Braincase of the Upper Devonian Shark Cladodoides Wildungensis (Chondrichthyes, Elasmobranchii), with Observations on the Braincase in Early Chondrichthyans

Home , Acanthodes, Chlamydoselachus, Cladodont, Cladodus, Cobelodus, Debeerius

BRAINCASE OF THE UPPER DEVONIAN SHARK CLADODOIDES WILDUNGENSIS (CHONDRICHTHYES, ELASMOBRANCHII), WITH OBSERVATIONS ON THE BRAINCASE IN EARLY CHONDRICHTHYANS

JOHN G. MAISEY Division of Paleontology American Museum of Natural History ([email protected])

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 288, 103 pp., 38 ®gures Issued March 2, 2005

CONTENTS Abstract ...... 3 Introduction ...... 3 Materials and Methods ...... 4 Anatomical Abbreviations ...... 10 Description ...... 11 General Features, Taphonomy, and Forensic Observations ...... 11 Ethmoid Region ...... 15 Extent of Polar Cartilage-Derived Region ...... 18 Ophthalmic and Efferent Pseudobranchial Arteries ...... 19 Anterior Palatoquadrate Articulation ...... 22 Bucco-Hypophyseal Fenestra and Internal Carotid Arteries ...... 22 Optic Pedicel, Optic Nerve, and Oculomotor Nerve ...... 25 Trochlear Nerve ...... 26 Abducent Nerve ...... 27 Profundal Nerve ...... 30 Trigeminal Nerve ...... 30 Facial Nerve ...... 31 Octavolateral Nerves ...... 34 Posterior part of Orbit ...... 35 Postorbital Process ...... 36 Dorsal Aortae ...... 39 Orbital Artery and Palatine Nerve ...... 40 Roof of Orbitotemporal Region ...... 41 Posterior Tectum and Dorsal Fontanelle ...... 41 Dorsal Otic Ridges ...... 42 Lateral Otic Process and Hyomandibular Fossa ...... 42 General Features of Skeletal Labyrinth ...... 42 Occipital Arch ...... 43 Cranial Endocast ...... 52 Discussion ...... 57 Ontogenetic Aspects of the Braincase ...... 57 Precerebral Fontanelle ...... 58 Efferent Pseudobranchial Artery ...... 58 Anterior Palatoquadrate Articulation ...... 59 Trabeculae, Polar Cartilage, and Suborbital Shelf ...... 64 Optic Pedicel and Position of Oculomotor Foramen ...... 67 Trigeminal, Facial, and Octavolateral Nerves ...... 68 Trigemino-Facial Recess ...... 71 Postorbital Process ...... 72 Postorbital Articulation ...... 77 The ``Extra Foramen'' Problem ...... 78 Remarks on the Dorsal Vein Sinus in Acanthodes ...... 81 Otico-Occipital Proportions ...... 81 Remarks on the Inner Ear in Chondrichthyans, Placoderms, and Agnathans ...... 81 Posterior Dorsal Fontanelle and Posterior Tectum ...... 86 Lateral Otic Process ...... 87 Hyomandibular Fossa ...... 88 Occipital Arch, Otico-Occipital Fissure, and Hypotic Lamina ...... 89 What Kind of Shark was Cladodoides? ...... 92 Phylogenetic Patterns in Paleozoic Sharks ...... 93 Conclusions ...... 95 Acknowledgments ...... 98 References ...... 98 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 3

ABSTRACT The braincase of an Upper Devonian shark (Cladodoides wildungensis) is investigated using high-resolution CT scanning, and its internal and external morphology is described from three- dimensional digital reconstructions. Many features are shared with modern elasmobranchs (neoselachians); these mostly represent conserved elasmobranch, chondrichthyan, or gnathostome characters. C. wildungensis resembles neoselachians in having a closed metotic ®ssure de®ning a glossopharyngeal canal, unlike some early chondrichthyans (e.g., Pucapampella, Orthacanthus) in which an open metotic ®ssure extends a considerable distance beneath the otic capsule. However, the glossopharyngeal canal in C. wildungensis is connected to a persistent dorsal ®ssure between the occipital arch and otic capsules, whereas this ®ssure becomes closed during neoselachian ontogeny. The embryonic polar cartilage apparently made a much greater contribution to the basicranium than in neoselachians and gnathostomes generally. This has affected the position of the efferent pseudobranchial foramen and the location of the orbital articulation in Cladodoides (both of which are located anteriorly in the orbit), as well as the size and extent of the dorsum sellae and the hypophyseal chamber inside the braincase. An- other unusual feature in Cladodoides is the postorbital position of its trigeminal and facial nerves; instead of being situated in a trigemino-facial fossa of the orbit as in neoselachians, their foramina are located behind the postorbital process, although the pituitary vein and the origin of the external rectus eye muscle (located within the trigemino-facial fossa in neoselachians) both lie in the posterior part of the orbit. The oblique orientation of the postorbital arcade (derived from the embryonic lateral commissure) may be related to the exclusion of the trigeminal and facial nerves from the orbit. A laterally oriented canal through the process probably contained the trigeminal maxillary ramus (perhaps accompanied by the buccal ramus of the anterodorsal lateral line nerve). By contrast, in neoselachians these rami pass directly into the orbit (sometimes forming a buccal-maxillary complex), but they do not pass through the postorbital process. The Cladodoides braincase probably represents a juvenile or submature individual and shows indications of incomplete development (e.g., presence of a wide bucco- hypophyseal fenestra and absence of a precarotid commissure; incomplete enclosure of the internal carotids and orbital arteries by basicranial cartilage; incomplete closure of the braincase roof along the dorsal midline). The braincase also lacks multiple-layered prismatic calci®cation, possibly representing another juvenile feature. The phylogenetic relationship of Cla- dodoides wildungensis is discussed. Dental, postcranial, and cranial features lend some support to the old hypothesis that cladodont sharks form a monophyletic group. While such a view is at variance with several recently published phylogenetic analyses, many of the characters noted in the present work have yet to be tested within a rigorous cladistic framework, and their phylogenetic signi®cance is uncertain.

INTRODUCTION drichthyans and elasmobranchs has helped Literature on the phylogeny of early gna- steer these hypotheses, whereas little atten- thostomes is dominated by descriptions and tion has been paid to extinct chondri- comparisons of osteichthyans, placoderms, chthyans. This stasis undoubtedly results and acanthodians, but the body of literature from the great rarity of morphologically in- concerned with fossil chondrichthyan anato- formative chondrichthyan fossils, as well as my is comparatively small. Despite this pau- from inadequate preservation and a limited city of reliable morphological data (or per- ability to observe crucial features even in haps because of it), chondrichthyans have better specimens. Whatever the reason, there sometimes been credited with considerable has never been a comprehensive survey of evolutionary importance (for example as pu- morphological diversity in early chondri- tative placoderm descendants; Holmgren, chthyans. 1942; StensioÈ, 1969), while at other times We need to look for more shark fossils, they have been treated rather summarily, as and the importance of new discoveries is am- a primitive sister group to osteichthyans (Ro- ply illustrated by recent discoveries of artic- sen et al., 1981; Gardiner, 1984a). Invariably, ulated and other well preserved remains of however, the morphology of modern chon- Lower and Middle Devonian chondri- 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 chthyans (e.g., Pucapampella, Doliodus, and and jaws presented by Gross (1938: ®g. 2) Gladbachus; Maisey, 2001b; Maisey and will be familiar to many readers. However, Anderson, 2001; Heidtke and KraÈtschmer, the braincase was never completely prepared 2001; Miller et al., 2003). However, we also and its occipital region has only recently need to utilize new technologies such as CT been revealed by CT scanning (Maisey, scanning to retrieve previously inaccessible 2001a). This paper presents a revised and morphological data from specimens we al- more complete account of cranial morphol- ready have. From such studies, we have al- ogy in this specimen, drawing heavily on CT ready learned that cranial morphology in ear- scanning and digital imaging. New data are ly chondrichthyans differed in many respects presented which should be useful to investi- from that in modern forms (Maisey, 2004a), gators of other early chondrichthyans such as requiring some radical reappraisal of char- Gladbachus and Doliodus, and also to those acter polarity not only in chondrichthyans, interested in the interrelationships and early but in osteichthyans and even gnathostomes radiations in gnathostomes. in general. For example, it is no longer possible to defend a relationship between pla- MATERIALS AND METHODS coderms and elasmobranchs on the basis of labyrinth morphology (Maisey, 2001a), or to The specimen under investigation (For- regard the palatobasal articulation or ventral schungsinstitut und Naturmuseum Sencken- otic ®ssure as osteichthyan synapomorphies berg, Frankfurt, P2468) is a late Devonian (paradoxically, the number of supposedly (Upper Frasnian) shark braincase, from the basal osteichthyan synapomorphies has fall- Wildungen Limestone, Wildungen, Germany en as our knowledge of early chondri- (for a modern stratigraphic summary of this chthyans has improved). unit, see Maisch, 1998). It is almost perfectly Modern elasmobranchs and holocephalans preserved, with little distortion and minimal are regarded here as monophyletic sister damage (®g. 1). External features of the groups within the crown group Chondrich- specimen were described in considerable de- thyes, and both possess autapomorphic traits tail by Gross (1937), and several of his an- suggesting they are divergently specialized. notated line drawings are reproduced here as Thus, a modern shark such as Squalus is no a single composite illustration, along with more representative of elasmobranchs in gen- their original German annotations (®g. 2). eral than a cow is of tetrapods; both retain Readers may ®nd it useful to refer to this conserved features of their respective groups, ®gure throughout the work. and both also possess an overlay of subse- It is important to note that the braincase quently acquired apomorphic characters, but described by Gross (1937) is not the holo- those in cows (e.g., as amniotes, therians, un- type of C. wildungensis. That taxon is found- gulates, etc.) have traditionally been more ed on jaw elements associated with teeth clearly identi®ed than those of modern elas- (Jaekel, 1921), and while Gross (1937) pre- mobranchs. With an improved knowledge of sented a circumstantial case for including the the selachian fossil record, however, we are braincase in this species, no diagnostic char- slowly gaining some insight into primitive acters can actually be compared. Thus, the elasmobranch morphology and how this may precise identity of the braincase is problem- have changed during their subsequent evo- atic, despite its importance as a source of lutionary history. data concerning early elasmobranch cranial The present investigation involves a shark morphology. Moreover, many nominal spe- braincase from the Upper Devonian lime- cies of Cladodus (including the type species, stone of Wildungen, Germany (®g. 1), ®rst C. mirabilis Agassiz, 1843) are founded on described and referred to Cladodus wildun- isolated teeth, but Cladodus-like teeth occur gensis by Gross (1937). The specimen has in several different kinds of Paleozoic ``cla- received considerable attention elsewhere dodont'' sharks including Cladoselache, ste- (e.g., Holmgren, 1941; Schaeffer, 1967, thacanthids, symmoriids, and ``ctenacanths''. 1981; Moy-Thomas and Miles, 1971; Jarvik This led Zangerl (1973) to declare Cladodus 1980), and the reconstruction of the cranium a nomen vanum (subsequently suppressed in 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 5

Fig. 1. Views of the Cladodoides braincase. (A) Dorsal view; (B) ventral view; (C) lateral view; (D) anterior view. 6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 2. Reproductions of original illustrations of the Cladodoides braincase, slightly modi®ed from Gross (1937) by combining several originally separate ®gures. (A) Dorsal view; (B) ventral view; (C) left lateral view; (D) scrap lateral view of orbit with postorbital process removed; (E) posterior view of postorbital process. For the convenience of the reader, the original abbreviations of the annotations in German are retained, so that features identi®ed by Gross and discussed in the text are more readily identi®ed; for further explanation, the original work should be consulted. These annotations differ from those used elsewhere in the present paper and are not explained in the list of abbreviations. The only other changes from the original published drawings are digital clari®cation of leader lines and some abbreviations, cleaned outlines, and improved contrast. 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 7 favor of the term nomen dubium; Chorn and Preliminary animated image processing at Whetstone, 1978; Williams, 1985). Some UTA used Voxblast to produce three-dimen- new genera have been founded on specimens sional resliced renderings, in which slices are in which ``Cladodus'' teeth are associated sequentially removed along the orthogonal with anatomically informative skeletal re- X±Y±Z axes while retaining the same slice mains (e.g., Gutturensis Sequeira and Coates, thickness generated in the original scan im- 2000: formerly Cladodus neilsoni Traquair, agery. Subsequent image processing by the 1898). Cladodus wildungensis was referred author utilized Imaris/Surpass surface- and to a new genus Cladodoides by Maisey volume-generating software, by means of (2001a), but this reassignment does little to which structural contours were plotted on resolve its relationships since the distinction each original 8-bit CT scan slice. Three-di- rests mainly on rather slight differences in mensional isosurfaces and volumes of both the teeth of Cladodus mirabilis and C. wil- the braincase and its endocast were created dungensis (e.g., the number and arrangement digitally from these contours, and individual of lateral cusps). Nevertheless, dental char- images were captured in either isometric or acters do provide a basis for distinguishing perspective mode. some cladodont sharks (Ginter and Ivanov, Two main kinds of three-dimensional im- 1992; Ginter, 1998 and in prep.), and the ge- ages are generated by the Imaris/Surpass nus Cladodus may therefore eventually re- software, namely surface and volume ren- gain systematic utility. For example, Clado- derings. A surface rendering is essentially a dus mirabilis teeth resemble those of the digital reconstruction of an object and is the ctenacanth sharks from the Cleveland Shale three-dimensional equivalent of a drawing, in the morphology of articular buttons on the whereas a volume rendering is more the lingual surface of the basal platform and in scanning equivalent of a low-resolution pho- the relationship of successional teeth in each tograph. Volume renderings are produced di- tooth family. This tooth gestalt differs from rectly from a stack of original CT scan slices that found in symmoriids and stethacanthids (transverse in this case) and provide an im- and may re¯ect a basic dichotomy within cla- mediate three-dimensional rough image of dodont sharks. Unfortunately, the morphol- the specimen, but internal features can only ogy of the basal platform is poorly known in be seen on a slice-by-slice basis. However, Cladodoides wildungensis. surface renderings are calculated from con- The specimen underwent considerable me- tours plotted manually on each original slice, chanical preparation prior to its original de- and points are then triangulated between ad- scription, and in view of its historic and mor- jacent slices by the computer, generating a phological importance no further preparation geodesic surface which is then digitally ®l- could be undertaken. Although the specimen tered and smoothed. Surface renderings are is extremely informative, there are limita- time-consuming to produce but create re- tions on what can be observed externally. markably detailed images which can be Computerized tomography (CT scanning) viewed in any orientation and in either per- provides an ideal noninvasive procedure to spective or standard orthographic projec- investigate aspects of the fossil that are oth- tions. More important, internal features oth- erwise inaccessible and also provides a use- erwise obscured by matrix and overlying ful test of earlier observations. A high-reso- structures are revealed by surface renderings. lution CT scan of the specimen was made at Surface renderings can also be sectioned by the Geology Department of the University of selecting an appropriate clipping plane, Texas at Austin (UTA). Scan parameters are which permits structures to be viewed in al- as follows: 180 kV, 0.133 mA, no ®lter, glass most any cross section (as in the series of wedge, 160% offset, slice thickness 5 lines slices through the postorbital process illus- (ϭ 0.262 mm), S.O.D. 92 mm, 2400 views, trating this work). Virtually any clipping 2 samples per view, interslice spacing 4 lines plane can be generated by selecting different (ϭ 0.2096 mm), ®eld of reconstruction 43.2 coordinates along the X±Y±Z axes of the ob- mm, reconstruction offset 750, reconstruc- ject, and the plane can also be tilted oblique- tion scale 33, scanned in three-slice mode. ly by rotating each axis independently. 8 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 3. Volume and surface renderings of the Cladodoides braincase. (A, B) Combined volume and surface renderings; (C, D) surface rendering only. Panels A and C are oblique left anterolateral views; panels B and D are oblique left posterolateral views. Note the marked lack of preparation in the occipital region and a thin but signi®cant covering of matrix in the ethmoid region and the anterior part of the supraorbital shelf (darker gray areas).

Many of the illustrations in this work rep- terpretation of scans and the plotting of con- resent views captured from Imaris/Surpass tours. Combining volume and surface ren- surface renderings and processed using com- derings also reveals interesting taphonomic mercially available imaging software such as and forensic details about the Cladodoides Adobe Photoshop. Only the left side of the braincase. Forensic observations include: the braincase was used to generate contours, lack of original preparation in the occipital since it is more complete. The views of the region; a thin layer of matrix covering the complete braincase are composites, made by anterior palatoquadrate articulation, which mirror-imaging the surface rendering with an was incorrectly considered to be broken by appropriate allowance for different illumi- Gross (1937); and incomplete preparation nation of each side. and cleaning of the supraorbital shelf and It is sometimes useful to combine volume other delicate structures (®g. 3). Such foren- and surface renderings although most mor- sic observations illustrate the great value of phological features are best revealed by the CT scans in evaluating the accuracy of ear- latter. Discrepancies between volume and lier reconstructions of fossils, particularly surface renderings are easily identi®ed by where further preparation is inadvisable or combining them in a single image, which prohibited. provides a valuable aid during the initial in- Another less complete shark braincase 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 9 from the Devonian of Wildungen is housed iobatis vetustus was erected originally by in the Humboldt University Museum (Ber- Eastman (1897), but his description is highly lin). This specimen was described by StensioÈ erroneous. The type specimen (NMNH 1717) (1937), who referred it to Cladodus wildun- is an isolated braincase, which lacks unam- gensis. His paper actually appeared ®rst, giv- biguous diagnostic characters although it car- ing Gross (1937) an opportunity to dispute ries the taxon name. The specimen was col- the identi®cation and refer StensioÈ's speci- lected in Powell County, Kentucky, but its men to a new species, C. hassiacus. The two exact provenance and horizon are uncertain. specimens are super®cially similar but differ Romer (1964) and Schaeffer (1981) consid- in size and matrix lithology, suggesting that ered it to be early Mississippian in age, but they may have come from different localities Williams (1998) suggested that it may be late and/or stratigraphic horizons. Schaeffer Devonian. When the specimen was originally (1981: 42) found the morphological differ- collected, it was embedded in a limestone ences between the two specimens ``mostly nodule with only its ventral surface exposed trivial or dif®cult to evaluate''. However, (identi®ed as dorsal by Eastman, 1897), but now that the internal morphology of the it was entirely removed from the matrix by Senckenberg specimen is known, profound acid preparation prior to Romer's (1964) de- differences from StensioÈ's (1937) description scription. While this facilitated its investi- of the otic region have emerged, although in gation, apparently no matrix residues were many other respects the two specimens seem retained, so samples are now unavailable for similar. Therefore, either StensioÈ's (1937) in- micropaleontological determination. terpretation of the otic capsule in the Berlin Schaeffer (1981) referred an isolated specimen is inaccurate, or two radically dif- braincase from the early Mississippian of ferent taxa are represented by the Sencken- Bedford (Indiana) to ``Tamiobatis sp.'' berg and Berlin specimens. In this paper, (AMNH 2140). Subsequently, Williams only the Senckenberg specimen is referred to (1998) referred a more complete fossil Cladodoides, pending a review of StensioÈ's (CMNH 9280, including a braincase, jaws, specimen. teeth, denticles, and the impression of a ®n- Throughout this work, comparisons are spine) from the late Devonian Cleveland made with braincases of other Paleozoic Shale of Bedford (Ohio) to the type species sharks, especially those referred to Tamio- T. vetustus. However, the emended diagnosis batis and Xenacanthus by Romer (1964) and provided by Williams (1998) consists only of Schaeffer (1981). There is little doubt that all dental characters, which are inappropriate the Permian specimens from Texas and both for the genoholotype NMNH 1717 and Oklahoma referred to Xenacanthus by for AMNH 2140. Thus, three specimens Schaeffer (1981) represent xenacanth sharks, from differing horizons and localities have as these are by far the most abundant elas- now been referred to Tamiobatis, yet there mobranch remains in those Permian red- are still no reliable criteria to distinguish the beds. Unfortunately, most of these skeletal holotype of the type species from other Pa- elements are isolated (apart from the spec- leozoic sharks with similar cranial morphol- tacular specimen collected by A.S. Romer ogy (e.g., Ctenacanthus). Furthermore, while and illustrated by Schaeffer, 1981: ®g. 1). the braincases CMNH 9280 and NMNH However, these are very similar to more 1717 are generally similar, the arrangements complete articulated fossils from Europe re- of their basicranial foramina differ slightly, ferred to Orthacanthus (e.g., O. senckenber- suggesting that they may represent different gianus), and much of the American xena- taxa. Williams (1998) compounded the prob- canth material should be referred to Ortha- lem further by referring additional species canthus rather than Xenacanthus (e.g., O. (all founded on isolated teeth) to the genus texensis; Heidtke, 1998, 1999; Johnson, Tamiobatis (T. succinctus, wachsmuthi, and 1999). springeri). While all the material investigated Unfortunately, the systematic status of by Williams (1998) could represent a distinct Tamiobatis is poorly resolved, although for genus based on dental morphology alone, its different reasons than for Cladodoides. Tam- inclusion in Tamiobatis is unjusti®ed and 10 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 cannot be supported by cranial synapomor- sions for tooth families. Cladodus elegans phies with NMNH 1717. Thus, there is no teeth are associated with the midorbital frag- compelling evidence that the three braincases ment and the isolated occipital region. These already referred to Tamiobatis are really con- fragments are almost identical to the corre- generic, although there is every reason to sponding regions of the ``Tamiobatis sp.'' suppose they pertain to quite similar sharks, braincase described by Schaeffer (1981), and given their overall morphological similarity. the referred palatoquadrate is ``cleaver- In the absence of generically diagnostic char- shaped'', with a strong postorbital process acters for Tamiobatis, the genus should prob- bearing an articular surface (like the ctena- ably be con®ned to the holotype of the type canth palatoquadrate described by Williams, species, NMNH 1717. 1998), suggesting that the cranial endoskel- Comparison of Cladodoides with other Pa- etons of Cladodus elegans (and C. mirabi- leozoic sharks is limited by paucity of data, lis?), Tamiobatis vetustus,``Tamiobatis sp.'', especially concerning internal features of the and Cladodoides were very similar. braincase (these have only been described in At various places in the text, references are two other forms, ``Tamiobatis sp.'' and Or- made to an isolated symmoriid braincase thacanthus; Schaeffer, 1981). The isolated (``Cobelodus'') from the Pennsylvanian Fay- ``Tamiobatis sp.'' braincase (AMNH 2140) is etteville Shales of Arkansas (Maisey, 2004a); of importance to this work because its inter- a detailed description of this specimen is still nal morphology is revealed in a series of hor- in preparation. In addition, Schaeffer (1981) izontal slabs that can be compared with scans tentatively referred three specimens with of Cladodoides. However, a detailed descrip- braincases from the Cleveland Shale to Cten- tion is beyond the scope of this work, and acanthus (CMNH 5965, 6919, 7852), but only some of its features are discussed be- these specimens have not yet been investi- low. AMNH 2140 is far larger and more gated or described and they are omitted from heavily calci®ed than the Cladodoides brain- further discussion here. case. Features of cranial morphology in Or- thacanthus are taken mainly from Schaeffer ANATOMICAL ABBREVIATIONS (1981) and Maisey (1983). aa anterior ampulla Recently, some fragmentary skeletal re- acv foramen for anterior cerebral vein mains associated with teeth identi®ed as Cla- adll passage for anterodorsal lateral line dodus elegans were discovered in the British nerve Geological Survey collections (M. Ginter, appr anterior postorbital process (arthrodire) personal commun., 2003). That material is asc anterior semicircular canal not necessarily from a single individual and avll? passage for anteroventral lateral line not all the fragments are associated with nerve teeth, but all are from the Lower Carbonif- cc crus commune erous Dockra Limestone of Lugton Quarry, cer cerebellar region dor dorsal otic ridge Ayrshire, Scotland. Teeth of C. elegans are ds dorsum sellae very similar to those of C. mirabilis (the type ea external (horizontal) ampulla species), and this discovery may therefore ebra1 efferent branchial artery help clarify the systematic position of the ge- eha efferent hyoidean artery nus. The remains (which will be described epsa efferent pseudobranchial artery elsewhere) include: the midorbital region of end f endolymphatic foramen a large braincase, with wide supraorbital esc external (horizontal) semicircular canal crests and most of the orbital wall; a large fm foramen magnum central section of another braincase; a com- g median basicranial groove plete occipital arch, with the posterior part of gc glossopharyngeal canal gr prof? groove for profundal nerve the otic capsule and a broken lateral otic pro- hl hypotic lamina cess; associated jaw elements, including the hm art position of hyomandibular articulation right palatoquadrate, Meckel's cartilage, and hyp ch hypophyseal chamber ceratohyal; and the anterior part of a left pal- hyp f bucco-hypophyseal fenestra atoquadrate, showing two or three depres- ic a internal carotid artery 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 11 iva foramen for intervertebral artery IV trochlear nerve jc jugular canal V trigeminal nerve lda lateral dorsal aorta V md mandibular ramus of trigeminal nerve lhr lateral hypophyseal ridge VI abducent nerve lof lateral otic fossa VII facial nerve lor lateral otic ridge VII ar anterior ramule of palatine ramus mdv? inferred position of mandibular vein VII h hyomandibular trunk of facial nerve med medullary region VII hy hyoidean ramus of facial nerve mes mesencephalic region VII pal palatine ramus of facial nerve met f metotic ®ssure VII pr posterior ramule of palatine ramus ms median suture in braincase roof VIII octaval (acousticovestibular) nerve not can notochordal canal IX glossopharyngeal nerve n soph notch for super®cial ophthalmic ramus X vagal nerve oc a occipital arch oc cot occipital cotylus Note that this list does not include abbre- olf c olfactory canal viations for the ®gures reproduced from oof otico-occipital ®ssure Gross (1937) and StensioÈ (1937); see cap- opa optic (retinal) artery tions for details. Terminology for cranial opha ophthalmic artery nerves here follows Northcutt and Bemis ora orbital artery (1993) in recognizing a distinct series of oc- or art orbital articulation of palatoquadrate tavolateral nerves (including the octaval and pa posterior ampulla lateral line nerves), instead of including them pac preampullary canal as parts of the facial, glossopharyngeal, and pbc posterior basicapsular commissure vagal nerves. The profundal is also treated as pcc precarotid commissure a separate nerve, rather than a branch of the pdf posterior dorsal fontanelle (ϭ parietal or endolymphatic fossa of authors) trigeminal nerve. All nerve names are angli- ped attachment area for optic pedicel cized. pf? possible pineal organ p fen perilymphatic fenestra DESCRIPTION pit v pituitary vein GENERAL FEATURES,TAPHONOMY, AND po art postorbital articulation for palatoquad- FORENSIC OBSERVATIONS rate popr postorbital process Comparison of the surface renderings pre- pppr posterior postorbital process (arthrodire) sented below with the illustrations in Gross prcf precerebral fontanelle (1937; see ®g. 2 here) shows that CT scan- prfc prefacial commissure ning has captured the entire morphology of prof profundal nerve foramen the Senckenberg Museum braincase in great psc posterior semicircular canal detail. Scanning provides information not psr presphenoid ridge pt posterior tectum only about the morphology of a fossil spec- re insertion of external rectus muscle imen, but also about its postmortem tapho- sac saccular chamber nomic history and its subsequent forensic snc subnotochordal ``chamber'' history (preparation, conservation, damage sns subnotochordal septum and repair). Diagenetic observations of inter- soc spino-occipital nerve canals (in the en- est in the Senckenberg braincase include the docast) lack of compression or subsequent distortion sof spino-occipital nerve foramina throughout the specimen; the presence of a soph super®cial ophthalmic complex (tri- complete occipital region still enclosed by geminal ϩ anterodorsal lateral line matrix; the uniform carbonate matrix com- nerves) pletely ®lling the braincase, leaving no voids, sub s suborbital shelf sup s supraorbital shelf secondary crystalline areas, or pyrite over- tel telencephalic region growths; and the absence of bioturbation. Fo- ur utricular recess rensic observations of interest include dam- vlc vesticulolateral (auricular) chamber age and slight repair to the right postorbital II optic nerve process (especially its tip), and unrepaired III oculomotor nerve damage to the lateral walls of both otic cap- 12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 4. Dorsal view of the Cladodoides braincase (surface rendering).

sules. There is no evidence of any dermal cess separates the orbital region from the structures (teeth, denticles) or parts of the slightly longer otico-occipital region. Miss- postcranial and visceral skeleton in the ing areas include most of the antorbital re- scanned matrix, and the braincase appears to gion (although the posterior part of the pre- be completely isolated. Only a very thin lay- cerebral fontanelle is still present) and the er of prismatically calci®ed cartilage covered lateral otic process farther posteriorly. These the inner and outer surfaces of the braincase, regions have not been reconstructed here. rather than heavy or multiple-layered calci- Anteriorly, combined surface and volume ®cation seen in other Paleozoic sharks such renderings reveal a matrix-®lled notch (not as Orthacanthus. The delicate nature of this observed by Gross, 1937) for the super®cial specimen suggests it underwent very little ophthalmic ramus (®g. 3). The occipital arch postmortem disturbance or transportation. is also covered by matrix in the specimen but Gross (1937) noted that the jaws ®gured by is revealed by the scan and is clearly sepa- Jaekel (1921) came from the same locality rated from the otic region by a persistent oti- and are in similar matrix, suggesting that co-occipital ®ssure. these specimens may pertain to a single tax- In ventral view (®g. 5), the basicranium on if not to one individual. forms an unbroken surface, with no ventral In dorsal view (®g. 4), the postorbital prootic ®ssure. Anteriorly, the ¯oor of the eth- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 13

Fig. 4 Continued.

moid region is missing. There is a large buc- tures exhibited here are obscured by matrix co-hypophyseal fenestra leading into a ma- and have never before been observed. The trix-®lled central opening for the hypophy- internal openings of all the major cranial seal chamber. nerves and blood vessels can be recognized, The lateral view (®g. 6) shows the major and many details of labyrinth morphology foramina within the orbit and also reveals the are also evident. The dorsum sellae is partic- occipital arch for the ®rst time, including ularly prominent, extending from below the several foramina for spino-occipital nerves oculomotor foramen and sloping posteriorly and intervertebral arteries. Features in the beneath the abducent foramen. A persistent posterior part of the orbit are obscured in this otico-occipital ®ssure is also seen, separating view by the postorbital process but are illus- the otic capsule from the lateral wall of the trated later in this work. occipital region. The medial view (®g. 7) dramatically il- In the anterior and posterior views of the lustrates the investigative power of CT scan- braincase (®g. 8) the dorsum sellae is again ning in paleontology, by reconstructing the observed deep within the cranial cavity, internal morphology of the braincase as if forming a prominent transverse wall. Also sliced through the sagittal plane. All the fea- visible are large openings in the orbital sur- 14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 5. Ventral view of the Cladodoides braincase (surface rendering). face of the postorbital process; these are Throughout the braincase, the prismatic cal- completely obscured by matrix and were not ci®cation is thin and there is no evidence of observed by Gross (1937) or Schaeffer multiple layers, thickened regions of pris- (1981). The extent of the otico-occipital ®s- matic tissue, or calci®ed ®brocartilage. This sure is particularly evident in posterior view. observation is important, because Schaeffer The braincase is quite small (its maximum (1981) claimed that multiple-layered calci®- length, including the unexposed occipital re- cation was one of the few characters puta- gion, is approximately 65 mm), but it is es- tively linking Cladodoides with Tamiobatis sentially intact and has suffered minimal dis- and Orthacanthus.IftheCladodoides brain- tortion. Despite its small size, the braincase case represents a juvenile or submature in- is calci®ed super®cially (both internally and dividual (a distinct possibility, discussed be- externally) with prismatic hard tissue (a low), multiple-layered calci®cation may not chondrichthyan characteristic), although yet have formed. Multiple prism layers have some regions may have been unmineralized, been reported in the braincase of a stetha- and internal features did not show up well canthid (Akmonistion; Coates and Sequeira, locally because of poor X-ray density con- 1998), but they are not as extensive or thick trast between the phosphatic calci®cation of as in Cladodus elegans, Tamiobatis, Ortha- the braincase and the limestone matrix. canthus, or ctenacanth sharks from the 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 15

Fig. 5. Continued.

Cleveland Shale. Note that in cladoselachi- foramen (®g. 7), forming a low presphenoid ans from the Cleveland Shale the braincase ridge or ledge (Praesphenoidvorsprung; Ge- is also thinly mineralized and less robust than genbaur, 1872). This ridge is well developed in ctenacanths, although it has not yet been in some neoselachians but is weak or absent determined whether the prismatic calci®ca- in others (Maisey, 2004b). tion in cladoselachians consists of more than There is no evidence of a ventral ethmoid- a single layer. The fragmentary cranial ma- al keel, although one may have been devel- terial from Scotland referred to Cladodus oped farther anteriorly. In Orthacanthus, the elegans has multiple-layered calci®cation. basicranium narrows abruptly in front of the ETHMOID REGION: Almost all the ethmoid anterior palatoquadrate articulation to form a region is missing in the Senckenberg brain- narrow base for the rostrum (Schaeffer, 1981: case, and only a small part of the left olfac- ®gs. 5, 6). In Hybodus basanus, the basicra- tory canal is recognizable. No parts of the nium is also narrowed anteriorly, forming a postnasal wall, ectethmoid chambers, eth- caudal internasal keel beneath paired ethmoid and/or ectethmoid processes, orbito- mopalatine processes (Maisey, 1983: ®gs. 8, nasal canals, or olfactory capsules are pre- 9). In both those forms, the ethmoidal keel served. The telencephalic region of the en- is con®ned to the region anterior to the optic docranial cavity extends between the orbits foramen, whereas very little of this region is and is widest immediately below the optic preserved in the Senckenberg braincase. 16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 6. Lateral view of the Cladodoides braincase (surface rendering).

Coates and Sequeira (1998) did not ®nd any Chlamydoselachus (see also Holmgren, evidence of an ethmoidal keel in Akmonis- 1941: ®g. 11). However, while Chlamydo- tion, but even in their most complete brain- selachus may occupy a basal position among case the region anterior to their ``posterior neoselachians (ϭ crown group elasmo- ethmoid process'' is not preserved. The phy- branchs), there is no evidence that the taxon logenetic signi®cance of the ethmoidal keel represented by the fossil was more closely in elasmobranchs was recently reviewed related to Chlamydoselachus than to any oth- elsewhere (Maisey, 2004b: 41) and will not er modern elasmobranch, and therefore such be discussed here. a reconstruction is not justi®ed phylogeneti- The entire preorbital region of the Hum- cally, although the general morphology of boldt University ``Cladodus'' hassiacus the ethmoid region may have been similar. In braincase is also missing, but StensioÈ (1937: Chlamydoselachus, the ¯oor of the telence- ®g. 6) reconstructed it by superimposing the phalic chamber slopes upward anteriorly and ethmoid region of the modern frilled shark the ¯oor of the precerebral fontanelle is el- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 17

Fig. 7. Medial view of the Cladodoides braincase sliced through the sagittal plane (surface rendering). evated almost to the same level as the roof the unexposed ¯oor of the precerebral region of the braincase (Allis, 1923a; Maisey, farther posteriorly is also smooth. The course 2004b). This arrangement is atypical of most of the terminal nerve is unknown. neoselachians and may represent an autapo- The olfactory capsules and the antorbital morphy of the genus. Judging from the depth (postnasal) wall are not preserved in the of the preserved part of the telencephalic Senckenberg braincase, and its olfactory ca- chamber in the Senckenberg braincase, its nals are represented only by a slight bulge in fontanelle ¯oor was probably much deeper the lateral wall of the braincase just above than in Chlamydoselachus. the anterior palatoquadrate articulation. The As in elasmobranchs generally, the endo- olfactory tracts were probably extensive an- cranial chamber in Cladodoides is open anteroposteriorly, as in neoselachians. In Or- teriorly, forming a large precerebral fonta- thacanthus, the lateral wall of the olfactory nelle that intrudes slightly between the olfac- canal bulges strongly into the orbit behind tory canals as in Notorynchus (Maisey, the preorbital process (Schaeffer, 1981: ®g. 2004b). The exposed part of the fontanelle 2; Maisey, 1983: ®g. 14B). In neoselachians, ¯oor is smooth and con¯uent with the wall the olfactory canal is usually located medial of the olfactory canal. Scanning shows that to the main antorbital cartilage, which can 18 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 8. Anterior (A) and posterior (B) views of the Cladodoides braincase (surface renderings).

form a prominent partition separating the or- GION:InCladodoides, there are separate fo- bit from the ectethmoid chamber. ramina for the efferent pseudobranchial ar- Just anterior to the optic foramen is a very tery and the pituitary vein. These vessels small canal (not detected by Gross, 1937), therefore entered the braincase separately (as which may have contained the anterior ce- in most neoselachians, although they share rebral vein, since this vessel occupies a sim- an opening in some torpedo rays; Holmgren, ilar position in neoselachians. The canal 1941). In modern gnathostomes, these fo- opens internally and cannot therefore have ramina are important ontogenetic landmarks, transmitted the profundal nerve onto the dor- indicating the anterior and posterior limits of sal surface of the snout. Holmgren (1941) the polar cartilage, respectively (De Beer, noted that the profundal nerve shares a com- 1931, 1937). In Cladodoides, the efferent mon opening with the anterior cerebral vein pseudobranchial and pituitary foramina are in some specimens of Chlamydoselachus, but spaced far apart, suggesting that the contri- in others they are separate (as in other neo- bution made by the polar cartilage to the ba- selachians). In Cladodoides, the profundal sicranium was extensive. This is also sup- nerve probably left the orbit via a shallow ported by the extensive dorsum sellae and groove dorsal to the olfactory canal, but no hypophyseal chamber in the endocast (see profundal foramen has been identi®ed. below). If the positions of these foramina are EXTENT OF POLAR CARTILAGE-DERIVED RE- reliable indicators of the original boundaries 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 19

Fig. 8. Continued. of embryonic tissues in Cladodoides, the po- cartilage is comparatively small and makes lar cartilage formed perhaps half the orbital only a minor contribution to the basicranium ¯oor plus a considerable part of the basicra- (®g. 10; also see discussion). nium (®g. 9), including an extensive postpi- OPHTHALMIC AND EFFERENT PSEUDOBRAN- tuitary commissure that extended behind the CHIAL ARTERIES: Separate foramina for the internal carotids, and perhaps also forming ophthalmic and efferent pseudobranchial ar- the lateral walls of the bucco-hypophyseal teries were found by Gross (1937) in Cla- fenestra. The area derived from the polar car- dodoides. StensioÈ (1937) also found separate tilage in Cladodoides presumably extends foramina for these vessels in the Humboldt from the postpituitary commissure posteri- University braincase. The ophthalmic fora- orly to the level of the efferent pseudobran- men in both specimens is located ventral and chial foramina anteriorly, because the inter- slightly posterior to the optic foramen, adja- nal carotids can be traced anteriorly around cent to the articular surface for the palato- the side walls of the hypophyseal chamber quadrate. The efferent pseudobranchial fo- (see below). By contrast, in neoselachians ramen lies even farther ventrally, within the the efferent pseudobranchial foramen lies in posterolateral margin of the ¯ange forming the posterior part of the orbit, and the polar this articular surface. In CT scans, a passage 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 9. The Cladodoides braincase in lateral, ventral, and medial views, showing the inferred polar cartilage-derived region (gray area) and important related landmarks. Compare with neoselachians (®g. 10). is seen extending ventrolaterally from the ranchial artery entered the cranial cavity lat- base of the optic canal. This passage branch- erally and above the trabeculae, as in modern es distally within the thickness of the orbit elasmobranchs. A similar arrangement has wall and opens externally at the ophthalmic also been recognized in Orthacanthus and and efferent pseudobranchial foramina (seen Hybodus (Schaeffer, 1981; Maisey, 1983). in the endocast and in transverse scan sec- In Orthacanthus, the ophthalmic and ef- tions through the braincase; ®gs. 11, 12). In ferent pseudobranchial foramina are situated Cladodoides, therefore, the efferent pseudob- much farther posteriorly than in Cladodoides 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 21

Fig. 10. The extent of the polar cartilage-derived region (gray area) in neoselachians. (A) Squalus 50-mm stage, lateral view; (B) Scyliorhinus 36-mm stage, lateral view; (C, D) Heterodontus 62-mm stage, lateral (C) and medial (D) views. Panels A and B are after De Beer 1931; panels C and D are after Holmgren 1941.

