, Romer, 1930, Although the posterior margin of the skull is p. 100 (only). modestly embayed in the area of the post- - Stegops newberryi (Cope, 1875); Hook and parietal-tabular suture, there is no suggestion Baird, in press. of an otic notch. The orbits are directed dor- 8 AMERICAN MUSEUM NOVITATES NO. 2770

A B

C FIG. 1. Colosteus scutellatus (Newberry). Restoration of subadult skull and mandible. A, skull in dorsal view; B, skull in ventral view; C, skull and detached mandible in lateral view. Scale equals 1 cm. sally with their lateral margins slightly lower act nature of the anterior portion of the pre- than the medial. In contrast to contemporary maxillary-nasal suture is uncertain. Although edopoids and trimerorhachoids, the external the evidence is equivocal, rostral fenestrae, nares are situated near the skull margin and as found occasionally in the loxommatid have no dorsal exposure. Megalocephalus pachycephalus (Beaumont, The premaxilla is a robust, tusk-bearing 1977), may have perforated the dorsal surface element that forms the bluntly rounded an- of the snout in this region to accommodate terolateral margin ofthe skull. Because crush- the tips of the parasymphysial tusks. ing has favored exposure ofthe more massive At about the level of the anterior margin vertical portions of this bone at the expense of the external naris, the premaxilla contacts ofthe dorsal and palatal components, the ex- the prefrontal. From this point on the dorsal 1983 HOOK: COLOSTEUS SCUTELLA TUS NEWBERRY 9

FIG. 2. Colosteus scutellatus (Newberry). Subadult skull, mandible, and anterior postcranial skeleton. Right counterpart block of holotype specimen, AMNH 6916. Scale equals 1 cm. surface, the premaxilla gives rise to a wedge- ries, nearly equaling the length of the skull shaped, posterolateral extension that over- table. It is also the narrowest element in this laps the anterior edge ofthe prefrontal before series, with a width less than one-fourth its passing laterally and downward to the exter- length. At a level slightly posterior to the or- nal naris. The anterior halfofthe narial open- bits, the frontal overlaps the parietal in an ing is bordered by the premaxilla. Immedi- interdigitating suture. ately above the premaxillary tusk, a short, Previous reconstructions of Colosteus (Ro- unsculptured flange forms the ventral rim of mer, 1945, 1947, 1966) have shown the pre- the naris; this flange obviously communicat- frontal as not reaching the naris because of ed with an anterior extension of the maxilla, an intervening septomaxilla. However, latex but the precise configuration is unclear. casts of AMNH 6916 (fig. 2) and 6945 (fig. The maxilla occupies about 75 percent of 3A) clearly indicate that a septomaxilla is not the lateral skull margin. It attains its greatest present and that the prefrontal continues an- height and concomitant dorsal exposure teriorly between the premaxilla and maxilla where the infraorbital lateral line sulcus in- to comprise about one-sixth of the external tersects the dorsal edge of the bone, near the narial border. This uncommon extension of level of the vomerine tusk. Immediately the prefrontal is identical to that of Greer- above the marginal tooth row, the external erpeton (Smithson, 1982). surface of the maxilla is nearly vertical and The configuration of the circumorbital only weakly sculptured. Posteriorly, the max- bones is modified from that seen in other illa appears to contact the quadratojugal, but colosteids to an arrangement that is unique the evidence for this is equivocal. among all amphibians. In the aforemen- The nasal is a flat bone that is preserved tioned restorations, a posterolateral exten- only in subadult specimens (AMNH 6916, sion of the prefrontal enters the orbital mar- fig. 2; 6945, fig. 3A; BM[NH] R.2547, fig. gin. However, close examination of several 3B). For halfofits length, the nasal is a broad well-preserved specimens (AMNH 6916, fig. element that contacts the premaxilla laterally 2; 6945, fig. 3A; FMNH UC 2001, fig. 4) while sharing a common medial suture with indicates that the prefrontal is precluded from its partner along the dorsal midline. The pos- the orbital margin by an anteriorly directed terior halfofthe bone forms a wedge between lacrimal-postfrontal suture. In a somewhat the frontal and prefrontal. similar fashion, a brief postfrontal-jugal con- Unlike the frontal of most temnospondyls tact on the posterior side ofthe orbit excludes with elongate skulls, the frontal of colosteids the postorbital from the circumorbital series. is the longest bone of the dorsal midline se- The resulting arrangement of only three 10 AMERICAN MUSEUM NOVITATES NO. 2770

B

FIG. 3. Colosteus scutellatus (Newberry). Subadult skull and mandibles of single individual in coun- terpart blocks. A, holotype of Sauropleura longidentata, AMNH 6945; B, BM[NH] R.2547. Scale equals 1 cm.

bones -jugal, lacrimal, and postfrontal- gin is formed by the triangular lacrimal. Be- forming the circumference of the orbit is in cause of the prefrontal-maxillary suture, the contrast to the standard five-bone configu- lacrimal fails to reach either the nasal or the ration of most other primitive labyrintho- external naris. The postfrontal extends an- donts. teriorly above the orbit and between the fron- The anterior one-third of the orbital mar- tal and lacrimal to meet the prefrontal in a 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 11

roughly transverse suture. Posteriorly, the postfrontal expands laterally to meet thejugal and form about one-half of the posterior or- bital margin. In adult specimens (AMNH 6826, fig. 5), this contact, as well as the ad- joining postfrontal-postorbital suture, is a strongly interdigitating feature that is rarely disrupted by crushing. The posterolateral margin of the orbit is formed by the jugal, the largest bone of the skull roof. Below the orbit, it reaches the posterolateral corner of the lacrimal while laterally overlying the maxilla. Crushing has revealed that the jugal also overlaps the quadratojugal and squa- mosal posteriorly while its own mediodorsal border is overlain by the lateral edge of the postorbital. The postorbital is a roughly pentagonal ele- ment whose margins overlap those ofall sur- rounding bones. As Smithson (1982) has not- ed in describing Greererpeton, the absence of FIG. 4. Colosteus scutellatus (Newberry). Por- an intertemporal in colosteids results in an tion of subadult skull roof in dorsal view. FMNH exceptional postorbital-parietal contact. Al- UC 2001. Scale equals 1 cm. though the intertemporal is lacking in Co- losteus, its presence is often mimicked by the truncation of the abbreviated temporal sul- distinct otic notches, to Colosteus (Romer, cus ofthe lateral line system near the anterior 1930, fig. 8, above right); this specimen is end ofthe parietal-supratemporal suture (see actually a large individual of Erpetosaurus figs. 3-6); this groove, usually constricted by radiatus (Baird, personal commun.). postmortem crushing, resembles the medial One subadult specimen (AMNH 6915, fig. boundary of a much-reduced intertemporal 7B) partially exposes the underside ofthe skull element. table to display the rudiments of the attach- The remainder ofthe skull roof is virtually ment structures for the braincase. These same identical with that of Greererpeton in all re- features are much better developed in mature spects. The parietal is the largest bone of the specimens of Greererpeton (Smithson, 1982, skull table and attains its greatest breadth at fig. 1 3B). Paired rugosities located posteriorly the level of the pineal foramen. The post- on each side of the dorsal midline indicate parietal is nearly two-thirds the length of the where the exoccipital reached the postpari- parietal and, with its partner, comprises about etal. Laterally, a ventral flange on the tabular one-third ofthe posterior skull margin as seen received the paroccipital process from the in dorsal view. In the reduced two-element braincase, as well as the epipterygoid from temporal series, the supratemporal contacts the palatal complex. An occipital "button" the postparietal posteromedially in typical such as is present on the chamfered posterior temnospondyl design, and a crescent-shaped edge ofthe tabular of Greererpeton and Pholi- tabular extends laterally to embay the pos- dogaster is not exposed in the material at teromedial corner of the squamosal. On the hand. An occipital expansion of the postpar- basis of material in which the cheek had sep- ietal and tabular, as characteristically devel- arated from the skull table (AMNH 6915, fig. oped in trimerorhachoids (Coldiron, 1978), 6; 6932, fig. 7A), Romer (1930) interpreted is absent in all colosteids. this scallop as the lower rim of an otic notch. The preservation of the skull table-cheek This misconception was reinforced by the at- junction ranges from undisrupted to com- tribution of AMNH 6918, a specimen with pletely detached. The latter state is not in- 12 AMERICAN MUSEUM NOVITATES NO. 2770

sa- qj ..W~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Z

FIG. 5. Colosteus scutellatus (Newberry). Postorbital region of adult skull and partial mandible. Paratype of Macrerpeton deani, AMNH 6826. Scale equals 1 cm. dicative of a partially kinetic skull (A. R. Mil- corner, the quadratojugal has a distinct flange ner, 1980a), but simply the poorly sutured that continues posteriorly onto the occipital condition of immature specimens. surface. Although the actual quadrate-quad- The squamosal is a large bone that overlaps ratojugal suture, as well as a paraquadrate the edge of the skull table from the posterior foramen, is not preserved in any specimen, skull margin to about the level of the pineal the quadratojugal appears to contribute to the foramen. Laterally, the squamosal joins the jaw articulation in a manner similar to that R. quadratojugal, a subrectangular element that of Eugyrinus (A. Milner, 1980a). forms the greater part of the lateral wall of PALATE the subtemporal fossa. Like the maxilla, the marginal portion of the quadratojugal is There is an unfortunate lack of reasonably weakly sculptured. Towards its posterolateral well-preserved specimens to furnish infor- 1983 HOOK: COLOSTEUS SCUTELLA TUS NEWBERRY 13

A B FIG. 6. Colosteus scutellatus (Newberry). A, B, postorbital region of subadult skull and mandible. Counterpart blocks of AMNH 6915. Scale equals 1 cm. mation on the structure of the palate and whereas the palatal flange of the premaxilla braincase. No single specimen approaches makes a minor contribution to the anterior completeness, most are subadult, and all are portion of the opening. This roughly circular disarticulated. These inadequacies, com- outline of the choana is unlike the elongate bined with the erroneous assumption that or irregular openings commonly possessed by Colosteus and Erpetosaurus were members primitive amphibians. ofthe same family, led Romer (1930, p. 104) In addition to the choanae, the anterior to assert that the palate of Colosteus was region of the palate is perforated by a second similar "in most respects" to the relatively set of openings that would have received the well-represented palate of Erpetosaurus. elongate parasymphysial tusks upon closure However, study of all palatal materials at- of the jaws (AMNH 6951, fig. 8). These an- tributable to either form indicates that pre- terior palatal fenestrae are close-set along the vious suppositions are unfounded and that midline and encircled by the premaxilla an- marked differences exist. In short, the palate terolaterally and the vomer posteromedially. of Colosteus (fig. 1B) is characterized by an- Although this construction differs from the terior palatal fenestrae, a formidable com- single median fenestra of ichthyostegids plement ofmarginal tusks, denticulated pter- (Jarvik, 1980) and some loxommatids (Beau- ygoids enclosing long but relatively narrow mont, 1977), it resembles the paired condi- interpterygoid vacuities, and an unfused bas- tion common to most trimerorhachoids al articulation. (Chase, 1965). The medial border of the choana is shared The margin of the palate is bounded an- equally by the vomer and palatine. As re- teriorly by the massive tusk-bearing premax- stored from the slightly disrupted left side of illa. As preserved on CM 25307 (fig. 7B), the AMNH 6932 (fig. 7A), the lateral margin of posterolateral portion of this bone is expand- the choana is enclosed mainly by the maxilla, ed beneath and thickened in front of the ex- 14 AMERICAN MUSEUM NOVITATES NO. 2770

