sign in sign up
<<
Home , Breviceps adspersus, Ceratodus, Choana, Cladoselache, Cleithrum, Cosmine

, , , AND PLESIOMORPHY

DONN E. ROSEN Curator, Department of American Museum of Natural History Adjunct Professor, City University of New York

PETER L. FOREY Principal Scientific Officer, Department of Palaeontology British Museum (Natural History)

BRIAN G. GARDINER Reader in Zoology, SirJohn Atkins Laboratories Queen Elizabeth College, London

COLIN PATTERSON Research Associate, Department of Ichthyology American Museum of Natural History Senior Principal Scientyifc Officer, Department of Palaeontology British Museum (Natural History)

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 167: ARTICLE 4 NEW YORK: 1981 BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Volume 167, article 4, pages 159-276, figures 1-62, tables 1,2 Issued February 26, 1981 Price: $6.80 a copy

ISSN 0003-0090

Copyright © American Museum of Natural History 1981 CONTENTS

Abstract ...... 163 Introduction ...... 163 Historical Survey ...... 166 Choana, , and Snout ...... 178 (A) Initial Comparisons and Inferences ...... 178 (B) Nasal Capsule ...... 182 (C) Choana and in Dipnoans ...... 184 (D) Choana and Nostril in Rhipidistians ...... 187 (E) Choana and Nostril in Tetrapods ...... 195 Nasolacrimal Duct, Labial Cavity, Rostral Organ, and Jacobson's Organ ...... 196 Paired Fins and Girdles ...... 202 (A) Paired Fin Skeleton ...... 202 (B) Pectoral Girdle ...... 216 Dermal Bones of the Skull ...... 221 Palate and Jaw Suspension ...... 227 (A) Historical ...... 227 (B) Dermal Bones of Palate ...... 227 (C) Palatoquadrate ...... 232 (D) Relation Between Palate and Braincase ...... 233 Hyoid and Arches ...... 235 Ribs and Vertebrae ...... 242 (A) Ribs ...... 242 (B) Interpretation of Gnathostome Vertebrae ...... 246 (C) Heterospondyly in Caturus furcatus, and Vertebral Homologies ...... 249 (D) Vertebrae of Rhipidistians, Lungfishes, and Tetrapods ...... 249 Dermal Skeleton ...... 252 (A) Cosmine ...... 252 (B) Folded Teeth ...... 254 Synapomorphy Scheme ...... 255 Conclusions ...... 262 Acknowledgments ...... 264 Literature Cited ...... 264

ABSTRACT We conclude that the internal (excurrent) nos- hypothesize, in agreement with most nineteenth- tril of Recent lungfishes is a true choana, as and many twentieth-century biologists, and in dis- judged by its comparison with (1) the internal nos- agreement with the current paleontological view, tril of a which opens that lungfishes are the sister group of tetrapods, through the bony palate internal to an arcade of and further that actinistians are the sister group maxillary and premaxillary teeth; (2) the choana of those two, and that is the sister of the Devonian ichthyostegid , and group of those three. We also conclude that the (3) nostril development in Recent urodeles. The characters used formerly to link Eusthenopteron idea that lungfishes might therefore be the sister with tetrapods either (1) are primitive for all bony group of tetrapods is compared with the compet- (including cladistians and actinopterygians) ing, deeply entrenched theory that rhipidistian or for living gnathostomes (including chondrich- fishes and eusthenopterids in particular include thyans); (2) are convergent with those of several the ancestor of tetrapods. Our own theory, de- groups of gnathostomes; (3) only justify the inclu- rived from study of Recent and material, sion of Eusthenopteron in a group with actinis- and an analysis of literature spanning 140 years, tians, dipnoans and tetrapods; or (4) are spurious. is framed in the context of a classification of the We attribute the century of confusion about the main groups of fossil and living gnathostomes: structure and position of lungfishes to the tradi- acanthodians, chondrichthyans, cladistians, ac- tional paleontological preoccupation with the tinopterygians, rhipidistians, actinistians, dip- search for ancestors, to the interpretation of Eus- noans, and tetrapods. In formulating our proposal thenopteron in the light of tetrapods and the re- we have reviewed the anatomy of the nasal cap- ciprocal interpretation of fossil amphibians in the sule, nostrils and related structures, paired fins light of Eusthenopteron, and to the paleontologi- and their girdles, dermal bones of the skull, palate cal predilection for using plesiomorphous char- and jaw suspension, hyoid and gill arches, ribs acters to formulate schemes of relationships. and vertebrae, and and tooth structure. We

"The convictions of our palaeontological colleagues are very real to them, and under the drive of these convictions they have quite honestly contended for their theories. The colour-blind man sees the scarlet robe and the green lawn the same colour, to him they are the same colour, but he is wrong." Kesteven, 1950, p. 99

INTRODUCTION Miles's (1977) descriptions and illustra- ture of the two outermost lip folds discussed tions of the head of the Devonian lungfish, by Allis, our attention was drawn to a large , caused the independent re- opening in the outermost fold of Neocerat- alization in London (BGG) and New York odus near the corner of the mouth that was (DER) that the internal nostrils of lungfishes first described by Gunther (1871, p. 515): are true choanae. This conviction meant to "At the angle of the mouth, and hidden be- us that the only synapomorphy of Miles's low a duplicature of the skin, there is an (1977, p. 316) Choanata (tetrapods and rhip- opening wide enough to admit an ordinary idistians; cf. Gaffney, 1979b, p. 93) is also quill (pl. XXX, fig. 2, a); it leads into a spa- present in lungfishes. It also meant to us that cious cavity (b), irregular in shape, clothed Allis's (1919, 1932a, 1932b) arguments for with a mucous membrane, and containing rejecting the internal lungfish naris as a coagulated mucous in which an immense choana, based partly on the interpretation of number of mucous corpuscles are deposited. certain folds of soft tissue as primary, sec- This cavity is separated from the cavity of ondary, and tertiary upper lips, might be the mouth by the membrana mucosa only, faulty. With the idea of reviewing the struc- and there is no direct communication be- 163 164 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 tween them; a branch cavity runs forward tians, and osteolepiforms in particular, are into the interior of the upper lip." the closest relatives, or direct ancestors, of The extent of this rostral structure, which tetrapods. This doctrine is so pervasive that Gunther called the labial cavity, is from the a researcher seeking the primitive anterior wall of the orbit forward to the an- condition of some structure turns automati- terolateral part of the nasal capsule (fig. 18). cally to the rhipidistians, usually to Eusthe- It is subtriangular, narrow anteriorly near the nopteron, forgetting that Eusthenopteron has nasal capsule and wider posteriorly in front been interpreted in the light of tetrapods and of the orbit where the internal wall of the that tetrapods have been interpreted in the cavity is thrown into a series of folds. The reciprocal light of Eusthenopteron. Our as- opening of the labial cavity to the outside is sault on the "rhipidistian barrier" begins through a short and broad tube that leads with an account of the history of opinion on directly to the pore in the outermost acces- lungfishes, tetrapods, rhipidistians and other sory lip (fig. 19). Visual examination of the crossopterygian fishes. We believe that this same region of the snout in premetamorphic account shows that current beliefs about of about 2 cm. in total length rhipidistians have grown up among a thicket (prior to development of pelvic fins) revealed of preconceptions and wishful thinking, es- neither a pore nor a labial cavity, but, in pecially the belief that processes can be dem- larger young, serial sections show these onstrated in the fossil record. To give one structures (fig. 20). The presence in larger classic example, Watson, in his Croonian young and adults of an apparently mucous- lecture on origins, wrote (1926, p. secreting organ extending between the nasal 189) "It is possible to view the problem as capsule and the orbit and its opening to the one of purely formal morphology, the estab- outside on the side of the head not far in lishment of a series of stages which show front of the eye, and its absence in premeta- intermediates between typical structures morphic young, suggested to us the anatom- and those of the homologous organs in Am- ical relations and ontogenetic history of the phibia. ... But such studies ... are now nasolacrimal organ of living amphibians. satisfying to few men: the centre of interest These two kinds of observations, and their has passed from structure to function, and it interpretations, that dipnoans may have a is in the attempt to understand the process choana and nasolacrimal structure synapo- by which the 's mechanism was so morphous with such features in tetrapods, profoundly modified ... that the attraction caused us to look critically at the dipnoans, of the problem lies." Watson emphasized and also the rhipidistians, Paleozoic fishes process both in morphology ("establishment now universally regarded as the closest rel- of ... intermediates") and in "the attempt atives of tetrapods. In searching for synapo- to understand the process" through which morphies that would justify the latter morphology is modified. Our approach is opinion, we have also reviewed the simpler. It rests on the assumption that mor- interrelationships of the main groups of fossil phological pattern must be recognized before and living gnathostomes. one can speculate about process. Schaeffer (1965, p. 115), discussing the We have accepted a number of assump- history of ideas about tetrapod origins, wrote tions or axioms. First, that tetrapods are that Cope (1892) "broke through the 'dip- monophyletic. Gaffney (1979a) listed 11 syn- noan barrier."' By "dipnoan barrier," apomorphies which characterize living and Schaeffer meant the late nineteenth-century fossil tetrapods as a monophyletic group. view that dipnoans are the closest relatives Four of these also occur in lungfishes (Gaff- ("nearest allies"-Huxley, 1876, p. 56) of ney's numbers 1, 8, 10, and 11: absence of tetrapods. Cope and his successors have suc- intracranial joint; pectoral girdle not joined ceeded so well that our problem has been to to skull; pelvic girdle with well-developed is- break through the "rhipidistian barrier," to- chiac ramus and pubic symphysis; well-de- day's unquestioned doctrine that rhipidis- veloped, ventrally directed ribs), but we re- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 165 gard the remainder, especially the carpus, accused of bias, but since most previous dis- tarsus, and dactyly as uncontradicted evi- cussions of the problem have relied largely dence of tetrapod monophyly. on one or another Paleozoic amphibian, we Second, we accept Gaffney's (1979a) char- shall select a living group, the urodeles. As acterization of as the sister is well known (Kerr, 1932; Holmgren, 1933, group of the remaining tetrapods (Gaffney's 1949a, 1949b; Kesteven, 1950; Lehman, Neotetrapoda). He listed five synapomor- 1956; Fox, 1965), urodeles share many fea- phies which distinguish all neotetrapods1 tures of soft anatomy with lungfishes. Our from Ichthyostega, and we accept all those selection of urodeles as primitive tetrapods but the last (ethmosphenoid and otico-occip- means that we rate these features as synapo- ital portions of braincase not separated by morphies of choanates, symplesiomorphies suture), preferring to wait until the reputed of tetrapods. division in the braincase of Ichthyostega is Third, rhipidistians are not a monophyletic described. Accepting Ichthyostega as the group, so that no unique statements can be sister group of all other tetrapods, we turn made about their structure. So far as we to it in seeking primitive tetrapod conditions know, the same is true of osteolepiforms. in the , nostrils, etc. But there are This means that osteolepiforms have to be many structures which are still unknown in treated as a series of nominal taxa, among ichthyostegids (e.g., manus, braincase, de- which Eusthenopteron foordi Whiteaves is tails of vertebral column and course of sen- most completely known, and is selected for sory canals, all soft parts), and we have to discussion. The porolepiforms are united by look elsewhere for information on primitive one derived character, dendrodont tooth tetrapod conditions. At present, there is no structure (Schultze, 1970a), known in Poro- acceptable phylogeny of tetrapods (see L0v- lepis, , , and trup, 1977, p. 176; Gaffney, 1979a, 1979b; Laccognathus. On the basis of this one syn- Carrol and Holmes, 1980). Lack of such a apomorphy, we regard the group as phylogeny leaves us three alternatives: to monophyletic. follow Gaffney (1979b) in regarding lepo- Fourth, Dipnoi are monophyletic. The spondyls as the sister group of other tetra- form ofthe snout and dentition (radiate tooth- pods; or to survey Paleozoic "amphibians" plates or marginal tooth ridges-Miles, 1977, and select what appear to be primitive con- p. 288) are synapomorphies of all those Re- ditions; or to select a living tetrapod group. cent and well-preserved fossil fishes currently In any case our selection could be justified placed in the Dipnoi. Whether other Paleo- only by outgroup comparison, i.e., compar- zoic fishes, such as Powichthys (Jessen, ison with non-tetrapod groups, and would 1975), are dipnoans we cannot say. therefore depend on a previous solution to Fifth, osteichthyans are monophyletic, the problem we are trying to solve. Of and their sister group is the . course, all previous attempts to solve that Sixth, actinopterygians are monophyletic, problem have depended on similar though and include the Chondrostei and Neoptery- rarely acknowledged circular arguments. gii; seventh, the Cladistia (=Brachiopterygii; Whichever alternative we choose we can be Polypterus and Erpetoichthys) are monophy- letic; and, eighth, actinistians () also are monophyletic. Finally, almost every topic discussed in ' We do not, however, endorse the use of Gaffney's this paper is already embedded in deposits term Neotetrapoda, or any other named taxon for Re- cent organisms formulated for, or because of, the cla- of theory and interpretation-vertebral the- distic position of a fossil; elsewhere (Patterson and Ro- ory of the skull, fin-fold theory, gill-arch con- sen, 1977) reasons were advanced -for believing that tributions to the skull and forelimb skeleton, allowing to affect the taxonomic structure of clas- bone fusion versus bone loss, constancy of sifications of extant organisms is inevitably self-defeat- bone/sensory canal relationships, and so on. ing. In dealing with these topics, we do not claim 166 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 to have eyes unclouded by theory. But we only changes in anatomical relationships that have tried to be guided by one principle, that are demonstrable are those which occur in agreed interpretations of soft parts are only the ontogeny of one or another living attainable among living , and that the species.

HISTORICAL SURVEY When living lungfishes were first discov- African , followed by a detailed ered, lepidosirenids (Fitzinger, 1837; Natter- account in 1841. Owen was the first (1839, p. er, 1837) and neoceratodontids (Krefft, 1870) 331) to notice the resemblance between Re- were both recognized as amphibians. The cent lungfish toothplates and the Mesozoic purpose of this review is to map the path Agassiz (1838). Owen disagreed from those first thoughts to recent opinions with Fitzinger's and Bischoffs opinion that such as that found in today's textbooks, that lungfishes were amphibians and differed lungfishes are more remote from amphibians from Bischoff on two particulars, the auricle than are coelacanths, or those of von Wah- and internal nostril, finding both the auricle lert (1968) and Bjerring (1977), that lungfish- and nostril to be single. He gave an interest- es are more remote from amphibians than are ing and thorough discussion of the system- actinopterygians, or of Jarvik (1968a), that atic position of Protopterus, considering they may be chondrichthyans. every system of the body in a search for the Lungfishes are now universally regarded essential character that distinguishes fishes as fishes (Kesteven, 1950, was the most re- from tetrapods. After surveying the skeleton cent exception), but in the middle of the and the respiratory, digestive, reproductive, nineteenth century there was a controversy and nervous systems Owen concluded that over the class of to which they none of these is essentially different in fishes belonged. The first detailed account of Lep- and tetrapods, and wrote (1841, p. 352) "In idosiren was Bischoffs (1840a, 1840b). He the organ of smell we have, at last, a char- agreed with Fitzinger that the animal was an acter which is absolute in reference to the amphibian, and since several zoologists dis- distinction of Fishes from . In every agreed, judging it to be a fish, Bischoff gave Fish it is a short sac communicating with the the reasons for his belief in full (table 1). external surface; in every it is a canal L0vtrup (1977, p. 43), in his axiomatiza- with both an external and an internal open- tion of phylogenetics, gave as one theorem ing. According to this test, the Lepidosiren "If, in a basic [i.e., three-taxon] classifica- is a Fish: by its it is known not to be a tion, one . .. taxon has several characters in Reptile . .. so that at the close of our anal- common with both of the other taxa, then ysis we arrive at this very unexpected result, one of the two sets of characters will repre- that a Reptile is not characterized by its sent the presence of plesiotypic characters lungs, nor a Fish by its , but that the only absent, and/or the absence of teleotypic [i.e., unexceptionable distinction is afforded by apomorphic] characters present, in the third the organ of smell." taxon." If the three taxa were Lepidosiren, Having seen Owen's 1839 paper, Bischoff a and a cow, we see that in table 1 added a supplement to his own, commenting the four salmon-like characters (fishlike, nos. on the differences between his own obser- 1, 2, 4, 5) consist of three primitive charac- vations and Owen's. In particular he reiter- ters absent in cows (1, 2, 4) and one derived ated his opinion on the divided auricle and character (5) absent in fishes, as L0vtrup's the internal nostril. Milne Edwards (1840) theorem predicts. added some remarks to the French transla- While Bischoffs work was in press, Owen tion of Bischoff's paper, pointing out that (1839) published a preliminary account of the dissection of a Paris specimen ofProtopterus 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 167

TABLE I Characters of Lepidosiren paradoxa (From Bischoff, 1840b, pp. 144-151) Fishlike Amphibian-like 1. Scales 1. Internal nostril 2. Sensory canals 2. Large paired lungs, connected with the oesophagus 3. No ossified vertebrae, and by a large glottis penetrating base of cranium 3. Gills much reduced, most blood bypasses them 4. Opercular bones 4. Heart with two auricles, one receiving blood 5. No external ear from lungs 6. Labial and 5. Conus with two valves which separate the pulmonary and branchial blood confirmed Bischoffs observation of a divid- Fishes and Reptiles . .. an order uniting fish- ed auricle and internal nostrils. es with the amphibian reptiles, seems to be In this first exchange in the controversy, the true mode of giving it its proper place." we see Bischoff correctly interpreting syn- By "systematizing zoologists" M'Donnell apomorphies, and Owen basing his opinion might have been referring to Owen (1859), partly on a mistaken observation, and partly who named a group Haemacrymes for cold- on primitive characters. Owen never (e.g., blooded vertebrates (equivalent to his 1866 1846, p. 256; 1866, p. 474) agreed that the Haematocrya) and wrote that within this auricle was divided in lungfishes. He later group "Archegosaurus conducts the march (e.g., 1846, p. 201) allowed that there were of development from the fish-proper to the two nostrils, but argued that neither opened Labyrinthodont type: the Lepidosiren con- into the mouth. This first exchange set the ducts it to the perennibranchiate batrachian pattern for the next 20 years. Those workers type." Or M'Donnell might have been refer- who believed lungfishes to be amphibians ring to Darwin and Wallace. In either case, (Milne Edwards, 1840; Hogg, 1841; Heckel, such views as his were readily adapted to the 1845; Melville, 1848, 1860; Vogt, 1845, 1851; evolutionary doctrine, and in Haeckel's first Asmuss, 1856; Gray, 1856) relied on the trees (1866) the Dipnoi are placed as the sis- same characters as Bischoff, and included ter group of the tetrapods, with a line of those who had observed living animals "Dipnoi ignoti" extending back into the De- (Gray, Melville). Those who put lungfishes vonian. In Haeckel's later versions (e.g., fig. among fishes (Muller, 1840, 1844; Oken, 1) lungfishes are placed as the direct ances- 1841; Fitzinger, 1843-a change of mind; tors of tetrapods. Thus with acceptance of Agassiz, 1843; Hyrtl, 1845; Newman, 1857; evolution, the problem of distinguishing tet- and others-see Dumeril, 1870) were, when rapods from fishes, i.e., the problem of char- they gave reasons, weighting primitive char- acterizing tetrapods, disappeared, since dif- acters. ficulties in distinguishing the two groups A third viewpoint was possible, that it was were agreeable to evolutionists. futile to search for an essential distinction In 1870, Krefft announced the discovery between fishes and amphibians. Heckel of Neoceratodus, "a gigantic Amphibian." (1851) took this view, as did M'Donnell As with the lepidosirenids a third of a cen- (1860, p. 18) in these words: "our present tury before, such opinions found no support knowledge . . . leads us to regard it in the British Museum, where Gunther wrote [Protopterus] as a form so transitional as to the first detailed account of Neoceratodus support the view taken by some systematiz- (1871). Before reviewing Gunther's conclu- ing zoologists in later days, that no distinctly sions, it is worth recalling that he was, like defined line of demarcation exists between most of the senior naturalists in the British 168 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

Anura Peromela Plectognathi Salamandrina Lophobranclhii Labyrinthodonta Cryptobranchia Stichobranchii ! Perennibranchia Stegocephlia Lissamphibla Enchelygenes Phractamphibia

Thrissogeiies Amphibia Physostomi Teleostel Semaec )pteri Dipneumones Fulcrati Thrissopides Monopneumones

Ctenodipterini EfU1Cri -'1 Amiades Rhombifeiri Dipneusta Leptolepides Crossopterygii (?) Sturiones

Cephalaspides Cycliferi

Placodermi Tabuliferl Gan4oides Rajacei Squalacei Chimaeracei Holocephali Plaglostomi Selachii Pisces Amphirhina CycloSistoma mono rhina . I1 Craniota Acrania FIG. 1. Phylogeny of lower vertebrates, from Haeckel (1889, p. 613). 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 169

Museum at that time, an opponent of evo- good arguments might be adduced for the lution. Gunther's most recent biographer re- assertion that it is an amphibian or a fish, or marks that in the memoir on Neoceratodus both, or neither." In this paper, Huxley pro- he "had taken the opportunity to trail his posed the names Chondrichthyes and Oste- coat before the evolutionists" (A. E. Gun- ichthyes, but included in the latter only the ther, 1975, p. 335). ganoids and , placing dipnoans in a Gunther (1871) found that Neoceratodus group Herpetichthyes named "for that par- was less amphibian-like than the lepidosiren- ticular stage of evolution of which ids, since it had ganoid valves in the conus, both the typical Fishes and the typical Am- a single lung, well-developed gills, and so on. phibia are special modifications." Huxley's Concerning the nostrils of lungfishes, it diagram seems to imply that he regarded the would be hard to find a better illustration of crossopterygians as descendants of dipnoan- the idea that all observation is theory-laden like fishes. Thus, in contrast to Gunther, than the contrast between Gunther's view Huxley sought to blend lungfishes with both (1871, p. 517) that both the anterior and pos- crossopterygians and tetrapods. terior nostrils open inside the mouth, and Similar uncertainties about the status, and Owen's (1866, p. 329) that neither opens content, of the Crossopterygii are evident in within the mouth. Gunther concluded that the publications of other workers during this lungfishes were ganoid fishes, and named a period (see Schaeffer's 1965 review). For ex- new subclass Palaeichthyes to include all ample, Gill (1872; pp. xxxii, xliii) had the fishes except teleosts. Gunther was the first crossopterygians as the sister group of coela- to realize that Paleozoic fishes such as Dip- canths in one of his genealogical trees, and terus and Phaneropleuron, included by Hux- as the sister group of actinopterygians, with ley (1861) in his Crossopterygii, were lung- a question mark, in the other (our fig. 2); fishes. Although his discovery carried the Haeckel's question mark against the cros- dipnoans back into the Devonian, as Haeck- sopterygians in figure 1 is also symptom- el's trees demanded, Gunther saw this not as atic. Cope (1872, 1885, 1887, 1892) vacillated evidence of evolution, but as "a proof of the on the content of the Crossopterygii [Wood- incompleteness of the palaeontological rec- ward (1891, p. xxii), whose Catalogue sta- ord" and as "the most remarkable example bilized things, complained of "the somewhat of persistence of organization, not in Fishes fluctuating classifications of Cope"]. In 1887 only, but in Vertebrates." Cope removed the (Tristichopter- Huxley, through whose Crossopterygii idae) and Actinistia (Coelacanthidae) from Gunther had trailed his coat, replied in 1876. the Crossopterygii, leaving in the latter only He distinguished autostylic, hyostylic, and the Holoptychiidae, Osteolepididae, Polyp- amphistylic skulls, showed that lungfishes teridae, and Phaneropleuron (a dipnoan that had the same autostyly as amphibians, and he had formerly excluded). By 1892 the hol- discussed the paired fins of Neoceratodus as optychiids and osteolepids had also joined forerunners of tetrapod limbs. On classifi- the rhipidistians and actinistians, leaving the cation, he argued against Gunther's Palae- polypterids as the only crossopterygians. ichthyes, and for Muller's original (1844) A passage from Cope's 1892 paper is cited groups (Dipnoi, Ganoidei, etc.). Huxley by Schaeffer (1965, p. 115) as the source of maintained that Dipnoi are the "nearest al- "our present understanding of amphibian or- lies of the Amphibia" (p. 56), whereas in igin." Like Gill, Huxley, Haeckel, and other 1861 (p. 27) he had argued that they are evolutionists of this period, Cope had pre- "next of kin" to the crossopterygian gan- viously believed that tetrapods were de- oids. This ambiguity was not cleared up in scended from dipnoans (e.g., Cope, 1884, p. Huxley's next contribution (1880). There he 1256, "The Batrachia have originated from wrote of Neoceratodus (p. 660), "This won- the sub-class of fishes, the Dipnoi, though derful creature seems contrived for the illus- not from any known form"). But in 1892 he tration of the doctrine of Evolution. Equally emphasized the lack of the maxillary arch in 170 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

14 fl--ELEOST SERIES (I.-EI.). p 0 >

- 0 (XIII.). x Om r L CYCLOGANOIDZAR}HOMBOGANOIDEA (XII.).

-?CROSSOPTERYGII (XI.). rAx

RHOLOCEPHALI (VI.).

-RAIAR (V.).

SQUALI (IV.).

W 'HYPXROTB:TA (IL). -P4

-LEPTOARDII.-cxuaHosTom (I.). FIG. 2. A genealogical tree of the relations of major groups of vertebrates, from Gill (1872, p. xliii).

Dipnoi, their large dorsal and anal fins, and (1891, p. 341). Pollard also asserted that the the fact that autostyly developed during on- posterior nares of dipnoans are not homolo- togeny in amphibians, and concluded that gous with the choanae of urodeles. Kingsley amphibians therefore originated from a hyo- (1892) argued in the same vein, that since stylic fish with a complete maxillary arch, autostyly developed during ontogeny in uro- feeble median fins, and amphibian-like deles, they could not be descended from dip- paired fins. The rhipidistians, and Eusthe- noans. Baur (1896) reached similar conclu- nopteron in particular, were offered as such sions from a study of labyrinthodonts, and fishes. was the first to suggest (p. 670) that the re- Pollard's (1891, 1892) papers on Polypt- semblances between dipnoans and amphibi- erus are a forthright attack on dipnoan-tetra- ans are merely parallelisms. pod relationships. He argued that Polypterus Dollo (1896) produced the first detailed resembles the urodeles in features of the in- phylogeny of dipnoans, and summarized ear- ner ear, braincase, palate, musculature, lier work on the interrelationships of dip- nerves and ribs, and concluded that "the an- noans, crossopterygians, and tetrapods. He cestry of the Urodela must be sought among expressed his conclusions in a diagram (fig. the Crossopterygian forms now represented 3) which places the dipnoans as the sister only by Polypterus and Calamoichthys" group of tetrapods, and crossopterygians as 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 171

MAMMIFtRES. OISEAUX. PHYSOCLYSTES.

REPTILES. PHYSOSTOXES.

BATRACIENS. DIPNEUSTES. Amioldes.

Vie terrestre. Vie en eau corrompue, puis dans la vase. Lepidost6oldes.

Crossopt&ygiens. Acipensboldes.

Continuation de la vie Retour progressif, et en eau douce. de plus en plus complet, a as mer. HOLOCfrHALIS. SLAC1ENS. ICHTHYOTOINS.

GANOIDES. PLEU1OPTi1YGIES. AcAmoDIsES. ? OSTRACODMES.

Chondropt6rygiens. Ost6opt6rygiew.

En geniral, sejour ininterrompu Adaptation generale dans la mer. a l'eau douce.

Gnathostomes.

Souche marine.

POISSONS.

FIG. 3. Phylogeny of gnathostome vertebrates, from Dollo (1896, p. 113). 172 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 the group ancestral to both. This was the terygians and labyrinthodonts (Wood- generally accepted scheme for the next 40 ward, 1898, p. 123-a synapomorphy of years, though with many variants. Thus, by bony vertebrates) the end of the nineteenth century, the dip- 6. Paired fins of Eusthenopteron more like noan ancestry of tetrapods favored by early tetrapod limbs than are those of dipnoans evolutionists (fig. 1) had been replaced by a (Cope, 1892; for our view see p. 209) 7. Polypterus has dorsal ribs (Pollard, 1892; crossopterygian ancestry (fig. 3). In cladistic Baur, 1896: for our view see p. 242) terms, figure 3 implies that dipnoans are the 8. Fossil crossopterygians and labyrintho- sister group of tetrapods (i.e., those two donts have an infradentary (splenial) in groups share synapomorphies), whereas the the (Woodward, 1898, p. 123- crossopterygians are a paraphyletic group, so have all non-actinopterygian some members of which share synapomor- osteichthyans) phies with dipnoans and tetrapods, but 9. The median fins are reduced in Paleozoic whose relationships are yet to be worked crossopterygians (Cope, 1892; Baur, out. Of course, we find nothing to quarrel 18%: no comment necessary) with in that statement. It is worth summariz- 10. Fossil crossopterygians and labyrintho- ing the evidence so far produced to cause the donts have folded teeth conceptual change from figure 1 to figure 3: As is clear from the above list, the char- the following characters were used by late acters so far offered in support of the rela- nineteenth-century biologists to relate tetra- tionship between crossopterygians and tet- pods, dipnoans, and crossopterygians in the rapods were a melange of primitive way shown in figure 3: characters, characters of Polypterus alone, and vague assertions about Paleozoic cros- I. Characters excluding dipnoans from tetrapod ancestry sopterygians. Concerning fin structure, 1. Dentition too specialized (Boas, 1882, p. Goodrich's (1901) account of the 557: cf. Miles, 1977, fig. 157-the tooth and girdle ends with the statements that Po- plates are a synapomorphy of higher dip- lypterus "probably belongs to the actinop- noans) terygian line" and that Eusthenopteron is 2. No maxillary arch (Cope, 1892: for our "very far removed from Polypterus." view see p. 180) Nevertheless, the belief that the Crossop- 3. Autostyly develops during ontogeny in terygii, containing polypterids and rhipidis- amphibians (Cope, 1892; Kingsley, 1892) tians, was a natural group, and that this group 4. Ribs not homologous in dipnoans and tet- gave rise to the tetrapods, continued to gain rapods (Pollard, 1892; Baur, 1896: for our Since view see p. 242) strength. Polypterus lacks many of the II. Characters relating crossopterygians to tetra- derived features in which dipnoans agree pods with tetrapods, it followed that these dip- 1. Large maxilla, dentition not modified noan/tetrapod similarities must have arisen (Cope, 1892; Pollard, 1892: cf. actinop- in parallel, as Baur (1896) had realized. This terygians-a synapomorphy of osteich- was acceptable to some (e.g., Boulenger, thyans?) 1901, p. 30; Bridge, 1904, p. 520), but not to 2. Hyostyly (Cope, 1892: cf. -a syn- others (e.g., Semon, 1901, who listed more apomorphy of gnathostomes?) than 20 resemblances between dipnoans and 3. Most of the bones can be homologized in amphibians; Jordan, 1905, p. 600; Goodrich, Polypterus, fossil crossopterygians and labyrinthodonts (Pollard, 1892; Baur, 1909, p. 230). Goodrich (1909) believed that 1896; Woodward, 1898, p. 123-for our dipnoans "present many striking points of view see p. 221, 227) resemblance to the Amphibia, which cannot 4. Many sclerotic plates in fossil crossopte- all be put down to convergence." In his phy- rygians and tetrapods (Woodward, 1898, logenetic diagram (p. 29) Goodrich reverted p. 123-a synapomorphy of sarcopteryg- to an unfashionable view in placing dipnoans ians: cf. Miles, 1977, p. 249) as the sister group of teleostome fishes, with 5. Pineal foramen present in fossil crossop- the tetrapods entered, with a query, as a tri- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 173

,COELACANTHS C O E LAC A N T H S

RHIPIDISTIANS R H I P I D I S T I A N S A B DIPNOANS DIPNOANS

TETRAPODS TETRAPODS

,DIPNOANS COELACANTHS

COELACANTHS DIPNOANS

CD D,

RH I PI DI ST IAN S , RHIPIDISTIANS

TETRAPODS ' TETRAPODS FIG. 4. Cladograms to show evolution of opinion on interrelationships of crossopterygians, lung- fishes, and tetrapods. A, Goodrich (1924); B, Watson (1926), Romer (1933); C, Save-Soderbergh (1935), Westoll (1961, etc.), Miles (1975); D, Miles (1977), Wiley (1979).

chotomy with those two groups. In that dia- mous with "rhipidistian," in a paper includ- gram, polypterids are also marked by a ing new information on Megalichthys, and query, since Goodrich's assessment was that the observation that the mode of tooth re- they are closer to actinopterygians than to placement was identical in rhipidistians and coelacanths and rhipidistians. Goodrich's labyrinthodonts. opinions were clearer by 1924. In that paper, Gregory (1915) published the first detailed he still insisted that the resemblances between review of the question of tetrapod origins. lungfishes and tetrapods "can scarcely be On dipnoans, Gregory believed that while due to convergence" and wrote "the earliest their similarity with tetrapods, and urodeles diverged into teleostome and in particular, "might be ascribed to conver- dipnoan branches and that the tetrapods gence" they implied "a similarity in the 'po- arose from the base of the latter" (cf. fig. tential of evolution,' that is, of structural 4A). Goodrich is often cited as one of the possibilities, in the forerunners of these architects of the theory of the crossopteryg- groups" (p. 322). Gregory argued that "The ian ancestry of tetrapods (e.g., Romer, 1956, only crossopterygians that can claim even p. 159; Schaeffer, 1965), but in our reading remote relationships with the Amphibia are his work shows nothing of the sort. the Devonian Rhipidistia, especially the Os- After Goodrich's 1909 work, Polypterus teolepidae and the nearly allied Rhizodonti- played a less central role in discussions of dae" (p. 326). In his review he concentrated tetrapod origins. For example, Watson on seeking homologies in the dermal bones (1912) used "'crossopterygian" as synony- of the skull and in the forelimb skeleton be- 174 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 tween rhipidistians and labyrinthodonts. mer to dipnoan/tetrapod similarities: they Concerning the forelimb, Gregory tried to may be parallelisms or retained primitive settle the much-discussed question of the rel- characters. ative status of the fin of Neoceratodus and In 1932 Save-Soderbergh announced the of rhipidistians. Unlike many of his prede- discovery of the Devonian ichthyostegids. cessors, Gregory rejected "the traditional These animals have had a profound effect on view that the 'archipterygia' of Dipnoi are subsequent discussions of tetrapod origins, primitive structures and [regarded] the im- because they are both stratigraphically older perfect archipterygium of the Devonian Os- and anatomically more primitive than the teolepis as more primitive" (p. 350). Thus amphibians that Watson had Gregory imagined that tetrapods are de- used to demonstrate similarities between os- scended from rhipidistians, and argued for a teolepids and early tetrapods. As Parrington "common origin of the Dipnoi and Crossop- (1967, p. 231) put it, "It was to be expected, terygii" (p. 325). therefore, that any Amphibia from the Upper Gregory did not describe exactly how he Devonian would be intermediate in their envisaged the interrelationships of rhipidis- structures between the Middle Devonian os- tians and dipnoans, but this was made clear teolepids and the Carboniferous labyrintho- by Watson (1926), the next major reviewer donts, but when discovered the ichthyo- of tetrapod origins. He wrote (p. 195) "the stegids did not conform at all well to this most primitive known Dipnoan is ... Dip- expectation. While their skulls showed some terus valenciennesi .. . this-fish presents so very primitive features which might have many resemblances in structure to the con- been expected, the pattern of the dermal temporaneous Osteolepids as to show that bones did not conform to plan." In attempt- the two groups arose from a common ances- ing to explain the differences in the skull roof tor not much earlier in date. It is from this between osteolepids and ichthyostegids, hypothetical fish that I believe the Amphibia Save-Soderbergh adopted an explanation in- to have arisen." Thus Watson envisages a volving differential fusions from an ancestral trichotomy of dipnoans, rhipidistians, and pattern which, as he said (1934, p. 5), is tetrapods (fig. 4B), this arrangement, rather "astonishingly" dipnoan-like. After the dis- than derivation of dipnoans and tetrapods covery of the ichthyostegids, discussions of from rhipidistians (fig. 3), being necessitated the dermal skull roof shifted away from the by the intracranial joint of rhipidistians (cf. search for homologies between rhipidistians Watson and Gill, 1923, p. 210), which Wat- and tetrapods, toward discussions of the fu- son regarded as a specialized feature. On the sion theory and of Westoll's (1938) alterna- resemblances between dipnoans and uro- tive explanation requiring changes in pro- deles, Watson wrote (1926, p. 190) "it is cer- portion (see Jarvik, 1967; Schmalhausen, tain that the existing resemblances depend 1968; and section on Dermal Bones). on the parallel evolution of not distantly re- Of his 1932 paper, Save-Soderbergh wrote lated stocks." Romer (1933) followed Wat- (1935, p. 201) "the investigations of the De- son in adopting a trichotomous diagram of vonian Ichthyostegids ... show that these the relationships between dipnoans, tetra- . . . are more nearly related to the Crossop- pods and rhipidistians (his fig. 51), but Ro- terygians [i.e., rhipidistians and coelacanths] mer preferred to regard the similarities be- than to the Dipnoans." Thus he had adopted tween dipnoans and amphibians as "retention the scheme of relationships shown in figure of characters present in their crossopteryg- 4C, though it is not clear to us what char- ian ancestors." Given the trichotomous pat- acters of ichthyostegids or crossopterygians tern of relationships inferred by Watson and prompted this conclusion. In any case, this Romer (fig. 4B), it should be pointed out that opinion was short-lived, for in 1933, after special similarities between rhipidistians and elaborating on the characters common to tetrapods are open to just the same interpre- dipnoans, crossopterygians, and tetrapods tations as those applied by Watson and Ro- and naming the group Choanata for them, 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 175

Save-S6derbergh suggested (p. 118) that the letic, and "that all known tetrapods can be tetrapods were polyphyletic, urodeles being derived from a single basic structural pat- related to dipnoans, and the labyrinthodonts tern, which resembles osteolepid crossopte- possibly being polyphyletic. The idea of tet- rygians in so many details that this group is rapod polyphyly, and of direct dipnoan/uro- the only possible source of proto-tetrapods" dele relationships, was taken up by Holm- (p. 79). This conclusion is not surprising, for gren (1933; see also 1939, 1949a, 1949b), at Westoll's discussion was much influenced by first on the structure of the limbs. He con- his (1938) discovery of Elpistostege, "an im- cluded (1933, p. 288) "regarding the limbs portant proto-tetrapod" (Westoll, 1943, p. there is a greater agreement between a dip- 78) now interpreted as a panderichthyid fish noan and an urodele, than between an uro- (Vorobyeva, 1977a), or, in Westoll's classi- dele and an anuran." fication, an osteolepid. Later, Westoll wrote Save-Soderbergh adopted Holmgren's (1958b, p. 100; 1961, p. 605) of "overwhelm- views on urodele/dipnoan relationships (1934, ing evidence" of the descent of tetrapods 1935) and so began the central theme of mod- from osteolepid rhipidistians. On lungfishes, em discussions of tetrapod origins, the dis- Westoll (1949; Lehmann and Westoll, 1952) pute between the advocates of polyphyly, did not accept Jarvik's argument that the and the advocates of monophyly. Holm- posterior nostrils of dipnoans are not choa- gren's original advocacy of urodele/dipnoan nae, and maintained that dipnoans, like tet- relationships can be seen as an attempt to rapods, are descended from osteolepids. reconcile the conflicting claims of neontolo- Westoll's scheme of relationships (e.g., gy (resemblances between Recent dipnoans 1961, fig. 1) is like that introduced by Siive- and amphibians, especially urodeles) and pa- Soderbergh (fig. 4C), with coelacanths closer leontology (resemblances between labyrin- to tetrapods than are dipnoans, but the coel- thodonts and rhipidistians). However, Holm- acanths are regarded as descendants of po- gren's views received little support (see rolepiforms, and, following Jarvik, the pos- references cited in Holmgren, 1949a), and sibility of urodele descent from porolepiforms the emphasis soon passed back to paleontol- is allowed. Romer's views (e.g., 1968) and ogy, with Jarvik's (1942) monumental work his trees (e.g., 1956, fig. 19) are similar to on the snout. In that monograph, Jarvik gave Westoll's; like Westoll he did not accept Jar- incomparable accounts of the morphology of vik's argument that dipnoans have no choa- the snout in Eusthenopteron and porolepi- nae, but unlike Westoll he never allowed the forms. Like Holmgren and Siive-Soder- possibility that tetrapods are non-monophy- bergh, he advocated a diphyletic origin of letic. tetrapods, but as urodele ancestors he re- On the relationships of dipnoans, crossop- placed the dipnoans by the porolepiforms. terygians, and tetrapods most authorities Jarvik's principal reason for excluding the have followed the pattern shown in figure 4C dipnoans from consideration was his argu- (e.g., Trewavas et al., 1955-"the closer re- ment that they have no choanae (see section lationship of [tetrapods to] Osteolepidoti and on Nostrils). Coelacanthini than to the Dipnoi"), and so Westoll's 1943 review of tetrapod origins, today's textbook view became established. though published after Jarvik's monograph, As for the dipnoans, they were free to was written without knowledge of it, because wander farther down the family tree, since of wartime conditions. Westoll's paper is in coelacanths, now known (Latimeria) to lack the tradition of Watson's (1926) review. Like choanae, and judged to show nothing in com- Watson, Westoll emphasized process, both mon with amphibians, were interposed be- in his discussion of the significance of the tween lungfishes and tetrapods. White (1965) change from fish to tetrapod, and in his mor- argued that dipnoans were distinct from oth- phology, where sequences of fossils are se- er osteichthyans; Schaeffer (1968) that they lected to illustrate trends. Westoll's general formed a trichotomy with actinopts and cros- conclusion was that tetrapods are monophy- sopts; Bertmar (1966) that they are related to 176 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