(Schaeffer, 1981: 19). In Tamiobatis vetustus attention, and very few features are visible (NMNH 1717) and ``Tamiobatis sp.'' laterally in the slabbed pieces. It is therefore (AMNH 2140), Schaeffer (1981) found only possible that a separate ophthalmic foramen a single foramen and suggested that the ef- is present in both fossils, but simply cannot ferent pseudobranchial and ophthalmic arter- be observed. The position of the efferent ies may have exited together and separated pseudobranchial in CMNH 9280 (referred to within the orbit. However, the opening he Tamiobatis vetustus by Williams, 1998) is found corresponds topographically to the ef- also uncertain, but it may be represented by ferent pseudobranchial foramen in Cladodoi- a semicircular indentation in the suborbital des and Orthacanthus. In fact, it cannot be cartilage slightly anterior to the bucco-hy- determined if there is only a single foramen pophyseal fenestra. This would place the ar- in NMNH 1717 because the lower part of the tery in a midorbital position, as in Orthacan- orbital wall (including the expected position thus. of an ophthalmic foramen) is crushed against Coates and Sequeira (1998) found a small the suborbital shelf. The ``Tamiobatis sp.'' triangular projection in the ethmoid region of braincase (AMNH 2140) was cut into thick Akmonistion, which they termed the posterior horizontal slabs before it came to scienti®c ethmoid process. In one specimen, they ob- 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

polar and trabecular cartilages in neoselachians (see discussion later in this work). In Orthacanthus, the articular surface is more complex than in Cladodoides and is scalloped by two or three transverse ridges separated by wide, shallow concavities (Schaeffer, 1981), and there are corresponding ridges and grooves on the medial articular surface of the palatoquadrate (Hotton, 1952). In the slabbed braincase AMNH 2140 (referred to ``Tamiobatis sp.'' by Schaeffer, 1981: ®gs. 22G, 23G), the articular surface includes at least two ridges and deep grooves. Unfortunately, in the holotype of Fig. 11. Cladodoides endocast; detail of pas- Tamiobatis vetustus (USNM 1717), the cor- sages for optic nerve and ophthalmic and efferent responding region is missing, although a sub- pseudobranchial arteries (surface rendering, orbital shelf is present. The anterior parts of oblique anterolateral view). the palatoquadrates and braincase are missing in the articulated specimen referred to Tamiobatis vetustus by Williams (1998; CMNH 9280), but its suborbital shelf widens served a canal in the broken cartilage just appreciably just behind the expected position anterior to this process, which they inter- of an articular surface. An articular process preted as possibly containing the prehypo- with ridges and grooves is clearly present in physeal branch of the internal carotid artery, isolated palatoquadrates from the Cleveland or the optic or ophthalmic artery. Compari- Shale (e.g., CMNH 9450, probably from a son with Cladodoides suggests that this canal ctenacanth) and in many isolated shark pal- may indeed have carried the ophthalmic ar- atoquadrates from the Pennsylvanian black tery. An efferent pseudobranchial foramen shales of Indiana and Nebraska (Zangerl and has not been identi®ed in Akmonistion; Case, 1976; Williams, 1985). In Cobelodus Coates and Sequeira (1998) suggested that aculeatus and Akmonistion zangerli, there is the artery entered the cranial cavity farther a prominent ¯ange on the palatine part of the posteriorly, but alternatively it could have palatoquadrate plus an articular surface on joined the ophthalmic artery within the orbit. the internasal plate (Zangerl and Case, 1976, ANTERIOR PALATOQUADRATE ARTICULA- Coates and Sequeira, 2001b). TION:InCladodoides, an articular surface is BUCCO-HYPOPHYSEAL FENESTRA AND IN- present on the lateral wall of the internasal TERNAL CAROTID ARTERIES: The large size of plate, extending from below the ethmoid re- the bucco-hypophyseal fenestra in Cladodoi- gion into the anterior part of the orbit, and des may indicate an immature condition meeting the suborbital shelf posteriorly. It (®gs. 5, 13), since this part of the basicrani- consists of a single transverse ridge ¯anked um is one of the last to chondrify in extant by wide, slightly concave areas forming a elasmobranchs (De Beer, 1931). However, connection with the anterior part of the pal- the basicapsular fenestra is closed and the atoquadrate (Gross, 1937, 1938) and has parachordal region is completely chondri®ed, been termed either an orbital articulation as in adult neoselachians. The precarotid (e.g., Gross, 1937) or an ethmoidal articula- commissure was not mineralized and proba- tion (e.g., Maisey, 1980; Schaeffer, 1981; bly was not chondri®ed, leaving the internal Coates and Sequeira, 1998). The efferent carotid arteries exposed beneath the brain- pseudobranchial foramen in Cladodoides is case before they entered the fenestra. This located at the boundary of the articular and also suggests immaturity, and the precarotid nonarticular regions of the suborbital carti- commissure may therefore have chondri®ed lage, suggesting that the boundary corre- later in ontogeny, ultimately creating sepa- sponds to the contact between the embryonic rate canals for the internal carotids. However, 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 23

Fig. 12. Transverse CT scan slices of the Cladodoides braincase (levels of slices indicated by white lines on lateral view). (A) Slice 68, through optic foramen; (B) slice 99, through oculomotor and trochlear foramina; (C) slice 106, through pedicel ridge and hypophyseal chamber; (D) slice 110, through hypophyseal chamber and bucco-hypophyseal fenestra. the bucco-hypophyseal fenestra is small in be described below, in the section on the en- the Humboldt University braincase (a larger docast). specimen), and separate canals for the inter- The internal carotid arteries evidently con- nal carotid arteries are present (StensioÈ, verged gradually but probably did not meet 1937). The bucco-hypophyseal fenestra in before they entered the braincase, because Cladodoides is pinched off from the overly- separate arterial grooves can be traced on ing hypophyseal chamber internally and each side of the bucco-hypophyseal fenestra. there is only a single opening for the internal By contrast, in neoselachians these vessels carotids and hypophyseal duct in its roof. converge almost at a right-angle to the mid- The fenestra is probably extracranial and its line, forming part of a ``bell-shaped'' vas- roof is probably formed from the posterior cular loop which is con®ned to the area be- end of the trabecular-polar cartilage as in hind the bucco-hypophyseal fenestra neoselachians (the hypophyseal chamber will (Schaeffer, 1981). The subsequent course of 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 13. Detail of the hypophyseal region in Cladodoides.(A) Medial view; (B) ventral view. Anterior to left in both views. The probable course of the internal carotid artery is indicated by a series of dots. After entering the bucco-hypophyseal fenestra (which is essentially an extracranial space), the artery continues around its margin before it enters the braincase via a central opening in the ¯oor of the hypophyseal chamber. these arteries cannot be traced in Cladodoi- des, but they presumably joined the orbital, ophthalmic, and efferent pseudobranchial arteries anteriorly and the basilar artery farther posteriorly. In Tamiobatis vetustus (NMNH 1717), the bucco-hypophyseal fenestra appears to be very large. Originally the fenestra may have been smaller, since the surface of the hypophyseal region has been lost by erosion (Romer, 1964; Schaeffer, 1981). It is also

physeal fenestra; (B) internal surface of basicra- → nium, with internal carotids entering cranial cavity separately. The arteries separated from each other Fig. 14. ``Tamiobatis sp.'' braincase AMNH and turned laterally within the thickness of the 2140, details of the hypophyseal region in slabbed basicranium. They did not extend medially across specimen, showing the relationship of the internal the bucco-hypophyseal fenestra nor around its lat- carotid canals and bucco-hypophyseal fenestra. eral walls as in Cladodoides, and they were en- (A) Ventral surface, with internal carotids entering closed by cartilage internally (the precarotid com- basicranium close together behind bucco-hypo- missure?). 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 25 possible that a precarotid commissure was doselache, the internal carotids probably lay originally present. The specimen has been re- in open grooves in the basicranium, not with- stored with its internal carotid canals either in enclosed canals (Williams, 1998; Coates completely enclosed (Schaeffer, 1981) or and Sequeira, 1998; cf. Schaeffer, 1981). only partly closed (Williams, 1998). Cobelodus aculeatus has been restored with The hypophyseal region is exposed on internal carotids in canals (Zangerl and Case, both sides of the lowermost slab of the 1976; Schaeffer, 1981; Coates and Sequeira, ``Tamiobatis sp.'' braincase (AMNH 2140; 1998), but the better preserved ``Cobelodus'' ®g. 14). On the lower surface of this slab, braincase casts doubt on these earlier recon- the bucco-hypophyseal fenestra is bisected structions, and the internal carotid arteries of externally by a narrow area of prismatic cal- symmoriid sharks probably did not follow a ci®cation which is continuous with the walls conventional arrangement (Maisey, 2004a). of the internal carotid canals. However, on In fact, there is little evidence that internal the internal surface of the same slab, the in- carotids were enclosed completely by basi- ternal carotid canals enter the cranial cavity cranial canals in any fossil or extant elas- much farther laterally. Presumably the canals mobranch. turn sharply outward within the thickness of In neoselachians, the internal carotids may the basicranial cartilage. Thus, the internal enter the braincase either separately or to- carotids in ``Tamiobatis sp.'' passed oblique- gether, but no clear phylogenetic pattern has ly through the basicranium on either side of emerged (Holmgren, 1941: char. 9). It is the bucco-hypophyseal fenestra, as in Cla- therefore doubtful whether much phyloge- dodoides (and in Tamiobatis vetustus?). The netic signi®cance can be attached to the pres- ventral median calci®cation probably repre- ence of single or paired internal carotid fo- sents part of the ¯oor of the overlying hy- ramina in extinct sharks (cf. Coates and Se- pophyseal chamber and may have contained queira,1998: char. 6), although paired foram- a median hypophyseal duct, unlike in Cla- ina are certainly more usual in Paleozoic dodoides, where the duct probably entered taxa. A single carotid foramen is present in the basicranium with the internal carotids. Tristychius and Hybodus (Dick, 1978; Mais- The precarotid commissure in ``Tamiobatis ey, 1983). Internal carotids are absent in chi- sp.'' was evidently extensive and heavily maeroids, and their brain receives its blood mineralized. The hypophyseal region is also supply entirely from the efferent ``pseudo- preserved in one of the fragments referred branchial'' artery (De Beer and Moy-Thom- tentatively to Cladodus elegans. Its internal as, 1935). carotid canals are extensive and converge an- OPTIC PEDICEL,OPTIC NERVE, AND OCU- teriorly, but the material is too fragmentary LOMOTOR NERVE:InCladodoides, the orbital to determine the original relationship of the wall contains a large optic foramen centrally vessels and the bucco-hypophyseal fenestra. (®gs. 1, 2, 6), which probably contained the Coates and Sequeira (1998: ®g. 5) de- optic (retinal) artery as well as the optic scribed paired internal carotid foramina in nerve (as in Chlamydoselachus; Allis, Akmonistion, separated by a thin median wall 1923a). Immediately behind this foramen is or septum similar to that in ``Tamiobatis sp.'' an extensive longitudinal ridge, which Gross Their observation that the hypophyseal and (1937) interpreted as the basal attachment internal carotid openings were ``coincident'' area for the optic pedicel. That interpretation in Akmonistion suggests that the precarotid is supported by the topographic position of commissure was absent. In Orthacanthus, the ridge, the absence of calci®cation distally there is a single median internal carotid fo- (where the pedicel would attach in life), and ramen separated from the persistent bucco- the fact that the internal surface of the car- hypophyseal fenestra by a broad precarotid tilage beneath the ridge is continuously cal- commissure. Schaeffer (1981) reconstructed ci®ed, with no foramina for nerves or blood its bucco-hypophyseal fenestra with a deep, vessels passing through it. Gross (1937) de- tapering canal connected to the hypophyseal picted the apical part of the pedicel ridge as chamber, although this region was not well rough and largely uncalci®ed, but CT scan preserved in the material available. In Cla- slices of the unprepared right orbit show that 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

this is not clear from the specimens he illustrated. In the slabbed ``Tamiobatis sp.'' braincase (AMNH 2140), the presumed pedicel attachment site lies in the same horizontal plane as the hypophyseal chamber, as in Cladodoides (Schaeffer, 1981: ®g. 23F). In Cladodus elegans, the orbital wall is in¯ated in the expected position of the optic pedicel behind a large foramen (probably for the optic nerve), but other features typically ac- companying the pedicel have not been found. In neoselachians, the base of the optic pedicel lies just behind and below the oculomotor foramen. Judging from the position of the foramen in Cladodoides, the anterior part of this ridge probably did not actually support the pedicel, which probably arose farther posteriorly near the abducent foramen and the presumed origin of the posterior rectus muscle (see below). An optic pedicel is pre- Fig. 15. Ventral region of the orbit in the sent in Pucapampella but is unknown in Humboldt University specimen (``Cladodus has- Gladbachus or Doliodus. siacus''). (A) From StensioÈ 1937; (B) with certain In Cladodoides, the oculomotor foramen features reidenti®ed following the present inter- lies just above the middle of the pedicel ridge pretation of Cladodoides. The unlabeled feature in panel B is simply the clipped wall of the otic ®g. 6). The oculomotor nerve probably fol- capsule. Differing interpretations include the ex- lowed the same course as in neoselachians, ternal rectus insertion (previously identi®ed as the leaving the brainstem near the midline of the trigeminal foramen), the inferred position of the mesencephalon, then entering the orbit and anteroventral lateral line nerve, and reinterpreta- dividing into a dorsal branch for the superior tion of the supposed hyomandibular facet as part and anterior rectus eye muscles, and a ventral of the lateral otic fossa. branch which subdivides into branches for the inferior oblique and inferior erectus eye muscles (e.g., Notorynchus; Daniel, 1934). the ridge is extensively calci®ed except at its This area is missing in the Humboldt Uni- very tip, suggesting that the left ridge was versity braincase, and StensioÈ (1937) did not damaged during preparation. A similar ar- identify the oculomotor foramen, which pre- rangement was described by StensioÈ (1937: sumably lay farther dorsally. ®g. 7) in the Humboldt University braincase. TROCHLEAR NERVE: In neoselachians, the He showed both an ``eye stalk area'' and an trochlear nerve arises in the mesencephalon, actual ``eye stalk process'' corresponding to and then its roots cross and the nerve leaves the longitudinal ridge in the Senckenberg the roof of the brain between the mesen- specimen (®gs. 15, 16). cephalon and cerebellum, entering the orbit Transverse CT scan sections show that the to supply the superior oblique eye muscle. pedicel ridge in Cladodoides is positioned The trochlear foramen in Cladodoides lies lateral to the hypophyseal chamber (®g. 12), almost directly above the oculomotor fora- unlike in modern sharks where the pedicel men, and the CT scan shows that both nerve attachment area is typically located above the openings lead directly into the cranial cavity. level of the chamber (e.g., Chlamydosela- The region of the orbit including the troch- chus; Allis, 1923a; Echinorhinus, Centros- lear foramen is probably missing in the Hum- cyllium, Squaliolus; Shirai, 1992). According boldt University braincase. Schaeffer (1981) to Schaeffer's (1981) reconstruction of Or- identi®ed the trochlear foramen of Orthacan- thacanthus, the pedicel seems to be posi- thus in an unusually anterior position (in tioned above the hypophyseal chamber, but front of the optic foramen), but it is located 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 27

Fig. 16. Surface rendering of the scanned Cladodoides braincase with clipping planes cutting through the postorbital process and suborbital shelf. Both views are reversed to facilitate comparison with ®gure 15. The clipping plane in panel A is almost parallel to the sagittal plane. In panel B the plane is more oblique, cutting the posterior part of the braincase farther laterally and the anterior part farther medially. Orientation data for clipping planes are as follows: Plane (axis, position, rotation); A (X, 14.08, 77.0; Y, 5.0, 109.0; Z, 11.34, 270.0); B (X, 10.56, 77.0, Y, 1.60, 121.0, Z, 0.80, 270.0). These values represent positions along and angular deviations from the surface rendering's original XYZ axes. farther posteriorly (i.e., in a more conven- the orbit, whereas the abducent nerve typi- tional position) in other Orthacanthus mate- cally arises caudoventrally below the trigem- rial (Maisey, 1983: ®g. 14B). The most likely inal and facial ganglia. Instead, the abducent explanation of this discrepancy is that nerve (assuming it had a separate foramen in Schaeffer's (1981) supposed trochlear fora- Cladodoides) may have exited much lower men is actually the anterior profundal fora- in the cranial wall via a small foramen ven- men. Hybodus basanus has been restored tral to the main trigeminal foramen. Gross with separate trochlear and profundal foram- (1937) considered this foramen to house an ina, both situated anteriorly in the orbit unspeci®ed branch of the facial nerve, but its (Maisey, 1983: ®g. 13A). passage has no obvious internal connection ABDUCENT NERVE: Gross (1937) consid- with the main facial foramen (see below). ered that the abducent nerve was separate in Furthermore, the foramen lies at the same Cladodoides and he identi®ed its foramen level in the orbit as the optic pedicel (al- within the jugular canal. However, such a po- though still some distance behind it), approx- sition would place the nerve rather high in imately where the abducent nerve leaves the 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 17. Postorbital region of Cladodoides braincase (surface rendering), oblique perspective detail views. (A) Posterolateral view; (B) anterolateral view; (C) anteromedial view; (D) posteromedial view. Differences from Gross (1937) involve the positions of the abducent and profundal nerve. The dorsum sellae overhangs the deep hypophyseal chamber, and the lateral hypophyseal ridge delimits the side wall of this chamber. The super®cial ophthalmic ramus passes through the postorbital process and emerges posteriorly just above the main trigeminal and facial nerve foramina. braincase in neoselachians. The supposed ab- scured by the postorbital process in lateral ducent foramen identi®ed by Gross (1937) view (®g. 6), but it can be seen in the oblique may instead have housed the profundal nerve views of the orbit and postorbital process (see next section). The foramen regarded (®g. 17B, D). here as housing the abducent nerve is ob- Several foramina are present in the ventral 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 29

Fig. 17. Continued. part of the orbit in the Humboldt University The foramen said to house an unspeci®ed fa- ``Cladodus'' hiassicus braincase (®g. 15) and cial ramus in Cladodoides by Gross (1937) their arrangement agrees closely with the apparently corresponds to StensioÈ's (1937) Senckenberg specimen (the corresponding foramen ``for the roots of the lateralis ®bres area in this specimen is shown in ®g. 16). to the trigeminus branches'' (Vl, ®g. 15A) in 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

``C.'' hiassicus. However, the corresponding though this is admittedly an unsatisfactory foramen in Cladodoides probably contained interpretation because the profundal nerve in the abducent nerve, and this is also possible neoselachians generally does not leave the in the case of ``C.'' hiassicus (®g. 11B). braincase separately (the condition in os- The position of the abducent foramen has teichthyans is more variable; Gardiner, not been determined in Tamiobatis vetustus 1984a: 238). The foramen in question is lo- or ``Tamiobatis sp.'', but in Orthacanthus it cated behind the postorbital process, in the is located next to a large trigemino-pituitary ventral part of a deep depression (``Gan- fossa high up in the orbit, within the opening gliongrube'' of Gross, 1937) which also con- of the jugular canal (Maisey, 1983: ®g. 14). tains the trigeminal foramen (®g. 17). Gross In Hybodus basanus, the abducent foramen (1937) suggested that this depression housed is lower in the orbit, sandwiched between the extracranial components of the trigeminal trigemino-pituitary fossa and the exit for the and anterodorsal lateral line ganglia. If that hyomandibular trunk of the facial nerve. is correct, the profundal ganglion presumably In gnathostomes, the abducent nerve arises lay farther ventrally, perhaps contacting the from a nucleus in the medulla and passes an- anterodorsal lateral line ganglion dorsally. terolaterally into the orbit, where it supplies Conceivably, the ganglia of both the facial's the external (ϭ posterior) rectus eye muscle. super®cial ophthalmic ramus and profundal In neoselachians, a separate abducent fora- nerve in Cladodoides were closely associated men is present in Oxynotus and Centropho- with each other, as in Mustelus (Norris and rus but not in Etmopterus (Holmgren, 1941). Hughes, 1920: ®g. 23). Both ganglia may In Chlamydoselachus, the abducent and tri- have occupied the supposed profundal fora- geminal nerves exit together in the upper part men, although such an arrangement would of the trigemino-pituitary fossa (Allis, separate them from the main trigeminal and 1923a), but in hexanchiforms the abducent anterodorsal lateral line trunks. Since all nerve has its own foramen. The abducent and these nerves typically occupy the prootic fo- trigeminal nerves also share a foramen in ramen in neoselachians, this interpretation Heterodontus, Torpedo, and in the chima- presumes that the embryonic prootic foramen eroid Hydrolagus (Gardiner, 1984a). Gardi- was partly closed in Cladodoides, leaving the ner (1984a: 248) concluded that in primitive two small openings observed. gnathostomes the abducent nerve left the cra- In neoselachians, the profundal ganglion is nial cavity below and behind the dorsum sel- usually distinct but its position is variable; in lae, and that in osteichthyans the dorsum sel- Squalus it is completely extracranial and lae was primitively ossi®ed by the basisphe- makes contact dorsally with the anterodorsal noid. The abducent foramen lies below and lateral line ganglion, but in Mustelus it is in- behind the dorsum sellae in Cladodoides ®g. tracranial (Norris and Hughes, 1920). In 7), and in chimaeroids and some neoselachi- Chlamydoselachus, the profundal nerve is- ans (e.g., Squalus), and this may represent sues through the trigeminal foramen (Allis, the primitive pattern for chondrichthyans. 1923a), not separately as stated by Gardiner However, in some modern sharks the abdu- (1984a: 240; this condition also occurs in cent foramen may be located above and be- many other neoselachians). The profundal hind the dorsum sellae, which may represent nerve usually passes deep within the orbit in an autapomorphic condition (e.g., Hetero- neoselachians, initially passing between the dontus). superior and inferior rectus muscles, then be- PROFUNDAL NERVE: Ever since Wijhe neath the anterior rectus and between the (1882) recognized that the profundal nerve oblique muscles (Daniel, 1934); the nerve in Scyliorhinus was distinct, there has been may have followed a similar course in Cla- a tendency to regard the profundal (V1) and dodoides. trigeminal (V2 ϩ V3) nerves as separate, cul- TRIGEMINAL NERVE: Gross (1937) associ- minating in the work of Northcutt and Bemis ated only one foramen with the trigeminal (1993). As stated above, the profundal nerve nerve in Cladodoides, within a deep depres- in Cladodoides may have lain within the sup- sion just behind the postorbital process and posed abducent foramen of Gross (1937), al- a considerable distance from the pituitary fo- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 31 ramen and from the pit for the external rectus ramina for branches of the facial nerve (``V muscle (both of which lie within the orbit; ϩ VII'', ``VII h'', ``RVII'', and ``rd''). How- ®g. 17). Schaeffer (1981: 45) found such a ever, his ``foramen V ϩ VII'' represents the trigeminal arrangement ``improbable'' and dorsal part of the jugular canal (see below) argued instead that the trunks of the trigem- and probably did not contain the facial nerve inal and facial nerves passed through the fo- proper, while his foramen ``RVII'' may be ramen ``RVII'' of Gross (1937). In fact, scan- the abducent foramen (see above). However, ning provides considerable support for his foramen ``VII h'' (located below and Gross' (1937) original observations. More- slightly behind the cartilage forming the ¯oor over, Schaeffer's (1981) proposals are incom- of the jugular canal in the postorbital pro- patible with the present interpretation of cess; ®g. 2) almost certainly contained the ``RVII'' as the abducent foramen. The main main facial (ϭ hyomandibular) trunk (®g. trunk of the trigeminal nerve apparently ex- 17). Adjacent to this foramen is another ited the braincase exactly where Gross smaller opening, which Gross (1937) identi- (1937) suggested, probably accompanied by ®ed as housing either a ramule of the facial the trunk of the anterodorsal lateral line nerve or an unspeci®ed blood vessel. Scan- nerve. Since the trigeminal nerve emerges in ning shows that this opening is con¯uent in- front of the postorbital wall in neoselachians, ternally with the main hyomandibular canal holocephalans, and osteichthyans, its more and that the two are merely separated super- posterior position in Cladodoides is unusual ®cially by a short cartilaginous bridge ex- (the mandibular ramus probably entered the tending between the postorbital process and postorbital process; see below and ®gs. 18B, otic capsule immediately below the prefacial 19). Nevertheless, in Acanthodes the man- commissure. The lowermost of the two ex- dibular ramus was apparently located even ternal openings de®ned by this bridge prob- farther posteriorly than in Cladodoides (be- ably contained the hyoidean ramus, while the hind the postorbital process; Miles, 1968, upper one probably housed the laterally di- 1973; Jarvik, 1977). rected mandibular ramus. The anteroventral The supposed trigeminal foramen in the lateral line nerve in Cladodoides probably braincase of ``Cladodus'' hiassicus (StensioÈ, left the braincase along with the hyomandib- 1937) corresponds closely with the pit for the ular trunk as in neoselachians, passing external rectus muscle in Cladodoides, which through a short passage immediately after raises three alternative possibilities. Either leaving the braincase (®g. 16). the trigeminal foramen is not preserved in Scanning reveals that the ``canal'' which ``C.'' hiassicus and lay instead farther dor- Schaeffer (1981) thought contained the or- sally (the preferred interpretation, since it bital artery does not in fact pass directly into agrees with the arrangement in Cladoidoides; the orbit as one might expect; instead, it runs compare ®gs. 15, 16); or the trigeminal and dorsally toward the hyomandibular foramen facial nerves may have shared a single fo- and then turns abruptly downward to enter ramen in ``C.'' hiassicus (a very different ar- the orbit at the base of the postorbital process rangement from Cladodoides and elasmo- (®g. 17). This ``canal'' therefore consists of branchs generally); or the supposed facial fo- two passages, one anterior to the other, which ramen in ``C.'' hiassicus might be for the tri- converge dorsally and probably contained geminal nerve (in which case the facial the anterior and posterior ramules of the pal- foramen presumably lay in a different posi- atine ramus (as proposed by Gross, 1937). tion than in Cladodoides, and has simply not The arrangement closely resembles that seen been found). in Squalus, where the palatine ramus passes The middle cerebral vein in Cladodoides through the ¯oor of the lateral commissure may have accompanied the trigeminal nerve ventrally (as in many squaloids; Norris and as in neoselachians (Gardiner, 1984a), al- Hughes, 1920; Holmgren, 1941). However, though there is no direct evidence of its po- the palatine ramus in Squalus divides into an- sition. terior and posterior ramules below the lateral FACIAL NERVE: Excluding the lateralis commissure, whereas in Cladodoides this di- components, Gross (1937) identi®ed four fo- vision occurs within the equivalent cartilage. 32 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 18. Transverse CT scan slices of the Cladodoides braincase (levels of slices indicated by white lines on lateral view). (A) Slice 126, through the postorbital process and the presumed origin of rectus muscle; (B) slice 139, through the jugular canal, showing the combined exit for the trigeminal and anterodorsal lateral line nerves, and the convergence of the anterior passage for the palatine ramus with the groove for the orbital artery; (C) slice 148, showing the separation of passages for the hyomandibular trunk and posterior branch of the palatine ramus; (D) slice 156, showing the exit of the hyomandibular trunk; (E) slice 161, showing the passage of the octaval nerve entering the otic capsule 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 33

Fig. 19. A series of clipping plane sections through the surface rendering of the left postorbital process in Cladodoides.(A) Lateral view, showing foramina for oculomotor, trochlear, and facial nerves and pituitary vein; (B) showing foramina for the anterior and posterior ramules of palatine ramus; (C, D) canal for mandibular ramus of trigeminal nerve (or main buccal-maxillary complex?); (E) showing alignment of trigeminal foramen with canal for mandibular ramus in postorbital process; (F) slice through jugular canal, with position of super®cial ophthalmic canal indicated by white dashed lines. This section just clips the anterior ampulla (in¯ated area behind trigeminal foramen). 34 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

In Squatina, the lateral commissure is more ramen lies behind the postorbital process. In extensively chondri®ed than in Squalus, but Tamiobatis vetustus (NMNH 1717) there is the palatine ramus arises behind the process a foramen in the lateral wall of the otic re- and does not pass through cartilage. gion slightly behind the postorbital process. In Cladodoides, the passage for the ante- According to Romer (1964) this contained rior ramule of the palatine ramus merges the middle cerebral vein, but comparison with a groove for the orbital artery just be- with Cladodoides, Orthacanthus, and ``Tam- hind the orbit. This opening apparently cor- iobatis sp.'' suggests that it housed the hyo- responds to ``foramen one'' of Williams mandibular trunk of the facial nerve. Romer (1998), while the posterior palatine foramen (1964) considered that the facial foramen lay corresponds to his ``foramen two''. Since higher on the lateral wall of the braincase previous interpretations of the palatine fo- immediately behind the dorsal part of the ramina in fossil sharks are inextricably inter- postorbital process, but comparison with woven with scenarios about the basicranial Cladodoides suggests it contained the tri- arteries, further discussion of the palatine geminal and anterodorsal lateral line nerves. nerve will be found in the section on the ar- The position of the facial foramen is un- terial circuit (see below). There is little doubt known in Cladodus elegans. that the palatine ramus in Cladodoides con- OCTAVOLATERAL NERVES: The anterodorsal tinued anteriorly to pass underneath the or- lateral line ganglion was probably extracra- bital articulation as in neoselachians. nial in Cladodoides, but it is not possible to A virtually identical arrangement is seen tell whether its super®cial ophthalmic and in StensioÈ's (1937) lateral view of the ``Cla- buccal portions were con¯uent (as in Squal- dodus'' hiassicus braincase (®g. 15A). Its us) or widely separated (as in Mustelus). main facial foramen is surrounded by three Gross (1937) found that the super®cial narrow cartilaginous projections forming ophthalmic ramus in Cladodoides entered the passages for the palatine rami and the main orbit via a foramen in the roof of the post- hyomandibular trunk. However, the latter orbital process before continuing anteriorly probably did not pass where StensioÈ (1937) in a shallow groove in the ventral surface of indicated (through a ``rostro-caudal facialis the supraorbital cartilage. Here, it gave off canal''). Instead, the hyomandibular trunk numerous dorsal ramules that pierced the was probably directed posteriorly and the cartilage to supply the supraorbital sensory ``facialis canal'' probably contained the an- canal. Scanning shows that the canal for this terior ramule of the palatine ramus and the nerve passes completely through the postor- abducent nerve (®g. 15B). The passage for bital process above the jugular canal and the posterior palatine ramule agrees in the emerges posteriorly just lateral to the pre- Senckenberg and Humboldt University spec- sumed exit of the trigeminal nerve (®gs. 16± imens. 19). However, the super®cial ophthalmic ra- The course of the facial nerve is known mus in ``Tamiobatis sp.'' (AMNH 2140) only in a few extinct sharks besides Clado- probably passed through the jugular canal doides. In the primitive neoselachian Syne- just below its roof, instead of through a sep- chodus there is a large foramen in the back arate canal (this was indicated as a shaded of the orbit, possibly for the hyomandibular area within the jugular canal by Schaeffer, trunk (Maisey, 1985). In Hybodus basanus, 1981: ®g. 23D). In both Cladodoides and there is also a large posteriorly directed ``Tamiobatis sp.'', the trigeminal and anter- opening for the hyomandibular trunk within odorsal lateral line nerves probably left the the jugular canal of the postorbital process braincase together, and the anterodorsal lat- (Maisey, 1983), and the palatine ramus may eral line nerve subsequently divided into su- have passed through a large opening in the per®cial ophthalmic and buccal rami behind chondri®ed ¯oor of the jugular canal (pre- the postorbital process. The super®cial oph- viously identi®ed as a foramen for the effer- thalmic ramus then passed through (``Tam- ent hyoidean artery). According to Schaeffer iobatis sp.'') or above (Cladodoides) the jug- (1981), in Orthacanthus and ``Tamiobatis ular canal, probably accompanied by the tri- sp.'' (AMNH 2140) the hyomandibular fo- geminal super®cial ophthalmic ramus. 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 35

Gross (1937) also suggested that the buc- contained the origin of the external rectus cal ramus in Cladodoides entered the post- muscle and are equidistant from the pituitary orbital process proximally (where it suppos- foramen anteriorly and the abducent foramen edly gave off dorsal ramules for part of the posteriorly, but they are remote from the tri- infraorbital sensory canal), although he did geminal and facial foramina. Gardiner not describe the course of the ramus in detail. (1984a) proposed that, in gnathostomes, the The present investigation supports his inter- external rectus muscles primitively originat- pretation in part, but the internal morphology ed posterior to the optic nerve on the inter- of the postorbital process is more complex orbital septum. Their inferred arrangement in than he supposed (see below). The course of Cladodoides agrees with this primitive ar- the buccal ramus has not been determined in rangement and there is certainly no evidence the Humboldt University specimen. of backward growth of the external rectus In both Cladodoides and ``Cladodus'' muscles to form a posterior myodome as in hiassicus, the anteroventral lateral line nerve actinopterygians. When internal and external may have passed through a small canal im- views of the braincase are compared, the rec- mediately anterior to the facial foramen in tus muscle pit and the dorsum sellae are seen the ¯oor of the jugular canal ®gs. 15, 16). to be aligned with each other on the lateral However, in ``C.'' hiassicus, StensioÈ (1937) and medial surfaces of the cartilage respec- simply identi®ed this canal as connecting the tively (e.g., ®gs. 6, 17). The pituitary fora- jugular and facial nerve canals (®g. 15). Nev- men, abducent foramen, and the pit for the ertheless, it is remarkably similar to a small external rectus muscle are all located in the canal thought to be for the anteroventral lat- ¯oor of the orbit at the anterior end of the eral line nerve in Cladodoides. jugular canal. Thus, all these structures are The octaval (acousticovestibular) nerve is separated from the trigeminal and facial fo- considered separately in the section on the ramina by the dorsal part of the postorbital otic region (see below). arcade. POSTERIOR PART OF ORBIT: According to The postorbital process also separates the Schaeffer (1981: 19), the external rectus trigeminal and facial foramina from the orbit musculature in Paleozoic sharks supposedly in Tamiobatis vetustus and ``Tamiobatis sp. originated in a large fossa, ``in part the tri- In ``Tamiobatis sp.'' (AMNH 2140), the fa- gemino-pituitary fossa of Allis (1923)''. cial, trigeminal, and anterodorsal lateral line However, the structure he identi®ed as a tri- nerves all left the braincase behind the post- gemino-facial recess in Cladodoides is in re- orbital process (Schaeffer, 1981: ®g. 23C, E, ality little more than a depression in the post- F), and the anterodorsal lateral line nerve orbital wall that may have housed the lateral may have accompanied the trigeminal as head vein or posterior rectus muscles (®g. postulated in Cladodoides. Accordingly, the 20A), but certainly not the trigeminal or fa- supposed trigemino-facial recess in ``Tam- cial ganglia, which were excluded from the iobatis sp.'' probably did not house these orbit by the postorbital process, as Gross nerves any more than in Cladodoides, al- (1937) originally recognized (®gs. 2, 6, 8A, though it probably housed the external rectus 17, 19). While it is true that the trigeminal muscle. It is therefore concluded that these foramen in Cladodoides lies within a small Paleozoic sharks lacked a trigemino-pituitary recess, this is located behind the postorbital recess like that described by Allis (1914, process and is completely separated from the 1923a) in neoselachians, or even a trigemino- pituitary foramen, rectus muscle insertion, facial recess of the kind envisaged by and abducent foramen (all of which are lo- Schaeffer (1981). cated below the ridge forming the optic ped- In fact, Orthacanthus is the only extinct icel; ®g. 17B). A deep, narrow, matrix-®lled shark investigated by Schaeffer (1981) which pit is present in the back of each orbit. Al- may have a neoselachian-like trigemino-pi- though these pits are not readily observed on tuitary fossa (®g. 20B). It de®nitely pos- the specimen, they can be seen in CT scan sessed a deep recess within the orbital end of sections and oblique surface renderings of the jugular canal, just below a large opening the orbit (re, ®gs. 17, 18). They probably for the combined trigeminal and facial nerves 36 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

