Peabody Museum of Natural History Yale University New Haven, CT 06511 Postilla umber ise 15 February 1982 An Early Triassic Hybodont Shark from Northern Madagascar Keith Stewart Thomson ita, a Sea (Received 2 April 1981) Abstract Material of the upper and lower jaws, together with teeth and other remains, of a Triassic hybodont shark from Madagascar is tentatively referred to the genus Acrodus. The material offers new evidence concern- ing the jaw suspension in hybodont sharks and its significance in the evolution of Elasmobranchii. Key Words Elasmobranchii, Hybodonti, jaw suspen- sion, Acrodus, Madagascar, Triassic. Introduction The evolutionary history and relationships of the cartilaginous fishes comprise one of the less known and more intractable areas of study in vertebrate paleontology. The Problems stem inmost part from the nature of the skeleton which does not (except for the dentition and spines) lend itself to pre- Servation in fossil form. Thus, whereas Many taxa of fossil chondrichthyan fishes have been described on the basis of minute differences in dental structure, the number Of taxa that are known from other cranial or Postcranial skeletal remains is frustratingly © Copyright 1982 by the Peabody Museum of Natu- ral History, Yale University. All rights reserved. No Part of this publication, except brief quotations for Scholarly purposes, may be reproduced without the written permission of the Director, Peabody Museum of Natural History. small. The present contribution.offers a de- scription of new cranial.material of a Trias- sic hybodont shark and a discussion of the evolution of certain features of head anato- my in the Elasmobranchii (sharks, skates and rays) and their immediate fossil rela- tives within the cartilaginous fishes. Two major models have been proposed for elasmobranch relationships. In a seminal work, Schaeffer (1967) synthesized availa- ble information on living and fossil forms into a three-part horizontal classification, recognizing three grades—“cladodont" (es- sentially Paleozoic), “hybodont” (essentially Mesozoic, and “modern level” elasmo- branchs. This rational organization was fol- lowed by a new surge of interest in the group, with descriptions of new taxa and new analyses of relationships eventually leading to the second model. Maisey (1975, see also Campagno, 1977) proposed a more cladal classification, realigning the “hybodont” sharks into two vertical assem- blages—“hybodontiform” (for example, Tristychius, Hybodus, Acrodus, Astera- canthus, Lissodus and Lonchidion) and “ctenacanthiform” (including Ctena- canthus, Spenacanthus, Goodrichthys, Nemacanthus). The ctenacanths were then linked formally with the modern level sharks (“euselachiforms”) while the hybo- donts, as thereby restricted, were removed from any relationship with modern sharks. Compagno (1977) further has reorganized schemes of relationships among the modern sharks and the three apparently primitive groups— Heterodontus, Chlamy- doselache and the hexanchoids—which previously had been thought to be indepen- dent relics of hybodont radiations, are now more securely incorporated into the radia- tion of euselachians. i Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 A nomenclatural note must be added here. Maisey (1975) uses the term “eusela- chian” for the ctenacanths plus modern sharks, skates and rays, whereas Compagno prefers the original use of Regan (1906) in which the ctenacanths plus modern forms are termed the “neoselachians” and the term “euselachian” is restricted to the modern level radiations which are consid- ered to be monophyletic. This latter use will be followed in the present paper. These two models of elasmobranch rela- tionships have had a great heuristic value in focusing attention on the important issues. Compagno’s work (1973, 1977) has concentrated upon the living groups and their immediate fossil relatives. Zangerl (1973), Zangerl and Williams (1975), Schaeffer and Williams (1977) and Schaef- fer (in preparation), inter alia, have brought important new information concerning the complex radiations of Paleozoic elasmo- branchs. Maisey (1975, 1976, 1977) and Dick (1978) have restudied some of the Mesozoic hybodont and ctenacanth mate- rials. Dick (1978) has also questioned the ctenacanth/hybodont separation, leaving this question still to be resolved. Much work remains to be done. Not only is there little solid information that helps assign relationships within and among the various groupings, the validity of current groupings still remains to be tested. In the present work, new material is described of the Trias- sic hybodont Acrodus and features of the evolution of the elasmobranch palate are discussed. Description In 1961, Professor Bernard Kummel of the Museum of Comparative Zoology (MCZ), Harvard University, made extensive collec- tions from the famous nodule-bearing beds of the Early Triassic of Northern Madagas- car (see, for example, Piveteau, 1934, and Lehman, 1952). Among the material