40, Al 0 a 0

FIG. 7. Colosteus scutellatus (Newberry). A, postorbital region of subadult skull and mandibles with disarticulated branchial arches. Paratype of Sauropleura newberryi, AMNH 6932. B, anterior portion of subadult skull and mandibles. Right counterpart block of CM 25307. Scale equals 1 cm. ternal naris to accommodate the tusk and arate the anterior palatal fenestrae. Poste- replacement pit. There are no specimens that riorly, along the midline, the medial corner delimit the medial extent ofthis palatal flange. of the vomer is extensively underlain (over- The maxilla forms much ofthe lateral mar- lain in ventral view) by a robust, finger-like gin of the palate. At the level of the choana, anterior process of the pterygoid. Crushing the ventral edge of the maxilla is expanded has revealed that the vomer-palatine suture, slightly to form the lateral rim ofthe opening. a diagonal contact directed posteriorly to- Posteriorly, this element probably contrib- ward the midline, is also constructed in a utes to the lateral wall of the subtemporal stepped manner. The area of the vomer that fossa. would support small palatal tusks, as seen in Two specimens (AMNH 6951, fig. 8; CM Greererpeton, is not exposed. A small but dis- 25307, fig. 7B) provide information on the tinct foramen appears in the central portion structure of the vomer. A thin anterior pro- of the bone; close scrutiny of Greererpeton cess extends medially from the main body of (Cleveland Museum of Natural History each bone to contact the premaxilla and sep- 1 1068) reveals the presence ofthe same struc- 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 15

ture. In Edops, this opening is thought to be I pm associated with a branch ofthe palatine nerve (Romer and Witter, 1942). The sutures between the remaining dermal bones of the palate-palatine, ectopterygoid and pterygoid-are indiscernible. Presum- ably, there was an extensive longitudinal con- tact shared by the pterygoid and the two lat- eral elements; that this area is not well preserved in subadult specimens of either Colosteus or Greererpeton is attributable to the thin nature of the palate in this area. Although the tusks of the palatine and ec- topterygoid are visible in several specimens, the palatal surfaces of these bones are poorly exposed. Immediately behind the choana, the palatine supports a large tusk pair, whereas the posterior half of the bone bears a row of smaller accessory teeth. Exposure ofthe dor- sal surface ofthe left palatine on AMNH 6915 (fig. 6B) indicates that it was overlapped pos- FIG. 8. Colosteus scutellatus (Newberry). An- teriorly by the ectopterygoid in a well-devel- terior halfofsubadult skull and mandibles in ven- oped tongue-and-groove suture. The palatine tral view. Left counterpart block of AMNH 6951. appears to have had little dorsal development Scale equals 1 cm. against the marginal elements ofthe skull roof. What appears to be the socketed ventral surface of the right ectopterygoid is exposed The width of the flat palatal ramus of the between the jaws of the holotype specimen pterygoid increases gradually from the nar- (AMNH 6916, fig. 2). This bone apparently row anterior process to the anterior border supported a tusk pair anteriorly at about the of the subtemporal fossa. At the latter level, mid-orbit level and a row of accessory teeth the medial portion of the bone is thickened posteriorly near the subtemporal fossa. The dorsally and recurved medially to form a tri- form of the anterior margin of the fossa and angular basal process (internal process of the development of the ectopterygoid in this Watson, 1956). The morphology ofthe basal area remains unknown (contra Coldiron, articulation, as judged from the partial me- 1974). dial exposure ofthe right pterygoid on AMNH The morphology of the pterygoid agrees in 6915 (fig. 6B), conforms to that of Greerer- all respects with that of Greererpeton. In con- peton (Smithson, 1982, fig. 13). A well-de- trast to Erpetosaurus and other trimero- fined concavity in the posterior wall of the rhachoids, the pterygoids of Colosteus join at basal process represents the conical recess into the midline, in front ofthe interpterygoid va- which the basipterygoid process ofthe brain- cuities, and extend forward to the postero- case inserts (Romer, 1947). This socket is medial corners of the vomers; thus, the in- essentially a U-shaped notch in the medial terpterygoid vacuities are completely enclosed wall of the central body; it is bounded ante- by the pterygoids. In addition, the sides of riorly and ventrally by the posteroventral the openings show little lateral expansion, flange of the basal process, and posteriorly unlike the more lyrate outlines exhibited by by a vertical process arising from the nar- most temnospondyls. Further analysis of the rowed ventral surface of the pterygoid. The proportions of the interpterygoid vacuities is low ridge present on the lateral wall of the inappropriate since their development is conical recess of Greererpeton (Smithson, largely size dependent and they are only ex- 1982) is not preserved in any specimen. Like- posed in subadults. wise, the dorsal portion of the central body 16 AMERICAN MUSEUM NOVITATES NO. 2770 is not exposed and a pterygoid-epipterygoid parasphenoid that apposed the underside of suture is not visible. the braincase is exposed only in AMNH 6915 Posterior to the basal articulation, the ver- (fig. 6), but is so poorly preserved that de- tically oriented quadrate ramus of the pter- scription is not possible. What appears as a ygoid continues posterolaterally to form the parasphenoid body associated with the left internal wall of the subtemporal fossa and pterygoid ofAMNH 6932 (fig. 7A) is actually contact the squamosal, tabular, and quad- the severely distorted posterior half of the rate. Immediately behind the conical recess, right pterygoid. an elongate depression develops in a hori- The predominantly subadult nature of the zontal plane above the rounded medioven- specimens at hand and coincident lack of en- tral lip of the quadrate ramus. This elongate dochondral ossification make identification pocket, commonly known as the excavatio ofbraincase structures tenuous at best. How- tympanica because of its supposed associa- ever, the situation is immeasurably im- tion with a middle ear cavity (Bystrow and proved in Greererpeton and the reader is re- Efremov, 1940), has been reinterpreted re- ferred to Smithson (1982) for a detailed cently as the homologue ofthe spiraculo-hyo- account of the colosteid braincase and asso- mandibular recess of rhipidistian fishes ciated structures. (Smithson, 1982). On the basis of Greerer- The lower half of the sphenethmoid is ex- peton, Smithson has proposed that this area posed in lateral view between the parted jaws ofthe quadrate ramus braced the ventral por- of the holotype specimen (AMNH 6916, fig. tion of the stapes which, in turn, supported 2). As reconstructed, the anterior end of this the braincase. Unfortunately, Colosteus can- structure terminated at about the level of the not be used to evaluate this hypothesis since anterior orbital margin. The structural rela- no stapes can be identified among the re- tionship between the sphenethmoid and oti- mains. co-occipital components cannot be deter- The remainder of the quadrate ramus and mined. its contact with the skull roof and quadrate Resting between the lateral margin of the are generally difficult to interpret. Only the skull table and right cheek of AMNH 6915 inturned posterior edges ofthe squamosal and (fig. 6) is an amorphous mass of unfinished tabular on the left side of AMNH 6932 (fig. bone. The outline of the left basipterygoid 7A) provide evidence that the construction process can be seen in dorsal aspect at the ofthis region does not differ significantly from broadened posterior end of the cultriform Greererpeton (Smithson, 1982, fig. 12). A dis- process, just opposite the parietal-supratem- crete quadrate ossification is not exposed on poral suture. Adjacent to this area, a conical any specimen; hence, the configuration ofthe structure bearing a concave face at its ex- jaw articulation remains unknown. panded end represents the remnants of the basioccipital. A second, much larger speci- PARASPHENOID AND BRAINCASE men (AMNH 6826, fig. 5) provides a glimpse of what is an Little can be said of the detailed mor- probably exoccipital. phology of the parasphenoid from the frag- mentary remains present. The cultriform MANDIBLE process conforms to the primitive temno- Each mandibular ramus can be divided into spondyl form ofan elongate, narrow structure three regions: a swollen posterior end that with a V-shaped cross section (Romer, 1947). accommodates the glenoid and adductor fos- On AMNH 6951 (fig. 8), the displaced an- sae, a medially fenestrated midsection that terior end of the process is reduced to a splint supports coronoid and dentary tooth rows, of bone that appears to have been sand- and a bluntly rounded anterior portion that wiched ventrally by the pterygoids at their bears the characteristic labial notch and large forward confluence. If the cultriform process parasymphysial tusks. Anteriorly, the rami reached the vomers dorsally, the contact are firmly sutured together at the jaw sym- would have been meager. physis. In lateral aspect (fig. 1C), the jaw ar- The expanded posterior section of the ticulation is in line with the tooth row and 1 983 HOOK: COLOSTE US SCUTELLA TUS NEWBERRY 17

* Sl'' ' ''1\~~~~~~~~t'^'; sa-- pa.f. -

FIG. 9. Colosteus scutellatus (Newberry). Posterior portion of adult mandible in medial view; most of prearticular is absent to expose medial surface of angular. Holotype of Macrerpeton deani, AMNH 2934. Scale equals 1 cm. the evenly rockered ventral outline attains its ally, and a lower splint that tapers against the greatest depth at the level ofthe anterior limit postsplenial posteriorly. Between the sym- of the adductor fossa. physis and the fenestra, the main body of the The dentary is the dominant bone of the splenial is perforated by two small foramina. lower jaw and contacts all other mandibular The splenial-prearticular suture above the dermal elements except the prearticular. Pos- Meckelian fenestra cannot be determined with teriorly, the dentary contributes to the an- certainty. On AMNH 6945 (fig. 3A), the terolateral border ofthe adductor fossa. Over splenial appears to end abruptly, but is ap- most ofits dorsal exposure, the bone consists posed internally by an anterior extension of of a raised tooth-bearing platform that is the prearticular. Following this interpreta- stepped lingually to adjoin the coronoid shelf. tion, the reduced lingual wall of the Meck- As seen on AMNH 6945 (fig. 3A), the width elian space consists almost exclusively of a of the platform varies with the diameter of considerably elongate prearticular. Poste- the alveoli. At the anterior end of the tooth riorly, where it encloses the adductor fossa, row, the dentary curves toward the midline the prearticular appears slightly thickened but and expands to contain the sizeable alveoli is thin elsewhere. A fairly large, ovoid para- of the parasymphysial tusks. Externally, at articular foramen (Romer and Witter, 1942) this level, the bone approaches its maximum for the chorda tympani is situated on the height and displays a striking U-shaped notch prearticular-articular suture (AMNH 2934, that afforded passage to the premaxillary tusk fig. 9). Although not completely preserved on upon closure of the jaws. The strongly punc- any specimen, the prearticular-angular suture tate anteromedial face of the dentary com- probably spanned the medial jaw surface be- prises at least three-quarters of the symphy- tween the articular and the Meckelian fenes- sial surface. tra. The splenial forms the anterior one-third The sutures subdividing the coronoid se- of the ventral jaw margin and makes a minor ries cannot be traced and the tusk-bearing contribution to the symphysis as it joins its anterior coronoid of Greererpeton is not pre- fellow below the dentaries. The elongate served. A narrow, elongate bone possessing Meckelian fenestra (foramen intermandibu- a single row of very small teeth, visible on laris communis of Heaton, 1980a) divides AMNH 6915 (fig. 7) and 6945 (fig. 3A), is the posterior halfofthe splenial into an upper undoubtedly a coronoid element. Although expanse that shoulders the coronoids later- its morphology differs markedly from the 18 AMERICAN MUSEUM NOVITATES NO. 2770 broad, densely denticulated coronoid pave- distinct condyloid processes. From above, the ment of Greererpeton, it is very similar to depression is an anteriorly open crescent that of ichthyostegids (Campbell and Bell, whose tangent is perpendicular to the skull 1977; Jarvik, 1980) and Doragnathus midline. In contrast to the majority of Car- (Smithson, 1980b). The only other coronoid boniferous amphibians, the jaw articulation segment is preserved on AMNH 2934, the is not raised or depressed relative to the tooth posterior third ofa very large mandible. Here, row, but virtually on the same horizontal the same general morphology persists, but the plane. minute teeth occasionally occur two abreast. In addition, the posterior edge of the coro- noid encloses only the anteriormost part of DENTITION the adductor fossa. No coronoid process is A complete description of the dentition is developed. not possible because the palatal margins and The sutures separating the splenial, post- lower jaws are incompletely exposed. In at- splenial, angular, and surangular form a pos- tempting to reconstruct dental batteries, con- teriorly directed en echelon pattern from the servative estimates were favored; therefore, ventral edge of the dentary to the bottom of it is likely that some tooth counts were ac- the mandible. On the lateral face, the tually greater than those shown in the res- postsplenial extends far anteriorly, nearly torations. Moreover, since postmortem reaching the posterior lip ofthe dentary notch. crushing has dislocated and distorted most This condition is more like that of Pholido- of the large teeth, their appearance in the gaster (Panchen, 1975, fig. 15) than Greerer- somewhat stylized reconstructions differs peton and may indicate that the antorbital from that of the specimen drawings. region ofeven the largest Colosteus skull was The marginal teeth of the premaxilla and proportionally shorter than that of Greerer- maxilla are simple acuminate pegs in lateral peton. The angular has an extensive lateral aspect ("bullet shaped" of Smithson, 1980b, exposure and wraps around the ventral mar- p. 920). It is difficult to say how tooth re- gin of the jaw to contact the prearticular and placement occurred since some marginal teeth enclose the posteroventral corner of the in- have undoubtedly fallen away. Where the termandibular fenestra. bases of the massive palatine tusks impinge In form and sculpture, the surangular re- upon the internal surface of the maxilla, the sembles the quadratojugal. Not only do both tooth row is discontinuous, as in Caero- elements enclose the outer face of the ad- rhachis (Holmes and Carroll, 1977); since the ductor chamber, but they also contribute to presence of this feature cannot be verified in the lateralmost portions of the jaw articula- adult material of either Colosteus or Caero- tion surfaces. The dorsal outline of the sur- rhachis, it is possible that this condition is angular is raised slightly above the upper sur- size-related. There is an increase (about 40%) face of the dentary in a subtle convexity. in tooth length anteriorly along the maxillary Crushing on AMNH 2934 has plainly re- row, but no discrete canine peaks. The max- vealed that the surangular constitutes about illary tooth count cannot be estimated with one-fourth ofthe jaw articulation surface; the any confidence because no specimens expose same holds true for Greererpeton (Smithson, more than a small portion of the tooth row 1982), Doragnathus (Smithson, 1980b), and and the posterior extent of the series is un- possibly Caerorhachis (see Holmes and Car- known. roll, 1977, fig. 5). The presence of even a The marginal teeth of the premaxilla are modest retroarticular process on Colosteus smaller than the anterior maxillary teeth. On appears unlikely. the premaxilla, there is room for a continuous The articular occupies the posterior por- row of 10 teeth anterior to the premaxillary tion of the mandible between the lateral and tusks. An isolated, smaller marginal tooth is medial dermal elements. As best represented usually located on the posterolateral corner by AMNH 2934, the glenoid fossa of the ar- of the palatal flange of the premaxilla. ticular appears as a simple trough between The occurrence of tusk pairs on the pre- 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 19 maxilla and palatine conforms to the stan- ofdentary and maxillary teeth from the same dard replacement pattern of a single tusk ac- level is about 5: 1. The dentary tooth row spans companied by a replacement pit. Some the distance from the labial notch to the ad- specimens (AMNH 6915, fig. 6B; 6951, fig. ductor fossa and affords space for an esti- 8) possess an intermediate condition in which mated 21 ± 2 teeth. both alveoli of the palatine are occupied by A sequence ofminute coronoid teeth is sit- tusks of disparate sizes. The length of the uated along a shelf below the dentary tooth premaxillary tusk is usually two and one-half row, on the lingual side ofthe mandible. An- times its basal diameter and about 10 percent teriorly, this presumably replaceable series is shorter than the palatine tusk. In labial as- limited to a single row, but posteriorly, near pect, the tusks ofboth bones appear as evenly the adductor fossa, it bears irregularly placed, tapered cones with sharply defined tips. side by side tooth pairs. Nowhere do the cor- The palatine also bears a series of at least onoids display the shagreen of denticles ex- three alveoli posterior to the tusk pair. These hibited by Greererpeton. alveoli housed accessory teeth that were two The development of denticles on the pal- to three times the length ofthe marginal teeth. atal surface is limited to the palatal ramus of On small specimens (AMNH 6932, fig. 7A), the pterygoid. Furthermore, the denticulated the alveoli are simple depressions that lack platelets commonly associated with the pal- the coarse, radiating ridges characteristic of ates ofother Linton temnospondyls (e.g., Er- larger replacement pits. petosaurus, Saurerpeton, Stegops) are not As previously noted, the small teeth pres- present in Colosteus. ent on the vomers of Greererpeton are not preserved in any specimens of Colosteus. However, the general agreement between the DERMAL SCULPTURE AND anterior portions of the palates of these two LATERAL LINE CANALS formsjustifies the reconstruction ofsuch teeth The dermal sculpturing ofthe skull roofand in Colosteus. There are clearly no enlarged lower jaws has been reproduced in all spec- vomerine tusks present in any colosteid. imen drawings. The predominant pattern, as The ectopterygoid appears to support at described originally by Romer (1930, p. 103), least two tusks of intermediate size that re- is one of "pits in the center of each element, semble the general dagger-like form of other from which ridges, bifurcating, radiate to- palatal tusks (BM[NH] R.2547, fig. 3B). The ward the periphery. The enclosed valleys anterior ectopterygoid tusk, situated just be- deepen centrally, where a pore is present." hind the palatine-ectopterygoid suture, is the In the absence of a sufficient number of larger of the two, with a length equal to ap- reasonably complete specimens, the devel- proximately two-thirds that of the premax- opment of the dermal sculpture provides a illary tusk. Although not exposed on any subjective means by which proportional specimens, it is likely that this tusk is asso- changes accompanying absolute growth may ciated with a replacement pit. At about the be discussed. More specifically, areas of the level of the jugal flexure, a second tusk pro- skull that exhibit a preponderance of elon- jects below the marginal tooth row. This gated pits are considered as zones ofintensive smaller tusk may be one of a series of acces- growth (Bystrow, 1935). At least three areas sory teeth borne by the posterior half of the of this type are evident in Colosteus: an ant- bone. orbital region posterior to the nares, a central The parasymphysial tusks of the lower jaw cheek region lying between the ossification are similar in both size and shape to the large centers of the jugal, squamosal, and qua- palatal tusks. They occur in the typical tusk- dratojugal, and a dorsal midline region ra- and-pit combination on each side ofthe sym- diating anteriorly from the pineal foramen to physis and are slightly recurved toward their about the level of the mid-orbit. The differ- apices. The other dentary teeth are less acu- ential growth of these zones indicates that minate but still of considerable size. On CM certain proportional changes are to be ex- 25307 (fig. 7B), the ratio between the lengths pected with an increase in size, namely an 20 AMERICAN MUSEUM NOVITATES NO. 2770

