Novitate MUSEUM

vitates

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY

CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY

Number 3429, 52 pp., 13 figures

10024 February 27, 2004

Morphology of the Braincase in the Broadnose Sevengill Shark Notorynchus (Elasmobranchi, Hexanchiformes), Based on CT Scanning

JOHN G. MAISEY!

ABSTRACT

A detailed description is presented of the neurocranium in the hexanchiform shark Noto- rynchus cepedianus, a primitive modern elasmobranch (neoselachian). The study is based on high-resolution CT scanning and digital imaging, which revealed both the external and internal morphology of a wax-impregnated braincase. Besides providing new data concerning Noto- rynchus and neoselachians generally, the investigation also provides a control for establishing the reliability of morphological observations of fossil elasmobranch braincases based on CT scans. Many of the features described here have considerable phylogenetic potential, although comparative CT scan data are still unavailable for most modern and extinct elasmobranchs.

INTRODUCTION

This work describes the morphology of a shark braincase, based almost entirely on digital imaging and analysis of high-resolu- tion computerized tomography (CT) scan- ning. Scanning provides a reliable, nonde- structive procedure for repeated observation of structures in original (and often unique) specimens (Rowe et al., 1997). Digital im- aging allows the three-dimensional recon-

struction of internal and external morpholog- ical features in ways that are difficult or im- possible with conventional serial sectioning or grinding techniques.

The principal goals of this work are two- fold: (1) To provide a description of the braincase in Notorynchus, a primitive mod- ern elasmobranch (neoselachian), including its external morphology and major internal features, together with an account of topo- graphic relationships between these struc-

' Division of Paleontology, American Museum of Natural History. e-mail: maisey@amnh.org

Copyright © American Museum of Natural History 2004

ISSN 0003-0082

z AMERICAN MUSEUM NOVITATES

tures within designated regions of the cranial walls. (2) To establish the reliability of CT scanning and digital imaging in morpholog- ical description of cranial morphology in a modern elasmobranch, thereby providing a control for the interpretation of CT scans when digitally reconstructing neurocranial features in fossil elasmobranchs whose mor- phology may differ from that of extant forms.

Most external features of the braincase seen in the CT scans are well documented in many modern elasmobranchs (e.g., Allis, 1923; Daniel, 1934; Iselst6ger, 1937; Holm- gren, 1941, Devillers, 1958), and are there- fore easily verified. Unfortunately, the same cannot be said for internal features, since available descriptions differ widely in their level of detail and reliability (especially in the earlier literature), and internal morphol- ogy of the braincase has only been described in a few extant neoselachians (nevertheless, these encompass a wide systematic range of taxa, including squaloids, galeomorphs, ba- toids, hexanchiforms, Heterodontus, and Chlamydoselachus). As might be expected, the most extensively studied form is Squalus, including descriptions of its internal cranial morphology and skeletal labyrinth (Wells, 1917; Devillers, 1958; Schaeffer, 1981), the relationship of the brain and other internal structures to the braincase (Marinelli and Strenger, 1959), and cranial development (De Beer, 1931; Holmgren, 1940; El-Toubi, 1949; Jollie, 1971). In general, however, ontoge- netic studies of the braincase in neoselachi- ans have focused on cladistically derived taxa (e.g., Squalus, Etmopterus, Scyliorhinus, Raja, Torpedo, Urolophus), and there is still no description of its development in primi- tive neoselachians such as hexanchiforms and Chlamydoselachus although later devel- opmental stages have been investigated in Heterodontus (De Beer, 1924; Holmgren, 1940), a putative sister taxon to modern gal- eomorphs (Shirai 1992, 1996; Carvalho, 1996).

The present study represents a direct ex- tension of J. Frank Daniel’s seminal early 20th-century work on elasmobranchs, be- cause the Notorynchus braincase scanned for this investigation (fig. 1) is supposedly one of two wax-impregnated specimens original-

NO. 3429

ly described (under the name Heptanchus) in his classic volume The Elasmobranch Fishes (first published in 1922; the 1934 second edi- tion was used in the present work). Unfor- tunately, his illustrations cannot be matched precisely with either of these specimens, sug- gesting either that his figures were based on another specimen or that they are composites based on more than one example. It never- theless seems appropriate that the present in- vestigation involves one of these historically well-documented specimens, and that mod- ern technology permits new observations to be tied to those made by J. Frank Daniel more than 80 years previously.

The broadnose sevengill shark Notoryn- chus cepedianus is unusual among modern hexanchiforms in favoring relatively shallow waters of the continental shelves, whereas six- gill and sharpnose sevengill sharks (Hexan- chus, Heptranchias) generally occur in deep- er water (up to 1900 m) on outer shelves and upper continental slopes. The preferred hab- itat of Notorynchus is clearly more accessible to ichthyologists, perhaps explaining why this form has become the best investigated member of the Order Hexanchiformes.

From a historical perspective, Daniel’s choice of a hexanchiform as a paradigm for elasmobranch anatomy was logical, because these sharks have long been considered ex- tremely primitive and have even been com- pared with some of the earliest known extinct sharks such as Cladoselache and Cladodo- doides from the Devonian (Holmgren, 1941; Romer, 1966; Jarvik, 1980). However, the hexanchiform fossil record can be reliably traced only to the Lower Jurassic, although some isolated shark teeth of lower and mid- dle Devonian age (Emsian-Eifelian) from Australia have been tentatively referred to the Hexanchiformes (WVcMurdodus; Turner and Young, 1987). The presence of a post- orbital palatoquadrate articulation, more than five gill slits, and an unconstricted notochord have all been cited as primitive elasmo- branch features retained by modern hexan- chiforms (Young, 1962), although these sup- posedly ancient and conserved evolutionary attributes do not withstand critical appraisal. The Jurassic hexanchiform Notidanoides has a well-developed vertebral column, with cen- tra constricting the notochord as in other

2004

Big 4l*

MAISEY: BRAINCASE OF NOTORYNCHUS 3

L

Mim »

© a 5

Notorynchus cepedianus braincase photographed in dorsal (A), ventral (B), and left lateral

(C) views. This illustration is included for comparison with digital images from CT scans in the re-

maining illustrations. Scale bar = 10 mm.

**modern-level”’ or crown-group (neoselachi- an) elasmobranchs, and both the absence of vertebral calcification and corresponding no- tochordal constrictions may be apomorphic features of Recent hexanchiforms (Maisey, 1986). The persistent notion that elasmo- branchs primitively had more than five gill slits is unsubstantiated by fossil evidence, al- though ironically it appears to have arisen from Dean’s (1909) pioneer observations of the Devonian shark Cladoselache. He iden- tified only five branchial arches in this form, but surmised that there may have been a sixth and even a seventh. Subsequently, how- ever, the presence of only five gill arches in Cladoselache has become widely accepted

(e.g., Blot, 1969; Moy-Thomas and Miles, 1971), and there are clearly only five gill clefts in a three-dimensional cladoselachian fossil described more recently (Maisey, 1989). Furthermore, there is no evidence of more than five branchial arches in hybodonts (Maisey, 1982), the putative sister group to neoselachians (Maisey et al., in press), and modern phylogenetic analyses of neoselachi- ans based on morphology consistently re- solve the higher number of gill clefts in mod- ern hexanchiforms as a derived condition (Shirai, 1992, 1996; Carvalho, 1996; Car- valho and Maisey, 1996). Finally, the num- ber of gill arches has never been determined in any fossil hexanchiform, and while it is

- AMERICAN MUSEUM NOVITATES

Oo art

NO. 3429

po art

Fig. 2. Notorynchus braincase and jaws in lateral view, showing their articular relationships (after

Daniel). No scale.

widely assumed that they all had more than five this may not necessarily have been the case.

The postorbital articulation (fig. 2) re- mains one of the most controversial aspects of hexanchiform anatomy, because it closely resembles the joint found in many extinct sharks (especially Paleozoic taxa). A post- orbital articulation is also present in some basal neoselachians and primitive extinct gal- eomorphs such as Synechodus; Maisey, 1985), so its presence in hexanchiforms does not necessarily support a more remote place- ment deeper in chondrichthyan phylogeny than at the neoselachian level. In modern morphologically based phylogenetic analy- ses, hexanchiforms have been resolved in a basal position among neoselachians, within a large hypnosqualean clade that also includes squaloids, squatinoids, pristiophoroids, and batoids. However, in an alternative molecular phylogeny based on the RAG-1 nuclear gene (Maisey et al., in press) hexanchiforms fall at the base of a clade comprising ‘“‘orbitos- tylic”’ sharks (sensu Maisey, 1980; essential- ly the hypnosqualeans minus batoids). De-

spite this fundamental disagreement between modern morphological and molecular analy- ses regarding the placement of batoids, they nevertheless agree in placing hexanchiforms firmly within the neoselachian clade (fur- thermore, they also agree that the frilled shark Chlamydoselachus is the closest living relative of hexanchiforms). Under these cir- cumstances, the postorbital articulation in hexanchiforms could represent a conserved, plesiomorphic neoselachian condition. Among hybodonts, however, a postorbital ar- ticulation is typically absent, suggesting that this feature was either lost independently in hybodonts and various neoselachian lineages, or that it was lost once in the common an- cestors of hybodonts and neoselachians and was reacquired in some neoselachians (see remarks below).

Thus, while hexanchiforms such as Noto- rynchus can be considered very primitive liv- ing neoselachians, perhaps they do not de- serve the icon status of basal elasmobranchs (in the sense of a taxonomically much broader group, including all crown-group elasmo- branchs plus many additional extinct selachi-

2004 MAISEY: BRAINCASE OF NOTORYNCHUS ,

an lineages such as hybodonts, cladoselachi- ans, etc.). Hexanchiforms are nevertheless of considerable phylogenetic importance, as one of the most primitive and geologically earliest appearing groups of crown-group elasmo- branchs (even if McMurdodus is excluded), and they may indeed have retained many primitive features of early neoselachians. Giv- en the interest historically shown in hexan- chiforms, and the fact that these sharks are comparatively well known, Notorynchus cer- tainly provides a useful starting point for mor- phological comparisons of cranial morpholo- gy in modern and fossil elasmobranchs.

MATERIALS AND METHODS

Notorynchus maculatus Ayres. Recent, provenance unknown, but probably from San Francisco Bay. Uncataloged specimen; wax- impregnated braincase, preserved length ca. 120 mm, reportedly one of two specimens used as a basis of J. Frank Daniel’s original early 20th-century studies on elasmobranch morphology and now housed in the Museum of Vertebrate Paleontology at the University of California, Berkeley (fig. 1). Wax impreg- nation has preserved the braincase essentially intact, with only minor damage to external features, while CT scanning reveals minimal internal damage and complete impregnation of internal structures (a remarkable testament to preparation skills of the early 20th-centu- ry). The braincase was scanned normal to its z (long) axis by R. Ketcham and M. Colbert (University of Texas at Austin, 3 Dec., 1999). RLS, 420 kV, 1.8 mA, no filter, air wedge, 130% offset, gain 8, integration time 32 ms, slice thickness 0.25 mm, S.O.D. 730 mm, 1000 views, 2 rays averaged per view, 1 sample per view, interslice spacing 0.25 mm, field of reconstruction 95 mm, recon- struction offset 400, reconstruction scale 1450. 8-bit export parameters: level 2047, width 4095. Original imaging by M. Colbert using Voxblast was adapted by the author for publication here. Additional images of the vestibular region were rendered by the author using Imaris/Surpass software. The CT slices used in this investigation are available on- line at http://research.amnh.org/vertpaleo/ maisey/ct.html

aa acv asc atfr

ba cer

ds

ea

ect ch ect pr end f eps

eSC

fm hc? hl hm art hyp ic inp 10c lag Ic Ir

mcw med mes myc not

oO art

oc cot oc cr olf can olf cap onc

or

ora

ot cap pa

pac

p can ped pff p fen p fos

ABBREVIATIONS

anterior ampulla

passage for anterior cerebral vein anterior semicircular canal acustico-trigemino-facialis recess (internal)

basal angle

cerebellar chamber

dorsum sellae

external ampulla

ectethmoid chamber

ectethmoid process

endolymphatic foramen

passage for efferent pseudobranchial artery

external (horizontal) semicircular canal

foramen magnum

hypophyseal canal

hypotic lamina

hyomandibular articulation hypophyseal chamber

passage for internal carotid artery internasal plate

infraorbital canal

lagenar chamber

labial cartilage

foramen for lateral ramule of buccal + maxillary ramus (= classical “buccal branch of facial nerve’’) Meckel’s cartilage (lower jaw) median capsular wall

medullary chamber

mesencephalic chamber myencephalic chamber

notochordal canal

foramen for otic lateral line nerve (= classical “‘otic ramus of trigem- inal nerve’’)

orbital articulation for palatoquad- rate

occipital cotylus

occipital crest

olfactory canal

olfactory capsule (position of) orbitonasal canal

orbit

passage for orbital artery

otic capsule

posterior ampulla

ascending pre-ampullary part of posterior semicircular canal perilymphatic canal

attachment area for optic pedicel prefrontal fontanelle

perilymphatic fenestra

parietal (endolymphatic) fossa

6 AMERICAN MUSEUM NOVITATES

pnw postnasal wall

po art postorbital articulation for palato- quadrate

po pr postorbital process

pot pr post-otic process

p pr preorbital process

pq palatoquadrate

pref precerebral fontanelle

prof passage for profundal nerve above olfactory capsule (= classical “‘pro- fundus branch of trigeminal nerve’’)

psc posterior semicircular canal

sac saccular chamber

Soc passages for spino-occipital nerves

soph passages for superficial ophthalmic ramus of anterodorsal lateral line nerve (= classical “‘superficial oph- thalmic branch of facial nerve’’)

sor subocular ridge

st supratemporal lateral line nerve (= classical “dorsal ramus of vagus’’)

sup cr supraorbital crest

tel telencephalic chamber

t med taenia medialis

tpf trigemino-pituitary fossa (external)

ur utricular recess

vic vestibulolateral (auricular) chambers

Vv pr vestibular process

Il optic nerve

I oculomotor nerve

IV trochlear nerve

Vv trigeminal nerve

vu facial nerve

Vilh hyomandibular trunk of facial nerve

VIII octaval (acousticovestibular) nerve

Ix glossopharyngeal nerve

x vagal nerve

Note: Terminology for cranial nerves fol- lows Northcutt and Bemis (1993); see text for details.