he col- lected was a single largish nodule (MCZ 13432, Ambilobe Bay Locality) that, on preparation, revealed the presence of the first shark material (except for scraps of denticles) recorded from Madagascar. The specimen Is preserved, as are all such nodules, in part and counterpart with the calcified material almost totally removed by solution, leaving a natural cast of the re- mains (Figs. 1 and 2). The upper and lower jaws and dental barriers of the left side, part of the right mandible, two (?) ceratohyals, a fragment of a possible hyomandibular and an indistinct indication of the posterior por- tion of the braincase are preserved and have been developed by very careful prepa- ration, further revealing the natural cast, fol- lowed by casting in various plastics. As the teeth in this specimen are com- parable with other teeth from around the world usually ascribed to the genus Acrodus Agassiz, the dentition of the new Madagascar specimen will be described first. The various described taxa of Acrodus differ from each other in rather minor fash- ion among the sizes and patterns of ridges on the dental plates and the shape and cur- vature of the crown. Typically each tooth is lozenge-shaped or rhomboidal with a single low crown.The maximum height of the tooth is less than half of the maximum length of the tooth. Each tooth is ornament- ed with a series of fine ridges which more or less radiate from the center of the crown (Figs. 3 and 4). There seem to be four rows of teeth in each dental battery, although the possibility of an extra row of small teeth at the anterior margin of the battery cannot be excluded. The teeth of the first row are distinctly smal- ler than the remaining three which are all Fig. 1 > Half nodule completely negatively prepared to pro- duce natural mold and then cast in Smooth-on molding compound to show head structures in mesial view. Scale in mm. Fig. 2 > Half nodule completely negatively prepared to pro- duce natural mold and then cast in Smooth-on molding compound to show head structures in later al view. Scale in mm. 3 Early Triassic Hybodont Shark Postilla 186 from Northern Madagascar 5 Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 longer and roughly of equal size to each other. The teeth of the first and second rows have a slightly more strongly curved crown than those of the last two rows. There seems to be no basic change in the outline of the base of the crown among the four rows. These characteristics of the dental battery seem to exclude the material from the genus Paleobates Von Meyer, 1849, which is described by Stensié (1921) as having more tooth rows with the third and fourth rows made up of teeth signifi- cantly longer, flatter, and more rectangular in shape than the other rows. Similarly, al- though the mandible of Pa/eobates polaris as described by Stensid (1921) is short and deep like that described here, the detailed shape is different and in the face of So little comparative material taxonomic comparisons are tenuous at best. The dimensions of the largest teeth in the Madagascar material are as follows: Average length 8 mm: average breadth 2.3 mm; average crown height 2 mm. The ridges on the teeth are relatively fine com- pared with those of described Acrodus Material and they show a pattern of bifurca- tion as they proceed from the center of the Crown. The general appearance is shown in Figures 1 and 2. A survey of described materials fails to Show any Triassic shark dental material with a pattern exactly comparable to that of the new material from Madagascar. It might be reasonable, therefore, to conclude that the taxon represented in Madagascar 'S distinct and that a new species should be Named for it. Here is a classic paleontolo- gist’s dilemma, for it is certain that not all the species of Acrodus or other genera dis- tinguished by their authors on the basis of dental ornamentation are true species (however that might be defined). It may be worthless to add another to this disreputa- Fig. 3 ’Acrodus sp. Tooth in side view. Fig. 4 ? ; crete ’Acrodus sp. Incomplete view of two teeth in side and occlusal views. ble list. Furthermore, although | have other morphological data upon which to base a description of the Madagascar specimen, | have had no opportunity for comparison of skull data with any other Acrodus material, let alone the type material. For the moment | will merely recognize the new material from Madagascar as ?Acrodus sp. with the note that, if | were willing to accept the * dental evidence as prima facie evidence (as lam not), it would be possible to distinguish the material as belonging to a “species” dis- tinct from other described materials. Palatoquadrate The two halves of the nodule show the pala- toquadrate from the medial (Fig. 5) and la- teral view (Fig. 6). The medial exposure of the palate is virtually perfect on one half- nodule; the whole mesial surface is ex- posed apparently undistorted. The other half principally shows the posterior portion of the