elongation of the snout, a broadening and jugal canal before forming the jugal flexure. deepening of the cheek accompanied by a The sulcus is usually "bridged over" several posterior shift in the jaw articulation, and a times in the vicinity of this right-angle bend. slight posterior migration of the pineal fo- Continuing anteriorly along the jugal-max- ramen. These trends are not unusual, how- illary suture, the infraorbital canal is flexed ever, for similar patterns have been noted in anteromedially at the level of the anterior a variety of other Paleozoic amphibians margin ofthe orbit to form a medially convex (Beaumont, 1977; Bossy, Ms; Boy, 1972; arch on the lacrimal. At the intersection be- Langston, 1953; Panchen, 1970; Romer, tween the lacrimal, maxilla and prefrontal, 1939; Watson, 1963). the canal turns laterally and meets the skull In addition to dermal sculpture, the skull margin at the level of the posterior rim ofthe surface is modified by a well-developed lat- choana. eral line system. Regardless of skull size, the The jugal sulcus is continuous from the sulci are expressed as continuous, steep-sid- posterior margin of the squamosal to the ed, flat-bottomed channels, not unlike the middle of the jugal, where it is joined to the grooves commonly found in other temno- infraorbital sulcus. spondyls; however, this condition contrasts The occipital commissure ofthe lateral line with the partially enclosed networks of Greer- system figured by Steen (193 1, fig. 6) cannot erpeton and Pholidogaster (Smithson, 1982; be traced on any specimens. Examination of Panchen, 1975). a latex peel ofher material (BM[NH] R.2547, The exact pattern of the lateral line system fig. 3B) reveals that the supposed sensory ca- is somewhat variable in the postorbital region nal is actually poorly preserved dermal sculp- and can even differ between opposite sides ture. ofthe same skull. The temporal sulcus is usu- The mandibular sulcus extends the length ally split into two short, subparallel branches ofthe jaw ramus and is continuous across the situated on the lateral third of the parietal. jaw symphysis. Posteriorly, the canal is su- These canals are connected anterolaterally perimposed upon the articular-surangular su- before joining the supraorbital and infraor- ture before turning inward slightly and con- bital canals in the center of the postorbital. tinuing anteriorly along the ventral rim ofthe Temporal grooves are not developed poste- jaw where it is frequently constricted or en- riorly as they are sometimes on the supra- closed. A faint longitudinal groove is present temporals of Greererpeton (Romer, 1969, fig. near the upper margin of the lateral face of 2). the dentary of some specimens (BM[NH] In its fullest development, the posterior part R.2547, fig. 3B; AMNH 6829). This may rep- of the supraorbital sulcus consists of a very resent a poorly developed oral sulcus of the short groove that is limited to the postorbital lateral line system. and conjoined to the temporal and infraor- bital canals posteriorly. After disappearing from the skull roof at the postorbital-post- BRANCHIAL ARCHES frontal contact, the supraorbital lyra is visible There are at least two subadult specimens again near the ossification center of the post- which exhibit remains ofthe partially ossified orbital; from this point the canal undulates branchial apparatus. AMNH 6932 (fig. 7A) anteriorly over the frontal, prefrontal, nasal, displays an association of three or four bran- and premaxilla. Close to the anterolateral chial elements among a disrupted patch of corner of the snout, the channel receives the scales. Two ofthese structures are subrectan- ethmoid commissure medially before ter- gular in outline with rough, unfinished sur- minating near the anterior border of the pre- faces and slightly expanded ends. These are maxillary tusk. probably hypobranchials. The identities of From the meeting of the temporal and su- the two more elongate structures are less ob- praorbital canals on the postorbital, the in- vious, but may be regarded as ceratobran- fraorbital sulcus extends anterolaterally across chials. Support for this interpretation is pro- the medial half of the jugal and receives the vided by AMNH 6917 (fig. 1 2A, B), a 1983 HOOK: COLOSTEUS SCUTELLA TUS NEWBERRY 21

.PC .. I.c , p

Z77.- j. F.. A B

FIG. 10. Colosteus scutellatus (Newberry). Posterior presacral vertebrae. A, central elements as pre- served on MCZ 2158; B, "exploded" restoration. Scale equals 1 cm.

specimen that exhibits an assemblage ofplain inadequate record of the axial skeleton and bladelike and minute toothed structures. a total lack ofinformation on the pelvic girdle When present in other juvenile temnospon- and hindlimbs. dyls, the toothed elements have been regard- AXIAL SKELETON: Although the uniqueness ed as gill rakers of the ceratobranchials (Ber- of the dermal scales aids in the identification man, 1973; Boy, 1972; Milner, 1982). of postcranial remains, their profusion fre- The significance of these partially ossified quently obscures the morphology of under- branchial elements is difficult to judge on such lying elements. Vertebrae, for example, are exiguous material. Although both specimens usually represented solely by bulges in the represent small subadults, comparably sized pavement of body armor. Consequently, individuals (e.g., AMNH 6915, 6916, 6945) much significant information, such as atlas- do not exhibit such remains. The retention axis construction, presacral length, and re- of branchial arches in adults cannot be ver- gional variation, remains inaccessible. ified because the pharyngeal region is not fa- The following description and reconstruc- vorably exposed on such specimens. Simi- tion of vertebral elements is based primarily larly, the mechanical techniques employed in on MCZ 2158 (fig. IOA). This specimen in- the preparation of Greererpeton largely pre- cludes parts of 14 ribs, 10 intercentra, and clude recognition of branchial remains. an undetermined number ofpleurocentra and neural arches, all exposed by postmortem disruption of the dorsal scalation. Direct POSTCRANIAL REMAINS comparison with articulated skeletons of The postcranial skeletons of Linton edo- Greererpeton indicates that the preserved area poids, loxommatids, and anthracosaurs are represents a far-posterior segment from the virtually unknown. This necessitates a degree dorsal series of a large individual. of conservatism in attributing isolated post- The general form ofthe vertebrae (fig. l OB) cranial materials to any particular labyrin- approaches the schizomerous condition first thodont genus. However, recognition of the elucidated by Romer (1964) in Pholidogaster. distinct colosteid scalation and study of the A moderately large notochordal space is par- similar postcranial skeleton of Greererpeton tially enclosed by a dorsally open intercen- facilitate the identification of postcranial re- trum and paired pleurocentra. In their crushed mains of Colosteus. Nonetheless, there is an and dissociated state, it is difficult to establish 22 AMERICAN MUSEUM NOVITATES NO. 2770

A B FIG. 11. Colosteus scutellatus (Newberry). A, anterior intercentrum, pleurocentrum, ribs, and inter- clavicle. Paratype specimen, holotype of Colosteus pauciradiatus, AMNH 6920; B, restored interclavicle and right clavicle in ventral view. Scale equals 1 cm. the proper alignment of the pleurocentra rel- ridges that denote the roots of the ascending ative to other vertebral components. Since processes. The periosteal field of the process all exposed edges of the pleurocentra are un- is deeply recessed except where it is gathers finished, however, it is probable that modest to form the anterior boundary of the protu- dorsal and ventral gaps were filled by carti- berant rib facet. This semicircular articular lage in life. This functionally diplospondy- surface is located about halfway up the pos- lous configuration is the same as that seen in terior side of the ascending process and is the more completely ossified Greererpeton. more pronounced anteriorly along the col- The intercentra of posterior presacral ver- umn, as well as in smaller individuals (AMNH tebrae conform to the general rhachitomous 6958, 7493, BM[NH] R.2664). design of a crescentic, ventrally robust husk, A single intercentrum is preserved in ven- with wedge-shaped, upturned sides. Ante- tral view on AMNH 6920 (fig. lIA). Al- riorly in the presacral series, the ascending though considerably smaller than those of processes extend farther dorsally than in the MCZ 2158, this element is noteworthy for its posterior sections of the column. Internally, buckled ventral midline and paired, postero- the bone surface is rough and unfinished. This laterally directed tubercles. These pro- texture continues around the edges and onto nounced bosses are located on the posterior the external surface, where it is modestly de- half of the base along the lateral ridges. Sim- veloped ventrally, but widely exposed on the ilar features are present on the anterior in- ascending process. Elsewhere, the external tercentra of Trimerorhachis (Case, 1935) and surface is covered with dense periosteal bone Greererpeton (personal observ.). that bears minute, irregular pits. The pleurocentra are developed to a greater The ventral surface of the intercentrum is extent than in typical rhachitomes (i.e., Ery- thickened near its anterior and posterior mar- ops; Moulton, 1974), but are not fused ven- gins to produce a generally concave profile. trally, as in Caerorhachis (Holmes and Car- A distinct longitudinal ridge is situated me- roll, 1977). Viewed laterally, the pleurocentra dially and is flanked by two less developed resemble half an inverted intercentrum: the 1 983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 23 margins are formed by wide, unfinished con- the vertebrae, the ribs are angled posteriorly tinuations ofthe internal surface and the peri- at about 600 to the column and curved mod- osteal field is concave. An articular facet is estly over their distal half. The most striking present on the anterior half of the pleurocen- costal feature is a stiletto-like uncinate pro- trum near its expanded dorsal edge. Like their cess located about halfway down the shaft. counterparts on the intercentrum, the round- This structure is well developed on all but ed side of this facet is enclosed by raised fin- the most posterior thoracic ribs. ished bone while the opposite face merges A nearly complete right thoracic rib is ex- with the unfinished border. Below the level cellently preserved in the counterpart blocks of the rib articulation, the sides constrict to of AMNH 6920 (fig. 1 IA). Between the ex- a blunt ventral point as the bone curves ven- panded proximal end and the uncinate pro- tromedially. cess, the rib is moderately constricted and a Most of the detailed morphology of the sharp keel is present along the anterodorsal neural arches has been obliterated by post- edge. Anterior to this keel, a deep groove is mortem crushing. Transverse processes, typ- apparent, but distorted by crushing. The un- ically separated from the pedicel at their roots, cinate process is borne on a posteriorly di- appear to be relatively long, broad structures, rected flange and is strengthened along its but a satisfactory view of the articular facet axis by a ridge that is contiguous proximally is not available. An articulated series of 10 with the shaft. Viewed medially, the posterior posterior presacral vertebrae is preserved on portion of the process is distinctly grooved; AMNH 6957. The anterior zygapophyses are as suggested by the anterior-most exposed tilted dorsomedially to receive the ventro- ribs of AMNH 6917, this groove may have laterally oriented posterior zygapophyses. The rested against the leading edge of the subja- retention of the supraneural canal, as report- cent rib shaft to form a slight costal overlap. ed in Greererpeton (Smithson, 1982), cannot Distal to the base of the process, the rib con- be verified. sists of a shaft that terminates in an unfin- Steen (193 1) cited two subadult postcranial ished expansion. specimens from Linton that are almost cer- A series of 14 partially exposed thoracic tainly attributable to Colosteus. BM[NH] ribs is visible on the right side ofMCZ 2158. R.2660, formerly identified as Erpetosaurus Anteriorly in this sequence, short uncinate (Steen, 1931), is assigned to Colosteus on the processes and broad, unfinished terminal ends basis of the morphology of the interclavicle are present. The brevity of the posterior and scales. Likewise, BM[NH] R.2664, con- flanges and uncinate processes precludes the sidered to be an indeterminate "phyllospon- possibility ofoverlap. After the sixth exposed dyl" by Steen (1931, p. 855 as "R. 2644" in rib, the distal ends appear finished and acu- error, pl. 4, fig. 2), is a juvenile specimen of minate, and uncinate processes are absent. A Colosteus featuring dorsal and ventral squa- rapid shortening is exhibited along the pos- mation and an articulated string of several terior half of the series, with the last exposed vertebrae. In considering the vertebrae ofEr- rib measuring only 25 percent the length of petosaurus, Watson (1956, p. 375, fig. 28) the fifth. The absence of ribs on the 10 artic- reviewed this material and concluded "that ulated vertebrae ofAMNH 6957 suggests that all the elements are paired, none fused into ribs were not developed in the posterior por- crescent-shaped intercentra." This bipartite, tion of the presacral series. unossified condition of the intercentra is ob- Because of the lack of specimens from the viously a juvenile feature. pelvic and caudal regions, no information on The salient features ofthe thoracic ribs can sacral ribs or haemal spines can be reported. be seen on AMNH 6917 (fig. 1 2A), 6920 (fig. APPENDICULAR SKELETON: The most com- lIA), and MCZ 2158 (fig. lOA). Although plete representation of the dermal shoulder the proximal ends are imperfectly exposed, girdle is that of AMNH 6917 (fig. 12A, C). the articulating heads of the anterior ribs ap- As reconstructed in figure 11 B, the shape of pear to be differentiated into discrete capit- the pectoral girdle is somewhat similar to that ular and tubercular facets. When joined to of other aquatic temnospondyls, such as tri- 24 AMERICAN MUSEUM NOVITATES NO. 2770