FIG. 5. Phylogeny of gnathostomes, modified from Jarvik (1960, fig. 28). actinopts; von Wahlert (1968) that they are ans and chondrichthyans (cf. fig. 5). Despite the sister group of all osteichthyans; and Jar- a few dissenting neontologists (e.g., Fox, vik (1968a, 1968b), citing resemblances be- 1965), it is now accepted that the resem- tween dipnoans and holocephalans, con- blances between dipnoans and amphibians cluded (1968b, p. 518) "if we ask which of are convergent, and that the resemblances the recent vertebrate groups is the sister between rhipidistians, especially osteolepi- group of dipnoans (holocephalians? coel- forms, and tetrapods are evidence of rela- acanthiforms?) no answer can be given." tionship (i.e., synapomorphies). What, then, Thus the fossils had taken over. Dipnoans, are these resemblances? Above, those used placed as amphibians or the closest relatives by nineteenth-century biologists are listed; of tetrapods by nineteenth-century neontol- the following list contains characters empha- ogists, had been ousted by the rhipidistians, sized by twentieth-century workers to relate and relegated to a limbo between osteichthy- osteolepiform rhipidistians and tetrapods, 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 177 additional to or amplifying those of nine- 1954, 1963, 1964, 1966, 1967, 1972, 1975) on teenth-century biologists: osteolepiforms. But it must be emphasized that Jarvik never pretended to cite ch'arac- 1. Detailed correspondence between bones of ters common to osteolepiforms and tetra- skull roof and cheek (e.g., Westoll, 1943; cf. Jarvik, 1967; Parrington, 1967) pods; instead, he cited characters common 2. Intracranial joint is immobilized in some os- to Eusthenopteron and . Our main task teolepiforms (e.g., Westoll, 1943; Vorobyeva, in this paper is to review the characters listed 1977a) above, in osteolepiforms, tetrapods, and dip- 3. Dermal bones of palate "practically identi- noans. Some of them (nos. 7, 9, 20) may be cal" in osteolepids and primitive tetrapods immediately discarded as primitive for os- (Westoll, 1943, p. 84) teichthyans or gnathostomes. Others may be 4. Choana present, with the same relationship discarded because of lack of published evi- to the dermal bones in both groups (Jarvik, dence (nos. 8, 19), or because dipnoans are 1942; Westoll, 1943) known to share the same features (nos. 2, 5. Mandible "almost identical" in early tetra- pods and osteolepids (Westoll, 1943, p. 85) 10, 12; cf. Miles, 1977). The remainder are 6. Tetrapod septomaxilla is found in osteolepi- discussed in subsequent sections. forms (Panchen, 1967) During the last few years, there has been 7. Opercular and preopercular present in ichthy- a changed approach to dipnoan relation- ostegids (Jarvik, 1952) ships, stemming mainly from cladism. Miles 8. Suture in braincase of ichthyostegids repre- (1975) reviewed the question and concluded sents intracranial joint (Jarvik, 1952) that the textbook solution (fig. 4C) was cor- 9. Notochord penetrates base of braincase in rect, but later (1977) changed his mind, after ichthyostegids (Jarvik, 1952) studying the Devonian Gogo lungfishes, and 10. Cranial cavity and its inferred contents simi- adopted a new scheme (fig. 4D). Meanwhile, lar in Ectosteorachis and amphibians (Romer, 1937) others had suggested that coelacanths be- 11. Osteolepiforms have ethmoid, basal, ascend- longed elsewhere (L0vtrup, 1977-sister ing and otic processes on the palatoquadrate group of chondrichthyans; Wiley, 1979-sis- (Jarvik, 1954) ter group of osteichthyans) on evidence from 12. Hyomandibular double-headed, like tetrapod the anatomy, physiology, and biochemistry , in osteolepiforms (Jarvik, 1954) of Latimeria. In other words, further inves- 13. Distribution of dentition, and mode of re- tigation of Latimeria repeatedly turns up placement of fangs, similar in the two groups primitive osteichthyan or chondrichthyan- (Westoll, 1943; Jarvik, 1954, 1963) like characters. Further investigation of Re- 14. Osteolepiforms have folded teeth of the same cent lungfishes, on the other hand, repeat- type as labyrinthodonts (Schultze, 1970a) 15. Vertebrae of osteolepiforms are "protora- edly produces characters which seem to be chitomous" (Andrews and Westoll, 1970a, synapomorphous with tetrapods; e.g., lens 1970b) proteins and bile salts (L0vtrup, 1977), cil- 16. Eusthenopteron has bicipital dorsal ribs (An- iation of the larva (Whiting and Bone, 1980) drews and Westoll, 1970a) and gill-arch muscles (Wiley, 1979). It is un- 17. Scapulocoracoid of Eusthenopteron like that fortunate that, so far as we know, no pro- of early tetrapods, with a screw-shaped glen- teins of lungfishes or Latimeria have yet oid (Andrews and Westoll, 1970a) been sequenced. In any case, the number of 18. Humerus of Eusthenopteron and Strepsodus apparent synapomorphies between lungfish- comparable with tetrapod humerus (Andrews es and to Pro- and Westoll, 1970a, 1970b) tetrapods continues increase. 19. Eusthenopteron may have a sacral attach- ponents of the rhipidistian ancestry of tet- ment (Andrews and Westoll, 1970a) rapods have continued to rate such characters 20. Ichthyostega has a fishlike tail (Jarvik, 1952) as convergent, following Baur (1896). Alter- natives are to rate these characters as syn- This list may seem short to those who re- apomorphies, either of tetrapods and dip- call the extraordinarily detailed comparisons noans, or of some more extensive group. with tetrapods in Jarvik's papers (1942, 1952, Since most of these characters are in soft 178 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 anatomy and embryology, they can, when Neo-Darwinian picture of the evolutionary assumed to have occurred in rhipidistians, process. Yet the sequences consist of noth- be rated as synapomorphies of a group com- ing more than abstractions from paraphyletic prising tetrapods, lungfishes, and rhipidis- groups such as rhipidistians, osteolepiforms, tians. As a paraphyletic group, rhipidistians and labyrinthodonts. As for the pattern, very might then include members of the stem- few plausible derived characters have yet groups of tetrapods, of dipnoans, and of tet- been presented to link some rhipidistians and rapods plus dipnoans. That statement recalls some tetrapods (e.g., folded teeth). For the our comment on Dollo's (1896) diagram (p. rest, the characters used by twentieth-cen- 170), implying that remarkably little progress tury biologists seem to us little different from has been made during the last eighty years. those of nineteenth-century workers-a me- In our view, the reason for that lack of lange of primitive characters, and assertions progress is plain enough. It is acceptance of of identity or general similarity whose pre- evolution, which has directed attention away cise content is still to be established. from the unsolved problem that occupied Acceptance of cladistic methodology dur- pre-Darwinians-what are the characters of ing the last few years has begun to direct tetrapods, how are they distinguished from attention back towards pre-Darwinian prob- fishes, and from lungfishes in particular-and lems, problems of pattern. How are choa- toward a search for sequences of fossils nates and tetrapods characterized (e.g., which will fit the evolutionary doctrine, and Miles, 1977; Szarski, 1977; Schultze, 1977a; toward an interest in process rather than pat- Gaffney, 1979a, 1979b)? With the realization tern. The search for fossils has produced su- that some generally accepted groups (rhipi- perficially acceptable sequences, as it was distians, osteolepiforms, labyrinthodonts) bound to, for few transformations, however are uncharacterized, or uncharacterizable, the fantastic, are forbidden by the Darwinian or way is open for a new attack on the problem.

CHOANA, NOSTRILS, AND SNOUT (A) INITIAL COMPARISONS AND which marginal jaw teeth are present lateral, INFERENCES rather than medial, to this opening (fig. 7). Our interpretation of the dipnoan internal In vertebrate systematics identification of nostril as a choana (fig. 6) has already been a choana must satisfy not only anatomical alluded to in the Introduction. This interpre- criteria, but other criteria of as tation has resulted from the rejection of a well. To do this, the lungfish choana either comparative anatomical argument by Allis must be indistinguishable from the choana of (1919, 1932a, 1932b) concerning living lung- tetrapods, or one must be regarded as a mod- fishes from which he deduced that the true ification of the other. Primary comparisons upper lip and excurrent, external nostril had are, therefore, made between Miles's (1977) migrated together into the oral cavity. Allis's excellent specimens of Griphognathus reasoning predicts that if marginal upper jaw whitei, as interpreted partly on the basis teeth were present in lungfishes, these teeth of our own dissections of Neoceratodus for- (which are absent in extant and most fossil steri and ichthyostegid amphibians, which species) would lie medial to the internal nos- we consider to be the sister group of all other tril. The prediction, however, is inconsistent tetrapods (see Introduction). Our comments with Miles's (1977) account of several more on ichthyostegids are based on Ichthyostega or less complete skulls of the Devonian lung- sp. as described by Save-Soderbergh (1932) fish, Griphognathus whitei Miles, in which and Jarvik (1952) and on two casts of well- the internal nostril is clearly marked and in preserved palates (fig. 10). 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 179

TONGU E ANTERIOR NARIS CHOANA

.4 FIG. 6. Neoceratodus forsteri (Krefft), approximiately 120 cm. total length, AMNH 40800, showing head and bases of pectoral appendages.

In Griphognathus the preoral region of the them. Although Miles seemed to have some snout medial to a pair of nasal recesses bears reservations that these elements are either a pair of tooth ridges fused with the rostrum tooth plates or premaxillae, for he identified and covered with enamel. Miles (1977, p. them in his figures only as preoral emi- 187) refers to these tooth ridges as premax- nences, their interpretation as dentigerous illae because, when the lower jaw is in place, on the one hand and as paired elements on the mandibular symphysis closes just behind the other is supported by the presence of a 180 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

ANTERIOR PALATAL RECESS PREMAXILLARY TEETH

VOMER SUBNASAL RIDGE "EXTRA"

PALATIN E

POSTERIOR LIMIT -MAXILLARY TEETH

CHOANA-

A

C

5mm

FIG. 7. Griphognathus whitei Miles. A, ventral view of skull, from BMNH P. 56054, after Miles (1977, fig. 6); B, camera lucida drawing of left choana and surrounding dermal bones in BMNH P. 56054; arrow indicates a crack not a suture; C, camera lucida drawing of left choana in BMNH P. 50996 (holotype), an individual in which the bones surrounding the choana are fused. similar but more distinct pair of toothed ele- have fused with the bone of the snout is ments medial to the nasal recesses in Gan- based on the observations that they are orhynchus woodwardi (fig. 8). Our conclu- paired, are in series with a more posterior sion that they represent premaxillae that marginal upper jaw dentition (as are premax- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 181

PREMAXILLARY TEETH

ANTERI OR NARIS .%. 11.

MAXILLARY SUBNASAL TEETH ,-R IDGE

- U

'1-02iGa 1.I _

FIG. 8. Ganorhynchus woodwardi Traquair. Snout of a Devonian "hard-snouted" lungfish in ventral view (holotype, BMNH 44627, whitened with ammonium chloride). A paired row of premaxillary teeth on the preoral area (Miles, 1977, p. 179) and a patch of maxillary teeth on the subnasal ridge lateral to the notch leading to the anterior nostril are shown. x 1.5. illae with maxillae), are the most anterior riorly by a series of dermopalatines (fig. 7). teeth in the upper jaw, and are immediately The series of dermopalatines and the in front of an anterior palatal recess that, in extend forward around the pterygoid and osteichthyans, normally follows the concave meet in the midline. An additional paired, oral border of the premaxillae, as it does in toothed ("extra" dermal bone, Griphognathus (figs. 7, 10, 14). Lateral to fig. 7) is present between the dermopalatines each premaxilla there is a sulcus or nasal re- and the inner, hind, edge of the nasal recess cess in the consolidated ethmoid block and and this element may represent a supernu- the latter is notched along its oral margin merary dermopalatine as suggested by Miles where the sulcus extends obliquely toward (1977) or a vomer (but see p. 230). Between the palate. Posterolateral to the sulcus is a the anterior end of the pterygoids and high tooth-bearing knob (the subnasal ridge, the encircling and the ethmoid block figs. 7, 8), also enamel-coated, and this is with its ankylosed premaxillae there is a followed in Griphognathus by a long, low long, shallow depression which, in other sar- tooth-bearing ridge (the maxilla) that is an- copterygian fishes and tetrapods, has been kylosed with the bones of the cheek (fig. 7B, termed the anterior palatal recess. The po- C). We interpret the high, tooth-bearing sition of the nasal capsule between the elon- knob as the anterior part of the maxilla since gate, oval posterior opening in the palate and its enamel coating continues posteriorly onto a region just behind the inner end of the nasal the low dentigerous surface behind. recess is suggested by the position and ter- The maxilla has a medially directed flange mination of the canal for the olfactory tracts posteriorly which is sutured to the pterygoid and by the course of various canals (fig. 9) (fig. 7). This flange forms the posterior that are comparable with those for nerve and boundary of an elongate, oval opening into blood vessels in extant fishes and amphibi- what was probably the nasal capsule. The ans. The posterior oval opening in the palate anterior part of the oval opening is bounded is therefore interpreted as the fenestra exo- laterally by the maxilla, medially and ante- choanalis, and the nasal recess, present in 182 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

PROFUNDUS CANAL-

VASCULAR CANAL-

FIG. 9. Plans of snout of two Devonian dipnoans to show major canals as interpreted by Miles. A, Griphognathus whitei Miles (after Miles, 1977, fig. 63); B, australis Miles (after Miles, 1977, fig. 66). some form in all fossil and extant lungfishes, the pterygoids anteriorly, and (5) the junc- as housing the anterior, external, incurrent tion of the broad pterygoids in the midline nostril. When the lower jaw is in place in under the anterior part of the parasphenoid. Griphognathus the mandibular symphysis is Each of these features may be seen also in seated in the anterior palatal recess just be- Ichthyostega (cf. figs. 10 and 46B). hind the premaxillae, as in all osteichthyans, and the anterior tip of the nasal recess is just visible beyond the lower jaw margin as in (B) NASAL CAPSULE living lungfishes. There are basically four major openings in The essential features of Griphognathus the wall of the endoskeletal nasal capsule of for comparison with tetrapods are (1) the gnathostomes that transmit water or air cur- ventrally opening anterior nostril and choana; rents, nerves, and blood vessels. One is for (2) the oval outline of the fenestra exochoa- the olfactory tract, and this is situated pos- nalis (choana); (3) the interruption of the teromedially. Another, anterolateral opening dentigerous border of the upper jaw (be- is for the nostrils extending to the outside. tween the premaxilla and maxilla) at the site This opening may be subdivided by soft tis- of the anterior nostril; (4) the formation of sue, dividing it into an incurrent and excur- the fenestra exochoanalis by the maxilla and rent port. These two ports may each form the dermal bones of the palate that encircle separate tubes that open remotely from each 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 183

E X T E R N A L NARI S

CHOANA

ANTERIOR PALATAL RECESS

FIG. 10. Casts of ventral surface of snout in specimens of Ichthyostega sp., to show choana. A, AMNH P. A756a; B, BMNH R. 9469. 184 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

angular opening (palatal fenestra, fig. 12) in the bony palate just below a large fenestra in the ventrolateral wall of the nasal capsule which can be seen in cleared and stained F E N ES T R A specimens, but not in fresh specimens unless E X ON ARI N A the covering soft tissues of the oral mucosa ANTERIOR are dissected away. The coincidence of the palatal fenestra in the nasal capsule and der- mal palate involving premaxilla, maxilla, and dermopalatine is similar to the "choana" of Eusthenopteron as described by Jarvik FENESTRA EXONAR I NA (1942). P O S T E R I O R FIG. 11. Snout of Polypterus senegalus Cu- (C) CHOANA AND NOSTRIL IN vier, after Jarvik (1942, fig. 84), showing bony lim- its of fenestra exonarina communis. DIPNOANS The internal nostril of living lungfishes was other, but at their bases they come together regarded as a choana by all nineteenth-cen- into the anterolateral wall of the nasal cap- tury anatomists, and by virtually all in the sule. The latter condition is found in cladis- first half of the twentieth (references in Jar- tians [Polypterus (fig. 11) and Erpetoichthys]. vik, 1942, p. 273). Allis (1932a, 1932b) argued A third opening, in the posterior wall of the that this nostril was homologous with the ex- capsule, transmits the maxillary branch of current nostril of chondrichthyans and actin- the fifth nerve and the palatine branch of the opterygians, but not with the choana, be- seventh. The fourth main opening,4 in the cause the choana lies medial to the secondary ventral or ventrolateral part of the capsule, upper lip, whereas the dipnoan excurrent is for the ramus buccalis lateralis and, often, nostril lies external to that lip (Allis, 1932a, one or more blood vessels. In some cases, p. 666). Allis here interpreted the anterior for example, Polypterus, the third and fourth toothplates of lungfishes, usually called vo- openings are represented by only a single fe- mers, as premaxillary, and so identified the nestra. In other cases there are numerous secondary upper lip, which always lies im- smaller perforations in the capsule for nerves mediately external to the premaxillary and and blood vessels. maxillary teeth. Allis had earlier (1919) ar- Of the various principal openings in the gued that dipnoans have no secondary upper capsule, the last two, for branches of the fifth lip, but a tertiary upper lip, unique to them, and seventh nerves, are most interesting lying external to both nostrils. here because of their relations to surround- Jarvik (1942) developed Allis's argument ing bony structures. In cladistians these further, bringing in interpretations of the nerves enter the capsule through a single, subnasal cartilage of lungfishes, the course large, posteroventral fenestra, but the ramus of the maxillary nerve, and the interruption buccalis and at least one of the main blood of the infraorbital sensory canal opposite the vessels pass together between the capsule nostrils. Jarvik concluded, like Allis, that and the anteroventral wall of the orbit dorsal dipnoan nostrils are ordinary fish nostrils to the bony palate. Where they turn dorsally which have migrated back so that one opens to enter the posteroventral capsule wall there into the roof of the mouth, and that dipnoans is a triple junction of the premaxilla, maxilla, are not choanates. The question has since and dermopalatine (fig. 12). The premaxilla been reviewed by Thomson (1965), Bertmar is notched posteriorly just below the fenestra (1966), Panchen (1967), and Miles (1977, p. in the nasal capsule and the maxilla and der- 147), each of whom disagreed with Jarvik mopalatine fail to cover completely the pre- over the interpretation of one structure or maxillary notch. The result is an oval or tri- another, while agreeing with his conclusion. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 185

NFRAORBITALS

FIG. 12. Polypterus ornatipinnis Boulenger, ventral view of roof of mouth ofjuvenile (4 cm. standard length), still with large external gills, to show fenestra beneath nasal capsule, between premaxilla, maxilla, and dermopalatine. Note that maxilla and infraorbital bones are still recognizably discrete. Eye and nasal capsule are indicated by broken lines. Cartilage stippled.

Thus Panchen (1967, p. 379) wrote "the evi- is interrupted anteriorly by a notch (fig. 7A) dence for the homology of the dipnoan pos- which is identifiable as the anterior margin terior nares with those of actinopterygian of the anterior nostril, by comparison with fishes and their non-homology with tetrapod Recent lungfishes. The same notch is evident choanae seems conclusive." Rather than re- in other fossil "hard-snouted" dipnoans view these interpretations yet again, we refer (e.g., White, 1962, pl. 2-, Rhino- the reader to pages 376-379 in Panchen's pa- dipterus, Ganorhynchus; Miles, 1977, fig. per, and suggest that the conclusion quoted 4-Holodipterus, Chirodipterus; fig. 8). As comes as a surprise if the whole passage is argued above (pp. 180-181), we identify the read with an open mind. teeth anteromedial to this notch (e.g., in Gri- Miles's (1977) figures of the Devonian phognathus and Ganorhynchus, figs. 7, 8) as lungfish Griphognathus show an outer dental premaxillary, as did Miles (1977, p. 187), arcade (premaxillary-maxillary dentition in since they bite outside the lower jaw, are the other osteichthyans), identifiable by its po- most anterior teeth in the upper jaw, and lie sition outside the lower jaw when the mouth immediately in front of an anterior palatal is closed. This tooth row is external to the recess similar to that which normally lies be- posterior nostril, which is completely en- hind the premaxillae in osteichthyans (figs. closed by dermal bone in some individuals 7, 10, 14). Posterolateral to the anterior nos- (Miles, 1977, fig. 80; fig. 7C). The tooth row tril, Griphognathus has a tooth row which 186 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 runs back on a series of bones which Miles viso-"it is very unlikely that there was ever (1977, fig. 57) interprets as lateral nasal in the ontogeny or phylogeny of the Dipnoi toothplates and ectopterygoid. These teeth an actual physical ventral migration of the bite outside the lower jaw (Miles, 1977, p. nares. Rather there was a failure of the dor- 181), and we therefore interpret them as sal migration of the nasal placode normal in maxillary, since in no known gnathostome the ontogeny of other bony fish." If the dip- do palatal teeth bite outside the lower jaw. noan posterior nostril is a choana, this ex- Maxillary teeth occur in some other Devo- planation should also apply to tetrapods. nian lungfishes (Holodipterus, Dipterus; Panchen (1967, p. 380) reviews this hy- Miles, 1977, p. 186), but in most hard-snout- pothesis: that the tetrapod choana is the ed dipnoans they are represented only by the homologue of a fish nostril. He specifies one subnasal ridge (Miles, 1977, p. 185), a den- prediction from it, that primitive tetrapods ticulated or enameled knob lateral to the an- would have "a single external naris on each terior nostril (fig. 8). side, situated at or very near the jaw margin The posterior nostril is, as noted above, and almost confluent with a laterally placed completely bone-enclosed in some speci- choana inside that margin. Thus the condi- mens of Griphognathus (fig. 7). In these tion in the amphibian Ichthyostegalia (Siive- specimens, the opening is a rostro-caudally Soderbergh, 1932; Jarvik, 1952) would be re- elongate ellipse, with smoothly inturned, garded as primitive." Later (1967, p. 407) toothed margins (fig. 7C). The opening is Panchen concluded that this condition was bordered laterally by Miles's ectopterygoid, indeed primitive for tetrapods, a view he still our maxilla, and medially by Miles's der- holds (Panchen, personal commun.). Pan- mopalatine 2, our palatine. Miles noted (p. chen also pointed out the choana/fish nostril 175) that the opening has the position of a homology would be refuted if any fish had choana, but rejected that homology in favor two external nostrils in addition to a choana, of "chance similarity." This is because he and noted that this was the basis for Wes- (p. 147) interpreted dipnoan nostrils to bring toll's (1943) doubt that porolepiforms have "dipnoans into line with other fishes." But two external nostrils, since porolepiforms since his overall conclusion (p. 313) was that are the only fishes described as having two dipnoans are the sister group of choanates, nostrils and a choana. Panchen went on to it is with them, not other fishes, that we summarize Jarvik's more recent interpreta- should expect dipnoans to fall in line. tions of porolepiforms as removing Westoll's The outer dental arcade ofGriphognathus, doubts, and refuting the choana/naris ho- which lies external to the posterior nostril mology. But even if two external nostrils and is notched or interrupted by the anterior were demonstrated beyond any possible nostril, meets Allis's (1919, 1932a) criteria doubt in porolepiforms, we do not agree that for choanates, and removes his reason for the choana/naris hypothesis would be refut- denying the homology of the lungfish poste- ed, for we would question the existence of rior nostril and the choana. This interpreta- a choana in porolepiforms. To explain why, tion carries the implication that the two nos- it is necessary to review the choanae of rhip- trils of choanates (lungfishes and tetrapods) idistians. Before doing so, we emphasize that are homologous with those of other gnatho- the varied interpretations of lungfishes show stomes, so that the choana is the homologue that even in living animals, with full onto- of a fish nostril. We note Bertmar's (1966, p. genetic information, identification of choa- 140) conclusion on the origin of the dipnoan nae can be a matter for debate. In fossils, condition-"The posterior nostril migrated where soft anatomy can only be interpreted ventrally simultaneously with a reduction of after some Recent model, and where vaga- the maxillary, the premaxillary and the an- ries of preservation lead to disputed inter- terior part of the infraorbital sensory pretation even of hard parts, there is endless canal"-and Panchen's (1967, p. 379) pro- room for debate. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 187

(D) CHOANA AND NOSTRIL IN specimens in the Hancock Museum. On this RHIPIDISTIANS evidence, he wrote (p. 198) "The internal nostril found in all Osteolepids is identical in Dollo (1896, p. 109) suggested that if os- position, size and borders with that of certain teolepid crossopterygians were ancestral to Embolomeri." lungfishes and tetrapods, they would have a Stensio (1932, fig. 30) figured two undeter- choana. Such an aperture was first reported mined crossopterygian snouts from the De- by Watson (1912, p. 9) in Megalichthys, as vonian of Spitsbergen, and described them an opening whose front margin is formed by as showing two external nostrils and an in- the vomer, and then by Watson and Day ternal nostril, the latter confluent with the (1916, p. 11, pl. 1, fig. 5) in Glyptopomus, posterior external nostril. Romer (1933, fig. where it was said to lie entirely within the 53B) illustrated the choana in Eusthenopte- palatine. Bryant (1919, p. 18), perhaps influ- ron, on data from Bryant, Watson, and Sten- enced by Watson and Day, described a sim- si6. Siive-S6derbergh (1933) mentioned in- ilar internal nostril, enclosed within the pal- ternal nostrils in Dictyonosteus and atine, in Eusthenopteron. Stensio (1918, p. , referring to Stensio (1922) and 120) suggested that a choana might be pres- Watson (1926), and named a group Choana- ent in Dictyonosteus, a Devonian fish since ta, comprising Dipnoi, Crossopterygii, and interpreted as a by Jarvik (1942, tetrapods. Nielsen (1936, p. 32, figs. 14, 15) p. 579, pl. 17), with the opening in question restored an internal nostril in an undeter- labeled as the fenestra endonarina commu- mined coelacanth from the of East nis. . Jarvik (1942, figs. 80, 81) repro- In 1921, Stensio referred (p. 136) to choa- duced Nielsen's figures and reinterpreted the nae in the rhipidistians Dictyonosteus, Os- "choana" as the canal for the buccal nerve. teolepis, Diplopterus (=Gyroptychius), Rhi- Holmgren and Stensio (1936) reproduced zodopsis and Megalichthys, but the last four Watson's (1926) figure of Eusthenopteron are clearly due to misreadings of Dollo (1896) and Stensio's (1932) figures of the Spitsber- and Watson and Day (1916), who mentioned gen crossopterygians, now determined as these names in discussing nostrils, but re- Porolepis, both showing an internal nostril. ferred only to their external nostrils. In 1922 In a footnote (p. 348) they referred to Watson Stensio (p. 195) suggested that the coel- and Day's identification of a choana in the acanth Diplocercides had a choana, and this middle of the palatine of Glyptopomus, and suggestion seems to be the only source of said that Save-Soderbergh has found that the Watson's (1926, p. 200) statement "The orig- opening "ihre normale Lage hat." inal conclusion that the internal nostrils of Westoll (1937, p. 28) wrote of crossopte- Osteolepids imply an air-breathing habit is rygians "there is typically a single external confirmed by thefact that whilst the Old Red nostril and an internal nostril on each side: Sandstone Coelacanth Diplocercides pos- in early forms (e.g., Thursius, Diplopterax sessed similar internal nostrils, the marine [=Gyroptychius]) the external and internal Triassic and Coelacanths pos- nares may be confluent, the former being sess only an external nostril" [italics ours; merely notches in the dermal bones of the air-breathing is, of course, not implied by the lip." Also in 1937, Jarvik described the nasal possession of internal nostrils-see Broman, capsule of two species of Eusthenopteron 1939, on lungfishes, and Szarski, 1977, on from the Baltic, and showed a large ventral teleosts]. In the same paper, Watson (1926, opening identified as the fenestra choanalis. fig. 4) restored a choana in Eusthenopteron, Finally, in 1937 Romer described the nasal on the basis (p. 249, figs. 34, 35) of two dis- capsule of Ectosteorachis with a ventral articulated specimens in the Royal Scottish opening "to the choana." He named a group Museum, and described (p. 246) a choana in Choanichthyes to include lungfishes and Megalichthys, on the basis of disarticulated crossopterygians. 188 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

So far as we know, this review covers all (series 1) lacked the palate, while the other the primary literature on crossopterygian was complete only on one side, and there the choanae prior to Jarvik's 1942 monograph. palate was thrust up into the nasal capsule Before commenting on that, it is worth (Jarvik, 1942, pls. 11, 12). Jarvik's restora- summing up this early work. By 1940, tion of the fenestra exochoanalis was based the literature contained references to inter- mainly on this second series, but that open- nal nostrils in 12 genera of Paleozoic ing is labeled in one other specimen (pl. 8, crossopterygians: Megalichthys, Glyptopo- fig. 1). mus, Dictyonosteus, Osteolepis, Gyropty- We have prepared four BMNH specimens chius, Rhizodopsis, Diplocercides, Eusthen- of Eusthenopteron foordi to show the nasal opteron, Porolepis, Nielsen's undetermined capsule and surrounding structures. The coelacanth, Thursius and Ectosteorachis. most informative of these is P. 60310 (figs. Three of these (Dictyonosteus, Diplocer- 13, 14), an acid-prepared specimen consist- cides and Nielsen's fish) are coelacanths, ing of a snout, both palates, and the maxilla and the openings concerned were reinter- and cheek bones of one side. Our specimens preted after the discovery of Latimeria. In confirm the accuracy of Jarvik's account of Osteolepis and Rhizodopsis no internal nos- these bones, and we will comment only on tril had been described-the references were the choana. mistaken. In Glyptopomus the only illustrat- Our restoration of the fenestra exochoa- ed specimen was wrongly interpreted. In nalis and surrounding bones (fig. 14) differs Gyroptychius and Thursius Jarvik later in several ways from Jarvik's (1942, fig. 56- (1948, p. 36) maintained that Westoll's ob- since much reproduced and redrawn, e.g., servations were based on broken specimens. Jarvik, 1954, fig. 25; 1966, fig. 17B; 1972, fig. In Porolepis, Ectosteorachis, and Jarvik's 73B; Romer, 1966, fig. 101; Panchen, 1967, Baltic Eusthenopteron all that had been de- fig. 3; Thomson, 1968, fig. 2B; Vorobyeva, scribed was a ventral opening of the endo- 1977a, fig. 3C). In his restoration the fenestra skeletal , not an opening in the exochoanalis is a drop-shaped opening roof of the mouth. But all these names surely bounded anteromedially and anterolaterally contributed to a climate of opinion in which by the notch between the vomer and pre- Watson could write "the internal nostrils maxilla, posterolaterally by the maxilla, and found in all osteolepids," Save-S6derbergh posteromedially by the dermopalatine. The and Romer could name the Choanata and opening has a maximum width (at the der- Choanichthyes, and so on. The only unchal- mopalatine/vomer junction) of about 12 lenged information that had been presented (1942) to 15 (1972) percent of the total width was Watson's restoration of Eusthenopteron of the snout at that point, and a maximum and his account of Megalichthys. And in the length about equal to the greatest width of museums of the world no single specimen the vomer (1942) or somewhat more than had been reported that might be shown to that (1966, 1972). In our restoration, the the skeptical enquirer anxious to judge the opening has the same boundaries as in Jar- size and position of this opening for himself. vik's, but its medial border is irregular, and Jarvik (1942) presented uniquely detailed it is much smaller, almost equal in length to accounts of snout structure in Porolepis and Jarvik's, but with a maximum width of only Eusthenopteron. In Porolepis no palate was about 5 percent of the width of the snout at described, so that the opening in the palate, the dermopalatine/vomer junction. This dif- the fenestra exochoanalis (p. 372) could not ference is due almost entirely to the form of be described. In Eusthenopteron he gave an the head of the dermopalatine in the two re- extraordinarily thorough and well-illustrated constructions. Jarvik (1942, p. 453, figs. 54A, description of the fenestrae endo- and exo- 56) described the head of the dermopalatine choanalis, and the structures surrounding as articulating with the vomer by one sur- them. This account was based primarily on face, a roughened depression facing antero- two serially-ground specimens, one of which ventrally, and overlapping the posterolateral 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 189

FIG. 13. Eusthenopteron foordi Whiteaves. Ventral view of an acid-prepared specimen (BMNH P. 60310, whitened with ammonium chloride) comprising snout, incomplete palates, and left cheek. Com- pare figure 14. corner of the vomer. He also argued (1942, our specimens (P. 60310, acid-prepared, P. p. 457), and has since insisted (1954; also 6807, mechanically prepared) the head of the Bjerring, 1967, 1973), that the joint between dermopalatine ends in a thumblike process the palate and the snout was immobile. In which bears three articular surfaces. The 190 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

ANTERIOR PALATAL RECESS-

PROC ESSUS IN TERMED IUS / X<

-N ASAL CAPSU LE

ECTOPTERYGOID FIG. 14. Eusthenopteron foordi Whiteaves. Snout with left palate and maxilla in articulation in ventral view, based mainly on camera lucida drawings of BMNH P. 60310 (fig. 13). Body of parasphenoid and ventral margin of postnasal wall (damaged in P. 60310) drawn from other BMNH specimens. Otherwise, the only restoration in figure is by matching two sides of specimen, restoring damaged teeth, and in articulation of palate and maxilla with snout. Inset, drawn on a slightly larger scale, is an earlier attempt at restoring maxilla and palate in articulation with snout, made before left palate and maxilla of specimen were completely free from matrix and each other. Head of dermopalatine is more closely articulated with vomer and dermopalatine is turned medially more than in the main figure. Numbers 1, 2, and 3 identify three articular facets on the head of the dermopalatine (see text); foramen in postnasal wall is inferred to have been for nerves and vessels (fenestra endonarina posterior of Jarvik); processus intermedius is endoskeletal lining of processus dermintermedius; pit in dermopalatine (labeled on inset) is inferred to have accommodated fang of first coronoid. For other structures, see Jarvik (1942). most dorsal of these (1, fig. 14) corresponds al corner of the vomer. The second articular with the one described by Jarvik. It faces surface (2, fig. 14) lies ventrolateral to the anteroventrally and articulates with a flat first, faces anteromedially, and articulates surface on the dorsal side of the posterolater- with a facet on the lateral face of the vomer, 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 191 separated from the dorsal surface by a crest. the palatoquadrate/cheek unit was movably The third surface (3, fig. 14) is the most articulated with the snout. ventral, faces anterolaterally, and articulates Thus our interpretation of the fenestra with a facet on the posteromedial face of the exochoanalis differs from Jarvik's in two vomer, above the hind end of the vomerine particulars: our opening is smaller, and is tra- tooth row. In Jarvik's figures, the external versed by a mobile joint-plane. Neverthe- and internal tooth rows on the dermopalatine less, the opening does communicate with the extend forward to the junction with the vo- nasal capsule in the fossils, and could still be mer, and are continuous with the corre- a choana (though we are aware of no choa- sponding tooth rows on the vomer so that nate with a joint through the choana). But if the fenestra exochoanalis has a toothed me- we attempt to restore a choanal tube leading dial margin, as in Ichthyostega (fig. 10; Jar- from the fenestra exochoanalis to the nasal vik, 1952). In our specimens, the external capsule, the passage is severely restricted by tooth row of the dermopalatine is absent on the processus intermedius (endoskeletal; fig. the anterior part of the bone, and the larger, 14) and the processus dermintermedius of internal teeth decrease in size and fade out the lateral rostral, which occlude the anterior on the upward-sloping part of the bone be- and widest part of the opening (cf. Jarvik, hind the fenestra exochoanalis, and the 1942, figs. 47, 48, 52, 53, 56, 58). This means thumblike articular process of the dermopal- that a choanal tube would have had to turn atine is toothless. medially to enter the nasal capsule beneath Because of these differences in the head the posterior part of the processus interme- of the dermopalatine, the fenestra exochoa- dius. But here we meet another problem. In nalis is much smaller in our reconstruction the lateral ethmoid (postnasal wall, Jarvik) than in Jarvik's. When the dermopalatine is of Eusthenopteron there is a large foramen placed in articulation with the vomer, it (fig. 14). In 1937 Jarvik interpreted this as rocks in the dorsoventral plane and in the having "most probably transmitted the trun- horizontal plane, pivoting on the articulation cus infraorbitalis (the ramus maxillaris tri- between the processus apicalis of the pala- gemini and the ramus buccalis of the nervus toquadrate and the lateral ethmoid (fig. 14; facialis) and its accompanying vessels" (Jar- Jarvik, 1942, figs. 54, 55). The lateral surface vik, 1937, p. 98, fig. 13). This is a reasonable of the dermopalatine and ectopterygoid was interpretation, since it brings Eusthenopte- closely and apparently immovably bound to ron into line with other primitive fishes, in the maxilla and lacrimal (cf. Jarvik, 1942, pl. which there is a large canal in the postnasal 13, figs. 3-5). If the rostral end of the palate wall (Porolepis, Jarvik, 1942, fig. 36; Glyp- was mobile, the rostral end of the maxilla tolepis, BMNH P. 47838), presumably trans- and lacrimal must also have been movably mitting the orbitonasal sinus, in addition to articulated with the snout. That this was so nerves. But in 1942 Jarvik reinterpreted the is indicated by the fact that the two joints opening in Eusthenopteron as the fenestra (dermopalatine/vomer and lacrimal + max- endonarina posterior, or nasolacrimal duct. illa/premaxilla + lateral rostral) are in the This interpretation raises certain difficulties same transverse plane (figs. 13, 14), and by (Thomson, 1964; Schmalhausen, 1968), prin- the form of the joint surfaces on the medial cipally, in our view, because it leaves no face of the anterior part of the lacrimal and room for the substantial venous drainage of on the lateral face of the lateral rostral and the snout which one would expect (cf. Jar- supraorbito-tectal (Jarvik, 1944a, fig. 7B, od. vik, 1942, p. 481; Thomson, 1964, p. 330). La + Mx). These two surfaces are a poor Furthermore, the nasolacrimal duct does not match, and that on the lateral rostral and su- pass through the endoskeleton in primitive praorbito-tectal shows blurred margins, tetrapods: in living amphibians, the duct quite different from the sharply defined over- passes through the fenestra endonarina com- lap areas of immobile dermal bones such as munis except in the plethodontids Pletho- the supraorbital. We therefore believe that don, Hydromantes, and Batrachoseps, and 192 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 in postmetamorphic adspersus tive tetrapods-temnospondyls, lepospon- among anurans (Jurgens, 1971; Presley, per- dyls, anthracosaurs, hynobiids). In these, sonal commun.); in primitive fossil tetrapods the two nostrils are separated only by a nar- the duct penetrates the lacrimal from end to row bar of tissue (in ichthyostegids, loxom- end, like a sensory canal (Watson, 1940, fig. matids, and anthracosaurs by the inturned 22; Panchen, 1977, fig. 8; Heaton, 1979, fig. anterior end of the maxilla; fig. 10, cf. Pan- 15). In Eusthenopteron, Jarvik's (1942, fig. chen, 1967, p. 407), and the infraorbital sen- 57) reconstruction of the cavities in the snout sory canal is interrupted. Instead, if Eus- shows the canal in the postnasal wall (labeled thenopteron is a choanate, it shows a derived f. enp.) to be directly in line with a canal of condition of the snout, with the external nos- roughly equal size, the nasobasal canal (c.n.- tril well above the edge of the mouth, and an b.) running into the rostrum from the anterior uninterrupted infraorbital canal beneath it, wall of the nasal capsule, and breaking up as in advanced amphibians (Schmalhausen, into three branches. It is the same in our 1968; see p. 226 on the infraorbital canal of acid-prepared specimen: looking forward Ichthyostega). through the opening in the postnasal wall, If Eusthenopteron had no choana, then one looks directly into the posterior opening both external nostrils must have opened of the nasobasal canal. If we restore the trun- through the single fenestra exonarina, as cus infraorbitalis and vessels running through Panchen (1967, p. 402) suggested. We note the ventrolateral part of the nasal capsule, that this is also the case in Polypterus (fig. from the opening in the postnasal wall to the 11), Laccognathus (Vorobyeva, 1980), and, nasobasal canal, they occlude the only space according to Jarvik (1966) in Holoptychius available for a choanal tube (cf. Jarvik, 1942, (fig. 17D). fig. 57). Since Jarvik's 1942 account of the fenestra We do not suppose that these remarks will exochoanalis in Eusthenopteron, a similar convince the reader that Eusthenopteron fenestra has been illustrated (see fig. 15) in had no choana, and that is not our intention. the osteolepiforms Glyptopomus kinnairdi We wish only to show that interpretation of (Jarvik, 1950, pl. 5, fig. 2; 1966, fig. 16C, D), the opening in the palate as a choana raises Eusthenodon waengsjoei (Jarvik, 1952, pl. certain difficulties, and is at least debatable. 16, fig. 2, text-fig. 29), rhom- And there are other, non-choanate fishes bolepis (Vorobyeva, 1975, fig. 2), Thursius with a comparable opening in the palate. In estonicus (Vorobyeva, 1977a, fig. 3) and Gy- Polypterus the opening (fig. 12, palatal fe- roptychius pauli (Vorobyeva, 1977a, figs. 3, nestra) lies beneath the anterior part of the 29). Among these five genera Glyptopomus, nasal capsule, between the premaxilla, max- Panderichthys, and Eusthenodon have fangs illa, and dermopalatine, and is large in young at the tip of the dentary and differ from Eus- fishes. In paleoniscoids such as Mimia (Gar- thenopteron, Latvius, and Thursius which diner and Bartram, 1977, fig. 2) there is a lack them (Jarvik, 1966, p. 57). The dentary notch between the premaxilla and vomer, fangs of Glyptopomus, Panderichthys and similar to the notch between those bones in Eusthenodon fitted into the anterior palatal osteolepiforms (fig. 15), and when the der- recess when the mouth was closed (Jarvik, mopalatine and maxilla are in place, there is 1966, fig. 16C), and this suggested the pos- a drop-shaped opening between the four sibility that the other opening in the palate, bones, as in Eusthenopteron. This opening the fenestra exochoanalis, might have been lies behind the nasal capsule, and does not occupied by the fangs of the anterior coro- communicate with the latter as it does in noid. We have examined the specimen of Eusthenopteron. But our main reasoning for Glyptopomus illustrated by Jarvik (1966, fig. questioning the existence of a choana in Eus- 16C), and find that the tip of the first coro- thenopteron is that the nose does not match noid fang is preserved beneath the dermo- the condition in other primitive choanates palatine in about the same position as the (dipnoans, ichthyostegids, and other primi- depression in the dermopalatine of Eusthe- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 193

