The identification of the dermal bones of the head

J. L~)z)z. SOC. (ZOOZ.), 47, 311, pp. 231-239 With 5 figures Printed in Cheat Britoin October, 1965 The identification of the dermal bones of the head BY F. R. PAKRINGTON F.R.S. University Musezr of Zoology, Cambridge The discovery of the ichthyostegid Amphibia in Upper Devonian rocks by Siive-Soderbergh (1932) introduced further difficultics into the already complex problems of the dermal bones of the skull roof. For some years previously ideas about the origin of the tetrapods had been dominated by Watson s (1926) Croonian Lecture in which he had demonstrated beyond reasonable doubt that the crossopterygian fishes and not the Dipnoi were their ancestors, and had attempted t,o show that many of the features of the Carboniferous labyrinthodonts were a direct inheritance from these fishes. It was to be expected, therefore, that any Amphibia from t he Upper Devonian would be intermediate in their structures between the Middle Devonian osteolepids and the Carboniferous lab,yrinthodonts, but when discovered the ichthyostegids did not conform at all well to this expectation. While their skulls showed some very primitive features which might have been expected, the patternof the dermal bones did not conform to plan, for these new animals had lost, it seemed, the intertemporals, bones found in both the osteolepids and nearly all early labyrinthodont,s, and had a single postparietal bone in place of the paired bones of all other early Amphibia. The osteolepid skull had many more bones than tjhese earliest Amphibia. In an attempt to reconcile the resulting difficulties Save-Soderbergh developed an idea which had been introduced by Stensio. Previously when bones appeared to be missing from skulls it was assumed that they had been lost; but Stensio assumed that t>hey had fused with neighbouring bones. This assumption of extensive bone fusions was applied by Siive-Soderbergh who put forward a theoretical pattern of bones which, by different fusions, could have given rise to those of bot li osteolepid fishes and a hypothetical tetrapod ancestor. From the latter, by further different fusions, the patterns found in ichthyostegids and labyrint,hodonts could be achieved (Save-Soderbergh, 1935, fig. 45). These assumptions of extensive bone fusions in early vertebrates have been maintained by Jarvik in spite of serious objections raised by other writers. Furthermore both Siive-Soderbergh and Jarvik accepted the once widely adoptcd assumption of the constancy of t8he relations of lateral-line canals to dermal bones and this dogma has also been maintained in spite of serious, unanswered objections. The early confusions were largely cleared up by Westoll who disputed SGve-Soderbergh s identifications, first on grounds of general topography (Westoll, 1936, 1937 a, 1938) but eventually by relating the midline dermal bones to the parts of the underlying neurocranium and showing that the so-called frontals and parietals of crossopterygian fishes had much the same relations as tlic parietals and postparietals of early tetrapods (Westoll, 1943). He described Elpistosteye watsoni, a skull table from the Upper Devonian, in which the proportions of the parietal and postparietal bones approached those of the early tet,rapods and showed that the lat,tcr bones diminished in size from the crossopt>erygian fishes to the Amphibia and the earliest reptiles with a corresponding backward movement of the parietals (1938). This obscrvation received support when Jarvik (1952) described the osteolepid fish Eusthenodou ~oungsjoi, a form in which the parietals are distinctly longer than the postparietals. This solution was immediately convincing to many workers (e.g. Romcr, 1941, 1945) but Jarvik (1952) made a direct comparison between the skull of an osteolepid fish and

232 F. R. PARRINGTON, F.R.S. that of an embryo rabbit and claimed that the original identifications of the crossopterygian bones were correct. However he made no comparison with early tetrapods or with other mammals. Parrington (1949, 1950, 195.6) discussed the basic assumptions which the Swedish workers have used when making their identifications of dermal bones and claimed that neither the assumption of the constancy of lateral-line canals nor the assumption of exten- Fig. 1. Ewthenoptcroti foordi x 1. Neurocranium and investing dermal bones. Data from Jarvik. IT, intertemporal; M.EXT, median extrascapular; N, nasal; P, parietal; PF, postfrontal; PP, postparietal; ST, supratemporal; T, tabular; bp.pr, basipterygoid