EXTERNAL MORPHOLOGY

As Daniel (1934) observed, the braincase of Notorynchus is a single-unit chondrocra- nium, like that of all modern chondri- chthyans (figs. 1-6). It has been suggested that such a continuous single-unit cartilagi- nous chondrocranium is primitive for gna- thostomes (Goodrich, 1930: 231), but there is mounting evidence that the modern elas- mobranch braincase is specialized rather than primitive (particularly in the otic and occip- ital regions; Maisey, 2001b and in press) and that the braincase in early chondrichthyans consisted of more than one component (Maisey and Anderson, 2001). The braincase

NO. 3429

of Notorynchus can be characterized as pla- tybasic, with a centrally located cranial cav- ity in contact with the basicranium and sep- arating the orbital cartilages. There is no ap- preciable deepening of the prehypophyseal (trabecular) part of the basicranium although it is certainly narrow in places (e.g., in the posterior part of the orbit; Holmgren, 1942). As in other modern elasmobranchs, the brain in Notorynchus contacts the basicranium for most of its length, an arrangement which Northcutt (1978) suggested may be primitive for craniates and gnathostomes.

In Notorynchus there is very little calcifi- cation of the braincase or the rest of the chondral skeleton (typical for modern hex- anchiforms, but unusual for neoselachians generally). An optic pedicel is present in No- torynchus, although it is not preserved in the CT scanned braincase. Its former position is marked by a low expansion of the orbital wall near the anterior edge of the trigemino- pituitary fossa. The braincase consists of a relatively thin-walled box to which the ol- factory and otic capsules are fused, and it bears articular surfaces for the palatoquadrate and epihyal (hyomandibular) cartilage later- ally and for the vertebral column posteriorly (fig. 1). In dorsal view, the braincase is re- markably similar to that of Chlamydosela- chus and Hexanchus (Allis, 1923; Holmgren, 1941). It is broadly pointed anteriorly and almost square posteriorly, although the oc- cipital region projects for some distance be- hind the otic capsules (fig. 3). Heptranchias differs from Notorynchus, Hexanchus and Chlamydoselachus in having a much narrow- er braincase and a shorter postorbital process.

In Notorynchus the cranial roof is slightly convex anteriorly, and extends above the in- ternal cranial cavity as far as the large dor- sally located opening of the anterior or pre- cerebral fontanelle (cavum praecerebrale of Allis, 1913). As in other modern sharks, there is no posterior fontanelle in the cranial roof (unlike in batoids, where there is often a large posterior fontanelle). There is a thin, delicate cartilaginous roof above the olfac- tory capsules, penetrated by a short canal for the profundal nerve, behind which there are one or two openings on each side for the dis- tal part of the superficial ophthalmic ramus of the anterodorsal lateral line nerve. A series

2004

oc Ci

Fig. 3.

of smaller foramina for ramules of the same nerve are arrayed along the wide supraorbital shelf farther posteriorly. The junction of the orbital and otic regions is marked laterally by the postorbital process (figs. 3—6). In the otic region there is a broad median dorsal depression (the parietal or endolymphatic fossa) containing the paired perilymphatic fe- nestrae and endolymphatic foramina (figs. 3, 6B). The positions of the anterior and pos- terior semicircular canals are marked by faint V-shaped ridges on the cranial roof on each side of the parietal fossa. Behind the fossa is a short occipital region with a medial crest.

MAISEY: BRAINCASE OF NOTORYNCHUS i

Dorsal view of Notorynchus cepedianus braincase; external virtual image rendered from CT scans. Anterior to top of page. Scale bar = 10 mm.

As in other modern hexanchiforms, the ventral surface of the braincase in Notoryn- chus is angular, with a large, ventrally di- rected process beneath the orbits (figs. 4, 5B; curiously, a ventral view was never figured in any editions of Daniel’s Elasmobranch Fishes). This process is formed within a thickened area of the basicranium termed the basal angle, which has been extensively stud- ied in Squalus (El-Toubi, 1949; Jollie, 1971). A basal angle is found in modern squaloids and hexanchiforms, but not in other adult ex- tant elasmobranchs (Holmgren, 1942). The cartilage forming the lateral part of the basal

8 AMERICAN MUSEUM NOVITATES

=

IX

NO. 3429

inp

olf cap

ect pr

or

ba

v pr

S oc cot

Fig. 4. Ventral view of Notorynchus cepedianus braincase; external virtual image rendered from CT

scans. Anterior to top of page. Scale bar = 10 mm.

angle in Notorynchus is very thick and forms a prominent articular surface for the orbital process of the palatoquadrate.

Like other modern hexanchiforms, Noto- rynchus has a postorbital articulation for the palatoquadrate, located on the chondrified upper part of the postorbital process. How- ever, a postorbital articulation is absent in Chlamydoselachus, the immediate sister tax- on to hexanchiforms, according to Shirai (1992, 1996) and Carvalho (1996). Further- more, in the Upper Jurassic hexanchiform Notidanoides the quadrate flange of the pal- atoquadrate is low and elongated (as in Chla- mydoselachus) and there is no evidence of

any articular surface for the palatoquadrate on the postorbital process (Maisey, 1986: 100). It is therefore possible that a postorbital articulation was primitively absent in hex- anchiforms, and may represent a synapomor- phy only of more derived members of this lineage (including the crown group). The postorbital articulation in Notorynchus 1s dis- cussed further below.

ROSTRAL AND ETHMOIDAL REGIONS

GENERAL: The neurocranium of Notoryn- chus is broad and somewhat blunt anteriorly, with a short rostrum beneath the precerebral

2004

V,VII

p pr ect pr Fig: -5:

eps

MAISEY: BRAINCASE OF NOTORYNCHUS 9

ped v pr SOC

Lateral views of Notorynchus cepedianus braincase; external virtual images rendered from

CT scans. Right side (A); left side (B). Scale bar = 10 mm.

fontanelle. There is a short, broad rostral plate, which encloses a small median open- ing (the prefrontal fontanelle of Holmgren, 1941). The anterior margin of this fontanelle is broken in the scanned specimen, but was shown intact in the braincase illustrated by Daniel (1934: fig. 45). In other modern hex- anchiforms (e.g., Hexanchus, Heptranchias) and in Chlamydoselachus the anterior margin of the rostral cartilage is smooth and round- ed. The prefrontal fontanelle is therefore unique to Notorynchus among crown group hexanchiforms, but there is some evidence that one was also present in the Jurassic hex- anchiform Notidanoides (Maisey, 1986). ROSTRAL CARTILAGE AND PRECEREBRAL FONTANELLE: In Notorynchus the cartilage

between the rostral plate and nasal capsules is extremely thin and is perforated by several small irregular openings (sometimes con- joined; figs. 1, 3). In Chlamydoselachus there is a distinct notch on either side of the ros- trum, housing an anterior extension of the infraorbital sensory canal (Allis, 1923: pls. IV, VI—VIII). In the scanned specimen of Notorynchus no such notch is evident, al- though one was suggested by Holmgren (1941: fig. 5) and there is also a small notch between the rostral plate and olfactory cap- sule in the Hexanchus braincase figured by Shirai (1992: pl. 2). A similar notch for the infraorbital canal seems to have been present in Notidanoides (Maisey, 1986).

The dorsal opening of the precerebral fon-

10 AMERICAN MUSEUM NOVITATES

onc off

V pr fm

oc cot

NO. 3429

prcf inp

olf can

xX |X

Fig. 6. Anterior (A) and posterior (B) views of Notorynchus cepedianus braincase; external virtual

images rendered from CT scans. Scale bar = 10 mm.

tanelle in Notorynchus is smooth and has an almost oval outline, apart from a slight bulge posteriorly. This opening lacks a small, me- dially-directed processes like that found in Chlamydoselachus (Allis, 1923: fig. 9). Ac- cording to Allis (1923), the cartilage forming the floor of the fontanelle in Chlamydosela- chus probably corresponds to the rostral plate of Sewertzoff (1899), and the same may be true in Notorynchus, although the earliest stages of cranial development have still not been described in any of these taxa. The ros- tral plate in these forms may represent an anterior extension of the trabeculae as in Squalus, rather than a separate rostral carti- lage as in galeomorphs (De Beer, 1937; El- Toubi, 1949; Devillers, 1958).

The floor of the fontanelle in the scanned Notorynchus braincase is confluent with the floor of the cranial cavity farther posteriorly (fig. 3). Although the precerebral and cranial spaces would have been separated in life by the membranous dura mater surrounding the brain, no obvious skeletal feature marks the former position of this membrane (a common situation in elasmobranchs). Instead, the car- tilage flooring the cranial cavity and the pre- cerebral area farther anteriorly is smoothly continuous, suggesting that the morphology of the anterior basicranium was not greatly affected by the anterior extent of the fore- brain or the position of the dura mater. Sim- ilarly, the posterior limit of the precerebral fontanelle located in the roof of the braincase

2004

does not correspond to the anterior limit of the cranial cavity in Notorynchus, because the pineal organ (which is located within the upper anterior part of the cranial cavity) is not enclosed by cartilage and lies instead within the fontanelle (as in Chlamydosela- chus). In many other elasmobranchs there is a separate pineal opening in the roof of the braincase, marking the position of the pineal organ behind the fontanelle (e.g., Hexanchus, Squalus). Thus, the extent to which the fon- tanelle is truly precerebral in elasmobranchs is variable and not easily determined from inspection of the braincase alone. When ‘‘soft”’ tissues are unavailable (as in fossils), the absence of noteworthy morphological landmarks tends to obscure the original boundary of internal cranial and extracranial spaces in the ethmoid region.

In Notorynchus the ventral surface of the cartilage flooring the precerebral fontanelle is concave from side to side, and there is no cartilaginous keel in the ventral midline of the nasal or ethmoid region (fig. 4). In life the precerebral fontanelle is separated from the olfactory capsules by the ectethmoid chamber (an extracranial space behind and medial to the capsules), but the olfactory capsule is missing in the scanned braincase and the precerebral fontanelle therefore ap- pears to be confluent with the olfactory chamber. In elasmobranchs generally, the na- sal septum is formed by fusion and upgrowth of the anterior part of the trabeculae (Good- rich, 1930: 232). In Notorynchus this area is slightly narrower than in Chlamydoselachus or Hexanchus, but is still much wider than in Heptranchias.

NASAL REGION: There appears to be a cor- relation between the width of the rostral plate/nasal septum and the lateral spacing of the olfactory capsules and ectethmoid cham- bers. The rostral plate and nasal septum are both broad in Chlamydoselachus and Noti- danoides, and their olfactory capsules and ectethmoid chambers are widely separated. In Hexanchus and Notorynchus the rostral plate and nasal septum are slightly narrower, but still separate the capsules and ectethmoid chambers quite widely. However, in Hep- tranchias (and some squaloids), the rostral plate and nasal septum are both very narrow, and the olfactory capsules and ectethmoid

MAISEY: BRAINCASE OF NOTORYNCHUS 11

chambers are located close to the ventral midline.

The olfactory capsules are missing in the CT scanned Notorynchus braincase, but their original position is clearly marked by large chambers (the cavum nasi of Gaupp, 1906) formed in the cartilage of the nasal septum. The posterior wall of the cavum nasi (antor- bital process of Goodrich, 1930) forms the postnasal or antorbital wall (= planum or- bitale or planum orbitonasalis) and separates the cavum nasi from the orbital opening.

In Notorynchus a large olfactory canal ex- tends posteromedially, and in life the nasal apertures are directed ventrally (and to some extent laterally; Holmgren, 1941). According to Goodrich (1930), the olfactory canal is first defined by cartilage forming the antor- bital process, which represents an outgrowth of the trabecular cartilage around the olfac- tory tract that then rejoins the nasal septum. In Notorynchus a small orbitonasal canal runs through the postnasal wall from the or- bit and opens anteriorly into the ectethmoid chamber below the opening of the olfactory canal (fig. 6A). In hexanchiforms and many squaloids, the ectethmoid chamber is filled in life with diffuse connective tissue (Holm- gren, 1941), but in Chlamydoselachus the chamber is covered by a tough, glistening membrane (Allis, 1923). In Notorynchus the roof of the ectethmoid chamber contains paired openings into the precerebral region. Whether these opened into the fontanelle or the cranial cavity behind the dura mater can- not be determined from the preserved brain- case alone, although they are tucked within the olfactory canals and are not exposed in dorsal view like the paired subnasal or rostral fenestrae (“‘basal communicating canals’) of other taxa such as Squalus, which do not contain blood vessels or nerves and are filled with connective tissue in life. In many squa- loids, the subnasal fenestrae open directly into the floor of the precerebral fontanelle.

ECTETHMOID AND PREORBITAL PROCESSES: Daniel (1934) identified two processes of the postnasal wall of Notorynchus: a preorbital process located on the posterolateral margin of the postnasal wall, and an antorbital (ect- ethmoid) process farther ventrally (fig. 5B). There is some evidence that both these pro- cesses were also present in the Jurassic hex-

1 AMERICAN MUSEUM NOVITATES

anchiform Notidanoides (Maisey, 1986: fig. =)

The ectethmoid process of Notorynchus clearly differs from the “‘antorbital process” discussed by Goodrich (1930), which is an embryonic precursor of the entire postnasal wall. An ectethmoid process is well devel- oped in all hexanchiforms, as well as in Chlamydoselachus, squaloids, and some gal- eomorphs (Allis, 1923; Holmgren, 1941; Schaeffer, 1981), and has been defined as forming from cartilage located lateral to the orbitonasal canal (De Beer, 1931). Unfortu- nately, this topographic/ontogenetic criterion can only be established reliably using an on- togenetic series, and since cranial develop- ment has never been adequately investigated in a modern hexanchiform the true identity of the supposed ectethmoid process in these forms remains untested (although it contin- ues to be regarded as such here). Similar problems surround the supposed ectethmoid process in certain extinct elasmobranchs (e.g., in hybodonts: Maisey, 1983; Coates and Sequeira, 1998; Maisey et al., in press).