- FP F P e

FIG. 15. Reconstructions of snouts of rhipidistians in ventral view to show variations in fenestra in palate (fp). A, Porolepis brevis Jarvik; B, Thursius estonicus Vorobyeva; C, Eusthenopteron foordi Whiteaves; D, Megalichthys hibberti Agassiz; E, Gyroptychius pauli Vorobyeva; F, Panderichthys stolbovi Vorobyeva. After Vorobyeva (1977a, fig. 3). nopteron which we assume housed a coro- mopalatine. The coronoid fangs must have noid fang (pit, fig. 14). However, the speci- fitted into the "choana" (fig. 16B). men is dorsoventrally flattened, and the jaws In porolepiforms, Jarvik had no informa- displaced laterally. Janvier (personal com- tion on the fenestra exochoanalis in 1942 (p. mun.) informs us that in an undetermined 372). In 1966, he restored the fenestra in Po- osteolepiform which he obtained in the Up- rolepis brevis (his fig. 2A), adding informa- per Devonian of Turkey the opening in the tion from Glyptolepis. Since the porolepi- floor of the nasal capsule is in the matching form dermopalatine is well known only in position to some fangs in the lower jaw. And Glyptolepis groenlandica (Jarvik, 1972, p. Jessen (1966, fig. SB, C) illustrated an un- 189) we assume that the restoration of Po- determined osteolepid lower jaw in which rolepis includes the dermopalatine of Glyp- the fangs of the first coronoid are placed lat- tolepis, so that the shape and size of the fe- erally, just inside the dentary tooth row, and nestra exochoanalis are conjectural. posterolateral to the pit for the vomerine In Glyptolepis, the fenestra exochoanalis fang. If Jessen's picture is accurate, the cor- (Jarvik, 1972, figs. 30, 31, pl. 22, fig. 1) is onoid fangs would surely fit into the fenestra restored as a triangular opening, with a exochoanalis (fig. 16A). In Panderichthys, smooth outer border formed by the premax- where the opening lateral to the dermopala- illa and maxilla, and an angled inner margin tine appears very large (Vorobyeva, 1973, pl. formed by the junction of the vomer and der- 36, fig. 4), the lateral articulation between the mopalatine. The opening is very small, with palatoquadrate and ethmoid is absent (Vo- a maximum width of about 3 percent of the robyeva, 1980) and there is no pit in the der- breadth of the snout at that level. This res- 194 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

FENESTRA EXOCHOANALIS A~~~~- PIT FOR VOMERINE FANGS t ~~~~~~~~~A GS5O F FIRS T1

PIT FOR VOMERINE FANGS PIT FOR DERMOPALATINE FANGS CORONOID FANG FIG. 16. Possible occlusal relationships between coronoid fangs and fenestra exochoanalis in osteo- lepiforms. A, reconstruction of anterior part of lower jaw, in dorsal view, of an undetermined osteo- lepidid from the of Bergisch Gladbach, West Germany, after Jessen (1966, fig. 5B). Note positional relationships of coronoid fangs and pit for vomerine fangs, compare figure 14; B, Pander- ichthys rhombolepis (Gross). Reconstruction of left half of snout in ventral view (after Vorobyeva, 1975, fig. 2) and of anterior part of right lower jaw in medial view (after Vorobyeva, 1962, fig. 30, reversed). Arrows mark approximate boundaries of pits or openings expected in opposing jaw to house fangs. Endoskeleton stippled in A and B; C, Eusthenopteron foordi Whiteaves, outline of head (after Jarvik, 1944a, fig. 2), with upper and lower fangs (represented by arrows) added (after Jarvik, 1944a, figs. 1iB, 13). Bracket covering tips of fangs of first coronoid is position of choana in Jarvik's recon- struction (1944a, fig. 13). toration is based primarily on Jarvik's seri- which a choana has been described is Pow- ally ground specimen. In the published sec- ichthys (Jessen, 1975). The palate of this fish tions, the right fenestra exochoanalis is not is unknown, and all that has been described present in no. 50 (1972, fig. 8B), is shown in is an opening in the floor of the nasal capsule. nos. 58 and 62 (1972, figs. 66C, 8C), and has In summary, we regard statements like disappeared by no. 67 (1972, fig. 32A). The "the choanate condition of the Rhipidistia left fenestra exochoanalis is not present in has been regarded as firmly established" no. 58, is shown in nos. 61, 62 and 67 (1972, (Panchen, 1967, p. 381) as wholly unwar- figs. 76B, 8C; 1963, fig. 13A) and has disap- ranted. No satisfactory evidence of a choana peared by no. 72 (1966, fig. IA). This gives in porolepiforms has been produced. In os- the opening a maximum rostrocaudal extent teolepiforms, the only well-known example of less than 1.5 mm. (section interval 100 ,.) is Eusthenopteron, and we find the fenestra in a fish of head length about 75 mm. We exochoanalis to be much smaller than in pre- note that the dermal bones are displaced on vious restorations. Possible interpretations both sides of the sectioned head, and suggest of that opening are that it was part of a mo- that one would expect some sort of notch in bile joint between the cheek/palate unit and the margin of the palate at the junction of the snout; and/or that it transmitted nerves four dermal bones (cf. Jarvik, 1972, pl. 22, and vessels branching from the adjacent fig. 1), and that there are no grounds for re- truncus infraorbitalis and its accompanying garding such a notch as the choana. vessels; and/or that it accommodated a cor- The only other supposed porolepiform in onoid fang (fig. 16). 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 195

FEC (E) CHOANA AND NOSTRIL IN TETRAPODS Our theory is that the choana is the homo- logue of a fish nostril. Dismissing that the- "s ory, Panchen (1967, p. 380) wrote "Argu- B ments against this theory follow closely those used in the rejection of the choanate condition of the Dipnoi and are reinforced by description of the snout in Rhipidistia and the embryology of tetrapods." Arguments concerning the choanate condition of the Dipnoi and the snout of Rhipidistia are dis- cussed in the two preceding sections. In his comment on "the embryology of tet- rapods" Panchen referred to his own sum- C D F EC mary of the facts (1967, pp. 389-391). He v F E P noted that the and apodan choana FIG. 17. External nostrils in rhipidistians. A, forms at the oral end of a nasobuccal groove Glyptopomus kinnairdi (Huxley) and B, Megalich- similar to that formed in the ontogeny of Dip- thys hibberti Agassiz, are osteolepiforms showing noi and Chondrichthyes. Urodeles and an- our interpretation of a fenestra exonarina com- urans differ in that their choanal tube devel- munis (fec, cf. fig. 11); C, Porolepis brevis Jarvik; ops in two portions, an anterior choanal and D, Holoptychius sp., are porolepiforms, the process, reasonably homologized with the first with separate anterior (fea) and posterior nasobuccal groove, and a gut process that (fep) fenestrae exonarinae, the second with a grows forward to meet the choanal process. fenestra exonarina communis, as in A and B Panchen accepts that this condition is sec- (A, B after Jarvik, 1966, fig. 14B, C; C, D after ondary, and cites Medvedeva's and Bert- Jarvik, 1972, fig. 7A, C). mar's work on the possible relationship be- tween this mode of development and the tive for osteichthyans, the dorsal migration influence of the infraorbital sensory canal. of the nasal placode during embryogenesis He concludes that the anuran and urodele displaces both nostrils external to the second- mode of development is "probably . .. due ary upper lip, which includes the dermal ... to a precocious development of the [in- bones of the maxillary-premaxillary arch and fraorbital sensory] canal in the larvae of the its dental arcade. This migration from the ancestral forms" (1967, p. 391). primitive ventral position of the placode ex- The theory of the tetrapod choana that we plains the position of the nostrils on the side advocate is that it is the homologue of the of the snout in front of the eye in cladistians, dipnoan choana, and hence of the excurrent actinopterygians, rhipidistians and actinis- nostril of chondrichthyans. The chondrich- tians.2 The choanate condition, (dipnoans, thyan excurrent (posterior) nostril lies exter- nal to the primary upper lip which, in turn, 2 We note Panchen's (1967, p. 380) summary of the borders the external edge of the dental ar- controversy over whether the teleostean nasal placode cade of the palatoquadrate. When a second- migrates up the tip of the snout, so reversing its rostro- ary upper lip extends forward from the angle caudal direction, the anterior nostril becoming the pos- terior and vice versa, or migrates up the side of the of the mouth to the nasal region, this fold of snout, so that the orientation of the nostrils is un- tissue will abut against the nasobuccal changed. Since, as Panchen says, it is the placode not groove that connects the excurrent (poste- the nostrils that migrates, and the nostrils differentiate rior) and incurrent (anterior) nostrils in chon- subsequently, we consider it immaterial whether the drichthyans. Given such a development of choana is homologized with the excurrent or incurrent primary and secondary upper lips as primi- nostril of other osteichthyans. 196 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 tetrapods) arose by a modification of this distance between the choana and the external primitive osteichthyan pattern, in which the nostril increased, and in those forms with placode failed to migrate, and the excurrent aquatic larvae, restoration of continuity of nostril came to open in the roof of the mouth, the infraorbital sensory canal entailed a fur- between the primary and secondary upper ther modification of choanal ontogeny, the lips. As one consequence of this modifica- development of a gut process. tion, the secondary upper lip and outer den- Other predictions of our hypothesis (see tal arcade were interrupted in primitive fig. 17) are that the porolepiforms, if cor- choanates. This condition is shown by the rectly restored with two external nostrils, "hard-snouted" dipnoans (figs. 7, 8), and is had no choana; and that osteolepiforms, if hardly modified in larval urodeles (e.g., correctly restored with one external narial Schmalhausen, 1968, fig. 90) and in primitive opening, either had two narial tubes emerg- fossil tetrapods such as Ichthyostega and ing from that opening (like Polypterus) and other genera reviewed by Panchen (1967, pp. no choana, or were choanates exhibiting an 402-407), where the choana and external advanced condition of the nostrils, with a nostril are separated only by an inturned secondarily intact infraorbital canal between prong of the maxilla (fig. lOB). A second the choana and external nostril. Since the consequence of this modification was inter- second interpretation demands the indepen- ruption ofthe infraorbital sensory canal, as in dent acquisition of an uninterrupted infraor- dipnoans and primitive tetrapods. One pre- bital canal, we favor the first. Panderichthys diction of our interpretation is therefore that is restored as an osteolepiform with two ex- the lateral rostral of Ichthyostega (Jarvik, ternal nostrils. If correct, this would corrob- 1952), a tubular ossicle carrying the infraor- orate our view that other osteolepiforms had bital sensory canal between the nostrils, does a fenestra exonarina communis. not exist. In more advanced tetrapods, the

NASOLACRIMAL DUCT, LABIAL CAVITY, ROSTRAL ORGAN, AND JACOBSON'S ORGAN As noted above in the Introduction, the posing that the nasolacrimal duct of tetra- nasolacrimal duct of amphibians and the la- pods is homologous with the posterior ex- bial cavity of Neoceratodus (figs. 18-20) ternal nostril in bony fishes. He saw in have similar anatomical relations and onto- Eusthenopteron a canal in the endocranial genetic histories (figs. 18, 19). The only other postnasal wall connecting the nasal capsule structure in gnathostomes anatomically simi- with the orbit approximating the course ofthe lar to these dipnoan and tetrapod features is nasolacrimal duct of tetrapods, that Eusthe- the rostral organ ofactinistians. The rostral or- nopteron has but a single external narial gan, which is known in some detail in Lati- opening in the dermal bones of the snout and meria (fig. 22) (Millot and Anthony, 1965) is, that Porolepis lacks the canal but has a nor- like the labial cavity in Neoceratodus, filled mally developed posterior external nostril as with a cavernous mucosa, partly fills the re- well as an anterior one. From this he con- gion of the snout between the orbit and the cluded that the posterior external nostril of nasal capsule, and opens by pores to the out- Eusthenopteron has sunken into the side. The rostral organ, however, is very ex- endocranium of the snout to form a primitive tensive, the right and left ones are intercon- nasolacrimal duct, but that in Porolepis the nected in the midline and, on each side, there posterior external nostril is essentially un- are three pores opening to the outside, one modified. What, to us, seems basically un- on the snout and two between the orbit and convincing about Jarvik's proposal is that the posterior external nostrils. the canal in the endocranium of Eusthenop- Jarvik (1942) followed Allis (1932a) in pro- teron exactly parallels the course of the 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 197

NASAL CAPSULE

A B FIG. 18. Dissected snouts, in dorsal views, of A, Neoceratodusforsteri (Krefft) and B; larval Triturus alpestris (Laurenti), to show position of lungfish labial cavity and tetrapod nasolacrimal duct (A, after Jarvik, 1942, fig. 1 1A; B, after Jarvik, 1942, fig. 19B). maxillary and buccal nerves and accom- hausen's theory, as well as to all others (e.g., panying vessels as they emerge from the Allis, 1932a; Jarvik, 1942; Bertmar, 1965, anterior orbital wall and pass forward into 1968) postulating a former occurrence of two the posterior wall of the nasal capsule in elas- pairs of posterior nares. One is that a theory mobranchs, Polypterus, Amia, dipnoans, in which the nasolacrimal duct is said to form and urodeles and that there might well have from a seismosensory canal and a nostril is been an anterior and posterior nostril sepa- inconsistent with the fact that the nostrils are rated only by soft tissue as was probably true fully developed in urodeles prior to meta- of Laccognathus (Vorobyeva, 1980) and morphosis, whereas the nasolacrimal duct Holoptychius (the nariodal bone shown by appears later and, except in salamandrids, Jarvik within the external nasal opening has becomes confluent with the nasal sac. been included in the nasal opening of Hol- Schmalhausen's and Allis's theory requires optychius by supposition, not by observa- the presence of a posterior external nostril tion). At best, Jarvik's proposal ignores a prior to metamorphosis and this is not ob- simpler interpretation of the Eusthenopteron served in any choanate animal. The second canal based on frequently observed and objection is that no vertebrate animal is widespread anatomical relations of gnatho- known to have had three pairs of nostrils. stomes. We see our own theory, in which the nasal Although Schmalhausen (1968), like Jarvik placode fails to migrate outward and upward (1942) before him, had accepted Allis's to the side of the snout as in other fishes, as (1932a) theory of the transformation of the more simply accounting for the presence of fish posterior nostril into the nasolacrimal the two nostrils (an anterior one and a pos- duct and the development of the choana as terior one, or choana) in tetrapods, and in a neomorph (hence, a third functional open- dipnoans. This theory, as outlined above ing in the nasal sac of choanates), Schmal- (pp. 195-196), interprets the external nostrils hausen disagreed with Jarvik and Allis by of actinopterygians, actinistians, and rhipi- proposing that the nasolacrimal duct incor- distians as a result of ontogenetic migration porates material from the cephalic of the nasal placode. system, resulting in an interrupted infraor- Nevertheless, we find much merit in bital canal. Schmalhausen's ideas, particularly as they We have two major objections to Schmal- pertain to the homologies of the nasolacrimal 198 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

ANTERIOR 5 \ NA RES -. LAB I AL CAVI TY

?b~ 11111!4S 1

MEMBRANOUS i ROOF OF4 LABIAL CAVITY -

FIG. 19. Neoceratodus forsteri (Krefft), approximately 120 cm. total length, AMNH 40800. Photo- graphs of head to show A, opening of labial cavity; and B, membranous roof of cavity with a probe inserted. duct. We find interesting, for example, the indications that the duct could have had a topological and dependent developmental re- seismosensory origin.3 If, as discussed lations between the duct and the lacrimal and septomaxillary bones (fig. 23), as well as the 3 Schmalhausen argued that the nasolacrimal duct fact that extensions of the infraorbital canal forms from a detached extension of the infraorbital lat- can become confluent with the nasal sac, as eral line canal that has become confluent with the nasal 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 199

ROOF OF NASAL C A P S U L E

PTERYGO ID T0 T H P L AT E EXT ERN AL /NASAL VEIN-

-BUCCAL NERVE- INFRAORBITAL SENSORY CANAL LAMI NA -ORBITONASALIS- ,LABIAL CAVITY-

-LOWER LIPFOLD- MAN D tBU LAR TOOTH PLATE _-TON GUE--

.MECKEL'S - CARTILAGE

0.5mm FIG. 20. Neoceratodus forsteri (Krefft). Transverse sections through deepest part of labial cavity in 27 mm. (A) and 34.5 mm. (B) larvae (series F120, F121, Hubrecht Laboratory, Utrecht; same series used by Fox, 1965). Labial cavity at these stages is a deep pit at angle of mouth, whose apex is close to distal end of lamina orbitonasalis, and is attached to latter by connective tissue. This relationship is maintained in the adult, where cavity is greatly enlarged by increase of gape.

above, the labial cavity of Neoceratodus and sense organs comparable with neuromasts the rostral organ of Latimeria might be and an appropriate innervation. So far as homologues of the tetrapod duct, and if the presently known, the rostral organ of Lati- duct has a seismosensory origin, then one meria, according to Millot and Anthony should expect to find in the labial cavity, ros- (1965), has an extensive innervation entirely tral organ, and in the primitive tetrapod duct, from the n. ophthalmicus superficialis (V) which is characteristically associated with sac because: (a) the formation of the lacrimal and sep- lateral line innervation (see, e.g., Strong, tomaxillary bones is dependent on the presence of the 1895; Norris, 1913). Fibers from at least the nasolacrimal duct (Schmalhausen cites the experiments ventral branches of the n. buccalis would not of Medvedeva, 1959, 1960, in which the anlage of the be expected in the rostral organ since Lati- duct was removed and the two bones failed to develop); meria has a well-developed and uninterrupt- (b) the lacrimal and septomaxillary bones are membrane ed infraorbital canal; n. buccalis innervates bones that, at metamorphosis, come to surround com- neuromasts pletely the nasolacrimal duct but underlie the duct at of the posterolateral snout re- first-thereby closely resembling the development of gion between the eye and anterior nostril. Do seismosensory canal bones in bony fishes (Schmalhau- the nerve fibers in Latimeria go to some sen's fig. 103 shows these relations in a larva of the form of seismosensory organ? Are there urodele Onychodactylus); and (c) parts of the infraor- such organs in the labial cavity of Neocerat- bital canal are known to become confluent with the nasal odus and are they innervated by n. ophthal- sac in some fishes (Schmalhausen illustrates this, fig. micus superficialis? Do the nasolacrimal 104, in an individual of Amia). ducts of the anuran Xenopus and apodans, 200 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

FIG. 21. Neoceratodus forsteri (Krefft). Snout of 15 mm. larva, AMNH 40801 (alcian blue/alizarin counterstained, cleared preparation), in ventral view. which are associated with tentacular organs Salamandridae, do not become confluent of presumed sensory function, have a similar with the nasal sac? Future work will decide. innervation and are there seismosensory or- If Schmalhausen's theory is correct, how- gans in the nasolacrimal ducts which, in the ever, the duct should be associated ontoge-

POSTERIOR EXTERNAL OPENINGS OF ROSTRAL ORGAN

LATERAL ROSTRAL B CANAL FOR BUCCAL NERVE FIG. 22. Latimeria chalumnae Smith. A, outline of head to show nostrils and openings of rostral organ; B, diagram of rostral organ and nasal sac of right side in dorsal view; C, snout in lateral view (A, after Millot and Anthony, 1958; B, C, after Millot and Anthony, 1965). 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 201

OUTLINE OF CHOANA FIG. 23. Nasal structure in hynobiid urodeles. A, Ranodon sibiricus Kessler, diagrammatic recon- struction of snout of a 22 mm. larva, ventral view; B, Onychodactylusfischeri (Boulenger), diagrammatic reconstruction of left nasal sac of 85 mm. (postmetamorphic) individual in dorsal view. After Schmal- hausen (1968, figs. 98A, 103). netically with the infraorbital system and, lungfishes. But even this difference appears hence, with n. buccalis, and, if that proves to be minimal when lungfishes are compared to be the case, it would appear to rule out with 22 mm. Ranodon larvae in which, ac- the n. ophthalmicus superficialis innervated cording to Schmalhausen's (1968) figure 98A rostral organ of Latimeria as a homologue. (fig. 23A), the external nostril opens ventrally We find ourselves in disagreement with on the margin of the lip as in lungfishes and one other feature of Schmalhausen's analy- ichthyostegids. It is the snout in modern sis, and that concerns the explanation of the urodeles, anurans, and that ap- interruption of the infraorbital canal. The in- pears to have turned upward, carrying the terruption typically is at the point where, in external nostril into a more dorsal position. lungfishes and tetrapods, the canal meets the One additional similarity between the na- nostril and does not appear to be associated sal apparatus of lungfishes and tetrapods with the ontogenetic history of the choana, concerns a lateral diverticulum of the nasal whatever that history might be. In both sac. In lungfishes it has generally been given groups, the infraorbital canal bends down- the neutral term of diverticulum laterale, ward toward the oral border [in dipnoans, whereas in tetrapods it is often identified as temnospondyls, lepospondyls, and in those a or Jacobson's organ. In anthracosaurs with a ventrally situated ex- both groups the diverticulum is a lateral out- ternal nostril, and in larval Necturus and pocketing of the posterior part of the nasal Ranodon (see, e.g., Stensio, 1947, figs. 2, sac (fig. 24) and consists largely of olfactory 24A; Schmalhausen, 1968, figs. 100, 101; epithelium. In the more terrestrial amphibi- Miles, 1977, figs. 18, 35, 111, 112)] and the ans Jacobson's organ is better developed and similarity in the courses of the supraorbital is associated with the nasolacrimal duct. In and infraorbital lateral lines is particularly some forms it includes a narrow strip of res- striking between lungfishes and such larval piratory epithelium along its dorsal part (Jur- hynobiid urodeles as Ranodon (Schmalhau- gens, 1971, fig. 31). In this dorsal position in sen, 1968, fig. 90). Differences between lung- at least some Protopterus the epithelium ap- fishes and urodeles in the external form of pears to be somewhat differentiated and is the snout appear to be associated primarily less nucleated (Bertmar, 1965, fig. 18D). A with turning under of the tip of the snout in comparison of Bertmar's (1965) figure 18C of 202 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

39A) (fig. 24). Bertmar (1965, 1968) has ar- gued that a large diverticulum laterale is cer- tainly formed in all three dipnoan genera but that "it develops into an olfactory fold as do other lateral diverticles of the nasal sac, and therefore it does not seem to be homologous with the vomeronasal organ . . . in tetra- pods." Since the lateral diverticulum is, in fact, an outpocketing of olfactory epithelium in both dipnoans and tetrapods, Bertmar's reasoning escapes us. Perhaps he is saying, FIG. 24. The nasal sac in A, Triturus alpestris however, that because dipnoans have other (Laurenti); and B, Protopterus annectens Owen. diverticula, the large posterolateral one can- A, diagram of right sac in dorsal view, after Jur- not be homologous with the tetrapod diver- gens (1971, fig. 39A); B, diagram of right sac, in ticulum in the same position. We are equally dorsal view, of 20 mm. larva, after Bertmar mystified by this argument. The question is: (1965, fig. 18B). do other gnathostomes besides dipnoans and tetrapods have a lateral diverticulum of this a transverse section of the nasal sac and lat- type and in this position on the nasal sac? If eral diverticulum of Protopterus with com- the answer is negative, and Bertmar has not parable sections in various urodeles (Jur- suggested that such a modification of the na- gens, 1971, fig. 32; Higgins, 1920, pl. 8, fig. sal sac occurs elsewhere, the structure in the 55; Medvedeva, 1968, fig. 1 showing Triton two groups is most simply regarded as syn- and Ranodon) makes it clear why the dip- apomorphous. Bertmar (1965) does not re- noan structure has been likened to the tet- fute Rudebeck's (1945) claim in his study of rapod vomeronasal organ. The similarity is the forebrain of Protopterus that Jacobson's heightened by a comparison of Bertmar's organ is present in the as a rudimen- (1965) reconstruction of the whole nasal sac tary accessory olfactory bulb and that a ru- of Protopterus (his fig. 18B) and with Jur- dimentary vomeronasal nerve may be gens's (1971) of the sac in Triturus (his fig. formed.

PAIRED FINS AND GIRDLES (A) PAIRED FIN SKELETON dorsal process that has been likened to a tet- Although the paired appendages of lung- rapod humeral process. The terms humerus, fishes have been cited as precursors of tet- ulna, radius and femur, tibia, and fibula are rapod limbs (Braus, 1901 and many earlier in general use for the three radial elements workers; Holmgren, 1933, 1949a, 1949b, as a in the osteolepiform pectoral and pelvic fins. precursor of urodele limbs), the osteolepi- Preference for the osteolepiform structures form crossopterygian fins are today assumed seemed to be reinforced by the observation to be closer to the tetrapod structure (Greg- that, at the base of lungfish fins, the principal ory, 1911; Watson, 1913; Westoll, 1943, radials are in a single series with smaller ones 1958a; Jarvik, 1964, 1965; Thomson, 1968, radiating outward from each side of the cen- 1972; Andrews and Westoll, 1970a, 1970b; tral axis as do the barbs on the rachis of a Rackoff, 1976, 1980). The favored compari- feather. Moreover, Neoceratodus fins were son of osteolepiforms is attributable to their seen to be very similar to those of Paleozoic possession of a pair of distal bones (one short xenacanth sharks and Gegenbaur (1895) had and stout, the other long and slender) artic- represented this "archipterygial" fin as the ulating on a single stout basal radial with a primitive type for gnathostome fishes. It ap- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 203 parently went unnoticed that, during Gegen- around the fanlike shoulder girdle, in one row baur's era, Semon (1898), in a study of the laterally and in three or more rows poste- embryology of the pectoral fin of Neocerat- riorly (Stensio, 1959, fig. 14A). In Pseudo- odus, showed that the second axial radial petalichthys problematicus Moy-Thomas, all element is ontogenetically double. that is known of the endoskeletal support is In reviewing the skeletal anatomy of the a pair of somewhat triangular basal elements paired appendages and their girdles in each of which articulates separately in a gnathostomes we have concluded that lung- notch in the shoulder girdle (Gross, 1962, fig. fish fins include synapomorphies with tetra- 7). This last fossil also shows that pelvic ra- pod limbs and that there are synapomorphies dials were present in placoderms, in this of the fins and girdles in all chondrichthyans case 10 or 11 rods parallel to each other and, and osteichthyans. as a group, parallel to a more medial elongate Osteichthyan paired fins, as pointed out by bone. Gross (1962, p. 76 and fig. 7) inter- Goodrich (1930), are of two kinds, those of preted this elongate bone as a basipterygium sarcopterygians in which pectoral and pelvic (=metapterygium). He described it as having fins are constructed according to the same a longitudinal ridge, which makes it unlike basic plan, and those of actinopterygians in the metapterygial axes of other fishes. It which the pectoral and pelvic differ. The seems to us that the row of parallel radials differences between the two groups of bony is most simply interpreted as a primitive se- fishes may be understood by examining the ries of basal radials, and the medial bone fin skeletons of other gnathostome fishes. In lying against the radials on one side, and at acanthodians all that is known about endo- the anterior end of the series of radials on skeletal fin supports is that in Acanthodes the other, as a pelvic girdle. A summary of bronni the pectoral fin has a few small nod- acanthodian and placoderm fin anatomy in- ules arranged irregularly at the fin base dicates that the former have nothing com- (Miles, 1973). In arthrodires a condition parable with the rodlike radials of other more nearly like that of chondrichthyans has gnathostomes and that the latter have basal been found, especially in Brachyosteus diet- radials but lack the details of endoskeletal richi Gross, in which the pectoral fin includes anatomy of sharks and bony fishes. a segmented metapterygial axis articulated The pectoral fin of living chondrichthyans with a posterior process on the scapulocor- has often been characterized as tribasal be- acoid and a series of radials arranged along cause there are three distinct cartilages artic- the articular surface of the scapulocoracoid ulating with the shoulder girdle that support anterior to the metapterygium (Stensio, three or more rows of distal radials. In fossil 1959, pl. 15, figs. 1-3). The term metapte- chondrichthyans the fin anatomy may be rygium as used here refers simply to one or much simpler in lacking a definite propteryg- more basipterygial cartilage or bone seg- ium along the fin's leading edge and a central ments aligned in series, joined to a posterior cartilage, the mesopterygium, between the articulatory process or surface on the scap- former and the metapterygium at the fin's ulocoracoid, and forming the posterior or trailing edge (fig. 26A, B). The tribasal fin trailing endoskeletal support of the fin. In the seems to be characteristic only of some eu- fin of Brachyosteus, as in all primitive selachians.4 Among non-euselachian chon- fins, the metapterygium appears to be a main drichthyans present in the Paleozoic the axis of fin support. An arrangement of fin modern tribasal condition did not exist (Zan- supports like that of Brachyosteus occurs gerl, in press). Instead, the most general con- also in the arthrodire Enseosteus jaekeli dition, as exemplified by species of Clado- Gross (Stensio, 1959, pl. 2). In another group selache (fig. 25A), Cladodus, Cobelodus (fig. of placoderms related to arthrodires, the pe- 25B), Fadenia, and Xenacanthus is the talichthyomorphs (Miles and Young, 1977), pectoral radials are also present. In Gemuen- 4Euselachians include all modern sharks and dina sturtzi Traquair, radials are arranged rays, and some fossil groups (Zangerl, in press). 204 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

gin of the scapulocoracoid in front of the metapterygium. Various morphologically transitional conditions in some non-eusela- chians suggest that anterior radials enlarge or fuse to form the propterygium and mesop- terygium of euselachians (cf. figs. 25-27). Correlated with the development of a tribasal fin is the shift in the main point of fin artic- ulation from the metapterygium in non-eu- selachians to the propterygium in euselachi- ans. It is clear, therefore, that cladistic comparisons involving chondrichthyan fin structure should be with a Cobelodus-like fin FIG. 25. Left pectoral fin skeleton of A, Cla- in which there are no more than four or five doselache brachypterygius Dean; and B, Cobel- segments in the metapterygium, seven or odus aculeatus Zangerl and Case (after Zangerl, eight segmented preaxial radials articulating in press, fig. 34). on the scapulocoracoid anterior to the me- tapterygial base, and the main axis of fin sup- port on the trailing (metapterygial) rather presence of a well-defined metapterygial axis than leading (propterygial) side of the fin. made of an elongate basal segment that ar- The pelvic fin skeleton of chondrichthyans ticulates with the girdle followed by one to has apparently had a history similar to, but 13 cuboidal or rectangular segments. On the less complex than, that of the pectoral. A preaxial side of the metapterygium (facing metapterygium, also the main axis of fin sup- the fin's leading edge) there are numerous port, has numerous radials attached along its parallel radials arranged at an angle between preaxial surface. In the pelvic skeleton, how- 70 to 90 degrees with the metapterygial axis. ever, only a few radials extend anterior to Zangerl (in press) regarded those species with the metapterygial base, the metapterygium unsegmented radials and a long whiplike me- has only one to three segments distal to the tapterygium as being somewhat specialized. basal articular segment, the metapterygium He constructed what he considered to be a retains its function as the main axis of sup- generalized non-euselachian pectoral (his fig. port for the fin even in euselachians in which 35A) in which there are only four metapter- a small propterygial element develops (fig. ygial segments, and proposed that the earli- 26B), and the fin is characteristically either est transition toward the pectoral of modern unibasal or (in some euselachians) dibasal. sharks involved segmentation of radials (his In male chondrichthyans the terminal metap- fig. 34). We would state the generalized con- terygial segments (fig. 26B) form internal dition of the pectoral somewhat differently support for a clasper (claspers are not known by observing that in Zangerl's cladogram of to be present in some cladodont sharks). the main groups of sharks (his fig. 52), in In the discussion that follows we will in- which euselachians and hybodonts are de- terpret actinopterygian paired fin structure picted as the sister group to all other sharks, as a transformation of a primitive (chon- segmented radials are present in both sister drichthyan) metapterygial fin into a propter- groups and that long whiplike metapterygia ygial type, paralleling the development of the occur only in groups with cladistically apo- euselachian fins, and sarcopterygian fin morph positions among non-euselachians. structure as an alternative transformation of Also characteristic of these primitive shark a primitive metapterygial fin into an exclu- pectorals is the presence of radials anterior sively metapterygial structure primarily by to the base of the metapterygium as in some the loss of all radial elements anterior to the arthrodires. As many as 10 of these anterior metapterygial base. We feel justified in pur- radials attach along the posteroventral mar- suing these comparisons because the basic 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 205

POSTAXIAL RADIALS

METAPTERYGIUM,_) Gt

A

FORAMINA FOR DIAZONAL NERVES

PROPTERYGIUM -PREAXIAL RADIALS FIG. 26. Left pectoral (A) and pelvic (B) fin skeleton, in dorsolateral view, of 17 cm. d Squalus acanthias Linnaeus, AMNH 40804. structural elements of the pectoral and pelvic transformation series between osteichthyan fins of osteichthyans are present in primitive fin structure and that of acanthodians, as sharks and placoderms as well (e.g., Brach- would be required by Miles's (1973) theory yosteus and Pseudopetalichthys). To infer a of relationships, requires the assumption that 206 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

both Acipenser and Eusthenopteron differ from those of chondrichthyans and resemble JUVENILE CLASPER one another in the lower number of distal METAPTERYGIUM radials anterior to the metapterygial base, and the alignment on the metapterygial axis of one basal preaxial radial per metapterygial segment. The metapterygial axes in Eusthe- nopteron may be distinguished from those of Acipenser by having posterodorsal process- 10 es (=entepicondyles of Andrews and Wes- toll, 1970a) in the first and third, second and third (AMNH P. 3871), or second (AMNH P. 8095) pectoral metapterygial segments (although some specimens have none at all; Holmgren, 1933, fig. 18) (fig. 29) and by hav- ing a greatly reduced pelvic skeleton in which there are only two metapterygial seg- ments with three basal radials attached to them. Hence, Acipenser appears to have re- FIG. 27. Left pelvic fin skeleton of 20 cm. d a Hydrolagus colliei (Lay and Bennett) in dorsolat- tained primitive chondrichthyan arrange- eral view, AMNH. ment of radials but has paralleled euselachi- ans by shifting part of the fin support to the propterygial side, and Eusthenopteron ap- pears to have retained the primitive chon- complex similarities between osteichthyans drichthyan pattern of four pectoral metap- and chondrichthyans were attained indepen- terygial segments and two pelvic ones and, dently. by losing all radials anterior to the metapter- The difference between actinopterygians ygium, to have retained a primitive metap- and sarcopterygians in pectoral and pelvic terygial support for the pectoral and pelvic structure can be appreciated by contrasting fins. the paired fin endoskeletons of Acipenser With respect to fin structure in general, and Eusthenopteron (fig. 28). Even a cursory Jessen (1972) has shown that Acipenser is comparison shows that the most complete of typical of the members of other plesiomorph the pectoral and pelvic fin skeletons of Eus- groups of actinopterygians, including various thenopteron illustrated by Andrews and paleoniscids (and see also Nielsen, 1942, Westoll (1970a, fig. 8J) have branching pat- 1949), as Andrews and Westoll (1970b) have terns that resemble in detail the arrange- shown that Eusthenopteron is similar to oth- ments of metapterygial elements and their er "rhipidistian" fishes such as preaxial radials in Acipenser. Acipenser fins (known only from part of a shoulder girdle appear to differ from those of Eusthenopte- and fin) and Megalichthys. According to ron only in the presence of three sets of ra- Thomson (1972) and Rackoff (1980) Sterrop- dials anterior to the metapterygial axis and terygion brandei, of which both pectorals of a heavy, fenestrated pectoral propterygial and pelvics are known, is, again, much like segment that Jessen (1972) identified as a Eusthenopteron. character of actinopterygians. It is also evi- Although for more than a half-century dent that the fins of Acipenser resemble, in Eusthenopteron fin structure has been con- considerable detail, the paired fins of chon- sidered antecedent to tetrapod limbs, and the drichthyans, differing from the primitive most recent interpreter of this doctrine has shark fins (fig. 25), but resembling those of even identified elbow and ankle joints in the euselachians (fig. 26), in the development of fins of Sterropterygion (Rackoff, 1980), we a strong propterygial articulation. The fins of believe that Goodrich's (1930, p. 161) cau- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 207

/ > ~~~~~~PROPTERYGIUM \/ PREAXIAL RADIALS

PELVIC GIR

FIG. 28. Left pectoral (A) and pelvic (B) fin skeleton, in dorsolateral view, of adult Acipenser sturio Linnaeus, BMNH E. 170-1, distal elements in A from Jessen (1972); C, D, left pectoral (C) and pelvic (D) fin skeleton ofEusthenopteronfoordi Whiteaves, in lateral view, after Andrews and Westoll (1970a, figs. 8H, J, 27).