FIG. 12. Colosteus scutellatus (Newberry). Anterior postcranial skeleton. A, dorsal view; B, cerato- branchial ossifications (cb in A) enlarged; C, ventral view; D, restoration ofleft manus. Paratype specimen, counterpart blocks of AMNH 6917. Scale equals 1 cm. except in B. merorhachids or metoposaurs. The clavicles poorly represented and were probably not ex- meet anteriorly below the interclavicle to form tensively ossified. a broad dermal shield that extends forward The largest and best preserved interclavi- under the skull to about the level of the basal cle, AMNH 6920 (fig. llA), resembles an articulation. The width of the shield is great- anteriorly directed spade with a very short est across the lateral corners of the clavicular posterior lappet. Ifthe interclavicle is divided plates and is equal to the maximum breadth longitudinally into three parts ofequal width, of the skull. The cleithral process of the clav- the lateral thirds are gently upturned to effect icle arises from the ventral shield at a low a subtle concavity (in dorsal view). The angle to join the cleithrum. The endochon- smooth dorsal surface shows only faint, ir- dral components of the shoulder girdle are regular grooves that may represent blood ves- 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 25

II 11 IV

III

D

FIG. 12. Continued. sel channels (Holmes, 1980). In contrast, the sternal process, as typically developed in em- ventral face exhibits a teardrop-shaped cen- bolomeres, is not present; rather, a tongue- tral area bearing a dermal sculpture of fur- shaped, medially recessed lappet is present. rows and bifurcating ridges. The pattern of The holotype specimen (AMNH 6916, fig. ornamentation indicates that the center of 2) includes an interclavicle that is about half ossification was situated along the midline, the size of the interclavicle of AMNH 6920. about two-thirds the length of the interclav- In comparison, the sculpture of the type is icle from the leading edge. The periphery of less distinct and the longitudinal grooves the sculptured field merges with faintly striat- leading to the fimbriated edge are more pro- ed, distally thinning extensions that apposed nounced. The clavicular facets also appear to either the clavicles or gastralia. Where the be more extensive on the smaller specimen. beveled margins of the two anterior quad- The type specimen of Colosteus foveatus rants join, a pectinate texture is developed (AMNH 6919), referred tentatively to Er- which extends to form the fimbriated leading petosaurus by Romer (1930), is an isolated edge of the interclavicle. Posteriorly, a para- interclavicle. As seen in ventral view (Cope, 26 AMERICAN MUSEUM NOVITATES NO. 2770

1875, pl. 36, fig. 1), the sculptured ventral appearance of the distal half of the left hu- surface is markedly embayed by clavicular merus indicates that the original mold was facets and finely sculptured by reticulate or- thoroughly pyritized, a condition usually as- namentation. Preparation of the dorsal sur- sociated with dense cartilaginous tissue at face has revealed a distinct sternal trabecula. Linton. Proximally, the preservation of the These features indicate that this specimen is humerus improves and some details of the attributable to Saurerpeton. dorsal surface are discernible. A small tuber- Each clavicle consists ofa flat ventral plate osity is developed near the incompletely os- and an ascending cleithral process. The ven- sified proximal end. A similar process, found tral surface of the clavicular plate and the in a variety of other primitive forms, includ- posterolateral face of the ascending process ing Greererpeton, Proterogyrinus, and Nel- bear the same type of sculpturing as the in- dasaurus, is generally regarded as the pro- terclavicle. The pattern indicates that the cessus latissimus dorsi (after Miner, 1925). center of ossification is at the lateral corner Anterolateral to this tuberosity, the outline of the plate, near the base of the ascending of the anterior flange (Holmes, 1980) is pre- process. Internally, faint ridges radiate out- served, but no other features of the humerus ward from the center of ossification to the are exposed. medial edge of the plate, where additional The radius and ulna are only slightly short- striae are also developed. er than the humerus as preserved. Otherwise, The thin medial edge ofthe clavicular plate the epipodials are unremarkable structures is slightly embayed where it abuts against the whose hollowed ends indicate that they were sculptured face of the interclavicle. Con- finished in cartilage. The radius is expanded versely, the anterolateral margin is thickened at both ends, with the distal end wider and and virtually straight. This thickened edge flatter than the proximal. Other surface mor- develops posterolaterally into a low ridge that phologies are masked by crushing. In its in- joins the posterior margin of the plate and completely ossified state, the length of the curves dorsolaterally to form the vertical ulna is about equal to that of the radius. A cleithral process. Although crushing has dis- modest distal expansion is visible, but the torted the attitude of this rodlike structure, proximal end is obscured by the radius and it is apparent that the process was inclined scales. posterolaterally at a relatively low angle to There are no carpals preserved with either the clavicular plate. forelimb. The carpus was probably unossified Other elements of the pectoral girdle are in subadults, if not mature individuals. poorly represented by the material at hand. The only element missing from the right On AMNH 6917 (fig. 12C), a curious trian- manus is the metacarpal I. The hand, there- gular element is partially exposed near the fore, can be restored confidently with a pha- right clavicle. Its surface is finished but un- langeal count of 2, 2, 3, 3; digit III has the sculptured, and bears a distinct groove on its greatest length, followed by IV, II, and I (fig. long side. This structure probably represents 12D). Although the three preserved meta- the expanded dorsal blade ofthe right cleith- carpals are virtually identical, the proximal rum in lateral aspect. No remains of the phalanges ofthe middle digits are longer than endochondral shoulder girdle are present. their counterparts on I and IV. Each digit Considering the feeble ossification of such ends in an elongate, bluntly pointed ungual elements in Greererpeton (the scapulocora- phalanx. coid figured by Romer, 1969, is actually a No pelvic remains can be attributed con- humerus), it is likely that these structures were fidently to Colosteus. also reduced in Colosteus. A single specimen, AMNH 6839 (fig. 13), Knowledge of the diminutive forelimb is figured by Cope as Pleuroptyx clavatus (1875, obtained exclusively from AMNH 6917 (fig. pl. 44, fig. 2a), preserves a partially articu- 12). A small portion of the distal ventral sur- lated hindlimb that is similar to that of Greer- face of the right humerus is insufficiently ex- erpeton. However, since no colosteid scales posed for description. The rather amorphous accompany this material, and such a design 1 983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 27

FIG. 13. Indeterminate temnospondyl. Femur, tibia, fibula(?), metatarsals, and phalanges. Paratype of Pleuroptyx clavatus, AMNH 6839. Scale equals 1 cm. [Drawing by P. Gaskill, courtesy of R. L. Carroll.] may well be generalized for temnospondyls, dorsal "scales" of Ophiderpeton (Baird, 1964, this otherwise informative specimen must be fig. 1C). The scales of the pharyngeal region regarded as indeterminate pending improved are variable in form (like those of Trimero- understanding of other forms. rhachis; Olson, 1979), but are probably mod- SCALES AND OSTEODERMS: Previous de- ified from the ventral series. scriptions of tetrapod "armor" have em- Both the dorsal and ventral scales usually ployed a variety of terms, often colloquially, possess distinctly crenulate to punctate pos- to denote both epidermal and dermal struc- terior borders. On the basis of AMNH 6917 tures. For the present purposes, the term scale (fig. 12), Romer mistakenly ascribed this refers to an ossification that was probably scale-type to Macrerpeton (1930, fig. 22). The derived from mesenchyme within the der- crenulations are expressed fairly uniformly in mis; "squamation" is a collection of scales the dorsal squamation, but are often reduced and "gastralia" are specifically those scales to simple grooves near the ventral midline, of the ventral surface of the body. "Osteo- and posteriorly in the gastralia (OSU 4481, derms" are structures which were formed in UMMP 3573). Similar ornamentation is the superficial dermal layers directly under- present on some ventral scales of Greerer- neath the epidermis; "squamulae" (Romer peton (Carroll, 1980, fig. 2), but differences and Witter, 1941) are the concentrically ar- in types of preservation and preparation in- ranged bony bands that comprise a particular validate direct comparison. type of osteoderm. In spite of the etymology The configuration of the gastralia is excel- ofthe animal in question, the use ofthe terms lently preserved in AMNH 6917 (fig. 12C). "scutes" and "scutellae" is avoided in that A series of highly sculptured scales rabbet these usually connote cornified epidermal onto the posteroventral face of the interclav- structures of a sort not seen in Colosteus. icle. Commencing at the posterior lappet of Excepting the skull, mandibles, pectoral this bone, an anteriorly directed chevron is girdle and limbs the entire body is clad in formed at the midline by opposing, postero- one of four types ofdermal ossifications. The laterally trending rows of scales. The number most extensive are the gastralia which cover of scales in each row increases from seven the underside of the body and continue onto near the shoulder girdle to 11 or more near the lower halfofthe flanks. Resembling these the pelvis; if one assumes that the body cir- in appearance, but less widespread poste- cumference does not increase significantly, riorly, are the dorsal scales. The remaining the contribution of the ventral squamation body surfaces-the upper flanks-are cov- to the dermal covering ofthe flanks increases ered by two varieties of osteoderms: ovoid caudally. Excluding those scales subjacent to forms similar to the "scales" of Trimero- the pectoral girdle, the anterior and medial rhachis (Colbert, 1955), and minuscule, gran- sides of each ventral scale are internal (over- ular aggregates similar to the pebble-shaped lapped in ventral perspective) to its neigh- 28 AMERICAN MUSEUM NOVITATES NO. 2770 bors. Further, because the rows are staggered tively wider and thinner dimensions of the at the midline, an interlocking pattern is former. Equally evident is a variation in the formed along the apex of the chevron. form ofthe medial articulating facet. On dor- The outline of individual ventral scales is sal scales, the lateral side of the notch is vir- essentially that ofa rhomboid with large acute tually straight, whereas the same feature on angles. As seen in AMNH 6958 and MCZ ventral scales is semicircular; this difference 2158, the internal surface bears a distinct lat- reflects the more angular posterolateral cor- eral recess which receives the medial flange ner of dorsal scales. ofthe adjacent scale. The anterior and medial Romer (1972) noted that the dorsal scal- edges are very thin, the former evenly bev- ation of Greererpeton consists ofan extensive eled and the latter abruptly notched from the mosaic of small, subcircular scales. In Co- thickened middle of the scale. It is the pos- losteus, a few ossifications of this type are terior portion of this somewhat inflated area present in the pharyngeal region and the lat- that is variously sculptured to produce the eral extremes of the dorsal squamation. Ex- distinct crenulations of the trailing edge. All amination ofRomer's material indicates that of these features become less apparent along dorsal rhomboids, as commonly seen in Co- the lateral extent of the ventral squamation; losteus, are totally absent in Greererpeton. thus, the ventrolateral scales are generally thin Unfortunately, a definitive account of the and unsculptured with rounded corners. Ad- scalation of Pholidogaster is lacking. ditional variation is seen on the scales that The remaining dermal ossifications are form the apex ofthe chevron; the medial end considered to be osteoderms on the basis of of these scales is usually enlarged or bent at Romer and Witter's (1941) work on the squa- an angle to the long axis of the scale to effect mation of Eryops and Olson's (1979) inves- the interlocking pattern of the midline. tigation ofthe integument of Trimerorhachis. The extensive dorsal squamation is pre- Little can be said about the disposition of served best on AMNH 6917 (fig. 12A). In either type ofosteoderm, other than that they general arrangement and individual form, occur together on the otherwise barren flanks, these scales are remarkably similar to the gas- between the posteriorly directed ventral and tralia: not only is the manner ofoverlap iden- dorsal series. The larger osteoderms (greatest tical, but the dorsal scales are also arrayed in diameter of 4.5 mm.) are formed by ex- anteriorly directed chevrons. Because the tremely thin squamulae and usually occur in dorsal midline is not intact on any specimen, patternless layers. In this respect, they resem- it is difficult to determine if the opposing di- ble the general "cycloid scale" morphology agonals met as seen in the ventral squama- (Colbert, 1955) and erratic arrangement (Ol- tion. There is no evidence to suggest that the son, 1979) of osteoderms in Trimerorhachis. dorsal scales were in any way joined to, or The second type ofosteoderm is a tiny (great- supported directly by the vertebrae. est diameter of0.4 mm.), subcircular granule On the basis of three specimens represent- that usually occurs in aggregates (see isolated ing various levels along the vertebral column, scales near symphysis of right mandible on it appears as though the lateral extent of the BM[NH] R.2547, fig. 3B). On small individ- dorsal squamation diminishes posteriorly as uals (AMNH 6917), these ossifications are the width of the ventral field increases. From poorly formed and appear as undefined, rough AMNH 691 7, a count ofseven or eight rhom- surfaces. Where the dorsal scales are reduced, bic scales per diagonal is estimated; in MCZ both types of osteoderms comprise the great- 2158, an approximation of five per row is er portion of the dorsal covering. established; and on AMNH 6957, no discrete The gastralia of Linton urocordylids, par- rows are present and a total of only eight ticularly Ctenerpeton remex (Moodie, 1916, irregularly shaped dorsal scales is observed. pl. 23, fig. 2), have been confused with the Concurrently in this material, an increase in squamation of Colosteus in the past because the number ofventral scales per row is noted. they also bear posterior pits. For instance, Isolated dorsal and ventral scales can be BM[NH] R.2548, a small patch of ventral differentiated from each other by the rela- scales in articulation, was referred to Cte- 1 983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 29

