Holotype

GP/2T-2 IG-USP. Complete fish lacking the caudal fin, figured by Brito and Ferreira (1989: fig. 1A).

Paratype

DGM-DNPM 1296-P. Skull with first fin spine and associated dermal elements, figured by Brito and Ferreira (1989: fig. 2B).

Type Species

Tribodus limae Brito and Ferreira, 1989.

Diagnosis

As for the type species, Brito and Ferreira, 1989. Hybodont shark, weakly heterodont. Teeth high crowned not exceeding 5 mm in length, crown with strong vascular striae, deeper than broad in anterior and posterolateral teeth, shallower than broad in anterolateral teeth, root hybodontoid; fin spines slender reaching about 12.5 cm, lateral faces with 8 to 10 sharp continuous ridges, single anterior ridge forming a keel in the anterior border, posterior denticles in two series; large dermal denticles with thornlike shape measuring approximately 2.5 to 3.0 mm.

An emended diagnosis for Tribodus limae was provided by Brito (1991), in agreement with that of Brito and Ferreira (1989) except for the following change:

General Features

The neurocranium of Tribodus limae was first described by Brito (1992), who reconstructed it as generally similar in overall morphology to that of Egertonodus basanus and other previously described hybodonts. However, that description was based on an incomplete specimen (lacking much of the ethmoid region and postorbital processes), the only material then available (fig. 2). Later, Maisey and Carvalho (1997) announced the discovery of a more complete specimen, AMNH 13958. Although this specimen (fig. 3) was mentioned in several later papers, revealing a selection of intriguing morphological details (Maisey, 2004a, 2005, 2007), a complete description of its morphology has not been presented until now. Additional specimens of Tribodus limae neurocrania have since been discovered and are included in the present description (figs. 4, 5). CT scans of the braincase of AMNH 13958 were examined for the present study (figs. 6–9), and were used to produce digitally rendered images of the braincase (Fig. 10figs. 10–14), cranial endocast (figs. 15–18), and skeletal labyrinth (figs. 19–22), described below.

Fig. 2

Reproduction of original illustrations of the Tribodus limae braincase, modified from Brito (1992), in A, dorsal; B, ventral, and C, lateral views. Abbreviations (translated from the original French): ald, foramina for lateral dorsal aortae; a.orb, orbital artery; cap.ot, otic capsule; c.j, jugular canal; c.occ, occipital cotylus; csca, anterior semicircular canal; csce, external semicircular canal; cscp, posterior semicircular canal; f.end, endolymphatic (parietal) fossa; f.m, foramen magnum; f.p, precerebral fontanelle; p.art, articular process for palatoquadrate; p.ot.lat, lateral otic process; ppo, postorbital process; q.int, internasal keel; V.opht, superficial ophthalmic ramus of nerve V; V + VII, foramen for main branch of trigeminal and facial nerves; VII hym, foramen for hyomandibular branch of facial nerve; IX + X, glossopharyngeal-vagus foramen. Not to scale.

Fig. 3

Braincase of Tribodus limae, AMNH FF 13958. A, ventral view; B, dorsal view; C, lateral view. Scale bar is 2 cm.

Fig. 4

Tribodus limae, UERJ-PMB-40. A, dorsal; B, ventral views. Scale bar is 2 cm.

Fig. 5

Tribodus limae, UERJ-PMB-114. A, dorsal; B, ventral views. Scale bar is 2 cm.

Fig. 6

Transverse CT scan slices of the Tribodus limae braincase, AMNH 13958. Ethmoidal region. A, slice 57, through the olfactory chamber and precerebral fontanelle; B, slice 72, through the precerebral fontanelle showing the ventral exit of the orbitonasal canal; C, slice 83, through the precerebral fontanelle showing the orbitonasal canal traveling through the cartilage; D, E, F, slices 87, 89, and 94, respectively, through the precerebral fontanelle showing the course of the profundus nerve and orbitonasal canal; G, slice 100, at the level of the anterior cerebral canal; H, slice 159, through the anterior of the orbit at the level of the trochlear nerve and median ventral basicranial process. No scale.

Fig. 7

Transverse CT scan slices through the Tribodus limae braincase. Orbital region. A, slice 187, at the level of the optic nerve; B, slice 209, at the level of the oculomotor nerve; C, slice 216, at the level of the oculomotor nerve showing the anterior semicircular canal originating anterior to the postorbital process; D, E, slices 222 and 226, respectively, at the level of the efferent pseudobranchial foramen and the foramen for the superficial ophthalmic complex and profundus nerve; F, slice 239, at the level of the utricular recess and abducens foramen; G, H, slices 241 and 242, respectively, showing the course of the superficial ophthalmic complex and abducens nerve within the cartilage of the orbit. No scale.

Fig. 8

Transverse CT scan slices through the Tribodus limae braincase. Otic region. A, slice 257, at the level of the hypophysis and facial nerve; B, slice 270, showing the pituitary vein canal and the anterior ramus of the octaval nerve; C, slice 285, showing the hypophyseal duct and internal carotids; D, slice 290, showing the posterior ramus of the octaval nerve and the exit of the orbital arteries from the braincase; E, slice 298, at the level of the hypophyseal foramen; F, slice 303, showing the exit of the efferent hyoidean arteries from the braincase; G, slice 307, showing the canals for the lateral dorsal aortae converging to form a single chamber medial to the efferent hyoidean foramina; H, slice 313, showing the lateral dorsal aortae diverging posteriorly. No scale.

Fig. 9

Transverse CT scan slices through the Tribodus limae braincase. Occipital region. A, slice 334, at the level of the preampullary canals and perilymphatic foramina; B, slice 369, at the level of the posterior ampullae and glossopharyngeal canal; C, slice 395, at the level of the vagus nerve; D, slice 400, at the level where the vagus canal exits into the glossopharyngeal foramen; E, slice 418, showing canals for the occipital arteries and spino-occipital nerves; F, slice 423, at the level of the occipital cotylus. No scale.

Fig. 10

Dorsal view of the Tribodus limae braincase, AMNH FF 13958, anterior at top (surface rendering). No scale.

Fig. 10

Continued

Fig. 11

Ventral view of the Tribodus braincase, anterior at top (surface rendering). No scale.

Fig. 11

Continued

Fig. 12

Anterior (A) and posterior (B) views of the Tribodus braincase (surface renderings). No scale.

Fig. 12.

Continued

Fig. 13

Lateral view of the Tribodus braincase, anterior to the right (surface rendering). No scale.

Fig. 14

Medial view of the Tribodus braincase sliced through the sagittal plane, anterior to the left (surface rendering). No scale.

Fig. 15

Dorsal view of the Tribodus limae endocast with otic labyrinth removed, anterior at top (surface rendering). No scale.

Fig. 16

Ventral view of the Tribodus limae endocast with otic labyrinth removed, anterior at top (surface rendering). No scale.

Fig. 17

Anterior (A) and posterior (B) views of the Tribodus endocast with otic labyrinth removed, (surface renderings). No scale.

Fig. 18

Lateral view of the Tribodus limae endocast with otic labyrinth removed, anterior to the right (surface rendering). No scale.

Fig. 19

Dorsal view of the Tribodus endocast and skeletal labyrinth, anterior at top (surface rendering). No scale.

Fig. 20

Ventral view of the Tribodus endocast and skeletal labyrinth, anterior at top (surface rendering). No scale.

Fig. 21

Anterior (A) and posterior (B) views of the Tribodus endocast and skeletal labyrinth (surface renderings). No scale.

Fig. 22

Lateral view of the Tribodus endocast and skeletal labyrinth, anterior to the right (surface rendering). No scale.

Tribodus limae has a platybasic braincase (Fig. 10fig. 10), wedge shaped to trapezoidal in lateral view (fig. 13), in which the dorsal surfaces of the occipital and ethmoid regions slope downward to form obtuse angles with the top of the orbitotemporal region. When viewed under a light microscope, the pattern of prismatic calcified cartilage is visible on the surface of the neurocranium as a thin layer composed of dark brown tesserae, which appear rounded to polygonal in shape. The prismatic cartilage is particularly visible on the ventral surface of the braincase, and in the area surrounding the parietal ( =  endolymphatic) fossa dorsally. This prismatic pattern is also visible in many of the other cartilaginous skeletal elements in the AMNH specimens of Tribodus, except those in which the surface was abraded during preparation (e.g., the jaws of AMNH FF 13958). As in other hybodonts and neoselachians (and unlike many Paleozoic sharks), most of the skeleton is covered with only a single layer of prismatic cartilage, which wraps around to cover both outer and inner surfaces including that of the endocranial space.

In dorsal view (fig. 10) the braincase is weakly triangular, with the otico-occipital region approximately 1.5 times wider than the ethmoid region. Although the distal parts of the postorbital processes (figs. 10–13: popr) are broken off, their lateral extent can be extrapolated based on the shape of the preserved portion of each process. Based on this, it can be concluded that the widest part of the braincase was across the otic region, rather than between the postorbital processes as in Egertonodus (Maisey, 1983). The maximum width of the braincase in Tribodus is 88.5 mm, 78% of its total length of 113 mm. As in Egertonodus (Maisey, 1983), the width of the neurocranium is only slightly less than its length.

The margins of the orbits form a more complete oval than in Egertonodus basanus (Maisey, 1983), in which the broad forward-pointed postorbital processes give the orbits a more anteriorly directed appearance. A major difference between Tribodus and Egertonodus is the lack in Tribodus of an anteroposterior ventral ethmoidal keel (the “caudal internasal keel”; Maisey, 1983: figs. 9B–D: cik; this paper, figs. 29B, 33, 34: cik). Instead, the ventral surface of the ethmoid region in Tribodus is smooth and very slightly convex both anteroposteriorly and transversely, and slopes posteroventrally before joining with a median ventral process (Fig. 11fig. 11: mvpr), located between and directly posterior to the antorbital processes. Tribodus also lacks the inflated ethmopalatine processes of Egertonodus. A raised supraorbital crest is present above the orbits in Tribodus. At the widest part of the neurocranium (at the level of the otic capsules) are paired, rectangular, laterally concave structures, which are fibrous and lack prismatic calcification (fig. 3: csa). These have been identified as the points of articulation for the cephalic spines (Maisey and Carvalho, 1997). The convexly curved cephalic spine bases fit neatly within these concave, bowl-shaped platforms, with the curved distal cusps of the spines directed posteriorly. Elevated ridges on the cranial roof immediately anterior and posterior to the parietal fossa correspond to the positions of the anterior and posterior semicircular canals (cf. Notorynchus; Maisey, 2004b). As in Egertonodus basanus (Maisey, 1983) and many other hybodonts, the otico-occipital region is “telescoped” forward between the postorbital processes, so that the otic capsules are situated between these processes (unlike many Paleozoic sharks, in which the otic capsules are located behind the postorbital processes). However, in Egertonodus the occiput projects somewhat farther posteriorly than it does in Tribodus and most neoselachians (cf. Maisey, 1983).

Proportions of the Ethmoid Region

The ethmoid region of Tribodus limae is proportionally wider across than it is in Egertonodus basanus, and the olfactory capsules are more widely spaced. The left olfactory chamber of AMNH 13958 is well preserved, and although its interior is filled with matrix, it can be observed via CT scanning (fig. 6A). The right olfactory chamber is not preserved. The ethmoid region of Tribodus comprises about one-third of the entire length of the braincase, unlike Egertonodus basanus and Hybodus reticulatus, in which the ethmoid region constitutes only one-quarter of its length (Maisey 1983, 1987). Maisey (1983) describes the ethmoid region in Egertonodus as shorter than in most living sharks, but similar in length to that of Tristychius, while in Cobelodus, Tamiobatis, and Xenacanthus, the proportion is even lower (Maisey, 2005, 2007). In the extant shark Notorynchus (Maisey, 2004b), the ethmoid region comprises approximately one-third of the total braincase length, resembling the condition in Tribodus. The proportional length of the ethmoid region in Tribodus thus more closely resembles that of some extant neoselachians than that of other hybodonts and outgroup chondrichthyan taxa.