process; why, region of attachment of the ventral foot of the hyomandibular (site of fenestra ovalis); 11, V, VII, foramina for the optic, trigeminal and facial nerves. sive bone fusions affords the best explanations of the recorded observations and the alternative views advanced have received support from Pritchard, Scott & Girgis (1956) and, from White (1965). They have not been discussed by Swedish workers. In 1965 Thomson compared the neurocrania and related dermal bones of Osteolepis and the labyrinthodont Edops and showed that Westoll's claims were fully justified. However in a lecture delivered in Paris in 1966 Jarvik returned to the matter. He now made a comparison between an osteolepid fish and two Triassic labyrinthodonts and reached the same equations that Westoll had made, but he maintained that the disputed bones were

Identification of dernial bones of head 233 the frontal and parietal by reason of his comparison of the osteolepid with a rabbit embryo. He concluded that the ' Reptiliomorpha ', i t. the amniotes and their amphibian ancestors, the seymouriamorphs and the anthracosaurs, had the true frontals and parietals of mammals and must have evolved independently from the temnospondylous labprinthodonts. Again, he made no examination of the relations of the neurocrania of the early ' Reptiliomorpha '. It cannot be too emphatically denied that Westoll identified the parietal bone in F c--. P p~ - n- PP ' bpt.pr Fig. 2. Skulls of Edops (A, C x f) and I'rtkrrogyriuus (B, D x 4) to show the relations of the dermal bones to the neurocrania. Data for Edops from Romer & Witter; PaZaeogyri,iiis from Panchen. F, Frontal; fen.oc., fenestra ovalis. Otlic~ letters as before.

234 F. R. PARRINGTON, F.R.S. osteolepid fishes by reason of the pineal foramen. It has long been known that this foramen can vary considerably in its position, and it is also known that to identify structures by reference to a single point is dangerous (see e.g. Parrington & Westoll, 1940: 321). What Westoll did was to point out that the so-called frontal bone of osteolepid fishes lay above that part of the neurocranium which formed the optic foramen, the pituitary fossa, and the basipterygoid process as well as the pineal foramen (Fig. 1) and that these four structures were covered by the parietal bone in early tetrapods (Westoll, 1943). He then went on to point out that, the so-called parietals of t hese fishes covered t,hat part of F P pin PP - A F P, pinp- I PP B F P C I \, ; fen.ov I Fig. 3. Neurocrania of A, Kotlussia x 2 ; B, Seytnouria x 14; C, Captorhinus x 1 to show the relations of the frontal, parietal and postparietal bones. Data for A from Bystrow, B from White, C from Price. SPH, sphenethmoid; pin, pineal foramen. Other letters as before. the neurocranium through which the trigeminal and facial nerves emerged and to which the otic capsule, with the hyomandibular, was attached, and he claimed that in early tetrapods it was the postparietals which had these relations. The truth of these claims is well demonstrated by Thomson s (1965) figure in which the brain cases of Eusthenopteron and the early temnospondylous labyrinthodont Edops and their relations to the dermal bones are compared. Examination of Romer & Witter s data of Edops suggests, however, that in this amphibian the parietal bone extends slightly more posteriorly than it did in the osteolepid fishes (where it is confined posteriorly by the hinge) and lies above the foramina for the exit of the trigeminal and facial nerves (Fig. 2). It appears that Jarvik accepts such a comparison, agreeing that the pineal bearing bones of

Idelztification of dermal bones of head 235 the two animals are homologous, as are tlie midline bones immediately behind them. He denies that the bones are the mammalian parietals and postparietals. If the condition in the anthracosaur amphibian Palaeogyrinus is now examined it is found to be almost identical with that of Edops, differing only in that the parietal extends slightly more posteriorly to a point clearly bchind the foramina for the trigeminal and facial nerves. It is not to be expected that such bones would have relations which are absolutely identical. Thus the metoposaur Eupelor fraasijonesi has postparietals which are about 50% longer than those of Eupelor browni and so they are unlikely to have precisely the same relations to the brain case (Colbert & Imbric, 1956, fig. 8). That the frontals, parietals and postparietals of the two labyrinthodonts are