In Notorynchus the ectethmoid process clearly arises from the posteroventral margin of the postnasal wall, as in Chlamydosela- chus and Hexanchus. In all these taxa the process is very large, extending posterolat- erally behind the olfactory capsules and lat- eral to the ectethmoid chamber. By contrast, in Heptranchias the ectethmoid process is much smaller, oriented more posteriorly and medially, and does not extend lateral to the ectethmoid chamber. In all modern hexan- chiforms, the ectethmoid process is separated from the ectethmoid chamber by a narrow cartilaginous band, whereas in Chlamydose- lachus the anterior margin of the process overhangs the posterior margin of the ecteth- moid chamber without any intervening car- tilage. In modern squaloids the ectethmoid process and chamber are usually separated by a broad cartilaginous area of the postnasal wall (= antorbital shelf of Wells, 1917), and the process is usually positioned some dis- tance behind the chamber (e.g., Squalus; De- villers, 1958: fig. 349). Overlap of the ect- ethmoid chamber by the ectethmoid process in Chlamydoselachus is therefore an unusual condition, apparently representing an auta- pomorphy of the genus.

NO. 3429

The ectethmoid process in Notorynchus is unusual in having a canal passing through it anteroposteriorly (fig. 6A). According to Holmgren (1941: fig. 5) this canal houses the buccal branch of the facial nerve (= buccal ramus of the anterodorsal lateral line nerve; Northcutt and Bemis, 1993). The canal is ab- sent in other hexanchiforms and Chlamydo- selachus. In Chlamydoselachus the buccal ramus of the anterodorsal lateral line nerve passes dorsal to the ectethmoid process, along with the maxillary artery, the facial vein, and the maxillary ramus of the trigem- inal nerve (Allis, 1923; Jarvik, 1942). The cartilage forming this process presumably extends slightly farther dorsally in Notoryn- chus than in Chlamydoselachus.

The elasmobranch preorbital process seems to lack any precise topographic defi- nition, although according to Daniel (1934) it is located even farther laterally on the post- nasal wall than the ectethmoid process. Since it is difficult to define, its systematic distri- bution is problematic to determine. A pre- orbital process is supposedly present in No- torynchus and Hexanchus but seems to be absent in Heptranchias and weakly devel- oped in Chlamydoselachus (Holmgren, 1941). There is apparently no equivalent pro- cess in Sgualus.

ETHMOIDAL “‘ARTICULATION’’: According to Wolfram (1984), in Notorynchus the an- terior ends of the palatoquadrates are strong- ly bound to each other by ligaments; this symphsis is held in place against the floor of the braincase by additional strong, ligamen- tous tissue which arises just anterior to the articulation (located near at the midline be- tween the antorbital processes) and extends above the tooth-bearing part of the palato- quadrate, eventually merging with a tendon associated with the adductor musculature. Unfortunately, there is no indication of where these ligaments arose on the cleaned braincase, and no articular surface for an eth- moidal articulation can be identified in the Notorynchus braincases examined here, nor in other hexanchiforms or Chlamydosela- chus.

In Orthacanthus (and apparently in many other Paleozoic sharks), the anterior part of the palatoquadrate has a well-developed pro- cess anteriorly, which articulated with the

2004

ethmoid region of the basicranium just below or behind the olfactory capsules (Hotton, 1952; Schaeffer, 1981). Wolfram (1984) not- ed that the anterior ends of the palatoquad- rates in Orthacanthus do not meet at the mid- line and lack a symphyseal surface, and in- stead are rounded as in Notorynchus, sug- gesting a loose symphyseal connection that would permit a degree of lateral eversion of the palatoquadrate. Furthermore, she noted that the anterior process in Orthacanthus is strongly angled inwards in the orbit, unlike the more vertical orbital articulation of No- torynchus and other hexanchiforms. Al- though the ethmoidal articulation in extinct sharks such as Orthacanthus differs from the orbital articulation of neoselachians in its to- pographic position within the orbit, its rela- tionship to the efferent pseudobranchial fo- ramen and polar cartilage are similar and the articulations could be homologous.

ORBITAL REGION

GENERAL: As in other modern elasmo- branchs, in Notorynchus the entire orbital wall is chondrified except for various foram- ina for nerves and vessels (fig. 5). The most prominent opening in the orbital wall is for the optic nerve, which is located more or less centrally in the orbit. Some distance behind this, at approximately the same height in the orbit, is a smaller opening for the oculomotor nerve. According to Sewertzoff (1899), the optic and oculomotor foramina provide im- portant developmental landmarks, marking the approximate line of fusion between the embryonic trabecular and orbital cartilages. The superficial ophthalmic branch of the an- terodorsal lateral line nerve leaves the orbit anteriorly via the preorbital canal, near to which is a small foramen for a deep branch of the trigeminal nerve. Closer to the roof of the orbit (behind the superficial ophthalmic foramen) is a small trochlear foramen, pro- viding innervation to the superior oblique eye muscle. In the front of the orbit, above the orbitonasal canal, the foramen for the profundal nerve leads into a short, antero- dorsally directed passage which opens on the braincase roof behind the ectethmoid cham- ber (figs. 3, 5B). A foramen for the anterior

MAISEY: BRAINCASE OF NOTORYNCHUS 13

cerebral vein is also situated here (fig. 3A), just above the profundal foramen.

TRIGEMINO-PITUITARY Fossa: In gnathos- tomes, the main branches of the trigeminal and facial nerves leave the braincase via fo- ramina situated low down in the back of the orbit, just in front of the otic capsule and postorbital process. In Notorynchus and many other modern elasmobranchs, the openings for these nerves are located in the posteroventral part of the orbit, within an ex- ternal embayment or recess (the orbital fis- sure of Daniel, 1934; in part the trigemino- pituitary fossa of Allis, 1914; trigemino-fa- cialis recess of Schaeffer, 1981). The anter- oventral margin of the trigemino-pituitary fossa also contains foramina for the abducens nerve and pituitary vein (fig. 5). This recess is distinct from the acustico-trigemino-faci- alis recess of Allis (1914), which is a related but internal feature of the endocranial wall that contains the exits not only of the trigem- inal and facial nerves, but also the passage of the acoustic nerve leading into the otic capsule (discussed below).

Within the trigemino-pituitary fossa in No- torynchus there is a narrow prefacial com- missure (an upgrowth of the embryonic basal plate extending to the otic capsule; Goodrich, 1930). The commissure separates the pala- tine ramus and hyomandibular trunk of the facial nerve from what used to be regarded as its ophthalmic and buccal branches (cor- responding to the anterodorsal lateral line nerve of Northcutt and Bemis, 1993), as well as separating the facial and trigeminal nerves. The prefacial commissure therefore takes on heightened morphological signifi- cance according to this interpretation, be- cause it more clearly separates distinct nerves (the facial, and the anterodorsal lat- eral line + trigeminal), rather than merely dividing the “‘facial’’ components.

SUBORBITAL REGION: In Notorynchus (as in other modern hexanchiforms and squaloids), a suborbital shelf is absent, although there is an inflated area of cartilage forming the me- dial surface of the orbital articulation on ei- ther side of the basal angle. In the Upper Jurassic hexanchiform Notidanoides, the ba- sicranium is broader than in Notorynchus, but a suborbital shelf still seems to be absent (Maisey, 1986: fig. 6). In Squatina, a sub-

14 AMERICAN MUSEUM NOVITATES

orbital shelf is present behind the orbital ar- ticulation, which is unusual in (1) extending obliquely across much of the orbit, and (2) not forming a well-defined articular surface as in other orbitostylic sharks (Iselst6ger, 1937; Holmgren, 1941). Unfortunately, on- togenetic data for Sguatina are mostly lack- ing, and consequently it is not possible to determine whether its suborbital shelf is ho- mologous with that of galeomorphs.

BASAL ANGLE AND ORBITAL ARTICULATION: In Notorynchus, the basal angle forms a prominent projection on the ventral surface of the braincase. Its topographic relationships to surrounding features are complex; it is po- sitioned below and slightly anterior to the tri- gemino-pituitary fossa, and behind the level of the optic foramen but anterior to the optic pedicel, rectus musculature, and efferent pseudobranchial foramen (figs. 4, 5). The lat- eral surface of the basal angle includes a smooth, almost vertical groove forming an articular surface for the orbital process of the palatoquadrate (= ‘“‘ethmopalatine groove”’ of Wilga, 2002). The dorsal part of this groove extends onto the orbital wall and is positioned approximately midway between the optic foramen and trigemino-pituitary fossa, but most of the groove lies entirely below the level of these features. The mar- gins of the articular groove are formed in life by fibrous connective tissue connecting it to the orbital process of the palatoquadrate (fig. 2), and these are loosely held together by a ligamentous sheet forming a sac which en- closes the entire joint (Wolfram, 1984).

HYPOPHYSEAL REGION: There is no hypo- physeal opening in the basicranium of No- torynchus, although there is evidence of a vestigial hypophyseal duct within the thick- ness of the basicranial cartilage (discussed below). The internal carotid foramina (which lie more or less between the postorbital pro- cesses in Notorynchus) nevertheless provide an important topographic and developmental landmark, since the internal carotids enter the braincase between the embryonic trabeculae and parachordals. During ontogeny, the paired trabeculae fuse to each other anteri- orly, and also to the basal (parachordal) plate posteriorly, leaving an anterior basicranial fe- nestra containing the hypophysis and internal carotids (Goodrich, 1930: 234). Fusion be-

NO. 3429

tween the trabecular and parachordal carti- lages may also involve a separate polar car- tilage (e.g., Squalus; van Wijhe, 1922), but it is unknown whether such a cartilage is pre- sent in Notorynchus. The polar cartilage has a widespread occurrence in many gnatho- stomes, and may even surround the internal carotids (e.g., in birds; Goodrich, 1930).

POSTORBITAL PROCESS

GENERAL: In Notorynchus the postorbital process is a prominent feature of the brain- case defining the posterior limit of the orbit, extending laterally from its midregion above the lateral head vein and the hyomandibular ramus of the facial nerve (figs. 1-6). The postorbital process is an important develop- mental landmark in gnathostomes because it arises from the anterolateral margin of the basal plate within the blastemic lateral com- missure, level with or just behind the embry- onic trabecular—parachordal junction. The lateral commissure develops secondarily as an upward extension of the basal plate that eventually fuses with the prootic region of the otic capsule (De Beer, 1937; Schaeffer, 1981). According to Holmgren (1940, 1941), in hexanchiforms the lateral commissure does not persist and only the primary post- orbital process (extending from the embry- onic supraorbital cartilage) is chondrified (fig. 5). Thus, the postorbital articulation (discussed below) is confined to the primary postorbital process and does not include car- tilage derived from the lateral commissure (an important difference from extinct am- phistylic sharks).

INNERVATION: In Notorynchus the dorsal surface of the postorbital process is flat or slightly concave, and bears a few small open- ings for branches of the buccal ramus of the anterodorsal lateral line nerve, innervating the overlying sensory canal (figs. 1A, 3). Similar openings occur in Chlamydosela- chus, but are far more numerous (Allis, 1923). There is a single opening near the base of the posterior surface of the postor- bital process in Notorynchus, corresponding to the foramen for the otic lateral line nerve in Chlamydoselachus (fig. 6B).

POSTORBITAL ARTICULATION: One of the most important features of the postorbital

2004

process in hexanchoids is the presence of an articular surface for the otic process of the palatoquadrate (figs. 1-6). In Notorynchus, the surface extends transversely across part of the posterior surface of the process. No comparable articulation is present in Chla- mydoselachus, nor has one been described in any other modern elasmobranch. It has been claimed that a postorbital articulation is pre- sent in the crocodile shark Pseudocarcharias (Compagno, 1977), but its structure has nev- er been described and the connection is ap- parently lost during jaw protraction. Curiously, the structure of the postorbital articulation in hexanchiforms has not been described in detail, despite its obvious phy- logenetic interest (although it has been in- vestigated in Notorynchus, as part of an un- published Master’s thesis; Wolfram, 1984). In Notorynchus, the articular surface of the postorbital process is little more than a flat surface extending posteroventrally (fig. 4). This surface contacted a corresponding an- terodorsally directed surface on the palato- quadrate. According to Wolfram (1984), these articulating surfaces are covered with fibrous connective tissue and are surrounded by a tough connective tissue which forms a bursa around the articulation. Garman (1913) noted that the postorbital articulation in Hep- tranchias is stronger than in Notorynchus or Hexanchus; Holmgren (1941) also noted that the articular surface for the palatoquadrate (“‘articular disc’’) on the postorbital process of Heptranchias was ‘‘well delimited’’. Wolfram (1984) concluded that, given the ligamentous connections and articular surfac- es she observed in the postorbital articulation of Notorynchus, only one pattern of palato- quadrate movement was possible, involving lateral eversion of the posterior end of the palatoquadrate as well as slight medial rota- tion of the anterior margin about its long axis. She found that such movement pro- duced (1) rotation of the palatoquadrate at the postorbital articulation about an axis run- ning parallel to its long axis; (2) translation along the groove forming the orbital articu- lation, and (3) posterior translation of the up- per jaw symphysis along the ventral surface of the braincase. Luther (1908) and Wolfram (1984) both concluded that no anterior trans-

MAISEY: BRAINCASE OF NOTORYNCHUS 15

lating or shifting of the jaws is possible in the hexanchoids they examined.

The postorbital articulation in extinct am- phistylic sharks such as Orthacanthus, Tam- iobatis, and Cladodoides is more complex than in hexanchiforms. The articular surface is about twice as wide as deep (1.e., much wider than in Notorynchus), and is usually curved into a complex saddle shape, convex transversely and concave dorsoventrally. Also, the articular surface typically has a pronounced anhedral angle, so its distal ex- tremity is somewhat lower than its proximal part. Furthermore the articular surface is lo- cated farther laterally on the postorbital pro- cess than in Notorynchus and other hexan- chiforms, and it clearly extends onto the lat- eral commissure unlike in modern hexanchi- forms. Thus, the postorbital articulation of hexanchiforms differs significantly from that of extinct amphistylic sharks both in its to- pographic extent and its relationship to the lateral commissure, although they are similar in their presumed relationship to the lateral head vein/jugular canal and involve identical skeletal components (e.g., postorbital pro- cess, otic flange of palatoquadrate). A post- orbital articulation is supposedly present in the extinct neoselachian Synechodus, but its lateral commissure was unchondrified and presumably the articular surface was con- fined to the primary postorbital process as in hexanchiforms (Maisey, 1985).