Precerebral Fontanelle, Rostral Bar, and Basicranial Process

The precerebral fontanelle of Tribodus limae is oval in cross section, and is slightly wider across than in Egertonodus basanus (fig. 3B, 6A–D). At the anteroventral tip of the precerebral fontanelle, the base of a forward-directed process is visible, although its anterior portion is missing (fig. 3B: rb). This is interpreted as a rostral bar, as in Egertonodus (Maisey, 1983). On the ventral medial surface of the rostral bar (Fig. 11fig. 11) there is a small, ventrally directed process, which appears V-shaped in frontal view. Posterior to this, the ventral surface of the internasal septum is broad and smooth, without a caudal internasal keel medially or paired ethmopalatine processes laterally such as those described in Egertonodus by Maisey (1983: fig.9: cik, ethppr).

Farther posteriorly, a large, medial, ventrally directed process is present on the ventral surface of the basicranium, originating anteriorly directly between the bases of the antorbital processes and extending posteriorly to just below the anterior margin of the optic foramen (figs. 3, 5, 6, Fig. 1111: mvpr). This process is located immediately anterior to the anterior margin of the palatoquadrate, as illustrated by UERJ-PMB-114 (fig. 5B). This median ventral process was referred to as the basicranial process by Maisey and Carvalho (1997). It is possible that the basicranial process is homologous to the caudal internasal keel of other hybodonts, due to its ventromedial position and location in the anterior region of the braincase (also originating as an outgrowth of the embryonic trabeculae), and its probable similar involvement in jaw suspension (perhaps as an attachment point for soft tissues involved in jaw support in the case of Tribodus, rather than directly abutting the palatoquadrates as in Egertonodus). In Egertonodus the caudal internasal keel extends from the tip of the rostrum to the anterior margin of the suborbital shelf. This includes the area where the ventral process is located in Tribodus, further suggesting that the two structures could be homologous. However, the basicranial process of Tribodus differs from the caudal internasal keel of Egertonodus in its much shorter anteroposterior extent. It is also possible that the median ventral process of Tribodus is homologous to a ventral interorbital septum, as in Squaliolus (Maisey, 2007), which like the process in Tribodus forms a solid wall of cartilage in the basicranium, and is located between the antorbital process and optic foramen.

The median basicranial process in Tribodus could alternatively be homologous to the keel process of some modern squalean sharks (e.g., Centroscyllium, Etmopterus; Maisey, 2007), which also typically terminates immediately anterior to the palatoquadrates and is located between the optic foramen and antorbital process. In extant elasmobranchs, the preorbitalis muscle ( =  suborbitalis of Shirai, 1992; Maisey, 2007) originates at the dorsal end of this process in Etmopterus and Centroscyllium. According to Maisey (2007), the interorbital septum of Squaliolus is probably not homologous to the keel process of the aforementioned sharks, because the latter forms later in ontogeny and differs in its ontogenetic point of origin. Because Tribodus is extinct and its ontogeny is unknown, the exact homology of its median ventral process to similar structures of modern forms remains uncertain. However, due to its location and probable function (confined to the anterior of the orbit between the antorbital processes and optic foramen; immediately anterior to palatoquadrate symphysis; not directly articulating with a forwardly extended palatoquadrate), it seems more likely that this process was homologous to the ventral orbital processes of modern elasmobranchs than to the caudal internasal keel of Egertonodus. A similar basicranial process occurs in embryos of the extant batoid Torpedo, but disappears later in ontogeny (Holmgren, 1940: fig. 162: m.a.). This process was described by Holmgren (1940: 169) as a “distinct rudiment… corresponding to the posterior part of the medial area in Etmopterus and Squalus,” and is positioned in front of the palatoquadrate symphysis. In Torpedo, this process is continuous with an anteroposteriorly elongated ridge or medial lamella (the “medial area” of Holmgren), which extends from nearly the anterior tip of the braincase to just in front of the palatoquadrate, but later disappears. Interestingly, this elongated ridge covers the same area as the caudal internasal keel of Egertonodus. The last part of the ridge to remain during ontogeny is the small, anteroposteriorly compressed median process (the “rudiment” described above), which superficially resembles the median ventral process of Tribodus. Whether the “medial area” of Torpedo is homologous to the median ventral process of Tribodus, and/or the caudal internasal keel of other hybodonts, is uncertain. However, it suggests that the two structures found in hybodonts could be homologues of each other, resulting from similar ontogenetic processes.

As in Egertonodus and Chlamydoselachus, the anteroventral margin of the precerebral fontanelle in Tribodus is defined by the slightly upturned forward part of the internasal septum (Maisey, 1983). The precerebral fontanelle is infilled with matrix, making structures on the interior walls and floor of the fontanelle difficult to examine; however, these structures can be observed with the aid of CT scanning (fig. 6A–D). In a three-dimensional image of the braincase made from CT scans of AMNH 13958, the olfactory nerve foramen is visible as a large opening in the internasal septum connecting the precerebral fontanelle with the olfactory chamber (fig. 14, I). The olfactory nerve canal is visible on the interior of the braincase as an elongated furrow on the ventrolateral wall of the ethmoid region (fig. 14: olf can).

Vascularization and Innervation of the Ethmoid Region

Allis (1923) and Jarvik (1942) noted three major groups of nerves and blood vessels in the anterior of the orbit in Chlamydoselachus: a set of dorsal and ventral foramina within the orbit, and an additional group of structures that pass superficially, lateral to the ectethmoid process (Maisey, 1983). These include the maxillary artery, the facial vein, and two nerve rami: the maxillary ramus of the trigeminal nerve, and the buccal ramus of the facial nerve (now revised as the buccal ramus of the anterodorsal lateral line nerve; Northcutt and Bemis, 1993). Based on comparison with modern elasmobranchs such as Chlamydoselachus, Maisey (1983) provided an interpretation of the innervation and vascularization of the ethmoid region of Egertonodus basanus. Based on this interpretation and on CT scan images of Tribodus limae (AMNH 13958), small, paired foramina on the ventral surface of the internasal lamina (located directly posterior to the nasal capsules and opening into them) are interpreted as anterior openings for the orbitonasal vein canals (figs. 3A, 6B, Fig. 1111: onc). These connect internally with corresponding small, ventrally located openings in the anterior of each orbit, where the orbitonasal veins presumably exited the orbit (figs. 2C, 6B–F, Fig. 1111, 13: onc). The positions of these foramina for the orbitonasal vein are consistent with those described in Egertonodus and other elasmobranchs (e.g., Chlamydoselachus and Notorynchus; Maisey, 2004b, 1983; Shirai, 1992; Allis, 1923). Allis (1923) described the orbitonasal vein in Chlamydoselachus as originating in the nasal capsule and traversing the ectethmoid chamber before entering the orbit via the orbitonasal foramen. The positions of the orbitonasal foramina in Tribodus are consistent with this description. Based on the inferred position of the orbitonasal canal, the part of the postnasal wall lateral to the orbitonasal foramen in Tribodus limae is then the planum antorbitale, or antorbital process (figs. 3A, C, 10–13: ant), originating from the lamina orbitonasalis of the embryonic trabecula, according to the interpretation given by De Beer (1931).

On the ventromedial surfaces of the antorbital processes of AMNH 13958 are small, paired, rounded structures, immediately lateral to a large notch separating the distal end of the antorbital process from the interorbital wall (figs. 3A, Fig. 1111, Fig. 1212: pr.art). The rounded shape of these paired structures, together with their lack of prismatic cartilage distally, indicate that they may have served as an attachment point for muscles or ligaments involved in jaw suspension. There are several examples of extant elasmobranchs in which the antorbital processes serve as muscle attachment sites: the suborbitalis ( =  preorbitalis, Maisey, 2007) muscle in Chlamydoselachus, hexanchoids, and the squalean Echinorhinus originates at the tip of the antorbital process, and the adductor mandibularis superficialis muscle in hexanchoids inserts on the antorbital process as well as behind the eye (Shirai, 1992). In extant carcharhinoids, the ventral head of the preorbital ( =  suborbitalis, Maisey, 2007) muscle originates in the anterior of the orbit or on the posterior of the nasal capsule (Compagno, 1988: fig. 8.1). Interestingly, small rounded processes similar in appearance to those mentioned in Tribodus occur on the medial part of the antorbital processes in embryonic Torpedo (the “short, caudally directed process” described by Holmgren, 1940: 166, figs. 157, 158). According to Holmgren, these small rounded processes act as insertion points for the musculus depressor rostri, as well as another muscle that Holmgren did not identify (possibly the preorbitalis; cf. Holmgren, 1940: fig. 157).

Immediately lateral to the paired rounded processes is a smaller notch, interpreted here as possibly housing the lateral ramule of the buccal and maxillary ramus (lr) as in Notorynchus (Maisey, 2004b). However, in Notorynchus, this ramus is more fully enclosed within the ectethmoid process (Maisey, 2004b). The maxillary artery and facial vein may also have passed through this notch, as these structures have also been described as superficially located along with the buccal and maxillary rami in extant elasmobranchs (Maisey, 1983). As these nerves and vessels generally pass lateral to the ectethmoid process, at least in extant elasmobranchs (Maisey, 1983), the small, rounded, paired processes in Tribodus described above (located medial to these structures) could be homologous to the ectethmoid processes of other elasmobranchs. Anterior to passing lateral to the ectethmoid process, it is likely that the maxillary artery and facial vein would have passed mesially into the nasal capsule via a small aperture in its lateral margin, as in Chlamydoselachus (Jarvik, 1942; Maisey, 1983). Egertonodus basanus was described by Maisey (1983) as having an ectethmoid process like that of Chlamydoselachus, in the form of a laterally directed, pointed flange extending outward from the anterior edge of the postnasal wall. Hybodus reticulatus was also reconstructed as having an ectethmoid process like that of Egertonodus, although this part of the specimen was not preserved (Maisey, 1987).

Orbitonasal Vein, Anterior Cerebral Vein, and Profundus Nerve

In the anterior of the orbit are several foramina, the ventralmost of which is considered to have contained the orbitonasal vein (see above). Dorsal to this is an additional foramen, interpreted here as possibly housing the anterior cerebral vein (figs. 6G, 13, 14: acv) as in Egertonodus basanus and Orthacanthus (Maisey, 1983). The anterior cerebral vein may have exited the orbit through this foramen, which opens into the precerebral fontanelle, and then passed along the interior edge of the fontanelle wall and into the nasal capsule via the olfactory foramen. Alternatively, the anterior cerebral vein may have passed through the profundus foramen alongside the nerve, as in some modern elasmobranchs such as Chlamydoselachus (Maisey, 1983; Jarvik, 1942), although this would leave this medial foramen (above the orbitonasal foramen and below the profundus foramen) without explanation. According to Maisey (1983), most extant sharks (except Chlamydoselachus and Squatina) have separate foramina for the anterior cerebral vein, orbitonasal vein, and profundus nerve. Unlike Egertonodus (Maisey, 1983), Tribodus lacks paired foramina in the floor of the precerebral fontanelle for the anterior cerebral vein.

Dorsal to the foramen for the anterior cerebral vein is an additional foramen, which appears to emerge on the dorsal surface of the braincase posterior to the nasal capsules (figs. 6B–F, 10, 13: prof). This foramen is here interpreted as housing the profundus nerve. As Maisey (1983) concluded to be the case in Egertonodus basanus, it seems unlikely that the profundus nerve passed through the orbitonasal canal, because this canal emerges on the ventral side of the braincase where it appears to enter the olfactory capsule (although according to Maisey, 1983, the possibility of a common foramen for the orbitonasal vein and profundus nerve in Egertonodus could not be ruled out). According to Allis (1923), in Chlamydoselachus the profundus nerve emerges on the dorsal surface of the nasal capsule. The profundus nerve in Tribodus may have exited the orbit from a separate foramen dorsal to the orbitonasal foramen, and emerged from a small opening visible on the dorsal surface of the nasal capsule in AMNH 13958. A large foramen located dorsal to the profundus foramen in the anterior part of the orbit of AMNH 13958 is here interpreted as the opening of the preorbital canal (figs. 10, 13: pre). Through this canal would have passed the orbital artery and vein and the superficial ophthalmic ramus of the anterodorsal lateral line nerve (originally referred to as the superficial or lateral ophthalmic ramus of the facial nerve; Jarvik, 1942: fig. 3: r.o.lat; Maisey, 1983: fig. 12: loph VII; Allis, 1923: 123: ramus ophthalmicus superficialis trigemini), as in Egertonodus and extant elasmobranchs (Maisey, 1983; Northcutt and Bemis, 1993).