even if it was continuous with the presumed attachment of the optic pedicel. While a jugular canal is almost certainly present in Ak- monistion, from the above remarks it cannot be assumed that a neoselachian-like trigemino-facial recess was also present, and it is possible that the main facial foramen lay behind the postorbital process as in Cladodoi- des and ``Tamiobatis sp.''. POSTORBITAL PROCESS: There is a massive postorbital process in Cladodoides, with an articular surface for the palatoquadrate posteriorly. The general morphology of the process as described by Gross (1937) and Schaeffer (1981) is well known, but many new features have been revealed by the CT scan. Its main features can be seen in the earlier views (®gs. 1±6, 8). In addition, oblique views of the process are shown in ®gure 17 and a series of clipping planes through the process is shown in ®gure 19. Fig. 20. Diagrammatic views of the postor- A postorbital process similar to that in bital process looking posteriorly, in Cladodoides Cladodoides is present in many Paleozoic (A) and Orthacanthus (B). Both panels are from sharks (e.g., Orthacanthus, Tamiobatis vetus- Schaeffer (1981), whose original abbreviations tus,``Tamiobatis sp.'', Akmonistion). In one are as follows; etha, ethmoid articulation (ϭ or- of the fragments referred to Cladodus ele- bital articulation in this work); fepsa, foramen for gans there is a small area of cartilage appar- efferent pseudobranchial artery; foph, foramen for ophthalmic artery; fpal, foramen for branch of ently forming the ¯oor of the jugular canal palatine nerve (ϭ trigeminal mandibular ramus in (presumably representing cartilage within the this work); fso, foramen for super®cial ophthal- embryonic lateral commissure), suggesting mic nerve; jc, jugular canal; tfr, trigemino-facialis that the postorbital process was as well de- recess (probably a recess in Orthacanthus, but not veloped here as in Cladodoides and Tamiob- in Cladodoides; see text for details); II, optic fo- atis vetustus. Very little is known about the ramen. postorbital process in ``Cladodus'' hiassicus, although StensioÈ (1937) depicted a curious forked distal extremity which suggests the (Maisey, 1983: ®g. 14B). Although this re- presence of a canal passing through the process is positioned farther laterally than the cess. A postorbital process is present in early trigemino-pituitary fossa in neoselachians, its chondrichthyans such as Pucapampella and proximity to the oculomotor and abducent Doliodius, but its morphology is quite vari- foramina suggests that it could have housed able. In Pucapampella, the postorbital arcade the external rectus musculature. is thin and delicate (Maisey, 2001b), but in The ventral part of the orbit is not pre- Doliodus it is more robust and much deeper served in the material referred to Cladodus anteroposteriorly (Miller et al., 2003). Coates elegans and nothing is known of its facial and Sequeira (1998) drew attention to the ap- and trigeminal arrangement or whether a fos- parently greater length of the postorbital pro- sa was present. Coates and Sequeira (1998) cess in Akmonistion and Cladoselache than identi®ed an opening in the anterior face of in many other fossil sharks. They suggested the postorbital process in Akmonistion as the that this may be correlated with presence of trigemino-facial recess and jugular canal. a wide orbital roof. However, if the supra- However, they were unable to ascertain the orbital cartilage is omitted from consider- positions of the trigeminal and facial nerves ation, the relative width of the otic region is or the pituitary vein and they were uncertain also a potentially important factor since it is 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 37 approximately twice the width of the inter- Chlamydoselachus and Heptranchias, but orbital part of the braincase in Cladodoides, that view is not widely accepted. Holmgren Orthacanthus, Tamiobatis vetustus, Hybod- (1940, 1941) nevertheless pointed out that us, and many neoselachians, but it is only the neoselachian jugular canal only contains slightly wider than the interorbital region in the lateral head vein and the main (hyoman- Akmonistion. A comparatively narrow otic dibular) trunk of the facial nerve (although region can emphasize the lateral extent of the the extent to which the latter is ``contained'' postorbital process, while a wider otic region by the canal depends on the extent of the behind the process would have the opposite lateral and prefacial commissures as dis- effect. Nevertheless, postorbital process cussed below). Thus, any interpretation in- length is not consistently related to the width volving additional nerves and/or vessels of the otic region, and considerable varia- within this canal in extinct elasmobranchs tions in its length and proportions are found would represent a difference from modern in living and fossil elasmobranchs. forms. Schaeffer (1981: 48) characterized the In Cladodoides, the jugular canal does not postorbital process in Paleozoic sharks as pass straight through the postorbital process. ``bounded anteriorly by the trigeminal fora- Instead, its orbital opening is situated in the men and posteriorly by the foramen for the ventral part of the orbit, behind the pituitary hyomandibular nerve''. However, that view vein foramen (®g. 19F). The canal then turns is no longer tenable since the trigeminal fo- posterodorsally and opens posteriorly adja- ramen in Cladodoides as well as in Tamiob- cent to the trigeminal foramen. The posterior atis vetustus and ``Tamiobatis sp.'' is clearly end of the jugular canal is therefore posi- located behind the process and does not lie tioned farther dorsally than the anterior end. in the orbit (see above). StensioÈ (1937: ®g. 7) showed that the medial The postorbital articulation for the pala- wall of the jugular canal follows a similar toquadrate in Cladodoides is directed poste- course in the Humboldt Museum braincase, riorly, with a distinct groove ¯anked laterally although its dorsal part is not preserved. In by a ¯ange projecting from the posterior rim Tamiobatis vetustus,``Tamiobatis sp.'', and of the postorbital process as in Orthacanthus, Cladodus elegans, the jugular canal is also Tamiobatis vetustus,``Tamiobatis sp.'', and curved upward posteriorly. A jugular canal Akmonistion (Hotton, 1952; Schaeffer, 1981; is also present in Pucapampella (Maisey, Coates and Sequeira, 1998). No articular sur- 2001b), but is unknown in Gladbachus and face is preserved in any of the braincase frag- Doliodus. ments referred to Cladodus elegans, but there In addition to the jugular canal, the post- is a postorbital articulation on the palato- orbital process in Cladodoides also contains quadrate. A postorbital articulation is present a transversely oriented canal. This opens in Pucapampella, oriented almost vertically proximally into the jugular canal and extends on the thin lateral wall of the postorbital ar- laterally through cartilage derived from the cade (Maisey, 2001b). The palatoquadrate is embryonic lateral commissure. The canal still articulated to the postorbital process in opens within the orbital surface of the post- one of the Doliodus specimens described by orbital process farther distally. Gross (1937) Miller et al. (2003). Although no details of identi®ed the proximal opening of this canal the postorbital articulation are known, the in the wall of the jugular canal (``R.b.'') and overall arrangement seems similar to that in also identi®ed smaller foramina in the pos- Cladodoides and many other Paleozoic terior wall of the postorbital process (``r.b.'') sharks. as well as a large opening in the tip of the It is now widely accepted that the main process (``OÈ ffnung am distalende des Pro- anteroposterior passage through the postor- cessus postorbitalis''). He did not recognize bital process in Paleozoic sharks is the jug- the full extent of the large passage through ular canal (see discussion in Schaeffer, the process, but nevertheless suggested that 1981). Holmgren (1941) considered that the the buccal ramus of the anterodorsal lateral canal in Cladodoides was homologous to a line nerve passed through the postorbital pro- passage for the otic lateral line nerve in cess and gave rise to small ramules, which 38 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 surfaced along a transverse groove on the the orbit rather than medially (although it posterior surface of the postorbital process. would still be more medial than originally The buccal nerve then supposedly exited the interpreted by Gross, 1937). process via the distal opening in the process It is also possible that the postorbital pro- (probably corresponding to the notch in the cess contained the mandibular ramus of the end of the process in the Humboldt Univer- trigeminal nerve (either alone or accompa- sity specimen illustrated by StensioÈ, 1937: nied by the buccal-maxillary complex). In ®g. 1). many neoselachians, the proximal part of the The strongest evidence presented by Gross mandibular ramus is directed laterally just (1937) that the buccal ramus entered the below or in front of the postorbital process postorbital process is the presence of small (e.g., Squalus; Marinelli and Strenger, 1959) openings on its posterior surface, said to in more or less the same orientation as the have contained buccal ramules. Unfortunate- passage in Cladodoides. However, the man- ly, only one of these openings is even faintly dibular ramus in Squatina and squaloids is suggested in the CT scan and it could not be located anterior to the postorbital process clearly traced to the main canal within the (which includes the lateral commissure) and process. Furthermore, the openings identi®ed does not pass through it. In ``long-jawed'' by Gross (1937) are located within the main neoselachians such as Chlamydoselachus, the articular surface for the palatoquadrate (as mandibular ramus passes close beneath the described by Gross, 1938), which is an im- process posterolaterally before continuing plausible location for buccal ramules. It posteriorly toward the mandibular joint and therefore seems more likely that any canals across the lateral surface of the palatoquad- emerging here were for small blood vessels rate adductor muscle (Luther, 1909; Allis, associated with the postorbital articulation. 1923a). In hexanchiforms (where the pala- This does not preclude the buccal ramus toquadrate articulates with the postorbital from passing through the postorbital process, process posteriorly), the mandibular ramus is but the evidence for buccal ramules is not directed laterally as it passes below and in compelling. front of the postorbital process. The ramus If the buccal ramus of the anterodorsal lat- then turns posterolaterally below the postor- eral line nerve entered the postorbital process bital articulation (Luther, 1909). Such a in Cladodoides as Gross (1937) proposed, it course would also be feasible in Cladodoides may have been accompanied by the trigem- and would place the mandibular ramus below inal maxillary ramus (forming a buccal-max- and in front of the postorbital articulation. illary complex, as in many neoselachians). Since the trigeminal foramen lies behind the This is suggested by the position of the inner postorbital process, the only alternative opening of the canal, which is aligned with would be for the mandibular ramus to pass the exit of the trigeminal/anterodorsal lateral above and behind the postorbital articulation, line nerve (®g. 19D±F). The buccal-maxil- which is implausible anatomically even lary complex does not pass through the post- though the postorbital articulation in hexan- orbital process in any modern elasmobranch chiforms is positioned farther dorsally and although some ramules from the buccal ra- medially on the process than in Cladodoides. mus may enter the cartilage to supply part of Alternatively, the buccal ramus (or buccal- the infraorbital sensory canal above the pro- maxillary complex) may have entered the or- cess (e.g., Chlamydoselachus). Instead, the bit directly via the jugular canal. In neose- neoselachian buccal-maxillary complex lachians, this canal contains only the lateral mostly passes anteriorly (and to some extent head vein and the hyomandibular trunk, but ventrolaterally) below the rectus musculature in ``Tamiobatis sp.'' (AMNH 2140) it also rather than laterally. Therefore, even if the contained the dorsal (super®cial ophthalmic) buccal-maxillary complex in Cladodoides component of the anterodorsal lateral line did pass through the postorbital process, its nerve. Conceivably, the buccal ramus may subsequent course into the ethmoid region also have passed through the canal farther would have been extremely circuitous be- ventrally. cause the canal emerges in the lateral part of To summarize, the lateral part of the post- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 39 orbital process in Cladodoides, Tamiobatis vetustus, and many other Paleozoic sharks probably contained the trigeminal mandibular ramus, perhaps accompanied by the buccal ramus or buccal-maxillary complex, although the latter may have entered the orbit via the jugular canal. The palatine nerve pierced the ¯oor of the portorbital process medial and ventral to the jugular canal. Other interpretations of Paleozoic sharks, showing the facial mandibular ramus or palatine ramus passing through the lateral part of the postorbital process (e.g., Romer, 1964; Schaeffer, 1981; Coates and Sequeira, 1998), are rejected on morphological grounds (see discussion). In Cladodoides, there is a small foramen in the ¯oor of the postorbital process near its orbital margin ®g. 17B). This may have contained a mandibular vein connected to the lateral head vein as Dick (1978) suggested in Tristychius. Although such a vessel is absent in modern sharks, one is present in rays such as Urolophus. Even if it were present in early Fig. 21. Ventral view of the braincase in Cla- dodoides (surface rendering, mirror-imaged), with sharks such as Cladodoides, the lateral extent a reconstruction of the basicranial arterial circuit. of the suborbital shelf would have prevented No scale. it from passing beneath the postorbital process. DORSAL AORTAE: The basic arrangement the cotylus and the bucco-hypophyseal fe- of major cranial arteries in Cladodoides out- nestra. There is no evidence of efferent bran- lined by Gross (1937) and Schaeffer (1981) chial or hyoidean arteries branching from the is con®rmed from CT scanning, since those aortae within the parachordal cartilage. The parts of the basicranial circuit which were efferent hyoidean artery probably branched enclosed by cartilage can be traced through farther anteriorly (see below), while the ef- successive sections and regions that lay in ferent branchial arteries probably arose far- well-de®ned grooves can also be reconstruct- ther posteriorly behind the occipital region as ed with a high degree of con®dence. Inter- in Chlamydoselachus (Allis, 1923a). vening regions are more problematic, al- The internal carotid and orbital arteries though some can be inferred from the ori- separated only after the aortae left the brain- entation and direction of adjacent canals and case anteriorly, and their positions are grooves. A new reconstruction of the main marked by shallow grooves on the basicra- basicranial circuit in Cladodoides is present- nium. The internal carotid artery passed into ed here, mapped onto a ventral view of the the braincase via the bucco-hypophyseal fe- braincase (®g. 21). nestra, and the orbital artery entered the orbit Below the occipital cotylus, the paired lat- via a notch or short canal in the suborbital eral dorsal aortae lay in grooves and were shelf, although the artery was evidently ex- presumably united behind the occiput. The posed beneath the basicranium and did not aortae became completely enclosed by car- pass through cartilage (cf. Schaeffer, 1981: tilage only anterior to the cotylus (between ®g. 12C). the level of the third and fourth spino-occip- Small canals arise from each main aortic ital foramina). The aortic canals continue an- canal dorsolaterally and open just below the teriorly within the parachordal cartilage, re- spino-occipital foramina in the lateral wall of surfacing approximately midway between the occipital cotylus (®g. 6). These canals are 40 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 provisionally identi®ed as passages for inter- aorta is primitive for gnathostomes, paired vertebral arteries, although it is possible that aortic canals are present in Pucapampella the anteriormost one contained a vertebral ar- and many other Paleozoic chondrichthyans tery. According to Allis (1923a), in Chla- (Maisey, 2001c). In fact, the only chondri- mydoselachus the dorsal aorta receives three chthyans in which a median aortic canal is intervertebral arteries on each side before di- known are symmoriids and stethacanthids viding into paired lateral dorsal aortae; each (although the paired lateral aortae are com- paired aorta then receives a single interver- bined within a common sheath in some mod- tebral artery before receiving the anterior- ern hexanchiforms; Jarvik, 1980). While the most vertebral artery. Although the aortae character is homoplaseous, within chondri- and most of the intervertebral arteries in chthyans a single aortic canal may represent Chlamydoselachus are not contained by car- a synapomorphy of symmoriid and stetha- tilage, Allis (1923a: ®gs. 11, 52) described canthid sharks. its anteriormost vertebral artery as traversing ORBITAL ARTERY AND PALATINE NERVE: the ventrolateral margin of the braincase and Scanning reveals that what Schaeffer (1981) he also showed canals for both the ®rst ver- took to be the posterior opening for the or- tebral and intervertebral arteries passing bital artery in Cladodoides is actually con- through cartilage on each side of the occiput. nected internally to the facial (hyomandibu- StensioÈ's (1937: ®g. 6) reconstruction of lar) foramen by a slotlike space and does not the basicranial circuit in ``Cladodus'' hassi- reach the expected position of the artery (see acus differs only slightly from the scheme above). Gross (1937) suggested that this fo- presented here. Its orbital arteries are more ramen contained the posterior palatine ra- completely enclosed by canals than in Cla- mus, which is supported by present obser- dodoides, and its internal carotids are also vations from the CT scan. It is very unlikely enclosed anteriorly. These differences are not that the orbital artery occupied any part of necessarily of taxonomic importance, and the this canal. Instead, the artery probably ran Humboldt University specimen may simply entirely beneath the braincase before it en- be a more mature individual with more ex- tered a partially roofed groove (which meets tensive calci®cation. StensioÈ (1937) depicted the passage for the anterior ramule of the pal- the internal carotids as joining at the midline atine ramus, immediately below the postor- in his specimen, but this cannot be observed bital process). Furthermore, the CT scan also (although scanning might reveal whether con®rms that the foramen for the orbital ar- these vessels remained separate as in the tery lies between the two palatine foramina Senckenberg braincase). The two braincases and refutes Schaeffer's (1981) interpretation agree in having: (1) enclosed canals for that it contained the palatine nerve; this pas- paired lateral dorsal aortae; (2) the regions sage simply does not connect to other parts where the internal carotid, orbital, and effer- of the facial trunk. The orbital and internal ent hyoidean arteries divide are exposed ven- carotid arteries in Cladodoides probably did trally; and (3) the efferent pseudobranchial not diverge immediately in front of the an- and ophthalmic arteries join the internal ca- terior opening of the lateral dorsal aorta as rotids dorsal to the trabeculae. suggested by Schaeffer (1981: ®g. 12). In- A single dorsal aorta enters the basicrani- stead, the orbital artery arose farther anteri- um in both Falcatus and Akmonistion (Lund, orly and passed anterolaterally across the ba- 1985; Coates and Sequeira, 1998), and in a sicranium to enter the orbit via a short three-dimensional Pennsylvanian braincase groove beneath the postorbital process (®g. referred to ``Cobelodus'' (Maisey, 2004a and 21). In Cladodoides, the efferent hyoidean in prep.). Interpretathions of the basicranial artery probably branched from the lateral circuit in Cobelodus aculeatus by Zangerl dorsal aorta somewhere between its emer- and Case (1976) and Schaeffer (1981) are gence from the basicranium and the diver- probably erroneous in depicting the aorta as gence of the internal carotid and orbital ar- paired and external to the basicranium. While teries. Coates and Sequeira (1998) may be correct As mentioned above, the orbital arteries in that a median groove or canal for the dorsal the Humboldt University braincase are more 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 41 completely enclosed in canals than in the case after leaving the orbit, but this region is Senckenberg specimen. The anterior palatine not preserved in Cladodoides. nerve ramule therefore probably entered the POSTERIOR TECTUM AND DORSAL FONTA- orbit within this canal, accompanied by the NELLE: As in many other fossil sharks, Cla- orbital artery as suggested by StensioÈ (1937). dodoides has a large, anteroposteriorly elon- ROOF OF ORBITOTEMPORAL REGION: Gross gated fontanelle in the roof of the braincase (1937) noted several small areas where the (®g. 4). It lies between the otic capsules and roof of the braincase is supposedly damaged is bordered posterolaterally by short, paired in the Senckenberg braincase (®g. 2: ``Ver''). dorsal otic ridges (see below). The fontanelle Scanning reveals that cartilage at the dorsal almost reaches the otico-occipital ®ssure but midline is considerably thinner than else- is separated from it by a weak posterior tec- where and is even open locally (®g. 4), leav- tum located below the level of the dorsal otic ing a median suture that is especially notice- ridge. The dorsal opening in Cladodoides able in scan sections 85±107 (®g. 12). In and primitive chondrichthyans such as Pu- neoselachians, the midregion of the braincase capampella is perhaps best envisaged as a roof is one of the last to chondrify (De Beer, persistent posterior dorsal fontanelle, homol- 1931), and it may even remain partly open ogous to that of osteichthyans (see discus- (as an anterior fontanelle) in batoids (Holm- sion). gren, 1940). The Cladodoides braincase A posterior tectum borders the fontanelle could represent a juvenile individual in posteriorly in Tamiobatis vetustus (NMNH which the roof was only just approaching 1717), ``Tamiobatis sp.'' (AMNH 2140), Ak- completion. There is also a distinct opening monistion (Coates and Sequeira, 1998), and approximately midway between the parietal ``Cobelodus'' (in prep.). However, in Ortha- and precerebral fossae, resembling a pineal canthus and Pucapampella, a tectum is ab- foramen, although it overlies the mesence- sent and the fontanelle is con¯uent with the otico-occipital ®ssure posteriorly (Schaeffer, phalic region of the endocast and therefore 1981; Maisey, 2001c; Maisey and Anderson, seems too far posterior (in neoselachians, the 2001). In adult neoselachians and hybodonts, epiphysis usually arises above the optic the otico-occipital ®ssure is closed, but in nerve and reaches the braincase roof above many neoselachian embryos a posterior tec- the telencephalon). A similar opening has tum separates the parietal fossa from the em- been described in the roof of the braincase bryonic otico-occipital ®ssure. in Acanthodes (Watson, 1937; Miles, 1973; Perilymphatic fenestrae have not been Jarvik, 1977, 1980). This may be related to identi®ed in Cladodoides. In neoselachians, the pineal organ and is also one of the last these fenestrae are located at the upper apex regions to ossify (Miles, 1973: 77). of the posterior semicircular canal. In Cla- In Cladodoides, the supraorbital shelf is dodoides, this point is located just behind the well developed. Paired rows of foramina for crus commune, but there is no evidence of ramules of the super®cial ophthalmic ramus an opening here. In Orthacanthus, the chon- (supplying the supraorbital sensory canal dri®ed lateral walls of the posterior dorsal above the braincase) pierce the cartilage fontanelle extend a considerable distance be- above the orbit. Anteriorly, a notch (still low the surface. Schaeffer (1981: 22) as- largely obscured by matrix) marks the posi- sumed that its perilymphatic openings ``must tion where this nerve passed through the su- have opened into the bottom of the endolym- praorbital cartilage and emerged onto the phatic fossa in close proximity to the poste- dorsal surface above the olfactory region as rior semicircular canal.'' However, as in Cla- in neoselachians (®gs. 3, 4). In Tamiobatis dodoides, there is no evidence that Ortha- vetustus (NMNH 1717), ``Tamiobatis sp.'' canthus possessed perilymphatic fenestrae. (AMNH 2140), and Cladodus elegans, the Moreover, fenestrae in such a deep position roof of the orbit is also pierced by numerous could not have had a tympanic function as in foramina for super®cial ophthalmic ramules. neoselachians. If perilymphatic fenestrae In neoselachians, the profundal nerve were present in Tristychius, as Dick (1978) emerges onto the dorsal surface of the brain- has suggested, that taxon would be unique 42 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 since its posterior semicircular canal is not otic capsule and involves only the lateral otic separated from the anterior one and a sinus process and lateral otic ridge. This is con- superior is still present. ®rmed in articulated Orthacanthus speci- Endolymphatic foramina have not been mens from Germany (Heidtke, 1998, 1999). identi®ed in Cladodoides although endolym- The hyomandibular fossa in the Senckenberg phatic ducts were probably present and pre- braincase was probably restricted to the same sumably passed from the posterior dorsal region, but the specimen is damaged on both fontanelle into the unchondri®ed medial sur- sides in the expected position of the fossa, face of the otic capsule, as hypothesized in exposing the posterior semicircular canals. some fossil osteichthyans (e.g., Moythoma- GENERAL FEATURES OF SKELETAL LABY- sia; Gardiner, 1984a). RINTH: The skeletal labyrinth in Cladodoides DORSAL OTIC RIDGES: Paired dorsal otic is seen in views of the endocast (®gs. 23± ridges are developed on each side of the pos- 26). The medial capsular wall is unchondri- terior dorsal fontanelle in several extinct ®ed and the vestibular chamber is therefore sharks (Schaeffer, 1981). Those of Clado- con¯uent with the cranial cavity. The hori- doides are small and weak, present only ad- zontal semicircular canal lies just below the jacent to the fontanelle, and are not directed lateral otic ridge, and the anterior and pos- laterally (®gs. 4, 6, 8, 22). The dorsal otic terior semicircular canals meet dorsally at a ridges in Akmonistion are also weakly de- crus commune, below which is a short sinus veloped and are restricted in their anteropos- superior. The sinus merges proximally with terior extent (Coates and Sequeira, 1998). the cranial cavity and vestibular chamber, They are far more extensive in Orthacanthus, and features deeper within the vestibular re- Tamiobatis vetustus, and ``Tamiobatis sp.'', gion cannot be discerned. There is a promi- where they extend from the postorbital pro- nent utricular recess (con¯uent with the sac- cess to the posterior end of the fontanelle and cular chamber medially) to which the ante- their dorsal surface is directed laterally. Dor- rior and external ampullae are connected. sal otic ridges are absent in Hybodus and The posterior ampulla is con¯uent with the neoselachians. In ``Cobelodus'', the cartilage saccular chamber and there is no evidence of surrounding the posterior dorsal fontanelle is a preampullary canal (sensu Maisey, 2001b). raised above the general surface of the brain- The main vestibular chamber is weakly di- case, but paired ridges are absent (in prep.). vided into dorsal (utricular) and ventral (sac- LATERAL OTIC PROCESS AND HYOMANDIB- cular) regions and there is little evidence of ULAR FOSSA: Most of the cartilage surround- a lagenar chamber. It is unknown whether the ing the posterior semicircular canals in the crista neglecta was enlarged as in neosela- Cladodoides braincase is missing, apart from chians, although this seems unlikely. The a small fragment illustrated by Gross (1937: medial (nonampullary) part of the external ®g. 3, ``Rest der fehlenden hinteren otical- semicircular canal turns inside the curve of wand''). A lateral otic process (sensu Schaef- the posterior canal before passing deeper into fer, 1981) was probably present in Clado- the labyrinth cavity. Its subsequent course doides, but its original extent is unknown and cannot be determined but presumably it en- the large process reconstructed by Schaeffer tered the sacculus just below the crus com- (1981: ®g. 25) is entirely speculative. Note mune, as in chimaeroids and osteichthyans. that the lateral otic process is only one of The anterior and external ampullae in Cla- several different processes that may be de- dodoides are clearly separated and probably veloped on the lateral wall of the otic capsule had separate openings into the utricular re- (see discussion). cess. A short passage for the anterior ramus According to Gross (1937), the lateral wall of the octaval nerve passes through cartilage of the otic capsule in Cladodoides includes forming the anteroventral part of the laby- an articular fossa for the hyomandibula al- rinth wall before entering the proximal end though its extent was not clearly established. of the utricular recess. Presumably, the an- Schaeffer (1981) showed that the hyoman- terior and posterior rami were already sepa- dibular fossa in Orthacanthus is con®ned to rated and the posterior ramus passed through a small area on the posterolateral wall of the an unchondri®ed part of the capsular wall. 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 43

The only other fossil shark in which the po- (Schaeffer, 1981; Coates and Sequeira, sition of the octaval nerve has been described 1998). The occipital arch contains the fora- is Cobelodus aculeatus (Zangerl and Case, men magnum and also forms a continuous 1976: ®gs. 8, 9), where it was apparently ac- roof to the dorsal nerve cord. By contrast, in companied by the facial and trigeminal Tamiobatis vetustus and ``Tamiobatis sp.'', nerves within a recess that may be homolo- the dorsal part of the occipital arch anterior gous to the acustico-trigemino-facial recess to the foramen magnum is uncalci®ed in neoselachians (see discussion). (Schaeffer, 1981). However, the margins of The semicircular canals of the Sencken- the calci®ed region are un®nished, suggest- berg braincase are arranged in a profoundly ing that a complete membranous occipital different way from the neoselachian-like pat- arch was present even though its dorsal part tern described by StensioÈ (1937) in the Hum- was unmineralized. Schaeffer (1981) argued boldt University specimen. Crucially, StensioÈ that the glossopharyngeal nerve of Ortha- (1937) argued for an almost completely sep- canthus and Tamiobatis vetustus passed arate posterior semicircular canal in the latter, through a slitlike opening in the ventrolateral forming a ring that communicates only with surface of the otic region (the ventral otic the vestibular region. If that is true, the Hum- notch, corresponding to the metotic ®ssure in boldt University braincase represents the old- embryonic neoselachians). In Orthacanthus, est known shark with such an arrangement the ventral otic notch extends anteriorly me- (a neoselachian-like labyrinth morphology is dial to the posterior semicircular canal. It also present in Mesozoic hybodonts; Maisey, then passes completely beneath the hyoman- 2004a and in prep.). dibular fossa and terminates more or less lev- The semicircular canals in Cladodoides el with the middle of the posterior dorsal fon- are comparatively narrow, as in Squalus.By tanelle. In the original Tamiobatis vetustus contrast, the canals in Notorynchus are much braincase (NMNH 1717), the ventral otic wider (®g. 27B, C), while those of Ortha- notch also extends below the hyomandibular canthus seem to fall somewhere in between fossa, but it terminates somewhat farther pos- (Schaeffer, 1981: ®g. 14). Canal diameter teriorly than in Orthacanthus (Schaeffer, may be correlated generally to behavior, 1981: ®g. 19). The occipital region is not since ®shes with low turning speeds tend to preserved in the Humboldt University brain- have larger semicircular canals than highly case. Among the material referred to Cla- maneuverable ®shes (Young, 1981). This dodus elegans is a complete occipital seg- suggests that Cladodoides was perhaps as ag- ment with a persistent otico-occipital ®ssure ile as Squalus in swimming maneuvers. and no signs of an uncalci®ed dorsal area. The anterior and posterior ampullae in Although the ®ssure is persistent in Cla- Cladodoides lie at approximately the same dodoides, its metotic part is distinctly wid- distance from the midline. However, in ened to form a glossopharyngeal canal, Squalus and Notorynchus, the posterior am- ¯oored by a continuous shelf (presumably pulla is closest to the midline (Maisey, derived from the embryonic hypotic lamina) 2004b: ®g. 11E). The functional signi®cance below the otic capsule. The arrangement re- of this difference is unknown but may be resembles that of neoselachians, except that in lated to decoupling of the posterior canal Cladodoides the canal is still connected to from the rest of the labyrinth. the otico-occipital ®ssure dorsally, instead of OCCIPITAL ARCH: Scanning reveals that the being completely isolated. Laterally, the oc- occipital region in the Cladodoides braincase cipital arch is connected to the parachordal is far more complete than was previously region by the posterior basicapsular commis- supposed (Maisey, 2001c). In dorsal view, sure and is not fused to the posterior wall of the occipital arch intrudes only slightly be- the otic capsule. The dorsal part of the otico- tween the otic capsules (although it also ex- occipital ®ssure is narrow, but it widens just tends behind them) and there is a persistent medial to the posterior ampulla in the ex- otico-occipital ®ssure similar to that de- pected position of the vagal nerve (®g. 28A, scribed in Orthacanthus, Tamiobatis vetus- C). Thus, in Cladodoides the positions of the tus,``Tamiobatis sp.'', and Akmonistion glossopharyngeal and vagal nerves are both 44 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 22. Otic region of the Cladodoides braincase (surface rendering), oblique perspective detail views. (A) Anteromedial view; (B) anterolateral view; (C) posterior view; (D) posteromedial view. This block overlaps slightly with that depicted in ®gure 17, repeating the exit of the hyomandibular trunk. Note the absence of a medial wall to the saccular chamber, utricular recess, and crus commune, although the semicircular canals are almost completely enclosed by cartilage. The position of the octaval (acus- ticovestibular) nerve is indicated, but is more clearly seen in sections (see ®g. 18E). The hypotic lamina extends beneath the otic capsule, forming a wide glossopharyngeal canal and a short metotic ®ssure. The lateral dorsal aorta is enclosed by cartilage posteriorly, then emerges ventrally. The uneven internal ¯oor and thickened cranial roof in front of the posterior dorsal fontanelle are artifacts of digital interpretation, as the braincase wall is poorly calci®ed in these areas; it is unlikely that the synotic tectum was originally as thick as shown here. 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 45

Fig. 22. Continued. much better de®ned within the otico-occipital 1931). In Orthacanthus, most of the hypotic ®ssure than in Orthacanthus.InCladodoi- lamina is separated from the capsule by a des, the ventral otic notch is either absent or wide ventral otic notch. However, in ``Tam- extremely short, unlike in Orthacanthus iobatis sp.'' (AMNH 2140), there is only a where there is an extensive notch (Schaeffer, short ventral otic notch which does not ex- 1981: ®g. 6). The lateral margin of the hy- tend anteriorly beyond the level of the hyo- potic lamina in Cladodoides therefore meets mandibular fossa. The arrangement in ``Tam- the otic capsule posteriorly as in neoselachi- iobatis sp.'' is therefore more like Cladodoi- ans and hybodonts. des than Orthacanthus. It is unknown wheth- In Cladodoides, the glossopharyngeal ca- er a metotic ®ssure was present in the nal merges with the lumen of the saccular Humboldt University braincase. chamber, but there is no evidence that the Cladodoides has a circular occipital coty- nerve actually entered it; instead, they were lus which includes a deep conical central probably separated by a membranous cap- cavity tapering into a short notochordal ca- sular ¯oor as in Scyliorhinus (De Beer, nal. The canal extends only a short distance 46 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 23. Lateral view (left side) of the Cladodoides endocast (surface rendering). into the parachordal region and does not Maisey, 1984b, 1985; Maisey et al., 2004). reach the dorsum sellae (®gs. 7, 28B, C). The In batoids, chordal centra are typically pre- inner surface of the cotylus appears smooth sent but a hemicentrum is absent and the no- in the reconstruction (®g. 8), although this tochord terminates within the synarcual car- may be an artifact generated by the imaging tilage some distance behind the occipital software. In Hybodus, this surface is formed arch. Thus, the presence of an occipital hem- in ®brocartilage, with a rough spongy ap- icentrum in neoselachians is probably corre- pearance (Maisey, 1987). The cotylar surface lated with their tendency to develop chordal also has a spongy appearance in other Paleo- centra. Such centra are absent in hybodonts zoic sharks. In modern sharks, the cotylus and many Paleozoic sharks, where the hem- includes an occipital ``hemicentrum'' or icentrum is also absent. The occipital region ``half centrum'' (Maisey, 1984a, 1984b), in gnathostomes is always perichordal, form- with a smooth inner surface like those of the ing outside the elastica externa (like the amphicoelous chordal centra forming the paired arcualia of the vertebral column; vertebral column. This smooth surface char- Goodrich, 1930: 10). The cotylus in extinct acterizes those neoselachians in which (1) the sharks such as Cladodoides is therefore al- ®brous notochordal sheath is invaded by most certainly of perichordal origin and is mesenchymatous cells to form chordal centra probably homologous to the perichordal car- and (2) the notochord extends into the basi- tilage surrounding the chordal occipital hem- cranium. Both these conditions are met in icentrum in modern sharks. modern sharks and presumably were also re- In Cladodoides, at least two foramina for alized in extinct forms where a hemicentrum intervertebral arteries pass through the lateral is present (e.g., Palaeospinax, Synechodus; wall of the cotylus, suggesting that the co- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 47

Fig. 24. Dorsal view of the Cladodoides endocast. Anterior to top. 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 25. Ventral view of the Cladodoides endocast. Anterior to top. 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 49

Fig. 26. Anterior (A) and posterior (B) views of the Cladodoides endocast. tylar region spans more than one vertebral line of fusion between the left and right para- segment. The presence of four or ®ve spino- chordals (®g. 7). The parachordal cartilage occipital foramina farther dorsally suggests therefore completely enclosed the notochord, that the occipital arch included several ver- as in Scyliorhinus and Squalus (De Beer, tebral segments. 1931; El-Toubi, 1949; see ®g. 29 here). Car- The occipital cotylus and notochordal ca- tilage surrounding the notochord in Clado- nal in Cladodoides lie within the thickness doides is calci®ed (presumably perichordal- of the basicranial cartilage at the presumed ly), although surrounding areas of the basal 50 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 27. Comparison of the cranial endocast in Cladodoides (A, D) and Notorynchus (B, C). The endocasts are positioned with their anterior ampullae aligned. Note the greater length of the mesencephalic chamber in Cladodoides, resulting in a more anterior location of the hypophyseal chamber and foramina for the oculomotor and optic nerves and efferent pseudobranchial artery. Despite these differences, the position of the trochlear foramen is similar, suggesting that the proportions of the mesencephalic chamber primarily involve its ventral part and may be related to differential development of the polar cartilage. Not to scale. plate are unmineralized. Posteriorly, the oc- A similar arrangement is seen in a frag- cipital cotylus completely separates the para- mentary occipital region of a large Tamiob- chordals, and the notochordal canal at its atis-orCtenacanthus-like braincase (USNM center is located a short distance above the 299647; ®g. 31), but here the notochordal ca- ¯oor of the braincase. As the cotylus tapers nal is considerably elevated above the ¯oor anteriorly, the calci®ed sheath of the noto- of the parachordal plate by a median septum chordal canal remains elevated but is still and there are massive paired subnotochordal connected to the ¯oor of the basal plate by a chambers. In the holotype of Tamiobatis ve- vertical septum. This represents a region of tustus (NMNH 1717), a subnotochordal sep- deep but localized endochondral calci®cation tum and paired chambers are revealed where beneath the notochord and presumably lies the basicranial cartilage has been stripped along the original line of fusion between the away by erosion (see Romer, 1964: pl. 1). paired parachordals (®g. 30A, B). On each The posterior part of a short median noto- side of the septum is a small uncalci®ed chordal septum is also visible in the slabbed ``chamber'' (probably cartilage-®lled in life). ``Tamiobatis sp.'' braincase, immediately an- Farther anteriorly, the median septum breaks terior to the occipital cotylus (AMNH 2140; down and these chambers are interconnected, Schaeffer, 1981: ®g. 24H). Externally, the separating the calci®ed notochordal canal position of the subnotochordal chambers is from the ¯oor of the basal plate (although the marked by paired convex areas lying be- canal probably remained within the thickness tween the median groove and the lateral aor- of the parachordal cartilage; ®g. 30C). tic grooves. These features conceivably be- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 51

Fig. 27. Continued. came enhanced during growth and may be anteriorly, passing through the parachordal better developed in larger individuals. Ac- cartilage between the ¯oor of the cranial cav- cording to Schaeffer (1981: ®g. 9), in Ortha- ity and the metotic ®ssure. Three or four spi- canthus the notochord is contained entirely no-occipital foramina are present along each within the thickness of the mineralized basal lateral surface in ``Cobelodus'' (in prep.) and plate and no subnotochordal septum or in an undescribed isolated occipital region chambers are present (as in neoselachians). referred tentatively to Cladodus elegans. Thus, a subnotochordal sheath and chambers Paired foramina are also present in the para- may not characterize all early sharks with an chordal region of another fragment referred elongate otico-occipital region. to C. elegans, corresponding to the more an- In Cladodoides, at least four and perhaps terior ones in ``Tamiobatis sp.'' In these ®ve spino-occipital foramina are present specimens, the cartilage containing the an- along each side of the occipital arch (®gs. 27, teriormost foramina is probably homologous 28). At least four spino-occipital nerves are to the dorsal process of the parachordal plate also present in the occipital block of ``Tam- in neoselachians, which forms the basicap- iobatis sp.'' (Schaeffer, 1981: ®g. 24F) and sular commissure connecting the hypotic there may even be an additional pair farther lamina to the ¯oor of the otic capsule (e.g., 52 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 28. Occipital region of the Cladodoides braincase (surface rendering), oblique perspective detail views: (A) posterolateral view; (B) posteromedial view; (C) anteromedial view; (D) anterolateral view. This block overlaps slightly with that shown in ®gure 22, repeating the tip of the dorsal otic ridge and posterior semicircular canal. The hypotic lamina is fused to the capsular wall laterally, forming a glossopharyngeal canal. The position of the vagal nerve is indicated by an expansion in the otico-occipital ®ssure above the glossopharyngeal canal.

Scyliorhinus; De Beer, 1931). The glosso- leozoic sharks, and this may represent the pharyngeal trunk passes through the dorsal primitive number for elasmobranchs. process in neoselachians, but in ``Tamiobatis CRANIAL ENDOCAST: Gross (1937) was sp.'' and Cladodus elegans it presumably en- able to observe very little of the cranial cav- tered the metotic ®ssure farther anteriorly. ity or skeletal labyrinth in Cladodoides, al- It is concluded that three to ®ve spino-oc- though he was able to determine the posi- cipital nerves were commonly present in Pa- tions of the posterior semicircular canals 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 53

Fig. 28. Continued.