homologous bones seems an inescapable conclusion (Fig. 2). If these bones are not true homologues there seems no test by which homologies can be decided. It is generally accepted that the next stage in the evolution of the amniotes is represented by the seymouriamorphs. Here the condition can be examined in Seymouria itself and also in Kotlassia (Fig. 3). The work of Bystrow (1944) showed that in Kothssia the pineal bearing parietal lies above the optic foramen, the pituitary fossa, and the basipterygoid processes (just as did the frontal of the osteolepid fishes) and, as in the anthracosaur, it extends posteriorly above the exits of the trigeminal and facial nerves. The condition in Seymouria is much the same but here the parietal extends still further covering part of the large fenestra ovalis. However, the ear structures are somewhat aberrant in these forms (White, 1939). In both seymouriamorphs much of the otic capsule was still covered by the postparietals. The condition in the earliest amniotes can be represented by the cotylosaur Captorhinus (Fig. 3C). Here there are still more marked changes. The frontal bone has developed posteriorly and now extends backwards over the basipterygoid processes, just about reaching the position of the pituitary fossa, and it must have lain above the optic foramen. The parietal extends from a position above the pituitary fossa almost to the back of the skull, covering the exits for the trigcrninal and facial nerves and also the otic capsules. The postparietals have moved down on to tlie occiput and cover only a small part of the back of the neurocranium. That the frontals, parietals and postparietals (or interparietals) of this form are properly equated with the same bones of synapsid reptiles and mammals has been demonstrated beyond reasonable doubt by the work of Broom, Watson, Romer, Olson, Brink and many others.,4 general picture of the condition of the midline bones in these animals is given by Parrington & Westoll (1940, fig. 10). The palaeontological evidence is clear and ample and shows beyond doubt that Westoll s identifications of the so-called frontal and parietal bones of osteolepid fishes as the tetrapod parietals and postparietals were correct. What then has gone wrong? Jarvik rests his case for the identification of the disputed bones of the osteolepids on his comparison with the embryo of Oryctolagus ( Lepus ), the rabbit. Many will think this strange since it paps no attention to the conditions in the numerous mammalian forebears which enable the evolution of the group to be followed in outline. And it is also strange that no consideration is given to the variable conditions found among mammalian embryos in general. Thus Jarvik found that in the rabbit embryo the frontal bone lies above the optic foramen, the pituitary fossa (i.e. the hypophysis), and the basipterygoid processes as did the pineal bearing bone which he calls the frontal in the osteolepids, but which, it has been shown, gradually shifted until it had the clearly recognizable relations of the parietal of amniotes. Since, therefore, there is a clash of opinion, it is surely worth examining the condition in mammals other than the rabhit, which, it may be noted, has a large frontal and a modest parietal. If the condition in such an insectivore as Talpa, the mole, is examined, it will be found that the relations of the critical bones are very different. Here the frontal is small and the parietal is large and it is the latter bone which overlies the optic foramen, the basipterygoid process and the pituitary fossa (Fig. 4). Had Jarvik chosen the

236 F. R. PARRINGTON, F.R.S. insectivore as his type and not the rabbit he could not have reached the conclusion he did. Plainly the evidence of mammalian embryos is variable and it is noteworthy that the condition of the parietal in the insectivore embryo, a primitive placental, is not greatly different from that found in early amniotes. To use the variable condition found among mammal embryos in place of the direct evidence of palaeontology seems quite unreasonable. Nor does consideration of the developing frog skull alter the position for, as Jarvik shows, the frontal and parietal bones of this form have the same relations as the frontals and parietals of the early labyrinthodont Amphibia. The identifications of frontal and parietal bones accepted here have been made on the evidence of their general relations to other structures, and modifications of these relations are accepted as the consequences of