lateral surface, with some details of the anterior tip of the palate. All the articular surfaces of the palate are clearly visible. The palatoquadrate (overall length = 9.0 cm; maximum depth = 2.4 cm) is elongate with a relatively small postorbital expansion and it lacks any significant deepening at the otic process. The anterior three-quarters of the palatoquadrate is formed as a straight, stout bar with a pronounced down- ward and mesial curvature of the tip, and there is a broad ventromesial flange bearing the dental battery. The most prominent features of the mesial surface of the palatoquadrate are three articular surfaces (Figs. 5 and 7). The largest of these is formed on the anterodor- sal extreme of the otic process and forms the articulation with the postorbital process of the braincase. This articular surface is a massive groove oriented not transversely but directed anterolaterally at an angle of about 17°. In the vertical transverse plane it is directed ventromesially at an angle of about 12° below horizontal. The whole ar- ticular facet is set off from the surface of the palatoquadrate by modest ridges. Early Triassic Hybodont Shark Postilla 186 from Northern Madagascar f Early Triassic Hybodont Shark Postilla 186 from Northern Madagascar bas post eth 20mm c Outline drawing of principal features of Figure 1, Restoration of the right palatoquadrate in lateral half nodule prepared to show mesial view of head view (above) and mesial view (below). For abbrevia- Structures. Abbreviations for this and following fig- tions, see Figure 5. ures: bas = basal articulation; c = condyle; cer = ceratohyal; eth = ethmoid articulation; ?hy = ?hyomandibular; m = mandible; post = postorbi- tal articulation; pq = palatoquadrate. 4Fig.6 Outline drawing of the principal features of Figure 2, half-nodule prepared to show lateral view of head Structures. For abbreviations, see Figure 5. 8 Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 Anteriorly there are two other major facets. A ventrally directed facet is formed as a broad groove located about one-third of the distance from the anterior tip of the suborbital ramus. This articulation faces ventrolaterally at an angle of about 40° below horizontal and thus, when seen from a directly anterior view, forms an angle of some 52° with the groove of the otic pro- cess. This facet is supported on a well- developed process formed as a flange on the mesial surface of the suborbital ramus and therefore, properly speaking, is as much a mesial as a ventral articulation. We may term this articulation the basal artic- ulation: it is Supported by the basal process. The third major articulation may be called the ethmoid articulation. This is a shallow, concave surface, formed as an oval, borne distinctly clear of the upper and slightly mesial surface of the anterior end of the palatoquadrate bar. This facet is oriented forwards, upwards and slightly mesially and evidently articulated with some sort of ectethmoid process of the postnasal wall. In addition to these three major facets, the mesial surface of the ventrally curved tip of the palatoquadrate is formed into a flange that apparently was ligamentously connected to the opposing structure of the other palatoquadrate. Between this flange and the ethmoid process the upper surface of the palatoquadrate is marked by shallow ridges and grooves, suggestive of a sliding connection with the underside of the postnasal wall. The upper dental battery was borne upon a deep thick flange of the palatoquadrate extending over the whole length of the suborbital ramus. The lateral surface of the palatoquadrate (Figs. 4 and 7) is relatively uncomplicated. The postorbital portion is deeply concave and massively thickened. In lateral view the upper margin of the palatoquadrate bar and the lower surface of the flange bearing the dental battery are parallel and horizontal. The palatomandibular articulation is typically double. The two parts of the joint lie along the posterior rim of the posterior process and make an angle of about 70° from the sagittal plane. The lateral portion of the joint, at the posterior tip of the palato- quadrate, is a narrow, convex, somewhat triangular process. The inner part of the joint is a larger, deep glenoid facet formed as an opposite triangle. Mesially, the inner- most part of the flange forming the posteri- or margin of the inner half of the joint is pro- duced into a slight ventral process contin- ued anterodorsally as a ridge on the mesial surface of the postorbital process. There is no obvious groove in the poste- rior rim of the palatoquadrate of the sort that would have marked the close apposi- tion of the hyomandibular. However, the la- teral and mesial angles marking the extent of the bicondylar jaw joint are both devel- oped significantly behind the curve of the posterior surface of the postorbital ramus and it is possible that the tip of the hyoman- dibular could have fitted alongside either of these. Mandible The mandible (overall length 8.3 cm; maxi- mum depth 3.4 cm) is well exposed in both mesial and lateral views. The main anterior part of the ramus is essentially flat, with no marked convexity. The mandible is relative- ly deep, the maximum depth being con- tained approximately 2.2 times in the over- all length. In lateral view the mandible shows a concavity in the posteroventral region but no other major features. The mesial surface is marked by a deep, long horizontal groove which evidently was the site for attachment of the lower dental bat- tery. Almost in the center of the mesial sur- face of the mandible there is a large scar, probably for muscle attachment. The bicondylar jaw joint is set at an oblique angle to the main mandibular ramus which is otherwise relatively straight. The more la- teral and anterior of the two portions of the joint are borne on a pronounced process which is continued as a ridge forming the angle in the posterior part of the mandible. This ridge and the ridge on the palatoqua- drate that leads to the upper articular facet 9 Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 form an essentially single line and both were evidently the site of a major ligamen- tous connection between the upper and lower jaws. The orientation of the two artic- ular facets in the mandible shows that the plane of the mandibular ramus was not vertical when the gap was closed but was inclined mesially at some 10°. Branchial skeleton Lying diagonally across the mesial surface of the mandible (Fig. 5) is a large element that is tentatively identified as the ceratohy- al. Its anterior margin, particularly the anter- Oventral part, is incomplete, but the posteri- Or portion is intact. The total length and Shape of the elements cannot be guessed. Another fragment lying above the palato- quadrate may possibly represent part of an €pibranchial. This fragment again only Shows the posterior portion. It is exposed in lateral view and shows a massive lingual Shelflike flange. Slightly inside the posterior rim of the Palatoquadrate (Fig. 6) is a rod-shaped sec- tion of an element that is preserved in the €xpected position of a hyomandibular. This rod does not extend as far as the mandibu- lar articulation and it is difficult to tell, if this 's the hyomandibular, what part it might have played in the jaw suspension. The fact that the element is circular in cross section, rather than being flattened so as to be Pressed to the palatoquadrate, is a small tem of evidence suggesting a minor role at best in the suspensorium for this element. Relationship of the Braincase to the Jaws No part of the braincase is well preserved, but the strongly developed articular facets On the palatoquadrate allow us to make Some tentative reconstructions, at least of the overall proportions of the braincase and Of its relationship to the jaws. First, we can Note that the distinctly posterior placement Of the otic articulation with the postorbital process and the oblique orientation of the “hyomandibular” strongly suggest that the otic region of the braincase was short. Fur- ther, the postorbital processes were well developed not only in the lateral extent but also were deep ventrally. The basal articula- tion between palate and braincase is inter- esting because it is relatively far forward and must be in an antorbital rather than suborbital position. There must have been paired rodlike basal processes on the antor- bital/suborbital shelf of the braincase, pro- jecting directly laterally. In addition, there must have been well-developed, paired ec- tethmoid or antorbital processes of the posterior nasal region for the articulaton of - the ethmoidal articular facets of the palato- quadrate. This must have been developed immediately behind and/or below the nasal capsule with a sliding articulation of the capsule. However, the palatoquadrate probably did not extend forward beneath the whole of the capsule, but only to the back of the capsule. Having delineated the relationship be- tween braincase and palatoquadrate, we can also ask what the mobility of the jaws was. It was clearly impossible for the jaws to move anteroposteriorly relative to the braincase. The postorbital and nasal artic- ulations are arranged to allow only lateral excursions of the palatoquadrate relative to the braincase, whereas the ethmoid artic- ulation suggests a rolling hinge. But it is dif- ficult to see what sort of lateral movement of the palatoquadrates occurred. The ob- lique orientation of the transverse basal and postorbital articulations is such that ex- cursion with close connection of palate and braincase at one joint would cause a sepa- ration of the two structures at the other joint. This result is heightened by the slight- ly anterior orientation of the groove of the postorbital articulation. If the joint were somewhat loose, with ligamentous bind- ings, it is possible that the palate was flared laterally from the braincase, with the eth- moid articulation forming the fulcrum and the