(which are partially exposed) and the hori- imperceptibly with the diencephalic and mes- zontal semicircular canals beneath a horizon- encephalic regions, as in neoselachians (®gs. tal crista on the lateral capsular wall. The 23, 24). A small protuberance on the dorsal cranial and labyrinth spaces have been re- surface marks the position of a small opening constructed digitally from surface renderings, in the roof of the braincase. It is tempting to as outlined at the beginning of this paper regard this as a pineal organ, but its posterior (®gs. 23±26). Lateral and dorsal views of the location suggests that it merely represents an endocast in Cladodoides and modern Noto- unchondri®ed area of the cranial roof. The rynchus are shown in ®gure 27. combined exit for the optic nerve and artery The telencephalic region is narrow from is located anterodorsal to the conjoined ef- side to side, but is quite deep and merges ferent pseudobranchial and ophthalmic ca- 54 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Fig. 29. Transverse sections of a Scyliorhinus embryo (after De Beer, 1931; no scale indicated). Note the continuity of the metotic and otico-occipital ®ssures beneath the otic capsule, and the position of the glossopharyngeal nerve between the capsule and hypotic lamina. nals and lies immediately in front of a small transverse constriction marking the position of the presphenoid ridge. The mesencephalic region widens posteriorly and the cerebellar chamber is well de- ®ned in lateral and dorsal views (®gs. 23, Fig. 30. Transverse sections through the oc- 24). In Cladodoides,``Tamiobatis sp.'', and cipital region of the Cladodoides braincase show- Orthacanthus, the mesencephalic chamber is ing the subnotochordal septum and ``chambers'' (presumably cartilage-®lled in life). (A) Slice 281; comparatively longer than in Squalus or No- (B) slice 280; (C) slice 276. All of these features torynchus.InCladodoides, the cerebellar extend over fewer than 10 slices (approx. 3 mm) chamber is broadly expanded, forming paired in this braincase, in contrast with the occipital re- vestibulolateral (auricular) chambers. As in gion shown in ®gure 31. Squalus and Notorynchus, these chambers 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 55

Fig. 31. Fragmentary acid-prepared occipital region of an undetermined (ctenacanth?) shark braincase, USNM 299647, St. Louis Limestone, Alton, Illinois. (A) Dorsal view, anterior to top; (B) anterior view. Note the extensive notochordal sheath supported by a median subnotochordal septum and large paired ``chambers'' forming prominent convexities ventrally between the median groove and paired aortic canals. Scale bar ϭ 5 cm. 56 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 are seen best in dorsal view (®g. 24). By con- postorbital process arising from the supraor- trast, the roof of the cerebellar chamber in bital cartilage. The trunk of the super®cial Chlamydoselachus slopes more abruptly ophthalmic complex then must have con- above the acustico-trigemino-facial recess, nected with that of the mandibular complex and its vestibulolateral chambers are more just above the posterior end of the jugular evident in lateral aspect (Allis, 1923a: ®g. canal. Four or ®ve paired spino-occipital 12). In Hydrolagus, modern osteichthyans nerves are present, although the anterior ones (e.g., Protopterus, Salmo), and lampreys, the are obscured by the vagal nerve except in cerebellum and fourth ventricle are common- ventral view (®g. 25). The spino-occipital ly situated behind the level of the pituitary nerves are paired, as in gnathostomes gen- organ (Jollie, 1962). The adenohypophysis erally and unlike in cyclostomes and many and oculomotor nerve were apparently locat- extinct agnathans where an alternating pat- ed some distance anterior to the otic capsule tern is observed. in early chondrichthyans (e.g., Pucapampel- The hypophyseal (pituitary) chamber can la; Maisey, 2001b) and in some placoderms be seen in lateral and ventral views of the (e.g., Tapinosteus; StensioÈ, 1963a). Collec- endocast (®gs. 23, 25). The chamber is de- tively, these observations suggest that the ®ned posteriorly and dorsally by the very ex- morphology of the cerebellum in Cladodoi- tensive dorsum sellae. Its lateral wall con- des and Orthacanthus was generalized, and tains the pituitary foramen, as in gnathosto- that the comparatively ``short'' mesencephal- mes generally (®g. 7). The chamber is much ic region in hybodonts and neoselachians is larger than in modern sharks, where the dor- an apomorphic feature. sum sellae projects only a short distance In neoselachians, the trochlear canal arises above the ¯oor of the cranial cavity. As in from the optic lobe (which may be well de- modern sharks, however, the roof of the hy- veloped, as in Squalus, or weak, as in No- pophyseal chamber curves above the pitui- torynchus; Maisey, 2004b). The trochlear ca- tary foramen and meets the medial surface nal presumably marks the position of the op- of the sidewall of the cranium just below the tic lobe in Cladodoides although the cranial oculomotor foramen, con®ning the dorsum cavity around it is not expanded and optic sellae entirely behind the efferent pseudob- lobes are not evident. ranchial foramen. It is concluded that the The scan does not show the full extent of dorsum sellae in Cladodoides probably arose the endocast between the otic capsules. The from the embryonic polar cartilage as in medullary region consequently cannot be de- modern sharks, and the considerable size of scribed in detail, although it was evidently the hypophyseal chamber and dorsum sellae constricted between the capsules as in neo- in Cladodoides supports the suggestion that selachians. The posterior dorsal fontanelle the polar cartilage made an extensive contri- becomes indistinct as it merges laterally with bution to the basicranium (see above and the the crus commune below the anterior and following discussion). In modern sharks, the posterior semicircular canals. The vagal hypophyseal chamber is long and is bounded nerve probably occupied the widest part of posteriorly by the dorsum sellae and anteri- the otico-occipital ®ssure (®gs. 25, 26B). orly by the preclinoid wall. In batoids, there Farther ventrally, this ®ssure is connected is no dorsum sellae, preclinoid wall, or pi- with the glossopharyngeal canal as discussed tuitary fossa, although there is a membranous earlier. The transverse passage for the trigem- pituitary sac below which there is a subpi- inal mandibular ramus inside the postorbital tuitary space containing the internal carotid process projects laterally from the endocast; arteries (Gegenbaur, 1872; Allis, 1928). its proximal region has been reconstructed as In Cladodoides, the anterior margin of the traversing the jugular canal and arising at the dorsum sellae separates the hypophyseal and trigeminal foramen. The super®cial ophthal- mesencephalic chambers almost directly be- mic complex passes through cartilage form- low the oculomotor foramen as in neosela- ing the roof of the jugular canal. Presumably, chians, but its oculomotor passage is located this cartilage formed either in the lateral much farther anteriorly relative to the trigem- commissure or between it and the primary inal nerve and utricular recess (this can be 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 57 seen by comparing the endocasts of Clado- could potentially have grown to a much doides and Notorynchus; ®g. 27). The size of greater size. Although the braincase is essen- the hypophyseal and mesencephalic cham- tially complete, there is some evidence that bers in Cladodoides may therefore re¯ect the it was not fully developed. Ontogenetic stud- greater extent of the polar cartilage-derived ies of the neoselachian braincase (e.g., De region, since the oculomotor foramen is lo- Beer, 1931; Holmgren, 1940) suggest that the cated farther anteriorly not only to structures narrow openings along the dorsal midline behind the level of the hypophysis (e.g., the and the large bucco-hypophyseal fenestra in trigeminal foramen) but also to features far- Cladodoides are juvenile or submature fea- ther dorsally (e.g., the trochlear foramen). In tures. Absence of a precarotid commissure Hybodus and Tribodus, the mesencephalic behind the bucco-hypophyseal fenestra is an- and hypophyseal chambers are proportioned other possible juvenile feature. Absence of a as in modern sharks (Maisey, 2004a: ®g. 5). chondri®ed medial capsular wall might be an The mesencephalic chamber in Orthacanthus ontogenetically primitive condition, but the is quite long although the polar cartilage ap- wall is unchondri®ed even in much larger parently made a smaller contribution to its (presumably adult) Paleozoic sharks (e.g., ¯oor than in Cladodoides (Schaeffer, 1981: ``Tamiobatis sp.''). ®g. 14). The absence of multilayered prismatic cal- A weak lateral hypophyseal ridge extends ci®cation may represent a juvenile character- from the dorsum sellae as far as the optic istic, but there is no modern paradigm (mul- foramen along the inner wall of the cranial tiple-layered calci®cation is rarely found in cavity in Cladodoides. This ridge separates neoselachians and typically only in high- the lateral wall of the cranial cavity from the stress regions of the jaws; Summers, 2000). upper margin of the hypophyseal chamber, as Nevertheless, multiple-layered calci®cation in some neoselachians (®g. 7; also see Mais- occurs throughout the braincases of Ortha- ey, 2004b). The anterior margin of the hy- canthus and many other Paleozoic sharks. pophyseal chamber is weakly de®ned by a Such calci®cation may have developed cu- low transverse presphenoid ridge (ϭ praes- mulatively with age, and the presence of suc- phenoõÈdvorsprung; Gegenbaur, 1872). A sim- cessive prism layers in such forms suggests ilar ridge is present in some neoselachians that calci®cation was locally periodic rather (e.g., Heptranchias, Squalus, Dalatias, than gradual. There is little evidence of sub- Deania, Mustelus) but not in others. sequent in vivo resorption or disruption of The hypophyseal chamber of Orthacan- the laminar arrangement once formed, al- thus is much shallower than in Cladodoides, though the layers are often broken up post- although its dorsum sellae is almost as prom- mortem. Adult Cladodoides may therefore inent (Schaeffer, 1981: ®g. 14). The extent of have been more heavily calci®ed than the these structures have not been determined in small specimen investigated. If multiple lay- Tamiobatis vetustus, but there is a large hy- ers of prismatic cartilage were developed in pophyseal chamber in ``Tamiobatis sp.'' The Cladodoides, this presumably commenced dorsum sellae and hypophyseal chamber are only after the braincase was fully formed. On both very deep in ``Cobelodus'' (Maisey, the other hand, multiple-layered calci®cation 2004a: ®g. 8), and the dorsum sellae extends is also absent or poorly developed in other much farther dorsally in modern chimaeroids much larger Paleozoic sharks (e.g., Clado- than in neoselachians (e.g., Callorhinchus; selache), so it is possible that Cladodoides De Beer, 1937: pl. 21). never possessed more than a single layer of prisms. It is not possible to infer any phy- DISCUSSION logenetic signi®cance to the absence of multiple-layered calci®cation in Cladodoides ONTOGENETIC ASPECTSOFTHEBRAINCASE: without a greater understanding of its wider The Cladodoides braincase probably repre- phylogenetic distribution and development in sents an individual just approaching its full elasmobranchs generally. ontogenetic maturity, but which may still Comparison with neoselachian embryos have been a small, submature shark and nevertheless indicates that in most respects 58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 the ontogenic development of the Cladodoi- lae) are already fused to the capsules before des braincase is fairly complete. For exam- the occipital arch is complete dorsally (El- ple, the postorbital process seems to be fully Toubi, 1949), implying some heterochrony in developed and the palatine ramus is enclosed development of the occipital arch and closure within its ¯oor, so ontogenetic fusion of the of the otico-occipital ®ssure in neoselachi- lateral commissure to the rest of the brain- ans. The occipital arch is complete in Cla- case was well advanced. In addition, most of dodoides (and in 62-mm Heterodontus), sug- the metotic ®ssure in Cladodoides is closed, gesting that the dorsal part of the otico-oc- apart from a small notch extending from the cipital ®ssure persisted later in ontogeny than ventrolateral margin of the glossopharyngeal in Squalus. canal. Consequently, most of the hypotic PRECEREBRAL FONTANELLE: The presence lamina is fused to the capsular ¯oor as in of a precerebral fontanelle has been regarded adult neoselachians, rather than leaving an as an elasmobranch synapomorphy (Maisey, elongate ventral otic notch as in Orthacan- 1984a; Gaudin, 1991), although some (e.g., thus. The interorbital septum is essentially Holmgren, 1942; Schaeffer, 1981) have sug- complete and the embryonic prootic foramen gested that the fontanelle is homologous to is absent. Moreover, the embryonic basicra- the ethmoid canal in holocephalans. How- nial cartilages (trabeculae, polar cartilages, ever, the ethmoid canal is an enclosed tube parachordals) are completely fused together, containing the anterior portion of the profun- and an extensive contribution to the basicra- dal nerve dorsal to the olfactory capsules and nium by the polar cartilage is suggested by is apparently a secondarily roofed-over ex- the position of the orbital articulation and ef- tracranial space (De Beer and Moy-Thomas, ferent pseudobranchial artery, by the extent 1935; Didier, 1995), not a remnant of the cra- of the hypophyseal chamber and dorsum sel- nial cavity as proposed by Holmgren (1942). lae, and by the position of both the prefacial It is nevertheless possible that absence of a commissure and the facial foramen behind precerebral fontanelle in chimaeroids is sec- the orbit (discussed further below). ondary, and that its presence is a chondri- Persistence of an otico-occipital ®ssure chthyan synapomorphy (Coates and Sequei- throughout ontogeny may be a conserved ra, 2001a). A precerebral fontanelle is pre- feature of crown-group gnathostomes. In sent in Gladbachus and Doliodus (Heidtke Cladodoides, much of the embryonic metotic and KraÈtschmer, 2001; Miller et al., 2003) ®ssure is presumably closed, and the hypotic but has not been observed in Pucapampella lamina/basicapsular commissure is inferred (although its ethmoid region has not yet been to have enclosed a wide glossopharyngeal described; Maisey, 2001b; Maisey and An- canal below the otic capsule. The otico-oc- derson, 2001). cipital ®ssure in Cladodoides is therefore less EFFERENT PSEUDOBRANCHIAL ARTERY:In extensive than in much larger (presumably neoselachians, the efferent pseudobranchial adult) Orthacanthus braincases in which the and internal carotid arteries meet inside the embryonic metotic ®ssure persisted as a ven- cranial cavity, dorsal to the trabecular-polar tral otic notch. The metotic ®ssure in Clado- plate (De Beer, 1924). Usually, the ophthal- doides presumably closed early in ontogeny, mic and efferent pseudobranchial arteries as in neoselachians, although its embryonic branch within the orbit, and there is only a otico-occipital ®ssure was persistent farther single exit for the combined vessel in the or- dorsally. From a phylogenetic perspective, bital wall (e.g., Chlamydoselachus; Allis, therefore, ontogenetic closure of the metotic 1923a: ®g. 52). In chimaeroids, the efferent ®ssure may have preceded obliteration of the ``pseudobranchial'' artery is specialized to dorsal otico-occipital ®ssure. Closure of the provide the main vascular supply to the brain metotic ®ssure precedes fusion of the occip- (e.g., Callorhinchus; De Beer and Moy- ital arch to the otic capsules in some neose- Thomas, 1935). The efferent vessel is unin- lachians (e.g., 62-mm Heterodontus; Holm- terrupted by a pseudobranch, and the spiracle gren, 1940). However, no consistent pattern closes early in ontogeny, while internal ca- is discernible and in 64-mm Squalus the lat- rotid and ophthalmic arteries are absent in eral rudiments of the arch (the occipital pi- the adult. In the primitive Devonian chon- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 59 drichthyan Pucapampella, the efferent pseu- atoquadrate). The con®guration of the effer- dobranchial artery apparently passed through ent pseudobranchial in neoselachians and the basicranium via a canal which opens near chimaeroids may therefore represent two dis- the posterior margin of the optic foramen, tinct apomorphic patterns, neither of which but there is no evidence of a separate oph- primitively characterizes chondrichthyans. thalmic foramen, suggesting that it met the Thus, a connection between the internal ca- efferent pseudobranchial within the orbit rotid and efferent pseudobranchial arteries (Maisey, 2001b). Nothing is known of their dorsal to the trabeculae is interpreted here as arrangement in Gladbachus or Doliodus. an apomorphic feature of elasmobranchs but In modern osteichthyans, the efferent not all chondrichthyans. pseudobranchial and internal carotid arteries ANTERIOR PALATOQUADRATE ARTICULA- are conjoined below the trabeculae and enter TION: In modern elasmobranchs, there may the basicranium together. The same arrange- be no anterior connection between the pala- ment has been identi®ed in primitive extinct toquadrate and braincase (e.g., batoids). actinopterygians (e.g., Ligulalepis; Basden Where an articulation is present, however, and Young, 2001). It is unlikely that such a two distinct patterns are recognizable. In one subcranial commissure existed in Pucapam- pattern, either an articular surface is present pella and the two vessels probably passed in the ethmoid region (e.g., Heterodontus,or- through the trabecular cartilage separately ectolobiforms), or the palatoquadrate has a (Maisey and Anderson, 2001). StensioÈ ligamentous attachment to the braincase in (1963a) and Goujet (1984) argued that the the corresponding position (e.g., Scyliorhin- efferent pseudobranchial arrangement in pla- us). In the other pattern, the dorsal margin of coderms resembled that of elasmobranchs the palatoquadrate bears an orbital process rather than osteichthyans, but Young (1980, extending a considerable distance into the or- 1986) has presented evidence that in placod- bit and usually resting against an articular erms the artery probably passed medially in surface located between the optic and effer- a basicranial groove and joined the internal ent pseudobranchial foramina (e.g., Chla- carotid before it entered the braincase. Con- mydoselachus, hexanchiforms, squaloids, sequently, the most parsimonious scenario is squatinoids, and pristiophoroids; Maisey, that the efferent pseudobranchial artery prim- 1980). itively passed through the trabeculae in gna- Holmgren (1942: 140) compared the an- thostomes, and this arrangement is conserved terior palatoquadrate articulation in Clado- by osteichthyans and primitive chondri- doides and Chlamydoselachus and noted that chthyans such as Pucapampella (Holmgren, while both are positioned anteriorly on the 1943; Young, 1986; Maisey and Anderson, palatobasal shelf, their topographic relation- 2001). Uni®cation of the internal carotid and ship to the optic nerve differs. He concluded efferent pseudobranchial arteries dorsal to that Cladodoides ``seems to have behaved as the trabeculae is apparently an apomorphic a squaloid shark, but with a shelf (crista su- condition shared by neoselachians and many bocularis) developed behind the orbital artic- extinct sharks including Cladodoides, Ortha- ulation.'' Maisey (1980) and Schaeffer canthus, and Hybodus. Although it has been (1981) recommended restricting the term or- claimed that the efferent pseudobranchial in bital articulation to the pattern found in mod- modern chimaeroids is located above the tra- ern orbitostylic sharks. However, while this beculae as in elasmobranchs (Schaeffer, recommendation certainly emphasizes the 1981), the condition in Callorhinchus as de- derived nature of their anterior palatoquad- scribed by De Beer and Moy-Thomas (1935) rate articulation, it does little to resolve its suggests that it is located at the lateral margin possible homology with `èthmoidal'' and of the trabeculae, an arrangement which `òrbital'' articulations in other neoselachians could have been derived from that found in or in elasmobranchs generally. Additionally, Pucapampella, since the efferent pseudob- there has been uncertainty as to whether any ranchial in Callorhinchus also enters the of the anterior palatoquadrate articulations in braincase via a short canal (supposedly at the sharks are homologous to those found in oth- former boundary of the trabeculae and pal- er gnathostomes, such as the palatobasal ar- 60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 ticulation of osteichthyans or the `òrbital'' anterior to the process, whereas in the squa- and ``basal'' articulations in placoderms. loids the growth principally takes place fron- Turning ®rst to the question of homology tally'', implying that migration of the artic- between ethmoidal and orbital articulations ulation has occurred in squaloids. However, in elasmobranchs, De Beer (1931) and Holm- the presumptive articulations initially do not gren (1941: char. 47) considered that an or- form in the same place and their relationships bital process was present in many modern to the polar cartilage are completely different and extinct elasmobranchs, including galeo- from their earliest appearance. The articula- morphs and all orbitostylic sharks. However, tion in Scyliorhinus ®rst appears beneath the the articulation in Heterodontus and galeo- anterior part of the orbit (Holmgren, 1940: morphs does not correspond topographically ®g. 103), whereas the orbital articulation in with the orbital articulation in orbitostylic Squalus and Etmopterus arises beneath its sharks. According to De Beer (1931, 1937) posterior part. Importantly, neither of these and Holmgren (1940), the orbital process in articulations shows appreciable anteroposte- Heterodontus and Scyliorhinus is con®ned to rior relocation within the orbit during sub- the anterior part of the trabecular region, and sequent ontogeny (Holmgren, 1940: ®gs. 55, any articular surface is developed far anterior 67, 79, 81). Thus, apart from the fact that to the polar cartilage and to the efferent pseu- both articulations are formed on the lateral dobranchial artery (e.g., Heterodontus,or- margin of the trabeculae, there is little to sug- ectolobiforms). By contrast, in orbitostylic gest that they are homologous to each other. sharks (e.g., Squalus, Etmopterus), the artic- In addition, these articulations differ in ular surface is located at the junction of the their topographic and ontogenetic relation- trabecular and polar cartilages and lies im- ships to the suborbital cartilage. Holmgren mediately anterior to the efferent pseudob- (1940: 253) noted that in Scyliorhinus, ``the ranchial foramen (Maisey, 1980). Holmgren membrane connecting the palatoquadrate (1940, 1942) found evidence for two adja- with the trabecula medially chondri®es, cent embryonic palatoquadrate connections forming the subocular cartilage or shelf. In in this part of the orbit in Squalus, one with Squalus and Etmopterus no such shelf is pre- the posterior end of the trabecular plate (ul- sent, the connecting membrane never chon- timately forming the orbital articulation) and drifying''. Note, however, that the cartilage another (which ultimately disappears) with forming the shelf in Scyliorhinus is posterior the polar cartilage slightly farther posteriorly. to the articulation, whereas the ``connecting The articulation in Heterodontus and galeo- membrane'' in Squalus is anterior to it. Thus, morphs lies on the trabecular margin but has even if the articulation is homologous in no obvious connection to the polar cartilage these taxa, the ``connecting membranes'' dif- and is located much farther anteriorly than fer topographically. The anterior articulation the orbital articulation in orbitostylic sharks. in Scyliorhinus resembles that of Cladodoi- Arguably, these articulations are not homol- des in lying anterior to the suborbital shelf, ogous on both topographic and ontogenetic and in both cases the orbital artery penetrates grounds. Support for that view is found in the shelf posteriorly. The articular surface in the different ontogenetic origins of the car- Cladodoides therefore resembles the orbital tilage connecting the palatoquadrate to the articulation of orbitostylic sharks in its rela- braincase; in Scyliorhinus, a separate chon- tionship to landmark arterial vessels and pre- dri®cation arises from the trabecular cartilage sumably to the polar cartilage, but also re- (the ``lateral trabecular process'' of Holm- sembles the ethmoidal articulation in galeo- gren, 1940) and fuses with the palatoquadrate morphs and Heterodontus in its topographi- to form the orbital process, but in Squalus cally anterior location within the orbit and the orbital process arises as part of the pal- the presence of a suborbital shelf (penetrated atoquadrate (Holmgren, 1940: 252). by the orbital artery) behind the articulation. Holmgren (1940: 253) argued that the rea- In Hybodus basanus, there is extensive lat- son the orbital articulation is located farther eral contact between the braincase and the anteriorly in Scyliorhinus is simply because anterior part of the palatoquadrate (Maisey, the trabeculae do not ``grow out very much 1983). However, only the anterior part of this 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 61 contact actually forms an articular surface in present in Chlamydoselachus, anterior to the the ethmoid region, a considerable distance orbital articulation and medial to the olfac- both from the efferent pseudobranchial fo- tory capsule and ectethmoid process. Actu- ramen and from the presumed site of the po- ally, no real articulation is present here; in- lar cartilage. In the primitive neoselachian stead (according to Allis, 1923a: 167), the Synechodus, a prominent articular surface for anterior part of the palatoquadrate in Chla- the palatoquadrate occupies the anterior part mydoselachus merely ``rests, when the of the orbit in front of the optic and efferent mouth is closed, against the lateral edge of pseudobranchial foramina, as in Heterodon- the solum nasi''. Holmgren (1941) also did tus and galeomorphs, but in Synechodus a not recognize a separate ethmoid articulation suborbital shelf is absent and there is no ev- in Chlamydoselachus. In fact, the membra- idence of any palatoquadrate articulation in nous anterior connection between the pala- the inferred site of the polar cartilage (Mais- toquadrate and braincase described by De ey, 1985: ®gs. 3, 4). The suborbital shelf ex- Beer (1931) in Scyliorhinus does not form a tends for a considerable distance anterior to cranio-quadrate articulation and probably the efferent pseudobranchial foramen in Hy- consists only of connective tissue, as in No- bodus basanus (Maisey, 1983). torynchus (Wolfram, 1984; Maisey, 2004b). Thus, in elasmobranchs there may be (1) In Orthacanthus, the anterior part of the pal- an articular surface situated immediately an- atoquadrate has a shallow depression dorsal- terior to the efferent pseudobranchial fora- ly (Hotton, 1952). In articulated specimens, men and to the inferred anterior margin of this depression is located below the olfactory the polar cartilage-derived region (either con- capsule (Schaeffer, 1981: ®g. 1). A similar ®ned to the orbital ¯oor, as in Cladodoides depression is also found on other Paleozoic and Orthacanthus, or extending into the orbit shark palatoquadrates (including material re- behind the optic foramen as in Chlamydo- ferred provisionally to Cladodus elegans). In selachus); (2) an articular surface situated Hybodus basanus, a short ¯at area on the farther posteriorly in the orbit, but still as- corresponding part of the palatoquadrate sociated with the efferent pseudobranchial makes contact with the ventral surface of an artery and located behind the optic foramen ethmopalatine process on the internasal plate (e.g., Squalus, Notorynchus); or (3) an ante- (Maisey, 1983). This may correspond to the riorly situated articular surface that is not as- ``rostral articulation'' in Orthacanthus sociated with either the efferent pseudobran- (Schaeffer, 1981: 56), but probably neither is chial foramen or with the polar cartilage homologous to the orbital articulation in neo- (e.g., modern Heterodontus and galeo- selachians. At best, the anterior depression morphs, and perhaps Synechodus and Hybod- may have formed a sliding surface (e.g., Or- us). A fourth (nonarticular) condition is rep- thacanthus) or a buttress (Hybodus) beneath resented by fully hyostylic elasmobranchs, in the olfactory capsule like that described by which an ethmoid or orbital articulation is Allis (1923a) in Chlamydoselachus.Itmay absent (e.g., batoids, Tribodus). These obser- be homologous to the ethmoidal articulation vations suggest that the anterior palatoquad- in galeomorphs and Heterodontus, although rate articulation in Paleozoic sharks such as in these forms the articulation is with the me- Cladodoides, Tamiobatis, and Orthacanthus dial surface of the palatoquadrate rather than is primitively homologous to the orbital ar- with its dorsal surface. ticulation in modern orbitostylic sharks, al- Primitive actinopterygians possess an au- though the orbitostylic condition is probably topalatine (or `èthmoid''; Janvier, 1996) ar- derived in extending farther dorsally into the ticulation anteriorly, plus a palatobasal artic- orbit and behind the optic foramen (Maisey, ulation farther posteriorly (e.g., Mimia, Moy- 1980). However, there is little evidence to thomasia; Gardiner, 1984a). The autopalatine suggest homology of these articulations to articulation in actinopterygians may be ho- the anterior one in Heterodontus, galeo- mologous to the ethmoidal articulation in morphs, Synechodus or Hybodus. chondrichthyans, since it is developed on the De Beer (1931) and Schaeffer (1981: 56) lateral margin of the embryonic trabeculae both claimed that an ethmoid articulation is and involves the anteriormost (autopalatine) 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 part of the upper jaw. An autopalatine artic- symphysis, similarities noted by Holmgren ulation is probably absent in Acanthodes and (1942) in the anterior part of the suspenso- placoderms (Gardiner, 1984a). rium of Acanthodes and sharks are probably Homology of the elasmobranch orbital ar- convergent. ticulation to the palatobasal articulation in Gardiner (1984a: 299) drew attention to osteichthyans was ®rst advocated by Huxley topographic and ontogenetic similarities be- (1876) and was supported by De Beer (1931: tween the osteichthyan palatobasal articula- 628), who argued for a separate ethmoid ar- tion and the neoselachian orbital articulation, ticulation anterior to the orbital one in some noting that ``the basipterygoid process is typ- modern sharks (e.g., Chlamydoselachus; but ically developed in osteichthyans, but is also as discussed above, there is actually little ev- recognizable in many selachians where it idence of such an ethmoid articulation in forms part of the subocular shelf as in Hep- Chlamydoselachus, hexanchiforms, or other tranchias, and in Squalus . . . where it arises orbitostylic sharks). De Beer (1931: 419) from the site of the polar cartilages.'' There noted that the elasmobranch orbital (his bas- are also compelling topographic similarities al) articulation and the osteichthyan basitra- in the palatobasal articulations of primitive becular articulation are both located ventral osteichthyans and Pucapampella (Maisey, to the lateral head vein and are both dorsal 2001b; Maisey and Anderson, 2001). In and anterior to the palatine ramus of the fa- some primitive sarcopterygians, there is a cial nerve (the same relationship may also be ``suborbital ledge'' with a palatal articular postulated in Cladodoides; see sections on surface (Yu, 1998). The osteichthyan pala- the jugular canal and palatine ramus later in tobasal articulation and the elasmobranch or- this work). bital articulation are both associated devel- Some authors have strongly contested De opmentally with the polar cartilage and with Beer's proposal (e.g., Holmgren, 1943; De- the efferent pseudobranchial artery, although villers, 1958). Miles (1965, 1973) interpreted in osteichthyans this vessel meets the internal Acanthodes as having an osteichthyan-like carotids below the basicranium rather than basipterygoid process and a palatobasal ar- above it as in elasmobranchs (but not nec- ticulation with the palatoquadrate, and he essarily in all chondrichthyans; Maisey and concluded that this was different from the or- Anderson, 2001: 710). The elasmobranch orbital articulation in elasmobranchs. Holm- bital articulation can therefore be considered gren's position is somewhat ambiguous, homologous to the palatobasal articulation of since he has argued that the orbital and pal- osteichthyans on the basis of its topographic atobasal articulations are not homologous relationship to the lateral head vein and the (Holmgren, 1943), but has suggested else- palatine ramus of the facial nerve (De Beer, where that the autopalatine process in Acan- 1931) and to the site of the embryonic polar thodes is homologous to the orbital process cartilage (Gardiner, 1984a). It is concluded in squaloids and hexanchiforms (Holmgren, that (1) a palatobasal articulation is primi- 1942), basing his argument on the supposed tively present in Acanthodes, osteichthyans, presence of an anterior symphysis anterior to and chondrichthyans, and therefore probably the process in Acanthodes (which he regard- represents a conserved gnathostome charac- ed as an `àlmost identical organization of the ter; (2) it has become highly specialized in palatoquadrate'' to that of squaloids and hex- modern orbitostylic sharks; and (3) it has anchiforms). According to Miles (1965, been secondarily lost (perhaps independent- 1973), however, Acanthodes lacks an anterior ly) in Heterodontus, galeomorphs, and ba- palatoquadrate symphysis. Perhaps even toids. more signi®cantly, a new reconstruction of It has been claimed that ``basal'' (i.e., pal- the head in the early chondrichthyan Puca- atobasal) and `òrbital'' articulations are both pampella suggests that the palatoquadrates present in some placoderms (e.g., Tapinos- were separated by a wide ethmoid region an- teus, Kujdanowiaspis; StensioÈ, 1963a). teriorly, and that no symphysis was present Clearly such conjunction would refute ho- (Maisey and Janvier, in prep.). If chondri- mology between them (Miles, 1968). How- chthyans primitively lacked a palatoquadrate ever, Goujet (1984) found no evidence of a 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 63 palatobasal articulation in placoderms and tion in Pucapampella, Debeerius, osteichth- pointed out that the palatoquadrate is con- yans, and Acanthodes suggests that its more ®ned to an area below the ``suborbital'' shelf anterior position in elasmobranchs such as (which includes several structures associated Cladodoides and Akmonistion is a derived with the orbital cartilage in other gnathosto- condition. Paradoxically, the anterior posi- mes; e.g., optic pedicel, eye muscle insertion is apparently plesiomorphic for orbitos- tions, and branches of the oculomotor nerve). tylic neoselachians (e.g., Chlamydoselachus) Furthermore, StensioÈ's (1963a: ®gs. 16, 17, and it occupies a more posterior position in 21, 22) illustrations reveal several features cladistically advanced orbitostylic taxa. In challenging the identity of these articula- many orbitostylic sharks, the palatoquadrate tions. He inferred the anterior limit of the lacks a ``palatine'' component; instead, the polar cartilage-derived region in Kujdanow- jaws are abbreviated and do not extend far iaspis from the supposed position of the ef- (if at all) beneath the ethmoid region (e.g., ferent pseudobranchial foramen, but this may Squalus, hexanchiforms). The posterior po- have been incorrectly identi®ed (see below). sition of the orbital articulation may be a de- In Tapinosteus, the supposed orbital articu- rived feature, conceivably related to the pres- lation is apparently located even farther an- ence of a ``shortened'' palatine region. teriorly, on the lateral margin of a region that Recent discoveries suggest that separate is presumably derived from the trabecular ethmoidal (or rostral) and palatobasal artic- cartilage (StensioÈ, 1963a: ®g. 59) and is ulations are present in Debeerius (a primi- widely separated from the region supposedly tive, nonholostylic paraselachian holocephal- derived from the polar cartilage (like the eth- an; Grogan and Lund, 2000) and in Puca- moidal articulation in galeomorphs and Het- pampella (Maisey and Anderson, 2001). At erodontus). The supposed basal articulation present, it is not known whether a corre- in placoderms is therefore unlikely to be ho- sponding articulation is present in Gladba- mologous to the palatobasal articulation in chus or Doliodus.InPucapampella, an artic- osteichthyans. However, Young (1986: 30) ular surface is present on the anteriomost part noted that in Buchanosteus there is a close of the palatoquadrate on either side of a wide association between one of the cranio-quad- ethmoidal region as in osteichthyans (there is rate articulations and the groove for the ef- apparently no palatoquadrate symphysis; ferent pseudobranchial artery, which passes Maisey and Janvier, in prep.). Although the between two articular areas (he also recog- ethmoid region is still unknown in Pucapam- nized a similar arrangement in Romundina). pella, the anterior palatoquadrate articulation Thus, while there is still considerable dis- evidently extended far anteriorly. Conjunc- agreement about the articular relations be- tion of the ethmoidal and palatobasal artic- tween the palatoquadrate and braincase in ulations in Pucapampella provides compel- placoderms (reviewed by Young, 1986), ling evidence that both articulations were there may be an articulation associated with primitively present in chondrichthyans. As the polar cartilage-derived region of the Schaeffer (1981: 57) noted, the palatoquad- braincase in some taxa. rate in modern chimaeroids is fused to the Coates and Sequeira (1998) suggested that braincase adjacent to the polar cartilage, be- presence of an ethmoidal articulation is an low the efferent pseudobranchial foramen elasmobranch synapomorphy and that the ar- (i.e., in the topographic position of the pal- ticulation was primitively restricted to the atobasal articulation). Collectively, these anterior part of the suborbital region (e.g., ®ndings support the hypothesis that ethmoid- Cobelodus aculeatus, Akmonistion zangerli). al and palatobasal articulations are primitive- Based on their description, however, the sup- ly present in crown group gnathostomes and posed ethmoidal articulation in Akmonistion were conserved in primitive chondri- probably corresponds to the orbital one in chthyans. Since the palatobasal articulation Cladodoides (although the extent to which was located in the posterior part of the orbit the polar cartilage contributed to the basicra- in primitive osteichthyans and chondri- nium in Akmonistion is admittedly uncer- chthyans, its more anterior articulation in Pa- tain). The posterior position of this articula- leozoic sharks such as Cladodoides is prob- 64 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 ably apomorphic. Although the orbital artic- passage leading to the hypophyseal chamber ulation of modern orbitostylic sharks may be is sometimes present (e.g., Echinorhinus; homologous to the osteichthyan palatobasal Shirai, 1992: pl. 3B, 18A). The mode of de- articulation, as claimed by Huxley (1876), velopment of the adenohypophysis in hag®sh De Beer (1931) and Gardiner (1984a), its un- is still poorly understood. usual dorsal extent into the orbit is also prob- In modern elasmobranchs, fusion between ably apomorphic (Maisey, 1980). The eth- the trabeculae anteriorly produces an elon- moidal articulation has presumably been lost gated median space (the polar fenestra; secondarily in Cladodoides and modern or- Holmgren, 1940). As development proceeds, bitostylic sharks, but is conserved in Heter- this fenestra becomes reduced, leaving only odontus and galeomorphs. the bucco-hypophyseal fenestra and internal TRABECULAE,POLAR CARTILAGE, AND SUB- carotid foramina. The ontogenetic timing of ORBITAL SHELF: According to De Beer (1931, its closure is variable, occurring early in ba- 1937), after the polar cartilage becomes toids and many sharks but later in Scylior- fused with the trabeculae in Scyliorhinus, its hinus. One of the earliest parts of the brain- position is still marked by notches on the tra- case ¯oor to form is the postpituitary com- becular-polar bar. The notch de®ning its an- missure, connecting the polar cartilages be- terior margin contains the efferent pseudob- low the level of the parachordals and behind ranchial artery while the posterior one con- the internal carotids (e.g., Scyliorhinus;De tains the pituitary vein, but these vessels be- Beer, 1931, 1937; Holmgren, 1940). This come enclosed by separate foramina as commissure separates the bucco-hypophyseal chondri®cation becomes more advanced. In fenestra (the remaining portion of the polar Etmopterus, the polar cartilage and trabecu- fenestra) from the basicapsular fenestra in the lae are united above and below the efferent parachordals. During early ontogenetic stag- pseudobranchial artery near the anterior mar- es, the internal carotids also enter the brain- gin of the polar cartilage (Holmgren, 1940). case via the bucco-hypophyseal fenestra. A The efferent pseudobranchial and pituitary precarotid commissure subsequently devel- vein foramina therefore represent important ops, separating the carotid foramen from the developmental and topographic landmarks in remainder of the bucco-hypophyseal fenestra neoselachians (and in gnathostomes gener- farther anteriorly. Just behind the carotid fo- ally) by indicating the original site and ramen, paired processes arise from the polar boundaries of the polar cartilage. cartilages and extend inward toward the no- The bucco-hypophyseal fenestra (often tochord, ultimately giving rise to the dorsum secondarily closed in neoselachians) is locat- sellae. ed in front of the internal carotid opening, According to Sewertzoff (1899), the optic which represents an important topographic and oculomotor foramina provide important landmark indicating the extent of the polar developmental landmarks indicating the ap- cartilage contribution to the basicranium. proximate line of fusion between the embry- During ontogeny, the hypophysis in lam- onic trabecular and orbital cartilages in the preys and gnathostomes becomes associated lateral walls of the braincase. Assuming that with the ventral tip of the notochord within the relationship of the optic and oculomotor a sac or chamber (pituitary fossa or hypo- foramina to the trabeculae and orbital carti- physeal chamber) formed by an invagination lage in Cladodoides was the same as in neo- of the cranial wall between the trabecular- selachians, during ontogeny its trabeculae polar cartilage and basal (parachordal) plate would have overlapped the polar cartilage (Goodrich, 1930). In gnathostomes, there is posteriorly in the vicinity of the bucco-hy- usually an additional extracranial space (the pophyseal fenestra and dorsal to the anterior subpituitary space of Allis, 1928) traversed margin of the parachordal plate. In Chlamy- by the pituitary vein and lying between the doselachus, the orbital articulation ®lls the posterior ends of the trabecular-polar carti- anterior part of the orbit (Allis, 1923a; Holm- lages and the parachordals. In modern adult gren, 1940, 1941), although the locations of elasmobranchs, the bucco-hypophyseal fe- the efferent pseudobranchial and pituitary fo- nestra is usually closed although a narrow ramina suggest that the polar cartilage-de- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 65 rived area was no greater than in other neo- is almost as long as the trabecula (e.g., Cal- selachians (unfortunately, the early develop- lorhinchus; De Beer and Moy-Thomas, mental stages in Chlamydoselachus have 1935), although it has a similar relationship never been described and the extent of its to the efferent pseudobranchial artery and pi- embryonic cranial cartilages is still conjec- tuitary vein as in neoselachians (StensioÈ, tural). 1963a). It is uncertain whether the polar car- The polar cartilage makes a relatively mi- tilage in chimaeroids is extensive from its nor contribution to the braincase in modern ®rst inception in the embryo or if it under- osteichthyans (e.g., Amia; Bjerring, 1978) goes enlargement during ontogeny, but in and it is reduced or even lost in some teleosts Hydrolagus, the telencephalon is apparently (where the pituitary vein passes ventral to the elongated secondarily during ontogeny, and basicranium instead of entering it). It has in earlier stages the brain is proportioned been suggested that the polar cartilage made much as in modern sharks (G. Northcutt, per- an extensive contribution to the basicranium sonal commun. 2003). It is also uncertain in some arthrodires (e.g., Kujdanowiaspis, whether presence of an extensive polar car- Tapinosteus; StensioÈ, 1963a), based on the tilage-derived region in the basicranium rep- belief that the efferent pseudobranchial ar- resents a synapomorphy of holocephalans tery was located anteriorly and passed dorsal and some Paleozoic sharks or a convergently to the trabecular region before meeting the acquired feature. A phylogenetic relationship internal carotids, as in neoselachians. How- between some ``cladodont'' sharks (especial- ever, if it joined the internal carotid below ly symmoriids and stethacanthids) and hol- the trabeculae and entered the braincase far- ocephalans has been proposed elsewhere ther posteriorly (as in osteichthyans), such an (e.g., Janvier, 1996: ®g. 4.13.9; Coates and interpretation is not justi®ed (Young, 1980, Sequeira, 2001a), although support for that 1986). proposal came mainly from equivocal post- The polar cartilage apparently made a far cranial features. greater contribution to the basicranium in In Cladodoides, the suborbital shelf ex- Cladodoides than in neoselachians. There is tends between the efferent pseudobranchial also some evidence of extensive basicranial artery and pituitary vein, suggesting that it is contribution by the polar cartilage in Tam- developed on the lateral margin of the polar iobatis vetustus and ``Tamiobatis sp.'' In all cartilage-derived region. In Orthacanthus, these forms, the braincase has a generalized the suborbital shelf is much shorter and the platybasic morphology and an elongated otic efferent pseudobranchial foramen is located region (including the region formed by the farther posteriorly (near the exit of the orbital parachordals). However, an extensive polar artery; Maisey, 1983: ®g. 13B), suggesting cartilage-derived region is also suspected in that its polar cartilage made a somewhat tropibasic Paleozoic sharks with a compara- smaller contribution to the basicranium than tively short otic region (e.g., ``Cobelodus''; in Cladodoides. According to De Beer in prep.), based on the position of the effer- (1937: 61), the suborbital shelf in 36-mm ent pseudobranchial foramen and the size of Scyliorhinus arises from two centers, one an- the hypophyseal chamber and dorsum sellae. terior (associated with the lamina orbitonas- On the other hand, in Orthacanthus the ba- alis and forming a mesenchymatous connec- sicranial contribution of the polar cartilage tion with the palatoquadrate) and one poste- was comparatively small, judging from the rior (enclosing the orbital artery). Holmgren posterior position of the efferent pseudobran- (1940: 167±168) also described a similar ar- chial foramen within its orbit, although the rangement of anterior and posterior subor- braincase is platybasic and its otic region is bital mesenchymatous centers in 30-mm Scy- elongated. The extent of polar cartilage and liorhinus. In 40-mm embryos, he described parachordal contributions to the basicranium the suborbital shelf as extending `àlong the therefore seems to have been considerably trabecula, beginning in front with the cranio- more variable in these Paleozoic sharks than quadrate connection, from which its anterior in modern elasmobranchs. part cannot be delimited.'' Behind this, the In modern chimaeroids, the polar cartilage shelf continues along the trabecular margin, 66 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 becoming thinner laterally and maintaining a capampella presents us with yet another var- membranous connection with the medial side iation on this theme, since it has a broad sub- of the palatoquadrate for most of its length. orbital shelf anterior to the palatobasal artic- The efferent pseudobranchial artery accom- ulation. Such an arrangement is unknown in panies `à posterior branch of the palatine other elasmobranchs but is similar to os- nerve'' through the suborbital shelf posteri- teichthyans, in which a shelf (usually much orly, then crosses the ¯oor of the orbit where narrower than in Pucapampella) is present it ®nally enters the braincase (Holmgren, anterior to the palatobasal articulation and is 1940: ®g. 118). This arrangement differs covered ventrally by the parasphenoid (e.g., from that in Cladodoides, where the efferent Amia, Moythomasia, Eusthenopteron, Hol- pseudobranchial enters the braincase via a optychius; Jarvik, 1980; Gardiner, 1984a). canal in the suborbital shelf, and only the There is undoubtedly much homoplasy in all ophthalmic artery passes through the orbital these variations, yet there may also be phy- wall. However, both arrangements are similar logenetically meaningful signals if the pat- in one important respect, namely the close tern in Pucapampella is primitive for chon- relationship of the efferent pseudobranchial drichthyans. artery to the inferred extent of the polar car- The relationship of the efferent pseudob- tilage, since in Scyliorhinus the vessel ®rst ranchial artery to the polar cartilage is con- passes through the suborbital shelf in the sistent in modern sharks, but its relationship posterior part of the orbit, adjacent to the to the suborbital shelf is variable; in Squalus, (very small) polar cartilage. If the efferent the efferent pseudobranchial is located in the pseudobranchial artery was to become more posterior part of the orbit (a suborbital shelf completely enclosed by the suborbital shelf is absent), and the site of the polar cartilage in Scyliorhinus, the vessel would not traverse is small; in Scyliorhinus, the artery ®rst pass- the orbit before entering the braincase and es through the suborbital shelf before enter- would instead have a single entrance in the ing the cranial cavity at the trabecular-polar back of the orbit, in a position similar to cartilage junction. In Cladodoides, the artery Squalus. passed through cartilage forming the anterior Some correlation is noted in neoselachians part of the suborbital shelf (presumably near between the presence and position of the or- the anterior border of the polar cartilage), bital/palatobasal articulation and the extent where it met the ophthalmic artery before en- of the suborbital shelf, which arises from an- tering the cranial cavity. terior and posterior centers (De Beer, 1937; Beginning with Huxley (1875), several in- Holmgren, 1940). In modern orbitostylic vestigators have proposed that the trabecular sharks, there is no suborbital shelf anterior to cartilages are derived evolutionarily from the this articulation, and a shelf is rarely devel- visceral arch skeleton. Allis (1923b) addi- oped farther posteriorly. One exception is tionally suggested that the polar cartilage is Squatina, whose orbital articulation is ori- of visceral arch origin. Three lines of evi- ented obliquely, with its lower part situated dence supported those early views (reviewed in the anterior part of the orbit (presumably in De Beer, 1937: 375); the location of the the suborbital shelf meets the margin of the subpituitary space (including the passage for polar cartilage-derived region of the basicra- the pituitary vein) below the dura mater and nium, although its ontogeny still has not been above the trabeculae, indicating that the latter described). Where there is no orbital/palato- are not part of the primitive cranial wall (Al- basal articulation, a suborbital shelf may be lis, 1923b); the position of the trabeculae an- absent (e.g., batoids), or it may extend anterior to the parachordal plate (and its asso- teriorly from the site of the polar cartilage ciated somites), which precludes it from be- along the lateral margins of the trabeculae ing derived from the axial skeleton; and ex- (e.g., Heterodontus and galeomorphs). In perimental removal of neural crest tissue led Cladodoides and many other Paleozoic to abnormal development of the trabeculae sharks, the suborbital shelf is only found be- and visceral arches, indicating that their de- hind the articulation (resembling the situation velopment was related. Both Allis (1923b) in Squatina). The early chondrichthyan Pu- and De Beer (1937) suggested that the tra- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 67 beculae were formerly parts of a premandi- tion (Cerny et al., 2004); no part of the jaws bular arch. Allis (1923b) also proposed that apparently forms in the supposed maxillary the polar cartilage represents a pharyngo- primordium. Thus, Schaeffer's (1981) predic- mandibular element, but De Beer (1937) ob- tion regarding ectomesenchymatic compe- jected because a supposed pharygomandi- tence seems to be met (at least as far as the bular has been identi®ed between the pala- trabeculae are concerned), although it is still toquadrate and braincase in some sharks uncertain what the ``certain conditions'' he (Sewertzoff and Disler, 1924; but see Holm- advocated might be; one possible factor is gren, 1940, 1943; Gardiner, 1984a), and be- the absence of Hox (homeobox transcription cause the polar cartilage supposedly is of ax- factor) expression in the trigeminal neural ial rather than visceral arch derivation. crest stream (Hox is expressed in the hyoid/ Holmgren (1943) claimed that certain dorsal preotic and common branchial/postotic elements of the prootic arches may have been streams farther posteriorly); this has already incorporated into the braincase, Bertmar been implicated in the origin of jaws (Cohn, (1959) and Jarvik (1954) both believed that 2002) and may also have played a role in the the ectomesenchymatic tissue of the trabec- evolution of the trabecular plate in gnathos- ulae in ®shes represent an infrapharyngo- tomes. Mandibular and premandibular re- mandibular, and Jarvik (1954) made elabo- gions of the trigeminal neural crest stream rate reconstructions of the braincase in have not so far been recognized, although the Eusthenopteron in which he identi®ed parts dorsal ``maxillary'' condensation extends of the premandibular and mandibular arches into the premandibular region anteriorly. on the basis of topographic (serial) similari- Moreover, Dlx1 cognates in the lamprey are ties. expressed throughout the mandibular and Schaeffer (1981: 21) was highly skeptical premandibular regions, but only in the man- of such scenarios, pointing out that ``there is dibular region of gnathostomes (Kuratani et . . . no evidence from studies of neural crest al., 2004). Thus, the absence of Hox expres- migration that elements of the visceral skel- sion in the entire prechordal region, and the eton were literally incorporated into the restricted expression of Dlx cognates (and re- chondrocranium'' (see also Chibon, 1974; lated ectodermally derived growth factors) in Schaeffer, 1975). Instead, Schaeffer (1981) mandibular ectomesenchyme, may be viewed argued that the apparent absence of visceral as developmental synapomorphies that are arch tissue in the side walls of actinoptery- linked to the evolution of trabeculae and jaws gian and neoselachian braincases, which in gnathostomes. Bertmar (1959: 328, 338) had attributed to OPTIC PEDICEL AND POSITION OF OCULO- ``modi®cations in early ontogeny'', simply MOTOR FORAMEN: An optic pedicel is absent meant that ectomesenchymatic tissues are in chimaeroids and in all modern osteichth- equally competent to form visceral or chon- yans (with the possible exception of Acipen- drocranial structures under certain condi- ser; Gardiner 1984b). Gardiner's (1984a, tions. In fact, it has now been shown that the 1984b) suggestion that the pedicel represents rostralmost (trigeminal or mandibular) a primitive gnathostome character is sup- stream of neural crest cells (which ultimately ported by recent discoveries of an eyestalk populates the entire anterior head region), attachment in early osteichthyans (e.g., Lig- eventually condenses into a dorsal ``maxil- ulalepis, Psarolepis; Basden et al., 2000; Zhu lary'' primordium below the eye, and a ven- et al., 2001). There is also evidence of a sim- tral ``mandibular'' primordium below the ilar attachment in many placoderms associ- oral cavity and serially aligned with neural ated with foramina for the optic and oculo- crest condensations of the visceral arches far- motor nerves, ophthalmic artery, and pitui- ther posteriorly (Francis-West et al., 1998; tary vein as well as attachments for various Chai et al., 2000; Meulemans and Bonner- extrinsic eye muscles (Young, 1986). The Fraser, 2002). Somewhat surprisingly, the pedicel is positioned behind the optic fora- trabeculae develop from the dorsal primor- men and beneath the oculomotor foramen in dium while the palatoquadrate and Meckel's both elasmobranchs and placoderms. Young cartilage both arise in the ventral condensa- (1986) concluded that the similarity could be 68 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 interpreted either as a synapomorphy of these separate entities (Northcutt and Bemis, 1993) groups or as a primitive gnathostome con- has added an additional layer of complexity dition. In Ligulalepis, the attachment area to the identi®cation of cranial nerve foramina lies above and behind the optic foramen and in fossils. below and anterior to the oculomotor fora- Excluding the ophthalmic profundal men (Basden et al., 2000), but in Psarolepis branch (V1), the gnathostome trigeminal the attachment area is behind the optic fo- nerve includes a maxillary ramus (V2, some- ramen as in elasmobranchs and placoderms times regarded as a pretrematic ramus, al- (Zhu et al., 2001), suggesting that it is a con- though its equivalence to a pretrematic bran- served gnathostome pattern and not a syna- chial ramus is conjectural) and a mandibular pomorphy of placoderms and elasmobranchs. ramus (V3, sometimes regarded as a posttre- Having the oculomotor foramen situated matic branch). In gnathostomes generally, slightly farther posteriorly in Ligulalepis may the trigeminal nerve also has a dorsal super- be autapomorphic, although its phylogenetic ®cial ophthalmic ramus (absent in Latimeria; signi®cance cannot be reliably determined Northcutt and Bemis, 1993). In neoselachi- given the absence of a pedicel in other prim- ans, this ramus can be separate (e.g., Squal- itive actinopterygians. According to Yu us; Norris and Hughes, 1920) or intimately (1998: ®g. 1C), in Psarolepis the oculomotor associated with the super®cial ophthalmic ra- nerve is located above a large pit (said to be mus of the anterodorsal lateral line nerve, for the pituitary vein); however, according to forming the super®cial ophthalmic complex the illustration of Psarolepis in Zhu et al. (e.g., Chlamydoselachus; Allis, 1923a). The (2001: ®g. 1d), this pit forms the pedicel at- ganglia of the trigeminal nerve and buccal tachment area, in which case the oculomotor ramus of the anterodorsal lateral line nerve foramen is dorsal to the attachment as in can be so closely associated as to be insep- elasmobranchs and placoderms. arable. TRIGEMINAL,FACIAL, AND OCTAVOLATERAL In Chlamydoselachus, the maxillary ramus NERVES: The arrangement of the trigeminal of the trigeminal nerve divides into three and facial nerves is often dif®cult to recon- principal ramules, a posterior one (extending struct in fossils, as exempli®ed by the differ- to the superior labial levator muscle and the ing interpretations of Cladodoides and Tam- ventral edge of the palatoquadrate) plus two iobatis in Gross (1937), Romer (1964), anterior ones that pass across the ectethmoid Schaeffer (1981), and earlier in the present process and extend ventral to the nasal open- work. In neoselachians, the ventral part of ing (Luther, 1909; Allis, 1923a). As in many the lateral commissure is rarely chondri®ed. other neoselachians, most of the maxillary In addition, very few soft structures pass ramus in Chlamydoselachus is fused with the through the postorbital process of neosela- buccal ramus of the anterodorsal lateral line chians, and consequently any interpretation nerve (formerly considered part of the facial of their arrangement in Cladodoides or other nerve; see below), an arrangement which Paleozoic sharks lacks a reliable modern par- may be primitive for gnathostomes (North- adigm. It is nevertheless possible to make cutt and Bemis, 1993). The mandibular ra- comparisons that are at least congruent with mus in Chlamydoselachus runs outward the general arrangement of major nerves and across the anterodorsal margin and outer sur- vessels in the postorbital region of neosela- face of the superior maxillary levator muscle, chians, even though their relationship to the then across the mandibular adductor muscle. process may differ. The interpretation of It typically supplies the dorsal mucosa as nerves in Cladodoides presented earlier well as connective tissues of the hyoid arch agrees in most part with the arrangements and the ¯oor of the mouth and therefore ex- seen in neoselachians, but certain features tends much farther laterally than the other (especially the arrangement and topographic branches of the trigeminal nerve. According position of the major nerve ganglia) vary in to Allis (1923a: 212), `ìn the dorsal part of modern forms and do not provide a satisfac- its course it ®rst sends a branch posteriorly tory paradigm for interpreting fossils. Fur- beneath the postorbital process to the outer thermore, treating the octavolateral nerves as surface of the levator maxillae superioris''. 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 69