PARACHORDAL PLATE AND OTIC CAPSULES

GENERAL: In craniates generally, the otic capsules develop above the basal (parachord- al) plate as independent structures dorsal to the hyoidean and first branchial arches, first appearing as invaginations (otic placodes) which subsequently undergo an unequal growth pattern to produce all the major struc- tural parts of the inner ear (Maisey, 2001a). Chondrification of the otic capsule suppos- edly begins at two different centers, one as- sociated with the anterior and horizontal am- pullae, the other with the posterior ampulla (Squalus; van Wijhe, 1922). The fact that the anterior and horizontal ampullae share a sin- gle chondrification center undoubtedly re- flects an underlying developmental relation-

16 AMERICAN MUSEUM NOVITATES

ship between them, since gene expression patterns of the anterior and horizontal cristae are similar (Morsli et al., 1998), and the an- terior and horizontal ampullae are both sup- plied by the anterior branch of the octaval (acousticovestibular) nerve whereas the pos- terior ampulla is not (Maisey, 2001a).

In elasmobranchs, a separate floor (the hy- potic lamina or lamina basiotica) develops from the parachordal cartilage beneath the capsule (Goodrich, 1930). The otic capsule eventually fuses with the hypotic lamina, al- though parts of the capsule floor may remain unchondrified (especially above the passage for the glossopharyngeal and middle lateral line nerves, as discussed below). Space be- tween the lamina and capsule (..e., the em- bryonic metotic or basicapsular fissure) is progressively obliterated by this fusion, leav- ing only the glossopharyngeal canal (Good- rich, 1930; Schaeffer, 1981). The metotic fis- sure is also closed by the occipital pila, which grows up and fuses with the wall of the otic capsule above the vagal nerve. The posterior semicircular canal is_ ultimately sandwiched between the glossopharyngeal canal medially and vagal canal laterally.

PARIETAL (ENDOLYMPHATIC) FOSSA: In No- torynchus, the mid-dorsal surface of the otic region includes the parietal fossa (figs. 1, 3, 6B), a shallow depression located between the otic capsules. A parietal fossa is well de- veloped in most modern elasmobranchs, al- though its extent and depth differ consider- ably. In Notorynchus, the fossa is well de- fined laterally, but merges smoothly with the cranial roof anteriorly and posteriorly. By contrast, in Chlamydoselachus the posterior wall of the fossa is steep although its anterior margin is smooth (Allis, 1923), and in Squal- us the fossa lacks a well-defined border ex- cept posteriorly (Devillers, 1958: fig. 345). During ontogeny, the anteroposterior extent of the elasmobranch parietal fossa is defined by two transverse bridges between the otic capsules, the synotic tectum anteriorly and posterior tectum posteriorly (although the distinction between them is not absolute; Gaupp, 1906; Goodrich, 1930).

The parietal fossa contains large paired perilymphatic fenestrae and smaller paired endolymphatic foramina. The perilymphatic fenestrae are covered by a membrane in life,

NO. 3429

below which the perilymphatic canals are connected (via a posterior canal duct) to the perilymphatic space surrounding the poste- rior semicircular canal (Corwin, 1989). De- velopmentally, the perilymphatic fenestra represents an unchondrified region in the me- dial wall of the otic capsule, where it meets the posterior semicircular canal (De Beer, 1931, 1937; Holmgren, 1940). The inner ear of modern elasmobranchs is highly special- ized toward semidirectional low-frequency phonoreception (Maisey, 200la), but the perilymphatic fenestrae represent the only external feature of the braincase that is as- sociated with this ability (many internal skel- etal features are also associated with it; see discussion below). The perilymphatic fenes- trae lie medially and posteriorly to the paired endolymphatic foramina, which are connect- ed with the saccular region of the inner ear via the endolymphatic ducts.

LATERAL SURFACE OF OTIC REGION: In No- torynchus this region is relatively featureless except posteriorly. The dorsal margin of the lateral surface forms a distinct ridge as in Chlamydoselachus, which Allis (1923) re- garded as corresponding to the spheno-pter- otic ridge of actinopterygians (although such putative homologies seem dubious given the lack of any ossification centers in elasmo- branchs comparable to those of osteichth- yans). A similar ridge extends posteriorly from the postorbital process in many modern elasmobranchs. In Notorynchus, the ridge is relatively straight, whereas in Chlamydose- lachus and Squalus it bears a short process (pterotic process; Wells, 1917: fig. 1; Allis, 1923: figs. 8-10; Devillers, 1958: fig. 345).

In Notorynchus, the head of the hyoman- dibula makes contact with the posterior part of the capsular wall at a weakly defined ar- ticular fossa (fig. 5B), unlike in Chlamydo- selachus where the dorsal margin of the fossa forms a distinct ridge on the lateral capsular wall. According to Gegenbaur (1872) and Gadow (1888), this connection between the hyomandibula and braincase in sharks is merely ligamentous rather than a true artic- ulation; Gadow (1888) even suggested that in Hexanchus there is no absolute contact be- tween the hyomandibula and cranium be- cause of intervening ligaments. However, Wolfram (1984) found that in Notorynchus

2004

the hyomandibula is held securely in the ar- ticular fossa by ligaments, and it contacts the medial capsular wall. She found only a lim- ited range of motion was permitted by these ligaments, including some rotation except anteriorly, with the greatest range of move- ment about an axis parallel to the long axis of the fossa (allowing the distal end of the hyomandibula to swing laterally).

The hyomandibular fossa in Notorynchus forms a deep embayment between two pro- cesses: ventrally there is a rather short, Square vestibular process (Gadow, 1888) and dorsally there is a longer postotic process containing the glossopharyngeal foramen (Holmgren, 1941), just above which is a small foramen housing what is classically re- garded as a dorsal branch of the vagal nerve, but now identified as the supratemporal lat- eral line nerve (figs. 3, 6B). According to Holmgren (1941: figs. 6,8), the correspond- ing foramen in Chlamydoselachus houses the same nerve, although this was not noted by Allis (1923).

Postotic and vestibular processes are both present in other modern hexanchiforms and at least the postotic process is present in No- tidanoides (= “‘lateral otic process”’ of Mais- ey, 1986). Both processes are also present in Squatina (“‘Dorsalrand”’ and “‘Basalrand des Hyoidgelenkes”’ of Iselst6ger, 1937: pl. 6). In many other modern elasmobranchs the glos- sopharyngeal canal is contained within a postotic process, but the vestibular process is highly variable in its occurrence and is weak or absent in some taxa.

In Notorynchus, the vestibular process is not associated with any foramen. By con- trast, in Hexanchus the vestibular process contains an opening (the “‘vacuity below hyomandibular fossa’’ of Shirai, 1992), and Gegenbaur (1872) illustrated a corresponding foramen in Heptranchias. A vestibular pro- cess is present in Chlamydoselachus, but it does not have a foramen.

The postotic process has been compared with the lateral otic process in other extinct sharks (e.g., Orthacanthus, Tamiobatis; Schaeffer, 1981), but that process is posi- tioned farther dorsolaterally (relative to the inferred position of the glossopharyngeal nerve) than the postotic process, and the lat-

MAISEY: BRAINCASE OF NOTORYNCHUS 17

eral otic process in these fossils does not contain the glossopharyngeal canal.

VENTRAL SURFACE OF OTIC REGION: The region behind the internal carotid foramina mostly represents cartilage of parachordal derivation. In Notorynchus, it is relatively featureless, apart from a median sulcus (per- haps emphasized by slight shrinkage of the cartilage), corresponding approximately to the former line of contact between the paired parachordals. Posteriorly the ventral surface is continuous with the occipital arch, as in other modern elasmobranchs.

VAGAL AND GLOSSOPHARYNGEAL CANALS: The topographic relationship of the glosso- pharyngeal and vagal openings differ in No- torynchus and Chlamydoselachus. According to Allis (1923), in Chlamydoselachus there is a large glossopharyngeal-vagal fossa, with the vagal foramen in its medial part and the glossopharyngeal foramen in its ventrolateral corner. By contrast, in Notorynchus the vagal foramen lies within the base of the notch formed by the posterior capsular wall and the occipital region, separate from the glosso- pharyngeal canal (as in Hexanchus; Shirai, 1992: pl. 21A). However, Allis (1923: fig. 10) illustrated two different arrangements in his specimen of Chlamydoselachus, with the right glossopharyngeal opening separated from the remainder of the fossa by a bridge of cartilage. If the posterior wall of the glos- sopharyngeal canal failed to develop com- pletely in Notorynchus or Hexanchus, it would result in a deep fossa containing both the vagal and glossopharyngeal openings, much as in Chlamydoselachus. Variation in the extent of the posterior wall of the glos- sopharyngeal canal may therefore be deci- Sive as to whether its exit also includes the vagal foramen in elasmobranchs. In Hetero- dontus and many other modern elasmo- branchs the vagal and glossopharyngeal openings are separate (Daniel, 1915), but in extinct hybodont sharks there may be a deep glossopharyngeal-vagal fossa containing both nerves (Maisey, 1983). In addition, the posterior lateral line nerve (classically iden- tified as the lateral line ramus of the vagal nerve) should also exit through this opening.

OCCIPITAL REGION

GENERAL: Anteriorly, the occipital block in Notorynchus is wedged firmly between the

18 AMERICAN MUSEUM NOVITATES

otic capsules as in other modern elasmo- branchs, with the vagal foramen positioned lateral to the foramen magnum. There are no “‘condylar ridges”’ for vertebral branches of the dorsal aorta like those described on either side of the occiput in Chlamydoselachus (A\- lis, 1923). Shirai (1992: pl. 2) illustrated large paired arches defining openings in the posterolateral margin of the basicranium in Hexanchus, but he did not identify them. They are positioned farther laterally than the condylar ridges in Chlamydoselachus, and it is uncertain whether they housed aortic ves- sels.

OccIPITAL CoTyLus: In WNotorynchus, a strong connection with the vertebral column is provided by a deep occipital cotylus (figs. 4, 6B). A similar cotylus is present in hy- bodonts and many Paleozoic sharks; Maisey, 1983). In most modern sharks (but not ba- toids), an occipital half-centrum is incorpo- rated into the braincase, occupying the co- tylus (= basioccipital fovea; Shirai, 1992). There are traces of an occipital half-centrum in Notorynchus, although it is poorly calci- fied and does not fill the entire cotylus. The lateral margins of the cotylus extend poste- riorly to flank the hemicentrum and at least the first neural arch of the vertebral column. In Chlamydoselachus, there is also a well de- veloped occipital connection (“‘condyle”’ of Allis, 1923), bounded ventrally by condylar ridges which have a slightly convex posterior surface forming an articulation with a cor- responding concavity of the first vertebra. Daniel (1934: fig. 47) depicted the lateral margin of the cotylus as contacting the first free basiventral in Notorynchus. Gegenbaur (1872) identified a condylar ridge in Hex- anchus and Notorynchus, but did not find any real articulation.

According to Shute (1972), in Squalus the cranio-vertebral articulation includes paired occipital condyles, supposedly formed by a basidorsal which has fused to the posterior end of the parachordal. Each condyle artic- ulates with the first free interdorsal, which is pierced by a dorsal nerve root as in more caudal elements, and its corresponding ven- tral root passes behind (not through) the con- dyle to supply the first metotic myotome. However the ‘“‘condyles’? to which Shute (1972) referred apparently do not correspond

NO. 3429

with the paired articular condyles on either side of the occipital region recognized by most other workers. These are situated in a ventrolateral position and have no connection with the dorsal arcualia, meeting instead with paired basiventrals (Compagno, 1988). The condyles are formed within cartilage of para- chordal derivation, within the posteromedial part of the basicranium lying medial to the vagal canal. Thus, in modern sharks the prin- cipal cranio-vertebral articulation is provided by the occipital cotylus, usually incorporat- ing an occipital half-centrum (one is absent in Pristiophorus; Shirai, 1992), and often buttressed by paired occipital condyles of parachordal origin, articulating with the an- teriormost free basiventral. In addition, the hypotic lamina may extend posteriorly as su- pravagal and subvagal plates above and be- low the vagal canal (Compagno, 1988), which meet or fuse with basiventrals of the first one or two vertebral centra (e.g., Car- charhinus).

INTERNAL MORPHOLOGY

GENERAL: Digital imaging permits the en- docranial and labyrinth cavities of Notoryn- chus to be rendered as negative morphospa- ce, generating a virtual endocast that reveals many morphological features which can be compared with data obtained from conven- tional endocasts (figs. 7, 8).

ETHMOID AND FOREBRAIN REGIONS: In fig- ure 7, the precerebral fontanelle is rendered arbitrarily as a solid area, because it has no physical boundary either with the endocra- nial cavity or the medial ends of the olfactory canals. The depicted upper limit of the fon- tanelle is also arbitrary. The transversely ridged surface appearance of the olfactory canals is merely an artifact of digital imaging from the CT scans. The internal shape of the fontanelle region is not rendered but can be seen in the horizontal, sagittal, and transverse slices depicted elsewhere in this work.

Anteriorly, each olfactory canal arises an- terolaterally from the front (telencephalic) part of the prosencephalic chamber and ex- tends outward toward the olfactory capsules (fig. 7A, D). The capsules themselves are missing from the scanned specimen, but the large space they formerly occupied is clearly

2004

discernible although it is confluent with that of the ectethmoid chamber. The olfactory ca- nals are more evident in dorsal than in ven- tral view, since they merge with the inter- nasal plate ventrally. By contrast, in the Squalus endocast illustrated by Schaeffer (1981: fig.15; see fig. 8 here) the olfactory canals are well defined both dorsally and ventrally, because the lateral walls of the pre- cerebral fontanelle intrude farther posteriorly between the olfactory capsules than in No- torynchus. In dorsal view, however, the ol- factory canals in Squalus appear to diverge from the telencephalon behind the precere- bral fontanelle, whereas in Notorynchus they arise from its lateral margins.