tionary comments are still appropriate. It must be met, including some of the points was clear to Goodrich that we should expect made by Goodrich: a fishlike relative of the tetrapods to possess "6pectoral and pelvic fins alike in structure, 1. An unpaired basal element (humerus or with outstanding muscular lobe, extensive femur) supporting entire appendage. endoskeleton with at least five radials [=me- 2. Paired, subequal subbasal elements (ulna- tapterygial segments], small web, and few if radius and fibula-tibia) that are function- any dermal rays. But. .. the fin (espe- ally joined distally. cially the pelvic fin) is too specialized [in 3. Two primary joints (shoulder or hip and Eusthenopteron] ... with short endoskele- elbow or knee) in each appendage. ton, large fin-web, and powerful dermal 4. Distal bony elements (carpals or tarsals rays. " and digits) arising on postaxial side of ap- We specify below seven general condi- pendage (see p. 211). tions of the tetrapod limb that we believe 5. Pectoral and pelvic appendages alike in 208 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

tapterygial axis, and the fifth is also, because the osteolepiform pectoral has three or four axial elements and the pelvic only two. The sixth criterion is rejected because osteolepi- forms have very short muscular lobes as judged by the posterior limits of the radials (when known), scale cover and position of the fin-ray bases, and the seventh criterion is also denied because of the extensive fin FIG. 29. Eusthenopteron foordi Whiteaves, web. left pectoral fin endoskeletons, showing variable development of processes on postaxial (dorsal) In porolepiform "rhipidistians" criteria margin of metapterygial segments (compare fig. six and seven are met with respect to the 28). Upper left, AMNH P. 8095; upper right, pectoral fins only. Internal skeletal anatomy AMNH P. 3871; lower, from Holmgren (1933, fig. of porolepiform paired fins is virtually un- 18). Note spread radials of upper two fins and known, however (see Jarvik, 1972). contracted radials of lower fin. Dipnoans, on the other hand, meet all of the seven criteria (fig. 30A, B), many of structure (i.e., structurally similar and which were previously discussed by Holm- with at least four or five primary seg- gren (1933). There are some minor points of ments). disagreement between Holmgren's and our 6. Extensive muscular lobe that can be ex- analysis which require some comment. The tended well below the ventral body wall. second criterion is not evident in adult lung- 7. Reduction or loss of dermal rays. fishes, though, because ontogenetically paired sub-basal elements fuse to form a single Of these seven criteria, osteolepiforms sat- broad cartilage (fig. 31). But there is no in- isfy only the first and part of the second, that dication either in early or late development is, they have an unpaired basal element (the of a significant inequality in length of these first segment of the metapterygial axis) and elements and the two sub-basal pelvic fin they have a second axial segment with the cartilages fuse incompletely and show no in- proximal end of which a large radial articu- dications of the divergent angle between the lates. The second criterion is only partly met second pelvic metapterygial segment and the because the second axial segment and asso- first preaxial radial that occurs in Eusthe- ciated radial are decidedly unequal in length nopteron (figs. 28, 29). In early development and the two bones diverge distally. In some of the pectoral fin, the radial associated with specimens of Eusthenopteron, the two bones the second metapterygial segment actually are more closely aligned than in others (e.g., curves down toward that segment at its distal in the specimens illustrated by Andrews and end (fig. 31; and see also Holmgren, 1933, Westoll, 1970a, as compared with two spec- fig. 11). It will be noticed by comparing fig- imens seen by us; see figs. 28D, 29) and this ures 30A and 30B of the pectoral and pelvic suggests to us that the Eusthenopteron fin, fins of Neoceratodus that the pectoral is like those of other fishes, could be spread structurally the obverse of the pelvic, al- and contracted, corresponding with a motion though the two fins are constructed accord- of the radial away or toward the second axial ing to exactly the same plan. In other words, segment. In turn, this means that the distal the larger radials, which stand in a one-to- tips of the two bones are not functionally one relationship with the metapterygial seg- joined. The third criterion is not met be- ments, are postaxial (dorsal) in the pectoral cause there is no joint of tetrapod type and preaxial (ventral) in the pelvic. Likewise between the first and second axial seg- the radial that is functionally and ontogenet- ments (but Rackoff, 1980, disagrees). The ically associated with the second metapter- fourth is rejected for osteolepiforms because ygial segment is postaxial in the pectoral and all radials are on the preaxial side of the me- preaxial in the pelvic. Semon (1898, p. 68), 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 209

/

ULNA + RADIUS HUMERUS PREAXIAL RADIALS IN POSTAXIAL POSITION

ANOCLEITHRUM\ SCAPULOCORACOID

A BALL AND SOCKET JOINTS POSTAXIAL RADIALS IN PREAXIAL POSITION

TIBIA POSTAXIAL RADIALS

FEMUR / FIBULA PREAXIAL RADIALS FIG. 30. Neoceratodusforsteri (Krefft), left pectoral (A) and pelvic (B) fin endoskeletons in positions of rest against body. AMNH 36982, 18 cm. total length.

as mentioned earlier, described the devel- prior to the formation of the prochondral an- opmental sequence at ontogenetic stage 45 lage of the radials in Neoceratodus and car- of how the pectoral anlage rotates dorsal- pals (tarsals) and digits in urodeles; and (2) ward 90 degrees from the horizontal, and the by identifying the segments of the primitive pelvic anlage ventralward 90 degrees from metapterygial axis in the developing urodele the horizontal, so that the preaxial margins limb, one could discover whether the carpals of the two fins have a 180 degree different (tarsals) and digits are pre- or postaxial and orientation. Part of Holmgren's argument for are therefore homologous with pre- or post- drawing a comparison between the two axial lungfish radials. Holmgren's argument paired appendages of Neoceratodus and uro- is complex, but reduces to a conclusion that deles is his inference that the carpals and tar- the limbs of urodeles include both pre- and sals and digits of urodeles are preaxial radials postaxial radials. We believe that Holm- now in the postaxial position (presumably gren's goals are worthy but that his analysis also requiring that in a lungfish-like ancestor is contrived, and that the problem can be re- of the urodeles the pelvic anlage has rotated duced to a set of simple comparisons. dorsalward together with the pectoral). He For our primary comparisons with Neo- made this inference for two reasons: (1) in ceratodus we select the same urodeles as did Neoceratodus and urodeles the unpaired Holmgren. We choose the hynobiid salaman- basal and paired sub-basal elements of the ders because the hynobiids are considered paired appendages (the first two metapter- to be among the most primitive of all liv- ygial segments and the radial associated with ing urodeles (e.g., in their reproductive meth- the second segment) in Neoceratodus and ods, arterial system, unfused lacrimal, sep- the humerus (femur) and radius and ulna (tib- tomaxilla, angular and prearticular bones, ia and fibula) of urodeles all appear in the size of columella and correspondence in po- primary blastema of the limb endoskeleton sition to the lungfish in ad- 210 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

PREAXIAL SIDE IN POSTAXIAL POSITION

1 V ULNA

JOINTS

:LAVICLE

FIG. 31. Neoceratodus forsteri (Krefft), left pectoral fin endoskeleton of 15 mm. larva in which pelvic fins are not yet present. AMNH 40801.

vanced larvae; Schmalhausen, 1968, pp. ton (figs. 31, 33). The existence of this sec- 184-187, figs. 110-113). We choose hynobiids ond joint in lungfishes is easily appreciated also because in the earliest stages of limb in cleared and stained preparations of small development the paired appendages appear specimens which show that the fin is cocked as leaf-shaped fins with a strong muscular at this juncture and in large and small spec- lobe as in larval Neoceratodus (fig. 32). imens by the presence of a ball and socket We agree with Holmgren that the occur- joint surface between the first and second rence in Neoceratodus and Hynobius of metapterygial segments (figs. 30A, 31). an unpaired basal and paired sub-basal ele- These observations therefore satisfy crite- ments in the primary blastema of the ap- rion number 3, above. pendicular endoskeleton prior to the for- The application of the fourth criterion, mation of more distal elements is probably whether the distal elements of the fin endo- significant since in primitive gnathostomes skeleton are postaxial, depends on discov- the fin endoskeleton first appears as a pro- ering the position in tetrapods of what re- cartilaginous plate continuous with the pri- mains, if anything, of the metapterygial fin mary girdle and later separates from the axis. In Neoceratodus, the metapterygium girdle and then subdivides into radials extends distally to the pointed tip of the fin. (Goodrich, 1930). But what strikes us as be- In developing Hynobius no skeletal element ing a more complete, and therefore more in- extehds to the tip of the fin membrane, but formative, comparison is that the radial as- instead the first digit extends ventrally to the sociated with the second metapterygial preaxial margin and digits two to four (in the segment of Neoceratodus and the ulna of pectoral) or five (in the pelvic) extend dor- Hynobius and other urodeles are joined to the basal segment (humerus) in the same sally to the postaxial fin margin. Holmgren way, viz., mainly along the postaxial (dorsal) assumed that three parts of the Hynobius surface of the humerus, and that their paired limb are important for this question, the ra- sub-basal segments (radius and ulna) form a dius (tibia), the ulna (fibula) and the central definite joint surface with the humerus not carpals (tarsals); the lateral carpals (tarsals) found between more distal segments or in and digits he considered to be secondary ele- any other gnathostome pectoral endoskele- ments. He reasoned, since there are three 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 211 primary branches and, since the urodele limb must have arisen phylogenetically from an archipterygium with preaxial and postaxial radials on a central metapterygium, that the middle of the three branches in Hynobius must be homologous with the dipnoan me- tapterygium. Holmgren's argument is clearly FIG. 32. Hynobius chinensis Gunther, left pel- loaded by the assumption of what is primary vic "fin" of 3 cm. larva. AMNH 30616. Positions and secondary for these comparisons and by of anlagen of first four of five digits shown by his assumption that each of the three ele- dashed lines. ments of a lungfish fin is represented in the tetrapod limb. Our own study of developing terion by being of similar form and construct- Hynobius limbs suggests to us that (1) there ed along virtually identical lines (although are only two morphological units, a ventral they are 1800 out of alignment with each oth- unbranched series of segments (radius + ra- er). The high number of axial elements in diale + prepollex) and a dorsal branched se- living and many fossil dipnoans and the pres- ries (ulna + carpals + digits) (fig. 33); (2) the ence of true postaxial radials are probably V'entral unbranched series is composed of derived, however, and are autapomorphous more robust segments; (3) a line drawn for these species. In more primitive lungfish- through the axis of the ventral unbranched es, such as Griphognathus, the muscular fin series intersects at or very near the pointed bases are stout and neither so long nor so tip of the larval fin; (4) the base of the slender as in Neoceratodus (Schultze, 1969a, first element of the dorsal series (the ulna) fig. 31). has a dorsal contact with the humerus as Criteria 6 and 7 relate to features that are does the dorsal radial in the primary fin blas- well known in lungfishes and their signifi- tema of Neoceratodus; and (5) the base of cance is enhanced by the use of the fins, par- the first element of the ventral series (the ra- ticularly in Neoceratodus, for walking across dius) has a socket for the ball joint of the a substrate by their alternate side-to-side humerus as does the second metapterygial motions in connection with sinusoidal mo- segment in Neoceratodus. We conclude, tions of the trunk and tail as in urodeles. as did Romer and Byrne (1934, p. 44), The close similarity in paired appendages that the ventral unbranched series (radius + between dipnoans and tetrapods does not im- radiale + prepollex) represents the primi- ply, of course, that Eusthenopteron fins are tive metapterygium, and that the dorsal without significance for this problem, but branched series (ulna + carpals + digits) rep- that, cladistically speaking, their significance resents preaxial radials that are in the rotated lies at a higher level ofgenerality which spec- postaxial position. The synapomorphy that ifies only that Eusthenopteron is the sister we therefore specify between Neoceratodus taxon to a group that includes dipnoans and and Hynobius is the ontogenetically rotated tetrapods and perhaps also porolepiforms as appendage with its preaxial radials on the the sister taxon to those two. postaxial side, and this would satisfy crite- Since osteolepiforms have a probably rion number 4, as given above. We must also primitive number of metapterygial segments point out, however, if we are wrong and in both fins but fewer preaxial radials than Holmgren is right, then our fourth criterion, chondrichthyans, actinopterygians, dip- restated in Holmgren's terms, would still re- noans and tetrapods, we infer that osteolep- late dipnoans and tetrapods and exclude Eus- iforms exhibit a trend in radial reduction. thenopteron. We simply feel that our inter- Radial reduction is especially evident in the pretation is more closely bound to simple pectoral fin of Panderichthys (Vorobyeva, morphological comparisons and carries a 1975, fig. 3). Coelacanths also have a single smaller burden of assumptions. radial associated with the first two metapter- Lungfish fins certainly satisfy the fifth cri- ygial segments, and these radials are reduced 212 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

ULNA

GIRDLE RAD IUS

RAD IUS

FIG. 33. Hynobius chinensis Gunther, left pectoral appendage of 3 cm. larva. AMNH 30616. Upper figure, diagram of girdle and limb showing four digits clustered into two distinct developmental fields. Lower figure, detail of cartilaginous anlagen of segments of digits within the two fields. Ulna and prepollex are hypothesized to be homologues of fish metapterygium, and radius, carpals and digits, homologues of fish preaxial radials now in postaxial position (compare figs. 30, 31). to tiny nubbins embedded in connective tis- nopteron in which the "humeral processes" sue (fig. 34). Such reduction suggests their for muscle attachment have the primitive os- membership in a group including osteolepi- teichthyan orientation (cf. figs. 28, 30, 34). forms. But contrasting with this interpreta- Thus, whereas one character (radial reduc- tion is the presence of four metapterygial tion) somewhat ambiguously aligns coela- segments in the pelvic fin and the mirror-image canths with osteolepiforms, a second (four complete similarity of pelvic and pectoral en- pelvic radials) and a third (pectoral rotation) doskeletons (fig. 34), features which are pre- align them as a sister group of dipnoans plus sumably derived and suggest that coelacanths tetrapods. Although one might wish to argue fit near dipnoans and tetrapods. Also indica- the exact cladistic placement of coelacanths tive of coelacanth-dipnoan-tetrapod relation- by using alternative interpretations of paired ship is the evident rotation ofthe preaxial side fin characters, the placement of these fishes ofthe coelacanth pectoral to the postaxial po- with the sarcopterygians is unambiguous: sition so that, like dipnoans, the pectoral and coelacanths completely satisfy the first cri- pelvic fins are out of phase. Millot and An- terion, as do osteolepiforms, and, like Eus- thony (1958) illustrated the pectoral fin of thenopteron, they have entirely metapteryg- Latimeria wrong way up in their illustration ial fins, thus rejecting theories of relationship of the skeleton, but their stated purpose was that would place them as the sister group to show its similarity to the fin of Eusthe- either of actinopterygians, of all other os- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 213

PREAXIAL SIDE IN POSTAXIAL POSITION

FIG. 34. Latimeria chalumnae Smith, left pectoral (above) and pelvic (below) fin endoskeletons of unborn juvenile of 34 cm. total length (AMNH 32949) shown in positions of rest against body. teichthyans (e.g., Wiley, 1979), or of chon- Pelvic anatomy also contributes to an un- drichthyans (as suggested by some workers derstanding of cladistian relationships. Sten- with a penchant for treating shared primitive sib (1921, 1925) was the first to recognize that characters as evidence of relationship; cf. the pelvic girdle of actinopterygians may Forey, 1980). We also reject the conclusion have a unique ontogenetic history. He sug- of Bjerring (1973) that the relationship of gested that the girdle has a compound struc- coelacanths to actinopterygians and sarcop- ture, consisting of the pelvic plate to which terygians, or to chondrichthyans and os- the metapterygium is fused. We agree with teichthyans, is unresolved. Stensio in substance if not in detail. Our own This discussion would be incomplete with- conclusion that actinopterygians might not out some comment on the position of cla- have a primary pelvic girdle derives from a distians within the Osteichthyes, since Po- comparison of the endoskeletal fin supports lypterus, as the "type" crossopterygian, has in gnathostomes generally. In acanthodians often been compared with tetrapods (e.g., the only known skeletal support of the pelvic Pollard, 1891, 1892), and Klaatsch (1896) fin is a heavy spine along the fin's leading even tried to relate its fin structure to the edge. All other gnathostomes have a com- tetrapod limb. Polypterus, however, has the plex internal fin skeleton primitively without propterygial margin of the pectoral fin a leading spine. In sharks the pelvic is struc- strongly developed (fig. 35) which, within the tured much like the pectoral; there are two, Osteichthyes, is a characteristically acti- or sometimes three, rows of radials joined to nopterygian feature. the preaxial side of the metapterygium (fig. 214 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

FIG. 35. Polypterus ornatipinnis Boulenger, left shoulder girdle and pectoral fin endoskeleton. AMNH 40802, 14 cm. standard length.

26B). In primitive actinopterygians such as inner row of long radials and an outer row of Polyodon (fig. 36A) and Acipenser (fig. 36B) very short ones) comparable with the preax- (and some paleoniscids as well; Lehman, ial radials of chondrichthyans and 1966), there is a series of radial-like carti- osteichthyans, we conclude that the surface lages that either fuse (in most adult forms) or along which they attach is at least, in part, remain separate (in juveniles and some adult metapterygial (fig. 38). forms). Two rows of radial elements are Goodrich (1930) objected to arguments joined to the outer surface of these internal like those of Stensio's on the grounds that cartilages, an inner row of elongate radials the pelvic girdles of actinopterygians are and an outer row of shorter ones, the whole pierced by foramina for diazonal (=spinal) thus closely resembling a metapterygium and nerves as are the girdles in other gnatho- its preaxial radials in chondrichthyans and stomes, thus indicating a common plan of sarcopterygians; no other recognizably sep- development. These foramina are explained arate structure lies anteromedial to these in- by the fact that in most gnathostomes the ternal cartilages. The various conditions in girdle is primarily outside the myotomes in actinopterygians suggest that there is a trend early development and is merely an exten- toward consolidation of the internal carti- sion inwards from the base of the fin skeleton lages both ontogenetically and phylogenet- with which it initially forms a continuous pro- ically, and a reduction in the length of the cartilaginous plate. As the girdle grows inward series and number of radials it supports. it comes to surround the nerves that pass Among the most primitive living and fossil from the myotomes outwards to supply the actinopterygians (see Nielsen, 1942, 1949) fin. But exactly the same developmental his- varying degrees of reduction are already ap- tory would be expected even if a girdle were parent (fig. 37A, B). Since Polypterus has a reduced, combined with the metapterygium simple triangular girdle along which two of the fin endoskeleton, or replaced entirely rows of radials are arranged in parallel (an by the metapterygium. In the latter case, the 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 215

B \

FIG. 36. Left pelvic girdle and fin endoskeleton in A, Polyodon spathula (Walbaum), juvenile, 12 cm. total length (AMNH 40803); and B, Acipenser oxyrhynchus Mitchill, juvenile, 11 cm. total length (AMNH 37296). metapterygium, in consolidating and enlarg- appear to lie on both sides of the body wall ing to form a support for the preaxial radials, (figs. 28B, 36B), and that, when the external would also come to surround one or more part of the metapterygium is finally lost al- diazonal nerves. together in more advanced actinopterygians The alternative explanation, that the par- (see Lehman, 1966), all that is left is a series allel series of radials lying outside the body of parallel radials such as those found in Po- wall is a remnant of the metapterygium (i.e., lypterus. We conclude, therefore, that the that they are not preaxial radials at all, but pelvic fin radials of Polypterus are of a de- metapterygial segments that have assumed rived, typically actinopterygian type and that the form of such radials) seems to us con- its girdle, which lies entirely between the trived, and is rejected by the observation body wall and the myotomes and lacks the that in sturgeons the external metapterygial dorsal processes that are primitively present axis is continuous with internal segmented in chondrichthyans (e.g., Cladoselache and cartilage so that the metapterygial segments chimaeras, fig. 27) and sarcopterygians, but 216 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

FIG. 37. Pelvic girdle and fin endoskeleton in A, Boreosomus piveteaui Nielsen; find B, Pteronisculus gunnari (Nielsen), both from Nielsen (1942).

absent in chondrosteans, is also of actinop- sarcopterygian groups and also to distinguish terygian type. one group from another (Jarvik, 1944b, 1948, 1972; Andrews and Westoll, 1970a, 1970b; (B) PECTORAL GIRDLE Andrews 1973). In osteolepiforms (the Rhizodontidae sen- Characters of the dermal bones of the pec- su Andrews and Westoll are very poorly toral girdle have been used by several work- known arnd will be dealt with separately), ers to support ideas of relationships between porolepiforms and primitive lungfishes such as Chirodipterus the girdle consists of five paired dermal bones arranged in an overlap- ping series (fig. 39C-G). A median interclav- icle is present ventrally. Primitive actinop- terygians show a series of four paired bones plus a median interclavicle (fig. 39A, B; Jes- sen, 1972). We may assume therefore that the primitive osteichthyan had a girdle which consisted of, at least, a post-temporal, su- pracleithrum, cleithrum, and inter- clavicle and that successive elements in the series overlapped one another from dorsal to ventral. The main lateral line runs through FIG. 38. Polypterus ornatipinnis Boulenger, the post-temporal and descends through the left pelvic girdle and fin endoskeleton. AMNH supracleithrum before running through the 40802, 14 cm. standard length. body scales. Polypterus is unusual in that the 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 217

4-4- '9 ~~~~~~4- EXTRACLEITHRUM E FCH -CLAVICL~E -_ FIG. 39. Diagrams of dermal shoulder girdles (interclavicle omitted) of osteichthyans to show de- velopment of anocleithrum (actinopterygian postcleithrum) in sarcopterygians and transformation series in the increasing importance of clavicle and dorsal migration offin insertion (arrow). Stipple distinguishes ornamented postcleithrum and anocleithrum. A, Mimia toombsi Gardiner and Bartram (original, BMNH specimens); B, Polypterus Saint-Hilaire (from Jarvik, 1944b); C, Eusthenopteronfoordi Whiteaves (from Jarvik, 1944b); D, Rhizodus hibberti (Agassiz) (from Andrews and Westoll, 1970b); E, Holop- tychius sp. (from Jarvik, 1972); F, Latimeria chalumnae Smith (from Millot and Anthony, 1958); G, Chirodipterus australis Miles (original, BMNH P. 52560, P. 52586); H, Neoceratodus forsteri (Krefft) (from Wiedersheim, 1892). sensory canal does not penetrate the supra- of the actinopterygians in the incorporation cleithrum but runs directly from the post- of the anocleithrum which lies between and temporal to the first lateral line scale. separates the cleithrum and supracleithrum The skull-girdle connection is provided by (where present) and is overlapped by both of the post-temporal which abuts against the these elements. The overlap relationships of hind wall of the braincase and is partially the anocleithrum suggest that it is not a part overlain by the extrascapular series. In Re- of the primitive series in the osteichthyan cent actinopterygians and in Polypterus girdle. The homologous bone in actinopte- there is a pronounced anteroventral process rygians is the dorsalmost postcleithrum on the post-temporal which provides a firm which lies slightly behind the supracleithrum union with the braincase. This process is ab- and cleithrum, which are in contact, but is sent from the post-temporal of Mimia and nevertheless overlapped by both of these Moythomasia and therefore it is considered bones (fig. 39A, B; Jessen, 1972). Although as a synapomorphy of all actinopterygians the postcleithrum may develop a ventral pro- except those most primitive genera. cess in some neopterygians it remains fun- The sarcopterygian girdle differs from that damentally scalelike, a condition well dis- 218 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 played in the paleoniscid Mimia where a group, where the posterior lateral lies wholly typical scale peg-and-socket is still present. behind the supracleithrum and cleithrum The anocleithrum of Eusthenopteron (Jar- equivalents, only reinforces the view that the vik, 1944b; fig. 3) shows a small ornamented actinopterygian postcleithrum exemplifies area which lay superficially but most of the the primitive character state. bone is represented as overlap surfaces. The It does, in fact, seem more reasonable to anocleithrum of porolepiforms is very poorly assume that the superficial, scalelike post- known but there are indications (Jarvik, cleithrum represents the primitive condition 1972, pl. 18, fig. 2) that in Glyptolepis it is and that the anocleithrum was developed by large and platelike and deeply embedded, a forward migration and sinking inward of bearing no ornament. In fact, it looks very the postcleithrum. The stages in this hypoth- similar to the anocleithrum of Chirodipterus esis would be represented by Polypterus- (fig. 39G), both showing anterior processes. Eusthenopteron-coelacanths, lungfishes, and In Recent lungfishes and in coelacanths the possibly porolepiforms. A similar hypothesis anocleithrum is the most dorsal functional can be suggested for another bone of the element in the girdle. It varies in shape from shoulder girdle, the interclavicle, which has an ovoid plate in Neoceratodus (fig. 39H), to also apparently sunk inward to become part splintlike in Latimeria (fig. 39F) and Protop- of the teleostean urohyal (Patterson, 1977). terus (Wiedersheim, 1892), while it is absent The coelacanth interclavicle has also sunk in Lepidosiren (Bridge, 1897). inwards to come to lie between the urohyal The incorporation of the anocleithrum/ and the . postcleithrum as a functional unit in the gir- Recent lungfishes have lost both the su- dle is a feature common to osteolepiforms, pracleithrum and the post-temporal and the coelacanths, porolepiforms and lungfishes, former element is absent from coelacanths and this stands in contrast to the scalelike (the anocleithrum is often mislabelled as the postcleithrum in actinopterygians. Jarvik supracleithrum). Furthermore, the anocleith- (1944b, p. 27) took the view that, since the rum lies deep within the epaxial musculature incorporation of an anocleithrum occurs in and this means that like tetrapods, which rhipidistians, then "it must be assumed to have lost the anocleithrum as well, there is have undergone a regressive development in no bony connection between the skull and Actinopterygians and Brachiopterygians." the girdle.5 The absence of such a connection In other words, according to Jarvik the an- cannot be regarded as a synapomorphy of ocleithrum condition is primitive for os- these three groups since at least Chirodip- teichthyans. We would argue that the con- terus retains a skull connection by way of a verse is more plausible; that is, the post-temporal and a supracleithrum (fig. 39G). anocleithrum condition is derived and can be considered as a synapomorphy for sarcop- I Watson (1926) maintained that a skull-girdle con- terygians, later lost in tetrapods. Which of nection exists in primitive tetrapods such as Pholider- the two character states is derived cannot be peton and Eogyrinus attheyi; the former was said to decided simply on outgroup comparison, possess a post-temporal and the latter a complete series since the only other group with a well-de- of connecting bones. However, Romer (1957, p. 112) veloped dermal shoulder girdle is the Placo- reinterpreted the post-temporal as a cleithrum of Phol- dermi. Despite attempts of Stensio (1959) to iderpeton and suggested that the supposed specimen list detailed homologies between the bones (DMSW.34) of E. attheyi is in fact a fish. The specimen of the trunk shield and the bones of the has been subsequently reinterpreted as a rhizodont, pos- osteichthyan girdle the differences seem too sibly Strepsodus, by Andrews (1973). At the very best, therefore, the evidence of a skull-girdle connection in great to allow any but the most tenuous com- tetrapods by way of a post-temporal, supracleithrum, parison. But even if we allow Stensio's sug- and anocleithrum is equivocal. Specialized secondary gestion that the placoderm posterior lateral connection between the dorsal tip of the cleithrum and is equivalent to the postcleithrum/anocleith- the "tabular" horn in some nectrideans has been re- rum, then the condition in the primitive out- ported (Angela Milner, personal commun.). 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 219

The absence of a skull-girdle connection re- (Schmalhausen, 1917). These figures dem- mains a similarity between coelacanths and onstrate that there is a progressive increase tetrapods but in the light of other synapo- in the size of the clavicle. On this basis, the morphies between lungfishes and tetrapods traditional position ofEusthenopteron as the this must be assumed to have developed in most tetrapod-like fish cannot be supported. parallel, in these two groups and in Recent The position of the fin insertion follows a lungfishes. The skull-girdle connection is also similar sequence, showing a dorsal migration absent in some teleosts such as . of the point of fin articulation through the The cleithrum and clavicle, the dominant series primitive actinopterygian-Eusthenop- elements in the primitive dermal shoulder teron-porolepiforms/lungfishes/coelacanths. girdle, always have a complex interlocking It is difficult to quantify the relative positions suture between them. In primitive members of fin insertion in fossil material, particularly of the , , and for round-bodied sarcopterygians in which the clavicle bears an ascend- preservation nearly always distorts the true ing process which fits along the inner face of position. the cleithrum. In coelacanths and lungfishes The rhizodontiforms (Sauripterus, Rhizo- the dorsal part of the clavicle is not clearly dus, and Strepsodus) are not our immediate demarcated from the main body of the bone concern here but since this group is,recog- but nevertheless it rises to wrap around the nized on distinctive features of the shoulder anterior margin of the cleithrum. It is there- girdle (Andrews and Westoll, 1970b) it is ap- fore misleading to consider that coelacanths propriate to add a note. The surface area ra- lack an ascending process (cf. Jarvik, 1944b, tio between the cleithrum and clavicle is 2.2 p. 27). (Rhizodus); that is a value between Eusthe- Jarvik (1944) sought to distinguish a par- nopteron and Holoptychius/Latimeria. The ticular type of double overlap between the fin insertion is high in Rhizodus, apparently clavicular ascending process and the cleith- comparable with that in Holoptychius, coel- rum in porolepiforms from a simple overlap acanths and lungfishes, but considerably of the clavicle by the cleithrum in osteolep- lower in Sauripterus. So, on the basis of iforms (Eusthenopteron). But Andrews and shoulder girdle morphology, rhizodontiforms Westoll (1970b, p. 407) rightly pointed out could be placed on our cladogram immedi- that Osteolepis shows an essentially porolep- ately above Eusthenopteron. iform pattern and intermediate conditions Our interpretation of the dermal shoulder are seen in Rhizodus. girdle can be summed up as follows. The The relative sizes of the cleithrum and incorporation of the anocleithrum as a func- clavicle differ considerably between primi- tional element in the girdle is a synapo- tive members of osteichthyan groups. A morphy of Eusthenopteron, Latimeria, transformation series of the ratio of exposed Holoptychius, and dipnoans. We assume this surface area of the cleithrum to exposed sur- to have been lost, with the skull-girdle con- face area of the clavicle would read: Polyp- nection, in tetrapods. The ratio of cleithrum terus, 7; Eusthenopteron, 3.7; Mimia, 3.4; to clavicle area follows a similar sequence Holoptychius and Latimeria, 1.5 (the latter except that it is not possible to discriminate combines the area of the extracleithrum with between Latimeria, Holoptychius, and dip- the cleithrum); Chirodipterus, 1.25. The ratio noans. The high fin insertion is a synapo- in a primitive tetrapod such as Archeria is morphy of Latimeria, Holoptychius, dip- approximately 0.9, measured from Romer's noans, and tetrapods (paralleled by many (1957) restoration. The ratio in Neocerato- acanthopterygians and paracanthopterygians dus is 0.93, whereas in adult Protopterus no among fishes). The anocleithrum of ratio can be calculated since there is only coelacanths, Holoptychius and dipnoans has one bone present which represents the result sunk beneath the surface and lost all trace of of ontogenetic fusion between an embryolog- ornament. ically separate clavicle and cleithrum The endoskeletal shoulder girdle sheds lit- 220 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

C LE ITHRUM

FIG. 40. "Screw-shaped" glenoid and scapulocoracoid in Eusthenopteron (A) and a Devonian lung- fish (B). Posterior view of scapulocoracoid of A, Eusthenopteron foordi Whiteaves (modified from Andrews and Westoll, 1970a, fig. 4a, reversed); and B, Chirodipterus australis Miles (BMNH P. 52576). Note especially strong processes, presumably for muscle attachment, developed on scapulocoracoid of Chirodipterus. Andrews and Westoll (1970a) commented on several similarities between scapulocora- coid of Eusthenopteron and that of early tetrapods, including a screw-shaped glenoid. Features of such a glenoid are that it is pear-shaped, with articular surface twisted, so that narrow lateral end faces downward, and broad medial end faces upward. In the lungfish, the glenoid fossa is pear-shaped, as in Eusthenopteron. It shows no transverse twist, and is much less thoroughly ossified than glenoid in large individual of Eusthenopteron figured by Andrews and Westoll (1970a, figs. 4, 6), where a twist is not easy to detect. In living lungfishes, glenoid is a convex knob (figs. 30, 31), not a hollow as it is in tetrapods (cf. Jarvik, 1968a, p. 228). In Gogo lungfishes (the only fossil lungfishes in which the endo- skeletal girdle is accessible) both glenoid and head of humerus are too weakly ossified for us to comment on the form of the joint, though figured specimen suggests that the glenoid was convex. tle light on the interrelationships of the bony lungfish, Chirodipterus (fig. 40). Here the fishes under discussion. Romer (1924) noted scapulocoracoid, which is attached to the the basic similarity in the construction of the cleithrum by three contact areas as in scapulocoracoid in primitive actinopterygi- Eusthenopteron, bears a large pear-shaped ans and primitive tetrapods, with Neocerat- glenoid facet entirely comparable with that odus differing in the possession of an unfen- in Eusthenopteron (Andrews and Westoll, estrated scapulocoracoid (the nerves, blood 1970a, fig. 4a). It seems to us therefore that vessels and muscles pass entirely outside the the same tetrapod-like conditions (Romer, skeleton). Latimeria shows an unfenestrated 1924) seen in Eusthenopteron can be scapulocoracoid similar to that of the Recent matched in Chirodipterus. The shoulder gir- lungfishes. dle of Chirodipterus is of further interest be- The scapulocoracoid of Eusthenopteron cause the scapulocoracoid is relatively much has been favored (e.g., Andrews and Wes- larger than that of Eusthenopteron and be- toll, 1970a) over that of other bony fishes as cause of the development of a broad trans- being of tetrapod type because the glen- verse flange, developed along the anterior oid fossa is said to be screw-shaped as in edge of the cleithrum (fig. 40). The rugose early tetrapods (Miner, 1925). The signifi- nature of the bone surface suggests that a cance of this resemblance is, however, very large muscle sheet, comparable with the brought into question in the light of the struc- deltoides scapularis of tetrapods, inserted at ture of the scapulocoracoid in the primitive this point. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 221

DERMAL BONES OF THE SKULL Attempts to derive a useful nomenclature skull table of the Devonian fossil Elpistos- for the dermal bones of the skull involve tege (now thought to be a panderichthyid some of the most confused, and confusing, fish), Westoll decided that the crossopteryg- episodes in the history of comparative stud- ian frontal was homologous with the tetra- ies of fishes and tetrapods. Today, consensus pod parietal and that both the classical ho- on bone homology is still lacking, as we try mologies and Save-Soderbergh's (1935) to make evident in the following brief re- revision were misleading. He concluded view. (1938, p. 127) that the tetrapod names of The nomenclature of the skull roof and bones have largely been misapplied in fish- cheek bones in osteichthyans has for the es, and suggested that the parietal of cros- most part been derived from that originally sopterygians was homologous with the post- applied to tetrapods, in particular, man. parietal of tetrapods and that the latter bones Thus, the early anatomists confidently ho- diminished in size passing from crossopte- mologized the frontals and parietals of mam- rygian fishes to amphibians and to reptiles. mals with similar bones in bony fishes. These He therefore imagined that there was a cor- homologies were based on positional rela- responding backward movement of the pa- tionships to other bones and underlying in- rietals. ternal structures, as well as embryonic de- Romer (1941, 1945) subsequently champi- velopment. Consequently, over several oned Westoll's theory and today many Brit- decades the terminology of the dermal bones ish and North American texts (Jollie, 1962; of advanced was applied to rep- Moy-Thomas and Miles, 1971) and authors tiles, amphibians, and fishes (with the excep- (Thomson, 1965; Parrington, 1967;. Andrews, tion of the Dipnoi-see later) without anyone 1973) have adopted Westoll's terminology doubting that the bones were homologous and consider the bone previously named as throughout these groups. frontal in crossopterygians to be the parietal Then in 1932 Save-Soderbergh proposed and the parietals to be post-parietals. that the dermal skull roof of fossil crossop- The underlying reason for Westoll's radi- terygians and ichthyostegids had been de- cal change in terminology was his belief rived from a common ancestor with two (1938, p. 128) that "the pineal foramen is ab- pairs of frontals and two pairs of parietals solutely homologous in position in all and so concluded (his fig. 5) that the tetrapod forms," a view with which Romer (1941, fig. parietal and post-parietal should be regarded 1) agreed, stating that in the transition from as frontoparietal and parieto-extrascapular, fish to amphibian to reptile there "has been respectively. He developed this terminology no shift in the pineal eye, but a gradual shift further in his 1935 paper on labyrinthodonts in bone positions." In order to substantiate where he introduced a set of compound his theory Westoll (1943) pointed out that the names for additional, inferred bone fusions, osteolepid frontal not only included the pi- viz., lacrimo-maxillary, supratemporo-inter- neal foramen but also lay above the optic temporal, supraorbito-dermosphenotic, etc. foramen, pituitary fossa, and basipterygeroid By the assumption of different fusion pat- process, all of which he claimed were cov- terns, Save-Sbderbergh was then able to ex- ered by the parietal in early tetrapods. More- plain the varying conditions of the skull roof over, the parietals of osteolepids covered the seen in osteolepid fishes, ichthyostegids, and otic capsule, hyomandibular facet, and the labyrinthodonts. exits of the trigeminal and facial nerves, all A year later Westoll (1936, p. 166) working of which in early tetrapods were covered by on Osteolepis suggested that the frontals and the post-parietals. nasals of this fossil fish included homologues The pineal foramen may occupy different of the tetrapod nasals, frontals, and pari- positions in vertebrates (Jarvik, 1967, fig. 3) etals. Then in 1938, with the discovery of the and even within the osteolepids it may be 222 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 situated between the eyes (Osteolepis; Jar- fig. 12: Trematosaurus; Bystrov, 1935, fig. vik, 1967, fig. 4a) or behind them (Eusthen- 25: Dolichopareias; Watson, 1926, fig. 24b) odon; Jarvik, 1967, fig. 4c). The foramen does the lateral line groove actually connect may lie between the parietals, or between the with that of the supratemporal canal. More- parietals and frontals, or between the fron- over, in the majority of stegocephalians the tals, or even anterior to the latter (Powich- groove for the cephalic division of the lateral thys; Jessen, 1975, fig. 2). In Captorhinus line terminates posteriorly in the supratem- (Parrington, 1967, fig. 3c) the frontals lie poral and only in such forms as Metopo- above the basipterygoid process and the op- saurus (Romer, 1937), Trematosaurus, tic foramen, whereas in the urodele Molge Lyrocephalus, Lonchorhynchus, and (Reynolds, 1913) and in Sphenodon (Siive- Dolichopareias (Bystrov, 1935) does the ca- S6derbergh, 1946) both this foramen and the nal leave a furrow on the anterior limits of hypophysis lie beneath the junction of fron- the tabular. Further, in many temnospondyls tals and parietals. the supratemporal canal is confined to the The frontal/parietal controversy is associ- tabular (Aphaneramma, Lyrocephalus; Siive- ated with the problem of whether or not Soderbergh, 1935, figs. 31, 60, 61), whereas homologues of the tetrapod tabulars and in others, where the furrow is continued onto post-parietals may be recognized in any oth- the post-parietal, it rarely if ever meets its er group of osteichthyans. There are two counterpart medially (Trematosaurus, Ben- points of view. The first owes its inception thosaurus; Save-Soderbergh, 1937, figs. 4, 9). to Watson and Day's 1916 paper in which the In his criticism of Westoll's (1936) theory, authors homologized the crossopterygian Jarvik (1967, p. 186) found it difficult to imag- median extrascapular with the post-parietal ine how the bones of the skull roof could and the lateral extrascapular with the tabular move backward, independently of the un- (their figs. 1, 5). These homologies were ac- derlying neural endocranium, yet in accept- cepted by Goodrich (1919, 1930), Schmal- ing in its place the theory of Watson and Day hausen (1950, 1968) and most Russian, (1916) or its modification proposed by Save-< French, and Swedish workers but with Save- Soderbergh (1936) he demanded the re- Soderbergh (1936, fig. 14) and Jarvik (1967, verse-that a transverse series of bones fig. 13B) later maintaining that the post-pa- moved forward onto the endocranium. rietal in temnospondyls represented a com- In osteichthyans many of the dermal bones posite parieto-extrascapular and the tabular of the skull roof are tightly attached to the a supratemporo-extrascapular. The second underlying neural endocranium by ventral hypothesis (Westoll, 1938) is that the cros- outgrowths ofmembrane bone from their cen- sopterygian parietal is a post-parietal and the ters of growth (descending laminae). The ex- posterior supratemporal a tabular. trascapulars are clearly recognizable as a Schmalhausen (1968, p. 82), in accepting transverse series of scalelike bones variable the theory of Watson and Day (1916), point- in number and arrangement (4 in Polypterus, ed out that the lateral line furrows in fossil 7 in Diplurus, 3 in Osteolepis and Dipterus) amphibians pass unchanged over the same and never intimately connected with the neu- bones as in fishes, whereas Westoll's (1938) rocranium. In many fossil amphibians the hypothesis presumed that the fish extrascap- tabulars and post-parietals show extensive ular series had vanished. Schmalhausen descending laminae (Benthosaurus; Bystrov (1968, p. 83) cited in support the works of and Efremov, 1940, fig. 23). And from this Watson (1926) and Bystrov (1935), in which we must conclude that he believed it had been demonstrated that there is no unambig- the cephalic division of the main lateral line uous way to recognize the homologues of the (infraorbital canal) joined the transverse su- tetrapod post-parietal and tabular in the skull pratemporal canal in the tabular. In none of roof of fishes. the examples chosen by Schmalhausen, Regarding lungfishes, Save-Soderbergh however (viz., Benthosaurus; Bystrov, 1935, (1932, pp. 97-98) commented that, in the 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 223