In Squalus, the mandibular ramus also ex- trematic ramus of a branchial nerve is spec- tends laterally in front of the postorbital pro- ulative) passes posteriorly at the mesial ven- cess (Norris and Hughes, 1920; Marinelli and tral border of the palatoquadrate. It then turns Strenger, 1959). ventrally, where it meets the spiracle and In many neoselachians, the trigeminal fo- runs along the dorsolateral surface of Meck- ramen lies either in the orbital opening of the el's cartilage. This subdivision of the palatine jugular canal (e.g., Oxynotus, Scymnodon)or ramus and the initial separation of the man- in front of it within the posterior part of the dibular and hyoidean rami are always extra- orbit (e.g., Etmopterus, Squatina, Rhinoba- cranial in neoselachians. tus). However, in some taxa the foramen lies The hyomandibular trunk leaves the brain- much farther posteriorly, behind the postor- case behind the postorbital process in many bital process (e.g., Carcharodon; Parker, neoselachians (e.g., Scymnodon, Oxynotus; 1887). Holmgren, 1941). Gardiner (1984a: 239) re- The facial nerve of gnathostomes includes garded this as a derived condition, possibly three principal rami after the sensory lateralis associated with the posterior position of the components are removed from consideration: hyomandibular fossa in elasmobranchs (in the palatine ramus (equivalent to the pharyn- contrast with osteichthyans, in which the fos- geal branch of a typical branchial nerve), the sa is located farther anteriorly). The mandib- mandibular ramus (equivalent to the pretre- ular ramus of this trunk often initially accom- matic branch?; note however that such ho- panies the maxillary ramus of the trigeminal. mologies with branchial nerves are specula- These may cross the ¯oor of the orbit to- tive), and the hyoid ramus (the posttrematic gether before the mandibular ramus angles equivalent?). The mandibular and hyoidean sharply backward (e.g., Laemargus), or the rami arise from the main (hyomandibular) mandibular ramus may separate sooner and trunk, from which the palatine ramus ®rst cross the posterior wall of the orbit to reach branches at the geniculate ganglion. The pal- the posterior angle of the mouth (e.g., Chla- atine ramus typically passes anteroventrally mydoselachus, Squalus; Daniel, 1934). to supply the dorsal oral epithelium, the in- Discussion of the trigeminal and facial ternal mandibular ramus supplies the internal nerves in Paleozoic sharks necessarily in- surface of the lower jaw, and the hyoidean volves consideration of the transverse pas- ramus supplies the branchial muscles of the sage through the postorbital process de- hyoid arch (Norris and Hughes, 1920). scribed earlier. Gross (1937) did not identify In Squalus and many other squaloids, the any opening from this passage within the or- palatine nerve arises intracranially from the bital surface of the process, but Romer hyomandibular trunk and then penetrates the (1964) found such an opening in Tamiobatis basal part of the lateral commissure. How- vetustus (which he interpreted as containing ever, in other neoselachians the palatine ra- the pretrematic or mandibular ramus of the mus often arises extracranially and there is facial nerve). This led Schaeffer (1981) to no separate palatine foramen (e.g., batoids, suggest that such an opening may also be Chlamydoselachus and hexanchiforms, present but not identi®ed in Cladodoides, Squatina, galeomorphs, and Heterodontus). which is now con®rmed by the CT scan. Holmgren (1941) observed that the palatine Schaeffer (1981) considered that the canal nerve is entirely extracranial in embryos of contained a laterally situated palatine ramus Squalus and Etmopterus, although it be- of the facial nerve in Tamiobatis, Orthacan- comes enclosed proximally by cartilage later thus, and Cladodoides. Coates and Sequeira in ontogeny. Upon emerging from the brain- (1998) proposed that the palatine ramus also case, the palatine ramus in Squalus divides penetrated the postorbital process in Akmon- into an anterior and a posterior ramule (Nor- istion; furthermore, they found evidence for ris and Hughes, 1920). Its anterior division two canals in one of their specimens and sug- passes anteriorly between the cranium and gested that at least one was big enough to the palatoquadrate, while the posterior have carried both nerves and blood vessels branch (sometimes termed the pretrematic ra- (perhaps including a branch of the orbital ar- mus, but again any equivalence to the pre- tery and a palatine ramule). Thus, several al- 70 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 ternative interpretations have been advanced not the main passage. It is therefore consid- regarding the function of this canal in Paleo- ered unlikely that the palatine nerve tra- zoic sharks. versed the postorbital process laterally in Romer's (1964) proposal that the mandib- Cladodoides or other Paleozoic sharks. ular ramus of the facial nerve passed through Coates and Sequeira (1998: 79) comment- the postorbital process in Tamiobatis vetustus ed that presence of distal foramina in the (or Cladodoides) is considered implausible. postorbital process is `à likely primitive In Chlamydoselachus, this ramus is directed chondrichthyan synapomorphy'', implying posterolaterally rather than laterally and ex- homology across a wide suite of taxa. In fact, tends far behind the postorbital process to if such foramina were for the trigeminal supply the medial surface of the lower jaw mandibular ramus or for the buccal-maxil- (as in gnathostomes generally; Luther, 1909; lary complex, their position may conceivably Allis, 1923a). Even in Squalus (which has re¯ect a more fundamental (e.g., gnathosto- short, transverse jaws), the facial mandibular me?) pattern of cranial nerves. However, the ramus is directed posterolaterally away from arrangement of the trigeminal nerve inferred the postorbital process (Norris and Hughes, here in Cladodoides apparently differs from 1920; Marinelli and Strenger, 1959). Indeed, that in Acanthodes, where the trigeminal if the mandibular ramus passed through the mandibular ramus lay behind the postorbital postorbital process as Romer (1964) pro- process and passed through the palatoquad- posed, it would emerge on the wrong (i.e., rate (Miles, 1973; see below). In Pucapam- lateral) surface of the mandibular arch. Ad- pella, there is no evidence of a canal in the ditionally, this ramus is normally associated lateral wall of the postorbital arcade as in with the spiracular opening behind the post- Cladodoides. It is possible that the peculiar orbital process (Dick, 1978). arrangement found in Cladodoides and Tam- Several criticisms can be leveled at iobatis vetustus is an apomorphic character, Schaeffer's (1981) proposal that the palatine but further analysis is necessary. ramus passed laterally through the postorbital The octavolateral nerves in gnathostomes process. First, in neoselachians the ramus (sensu Northcutt and Bemis, 1993) include passes ventrally toward the roof of the oral the octaval (acousticovestibular) nerve plus cavity and does not extend very far laterally. all the lateralis components classically con- Therefore, while its course may carry it sidered to be parts of the facial, glossopha- through the unchondri®ed ¯oor of the lateral ryngeal, and vagal nerves. Two classical la- commissure (e.g., Squalus, Etmopterus), the teralis components of the facial nerve (the ramus never enters the postorbital process super®cial ophthalmic and buccal branches) farther laterally. Second, the course of the form the anterodorsal lateral line nerve, palatine ramus in Cladodoides (including its whose main trunk usually accompanies the distal ramules) is clearly established and trigeminal through the prootic foramen. Dur- agrees essentially with the arrangement in ing ontogeny, the trigeminal and anterodorsal neoselachians. It penetrated the chondri®ed lateral line nerves frequently become sepa- ¯oor of the postorbital process medial to the rated from the facial nerve by the prefacial jugular canal and emerged beneath the brain- commissure (discussed below). The super®- case without passing laterally. Since this cial ophthalmic components of the antero- nerve supplies the roof of the oral cavity in dorsal lateral line and trigeminal nerves may gnathostomes, it is unlikely that any addi- combine, forming the super®cial ophthalmic tional palatine ramules passed farther later- complex. The buccal ramus is often fused ally through the postorbital process. Third, with the maxillary ramus of the trigeminal to the main facial (hyomandibular) foramen is form the buccal ϩ maxillary complex. In not aligned with the transverse canal in the Chlamydoselachus, the supraorbital sensory postorbital process. Finally, scanning shows canal is innervated by ramules of the super- that the supposed foramen for the palatine ®cial ophthalmic ramus passing through the nerve at the distal end of the postorbital pro- supraorbital cartilage, while the posterior part cess in Cladodoides shown by Schaeffer of the infraorbital canal is supplied by ra- (1981: ®g. 7C) is merely a minor branch and mules from the buccal ramus (some of which 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 71 pass posterodorsally through the roof of the have been traced in CT scans of Notorynchus postorbital process; Allis, 1923a: 197). The (Maisey, 2004b), where a small dorsal canal anteroventral lateral line nerve (classically, (probably for the middle lateral line nerve) the external mandibular ramus of the facial branches from the glossopharyngeal canal nerve) is closely associated (or even fasci- and passes dorsolaterally through the otic cularized) with the hyomandibular trunk, and capsule into the saccular region. Another these nerves may leave the braincase togeth- small canal (probably for the supratemporal er. lateral line nerve) branches from the vagal Additionally, there is an otic lateral line canal and passes through the posterior wall nerve (classically the otic ramus of the facial of the otic capsule, just behind the posterior nerve) supplying the posterior continuation semicircular canal. of the supraorbital canal and the spiracular TRIGEMINO-FACIAL RECESS: In craniates organ (where present) plus three postotic lat- generally, the trigeminal and facial ganglia eral line nerves (middle, supratemporal, and become enclosed by a chamber formed in the posterior). In Chlamydoselachus, several ra- membranous lateral wall of the cranium dur- mules of the otic lateral line nerve pass ing ontogeny. Allis (1928) envisioned that through the postorbital process and there is `ìf the mesial wall of this chamber were to a large canal opening onto the roof of the be broken down, the chamber would become braincase medial to the spiracle (Allis, a trigemino-facialis recess, while if this wall 1923a). In Squatina, there is large opening persisted and the lateral wall were broken in the orbital roof which was said to transmit down, the trigeminal and facialis ganglia the otic ramus by Gegenbaur (1872), but this would lie on the external surface of the prim- is mainly occupied by the orbital process of itive skull.'' Allis (1914, 1923a) had already the palatoquadrate (IselstoÈger, 1937). The described what he termed the acustico-trige- middle lateral line nerve accompanies the mino-facialis recess in Chlamydoselachus,an glossopharyngeal nerve, while the supratem- internal feature located immediately anterior poral and posterior lateral line nerves accom- to the otic capsule within the lateral wall of pany the vagal nerve. Among neoselachians, the cranial cavity and containing the roots of some (perhaps all) batoids lack any glosso- the trigeminal, facial, and octaval nerves. He pharyngeal lateralis component (Norris and also described the trigemino-pituitary fossa, Hughes, 1920), but in Squalus a minute an external depression in the posterior wall nerve arises from a small lateral line gangli- of the orbit which contains a foramen for the on at the root of the glossopharyngeal nerve pituitary vein and another for the trigeminal and passes through a short canal in the lateral and abducent nerves. The trigeminal gangli- wall of the otic capsule to the dorsal side of on lies in this fossa (which is closed exter- the head (the ramus supratemporalis IX of nally by connective tissue surrounding the Norris and Hughes, 1920). Also in Squalus, lateral head vein). The external rectus muscle the supratemporal nerve (their ramus supra- (supplied by the abducent nerve) also arises temporalis X) leaves the vagal canal and in this fossa, just behind the pedicel. Thus, passes through the posterior wall of the otic in Chlamydoselachus the trigemino-pituitary capsule, but the posterior nerve (their ramus fossa is an important landmark that contains dorsalis X) does not pass through any carti- the ganglia of the trigeminal and facial lage. Allis (1923a) did not describe the ar- nerves, the abducent nerve, the pituitary rangement of these nerves in Chlamydose- vein, and the origin of the external rectus lachus, but he noted that parts of the supra- muscle. temporal and anterior lateral line canals are In neoselachian embryos, the trigemino- innervated by branches extending from the pituitary fossa develops at the junction of the glossopharyngeal and vagal nerves. From his parachordal and the trabecular-polar cartilag- illustrations, it is likely that the middle and es, with the pituitary vein sandwiched be- supratemporal lateral line nerves pass tween them inside an extracranial space through cartilage forming the lateral and pos- formed at the posterior end of the polar car- terior walls of the otic capsule, as in Squalus. tilages (Allis, 1928; De Beer, 1931, 1937). The passages for two otic lateral line nerves The embryonic prootic foramen, acustico-tri- 72 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 gemino-facial recess, prefacial commissure, opterygians `ìs not very different from that and lateral commissure are all located im- of selachians such as Oxynotus, Squatina and mediately behind this region. This may also Squalus.'' The presence of a trigemino-facial have been the case in hybodonts and Ortha- chamber in some placoderms (e.g., Brinda- canthus, but not in Cladodoides (see above). bellaspis; Young, 1980) suggested to Gardi- It is concluded that presence of a trigemino- ner (1984a) that it represents a synapomor- pituitary fossa (sensu Allis, 1914) is not a phy of gnathostomes generally. However, in synapomorphy of elasmobranchs generally, arthrodires the facial nerve passes through but may be a synapomorphy of a group com- the anterior postorbital process and does not prising neoselachians and some extinct elas- lie within a trigemino-facial chamber as in mobranchs. osteichthyans and neoselachians. It is unclear In neoselachians, the medial wall of the whether this pattern is primitive for gnathos- otic capsule becomes secondarily fused to tomes or apomorphic for arthrodires. the parachordal plate in the vicinity of the In Pucapampella, a large space extends prefacial commissure (De Beer, 1937; Holm- posteriorly from a narrow prefacial commis- gren, 1940). The acustico-trigemino-facial sure, passing medial to the postorbital pro- recess (an internal space, sensu Allis, 1914) cess and the jugular canal. The space extends forms mainly from the medial capsular wall, posterior to the postorbital process and below although it is contiguous with the prefacial the anterior ampulla (Maisey, 2001c: ®g. commissure (which separates the facial nerve 16.1). I originally termed this space a trige- proper from the trigeminal and anterodorsal mino-facial recess, but it probably contained lateral line nerve trunks). This recess lies be- the profundal, trigeminal, abducent, and fa- hind the trigeminal and facial ganglia and cial nerves (including the hyomandibular also contains the octaval (acousticovestibu- trunk) and perhaps even the pituitary vein. In lar) nerve. In Cladodoides, the hyomandib- fact, the space clearly extends behind the ular and trigeminal foramina are separated by con®nes of the orbit (as de®ned by the post- the prefacial commissure, which connects the orbital process) and is almost as extensive as ¯oor of the anterior ampulla to the para- the embryonic prootic foramen of modern chordal cartilage and forms a wall behind the gnathostomes. The optic and prootic foram- trigeminal and facial ganglia (like the inner ina in Pucapampella are separated by a nar- face of the acustico-trigemino-facial recess in row bar of cartilage (corresponding to the neoselachians). In Chlamydoselachus and pila antotica, although it was originally mis- hexanchiforms, the hyomandibular trunk identi®ed as the prefacial commissure). The emerges just below and slightly behind the condition in Pucapampella supports the sug- postorbital process (the postorbital process gestion that cladistically primitive elasmo- only chondri®es above the lateral head vein branchs lacked a trigemino-facial chamber as and does not form a complete jugular canal; in neoselachians. Allis, 1923a; Maisey, 2004b). In many adult POSTORBITAL PROCESS: The morphology of squaloids, the facial (hyomandibular) fora- the postorbital process in most neoselachians men is located even farther posteriorly (on is considerably simpler than in Cladodoides. the lateral wall of the otic capsule; Holm- Even in neoselachians, however, the ontog- gren, 1941). In Heterodontus, the foramen eny of the process may involve tissues from lies in the posterior part of the orbit, and in different parts of the head skeleton. Accord- many galeomorphs the hyomandibular trunk ing to Holmgren (1940: 54, ®g. 55), the ®rst and trigeminal nerve emerge together (Good- part of the process to develop (in 38±39-mm rich, 1930; Holmgren, 1940, 1941). Squalus embryos) is a small blastemic ele- Schaeffer (1971) suggested that in actin- ment located just in front of the prootic fo- opterygians the term ``trigemino facialis ramen, which he termed the primary post- chamber'' should be restricted to the extra- orbital process. In later (48±49-mm) embry- mural cavity between the lateral cranial wall os, this element is connected to the blastemic and the region formed by the lateral com- supraorbital crest and these chondrify togeth- missure. Gardiner (1984a: 236) further con- er to form what he termed the secondary cluded that the chamber in primitive actin- postorbital process (in Scyliorhinus, however, 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 73 his ``primary'' postorbital process develops leading to chondri®cation of the lateral com- directly from the supraorbital blastema, blur- missure in primitive extinct sharks and in ring its distinction from the secondary pro- neoselachians such as Squatina and Centro- cess). At the same time, a laterally directed phorus is suppressed or absent in most neo- shelf is formed on the ventral part of the otic selachians. Chondri®cation patterns in the capsule below the main (hyomandibular) fa- neoselachian postorbital process thus include cial foramen and behind both the palatine ra- (1) lateral commissure is present and all parts mus and the orbital artery. In Squatina and of the postorbital process (including the pri- some squaloids, a connective tissue lamella mary process and the ventral shelf) are fully (which does not invariably become chondri- chondri®ed (e.g., Squatina, Centrophorus); ®ed) joins this shelf to the primary postor- (2) lateral commissure and ventral shelf are bital process, covering the hyomandibular present but unchondri®ed; the (primary) trunk and lateral head vein laterally. There is postorbital process is chondri®ed only above general consensus that this lamella is ho- the lateral head vein (e.g., Squalus, Etmop- mologous to the lateral commissure in os- terus, Rhinobatus); (3) lateral commissure teichthyans, which is also lateral to the jug- and ventral shelf are absent, (primary) post- ular canal and forms the sidewall of the tri- orbital process is chondri®ed only above the geminofacial chamber (De Beer, 1937; lateral head vein (e.g., Carcharhinus, Mus- Holmgren, 1940; Schaeffer, 1981; Gardiner, telus); and (4) postorbital process is absent, 1984a). with no evidence of a primary postorbital The extent to which the postorbital process process, ventral shelf, or lateral commissure becomes chondri®ed in neoselachians is (e.g., Heterodontus, Orectolobus, Raja, Tor- highly variable. There may be a well-devel- pedo, Discobatus). oped process behind the orbit, even in forms The postorbital process in Cladodoides is in which the lateral commissure is said to be presumed to include cartilage derived from absent (or at least does not persist into the the lateral commissure, primary postorbital adult; e.g., hexanchiforms, galeomorphs, process, and lateral otic shelf. There is no many batoids; Holmgren, 1941). Only rarely evidence to suggest that the lateral commis- is the lateral commissure fully chondri®ed, sure fused farther ventrally with the para- forming a complete arch enclosing the jug- chordal plate as in osteichthyans, and the car- ular canal and connected both dorsally and tilage behind the palatine ramus and below ventrally to the braincase (e.g., Squatina, the hyomandibular foramen in Cladodoides Centrophorus). In many neoselachians, there corresponds closely to the lateral shelf aris- is no postorbital cartilage lateral or ventral to ing from the otic capsule in Squalus as de- the lateral head vein, only above it. scribed by Holmgren (1940). The orbital ar- As described by Holmgren (1940), the tery and palatine ramus in Cladodoides (in- adult postorbital process in Squalus, Etmop- cluding its anterior and posterior ramules) terus, and Scyliorhinus consists only of the apparently passed through the lateral com- primary process and does not include the lat- missure and were subsequently enclosed in eral commissure, which remains membra- the ¯oor of the process medial and ventral to nous and unchondri®ed. In some galeo- the jugular canal as chondri®cation pro- morphs (e.g., Carcharhinus), a supraorbital gressed. cartilage is absent but a postorbital process The position of the postorbital process on is still present on the sidewall of the otic cap- the lateral surface of the braincase is highly sule (although the process is often poorly constrained during ontogeny. Nevertheless, chondri®ed). These observations suggest some variation is noted in the position of the that, while the primary postorbital process process relative to other features of the brain- (formed from blastemic tissue primitively as- case in modern and fossil elasmobranchs. sociated with the supraorbital shelf) is often The postorbital processes lies entirely ante- chondri®ed in neoselachians, the lateral com- rior to the labyrinth region in Cladodoides, missure and the ventral shelf arising from the in Orthacanthus and the ``Tamiobatis sp.'' otic capsule only rarely become chondri®ed. braincase described by Schaeffer (1981: ®gs. Presumably, the developmental mechanism 14, 22, 23), and in chimaeroids and Puca- 74 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 pampella (Maisey, 2001c). By contrast, in many neoselachians the postorbital process is lateral to the anterior ampulla, and in some carcharhiniforms the ampulla is positioned even farther anteriorly. The anterior ampulla also extends in front of the postorbital process in hybodonts (e.g., Tribodus, Hybodus). The anterior ampulla is situated between the anterior postorbital process in placoderms such as Kujdanowiaspis (StensioÈ, 1969). According to De Beer (1937) and Holm- gren (1940), the medial wall of the jugular Fig. 32. Diagrammatic representation of pos- canal in Squalus is formed by late chondri- tulated topographic differences in the dorsal at- ®cation of a connective tissue bridge be- tachment point of the lateral commissure in (A) tween the lateral capsular wall and the sub- Cladodoides and (B) a generalized hybodont or neoselachian. The positions of the trigeminal and orbital cartilage. The position of this bridge facial nerves relative to the otic capsule and eye relative to the labyrinth probably determines are identical, but they lie behind the orbit in panel the location and extent of the inner wall of A and within it in panel B. This probably re¯ects the jugular canal, and little topological vari- differences in the anteroposterior extent of the su- ation relative to the otic capsule and para- praorbital shelf (including the primary postorbital chordal plate is therefore possible. However, process, to which the lateral commissure is sec- the location at which the lateral commissure ondarily fused). In many neoselachians, the lateral is fused to the braincase dorsally (the pri- commissure is unchondri®ed or absent, but the mary postorbital process sensu Holmgren, primary postorbital process is commonly present. 1940) is determined by the posterior extent Anterior to left. No scale. of the supraorbital cartilage, which is potentially more variable. In Cladodoides, the en- and osteichthyans certainly develops in tire postorbital arcade is inclined obliquely in broadly similar fashion, with dorsal and ven- lateral view, with its dorsal part extending tral projections from the braincase connected farther anteriorly than its ventral part. A by the lateral commissure. However, accord- more posteriorly located dorsal attachment ing to De Beer (1937: 391) the embryonic would impart a more vertical orientation to derivations of the dorsal and ventral process- the arcade behind the trigeminal and facial es differ; the dorsal process in osteichthyans foramina, which would then lie either within is formed by the prootic process (a lateral the jugular canal or even within the orbit (®g. projection from the wall of the otic capsule) 32). Thus, the seemingly profound topo- rather than from an independent blastemic el- graphic differences noted earlier in the con- ement as in neoselachians; and the ventral ®guration of these nerves may result from (basitrabecular) process in osteichthyans differing orientations of the lateral commis- (and the postpalatine process farther ventral- sure during ontogeny, possibly resulting ly) is derived from the margins of the tra- from variation in the anteroposterior position beculae and parachordal plate, respectively, at which it is fused to the primary postorbital not from the ventral part of the otic capsule process arising from the supraorbital carti- as in neoselachians. Thus, while the lateral lage. Given the general morphological agree- commissure of neoselachians is probably ho- ment between the trigemino-facial chamber mologous to that of osteichthyans, the adult in modern elasmobranchs and early actin- postorbital process is not identical in these opterygians noted by Gardiner (1984a), the groups because its dorsal and ventral attach- postorbital position of the facial foramen, the ments to the lateral wall of the braincase in- inclination of the postorbital process, and the volve tissues of different embryonic origins. upturned jugular canal may be apomorphic The postorbital process in Acanthodes characters shared by Cladodoides, Tamiob- forms the lateral part of a large dorsal ossi- atis vetustus, and Akmonistion. ®cation and has been described in some de- The postorbital process in neoselachians tail (Watson, 1937; Miles, 1965, 1973; Jar- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 75