As in Egertonodus basanus (Maisey, 1983) and Tristychius arcuatus (Dick, 1978), a basal communicating canal (rostral fenestra of Schaeffer, 1981) is absent in Tribodus limae. According to Maisey (1983), this opening is found in squaloids, Heptranchias, and Pristiophorus (Holmgren, 1941), but not in other extant elasmobranchs, and it is also absent in Xenacanthus, Tamiobatis, “Cladodus” (Schaeffer, 1981; Maisey, 1983), and “Cobelodus” (Maisey, 1983). It has been proposed that a basal communicating canal may be a synapomorphy of some modern sharks (Maisey, 1980, 1983), and its absence in Tribodus and other hybodonts appears to be a conserved feature.

An epiphyseal foramen, if one had been present, is not preserved in AMNH 13958, as the dorsal midline of the braincase has been worn away from the posterior edge of the precerebral fontanelle to approximately the middle of the orbit. However, no evidence of an epiphyseal foramen was observed in any of the other specimens of Tribodus observed for the present paper, or in those described by Brito (1992). The epiphyseal foramen in Egertonodus basanus has been described as located in the dorsal midline just posterior to the margin of the precerebral fontanelle (Maisey, 1983). In the absence of better specimens of Tribodus, it remains possible that an epiphyseal foramen may have been present and positioned similarly to that of Egertonodus. Alternatively, the epiphysis may have been located within the precerebral fontanelle, as in some extant elasmobranchs, in which case no foramen would be visible. According to Maisey (1983), a separate epiphyseal foramen is present in extant squaloids, hexanchoids, and scylliorhinids, while it is absent in Heterodontus, triakids, most carcharhinids, lamnoids, and orectoloboids. In these latter forms, the epiphysis is located near the posterior margin of the precerebral fontanelle. Maisey (1983) states that an epiphyseal foramen is also absent in Cladodoides (“Cladodus”), Tamiobatis, Orthacanthus (“Xenacanthus”), the hybodontiform shark Tristychius, and “Cobelodus,” and he suggests that the epiphysis in these taxa was probably also located toward the back of the precerebral fontanelle. In the absence of any evidence of an epiphysis in Tribodus, even with well-preserved cranial material available, it seems very possible that the epiphysis in Tribodus did not have a separate foramen, and was located within the precerebral fontanelle as in the aforementioned taxa.

Roof of the Orbitotemporal Region

The cranial roof in the orbitotemporal region (figs. 3B, 10, 12) is convexly domed dorsomedially, as in Egertonodus (Maisey, 1983). Lateral to this on either side of the cranial roof are pronounced supraorbital crests overhanging the orbits, which slope upward dorsolaterally in cross-sectional view (figs. 3B, 6C–H, 7A–H, 10: supc). Between the supraorbital crests and the dorsomedial dome of the braincase are narrow, anteroposteriorly elongated concave regions, which contain the foramina (figs. 3B, 7A, 10, 13: soph) for the branches of the superficial ophthalmic nerve complex (trigeminal and anterodorsal lateral line nerves) supplying the supraorbital sensory canals. These foramina form a single row extending anteroposteriorly, and this row forms a laterally concave curve, following the shape of the supraorbital crest. In most of the specimens examined the median dorsal roof of the braincase is not preserved anteriorly.

Optic, Trochlear, and Oculomotor Nerves

The optic foramen is a large, round opening ventral to the center of the orbit, and immediately posterior and dorsal to the median ventral basicranial process (figs. 3C, 7A, Fig. 1111, 13, 14: II). The optic nerve and artery likely passed through this foramen together, as in many other elasmobranchs (e.g., Chlamydoselachus, Allis, 1923; Cladodoides, Maisey, 2005). The optic foramen marks the boundary between the embryonic orbital and trabecular cartilages in extant elasmobranchs (El-Toubi, 1949). The foramen for the trochlear nerve (IV), which innervates the dorsal oblique eye muscle in gnathostomes, is visible on the interorbital wall near the base of the supraorbital crest, dorsal and anterior to the optic foramen (figs. 6H, 13, 14: IV). This position for the trochlear foramen (anterodorsal to the optic foramen) also occurs in Egertonodus basanus and other hybodonts, and may be of phylogenetic importance as a shared hybodont feature (Maisey, 2004a). Ontogenetically, the trochlear foramen forms within the embryonic orbital cartilage and is surrounded by it on all sides (El-Toubi, 1949).

Posterior and slightly dorsal to the optic nerve foramen is a much smaller foramen, here interpreted as housing the oculomotor (III) nerve (figs. 7B–C, 13, 14). This interpretation is further supported by CT scan images and digital reconstructions of the cranial endocast, which show a small outgrowth from the endocast at the approximate position of this foramen (figs. 16, 18), located behind the optic nerve and immediately dorsal to the anterior of the hypophyseal chamber (the location of the oculomotor nerve in Notorynchus and many other neoselachians; Maisey, 2004b). The oculomotor foramen is well preserved in UERJ-PMB-40, in which it is visible as a small foramen with an anteriorly directed opening, located posterodorsal to the optic foramen and immediately anteroventral to the prootic foramen (see section on trigeminal and facial nerves, below). The oculomotor foramen in Tribodus is located slightly anterior to its inferred position in Egertonodus basanus (Maisey, 1983: fig. 14A), in which it is identified as dorsal and slightly posterior to the efferent pseudobranchial foramen (confirmed by CT scans of Egertonodus; see section on endocranial morphology of Egertonodus basanus, below).

There is no conclusive evidence of an optic pedicel ( =  eyestalk; optic stalk) in Tribodus, although it remains possible that one could have been present in the region of the oculomotor nerve, as in many other elasmobranchs. However, there is also no evidence of an optic pedicel in any other known hybodont. Optic pedicels are known in extant neoselachians (e.g., Chlamydoselachus, Squalus, Allis, 1923; Shirai, 1992; Maisey, 2007; batoids, many squaloids and galeomorphs, Shirai, 1992), as well as extinct chondrichthyans (e.g., “Cobelodus,” Cladodoides, Maisey, 2005, 2007; Tamiobatis, Schaeffer, 1981; Synechodus, Maisey, 1985), extinct osteichthyans (e.g., Ligulalepis, Psarolepis, Basden et al., 2000; Zhu et al., 2001; Maisey, 2007) and placoderms (Young, 1986). Presence of an optic stalk may be primitive for gnathostomes, although this feature is absent in extant osteichthyans and chimaeroids (Maisey, 2005) as well as in some extant neoselachians (e.g., many carcharhinoids and the squalean Isistius; Shirai, 1992). In elasmobranchs and placoderms, the optic pedicel is typically located posterior to the optic foramen and ventral to the oculomotor foramen (Maisey, 2005). Ontogenetic studies of extant elasmobranchs such as Squalus acanthias show that the optic pedicel forms as a separate cartilage independent of the developing braincase, with which it later joins (El-Toubi, 1949). In AMNH 13958, a laterally convex, uncalcified region is present posterior to the optic foramen and posteroventral to the oculomotor foramen on both sides of the braincase, and could possibly represent the attachment site for an optic pedicel. A similar raised, rounded, uncalcified area posteroventral to the oculomotor foramen and ventral to the foramen for the superficial ophthalmic complex is also present in UERJ-PMB-40. However, in Egertonodus no evidence of an eyestalk has been reported (Maisey, 1983, 1986, 1987), and in other hybodonts (e.g., Hybodus reticulatus and Hamiltonichthys mapesi), as a result of poor preservation, not enough evidence exists to determine whether an optic pedicel was present.

Trigeminal, Facial, and Anterodorsal Lateral Line Nerves

According to Goodrich (1930), the foramen for the main branch of the facial nerve is closed off in adult gnathostomes by a prefacial commissure, a narrow strip of cartilage that extends from the basal plate to the auditory capsule. This structure separates the foramina for nerves V and VII, creating an anterior (prootic) foramen and a posterior (facial) foramen. The prefacial commissure is absent in some extant chondrichthyans (Scylliodei and several others, Shirai, 1992), and also in some osteichthyans (Goodrich, 1930).

Developmentally, in extant elasmobranchs, the embryonic fenestra prootica (an opening between the orbital cartilage and otic capsules) contains the profundus, trigeminal, and abducens nerves, part of the facial nerve, and the anterior lateral line nerves (Goodrich, 1930; Holmgren, 1940; El-Toubi, 1949; Northcutt and Bemis, 1993). Generally, the prootic foramen in adult gnathostomes contains the ophthalmic rami of the trigeminal and anterodorsal lateral line nerves, the entry of the profundus nerve into the orbit, the buccal ramus of the anterodorsal lateral line nerve, and the maxillary and mandibular branches of the trigeminal nerve (V), whereas the palatine and hyomandibular branches of the facial nerve (VII) and the anteroventral lateral line nerve emerge from the facial foramen (Goodrich, 1930; Northcutt and Bemis, 1993). This arrangement, with two foramina for the above-mentioned nerves, occurs in squaleans, Heterodontus, and some fossil outgroups (Shirai, 1992). However, according to Shirai (1992) at least three other patterns are observed in extant neoselachians. Among squaleans, a single trigeminofacial foramen is found in Echinorhinus, whereas Zameus has three separate foramina: an anterior, containing the profundus nerve and the maxillary and mandibular rami of the trigeminal nerve; a central, containing the superficial ophthalmic complex and the buccal ramus of the anterodorsal lateral line nerve; and a posterior foramen, containing the hyomandibular ramus of the facial nerve. In many galeomorphs, two foramina are present, but the more dorsal foramen contains only the superficial ophthalmic complex (Shirai, 1992).

In Tribodus limae, three foramina for the trigeminal, facial, profundus and anterior lateral line nerves appear to have been present: a moderately sized anterodorsal foramen (figs. 3C, 7 D–H, 13, 14: V oph, alln, prof), a larger foramen ventrolaterally (figs. 3A, 8A, 11, 14: V, VII), and a small foramen immediately posterior to the large foramen (figs. 11, 14: VII hm). The anteriorly directed foramen posterior and dorsal to the oculomotor foramen is here interpreted as having probably contained the superficial ophthalmic complex and profundus nerve. The larger foramen immediately posteroventral to this corresponds to the “trigemino-facial” (V–VII) foramen of previous authors (e.g., Maisey, 1983; Brito, 1992), and likely contained the main trigeminal-anterodorsal lateral line complex (formerly considered the trigeminofacial complex; revised by Northcutt and Bemis, 1993), including the maxillary and mandibular rami of nerve V, and the buccal ramus of the anterodorsal lateral line nerve. In the present paper the abbreviation “V, VII” is retained for this foramen in the figures for clearer comparison with previous works. Finally, the small posteriormost foramen likely contained the main trunk of the anteroventral lateral line and facial nerves, including the hyomandibular ramus of the facial nerve (and possibly also its palatine ramus, as in Heterodontus; Goodrich, 1930: fig. 284), and the mandibular ramus of the anteroventral lateral line nerve. This interpretation of the trigeminofacial foramina in Tribodus, if correct, indicates a noticeably different pattern from that previously described in other fossil elasmobranchs, including previous descriptions of Tribodus (e.g., Brito, 1992), and may indicate a shared feature of hybodonts. Comparison with Egertonodus (see section on endocranial morphology of Egertonodus basanus, below) lends some support to this hypothesis. The presence of a separate superficial ophthalmic foramen in hybodonts is at least superficially similar to the condition in some galeomorphs, although these forms lack an additional foramen for the hyomandibular ramus. However, variations in closure of the embryonic incisura prootica can produce at least four separate patterns of foramina in modern elasmobranchs, differing even among closely related taxa (e.g., Zameus, Echinorhinus; Shirai, 1992). Therefore, an additional pattern such as that found in Tribodus could probably be arrived at fairly easily, at least from an embryological perspective.