nerpeton by Steen (1931, p. 881); as first rec- ficult to account for the absence of colosteids ognized by Baird (Bossy, Ms), these gastralia in other Carboniferous assemblages. These are actually those of Colosteus. Nectridean include the Linton-like thanatocoenosis of scales are generally fusiform and more deli- Nyrany (A. R. Milner, 1980b), the meager cate than those of colosteids, making con- vertebrate records ofMazon Creek (Gregory, fusion upon critical examination unlikely. 1950; Milner, 1982) and Cannelton (Baird, However, it should be noted that isolated 1978), the embolomere-dominated collec- ventrolateral scales of Colosteus are very dif- tion of Newsham (Land, 1974), and the Na- ficult to differentiate from the flat, plain ven- murian-age assemblages of Scotland (Smith- tral scales of Erpetosaurus (Steen, 1931) and son, 1 980a; the preliminary report ofcolosteid Saurerpeton. remains from Cowdenbeath in Andrews et In summary, the thoracic integument of al., 1977, is unfounded [Smithson, personal Colosteus was armored with ventral and dor- commun.]). Although it is beyond the scope sal pavements of rhombic, posteriorly sculp- ofthis paper to entertain the reasons for these tured scales. As the lateral extent of the ven- absences, I suggest that the chances of pres- tral squamation increased posteriorly, the ervation and recognition of colosteid re- width of the dorsal series decreased. In the mains, particularly the scalation, rate among absence of dermal scales, two types of osteo- the highest ofany Paleozoic vertebrate. Thus, derms covered the dorsolateral flanks of the their non-appearance may merit greater pa- body. With reduction of the dorsal squa- leoecologic and paleogeographic recognition mation posteriorly, these superficial dermal than is usually accorded to such negative evi- ossifications formed the dorsal covering. dence. CHARACTERISTICS AND OCCURRENCES OF COLOSTEID COMPOSITION OF THE REMAINS COLOSTEIDAE Seven localities spanning a stratigraphic interval from the mid-Visean to the upper- SHARED DERIVED CHARACTERS most Westphalian D have produced colos- OF THE FAMILY teid remains (fig. 14). The oldest colosteid, Pholidogaster, Greererpeton, and Colos- and indeed the first non-ichthyostegid laby- teus comprise a clade of primitive amphibi- rinthodont, is Pholidogaster [Otocratia] pis- ans. They are united by the following syn- ciformis, known from the Burdiehouse Lime- apomorphies: stone and Gilmerton Ironstone ofthe Visean (1) Elongate prefrontal entering the mar- of Scotland (Panchen, 1975). Greererpeton gin ofthe external narial opening, contacting occurs in the Bickett and Hinton Shales the premaxilla and maxilla, thus excluding (roughly equivalent to the upper Visean and the lacrimal and nasal from the naris. As Namurian A, respectively) of the Mauch originally noted by Smithson (1982), this fea- Chunk Group of West Virginia (Smithson, ture is unique to colosteids. Although the 1982). An additional Mississippian-age co- prefrontal may reach the naris in some am- losteid has been reported from the Point Ed- phibians (adelogyrinids, nectrideans, Acher- ward Formation (Namurian A) of Nova Sco- ontiscus, Micraroter, some trematopsids), it tia (Carroll et al., 1972; PU 20100). Poorly is almost universally separated from the pre- preserved colosteid material is present in the maxilla and maxilla by the nasal and lacri- famous Westphalian A amphibian assem- mal, respectively (the widely reproduced re- blage known from Jarrow, Ireland (A. R. Mil- construction of Sauropleura scalaris from ner, 1980a). The youngest stratigraphic rec- Steen, 1938, is in error; Bossy, MS; A. C. Mil- ord of the family is Colosteus scutellatus, ner, 1980). Only in the highly derived kera- known solely from the classic Pennsylvanian terpetontid Diploceraspis (Beerbower, 1963), locality of Linton, Ohio. in which both the nasal and lacrimal are ab- In view of the longevity and geographic sent, does the prefrontal reach the naris and distribution ofthe family, it is somewhat dif- contact both marginal elements. 30 AMERICAN MUSEUM NOVITATES NO. 2770

EUROPEAN STANDARD EASTERN U.S.

a-=i MONONGAHELA U.' I- CO) CONEMAUGH Colosteus 4 Freeport Coal Ohio ALLEGHENY z

z r - _L cn CO) 3- z LI.' CU z CL 0: POTTSVILLE CD UNDESCRIBED SPECIMENS 4 Jarrow Coal Ireland CO,0

LL z 0 co 47 Greererpeton cs z Hinton Shale- West Virginia C. NC MAUCH CHUNK < UNDESCRIBED SPECIMEN 2! CL. Pt. Edward Fm.- Nova Scotia Greererpeton CD. 4 Bickett Shale Pholidogaster COIcn) West Virginia 4 Gilmerton Ironstone CO Scotland cO) F! Pholidogaster Iz 4 Burdiehouse Limestone Mcn Scotland

o-0

FIG. 14. Stratigraphic correlation of colosteid localities. Nova Scotia material appears with eastern United States for convenience. Adapted from Panchen, 1970.

(2) Broad postorbital-parietal suture. prim us and Tupilakosaurus wetlugensis Within the Temnospondyli, this character is (Shishkin, 1973); brief postorbital-parietal also seen in Erpetosaurus and the brachy- contacts appear in the rhinesuchoid Chom- opoids (sensu Warren, 198 1) Dvinosaurus atobatrachus (Cosgriff, 1974) and the poorly 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 31 known Broomulus (Romer, 1947). Other Erpetosaurus is not regarded as evidence of temnospondyls possess either an intertem- a close relationship between Erpetosaurus and poral-postfrontal suture or, in the absence of the Colosteidae. Several shared derived char- an intertemporal, develop a supratemporal- acters indicate that Erpetosaurus is more postfrontal contact. Ichthyostegids, nectri- closely related to Saurerpeton than to the co- deans, some adelogyrinids (assuming the losteids. These include moderately devel- "postorbital" is indeed that), and some mi- oped tympanum-supporting structures; pal- crobrachomorph microsaurs are among the atine incorporated in skull roof; broad non-temnospondyl stocks that exhibit this interpterygoid vacuity bordered by vomer, feature. palatine, and pterygoid; lateral border of (3) Extensive tabular-squamosal contact. choana formed mainly by palatine; accessory In the presumed absence ofa tympanic mem- tusks only on vomer; and reduced lateral line brane, colosteids have completely closed the system. Further consideration ofthe affinities quadrate angle (see below) by extending the ofthe Saurerpetontidae (Chase, 1965) awaits tabular-squamosal contact posteriorly to the redescription of these Linton genera. occipital surface. A similar condition is found in trimerorhachids and some brachyopoids, but in these instances there is a slight occipital RELATIONSHIPS WITHIN THE FAMILY recess which probably supported a tympa- It is difficult at this time to evaluate the num. Microsaurs and nectrideans also elab- validity ofthe generic distinction afforded the orate the tabular-squamosal contact in the Lower Carboniferous colosteids. Whereas presumed absence of a tympanum. Greererpeton is represented by numerous ex- (4) Premaxillary tusk borne on postero- quisitely preserved and virtually complete in- lateral palatalflange and accommodated by dividuals, Pholidogaster is based on only two dentary notch. Excluding Erpetosaurus, which specimens whose preservation leaves much differs further in having an additional tusk to be desired. Because most of the cranial pair shared by the premaxillae at the midline, detail ofthe latter genus is obscured, any crit- premaxillary tusks are not known in any oth- ical comparison is necessarily limited to the er amphibians. Some forms (Ichthyostega, more informative postcranium. It is expected Edops, most embolomeres) exhibit large, that Stephen Godfrey (McGill University) marginally situated premaxillary teeth, but will address this matter upon completion of do not develop the tusk-replacement pit com- his current study of the postcranial skeleton bination typical of palatal tusks. of Greererpeton. (5) Elongate Meckelian fenestra. No tem- By the same token, analysis of the differ- nospondyls other than Colosteus and Greer- ences between Colosteus and earlier colos- erpeton are known to possess a single, long, teids must be based largely on comparisons narrow fenestra on the medial face of the with Greererpeton. Many potential differ- mandible (this surface is unknown in Pholi- ences seen in Colosteus, such as relatively dogaster). A single anthracosaur, Anthraco- larger dentary teeth, shorter antorbital length saurus (Panchen, 1981), and several diadec- and smaller forelimbs, cannot be considered tomorphs (Heaton, 1 980a) possess large, in the absence of more representative adult solitary Meckelian fenestrae. material. Although the presence of a single (6) Approximately 40 presacral vertebrae. row of coronoid teeth cannot be verified in Although the trunk length of Colosteus is adults, this feature is regarded as a valid char- unknown, Greererpeton, Pholidogaster, and acter since drastic change in coronoid den- the undescribed Jarrow colosteid appear to tition is not seen in other amphibian growth have approximately 40 presacral vertebrae series. Among early labyrinthodonts, only (Panchen, 1975; A. R. Milner, personal com- ichthyostegids (Campbell and Bell, 1977; Jar- mun.). This exceptional length is known in vik, 1980) and Doragnathus (Smithson, only one other temnospondyl, the proble- 1 980b) possess a single row ofcoronoid teeth. matical Kourerpeton bradyi (Olson and Lam- Highly denticulated coronoids similar to those mers, 1976). of Greererpeton occur in several primitive The occurrence of characters 2 and 4 in forms, including Megalocephalus, Caero- 32 AMERICAN MUSEUM NOVITATES NO. 2770 rhachis, and Edops. Because the densely den- itive labyrinthodonts. In loxommatids and ticulated condition is more common and the anthracosaurs, the braincase is attached to affinities of ichthyostegids and Doragnathus the skull roof by areas derived from the au- are unresolved, the single-row coronoid den- ditory capsules; primitive temnospondyls tition of Colosteus is considered apomorphic (colosteids, edopoids, trimerorhachoids) re- within the Colosteidae. duce these attachment surfaces and elaborate The characteristic circumorbital configu- a connection between the exoccipitals and ration of Colosteus is seen in specimens of postparietals (Smithson, 1982, fig. 22). all sizes and is certainly a valid generic char- Smithson regards the temnospondyl condi- acter. The exclusion of the prefrontal from tion as derived relative to the rhipidistian- the orbital margin appears to be unique among like condition shared by loxommatoids and Paleozoic amphibians; less exceptional, but anthracosaurs. In addition, both loxomma- equally derived among labyrinthodonts, is the tids (Beaumont, 1977) and anthracosaurs omission of the postorbital from the same (Panchen, 1970) appear to have retained a margin. The resulting three-bone configura- kinetic basal articulation characterized by tion enclosing the orbit is an autapomorphy opposing articular facets on the basipterygoid of Colosteus. process and the conical recess; such facets are An assessment of the character states rep- not present in primitive temnospondyls resented by the different types of dorsal (Smithson, 1982). scalation possessed by Colosteus and Greer- Further consideration of a possible lox- erpeton is impaired by a general lack of in- ommatoid-temnospondyl relationship indi- formation on such ossifications in other cates that the characters common to both forms. groups represent generalized amphibian con- ditions. These plesiomorphic characters in- RELATIONSHIPS OF THE clude (1) supratemporal-postparietal contact; COLOSTEIDAE (2) quadratojugal-maxilla contact; (3) orbital Consideration ofthe phylogenetic relation- lacrimal (conceivably secondary in loxom- ships ofthe colosteids raises two fundamental matoids); (4) marginally situated, laterally di- problems. The first is the difficulty encoun- rected external nares; (5) widely separated tered in attempting a diagnosis of the Tem- choanae; (6) pterygoids joined anteromedi- nospondyli; that is, while it may be possible ally; (7) paired tusks on vomers, palatines and at present to delimit the order, the number ectopterygoids; (8) parasymphysial tusks ac- ofinadequately described taxa makes cladis- commodated by palatal fenestrae; and (9) tic analysis infeasible. Related to this is the vertebrae with compound centra. Thus, with second problem of the development of the fundamentally different braincase-skull roof otic notch within the order. These general connections and basal articulations, and an areas will be addressed before apparent absence of shared derived charac- discussing the ters, loxommatoids do not appear to share a more particular concern of the taxonomic unique common ancestry with other am- placement of the Colosteidae. phibians classified as temnospondyls. With the loxommatids and spathicephalids THE STATUS OF THE ORDER removed, Smithson (1982) asserts that the TEMNOSPONDYLI Temnospondyli represent a monophyletic The composition of the Temnospondyli group "in the context ofPaleozoic tetrapods" and the affinities ofthe Colosteidae have been (acknowledging that anurans are generally re- considered most recently by Smithson (1982). garded as temnospondyl descendants, but ex- Building upon the suggestion that loxom- cluded from the order). But in the same pa- matoids (loxommatids and spathicephalids; per, it is demonstrated that the very character Beaumont, 1977) should not be regarded as used to consolidate the temnospondyls-an temnospondyls (Panchen, 1980), Smithson exoccipital-postparietal contact-also occurs has drawn attention to the manner in which in the order Microsauria. Smithson further the braincase is held within the skull ofprim- notes that temnospondyls and microsaurs 1983 HOOK: COLOSTEUS SCUTELLATUS NEWIBERRY 33 share additional features that are derived rel- Such a conclusion, however, largely ignores ative to other primitive amphibians: akinetic the manner by which cranial kinesis appears skull roof, immobile basal articulation, and to be reduced in labyrinthodonts, as well as four digit manus. Ifthese features, along with the wide variety of otic configurations seen the mode of braincase attachment, are re- in primitive temnospondyls. garded as synapomorphies uniting the two There is no conclusive evidence to indicate orders, then the claim of temnospondyl that an extensive tabular-squamosal contact monophyly appears to lack the support ofany is requisite to the stabilization of the skull. unique characters. In Ichthyostega, loxommatoids, most prim- Consideration of a close relationship be- itive temnospondyls, and advanced embolo- tween microsaurs and temnospondyls is re- meres which lose skull table-cheek kinesis tarded by the chronic lack ofpre-Westphalian (Anthracosaurus; Panchen, 1977b), this su- specimens. Nonetheless, the proposal clearly ture is either short or absent. Yet, these forms merits further attention, and study of the at- appear to have consolidated the skull roof las-axis complex of Greererpeton seems par- mainly through a posterior shift in the po- ticularly warranted in light of the specialized sition of the parietal-postparietal suture and atlas-axis configuration of microsaurs (Car- a change in the relative positions and sizes roll and Baird, 1968). Equally important is of the postfrontal, postorbital, and intertem- the need for accurate analysis of character poral bones (Panchen, 1972). It is, therefore, distributions among other lepospondyl difficult to argue that the broad tabular-squa- groups: in genera in which the occiput is mosal contact seen in colosteids represents known, both adelogyrinids (Adelospondylus; an archetypical temnospondyl design for the Carroll, 1967) and nectrideans (Diploceras- reduction of cranial kinesis. pis; Beerbower, 1963) exhibit exoccipital- The otic notch has been defined as "the postparietal contacts. Colosteids further re- bony emargination enclosing (usually ante- semble adelogyrinids, a group once suggested riorly) the tympanum" (Thomson and Bossy, as "an early-divergent family of microsaurs" 1970, pp. 18-19); thus, the recognition of an (Baird, 1965, p. 292), as well as the enigmatic otic notch postulates the existence of a tym- Acherontiscus (Carroll, 1969), in the posses- panic membrane. A second type of posterior sion oflateral lines (uncommon in lepospon- embayment of the skull roof is the space be- dyls), prefrontal that contributes to both the tween the lateral edge of the skull table and orbital and narial margins, postorbital-pari- the margin of the squamosal. This space, etal contact, and no otic notch (the "notch" termed the quadrate angle by Thomson and of adelogyrinids is probably an elaborated Bossy, often coincides with the topological surface for muscle attachment; Carroll, 1967). position of the otic notch. A re-examination of these problematical Colosteids may be considered as notchless Lower Carboniferous forms may well be of forms with no distinct quadrate angle because use in testing a possible microsaur-temno- the morphology of the stapes indicates that spondyl association. a tympanum was probably absent (Smithson, 1982), and an extensive tabular-squamosal suture supersedes any posterior embayment THE DEVELOPMENT OF THE of the skull roof. The only other primitive TEMNOSPONDYL OTIC NOTCH temnospondyl that has been cited as notch- The suggestion that the temnospondyl otic less is Caerorhachis bairdi, known only from notch was derived from a notchless ancestor a single specimen in which the skull roof is (see Panchen, 1975, 1977a; Carroll, 1980; A. poorly preserved. On the basis of a natural R. Milner, 1980a), combined with the con- cast that represents the ventral surface of the vincing argument that the colosteid stapes right tabular and squamosal, Holmes and represents the primitive temnospondyl mor- Carroll (1977, p. 493) stated "that the exis- photype (Smithson, 1982), seems to indicate tence of a well developed otic notch is un- that an extensive tabular-squamosal suture likely." This description, and an equally as seen in colosteids is primitive for the order. equivocal figure of the skull, have been in- 34 AMERICAN MUSEUM NOVITATES NO. 2770