FIG. 41. Skull roofs. A-C, Actinopterygians: A, Kentuckia deani (Eastman) (from Rayner, 1951); B, Moythomasia nitidus Gross (from Moy-Thomas and Miles, 1971); C, Pteronisculus magna (Nielsen) (from Nielsen, 1942). Cladistian: D, Polypterus sp. (from Schmalhausen, 1968). E-I, Rhipidistians: E, Porolepis brevis Jarvik (from Jarvik, 1972); F, Eusthenopteron foordi Whiteaves (from Jarvik, 1967); G, Eusthenopteron, composite, showing variations in bone patterns (from Jarvik, 1944a); H, composite, showing maximum number of separate bones observed in Osteolepis, Thursius, and Gyroptychius (from Jarvik, 1948); I, Holoptychius sp. (from Jarvik, 1972). Arrows show position of eye.

morphology of the skull roof, Dipterus ap- 1971; Miles, 1975, 1977) have given up at- peared more nearly related to ichthyostegids tempts at dermal bone homology and use in- than were crossopterygians. But most Brit- stead a system of letters and numbers. This ish and American authors (White, 1965; Den- custom stems from Romer (1936) who, find- ison, 1968a, 1968b; Thomson and Campbell, ing it impossible to reconcile the two mu- 224 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

A B C

F G FIG. 42. Skull roofs. A, tetrapod; B, D, E, F, G, dipnoans (all but E from Lehman, 1959); C, actinistian. A, Ichthyostega (from Romer, 1966); B, groenlandica Lehman; C, Diplo- cercides kayseri (von Koenen), after Stensi6, showing condition intermediate between A, B, D-G (this figure) and fig. 41; D, Soederberghia sp.; E, (various sources); F, Nielsenia nordica Lehman; G, Oervigia nordica Lehman. Arrows show position of eye.

tually contradictory systems of Watson and (e.g., bone fusion, relation of bones to pineal Day (1916) and Goodrich (1925, 1930), aban- position or to a part of the cephalic lateral- doned the use of names. He was subscribing line canal system). Seemingly irresolvable to a view put forward by Gregory (1915, p. differences have arisen when anatomists 327) that the primitive dipnoan had a skull have applied the same criterion (the lateral consisting of a mosaic of small and variable line) and yet obtained different results. A fur- plates. Forster-Cooper (1937) formulated a ther difficulty has been that the evidence for similar nomenclature of letters and numbers choosing among the different criteria, or for which was used by Westoll (1949), White applying any one criterion in specific cases, (1965), and Thomson and Campbell (1971). has been of a very indirect, circumstantial Disagreements and uncertainties in the use kind. The last, perhaps the major source of of bone names have several different causes. difficulty, is that the exposition of pattern The most obvious source of disagreement is and process have been inextricably bound application of different criteria of homology together so that assumptions about certain 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 225

QUADRATOJUGAL FIG. 43. Cheek in A, Eusthenopteron foordi Whiteaves (after Jarvik, 1944a, fig. 18); B, Porolepis brevis Jarvik (after Jarvik, 1972, fig. 43); C, Ichthyostega sp. (after Jarvik, 1952, fig. 35); D, Griphognathus whitei Miles (after Miles, 1977, fig. 112). In A and C there are seven bones in the cheek plate, and the same bone names can be applied. This pattern is usually treated as synapomorphous, uniting osteo- lepiforms with tetrapods. In B there are 11 bones in the cheek, and in D 14. Using A as a model, bone names can be applied in B and D only by giving the same name to more than one bone (e.g., squamosals 1-3 in B; preoperculars 1-6 in D). The different patterns in B and D must be autapomorphous or primitive, given that the pattern in A and C is synapomorphous. That synapomorphy may be tested by considering other patterns-see figure 44. processes have had the effect of denying pat- with it. The issue then degenerates into an terns that were readily apparent to the ear- irresolvable argument over process and the liest investigators of the problem. For ex- reality of the observed pattern becomes sec- ample, if we assert that a pattern exists with ondary. Thus, Save-Soderbergh suggested respect to the position of the eye in relation that a pattern exists to unite osteolepids and to the junction between two pairs of large tetrapods if one accepts a specific process of bones on the skull roof, actinopterygians, bone fusion. Or, in the case of Westoll's as- cladistians, and "rhipidistians" form one sertion of pattern based on pineal position, group and primitive dipnoans, and tetra- one must accept a process of bones shifting pods form another (figs. 41, 42). But if backward, or, in the case of Jarvik's theory, we also assert that the bone pattern has their shifting forward. Jarvik's hypothesis of been more or less stable and that in early pattern also involves the assumption of a development the eye can migrate anteriorly process of interaction of the placodes of the or posteriorly, we imply that accepting the cephalic lateral-line system and dermal bone pattern is dependent upon accepting this de- formation such that the latter is invariably velopmental explanation. Our confidence in dependent on the former. There is, of the significance of the pattern then comes course, evidence that, in some instances, into question because we lack information on eyes can assume different relative positions whether one process (shifting of the eye in in the head (), that bones can fuse early development) explains the pattern bet- in ontogeny or fragment in phylogeny (in- ter than another process, say elongation of fraorbital bones of characoid fishes), that the the snout, which carries the roofing bones pineal maintains a stable relationship to cer- 226 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

DERMOSPHENOTIC C S PIRACU LARS

POSTORB ITALT LACRIMAL - OPERCULAR ) LERU A

INF RAORB ITO - MAX ILLA QUADRATOJUG AL SUPRAMAXILLA SUBOPERCULAR

B

k SUBOPERCULARS B0 QUADRATOJUGAL ?PR EO P EERC U LAR FIG. 44. Cheek bones and in coelacanth (A), an onychodont (B), a cladistian (C), and a Devonian actinopterygian (D). A, Rhabdoderma elegans (Newberry) (after Forey, in press); B, Strunius walteri Jessen (after Jessen, 1966, fig. 6); C, Polypterus bichir Saint-Hilaire (after Daget, 1950, fig. 26); D, Cheirolepis trailli Agassiz (after Pearson and Westoll, 1979, fig. 20A). Bone names are those used by authors cited. The pattern in Eusthenopteron and Ichthyostega (fig. 43A, C) is best matched in B (quadratojugal missing) and D (squamosal missing). The quadratojugal (present in C and D) is apparently a primitive osteichthyan feature, lost or not found in onychodonts (Andrews, 1973, p. 146); and the squamosal is a primitive sarcopterygian feature (present in A, B and fig. 43). The features common to the cheek plate of Eusthenopteron and Ichthyostega (fig. 43A, C) seem therefore to be primitive for sarcopterygians; the extra bones in porolepiforms (fig. 43B) and dipnoans (fig. 43D) are autapo- morphous, as is the absence of maxilla and quadratojugal in coelacanths (A). tain bones (some tetrapods), and that the gardless of what causal agents might be in- neuromasts can influence dermal bone on- voked to explain them. Two of these patterns togeny or fail to do so. But the occasions on concern dermal roofing bones, and, one, the which we should call forth these various cephalic lateral-line system. The first is the "6explanations" are unknown, especially in position of the eye in retation to the large Devonian fishes and tetrapods where exper- paired bones on the skull roof which, as men- imental investigation is not possible. As we tioned above, groups dipnoans and tetrapods. see the problem, the patterns suggested by The second is a cluster of five bones in the Save-Soderbergh, Westoll, and Jarvik are occipital region of the skull of dipnoans and not actually observable and can be said to Ichthyostega. And the third is the interrup- exist only if one first assumes the reality and tion of the infraorbital canal near the incur- general applicability of a process of ques- rent (external) nostril in dipnoans and tetra- tionable relevance to the problem at hand. pods (see p. 196). An interpretation of the We therefore eschew such lines of argument dermal cheek bones is suggested in figs. 43, and suggest, instead, that there are three ob- 44. servable patterns that can be recognized re- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 227

PALATE AND JAW SUSPENSION

(A) HISTORICAL was the arrangement of the bones of the pal- ate very similar but also that "the ordinary The earliest comparisons of Recent lung- idea of the autostylism of the Tetrapoda is fishes with Recent amphibians, particularly incorrect in postulating a connection be- urodeles, noted an obvious similarity in the tween the palato-quadrate cartilage and the palates and braincases (Huxley, 1876; Cope, otic region. It is, I think, quite certain that 1885). But the late nineteenth century search there never was such a connection in primi- for amphibian ancestors meant that the many tive forms, except through the dermal bones derived features of dipnoans excluded these of the temporal region." fishes from immediate relationship with tet- That is still the current view (Kesteven, rapods (see Historical Survey) and placed, 1950) and may be summed up as follows: in our view, unwarranted emphasis on the crossopterygians (with or without Polypter- fossils. It was reasoned that lungfishes could us), rather than lungfishes, are related to am- not be ancestral to Recent Amphibia be- phibians because of similarities in palatal cause, among other features, they lack a maxillary arch (Cope, 1885) and the dentition dentition, a closer correspondence with fos- is too highly specialized (Baur, 1896; Greg- sil amphibians6 in jaw suspension and the ory, 1915). Pollard (1891, p. 343) argued that persistence of palatal kineticism, and be- ". .. Presuming the Dipnoi to be in the an- cause it is easy to draw homologies between cestral line of the Urodela such [autostyly the dermal bones of the palate of a crossop- sensu Huxley, 1876] should be the condition terygian and those of both Recent and fossil in young and primitive Urodeles. Such is not amphibians. the case." He believed, like Jarvik (1968a), Yet, Dollo (1896), Kesteven (1950), and that lungfishes were close relatives of holo- Thomson and Campbell (1971) recognized cephalans. Kingsley (1892, p. 679) main- that the tritoral dentition and the absence of tained that "the. autostylic condition [in maxillae, dermopalatines, and ectopterygoids urodeles] arises comparatively late in devel- in Recent lungfishes are simply unique dip- opment, and never attains the completeness noan specializations and are therefore irrele- which a Dipnoan ancestry would require." vant for establishing their relationships. The embryology of lungfishes was not known to either Kingsley or Pollard so this view ex- (B) DERMAL BONES OF PALATE presses the same opinion as that given by We are concerned here with the presence Dollo (1896), that the dipnoans are more au- or absence and the pattern formed by the tostylic than urodeles and cannot therefore paired pterygoid (entopterygoid or endopter- be considered ancestral to Recent amphibi- ygoid), ectopterygoid (transpalatine or trans- ans (see also Schmalhausen, 1968, and pp. verse of tetrapods), dermometapterygoid 233-234 below). (dermal metapterygoid), and the dermopala- Pollard (1892) went on to suggest a scheme tine (palatine of tetrapods) which are asso- of homology between the palatal bones of ciated with the palatoquadrate, and the vo- Polypterus (then considered as a crossopte- mer (prevomer) and the median parasphenoid rygian) and those of a . which are associated with the neurocranium The paleontological arguments for dis- (figs. 45, 46). A variety of names has been missing lungfish as amphibian ancestors used for the bones of the dermal palate; these were taken up by Watson from Baur (1896) and Cope (1892). Watson (1912, p. 10) com- 6 For our purposes in this section we take fossil am- pared the palate ofMegalichthys with that of phibians to include ichthyostegids, lepospondyls, tem- the temnospondylous amphibians Loxomma nospondyls, and anthracosaurs. It is unlikely that they and Eogyrinus. He concluded that not only form a monophyletic group (Gaffney, 1979a). 228 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

ENDOPTERYGOID

A

6 ( QUADRATE D FIG. 45. Palates in ventral view and suggested homology of dermal bones and their relation to palatoquadrate. Actinopterygian: A, Mimia toombsi Gardiner and Bartram (BMNH specimens). Cla- distian: B, Polypterus bichir Saint-Hilaire (BMNH specimens). Rhipidistians: C, Eusthenopteron foordi Whiteaves (from Jarvik, 1954); D, Glyptolepis groenlandica Jarvik (from Jarvik, 1972). Compare fig- ure 46. Code 2 refers to dermopalatine in all figures. are listed in table 2. The dermal palatal bones ilarity in number and arrangement of the are thought to have been derived from con- bones between the palate of Eusthenopteron densations within a uniform lining of denti- and primitive amphibians. In both (figs. 45C, cles within the buccal cavity (Nelson, 1970). 46B, C), the parasphenoid is flanked by a We regard the presence of such tooth-bear- large pterygoid which extends from the pos- ing bones (except the parasphenoid, present terolateral contact with the quadrate to the in placoderms) as a synapomorphy of os- vomer. The dermopalatine and ectoptery- teichthyans. goid are sutured to the lateral edge of the Many authors have stressed the close sim- pterygoid but a considerable portion of the 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 229

1I

2--

C

D (PE

FIG. 46. Palates in ventral view and suggested homology of dermal bones and their relation to palatoquadrate. Dipnoan: A, Griphognathus whitei Miles (from Miles, 1977). Tetrapods: B,Ichthyostega (from Romer, 1966); C, Eogyrinus attheyi Watson (from Panchen, 1972a); D, Benthosuchus sushkini Efremov (from Bystrov and Efremov, 1940); E, Tylototriton verrucosus Riese (from Noble, 1931). Com- pare figure 45 for bone names. latter extends posteriorly beyond the ecto- and ectopterygoid of Eusthenopteron and pterygoid. The dermopalatine and ectopte- primitive amphibians. This difference is of rygoid lie in series with the vomer and are unknown significance since actinopterygians securely attached to the maxilla laterally. have a variable number of bones in this In fossil amphibians there is no possibil- position (below). Also, fossil amphibians ity of movement between the dermopala- have been described with two ectopterygoids tine, ectopterygoid and maxilla. The choana, (Megalocephalus, Watson, 1926) and some or its presumed homologue in Eusthenop- have a single bone occupying the territory teron, is bordered medially by the vomer of the vomer and dermopalatine [Tremato- and dermopalatine. Achoanate Latimeria saurus brauni (Owen)-Watson, 1919; Ba- is very similar except that there are three trachosuchus-Sive-S6derbergh, 1935]. The bones in series in place of the dermopalatine palate of anurans and urodeles (fig. 46E) 230 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 differs from that described above chiefly larity of these bones is only partly disguised in the development of large palatal vacui- by the peculiar teeth on those of Dipnoans, ties and absence of the ectopterygoid. The and this difference in the teeth should not be palatal vacuities are present in more ad- deemed of phylogenetic importance because vanced temnospondyls; they are developed we observe marked differences in the teeth between the pterygoid-dermopalatine and of relatively closely related Elasmobranchs." the parasphenoid, in contrast to those found Since Kesteven wrote that sentence we have in amniotes between the pterygoids and the discovered a great deal about the primitive maxilla-jugal. Palatal vacuities and absence lungfish palate (Denison, 1968a, 1968b, 1974; of the ectopterygoid are both derived fea- Thomson and Campbell, 1971; Miles, 1977). tures within certain tetrapod subgroups. Our current knowledge strengthens the cor- There are two important differences be- rectness of Kesteven's view. The specialized tween the palatal structure of crossopteryg- dentition of the Recent lungfishes has clearly ians (including coelacanths) and tetrapods. arisen within the group (Denison, 1974) and In crossopterygians the pterygoids do not there are several genera (Griphognathus, meet and suture with one another in the mid- Uranolophus, Holodipterus, Fleurantia and line and the parasphenoid ends posteriorly at Conchopoma) in which the palate is covered the ventral fissure, as in primitive actinopte- by a shagreen of tiny teeth. The pterygoids rygians (fig. 45). This contrasts with medially are also primitively long (Miles, 1977, p. 173) united pterygoids and a long parasphenoid in as in crossopterygians and primitive tetra- tetrapods (we are uncertain about the con- pods. Griphognathus whitei Miles shows dition of the parasphenoid in Ichthyostega; long pterygoids which, in some specimens see Jarvik, 1952). (e.g., BMNH P. 56054) are flanked anteriorly In all dipnoans, like tetrapods, the ptery- by two paired bones identified here as vomer goids and vomers meet in the midline and and dermopalatine (figs. 7A, 46A). They are the parasphenoid is long.7 Kesteven (1950) equivalent to dermopalatine 1 and 2 of Miles. noted the close similarity in shape and ar- The vomers meet in the midline anteriorly and rangement of the vomers and pterygoids be- the dermopalatine forms the medial margin of tween Dipterus and Ichthyostega and went the choana. There is a third paired palatal bone

on to say (p. 125) ". . . The essential simi- which lies anterolateral to the vomer (der- mopalatine 3 of Miles; "extra" dermal bone, fig. 7). This could be interpreted as a der- 7 Three Lower Devonian lungfishes have been de- mopalatine, in which case the bone we have scribed as having short ("stalkless") parasphenoids and labeled by that name would be interpreted as this is presumed primitive for dipnoans (Miles, 1977). the Be that as it The precise condition of the parasphenoid in each of ectopterygoid. may, there are these forms is, however, uncertain. For marginal palatal bones and at least the poste- lehmanni Lehman and Westoll (1952, p. 410) stated that rior member of this series is firmly sutured to "The details, however, are not well shown, and a very a maxilla (see p. 181). It is possible that Hol- small posterior expansion may be present." In Dipno- odipterus gogoensis also shows a separate rhynchus sussmilchi there is no separate parasphenoid palatine and ectopterygoid since there are (Thomson and Campbell, 1971) although it is presumed several small tooth plates in this area (Miles, to have fused with the pterygoids (Miles, 1977). The 1977, fig. 72). We do, however, agree with parasphenoid of Uranolophus has a large toothed area, Miles that it is very difficult to restore these behind which is a smooth flange, broken posteriorly plates to their mutual position and they (Denison, 1968b, figs. 8, 9). Here too, there is doubt equally well might represent fragments of the about the posterior limit of the parasphenoid. The cru- maxilla or premaxilla as in cial fact is, not so much whether there is a "stalk" or seen Griphogna- not, but whether the parasphenoid has grown back suf- thus (fig. 7B). ficiently to occlude the ventral cranial fissue, as it has In Griphognathus therefore, we can see in all other dipnoans. The relative position of the ventral the same so-called special similarities be- fissue and parasphenoid is unknown in both D. lehmanni tween the crossopterygian palate and the pal- and Uranolophus. ate of a primitive tetrapod. The lungfishes 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 231

TABLE 2 Synonymy of Bone Names for Dermal Palate (Those in brackets are less commonly used.) Actinopterygians Crossopterygians This paper + and (figs. 45, 46) Polypterus Lungfishes Tetrapods Dermopalatine (2) dermopalatine dermopalatine palatine Ectopterygoid (3) posterior dermopalatine ectopterygoid transpalatine (transverse, ectopterygoid) Pterygoid (4) ectopterygoid entopterygoid pterygoid (entopterygoid) (endopterygoid) Dermometapterygoid (8) dermometapterygoid - (dermal metapterygoid) Endopterygoid (7) endopterygoid Vomer (1) vomer (prevomer) vomer (prevomer) prevomer (vomer) show an additional feature of tetrapods, ploy terminology commonly used for fishes namely the medially united pterygoids. (table 2). In porolepiforms (fig. 45D) the palate, as Actinopterygians and Polypterus show a restored by Jarvik (1972), is similar to that rather different arrangement of dermal bones of osteolepiforms. The vomers do not appar- on the palate: some homologies can be ently meet in the midline, in at least Poro- drawn between these and the palatal bones lepis or Glyptolepis, and this is a reflection discussed above but this group has additional of the broad snout of porolepiforms. Unlike elements. The Gogo paleoniscid Mimia dipnoans and tetrapods the pterygoids do not shows what we believe to be the primitive meet in the midline. condition for the group. There is a paired The vertebrates dealt with so far show a vomer9 followed by two series (fig. 45A); an similar number and arrangement of palatal outer row of dermopalatines and an inner dermal bones to which a consistent system row of endopterygoid followed by one or of names can be applied. The varied names more dermometapterygoids. The dermomet- which have been used resulted from the ap- plication of names coined for actinopterygi- viewed this problem with particular reference to ans or, in the case of the prevomer, from a Ornithorhynchus and showed that the palatine process presumed difference in mammals.8 We em- (prevomer) is ontogenetically part of the premaxilla. They concluded (p. 104) "There is therefore in no mam- 8 In non-mammalian vertebrates the bones which are mal an additional pair of elements between the vomer clearly homologues of fish vomers are often called pre- and the premaxilla, and thus the vomer(s) in all mam- vomers (Watson, 1926; de Beer, 1937) following a rec- mals occupy an homologous position to those of rep- ommendation by Broom (1895). The history of the prob- tiles" [italics theirs]. lem has been outlined by de Beer (1937, pp. 433-435) 9 A special problem surrounds the identification of and Parrington and Westoll (1940). Broom noted that the vomer of Polypterus. Several authors identify the the premaxilla of several primitive mammals developed vomer as that bone here called dermopalatine (2, fig. as two parts; a marginal tooth bearing portion and a 45B). It resembles the vomer of Griphognathus (cf. palatine process. The latter remains separate in Or- Miles, 1977, fig. 81; Allis, 1922, fig. 25) and lies close nithorhynchus (the dumbbell bone) and apparently fuses behind and parallel to the premaxilla. It is here identified very late in ontogeny in several others (e.g., Erinaceus, as a dermopalatine because it is associated with the pal- Tatusia). Since this separate process lay anterior to the ate in disarticulated skulls (a true vomer is always mammalian vomer Broom called this the prevomer. closely attached to the floor of the nasal capsule) and Broom considered this palatine process to be the topo- because in early ontogenetic stages of Polypterus an graphic homologue of the vomer of non-mammalian ver- identifiable vomer is recorded (Holmgren and Stensi6, tebrates. However, Presley and Steel (1978) have re- 1936; Pehrson, 1947). 232 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 apterygoid is a primitive characteristic of phies of Polypterus + actinopterygians for the group Polypterus + actinopterygians. two reasons. First, a dermometapterygoid One or several are found in paleoniscids has been figured in a specimen of Eusthen- (Watson, 1925; Nielsen, 1942, 1949). Polyp- opteron (BMNH P. 6807, Woodward, 1922). terus is the only living fish with a (single) We have examined this specimen and remain dermometapterygoid. undecided whether or not there is a separate The endopterygoid lies anterior to the der- bone. Second, if we assume that the group mometapterygoid and traditionally is the Polypterus + actinopterygians are the most bone considered homologous with the pter- primitive osteichthyans and if we also as- ygoid (sometimes called entopterygoid) of sume that the dermal skeleton has taken an rhipidistians and coelacanths, lungfishes, assimilative route (Nelson, 1970) then the and tetrapods. However, there is reason to extra bones in primitive actinopterygians believe that the endopterygoid is unique to could be the primitive osteichthyan condi- Polypterus and actinopterygians and that the tion. If so, the absence of an endopterygoid homologue of the pterygoid is the bone and dermometapterygoid are sarcopterygian called ectopterygoid in actinopterygians and synapomorphies. Polypterus. The ectopterygoid of actinopterygians and Polypterus (bone 4; fig. 45B) is the posterior (C) PALATOQUADRATE member of a series with the dermopalatines to The gnathostome palatoquadrate cartilage which it is related embryologically (Pehrson, primitively ossifies from three centers; an 1922, 1947). It reaches back to the quadrate, anterior autopalatine; a posteroventral quad- borders the adductor fossa and primitively rate at the lower jaw articulation; and a dor- contacts the maxilla anterolaterally. These are sal metapterygoid (epipterygoid of tetrapods) precisely the relationships of the pterygoid at the basal articulation. This primitive con- of Eusthenopteron (Jarvik, 1954), porolepi- dition may be seen in diverse fishes: Acan- forms (Jarvik, 1972), lungfishes (Miles, thodes (Miles, 1973) where these ossifica- 1977), urodeles (Goodrich, 1930), fossil am- tions are perichondral only; juvenile Mimia, phibians (e.g., Panchen, 1964; Beaumont, Moythomasia, some specimens of Pteronis- 1977; Save-Soderbergh, 1935, 1936), and culus (Nielsen, 1942), and Acropholis (Aldin- Latimeria (Millot and Anthony, 1958) al- ger, 1937), many teleosts, and coelacanths. though in the last a maxilla is absent. In the In osteolepiforms and porolepiforms, as, in pterygoid of all these forms and in the actin- large specimens of the actinopterygians opterygian ectopterygoid, the ossification Mimia, Moythomasia, and Pteronisculus, the center lies near the ventral (lateral) margin. In palatoquadrate is ossified throughout as a sin- contrast, the ossification center of the endop- gle bone. Our knowledge of the growth stages terygoid lies near the dorsal or medial margin. of the actinopterygians mentioned and the The pterygoid of sarcopterygians is very condition in coelacanths suggests that the large and reaches along the medial edge of palatoquadrate of rhipidistians also ossified the more anterior members of the series from three centers representing sites of artic- where it may overlap the ectopterygoid ulation between the palate and the braincase (coelacanths) or the ectopterygoid and the and lower jaw. dermopalatine (Eusthenopteron, porolepi- Reduction or loss of ossification centers is forms, Ichthyostega and many temnospon- therefore regarded as a derived condition. dyls, and anthracosaurs; similar overlaps are Thus, among actinopterygians, Lepisosteus shown by the ectopterygoid of Polypterus and some teleosts have lost the autopalatine and actinopterygians such as Lepisosteus, (Patterson, 1973) and only the quadrate Amia, and Saurichthys). seems to have survived in Birgeria (Nielsen, It is difficult to decide if the endopterygoid 1949). Polypterus has lost the metapterygoid and\ dermometapterygoid are synapomor- (Allis, 1922). 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 233

The palatoquadrate of Neoceratodus con- effected by three processes: basal,10 ascend- tains only a quadrate ossification, as in frogs ing and otic (defined by de Beer, 1937) and and urodeles, and even this is absent in Lep- the anterior projection of the palatoquadrate idosiren and Polypterus. Fossil lungfishes beneath the neurocranium (pterygoid pro- appear to show more extensive ossification cess) is reduced or absent (Fox, 1965; Bert- of the palatoquadrate in which it is not pos- mar, 1968). The relationships of the nerves sible to recognize centers of ossification and blood vessels to these processes and to (Miles, 1977, p. 24) but, since an ascending the extracranial cavities which they enclose process is well ossified in many of these lung- (cavum epiptericum and cranioquadrate pas- fishes, we assume that at least a metaptery- sage) in the two groups and the form and size goid center was present. Temnospondyls of these processes are very similar (Good- (Bystrov and Efremov, 1940), anthracosaurs rich, 1930; de Beer, 1937; Fox, 1965, figs. 21, (Panchen, 1964, 1970), , lizards, and 34, 38). The otic and ascending processes are Sphenodon have both a quadrate and a met- fused to the neurocranium; the basal process apterygoid (epipterygoid). is fused in dipnoans and some urodeles, but The picture which emerges is that choa- not in anurans and apodans. The hyoman- nates (lungfishes and tetrapods) have the pal- dibular is reduced in comparison with the atoquadrate foreshortened anteriorly so that first epibranchial and takes no part in jaw it no longer contacts the ethmoid region. suspension. This is associated with the loss of an auto- The autostyly of holocephalans is differ- palatine. Loss of an autopalatine is also seen ent, a fact which led Gregory (1904) to pro- in some actinopterygians but it is never as- pose the term holostyly for these fishes. The sociated with a foreshortening of the pala- palatoquadrate is attached throughout its toquadrate. length from the nasal capsule to the occiput, (D) RELATION there is no ascending process and the iden- BETWEEN PALATE tification of the otic process is questionable AND BRAINCASE (Goodrich, 1930, fig. 444). The hyomandib- The similarity between lungfishes and am- ular takes no part in jaw suspension but is phibians in the relation between the palate unreduced. With Gregory we agree there is and the braincase was pointed out long ago little ground for comparison between the au- by Huxley (1876) and Cope (1885), and more tostyly of holocephalans and dipnoans + tet- recently by Kesteven (1950) and Fox (1965). rapods and accept that fusion of the palate This type of jaw suspension was designated with the braincase has occurred convergent- autostylic by Huxley (1876, p. 40) who noted ly. that autostyly also occurred in holocepha- Consideration of the palate/braincase re- lans and Recent agnathans. These authors lationship in primitive amphibians and in regarded the union of palate with braincase crossopterygians has profoundly favored as evidence of relationship between lung- crossopterygians over lungfishes as tetrapod fishes and amphibians. Other authors have relatives since it is believed that the primitive regarded autostyly as evidence of a lung- amphibian palate is kinetic and that this fish-holocephalan relationship (Jarvik, 1968a; could only have been inherited from a cros- Edgeworth, 1935). Jarvik's view can be sopterygian ancestor. It is assumed that the falsified by several other characters which palate of primitive amphibians is movable on suggest that elasmobranchs and holocepha- lans are sister groups (Schaeffer and Wil- 10 There has been considerable argument (see de liams, 1977). The "autostyly" of holoceph- Beer, 1937, pp. 207, 214; Jarvik, 1954) as to whether alans differs from that of there is a true basal process in anurans (except Asca- considerably phus), since the relation of the palatine nerve to the lungfishes which, in turn, is very similar to process is rather different to that in dipnoans and uro- that of amphibians. deles. This is of little significance to the arguments here The contact between the palate and the since, if one wishes to argue for non-homology this braincase of dipnoans and lissamphibians is would simply be an autapomorphy of anurans. 234 BULL,ETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 the braincase (e.g., Watson, 1912; Panchen, bone-bending in the longitudinal plane. 1970; Beaumont, 1977). The palate is cer- These remarks are not intended to be face- tainly movable in Latimeria (Thomson, 1966) tious but there does seem to be a case for and this was probably true of Eusthenopte- questioning the existence of palatal kineti- ron and porolepiforms. In fact, in Eusthe- cism in primitive amphibians. Perhaps there nopteron, there is a very complex and, in our was some movement but the alleged remnant view, movable joint between the dermopal- of kineticism was not of the crossopterygian atine and the vomer (fig. 14). The palate and kind. ectopterygoid are keyed to the maxilla and To sum up this section on the palate and must have moved with it. The condition of jaw suspension we propose the following our Eusthenopteron specimen suggests that outline of character phylogeny. The primi- the maxilla, palate, and the circumorbital tive gnathostome condition is an amphistylic bones moved as a unit against the snout: jaw suspension with a fully developed sus- there are distinct sliding joints between the pensory hyomandibula and a palate which lacrimal and the supraorbito-tectal/lateral articulates with the braincase at the basal rostral. articulation and rises dorsally from this point A kinetic palate in primitive amphibians is as a cleaver-shaped process (Schaeffer, more difficult to believe. Anteriorly the skull 1975). The palatoquadrate of either side is very solid, with dermopalatine, ectopter- curves inward anteriorly and meets its anti- ygoid, vomers, and pterygoids firmly keyed mere beneath the nasal capsules. The pala- to one another and to the cheek bones, and toquadrate ossifies from three centers and is the cheek bones are keyed to the skull roof- covered with a shagreen of teeth; a vomer is ing bones, at least to those lying anterior to absent. These conditions are met in Acan- the pineal. There is no evidence of sliding thodes and in primitive sharks, except that joints. So, with the palate firmly fixed ante- sharks have an unossified palatoquadrate riorly the kineticism must be located poste- (derived sharks show hyostyly and rays eu- riorly, at the basal articulation, or at the hyostyly; holocephalans are holostylic). point where the ascending process contacts Synapomorphies of osteichthyans are: pala- the braincase. Ichthyostega may or may not toquadrate of either side not meeting in the retain the intracranial joint but kinesis at this midline anteriorly (chondrosteans parallel point is apparently impossible because of more plesiomorph groups in this respect) but ''a remarkably precocious consolidation of articulating with the postnasal wall; paired the skull roof' (Panchen, 1972b, p. 78). The vomer, dermopalatine, ectopterygoid and comments of two amphibian workers are pterygoid tooth plates present (actinopte- pertinent at this point. Panchen (1970), writ- rygians + Polypterus show additional tooth ing about anthracosaurs, denies the exis- plates and methyostyly is developed within tence of kineticism in temnospondyls (prim- the group). Sarcopterygians show an articu- itive relatives of lissamphibians, Gaffney, lation between the metapterygoid (epiptery- 1979b). Certainly the members of this group goid) and the dorsal portion of the orbitotem- generally accepted as advanced, the stereo- poral region of the braincase, hyomandibular spondyls, have the palate firmly fixed at the straddling the jugular canal, pterygoid en- basal articulation (Siive-Soderbergh, 1935, larged and forming the chief bone of the pal- 1936). But according to Beaumont (1977) the ate. Choanates show a reduced hyomandib- more primitive loxommatids have palatal ki- ula playing no direct part in jaw suspension, neticism which to us requires bone-bending. absence of an autopalatine associated with The bending would take place in the trans- an anterior reduction of the palatoquadrate, verse plane. In anthracosaurs (Panchen, an immobile dermopalatine-snout contact, 1970) a gap between the squamosal and the pterygoids meeting in the midline, and au- supratemporal is thought to enable the palate tostyly (but not necessarily involving fusion to move up and down but since the palate is of bone as opposed to cartilage fusion). fixed anteriorly this may mean no more than 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 235

HYOID AND GILL ARCHES In comparing the gill arches of the main gobranchials extending forward as in bony groups of gnathostomes we have accepted, fishes, whereas Nelson (1968), by studying in part, Nelson's (1968), rather than Miles's the details of paired articular surfaces on the (1964, 1965) and Moy-Thomas and Miles's epibranchials in relation to corresponding (1971), interpretation of the branchial skele- points on the pharyngobranchials, concluded ton ofAcanthodes, and we have accepted Jar- that the pharyngobranchials are directed vik's (1954) reconstruction of the parts of the posteriorly as in chondrichthyans. From Eusthenopteron skeleton preserved in his these, and other observations of the hyoid specimens (viz., the first three arches and arch, Nelson (1969) concluded that "the vis- part of the fourth). Nelson's and Miles's in- ceral apparatus of acanthodians appears to terpretations of the Acanthodes visceral be more primitively organized than that of skeleton are out of phase by one element, any other gnathostome group." namely whether the gill arches are supported If Nelson's interpretation of acanthodian ventrally on a copula with two ossified ba- gill arch structure is correct, then Acan- sibranchials between arches one and two and thodes resembles chondrichthyans in three two and three (Nelson's inference, fig. 47A) important respects: the ventral articulation or whether the basibranchials identified by of the first two gill arches fore and aft of a Nelson are really hypobranchials of the sec- small basibranchial element, the absence of ond and third arches, and the hypobranchials a suprapharyngobranchial element articulat- and ceratobranchials of Nelson are really the ing with the basicranium, and the posterior two parts of a segmented ceratobranchial orientation of the pharyngobranchials (fig. (Moy-Thomas and Miles's inference). We fa- 47). Further, if these are primitive conditions vor Nelson's view because (1) there is no of gnathostome gill arches, then the Osteich- particular reason to postulate an unusual thyes are defined by three derived condi- condition of a bipartite ceratobranchial (see, tions: the ventral articulation of at least the below, comments on Miles's postulate of a first two gill arches on a single basibranchial, bipartite epibranchial); (2) the structures the presence of a suprapharyngobranchial identified as basibranchials by Nelson are articulation with the basicranium, and the small elements of about equal length oriented forward orientation of the infrapharyngo- at an angle to the gill arch as one might ex- branchials. These three conditions are pect them to be ifthey were median elements; illustrated (fig. 48) in the sarcopterygian and (3) in a photograph published by Miles Eusthenopteron (in which the first (1973, plate 7), the plainly visible gill rakers suprapharyngobranchial apparently arises on the inner side of the arches are present all from the proximal end of the first infrapha- along what Nelson identified as hypobran- ryngobranchial), the paleoniscid Pteroniscu- chials and ceratobranchials but are absent on lus (in which there is a suprapharyngobran- the small ventral elements, as would be true chial on the first and second epibranchials, of ceratobranchials, hypobranchials, and ba- as in Recent chondrosteans), and the Recent sibranchials in other fishes. Nelson (1968) cladistian Polypterus (in which the only su- commented on the dorsal gill arches ofAcan- prapharyngobranchial present appears to thodes: "Watson and Miles apparently erred have fused with the first epibranchial). Nel- in showing the epibranchial divided in two son (1969) suggested that a condition not un- pieces, and in terming one of these either a like that of Eusthenopteron may be present pharyngobranchial (Watson) or an infrapha- on the first arch of Latimeria in which the ryngobranchial (Miles) . . . the suprapha- pharyngobranchial is a slender, dorsally di- ryngobranchial of Miles appears to be part rected rod (fig. 49A); it suggests the inter- ... of the same pharyngobranchial figured pretation that a compound supra-infrapha- by Reis and possibly also by Dean." In addi- ryngobranchial has been reduced distally, tion, Miles (1964, 1965) figured the pharyn- leaving only the erect dorsal process proxi- 236 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

FIG. 47. Gill arches, dorsal view. A, Acanthodes bronni Agassiz (from Nelson, 1968; compilation of figs. 3 and 4a; fifth arch not shown). B, Squalus acanthias Linnaeus (AMNH 40804); C, Hydrolagus colliei (Lay and Bennett) (AMNH 40805). Dorsal gill arches unshaded. mally. The suprapharyngobranchial of the elements of one side extending forward and second arch in Latimeria, like that in Eus- downward more or less parallel with each thenopteron and Pteronisculus, is articulat- other. In Polypterus the arches also run par- ed, not fused, with the epibranchial. allel and articulate against a copula, but the Eusthenopteron and Pteronisculus and copula, as in Mimia, is undivided into sepa- other actinopterygians share with acantho- rate basibranchials (Sewertzoff, 1927, figures dians and chondrichthyans the apparently a specimen of P. delhezi as having two ba- primitive condition of having the right and sibranchials, but our material of this species, left gill arches coming together in the ventral AMNH 31013, shows only a single basi- midline on a long, narrow copula, the arch branchial); the single basibranchial is also 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 237

FIG. 48. Gill arches, dorsal view. A, Eusthenopteronfoordi Whiteaves (modified fromJarvik, 1954); B, Polypterus ornatipinnis Boulenger (AMNH 40802); C, Pteronisculus stensioei (Nielsen) (modified from Nielsen, 1942). Dorsal gill arches unshaded. somewhat broader than in actinopterygians In Latimeria and Neoceratodus (fig. 49A, but is perhaps similar in outline to what is B) five arches are present, but the copula is known of the basibranchials of Eusthenop- greatly foreshortened (Latimeria) or reduced teron (fig. 48A). In addition, the copula of (Neoceratodus) and hypobranchials are ab- Polypterus may be secondarily foreshor- sent (Latimeria) or absent anteriorly and tened in connection with the loss of a gill greatly reduced posteriorly (Neoceratodus). arch (Nelson, 1969, has argued that an inter- The supposition that Neoceratodus has small mediate arch has been lost and his proposal hypobranchials present on arches three and is supported by the similarity in structure four is suggested by the presence of small, and associated dorsal arch elements on the forwardly directed proximal processes which first arch and the absence of dorsal elements closely resemble similar elements in the hy- on the fourth and last arch with the first and pobranchial position of larval urodeles (see, fifth arches of other fishes). for example, Hynobius, fig. 49C) and by the 238 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

D

FIG. 49. Gill arches, dorsal view. A, Latimeria chalumnae Smith (AMNH 32949); B, Neoceratodus forsteri (Krefft) (AMNH 36982); C, Hynobius leechi Boulenger (AMNH 58792); D, Glyptolepis groen- landica Jarvik (modified from Jarvik, 1972). Dorsal gill arches unshaded.

fact that one of these processes is free from copular region.11 Latimeria, Neoceratodus the ceratobranchial on one side of one spec- and urodeles also show an apparent transi- imen. Latimeria, Neoceratodus, and Hyno- bius have the copular region reduced to a 11 The gill arch bases of Griphognathus, as restored short triangular area with the apex of the tri- by Miles (1977, fig. 137) are not farther apart anteriorly so that the anterior arch than posteriorly, but Miles has restored the bases of the angle posterior, gill first two arches on the posterior end of a specialized bases are noticeably farther apart than those basihyal. If the known arches were restored entirely on of the posterior arches, and consistently the triangular basibranchial (on which they would all have the posteriormost arch articulating di- fit), the arrangement of gill arch bases would be similar rectly with the base of the preceding arch to those of Neoceratodus. We have no good clue as to rather than lying free or articulating with the which is the preferable solution. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 239

tion series in reduction of gill arch elements. In Latimeria the anterior epibranchials are reduced in length relative to posterior ones, the first pharyngobranchial is represented by a slender rod (cf. the first arch in Eusthe- nopteron), the second pharyngobranchial is a simple, round nodule loosely associated with the epibranchial, and the third pharyngo- branchial is either absent or, if present, is represented only by one or two tiny, irreg- ular bits of cartilage. In Neoceratodus the anterior epibranchials are reduced relative to the posterior ones and there are no pharyn- gobranchials. In urodeles there are neither epibranchials nor pharyngobranchials. Eus- thenopteron foordi, if correctly interpreted by Jarvik (1954), might be considered to be part of this transformation series in lacking a posterior pharyngobranchial. This loss may be a unique reductional feature of the species because it is associated with a very small FIG. 50. Hyoid arch in relation to palatoquad- third and no fourth epibranchial in combi- rate. Above, Acanthodes bronni Agassiz (various nation with very well-developed epibranchi- sources). Below, Chlamydoselachus anguineus als and pharyngobranchials similar to those Garman (from Allis, 1923). of actinopterygians on the first and second arches. Or the loss may be apparent and re- flect only poor preservation (see Jarvik, va, 1980). The Devonian dipnoan Gripho- 1954, pp. 22-23). A fifth arch was postulated gnathus is also described (Miles, 1977) as hav- by Jarvik, but is not actually known. Jarvik ing only four gill arches, but, as in modern (1972) has also reconstructed the ventral part lungfishes, the last arch articulates with the of the gill arches of the porolepiform Glyp- base of the preceding arch and there are no tolepis groenlandica, in which, again, ele- hypobranchials. In general, it does seem to ments of only four arches have been found us that there exists reason here, as in a num- (fig. 49D). In the porolepiform Laccogna- ber of other features discussed in other sec- thus, however, there are five (Vorobyeva, tions, to place porolepiforms in a group in- 1980). Jarvik shows the copula as suboval, cluding coelacanths, lungfishes, and greatly foreshortened, and supporting all but tetrapods. On the basis of its primitive vis- the last arch; this reconstruction strongly re- ceral arch anatomy, Eusthenopteron is ex- sembles the condition in Latimeria, except cluded from the above group and is not even that Jarvik has identified hypobranchials in unambiguously assignable to the Sarcopter- Glyptolepis. A further resemblance between ygii. Polypterus likewise cannot be placed Jarvik's reconstruction and the ventral gill within a subgroup of the Osteichthyes on the arch structure in Latimeria, Neoceratodus basis of its visceral arch structure. Living and urodeles is the articulation of the pos- and fossil chondrosteans appear to have the terior arch base with the base of the preced- most primitive osteichthyan gill arches in ing arch rather than with the copular region, showing the greatest general similarity to but unlike those three groups and other os- chondrichthyans and Acanthodes. teichthyans, the posterior arch has a hypo- Reorientation and reduction of dorsal ele- branchial. In Laccognathus the bases of ments is characteristic also of the hyoid arch each of the last two arches (fourth and fifth) of coelacanths, dipnoans, and tetrapods but articulate with the preceding arch (Vorobye- not of Eusthenopteron, actinopterygians, 240 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

FIG. 52. Hyoid arch in relation to palatoquad- FIG. 51. Hyoid arch in relation to palatoquad- rate. Above, Latimeria chalumnae Smith (modi- rate. Above, Pteronisculus stensioei (Nielsen) fied from Millot and Anthony, 1958). Below, Neo- (from Nielsen, 1942). Below, Eusthenopteron ceratodus forsteri (Krefft) (various sources; foordi Whiteaves (from Jarvik, 1972). showing fusion of palatoquadrate with neurocra- nium). chondrichthyans, or acanthodians. In the lates with a rodlike interhyal (stylohyal of au- last four groups (figs. 50, 51) the hyo- thors). The interhyal, which in Eusthenopte- mandibula is long and slender, closely asso- ron and actinopterygians is in series with and ciated with both the ceratohyal and palato- directly articulates between the hyomandibula quadrate, and conforms in orientation with and hyoid bar, is in Latimeria laterally offset the sharp forward slope of the posterodorsal from the distal end of the hyoid bar and at- margin of the latter. In coelacanths, dip- tached to it only by a bed of ligaments. In noans, and tetrapods the hyomandibula and Neoceratodus (fig. 52) the hyomandibula is the posterior part of the palatoquadrate have also a small, flat, triangular plate, but a more nearly vertical or backward orienta- the interhyal is absent and a connection be- tion, and the hyomandibula shows a pro- tween the ventral part of the hyomandibula gressive reduction in size in the series, and and the hyoid bar is made only by an oblique evidence of increasing functional decoupling ligament. But in dipnoans (figs. 52, 53) from the hyoid bar and palatoquadrate. In the hyomandibula is completely reoriented, Latimeria (fig. 52) the hyomandibula is a rela- its ventral end directed anteriorly toward the tively small, flat, subtriangular element that quadrate and, at least in Neoceratodus, its is displaced posteriorly away from the palato- dorsomedial process fused with the capsular quadrate. Ventrally the hyomandibula articu- wall of the otic region of the cranium (de 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 241

FIG. 53. Hyoid arch in relation to palatoquadrate and neurocranium. Above, Griphognathus whitei Miles (from Miles, 1977). Below, hynobiid urodele (modified from Schmalhausen, 1968; Hynobius leechi Boulenger, AMNH 58792, and H. chinensis Gunther, AMNH 30616).