gradual processes for which, it is claimed, there is satisfactory palaeontological evidence. What other tests of identity can be made? Fig. 4. Developing skulls of A, a young rabbit, Oryetolagus (data from Voit) and B, a young mole, Talpa (data from Fischer) to show the different relation of the parietal bones to the neurocrania. Not to scale. hyp,hypophysis. Other letters as before. One other test which Jarvik applies when making bone identifications is the dogma of the supposed constancy of the relations of the lateral-line canals. This view was introduced by Allis (1898) and it was adopted by most students of early fishes during the thirties. But experimental investigations threw doubt on the theory (Moy-Thomas, 1941) and further examination of the basis on which it rested showed that it is untenable. Allis started by identifying the dermal bones of fish embryos by their topography, and reached the conclusion that the lateral-line canals always had the same relations to these bones. He assumed that he had established a law (from which there should be no exceptions) rather than shown that, as would be expected, the lateral-line system followed the general rule of constancy of morphological relations, to which rule exceptions are to be expected. Allis therefore proclaimed that ' a bone or a part of a bone developed in any particular fish in relation to a particular part of the lateral-line system is always the homologue of the bone, or part of a bone, developed in relation to the same part of the lateral-line system in any

Identi catiot~ of der tnal bones of head 237 other fish or animal (Allis, 1898). But his rcasoning is unsound. Assuming that hc could identify bones on their topography he clainicd to show completely fixed relations of the canals When this idea nas applied, e.g to the bones of the skull table of osteolepid and holoptychiid fishes, the result was to deny that the apparently identical midline parietals of these two forms mere the same. 111 the osteolepid the lateral line canal passes throiigli the two laterally placed boneb, t lit. tabular and supratemporal, but in tlie holoptychiids it passes from the laterally placcd tabular on to the midline parietal 11hich, if canal relations are absolutely constant, should therefore be called the supratemporal (Westoll, 1937b). But the midline bones have the same essential topography 111 the two formi and if topography is not a reliable test of identity then the constancy of lateralline canals cannot be shown in the first place The conclusion destroys the premiw. In fact there are no grounds whatever on which it can be asserted that, unlike all other organ systems, the relations of the lateral line system are governed by an inviolable law and not by the general rule of constancy of morphological relations. That lateral-line canals are in fact movable has been claimed by Parrington (1950) and this has been confirmed quite recently by White (1965) in a most thorough study of the hcad of the frequently discussed lung fish, Dipterus valenciennesi. Thcre is one final matter \I hich must be considered; Jarvik s assumptions of extensive bone fusions. The theory of multiple bont. fusions could originally be justified on two grouiids. First, dermal bones are knomn to fuse in modern animals (e.g. the frontoparietal of tlic frog) and there are occasional indications of double centres of ossification in the single bones of some fossils (e.g. the postorbital-intertemporal of the labyrinthodont Bwiuasauius), and the theory was based, therefore, on a known occurrence. Second, the interpretation of the bone patterns of the hliulls of early fishes and their cquations with those of early tetrapods liad proved a difficult and complex matter and this justihed the introduction of new methods in an attempt to reconcile the difficulties. S~~ve-Soderbergh s attempted solution was a bold one which mrrited serious consideration. Against the theory of multiple bone fiisioiis are a series of facts. First, there is evidence that bones sometimes become reduced until they are very small, and the supposition that they may continue the reduction process to its final limit is logical. To suppose that dermal bones may be reduced in size but never lost is unreasonable. Thus unless extraneous assumptions are made, the bone A (Forhter C ooper notation) of early lung fishe..; is seen to be reduced in size and lost. It is a rensonably well-developed bone in Dipmrhynchus and Dipterus from the Middle Devoniaii, is smaller in Scaumenacia from the Upper