This interpretation of the foramina for the trigeminal, facial, profundus, and anterior lateral line nerves in Tribodus differs from that of Brito (1992), who identified the more anterodorsal foramen as the prootic foramen, as did Maisey in his description of Egertonodus (Maisey, 1983). In Tribodus, the prootic foramen (if one can be identified) would more properly be the largest of the three foramina, containing the main trigeminal-anterodorsal lateral line complex, identified as the hyomandibular foramen by Brito (1992). The prefacial commisure would then be the narrow strip of cartilage separating it from the smaller hyomandibular (facial) foramen immediately posterior to it.

However, the position of the prootic foramen in Tribodus and Egertonodus differs from that of Cladodoides (Maisey, 2005) in being located anteromedial, rather than posteromedial, to the postorbital process. Directly posterior to the dorsal edge of the superficial ophthalmic foramen in UERJ-PMB-40 is a tiny foramen whose opening is directed posterodorsally. From its opening, a narrow indented furrow is distinctly visible extending dorsally, terminating just anteroventral to one of the superficial ophthalmic foramina on the ventral side of the supraorbital crest. A similar narrow furrow is visible extending dorsally from the middle of the dorsal surface of the main prootic foramen. These furrows probably represent the paths of individual rami of the superficial ophthalmic complex.

Abducens Nerve

A small opening at the posterodorsal end of the trigeminofacialis recess is here interpreted as having contained the abducens (VI) nerve (figs. 7F–H, 13: VI). This position for the abducens foramen in Tribodus limae is supported by evidence from extant elasmobranchs, in which it is usually located in the vicinity of the prootic foramen (in the latter, the nerve typically exits either through the foramen (e.g., Heterodontus) or anterior to it (e.g., Squalus) (Goodrich, 1930: fig. 284). This interpretation is also supported by CT scan evidence, which shows that this foramen leads internally to a small canal that joins posteriorly with the larger canal leading to the superficial ophthalmic foramen. This canal originates at the base of the trigeminal-facial-anterodorsal lateral line nerve trunk farther posteriorly, a location consistent with the abducens nerve leaving the cranial cavity. The abducens nerve typically originates on the ventral side of the brain slightly mesial to the octaval and hyomandibular trunks, and then proceeds anterolaterally, following the pathway of the superficial ophthalmic complex and nerves V and VII (which typically exit through the prootic foramen) (Wischnitzer, 1993: fig. 8-2). Additional evidence in support of this conclusion is the position of this foramen just below the level of the oculomotor foramen, on a level with the presumed location where the optic pedicel would be if one were present. This position is consistent with the location of the abducens foramen in neoselachians and Cladodoides (Maisey, 2005). It is also within the trigeminofacialis recess, close to the presumed origin of the external rectus eye muscle, which it innervated. However, although the abducens foramen was identified in a similar location (anterior to the facial foramen, in the posterodorsal rim of the trigeminofacialis recess) in Egertonodus (Maisey, 1983), CT scans indicate that this foramen in Egertonodus probably housed the pituitary vein instead. It is possible that the abducens nerve in Egertonodus was contained within the prootic foramen, as in extant Heterodontus.

Alternatively, the fact that the above-mentioned foramen is internally connected with the superficial ophthalmic foramen could suggest a different interpretation in which this foramen is a branch of nerves V, VII, or the anterodorsal lateral line nerve (such as the mandibular ramus of nerve V), or perhaps a separate exit for the profundus nerve. An interpretation of this foramen as the profundus foramen may be supported by CT scan images, which could instead show the ophthalmic rami of the trigeminal and anterodorsal lateral line nerves diverging into two separate canals as they leave the cranial cavity (as in extant neoselachians such as Squalus, Wischnitzer, 1993: fig. 8-2) and exiting separately from the braincase. The larger canal (interpreted as housing the superficial ophthalmic complex and its associated nerves) exits through the dorsal foramen, and the smaller canal (possibly the profundus nerve) exits through the more posteroventral foramen, located in the rear of the trigeminofacialis recess. This alternative interpretation is further supported by the fact that the nerve trunk that diverges into these two canals is positioned anteriorly on the cranial endocast with respect to the large trigeminofacial-anterodorsal lateral line nerve trunk that branches from the endocast laterally, and appears to correspond to the location of the superficial ophthalmic, infraorbital, mandibular, and profundus nerves in extant neoselachians such as Squalus (Wischnitzer, 1993: fig. 8-2). A separate profundus foramen is also present in Cladodoides, located anteroventral to the prootic foramen (Maisey, 2005), and one may also have existed in Orthacanthus (Schaeffer, 1981). This arrangement also occurs in chimaeroids (e.g., Callorhinchus: De Beer, 1937: pl. 22). However, a separate profundus foramen was not reported in Egertonodus (Maisey, 1983), and is absent in most neoselachians (Maisey, 2005). Also, this foramen in Tribodus is located posteroventral to the prootic foramen, rather than anteroventral to it as in Cladodoides (see Maisey, 2005).

The latter interpretation would also necessitate a different location for the abducens foramen. One alternative is that the abducens nerve exited the braincase through the prootic foramen as in Heterodontus (and possibly Egertonodus). It is additionally possible that a small foramen located directly posterior and ventral to the prootic foramen (discussed above, and identified as the hyomandibular foramen) is instead the abducens foramen. CT scans of the interior of the braincase show that this foramen originates on its ventrolateral wall (dorsal to the dorsum sellae and about halfway along its length), and is located between and ventral to two larger foramina, here interpreted as the anterior and posterior rami of the octaval (VIII) nerve (see section on octaval, glossopharyngeal, vagus, and spino-occipital nerves, below). This interpretation would be supported by the fact that the abducens nerve exits the braincase in this position in Cladodoides (Maisey, 2005: figs. 7, 17). Also, according to Maisey (2005), having the abducens foramen located below and behind the dorsum sellae may be a primitive gnathostome feature. This feature is also found in chimaeroids and some neoselachians, and may be primitive for chondrichthyans (Maisey, 2005). However, in Cladodoides this foramen is level with the eyestalk (as is the abducens nerve in neoselachians), whereas in Tribodus this more ventral foramen is not on a level with the region where an eyestalk typically would be found. Also, in Cladodoides this position is within the orbit, ventral to the prootic foramen and anterior to the hyomandibular foramen, whereas in Tribodus the aforementioned foramen is posterior to the presumed hyomandibular foramen and far posterior to the “prootic” foramen (interpreted here as the superficial ophthalmic foramen, as discussed above). Because the abducens nerve in gnathostomes innervates the external rectus eye muscle, it seems unlikely that it would emerge this far back from the trigeminofacialis recess, where that muscle would have originated. The foramen in the posterodorsal rim of the trigeminofacialis recess is thus here considered more likely to be the abducens foramen, and the more posterior foramen is interpreted as the hyomandibular (facial) foramen.

Efferent Pseudobranchial Artery

A foramen located centrally within the trigeminofacialis recess in Tribodus and near its ventral margin is interpreted here as the foramen for the efferent pseudobranchial artery (figs. 7D, 13, 14, 24: epsa). Maisey (1983) identified the efferent pseudobranchial artery as occupying a similar location in Egertonodus basanus, also in the midventral part of the trigeminofacialis recess. However, CT scans show that the efferent pseudobranchial foramen in Egertonodus was located somewhat farther anteriorly (see section on Egertonodus endocranial morphology, below). As in Tribodus, smaller foramina occur anterodorsally and posterodorsally within this recess in Egertonodus. The more anterior of these foramina in Egertonodus has been interpreted as the oculomotor foramen, as in Tribodus (Maisey, 1983). The more posterior foramen in Egertonodus was interpreted by Maisey (1983) as the abducens foramen. It is here interpreted that the efferent pseudobranchial and ophthalmic arteries in Tribodus diverged external to the braincase, after exiting together through a common foramen (as in many extant neoselachians).

This location for the efferent pseudobranchial foramen is supported by several factors. First, because the efferent pseudobranchial foramen marks the anterolateral margin of the embryonic polar cartilage in elasmobranchs, whereas the pituitary vein foramen marks its posterolateral end, it is generally considered that the efferent pseudobranchial artery must be located anterior to the pituitary vein (although during a brief developmental stage in embryonic Torpedo, the two vessels share a common foramen; Holmgren, 1940). This is certainly the case in Tribodus, in which the efferent pseudobranchial artery is located well in front of the dorsum sellae and pituitary vein. Second, CT scan images reveal paired foramina (fig. 14: ica) in the floor of the cranial cavity, several millimeters anterior to the anterior tip of the dorsum sellae and slightly anterior to the foramen just described. CT scans indicate that the internal carotid arteries entered the cranial cavity through these foramina, after entering the braincase farther posteriorly and tunneling through the basicranial cartilage (see section on lateral dorsal aortae, internal carotid arteries, hypophyseal foramen, and pituitary vein, below).

Facial Nerve and Acoustico-Trigeminofacialis Recess

Brito (1992) identified a large foramen posteroventral to his “prootic” foramen on the exterior of the braincase as representing the foramen for the hyomandibular ramus of the facial nerve (fig. 2C: VII hym). Brito's “prootic” foramen is here interpreted as housing the superficial ophthalmic complex, and the larger, more posterior foramen is interpreted as the main trigeminofacial (V, VII) foramen (which probably contained the maxillary and mandibular rami of the trigeminal nerve, and the buccal ramus of the anterodorsal lateral line nerve, in Tribodus, following cranial nerve terminology of Northcutt and Bemis, 1993). Additional support for this hypothesis is provided by CT scans and digital images of the interior of the braincase. These scans show that this foramen (fig. 14: V, VII) opens internally in the anterior part of a large recess (interpreted as the acoustico-trigeminofacialis recess), immediately posteroventral to another foramen that leads anteriorly to the superficial ophthalmic foramen (fig. 14: V oph, alln, prof) and anterior to a larger foramen interpreted as the anterior ramus of the octaval nerve (fig. 14: VIIIa). This position is consistent with the location of the trigeminofacial foramen in many neoselachians. In addition to the superficial ophthalmic and trigeminofacial foramina, the acoustico-trigeminofacialis recess contains two additional large foramina posteriorly, both of which exit the cranial cavity and enter the otic capsule. These foramina likely contained the anterior and posterior branches of the octaval (VIII) nerve (see section on endocranial space, cranial endocast, and skeletal labyrinth of Tribodus, below).

An additional, smaller foramen in the region of the acoustico-trigeminofacialis recess, located between and immediately ventral to the trigeminofacial-anterodorsal lateral line and octaval foramina, is here interpreted as the hyomandibular (facial) foramen (fig. 14: VII hm). However, it is alternatively possible that this foramen housed the abducens nerve, as discussed above, or perhaps the palatine ramus of the facial nerve (unlikely, because this is usually directed anteriorly, not posteriorly, and generally diverges from the hyomandibular ramus external to the cranial cavity). The foramen arises in the correct location on the floor of the cerebellar region to be consistent with the abducens nerve (above and behind the anterior of the dorsum sellae; slightly medial to the acoustico-trigeminofacialis recess). The abducens foramen is located at this position in Squalus (Marinelli and Strenger, 1959), Cladodoides (Maisey, 2005), and “Cobelodus” (Maisey, 2007). However, the abducens nerve usually exits the braincase farther anteriorly, in the region of the external rectus eye muscle, which it innervates (Shirai, 1992).

In Cladodoides, the palatine ramus of the facial nerve diverges to form separate anterior and posterior ramuli, which exit through the floor of the braincase immediately below the base of the postorbital processes (Maisey, 2005). Although in Cladodoides the rami of the facial nerve diverge after exiting the braincase, it is possible that during ontogeny in Tribodus they could have been individually enclosed in cartilage to form separate foramina. In Squalus, there are separate foramina for the palatine and hyoid rami of nerve VII in the wall of the braincase (Goodrich, 1930: fig. 284). In any case, due to its position and relationship to the intracranial space (opening into the acoustico-trigeminofacialis recess), it seems fairly certain that this foramen was for a nerve ramus (probably of nerves V, VII, or the anterior lateral line nerves), rather than for an artery or vein.