posterior part of the palate making the greatest excursion, rolling outwards, and 10 Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 slightly forwards as the basal articulation slid outwards on the basal process. This was accompanied by depression of the mandible, the bicondylar jaw joint being ar- ranged so that as the mandible was de- pressed it rotated slightly, bringing the mandibular ramus into a more vertical plane. The complex articulations between pala- toquadrate and braincase and the specific nature of the mechanical connection of the two, with the palate very firmly braced against the braincase by the two major transverse articulations, make it unlikely that the hyomandibular had a prominent role in movements of the jaws. Discussion The palate of 7Acrodus shows many im- portant differences from that of other sharks, and these lead naturally to a discus- sion of the plate and neurocranium in sharks in general. It is widely agreed that there have been important changes in the nature of the jaw suspension articulation in the evolution of sharklike fishes, particularly in a general de- velopment of a hyostylic jaw suspension from an (ancestral) amphistylic condition. The data are well summarized by Schaeffer (1967) and Maisey (1980). Here, unfortu- nately, little progress has been made in refining this useful but broad generalization, the reason being that scant new informa- tion has come available concerning the nature of the jaws in fossil elasmobranchs. This being the case, it is frustrating in the extreme to discover that the structure of the palatroquadrate in ?Acrodus, so beauti- fully demonstrated in the material de- scribed here, is totally unlike that of any other shark. By drawing together the recent descrip- tions of Cobe/odus by Zanger|l and Wil- liams (1975), Denaea by Schaeffer and Williams(1977), and the older work on C/a- dodus by Gross (1937, 1938), we can begin to define the nature of the palatoqua- drate in Paleozoic “cladodont” elasmo- branchs (see also the morphotype defined by Zangerl, 1973). The palate seeems to have been basically quite simple. The pos- torbital ramus is very large having the typi- cal primitive “cleaver” shape described by Schaeffer (1975) and Schaeffer and Wil- liams (1977). The postorbital articulation is well developed in these sharks and this is a primitive characteristic for all gnatho- stomes (Schaeffer and Williams, 1977). The nature of the actual articulation which is borne on the ventral and posterior portion of the postorbital process and a massive otic process of the palatoquadrate is not completely clear. The articulation was es- sentially in a vertical sagittal plane and al- lowed no fore-and-aft movement of the palate except possibly through a rotatory movement in the plane of the palate. The suborbital ramus is relatively slender and has a well-developed basal articulation with the subocular shelf of the braincase. The subocular shelf shows a lozenge- shaped process which extended clear of the subocular shelf and the articular shelf. The articular surface between palate and braincase is somewhat elongate anteropos- teriorly. The basal articulation is developed rather anteriorly in the orbit and it is not necessarily homologous with the “basipte!- ygoid articulation” developed between palate and braincase in teleostome fishes and tetrapods, which typically is formed at the transverse level of the foramen for the hypophysial opening (see discussion in Jarvik, 1977, inter alia). There is no develop- ment of ethmoidal processes between the tip of the palatoquadrate and the nasal capsule. The two halves of the palates possibly met in the midline, posterior to the nasal capsule, except in Cladoselache where the mouth was terminal (Zangerl, 1973). It has been claimed that Cladodus had an orbital process and articulation, and als° a “basal angle” in the floor of the braincas® as in some modern sharks (see Jarvik, 1977). However, the material described by Gross (see photograph in Gross, 1938, pl. |. 11 Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 20 mm fig. 2A) shows merely a slight thickening of Fig. 8 the tip of the suborbital ramus of the palato- quadrate. In Gross’s reconstruction (1938, fig. 2) this expansion has been slightly exag- gerated (see also Jarvik, 1977, fig. 4D). There seems to be a fundamental difference between this sort of thickening of the Suborbital ramus and a true orbital process (see below). Further, the structure identified by Jarvik (1977) as the articular surface on the orbital wall for the reception of this pro- Cess does not seem to fit the process and is Probably no more than the angle produced behind the postnasal wall. An orbital pro- C€Ss is definitely absent in Cobe/odus and Denaea. Four sharks that would fall into the hybo- dontiform assemblage of Maisey’'s (1975) Classification have been described: Hy- Lodus (see Woodward, 1916; and Maisey, 'N preparation), Asteracanthus (Peyer, 1946), Tristychius (Woodward, 1924, and Dick, 