Devonian, and is missing, apparently, from the ( arboniferous forms Sagenodus, Ctenodus and llronemus. But it survives as a very sinall bone in the early Permian form Conchopoma (Westoll, 1949). And in labyrinthodonts 1 he iiitertemporal is usually well developed in early forms, is seen to be small in e g. Losoirir~m from the Carboniferous (Watson, 1926) and Edop from the Lower Permian (Romer S. Wittcr, 1942) and is commonly missing. particularly in later forms. Xrxt, tlie supposition by Save-Soderbergli that the parietal bone of typical oqtcolepids n as the product of the fusion of two parietal bones became gravely suspect whenlater it was shonn that these bones developed froin single centres (Jarvik, 1948). To persist in Save-SGderbergh s original interpretation then required the further assumption that the size and form of dermal bones are intrinsic properties, and experimental evidcncr denies that this is so. Thus, when Tatarko excisccl the subopercular bone of the carp tlie neighbouring opercular and interopercnlar bones took over the vacated territory (Tatarlio, 1934). Girgis & Pritchard (19*73, 1938) liavc. shown very similar happenings in the skull of the rat. The extirpation (and so presumabl? the loss) of a dermal bone is followed by the capture of the vacated territory by the ncy$boixring bones. Parrington has shown that if simple assumptions concerning bone gron t 11 are accepted, the changing bone patterns of early fishes and early tetrapods are easilj arcounted for assuming loss of bones rather than bone furions Thus. n hen the anterior pat ictals of the osteolrpids arc> lost on one or both

238 F. R. PARRINGTON, F.R.S. sides, the vacated territory is largely captured by the posterior parietals but the neighbouring supratemporal bones crowd the growing posterior parietals, taking over part of the vacated territory as would be expected. And when in labyrinthodonts the intertemporal bone is lost, the vacated territory appears in general to have been taken over by both the postfrontal and the supratemporal; it does not appear that the intertemporal has fused with the supratemporal (Parrington, 1956). Finally the application of the two theories, that of the constancy of the relations of lateral-line canals and that of multiple bone fusions, results in conclusions which are often extlwnely difficult to accept. For example according to orthodox nomenclature the three bones which protect the otico-occipital component ofthe neurocranium of both osteolepid and holoptychiid fishes are the postparietals, the tabulars and the supratemporals (Fig. 5). Fig. 5. Dermal bones of t,he otico-occipital component of the neurocrania of A, Osteokpis and B Holoplychius. The courses of the lateral-line canals are shown on the left. Centres of ossification of the postparietals lie at the junctions of the pit-lines. Not to scale. Data from Jarvik. The two forms differ in that the holoptychiids are broader fishes and have the postparietals ossifying from centres near the middle ofthe bones whereas those of osteolepid fishes ossify from a centre near the back. There is a corresponding difference in the suture pattern which is to be expected (Parrington, 1956, fig. 4A, C). Also the lateral-line canal, which passes through the tabular and supratemporal bones in the osteolepid, passes from the tabular to the postparietal in the holoptychiids, apparently having been captured by reason of the anterior position of the centre of ossification. To satisfy his basic theories Jarvik calls the tabular the supratemporal, and the postparietal the parietal-intertemporal. This leaves the supratemporal unidentified and the dilemma is overcome by coining a new name for it. It is called the prespiracular plate (Jarvik, 1950). The conclusions which must be draxn from the foregoing discussion are that Westoll correctly identified the parietal and postparietal bones of the primitive crossopterygisn skull as the bones previously called the frontal and parietal, and that his methods for doing so were sound. And the assumptions adopted by various other workers in their attempts to elucidate the details of the evolution of the bony skull, which have resulted in wrong identifications, are unsound and should be abandoned. I wish to record my thanks to Mr R. D. Norman for the preparation of the figures REFERENCES ALLIS, E. P., 1898. On the morphology of certain bones of the cheek and snout of Ainia ca2vn.j. Morph., 14: 425. BYSTROW, A. P., 1944. Kotlnssia prima Amalitzky. Bull. geol. SOC. Ant., 55: 379.