Lateral Dorsal Aortae, Internal Carotid Arteries, Hypophyseal Foramen, and Pituitary Vein

In Tribodus limae (UERJ-PMB-40 and AMNH 13961), paired foramina are visible immediately dorsal to the paired ventral processes in posterior view (located on the ventromedial side of the base of each process, figs. 11, 23: lda). They are especially well preserved in UERJ-PMB-40 (fig. 23). These probably represent the points of entry of the lateral dorsal aortae into the braincase.

If the lateral dorsal aortae entered the braincase here and continued anteromedially within the basicranial cartilage, then the median basicranial foramen is not an entry point for the internal carotids, and requires further explanation. The most obvious explanation is that this latter opening is a hypophyseal foramen. This explanation is supported by CT scans and digital reconstructions of the braincase, which show that this foramen opens internally into a large cavity posteroventral to the main endocranial space (in the rear of the hypophyseal chamber) (figs. 8E, F, 14). This cavity is connected anteriorly with the endocranial space by a narrow medial canal (figs. 8C, Fig. 1212, 16, 18: hd), indicating that it may represent the posterior lobe of a bipartite hypophysis as in some extant elasmobranchs (Daniel, 1934: 233). According to Maisey (1983), an open hypophyseal duct occurs in Egertonodus basanus, and also in Xenacanthus, Tamiobatis, “Cladodus,” and Cladoselache, but is lacking in adult neoselachians.

A small foramen located beneath the anterior part of the dorsum sellae and visible in digitally reconstructed sagittal sections of the braincase likely represents the interorbital canal, through which the pituitary vein passed to exit the braincase laterally through a small foramen ventromedial and posterior to the opening for the facial and anteroventral lateral line nerves (figs. 8B, 14, 16, 24: pit v). This location for the pituitary vein is in keeping with the typical elasmobranch arrangement, in which the pituitary vein foramen is located posterior to the efferent pseudobranchial foramen and marks the boundary between the embryonic polar cartilage and the parachordal cartilage posteriorly.

Efferent Hyoidean and Orbital Arteries

In Tribodus limae, anterior to the paired ventral processes and posterolateral to the hypophyseal foramen are two additional, larger pairs of foramina, best preserved in UERJ-PMB-40 (fig. 24). The more posterior pair of foramina exit the floor of the braincase posterolateral to the hypophyseal foramen, and open into the anterior part of an elongated, posterolaterally extending groove (fig. 24: eha). These foramina are here interpreted as housing the efferent hyoidean arteries, which passed posterolaterally from the lateral dorsal aortae to supply the hyoid arch. The more anterior pair of foramina is directed anterolaterally, and CT scan images of AMNH 13958 show that both these and the aforementioned pair of foramina is connected internally to the posterior foramina dorsal to the paired ventral processes. Based on these considerations, the anterolateral pair of foramina are here interpreted as representing the exits of the orbital ( =  stapedial) arteries from the braincase (fig. 24: ora). Narrow grooves in the cartilage extending anteriorly from these foramina toward the orbit further support the interpretation that the passage for the orbital arteries was located here. However, this interpretation is unusual in comparison with the typical pattern of cranial circulation in other elasmobranchs, and requires further explanation (see Discussion). It is unlikely that these foramina contained any cranial nerve rami, because CT scans do not show them connecting internally with any nerve passages, and they are located beneath the dorsum sellae where no cranial nerves originate. The only possible alternative is that these paired lateral foramina could represent the interorbital canal containing the pituitary vein. However, CT scans indicate that the pituitary vein was probably located slightly farther anteriorly, in front of the narrow canal leading to the posterior chamber of the hypophysis (as described above), and the presence of external foramina in that location interpreted as the pituitary vein foramina in UERJ-PMB-40 further support that hypothesis. Additionally, because the pituitary vein typically joins the lateral head vein anterior to the postorbital processes (through which the latter vein passes on its way to the orbit), the large lateral foramina mentioned previously may well be located too far posteriorly to represent the pituitary vein.

In UERJ-PMB-40 (fig. 4), the posteroventral part of the braincase is well preserved, including both of the paired ventral processes. In this specimen, small, paired foramina are present just posterior to the occiput in ventral view, at the same level as the base of the small pointed posterior processes located immediately lateral to the occiput. Similar paired foramina are also visible in UERJ-PMB-114 and AMNH 13961. Although it may appear at first glance that the paired lateral dorsal aortae entered the braincase here, as previously reconstructed in Egertonodus (Maisey, 1983), CT scan slices of Tribodus show no evidence of paired canals for the lateral dorsal aortae passing within the basicranial cartilage in the occipital region. However, CT scans do reveal the presence of paired narrow canals extending dorsomedially within the occipital region from these paired foramina, and opening dorsally into the posteromedial floor of the glossopharyngeal-vagus foramen on both sides of the braincase (figs. 9E, F, 11, 15–18: oc art). These canals may represent occipital arteries branching from the main trunk of the lateral dorsal aorta ventrally and entering the cartilage of the braincase to supply structures located farther dorsally. The paired ventral foramina just anterior to the occiput in Tribodus are thus interpreted here as entry points for the occipital arteries into the braincase. CT scans of Egertonodus show that its corresponding set of paired foramina also open into a similar pair of dorsally extending canals that also enter the glossopharyngeal-vagus foramen, changing the previous interpretation and indicating that at least two hybodonts (Tribodus and Egertonodus) have this arrangement of the occipital arteries and their foramina. Similarly positioned foramina for the occipital arteries have also been reported in the Carboniferous (Viséan-Namurian) hybodontiform Tristychius arcuatus (Dick, 1978). In Tristychius arcuatus, according to Dick (1978: 80; fig. 2), the ventral foramina of the occipital arteries are located within anteroposteriorly elongated, narrow indents for the “common carotid arteries” or lateral dorsal aortae. This is very similar to the arrangement found in Egertonodus (see Discussion), in which similar indents for externally located lateral dorsal aortae also contain the occipital artery foramina. In Tribodus, no external indents for the lateral dorsal aortae are visible, but it is likely that they traveled exterior to the braincase in the occipital region, connecting with the occipital arteries in an arrangement similar to that of Egertonodus and Tristychius. In Tristychius, the occipital arteries emerge dorsolaterally posterodorsal to the vagus nerve foramina (Dick, 1978).

Postorbital Processes

The postorbital processes in Tribodus limae (figs. 3A, C, 11–13: popr) are long, narrow, and tapered toward their distal ends; and are generally much more slender and gracile than those of Egertonodus (Maisey, 1983). Most of the left postorbital process is preserved in Tribodus limae, AMNH 13958 (although its distal part is missing), whereas the right postorbital process is not preserved except for its base. This is typical of most fossil hybodont specimens, in which the postorbital processes are often broken off or missing, and may reflect that these processes were only weakly attached to the braincase (Maisey, 1982, 1983). The postorbital processes originate on the lateral wall of the otic capsule, and extend anteroventrally at an angle of approximately 45° to the horizontal aspect of the braincase, defining the posterior margin of the orbit. These processes also extend farther laterally than the antorbital processes. Unlike Egertonodus, the widest part of the braincase in Tribodus is not the postorbital processes but the otic region farther posteriorly, at the point where the cephalic spines articulated. A large fenestra in the proximal part of the postorbital process is interpreted here as the jugular canal (figs. 3A, 8B, Fig. 1111, Fig. 1212: jc), through which passed the jugular or lateral head vein, and possibly also the hyomandibular ramus of the facial nerve (after exiting the braincase through the facial foramen). The supraorbital crest ends dorsal to the anterior margin of the base of the postorbital process, and is separated from it by a rounded notch.

According to Maisey (2008) and Holmgren (1940), the postorbital process in chondrichthyans forms ontogenetically from two structures: a primary process (which is a posterolateral outgrowth of the supraorbital shelf), and the lateral commissure (fused in part with the side wall of the otic capsule, and fused dorsally either with the otic capsule to enclose the hyomandibular trunk of the facial nerve, or with the ventrolateral margin of the primary process to enclose the lateral head vein). Based on the structure and position of the postorbital process in Tribodus, it is concluded here that it was formed at least in part by the embryonic lateral commissure, which was chondrified in the adult. This interpretation is based on the fact that the postorbital process in Tribodus is fused dorsally and medially with the otic capsule and encloses the jugular canal, which presumably housed the lateral head vein. In the same way, a chondrified lateral commissure was probably also present in Egertonodus, as was suggested previously by Maisey (1983).

A postorbital articulation was absent in most hybodontiforms (including Tribodus), whose jaw suspension is hyostylic (except for Tristychius, which had a postorbital articulation; Dick, 1978; Maisey, 2008). Thus, unlike other fossil chondrichthyans with a lateral commissure (e.g., Cladodoides, Maisey, 2005; “Cobelodus,” Maisey, 2007), the postorbital process in most hybodonts lacks an articular facet for the palatoquadrate.

Octaval, Glossopharyngeal, Vagus, and Spino-Occipital Nerves

CT scan images of the interior of the braincase show two large foramina located posterior to the foramen for the hyomandibular ramus of the facial nerve, both of which exit into the otic capsule. These are here interpreted as having housed the anterior and posterior rami of the octaval nerve, respectively (figs. 8B, D–G, 14: VIII a, p). Posterior to the octaval nerve foramina and located ventrolaterally on the endocranial wall is a much smaller foramen (fig. 14: IX). This foramen is here interpreted as having housed the glossopharyngeal nerve, and joins the glossopharyngeal-vagus canal at a location ventral to the preampullary canal. It is positioned similarly to the glossopharyngeal nerve of neoselachians (e.g., Squalus, Marinelli and Strenger, 1959; Gegenbaur, 1872). After exiting the cranial cavity, the glossopharyngeal nerve traveled through a narrow canal beneath the otic capsule, which widens posteriorly (figs. 15, 16, 17B, 18: IX). This canal exits on the posterior of the braincase together with the vagus nerve through a single, shared foramen lateral to the foramen magnum on either side (figs. 3B, 9D–F, 10, 12B: IX, X). Each of these shared foramina is located immediately anterior to the hyomandibular articulation, and is oval in outline.

In sagittal view, foramina for rami of the vagus nerve are visible as two small openings in the mesial wall of the braincase, posterior to the glossopharyngeal foramen (fig. 14: X). They are located beneath the domed roof of the medullary chamber, similar to the position of the internal vagus foramen in neoselachians (e.g., Notorynchus, Maisey, 2004b), and are arranged one above the other. These foramina open into canals that exit the braincase through the glossopharyngeal-vagus foramen.

Roof of the Otic Region

The otic region of Tribodus limae (AMNH 13958) is relatively complete, although the right postorbital process is missing. As in Egertonodus basanus, the paired otic capsules in Tribodus constitute approximately one-third of the total length of the braincase. The parietal (endolymphatic) fossa is large, elongated, and ovate, and is located centrally between the otic capsules (figs. 3B, 10: p fos). An anterior pair of small, rounded foramina located laterally within the parietal fossa represent the endolymphatic foramina (figs. 10, 14: end f), and a more posterior pair represent the perilymphatic fenestrae (figs. 9A, 14: pf). Other than these two pairs of foramina, the parietal fossa has a continuous cartilaginous floor, probably formed by an embryonic taenia medialis, as in extant elasmobranchs (Holmgren, 1940) (fig. 8B: t med). In extant elasmobranchs, this structure can be produced either as a posterior outgrowth from the synotic tectum (e.g., Squalus), or as an anterior outgrowth from the posterior tectum (where one is present, e.g., Scyllium) (Holmgren, 1940). Paired elevated ridges in the cartilage located laterally, anterior and posterior to the parietal fossa, indicate the location of the anterior and posterior semicircular canals, respectively. The ridge overlying the anterior semicircular canal extends anterolaterally at an angle of approximately 35° from the midline of the specimen, while the posterior ridge extends laterally at about 45° to the midline.