1978) and Onchoselache (Dick and a 1980). All four forms agree with *Acrodus in that the postorbital ramus of the palatoquadrate is relatively reduced Asteracanthus. Sketch of the palatoquadrate in la- teral view from British Museum (Natural History) specimen 12614. compared with the Paleozoic forms. It does not have the “cleaver” shape and massive otic process, being on the contrary low and elongate. Similarly, in all four forms the suborbital ramus is rather broader than in the Paleozoic forms. Asteracanthus has been described in some detail by Peyer, but unfortunately his interpretations are diffi- cult to follow and, after study of material in the British Museum (Natural History), par- ticularly specimen P. 12614, | believe that he had worked with an incorrect orientation of the materials. As shown in Figure 8, the overall proportions of the palatoquadrates of Asteracanthus are very similar to those in ?Acrodus. However, the points of artic- ulation with the braincase are completely different. Specifically, the prominent trans- verse otic-postorbital articulation of ?Ac- rodus are completely lacking and no spe- cial articular surfaces are developed in 12 Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 either position. In this respect, Astera- canthus agrees far more with Hyboadus : in these two genera the articulations be- tween braincase and palate were arranged to produce a fore-and-aft sliding movement. The articular surfaces of the palatoquadrate are therefore merely the upper rim of the postorbital ramus which fitted into an anteroposterior groove on the under side . of the massive postorbital process (Maisey, personal communication) and a similar slid- ing contact between the dorsomesial rim of the suborbital ramus and the side of the subocular shell of the braincase — probably continuing directly into a similar ethmoidal articulation with the undersurface of the nasal capsule. The palatoquadrate of Hy- bodus is far more massively developed es- pecially much deeper in proportion com- pared with that of ?Acrodus. All three agree, however, in the absence of any basal angle in the braincase and in the absence of an orbital process. As recently redescribed by Dick (1978) and Dick and Maisey (1980), the overall pro- portions of the palatoquadrate in /risty- chius and Onchoselache are again quite similar to those of ?Acrodus and Astera- canthus, particularly in the low nature of the postorbital ramus. The nature of the postorbital otic articulation and basal artic- ulation are not clear, but probably allowed transverse movements of the palate as in ?Acrodus. An interesting feature is that both have been restored with the anterior part of the suborbital ramus showing a small dorsal development that is identified as an orbital process. However, at least in Onchoselache, this is probably misinter- preted and represents the relatively deep anterior end of the palate which has become flattened out. Three other Meszoic sharks have been described from material showing the skull: Paleospinax (Woodward, 1889; Maisey, 1975); Synechodus (Woodward, 1886); and Squalogaleus (Maisey, 1976), all of which Maisey includes in the Paleospinaci- dae as relatively primitive euselachians. All three show features that allow the palato- quadrates to be restored essentially as in the apparently primitive living sharks Chlamydoselachus and Heptranchias. The postorbital ramus of the palatoquadrate is relativley reduced and in all these forms there is a well-developed orbital process of the palatoquadrate. This is a dorsal projec- tion from the upper and mesial! surfaces of the suborbital ramus of the palate; it rises in the orbit in front of the level marked by the optic foramen in the orbital wall and there is a sliding articular contact between this orbital process and the anterior orbital wall. The condition of the basal articulation in these early fossil euselachians is not available and therefore it is not possible to tell to what extent the development of an orbital process is correlated with the basal articulation. In the modern Ch/amydose- lachus, the orbital and basal articulations are quite separate from each other, the former being far forward in the orbit and the latter far back in the posterior part of the orbit. In modern Heptranchias, on the other hand, the two articulations are essen- tially confluent. An orbital process is found in many lines of modern sharks (for example, hexan- choids, squaloids, lamnoids, carcharinoids, and squatinoids) according to Compagno (1977), but in several of these groups the orbital process has become considerably specialized, forming a major articulation with the back of the nasal capsule rather than a vertical flange within the anterior orbit. Apart from the three paleospinacids mentioned above, the orbital process is not described with complete certainty in fossil forms. On the basis of the limited amount of evidence available, two possibilities exist. First, the orbital process may be a primitive