The precerebral fontanelle extends farther posteriorly between the olfactory canals in Notorynchus than in Squalus, and farther an- teriorly in Squalus than in Notorynchus. Thus, it would be misleading simply to char- acterize the rostrum of Squalus as “long” and that of Notorynchus as “‘short’’, because the relative position of the precerebral fon- tanelle and olfactory canals accounts for some of the topological difference in the snout of these taxa. Nevertheless, the precer- ebral fontanelle of Squalus is more than dou- ble the length of its olfactory capsules, whereas in Notorynchus these structures are of approximately equal length, so the differ- ing proportions of the ethmoid region in these taxa are both dimensional as well as topological. These differences can be sum- marized as follows: in Squalus, the precere- bral fontanelle is elongated anteroposteriorly but does not intrude between the olfactory canals, although the rostral plate separates the canals from the fontanelle ventrally; in Notorynchus (and Chlamydoselachus; Allis, 1923), the fontanelle is not elongated antero- posteriorly, but intrudes between the olfac- tory canals dorsally, and the rostral plate does not separate the canals from the fonta- nelle floor.

The orbitonasal canal of Notorynchus opens anteriorly within the ectethmoid cham- ber (the usual neoselachian arrangement; Holmgren, 1941), but then passes posterolat- erally through the postnasal wall, with its posterior end lying farther laterally, and the orbitonasal canals are therefore slightly con- vergent anteriorly (fig. 7A). Unlike in Noto-

MAISEY: BRAINCASE OF NOTORYNCHUS 19

rynchus, the orbitonasal canals in Sgualus pass posteromedially through the postnasal wall and are therefore strongly divergent an- teriorly (fig. 8). Variation in the orientation of this canal cannot be accounted for simply by differences in the relative spacing of the olfactory capsules, but may be related to the angle at which the olfactory canals diverge, which in turn is related to the topological ar- rangement of the olfactory capsule with re- spect to the orbit. Thus, in Notorynchus the capsule lies entirely anterior to the orbit, and the orbitonasal canal simply passes anteriorly toward the ectethmoid chamber (fig. 7A, B). In Squalus, however, the anterior part of the orbit is expanded anteriorly to overlap part of the olfactory capsule, and the ectethmoid chamber is situated somewhat lateral to the front of the orbit (fig. 8B).

In Notorynchus, the floor of the telence- phalic chamber rises anteriorly (fig. 7C), im- parting a strong taper to this region, unlike the corresponding region in Squalus where the telencephalic chamber remains fairly deep for much of its length and is actually slightly deeper anterior to the optic foramen than farther posteriorly. In Chlamydosela- chus, the telencephalic chamber is also ta- pered anteriorly, and its floor rises even far- ther dorsally than in Notorynchus. The floor of the precerebral fontanelle between the ol- factory capsules is at almost the same level as the roof of the neurocranium behind the fontanelle in Chlamydoselachus, which is ex- tremely shallow (Allis, 1923).

In modern elasmobranchs, the epiphysis arises from the membranous roof of the di- encephalon between the optic lobes, but only reaches the cranial vault above the telen- cephalon and its opening therefore appears to be within the telencephalic chamber. From the CT scan there is no evidence of an epiph- yseal canal in the roof of the cranium in No- torynchus. Schaeffer (1981) showed an epiphyseal canal in his Squalus endocast, and there is an epiphyseal foramen in the cranial roof of many squaloids, Heptranchias, and Hexanchus. A foramen is typically absent in Chlamydoselachus, Squatina, Pristiophorus, batoids, and many galeomorphs, but the epiphysis may form a notch in the posterior margin of the precerebral fontanelle (e.g., Mustelus, Hemigaleus), and there is a fora-

20 AMERICAN MUSEUM NOVITATES NO. 3429

Figen7:

olf can

[IX

Virtual endocast of Notorynchus cepedianus braincase rendered as negative morphospace

from CT scans. Dorsal view (A); ventral view (B); lateral view, left side, with skeletal labyrinth in place (C); lateral view, right side, with skeletal labyrinth removed (D). Scale bar = 10 mm.

men in some galeomorphs (e.g., Galeorhin- us). Whether an epiphyseal opening is pre- sent seems to depend on the anteroposterior extent of the cartilage forming the cranial vault, as the epiphysis may extend anteriorly to this and appear to lie within the precere- bral fontanelle (in such cases the fontanelle cannot be considered entirely precranial; e.g., the Orectolobus braincase illustrated by Holmgren, 1941: fig. 42). It is uncertain whether the distribution of the epiphyseal fo- ramen is of phylogenetic significance, and it can be ambiguous at species level because of intraspecific variation (e.g., in some squa- loids; Shirai, 1992).

MESENCEPHALIC REGION: In Notorynchus, there is almost no discernible separation be- tween the prosencephalic and mesencephalic

chambers (fig. 7C, D), as in Heterodontus, Squatina, Hexanchus, Notorynchus and ba- toids. In Squalus, by contrast, these two re- gions are separated by a slight constriction, dorsal to the exit of the optic nerve (fig. 8C). The position of the optic lobes is indicated by the trochlear foramen, but in Notorynchus the lobes themselves are indistinct (they are better defined in Squalus). Thus, very few features of the brain are represented by mor- phological features in this part of the endo- cranial cavity dorsally or laterally. By con- trast, the floor of the mesencephalic chamber contains a well-defined hypophyseal cavity, which in life housed the hypophysis, pitui- tary vein, median cephalic sinus of the inter- nal carotids, and the efferent pseudobranchi- als (fig. 7B—D). A hypophyseal cavity is pre-

2004

MAISEY: BRAINCASE OF NOTORYNCHUS 21

IC

soc med. attr

sent in most sharks, but this chamber is only weakly developed in Pristiophorus and is ab- sent in batoids. As in Chlamydoselachus, the pituitary vein of Notorynchus leaves the cra- nial cavity via a foramen in the lateral wall of the hypophyseal chamber, emerging ex- ternally within the deep trigemino-pituitary fossa.

In some modern elasmobranchs the ante- rior part of the trabecular region forming the basis cranii is inflated internally to form a transverse presphenoid ledge (praesphenoid- vorsprung; Gegenbaur, 1872), often reaching the inner surface of the interorbital wall. However, such a ledge is not evident in No- torynchus (discussed further below).

LABYRINTH REGION: Parts of the labyrinth region (especially its medial wall) are not visible in the complete endocast of Notoryn- chus, and the structure has therefore been il- lustrated separately (fig. 9). As in Squalus (fig. 8), the vestibular chamber does not con- form closely to the architecture of the mem- branous labyrinth it encloses, and it is not clearly differentiated into utricular or saccu- lar regions (although the lagenar area is rep-

hyp Il tel

resented by a small bulge in the floor of the vestibular chamber).

The labyrinth chambers in modern elas- mobranchs display many specializations to- ward low-frequency phonoreception, includ- ing isolation of the posterior semicircular ca- nal, ascending preampullary canal, large perilymphatic fenestrae, a posterior canal duct, separation of the posterior utriculus containing the macula neglecta, and devel- opment of a medial capsular wall separating the skeletal labyrinth and main endocranial cavity (Maisey, 2001a). In Notorynchus, the ampullae of the anterior and external semi- circular canals meet the utricular recess sep- arately (fig. 9A, D). The floor of the saccular region lies in approximately the same plane as the cranial cavity, not below it as in os- teichthyans and some extinct elasmobranchs. A short perilymphatic canal merges with the upper part of the posterior semicircular canal. In life, the membranous posterior utriculus is completely separated from the utricular re- cess and opens into the posterior semicircular canal via the posterior utriculo-saccular opening containing the macula neglecta (the

22 AMERICAN MUSEUM NOVITATES

Fig. 8.

NO. 3429

olf cap

olf can

Principal endocranial features in Squalus acanthias superimposed on outlines of the brain-

case. Dorsal view (A); ventral view (B); right lateral view (C). After Schaeffer (1981). No scale.

main phonoreceptor organ in elasmo- branchs). By contrast, in osteichthyans and holocephalans both the utriculo-saccular opening and macula neglecta lie within the utriculus.

GLOSSOPHARYNGEAL AND VAGAL CANALS: The courses of the vagal and glossopharyn- geal nerves can be partially traced through the virtual endocast of Notorynchus, except where their passages merge with the floor of the vestibular chamber (figs. 7B, C, 8A, C). The posterior semicircular canal provides a

useful landmark, since it is located lateral to the vagal canal and medial to the glossopha- ryngeal canal. The latter arises at about the same level as the perilymphatic fenestrae far- ther dorsally, and is small as it exits the en- docranial cavity. The canal then expands be- fore passing in front of the ascending (preampullary) part of the posterior semicir- cular canal, where it gives off a dorsal branch that passes into the saccular region (probably representing the middle lateral line nerve). The glossopharyngeal canal then becomes

2004

MAISEY: BRAINCASE OF NOTORYNCHUS 29

C E

Fig. 9.

F D

Virtual endocast of the left labyrinth region in Notorynchus. (A) lateral view; (B) medial

view; (C) dorsal view; (D) ventral view; (E) posterior view; (F) anterior view. Imaris/Surpass surface

rendering.

indistinct as it enters a pocket housing the lagena. The canal separates from the pocket posteriorly (presumably within the passage formed by fusion of the hypotic lamina to the floor of the otic capsule, as discussed by Schaeffer, 1981) and then passes between the posterior semicircular canal (medially) and vestibular process (laterally) before turning dorsally to open on the posterolateral surface of the otic region.

The vagal canal is fairly uniform in di- ameter, and its path is considerably simpler than that of the glossopharyngeal canal. It first appears in transverse sections just be- hind the level of the horizontal semicircular canal, then extends posterolaterally between

the medullary chamber and posterior semi- circular canal. A small canal branches from it dorsally and then turns anteromedially, but this disappears before reaching the endocra- nial cavity. The identity of the small branch is uncertain, but it may correspond to the ca- nal described in Chlamydoselachus by Allis (1923), in which there is a small vein coming from a plexus on the dorsal surface of the brain, considered by Gegenbaur (1872) to represent a primitive internal jugular vein. Slightly farther posteriorly, just behind the posterior semicircular canal, the vagal canal gives rise to a small dorsal passage (probably for the supratemporal lateral line nerve; figs. 3, 6B) which reaches the dorsal surface of

24 AMERICAN MUSEUM NOVITATES

the braincase behind the glossopharyngeal canal.

CEREBELLAR AND MEDULLARY REGIONS: The cerebellar chamber encloses the dorsal part of the hindbrain and includes a large acustico-trigemino-facialis recess (figs. 10K, 11J, 12H). As in Chlamydoselachus (Allis, 1914), this recess is located in the lateral wall of the cerebellar chamber, adjacent to the dorsum sellae and just behind the level of the trigemino-pituitary fossa occupying the pos- teroventral part of the orbital wall externally (figs. IOI, 11A, 12G, H). The cerebellar chamber corresponds both topographically and morphologically to the large “‘bay”’ de- scribed by Allis (1923: 162) in the cranial cavity of Chlamydoselachus. This chamber is situated dorsal to the acustico-trigemino-fa- cialis recess and immediately internal to the postorbital process. In Chlamydoselachus, the cerebellar chamber is evident in sagittal and horizontal slices through the cranial cav- ity published by Allis (1923: figs. 12, 58).

The cerebellar chamber in Notorynchus gradually widens dorsally into paired vesti- bulolateral (auricular) chambers housing the paired auricles of the cerebellum (fig. 7A). Vestibulolateral chambers are also well de- veloped in Squalus (fig. 8A). In both Noto- rynchus and Squalus, the vestibulolateral chambers are prominent in dorsal view, but the ceiling of the cerebellar chamber in front of the parietal fossa decreases in height more gradually in these taxa than in Chlamydose- lachus, making them less distinct in sagittal view (figs. 7D, 8C).

In Notorynchus, the medullary chamber becomes increasingly constricted as it passes beneath the parietal fossa farther posteriorly, and reaches its minimum height directly be- neath the perilymphatic fenestrae (fig. 7D). As in other modern elasmobranchs, the med- ullary chamber is separated from the paired labyrinth chambers farther laterally by the chondrified medial wall of the otic capsule, and the two are connected only by passages for the octaval and glossopharyngeal nerves. For this reason, the entire labyrinth cavity appears as an almost separate object in the endocasts of both Notorynchus and Squalus (figs. 7, 8).

Just behind the otic capsules (and above the origin of the vagal nerve) in Notorynchus

NO. 3429

is the myelencephalic (medullary) chamber, marking the transition to the spino-occipital part of the brain stem (fig. 7D). A distinct medullary chamber is also present in other hexanchiforms, Squatina and many squa- loids, but it is poorly developed in Chlamy- doselachus, Heterodontus, galeomorphs, and batoids. Allis (1923: 163) remarked that in Chlamydoselachus the cranial cavity in this region “‘is slightly enlarged, and the large va- gal foramen lies in the middle of the lateral wall of this part of the cavity.”

MORPHOLOGY OF THE CRANIAL WALL

GENERAL: Originally, the neurocranium was scanned in the transverse plane, which clearly limits the extent of observations re- garding variation within the cranial walls. Digital manipulation of the original trans- verse sections allows the virtual braincase to be resliced in any direction (figs. 10-12). In the following description, the braincase is ex- amined by means of slices in the transverse, sagittal, and horizontal planes. In each case, the first image in each series of slices depicts the external form of the braincase and cor- responds to one of the earlier views of the. Since only a small sample of representative slices can be published, some minor struc- tures discussed in the following sections may not be evident in these views. Throughout this part of the description, references to the appropriate illustrations are simply indicated by key letters (A, B, etc.).

TRANSVERSE SECTIONS: All key letters in this part of the text refer to a series of trans- verse slices (fig. 10), in which the sequence passes anteroposteriorly (fig. 1OA corre- sponds to fig. 6A). These slices reveal chang- es in the cross-sectional shape of the endo- cranial cavity, as well as variation in the thickness of its walls and the topographic re- lationship of certain landmark features inside and outside the braincase.

The main endocranial cavity (beginning immediately behind the precerebral fonta- nelle) is low and wide at the level of the olfactory canals (D), but is appreciably nar- rower farther posteriorly. The telencephalic chamber is roughly triangular in section and widest dorsolaterally (F), but farther poste-

2004

riorly the mesencephalic region of the en- docranial cavity deepens ventrally, until it becomes abruptly shallower again at the dor- sum sellae (H). Behind the dorsum sellae the cerebellar chamber has a squarish cross sec- tion, although this shape is modified where passages for the octaval and glossopharyn- geal nerves extend laterally. The cross-sec- tional shape gradually becomes more round- ed farther posteriorly, and the medullary chamber has an oval cross section for much of its length between the main vestibular chambers, becoming almost circular as it passes through the occipital region, where it rapidly narrows to match the diameter of the dorsal nerve cord.