Fig. 33. Comparison of the postorbital process in anterior view: (A) Acanthodes (after Miles, 1973); (B) Cladodoides. See text for discussion. Not to scale. vik, 1977, 1980). Unfortunately, certain as- farther ventrally. In Acanthodes, the ventral pects of its morphology are controversial and margin of the postorbital process contains a a reinvestigation in Acanthodes is necessary wide embayment. According to Jarvik before some of the inconsistencies noted here (1977) this contained the trigeminal nerve, can be resolved. The postorbital process in but according to Miles (1973) it represents Cladodoides resembles that of Acanthodes part of a trigeminofacial chamber. There is although there are important diferences; for general agreement that another foramen lo- example, the postorbital process in Clado- cated farther medially and anteriorly is for doides probably contained the mandibular ra- the oculomotor nerve. Signi®cantly, no jug- mus of the trigeminal nerve (see above), ular canal was recognized either by Holm- whereas there is general agreement that in gren (1942) or Miles (1973). Acanthodes this ramus passed through the Apart from the postorbital articulation otic process of the palatoquadrate farther (discussed below), these features are virtually posteriorly. the only landmarks of the postorbital process The orbital surface of the process in Acan- that can be compared in Acanthodes and Cla- thodes contains a canal. Watson (1937) iden- dodoides. Simply from a topographic view- ti®ed it a jugular canal, Holmgren (1942: point, the ventral embayment in Acanthodes 137) regarded it as a canal for the otic lateral super®cially resembles the dorsal margin of line nerve (`òtical part of the lateral sensory the jugular canal in Cladodoides (®g. 33). line''), Miles (1973) suggested it may have Unfortunately, the course of the lateral head housed the middle cerebral vein, and Jarvik vein in Acanthodes has only been reliably (1977) considered that it contained a supra- determined farther posteriorly, where it lay orbital vein connected to the lateral head vein in a shallow groove on the lateral surface of 76 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 the otic region, dorsal to the glossopharyngeal foramen and angled toward the embayment (Miles, 1973: ®g. 2; Jarvik, 1980: ®g. 260). The ventral embayment in Acanthodes was identi®ed by Miles (1973) as the dorsal part of a trigemino-facial recess, but it could conceivably have contained the lateral head vein if this had the same relationship to the embryonic lateral commissure as in modern elasmobranchs (and as inferred in Cladodoi- des). In fact, since Miles (1973: ®g. 4) considered the lateral wall of the embayment to be formed from the lateral commissure, it is dif®cult to envisage where else the lateral head vein might have passed (except perhaps through the small canal farther dosolaterally, which Miles regarded as possibly for the middle head vein). Miles' (1973) interpretation implies that the lateral head vein lay much farther ventrally, and/or that the lateral Fig. 34. Braincase of the arthrodire Dickso- commissure extended much farther dorsally nosteus (illustrations derived from Goujet, 1984: (relative to the lateral head vein) than in ®gs. 6 and 26, and combined here). The left side modern gnathostomes. However, if either the shows the braincase in ventral view, and the right side shows the endocast in dorsal view. The an- large ventral embayment or the smaller fo- terior postorbital process bears the hyomandibular ramen farther dorsolaterally in Acanthodes fossa and surrounds the jugular canal and facial contained the jugular canal, the process nerve. No scale. would agree morphologically with that of modern gnathostomes in having the lateral head vein enclosed by the lateral commis- derived ontogenetically from the embryonic sure. The position of the embayment relative lateral commissure, like the postorbital pro- to the trochlear foramen is relatively unin- cess in crown-group gnathostomes (StensioÈ, formative, since it lies at the same horizontal 1963a). However, the arthrodire anterior level as the trigeminal foramen and the dor- postorbital process differs from the elasmo- sal margin of the jugular canal in Cladodoi- branch postorbital process in at least two im- des, and at the same level as the ventral em- portant respects: (1) the anterior postorbital bayment in Acanthodes. The ossi®cation sup- process bears a hyomandibular fossa, unlike porting the main otic condyle in Acanthodes in elasmobranchs; (2) the anterior postorbital is probably homologous to the sphenotic of process contained the hyomandibular trunk osteichthyans (Gardiner, 1984a). Only the of the facial nerve and is located far behind `àuxiliary otic condyle'' (Miles, 1973) is lo- the main trigeminal foramen (which opens cated on the postorbital process farther an- into the medial side of a jugular groove or teriorly. Its position provides circumstantial pit). The prefacial commissure is positioned support for reinterpreting the ventral embay- medial to the lateral commissure in elasmo- ment as the roof of a jugular canal because branchs, but is presumably posterior to it in it is apparently the only part of the postor- arthrodires. Thus, even if the placoderm an- bital articulation that conceivably formed on terior postorbital process is derived from the the lateral commissure. lateral commissure, its relationships to sur- In arthrodires, the anterior postorbital pro- rounding structures would have been differ- cess encloses the jugular canal and is located ent from those found in modern osteichth- on the lateral wall of the otic capsule, slightly yans and chondrichthyans. anterior to the anterior ampulla (e.g., Dick- Several investigators considered that the sonosteus; ®g. 34). The arthrodire anterior lateral commissure is of neurocranial origin postorbital process was therefore probably (Swinnerton, 1902; De Beer, 1937; Hammar- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 77 berg, 1937; Hubendick, 1943; Daget and in hexanchiforms apparently lies on the pri- d'Aubenton, 1957; Bertmar, 1959, Schaeffer, mary postorbital process rather than on the 1981), but others have suggested a visceral lateral commissure (Maisey, 2004b), and it is arch origin (e.g., in actinopterygians; Holm- therefore farther dorsal (relative to the lateral gren, 1943, and in sharks; Jollie, 1971; see head vein) than in Cladodoides, Tamiobatis, discussion in Gardiner, 1984a). There has or Orthacanthus (in this regard, it seems to been little modern developmental work on resemble the `àuxiliary otic condyle'' in this part of the braincase, but according to Acanthodes). Furthermore, the articular sur- Bertmar (1959) the lateral commissure in face in hexanchiforms is oriented posterov- Neoceratodus is derived from the hyoid arch entrally rather than posteriorly. These differ- (which suggests it may have formed in the ences suggest that the hexanchiform postor- hyoid/preotic stream of neural crest tissue). bital articulation has perhaps evolved inde- On the other hand, Holmgren (1940, 1943) pendently from the primitive elasmobranch considered that the elasmobranch lateral articulation. In this case, the presence of a commissure arose in the mandibular arch postorbital articulation in hexanchiforms is (which suggests that it originated in the tri- not an `àrchaic'' feature but may instead be geminal stream). If the lateral commissures a secondary specialization of the group (and in Neoceratodus and elasmobranchs arise in it may be absent in early hexanchiforms such different neural crest streams, they arguably as Notidanoides; Maisey, 2004b). A postor- fail Paterson's (1982) test of ontogenetic sim- bital articulation is absent in Chlamydosela- ilarity and could be considered nonhomolo- chus (the putative sister taxon of hexanchi- gous. Theoretically, putatively homologous forms in some cladistic analyses), further structures (e.g., lateral commissures, or tra- suggesting that the articulation may be a syn- beculae and visceral arches) could fail this apomorphy of crown-group hexanchiforms. test and still pass Patterson's (1982) other El-Toubi (1949) found evidence of an ex- tests of homology (congruence, conjunction), traotic cartilage in Squalus, plastered against if their developmental site shifted during the lateral capsular wall to enclose part of the evolution (heterotopy; Haeckel, 1875; Hall, otic lateral line nerve and also fused to the 1998; Kuritani et al., 2004). To some extent, postorbital process dorsally. He suggested heterotopy is tautological since it makes a that the extraotic cartilage is homologous to priori homology assumptions (presumably the palatoquadrate otic process in hexanchi- based on congruence and conjunction) about forms, and he claimed that it is lateral to both structures that display ontogenetic differenc- the hyomandibular trunk and the jugular vein es. On the other hand, ontogeny is the de- (although this is not evident from his illus- velopmental phylogeny of an organism, and trations). The cartilage in question appears related taxa have congruent early ontogenies ®rst as a mesenchymatous band continuous up to their points of divergence (von Baer's with tissue surrounding the palatoquadrate law, from the more similar to the less simi- farther ventrally. It then grows and chondri- lar), so heterotopy may represent one of the ®es dorsally while atrophying ventrally, fus- key processes involved in stepwise phylo- ing secondarily with the otic capsule and genetic transformations of ontogeny. trapping the otic lateral line ramus between POSTORBITAL ARTICULATION: The postor- them. It is unclear whether the extraotic car- bital articulation in Cladodoides is located tilage is fused with the primary postorbital entirely on cartilage presumably derived process or with the lateral commissure. Ho- from the embryonic lateral commissure. The mology between the extraotic cartilage and lateral head vein presumably passed medial part of the palatoquadrate is also doubtful, as or dorsomedial to this articulation. Such an Holmgren (1941: ®g. 2) found a cartilage- arrangement has no exact modern counter- enclosed canal for the otic lateral line nerve part, since a postorbital articulation is absent in Heptranchias, where the postorbital artic- in neoselachians with a chondri®ed lateral ulation is present. commissure, but is present in hexanchiforms A postorbital articulation is typically ab- where the commissure is said to be absent sent in osteichthyans. One is present in Acan- (Holmgren, 1941). Thus, the articular surface thodes, but it apparently differs from that of 78 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 sharks in being located on the lateral wall of tion of suitable fossils (e.g., the recent dis- the sphenotic (Miles, 1965, 1973; Gardiner, covery of paired foramina for the lateral dor- 1984b). By contrast, in sharks the postorbital sal aortae in Cladoselache; Williams, 1998); articulation is located on cartilage presum- (2) inability to see whether particular foram- ably either formed in the lateral commissure ina are connected in fossils; and (3) disagree- (e.g., Cladodoides) or on the primary post- ment over the identity of canals and foramina orbital process (e.g., modern hexanchiforms, (e.g., for the palatine nerve, efferent hy- which lack the lateral commissure in the oidean artery). adult; Holmgren, 1941; Maisey, 2004b). Al- Williams (1998) presented an analysis of though Miles (1973: ®gs. 4, 5) identi®ed the the basicranial arteries in several Paleozoic lateral commissure in Acanthodes as lying sharks in which foramina were simply iden- medial rather than lateral to the postorbital ti®ed numerically rather than by implied ho- articulation, such an arrangement is anatom- mologies (although any interpretation im- ically improbable if the articular surface is in plies homology at some level). He drew at- fact located on the sphenotic (see above). In tention to controversies surrounding previous Acanthodes, there are two paired postorbital interpretations and outlined what he termed articular surfaces. One pair is located on the the `èxtra foramen'' problem in these early ventral part of the ossi®ed postorbital pro- sharks (although much of this ``problem'' cess, and the second pair is slightly posterior seems to be a product of differing opinions on the lateral wall of the otic region (Miles, rather than extra openings). The problem was 1968, 1973; Jarvik, 1977). The articulation ®rst articulated by Schaeffer (1981: 36), who therefore extends over a much wider region suggested that the internal carotid and orbital in Acanthodes than in sharks, possibly en- arteries in Tamiobatis vetustus were origi- compassing part of the lateral commissure as nally enclosed completely by cartilage and well as the sphenotic. that the efferent hyoidean emerged from the The mandibular ramus of the trigeminal posteriormost of the paired basicranial fo- nerve in Acanthodes passed behind the artic- ramina between the postorbital processes. ulation and penetrated the otic process of the Nevertheless, he admitted that an important palatoquadrate in order to reach the mandib- dif®culty for his interpretation is that ``there ular joint (Miles, 1965, 1971; Jarvik, 1977, appears to be an extra foramen . . . between 1980). However, in sharks this ramus either the palatine foramen and the one for the or- passes anterior to the postorbital process bital artery on the rim of the suborbital (hexanchiforms) or through it (in Cladodoi- shelf''. Williams (1998) noted that two of the des; see above). There is no evidence that the most controversial issues were (1) the sepa- mandibular ramus penetrates the palatoquad- ration of the orbital and internal carotid ar- rate in any modern or extinct elasmobranchs. teries, and (2) the separation of the efferent The postorbital articulation is therefore an- hyoidean artery. To these can now be added terior to the mandibular ramus in Acanthodes a third issue, that is, the position(s) of the but posterior to it in elasmobranchs, and it is facial nerve palatine ramus and its anterior concluded that the postorbital articulation in and posterior ramules. Gross (1937) and Acanthodes differs from that of Paleozoic Schaeffer (1981) reached very different con- sharks and modern hexanchiforms in its ex- clusions regarding the identity of foramina tent, complexity, and relationship to the tri- for the orbital artery and the palatine branch- geminal mandibular ramus (irrespective of es of the facial nerve in Cladodoides, while whether this articulation is homologous in all Williams (1998) has presented a novel inter- elasmobranchs). pretation of the basicranial circuit in Tam- THE `ÈXTRA FORAMEN'' PROBLEM: Inter- iobatis vetustus which differs radically from pretation of arterial vessels in Paleozoic that of Schaeffer (1981). Much of the dis- sharks is always problematic. Perennial dif- agreement in naming basicranial foramina in ®culties (some of which are now potentially Paleozoic sharks seems to involve misinter- overcome by scanning technology) include: pretation of the passages for the facial (1) incomplete or inaccurate data, stemming nerve's palatine ramules. Earlier interpreta- from inadequate preservation and prepara- tions are beset with assumptions about the 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 79

Fig. 35. Tamiobatis vetustus NMNH 1717. Ventral view of braincase, depicting four alternative scenarios of basicranial arteries discussed in the text. Note disagreements over the foramen for the orbital artery/anterior palatine ramus, site of efferent pseudobranchial foramen, and the extent to which internal carotid and orbital arteries were enclosed by cartilage. The preferred arrangement is similar to that reconstructed in Cladodoides (®g. 21). No scale. extent of missing cartilage and the supposed er channels around passages for nerves and directions of hidden passages. blood vessels. Romer (1964) recognized that Three different interpretations of the ba- the paired lateral aortae (``common carot- sicranial circuit in Tamiobatis vetustus have ids'') were largely enclosed by cartilage. He been presented by Romer (1964), Schaeffer also suggested that the internal carotids (1981), and Williams (1998). These are de- emerged from the braincase before passing picted diagrammatically, together with a nov- toward the bucco-hypophyseal fenestra. Far- el fourth interpretation, in ®gure 35. Romer ther laterally, he identi®ed an opening for (1964) did not illustrate the circuit, and the ``the hyomandibular (pseudobranchial) ar- interpretation shown here is therefore based tery'' (i.e., the efferent hyoidean) and con- only on his description. These hypotheses sidered that the orbital artery lay in a deep differ principally in the interpretation of what groove, leading forward and outward to the Williams (1998) termed ``foramen two'' and orbit. ``foramen three''. Unfortunately, the ventral Williams (1998) addressed the `èxtra fo- surface of NMNH 1717 is damaged and ramen'' problem in Tamiobatis vetustus and many of its features are lost apart from deep- offered an alternative interpretation in which 80 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 the efferent hyoidean and ®rst efferent bran- 1964) is not feasible in Cladodoides, and in chial arteries branched from the lateral aorta my view the corresponding foramen in Tam- within the ¯oor of the braincase and emerged iobatis vetustus (and many other Paleozoic via separate foramina. The lateral aorta then sharks, including ``Cladodus'' hassiacus and supposedly emerged from the cartilage be- Cladoselache) probably also contained the fore branching into the internal carotid and posterior ramule of the palatine ramus. Sten- orbital arteries, which then reentered the car- sioÈ (1937) identi®ed ``posterior'' and ``pala- tilage farther anteriorly. Two criticisms may tine'' passages in the Humboldt University be leveled at his reconstruction: ®rst, it is braincase, suggesting that here too the ante- based on a different specimen, which while rior and posterior ramules of the palatine ra- well preserved may not represent Tamiobatis mus both passed through the basicranium. vetustus; and second, his reconstruction also An alternative basicranial reconstruction requires several assumptions about hidden in Tamiobatis vetustus is therefore presented passages and the `èxtra'' foramen. His inter- here, with (as in Cladodoides) separate an- pretation of CMNH 9280 has been super- terior and posterior ramules rather than a sin- imposed on NMNH 1717, but this is not al- gle palatine ramus (explaining the `èxtra'' together satisfactory because the orbital ar- foramen in T. vetustus). In addition, contrary tery would have to enter the braincase at pre- to popular opinion, no trace of a canal or cisely the point it seems to leave in Romer's groove for the orbital artery is preserved in (1964) version, and the supposed efferent NMNH 1717. The supposed groove for the branchial and efferent hyoidean foramina in orbital artery is very deep and probably rep- CMNH 9280 could not have housed these resents the roof of a canal for the anterior vessels in NMNH 1717. The arrangement of palatine ramule, whose ¯oor has probably basicranial openings in CMNH 9280 differs been stripped away by erosion. Crucially, slightly from NMNH 1717, and the arterial this passage is directed posterodorsally to- circuit may not be identical, but examination ward the foramen for the posterior palatine of CMNH 9280 nevertheless suggests that ramule. The orbital artery in T. vetustus the supposed efferent hyoidean foramen ac- probably lay below the anterior palatine ra- tually housed the posterior ramule of the pal- mule, and both entered the orbit one above atine ramus and that the anterior ramule the other, as in Cladodoides. It is also pos- joined the orbital artery in a notch at the sible that the orbital artery was enclosed by point where the suborbital shelf meets the a canal for at least part of its length (possibly postorbital process. a growth-related feature as in the Humboldt The branching pattern of the palatine ra- University braincase). mus passage in Cladodoides is crucial to un- Some differences are noted in the arrange- derstanding the extra foramen problem in ment of basicranial foramina in NMNH 1717 Tamiobatis vetustus and other Paleozoic and the specimen referred to Tamiobatis ve- sharks. The position of the orbital artery fo- tustus by Williams (1998: CMNH 9280), ramen in Cladodoides is very similar to that suggesting there was some variation in the proposed by Williams (1998: ``foramen courses of the orbital and efferent hyoidean one'') in T. vetustus.InCladodoides, the an- artery and of the palatine ramus of the facial terior palatine ramule and orbital artery con- nerve, although it is uncertain whether this verged just as the former emerged from the is because different taxa are represented. braincase in front of the postorbital process, Some aspects of the basicranial pattern pos- so their passages share a common exit al- tulated by Williams (1998) in CMNH 9280 though they remained separate (the canal may have been incorrectly identi®ed, or other shows that one lay above the other). In Cla- factors (e.g., ontogenetic or preservational dodoides, the anterior and posterior ramules differences) may be involved. of the palatine ramus probably occupied fo- Akmonistion has been interpreted with a ramen one and foramen two, respectively, of separate efferent hyoidean foramen, suggest- Williams (1998). His interpretation that fo- ing that the artery arose from the lateral aorta ramen two housed the efferent hyoidean (as within the basicranium (Coates and Sequeira, suggested in Tamiobatis vetustus by Romer, 1998). However, there is no evidence of such 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 81 an arrangement in the ``Cobelodus'' brain- apomorphic or plesiomorphic character with- case (Maisey, 2004a and in prep.). In chi- in elasmobranchs). maeroids, the branchial arteries all meet the OTICO-OCCIPITAL PROPORTIONS: Romer aortic circuit below the braincase (De Beer (1964) and Coates and Sequeira (1998) sug- and Moy-Thomas, 1935), but in Pucapam- gested that a comparatively long otico-occip- pella an efferent hyoidean or branchial artery ital region is probably a primitive gnathos- may have branched from the lateral aorta tome condition, but they disagreed as to within the parachordal plate (Janvier and whether the elongate otic region in Tamio- SuaÂrez-Riglos, 1986; Maisey, 2001c). Re- batis vetustus is primitive (Romer, 1964) or constructions of sharks such as Tamiobatis a reversal (Coates and Sequeira, 1998). A vetustus showing an efferent hyoidean fora- relatively short otico-occipital region com- men are probably erroneous (e.g., Romer, monly occurs in actinopterygians, Pucapam- 1964; Schaeffer, 1981; Williams, 1998). pella, chimaeroids, neoselachians, and hy- REMARKS ON THE DORSAL VEIN SINUS IN bodonts, lending circumstantial support to ACANTHODES:InAcanthodes, the anterior the second alternative. The apparent length ventral ossi®cation (generally considered ho- of the otico-occipital region is affected by the mologous to the basisphenoid in osteichth- position of the postorbital process, although yans; Miles, 1973; Gardiner, 1984a; Janvier, other factors are also involved, including (1) 1996) contains a median basal fenestra in the the anteroposterior extent of the semicircular center of a shallow hypophyseal fossa. Miles canals, (2) the angle at which the anterior and (1973) and Jarvik (1977) agreed that the in- posterior canals diverge, and (3) the relative ternal carotids converged on the basal fenes- extent to which the occipital arch intrudes tra ventrally, then entered the ossi®cation and between the capsules. Schaeffer (1981: ®g. emerged on its dorsal surface. However, they 16) attempted to illustrate this by superim- differed in their interpretations of the posi- posing the endocasts of Squalus and Ortha- tion of the efferent pseudobranchial. Accord- canthus, but it is not possible to characterize these differences in a straightforward man- ing to Jarvik (1977), the internal carotids met ner. The otico-occipital proportions in Ortha- the efferent pseudobranchial (and ophthalmic canthus, Tamiobatis vetustus, and ``Tamiob- arteries) above the trabecular-polar plate in atis sp.'' are nevertheless very similar and elasmobranch fashion, but Miles (1973) Schaeffer (1981) suggested that this may rep- identi®ed a groove for the efferent pseudob- resent a synapomorphy of these taxa. In all ranchial on the ventral surface of the basi- these forms, the otico-occipital region is lon- sphenoid, suggesting that it met the internal ger (with respect to orbit length) than in Cla- carotid ventral to the trabeculae, as in os- dodoides. The hypotic lamina was also long teichthyans. in Cladodus elegans, and its otic region may Jarvik's (1977) interpretation of features therefore have been proportioned as in Tam- on the ventral surface of the basisphenoid iobatis vetustus and ``Tamiobatis sp.'' The was strongly in¯uenced by his proposal that posterior extremity of the Humboldt Univer- a neoselachian-like dorsal median vein sinus sity braincase is missing and its proportions was present in Acanthodes. However, his in- cannot be determined. terpretation is equivocal and assumes that REMARKS ON THE INNER EAR IN CHON- paired basicranial grooves originally housed DRICHTHYANS,PLACODERMS, AND AGNATHANS: buccopharyngeal veins. There is no direct In craniates generally, the semicircular canals evidence for the presence of a dorsal median initially develop within the thickness of the vein sinus in Cladodoides or any other ex- capsular walls, whereas the major vestibular tinct shark and it is therefore impossible to chambers (utriculus, sacculus, lagena) lie in- determine the distribution of the sinus outside the capsule. This represents a funda- side modern neoselachians. Even if a dorsal mental morphological and developmental vein sinus were present in Acanthodes, its characteristic of the labyrinthine and vestib- value as a potential synapomorphy of Acan- ular parts of the inner ear. The ampullae (and thodes and elasmobranchs would remain du- the utricular recess, where present) represent bious (irrespective of its signi®cance as an major points of entry for the canals into the 82 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 main vestibular chambers. Sensory cristae crown-group gnathostomes and could there- are primitively located at or near these points fore be regarded as a plesiomorphic feature of entry and the octaval nerve typically does at this level. Nevertheless, the medial cap- not extend beyond the ampullae inside the sular wall is fully ossi®ed in many placod- capsular wall. The external (horizontal) erms (®g. 36I), osteostracans, and galeaspids, semicircular canal is found only in gnathos- suggesting somewhat paradoxically that the tomes (including placoderms) and its absence wall is primitively present in stem gnathos- in cyclostomes (and all ostracoderms where tomes. Presence of a chondri®ed or ossi®ed the inner ear is known) presumably repre- medial capsular wall therefore has a disjunct sents the primitive craniate condition. distribution among modern and extinct cra- In neoselachians, the medial capsular wall niates and it may have arisen more than once is fully chondri®ed (and usually well calci- in craniate history, but its presence in phy- ®ed) apart from the octaval and glossopha- logenetically advanced elasmobranchs (neo- ryngeal foramina. However, its development selachians, hybodonts) is unique among is completed fairly late in ontogeny (De Beer, crown-group gnathostomes. 1931; Holmgren, 1940; Jollie, 1971). The In neoselachians, the medial capsular wall medial capsular wall is also chondri®ed in forms mainly in the taenia medialis, which Mesozoic hybodonts (e.g., Hybodus, Tribod- arises from the synotic tectum and extends us) although its extent in earlier hybodonts down between the brain and labyrinth to fuse such as Hamiltonichthys is unknown. Prob- with upgrowths from the parachordal plate ably the medial wall was partly or even com- (e.g., Scyliorhinus; De Beer, 1931). Thus, ab- pletely chondri®ed in Tristychius, since Dick sence of a medial capsular wall in extinct (1978: ®gs. 2, 4) showed extensive chondri- sharks may perhaps be more accurately de- ®cation on each side of the ``median supraot- ®ned as absence of chondri®cation in the tae- ic fossa'' extending into the braincase medial nia medialis. As discussed earlier, the inter- to the capsules. Chondri®cation of the medial nal wall of the acustico-trigemino-facial re- capsular wall effectively isolates the skeletal cess in Cladodoides is part of the prefacial labyrinth from the main cranial cavity, so commissure, not the taenia medialis. This their endocasts are essentially separate. Dig- may also be the case in Cobelodus aculeatus, itally generated endocasts of the skeletal lab- where the ganglia of the octaval, trigeminal, yrinth in the modern broadnose sevengill shark Notorynchus and the Cretaceous hy- and facial nerves all lie within a shallow re- bodont Tribodus are illustrated in ®gure cess located in a vertical septum separating 36A±F. the anterior part of the vestibular chamber By contrast, most of the medial capsular from the main cranial cavity (Zangerl and wall is unchondri®ed in Cladodoides, apart Case, 1976: ®g. 9B). Transverse sections of from the region immediately adjacent to the the braincase in Orthacanthus (Schaeffer, octaval nerve (®g. 36G). It could be argued 1981: ®g. 9) suggest very limited chondri®- that lack of chondri®cation in this particular cation in the ¯oor of the sacculus and the braincase is a juvenile characteristic, espe- lateral walls of the posterior dorsal fonta- cially since chondri®cation is also incom- nelle. There is no evidence of a continuous plete elsewhere. However, the medial cap- cartilaginous taenia medialis extending from sular wall is also unchondri®ed or weakly the fossa to the basicranium, but traces of chondri®ed in other Paleozoic sharks (e.g., calci®ed cartilage between the brain and lab- the Pennsylvanian braincase referred to yrinth cavities suggest that the membranous ``Cobelodus''; ®g. 36H, and in Cobelodus taenia in Orthacanthus was as extensive as aculeatus,``Tamiobatis sp.'', Orthacanthus; in neoselachians. In the ``Tamiobatis sp.'' Zangerl and Case, 1976; Schaeffer, 1981), as braincase (AMNH 2140, unfortunately stud- well as in Pucapampella (Maisey, 2001c; ied only in horizontal sections which are dif- Maisey and Anderson, 2001), in modern hol- ®cult to compare with the transverse slices ocephalans, and in osteichthyans (®g. 37A). of Orthacanthus), the synotic tectum is chon- Absence of a chondri®ed medial capsular dri®ed and forms the anterior margin of the wall is therefore a widespread feature among posterior dorsal fontanelle, but there is little 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 83

Fig. 36. Endocasts of the left skeletal labyrinth in: (A±C) modern Notorynchus, lateral (A), medial (B), and dorsal (C) views (after Maisey, 2004); (D±F) a Cretaceous hybodont, Tribodus, lateral (D), medial(E), and dorsal (F) views (original); (G) Cladodoides, lateral view; (H) a Pennsylvanian symmoriid, ``Cobelodus'', lateral view; (I) the arthrodire Kujdanowiaspis, dorsal view (after StensioÈ, 1963a). Not to scale. 84 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Gardiner, 1984a: ®g. 26, fhm). The trigeminal nerve and anterodorsal lateral line trunk left the cranial cavity farther dorsally; in Mimia these had separate foramina, but in Cladodoides they apparently shared a single opening (V, adll, ®g. 7; Gardiner, 1984a: ®g. 26, VII lat, V). In both Mimia and Clado- doides, the septum is notched for the utricular recess posteroventrally (®g. 37). In Mim- ia, the septum and zygal plates help de®ne the dorsal and ventral margins of a wide fenestra leading to the utricular recess, through which the octaval nerve must have passed. Although the ventral part of this fenestra is poorly de®ned in Cladodoides, it is nevertheless possible to envisage a comparable arrangement to Mimia since its octaval nerve apparently entered the recess in a corresponding position (this can be seen in CT scan section 161; ®g. 18E). In Cladodoides, therefore, this entire region corresponds to the chondri®ed acustico-trigemino-facial recess described in modern sharks by Allis (1914, 1923a). However, in neoselachians this septum forms a more extensive medial Fig. 37. Otico-occipital regions in medial capsular wall, almost completely separating view as if sliced sagitally, in (A) an early actin- the utricular recess from the metencephalic opterygian, Mimia (after Gardiner, 1984a); (B) chamber. Such an arrangement probably rep- Cladodoides. Note general similarity in vestibular resents a derived condition within crown- chamber morphology, including presence of a group gnathostomes, whereas the less exten- septum incompletely separating the utricular recess and medullary chamber dorsally. Not to sive septum in Cladodoides and Mimia prob- scale. ably represents a conserved gnathostome character. The view of the endocranial wall in Cob- evidence of a cartilaginous taenia medialis elodus aculeatus depicted by Zangerl and deeper within the braincase. Case (1976: ®g. 9B) also shows a septum There is a rather wide communication be- (presumably between the utricular and met- tween the cranial and labyrinth cavities in encephalic chambers and supposedly con- Cladodoides (®gs. 7, 21A, D), the primitive taining the trigeminal, abducent, facial, and actinopterygian Mimia (Gardiner, 1984a: ®g. octaval nerves). However, in the ``Cobelo- 26), and Eusthenopteron (Jarvik, 1980), al- dus'' braincase, the septum is chondri®ed though in all these forms parts of the laby- only dorsally and does not extend to the fa- rinth are enclosed by the walls of the otic cial foramen (in prep.). Without going into region. There is also an extensive saccular details, the arrangement depicted by Zangerl recess, but in Mimia this is partly separated and Case is probably incorrect and an alter- from the cranial cavity ventrally and slightly native interpretation will be presented else- posterior to the level of the utricular recess where. (Gardiner, 1984a: 227). In both Cladodoides The neoselachian posterior semicircular and Mimia, the utricular recess and meten- canal displays several unique or unusual fea- cephalic chamber are partly separated by a tures among extant craniates (Maisey, 2001b; septum, at the base of which the hyomandib- see also ®g. 36A±C here), including: ular ramus of the facial nerve probably left the cranial cavity (®gs. 7, 35, 37; see also 1. The ampullary and nonampullary ends of the 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 85

posterior semicircular canal are connected by chondrichthyans (e.g., Cladodoides, Pucapam- an almost vertical preampullary canal (ϭ pella; Maisey, 2001c), lampreys, and many os- ``posterior oblique semicircular canal'' of Dan- teostracans (Janvier, 1981a, 1981b, 1985). The iel, 1934), which extends anteriorly and up- hag®sh labyrinth includes a single canal con- ward from the posterior ampulla. Consequent- taining both anterior and posterior sensory ly, the posterior ampulla no longer represents cristae, suggesting that it represents the com- the point at which the semicircular canal enters bined anterior and posterior canals (Maisey, the vestibular chamber. A short preampullary 2001b). A crus commune is supposedly absent canal is also present in some other crown- in placoderms (StensioÈ, 1950, 1963a, 1963b, group gnathostomes including chimaeroids and 1969). Acipenser (Retzius, 1881), but it never extends 6. The neoselachian anterior and external ampul- very far dorsally and does not form a complete lae are said to share a single utriculo-saccular loop as in neoselachians. A preampullary canal connection with the utricular recess, instead of is well developed in some placoderms (e.g., having separate openings as in other gnathos- Kujdanowiaspis, Tapinosteus; StensioÈ, 1963a) tomes (Torrey, 1962). However, it is uncertain but not in others (e.g., Brindabellaspis, Dick- whether this feature characterizes all or only sonosteus; Young, 1980; Goujet, 1984). some neoselachians. 2. In neoselachians, the posterior semicircular canal is contained almost completely in the chon- StensioÈ (1950, 1963a) found several sim- dri®ed medial capsular wall instead of lying ilarities in the inner ear of arthrodires and partly within the vestibular chamber (as in chi- neoselachians, including their widely sepa- maeroids and osteichthyans). A remarkably rated anterior and posterior semicircular ca- similar arrangement occurs in placoderms and nals, enclosure of the posterior canal by the ostracoderms (e.g., Mimetaspis [Cephalaspis], medial capsular wall, and presence of an as- Kiaeraspis; StensioÈ, 1927), where the medial cending preampullary canal (e.g., Kujdanow- capsular wall is also fully ossi®ed apart from iaspis, Tapinosteus). However, several im- the glossopharyngeal and octavolateral canals. portant differences can also be identi®ed in 3. The neoselachian posterior semicircular canal has only a single connection with the vestibu- the semicircular canal arrangement in arthro- lar chamber via the posterior canal duct asso- dires and neoselachians, including: ciated with the macula (crista) neglecta (Cor- 1. In some arthrodires there is a crus between the win, 1989). This duct is sometimes quite long external and anterior semicircular canals (as in and separates the canal from the remainder of neoselachians), but the external canal is also the labyrinth (e.g., batoids; Bleckman and connected medially to the posterior one by a Hoffman, 1999). In osteichthyans, chima- narrow anteriorly directed commissure (e.g., eroids, osteostracans, and galeaspids the pos- Kujdanowiaspis; ®g. 36I). Young (1986: 41) terior semicircular canal enters the sacculus concluded that the dorsomedial ``ridge-like ex- ventrally and meets the anterior semicircular pansion of the sacculus in Kujdanowiaspis and canal dorsally, but in some arthrodires the pos- Tapinosteus is more suggestive of a semicir- terior canal may describe a full circle as in cular canal extension, [and] it could . . . have neoselachians. contained an anterior extension of the posterior 4. A short section of the neoselachian posterior canal''. If that is correct, the posterior canal semicircular canal is exposed in the lateral wall was not isolated as in neoselachians and in- of the parietal fossa and may even bulge into stead continued forward to meet the anterior the perilymphatic fenestra. Perilymphatic fe- one. The arthrodire dorsomedial canal is not nestrae are absent in modern osteichthyans, homologous to the neoselachian posterior ca- chimaeroids, and cyclostomes and have not nal duct, since it does not open into the upper been identi®ed in placoderms or ostracoderms. part of the vestibular chamber (``posterior According to Dick (1978), a posterior opening utriculus''). Thus, even if the posterior semi- in the dorsal fontanelle of Tristychius may circular canal describes a complete circle in have contained perilymphatic fenestrae, but the both neoselachians and arthrodires, its connec- opening in question is median rather than tion to the rest of the labyrinth is different. paired and may simply represent an unminer- 2. The super®cial position of the external semi- alized area of the fontanelle ¯oor. circular canal around the dorsomedial part of 5. The neoselachian anterior and posterior semi- the capsule in arthrodires (e.g., Kujdanowias- circular canals do not meet at a crus commune pis, Tapinosteus; StensioÈ, 1963a) contrasts dorsally. By contrast, a crus commune is pre- with its much deeper position in crown-group sent in chimaeroids and osteichthyans, extinct gnathostomes, where the external semicircular 86 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

canal turns inside the loop of the posterior ca- lary canal is absent, the medial capsular wall nal, and its nonampullary end cannot be traced is membranous, the anterior and external am- deeper in the endocast. In modern osteichth- pullae meet the utricular recess separately, yans and chimaeroids (Retzius, 1881), as well the posterior canal meets the sacculus ven- as in Cladodoides,``Cobelodus'', and Puca- trally and meets the anterior canal at a crus pampella, the external semicircular canal meets the utriculus below the crus commune commune, and perilymphatic fenestrae are (e.g., ®g. 36G, H). In arthrodires, therefore, the absent. Importantly, no apomorphic features medial end of the external semicircular canal of the inner ear have been identi®ed that are is positioned more super®cially (relative to the shared only by placoderms and cladistically rest of the labyrinth) than in crown-group gna- primitive chondrichthyans, while the few un- thostomes, and it can be traced for its entire usual placoderm features (e.g., the super®cial length ®g. 36I). In many osteostracans and gal- position of the dorsomedial canal, anterior easpids, the anterior and posterior semicircular preampullary canal) are absent in early elas- canals are again super®cial and can be traced mobranchs. It is concluded that those simi- from their ampullae all the way to the crus larities noted in the past between the laby- commune and saccular chamber (e.g., Cephal- rinth of arthrodires and neoselachians include aspis hoeli, Norselaspis, Duyunolepis, Xiu- only conserved craniate and gnathostome shuiaspis, Benneviaspis,``Boreaspis''; StensioÈ, 1927; P'an and Wang, 1978; Janvier, 1981a, characters, plus some convergent features 1981b, 1984, 1985; Wang, 1991). that are absent in more primitive chondri- 3. No perilymphatic fenestrae have been found in chthyans. Arthrodire inner ear morphology placoderms. These fenestrae form an essential provides no convincing morphological evi- part of the phonoreceptor system in neosela- dence of a phylogenetic relationship with chians and so far have not been identi®ed in elasmobranchs or chondrichthyans, and other craniates (Maisey, 2001a). while it may eventually provide informative 4. In some arthrodires, the anterior semicircular characters for resolving arthrodire or placod- canal meets the utricular recess indirectly via erm phylogeny, it does not help resolve their a short anterior preampullary canal (e.g., Ku- wider relationships other than to suggest that jdanowiaspis, Tapinosteus). In other arthrodi- arthrodires are plesiomorphic stem gnathos- res the anterior preampullary canal is absent (e.g., Brindabellaspis, Dicksonosteus). No tomes. such canal has been found in extant or fossil POSTERIOR DORSAL FONTANELLE AND POS- chondrichthyans. Presence of this canal in ar- TERIOR TECTUM: The dorsal fontanelle in ex- throdires may have phylogenetic importance. tinct sharks is commonly identi®ed as a parietal or endolymphatic fossa, like that of The following scenario is suggested by modern elasmobranchs. However, the neo- these observations: the anterior and posterior selachian parietal fossa forms part of a highly semicircular canals were primitively situated specialized, low-frequency, semidirectional entirely within the walls of the otic capsule phonoreceptor system that also involves in early craniates (a pattern that is conserved many other structures in the otic region (Cor- in lampreys, osteostracans, and galeaspids); win, 1989; Maisey, 2001b). The fossa is the external semicircular canal of gnathos- ®lled with a sound-transmitting gel and com- tomes subsequently arose in a similarly su- municates directly with the posterior semi- per®cial position (a pattern that is conserved circular canal (which is almost completely in arthrodires); the medial part of the external isolated from the remainder of the labyrinth; semicircular canal became located deeper in see below) via paired perilymphatic fenes- the labyrinth in the common ancestor of trae. These are actually unchondri®ed re- crown-group gnathostomes (a pattern that is gions in the medial capsular wall through conserved in osteichthyans, chimaeroids, and which the posterior semicircular canal pro- early sharks); and decoupling of the posterior trudes, forming a tympanic membrane that semicircular canal occurred late in elasmo- transmits low-frequency soundwaves to the branch history (a pattern that is conserved in crista neglecta (De Beer, 1931, 1937; Holm- neoselachians and hybodonts). gren, 1940; Corwin, 1989). Short perilym- In Cladodoides, osteichthyans, and chi- phatic ducts may extend from the fenestrae maeroids, an ascending posterior preampul- and are sometimes evident in the endocast of 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 87 the skeletal labyrinth (®g. 36B, C, E, F). coderms; a posterior dorsal fontanelle was There is no evidence of perilymphatic fenes- connected to a persistent otico-occipital ®s- trae or ducts in Cladodoides. A low-frequen- sure without an intervening posterior tectum cy phonoreceptor system was probably ab- in more crownward gnathostomes (a pattern sent in many Paleozoic sharks, and these that was conserved in Acanthodes, osteichth- forms probably relied on lateral line recep- yans, and early chondrichthyans such as Pu- tors to detect pressure waves (as in most os- capampella and Orthacanthus); the posterior teichthyans). dorsal fontanelle and the otico-occipital ®s- Maisey and Anderson (2001) noted that sure became ontogenetically closed indepen- the fontanelle in Pucapampella closely re- dently in several osteichthyan lineages and in sembles the osteichthyan posterior dorsal chimaeroids; the posterior dorsal fontanelle fontanelle (see Gardiner, 1984a) and sug- became separated from the otico-occipital gested that presence of the posterior dorsal ®ssure by a posterior tectum in elasmo- fontanelle may be a synapomorphy of crown- branchs (a pattern that is conserved in Cla- group gnathostomes. No comparable fonta- dodoides and Tamiobatis vetustus); and the nelle has been identi®ed in placoderms. In posterior dorsal fontanelle became conscript- Acanthodes, there is a median ``anterior dor- ed into a low-frequency phonoreceptor sys- sal fontanelle'' between the postorbital pro- tem, becoming the parietal fossa, and the oti- cesses (i.e., farther anterior than the posterior co-occipital ®ssure became closed in ontog- dorsal fontanelle), that is supposedly related eny (e.g., neoselachians, hybodonts). to the pineal organ (Miles, 1973: ®g. 3). A The ontogenetic origin of the posterior tec- posterior dorsal fontanelle may be represent- tum is unclear. Holmgren (1940: ®g. 122) il- ed by a slight widening of the dorsal part of lustrated the 62-mm stage of Heterodontus in the otico-occipital ®ssure (Miles, 1973: ®g. dorsal aspect, in which the otico-occipital ®s- 3; Jarvik, 1980: ®g. 258). However, Jarvik's sure separates the occipital pila from the otic (1977, 1980) reconstruction showing a neo- capsules and there are large paired fenestrae selachian-like endolymphatic fossa separated (presumably representing the combined en- from the otico-occipital ®ssure by a wide tec- dolymphatic and perilymphatic openings tum in Acanthodes is entirely speculative and within the parietal fossa). The posterior part seems to have resulted from over-reliance of of the fossa is still connected to the otico- morphological data obtained from Recent occipital ®ssure even though a medial tectum rather than from early elasmobranchs. The connects the synotic and occipital tectum parietal fossa is a specialized structure and is posteriorly. The medial tectum forms the known with certainty only in neoselachians ¯oor of the parietal fossa and is eventually and hybodonts, suggesting that the posterior connected to the otic capsules. It is possible dorsal fontanelle probably became second- that the posterior tectum in Cladodoides rep- arily conscripted as part of the phonoreceptor resents a medial tectum which has expanded system comparatively late in elasmobranch laterally to connect the otic capsules in front evolution (Maisey and Anderson, 2001). En- of the occipital ®ssure. Diagrammatic repre- dolymphatic ducts are present in chimaeroids sentations of these conditions are shown in (as in gnathostomes generally), but chima- ®gure 38. eroid inner ear morphology lacks numerous LATERAL OTIC PROCESS: The lateral otic specialized features found in neoselachians. process represents an extension of the pos- Consequently, there is no reason to suppose terolateral capsular wall and contains the that chimaeroids possessed a parietal fossa at posterior semicircular canal (Schaeffer, any stage in their history, although they may 1981). This process is well developed in have primitively possessed a posterior dorsal many Paleozoic sharks, including Orthacan- fontanelle. thus, Tamiobatis vetustus,``Tamiobatis sp.'', From these observations, it is suggested and Akmonistion. It was probably present in that: a posterior dorsal fontanelle and otico- Cladodoides although its extent is unknown occipital ®ssure were primitively absent in (see above). Part of the lateral otic process is craniates, and that this pattern was conserved preserved next to the occipital arch in one of by some early gnathostomes including pla- the specimens referred to Cladodus elegans. 88 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