feature for the elasmobranch fishes, presen! in Paleozoic cladodonts, Tristychius and Onchoselache plus modern sharks, and present also in acanthodians (see Jarvik, 1977). In this case, the absence of the orbi- | tal process in Hybodus, ?Acrodus, and AS Y teracanthus is asecondary feature and perhaps a specialization linking these threé within the hybodonts. In this case also, the Is Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 absence of an orbital process in xenacanth Sharks would be a second and independent instance of secondary loss of this feature. The second possibility is that the orbital Process is incorrectly identified in Paleozoic Cladonts (where the evidence is extremely limited) and possibly in 7ristychius and Onchoselache where the evidence is yet more slight. In this case, the orbital process Should be considered a specialization of Certain modern sharks (Maisey, 1980) and Its absence in the forms just mentioned, together with xenacanths, ?Acrodus, Hy- bodus, and Asteracanthus, would all re- flect a primitive condition. The matter re- quires considerable further research for cla- rification and not least among such studies Must be a careful examination of the rela- tlonship between orbital and basal pro- cesses and articulation. Finally the structure of the palate in ?Acrodus (and to a lesser extent, Hybodus and Asteracanthus) shows certain general resemblances to that of modern heterodont Sharks, particularly in the low postorbital ramus, absence of an orbital process, well- developed ethmoidal articulations and ab- sence of a basal angle. In the past, this would have been enough to allow one to Suggest a close relationship between hybo- donts and heterodonts. However, Com- Pagno (1977) has recently attempted to Show that hybodonts belong to a more de- "Ved position within the euselachians, spe- Cifically being allied with the galeoid oryc- toloboids and lamnoids. If this is the case, the absence of the orbital process in hetero- donts might be considered a highly derived Condition and the overall close similarity of the palates of the two groups a conver- Jence due perhaps to a common pattern of fore-and-aft jaw movements. The present inadequate state of knowledge of detailed Nybodont anatomy prevents us from resolv- 'Ng this problem. The result of the present study, therefore, ; to characterize part of the head in ‘Acrodus from the Early Triassic of Mada- Yascar and to demonstrate the diversity of Structure in early sharks. This diversity serves to confuse rather than clarify the phylogenetic relationships among Meso- zoic sharks and among hybodonts, ctena- canths, and euselachians. An orbital process is found in many lines of modern sharks (for example, hexan- choids, squaloids, lamnoids, carcharinoids, and squatinoids) according to Compagno (1977), but in several of these groups the orbital process has become considerably specialized, forming a major articulation with the back of the nasal capsule rather than a vertical flange within the anterior orbit. Apart from the three paleospinacids mentioned above, the orbital process is not described with complete certainty in fossil forms. On the basis of the limited amount of evidence available, two possibilities exist. First, the orbital process may be a primitive feature for the elasmobranch fishes, present in Paleozoic cladodonts, 7ristychius and Onchoselache plus modern sharks, and present also in acanthodians (see Jarvik, 1977). In this case, the absence of the orbi- tal process in Hybodus, ?Acrodus, and As- teracanthus isasecondary feature and perhaps a specialization linking these three within the hybodonts. In this case also, the absence of an orbital process in xenacanth sharks would be a second and independent instance of secondary loss of this feature. The second possibility is that the orbital process is incorrectly identified in Paleozoic cladonts (where the evidence is extremely limited) and possibly in 7ristychius and Onchoselache where the evidence is yet more slight. In this case, the orbital process should be considered a specialization of certain modern sharks (Maisey, 1980) and its absence in the forms just mentioned, together with xenacanths, ?Acrodus, Hy- bodus, and Asteracanthus, would all re- flect a primitive condition. The matter re- quires considerable further research for cla- rification and not least among such studies must be a careful examination of the rela- tionship between orbital and basal pro- cesses and articulations. Finally the structure of the palate in ?Acrodus (and toa lesser extent, Hybodus 14 Early Triassic Hybodont Shark Postilla from Northern Madagascar 1 86 and Asteracanthus) shows certain general resemblances to that of modern heterodont sharks, particularly in the low postorbital ramus, absence of an orbital process, well- developed ethmoidal articulations and ab- sence of a basal angle. In the past, this would have been enough to allow one to suggest a close relationship between hybo- donts and heterodonts. However, Com- pagno (1977) has recently attempted to show that hybodonts belong to a more de- rived position within the euselachians, spe- cifically being allied with the galeoid oryc- toloboids and lamnoids. If this is the case, the absence of the orbital process in hetero- donts might be considered a highly derived condition and the overall close similarity of the palates of the two groups a conver- gence due perhaps to a common pattern of fore-and-aft jaw movements. The present inadequate state of knowledge of detailed hybodont anatomy prevents us from resolv- ing this problem. The result of the present study, therefore, is to characterize part of the head in ?Acrodus from the Early Triassic of Madagascar and to demonstrate the di- versity of structure in early sharks. This di- versity serves to confuse rather than clarify the phylogenetic relationships among Mesozoic sharks and among hybodonts, ctenacanths, and euselachians. Acknowledgments | am extremely grateful to Paul E. Olsen for spotting this specimen in the Harvard col- lections and for preparation of the delicate material.| am grateful to Mr. Olsen, Amy R. McCune and William Kohlberger for valua- ble comments on the manuscript. This work is part of a project on Triassic fishes supported by the National Science Founda- tion (grants DEB 77-08412 and DEB 79-21746). The drawings were prepared by Linda Price Thomson and the photographs by William K. Sacco. 15 Early Triassic Hybodont Shark Postilla 186 from Northern Madagascar Literature Cited Compagno, L.V. 1973. Interrelationships of living elasmobranchs. /n P.H. Greenwood, R:S. Miles, C. Patterson [eds.] Interrelationships of Living Fishes. Zool. J. Linn. Soc., Suppl. 1,53:15-61. 1977. Phyletic relationships of living sharks. Am. Zool. 17:303-322. Dick, J.R.F. 1978. On the Carboniferous shark 7ristychius arenatus Agassiz from Scotland. Trans. R. Soc. Edinburgh 70:63-109. Dick, J.R.F. and J.G. Maisey. 1980. The Scottish Lower Carboniferous shark Onchoselache tra- quari. Palaeontology 23:363-374. Gross, W. 1937. Das Kopfskelett von Cladodus wildungensis. |. Endocranium und Palatoquadra- tum. Senckenbergiana, 19:80-107. PF 1938. Das Kopskelett von Cladodus wildungensis. ||. Der Kieferbogen. Senckenbergiana, (123-145. Jarvik, E. 1977. Systematic position of acanthodian fishes. /n S.M. Andrews, RS. Miles, A.D. Walker [eds.] Problems in Vertebrate Evolution. Linn. Soc., London, Symp. Ser. No. 4:199-225. Lehman, J-F. 1952. Etude complémentaire des Poissons de |’Eotrias de Madagascar. K. Sven. Vetenskapakad. Hand. Ser. 4:1-201. Maisey, J.G. 1975. The interrelationships of phalacanthous selachians. Neues Jahrb. Geol. Palaeontol. Monatsh., Part 9, 1975:553-567. 1976. The Jurassic selachian fish Protospinax Woodward. Palaeontology 19:733-747. ———— 1977. The fossil selachian fishes Palaeospinax Egerton 1872 andNemacanthus Agassiz 1837. Zool. J. Linn. Soc. 60:259-273. ————. 1980. An evaluation of jaw suspension in sharks. Am. Mus. Novitates No. 2706:1-17. Peyer, B. 1946. Die schweizerischen Funde von Asteracanthus (Strophodus). Schweitz. Palaeontol. Abh. 64:1-101. ail J. 1934. Paléontologie de Madagascar: les Poissons du Trias inférieur. Ann. Paleontol. ‘83-178. Regan, C.T. 1906. A classification of the selachian fishes. Proc. Zool. Soc., 1906: 722-758. Schaeffer, B. 1967. Comments on elasmobranch evolution, /n P.W. Gilbert, R.G. Mathewson, D.P. Rall leds.] Sharks, Skates, and Rays. Johns Hopkins Univ. Press, Baltimore. ———— 1975. Comments on the origin and basic radiation of the gnathostome fishes with Particular reference to the feeding mechanism. /n Problémes actuels de Paléontologie (Evolution des Vertébrés) Colloq, Int. C.N.R.S., No. 218:101-109. Schaeffer, B. and M. Williams, 1977. Relationships of fossil and living elasmobranchs. Am. Zool., 17:293-302. Stensié, E.W. 1921. Triassic Fishes of Spitzbergen. Part |. Adolf Holtzhausen, Vienna. Woodward, A.S. 1886. On the relations of the mandibular and hyoid arches in a Cretaceous Shark (Hybodus dubrisiensis Mackie). Proc. Zool. Soc., London. 218-224. “———— 1889. Catalogue of the Fossil Fishes of the British Museum (Natural History). Part |. Brit. Mus. (Nat. Hist), London. ————. 1916. The fossil fishes of the English Wealden and Purbeck Formations. Palaeontol. Soc. Monogr. 69, Part |: 1-48. ———— 1924. On a hybodont shark ( 7ristychius) from the Calciferous Sandstone Series of Eskdale (Dumfriesshire). Q.J. Geol. Soc. 80:338-342. Zangerl, R. 1973. Interrelationships of early chondrichthyans. /n P.H. Greenwood, R.S. Miles, and - Patterson [eds] Interrelationships of Early Fishes. Zool. J. Linn. Soc., Suppl. 1, 53:1-14. Zangerl, R. and M.E. Williams. 1975. New evidence on the nature of the jaw Suspension in Palaeozoic anacanthous sharks. Palaeontology 18:333-341. Early Triassic Hybodont Shark from Northern Madagascar Postilla 186 The Author Keith Stewart Thomson. Department of Biology and Peabody Museum of Natural History, Yale University, 170 Whitney Avenue, PO Box 6666, New Haven, CT 06511.