Beer, 1937, p. 412). The hyoid bar is ex- with coelacanths as the sister group of those panded posteriorly and turned sharply up- two. Eusthenopteron appears to have a prim- ward toward the otic region, a structural itive hyoidean and gill arch anatomy similar modification also present to some degree in to that of actinopterygians. Porolepiforms Latimeria. In hynobiids (fig. 53), the hyo- appear to have retained the primitive hyo- mandibula has exactly the same orientation mandibular orientation, as judged by the ar- as in dipnoans except that here the hyoman- rangement of dermal cheek bones, and thus dibula is fused anteroventrally with the quad- lack one of the two classes of visceral arch rate and its dorsomedial process is greatly synapomorphies that unite coelacanths with expanded and occludes an unossified region, dipnoans and tetrapods. Polypterus shows a the fenestra ovalis, in the capsular wall. jaw suspension of intermediate type: the hyo- Again, as in dipnoans, there is a hyoquadrate mandibula is only slightly expanded dorsal- ligament between the sharply upturned distal ly, is strongly oblique dorsally and nearly end of the hyoid bar and the ventral part of vertical ventrally where it lies against the the hyomandibula that fuses with the quad- vertical posterior face of the palatoquadrate, rate (this ligament is all that remains of the but remains intimately associated with both primitive osteichthyan interhyal joint). the hyoid bar and palatoquadrate as in chon- Thus, the anatomy of both the hyoidean drosteans. The cladistic position of Polyp- and branchial arches supports a sister group terus is therefore problematical not only with relationship between dipnoans and tetrapods respect to the branchial apparatus but with 242 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 the hyoid arch as well. Hence, the combined rachotomy with those four, that chondrich- information derived from hyoid and gill arch thyans are the sister group of bony fishes and anatomy specifies that porolepiforms are the that acanthodians, using Nelson's (1969) ar- sister group of coelacanths and dipnoans/tet- gument, are the sister group of other gnatho- rapods, that actinopterygians, cladistians stomes. and Eusthenopteron form an unresolved tet-

RIBS AND VERTEBRAE

(A) RIBS nous anlagen beneath the lateral line, at the One of the earliest reasons given for ex- outer junction of the horizontal and trans- cluding lungfishes from close relationship verse septa. According to this criterion, the with tetrapods was that lungfishes have pleu- ribs of elasmobranchs, previously thought to ral ribs, whereas tetrapods have dorsal ribs be dorsal, are ventral ribs, as are those of (Pollard, 1892; Goodrich, 1909). And the dipnoans and non-teleostean actinopterygi- identification of dorsal ribs, possibly bicipi- ans; the ribs of urodeles, previously taken to tal, in Eusthenopteron (Jarvik, 1952; An- be dorsal, are indeed dorsal; and those of drews and Westoll, 1970a) seemed an impor- apodans and amniotes, previously thought to tant confirmation of the relationship between be dorsal, are ventral. However, in urodeles osteolepiforms and tetrapods. the characteristic centripetal development of Devillers (1954) gave an excellent account dorsal ribs is seen only in the anterior ver- of the interpretation of ribs, on which the tebrae. Passing back down the trunk, the lat- following summary is based. The presence er the ribs appear, the closer their anlagen of two kinds of ribs on each vertebra in the are to the vertebral column, so that the most middle part of the trunk of Polypterus was posterior ribs develop as ventral ribs. Be- the foundation for the recognition of two dis- cause of this, Devillers doubted that Emeli- tinct categories of ribs, dorsal ribs in the hor- anov's criterion was as trenchant as its au- izontal septum, and ventral (pleural) ribs in thor supposed and noted that the anterior the wall of the coelom. Many teleosts also ribs of urodeles are the only exception to the have these two types of ribs, and in addition, universal presence of pleural ribs. epineural ribs in the epaxial musculature. In From this summary, it is clear that iden- other groups of gnathostomes, with only one tification of ribs as dorsal or ventral is no series of ribs, the problem is to find decisive simple matter. Topographic criteria (position criteria for allocating the ribs to one series of the articulation with the vertebra; orien- or another. Topographic criteria fail, be- tation of the rib; relationship to the muscu- cause in actinopterygians and chondrichthy- lature) do not suffice. In fossils, unless more ans the ribs vary in position, from species to than one series of ribs is present, the ribs can species, and from one part of the vertebral therefore be identified as dorsal or ventral column to another in the same fish, "wan- only by a comparative argument (showing dering" within the transverse septum. In tet- that the fossil is a member of a group char- rapods the problem is aggravated by the fact acterized by one type of rib), or by evidence that the horizontal septum is absent in the of mode of growth (centripetal or centrifu- trunk of all except urodeles. gal). Seeking a more reliable criterion, Emeli- The original reason for recognizing two anov (1935, 1936) found that the ventral ribs categories of ribs was the presence of both of Polypterus and teleosts develop centrifu- types in cladistians and teleosts. In Cladistia gally, from cartilaginous anlagen close to the the dorsal ribs are peculiar, borne on hyper- vertebra, whereas the dorsal ribs of those trophied parapophyses and firmly tied dis- fishes develop centripetally, from cartilagi- tally to the lateral line scales, an arrangement 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 243

IOmm

CENTRUM 2 (AXIS) FIG. 54. A, Griophognathus whitei Miles. Sketch of three vertebrae (approx. nos. 9-11) in left lateral view, based on BMNH P. 60420. B, Gyrinophilus porphyriticus (Green), vertebrae. Ribs, black. Unos- sified areas hatched. interpreted by Pearson (in press) as a spe- itive osteichthyans to have ribs of cialization enhancing the flexibility of the chondrichthyan type-short, oriented like trunk. In teleosts, dorsal (epipleural) ribs are dorsal ribs, but developing like ventral ribs. now thought to be a derived feature charac- In our view, this is the simplest interpreta- terizing higher (elopocephalan) teleosts (Pat- tion of the ribs of Eusthenopteron (figs. 55, terson and Rosen, 1977, p. 126) and certain 56). The thickening or anterior process on osteoglossomorphs (Heterotis and notopter- the ribs in the middle part of the trunk of ids; Taverne, 1979, p. 61). The alternative, Eusthenopteron, identified by Andrews and to regard dorsal ribs as a primitive attribute Westoll as a tuberculum, is matched in var- of actinopterygians, would be unparsimo- ious elasmobranchs (e.g., Notorynchus, nious, requiring their independent loss in Daniel, 1928, fig. 51; Somniosus, Dalatias, many lineages. By the same argument, dor- Helbing, 1904, figs. 27, 28), and the absence sal ribs are not a primitive attribute of os- of ribs on the first three to seven vertebrae teichthyans. Instead, we should expect prim- (Andrews and Westoll, 1970a, fig. 22; Jarvik, 244 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

SPACE FOR LONGITUDINAL : ,,° LIGAMENT

A B C

LONGITUDINAL LIGAMENIT HORIZONTAL SEPTUM

IB...... R LATER NERVE ~~~~RIB E FIG. 55. Form of ribs in Eusthenopteron and elasmobranchs. A-C, restorations of three vertebrae of Eusthenopteron foordi Whiteaves. A, fifth in posterior view, after Jarvik (1975, fig. 10); B, about eighteenth in anterior view, after Andrews and Westoll (1970a, fig. 21C); C, about twenty-sixth in anterior view, after Jarvik (1952, fig. 13); D, E, reconstructed thick sections of a mid-trunk vertebra of D, Mustelus laevis Risso, 100 mm.; and E, Torpedo sp., 30 mm., after Emelianov (1935, figs. 49, 52). In D and E muscle is stippled, nerve cord hatched.

1975, fig. 10) also agrees with sharks (fig. neural arches of the Gogo paleoniscoids). 56A, B). Such outgrowths are also present in the dip- An assessment of the primitive condition noan Griphognathus (fig. 54A), and appear of the ribs in tetrapods (dorsal ribs, pleural to be independently ossified. They are absent ribs, or both) requires a reasoned phylogeny in Eusthenopteron (figs. 55A-C, 56B), as are of tetrapods, and, as Devillers (1954) sug- any identifiable diapophyses on the rib-bear- gested, investigation of a more extensive ing vertebrae (Andrews and Westoll, 1970a). suite of urodeles than the two genera studied These diapophysial outgrowths in Griphog- by Emelianov. But one synapomorphy of tet- nathus and primitive actinopterygians are rapods is the bicipital rib, articulating with most completely developed on the foremost a diapophysis on the base of the neural arch, vertebrae; that is, they show a morphoge- and a parapophysis on the centrum. The dia- netic gradient which decreases with distance pophysis of tetrapods recalls the articulation from the occiput. The ribs of Eusthenopte- of the epineurals in teleosts. The separate ron, sharks and primitive actinopterygians epineurals of teleosts are a derived feature (fig. 56C) show a different gradient, which within actinopterygians, where the primitive decreases rostrally and caudally from the condition seems to be an outgrowth or pro- middle of the trunk. The ribs of primitive tet- cess of the neural arch in the diapophysial rapods (fig. 54B) and Recent lungfishes (fig. position (Patterson, 1973, p. 237; these out- 57) show the first type of gradient, decreas- growths are also present on the anterior ing from the occiput. The expanded head of 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 245

DORSAL ROOT SUPRANEURAL NEURAL ARCH VENTRAL ROOT

VENTRAL ARCH A

'TERY

I NT ER VENTRAL

PARASPHENOID\\\I\ FIG. 56. Occiput and anterior vertebrae in left lateral view of A, Notorynchus maculatus Ayres, after Daniel (1928, figs. 47, 51); B, Eusthenopteron foordi Whiteaves, after Jarvik (1975, fig. 10); C, Acipenser oxyrhynchus Mitchill, 150 mm. total length, from AMNH 37296. Cartilage stippled in A and C, bone stippled in B. 246 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

ARC H SPINE

FIG. 57. Neoceratodus forsteri (Krefft). A, anterior vertebrae and ribs, left lateral view; B, skull- vertebral articulation, dorsal view. Small bone in front of first neural arch, in the position of the tetrapod "preatlas," is present on only left side of this specimen, AMNH 36982. the first rib of young Neoceratodus (fig. 57; holocephalans (Callorhynchus), and, within this rib becomes the cranial rib in full-grown the -Osteichthyes, they are represented by animals) is evidence of this gradient in lung- ossified or cartilaginous elements in at least fishes. In urodeles (fig. 54B) modification of a part of the vertebral column: e.g., in actin- the heads of the foremost ribs is more pro- opterygians (Acipenser, fig. 56C; Polyodon, nounced, with the dorsal struts to the dia- fig. 58B; Pteronisculus, Caturus, fig. 59; pophysis decreasing in size away from the Pholidophorus, Patterson, 1968), in actinis- occiput. We suggest that the anterior verte- tians (Latimeria, fig. 58C), in dipnoans (Neo- brae and ribs of Neoceratodus fit the tetra- ceratodus, fig. 58D, E) and in tetrapods (Ar- pod morphocline far better than those of chegosaurus, Chelydosaurus and "Spheno- Eusthenopteron, which seem to be primitive saurus" sternbergii, Fritsch, 1883-1895, in all respects. pp. 25, 28). In many bony fishes and primi- tive tetrapods (loxommatids, temnospon- dyls) however, the ventral, caudal pair of (B) INTERPRETATION OF arcualia (interventrals) are wanting, they GNATHOSTOME VERTEBRAE have either failed to develop or have fused Four pairs of arcualia are present in the with the more anterior basiventrals. Thus, in development of many elasmobranchs, holo- the trunk and abdominal regions of Caturus cephalans, actinopterygians, unborn juvenile (fig. 59A, B), Pholidophorus, Latimeria, and Latimeria, and Neoceratodus. These arcu- the temnospondyls Archegosaurus, Chely- alia are also retained in some adult elasmo- dosaurus, and "Sphenosaurus" there is only branchs (Chlamydoselachus, fig. 58A) and one pair of ventral elements, the ventral ver- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 247

FIG. 58. Vertebrae. A, Chlamydoselachus anguineus Garman, trunk region (from Goodey, 1910); B, Polyodonfolium Lacepede, region of trunk-caudal transition (from Schauinsland, 1906); C, Latimeria chalumnae Smith, caudal region (from Andrews, 1977); D, Neoceratodusforsteri (Krefft), anterior trunk region (from Fiirbringer, 1904); E, N. forsteri, posterior trunk region (from Andrews, 1977).

tebral arch, in each segment. In Neocerato- ured the first 26 vertebrae. In all of these dus (fig. 58D), on the other hand, interven- reconstructions the interdorsal is interpreted trals have only been described in one small as being smaller than that of Ichthyostega portion of the thoracic region (Fiirbringer, (Jarvik, 1952, fig. 14A, B, C). In addition, 1904), whereas in Glyptolepis (Andrews and the occurrence of a notch in the interdorsal Westoll, 1970b, fig. 23), Eusthenopteron for the presumed passage of the ventral (Andrews and Westoll, 1970a, fig. 23) and nerve root in Eusthenopteron (figs. 56B, Osteolepis (Andrews and Westoll, 1970b, 60B) and Ichthyostega (Jarvik, 1952, fig. fig. 5) ossified interventrals are invariably 14C) is matched by a similar notch for the absent. passage of the ventral root in the interdorsal Previously (Jarvik, 1952, 1975; Andrews of Neoceratodus (fig. 58D). and Westoll, 1970a), the vertebral column of The pre- and postzygapophyses of Eus- Eusthenopteron was considered very similar thenopteron (Jarvik, 1952, fig. 13, pr.a, to that of Ichthyostega. For Eusthenopteron pr.p.) are here presumed to be no more than there are four very different reconstructions expansions of the neural spine associated of the vertebral column. The earliest is that with the passage of the dorsal ligament and of Gregory, Rockwell and Evans (1939); Jar- correspond with similar projections in Gri- vik (1952, fig. 13C) figured vertebrae from phognathus (fig. 54A), Latimeria (Andrews, the anterior part of the tail and from the most 1977, fig. 1) and many actinopterygians (Ca- anterior trunk region (Jarvik, 1975, fig. 1OA); turus, fig. 59). Zygapophyses, in the sense of Andrews and Westoll (1970a, figs. 19, 20) fig- processes with articular surfaces that join 248 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

A B

D 5mm FIG. 59. Vertebrae of Caturusfurcatus (Agassiz), as preserved in BMNH P. 20578, anterior to right. A, anterior trunk; B, trunk; C, posterior trunk; D, anterior caudal.

successive neural arches, are found only in occasions (Polypterus, Lepisosteus, Amia, tetrapods. and teleosts), whereas within the tetrapods Complete centra occur within the elas- we conclude that true centra have formed on mobranchs (Chondrenchelys, euselachians), at least two occasions (Recent amphibians holocephalans, actinopterygians, osteolepi- and amniotes). This latter conclusion is forms (Strepsodus, Rhizodus, Andrews and based on the fact that the amniote vertebral Westoll, 1970b, fig. 7), dipnoans (Gripho- column, unlike that of anurans, apodans, and gnathus, fig. 54, Protopterus, Mookejee, urodeles, is often composed of more than Ganguly, and Brahma, 1954), and tetrapods. one part per segment (viz., intercentrum and Within the actinopterygians true (complete) centrum) and is ossified not only perichon- centra have arisen on at least four separate drally, but endochondrally as well. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 249

(C) HETEROSPONDYLY IN tra form in association with the fused basi- CATURUS FURCATUS, AND ventrals and interventrals, but the ossifica- VERTEBRAL HOMOLOGIES tions of either side remain separate as in Glyptolepis, Osteolepis (Andrews and Wes- An almost complete specimen of Caturus toll, 1970b, figs. 4, 20), and Eusthenopteron furcatus (BMNH P. 20578) was acid-pre- (Jarvik, 1952, fig. 13D). Farther back along pared by one of our colleagues, Alan Bar- the trunk the fused ossifications of either side tram. This specimen has a complete and (basiventral + interventral) are joined into a uncrushed vertebral column (fig. 59). hemicentrum by the addition of a median Throughout the column individual ossifica- ventral crescentic calcification in the sheath tion centers comprising perichondral shells of the notochord (chordacentrum). Thus, surrounding endochondral bone are clearly these ventral ossifications resemble the con- recognizable and there are four such pairs dition seen in the trunks of many holosteans per segment from the thirty-ninth vertebra and Pholidophorus. Fusion of the ventral posteriorly. Furthermore, these ossifications elements in the midline has also occurred in may be directly compared with the four pairs the tail region of Eusthenopteron (Jarvik, of cartilages known to be present in Chlam- 1952, fig. 13E) and anterior trunk (Jarvik, ydoselachus and in the developing vertebral 1975, fig. 10) and throughout almost the en- column of Amia. We shall therefore call tire column in temnospondylous amphibians. these ossifications basidorsals, basiven- More posteriorly in the tail region of Ca- trals, interdorsals, and interventrals in ac- turus (vertebra 39 onward) two ventral hemi- cordance with Gadow's (1933) terminology. centra are formed in each segment. An an- Throughout the column the basidorsals are terior chordacentrum links the basiventrals continuous with the bases of the neural of either side while a posterior chordacen- spines and sit between the interdorsals. The trum joins the two interventrals. Thus, the pairs of interdorsals are joined to one middle part of the caudal region of Caturus another by half-rings of calcification of the shows diplospondylous hemicentra as in notochordal sheath as in the caudal region of Pholidophorus, Australosomus (Stensib, Pholidophorus (Patterson, 1968, fig. 3). Sim- 1932, pls. 35-37) and many holosteans and ilar calcifications of the notochordal sheath parallels the condition seen in the entire ver- are seen in the development of Clupea (Ra- tebral column of primitive amniotes. From manujam, 1929, p. 377) and Salmo (Fran- this study we conclude that Schaeffer (1967) gois, 1966, p. 292). Elsewhere calcifications and Thomson and Vaughn (1968) were prob- of the notochordal sheath are seen in some ably correct in homologizing the pleurocen- holocephalans, Chondrenchelys, various tra of crossopterygians with the dorsal inter- elasmobranchs, and Australosomus. calaries of Amia and furthermore that these In the trunk region of Caturus (anterior to dorsal intercalaries may be directly homolo- vertebra 39) the basiventrals and interven- gized with temnospondylous pleurocentra. trals of each side are presumed to have fused together, their point of junction being marked by the foramen for the intersegmen- (D) VERTEBRAE OF RHIPIDISTIANS, tal artery. A similar fusion of basiventral LUNGFISHES, AND TETRAPODS with interventral is deduced to have also oc- curred in the ventral element of Eusthenop- As was the case with the interpretation of teron (Jarvik, 1952, fig. 13E) where a dor- dipnoan ribs, dipnoan vertebrae were also soventral groove marks the passage of the taken to be of an extremely primitive type intersegmental artery. that has no bearing on the question of tet- In the caudal region of Caturus these two rapod affinities. This was partly based on the elements remain separate as in the tail of supposition that dipnoans lack ossified cen- Pholidophorus (Patterson, 1968, fig. 1). In tra and that their neural and hemal arch com- the anterior trunk region partial chordacen- ponents are extremely simple and primitive 250 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 for gnathostomes generally. Schmalhausen larly notched for the dorsal and ventral roots (1968) has written, for example, that the of the spinal nerves. But Jarvik also noted "primitiveness of the Dipnoi is displayed in (p. 36) that "The vertebral column in the ich- the retention of a permanent notochord even thyostegids . . . is on the whole imperfectly in the modern forms" and the "vertebrae preserved in the material available." When consisted only of cartilaginous elements, the one compares the published photographs of bases of the upper and lower arches." This the columns of ichthyostegids and Eusthe- widespread opinion is due, in large part, to nopteron there is, in.our opinion, little spe- the repetition of an old illustration in Good- cific resemblance between the trunk and tail rich (1909) showing several poorly differen- vertebrae of Eusthenopteron (Andrews and tiated vertebral segments from the posterior Westoll, 1970a, pls. 4, 5) and Ichthyostega part of the axis in Neoceratodus, and, in (Jarvik, 1952, figs. 2, 3, 14, 15) except insofar part, from failure to realize that the vertebral as they are all unconsolidated vertebrae with centra, ossified in some Devonian forms ossified neural and hemal arches, bits of such as Griphognathus, are represented in hemicentra, and structures of questionable modern forms by a few barely ossified ring identity called interdorsals: "The interdorsal centra or hemicentra in the posterior part of [of Ichthyostega] is well shown [in two the skeleton. Nevertheless, a study of the specimens] whereas in other specimens first few vertebrae of young Neoceratodus showing the neural arches it generally is im- shows that the column articulates with the perfectly preserved or missing [Jarvik, 1952, skull on two exoccipital processes (fig. 57B) p. 36]." Only one of these structures is iden- that closely resemble the exoccipital con- tified in one specimen and perhaps three or dyles of the bicipital first vertebral articula- four can be made out in the second; they are tion of modern and fossil amphibians (see of varying sizes and shapes, although some- Schmalhausen, 1968, figs. 136, 137). In con- what hemispherical and, to us, their relation- trast, actinistian and rhipidistian fishes, so ship to other structures is unclear. These fea- far as known, have only the primitive, single, tures in the two specimens of Ichthyostega basioccipital contact of the first vertebra that contrast with the positionally similar struc- also characterizes chondrostean actinopte- tures of Eusthenopteron which are smaller, rygians and chondrichthyans. more dorsal in position (behind the neural Until now, however, it has been assumed spine base), and round, nodular or hemi- that the vertebrae of Eusthenopteron and spherical (but with a different orientation tetrapods (as represented by Ichthyostega, from those of Ichthyostega). Hence, both fig. 60A) are significantly alike and that the taxa seem to be similar at most in having similarity signifies close relationship. The as- present in the column three of the four pos- sumption goes back to several statements sible arcual elements, basidorsals and basi- made by Jarvik (1952) in his review of the ventrals (neural and hemal arches) and inter- ichthyostegids: "each vertebra, both in the dorsals, and incomplete ring (Ichthyostega) trunk and in the tail, is composed of three or hemicentra (Eusthenopteron) that are elements, two dorsal and one ventral very primitive for chondrichthyans and osteich- suggestive of those in Eusthenopteron" (p. thyans. The only derived character they 36), and "neural spines [in the sacral region] might share, therefore, is the absence of gradually become narrower and more in- interventrals (see, for example, fig. 56C, of clined backwards and are converted into the Acipenser). comparatively narrow spines, round in sec- Panchen (1977), however, rejected Jar- tion, which in the tail articulate with the ra- vik's comparisons because he believed that dials. The latter strongly resemble the neural the interdorsal of Osteolepis is a better spines in Eusthenopteron" (p. 38). Jarvik match with those of ichthyostegids than that also compared the interdorsal vertebral seg- of Eusthenopteron in being larger. But in ments in Eusthenopteron and ichthyostegids Osteolepis the interdorsals, in being large tri- and concluded that their segments are simi- angular bones, are not only larger and more 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 251

A B

VENTRAL ELEMENT FIG. 60. Trunk vertebrae. A. Ichthyostega sp. (from Jarvik, 1952, fig. 14B, anterior to right); B-D, Eusthenopteron foordi Whiteaves, anterior to left; B, anterior trunk vertebrae, from Jarvik, 1975; C, from Andrews and Westoll, 1970a; D, from Jarvik, 1952.

heavily ossified than those of Eusthenopte- "is generally regarded as [a] primitive ron, they are also larger, better ossified, and [osteolepiform], characterized by heavy os- better differentiated than those of Ichthyo- sification" and is therefore a better ancestor, stega. In other words, Osteolepis has a con- and that the convergent development of ring dition of the interdorsals not present in either centra in some more advanced members of of the other taxa. Interestingly, Panchen's the Osteolepididae and Tetrapoda indicates argument, unlike Jarvik's apparent use of a relationship between them. Perhaps Pan- symplesiomorphous features for formulating chen's line of argument might also be used to relationships, is also an appeal to stratigraph- propose a relationship between those two ic position (Osteolepis is older than Eusthe- assemblages and the halecomorph acti- nopteron and is therefore a better ancestor) nopterygian Caturus furcatus (Agassiz), and convergence (Osteolepis resembles since different parts of the column of Catu- more advanced tetrapods with large, trian- rus show most of the main features of the gular interdorsals). He writes that Osteolepis column of osteolepids and primitive tetra- 252 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 pods and amniotes (fig. 59; Baur, 1896; An- ians, actinistians, dipnoans, and tetrapods drews, 1977; and p. 249 above). Baur (p. 663) (as represented by Ichthyostega) are primi- wrote: "In the Crossopterygii we find unos- tively with unconsolidated cartilaginous or sified and well ossified vertebrae . .. it is ossified arcualia and unossified or poorly os- very probable that some of the crossopteryg- sified centra. We note that the same is true ians showed a condition like Archegosaurus of Acanthodes, Osteolepis, and Eusthenop- and Caturus." And Andrews (p. 284) ob- teron. The Cladistia are the single exception. served: "Fossil chondrosteans and especial- Features that have been used in the past to ly holosteans show very diverse conditions unite groups representing patterns of fusion of the vertebrae, but in some forms at least of arcual elements and their co-ossification (Caturus, Eurycormus) the centra appear to with centra are therefore secondarily derived be composed, in a similar manner to Osteo- for each group (i.e., convergent). We are left lepis, from 'alternating dorsal and ventral in doubt about the structure of the vertebrae crescentic hemicentra' " (Schaeffer, 1967, p. in Ichthyostega, and we reject Panchen's 192). reasoning which uses every undesirable fea- Each of the proposals, by Jarvik and ture of past phylogenetic argumentation in Panchen, reminds us of our methodological paleontology and neontology-symplesio- goal in undertaking this study: to avoid the morphy, stratigraphic position, and conver- loose style of argument that has often char- gence-as evidence ofrelationship. We do not acterized investigations of gnathostome re- rule out the possibility that some unique de- lationships (Nelson's 1969 work is one ex- rived characters will be discovered in better ception), and to approach the problem in an material to unite Ichthyostega with one or explicitly cladistic fashion. To accomplish another "rhipidistian," but our own study this in dealing with the vertebrae we must has revealed only a single derived vertebral first specify the condition of the vertebrae character, a bicipital type of exoccipital joint primitive for each group. In attempting this, with the first vertebra (fig. 57B), that unites we observe that all but one of the recogniz- only dipnoans and tetrapods. able groups, chondrichthyans, actinopteryg-

DERMAL SKELETON (A) COSMINE dians (Poracanthodes), but the architecture Cosmine is a superficial hard tissue con- of the system is so different in cephalaspids, taining a pore-canal system. It is found on acanthodians, and osteichthyans (Gross, the dermal bones of some cephalaspido- 1956) that it is difficult to recognize any ho- morphs, rhipidistians and Paleozoic dip- mologies. We therefore regard the cosmine noans. No living fish has cosmine or a pore- and pore-canal system of dipnoans and rhip- canal system, and the function is unknown idistians as a synapomorphy (Schultze, (Thomson, 1977). The cosmine of cephalas- 1977b, fig. 1, seems now to agree, indicating pids differs from that of osteichthyans in so the pore-canal system as a synapomorphy in many ways (0rvig, 1969a; Schultze, 1969a, his cladogram). The problem is to decide p. 56) that we agree with 0rvig (1969a) that what group is characterized by this synapo- cosmine is not a primitive feature of a group morphy. containing cephalaspids and osteichthyans; In rhipidistians, cosmine occurs in poro- in other words, we do not regard cephalaspid lepiforms (Porolepididae-Porolepis, Heim- and osteichthyan cosmine as homologous. enia-and Powichthys, if Powichthys is a Schultze (1969a) and Thomson (1975, 1977) porolepiform) and in osteolepiforms accept that the pore-canal system (not nec- (Osteolepididae sensu Vorobyeva, 1977a; essarily enclosed in cosmine) is a primitive Rhizodopsis, Andrews and Westoll, 1970b, feature of vertebrates. It is true that a pore- p. 403). In dipnoans cosmine occurs in Ura- canal system also occurs in some acantho- nolophus, Dipnorhynchus and dipterids (sen- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 253 su lato). In onychodonts the only report of tween his interpretation and Jarvik's is a cosmine is a small patch on the premaxilla matter of how Westoll-lines are defined, and (Jessen, 1966, pl. 16, fig. 4) and parietal (Jes- depends on some theory of the ontogeny of sen, 1967, pl. 2B) of Strunius. This tissue cosmine. In any case, such lines have been has not been studied histologically, and reported, and observed by us, only on skull might be only superficially cosmine-like, as bones of Megalichthys. They are unknown are patches of porous enameloid on the head on osteolepiform scales. Evidence of re- of the dipnoan Holodipterus (Smith, 1977, p. sorption of cosmine (as distinct from West- 48, fig. 66). Cosmine has not been reported oll-lines) is found in dipnoans and osteo- in coelacanths, though Westoll (e.g., 1979, lepiforms, but not in cosmine-bearing p. 346) has referred to "modified cosmine" porolepiforms (Qrvig, 1969b). in undescribed Devonian specimens from According to Miles's (1977) phylogeny of Scaumenac Bay. We have examined those dipnoans, cosmine is a primitive feature of specimens, and did not see anything recog- the group, independently lost at least three nizable as cosmine. times. In any alternative dipnoan phylogeny, The pore-canal system of cosmine-bearing it seems that cosmine must either be lost or dipnoans, osteolepiforms, and porolepiforms developed more than once, or dipnoans can- has the same layout (Schultze, 1969a, fig. not be treated as monophyletic. In osteolep- 42). In some osteolepiforms the mesh canals iforms, Vorobyeva' s recent phylogenies linking the flask-shaped pore-canals are di- (1977a, 1977b) require that cosmine be in- vided by a horizontal partition into upper and dependently lost five or six times, or that it lower divisions (Gross, 1956). Schultze develop more than once within the group. In (1969a) and Thomson (1975, 1977) have tak- porolepiforms, it is generally assumed (large- en that condition to be primitive, since it re- ly on stratigraphic grounds) that cosmine is calls the perforated "sieve-plate" dividing primitive, and is lost in holoptychiids. the mesh-canal in cephalaspids. We regard In dipnoans, Smith (1977) has made thor- this as a far-fetched homology, since it de- ough comparisons of the histology of cos- pends on the assumption that the pore-canal mine-bearing (Chirodipterus) and cosmine- system is primitive for vertebrates. Horizon- free (Griphognathus, Holodipterus) dermal tal division of the mesh-canals, which has bones. Of particular interest is her report on not been found in any dipnoan or porolepi- superposed generations of tubercles in Gri- form, might be a synapomorphy of osteolep- phognathus and Holodipterus. Unless cos- iforms, or its absence might be a synapo- mine developed independently on several morphy of dipnoans and porolepiforms. In occasions, these genera have to be regarded some porolepids the enamel extends inward as having secondarily lost cosmine. Yet they to line the pore-canal (Qrvig, 1969b, p. 294). show a pattern of superposed tubercles (gen- This condition is also found in some dip- erations of odontodes; 0rvig, 1969a) resem- noans (Gross, 1956, fig. 69) but not in others bling the tubercles of Latimeria (Smith, 1977, (Smith, 1977, fig. 18). p. 41) and like that in other primitive fishes "Westoll-lines" are a prominent feature of (Schultze, 1977b). We therefore lack any an- the cosmine of dipnoans (Gross, 1956; 0rvig, atomical criterion which will distinguish 1969a). Such naked strips are not recorded "pre-cosmoid" (cosmine assumed never to in porolepiform cosmine, or, according to have been present) and "post-cosmoid" Gross (1956) and Jarvik (1968a) in any osteo- (cosmine lost) dermal bones. Discrimination lepiform. However, Thomson (1975) be- between those two conditions depends on lieves that Westoll-lines occur in some os- the position of the animal in question in a teolepiforms, and illustrates skull bones of phylogeny, and that phylogeny must be es- Megalichthys laticeps. Jarvik (1966, pl. 4, tablished on grounds other than cosmine. We fig. 3) has also illustrated these lines on a also therefore reject arguments that Lati- maxilla of Megalichthys, and said (1968a, p. meria is the primitive sister group of other 227) that their "nature ... is uncertain." osteichthyans because it possesses scale Thomson suggests that the difference be- odontodes. 254 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

To sum up, we believe that cosmine is a rus). Among osteolepiforms with polyplo- synapomorphy of some taxon, but cosmine codont teeth, only Panderichthys has the alone cannot tell us whether animals that same type as primitive tetrapods (Vorobye- lack it (such as coelacanths and tetrapods) va, 1977c, table 1). belong in that taxon or not, because we can- Schultze (1969b, p. 128) summarized the not distinguish pre-cosmoid and post-cos- significance of these facts neatly: either one moid dermal bones. Cosmine is therefore ir- must accept the detailed correspondence in relevant to the monophyly of , tooth structure between Panderichthys and or any other taxon. Resorption of cosmine, Ichthyostega as evidence of close relation- found only in cosmine-bearing dipnoans and ship (or direct ancestry), or one must push osteolepiforms, could be a synapomorphy tetrapod ancestry back to animals with sim- relating those taxa. But if the primitive dip- ple, unfolded teeth, and interpret the folded noan pattern is shown by Uranolophus teeth of tetrapods and osteolepiforms as con- (scales and skull bones with Westoll-lines, vergent. Schultze preferred the first of these superposed generations of tubercles in the alternatives, as does Westoll (1979), who en- scales, mesh-canals without a horizontal par- visaged a direct stratigraphic lineage from tition, Denison, 1968a, 1968b), it does not Panderichthys, through Elpistostege, to match the pattern generally assumed to be Ichthyostega, and writes (p. 347) "Late De- primitive for cosmine-bearing osteolepiforms vonian intermediates are confidently await- (no Westoll-lines, no buried tubercles, mesh- ed." canals with horizontal partition). We have to When Schultze discussed the teeth of Pan- conclude that, as yet, cosmine provides no derichthys, the animal was known only by phylogenetically useful information. isolated fragments-lower jaws, scales and (B) FOLDED TEETH bones of the skull roof and pectoral girdle. Since then, more or less intact individuals Schultze (1969b, 1970a) made a very thor- have been found (Vorobyeva, 1973, 1975, ough survey of the histology of folded teeth 1980). These serve as a test of Schultze's and in fishes and tetrapods. He found that true Westoll's hypothesis, which predicts that plicidentine occurs only in porolepiforms, Panderichthys will be more tetrapod-like osteolepiforms, labyrinthodont amphibians, than other osteolepiforms. In some respects, a few reptiles, and in lepisosteids. He this is true: the head is ichthyostegid-like in showed that these teeth could be separated dorsal view (Vorobyeva, 1980, fig. 1), with into several categories according to the mode "brow-ridges" and a long snout, and the and complexity of folding. All porolepiforms skull roof pattern is somewhat ichthyostegid- (except Powichthys, Vorobyeva, 1977c) like (Vorobyeva, 1973, fig. 1). But in other have dendrodont teeth, with very complex characters Panderichthys does not fit. It has folding and the pulp cavity filled with osteo- two external nostrils (Vorobyeva, 1973), a dentine. The most primitive amphibians (ich- very short pectoral fin skeleton, of only three thyostegids, loxommatids) have polyploco- segments (Vorobyeva, 1975, fig. 3), pelvic dont teeth, with simple branching of the folds fins which are "very small, and set far back in the dentine. Among fishes, polyplocodont close to the caudal fin" (Vorobyeva, 1980, teeth are found only in osteolepiforms, Pow- p. 194), and ring-vertebrae in the thoracic and ichthys and lepisosteids. Within osteolepi- caudal regions. Vorobyeva's conclusions forms, there are genera with simple, unfold- from her work on Panderichthys, and on ed teeth (Osteolepis, Thursius, Latvius), folded teeth (1977c), are that "at present with polyplocodont teeth (Gyroptychius, Panderichthys and Ichthyostega are practi- Megalichthys, Tristichopterus, Eusthenop- cally proved to belong to divergent evolu- teron, Panderichthys, Rhizodopsis, Rhizo- tionary lines" (1977c, p. 17), and that the dus, Strepsodus), and with the more compli- resemblances between the two are conver- cated, eusthenodont teeth (Eusthenodon, gent (1974). Platycephalichthys, Litoptychus, Sauripte- Thus, after studying more complete spec- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 255 imens of Panderichthys, Vorobyeva has dis- and tetrapods do not, therefore, appear to be carded her earlier opinion (1962, fig. 34) that synapomorphous. Nor are polyplocodont Panderichthys is the closest relative of tet- teeth synapomorphous within osteolepi- rapods among fishes, and now concludes that forms, since their distribution (Vorobyeva, tetrapods are the sister group of Osteolepi- 1977c, table 1: Gyroptychiinae, Megistole- didae (1977a, fig. 22). The same conclusion, pididae, Eusthenopteridae, Rhizodopsidae, that tetrapods are best compared with osteo- Rhizodontidae) cuts across every classifica- lepidids rather than more specialized osteo- tion (e.g., Andrews, 1973; Vorobyeva, lepiforms, has been reached by Panchen 1977a). Occurrence of such teeth in lepisos- (1977) on vertebral structure, Rackoff (1980) teids and other actinopterygians such as An- on fin structure, and by Andrews and Thom- arhichas (personal observ.) only emphasizes son (personal commun.). this. The polyplocodont teeth of Panderichthys