Slices through the rostral region reveal the depth of the precerebral fontanelle, as well as its flat floor (B). The lateral walls of the fontanelle extend dorsally and laterally around the ectethmoid chamber and olfactory capsule, and they also extend ventrally for a short distance to separate a concave area from the olfactory capsules ventrally (also seen in slice C). The floor of the fontanelle consists of the rostral and internasal plate (both probably representing anterior exten- sions of the trabeculae) and gradually thick- ens posteriorly. Although the dorsal and ven- tral surfaces of the fontanelle floor are both relatively flat, there is litthe correspondence between their features. The ethmoid region lacks a ventral keel in the floor of the inter- nasal region. The ventral extensions of the internasal plate in Notorynchus become low- er and wider posteriorly and eventually merge with the ventral margin of the ecteth- moid process extending from the postnasal wall (C, D). Cartilage overlying the olfactory capsule becomes thicker posteriorly, and the ectethmoid chamber, olfactory canal, and or- bitonasal canal are mostly developed within the thickness of the postnasal wall. The ect- ethmoid process extends ventrolaterally from the posterolateral margin of the postnasal wall and contains a large canal (discussed earlier) for the lateral ramule of the buccal + maxillary ramus (EB).

In the anterior part of the orbit, the brain- case wall is roughly triangular in transverse section, with a thick but narrow basicranium below the endocranial cavity (F). The inter- obital wall widens dorsally and expands lat-

MAISEY: BRAINCASE OF NOTORYNCHUS 25

erally, forming a broad supraorbital shelf. There is a distinct lateral thickening of the interorbital wall just below the optic fora- men, termed here the subocular ridge. This ridge becomes indistinct anteriorly and merg- es with the base of the ectethmoid process (E). In Chlamydoselachus, a corresponding ridge extends anterolaterally from behind and below the optic foramen to meet the post- nasal wall, where it merges with the posterior margin of the ectethmoid process, and part of the ridge defines the upper anterior margin of the orbital articulation behind the optic fo- ramen. In Notorynchus, this articular surface lies farther posteriorly and is not associated with the subocular ridge. The optic pedicel is not associated with this ridge but is situ- ated farther posteriorly, where it arises lat- erally in the same transverse plane as the dor- sum sellae (G, H). According to Allis (1923: pls. VII, IX), the optic pedicel in Chlamy- doselachus arises in a slightly more anterior position relative to the dorsum sellae. Supra- orbital shelves extend the entire length of the orbit and are pierced by canals for dorsal rami of the superficial ophthalmic nerve from the level of the precerebral fontanelle to the postorbital processes (E—G).

The orbital articulation forms a massive, laterally directed thickening at the basal an- gle in the basis cranii in the posterior half of the orbit, and the external form of the basi- cranium here does not correspond with the shape of the endocranial cavity. The floor of the cranium is thinner between the paired processes forming the basal angle, where the cartilage is almost three times thicker (G). The basicranial cartilage again thickens at the level of the dorsum sellae farther posteriorly, where it is penetrated by passages for the in- ternal carotids and the interorbital canal, and also contains the hypophyseal cavity (H, I).

In Notorynchus, the anterior ampulla is the anteriormost part of the skeletal labyrinth and is the first part of the vestibular system to appear in this series of transverse sections (1). In Squalus the anterior ampullae are again located between the postorbital pro- cesses, not behind them (fig. 8). Wells (1917: figs. 11-22) illustrated a series of transverse sections through the otic capsule of an adult Squalus braincase, some of which corre- spond to slices shown here. For example, her

26 AMERICAN MUSEUM NOVITATES

NO. 3429

Fig. 10. Notorynchus cepedianus neurocranium digitally resliced in transverse plane (corresponding to x—y axes of original CT scans). Sequence (A—U) passes from front to back of braincase. View A

corresponds to figure 4A. Scale bar = 10 mm.

anteriormost section (her fig. 11) shows the anterior ampulla connected to the ascending part of the anterior canal as in section I here.

In Notorynchus (as in modern elasmo- branchs generally), the medial capsular wall is chondrified throughout the length of the medullary region, separating the labyrinth al- most entirely from the endocranial cavity (fig. 9). According to Goodrich (1930) this wall develops mainly as an upgrowth from the parachordal. In Notorynchus, the utricular recess lies immediately behind the anterior ampulla, meeting it between the postorbital processes (J, K). A similar arrangement is found in Squalus (fig. 8) although it was not illustrated by Wells (1917). The octaval canal in Notorynchus branches away from the en- docranial cavity ventrally, passing through the medial capsular wall (L; see also fig. 9A). Wells (1917: fig. 13) showed a similar ar- rangement in Sgualus. However, her section

also shows a passage for the hyomandibular ramus passing posteriorly through the basi- cranial cartilage below the capsule at this level, whereas in Notorynchus there is no corresponding passage. The glossopharyn- geal canal seems to merge with the saccular chamber ventrally (M—O); in life, however, the glossopharyngeal nerve is separated from the vestibular apparatus by the membranous floor of the saccular chamber (Norris, 1929). Wells (1917: fig. 19) similarly identified the position of the glossopharyngeal canal within the floor of the chamber in Squalus.

In the midregion of the Notorynchus lab- yrinth, the medial capsular wall is penetrated by openings for the perilymphatic fenestrae dorsally (Q), much as Wells (1917: figs. 17, 18) showed in Sgualus. In Notorynchus, the roof of the medullary chamber (formed by the taenia medialis) is considerably lower than the capsules on either side, giving rise

2004

MAISEY: BRAINCASE OF NOTORYNCHUS 27

olf can

Fig. 10. Continued.

to the parietal fossa (L—S). By contrast in Squalus the parietal fossa is relatively shal- low (Wells, 1917: figs. 15,16). The ascending (preampullary) part of the posterior semicir- cular canal is seen between the endocranial cavity and glossopharyngeal canal (which begins to reemerge ventrally from the sac- cular floor; Q). It is not easy to identify the preampullary part of this canal in the sec- tions of Squalus shown by Wells (1917); comparison with slice Q suggests that the ca- nal labelled “‘VIII’’ in her figure 18 is the ascending preampullary region of the poste- rior canal, since the octaval nerve should not extend so far posteriorly. The canal shown immediately lateral to this in her section is probably the glossopharyngeal canal passing lateral to the canal. In Notorynchus, the la-

genar chamber forms a small recess in the floor of the saccular region just dorsolateral to the glossopharyngeal canal (Q). In Sgual- us, this seems to be represented by the later- almost part of the vestibular chamber de- picted by Wells (1917: fig. 18).

Transverse sections through the posterior part of the otic region in Notorynchus show the returning path of the external semicircu- lar canal between the upper and lower parts of the posterior canal, and the glossopharyn- geal canal is completely surrounded by car- tilage (S). In corresponding slices through the capsule in Squalus, Wells (1917: figs. 21, 22) showed the posterior semicircular canal sandwiched between the glossopharyngeal and vagal canals, but in Notorynchus the va- gal nerve leaves the cranial cavity slightly

28 AMERICAN MUSEUM NOVITATES

NO. 3429

Fig. 10. Continued.

farther posteriorly, and its canal does not therefore appear in section S. For this reason, the posterior semicircular canal is located in- stead between the medullary chamber and glossopharyngeal canal. Unfortunately, the origin of the vagal canal cannot be seen in the sections shown here, but its subsequent passage and the canal for its small dorsal ra- mus are visible (T). Canals for some of the spino-occipital nerves can also be observed, passing posterolaterally from the endocranial cavity. The anteriormost spino-occipital nerves emerged into the glossopharyngeal canal and shared this common exit from the braincase (T), but those situated farther pos- teriorly lay behind the canal and their pas- sages open directly onto the external surface of the braincase. Wells (1917) did not illus- trate the equivalent region in Squalus. SAGITTAL SECTIONS: All key letters in this part of the text refer to a series of slices in the sagittal plane (fig. 11). These slices are

sequenced beginning on the left side of the braincase (A, corresponding to fig. 5B), pass- ing progressively deeper to reach the midline (G), and then continuing partway into the right side. Slices B—H therefore reveal fea- tures at increasing depths within the brain- case, while slices I and J are the approximate equivalents of G and E respectively, the prin- cipal difference being that views G and F are directed medially, whereas I and J are di- rected laterally.

Slice B cuts the postnasal wall through the canal for the buccal branch of the anterodor- sal lateral line nerve. The postnasal wall meets the cartilage forming the roof of the olfactory and orbital regions dorsally, com- pletely separating the canals for the olfactory and optic nerves. The same slice also cuts through the supraorbital shelf above the or- bit, as well as the tip of the postorbital pro- cess. In slices closer to the midline, the or- bitonasal canal is seen passing through the

2004

MAISEY: BRAINCASE OF NOTORYNCHUS 29

Fig. 10. Continued.

postnasal wall from the orbit into the ecteth- moid chamber of the olfactory region (C). The main passage for the superficial oph- thalmic ramus penetrates the supraorbital shelf anterodorsally, and canals for several ramules of this nerve also extend upward (D). The anterior opening for the olfactory canal is visible in several of these slices but is not actually cut until slice E, and in sub- sequent slices the canal of the opposite side can be seen lateral to the precerebral fonta- nelle.

In slice B, the otic capsule is cut superfi- cially, revealing the outer part of the antero- posterior ramus of the external semicircular canal dorsally, and the lateral part of the sac- cular chamber farther ventrally. Behind this chamber, the glossopharyngeal canal is sec- tioned near to its exit on the posterolateral margin of the otic region. The vestibular pro- cess forms a distinct ventral projection be-

tween the exposed parts of the sacculus and glossopharyngeal canal. Slice C also reveals some features of the vestibular region, in- cluding sections through all three semicir- cular canals. The external semicircular canal (which is cut at two levels) is bracketed by the anterior and posterior canals and under- lain by the saccular and lagenar chambers (the latter is seen better in slice D). The dor- sal ramus of the vagal canal lies either im- mediately above the posterior semicircular canal (B) or just behind it (C, D), but the main vagal canal lies slightly more medially (its posterolateral wall is clipped in slices C and D and the entire canal is cut in E). Slice D exposes the external and anterior ampul- lae, as well as the utricular recess to which both ampullae are connected (in this section the external ampulla connects ventrally with the recess). The glossopharyngeal canal has

30 AMERICAN MUSEUM NOVITATES

NO. 3429

Fig. 10. Continued.

merged with the floor of the saccular cham- ber in slice D and is no longer distinct.

The posterior ampulla is the closest of the three to the midline and is not seen until slice E; it lies between the glossopharyngeal canal anteriorly and vagal canal posteriorly. The external semicircular canal merges with the vestibular chamber posteriorly, and its sub- sequent course through this region is indis- tinct (in life the membranous canal actually extends anteriorly within the vestibular chamber to meet the anterior semicircular ca- nal, as in all modern elasmobranchs, but this cannot be seen in endocasts; Schaeffer, 1981; Maisey, 2001A). Farther medially, cartilage forming the inner wall of the utricular recess

helps define the acustico-trigemino-facialis recess, which marks the point at which the octaval nerve enters the vestibular chamber (E). The glossopharyngeal canal is partly separated from the floor of the chamber and extends toward its medial wall. The ascend- ing (preampullary) ramus of the posterior semicircular canal is positioned deep beneath the main vestibular chamber (K J). The re- lationship of this canal to the perilymphatic fenestrae dorsally is not particularly evident in sagittal view, and it is seen better in trans- verse and horizontal sections.

The main endocranial cavity is seen only in deeper sagittal sections. The medial cap- sular wall bulges medially, obscuring the

2004

medullary chamber from view (F). The basis cranii is extraordinarily thick in the vicinity of the basal angle, enclosing the oculomotor canal and hypophyseal cavity, and the pas- sage for the optic nerve also passes through this cartilage just anterior to the basal angle. The hypophyseal pit is broad from side to side, but its anteroposterior extent is short (F—J). Its posterior limit is defined by the prominent dorsum sellae which overhangs the posterior part of the pit and the canals for the internal carotid arteries (G, I). Farther posteriorly, passages for several spino-occip- ital nerves pass through the basicranial car- tilage (EK J), and in slices close to the midline the notochordal canal extends anteriorly from the occipital cotylus (H). The medullary chamber is expanded dorsally before reach- ing the cotylus, forming the prominent myencephalic chamber. This chamber con- tains the exit of the vagal nerve in the midlle part of its lateral wall and the internal open- ings of the spino-occipital nerve canals along the lateral margins of its floor.

A narrow canal extends from the endocra- nial cavity into the basicranial cartilage at the ventral midline, just anterior to the main part of the hypophyseal cavity and near the pre- sumed anterior limit of the polar cartilages (H). The canal is tentatively identified as a vestigial hypophyseal duct, although it is separated by cartilage from the main hypo- physeal cavity farther posteriorly. In modern adult elasmobranchs, the hypophyseal fenes- tra typically becomes obliterated during on- togeny (this area is among the last to chon- drify; De Beer, 1931), although a fenestra is present in some adult Etmopterus (Holm- gren, 1940). An open hypophyseal fenestra and duct is present in many fossil elasmo- branchs (e.g., hybodonts, Orthacanthus; Schaeffer, 1981; Maisey, 1982, 1983), and closure of the duct in adults has been con- sidered a neoselachian synapomorphy (Mais- ey, 1984). A similar canal has also been il- lustrated in sagittal views of the braincase in Heptranchias and Hexanchus (Gegenbaur, 1872) and Echinorhinus (Shirai, 1992). In modern elasmobranchs, therefore, persis- tence of the hypophyseal canal has a very restricted systematic distribution and has so far been found only in orbitostylic sharks with a prominent basal angle. In modern hex-

MAISEY: BRAINCASE OF NOTORYNCHUS at

anchiforms, the hypophyseal fenestra is closed in the adult and the hypophyseal canal extends only partway into the basis cranii, but clearly in some squaloids the canal re- mains open. There is no evidence of a similar canal in the sagittal section of the adult braincase in Chlamydoselachus (Allis, 1923), although an open hypophyseal fenestra seems to be present in earlier stages of de- velopment (e.g., the 127-mm embryo figured by Holmgren, 1941: fig. 7).