As discussed previously, paired rectangular laterally concave areas located on the lateral part of the otic region represent the points of articulation for the cephalic spines (Maisey and Carvalho, 1997), indicating that AMNH 13958 was a male (female specimens of Tribodus lack cephalic spines, as well as the paired platforms for their articulation; the posterolateral part of the braincase is instead smoothly rounded). The convex platforms for the cephalic spines in AMNH 13958 lack a covering of prismatic cartilage, and appear fibrous. A photograph of the original specimen immediately after acid preparation shows that the cephalic spines sat in these convex platforms with their cusps directed posteroventrally (fig. 25: csp). Posteromedial to the cephalic spine platforms are two lateral processes, here interpreted as lateral otic processes as in Egertonodus (Maisey, 1983) (figs. 3B, C, 10: lotpr). It has been suggested that the hyomandibula articulated on these processes (Maisey and Carvalho, 1997). This interpretation is supported by the new material of Tribodus, which shows a rounded concave notch for the dorsal head of the hyomandibula on the medial surface of the lateral otic process on each side of the specimen, between the lateral otic process and a small, triangular, posteriorly directed process posteromedial to it (figs. 3A, 11, 12: hm art). The hyomandibular articulation is posterior and lateral to the glossopharyngeal-vagus foramen, and the lateral otic process is located entirely lateral to this foramen. Paired posterior triangular processes are located immediately posterior to the glossopharyngeal-vagus foramina, flanking the occiput laterally.

Fig. 25

Close-up of the head region of Tribodus limae (AMNH 13959) showing right cephalic spine in situ, after acid preparation. Anterior to the left. Scale bar is 1.5 cm.

Cephalic Spines

Cephalic spines of Tribodus limae have not previously been described or illustrated in any of the published literature (e.g., Brito and Ferreira, 1989; Brito, 1992), and the present paper represents the first description of the cephalic spines in this taxon (figs. 25–28). Spines are preserved in AMNH 13957 (figs. 27, 28A–C) and 13959 (figs. 25, 28D–E), and in UERJ-PMB-40. Cephalic spines of Tribodus limae are unicuspid, with a single, large, posteriorly recurved principal cusp and no accessory cusps (fig. 25–28). In AMNH FF 13959 (an exceptionally well-preserved right cephalic spine associated with a fragmented cranium and partial scapulocoracoid), the spine base is constricted and triradiate, with paired anterior lobes (medial and lateral) and a narrow posterior lobe (figs. 25, 28D–E). The cephalic spine base is also well preserved in UERJ-PMB-40, although the spine cusp is missing (fig. 26). When viewed from above, the spine base can best be described as anchor shaped. All three basal lobes are of approximately equal length and are slender and elongate in shape. The medial (fig. 26: lm) and lateral (fig. 26: ll) lobes taper distally, whereas the distal part of the posterior lobe (fig. 26: lp) first narrows, then expands into a broader rounded distal tip. The medial and lateral lobes are posteriorly recurved, so that the entire spine when viewed from above (e.g., lateral view, if the spine were in situ) appears nearly symmetrical on both sides. The base of the cephalic spine is composed of a porous, spongy tissue, presumably osteodentine, whereas the cusp is covered with a thin, glossy layer of enameloid, as in most other hybodonts (Maisey, 1982, 1983).

Fig. 26

(A) top, (B) bottom, and (C) lateral views of left cephalic spine of Tribodus limae (UERJ-PMB-40). Scale bar is 1 cm.

Fig. 27

SEM of Tribodus limae cephalic spine, AMNH 13959. Scale bar is 1 mm.

Fig. 28

Tribodus limae cephalic spines. A–C, AMNH 13957, left cephalic spine. A, lateral; B, dorsal; C, mesial views. D–E, AMNH 13959, right cephalic spine. D, mesial; E, lateral views. Scale bar is 1 cm.

The cephalic spine cusp in Tribodus is slender but not acuminate, in that it shows a distal barb apically (fig. 27: ba). The proximal part of the cusp is strongly ornamented with a series of vertical ridges anteriorly, some of which bifurcate toward the cusp base. These ridges converge approximately halfway up the spine cusp, where they join a sharp, narrow ridge, located slightly off center, which extends from here to the spine apex. The posterior and laterodistal sides of the spine cusp are unornamented except for a few faint parallel ridges on the posterior proximal end of the cusp, just above the spine base. In AMNH FF 13957, a left cephalic spine associated with a nearly complete fish, most of the spine base is lacking, showing the presence of a large pulp cavity in the interior of the cusp. In side view, all three lobes of the spine base slope downward posteroventrally. In Tribodus, the posterior lobe of the spine base extends slightly farther posteriorly than the spine apex. The distal part of the cusp is approximately horizontal in side view.

As originally preserved (prior to preparation), the cephalic spine of AMNH FF 13959 was located nearly in situ, allowing its approximate life position to be determined (fig. 25). This specimen shows that the cephalic spine in Tribodus limae was located on the lateral surface of the otic region just behind the postorbital process, in a broad, rectangular, concave platform (best preserved in AMNH FF 13958). Its location most closely approximates that of the cephalic spines located above the lateral otic process in Egertonodus basanus, Egertonodus fraasi, Hybodus hauffianus, and H. delabechei, although in Tribodus the spine platform is located much farther ventrally, possibly representing a ventral elongation of the lateral otic process. There is no evidence of more than a single pair of cephalic spines in Tribodus, as is also the case in Egertonodus basanus.

Cephalic spines of Tribodus limae differ from those of Hybodus, Egertonodus, Acrodus, and Asteracanthus (e.g., genera having “Sphenonchus”-type cephalic spines) in that the latter taxa have a much thicker and more robust spine base, in which the lobes are much less constricted. However, Tribodus cephalic spines are similar to spines of these taxa in having a barbed tip, and a dorsal, mesial, lateral, and posterior crest (see Maisey, 1982, 1987). They also differ in that the spine base in Tribodus is more bilaterally symmetrical than in these taxa. Tribodus cephalic spines are morphologically very different from those of Hamiltonichthys mapesi, which has an unconstricted spine base, multiple accessory cusps, and no ornamentation (Maisey, 1989). They also differ considerably from cephalic spines of Onychoselache, which have more than one principal cusp and a broad, unconstricted spine base.

Foramen Magnum and Occiput

Situated directly between the paired glossopharyngeal-vagus foramina on the posterior of the braincase is the foramen magnum, through which the spinal cord exited the braincase (figs. 3B, 10, 12B, 13: fm). The foramen magnum is a rounded opening, slightly elongated dorsoventrally. Above the foramen magnum is a dorsally extending occipital crest, which terminates at the posterior margin of the parietal fossa, representing the point of closure of the occipital arch during ontogeny. A similar crest is seen in Notorynchus (Maisey, 2004b). The occipital cotylus, located directly ventral to the foramen magnum (figs. 3A, 9C–F, Fig. 1011, 12B, 13: oc cot), is fibrous and has a concave posterior surface, as in Egertonodus (Maisey, 1983). The lateral sides of the occiput are fully formed, but the dorsal and ventral sides are not well developed and the occipital cotylus opens dorsally into the foramen magnum. Like other hybodonts (such as Egertonodus, Maisey, 1983), and unlike the condition in extant neoselachians, there is no occipital hemicentrum in Tribodus (and this form also lacks calcified vertebral centra, as do other non-neoselachian elasmobranchs). The occipital region in Tribodus is broader than in Egertonodus, and the occiput does not extend as far posteriorly behind the otic capsules as it does in Egertonodus. This feature of Tribodus is similar to many modern sharks, in which the occiput is also situated between the otic capsules (Maisey, 1983).

Cranial Endocast and Endocranial Space

A generalized description of endocranial morphology in Tribodus was provided by Maisey (2004a) as part of a comparative study of elasmobranch neurocrania. However, a detailed description of endocranial morphology in Tribodus has not yet been provided, and a reconstruction of the cranial endocast in Tribodus (or any other hybodont) has not been presented until now. The cranial endocast of Tribodus is similar in many respects to that of some extant neoselachians such as Notorynchus, particularly in dorsal and lateral views. As in Notorynchus and Squalus, the telencephalic chamber broadens anteriorly, with proportionally thick olfactory canals and widely spaced olfactory chambers (figs. 15, 16: tel). This condition is unlike that of Cladodoides (Maisey, 2005), in which the telencephalic region is more narrow and uniform in width from side to side, and actually narrows anterior to the cerebellar chamber. It is also considerably different from “Cobelodus” (a tropibasic form), in which the endocast is extensively foreshortened and the telencephalic region also tapers anteriorly in dorsal view. As Maisey (2004a) also pointed out, in both Tribodus and Egertonodus the floor of the telencephalic chamber above the median ventral process is flat to slightly convex and does not follow the shape of the underlying external morphology (fig. 14), a feature shared with neoselachians. As noted by Maisey (2004a), no constriction between the telencephalic and mesencephalic regions (such as that found in Squalus) is present in Tribodus. Also, the myelencephalic (medullary) chamber of Tribodus is domed as in Notorynchus and Squalus, as pointed out by Maisey (2004a). The vagus (X) nerve exits the cranial cavity in a similar location to that of Notorynchus and Squalus, near the center of the medullary chamber (figs. 14, 15, 17, 18: X).

Maisey (2004a) noted that in both Tribodus and Egertonodus, the optic foramen is positioned centrally within the orbit and the trochlear foramen is located anterior and dorsal to the optic foramen, which he describes as an unusual configuration for elasmobranchs (in which the trochlear foramen is typically dorsal, but posterior to the optic foramen). This is confirmed by CT scans and digitally reconstructed endocasts of both Tribodus and Egertonodus that show a lateral outgrowth in the correct position on the endocast to correspond to the trochlear nerve (figs. 15, 16, 18: IV). The optic artery also probably exited the braincase through the optic foramen, alongside the optic nerve.

Posterodorsal to the optic foramen on the Tribodus endocast is a smaller outgrowth, located directly above the anterior part of the infundibular region. This is in the correct position to correspond to the oculomotor nerve of many extant neoselachians (figs. 16, 18: III). It exits the braincase through a small foramen immediately anterior to the trigeminofacialis recess. Ventral to this, the infundibular region extends posteriorly to form a hypophyseal chamber (figs. 8A, B, 12A: hyp), visible in ventral and lateral views of the endocast (figs. 16, 18). At the rear of the infundibular/hypophyseal region immediately anterior to the dorsum sellae are paired lateral extensions, here interpreted as the pituitary veins (fig. 16: pit v), and comparable to those of Cladodoides (Maisey, 2005). Posterior to this, the hypophyseal chamber constricts to form a narrow medial tube, which joins posteriorly with a large chamber beneath the dorsum sellae (figs. 16, 18: hd). A median ventral extension from this chamber (fig. 16: hf) may represent either an open hypophyseal duct or a foramen for the internal carotid arteries, although the former interpretation is considered here more likely (see above discussion). From the anterolateral sides of this chamber, paired elongated tubes extend forward, joining the cranial endocast posterior to the optic nerve and below the oculomotor nerve (figs. 16, 18: ica). These paired canals were probably for the internal carotid arteries, which tunnel through the basicranial cartilage before entering the braincase and joining two additional vessels, the efferent pseudobranchial/ophthalmic arteries and the optic arteries (in order anteriorly). From the lateral sides of this chamber extend two additional sets of openings that exit the braincase laterally. Of these, the anterior pair of openings is here interpreted as the orbital arteries (figs. 16, 18: ora), and the posterior pair is interpreted as the efferent hyoidean arteries (fig. 16: eha). From the posterior part of this chamber, paired canals extend posterolaterally (figs. 16, 18: lda). These are here considered to represent the lateral dorsal aortae, which entered the braincase through paired posterior foramina (at the base of the paired ventral basicranial processes) and traveled anteriorly to enter the aforementioned chamber. They then joined the orbital arteries, and continued anteromesially as shortened internal carotid arteries before joining inside the chamber and dividing again to progress anteriorly.