lateral capsular wall, and both the process and the hyomandibular fossa are separated from the hypotic lamina by the ventral otic notch. Based on these topographic and ontogenetic differences, the postotic process in neoselachians is not considered homologous to the lateral otic process in Paleozoic sharks. The supposed lateral otic process in Hy- bodus resembles the postotic process in neoselachians in lying dorsal to the hyomandibular fossa, but it does not contain the glossopharyngeal nerve or the posterior semicircular canal (Maisey, 1983). On that basis, it probably does not represent either a lateral Fig. 38. Diagrammatic representations of the otic process or a postotic process; instead, it posterior dorsal fontanelle, otic capsules, synotic merely seems to be an outgrowth of the lat- tectum, and occipital arch in various elasmo- eral capsular wall which supports a dermal branchs. Dorsal views, anterior to top: (A) Pu- cephalic spine in males. capampella or Tamiobatis; posterior tectum ab- In some neoselachians, the hyomandibular sent, fontanelle con¯uent with otico-occipital ®s- fossa may be sandwiched between the pos- sure, occipital arch complete; (B) Cladodoides; totic process dorsally and a rather featureless posterior tectum present, fontanelle separated vestibular process ventrally (Gadow, 1888). from ®ssure, occipital arch complete; (C) 62-mm A vestibular process has not been identi®ed Heterodontus; medial tectum extends from synotic tectum and meets complete occipital arch pri- in Cladodoides or any other Paleozoic shark. or to its fusion with otic capsules (D) 48-mm The vestibular process may contribute to the Squalus; occipital pilae fused to otic capsules pri- ventral margin of the hyomandibular fossa or to closure of occipital arch dorsally. Panels C (e.g., the ``Basalrande des Hyoidgelenkes'' in and D are based on ®gures in Holmgren (1940). Squatina; IselstoÈger, 1937), but it does not Not to scale. contain the posterior semicircular canal or the glossopharyngeal nerve. To summarize, several different kinds of The process may also form part of the hyo- process may be developed on the lateral or mandibular fossa. posterolateral wall of the otic capsule in elas- It has been claimed that a lateral otic pro- mobranchs, but the lateral otic process cess is present in neoselachians (Schaeffer, uniquely encloses part of the posterior semi- 1981) and in Hybodus (Maisey, 1983), while circular canal. This process has a more re- Coates and Sequeira (1998: 79) regarded it stricted distribution than previously supposed as ``almost ubiquitous'' among chondri- and may help de®ne a monophyletic group chthyans. However, according to the above within elasmobranchs. de®nition there is no lateral otic process in HYOMANDIBULAR FOSSA: In modern batoids neoselachians or hybodonts. Instead, in neo- and orbitostylic sharks, the hyomandibular selachians there is a postotic process (sensu fossa is located on the posterior part of the Holmgren, 1941) that lies dorsal to the hyo- otic capsule, immediately in front of the mandibular fossa and contains the glosso- glossopharyngeal foramen and below the pharyngeal nerve but not the posterior semi- ridge for the external semicircular canal. The circular canal. A postotic process only occurs exact position of the fossa is uncertain in in elasmobranchs with a glossopharyngeal Cladodoides (see above), but in Hybodus, canal and is absent in forms with a persistent Tribodus, Orthacanthus, Tamiobatis vetus- metotic ®ssure (e.g., Orthacanthus, Tamiob- tus,``Tamiobatis sp.'', and Akmonistion it is atis). Its ontogeny in neoselachians involves located on the posterior part of the capsular fusion of the ventral part of the capsular wall wall. However, in modern Heterodontus, gal- and the hypotic lamina. By contrast, the lat- eomorphs, and in the extinct neoselachian eral otic process is formed entirely on the Synechodus (also perhaps a galeomorph; 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 89

Maisey, 1985; Maisey et al., 2004), the fossa or adjacent to the inferred position of the lat- is farther anterior, on the side wall of the otic eral commissure. Gardiner (1984a) consid- capsule. In Tristychius, the hyomandibular ered this to be the primitive position for gna- fossa is directly behind the postorbital pro- thostomes generally and that its posterior po- cess (Dick, 1978), and Falcatus supposedly sition in elasmobranchs is derived. has the articular fossa on the postorbital pro- A classical distinction between osteichth- cess (Lund, 1985). StensioÈ (1937) identi®ed yans and neoselachians involves the position the hyomandibular fossa in the Humboldt of the hyomandibula relative to the lateral University braincase in an anterior position head vein; the articulation is ventromedial to on the otic capsule, immediately posterior to the vein in elasmobranchs and lateral to it in the facial foramen. However, it seems more osteichthyans (De Beer, 1937). Although its likely that this area corresponds to part of the phylogenetic signi®cance is unclear, this dif- lateral otic fossa in the Senckenberg brain- ference is a highly conserved trait in both case and that the hyomandibular fossa prob- groups. Gardiner (1984a: 241) argued that ably lay much farther posteriorly in the Hum- this difference is not signi®cant on the boldt University specimen. The arrangement grounds that the fossa lay above the vein in in Tristychius is remarkable since the sus- some extinct sharks such as Hybodus and pensorial arrangement restored by Dick Tristychius.InHybodus and Tribodus, how- (1978: ®g. 14) is clearly present in several ever, the lateral head vein probably passed articulated specimens and differs profoundly above the hyomandibula, as in neoselachians. from modern elasmobranchs. OCCIPITAL ARCH,OTICO-OCCIPITAL FIS- Zangerl (in Zangerl and Williams, 1975; SURE, AND HYPOTIC LAMINA: The occipital Zangerl and Case, 1976) interpreted the sym- pila in gnathostomes arises during ontogeny moriid shark Cobelodus aculeatus as aphe- behind the otic capsules, from which it is ini- tohyoidean (sensu Watson, 1937) and pro- tially separated by the otico-occipital ®ssure posed that the hyomandibula was nonsuspen- (e.g., Scyliorhinus; De Beer, 1931, 1937: sory in primitive sharks. However, Williams Squalus; Holmgren, 1940; El-Toubi, 1949). (1985) noted that in Denaea meccaensis De Beer (1931) found that the glossopharyn- (which he considered to be another symmo- geal nerve in Scyliorhinus lies within the me- riid), the proximal end of the hyomandibula totic ®ssure (a transient embryonic space be- almost certainly articulated with the postero- tween the ¯oor of the otic capsule and the lateral part of the otic capsule. Subsequently, hypotic lamina; ®g. 29). He demonstrated Zangerl (1990: 119 et seq.) described the hy- that ontogenetic obliteration of the ®ssure omandibula in Stethacanthulus (a small, leaves only the glossopharyngeal canal in somewhat unusual stethacanthid) as the adult neoselachians. This closure may occur ``standard symmoriid type'', with a ``proxi- earlier in some taxa than in others; for ex- mal joint surface . . . attached to the lateral ample, there are indications of a persistent otic (epiotic) process on the neurocranium'', ®ssure in the 62-mm Heterodontus ®gured and he characterized stethacanthids and sym- by Holmgren (1940: ®g. 122). In all adult moriids as having the hyomandibula `àrtic- neoselachians, the occipital arch is fused to ulating at the otic process of the neurocra- the otic region. An embryonic otico-occipital nium.'' ®ssure was presumably also present in hy- Aside from the few controversial ®ndings bodonts (e.g., Hybodus, Tribodus), where the discussed above, the hyomandibular fossa in adult occipital arch is continuous with the fossil and modern elasmobranchs is usually otic region as in neoselachians. located on the posterolateral wall of the otic In many neoselachians, the occipital arch capsule. It is also in this position in the prim- is wedged almost completely between the itive chondrichthyan Pucapampella (Maisey otic capsules (e.g., Chlamydoselachus, Hep- and Anderson, 2001). The position of the tranchias, Squalus; Allis, 1923a; Holmgren, fossa differs in osteichthyans, placoderms, 1941; Schaeffer, 1981), but in others it may and Acanthodes (Miles, 1964, 1973; Jarvik, extend behind the otic region (e.g., Notoryn- 1977, 1980; Schaeffer, 1981; Gardiner, chus; Daniel, 1934; Maisey, 2004b). Where 1984a; Goujet, 1984), but it usually lies on the occipital arch is short, all the spino-oc- 90 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 cipital nerves may be con®ned to the glos- also see ®g. 38). The ®ssure is never com- sopharyngeal canal, but where it is longer the pletely ``lost'' in elasmobranchs, since it is more posterior nerves tend to emerge onto represented at least transiently in the embryo, the lateral surface of the occipital region, al- although its extent is often reduced following though the anterior ones may still enter the ontogenetic fusion of the otic capsule to the glossopharyngeal canal. occipital pila dorsally and to the hypotic lam- Schaeffer (1981) considered that persis- ina ventrally. tence of an otico-occipital ®ssure is an apo- The otico-occipital ®ssure is similar in ear- morphic character of adult Orthacanthus and ly elasmobranchs and osteichthyans, sug- Tamiobatis. However, the ®ssure has subse- gesting that it is a conserved feature of quently been identi®ed in several other Pa- crown-group gnathostomes. Furthermore, in leozoic chondrichthyans, including Clado- both groups it may be connected with a per- doides, Akmonistion, Cladodus elegans, sistent ventral otic ®ssure ventrolaterally, ``Cobelodus'', and Pucapampella (Coates separating the parachordal plate from the rest and Sequeira, 1998, 2001a, 2001b; Maisey, of the basicranium (for osteichthyans see 2001a, 2001b, 2004a; Maisey and Anderson, Rayner, 1951; Nielsen, 1942; Patterson, 2001). 1975; Maisey, 1999; Basden et al., 2000; The otico-occipital ®ssure in Orthacanthus Basden and Young, 2001: for chondri- consists of two distinct but interconnected re- chthyans see Maisey, 2001b; Maisey and An- gions, one located dorsally, between the oc- derson, 2001). Nielsen (1942) and Patterson cipital arch and otic capsules (regarded here (1975) recognized that the otico-occipital and as the otico-occipital ®ssure proper), and the ventral otic ®ssures of osteichthyans differ other lying below the capsule and above the from each other developmentally, morpho- hypotic lamina (the metotic ®ssure). The ex- logically, and evolutionarily, and this may tent of the metotic ®ssure seems to be gov- also be true for chondrichthyans (Maisey, erned by the degree to which the hypotic 2001b; Maisey and Anderson, 2001). lamina becomes secondarily fused to the Young (1986) suggested that the glosso- ¯oor of the otic capsule during ontogeny. In pharyngeal nerve in gnathostomes is primi- Orthacanthus, the metotic ®ssure is exten- tively located beneath the otic capsule (a sive and forms a deep ventral otic notch view ®rst advocated by De Beer, 1937) and (Schaeffer, 1981: ®gs. 5, 6), whereas in Cla- that its passage through the capsule is a de- dodoides, the metotic ®ssure is mostly closed rived feature of arthrodires and osteichthyans and there is no ventral otic notch (®gs. 5, 6, (most parsimoniously regarded as an inde- 8B, 21, 27), although in both taxa the dorsal pendently developed feature instead of a syn- part of the ®ssure is persistent. In neosela- apomorphy). The glossopharyngeal nerve chians and hybodonts, part of the metotic ®s- also ran through the capsule in osteostracans, sure can persist as a glossopharyngeal canal, although this is probably because all the vis- following closure of the dorsal part of the ceromotor cranial nerves had a strongly an- ®ssure (De Beer, 1937; Schaeffer, 1981). terolateral orientation resulting from the con- Moreover, in many neoselachians, the occip- siderable forward shifting of the gill com- ital pila is typically fused to the otic capsule partments relative to the brain (Janvier, before the arch is complete dorsally (Holm- 1981b, 1985). Presence of a hypotic lamina gren, 1940; El-Toubi, 1949), whereas in ex- might also represent a primitive gnathostome tinct sharks such as Cladodoides the ®ssure character, since Young (1986: 42) observed remained open dorsally even after the me- that in many placoderms (e.g., acanthocora- totic ®ssure became closed and the occipital cids, rhenanids, and petalichthyids) the para- arch was complete. However, in 62-mm Het- chordal plate is expanded beneath the otic erodontus, a complete occipital arch is pre- capsule, effectively forming a hypotic lamina sent and a ®ssure continues to divide the pila beneath the glossopharyngeal nerve as in from the otic capsule laterally, although the elasmobranchs. However, he also noted that arch is connected to the synotic tectum dor- in arthrodires the glossopharyngeal nerve ap- sally, via the medial tectum forming the ¯oor parently passed through the labyrinth cavity of the parietal fossa (Holmgren, 1940: 174; (e.g., Buchanosteus; Young, 1979). Con- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 91 versely, Schaeffer (1981) concluded that the may be recognized in the gnathostome otico- hypotic lamina is a derived feature of elas- occipital ®ssure: (1) hypotic lamina absent: mobranchs and implied that its ontogenetic (a) dorsal and metotic parts of the ®ssure fusion to the capsular ¯oor (to form the glos- open, including a posterior dorsal fontanelle sopharyngeal canal) is a further derived fea- (some early osteichthyans and perhaps Acan- ture of neoselachians (see also Maisey, thodes); (b) dorsal and metotic parts of the 1984a, 2001b; Coates and Sequeira, 1998; ®ssure completely closed; no vestibular fon- Maisey and Anderson, 2001). This con¯icts tanelle (chimaeroids, some osteichthyans, ar- with Young's (1986) contention that the glos- throdires); (c) ®ssure closed apart from the sopharyngeal nerve is primitively enclosed vestibular fontanelle (some osteichthyans). by a canal in gnathostomes, but not with De (2) hypotic lamina present; (a) dorsal and Beer's (1937) more general observation re- metotic parts of the ®ssure open, medial tec- garding the position of the nerve relative to tum absent (Pucapampella, Orthacanthus); the otic capsule. An enclosed glossopharyn- (b) metotic part of the ®ssure closed apart geal canal may therefore have been acquired from glossopharyngeal canal; dorsal part independently by arthrodires and elasmo- open, medial tectum present (Cladodoides, branchs. ``Cobelodus'', Tamiobatis vetustus); (c) ®s- The hypotic lamina (ϭ ``lateral pharyn- sure completely closed except for glossopha- gohyal plate'' of Jollie, 1971) in neoselachi- ryngeal canal and parietal fossa (neoselachians is a lateral extension of the basal (para- ans, Hybodus, Tribodus); (d) ®ssure com- chordal) plate that secondarily fuses with the pletely closed except for glossopharyngeal otic capsule at the posterior basicapsular canal; no posterior dorsal fontanelle (placod- commissure. In osteichthyans and chima- erms). eroids, a hypotic lamina is absent and the In modern chondrichthyans and osteichth- glossopharyngeal nerve typically leaves the yans, the cartilage containing the posterior braincase ventrally rather than ventrolateral- semicircular canal and its ampulla also forms ly. Gardiner (1984a: 204) considered that the the outer wall of the otico-occipital ®ssure. By ventral otic notch in Orthacanthus and Tam- contrast, there is no occipital segment in mod- iobatis vetustus is homologous to the vestib- ern agnathans, and the otic capsule de®nes the ular fontanelle of osteichthyans. However, posterior limit of the braincase (Janvier, the osteichthyan vestibular fontanelle repre- 1996). Thus, the otico-occipital ®ssure in sents only one part of the primitive gnathostome metotic ®ssure and corresponds to the crown-group gnathostomes effectively marks embryonic basicapsular fenestra (which may the posterior boundary of the primitive cra- remain open even after the rest of the metotic niate braincase (in lampreys, the glossopha- ®ssure is closed by the posterior basicapsular ryngeal and vagal cranial nerves lie behind commissure; De Beer, 1937). In osteichth- the otic capsule). However, in both placod- yans, the vestibular fontanelle is situated an- erms and osteostracans, the glossopharyngeal, terior or lateral to the glossopharyngeal fo- vagal, and anteriormost spino-occipital nerves ramen (Gardiner, 1984a), but the metotic ®s- are enclosed by the occipital region. In pla- sure in elasmobranchs extends farther later- coderms, the occipital region is often very ex- ally because the parachordal plate is tensive and may include numerous spino-oc- broadened to form the hypotic lamina. Since cipital nerves (e.g., Buchanosteus). The brain- the metotic ®ssure remains completely open case of placoderms and osteostracans there- in Orthacanthus and ``Tamiobatis sp.'', a fore extends farther posteriorly than in posterior basicapsular commissure was pre- modern cyclostomes, although there is no ev- sumably absent in these forms. In Cladodoi- idence of an otico-occipital ®ssure as in des, Hybodus, Tribodus, and neoselachians, crown-group gnathostomes. If osteostracans fusion of the hypotic lamina to the capsular and placoderms are successive sister groups ¯oor by the commissure does not leave a per- to crown-group gnathostomes (as suggested sistent vestibular fontanelle like that of os- by modern phylogenetic hypotheses; e.g., Jan- teichthyans. vier, 1996), incorporation of the occipital re- To summarize, several different conditions gion into the braincase is most parsimoniously 92 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 interpreted as a primitive conserved feature of vik (1977, 1980), the paired canals contained crown-group gnathostomes. metameric arteries arising from the dorsal In craniate embryos, the notochord usually aorta. In either case, the ventral occipital os- extends beyond the occipital region and is si®cation in Acanthodes presumably formed ¯anked by the parachordal cartilages, al- in the parachordal plate and the notochord though some variation is noted in the relative terminated at or behind the ventral otic ®s- positions of the parachordals and notochord sure. in gnathostomes (De Beer, 1937: 379). The WHAT KIND OF SHARK WAS CLADODOIDES?: notochord may be closely associated with the Schaeffer (1981) drew attention to similari- parachordals throughout their length (e.g., ties between the Cladodoides braincase and urodeles, anurans, mammals) or it may di- three specimens tentatively referred to Cten- verge anteriorly to leave a basicranial fenes- acanthus from the Cleveland Shale, and he tra (e.g., neoselachians, osteichthyans, non- also noted their association with ``clado- mammalian amniotes). In modern cyclo- dont'' teeth. At present, there are four known stomes, chondrichthyans, sturgeons, and or suspected associations of Cladodus-like lung®shes the notochord is usually persistent teeth with Tamiobatis-like braincases: (1) into the adult. In other gnathostomes, it may Cladodoides wildungensis from Wildungen become reduced to the posterior part of the (Jaekel, 1921; Gross, 1937, 1938); (2) Cten- parachordal plate (e.g., Polypterus), partly acanthus (compressus?) from the Cleveland chondri®ed by notochordal cartilage (e.g., Shale (Dean, 1909; Schaeffer, 1981); (3) anurans, urodeles) or even obliterated (e.g., ``Tamiobatis vetustus'' from the Cleveland mammals). The parachordals are largely sep- Shale (Williams, 1998, although not demon- arated by the notochord in Amia and teleosts, strably referable to Tamiobatis); and (4) Cla- but in neoselachians, Polypterus, and lung- dodus elegans from the Lower Carboniferous ®shes, the parachordals fuse above and be- of Scotland (in prep.). low the notochord, completely enclosing it. Ctenacanthus-like ®nspines (in the re- The parachordals fuse below the notochord stricted sense outlined by Maisey, 1981) are in chimaeroids, Lepisosteus, urodeles, and associated with both taxa from the Cleveland lacertilians, leaving the notochord dorsal to Shale (Dean, 1909; Williams, 1998). Clado- the basal plate. In anurans, however, the re- dont teeth are also present in ``Ctenacan- verse occurs and the notochord lies below the thus'' costellatus (Lower Carboniferous of basal plate. The notochord is enclosed by Scotland; Traquair, 1884), although its brain- cartilage apparently of parachordal deriva- case is unknown and its ®nspine ornament tion in placoderms (e.g., Kujdanowiaspis; differs from that of Ctenacanthus. Cladodus StensioÈ, 1963a: ®g. 29), primitive actinopter- elegans may have possessed large ®nspines ygians (e.g., Mimia; Gardiner, 1984a: ®g. like those of Ctenacanthus major, but at pre- 26), and primitive chondrichthyans (e.g., Pu- sent this cannot be con®rmed. There is no capampella; Maisey and Anderson, 2001), proven association between the holotype of suggesting that it represents a primitive gna- Cladodoides wildungensis (jaw cartilages as- thostome arrangement. sociated with teeth) and the Senckenberg According to Miles (1973), in Acanthodes braincase. It is also unknown whether this the ventral occipital ossi®cation (also pre- taxon possessed ®nspines, and the only re- sumably of parachordal derivation) enclosed ported ®nspine from Wildungen (Ctenacan- the notochord in a canal, and the dorsal aorta thus jaekeli Gross, 1933) is not typical of the passed beneath the bone. However, Jarvik genus and was subsequently referred to (1977, 1980) considered that the aorta lay in- Homacanthus (thought to be an acanthodian; side the canal (implying that the notochord Denison, 1979); unfortunately, this specimen was positioned farther dorsally). Miles' is now lost (Maisey, 1984c). Paleozoic (1973) interpretation is compromised by the sharks with a Tamiobatis-like braincase, Cla- presence of paired lateral canals arising from dodus teeth, and Ctenacanthus ®nspines may the central canal (said to house occipital ar- be characterized loosely as a ``ctenacanth'' teries), suggesting that it contained a vessel gestalt, which is also represented by numer- rather than the notochord. According to Jar- ous disarticulated fossils (probably including 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 93 much of the material classically referred to itive feature for gnathostomes generally and Ctenacanthus, Tamiobatis, and Cladodus). noted a similar arrangement in many placod- However, it is uncertain whether this gestalt erms and early rhipidistians. He concluded represents a monophyletic group of elasmo- that there has been a parallel evolutionary branchs or a paraphyletic assemblage of trend toward relative reduction of the otico- phalacanthous sharks that also includes occipital region in ``higher'' ®shes and tet- ``higher'' forms such as hybodonts and neo- rapods. selachians. Schaeffer (1981) reached an opposite con- Do the Senckenberg and the Humboldt clusion and proposed that the elongate otico- University braincases represent a single tax- occipital region in the Paleozoic sharks he on? This question cannot be adequately ad- investigated represents a synapomorphy of a dressed until the Humboldt Museum speci- monophyletic group of early elasmobranchs men is thoroughly reinvestigated. At present, that also supposedly included Ctenacanthus all that can be said is that all features in the (although ``ctenacanths'' have not been char- ventral part of the orbit agree closely and that acterized by any unique features, so their many features of the basicranium also agree monophyly has not been demonstrated, and (with the proviso that the two individuals Ctenacanthus itself is founded on isolated ®n may have different ontogenetic ages). The spines). Coates and Sequeira (1998) were un- major difference concerns labyrinth mor- able to retrieve Schaeffer's (1981) grouping phology, since its arrangement in the Hum- in their cladistic analysis of 23 cranial char- boldt University braincase as described by acters, although they found a large mono- StensioÈ (1937) is completely at odds with ob- phyletic clade comprising C. wildungensis, servations of the scanned Senckenberg ex- Tamiobatis, Xenacanthus, Cladoselache, Hy- ample. bodus and Tristychius (presumably this clade PHYLOGENETIC PATTERNS IN PALEOZOIC would also include crown-group elasmo- SHARKS: When Gross (1937) ®rst described branchs although these were not speci®cally the Cladodoides braincase, elasmobranch part of their analysis). phylogeny was still very much in its infancy. Schaeffer (1981) united Orthacanthus and (It was the ®rst Paleozoic shark braincase Tamiobatis on the basis of a single suppos- ever described, and would remain the only edly derived character (presence of an otico- one for almost three decades.) He left its sys- occipital ®ssure). The ®ssure is now known tematic position unresolved beyond a general in several additional Paleozoic elasmo- statement that it was more closely related to branchs, including Akmonistion, Cladodoi- modern elasmobranchs than to holocephalans des, Pucapampella, and ``Cobelodus'' or acanthodians. Holmgren (1941: 79) made (Coates and Sequeira, 1998, 2001a; Maisey, far more speci®c proposals and considered 2001c, 2004a) and is clearly more wide- that modern Chlamydoselachus may be `à spread than Schaeffer (1981) suspected, al- descendant of a Cladodus-like stock''. Such though Coates and Sequeira (1998) still a view is incompatible with modern phylo- maintained that it could be a derived char- genetic analyses in which neoselachians in- acter in elasmobranchs generally. However, cluding Chlamydoselachus are resolved as a this seems unlikely since an otico-occipital monophyletic group that is only distantly re- ®ssure is present in modern Polypterus, gars, lated to most Paleozoic sharks (Shirai, 1982; and Amia (®lled by cartilage; Patterson Maisey, 1984a, 1984b; Maisey et al., 2004). 1975), and in primitive actinopterygians Romer (1964: 104) suggested that ``Cla- (e.g., Pteronisculus, Kentuckia, Mimia, Moy- dodus'', Tamiobatis, and xenacanths all pos- thomasia, Ligulalepis; Nielsen, 1942; Ray- sessed `à truly primitive elasmobranch type ner, 1951; Gardiner, 1984a; Basden and of braincase, characteristic of ancestral shark Young, 2001), primitive sarcopterygians types in the Devonian'', which was distin- (e.g., Eusthenopteron, Diabolepis, Youngo- guished by having a relatively long otico-oc- lepis; Jarvik, 1980; Chang, 1995), and Acan- cipital region. He proposed that the elongate thodes (Watson, 1937; Miles, 1965, 1973; otico-occipital region of forms such as Tam- Jarvik, 1977; Gardiner, 1984a). The ®ssure is iobatis and Orthacanthus represented a prim- closed in modern chondrichthyans, Latimer- 94 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288 ia, dipnoans, and tetrapods (Gardiner, 1984a; Orthacanthus could be an apomorphy of Forey, 1998), and it is absent in all placod- some (or even all) xenacanths. Alternatively, erms, osteostracans, and galeaspids where if the glossopharyngeal canal is a primitive the braincase is known. An otico-occipital gnathostome character as Young (1986) sug- ®ssure therefore seems to be restricted to gested, a persistent metotic ®ssure may be a crown-group gnathostomes (Maisey, 2001c), synapomorphy of some elasmobranchs (al- and its presence is clearly apomorphic though this seems doubtful in view of the among craniates generally, whether it repre- situation in Pucapampella, which is appar- sents a synapomorphy of crown-group gna- ently a cladistically primitive chondri- thostomes or appeared independently in sev- chthyan; Maisey, 2001b; Maisey and Ander- eral gnathostome lineages. son, 2001). Schaeffer (1981) united Cladodoides, In Coates and Sequeira's (1998) phyloge- Tamiobatis, and Orthacanthus by one char- netic analysis, symmoriids and stethacanthids acter (otico-occipital region equal to or lon- (represented by Cobelodus aculeatus and Ak- ger than ethmo-orbital region). Unfortunate- monistion zangerli) consistently fall outside ly, the length of the ethmo-orbital region can- a group comprising xenacanths, Tamiobatis, not be accurately determined in Cladodoides, Cladodoides, Hybodus, and Tristychius. but its otic region is certainly longer than the However, no cranial synapomorphies were orbit (although less so than in Tamiobatis identi®ed for Cobelodus and Akmonistion, and Orthacanthus). In Coates and Sequeira's and these taxa were not resolved as a mono- (1998) analysis, this character helped support phyletic group. Coates and Sequeira (1998) a node including all these taxa plus Hybodus nevertheless recognized a large monophylet- and Tristychius, but it was reversed at the ic group of elasmobranchs including ``Tam- node including the last two taxa. Thus, the iobatis sp.'', T. vetustus, Cladodoides wil- feature may characterize a group of Paleo- dungensis, Hybodus, and Tristychius (and, by zoic sharks, but there is uncertainty whether inference, modern neoselachians, although this group is monophyletic. A long otic re- these were not speci®ed in their analysis). gion is also present in other Paleozoic phal- This group was characterized by several de- acanthous sharks that may be closely related rived features, including: dorsal aorta divid- to Cladodoides, Tamiobatis, and/or Ortha- ing into paired lateral aortae behind the oc- canthus but which were not included in ciput; a single internal carotid foramen; oti- Schaeffer's (1981) analysis (e.g., Goodri- co-occipital length equal to or greater than chthys, Ctenacanthus). The otic region is ethmo-orbital region; slot-shaped endolym- also quite long in Wodnika (which may be a phatic fossa; anteroposteriorly expanded primitive neoselachian; Maisey, 1984a). postorbital process; `èthmoid'' (ϭ orbital) Schaeffer (1981) united Ctenacanthus articulation extends anteriorly from the level with Cladodoides, Tamiobatis and Orthacan- of the optic foramen, with a short suborbital thus by two additional characters, but in the shelf (note however that this is not the con- Cladodoides braincase one of these (multi- dition in orbitostylic sharks); and presence of layered prismatic calci®cation) is absent and a precerebral fontanelle. the other character is unknown (lateral otic The concept of an ancient group of cla- process ``pronounced''). Thus, the similari- dodont sharks is entrenched as far back as ties he noted are only applicable to cranial the late 19th century, although earlier diag- morphology in Cleveland Shale ``ctena- noses are unsatisfactory. Woodward (1889: canths,'' Tamiobatis, and xenacanths. 16) summarized the Cladodontida as `àn un- In Orthacanthus, the ventral otic notch diagnosable family, apparently closely allied (representing a persistent metotic ®ssure) is to the Pleuracanthidae'', noting that the pec- very long, but the notch is considerably toral ®n was apparently `à uniserial archip- shorter in Cladodoides, Tamiobatis vetustus terygiumÐintermediate between the truly (NMNH 1717), and ``Tamiobatis sp.'' biserial one of Pleuracanthus and the pec- (AMNH 2140). Closure of the ®ssure (as in toral ®n of modern sharks''. Blot (1969) for- neoselachians and hybodonts) may be a de- mally diagnosed the Cladodontidae (includ- rived character, but the elongated ®ssure in ing Cladodus) by the presence of a long, 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 95 straight pectoral metapterygial axis, although chimaeroids (in which the otic region can this is unknown in C. mirabilis and C. wil- also be characterized as ``short'') is unre- dungensis. Miles (in Moy-Thomas and solved. Miles, 1971) also grouped Paleozoic sharks with such an axis in the order Cladodontida CONCLUSIONS and again included C. wildungensis despite the lack of postcranial data. Both classi®ca- 1. The original description of cranial morphol- tions imply that a close relationship exists ogy in Cladodoides presented by Gross (1937) is essentially accurate, and very few between C. wildungensis, stethacanthids, and points of disagreement have been found (e.g., symmoriids, which is not supported by sub- the position of the abducent foramen and the sequent investigations (Schaeffer, 1981; trigeminal mandibular ramus). Scanning has Maisey, 1984a; Coates and Sequeira, 1998), provided new information about the internal and the classical view that ``cladodont'' morphology of the braincase, especially the sharks form a monophyletic group is gener- arrangement of cranial nerves, blood vessels, ally considered obsolete. However, if the tri- and the skeletal labyrinth. These new data basal pectoral endoskeleton (with anterior mean that many aspects of cranial morphol- basals homologous to the propterygium and ogy in other extinct elasmobranchs such as mesopterygium plus a metapterygium) is the Tamiobatis vetustus and Orthacanthus can now be interpreted with greater con®dence. primitive condition for gnathostomes (Jan- 2. The braincase is probably from a juvenile or vier, 1996), the condition represented in ex- submature individual. It shows indications of tinct sharks such as Cladoselache, symmo- incomplete development (e.g., the presence of riids, and stethacanthids (with a series of pre- a wide bucco-hypophyseal fenestra and ab- metapterygial radials plus a segmented mesence of a precarotid commissure; incomplete tapterygial `àxis'') may represent a derived enclosure of the internal carotids and orbital condition. Cladodoides,``Cobelodus'', and arteries by basicranial cartilage; incomplete ``Tamiobatis sp'' are united by an extensive closure of the braincase roof along the dorsal polar cartilage-derived region, anterior posi- midline). However, only features which de- tion of the orbital articulation within the or- velop relatively late in neoselachian ontogeny are implicated, suggesting that cranial devel- bit, inclined orientation of the postorbital opment was essentially complete. In addition, process, and exclusion of the trigeminal and the braincase lacks multilayered prismatic facial foramina from the orbit. These features calci®cation (cf. Schaeffer, 1981), which may may eventually be recognized in other ``cla- represent a further juvenile condition. dodont'' sharks, and it is interesting that all 3. The efferent pseudobranchial and internal ca- Paleozoic sharks implicated as possessing an rotid arteries meet above the trabeculae in extensive polar cartilage contribution proba- neoselachians, hybodonts, and many Paleo- bly had cladodont teeth. It is even conceiv- zoic sharks, including Cladodoides, Tamiob- able that cladodont sharks really do form a atis and Orthacanthus. This probably repre- monophyletic group (an old idea that has sents a derived condition, in contrast with osteichthyans, placoderms, Acanthodes, and long been out of favor), characterized by primitive chondrichthyans such as Pucapam- apomorphic dental, cranial, and postcranial pella. The efferent pseudobranchial arrange- features, although it would be rash to present ment in modern chimaeroids is also consid- this in a formal classi®cation based on the ered to be derived and may have arisen from available data. Of particular interest is the a condition similar to that found in Pucapam- fact that these apomorphic characters cut pella. across other groupings that have been pro- 4. A ``basal'' (i.e., palatobasal) articulation be- posed in the past; for example, an extensive tween the palatoquadrate and cranium is pre- polar cartilage contribution is suggested in sent in crown-group gnathostomes (osteichth- ``Tamiobatis'' sp. and Cladodoides, both of yans plus chondrichthyans), Acanthodes, and perhaps in placoderms. The orbital articula- which have a ``long'' otic region, as well as tion in neoselachians is probably homologous in ``Cobelodus'', where the otic region is to the palatobasal articulation found in os- ``short'', but not in Orthacanthus, where the teichthyans and early chondrichthyans such otic region is ``long''. The phylogenetic sig- as Pucapampella. However, the articulation ni®cance of the extensive polar cartilage in in orbitostylic neoselachians clearly repre- 96 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