terpreted as a probable indication of the early apomorphic group that closed the quad- notchless or tympanumless condition rate angle while retaining the stapes as a (Panchen, 1977a; A. R. Milner, 1980a; braincase support, then they constitute a Smithson, 1982). However, since no stapes monophyletic taxon that may be the plesio- accompany this material, it seems more ju- morphic sister group ofall other temnospon- dicious to exclude Caerorhachis from con- dyls. sideration. A survey of the remaining early temno- spondyls does not reveal any compelling evi- COLOSTEIDS, EDOPOIDS, AND dence for a common ancestor that lacked both TRIMERORHACHOIDS the otic notch and quadrate angle. A plethora As previously noted, colosteids have been of notch types occurs in Westphalian tem- associated most commonly with either the nospondyls, ranging from slight (Saurerpe- edopoids or the trimerorhachoids, two groups ton), to moderate (Eugyrinus, Erpetosaurus), that have been regarded traditionally as to markedly embayed (Stegops, Cochleosau- primitive temnospondyl grades. Although it rus, Amphibamus, Dendrerpeton). In fact, the is not the intent of this section to diagnose majority ofWestphalian forms possesses well- these groups (the edopoids are being revised developed otic notches and quadrate angles by A. R. Milner, and a statement on the tri- in which the tabular is almost completely sep- merorhachoids awaits redescription of sev- arated from the squamosal on the skull roof. eral saurerpetontid species), members from If an abbreviated tabular-squamosal con- these two superfamilies can be used to con- tact is common among early temnospondyls, sider possible relationship to the colosteids. then colosteids may be regarded as a primi- As used here, the Edopoidea contains two tive group of aquatic specialists that main- families, the Edopidae and Cochleosauridae, tained the supporting function of the stapes, with Dendrerpeton and Caerorhachis re- never developed a tympanic membrane, and moved to separate families of uncertain po- secondarily reinforced the skull roof through sition (A. R. Milner, 1980a). The Trimero- closure of the quadrate angle. The precursors rhachoidea is considered to consist of three of other early forms, such as Dendrerpeton, families, the Eugyrinidae (A. R. Milner, Amphibamus, and Edops, would have elab- 1980a), Saurerpetontidae (Chase, 1965), and orated the anterior portion of the quadrate Trimerorhachidae, with the monogeneric angle into a bony, tympanum-supporting otic Dvinosauridae removed to the Brachyopoi- notch while modifying the stapes for trans- dea (Shishkin, 1973). mission of airborne vibrations (Smithson, Edopoids are known from as early as the 1982). Further, the problematical "otic slits" Westphalian A of Jarrow. In addition to a of saurerpetontids do not represent tympa- well-developed otic notch and stapes adapted num-supporting structures at all (Milner, for an impedance matching system (Romer 1982), but more plausibly an intermediate and Witter, 1942), a diagnostic complex of condition in which the quadrate angle is not characters is also present that distinguishes completely occluded by a tabular-squamosal the superfamily from other early temnospon- contact. If a tympanum were present, as sug- dyls (A. R. Milner, 1980a): no lateral lines, gested by the form of the stapes in these gen- lacrimal excluded from orbit by jugal-pre- era, it would have been supported by a slight frontal contact, septomaxilla incorporated in recess on the occipital expansion of the skull dermal skull roof, and broad separation of (see Warren, 1981, for further development the maxilla and quadratojugal. Only on the of the saurerpetontid plan in brachyopids). palate, where narrow interpterygoid vacui- It is concluded that the extensive tabular- ties, thin cultriform process not sutured to squamosal suture ofcolosteids is derived rel- the vomers, and reduced vomerine tusks oc- ative to other primitive temnospondyls, and cur, are any similarities to colosteids evident. that the otic notch developed from a non- These may be considered as primitive tem- colosteid form that possessed a short tabular- nospondyl features and do not in themselves squamosal contact and distinct quadrate an- indicate close relationship. gle. Finally, if colosteids are regarded as an Colosteids have most recently been asso- 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 35 ciated with trimerorhachoids (Carroll and of specialized features that precludes them Winer, 1977). Although the otic notch is from being direct ancestors ofany other pres- poorly expressed in trimerorhachoid genera, ently known temnospondyls. Because they can their slender, dorsolaterally oriented stapes be associated with the Trimerorhachoidea or suggest that an impedance-matching system Edopoidea only on the basis of characters of was developed. Other cranial features, in- questionable polarity or certain primitive- cluding dorsally oriented external nares, broad ness, the colosteids are placed in a separate cultriform process sutured to vomers, and superfamily, the Colosteoidea (Tatarinov, enlarged interpterygoid vacuities bordered in 1964), that is possibly the plesiomorphic sis- part by vomers, represent character states that ter group of all other temnospondyls. A re- are derived relative to those seen in colos- vised classification of primitive temnospon- teids. However, similarities exist between the dyls based on these conclusions and those of postcranial skeletons of the two groups: the A. R. Milner (1980a) is as follows: pectoral girdle is extended anteriorly to form Superfamily: Colosteoidea Tatarinov, 1964 a dermal plate beneath the braincase, and the Family: Colosteidae Cope, 1875 limbs are generally reduced. In addition, the Pholidogaster pisciformis Huxley, 1862 vertebrae of Neldasaurus (Chase, 1965) re- Greererpeton burkemorani Romer, 1969 semble the schizomerous condition of colos- Colosteus scutellatus (Newberry, 1856) teids. But in the absence ofcomparative ma- Superfamily: Trimerorhachoidea terial from other early amphibians, the Families: Saurerpetontidae, Eugyrinidae, Tri- taxonomic significance ofthese similarities is merorhachidae difficult to evaluate objectively for the rec- Superfamily: incertae sedis of within groups comes to Caerorhachis bairdi Holmes and Carroll, 1977 ognition polarities Superfamily: Edopoidea rely heavily upon ad hoc adaptational or Families: Edopidae, Cochleosauridae functional interpretations. Superfamily: incertae sedis In the apparent absence of derived cranial Family: Dendrerpetontidae features linking colosteids to trimerorha- choids or any other group, it is proposed that APPENDIX the superfamily Colosteoidea, first erected by Tatarinov (1964) and recently employed by Hypodigm of Colosteus scutellatus (Newberry) Shishkin (1973), be utilized. Like some Per- AMNH 2934: J. S. Newberry Collection. Pos- mo-Triassic temnospondyl superfamilies, this terior portion of adult mandible; association with taxon would consist solely of the monophy- AMNH 6826 (Moodie, 1916, p. 185) unlikely. Ho- letic type family, and the familial diagnosis lotype ofMacrerpeton deani. Moodie, 1916, figure would be elevated accordingly. Although this 40; plate 21, figure 1; figure 9 of this paper. may not be as economical as tentatively as- AMNH 6826: formerly 8535G, Newberry Col- lection. Postorbital region of adult skull and left sociating the colosteids with the Trimerorha- mandible; association with AMNH 2934 (Moo- choidea, it does not handicap a higher taxon die, 1916, p. 185) unlikely. Paratype of M. deani; that is slowly staggering toward monophyly erroneously cited as holotype by Romer (1930, pp. (see Coldiron, 1978). In an even less Hen- 102-103); erroneously cited as holotype of Mega- nigian perspective, this prudence underscores locephalus lineolatus by Beaumont (1977, p. 75). the difficulties of establishing the interrela- Moodie, 1916, plate 21, figure 2; Romer, 1930, tionships of Westphalian temnospondyls figure 9 right, below; figure 5 of this paper. when so few Lower Carboniferous represen- AMNH 6829: formerly 1088G, Newberry Col- tatives are known. lection. Right dentary of subadult. Erroneously cited as holotype of Leptophractus lineolatus by Moodie (1916, p. 169). SYSTEMATIC CONCLUSIONS AMNH 6833: formerly 8563G, 8570G, 8673G, 8693G, 9031 G, Newberry Collection. Isolated Colosteus, Greererpeton, and Pholidogas- dorsal and ventral scales, not associated. 8673G ter represent the family Colosteidae, a clade referred to Sauropleura by Moodie (1916, p. 160). of closely related, primitive temnospondyls. AMNH 6915: formerly 8587G (or possibly Despite a preponderance of plesiomorphic 8687G), Newberry Collection. Postorbital region characters, these forms also possess a suite of subadult skull, palate and mandibles; in coun- 36 AMERICAN MUSEUM NOVITATES NO. 2770 terpart blocks. Romer, 1930, figure 9 left, above AMNH 6949: formerly 951G, Newberry Col- and below; figure 6 of this paper. lection. Anterior portion of an adult palate and AMNH 6916: formerly 8584G and 8666G, mandibles. Newberry Collection. Subadult skull, mandibles, AMNH 6950: formerly 8572G, Newberry Col- and anterior portion ofbody, including partial ribs, lection. Anterior portion of a subadult skull and pectoral girdle, and scalation; in counterpart blocks. left mandible, very poorly preserved; in counter- Holotype of Pygopterus scutellatus, Colosteus part blocks. Referred to M. huxleyi by Moodie crassiscutatus, Sauropleura scutellata, and C. scu- (1916, p. 184), but not associated with 8532G tellatus. Cope, 1875, plate 33, figure 1; Moodie, (=AMNH 6954, an edopid) as stated. 1916, plate 21, figure 5 as S. scutellata; Romer, AMNH 6951: formerly 8709G, Newberry Col- 1930, figure 8 left, above and below; figure 2 of lection. Antorbital region of subadult skull and this paper. mandibles; in counterpart blocks. 6951 was cited AMNH 6917: formerly 8657G and 9027G, in error by Romer (1930, pp. 127-128) under L. Newberry Collection. Posterior portion of sub- lancifer, 6954 was intended. Figure 8 ofthis paper. adult skull and mandibles, and anterior segment AMNH 6957: formerly 8554G, Newberry Col- of body, including pectoral girdle, forelimbs, and lection. Adult posterior presacral vertebrae; coun- scalation; in counterpart blocks. Paratype of C. terpart of OSU 4481. Referred to S. pauciradiata crassiscutatus and C. scutellatus. Cope, 1875, plate by Moodie (1916, p. 160). Possibly from same 34, figure 2; Moodie, 1916, plate 14, figure 3 as individual represented by PU 16532 or AMNH S. scutellata, but also S. pauciradiata (pp. 159- 6958. 160); Romer, 1930, figure 22 as M. huxleyi; figure AMNH 6958: formerly 8668G, Newberry Col- 12 of this paper. lection. Section of adult presacral vertebral col- AMNH 6920: formerly 8670G, Newberry Col- umn, ribs, and scales. Referred to S. pauciradiata lection. Isolated interclavicle, ribs, scales, inter- by Moodie (1916, p. 160). Possibly from same centrum, and pleurocentrum; in counterpart individual represented by AMNH 6957, OSU blocks. Paratype of C. crassiscutatus, holotype of 4481, or PU 16532. C. pauciradiatus and S. pauciradiata. Cope, 1875, AMNH 7491: Newberry Collection. Incomplete plate 40, figure 1 as C. pauciradiatus; Moodie, subadult mandible; in counterpart blocks. 1916, figure 34A as S. pauciradiata, but also S. AMNH 7493: formerly 8333G, Newberry Col- scutellata (p. 156); figure 1 A of this paper. lection. Portion of presacral dorsal and ventral AMNH 6921: formerly 8671G, Newberry Col- scalation, intercentra partially exposed; in coun- lection. Isolated left clavicle in counterpart blocks; terpart blocks. not associated with AMNH 6920. Paratype of C. AMNH 20866: formerly 8558G (incorrectly as- pauciradiatus; erroneously designated as holotype sociated with AMNH 6965), Newberry Collec- ofS. pauciradiata by Moodie (1916, p. 158). Cope, tion. Isolated left subadult clavicle. 1875, plate 40, figure 2 as C. pauciradiatus; Moo- BM[NH] R.2547: T. Stock Collection, pur- die, 1916, figure 34B as S. pauciradiata. chased by Museum in April 1895 (A. C. Milner, AMNH 6932: formerly 8612G, Newberry Col- personal commun.). Subadult skull, mandibles, and lection. Incomplete subadult skull, mandibles, partial interclavicle; counterpart of AMNH 6945. branchial arches, and scalation. Paratype of S. Steen, 1931, figure 6; plate 2, figure 1; figure 3B newberryi; erroneously designated as type by of this paper. Moodie (1916, p. 158) and Romer (1930, p. 101). BM[NH] R.2548: Stock Collection (see BM[NH] Cope, 1875, plate 41, figure 5 as S. Newberryiana R.2547). Segment of gastralia. Referred to Cte- in error; Romer, 1930, figure 8 right, below; figure nerpeton remex by Steen (1931, p. 881). 7A of this paper. BM[NH] R.2660: J. W. Davies Collection, pur- AMNH 6936: formerly 8674G, Newberry Col- chased by Museum in October 1895 (A. C. Milner, lection. Segment of subadult gastralia. Referred to personal commun.). Disarticulated subadult post- S. pauciradiata by Moodie (1916, p. 160). cranial remains, including vertebrae, ribs, inter- AMNH 6937: formerly 2412G, Newberry Col- clavicle, and scales. Referred to Erpetosaurus ra- lection. Segment of subadult gastralia. diatus by Steen (1931, pp. 852, 856). AMNH 6945: formerly 8619G, Newberry Col- BM[NH] R.2664: Davies Collection (see lection. Partial subadult skull and mandible; coun- BM[NH] R.2660). Segment of subadult postcra- terpart of BM[NH] R.2547. Holotype of S. Ion- nial skeleton, including vertebrae, ribs, and dorsal gidentata; erroneously cited as S. dentatus by and ventral scalation in counterpart blocks. Steen, Romer (1930, p. 103). Moodie, 1909b, figure 18; 1931, plate 4, figure 2 as a phyllospondyl. 1916, plate 16, figures 2, 3; Romer, 1930, figure BM[NH] R.9949: formerly BM[NH] R.2675 9 right, above; figure 3A of this paper. (B.M.63), Davies Collection (see BM[NH] R.2660). 1983 HOOK: COLOSTEUS SCUTELLATUS NEWBERRY 37