SYNAPOMORPHY SCHEME Comparisons are among cladistically ple- al, ceratobranchial, epibranchial siomorphous or the only well-known mem- and pharyngobranchial elements. bers of each group: Acanthodes bronni (for 4. A cephalic lateral-line system that acanthodians), Eusthenopteron foordi (for includes the following canal sec- osteolepiforms), Latimeria chalumnae (for tions: a supraorbital, infraorbital, actinistians), Polypterus (for cladistians), supratemporal, mandibular, su- chondrosteans (for actinopterygians), Hy- pramaxillary (jugal) that joins nobiidae and Ichthyostegidae (for tetrapods), the infraorbital (absent in cladis- Neoceratodus forsteri (for living lungfishes) tians and actinopterygians), and and Griphognathus whitei (for fossil lung- preopercular that rises toward the fishes). Chondrichthyan characters are se- supratemporal canal (connection lected from both sharks and chimaeras, ex- absent in sarcopterygians). cluding from the comparisons apparently 5. Paired pectoral and pelvic append- derived characters of each (pelvic bridge and ages with internal supporting gir- posteriorly directed hypobranchials in sharks; dles. fin endoskeleton in chimaeras; etc.). Placo- 6. Three semicircular canals in the derms are not included; at least one, Pholi- otic capsule. dosteus, has a complex pectoral fin endo- B. Chondrichthyans also have the follow- skeleton without a definite metapterygial ing derived feature which they share axis (part of character B.7.), and another, with osteichthyans: Coccosteus, has a sclerotic ring of four 7. Complex endoskeletal support of plates (character C-I. 13.) (Moy-Thomas and the paired appendages consisting Miles, 1971). of a metapterygial axis and radials articulating with the metapteryg- A. Acanthodes shares with other gnatho- ium and girdle. stomes: C-I. Actinopterygians have the foregoing 1. The presence of a lower jaw sup- characters and share with other os- ported by a palatoquadrate and teichthyans: hyomandibula. 8. Dermal skull bones with internal 2. A hyoid bar connecting the bran- processes (descending laminae) chial apparatus with the hyoman- attaching to endocranium, and vari- dibula. ous toothed dermal bones on the 3. Anterior branchial arches consist- palate (see pp. 222, 227ff). ing of basibranchial, hypobranchi- 9. Endochondral bone. 256 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

10. The presence of lepidotrichia in 24. An exclusively metapterygial pec- addition to ceratotrichia in the toral and pelvic fin (loss of all ra- fins. dials anterior to base of metapter- 11. Radials of the fins never extending ygium) supported by a single basal to the fin margin. element, and an anocleithrum in 12. With two or fewer preaxial radials the shoulder girdle suspension associated with the first two meta- (see pp. 206, 207). pterygial segments of the pectoral 25. Absence of the connection be- fin endoskeleton. tween the preopercular and supra- 13. Dermal sclerotic ring (primitively temporal lateral-line canal. of four plates (see below, section 26. Enamel on teeth (Smith, 1979). D). 27. Sclerotic ring composed of more 14. The presence of marginal upper than four plates (see Miles, 1977, and lower jaw teeth (dental ar- pp. 249-250). cades) on dermal bones lateral to E. Actinistians have the characters of A, palatoquadrate (as contrasted with B, C-I, and D and share with dipnoans teeth fused with, or attached to, and tetrapods: the palatoquadrate and Meckel's 28. Anocleithrum of dermal shoulder cartilage in A and B). girdle sunken beneath dermis and 15. Interhyal (=stylohyal) present. with no trace of surface ornamen- 16. A forward, rather than backward, tation. orientation of the infrapharyngo- 29. Clavicle large relative to cleithrum, branchials. and pectoral appendage insertion 17. Suprapharyngobranchials present high as compared with these con- on first two gill arches. ditions in C-I, C-TI, and D. 18. Gill arches 1 and 2 articulating on 30. Pectoral and pelvic appendages the same basibranchial. each with long, muscular lobes 19. A pneumatic or buoyancy organ as that extend well below ventral an outpocketing of the anterior gut body wall and with structurally (swimbladder, lung). similar endoskeletal supports. C-II. Cladistians (=Brachiopterygii) also 31. Preaxial side of pectoral fin endo- uniquely share with primitive actin- skeleton rotated to postaxial po- opterygians: sition (Braus, 1901; Romer and 20. The absence of the supramaxillary Byrne, 1931). (jugal) canaljoining the infraorbital 32. A series of canal-bearing bones canal. (for the supraorbital canal) lateral 21. A differentiated propterygium in to large, paired, roofing bones be- the pectoral fin. tween orbits. 22. Modification of the pelvic fin me- 33. Absence of hyostyly in the jaw tapterygium to form part or all of suspension (the reduced, subtrian- the pelvic girdle, and the endo- gular hyomandibula and interhyal skeletal support of the pelvic fin decoupled from, and not in direct exclusively by preaxial pelvic ra- sequence with, the hyoid bar, and dials. only loosely connected to the 23. Acrodin caps on all teeth (0rvig, symplectic). 1976). 34. Posterior margin of palatoquad- D. Eusthenopteron has the characters of rate erect or sloping backward A, B and C-I (but character 19 not rather than sharply forward from known) and shares the following de- the jaw articulation. rived features with actinistians, poro- 35. Presence of a rostral organ (a pos- lepiforms, dipnoans and tetrapods: sible synapomorphy with the la- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 257

bial cavity of dipnoans and naso- They agree only with dipnoans in having a lacrimal duct of tetrapods, as long, leaf-shaped pectoral, and in details of discussed above, but recognized the cosmine pore-canal system (in a compar- here as an ambiguous character). ison of Porolepis and Dipterus; Schultze, 36. With only a single, broad, trian- 1969a, fig. 42). At most, one can say that gular basibranchial (cladistians they form an unresolved trichotomy with ac- have a similar basibranchial, per- tinistians and dipnoans + tetrapods. haps associated with the loss of one gill arch; a single elongate ba- F. Dipnoans have all the foregoing char- sibranchial is present also in prim- acters excluding those of C-TI, and itive paleoniscids; Gardiner, 1973). uniquely share with primitive tetra- 37. Last gill arch articulates with base pods: of preceding arch rather than with 42. A choana. basibranchial. 43. A labial cavity (a possible synap- 38. Reduction or loss of hypobranchi- omorphy of the nasolacrimal duct, als. as discussed above, but regarded 39. Reduction or loss of pharyngo- here as an ambiguous character). branchials. 44. Second metapterygial segment of 40. An inferior vena cava (Lauder, paired appendages composed of 1979). paired, subequal elements that are 41. A pulmonary vein. functionally joined distally. 45. Two primary joints in each paired appendage, between endoskeletal Porolepiforms (Porolepis, Glyptolepis, Lac- girdle and unpaired basal element cognathus, and Holoptychius are the best and between basal element and known of these fishes) are very incompletely paired elements of second seg- known but have characters reported in the ment. In pectoral appendage, literature some of which would place these preaxial member of paired ele- fishes in a position plesiomorphous, and oth- ments with a ball and socket joint ers, apomorphous, to actinistians. Nothing with basal element and postaxial is known of their dorsal gill arches and al- member articulating on dorsal most nothing is known of their endoskeletal (postaxial) margin of basal ele- fin supports. They might be regarded as ple- ment. siomorphous to actinistians because of eye 46. Reduction in ratio of dermal fin position in relation to paired dermal roofing rays to endoskeletal supports in bones and because the pelvic appendage is paired appendages. very much shorter than the pectoral, al- 47. Tetrapodous locomotion in living though the scale-covered (and presumably representatives (Dean, 1903; fleshy) fin bases are longer than those of os- Lindsey, 1978; personal observ.). teolepiforms and extend well below the ven- 48. Muscles in paired fins segmented tral body wall. They agree with actinistians, (Braus, 1901, pl. 21; in contrast to dipnoans and primitive tetrapods (e.g., ich- the non-segmented muscles in ac- thyostegids) in having a relatively large clav- tinistians, actinopterygians, cladis- icle and an embedded unornamented ano- tians and chondrichthyans). cleithrum, a high pectoral appendage 49. Fusion of right and left pelvic gir- insertion, an expanded parasphenoid, only a dles to form pubic and ischial pro- single, compact basibranchial, and the last cesses (fig. 61). gill arch (or the last two in Laccognathus; 50. Hyomandibula plays no part in Vorobyeva, 1980) articulating with the base jaw suspension. De Beer (1937) of the preceding arch, not with the basi- described the condition of Neo- branchial where these structures are known. ceratodus, as follows: "The orbi- 258 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

PUBIC CARTILAGE

A

FIG. 61. Pelvic girdle of A, Neoceratodusforsteri (Krefft), 18 cm. total length (AMNH 36982); and B, Necturus maculosus (Rafinesque), (from Gilbert, 1973). A is entirely cartilaginous.

totemporal region is different from geminal nerve (V. and V3), and to that of any other (non-Dipnoan) the orbital artery. All these rela- fish owing to the autostylic attach- tions are typical and practically ment of the quadrate by means of constant throughout Tetrapods." basal, otic, and ascending pro- 51. Loss of interhyal (=stylohyal). cesses. The basal connexion is 52. Hyomandibula reduced and, at ventral to the head vein and dorsal least in extant forms, joined dor- and anterior to the palatine branch sally by ligament and closely as- of the facial nerve. The otic con- sociated with an otic recess in the nexion is lateral to the head vein braincase, and ventrally with and orbital artery, posterior to the quadrate. trigeminal nerve (and ophthalmic 53. Loss of dorsal gill arch elements and buccal branches of the facial (pharyngobranchials). nerves), and anterior to the hyo- 54. Opercular bone, in extant forms, mandibular branch of the facial reduced and joined posteriorly by nerve. The ascending connexion a division of the epaxial muscula- is lateral to a branch of the head ture to the upper half of the pri- vein, postero-lateral to the abdu- mary shoulder girdle. cens nerve and the ophthalmic pro- 55. Right and left pterygoids joined in fundus nerve (V1), but antero-me- midline, excluding the parasphe- dial to the maxillary and noid anteriorly from the roof of mandibular branches of the tri- the mouth. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 259

56. Loss of autopalatine. trabecular (anterior) and parachordal (pos- 57. Elongation of the snout region. terior) cartilaginous primordia. In "rhipidis- 58. Eye somewhat posterior in skull, tian" crossopterygians, the joint may be at level of junction between two kinetic (as in Ectosteorhachis, where a peg- principal bones in skull roof. and-socket are present in the skull roof), aki- 59. Pattern of five dermal bones cov- netic (as in Eusthenopteron, where no hinge ering otic and occipital regions of mechanism has been found), or perhaps even braincase. absent (as in Panderichthys, Elpistostege 60. Resorption of calcified cartilage and Powichthys, where there is no external and subsequent deposition of indication that a joint exists; Jessen, 1975). bony tissue in the vertebrae of the In actinistians there is a well-developed ki- Devonian Griphognathus is iden- netic joint that moves on a ventrally situated tical with the formation of bony groove and track, but dorsally the joint sep- tissue in uncalcified cartilage in arates the cranium more posteriorly than it Tetrapoda, according to Schultze does in rhipidistians (Bjerring, 1973), and the (1970b, p. 329). claim has been made that the actinistian joint 61. There are, of course, numerous arose independently of that in "rhipidis- features of soft anatomy, devel- tians." Gardiner and Bartram (1977; also opment, and physiology in which Gardiner 1973), on the other hand, in a dis- dipnoans uniquely resemble uro- cussion of the primitive subdivision of the deles and other tetrapods. Some actinopterygian cranium along a cartilage- of the more striking of these fea- filled, staggered groove, claimed that the tures are the internal structure of ventral part of this groove (the ventral otic the lung (Johansen, 1970), the fissure) is homologous with the ventral part presence of a pulmonary circula- of the intracranial joint. Both a ventral otic tion, the development of a two- fissure and a lateral occipital fissure are also chambered auricle and the ventral present in Acanthodes so that the actinop- aorta as a truncus, the telolecithal terygian condition might be primitive for development of the jelly-coated gnathostomes. Bjerring's claim that the bipolar egg, ciliation of the larva joints are fundamentally different in actinis- (Whiting and Bone, 1980), pres- tians and rhipidistians is unconvincing be- ence of a glottis and epiglottis, cause in both groups the dorsal part of the flask glands, pituitary structure joint is just anterior to the emergence of the and a tetrapod neurohypophysial fifth and seventh nerves from the endocra- hormone (Perks, 1969; Archer, nium and because, ventrally, thejoint is either Chauvet, and Chauvet 1970), lens at or slightly in front of the end of the no- proteins and bile salts (L0vtrup, tochord. Differences between the two are 1977), and gill arch muscles (Wi- perhaps most easily related to the probably ley, 1979). different degrees of, and certainly radically different mechanisms for, kineticism. The two kinds of joints are either synapomor- Within the framework of this theory of re- phous, thus specifying a sister-group rela- lationships (fig. 62), three features of sarcop- tionship between "rhipidistians" and acti- terygians deserve some comment: the intra- nistians, or some form of a joint is primitive cranial joint, ventral gill arch muscles, and for sarcopterygians and has been lost in dip- the rhipidistian "choana." noans and tetrapods (the claim that a rem- When present, the intracranial joint sepa- nant of the joint is present in Ichthyostega rates the anterior part of the neurocranium, is not yet supported by published evidence). including the orbit, from the posterior part, We cannot ourselves resolve these questions including the brain minus the olfactory lobes, on the basis of a study of the anatomical fea- and it corresponds, at least ventrally, to an tures of the joint in different animals, but we embryonic division of the neurocranium into observe that if there are different degrees of 260 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

GNATHOSTOMATA

0 STE IC HT HYES 1

I ACTINOPTERYGII SARCOPTERYGI I

ACT IN OPTERI

8 -41

.24 - 27

I

7 1 = t Acanthodes bronni

2 = t Eusthenopteron foordi *1-6 3 = t Porolepiformes

FIG. 62. Character-state tree of major groups of gnathostomes. Numbered characters refer to syn- apomorphy scheme in text. Unresolved cladistic position of porolepiforms is discussed in synapomorphy scheme. joint development in rhipidistians, and rhip- cranial joint is non-homologous and, as a idistians are not a monophyletic group as we matter of parsimony, Miles is incorrect in have every reason to suspect, then actinis- asserting that the entire intracranial joint tians would simply represent the last of a is homologous." perhaps large series of sister groups whose Wiley (1979) also proposed in a study of positions relative to each other in the hier- ventral gill arch muscles that dipnoans and archy would be specified by the degree of tetrapods are sister groups but that actinis- kineticism. A parsimony argument would, tians are excluded from membership in either therefore, require that the joint be consid- the Sarcopterygii or Actinopterygii. He ad- ered either an autapomorphy of actinistians, vocated that the Actinistia are instead the as suggested by Bjerring (1973), or a primitive sister group of all other osteichthyans. The feature of sarcopterygians (Miles, 1977). Wi- latter conclusion was based on his observa- ley (1979) also reviewed the problem and tion that three conditions of the musculature concluded (p. 171) "that both Bjerring and are present as a synapomorphy of actinop- Miles are partly correct. Bjerring (1973) is cor- terygians plus sarcopterygians but are rect in his assertion that the dorsal part of the absent in actinistians. These three conditions intracranial joint is non-homologous [in rhipi- are (1) an obliquus ventralis 1; (2) a trans- distians and actinistians]. Miles (1977) is versus ventralis 4; and (3) the pharyngocla- correct in accepting the ventral part of the viculares. Although we have not ourselves intracranial joint as homologous. Bjerring is studied this musculature, we admit to some incorrect in asserting that the entire intra- surprise at the absence of three myological 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 261 features of Latimeria that occur in dipnoans porolepiforms in this sequence if more could and urodeles, primarily because four of the be learned about details of their anatomy- 14 synapomorphies we propose for uniting and that porolepiforms belong somewhere in actinistians with dipnoans and tetrapods are this sequence seems evident to us. What this in gill arch characters, three of which are fea- has meant to us, since actinistians and some tures of the ventral gill arch skeleton. For porolepiforms have two external narial open- the present, it is parsimony rather than a ings in the snout, is that the single narial study of Wiley's "absence" characters that opening in the snout of osteolepiforms, and forces us to interpret the absence of these Eusthenopteron in particular, is reasonably muscles in Latimeria as autapomorphic. Wi- interpreted as having housed a common ley's evidence that "refutes" a link between chamber for superficial membranous canals Latimeria and our Choanata includes not of two external nostrils, as in Recent Polyp- only the three "absence" characters, but terus (compare figs. 11 and 17A, B). Ac- also two muscles uniquely shared by dip- cepting this interpretation, there is no longer noans and urodeles: a subarcualis rectus and a reason to search for a missing fenestra exo- multiple pharyngoclaviculares. For us, how- narina posterior in the palate of Eusthe- ever, these last two conditions of the mus- nopteron and other osteolepiforms. Appli- culature only further corroborate our Choa- cation of parsimony in constructing theories nata and say nothing about the position of of relationships among tetrapods, dipnoans, Latimeria. actinistians and "rhipidistians" could long Finally, a word about the "choana" of ago have resolved the question of the rhipi- Eusthenopteron is in order. We have tried to distian "choana" with the ample data al- show that study of new material of the Eu- ready in the literature. sthenopteron palate allows other interpreta- Our conclusions, based on the foregoing tions of the palatal fenestra which is notably synapomorphy scheme are depicted in figure smaller in our specimens than in Jarvik's: 62, and summarized in the classification be- (1) as a space for the passage of nerves and low; fossils are incorporated according to the vessels as seen in some Recent fishes; (2) as method described by Patterson and Rosen a space allowing movement along a joint (1977). surface; and (3) as a space for receiving the tip of a coronoid fang when the mouth SUPERCLASS Gegenbaur, 1874 is closed. One might even suppose, since plesion tAcanthodes bronni Agassiz, 1832 the palatines and pterygoids were dis- CLASS Chondrichthyes Huxley, 1880 SUBCLASS Selachii Latreille, 1825 placed anteriorly under the vomer in Jarvik's SUBCLASS Holocephali Bonaparte, 1832 specimen, probably a result of dorsoventral CLASS Osteichthyes Huxley, 1880 crushing, that the much larger palatal fenes- SUBCLASS Actinopterygii Cope, 1891 tra observed by Jarvik was caused mechan- INFRACLASS Cladistia Cope, 1871 ically by an abnormally deep penetration of INFRACLASS Actinopteri Cope, 1871 a coronoid fang into the palate. Whatever the SERIES Chondrostei Muller, 1844 reality of the Eusthenopteron palate might SERIES Regan, 1923 have been, there is another view of the prob- DIVISION Ginglymodi Cope, 1871 lem which it is helpful to remember: to ac- DIVISION Halecostomi Regan, 1923 cept the position of Eusthenopteron as a sis- SUBCLASS Sarcopterygii Romer, 1955 Sarcopterygii incertae sedis t Porolepiformes ter group of tetrapods based on this feature Jarvik, 1942 and even the alleged "septomaxilla" and plesion tEusthenopteron foordi Whiteaves, "nasolacrimal duct," it would be necessary 1881 to reject at least 33 characters (nos. 28 INFRACLASS Actinistia Cope, 1871 through 61 in our synapomorphy scheme) INFRACLASS Choanata Save-Soder- which unite actinistians, dipnoans and tet- bergh, 1934 rapods. Some of these 33 characters would SERIES Dipnoi Muller, 1844 also dichotomously resolve the position of SERIES Tetrapoda 262 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

In the classification above we use two very include Cladistia. The distinction we suggest similar names intentionally: Actinopterygii, will allow continued use of the vernacular Actinopteri. Actinopteri is Cope's original "actinopt," and it will only be necessary to name for actinopterygians sensu stricto, ex- specify Actinopterygii or Actinopteri when cluding Cladistia. Actinopterygii is today the inclusive (+Cladistia) and exclusive commonly used sensu Goodrich (1930), to (-Cladistia) groups are to be distinguished.

CONCLUSIONS Goodrich (1909, p. 230), writing of the group relationships, there is nothing to say "many striking points of resemblance" be- about osteolepiforms as a whole, or osteo- tween lungfishes and amphibians, comment- lepidids, since they seem to lack synapo- ed that they "cannot all be put down to con- morphies, and therefore present the prob- vergence." Goodrich was careful to lems peculiar to extinct paraphyletic groups. distinguish primitive and derived characters, Previous attempts to specify the relationship and the "many striking points of resem- between osteolepiforms and tetrapods in blance" were, in his view, derived. Since terms of synapomorphies (Szarski, 1977; Goodrich's time, the number of apparent Schultze, 1977a; Gaffney, 1979a, 1979b) synapomorphies between lungfishes and tet- have avoided the problem of characterizing rapods has increased rather than decreased, osteolepiforms, and have relied almost ex- and the question still remains, can they all clusively on the choana, a character which be put down to convergence? The consensus we have found to be open to other interpre- today is that these resemblances are conver- tations. If we select Eusthenopteron foordi gent or adaptive, since they are outweighed as the best known osteolepiform, we can by detailed correspondence between Paleo- take that paleospecies, lungfishes, and tet- zoic amphibians and osteolepiform fishes. rapods as three monophyletic groups, and Our purpose in this paper has been to oppose list the apparent synapomorphies between that view, by reemphasizing the shared de- each pair. Eusthenopteron and lungfishes rived characters of lungfishes and tetrapods, share no plausible synapomorphies apart and by proposing alternative explanations of from the elongate basihyal (sublingual rod, the similarities between osteolepiforms and also found in Griphognathus). Among the tetrapods-that they are convergent or prim- apparent synapomorphies between Eusthe- itive. nopteron and tetrapods (p. 177), those Eusthenopteron is, after all, a very primi- that have withstood our criticism best are the tive fish, as Jarvik (e.g., 1968b, p. 506) has pattern of dermal bones on the cheek (fig. often said. Even so, it seems not to be prim- 43), and the structure of the humerus. Lung- itive enough to serve as a morphotype of the fishes and tetrapods share derived features ancestral tetrapod, since others who have of the choana, dermal roofing bones of the surveyed a wide range of osteolepiforms skull, palate and jaw suspension, hyoid arch, (Vorobyeva, Andrews, Thomson, Panchen) gill arches, skull-vertebra articulation, ribs, have concluded that it is osteolepiforms as dermal shoulder girdle, and paired append- a whole, or the earliest osteolepidids, that ages, in addition to numerous features of soft are closest to tetrapods, rather than any de- anatomy, physiology, embryology, and be- rived osteolepiform subgroup. These theo- havior which are not checkable in Eusthe- ries of tetrapod origins follow the traditional nopteron. The principle ofparsimony leads us paleontological method of the search for to accept these numerous derived features as ancestors-reliance on paraphyletic groups. synapomorphies. It follows that the charac- In the approach we have followed, the ters shared by Eusthenopteron and tetrapods search for synapomorphies specifying sister- are primitive for sarcopterygians, conver- 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 263 gent or spurious. Abstracting from the list of ison, 1968a); so do actinopterygians (Wes- those characters which were claimed to be toll, 1979, p. 345). Ancestors are slippery exclusive to Eusthenopteron and tetrapods things, but ancestral groups are very accom- (p. 177), our assessment is as follows: primi- modating, for their only attributes are plesio- tive-cheek pattern, "protorhachitomous" morphous. vertebrae; convergent-polyplocodont teeth, It seems fitting, therefore, to end our as- structure of humerus; spurious-fenestra sault on the "rhipidistian barrier" by record- exochoanalis, septomaxilla. ing some personal observations, made near In those assessments, we mean to argue the completion of this study, of five partic- an alternative view, not to insist that we are ular locomotor and feeding behaviors of two right. Those who are still impressed by re- large lungfishes (a Protopterus annectens semblances between Eusthenopteron, or and Neoceratodus forsteri, each over a me- other osteolepiforms, and tetrapods, may ter long)12 and a small Neoceratodus of amplify and enumerate those resemblances, about 5 centimeters.13 Needless to say, and seek to show that they outnumber or we record them because of their similarity outweigh the lungfish/tetrapod synapomor- to behaviors of aquatic urodeles. (1) Swim- phies that we cite. We hope they will do so. ming is accomplished by undulations of the In the course of this paper we have criticized caudal peduncle while the paired fins are many people and many ideas. Our criticisms held loosely near the body rather than being stem from our viewpoint, that of phyloge- used as maneuvering planes, although Neo- netic systematics. Some of those we criticize ceratodus will sometimes extend the pecto- have already published their opinion of that rals somewhat. (2) Stopping is accomplished viewpoint (e.g., Panchen, 1979; Westoll, by sinking to the substrate on the extended 1979, p. 346). We suggest that little will be paired fins; the pectorals are moved forward gained by denigrating logic (Westoll), or de- to an angle of 90 degrees, or slightly more, nying the principle of parsimony (Panchen). with the body axis, and the pelvics are often Nor will progress come from simply reas- moved forward sharply (as much as 1200). serting general similarities between particu- Thus at rest, with the pelvics firmly down on lar osteolepiforms and particular tetrapods, the substrate, the fish may raise the entire or between an osteolepiform morphotype precaudal region, sitting in this raised posi- and a tetrapod morphotype, unless those tion for some minutes, with the pectorals similarities are justified as synapomorphies hanging loosely downward. (3) When pro- by means of an explicit scheme of relation- gressing forward slowly, the weight of the ships, covering tetrapod subgroups and other body is transferred to the paired fins which relevant groups such as lungfishes and coel- lift the body and pull it forward by moving acanths. in a side-to-side alternating diagonal pattern. In the 140 years since living lungfishes During each cycle of fin movement, the body were first discovered, there has been evi- changes in curvature (when LF-RH fins are dence of a consistent pattern of derived char- protracted the right side of the body is con- acters shared by lungfishes and tetrapods. cave, whereas when RF-LH fins are pro- Nevertheless, during this century, that pat- tracted the right side of the body is convex). tern has been treated as irrelevant or has (4) When performing special movements been explained away, since the problem of across the substrate, such as turning or back- tetrapod relationships has become one of ing, right and left fins can be moved forward origins, the property of paleontology, solved and backward synchronously or one fin can by the process approach: following back be moved independently of the other three. stratigraphic series of tetrapods and finding (5) When feeding from the substrate, the fore- that they converge on a morphotype plausi- 12 In the New York Aquarium of the New York Zoo- bly attributed to osteolepiforms. But when logical Society. lungfishes are treated by the same method, 13 In the Department of Ichthyology of the American they also converge on osteolepiforms (Den- Museum of Natural History. 264 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 part of the body is elevated on the pectoral gene Gaffney, Philippe Janvier, Arnold fins and the head is bent down toward the Kluge, John Maisey, Roger Miles, Angela food; if the food consists of live worms in Milner, Charles Myers, Alec Panchen, Bob the substrate, small Neoceratodus will grasp Presley, Wolf-Ernst Reif, Bobb Schaeffer, the worms in its mouth and push backward Hans-Peter Schultze, Keith Thomson, E. 0. with its pectorals until the worms are dis- Wiley, and Richard Zweifel for providing vari- lodged. None of the above observations is ous significant kinds of information and ad- original with us; the same or similar behav- vice. Mr. Guido Dingerkus, Ms. Norma Fein- iors were noted and recorded many times by berg, Ms. Lynne Parenti, Ms. Amy Rudnick late nineteenth-century and early twentieth- and, in particular, Mr. Peter Whybrow provid- century biologists who were no less im- ed countless hours oftechnical assistance. We pressed than we by the remarkable behav- are grateful to Dr. Anne Kemp for supplying ioral likenesses of lungfishes and urodeles. living and preserved juvenile specimens of We record our own fascination with these Neoceratodus, and to Mr. Guido Dingerkus likenesses for a particular reason-the con- and the New York Aquarium of the New viction that, had the study of lungfish rela- York Zoological Society for making possible tionships developed directly from compari- observations of large, living African and sons among living gnathostomes without Australian lungfishes. We also thank Dr. P. interruption by futile paleontological search- Humphry Greenwood and Mr. Gordon es for ancestors, there would have been no Howes, Division of Fishes, British Museum "rhipidistian barrier" to break through, this (Natural History) (BMNH), Dr. William review would have been written decades Fink, Museum of Comparative Zoology, ago, and we would all have been spared a Harvard University, Dr. Stanley Weitzman, generation of anatomical misconstruction, United States National Museum, Drs. Bobb dispensable scenarios about the fish-amphib- Schaeffer, John Maisey, and Richard Zwei- ian transition, and hapless appeals to plesio- fel, all of the American Museum of Natural morphy. Finally, it should be obvious from History (AMNH), Dr. William N. Eschmeyer, the nature of this review that our argument California Academy of Sciences, and Dr. E. is not with the use of fossils, but with the use C. Boterenbrood, Hubrecht Laboratory, of the traditional paleontological method. Utrecht, for lending specimens in their care. The photographs were taken by Mr. Tim ACKNOWLEDGMENTS Parmenter, British Museum (Natural His- We especially thank Dr. Gareth Nelson for tory) and the staff of the Department of numerous discussions of the general prob- Photography, American Museum of Natural lem, Drs. Mahala Andrews, James Atz, Eu- History.

LITERATURE CITED Agassiz, J. L. R. special reference to Polypterus bichir. 1838. Recherches sur les poissons fossiles. Jour. Anat., vol. 56, pts. 3 and 4, pp. Neuchatel, vol. 3, p. 129. 190-294. 1843. Recherches sur les poissons fossiles. 1923. The cranial anatomy of Chlamydosela- Neuchatel, vol. 2, pp. 2-46. chus anguineus. Acta Zool. Stockholm, Aldinger, H. vol. 4, pp. 123-221. 1937. Permische Ganoidfische aus Ostgr0n- land. Meddel. Gr0nland, vol. 102, pp. 1932a. Concerning the nasal apertures, the 1-392. lachrymal canal and the buccopharyn- Allis, E. P., Jr. geal upper lip. Jour. Anat., vol. 66, pt. 1919. The lips and the nasal apertures in the 4, pp. 650-658. gnathostome fishes. Jour. Morph., vol. 1932b. The pre-oral gut, the buccal cavity and 32, no. 1, pp. 145-205. the buccopharyngeal opening in Cerat- 1922. The cranial anatomy ofPolypterus, with odus. Ibid., vol. 66, pt. 4, pp. 659-668. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 265

Andrews, S. M. 1840b. Description anatomique du Lepidosiren 1973. Interrelationships of crossopterygians. paradoxa. Ann. Sci. Nat., ser. 2, vol. In Greenwood, P. H., R. S. Miles and 14, no. 2, pp. 116-159. C. Patterson (eds.), Interrelationships Bjerring, H. C. of fishes. London, Academic Press, pp. 1967. Does a homology exist between the 137-177. basicranial muscle and the polar carti- 1977. The axial skeleton of the coelacanth lage? Colloques Internatl. Centre Natl. Latimeria. In Andrews, S. M., R. S. Rech. Sci., vol. 163, pp. 223-267. Miles and A. D. Walker (eds.), Prob- 1973. Relationships of coelacanthiforms. In lems in vertebrate evolution. London, Greenwood, P. H., R. S. Miles and C. Academic Press, pp. 271-288. Patterson (eds.), Interrelationships of Andrews, S. M., and T. S. Westoll fishes. London, Academic Press, pp. 1970a. The postcranial skeleton of Eusthenop- 179-205. teron foordi Whiteaves. Trans. Roy. 1977. A contribution to structural analysis of Soc. Edinburgh, vol. 68, no. 9, pp. 207- the head of animals. Zool. 329. Scripta, vol. 6, pp. 127-183. 1970b. The postcranial skeleton of rhipidistian Boas, J. E. V. fishes excluding Eusthenopteron. Ibid., 1882. Uber den Conus arteriosus und die Ar- vol. 68, no. 12, pp. 391-489. terienbogen der Amphibien. Morph. Archer, R., J. Chauvet, and M. T. Chauvet Jahrb., vol. 7, pp. 488-572. 1970. A tetrapod neurohypophyseal hormone Boulenger, G. A. in African lungfishes. Nature, vol. 227, 1901. Les poissons du Bassin du Congo. no. 5254, pp. 186-187. Brussels, lxii + 532 pp. Asmuss, H. Braus, H. 1856. Das vollkommenste Hautskelet der 1901. Die Muskeln und Nerven der Cerato- bisher bekannten Thierreihe. Abhandl. dusflosse. In Semon, R., Zoologische Erlangung der Magisterwurde, Dorpat., Forschungsreisen in Australien und 40 pp. dem Malayischen Archipelago. I. Ce- Baur, G. ratodus. Pt. 3. Jena, G. Fischer Verlag, 1896. The Stegocephali. A phylogenetic study. pp. 139-300. Anat. Anz., vol. 11, pp. 657-675. Bridge, T. W. Beaumont, E. 1897. The morphology of the skull in the Par- 1977. Cranial morphology of the Loxomma- aguayan Lepidosiren and other Dipnoi. tidae (Amphibia: Labyrinthodontia). Trans. Zool. Soc. London, vol. 14, pp. Phil. Trans. Roy. Soc. London (B), vol. 325-376. 280, pp. 29-101. 1904. Fishes (exclusive of the systematic ac- Beer, G. R., de count of Teleostei). In Harmer, S. F., 1937. The development of the vertebrate and A. E. Shipley (eds.), Cambridge skull. Oxford Univ. Press, pp. xxiv + natural history. London, Macmillan and 552. Co., vol. 7, pp. 139-537. Bertmar, G. Broman, I. 1965. The olfactory organ and upper lips in 1939. Uber die Entwicklung der Geruchsor- Dipnoi, an embryological study. Acta gane bei den Lungenfischen. Morph. Zool. Stockholm, vol. 46, pp. 1-10. Jahrb., vol. 83, pp. 85-106. 1966. On the ontogeny and homology of the Broom, R. choanal tubes and choanae in Urodela. 1895. On the homology of the palatine process Ibid., vol. 47, pp. 43-59. of the mammalian premaxillary. Proc. 1968. Lungfish phylogeny. In 0rvig, T. (ed.), Linnean Soc. New South Wales, vol. Current problems of lower vertebrate 10, pp. 477-485. phylogeny. Stockholm, Almqvist and Bryant, W. L. Wiksell, pp. 259-283. 1919. On the structure of Eusthenopteron. Bischoff, T. L. W., von Bull. Buffalo Soc. Nat. Sci., vol. 13, no. 1840a. Lepidosiren paradoxa. Anatomisch un- 1, pp. 4-22. tersucht und Beschrieben. Leipzig, vi Bystrov, A. P. + 34 pp. 1935. Morphologische Untersuchungen der 266 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

Dekknochen des Schadels der Wirbel- Soc. Belge Geol., Paleont., Hydrologie, tiere. Acta Zool. Stockholm, vol. 16, pp. vol. 9, no. 2, pp. 79-128. 65-141. Dumeril, A. H. A. Bystrov, A. P., and J. A. Efremov 1870. Le Lepidosiren et le Protoptere apartien- 1940. Benthosuchus sushkini-a labyrintho- nent a la classe des poissons ou ils sont dont from the Eotriassic of Sharzhenga les types de la sous-classe des dipnes. River. Trudy Paleozool. Inst., vol. 10, Ann. Soc. Linneenue Maine-et-Loire, pp. 1-152. [In Russian, English vol. 12, pp. 140-149. summary]. Edgeworth, F. H. Carrol, R. L., and R. Holmes 1935. The cranial muscles of vertebrates. 1980. The skull and jaw musculature as guides Cambridge Univ. Press, pp. ix + 493. to the ancestry of . Zool. Edwards, H. Milne Jour. Linnean Soc., vol. 68, pp. 1-40. 1840. Remarque sur les affinites naturelles du Cope, E. D. Lepidosiren. Ann. Sci. Nat. Ser. 2, vol. 1872. Observations on the systematic rela- 14, no. 2, pp. 159-162. tions of the fishes. Proc. Amer. Assoc. Emelianov, S. W. Adv. Sci., vol. 20, pp. 317-343. 1935. Die Morphologie der Fischrippen. Zool. 1884. Note on the phylogeny of the Vertebra- Jahrb. (Anat.), vol. 60, pp. 132-262. ta. Amer. Nat., vol. 18, pp. 1255-1257. 1936. Die Morphologie der Tetrapodenrippen. 1885. On the evolution of the Vertebrata, pro- Ibid., vol. 62, pp. 173-274. gressive and retrogressive. Ibid., vol. Fitzinger, L. J. F. J. 19, pp. 140-148; 234-247. 1837. Vorlaufiger Bericht uber eine h6chst in- 1887. Geology and palaeontology. Ibid., vol. teressante Entdeckung Dr. Natterers in 21, pp. 1014-1019. Brasil. Isis, Jena, pp. 379-380. 1892. On the phylogeny of the Vertebrata. 1843. Systema reptilium. Vienna, Braumuller Proc. Amer. Phil. Soc., vol. 30, pp. 278- et Seidel, pp. vi + 106. 281. Forey, P. L. Daget, J. 1980. Latimeria: a paradoxical fish. Proc. Roy. 1950. Revision des affinites phylogenetiques Soc. London, B. Biol. Sci., vol. 208, pp. des Polypterides. Mem. Inst. Fr. Afr. 369-384. Noire, vol. 11, pp. 1-178. [In press] The coelacanth Rhabododerma from Daniel, J. F. the Carboniferous of the British Isles. 1928. The elasmobranch fishes. Second edi- Palaeontology, vol. 24, pt. 1, pp. 203- tion. Berkeley, Univ. California Press, 229. xi + 332 pp. Forster-Cooper, C. Dean, B. 1937. The Middle Devonian fish fauna of 1903. Obituary notice of a lung-fish. Pop. Sci. Achanarras. Trans. Roy. Soc. Edin- Monthly, vol. 43, pp. 33-39. burgh, vol. 59, pp. 223-239. Denison, R. H. Fox, H. 1968a. The evolutionary significance of the ear- 1965. Early development of the head and liest known lungfish, Uranolophus. In ofNeoceratodus with a consid- 0rvig, T. (ed.), Current problems of eration of its phylogeny. Jour. Zool. lower vertebrate phylogeny. Stock- London, vol. 146, pp. 470-554. holm, Almqvist and Wiksell, pp. 247- Francois, Y. 257. 1966. Structure et developpement de la ver- 1968b. Early Devonian lungfishes from Wyo- tebre de Salmo et des teleosteens. Paris, ming, Utah and Idaho. Fieldiana: Geol., Archs. Zool. Exp. Gen., vol. 107, pp. vol. 17, pp. 351-413. 287-328. 1974. The structure and evolution of teeth in Fritsch, A. lungfishes. Ibid., vol. 33, pp. 31-58. 1883-1895. Fauna der Gaskohle und der Kalk- Devillers, C. steine der Permformation Bohmens. 1954. Les cotes. In Grasse, P. P. (ed.), Traite Prag, 1885, vol. II, pt. 1, pp. 1-32. de zoologie. Paris, Masson, vol. 12, pp. Fiurbringer, K. 673-697. 1904. Beitriige zur Morphologie des Skelettes Dollo, L. der Dipnoer, nebst Bemerkungen uber 1896. Sur la phylogenie des dipneustes. Bull. Pleuracanthiden, Holocephalen und 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 267