sents an apomorphic condition. A palatobasal to the palatobasal articulation (e.g., Pucapam- articulation is inferred to be present in Paleo- pella). Presence of an extensive suborbital zoic chondrichthyans such as Cladodoides, shelf behind this articulation may be a de- based on its topographic relationship to that rived feature of some Paleozoic sharks such region of the basicranium inferred to be de- as Cladodoides. It is uncertain whether the rived from the embryonic polar cartilage. extensive suborbital shelf in modern galeo- However, this area extends anteriorly within morph sharks is primitive or if it represents the orbit (probably representing an apo- an independently derived feature that ap- morphic condition). In Heterodontus and peared following the loss of a palatobasal ar- modern galeomorphs, an ethmoidal articula- ticulation. Absence of a shelf anterior to the tion is developed on the trabecular cartilage articulation may be a derived feature of or- anteriorly, but this is unrelated ontogenetical- bitostylic sharks and of Paleozoic taxa such ly to the polar cartilage and here is not con- as Cladodoides, but it is unclear whether this sidered an orbital articulation. An ethmoidal represents a synapomorphy of a larger elas- articulation seems to be present in extinct mobranch group. Mesozoic sharks such as Synechodus and Hy- 7. The mesencephalic chamber is comparatively bodus, as well as in much earlier chondri- longer in Cladodoides and Orthacanthus than chthyans such as Pucapampella and Debeer- in neoselachians such as Squalus and Noto- ius. The palatobasal and ethmoidal articula- rynchus and in hybodonts (e.g., Hybodus, Tri- tions are presumably obscured by fusion of bodus). The differing positions of the hypo- the palatoquadrate to the braincase in modern physeal chamber and oculomotor nerve (rel- chimaeroids. ative to the otic capsule and postorbital pro- 5. A very large area of the basicranium was cess) in these forms may be related to the probably derived from the polar cartilages in differing proportions of the mesencephalic Cladodoides, since: (1) two landmark fea- chamber. However, the length of this chamber tures marking the anterior and posterior mar- is apparently unrelated to the extent of the gins of the polar cartilage in neoselachians polar cartilage, since the chamber is long in (the efferent pseudobranchial and pituitary both Cladodoides and Orthacanthus, but the vein foramina) are widely separated; (2) the inferred polar cartilage-derived region is far internal carotid grooves continue around the less extensive in Orthacanthus. The shorter lateral walls of the bucco-hypophyseal fenes- mesencephalic chamber of hybodonts and tra, suggesting that the postpituitary commis- neoselachians may represent an apomorphic sure (formed by the polar cartilage in neose- condition. lachians) may also have extended around the 8. In Cladodoides, there is a small ventral fossa fenestra; and (3) the dorsum sellae and the containing the pituitary and abducent foram- hypophyseal chamber (closely associated ina as well as a pit for the external rectus with the polar cartilage in neoselachians) are muscle insertion. The fossa does not include both very extensive. The polar cartilage prob- either the trigeminal or facial foramina and ably also made an extensive contribution to therefore does not correspond exactly to the the basicranium in Tamiobatis,``Cobelodus'', trigemino-facial recess in neoselachians. Two and perhaps Akmonistion, but not in Ortha- mutually exclusive phylogenetic hypotheses canthus. In modern elasmobranchs and os- are suggested: the arrangement found in Cla- teichthyans, the polar cartilage is small (or dodoides may represent the plesiomorphic ar- even unrecognizable) and at best forms only rangement for elasmobranchs (in which case a minor part to the basicranium, although it the trigemino-facial recess arose indepen- contributes to the hypophyseal chamber, dor- dently in modern elasmobranchs and early ac- sum sellae, and basal angle. In batoids, the tinopterygians); alternatively, if the trigemi- embryonic polar cartilage is small but the no-facial recess is a conserved feature of dorsum sellae and hypophyseal chamber are crown-group gnathostomes (Gardiner, reduced or absent. Reconstructions showing a 1984a), the arrangement in Cladodoides may large polar cartilage-derived region in placod- be derived. The facial foramen is located be- erms (e.g., StensioÈ, 1963a) may be erroneous, hind the orbit in some (perhaps all) sharks since the inferred position of the efferent with an extensive polar cartilage, and these pseudobranchial artery is probably incorrect features may be related developmentally. and it probably lay farther posteriorly 9. The postorbital arcade in Cladodoides is in- (Young, 1986). clined anterodorsally. Consequently, its dor- 6. In chondrichthyans, a broad suborbital shelf sal and ventral attachments do not lie at the may have been present primitively, anterior same level on the lateral surface of the brain- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 97

case, the dorsal part of the process is anterior spaces are similarly separated anteriorly by a to the otic capsule, and the trigeminal and fa- septum ending just above an unchondri®ed cial trunks left the braincase behind the post- acustico-trigemino-facial fossa. The latter orbital process (as did the anterodorsal lateral represents a conserved feature of crown- line nerve). Presumably, the orientation of the group gnathostomes. Similarities noted by embryonic lateral commissure in Cladodoides earlier investigators in the inner ear of neo- affected its relationship to the facial and tri- selachians and arthrodires probably represent geminal nerves. This may represent an apo- a mixture of primitive (craniate, gnathostome) morphic feature in Cladodoides. and convergent features. 10. The mandibular ramus of the trigeminal nerve 13. Labyrinth morphology in Cladodoides differs probably passed through the postorbital pro- profoundly from that described by StensioÈ cess, unlike in neoselachians (where it passes (1937) in the Humboldt University ``Clado- anterior to the process), and also unlike Acan- dus'' specimen, in which the semicircular ca- thodes (where it passed posterior to the pro- nals are purportedly arranged as in neosela- cess). The super®cial ophthalmic ramus of the chians. Further investigation of the Humboldt anterodorsal lateral line nerve in Cladodoides University specimen is required in order to probably passed through the postorbital pro- establish the validity of StensioÈ's interpreta- cess dorsally before entering the orbit, while tion. If StensioÈ's claims are substantiated, its buccal branch either passed anteriorly via ``Cladodus'' hassiacus and Cladodoides wil- the jugular canal or else accompanied the tri- dungensis are very different taxa. geminal nerve through the postorbital process 14. The hyomandibular fossa is not preserved in (considered here the most plausible arrange- the Cladodoides braincase but was probably ment). The palatine ramus of the facial nerve located on the posterolateral wall of the otic in Cladodoides divided into anterior and pos- capsule, as in Orthacanthus. The position of terior ramules before it left the braincase, and the fossa in elasmobranchs is probably apo- both ramules passed separately through the morphic. ¯oor of the lateral commissure. Internally, the 15. The basicranial arteries in Tamiobatis vetus- prefacial commissure forms an acustico-tri- tus were probably arranged as in Cladodoides gemino-facial recess and isolates the facial wildungensis, and many discrepancies in the nerve proper from the trigeminal and anter- earlier literature can now be explained. Paired odorsal lateralis trunks, as in gnathostomes lateral aortae are present in many early chon- generally. drichthyans and sarcopterygians. A median 11. The large dorsal opening between the otic aortic canal is present in Acanthodes, actin- capsules of Cladodoides is identi®ed as a pos- opterygians, early holocephalans (e.g., Helo- terior dorsal fontanelle, not a parietal (endo- dus; Moy-Thomas, 1936), the stethacanthid lymphatic) fossa. In primitive chondri- shark Akmonistion (Coates and Sequeira, chthyans, this fontanelle does not form part 1998, 2001b), and in ``Cobelodus'' (in prep.). of a low-frequency phonoreceptor system like Gardiner (1984a: 207) was undecided wheth- the parietal fontanelle of neoselachians. A er presence of a median aortic canal is a prim- posterior dorsal fontanelle is also present in itive gnathostome character or if it arose in- Acanthodes, early actinopterygians, modern dependently (as suggested by its phylogenet- chondrosteans (e.g., Acipenser), and in early ically restricted occurrence). Presence of a sarcopterygians (e.g., Eusthenopteron). Pres- median aorta is probably apomorphic, but fur- ence of the fontanelle is here regarded as a ther investigation is necessary to determine synapomorphy of crown-group gnathostomes. whether it represents a synapomorphy of hol- Primitively, this fontanelle is connected to the ocephalans and some Paleozoic sharks. dorsal part of the otico-occipital ®ssure. In 16. Persistence of the otico-occipital ®ssure is re- placoderms and ostracoderms both the fon- garded here as a synapomorphy of crown- tanelle and ®ssure are absent (also see be- group gnathostomes plus Acanthodes. The low). ®ssure corresponds to the posterior end of the 12. The otic capsule of Cladodoides primitively braincase in hag®shes and lampreys, which resembles that of Acanthodes, osteichthyans, lack an occipital region. An occipital region and chimaeroids in several respects: a crus is well developed in placoderms and ostra- commune is retained, the posterior semicir- coderms (suggesting that the modern cyclo- cular canal is not isolated, the medial capsular stome arrangement may be derived), but there wall is unchondri®ed, and there is no evi- is no evidence of an otico-occipital ®ssure. In dence of perilymphatic fenestrae. In Clado- Cladodoides, the ventral (metotic) part of the doides and Mimia, the cranial and labyrinth otico-occipital ®ssure (corresponding to the 98 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

embryonic basicapsular fenestra) is closed ware licenses, and scanning expenses) was and the hypotic lamina is fused to the over- provided by the Herbert and Evelyn Axelrod lying otic capsule, forming a glossopharyn- Research Chair in Paleoichthyology (an en- geal canal as in neoselachians. However, the dowed fund in the Division of Paleontology). canal is continuous with the dorsal part of the Finally, I thank Maria da Gloria P. de Car- otico-occipital ®ssure (containing the vagal nerve), which contradicts Gardiner's (1984a) valho for her support during the preparation contention that an actinopterygian-like vestib- of this work. ular fontanelle bounded by a posterior basicapsular fenestra is present in elasmobranchs. REFERENCES A vestibular fontanelle is absent in many pla- Agassiz, J.L.R. 1833±44. Recherches sur les pois- coderms and all osteostracans and galeaspids sons fossiles. 5 vols. Neuchatel: Imprimerie Pe- where the braincase is known, but is present titpierre. in sarcopterygians (including tetrapods, Allis, E.P. 1914. The pituitary fossa and trigemi- where part of it forms the fenestra ovalis; no-facialis chamber in selachians. Anatomisch- Gardiner, 1984a: 204). However, in actinop- er Anzeiger 46: 225±253. terygians and sarcopterygians it represents a Allis, E.P. 1923a. The cranial anatomy of Chla- different part of the metotic ®ssure from the mydoselachus anguineus. Acta Zoologica 4: glossopharyngeal canal of neoselachians. 123±221. Neither Gardiner's (1984a) claim that a ves- Allis, E.P. 1923b. Are the polar and trabecular car- tibular fontanelle is present in Acanthodes, tilages of vertebrate embryos the pharyngeal el- nor his conclusion that the fontanelle repre- ements of mandibular and premandibular arch- sents a conserved gnathostome condition (im- es? Journal of Anatomy 58: 37±51. plying that its absence in sharks is secondary) Allis, E.P. 1928. Concerning the pituitary fossa, can be con®rmed. the myodome and the trigemino-facialis cham- 17. A persistent ventral otic ®ssure at the junction ber in recent gnathostome ®shes. Journal of of the parachordal and trabecular-polar carti- Anatomy 63: 95±141. lages is present in crown-group gnathostomes Basden, A.M., G.C. Young, M. Coates, and A. and probably in Acanthodes, but it is absent Ritchie. 2000. The most primitive osteichthyan in cyclostomes, placoderms, and ostraco- braincase? Nature 403: 185±188. derms. Basden, A.M., and G.C. Young. 2001. A primitive actinopterygian neurocranium from the early ACKNOWLEDGMENTS Devonian of southeastern Australia. Journal of Vertebrate Paleontology 21: 745±766. I thank G. Plodowsky (Forschungsinstitut Bertmar, G. 1959. On the ontogeny of the chon- und Naturmuseum Senckenberg, Frankfurt) dral skull in Characidae, with a discussion on for permitting me to borrow the Cladodoides the chondrocranial base and the visceral chon- braincase for study. Timothy Rowe, Richard drocranium in ®shes. Acta Zoologica (Stock- Ketchum, and Matthew Colbert (University holm) 40: 203±364. of Texas at Austin) scanned the specimen Bjerring, H. 1978. The ``intracranial joint'' versus and provided preliminary imaging. David the ``ventral otic ®ssure''. Acta Zoologica (Stockholm) 59: 203±214. Reddy (formerly at the American Museum of Bleckman, H., and M.H. Hoffman. 1999. Special Natural History) gave invaluable help in im- senses. In W.C. Hamlett (editor), Sharks, skates aging and visualization. Chester Tarka and and rays; the biology of elasmobranch ®shes: Lorraine Meeker (AMNH) provided conven- 300±328. Baltimore: Johns Hopkins University tional photography and made the reproduc- Press. tion of Walter Gross' original illustrations. I Blot, J. 1969. SysteÂmatique. In J. Pivetau (editor), also thank M. Ginter, G. Northcutt, B. TraiteÂ de paleÂontologie, IV (2), Gnathostomes, Schaeffer, and the late M. Williams for many acanthodiens, placodermes, eÂlasmobranches: useful discussions about chondrichthyan and 702±776. Paris: Masson. gnathostome morphology. The author is in- Cerny, R., P. Lwigale, R. Ericsson, D. Meule- mans, H.-H. Epperlein, and M. Bonner-Fraser. debted to P. Janvier and M. Coates for their 2004. Developmental origins and evolution of careful and informative reviews of this paper. jaws: new interpretation of ``maxillary'' and This research was undertaken almost entirely ``mandibular''. Developmental Biology 276: at the American Museum of Natural History. 225±236. Funding (including computer hardware, soft- Chai, Y., X. Jiang, Y. Ito, P. Bringas, J. Han, D.H. 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 99

Rowitch, P. Soriano, A.P. McMahon, and H.M. actions of the Royal Society of London (B) Sucov. 2000. Fate of the mammalian cranial 224: 287±312. neural crest during tooth and mandibular mor- Denison, R. 1979. Acanthodii. In H.-P. Schulze phogenesis. Development 127: 1671±1679. (editor), Handbook of paleoichthyology 5: 1± Chang, M.M. 1995. Diabolepis and its bearing on 62. New York: Fischer. the relationships between porolepiforms and Devillers, C. 1958. Le craÃne des poissons. In P. - dipnoans. Bulletin du MuseÂum National P. GrasseÂ (editor), TraiteÂ de zoologie 13(1): d'Histoire Naturelle (Paris) 4e seÂrie 17: 235± 551±687. Paris: Masson. 268. Dick, J.R.F. 1978. On the Carboniferous shark Chibon, P. 1974. Un systeÁme morphogeÂneÂtique re- Tristychius arcuatus Agassiz from Scotland. marquable; la creÃte neurale des verteÂbreÂs. An- Transactions of the Royal Society of Edinburgh nales de Biologie 13: 459±480. 70: 63±109. Chorn, J., and K.N. Whetstone. 1978. The use of Didier, D. 1995. Phylogenetic systematics of ex- the term nomen vanum in taxonomy. Journal of tant chimaeroid ®shes (Holocephali, Chimaero- Paleontology 52: 494. idei). American Museum Novitates 3119: 1±86. Coates, M.I., and S.E.K. Sequeira. 1998. The Eastman, C.R. 1897. Tamiobatis vetustus: a new braincase of a primitive shark. Transactions of form of fossil skate. American Journal of Sci- the Royal Society of Edinburgh Earth Sciences ence 4th. ser. 4: 85±90. 89: 63±85. El-Toubi, M. R. 1949. The development of the Coates, M.I., and S.E.K. Sequeira. 2001a. Early chondrocranium of the spiny dog®sh, Acanthias sharks and primitive gnathostome interrelation- vulgaris (Squalus acanthias). Part 1. Neurocra- ships. In P.E. Ahlberg (editor), Major events in nium, mandibular and hyoid arches. Journal of early vertebrate evolution: palaeontology, phy- Morphology 84: 227±280. Forey, P. 1998. History of the coelacanth ®shes. logeny, genetics and development: 241±262. London: Chapman and Hall. New York: Taylor and Francis for the System- Francis-West, P., R. Ladher, A. Barlow, and A. atics Association. Graveson. 1998. Signalling interactions during Coates, M.I., and S.E.K. Sequeira. 2001b. A new facial development. Mechanisms of Develop- stethacanthid chondrichthyan from the Lower ment 75: 3±28. Carboniferous of Bearsden, Scotland. Journal Gadow, H. 1888. On the modi®cations of the ®rst of Vertebrate Paleontology 21: 438±459. and second visceral arches, with especial ref- Cohn, M.J. 2002 Lamprey Hox genes and the or- erence to the homologies of the auditory ossi- igins of jaws. Nature 416: 386±387. cles. Philosophical Transactions of the Royal Corwin, J.T. 1989. Functional anatomy of the au- Society of London (B) 179: 451±485. ditory system in sharks and rays. Journal of Ex- Gardiner, B.G. 1984a. The relationships of the pa- perimental Zoology. 2(suppl.): 62±74. laeoniscid ®shes, a review based on new spec- Daget, J., and F. d'Aubenton. 1957. DeÂveloppe- imens of Mimia and Moythomasia from the Up- ment et morphologie du craÃne d'Heterotis nil- per Devonian of Western Australia. Bulletin of oticus Ehr. Bulletin de l'Institut Fondamental the British Museum (Natural History) Geology d'Afrique Noire 11: 1±178. 37(4): 173±428. Daniel, J.F. 1934. The Elasmobranch ®shes. Gardiner, B.G. 1984b. The relationships of pla- Berkeley: University of California. coderms. Journal of Vertebrate Paleontology 4: Dean, B. 1909. Studies on fossil ®shes (sharks, 379±395. chimaeroids and arthrodires). Memoirs of the Gaudin, T. 1991. A re-examination of elasmo- American Museum of Natural History 9: 209± branch monophyly and chondrichthyan phylog- 287. eny. Neues Jahrbuch fuÈr Geologie und PalaÈon- De Beer, G.R. 1924. Contributions to the study of tologie Abhandlungen 182: 133±160. the development of the head in Heterodontus. Gegenbaur, C. 1872. Untersuchungen zur verglei- Quarterly Journal of Microscopical Science 68: chenden Anatomie der Wirbelthiere. III. Das 39±65. Kopfskelet der Selachier, ein Beitrag zur Er- De Beer, G.R. 1931. The development of the skull kenntniss der Genese des Kopfskeletes der Wir- in Scyllium (Scyliorhinus) canicula L. Quarter- belthiere. Leipzig: Engelmann. ly Journal of Microscopical Science, n. ser. 74: Ginter, M. 1998. Taxonomic problems with Car- 591±652. boniferous ``cladodont-level'' sharks. Ichthyol- De Beer, G.R. 1937. The development of the ver- ith Issues Special Publication 4: 14±16. tebrate skull. Oxford: Clarendon Press. Ginter, M., and A. Ivanov. 1992. Devonian phoe- De Beer, G.R., and J.A. Moy-Thomas. 1935. On bodont shark teeth. Acta Palaeontologica Po- the skull of Holocephali. Philosophical Trans- lonica 37: 55±75. 100 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

Goodrich, E.S. 1930. Studies of the structure and Holmgren, N. 1943. Studies on the head in ®shes. development of vertebrates. London: Macmil- Part IV. General morphology of the head in lan. ®sh. Acta Zoologica (Stockholm) 23: 1±188. Goujet, D. 1984. Les poissons Placodermes du Hotton, N. 1952. Jaws and teeth of American xen- Spitzberg. Arthrodires Dolichothoraci de la acanth sharks. Journal of Paleontology 26: Formation de Wood Bay (DeÂvonien Inferieur). 489±500. Paris: Cahiers de PaleÂontologie, Centre Nation- Hubendick, B. 1943. Zur Kenntnis der Entwick- al de la Recherche Scienti®que. lung des Primordialcraniums bei Leusiscus ru- Grogan, E.D., and R. Lund. 2000. Debeerius el- tilus. Arkiv fuÈr Zoologi 34(7): 1±35. lefseni (Fam. nov., gen. nov., spec. nov.), an Huxley, T.H. 1875. Preliminary note on the brain autodiastylic chondrichthyan from the Missis- and skull of Amphioxus lanceolatus. Proceed- sippian Bear Gulch Limestone of Montana ings of the Royal Society of London 23: 127± (USA), the relationships of Chondrichthyes, 132. and comments on gnathostome evolution. Jour- Huxley, T.H. 1876. Contributions to morphology. nal of Morphology 243: 219±245. Ichthyopsida. No. 1. On Ceratodus forsteri, Gross, W. 1933. Die Wirbelthiere des Rheinischen with observations on the classi®cation of ®shes. Devons. Abhandlungen der Preussischen Geo- Proceedings of the Zoological Society of Lon- logischen Landesanstalt 154: 1±83. don 1876: 24±59. Gross, W. 1937. Das Kopfskelett von Cladodus IselstoÈger, H. 1937. Das Neurokranium von Rhina wildungensis Jaekel. 1. Teil. Endocranium und squatina und einige Bemerkungen uÈber ihre Palatoquadratum. Senckenbergiana 19: 80±107. systematische Stellung. Zoologische Jahrbucher Gross, W. 1938. Das Kopfskelett von Cladodus Abteilung fuÈr Anatomie und Ontogenie der Ti- wildungensis Jaekel. 2. Teil. Der Kieferbogen. ere 62: 349±394. Senckenbergiana 20: 123±145. Jaekel, O. 1921. Die Stellung der PalaÈontologie Haeckel, E.H.P.A. 1875. Die Gastrula und die Ei- zu einigen Problemen der Biologie und Phylo- furchung der Tiere. Jenaische Zeitschrift fuÈr genie. PalaÈontologische Zeitschrift 3: 213±239. Naturwissenschaft 9: 402±508. Janvier, P. 1981a. The phylogeny of the Craniata, Hall, B.K. 1998. Evolutionary developmental bi- with particular reference to the signi®cance of ology, 2nd. ed. London: Chapman and Hall. fossil ``agnathans''. Journal of Vertebrate Pa- Hammarberg, F. 1937. Zur Kenntniss der onto- leontology 1: 121±159. genetischen Entwicklung des SchaÈdels von Lepidosteus platostomus. Acta Zoologica Janvier, P. 1981b. Norselaspis glacialis n.g., n.sp. (Stockholm) 18: 209±337. et les relations phylogeÂneÂtiques entre les Kia- Heidtke, U.H.J. 1998. Revision der Gattung Or- eraspidiens (Osteostraci) du DeÂvonien infeÂrieur thacanthus Agassiz 1843 (Chondrichthyes: du Spitzberg. Palaeovertebrata 11(2, 3): 19± Xenacanthida). PalaÈontologische Zeitschrift 72: 131. 135±147. Janvier, P. 1984. The relationship of the Osteos- Heidtke, U.H.J. 1999. Orthacanthus (Lebacha- traci and Galeaspida. Journal of Vertebrate Pa- canthus) senckenbergianus Fritsch 1889 (Xen- leontology 4: 344±358. acanthida: Chondrichthyes): revision, organi- Janvier, P. 1985. Les ceÂphalaspides du Spitzberg. sation und phylogenie. Freiburger Forschung- Cahiers de paleÂontologie. Paris: Centre Nation- sheft C481: 63±106. al de la Recherche Scienti®que. Heidtke, U.H.J., and K. KraÈtschmer. 2001. Glad- Janvier, P. 1996. Early vertebrates. Oxford: Clar- bachus adentatus nov. gen et sp., ein primitiver endon Press. Hai aus dem Oberen Givetium (Obers Mitteld- Janvier, P., and M. SuaÂrez-Riglos. 1986. The Si- evon) der Bergische Gladbach-Paffrath-Mulde lurian and Devonian vertebrates of Bolivia. (Rheinisches Schiefergebirge). Mainzer Geow- Bulletin de l'Institut FrancËais d'Etudes Andines issenschaftliche Mitteilungen 30: 105±122. 15(3, 4): 73±114. Holmgren, N. 1940. Studies on the head in ®shes. Jarvik, E. 1954. On the visceral skeleton in Eusth- Part I. Development of the skull in sharks and enopteron with a discussion of the parasphen- rays. Acta Zoologica (Stockholm) 21: 51±267. oid and palatoquadrate in ®shes. Kungliga Holmgren, N. 1941. Studies on the head in ®shes. Svenska VetenskapsAkademiens Handlingar 5: Part II. Comparative anatomy of the adult se- 1±104. lachian skull with remarks on the dorsal ®ns in Jarvik, E. 1977. The systematic position of acan- sharks. Acta Zoologica (Stockholm) 22: 1±100. thodian ®shes. In S.M. Andrews, R.S. Miles, Holmgren, N. 1942. Studies on the head in ®shes. and A.D. Walker (editors), Problems in verte- Part III. The phylogeny of elasmobranch ®shes. brate evolution: 199±224. London: Academic Acta Zoologica (Stockholm) 23: 129±261. Press. 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 101

Jarvik, E. 1980. Basic structure and evolution of lachian genus Ctenacanthus Agassiz. No. 3. vertebrates. London: Academic Press. Nominal species referred to Ctenacanthus. Johnson, G.D. 1999. Dentitions of late Paleozoic American Museum Novitates 2774: 1±20. Orthacanthus species and new species of? Xen- Maisey, J.G. 1985. Cranial morphology of the fos- acanthus (Chondrichthyes: Xenacanthiformes) sil elasmobranch Synechodus dubrisiensis. from North America. Acta Geologica Polonica American Museum Novitates 2804: 1±28. 49(3): 215±266. Maisey, J.G. 1987. Cranial anatomy of the Lower Jollie, M. 1971. Some developmental aspects of Jurassic shark Hybodus reticulatus (Chondrich- the head skeleton of the 35±37 Squalus acan- thyes: Elasmobranchii), with comments on thias foetus. Journal of Morphology 133: 17± elasmobranch systematics. American Museum 40. Novitates 2878: 1±39. Jollie, M. 1962. Chordate morphology. London: Maisey, J.G. 1999. The supraotic bone in neop- Chapman and Hall. terygian ®shes (Osteichthyes, Actinopterygii). Kuritani, S., Y. Murakami, Y. Nobusada, R. Ku- American Museum Novitates 3267: 1±52. sakabe, and S. Hirano. 2004. Developmental Maisey, J.G. 2001a. Remarks on the inner ear of fate of the mandibular mesoderm in the lam- elasmobranchs and its interpretation from skel- prey, Lethenteron japonicum: conservative etal labyrinth morphology. Journal of Mor- morphology and development of the gnathos- phology 250: 236±264. tome jaw with special reference to the nature Maisey, J.G. 2001b. A primitive chondrichthyan of the trabecula cranii. Journal of Experimental braincase from the Middle Devonian of Boliv- Zoology 302B: 458±468. ia. In P.E. Ahlberg (editor), Major events in ear- Lauder, G.V. 1980. On the evolution of the jaw ly vertebrate evolution: paleontology, phyloge- adductor musculature in primitive gnathostome ny, genetics, and development: 263±288. New ®shes. Breviora 460: 1±10. York: Taylor and Francis, for the Systematics Lund, R. 1985. The morphology of Falcatus fal- Association. catus (St. John and Worthen), a Mississippian Maisey, J.G. 2001c. CT-scan reveals new cranial stethacanthid chondrichthyan from the Bear features in Devonian chondrichthyan ``Clado- Gulch Limestone of Montana. Journal of Ver- dus'' wildungensis. Journal of Vertebrate Pale- tebrate Paleontology 5: 1±19. ontology 21(4): 807±810. Luther, A. 1909. Untersuchungen uÈber die vom N. Maisey, J.G. 2004a. Endocranial morphology in trigeminus innervierte Muskulatur der Sela- chier (Haie und Rochen). Acta Societatis Scien- fossil and Recent chondrichthyans. In G. Ar- tiarum Fennicae 36: 1±168. ratia, M. Wilson, and R. Cloutier (editors), Re- Maisch, M.W. 1998. Wildungenichthys grossi n. cent advances in the origin and radiation of ear- gen. n. sp., a new selenosteid arthrodire (Pla- ly vertebrates. 139±170. Munich: Pfeil. codermi, arthrodira) from the Kellwasserwerk Maisey, J.G. 2004b. Morphology of the braincase (late Frasnian, Upper Devonian) of Bad Wil- in the broadnose sevengill shark Notorynchus dungen (Hessen, W. Germany). Palaeontologis- (Elasmobranchii, Hexanchiformes), based on che Zeitschrift 72: 373±382. CT scanning. American Museum Novitates Maisey, J.G. 1980. An evaluation of jaw suspen- 3351: 1±52. sion in sharks. American Museum Novitates Maisey, J.G., and M.E. Anderson. 2001. A prim- 2706: 1±17. itive chondrichthyan braincase from the early Maisey, J.G. 1981. Studies on the Paleozoic se- Devonian of South Africa. Journal of Verte- lachian genus Ctenacanthus Agassiz. No. 1. brate Paleontology 21(4): 702±713. Historical review and revised diagnosis of Maisey, J.G., and M.R. de Carvalho. 1997. A new Ctenacanthus, with a list of referred taxa. look at old sharks. Nature 385: 779±780. American Museum Novitates 2718: 1±22. Maisey, J.G., G.J.P. Naylor, and D.J. Ward. 2004. Maisey, J.G. 1983. Cranial anatomy of Hybodus Mesozoic elasmobranchs, neoselachian phylog- basanus Egerton from the Lower Cretaceous of eny and the rise of modern elasmobranch di- England. American Museum Novitates 2758: versity. In G. Arratia and A. Tintori (editors), 1±64. Mesozoic Fishes 3±Systematics, Paleoenviron- Maisey, J.G. 1984a. Chondrichthyan phylogeny: a ments and Biodiversity. 17±56. Munich: Pfeil. look at the evidence. Journal of Vertebrate Pa- Marinelli, W., and A. Strenger. 1959. Vergleichen- leontology 4(3): 359±371. de Anatomie und Morphologie der Wirbelthi- Maisey, J.G. 1984b. Higher elasmobranch phylog- ere. III. Lieferung, pt. 2, Squalus acanthias: eny and biostratigraphy. Zoological Journal of 179±308. Vienna: Franz Deuticke. the Linnean Society 82: 33±54. Meulemans, D., and M. Bonner-Fraser. 2002. Am- Maisey, J.G. 1984c. Studies on the Paleozoic se- phioxus and lamprey AP-2 genes: implications 102 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 288

for neural crest evolution and migration pat- homology. In K.A. Joysey, and A.E. Friday terns. Development 129: 4953±4962. (editors), problems of phylogenetic reconstruc- Miles, R.S. 1964. A reinterpretation of the vis- tion: 21±74. London: Academic Press. ceral skeleton in Acanthodes. Nature 204: 457± Rayner, D.H. 1951. On the cranial structure of an 459. early palaeoniscid, Kentuckia gen. nov. Trans- Miles, R.S. 1965. Some features of the cranial actions of the Royal Society of Edinburgh 62: morphology of acanthodians and the relation- 58±83. ships of the Acanthodii. Acta Zoologica (Stock- Retzius, G. 1881. Das GehoÈrorgan der Wirbelthi- holm) 46: 233±255. ere. Vol. 1. Stockholm: Samson and Wallin. Miles, R.S. 1968. Jaw articulation and suspension Romer, A.S. 1964. The braincase of the Palaeo- in Acanthodes and their signi®cance. Nobel zoic shark Tamiobatis. Bulletin of the Museum Symposium 4: 109±127. of Comparative Zoology at Harvard University Miles, R.S. 1973. Relationships of acanthodians. 131: 89±105. In P.H. Greenwood, R.S. Miles, and C. Patter- Rosen, D.E., P.E. Forey, B.G. Gardiner, and C. son (editors), Interrelationships of ®shes: 63± Patterson. 1981. Lung®shes, tetrapods, paleon- 103. London: Academic Press. tology and parsimony. Bulletin of the American Miller, R.F., R. Cloutier, and S. Turner. 2003. The Museum of Natural History 167: 159±276. oldest articulated chondrichthyan from the Ear- Schaeffer, B. 1967. Comments on elasmobranch ly Devonian period. Nature 425: 501±504. evolution. In P.W. Gilbert, R.F. Mathewson, and Moy-Thomas, J.A. 1936. On the structure and af- D.P. Rall (editors), Sharks, skates and rays: 3± ®nity of the Carboniferous cochliodont Helodus 35. Baltimore: Johns Hopkins University Press. simplex. Geological Magazine 73: 488±503. Schaeffer, B. 1971. The braincase of the holostean Moy-Thomas, J.A., and R.S. Miles. 1971. Palaeo- ®sh Macrepistius, with comments on neurocra- zoic ®shes. Philadelphia: Saunders. nial ossi®cation in the Actinopterygii. Ameri- Nielsen, E. 1942. Studies on Triassic ®shes from can Museum Novitates 2459: 1±34. East Greenland. 1. Glaucolepis and Boreoso- Schaeffer, B. 1975. Comments on the origin and mus. Meddelelser om Grùnland 138: 1±403. basic radiation of the gnathostome ®shes with Norris, H.W., and S.P. Hughes. 1920. The cranial, particular reference to the feeding mechanism. occipital and anterior spinal nerves of the dog- Colloque International Centre National de la ®sh, Squalus acanthias. Journal of Comparative Recherche Scienti®que 218: 101±109. Neurology 31: 293±402. Schaeffer, B. 1981. The xenacanth shark neuro- Northcutt, R.G., and W.E. Bemis. 1993. Cranial cranium, with comments on elasmobranch nerves of the coelacanth Latimeria (Ostei- monophyly. Bulletin of the American Museum chthyes: Sarcopterygii: Actinistia) and compar- of Natural History 169: 3±66. isons with other Craniata. Brain, Behavior and Sequeira, S.E.K., and M.I. Coates. 2000. Reas- Evolution 42(suppl. 1): 1±76. sessment of ``Cladodus'' neilsoni Traquair: a Oelofsen, B.W. 1986. A fossil shark neurocranium primitive shark from the Lower Carboniferous from the Permo-Carboniferous (lowermost of East Kilbride, Scotland. Palaeontology 43: Ecca Formation) of South Africa. In T. Uyeno, 153±172. R. Arai, T. Taniuchi, and K. Matsuura (editors), Sewertzoff, A.N. 1899. Die Entwickelung des Se- Indo-Paci®c ®sh biology: Proceedings of the lachierschaÈdels. Ein Beitrag zur Theorie der Second International Conference on Indo-Pa- korrelativen Entwickelung. Festschrift Carl von ci®c Fishes: 107±124. Tokyo: Ichthyological Kuppfer. Jena: Fischer. Society of Japan. Sewertzoff, A.N., and J. Disler. 1924. Das Phar- P'an, K., and S.T. Wang. 1978. Devonian Agnatha yngomandibulare der Selachier. Anatomischer and Pisces of South China. In Symposium on Anzeiger 58: 345±349. the Devonian System of South China: 299±333. Shirai, S. 1992. Squalean phylogeny: a new Beijing: Geological Publishing House. [in Chi- framework of ``squaloid'' sharks and related nese] taxa. Sapporo: Hokkaido University Press. Parker, T.J. 1887. Notes on Carcharodon ronde- StensioÈ, E. 1927. The Downtonian and Devonian letii. Transactions of the Zoological Society of vertebrates of Spitzbergen, 1. Family Cephal- London 1887: 27±40. aspidae. Skrifter om Svalbard og Nordishavet Patterson, C. 1975. The braincase of pholido- 12: 1±391. phorid and leptolepid ®shes, with a review of StensioÈ, E. 1937. Notes on the endocranium of a the actinopterygian braincase. Philosophical Devonian Cladodus. Bulletin of the Geological Transactions of the Royal Society of London Institute of Uppsala 27: 128±144. B269: 275±579. StensioÈ, E. 1950. La caviteÂ labyrinthique, Patterson, C. 1982. Morphological characters and l'ossi®cation scleÂrotique et l'orbite de Jagori- 2005 MAISEY: CRANIAL MORPHOLOGY OF CLADODOIDES 103

na. Colloques Internationaux du Centre Nation- Williams, M.E. 1985. The ``cladodont level'' al de la Recherche Scienti®que 21: 9±1. sharks of the Pennsylvanian black shales of StensioÈ, E. 1963a. Anatomical studies of the ar- central North America. Palaeontographica, throdiran head. 1. Preface, geological and geo- Stuttgart (A) 190: 83±158. graphical distribution, the organisation of the Williams, M.E. 1998. A new specimen of Tam- arthrodires, the anatomy of the head in the Dol- iobatis vetustus (Chondrichthyes, Ctenacanthoi- ichothoraci, Coccosteomorphi and Pachyosteo- dea) from the late Devonian Cleveland Shale of morphi. Taxonomic appendix. Kungliga Sven- Ohio. Journal of Vertebrate Paleontology 18(2): ska Vetenskapsakademiens Handlingar 9: 1± 251±260. 419. Wolfram, K.E. 1984. The functional anatomy of StensioÈ, E. 1963b. The brain and the cranial the jaw suspension of Notorynchus cepedianus nerves in fossil, lower craniate vertebrates. (Chondrichthyes, Selachii), with application to Skrifter av Det norske Videnskaps-Akademi i fossil forms. Unpublished M.Sc. thesis, Uni- Oslo 1, N.S. 13: 5±120. versity of Nebraska. 249 pp. StensioÈ, E. 1969. Elasmobranchiomorphi; Placod- Woodward, A.S. 1889. Catalogue of the fossil ermata; Arthrodires. In J. Piveteau (editor), ®shes in the British Museum (Natural History). TraiteÂ de paleÂontologie 4(2): 71±92. Paris: Mas- London: British Museum (Natural History). son. Young, G.C. 1979. New information on the struc- Summers, A.P. 2000. Stiffening the stingray skel- ture and relationships of Buchanosteus (Placod- etonÐan investigation of durophagy in myliob- ermi, Euarthrodira) from the early Devonian of atid stingrays (Chondrichthyes, Batoidea, My- New South Wales. Zoological Journal of the liobatidae). Journal of Morphology 243: 113± Linnean Society 66: 309±352. 126. Young, G.C. 1980. A new early Devonian pla- Swinnerton, H.H. 1902. A contribution to the coderm from New South Wales, Australia, with morphology of the teleostomian head skeleton, a discussion of placoderm phylogeny. Palaeon- tographica, Stuttgart (A) 167: 10±76. based upon a study of the developing skull in Young, G.C. 1986. The relationships of placod- the three-spined stickleback (Gasterosteus acu- erm ®shes. Zoological Journal of the Linnean leatus). Quarterly Journal of Microscopical Sci- Society 88: 1±57. ence 45: 503±593. Young, J.Z. 1981. The life of vertebrates, 3rd. ed. Torrey, T.W. 1962. Morphogenesis of the verte- Oxford: Oxford University Press. brates. New York: Wiley. Yu, X. 1998. A new porolepiform-like ®sh, Psa- Traquair, R.H. 1884. Description of a fossil shark rolepis romeri, gen. et sp. nov. from the Lower (Ctenacanthus costellatus) from the Lower Car- Devonian of Yunnan, China. Journal of Verte- boniferous rocks of Eskdale, Dumfriesshire. brate Paleontology 18: 261±274. Geological Magazine 3: 3±8. Zangerl, R. 1973. Interrelationships of early chon- Traquair, R.H. 1898. On Cladodus neilsoni (Tra- drichthyans. In P.H. Greenwood, R.S. Miles, quair) from the Carboniferous Limestone of and C. Patterson (editors), Interrelationships of East Kilbride. Transactions of the Geological ®shes: 1±14. London: Academic Press. Society of Glasgow 11: 41±50. Zangerl, R. 1990. Two new stethacanthid sharks Wang, N.Z. 1991. Two new Silurian galeaspids (Stethacanthidae, Symmoriida) from the Penn- (jawless craniates) from Zhejiang Province, sylvanian of Indiana, U.S.A. Palaeontographi- China, with a discussion of galeaspid-gnathos- ca, Stuttgart (A) 213: 115±141. tome relationships. In M.M. Chang, Y.H. Liu, Zangerl, R., and G.R. Case. 1976. Cobelodus acu- and G.R. Zhang (editors), Early vertebrates and leatus (Cope), an anacanthous shark from related problems of evolutionary biology: 41± Pennsylvanian black shales of North America. 65. Beijing: Science Press. Palaeontographica, Stuttgart (A) 154: 107±157. Watson, D.M.S. 1937. The acanthodian ®shes. Zangerl, R., and M. E. Williams. 1975. New ev- Philosophical Transactions of the Royal Society idence on the nature of the jaw suspension in of London (B) 228(549): 49±146. Palaeozoic anacanthous sharks. Palaeontology Wijhe, J. W. van. 1922. FruÈhe Entwickelungsstu- 18(2): 333±341. dien des Kopf- und Rumpfskeletts von Acan- Zhu, M., X. Yu, and P. Ahlberg. 2001. A primitive thias vulgaris. Bijdragen tot der Dierkunde 22: sarcopterygian ®sh with an eyestalk. Nature 271±298. 410: 81±84.