Portion ofpostorbital region ofadult skull; adjoins 1965. Paleozoic lepospondyl amphibians. OSU 4545. Steen (1931, p. 886) as unidentified Amer. Zool., vol. 5, pp. 287-294. labyrinthodont. 1978. Studies on Carboniferous freshwater CM 25307: R. Lund, collector. Preorbital region fishes. Amer. Mus. Novitates, no. 2641, of subadult skull and mandibles; in counterpart pp. 1-22. blocks. Figure 7B of this paper. Beaumont, E. H. FMNH UC 2001: A. S. Romer, collector(?), 1977. Cranial morphology of the Loxomma- 1932. Portion of subadult skull roof. Figure 4 of tidae (Amphibia: Labyrinthodontia). this paper. Phil. Trans. Roy. Soc. London (B), vol. MCZ 2136: D. and L. Baird, collectors, 1953, 280, pp. 29-101. 1954. Lateral portions of dorsal and ventral scal- Beerbower, J. R. ation, vertebrae and ribs partially exposed.- Two 1963. Morphology, paleoecology and phylog- specimens from possibly the same individual. eny of the Permo-Pennsylvanian am- MCZ 2158: R. N. Fearon Collection, 1883. Sec- phibian Diploceraspis. Bull. Mus. Comp. tion of adult posterior presacral region, including Zool., vol. 130, pp. 31-108. vertebrae, ribs, and dermal ossifications; counter- Berman, D. S. part of UMMP 3573. Figure IOA of this paper. 1973. A trimerorhachid amphibian from the MCZ 2159: Fearon Collection, 1883. Incom- Upper Pennsylvanian of New Mexico. plete adult mandibles. Jour. Paleont., vol. 47, pp. 932-945. OSU 4481: Newberry Collection. Adult poste- Bossy, K. V. H. rior presacral gastralia; counterpart of AMNH [Ms] Morphology, paleoecology and evolu- 6957. Possibly from same individual as repre- tionary relationships of Pennsylvanian sented by AMNH 6958 or PU 16532. urocordylid nectrideans (subclass Lepo- OSU 4545: Newberry Collection. Small portion spondyli, class Amphibia). Yale Univ., of adult skull roof; adjoins BM[NH] R.9949. Ph.D. dissertation, 1976. OSU 4560 and 4562: Newberry Collection. Boy, J. A. Poorly preserved squamation; in counterpart 1972. Die Branchiosaurier (Amphibia) des blocks. saarpfalzischen Rotliegenden (Perm, PU 16532: Newberry Collection. Section ofadult SW-Deutschland). Abh. Hess. L-Amt. postcranial skeleton including vertebrae, ribs, and Bodenforsch., vol. 65, pp. 5-137. scalation; in counterpart blocks. Possibly from Bystrow, A. P. same individual as represented by AMNH 6957, 1935. Morphologische Untersuchungen der 6958, or OSU 4481. Deckknochen des Schadels der Wirbel- PU 16540: D. Baird, collector, 1957. Subadult tiere. I. Mitteilung. Schadels der Ste- skull. Smallest known individual, skull length gocephalen. Acta Zool., vol. 16, pp. 65- 7.5 mm. 141. UMMP 3573: Fearon Collection, 1883. Portion Bystrow, A. P., and J. A. Efremov of gastralia; partial counterpart of MCZ 2158. 1940. Benthosuchus sushkini Efremov-a lab- YPM 4604: F. H. Bradley, collector. Postcranial yrinthodont from the Eotriassic ofShar- remains, including scales and vertebrae, of two jenga River. Trudy Paleozool. Inst., vol. individuals ofdisparate size on same bedding plane; 10, pp. 1-1 52 (in Russian, English sum- in counterpart blocks. mary). Camp, C. L., and H. J. Allison LITERATURE CITED 1961. Bibliography offossil vertebrates, 1949- 1953. Geol. Soc. Amer., Mem. 84, Andrews, S. M., M. A. E. Browne, A. L. Panchen, xxxviii + 532 pp. and S. P. Wood Camp, C. L., H. J. Allison, and R. H. Nichols 1977. Discovery of amphibians in the Na- 1964. Bibliography offossil vertebrates, 1954- murian (Upper Carboniferous) of Fife. 1958. Geol. Soc. Amer., Mem. 92, Nature, vol. 265, pp. 529-532. xxv + 647 pp. Baird, D. Camp, C. L., H. J. Allison, R. H. Nichols, and H. 1955. Latex micro-molding and latex-plaster McGinnis molding mixture. Science, vol. 122, p. 1968. Bibliography offossil vertebrates, 1959- 202. 1963. Geol. Soc. Amer., Mem. 117, 1964. The aYstopod amphibians surveyed. xlii + 644 pp. Breviora Mus. Comp. Zool., no. 206, Camp, C. L., R. H. Nichols, B. Brajnikov, E. Ful- pp. 1-17. ton, and J. A. Bacskai 38 AMERICAN MUSEUM NOVITATES NO. 2770