Closure of the hypophyseal duct may be ontogenetically delayed in sharks with a strong basal angle, as the fenestra is still open in the latest embryonic stages of Squal- us (59 mm) and Etmopterus (55 mm) inves- tigated by Holmgren (1940). However, there seems to be no correlation between the tim- ing of closure and the elimination of the em- bryonic angle between the trabeculae and parachordals in taxa where the basal angle is absent. For example, according to Holmgren (1940) in “Scyllium” (= Scyliorhinus) the fenestra is closed in the 40-mm stage, and in Torpedo it disappears in the 3l-mm stage, although in both cases basicranial flexure be- tween the trabeculae and parachordals is not eliminated until later stages. In Raja, how- ever, closure of the fenestra is delayed until the 60-mm stage, after the basis cranii be- comes straightened. From a _ phylogenetic perspective, persistence of the canal into the adult is undoubtedly a primitive state for elasmobranchs generally, and presence of the canal could be considered a plesiomorphic character of orbitostylic sharks that was re- tained by hexanchiforms and Echinorhinus and lost in various other lineages. Since the duct is only present in modern elasmo- branchs with a basal angle (which is itself a unique character of hexanchiforms and squa- loids), however, its persistence here may be secondarily related to retention of the angle into adulthood, and therefore could be re- garded as an apomorphic (though homopla- seous) feature.

HORIZONTAL SECTIONS: All key letters in this part of the text refer to a series of slices in the horizontal plane (fig. 12). These are sequenced beginning dorsally (A, corre- sponding to fig. 3), and then pass progres- sively farther ventrally. These views provide a useful “‘plan view”? of the braincase and

a2 AMERICAN MUSEUM NOVITATES

NO. 3429

Fig. 11.

Notorynchus cepedianus braincase, digitally resliced in sagittal plane (corresponding to x—

z axes of original CT scans). Sequence (A—J) passes from left to right of braincase. View A corresponds

to figure 3B. Scale bar = 10 mm.

are particularly informative about the antero- posterior topographic relationship of endo- cranial structures to external morphology. Only the left side of each slice is shown, al- though each view extends slightly beyond the midline to ensure that no structures were inadvertently cropped.

The most superficial horizontal slices pass through the thickness of the supraorbital shelf, cutting passages for several branches

of the superficial ophthalmic ramus, as well as the olfactory canal and trochlear nerve (B). Canals for additional branches of the su- perficial ophthalmic ramus are also present in the postorbital process (C, D). Farther an- teriorly, features revealed in horizontal sec- tions include the orbitonasal canal within the postnasal wall (D, E), the optic and oculo- motor canals, and the trigemino-pituitary fossa in the back of the orbit (G). Deeper

2004

MAISEY: BRAINCASE OF NOTORYNCHUS 33

Fig. 10,

slices through the ethmoid region pass through the buccal canal for the anterodorsal lateral line nerve in the ectethmoid process (M).

In the otic region, the anterior semicircular canal is the first part of the labyrinth to be exposed in horizontal sections (B); by con- trast, the posterior canal is seen only in deep-

Continued.

er slices (D). At this level the anterior am- pulla and utricular recess have already been cut anteriorly, showing that the entire laby- rinth system is inclined anterodorsally rela- tive to the long axis of the braincase. The anterior ampulla is located adjacent to the base of the postorbital process (C, D). The paired perilymphatic canal connects with its

34 AMERICAN MUSEUM NOVITATES

Fig. 11.

corresponding posterior semicircular canal medially, within the lateral walls of the pa- rietal fossa (D, E). The ascending (pream- pullary) part of this canal lies immediately lateral to the perilymphatic fenestra. Thus, the posterior semicircular canal in Notoryn- chus describes a virtually complete circuit (Maisey, 2001a), and is completely separated from the anterior one as in Squalus (figs. 8, 9). The posterior semicircular canal extends into the cartilage forming the base of the pos- totic process, which also contains the glos- sopharyngeal canal (I, J). The external semi- circular canal curves progressively deeper around the exterior of the saccular region (E— J) and eventually passes between the saccular space and posterior semicircular canal.

The floor of the vestibular chamber has a very complex morphology, incorporating the lagenar chamber and the passage for the oc- taval nerve, as well as parts of the glosso- pharyngeal canal (technically the latter lies below the capsule, but now appears confluent with the chamber because the membranous floor to the capsule is unchondrified). The octaval canal enters the vestibular chamber

NO. 3429

Continued.

from the acustico-trigemino-facialis recess anteromedially, immediately behind the dor- sum sellae (H—J). A similar arrangement has been described in Chlamydoselachus (Allis, 1914, 1923), except that its recess is located slightly farther anteriorly with respect to the dorsum sellae, which lies immediately below the anterior margin of the recess rather than in front of it as in Notorynchus. The octaval canal branches within the thickness of the cartilage in the scanned specimen, suggesting that the anterior (utricular) and posterior (vestibular) branches of the nerve divided within the capsular wall (Maisey, 2001a). The octaval canals in the Squalus endocasts prepared by Schaeffer (1981) do not branch, suggesting that either the nerve divided with- in the labyrinth, or the two branches shared a common canal.

In progressively deeper slices (J—M) the dorsum sellae separates the hypophyseal chamber from the posterior part of the en- docranial cavity. The passage for the efferent pseudobranchial artery is situated in the lat- eral wall of the hypophyseal chamber (J). The canal enters the orbit immediately in

MAISEY: BRAINCASE OF NOTORYNCHUS 35

2004

‘MUU QT = eq 9[BOS “[ sINSY 0} spuodsaI0D VY MIA ‘A[UO Opts JOT ‘aSvOUTeIG JO W0}0q 0} do} WoT sassed (q-V) souonbag “(suvds JD [eurIstIo Jo saxe Z—A 0} Sutpuodssi109) oueyd [eyUOZLIOY UI padT[sor AT[eWSIp osvouTeIg snuvipadad snyIUKLOJON ‘TI “SIA

UBD JJO

36

AMERICAN MUSEUM NOVITATES

Fig. 12. Continued.

NO. 3429

MAISEY: BRAINCASE OF NOTORYNCHUS Bt

2004

‘ponunuoy $7] “SI

NO. 3429

AMERICAN MUSEUM NOVITATES

38

‘ponunuoy “ZI “SI

2004

front of the trigemino-pituitary fossa and di- rectly above the basal angle (see also fig. 5B). The horizontal slices clearly show that the floor of the vestibular chamber and med- ullary chamber lie in approximately the same plane (K, L). Slightly deeper within the hy- pophyseal chamber is the interorbital canal (M). Passages for the internal carotid arteries are located within the floor of the chamber (N, O). The vestigial hypophyseal canal dis- cussed earlier is difficult to detect in hori- zontal sections.

The course of the vagal nerve and its dor- sal ramus can be seen in slices F—L. The no- tochordal canal is exposed in deeper slices through the basis cranii (its slightly asym- metric appearance is probably caused by slight shrinkage and internal tearing of car- tilage when the specimen was originally pre- served). The glossopharyngeal canal passes deeper within the cartilage of the basicrani- um than any of the other cranial nerves, but it does not extend laterally into the vestibular process (M-P).

DISCUSSION

INTERPRETATION OF ENDOCRANIAL MORPHOLOGY FROM CT SCANS

The present work demonstrates that cranial morphology in modern elasmobranchs can be studied successfully using digitally pro- cessed images obtained from CT scans, re- vealing otherwise inaccessible details. Im- ages of the external cranial surface earlier in this work are certainly as informative as con- ventional photographs, and in some respects they are clearer (for example in emphasizing the positions of small foramina). Internal cra- nial features are also clearly revealed, and digitally processed images of the endocrani- um provide new morphological data that can contribute enormously to phylogenetic inves- tigations. Clearly, CT scanning cannot be construed as a replacement or an alternative for traditional dissected or cleared-and- stained preparations. However, the ease with which scans can be obtained, coupled with advances in computerized imaging, makes CT scanning an increasingly valuable re- source to comparative morphologists, espe- cially when rare or irreplaceable type and/or voucher specimens are involved.

MAISEY: BRAINCASE OF NOTORYNCHUS 39

One of the great advantages virtual imag- ing from CT scans has over conventional se- rial sectioning or grinding techniques is the potential to investigate the three-dimensional relationships between internal and external morphospace, in addition to those aspects which are only evident in sections. For ex- ample, the cranial endocast of Notorynchus has been combined with external views of the braincase in figure 13 to create composite dorsal and ventral views. The topographic re- lationships of certain endocranial and exter- nal features are readily seen, including the alignment of the anterior ampulla with re- spect to the postorbital process and hypophy- seal chamber, and the topographic relation- ship of external features in the otic region to the semicircular, vagal, and glossopharyngeal canals.

As might be expected, endocranial fea- tures of Notorynchus revealed by CT scan- ning agree closely with many of Schaeffer’s (1981) observations of silicone endocasts prepared from Squalus braincases (fig. 8). However, some differences are also noted in the topographic arrangement of certain struc- tures in the braincases of Notorynchus and Squalus. For example, in Squalus the hypo- physeal chamber is positioned farther ante- riorly relative to the postorbital process than in Notorynchus, although both taxa agree in the position of the process with respect to the anterior ampulla (cf. figs. 7, 8). The thickest part of the basicranium (forming the basal angle) is also positioned differently in these taxa; in Squalus it lies more or less directly below the hypophyseal chamber, a consid- erable distance behind the exit of the optic nerve, but in Notorynchus it is positioned farther anteriorly, below and just behind the optic nerve canal. The differing relative po- sitions of these structures invites further de- velopmental investigation and phylogenetic analysis. There may well be a developmental correlation between the size and extent of the basal angle and position of the hypophyseal chamber in orbitostylic sharks, and this may also have some bearing on the position and extent of the hypophyseal duct. Unfortunate- ly, preparing silicone endocasts obliterates the braincase and destroys the original rela- tionships between internal and external fea-

NO. 3429

AMERICAN MUSEUM NOVITATES

40

MUU QT = Ieq o[eo¢g do} 0} JOLIQUY “SMOIA (Gq) [BUDA pue (¥) [esIOp UI JSBOOPUD [LNIIA pUL (POAOUIOI OPIS IO] YIM BISCUIT [ENIITA [BUIO}X9) BsvoUTeIQ SNUDIpad|ad SNYIUKIOJON JO SISCUIT [eUSIp osytsoduroy ‘ET “BIA

USO

USe

Al ued JJO

2004

tures, whereas CT scanning and appropriate imaging protocols can easily document them.

The difficulties inherent in reconstructions of endocranial morphology in fossils have al- ready been discussed at length elsewhere (Stensi6, 1963; Maisey, 2001a) and will not be repeated here. Historically, the compara- tive basis for interpreting endocranial fea- tures in extinct vertebrates has been extreme- ly limited. For example, in many earlier stud- ies of placoderm braincases a modern elas- mobranch paradigm was followed, but ongoing investigations of the braincase in several extinct elasmobranchs (especially Pa- leozoic taxa) cast doubt on the validity of many morphological and phylogenetic state- ments in those earlier studies. Technological improvements over the past century (begin- ning with serial grinding, sectioning, and acid preparation, and now scanning and dig- ital preparation) have gradually increased the availability of phylogenetically informative data about endocranial morphology in early craniates, and they hold great promise for improving our knowledge even further in the future.

COMPARATIVE REMARKS

The following remarks are not intended to provide a detailed comparison or analysis of cranial morphology in elasmobranchs, or even in neoselachians; instead, they merely expand on some issues noted during the course of preparing this work. Since several of the features discussed here represent im- portant landmarks developmentally, a discus- sion of their systematic distribution and mor- phological relationships to other structures may be useful.

VENTRAL NASAL/ETHMOID KEEL: In Noto- rynchus there is no cartilaginous Keel in the ventral midline of the nasal or ethmoid re- gion (figs. 1B, 4, 6A). There is also no keel in the Upper Jurassic hexanchiform Notida- noides (Maisey, 1986), nor in the extinct neoselachian Synechodus (Maisey, 1985). By contrast, a well-developed median keel is present within the floor of the nasal septum in some extinct sharks (e.g., Hybodus, Or- thacanthus: Schaeffer, 1981; Maisey, 1983), and is closely associated with the anterior (ethmopalatine) part of the palatoquadrates.

MAISEY: BRAINCASE OF NOTORYNCHUS 41

Absence of such a keel has been considered a neoselachian synapomorphy (Maisey, 1984) but the situation is actually more com- plicated. In Squalus a narrow keel (rostral carina of Wells, 1917; ‘‘caréne sous-rostral’”’ of Devillers, 1958) extends along the ventral midline of the rostrum onto the nasal plate, and an extremely narrow keel-like nasal plate is present in some other squaloids (e.g., Scymnorhinus, Etmopterus, Deania; Holm- gren, 1941). Holmgren (1940) described a median “‘keel-process of the basis cranii’’ in Etmopterus embryos, and he also identified a raised *“‘median area”’ in the embryonic eth- moid region of many sharks and _ batoids (supposedly formed at the anterior contact between the paired trabecular plates). Addi- tionally, Shirai (1992) noted the presence of a Slender ‘“‘suborbital keel-process”’ in adult Centroscyllium. Thus, a median ethmoidal keel is not absent in all modern elasmo- branchs, although it may be absent in all gal- eomorphs.

Further complicating this issue, a median ethmoidal keel is apparently absent in some fossil non-neoselachian sharks (e.g., Tristy- chius, Akmonistion, Tribodus; Dick, 1978; Coates and Sequeira, 1998; Maisey and De Carvalho, 1997) and also in the primitive stem chondrichthyan Pucapampella (Maisey, 2001b; Maisey and Anderson, 2001). If a keel is primitively absent in chondrichthyans and early elasmobranchs, its presence may represent a synapomorphy of an extensive elasmobranch clade including (but not nec- essarily restricted to) xenacanths, hybodonts, and neoselachians, with the keel becoming independently reduced or lost in several neo- selachian lineages. Reduction or absence of the ethmoid keel in neoselachians (and in hy- bodonts such as Tribodus) is often (though not invariably) associated with the presence of a highly kinetic (orbitostylic or fully hyos- tylic) mandibular apparatus (sensu Maisey, 1980). It is also possible that the keel became secondarily better developed in certain taxa (e.g., Hybodus, Orthacanthus).