An acoustico-trigeminofacialis recess (De Beer, 1937: 56) is visible on the internal lateral wall of the sagittally sectioned half braincase of Tribodus (fig. 14), and a corresponding lateral extension appears on each side of the cerebellar region in the endocast (figs. 15, 16, 18). This recess contained the proximal trunks of the trigeminal, facial, anterior lateral line, and octaval nerves. According to De Beer (1937) and Allis (1914), this space is intramural (mesial to the wall of the braincase) in modern Squalus but not intracranial, because it is located external to the dura mater, which may have also been the case in Tribodus. In the anteriormost part of this recess on the endocast is a forward-directed canal. This canal divides immediately anterior to exiting the cranial cavity, forming a large dorsal canal and a narrower ventral canal. The larger canal is here interpreted as having housed the superficial ophthalmic complex, and possibly also the profundus nerve (figs. 15, 16, 18: V oph, alln, prof).

The smaller canal may represent the abducens nerve (figs. 16, 18: VI), originating on the ventral surface of the brain and extending anterodorsally to exit the braincase into the orbit in the vicinity of the trigeminofacialis recess. It may alternatively represent the profundus nerve (discussed above); however, this interpretation is considered less likely because the abducens nerve would then have to exit the braincase farther posteriorly, away from the external rectus muscle, which it innervates. Posteroventral to the superficial ophthalmic and abducens canals is a second lateral extension from the acoustico-trigeminofacialis recess, interpreted as the main trunk of the trigeminal and anterodorsal lateral line nerves (which exit through the trigeminofacial foramen; figs. 15, 16, 18: V, VII). Two larger lobes immediately posterior to this on the endocast are here considered to be the anterior and posterior branches of the octaval nerve, respectively (figs. 15, 16: VIII).

Posterior to the acoustico-trigeminofacialis recess and hypophyseal chamber, the dorsal surface of the endocast slopes downward, forming a saddlelike depression into which the skeletal labyrinth fits (figs. 15, 18, 19, 21, 22). Posterior to this, the medullary chamber again expands dorsally, forming a rounded dome (figs. 14, 15, 18). The notochord is visible as an elongated median extension from the ventral surface of the medullary region (figs. 16–18: nt). Dorsolateral to the notochord, at the anterior end of the medullary dome, a small, paired extension from each side of the endocast represents the glossopharyngeal nerve (figs. 15, 16, 18: IX). Posterior to these, two additional pairs of canals extending from the dorsal and lateral sides of the medullary dome represent the vagus nerve (figs. 15, 17B, 18: X). These canals extend posterolaterally and exit the braincase posteriorly together with larger, more laterally located canals (emerging from the ventral sides of the skeletal labyrinth, and containing the glossopharyngeal nerves) inside the large glossopharyngeal-vagus foramina. However, although the two nerves share a common foramen to exit the braincase, they do not share a single canal opening into this foramen, and instead enter the lumen of the glossopharyngeal-vagus foramen through separate smaller foramina. Posteromedial to the glossopharyngeal-vagus foramina, the medullary chamber narrows, terminating posteriorly at the foramen magnum. At least three pairs of spino-occipital nerve canals extend laterally from the sides of the medullary chamber (figs. 14–17: soc), and several of them enter the glossopharyngeal-vagus foramen. Posteriorly, small canals extending dorsolaterally from the floor of the braincase and opening into the glossopharyngeal-vagus foramina represent paired occipital arteries (figs. 15–18: oc art).

Skeletal Labyrinth

The skeletal labyrinth of Tribodus limae was figured and briefly described by Maisey (2005). Digital imaging of CT scans for the present work confirm Maisey's description for the most part, except that the foramen for the octaval nerve appears to have been divided by a narrow band of cartilage to form separate foramina for the anterior and posterior rami of the octaval nerve (figs. 8B–G, 14, 15–16, 18, 20). As in neoselachians and Egertonodus (and unlike many Paleozoic chondrichthyans such as Cladodoides and Cobelodus; Maisey, 2005, 2007), a medial capsular wall is present in the otic region, separating the skeletal labyrinth from the main endocranial cavity housing the brain (figs. 8B–H, 14: mcw). This wall is fused with the parachordal-derived floor of the braincase ventrally, and with the roof of the otic region, except where it is penetrated by foramina. Anteriorly, two large foramina in the mesial wall of the braincase open into the otic capsule. These are here interpreted as separate passages for the anterior and posterior branches of the octaval nerve, which supplies the inner ear. In most gnathostomes, the anterior branch of this nerve innervates the utricular macula and the cristae of the anterior and external ampullae, whereas its posterior ramus innervates the posterior of the ear, including the preampullary canal, saccular and lagenar maculae, and macula neglecta (Cordier, 1954: 228; Maisey, 2001a).

As in all gnathostomes (Maisey, 2001a; Liem et al., 2001: 414), the skeletal labyrinth of Tribodus limae includes three semicircular canals: the anterior, external (or horizontal), and posterior canals (figs. 19–22). These canals are narrow and elongate. As in other gnathostomes, the ampullae of the anterior and external canals are directly connected to the utriculus anteriorly. As in neoselachians and most gnathostomes, a utricular recess is present (fig. 20, 21A, 22: ur). The saccular chamber is large and rounded, and a small lagena is present posteroventrally (figs. 9B, 22: lag). Dorsally, the labyrinth opens externally via two foramina, the endolymphatic foramen anteriorly (fig. 19: end f) and the perilymphatic foramen posteriorly (figs. 19, 21B: pf). These are paired structures, occurring on either side of the parietal fossa. The positions of the endolymphatic and perilymphatic ducts are similar to those of neoselachians, with the endolymphatic duct originating near the base of the anterior semicircular canal dorsally, and the perilymphatic duct originating from the dorsal margin of the posterior semicircular canal. As in neoselachians, the posterior semicircular canal of Tribodus forms an almost complete circuit, with a narrow preampullary canal (fig. 20: pac) connecting the posterior ampulla with the sacculus and perilymphatic fenestra. However, the preampullary canal is tilted upward anterodorsally, rather than almost vertical as it is in Notorynchus (Maisey, 2005: fig. 36). As in most gnathostomes, the semicircular canals are arranged at right angles to one another, with the two vertical canals aligned at a 90° angle to each other and both of them perpendicular to the horizontal (external) canal. The external canal extends laterally beyond the margin of the saccular chamber, leaving a narrow space between it and the sacculus. The semicircular canals appear proportionally narrower than those of neoselachians such as Notorynchus (Maisey, 2005: fig. 36), whereas the saccular chamber is proportionally larger, forming nearly half the anteroposterior length of the skeletal labyrinth. Like neoselachians and unlike most other craniates, a crus commune and sinus superior are absent in Tribodus, as Maisey (2001a, 2005) previously pointed out.

Skeletal Labyrinth of Egertonodus

The otic region of Egertonodus basanus examined for the present paper is not as well preserved as that of Tribodus limae. The skeletal labyrinth is best preserved on the right side of the specimen, but in the CT scans its lining of prismatic cartilage is faint and difficult to trace, particularly in the ventrolateral part of the saccular chamber and the proximal part of the preampullary canal. Nevertheless, it is well enough preserved to allow the skeletal labyrinth of Egertonodus to be reconstructed via digital imaging (fig. 36A–C). As in Tribodus, a medial capsular wall is present in the otic region of Egertonodus (fig. 31F–H, 34: mcw), separating the skeletal labyrinth from the rest of the endocranial cavity and containing foramina for the octaval, glossopharyngeal, and vagus nerves. Like Tribodus, there are two large foramina anteriorly that open into the otic capsule, which probably housed the anterior and posterior branches of the octaval nerve. The presence of two octaval foramina in Tribodus and Egertonodus may indicate a hybodont synapomorphy, and is lacking in neoselachians. As in Tribodus and neoselachians, the glossopharyngeal nerve passes through the wall of the endocranium in the medullary region, ventral to the preampullary canal and anterior to the vagus nerve. As in Tribodus and neoselachians, the glossopharyngeal nerve is located ventrally in the endocranial wall. After exiting the endocranial space, it passes via a narrow canal into the glossopharyngeal-vagus foramen, as in Tribodus. Posterior to the glossopharyngeal foramen is a pair of additional foramina, here interpreted as housing rami of the vagus nerve. One of these is located dorsally near the dome of the medullary region, and the other is located ventrally below it. Both of these foramina join canals that connect with the glossopharyngeal-vagus foramen, and the more dorsal canal continues farther laterally, exiting the braincase dorsal to the glossopharyngeal-vagus canal.

The skeletal labyrinth of Egertonodus (fig. 36A–C) is similar to that of Tribodus (fig. 36D–F) and other gnathostomes (fig. 37) in its overall structure. The anterior, external, and posterior semicircular canals are arranged as in Tribodus, and a utricular recess is present anteriorly communicating directly with both the anterior and external ampullae as in many gnathostomes. A crus commune and sinus superior are absent in Egertonodus, as in Tribodus and neoselachians, and the anterior and posterior semicircular canals are widely spaced dorsally. Like Tribodus and neoselachians, in Egertonodus an endolymphatic foramen is present anterodorsally, and a perilymphatic foramen is present posterodorsally (fig. 36B: end f, pf). The posterior semicircular canal forms a complete circuit, with a preampullary canal (figs. 32A, B, 36C: pac) connecting the posterior ampulla with the saccular chamber and perilymphatic fenestra. A small lagenar chamber is present in the posterior of the saccular region (figs. 32A, B: lag), as in Tribodus. In both Egertonodus and Tribodus, the external semicircular canal passes mesial to the posterior canal.

Fig. 37

Skeletal labyrinth structure in other elasmobranchs. A–C, extant Notorynchus, A, lateral; B, dorsal; C, medial views; D, Cladodoides, lateral view. After Maisey, 2005. No scale.

Extant Elasmobranchs

In most extant elasmobranchs, the cranial circulation follows what is probably a highly conserved pattern, in which the dorsal aorta divides into left and right branches ( =  lateral dorsal aortae), each of which gives off an efferent hyoidean artery posteriorly, and an orbital artery more anteriorly, before anastomosing to form a single internal carotid artery. This artery generally enters the braincase either after or just prior to anastomosing, producing either a single median internal carotid foramen (e.g., Raja: Allis, 1912; Heterodontus: Maisey, 1983; Shirai, 1992; Squalus: Ashley and Chiasson, 1988; Squatina; Pristiophorus; batoids: Rhynchobatus; hexanchoids: Hexanchus; many squaloids; Shirai, 1992) or a pair of foramina (e.g., Chlamydoselachus, Allis, 1911, 1923; many squaloids; galeomorphs: Proscyllium; Shirai, 1992; Compagno, 1988). As the above-listed taxa show, this feature (presence of one vs. two internal carotid foramina) is highly variable in extant neoselachians, and does not appear at present to have much taxonomic significance. Soon after anastomosing and entering the braincase, the internal carotid artery again divides into left and right branches. Each of these gives off a lateral branch (which divides to produce the efferent pseudobranchial and ophthalmic arteries), and an anterior branch, which gives rise to the optic and cerebral arteries. In neoselachians, the posterior part of the basicranial arterial system forms what has been referred to as a “bell-shaped” circuit (Schaeffer, 1981), due to the outwardly/laterally curving position of the lateral dorsal aortae, which form the concave “base” of the “bell,” while the V-shaped proximal area where the lateral dorsal aortae merge posteromedially as the median dorsal aorta forms the bell's “clapper.”

This bell-shaped area is mainly external to the braincase in neoselachians, particularly in its posterior part. However, Allis (1911) describes the efferent hyoidean artery of Chlamydoselachus as passing briefly through a short canal in the cranial cartilage before emerging again to join the lateral dorsal aorta. More anteriorly, the orbital artery can be external to the braincase (e.g., Chlamydoselachus, Allis, 1923), or can pass through the cranial cartilage, as in Heterodontus (Maisey, 1983). Compagno (1988: fig. 6.8) illustrates several extant galeomorphs in which the efferent hyoidean and orbital (stapedial) arteries meet the lateral dorsal aortae within the basicranial cartilage, anterior to the point where the lateral dorsal aortae enter the braincase (i.e., hemigaleine hemigaleids; Hemipristis). Anterior to the entry of the internal carotids into the braincase, the efferent pseudobranchial and ophthalmic arteries typically exit the braincase through a single foramen, and diverge afterward (Allis, 1911, 1912, 1923; Holmgren, 1942: fig. 18). The optic artery typically exits the braincase farther anteriorly, alongside the optic nerve.