1972. Bibliography offossil vertebrates, 1964- Coldiron, R. W. 1968. Geol. Soc. Amer., Mem. 134, 1974. Possible functions of ornament in lab- lxi + 11 19 pp. yrinthodont amphibians. Occ. Papers Campbell, K. S. W., and M. W. Bell Mus. Nat. Hist. Univ. Kansas, no. 33, 1977. A primitive amphibian from the late pp. 1-19. Devonian of New South Wales. Al- 1978. Acroplous vorax Hotton (Amphibia, cheringa, vol. 1, pp. 369-381. Saurerpetontidae) restudied in light of Carroll, R. L. new material. Amer. Mus. Novitates, 1964. Early evolution of the dissorophid am- no. 2662, pp. 1-27. phibians. Bull. Mus. Comp. Zool., vol. Cope, E. D. 131, pp. 161-250. 1868. Synopsis of the extinct Batrachia of 1967. An adelogyrinid lepospondyl amphibi- North America. Proc. Acad. Nat. Sci. an from the Upper Carboniferous. Ca- Philadelphia, 1868, pp. 208-221. nadian Jour. Zool., vol. 45, pp. 1-16. 1869. Synopsis of the extinct Batrachia, Rep- 1969. A new family of Carboniferous am- tilia, and Aves ofNorth America. Trans. phibians. Palaeontology, vol. 12, pp. Amer. Phil. Soc., vol. 14, pp. 1-252. 537-548. 1871a. Supplement to the synopsis of the ex- 1980. The hyomandibular as a supporting ele- tinct Batrachia and Reptilia of North ment in the skull ofprimitive tetrapods. America. Proc. Amer. Phil. Soc., vol. In Panchen, A. L. (ed.), The terrestrial 12, pp. 41-52. environment and the origin ofland ver- 1871b. Observations on the extinct Batrachia tebrates. London, Academic Press, pp. of the Carboniferous of Ohio (Linton). 293-317. Ibid., p. 177. Carroll, R. L., and D. Baird 1873. Remarks on fishes from the Coal Mea- 1968. The Carboniferous amphibian Tudita- sures of Linton, Ohio. Proc. Acad. Nat. nus and the distinction between micro- Sci. Philadelphia, 1873, pp. 417-419. saurs and reptiles. Amer. Mus. Novi- 1874. Supplement to the extinct Batrachia and tates, no. 2337, pp. 1-50. Reptilia ofNorth America. I. Catalogue Carroll, R. L., E. S. Belt, D. L. Dineley, D. Baird, of air-breathing Vertebrata from the and D. C. McGregor Coal Measures of Linton, Ohio. Trans. 1972. Field excursion A59: vertebrate paleon- Amer. Phil. Soc., vol. 15, pp. 261-278. tology of eastern Canada. Guidebook, 1875. Synopsis of the extinct Batrachia from 24th Int. Geol. Congress, 113 pp. the Coal Measures. Ohio Geol. Surv., Carroll, R. L., and P. Gaskill vol. 2, pt. 2, pp. 348-411. 1978. TheorderMicrosauria.Amer.Phil.Soc., 1877. A continuation ofresearches among the Mem. 126, 211 pp. Batrachia ofthe Coal Measures ofOhio. Carroll, R. L., and L. Winer Proc. Amer. Phil. Soc., vol. 16, pp. 573- 1977. Appendix to Carroll, R. L., Patterns of 579. amphibian evolution: an extended ex- 1885. Second continuation of researches ample ofthe incompleteness ofthe fossil among the Batrachia of the Coal Mea- record, Chapter 13 in Hallam, A. (ed.), sures of Ohio. Ibid., vol. 22, pp. 405- Patterns of evolution. Amsterdam, El- 408. sevier (privately circulated). 1897. On new Paleozoic vertebrates from Il- Case, E. C. linois, Ohio, and Pennsylvania. Ibid., 1935. Description of a collection ofassociated vol. 36, pp. 71-90. skeletons of Trimerorhachis. Contrib. Cosgriff, J. W. Mus. Paleont. Univ. Mich., vol. 4, pp. 1974. Lower Triassic Temnospondyli of Tas- 227-274. mania. Geol. Soc. Amer., Sp. Paper 149, Chase, J. N. vi + 134 pp. 1965. Neldasaurus wrightae, a new rhachito- Gregory, J. T. mous labyrinthodont from the Texas 1950. Tetrapods of the Pennsylvanian nod- Lower Permian. Bull. Mus. Comp. Zool., ules from Mazon Creek, Illinois. Amer. vol. 133, pp. 156-225. Jour. Sci., vol. 248, pp. 833-873. Colbert, E. H. Hay, 0. P. 1955. Scales in the Permian amphibian Tri- 1902. Bibliography and catalogue of fossil merorhachis. Amer. Mus. Novitates, no. Vertebrata of North America. U.S. 1740, pp. 1-16. Geol. Surv. Bull. 179, 868 pp. 1983 HOOK: COLOSTE US SCUTELLA TUS NEWBERRY 39

Heaton, M. J. de paleontologie. Paris, Masson, vol. 5, 1980a. The Cotylosauria: a reconsideration of pp. 67-125. a group of archaic tetrapods. In Pan- Milner, A. C. chen, A. L. (ed.), The terrestrial envi- 1980. A review of the Nectridea (Amphibia). ronment and the origin of land verte- In Panchen, A. L. (ed.), The terrestrial brates. London, Academic Press, pp. environment and the origin ofland ver- 497-551. tebrates. London, Academic Press, pp. 1980b. New advances in latex casting tech- 377-405. niques. Curator, vol. 23, pp. 95-100. Milner, A. R. Holmes, R. 1980a. The temnospondyl amphibian Den- 1980. Proterogyrinus scheelei and the early drerpeton from the Upper Carbonifer- evolution of the labyrinthodont pecto- ous of Ireland. Palaeontology, vol. 23, ral limb. In Panchen, A. L. (ed.), The pp. 125-141. terrestrial environment and the origin 1980b. The tetrapod assemblage from Nyrany, of land vertebrates. London, Academic Czechoslovakia. In Panchen, A. L. (ed.), Press, pp. 351-376. The terrestrial environment and the or- Holmes, R., and R. L. Carroll igin of land vertebrates. London, Aca- 1977. A temnospondyl amphibian from the demic Press, pp. 439-496. Mississippian of Scotland. Bull. Mus. 1982. Small temnospondyl amphibians from Comp. Zool., vol. 147, pp. 489-511. the Middle Pennsylvanian of Illinois. Hook, R. W. Palaeontology, vol.25, pp. 635-664. 1980. The cranial anatomy of Erpetosaurus Miner, R. W. radiatus (Cope). Res. Ohio Geol., 1979, 1925. The pectoral limb of Eryops and other p. 26. primitive tetrapods. Bull. Amer. Mus. 1981. The Pennsylvanian tetrapod fauna of Nat. Hist., vol. 51, art. 7, pp. 145-312. Linton, Ohio. Ohio Jour. Sci., vol. 81, Moodie, R. L. April Prog. Abstr., p. 41. 1909a. New or little known forms of Carbon- Hook, R. W., and D. Baird iferous Amphibia in the American Mu- [in press] Ichthycanthus platypus Cope, 1877, seum of Natural History. Ibid., vol. 26, reidentified as the dissorophoid am- art. 25, pp. 347-357. phibian Amphibamus lyelli. Jour. Pa- 1909b. A contribution to a monograph of the leont. extinct Amphibia of North America. Huxley, T. H. New forms from the Carboniferous. 1862. On new labyrinthodonts from the Edin- Jour. Geol., vol. 17, pp. 38-82. burgh Coal-field. Quat. Jour. Geol. Soc., 1915. Some methods of studying fossil Am- vol. 18, pp. 291-296. phibia embedded in coal. Kansas Univ. Jarvik, E. Sci. Bull., vol. 9, pp. 187-193. 1948. Note on the Upper Devonian vertebrate 1916. The Coal Measures Amphibia ofNorth fauna of East Greenland and on the age America. Carnegie Inst. Washington, of the ichthyostegid stegocephalians. Publ. no. 238, x + 222 pp. Arkiv Zool., vol. 41, pp. 1-8. Moulton, J. M. 1952. On the fish-like tail in the ichthyostegid 1974. A description of the vertebral column stegocephalians. Meddel. Gr0nland, vol. of Eryops based on the notes and draw- 114, pp. 1-90. ings of A. S. Romer. Breviora Mus. 1980. Basic structure and evolution of verte- Comp. Zool., no. 428, pp. 1-44. brates. London, Academic Press, vol. 1, Murphy, J. L. xvi + 576 pp. 1980. The Ohio Diamond Mine-paleonto- Land, D. H. logical treasure trove. Ohio Hist. Soc. 1974. Geology of the Tynemouth District. Echoes, vol. 19, pp. 4-5. Mem. Geol. Surv. Gt. Br., no. 15, x + Newberry, J. S. 176 pp. 1856. Description of several new genera and Langston, W. species of fossil fishes from the Carbon- 1953. Permian amphibians from New Mexi- iferous strata of Ohio. Proc. Acad. Nat. co. Univ. Calif. Publ. Geol. Sci., vol. Sci. Philadelphia, vol. 8, pp. 96-100. 29, pp. 349-416. 1871. Report of progress in 1870. Columbus, Lehman, J.-P. Ohio Geol. Surv., 586 pp. 1955. Rachitomi. In Piveteau, J. (ed.), Traite 1873. The classification and geological distri- 40 AMERICAN MUSEUM NOVITATES NO. 2770

bution of our fossil fishes. Ohio Geol. 1939. Notes on branchiosaurs. Amer. Jour. Surv., vol. 1, pt. 2, pp. 245-355. Sci., vol. 237, pp. 748-761. 1874. The Carboniferous System. Ibid., vol. 1945. Vertebrate paleontology. Second ed. 2, pt. 1, pp. 81-180. Chicago, Univ. Chicago Press, ix + 687 Olson, E. C. pp- 1979. Aspects of the biology of Trimero- 1947. Review of the Labyrinthodontia. Bull. rhachis (Amphibia: Temnospondyli). Mus. Comp. Zool., vol. 99, pp. 1-368. Jour. Paleont., vol. 53, pp. 1-17. 1952. Late Pennsylvanian and early Permian Olson, E. C., and G. E. Lammers vertebrates of the Pittsburgh-West Vir- 1976. A new brachyopid amphibian. In ginia area. Ann. Carnegie Mus., vol. 33, Churcher, C. S. (ed.), Athlon, essays on pp. 47-113. paleontology in honour of Loris Shano 1964. The skeleton of the Lower Carbonifer- Russell. Roy. Ontario Mus. Life Sci. ous labyrinthodont Pholidogaster pis- Misc. Publ., pp. 45-57. ciformis. Bull. Mus. Comp. Zool., vol. Panchen, A. L. 131, pp. 129-159. 1970. Anthracosauria. In Kuhn, 0. (ed.), 1966. Vertebrate paleontology. Third ed. Chi- Handbuch der Palaeoherpetologie. cago, Univ. Chicago Press, x + 468 pp. Stuttgart, Fischer, pt. 5a, vii + 84 pp. 1968. Notes and comments on vertebrate pa- 1972. The interrelationships ofthe earliest tet- leontology. Chicago, Univ. Chicago rapods. In Joysey, K. A., and T. S. Kemp Press, viii + 304 pp. (eds.), Studies in vertebrate evolution. 1969. A temnospondylous labyrinthodont Edinburgh, Oliver & Boyd, pp. 65-87. from the Lower Carboniferous. Kirtlan- 1973. Carboniferous tetrapods. In Hallam, A. dia, no. 6, pp. 1-20. (ed.), Atlas of paleobiogeography. Am- 1972. A Carboniferous labyrinthodont am- sterdam, Elsevier, pp. 117-125. phibian with complete dermal armor. 1975. A new genus and species of anthraco- Ibid., no. 16, pp. 1-8. saur amphibian from the Lower Car- Romer, A. S., and R. V. Witter boniferous of Scotland and the status of 1941. The skin of the rhachitomous amphib- Pholidogaster pisciformis Huxley. Phil. ian Eryops. Amer. Jour. Sci., vol. 239, Trans. Roy. Soc. London (B), vol. 269, pp. 822-824. pp. 581-640. 1942. Edops, a primitive rhachitomous am- 1977a. The origin and early evolution of tet- phibian from the Texas red beds. Jour. rapod vertebrae. In Andrews, S. M., R. Geol., vol. 50, pp. 925-960. S. Miles, and A. D. Walker (eds.), Prob- Shishkin, M. A. lems in vertebrate evolution. London, 1973. The morphology of the early Amphibia Academic Press, pp. 289-318. and some problems of the lower tetra- 1977b. On Anthracosaurus russelli Huxley and pod evolution. Trans. Palaeont. Inst., the family Anthracosauridae. Phil. Akad. Nauk. S.S.R., vol. 137, pp. 1-260 Trans. Roy. Soc. London (B), vol. 279, (in Russian). pp. 447-512. Smithson, T. R. 1980. The origin and relationships of anthra- 1980a. An early tetrapod fauna from the Na- cosaur Amphibia from the late Paleo- murian of Scotland. In Panchen, A. L. zoic. In Panchen, A. L. (ed.), The ter- (ed.), The terrestrial environment and restrial environment and the origin of the origin of land vertebrates. London, land vertebrates. London, Academic Academic Press, pp. 407-438. Press, pp. 319-350. 1980b. A new labyrinthodont amphibian from 1981. A jaw ramus of the Coal Measure am- the Carboniferous ofScotland. Palaeon- phibian Anthracosaurus from North- tology, vol. 23, pp. 915-923. umberland. Palaeontology, vol. 24, pp. 1982. The cranial morphology of Greererpe- 85-92. ton burkemorani Romer (Amphibia: Romer, A. S. Temnospondyli). Zool. Jour. Linnean 1930. The Pennsylvanian tetrapods ofLinton, Soc., vol. 76, pp. 29-90. Ohio. Bull. Amer. Mus. Nat. Hist., vol. Steen, M. C. 59, art. 2, pp. 77-147. 1931. The British Museum collection of Am- 1935. Early history of Texas redbeds verte- phibia from the Middle Coal Measures brates. Bull. Geol. Soc. Amer., vol. 46, of Linton, Ohio. Proc. Zool. Soc. Lon- pp. 1597-1658. don, 1931, pp. 465-504. 1983 HOOK: COLOSTEUS SCUTELLA TUS NEWBERRY 41

1938. On the fossil Amphibia from the Gas Watson, D. M. S. Coal of Nyrany and other deposits in 1921. On Eugyrinus wildi (A.S.W.), a bran- Czechoslovakia. Proc. Zool. Soc. Lon- chiosaur from the Lancashire Coal- don (B), vol. 108, pp. 205-283. measures. Geol. Mag., vol. 58, pp. 70- Tatarinov, L. P. 74. 1964. Superfamily Colosteoidea. Osnovy Pal., 1929. The Carboniferous Amphibia of Scot- land. Pal. Hungarica, vol. 1, pp. 219- vol. 12, pp. 67-69 (in Russian). 252. Thomson, K. S., and K. H. Bossy 1956. The brachyopid labyrinthodonts. Bull. 1970. Adaptive trends and relationships in Brit. Mus. Nat. Hist. (Geol.), vol. 2, early Amphibia. Forma Functio, vol. 3, pp. 315-392. pp. 7-31. 1963. On growth stages in branchiosaurs. Pa- Warren, A. laeontology, vol. 6, pp. 540-553. 1981. A homed member of the labyrintho- Wyman, J. dont superfamily Brachyopoidea from 1857. On a batrachian reptile from the Coal the early Triassic of Queensland. Al- formation. Proc. Amer. Assoc. Adv. Sci., cheringa, vol. 5, pp. 273-288. 10th meeting, 1856, pp. 172-173.