Identification of derm,al bones of head 239 COLBERT, E. H. & IMBRIE, J., 1956. Triassic inetoposaurid amphibians. Bull. Am. MZM. nat. Hist., 110: 405. GIRGIS, F. G. & PRITCHARD, J. J., 1955. Morpliological status of cranial sutures. J. Anat. Lod., 89: 577. GIRGIS, P. G. & PRITCHARD, J. J., 1958. Effects of skull da,mage on the developiiient of sutural patterns in the rat. J. dizat. Loitd., 92: 39. JAKVIP, E., 1948. On the morphology and taxoiloniy of the Middle Devonian osteolepid fishes of Scot,laiid. K. svenska Vetenskdkad. Hard., 25 : 1. JARVIK, E., 1950. Middle Devonian vertebmtex from Canning Land and Wegener s Halvii (East Greenland). Meddr. Grenland, 96: 1. JARVIIC, E., 1952. On the fish-like tail in the ichthyost,egid stegocephalians. Meddr. Gr0nlnnd, 114: 1. JARVIG, E., 1967. The homologies of frontal and parietal bones in fishes and tetrapods. C olloyues irbt. C enf. mtts. Rech. Scient., 163. MOY-THOMAS, J. A., 1941. Development of the froucal bones of the rainbow trout. Xoture, Lod., 147: 681. PARRINGTON, F. R., 1949. A theory of the relations of lateral lines to dermal bones. Proc. 3001. Sot. Lond., 119: 65. PARRINGTON, F. R., 1950. The skull of Dipterus. Ann. Mag. nat. Hist. (12) 3: 534. PARRIKGTON, F. R., 1956. The patterns of dermal hones in primitive vertebrates. f roc. zool. S oc. Lond., 127: 389. PARRINGTON, F. R. & WESTOLL, T. S., 1940. On the evolution of the mammalian palate. Phil. Trans. R. Soc. B, 230: 305. PRITCIIARD, J. J., SCOTT, J. H. & GIRGIS, F. G., 1956. The structure and development of cranial arid facial sutures. J. And., 90, 73. ROWER, A. S., 1941. Notes on the crossopterygiau Ii~wmrtndibular and braincsse. J. Morph., 69: 141. ROMER, A. S., 1945. Vertebrute Paleontology. Cliirago. ROMER, A. S. & WITTER, R. V., 1942. Edops, a primitive rhachitomous amphibian. J. Geol. 50: 925. SAVE-S~DERBERGH, G., 1932. Preliminary note on Devonian stegocephaliana from East Greenland. Meddr. Gronla,nd, 94: 1. SAVE-SODERBERGH, G., 1935. On the dermal bones of the head in labyrinthodont stegocephalians and primitive Reptilia. Meddr. Gre~nland, 98 : 1. TATARKO, K., 1934. Restitution des Kiemenderkels des Karpfens. 2002. Jb. (.4llq. Zoo/.) 53; 4ti I. THOMSON, K. S., 1965. The endocranium and associated structures in the Middle Devonian rliipidistian fish Osteokpis. Proc. Linn. SOC. Lond., 176: 181. WATSON, D. 1LI. S., 1926. Croonian Lecture. The evolution and origin of t,he Amphibia. Phil. Tmns. R. Soc. U, 214: 189. WESTOLL, T. S., 1936. On the structure of the derrnd ethmoid shield of Osteolepis. Geol. M/q., 73: 157. WESTOLL, T. 8.. 1937n. On a specimen of Euslheuopteron from the Old Red Sandstone of Scotland. Geol. Mag., 74: 507. WESTOLL, T. S., 1937b. The Old Red Sandstone fishes of the north of Scotland, particularl>- of Orkney and Shetland. PTOC. Geol. Ass., 48: 13. WESTOLL, T. S., 1938. Ancestry of the tetrapods. Snture, Lond., 141: 127. WESTOLL, T. S., 1943. The origin of tetrapods. Hiol. Reu., 18: 78. WESTOLL, T. S., 1949. On the evolution ofthe Dipnoi. InGenetics,Pa~eontoloyynnrlEuoZutio~i. Princeton. WHITE, E. I., 1965. The head of Dipterus tdeitcietttiesi Sedgwick and Murchison. Bull. Br. Mtts. nat. Hist. (Geol.), 11: 1. WHITE, T. E., 1939. Osteology of Aeyninlrritr hr!ylowttsis Rroili. Bull. Mus. comp. Zool., 85. 325.