Fossil Chondrichthyans

In fossil sharks, the pattern of cranial vascularization has been reconstructed based on the arrangement of basicranial foramina (and, where present, external grooves on the surface of the braincase) (e.g., Orthacanthus, Cladoselache, Schaeffer, 1981; Cladodoides, Tamiobatis, Maisey, 2005; “Cobelodus,” Maisey, 2007). The pattern in these fossil forms is broadly similar to that of extant neoselachians in having a posterior circuit formed from the lateral dorsal aortae and internal carotids, and in the positions of the efferent hyoidean and orbital arteries relative to this circuit. In general, the basicranial circuit in fossil chondrichthyans is narrower and more elongate than it is in neoselachians, and lacks the bell-shaped curve posteriorly. In many forms (e.g., Orthacanthus, Cladoselache, Tamio batis, Cladodoides; Schaeffer, 1981; Maisey, 2005, 2007), the lateral dorsal aortae enter the basicranial cartilage posteroventrally through paired foramina and travel anteriorly through enclosed canals. In others (e.g., “Cobelodus”; Maisey, 2007), the dorsal aorta enters the braincase through a single posteroventral foramen and divides within the cartilage, but the lateral dorsal aortae also travel anteriorly through enclosed canals. This is notably different from the pattern in hybodonts and many neoselachians, in which the posterior part of the lateral dorsal aortae and arterial circuit are external to the cartilage. Also, the part of the arterial loop posterior to the efferent hyoidean artery is longer in xenacanths and “ctenacanths” than in other chondrichthyans, a feature that is probably related to the elongated otico-occipital regions of these taxa. Another noticeable difference between many fossil and extant chondrichthyans is the anterior extent of the arterial system: in fossil forms including hybodonts and Paleozoic sharks, the cranial arterial system (from lateral dorsal aortae to optic arteries; based on the positions of foramina) appears to have extended nearly the entire length of the braincase, whereas in extant elasmobranchs (e.g., Heterodontus, Chlamydoselachus; Maisey, 1983) it extends anteriorly only about two-thirds the length of the braincase. This may also be related to the general shift in braincase proportions in neoselachians, with the otico-occipital regions shortening and the orbital and ethmoidal regions forming a larger proportion of overall braincase length. In many fossil sharks, the orbital and efferent hyoidean arteries have been reconstructed branching from the lateral dorsal aorta in very close proximity (Tamiobatis, Orthacanthus, Egertonodus, Cladoselache, “Cobelodus”; Schaeffer, 1981; Maisey, 1983, 2005, 2007), whereas in Cladodoides, Tribodus, and neoselachians there is apparently some distance between their branch points (Maisey, 1983, 2005; and the present paper). However, many of these observations rely on the accuracy of previous reconstructions, which necessarily involve at least some degree of speculation.

Tribodus

Based on CT scans and digital images of the braincase of Tribodus (AMNH 13958), its cranial circulation can be reconstructed according to several possible scenarios, described below. In all of these, it is considered likely that the optic artery passed through the optic foramen together with the optic nerve, as in Cladodoides and extant neoselachians (Maisey, 2005). It is also considered probable that the efferent pseudobranchial and ophthalmic arteries divided external to the cranial wall, as in many neoselachians and as Maisey (1983) reconstructed in Egertonodus (and confirmed in Egertonodus by CT scan images in the present study). The efferent pseudobranchial arteries are here interpreted as having passed through paired canals in the basicranial cartilage (visible in CT scan images), which emerge into the cranial cavity just anterior to the dorsum sellae. They then would have curved posterolaterally and slightly dorsally, to exit the cranial wall through a small foramen in the lateral wall of the orbit immediately anterior and ventral to the lateral hypophyseal ridge, before diverging into separate efferent pseudobranchial and ophthalmic arteries external to the braincase.

The remaining aspects of the cranial circulation are more problematic, due to the presence of several foramina not found in other fossil or extant elasmobranchs whose significance is difficult to interpret. However, the likelihood of each morphological scenario can be weighed according to a number of ontogenetic and phylogenetic considerations, allowing the most likely configuration of cranial vascularization to be identified.

On the basicranium there is a single median foramen located between the paired postorbital processes. This might normally be considered a foramen for the internal carotid arteries, as in some extant neoselachians (e.g., Heterodontus) in which the internal carotids meet external to the basicranial cartilage and enter the cartilage as a single vessel. This interpretation has also been suggested for Egertonodus (Maisey, 1983). However, unlike Egertonodus, Tribodus lacks any obvious indentations on the basicranium that could be interpreted as external arterial pathways. Also, Tribodus has several additional foramina in the basicranium that require explanation, casting doubt on the interpretation of the median foramen as an internal carotid opening. These include a pair of posterior foramina (located on the undersides of the paired posterior basicranial processes) and two pairs of lateral foramina (immediately lateral to the median foramen). CT scans show that all of these foramina are interconnected by internal canals. The posterior foramina lead to paired canals which continue anteromedially, converging to open into a large chamber overlying the median basicranial foramen, and continuous with it. This chamber is also continuous with the two pairs of lateral foramina, which open ventromedial to the postorbital processes. This configuration indicates that the structures contained within these foramina may have been continuous with one another, and suggest a system of branched arteries.

In most extant elasmobranchs, the basicranial circulation forms a loop or circuit posteriorly. The dorsal aorta divides into paired lateral dorsal aortae, which diverge to form the posterior part of the loop. They meet the paired posterolaterally directed efferent hyoidean arteries slightly farther anteriorly, continue forward as the common carotid arteries, and meet with the paired anterolaterally directed orbital arteries before turning medially as the internal carotid arteries to converge and enter the basicranial cartilage (whether the internal carotids meet immediately prior to, or after, entering the braincase is variable in many extant elasmobranchs). Indentations on the basicranium in Egertonodus strongly suggest a similar configuration, as proposed by Maisey (1983). Moreover, most fossil chondrichthyans currently known exhibit a more or less similar pattern (e.g., Orthacanthus, Tamiobatis, Schaeffer, 1981; Cladodoides, Maisey, 2005; “Cobelodus,” Maisey, 2007), with a comparable circuit and arterial arrangement (the main differences being in the shape of the circuit—bell shaped in neoselachians vs. more narrow and elongated in other elasmobranchs—and whether the circuit is entirely external or partly contained in cartilage, which varies greatly even among extant neoselachians; e.g., Compagno, 1988). Thus, the general pattern of basicranial circulation in elasmobranchs appears to be highly conserved, and it would seem most likely that Tribodus would conform to that paradigm, at least in terms of overall structure.

The internal carotid arteries extend forward within the basicranial cartilage beginning immediately anterior to the median foramen, and would have to enter the braincase somewhere posteriorly, either through the median foramen or through the foramina anterodorsal to the paired basicranial processes. The orbital arteries and efferent hyoidean arteries must also be accounted for.

In the first possible scenario of cranial circulation in Tribodus, the median basicranial foramen is interpreted as the internal carotid foramen (as in many extant neoselachians and Egertonodus as interpreted by Maisey, 1983), located where the internal carotid arteries converge to enter the braincase. Before entering the braincase, the internal carotid arteries diverge from a single median dorsal aorta posteriorly, which then divides into paired lateral dorsal aortae prior to anastomosing immediately external to the internal carotid foramen. According to this scenario, paired canals diverging posteriorly from the chamber above the internal carotid foramen are then explained as containing the efferent hyoidean arteries, the first pair of vessels to branch from the (now merged) internal carotid artery after its entry into the braincase. These canals emerge externally at the foramina above the paired basicranial processes. Immediately anterior to the junction with the efferent hyoidean arteries, the pair of canals originating in the space above the internal carotid foramen and exiting the braincase anterolaterally are then interpreted as having contained the orbital arteries. However, the posterolateral pair of foramina are then left without explanation. Anterior to the origin of the orbital arteries, the internal carotid artery would have again divided before continuing anteriorly as paired arteries tunneling through the basicranial cartilage beneath the hypophyseal chamber, which emerge on the floor of the cranial cavity to meet the efferent pseudobranchial/ophthalmic arteries, and the optic arteries farther anteriorly.

After entering the cranial cavity and giving rise to the efferent pseudobranchial and ophthalmic arteries, the paired internal carotid arteries again continue anteriorly, diverging one more time to produce first the optic arteries, which curved laterally and exited the braincase through the optic foramen alongside the optic nerve, and finally the cerebral artery, which curved dorsally to supply the brain. The difficulty with accepting this scenario is that the efferent hyoidean and orbital arteries in chondrichthyans do not typically meet the internal carotids after they merge and enter the braincase, and such a pattern is unknown in any other fossil or extant elasmobranch. Also, the posterolateral foramina in the hypophyseal region remain unexplained. Additionally, there are no observable indentations on the external surface of the basicranium that could be consistent with the path of the internal carotids entering the median ventral foramen, although this does not necessarily mean they were not present.

In the second scenario of cranial arterial circulation, the paired posterior canals converging above the paired ventral processes are interpreted as containing the lateral dorsal aortae. After entering the braincase, the lateral dorsal aortae continue forward anteromedially and enter the median basicranial chamber above the hypophyseal foramen, giving rise first to the efferent hyoidean arteries (posteriorly) and then the orbital arteries (anteriorly) before turning mesially and anastomosing above the median ventral foramen (the “internal carotid” foramen of the previously described scenario). The efferent hyoidean arteries then curve outward posterolaterally, exiting the braincase and passing along the posterolaterally directed grooves anterolateral to the paired ventral processes on their way to the hyoid arch. The orbital arteries exit the braincase through the paired foramina anterolateral to the median ventral foramen, as in the previous interpretation, and turn outward anterolaterally to enter the orbit. After anastomosing, the internal carotid arteries again diverge anteriorly, meeting the efferent pseudobranchial/ophthalmic, optic, and cerebral arteries as in the previous scenario. Because in this second scenario the internal carotid arteries enter the braincase through the paired posterior foramina, the medial basicranial foramen is left without explanation. Two possible interpretations are: (1) this foramen represents a hypophyseal duct that remained open externally; (2) the specimen is a juvenile in which the embryonic hypophyseal fenestra has not yet closed. Because other features of the braincase suggest that the specimen more likely represents an adult, the explanation that an open hypophyseal duct was present in Tribodus seems on the surface more likely. This interpretation is supported by the presence of a narrow internal canal (visible in CT scans) connecting the main hypophyseal chamber with the more posteroventral space beneath the dorsum sellae that opens ventrally through the median ventral foramen.

Nevertheless, each of the above-described scenarios is problematic. In scenario (1), although the internal carotid arteries enter the braincase through a central foramen as previously hypothesized in other hybodonts (i.e., Egertonodus, Maisey, 1983), the efferent hyoidean and orbital arteries join the internal carotid artery only after it anastomoses and enters the braincase, a key difference from most extant neoselachians and fossil chondrichthyans. Also, the pair of posterolateral foramina remain unexplained. In scenario (2), an open hypophyseal foramen is considered to have been present (unlike adult extant neoselachians). However, although the efferent hyoidean and orbital arteries join the lateral dorsal aortae after the latter structures enter the braincase (unlike most neoselachians), they do so posterior to the anastomosis of the internal carotids (as in most elasmobranchs). Therefore, the second scenario of cranial circulation in Tribodus is accepted here as the more plausible of the two.

Egertonodus

Digital reconstructions of the braincase of Egertonodus basanus made from CT scan slices as part of the present paper confirm Maisey's (1983: fig. 25) reconstruction of its basicranial circulation. These images show that the internal carotid arteries in Egertonodus most likely entered the braincase through the median basicranial foramen (Maisey, 1983: fig. 9b; this paper, fig. 40A), converging into a single vessel just prior to entering the braincase. They then diverged anteriorly into left and right branches. These branches then diverged again a few millimeters anterior to the internal carotid foramen, producing a laterally directed branch (the combined ophthalmic/efferent pseudobranchial artery, which exited the braincase via a common foramen, and then presumably diverged as in extant neoselachians), and an anterior branch (giving rise to the optic artery, which exited the braincase via the optic foramen alongside the optic nerve).