Squaliden. Denkschr. Med.-Naturwiss. 1925. Cranial roofing-bones in Dipnoi. Jour. Gesell. Jena, vol. 4, pp. 425-510. Linnean Soc. (Zool.), vol. 36, pp. 79- Gadow, H. F. 86. 1933. The evolution of the vertebral column. 1930. Studies on the structure and develop- Cambridge Univ. Press, 356 pp. ment of vertebrates. London, Macmil- Gaffney, E. S. lan and Co., pp. xxx + 837. 1979a. Tetrapod monophyly: a phylogenetic Gray, J. E. analysis. Bull. Carnegie Mus. Nat. 1856. Observations on a living African Lepido- Hist., vol. 13, pp. 92-105. siren in the Crystal Palace. Proc. Zool. 1979b. An introduction to the logic of phylog- Soc. London, Pt. 24, pp. 342-346. eny reconstruction. In Cracraft, J., and Gregory, W. K. N. Eldredge (eds.), Phylogenetic anal- 1904. The relations of the anterior visceral ysis and paleontology. Columbia Univ. arches to the chondrocranium. Biol. Press, pp. 79-111. Bull. Marine Biol. Lab. Woods Hole, Gardiner, B. G. vol. 7, pp. 55-69. 1973. Interrelationships of teleostomes. In 1911. The limbs of Eryops and the origin of Greenwood, P. H., R. S. Miles and C. paired limbs from fins. Science, new Patterson (eds.), Interrelationships of ser., vol. 33, pp. 508-509. fishes. London, Academic Press, pp. 1915. Present status of the problem of the or- 105-135. igin of the Tetrapoda, with special ref- Gardiner, B. G., and A. W. H. Bartram erence to the skull and paired limbs. 1977. The homologies of ventral cranial fis- Ann. New York Acad. Sci., vol. 26, pp. sures in osteichthyans. In Andrews, S. 317-383. M., R. S. Miles and A. D. Walker Gregory, W. K., H. Rockwell, and F. G. Evans (eds.), Problems in vertebrate evolu- 1939. Structure of vertebral column in Eus- tion. London, Academic Press, pp. 227- thenopteron foordi Whiteaves. Jour. 245. Paleont., vol. 13, pp. 126-129. Gegenbaur, C. Gross, W. 1895. Das Flossenskelet der Crossopterygier 1956. Uber Crossopterygier und Dipnoer aus und das Archipterygium der Fische. dem baltischen Oberdevon im Zusam- Morph. Jahrb., vol. 22, pp. 119-160. menhang einer vergleichenden Unter- Gilbert, S. G. suchung des Porenkanalsystems paliio- 1973. Pictorial anatomy of the Necturus. zoischer Agnathen und Fische. K. Univ. Washington Press, 47 pp. Svenska Vetensk. Akad. Handl., ser. 4, Gill, T. vol. 5, no. 6, pp. 1-140. 1872. Arrangement of the families of fishes. 1962. Neuuntersuchung der Stensioellida (Ar- Smithsonian Misc. Coll., vol. 247, pp. throdira, Unterdevon). Notizbl. Hess. xlvi + 49. Landesanst. Bedenforsch. Wiesbaden, Goodey, T. vol. 90, pp. 48-86. 1910. A contribution to the skeletal anatomy Gunther, A. C. L. G. of the , Chlamydoselachus 1871. Description of Ceratodus, a genus of anguineus Gar. Proc. Zool. Soc. Lon- ganoid fishes, recently discovered in don, vol. 19, pp. 340-571. rivers of Queensland, . Phil. Goodrich, E. S. Trans. Roy. Soc. London, vol. 161, pp. 1901. On the pelvic girdle and fin of Neoce- 511-571. ratodus. Quart. Jour. Micros. Sci., vol. 45, pp. 311-324. Gunther, A. E. 1909. Vertebrata craniata. First fascicle: cy- 1975. A Century of zoology at the British Mu- clostomes and fishes. In Lankester, E. seum. London, Dawsons, 533 pp. R. (ed.), A treatise on zoology. London, Haeckel, E. A. and C. Black, vol. 9, xvi + 518 pp. 1866. Generelle Morphologie der Organis- 1919. Restorations of the head of Osteolepis. men. 2 vols. Berlin, G. Reimer, 574 + Jour. Linnean Soc. (Zool.), vol..34, pp. 447 pp. 181-188. 1889. Naturliche SchopfungsGeschichte. 1924. The origin of land vertebrates. Nature, Eighth edition. Berlin, G. Reimer, xxx London, vol. 114, pp. 935-936. + 832 pp. 268 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

Heaton, M. J. Hyrtl, J. 1979. Cranial anatomy of primitive captorhi- 1845. Lepidosiren paradoxa: Monographie. nid reptiles from the late Pennsylvanian Abhandl. Bohmens Ges. Wiss., vol. 3, and early ofOklahoma and Tex- no. 5, pp. 1-64. as. Bull. Oklahoma Geol. Surv., no. Jardine, N. 127, iv + 84 pp. 1970. The observational and theoretical com- Heckel, J. ponents of homology: a study based on 1845. Bemerkung uber Lepidosiren paradoxa. the morphology of the dermal skull- Arch. Anat. Physiol., pp. 534-535. roofs of rhipidistian fishes. Biol. Jour. 1851. Uber einer neue Fisch-Species aus dem Linnean Soc., vol. 1, pp. 327-361. Weissen Nil, Protopterus aethiopicus. Jarvik, E. Sitzber. Akad. Wiss. Wien, vol. 6, pp. 1937. On the species ofEusthenopteron found 685-689. in Russia and the Baltic States. Bull. Helbing, H. Geol. Inst. Uppsala, vol. 27, pp. 63-127. 1904. Beitrage zur Anatomie und Systematik 1942. On the structure of the snout of cros- der Laemargiden. Nova Acta Leopol- sopterygians and lower gnathostomes in dina, vol. 82, pp. 335-524. general. Zool. Bidrag frtn Uppsala, vol. Higgins, G. M. 21, pp. 235-675. 1920. The nasal organ in Amphibia. III. Biol. 1944a. On the dermal bones, sensory canals Monogr., vol. 6, no. 1, 91 pp. and pit-lines of the skull in Eusthenop- Hogg, J. teron foordi Whiteaves, with some re- 1841. On the existence of branchiae in the marks on E. save-soderberghi Jarvik. young Caeciliae; and on a modification K. Svenska Vetensk. Akad. Handl., and extension of the branchial classifi- ser. 3, vol. 21, no. 3, pp. 1-48. cation of the Amphibia. Ann. Mag. Nat. 1944b. On the exoskeletal shoulder-girdle of te- Hist., vol. 7, pp. 353-363. leostomian fishes, with special refer- Holmgren, N. ence to Eusthenopteron foordi White- 1933. On the origin of the tetrapod limb. Acta aves. Ibid., ser. 3, vol. 21, no. 7, pp. 1- Zool. Stockholm, vol. 14, pp. 185-295. 32. 1939. Contribution to the question of the ori- 1948. On the morphology and of gin of the tetrapod limb. Ibid., vol. 20, the Middle Devonian osteolepid fishes pp. 89-124. of Scotland. Ibid., ser. 3, vol. 25, no. 3, 1949a. Contributions to the question of the or- pp. 1-301. igin of tetrapods. Ibid., vol. 30, pp. 1950. Middle Devonian vertebrates from Can- 459-484. ning Land and Wegeners Halvo (East 1949b. On the tetrapod limb problem-again. Greenland). Pt. II. Crossopterygii. Ibid., vol. 30, pp. 485-508. Meddel. Gr0nland, vol. 96, no. 4, pp. Holmgren, N., and E. Stensio 1-132. 1936. Kranium und Visceral skelett der Ak- 1952. On the fish-like tail in the ichthyostegid ranier, Cyclostomen und Fische. In L. stegocephalians with descriptions of a Bolk (ed.), Handb. Vergl. Anat., vol. 4, new stegocephalian and a new crossop- Berlin, Urban u. Schwarzenberg, pp. terygian from the Upper Devonian of 233-500. East Greenland. Ibid., vol. 114, no. 12, Huxley, T. H. pp. 1-90. 1861. Preliminary essay upon the systematic arrangement of the fishes of the Devo- 1954. On the visceral skeleton in Eusthenop- nian epoch. Mem. Geol. Surv. United teron with a discussion of the parasphe- K. Kingdom, decade 10, pp. 1-40. noid and palatoquadrate in fishes. 1876. Contributions to morphology. Ichthy- Svenska Vetensk. Akad. Handl., ser. 4, opsida No. 1. On Ceratodus forsteri, vol. 5, no. 1, 104 pp. with observations on the classification 1960. Theories de l'evolution des vertebres of fishes. Proc. Zool. Soc. London, pp. reconsiderees 'a la lumiere des recentes 24-59. decouvertes sur les vertebres inferieurs. 1880. On the application of the laws of evo- Paris, Masson, 104 pp. lution to the arrangement of the Verte- 1963. The composition of the intermandibular brata and more particularly of the Mam- division of the head in fish and tetrapods malia. Ibid., pp. 649-662. and the diphyletic origin of the tetrapod 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 269

tongue. K. Svenska Vetensk. Akad. ogy. vol. 4, London and New York, Handl., ser. 4, vol. 9, no. 4, pp. 1-74. Academic Press, pp. 361-413. 1964. Specializations in early vertebrates. Jollie, M. Ann. Soc. Roy. Zool. Belgique, vol. 94, 1962. morphology. New York, Rein- no. 1, pp. 11-95. hold, 478 pp. 1%5. On the origin of girdles and paired fins. Jordan, D. S. Israel Jour. Zool., vol. 14, pp. 141-172. 1905. A guide to the study of fishes. 2 vols., 1966. Remarks on the structure of the snout New York, H. Holt and Co. 624 + 599 in Megalichthys and certain other rhip- PP. idistid crossopterygians. Arkiv Zool., Jurgens, J. D. vol. 19, no. 2, pp. 41-98. 1971. The morphology of the nasal region of 1967. The homologies of frontal and parietal Amphibia and its bearing on the phylog- bones in fishes and tetrapods. Colloques eny of the group. Ann. Univ. Stellen- Internatl. Centre Natl. Rech. Sci., vol. bosch, ser. A, vol. 46, no. 2, 146 pp. 163, pp. 181-213. Kerr, J. G. 1968a. The systematic position of the Dipnoi. 1932. Archaic fishes-Lepidosiren, Protopte- In 0rvig, T. (ed.), Current problems of rus, Polypterus-and their bearing upon lower vertebrate phylogeny. Stock- problems of vertebrate morphology. holm, Almqvist and Wiksell, pp. 223- Jena Zeitschr. Naturwiss., vol. 67, pp. 245. 419-433. 1968b. Aspects of vertebrate phylogeny. In Kesteven, H. L. 0rvig, T. (ed.), Ibid., pp. 497-527. 1950. The origin of the tetrapods. Proc. Roy. 1972. Middle and upper Devonian Porolepi- Soc. Victoria, new ser., vol. 59, no. 2, formes from East Greenland with special pp. 93-138. reference to Glyptolepis groenlandica Kingsley, J. S. n.sp. and a discussion on the structure 1892. The head of an embryo Amphiuma. of the head in the Porolepiformes. Med- Amer. Nat., vol. 26, pp. 671-680. del. Gr0nland, vol. 187, no. 2,307 pp. Klaatsch, H. 1975. On the saccus endolymphaticus and ad- 1896. Die Brustflosse der Crossopterygier. jacent structures in osteolepiforms, an- Ein Beitrag zur Anwendung der Archip- urans and urodeles. Colloques Internatl. terygium-Theorie auf die Gliedmassen Centre Natl. Rech. Sci., vol. 218, pp. der Landwirbelthiere. In Festschrift 191-211. zum siebenzigsten Geburtstage von Carl Jessen, H. Gegenbaur. vol. 1, Leipzig, pp. 261- 1966. Die Crossopterygier des Oberen Plat- 391. tenkalkes (Devon) der Bergisch-Glad- Krefft, J. bach-Paffrather Mulde (Rheinisches 1870. Description of a gigantic amphibian al- Schiefergebirge) unter Beriicksichti- lied to the genus Lepidosiren, from the gung von amerikanischem und euro- Wide Bay District, Queensland. Proc. piiischem -Material. Arkiv Zool. Soc. London, pp. 221-224. Zool., vol. 18, no. 2, pp. 305-389. Lauder, G. V. 1967. The position of the Struniiformes Strun- 1979. Review of J. Millot, J. Anthony and ius and Onychodus among the crossop- D. Rabineau, Anatomie de Latimeria terygians. Colloques Internatl. Centre chalumnae, vol. 3 (1978). Copeia, no. Natl. Rech. Sci., vol. 163, pp. 173-180. 3, pp. 560-562. 1972. Schultergiirtel und Pectoralflosse bei Lehman, J.-P. Actinopterygien. Fossils and Strata, 1956. L'evolution des dipneustes et l'origine vol. 1, pp. 1-101. des urodeles. Colloques Internatl. Centre 1975. A new choanate fish, Powichthys thor- Natl. Rech. Sci., vol. 60, pp. 69-76. steinssoni n.g., n.sp., from the early 1959. Les Dipneustes du Devonien superieur Lower Devonian of the Canadian Arctic du Groenland. Meddel. Gr0nland, vol. Archipelago. Colloques Internatl. Centre 160, pp. 1-58. Natl. Rech. Sci., vol. 218, pp. 213-225. Lehmann, W. M., and T. S. Westoll Johansen, K. 1952. A primitive dipnoan fish from the Lower 1970. Air breathing in fishes. In Hoar, W. S., Devonian of Germany. Proc. Roy. Soc., and D. J. Randall (eds.), Fish physiol- B, vol. 140, pp. 403-421. 270 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

Lindsey, C. C. ered in the light of new ptyctodontids 1978. Form, function, and locomotory habits from Gogo, Western Australia. In An- in fish. In Hoar, W. S., and D. J. Ran- drews, S. M., R. S. Miles and A. D dall (eds.), . vol. 7. Lon- Walker (eds.), Problems in vertebrate don and New York, Academic Press, evolution. London, Academic Press, pp. 1-100. pp. 123-198. L0vtrup, S. Millot, J., and J. Anthony 1977. The phylogeny of Vertebrata. London, 1958. Anatomie de Latimeria chalumnae. J. Wiley, xii + 330 pp. Tome I. Squelette, muscles et forma- M'Donnell, R. tions de soutien. Paris, Centre Natl. 1860. Observations on the habits and anatomy Rech. Sci., 122 pp. of the Lepidosiren annectens. Jour. 1965. Anatomie de Latimeria chalumnae. Roy. Dublin Soc., vol. 2, pp. 3-20. Tome II. Systeme nerveux et organes Medvedeva, I. M. des sens. Ibid., 120 pp. 1959. The naso-lachrymal duct and its rela- Miner, R. W. tionships to the lachrymal and septo- 1925. Pectoral limb ofEryops, and other prim- maxillary dermal bones in Ranodon si- itive tetrapods. Bull. Amer. Mus. Nat. biricus. Dokl. Akad. Nauk SSSR, vol. Hist., vol. 51, pp. 145-3 12. 128, pp. 425-428. 1960. On the relationships of the developing Mookerjee, H. K., D. N. Ganguly, and S. K. naso-lachrymal duct to the lachrymal Brahma and septomaxillary dermal bones in Hy- 1954. On the development of the centrum and nobius keyserlingii. Ibid., vol. 131, pp. its arches in the Dipnoi Protopterus an- 1209-1212. nectens. Anat. Anz., vol. 100, pp. 217- 1968. Die Homologie des Jacobsonschen Or- 230. gans bei Anura und Urodela. In 0rvig, Moy-Thomas, J. A., and R. S. Miles T. (ed.), Current problems of lower ver- 1971. Palaeozoic fishes. Second edition. Phila- tebrate phylogeny. Stockholm, Alm- delphia and Toronto, W. B. Saunders qvist and Wiksell, pp. 331-340. Co., 259 pp. Melville, G. Muller, J. 1848. On the Lepidosiren. Rept. Brit. Assoc. 1840. Bericht uber die Fortschritte der ver- Adv. Sci., vol. 17, p. 78. gleichenden Anatomie im Jahre 1839. 1860. Appendix to M'Donnell's paper. Jour. Arch. Anat. Physiol., pp. 159-198. Roy. Dublin Soc., vol. 2, pp. 405-406. 1844. Ueber den Bau und die Grenzen der Miles, R. S. Ganoiden. Ber. Akad. Wiss. Berlin, 1964. A reinterpretation of the visceral skel- 1846, pp. 67-85. eton of Acanthodes. Nature, London, Natterer, J. vol. 204, pp. 457-459. 1837. Lepidosiren paradoxa, eine neue Gat- 1965. Some features in the cranial morpholo- tung aus der Familie der fischahnlichen gy of acanthodians and the relationships Reptilien. Ann. Naturhist. Mus. Wien, of the Acanthodii. Acta Zool. Stock- vol. 2, pp. 165-170. holm, vol. 46, pp. 233-255. 1973. Relationships of acanthodians. In Green- Nelson, G. J. wood, P. H., R. S. Miles and C. Patter- 1968. Gill-arch structure in Acanthodes. In son (eds.), Interrelationships of fishes. 0rvig, T. (ed.), Current problems of London, Academic Press, pp. 63-104. lower vertebrate phylogeny. Stock- 1975. The relationships of the Dipnoi. Col- holm, Almqvist and Wiksell, pp. 129- loques Internatl. Centre NatI. Rech. 143. Sci., vol. 218, pp. 133-148. 1969. Gill arches and the phylogeny of fishes, 1977. Dipnoan (lungfish) skulls and the rela- with notes on the classification of ver- tionships of the group: a study based on tebrates. Bull. Amer. Mus. Nat. Hist., new species from the Devonian of Aus- vol. 141, art. 4, pp. 475-552. tralia. Zool. Jour. Linnean Soc., vol. 61, 1970. Pharyngeal denticles (placoid scales) of pp. 1-328. sharks, with notes on the dermal skel- Miles, R. S., and G. C. Young eton of vertebrates. Amer. Mus. Novi- 1977. Placoderm interrelationships reconsid- tates, no. 2415, pp. 1-26. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 271

Newman, F. sure anthracosaurs. Phil. Trans. Roy. 1857. Note on Lepidosiren annectens. Proc. Soc. London (B), vol. 247, pp. 593-637. Linnean Soc. London, 1856, pp. 73-74. 1967. The nostrils of choanate fishes and early Nielsen, E. tetrapods. Biol. Rev., vol. 42, pp. 374- 1936. Some few preliminary remarks on Tri- 420. assic fishes from East Greenland. Med- 1970. Anthracosauria. In Kuhn, 0. (ed.), del. Gr0nland, vol. 112, no. 3, pp. 1-55. Handbuch der Palaoherpetologie, Stutt- 1942. Studies on Triassic fishes from East gart, Fischer, pt. 5a, pp. vii + 84. Greenland. I. Glaucolepis and Boreo- 1972a. The skull and skeleton of Eogyrinus at- somus. Palaeozool. Groenlandica 1, 403 theyi Watson (Amphibia: Labyrintho- PP. dontia). Phil. Trans. Roy. Soc. London 1949. Studies on Triassic fishes from East (B), vol. 263, pp. 279-326. Greenland. II. Australosomus and Bir- 1972b. The interrelationships of the earliest tet- geria. Ibid., 3, 309 pp. rapods. In Joysey, K. A., and T. S. Noble, G. K. Kemp (eds.), Studies in vertebrate evo- 1931. The biology of the Amphibia. New lution. Edinburgh, Oliver and Boyd, pp. York, London, McGraw-Hill Book Co., 65-87. 577 pp. 1977. The origin and early evolution of tetra- Norris, H. W. pod vertebrae. In Andrews, S. M., R. 1913. Cranial nerves of Siren lacertina. Jour. S. Miles, and A. D. Walker (eds.), Prob- Morph., vol. 24, no. 2, pp. 245-338. lems in vertebrate evolution. London, Oken, L. Academic Press, pp. 289-318. 1841. Lepidosiren paradoxa. Isis, vol. 34, pp. 1979. The cladistic debate continued. Nature, 462-469. vol. 280, p. 541. 0rvig, T. Parrington, F. R. 1969a. Cosmine and cosmine growth. Lethaia, 1967. The identification of the dermal bones vol. 2, pp. 241-260. of the head. Zool. Jour. Linnean Soc., 1969b. Vertebrates from the Wood Bay Group vol. 47, pp. 231-239. and the position of the Emsian- Parrington, F. R., and T. S. Westoll boundary in the Devonian of Vestspits- 1940. On the evolution of the mammalian pal- bergen. Ibid., vol. 2, pp. 273-328. ate. Phil. Trans. Roy. Soc. London (B), 1976. The interpretation of osteodentine, with vol. 230, pp. 305-355. remarks on the dentition in the Devo- Patterson, C. nian dipnoan Griphognathus. Zool. 1968. The caudal skeleton in Lower Liassic Scripta, vol. 5, pp. 79-96. pholidophorid fishes. Bull. Brit. Mus. Owen, R. Nat. Hist. (Geol.), vol. 16, pp. 201-239. 1839. On a new species of the genus Lepido- 1973. Interrelationships of holosteans. In siren of Fitzinger and Natterer. Proc. Greenwood, P. H., R. S. Miles, and C. Linnean Soc. London, vol. 1, pp. 27- Patterson (eds.), Interrelationships of 32. fishes. London, Academic Press, pp. 1841. Description of the Lepidosiren annec- 233-305. tens. Trans. Linnean Soc. London, vol. 1977. Cartilage bones, dermal bones and 18, pp. 327-361. membrane bones, or the exoskeleton 1846. Lectures on the comparative anatomy versus the endoskeleton. In Andrews, and physiology of the vertebrate ani- S. M., R. S. Miles, and A. D. Walker mals. Pt. I. Fishes. (eds.), Problems in vertebrate evolu- London, Longman, tion. London, Academic Press, pp. 77- Brown, Green and Longmans, xi + 308 121. PP. Patterson, C., and D. E. Rosen 1859. Palaeontology. In Encyclopaedia Bri- 1977. Review of ichthyodectiform and other tannica. Edinburgh, vol. 17, pp. 91-176. Mesozoic teleost fishes and the theory 1866. The anatomy of vertebrates. Vol. I. and practice of classifying fossils. Bull. Fishes and reptiles. London, Long- Amer. Mus. Nat. Hist., vol. 158, pp. mans, Green and Co., xlii + 600 pp. 81-172. Panchen, A. L. Pearson, D. M. 1964. The cranial anatomy of two Coal Mea- [In press] Functional aspects of the integument 272 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

in polypterid fishes. Zool. Jour. Linnean 1936. The dipnoan cranial roof. Amer. Jour. Soc. Sci., vol. 32, pp. 241-256. Pearson, D. M., and T. S. Westoll 1937. The braincase of the Carboniferous 1979. The Devonian actinopterygian Cheiro-' crossopterygian Megalichthys nitidus. lepis Agassiz. Trans. Roy. Soc. Edin- Bull. Mus. Comp. Zool., vol. 82, no. 1, burgh, vol. 70, pp. 337-399. 73 pp. Pehrson, T. 1941. Notes on the crossopterygian hyoman- 1922. Some points in the cranial development dibular and braincase. Jour. Morph., of teleostomian fishes. Acta Zool. vol. 69, no. 1, pp. 141-160. Stockholm, vol. 3, pp. 1-63. 1945. Vertebrate paleontology. Second edi- 1947. Some new interpretations of the skull in tion. Univ. Chicago Press, 687 pp. Polypterus. Ibid., vol. 28, pp. 399-455. 1956. The early evolution of land vertebrates. Perks, A. M. Proc. Amer. Phil. Soc., vol. 100, pp. 1969. The neurohypophysis. In Hoar, W. S., 157-167. and D. J. Randall (eds.), Fish physiol- 1957. The appendicular skeleton of the Perm- ogy. Vol. 2. London and New York, ian embolomerous amphibian Archeria. Academic Press, pp. 112-206. Contrib. Mus. Paleo. Univ. Michigan, Pollard, H. B. vol. 13, no. 5, pp. 103-159. 1891. On the anatomy and phylogenetic po- 1966. Vertebrate paleontology. Third edition. sition ofPolypterus. Anat. Anz., vol. 6, Univ. Chicago Press, ix + 468 pp. pp. 338-344. 1968. Notes and comments on vertebrate pa- 1892. On the anatomy and phylogenetic po- leontology. Univ. Chicago Press, 304 sition of Polypterus. Zool. Jahrb., vol. PP. 5, pp. 387-428. Romer, A. S., and F. Byrne Presley, R., and F. L. D. Steel 1931. The pes ofDiadectes: notes on the primi- 1978. The pterygoid and ectopterygoid in tive tetrapod limb. Palaeobiology, vol. 4, mammals. Anat. Embryol., vol. 154, pp. 25-48. pp. 95-110. Rudebeck, B. Rackoff, J. S. 1945. Contributions to forebrain morphology 1976. The osteology ofSterropterygion (Cros- in Dipnoi. Acta Zool., vol. 26, pp. 9- sopterygii: Osteolepidae) and the origin 156. of tetrapod locomotion. Ph.D. thesis Save-Soderbergh, G. Yale Univ., 292 pp. 1932. Preliminary note on Devonian stegoce- 1980. The origin of the tetrapod limb and the phalians from East Greenland. Meddel. ancestry oftetrapods. In Panchen, A. L. Gr0nland, vol. 94, no. 7, pp. 1-107. (ed.), The terrestrial environment and 1933. The dermal bones of the head and the the origin of land vertebrates. London, lateral line system in Osteolepis macro- Academic Press, pp. 255-292. lepidotus Ag. with remarks on the ter- Ramanujam, S. G. M. minology of the lateral line system and 1929. The study of the development of the on the dermal bones of certain other vertebral column in teleosts, as shown crossopterygians. Nova Acta Roy. Soc. in the life-history of the . Proc. Sci. Uppsala, ser. 4, vol. 9, no. 2, pp. Zool. Soc. London, pp. 365-414. 1-130. Rayner, D. H. 1934. Some points of view concerning the 1951. On the cranial structure of an early pa- evolution of the vertebrates and the leoniscid, Kentuckia, gen. nov. Trans. classification of this group. Arkiv Roy. Soc. Edinburgh, vol. 62, pp. 53- Zool., vol. 26, no. 17, pp. 1-20. 83. 1935. On the dermal bones of the head in lab- Reynolds, S. H. yrinthodont stegocephalians and primi- 1913. The vertebrate skeleton. Second ed. tive Reptilia. Meddel. Gr0nland, vol. Cambridge Univ. Press, xvi + 535 pp. 98, no. 3, pp. 1-211. Romer, A. S. 1936. On the morphology of Triassic stego- 1924. Pectoral limb and shoulder girdle in fish cephalians from Spitzbergen, and the and tetrapods. Anat. Rec., vol. 27, no. interpretation of the endocranium in the 2, pp. 119-143. Labyrinthodonta. K. Svenska Vetensk. 1933. Vertebrate paleontology. First edition. Akad. Handl., ser. 3, vol. 16, no. 1, Univ. Chicago Press, 491 pp. pp. 1-181. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 273

1937. On the dermal skulls of Lyrocephalus, Palaeontographia Ital., vol. 65, pp. 60- Aphaneramma, and Benthosaurus, lab- 137. yrinthodonts from the Triassic of Spitz- 1970a. Folded teeth and the monophyletic ori- bergen and N. Russia. Bull. Geol. Inst. gin of tetrapods. Amer. Mus. Novitates, Uppsala, vol. 27, pp. 189-208. no. 2408, 10 pp. 1946. On the fossa hypophyseos and the at- 1970b. Die Histologie der Wirbelkorper der tachment of the retractor bulbi group in Dipnoer. Neues Jahrb. Geol. u. Palaeont. Sphenodon, Varanus, and Lacerta. Ar- Abhandl., vol. 135, no. 3, pp. 311-336. kiv Zool., vol. 38 (A), no. 11, pp. 1-24. 1977a. The origin of the tetrapod limb within Schaeffer, B. the rhipidistian fishes. In Hecht, M. K., 1965. The evolution of concepts related to the P. C. Goody, and B. M. Hecht (eds.), origin of the Amphibia. Syst. Zool., vol. Major patterns in vertebrate evolution, 14, pp. 115-118. New York, Plenum Press, pp. 541-544. 1967. Osteichthyan vertebrae. Zool. Jour. 1977b. Ausgangsform und Entwicklung der Linnean Soc., vol. 47, pp. 185-195. rhombischen Schuppen der Osteichthy- 1968. The origin and basic radiation of the es (Pisces). Palaeont. Zeitschr., vol. 51, Osteichthyes. In 0rvig, T. (ed.), Cur- pp. 152-168. rent problems in lower vertebrate phy- Semon, R. logeny. Stockholm, Almqvist and Wik- 1898. Die Entwickelung der paarigen Flossen sell, pp. 207-222. des Ceratodus forsteri. In Semon, R., 1975. Comments on the origin and basic ra- Zoologische Forschungsreisen in Aus- diation of the gnathostome fishes with tralien und dem Malayischen Archipel- particular reference to the feeding ago. I. Ceratodus, pt. 2. Jena, G. Fisch- mechanism. Colloques Internatl. Centre er Verlag, pp. 59-111. Natl. Rech. Sci., vol. 218, pp. 101-109. 1901. Uber das verwandtschaftsverhaltniss Schaeffer, B., and M. Williams der Dipnoer und Amphibien. Zool. 1977. Relationships of fossil and living elas- Anz., vol. 24, pp. 180-188. mobranchs. Amer. Zool., vol. 17, pp. Sewertzoff, A. N. 293-302. 1927. Der Ursprung der Quadrupeda. Palae- Schauinsland, H. H. ont. Zeitschr., vol. 8, pp. 75-95. 1906. Die Entwickelung der Wirbelsaule nebst Smith, M. M. Rippen und Brustbein. In Hertwig, 0. 1977. The microstructure of the dentition and (ed.), Handbuch der Vergleichenden dermal ornament of three dipnoans from und experimentellen Entwickelungs- the Devonian of western Australia: a lehre der Wirbeltiere. Jena, Gustav contribution towards dipnoan interrela- Fischer, vol. 3, pt. 2, pp. 339-572. tions and morphogenesis, growth and Schmalhausen, I. I. adaptation of the skeletal tissues. Phil. 1917. On the dermal bones of the shoulder- Trans. Roy. Soc. London (B), vol. 281, girdle of the Amphibia. Revue Zool. pp. 29-72. Russe, no. 2, pp. 84-110. 1979. Structure and histogenesis of tooth 1950. On the homology of the skull roof bones plates in inaequalis Owen of fishes and terrestrial vertebrates. considered in relation to the phylogeny Zool. Jour. Akad. Nauk SSSR, vol. 29, of post-Devonian dipnoans. Proc. Roy. no. 2, pp. 176-186. Soc. London (B), vol. 204, pp. 15-39. Stensi6, E. A. 1968. The origin of terrestrial vertebrates. 1918. Notes on a crossopterygian fish from New York, Academic Press, xxi + 314 the Upper Devonian of Spitzbergen. PP. Bull. Geol. Inst. Uppsala, vol. 16, pp. Schultze, H.-P. 115-124. 1969a. Griphognathus Gross, ein langschnau- 1921. Triassic fishes from Spitzbergen. Pt. I. ziger Dipnoer aus dem Oberdevon von Vienna, A. Holzhausen, 307 pp. Bergisch-Gladbach (Rheinisches Schie- 1922. Uber zwei Coelacanthiden aus dem fergebirge) und von Lettland. Geol. et Oberdevon von Wildungen. Palaeont. Paleont., vol. 3, pp. 21-79. Zeitschr., vol. 4, nos. 2/3, pp. 167-210. 1969b. Die Faltenzahne der rhipidistiiden Cros- 1925. Triassic fishes from Spitzbergen. Pt. II. sopterygier, der Tetrapoden und der K. Svenska Vetensk. Akad. Handl., Actinopterygien-Gattung Lepisosteus. ser. 3, vol. 2, no. 1, 261 pp. 274 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

1932. Triassic fishes from East Greenland col- In Andrews, S. M., R. S. Miles, and A. lected by the Danish Expeditions in D. Walker (eds.), Problems in verte- 1929-1931. Meddel. Gr0nland, vol. 83, brate evolution. London, Academic no. 3, pp. 1-305. Press, pp. 247-270. 1947. The sensory lines and dermal bones of Thomson, K. S., and K. S. W. Campbell the cheek in fishes and amphibians. K. 1971. The structure and relationships of the Svenska Vetensk. Akad. Handl., ser. 3, primitive dipnoan lungfish-Dipnorhyn- vol. 22, no. 3, pp. 1-70. chus sussmilchi (Etheridge). Bull. Pea- 1959. On the pectoral fin and shoulder girdle body Mus. Nat. Hist., no. 38, pp. vi + of the arthrodires. Ibid., ser. 4, vol. 8, 109. no. 4, pp. 1-229. Thomson, K. S., and P. P. Vaughn Strong, 0. S. 1968. Vertebral structure in Rhipidistia (Os- 1895. Cranial nerves of Amphibia. Jour. teichthyes, Crdssopterygii) with de- Morph., vol. 10, no. 1, pp. 101-230. scription of a new Permian genus. Pos- Szarski, H. tilla (Yale Peabody Mus.), no. 127, pp. 1977. Sarcopterygii and the origin of tetra- 1-19. pods. In Hecht, M. K., P. C. Goody, Trewavas, E., E. I. White, N. B. Marshall, and and B. M. Hecht (eds.), Major patterns D. W. Tucker in vertebrate evolution. New York, Ple- 1955. Reply to A. S. Romer. Nature, London, num Press, pp. 517-540. vol. 176, pp. 126-127. Taverne, L. Vogt, C. 1979. Osteologie, phylogenese et systema- 1845. Translator's comments on Muller's pa- tique des teleosteens fossiles et actuels per. Ann. Sci. Nat. (Zool.), ser. 3, vol. du super-ordre des Osteoglosso- 4, no. 3, p. 31. morphes. 3. Evolution des structures 1851. Zoologische Briefe. Frankfurt-a.-M., osteologiques et conclusions generales vol. 2, 640 pp. relatives a la phylogenese et a la syste- Vorobyeva, E. I. matique du super-ordre. Mem. Acad. 1962. Rhizodontid crossopterygians of the Roy. Belg. Cl. Sci., 8°, ser. 2, vol. 43, main Devonian field in the USSR. Tru- no. 3, pp. 1-168. dy Paleont. Inst. SSSR, vol. 94, pp. 1- Thomson, K. S. 139. [In Russian.] 1964. The comparative anatomy of the snout 1973. Einige Besonderheiten im Schadelbau in rhipidistian fishes. Bull. Mus. Comp. von Panderichthys rhombolepis (Gross) Zool., vol. 131, pp. 313-357. (Pisces, Crossopterygii). Palaeonto- 1965. The nasal apparatus in Dipnoi, with spe- graphica, vol. 142A, pp. 221-229. cial reference to Protopterus. Proc. 1974. On the formation of tetrapod features in Zool. Soc. London, vol. 145, pp. 207- crossopterygians. Paleont. Zh., no. 2, 238. pp. 110-116. 1966. Intracranial mobility in the coelacanth. 1975. Bemerkungen zu Panderichthys rhom- Science, vol. 153, pp. 999-1000. bolepis (Gross) aus Lode im Lettland 1968. A critical review of the diphyletic the- (-Schichten, Oberdevon). Neues ory of rhipidistian-amphibian relation- Jahrb. Geol. Palaeont. Mh., pp. 315- ships. In 0rvig, T. (ed.), Current prob- 320. lems in lower vertebrate phylogeny. 1977a. Morfologiya i osobennosti evolyutsii Stockholm, Almqvist and Wiksell, pp. kisteperykh ryb. Trudy Paleont. Inst. 285-305. SSSR, vol. 163, pp. 1-240. 1972. New evidence on the evolution of the 1977b. Filogeneticheskie svazi osteolepiform- paired fins of Rhipidistia and the origin nykh Crossopterygii i ikh polozhenie v of the tetrapod limb, with description of sisteme. In Ocherki po filogenii i siste- a new genus of Osteolepidae. Postilla matike iskopaemykh ryb i beschelyustt- (Yale Peabody Mus.), no. 157, 7 pp. nykh. Paleont. Inst. Akad. Nauk SSSR, 1975. On the biology of cosmine. Bull. Pea- pp. 71-88. body Mus. Nat. Hist., vol. 40, pp. 1-59. 1977c. Evolutionary modifications of the teeth 1977. On the individual history of cosmine structure in the Palaeozoic Crossopte- and a possible electroreceptive function rygii. Jour. Paleont. Soc. India, vol. 20, of the pore-canal system in fossil fishes. pp. 16-20. 1981 ROSEN ET AL.: LUNGFISHES AND TETRAPODS 275

1980. Observations on two rhipidistian fishes toral fins of ostracoderms and early fish- from the Upper Devonian of Lode, Lat- es. In Westoll, T. S. (ed.), Studies on via (Lode). Zool. Jour. Linnean Soc., fossil vertebrates. Univ. London Press, vol. 70, no. 2, pp. 191-201. pp. 180-211. Wahlert, G., von 1958b. The origin of continental vertebrate fau- 1968. Latimeria und die Geschichte der Wir- nas. Trans. Geol. Soc. Glasgow, vol. belthiere. Eine evolutionsbiologische 23, pp. 79-105. Untersuchung. Stuttgart, G. Fischer 1961. A crucial stage in vertebrate evolution: Verlag, vii + 125 pp. fish to land animal. Proc. Roy. Inst. Watson, D. M. S. Great Britain, vol. 38, pp. 600-618. 1912. The larger Coal Measure Amphibia. 1979. Devonian fish biostratigraphy. In The Mem. Proc. Manchester Lit. Phil. Soc., Devonian system. Spec. Papers in Pa- vol. 57, pp. 1-14. laeontology, no. 23, pp. 341-353. 1913. On the primitive tetrapod limb. Anat. White, E. I. Anz., vol. 44, nos. 1/2, pp. 24-27. 1962. A dipnoan from the Assise de Mazy of 1919. The structure, evolution and origin of Hingeon. Bull. Inst. Roy. Sci. Nat. Bel- the Amphibia. The "Orders" Rhachi- gique, vol. 38, no. 50, pp. 1-8. tomi and Stereospondyli. Phil. Trans. 1965. The head of Dipterus valenciennesi Roy. Soc. London (B), vol. 209, pp. 1- Sedgwick and Murchison. Bull. Brit. 73. Mus. Nat. Hist. (Geol.), vol. 11, pp. 1- 1925. The structure of certain palaeoniscids 45. and the relationships of that group with Whiting, H. P., and Q. Bone other bony fish. Proc. Zool. Soc. Lon- 1980. Ciliary cells in the epidermis of the lar- don, 1925, pp. 815-870. val Australian dipnoan Neoceratodus. 1926. The evolution and origin of the Am- Zool. Jour. Linnean Soc., vol. 68, pp. phibia. Phil. Trans. Roy. Soc. London 125-137. (B), vol. 214, pp. 189-257. Wiedersheim, R. E. E. 1940. The origin of frogs. Trans. Roy. Soc. 1892. Das Gliedmassenskelet der Wirbel- Edinburgh, vol. 60, pp. 195-231. thiere, mit besonderer Berucksichtigung Watson, D. M. S., and H. Day des Schulter- und Beckengurtels bei 1916. Notes on some Palaeozoic fishes. Mem. Fischen, Amphibien und Reptilien. Proc. Manchester Lit. Phil. Soc., vol. Jena, 266 pp. 60, pp. 1-52. Wiley, E. 0. Watson, D. M. S., and E. L. Gill 1979. Ventral gill arch muscles and the inter- 1923. The structure of certain Palaeozoic Dip- relationships of gnathostomes, with a noi. Jour. Linnean Soc. (Zool.), vol. 35, new classification of the Vertebrata. pp. 163-216. Zool. Jour. Linnean Soc., vol. 67, pp. Westoll, T. S. 149-179. 1936. On the structure of the dermal ethmoid Woodward, A. S. shield of Osteolepis. Geol. Mag., vol. 1891. Catalogue of the fossil fishes in the Brit- 73, pp. 157-171. ish Museum (Natural History), vol. 2. 1937. On a specimen of Eusthenopteron from London, British Museum (Natural His- the Old Red Sandstone of Scotland. tory), xliv + 567 pp. Ibid., vol. 74, pp. 507-524. 1898. Outlines of vertebrate palaeontology for 1938. Ancestry of the tetrapods. Nature, Lon- students of zoology. Cambridge Univ. don, vol. 141, pp. 127-128. Press, xxiv + 470 pp. 1943. The origin of the tetrapods. Biol. Rev., 1922. Observations on crossopterygians and vol. 18, pp. 78-98. arthrodiran fishes. Proc. Linnean Soc. 1949. On the evolution of the Dipnoi. In Jep- London, vol. 134, pp. 27-36. sen, G. L., G. G. Simpson, and E. Mayr Zangerl, R. (eds.), Genetics, paleontology and evo- [In press] Elasmobranchii. In Schultze, H.-P. lution. Princeton Univ. Press, pp. 121- (ed.), Handbook of paleoichthyology. 184. Vol. 3. Stuttgart, Gustav Fischer Verlag. 1958a. The lateral fin-fold theory and the pec-