POSTNASAL WALL: A postnasal (antorbital) wall is well developed in many modern elas- mobranchs including Notorynchus (Fig. 5B), as well as in Hybodus and many Paleozoic sharks (Schaeffer, 1981; Maisey, 1983; Wil- liams, 1998). In some Paleozoic sharks the

42 AMERICAN MUSEUM NOVITATES

postnasal wall is not preserved although the olfactory canals are well developed imme- diately in front of the orbits and the wall may therefore have been weak or absent. PRESPHENOID LEDGE: A presphenoid ledge is absent in Notorynchus, but one is present in Squalus, immediately below the optic nerve and the posterior part of the telenceph- alon (Marinelli and Strenger, 1959: fig. 162). Neither Goodey (1910) nor Allis (1923) identified a presphenoid ledge in Chlamy- doselachus, although the basis cranii is dis- tinctly thickened at its expected position. Ac- cording to Gegenbaur (1872) a presphenoid ledge is absent in Squatina, batoids, Heter- odontus and some galeomorphs (e.g., Mit- sukurina) but is present in Heptranchias, Squalus, Dalatias, Deania, Galeus, and Mus- telus. He did not identify the ledge in Hex- anchus, although he showed a slight thick- ening of the basis cranii just anterior to the optic foramen, apparently corresponding to the ledge in Heptranchias. Shirai’s (1992) Sagittal sections suggest that a presphenoid ledge is present in Centroscyllium and Squal- iolus but not in Echinorhinus or Pristiopho- rus. It is concluded that presence of a pre- sphenoid ledge is highly variable in modern elasmobranchs and its distribution lacks a clear phylogenetic distribution, although it tends to be (1) present in most (but not all) orbitostylic sharks and (2) absent in batoids and most (but not all) galeomorphs. POSITION OF OPTIC FORAMEN: The floor of the endocranial cavity in modern elasmo- branchs sometimes includes a pocket be- tween the presphenoid ledge and dorsum sel- lae (e.g., Squalus: Marinelli, and Strenger, 1959: fig. 162). Where such a pocket is de- veloped it usually contains the endocranial opening of the optic foramen, low down on the interorbital wall (e.g., Dalatias, Deania, Squaliolus, Galeus). In Heptranchias, Squal- us, Centroscyllium and Mustelus the optic fo- ramen lies just at the anterolateral border of the pocket, and in Chlamydoselachus the fo- ramen is positioned just above the presphe- noid ledge, not within a pocket. The pre- sphenoid ledge and a pocket are both absent in batoids and some other modern elasmo- branchs (e.g., Sqguatina). In forms lacking a presphenoid ledge the pocket is poorly de- fined anteriorly and the optic foramen is

NO. 3429

more variable in position (e.g., in Squatina the optic foramen has an unusual position in the upper part of the interorbital cartilage, anterior to the trochlear foramen). As with the presphenoid ridge, no clear phylogenetic picture emerges from these observations, but there may be some correlation between the location of the optic foramen and the pres- ence or absence both of the pocket and the presphenoid ledge.

The arrangement of the optic, trochlear, and efferent pseudobranchial foramina in Chlamydoselachus differs from that of other modern elasmobranchs. Its optic foramen is located anterior to the exit of the superficial ophthalmic ramus, instead of below it (as in Notorynchus) or behind it (as in Heptran- chias and most other modern elasmo- branchs). The optic nerve describes an un- usually circuitous path in the orbit, curving anteriorly around the orbital process to reach the eyeball (as does the central retinal artery, which exits with the optic nerve; Allis, 1923: pls. 4, 19).

POSITION OF TROCHLEAR AND OCULOMOTOR FORAMINA: In modern elasmobranchs, the oc- ulomotor foramen is invariably located near the margin of the dorsum sellae, often at its junction with the interorbital cartilage, and it therefore provides a relatively fixed devel- opmental and morphological reference point. In many orbitostylic elasmobranchs the en- docranial exits for the optic, trochlear, and oculomotor nerves are almost equidistant from each other, describing an equilateral tri- angle with the optic foramen anteriorly, the oculomotor foramen posteriorly, and the trochlear foramen at its apex (e.g., Chlamy- doselachus, hexanchiforms, Squalus, Squal- iolus, Pristiophorus). An imaginary line drawn between the optic and oculomotor fo- ramina at the base of this triangle would pass through the foramen magnum in all these taxa. Different configurations of these open- ings occur in other taxa. For example, in Squatina the optic foramen is elevated to the same level as the trochlear foramen farther posteriorly, and in Galeus and Mustelus the optic foramen lies farther ventrally, below the level of the oculomotor foramen. In both cases, an imaginary line drawn between the optic and oculomotor foramina will pass obliquely and will miss the foramen magnum

2004

(passing below it in Sguatina and above it in Galeus and Mustelus).

There are important discrepancies between Goodey’s (1910) and Allis’ (1923) descrip- tions regarding the positions of the trochlear and oculomotor foramina in Chlamydosela- chus, apparently because the earlier author failed to observe the trochlear foramen cor- rectly. Thus, the trochlear foramen of Good- ey (1910) is the oculomotor foramen of Allis (1923), and the oculomotor foramen in Goodey’s (1910) account probably corre- sponds to the efferent pseudobranchial fora- men of Allis (1923). The arrangement shown by Allis (1923) is certainly more in line with that of many other elasmobranchs and is con- firmed by the embryonic Chlamydoselachus material illustrated by Holmgren (1941: fig. 8).

BASAL ANGLE AND ORBITAL ARTICULATION: A basal angle (figs. 4, 5) is present in all extant hexanchiforms and squaloids, and its lateral surface characteristically articulates with the palatoquadrate orbital process in the posterior part of the orbit (fig. 2). The on- togenetic development of the basal angle is best known in Squalus (Holmgren, 1941; EI- Toubi, 1949; Devillers, 1958, Jollie, 1971), where it supposedly results from retention of the original 45° orientation of the polar car- tilages with respect to the parachordals after the trabeculae become reoriented approxi- mately parallel with the parachordals, prob- ably because the embryonic basicranial car- tilages become fused before cephalic flexure is completely eliminated. The basal angle in hexanchiforms probably formed in similar fashion, since it is located at the presumed junction between the trabeculae, parachord- als, and polar cartilages Gf present), but this requires confirmation from ontogenetic in- vestigations.

The orbital articulation is positioned be- tween the optic foramen and efferent pseu- dobranchial foramina in all orbitostylic sharks, regardless of whether a basal angle is present (Maisey, 1980). Since the ophthalmic artery leaves the braincase via the optic fo- ramen, the orbital process also effectively al- ways Separates this vessel from the efferent pseudobranchial and ophthalmic arteries. Ac- cording to Allis (1923), in Chlamydosela- chus the optic nerve is confined to a groove

MAISEY: BRAINCASE OF NOTORYNCHUS 43

in the surface of the orbital articulation, into which it is pressed by the orbital process of the palatoquadrate. Conversely, Heptran- chias differs from other hexanchiforms and Chlamydoselachus in having a much narrow- er interorbital septum.

The orbital articulation is restricted to the posterior region of the orbit (1.e., at the junc- tion of the trabecular-polar cartilage and the parachordal) only in those orbitostylic elas- mobranchs in which a basal angle is present (e.g., hexanchiforms, squaloids; fig. 2). Un- fortunately, the development of the head in orbitostylic sharks has only been investigated in taxa possessing a basal angle. It has some- times been suggested that the orbital articu- lation in these forms is homologous with the basitrabecular process in other gnathostomes (e.g., Gardiner, 1984). The transverse shelf- like process of the polar cartilage in Squalus and Etmopterus has been identified as a bas- itrabecular process (Holmgren, 1940), but it clearly lies some distance behind the pala- toquadrate orbital process, which borders the trabeculae and does not extend posteriorly as far as the polar cartilage. In osteichthyans (including tetrapods), the basitrabecular pro- cess is positioned anterior to the palatine nerve (Goodrich, 1930: fig. 284). In Squalus, however, the orbital articulation is located much farther anteriorly relative to the pala- tine nerve, and in Chlamydoselachus and Squatina (both of which lack a basal angle) the articulation is even more remote from the palatine nerve. Furthermore, in Chlamydo- selachus the anterior end of the orbital artic- ulation overlaps the ectethmoid process, giv- ing rise to a broad suborbital platform in the anterior part of the orbit (Maisey, 1980).

Although the orbital articulation in orbi- tostylic sharks resembles the palatobasal one in osteichthyans, topographically, confine- ment of the orbital articulation to the poste- rior part of the orbit in hexanchiforms and squaloids may represent a derived state, probably related to development of the basal angle, secondarily resembling the topography of the basitrabecular process of other gna- thostomes. All modern sharks with a basal angle are orbitostylic, but not all orbitostylic sharks have a basal angle (e.g., Chlamydo- selachus, Squatina, Pristiophorus). In the Upper Jurassic hexanchiform Notidanoides,

tt AMERICAN MUSEUM NOVITATES

the basicranium is flat (even allowing for compression of the fossil during preserva- tion), and there is no sign of a basal angle. However, Notidanoides probably had an or- bital articulation, as the palatoquadrate has a strong orbital process; Maisey, 1986).

The consistent topographic relationship of the orbital articulation with respect to the op- tic foramen, efferent pseudobranchial fora- men, optic pedicel, and rectus musculature (whether or not a basal angle is present) has been regarded as a unique feature of orbito- stylic neoselachians (Maisey, 1980). The an- terior palatoquadrate articulation of other neoselachians (e.g., galeomorphs, Hetero- dontus) certainly does not have the same configuration and lies far anterior to the ef- ferent pseudobranchial foramen and _ polar cartilage. Conversely, modern orbitostylic sharks have (at best) only a ligamentous con- nection between the symphyseal region of the palatoquadrates and ethmoid region. However, the orbital articulation of Notoryn- chus and other orbitostylic sharks is widely regarded as homologous with the anterior (ethmopalatine) articulation of other elas- mobranchs. For example, Holmgren (1941) repeatedly referred to a “‘palatobasal pro- cess” with an articular surface for the pala- toquadrate not only in Chlamydoselachus, hexanchiforms, and squaloids, but also in Heterodontus and galeomorphs, and he clear- ly regarded the articulations as homologous in all these taxa. Wilga (2002) identified the articular surface for the palatoquadrate orbit- al process as an “‘ethmopalatine groove’’, even though in many orbitostylic sharks the articulation is far removed from the ethmoid region. Admittedly, the position of the orbital articulation in orbitostylic sharks is highly variable (Maisey, 1980); in Notorynchus it is located almost level with the postorbital pro- cesses, but in other modern hexanchiforms, Notidanoides, and Chlamydoselachus, it lies slightly farther anteriorly in the orbit. Thus, the orbital articulation is confined to the pos- terior part of the orbit only in sharks with a pronounced basal angle (hexanchiforms and squaloids). From an evolutionary viewpoint, therefore, it might be argued that this ar- rangement was primitively derived from a more anterior articulation and subsequently displaced toward the back of the orbit in

NO. 3429

those sharks which acquired a strong basal angle. Some morphological support for this is found in the position of the efferent pseu- dobranchial foramen, which is always close to the posterior margin of the orbital articu- lation in orbitostylic sharks and lies imme- diately behind the anterior articulation in ex- tinct sharks such as Cladodoides, Tamioba- tis, and Orthacanthus (Schaeffer, 1981; Maisey, 1983), even though the articulation occupies different positions in the orbit.

Several recently published phylogenies based on morphology have postulated a close phylogenetic relationship between orbitostyl- ic sharks and batoids (which are completely hyostylic and have no palatoquadrate artic- ulations with the braincase; Shirai, 1992, 1996; Carvalho, 1996; Carvalho and Maisey, 1996). According to those hypotheses, ba- toids are inferred to have lost the orbitostylic suspensorial arrangement. However, there is molecular evidence that batoids are not “‘de- rived sharks’? (Douady et al., 2003), and there is strong molecular support for an al- ternative phylogenetic hypothesis in which batoids and modern sharks are sister groups (Maisey et al., in press). In that scheme, or- bitostyly is a cladistically derived pattern that characterizes a monophyletic group of neo- selachians (orbitostylic sharks; Maisey, 1980), which includes hexanchiforms, Chla- mydoselachus, squaloids, squatinoids and pristiophoroids.

POSTORBITAL PROCESS, LATERAL COMMIS- SURE, AND PREFACIAL COMMISSURE: In actin- opterygians, the lateral commissure sur- rounds the lateral head vein and orbital artery (De Beer, 1937). By contrast, In modern elas- mobranchs the commissure encloses the lat- eral head vein but not the orbital artery (Holmgren, 1940, 1941). The extent of the lateral commissure, the degree to which it becomes chondrified, and its relationship to other structures (apart from the lateral head vein) all vary considerably in neoselachians. In Squatina and many squaloids (e.g., Oxy- notus, Scymnodon, Centrophorus, Somnio- sus), the lateral commissure completely sur- rounds the lateral head vein, although the jugular canal may be quite small (most squa- loids) or inflated (Sguatina). However, in Squalus and Centrophorus the lateral head vein and hyomandibular ramus pass through

2004

separate canals, and (according to De Beer, 1937) the medial wall of the jugular canal in Squalus is formed from the prefacial com- missure (derived ontogenetically from the basal plate), while its outer wall is formed by the lateral commissure. The prefacial com- missure in this form therefore separates the hyomandibular trunk from the jugular canal and the lateral commissure. Different condi- tions of the hyomandibular trunk, lateral head vein, the lateral and prefacial commis- sures in orbitostylic sharks include the fol- lowing: (1) Prefacial commissure and hyo- mandibular foramen located medial and slightly anterior to a “‘complete’’ lateral com- missure; the hyomandibular trunk accompa- nies the lateral head vein posteriorly through a jugular canal (e.g., Squatina). (2) Prefacial commissure and hyomandibular foramen lie ventral and medial to a “‘complete”’ lateral commissure; the hyomandibular trunk passes behind the lateral commissure and the pala- tine nerve exits below it (e.g., Squalus). (3) Prefacial commissure and hyomandibular fo- ramen are both ventral and medial to primary postorbital process (lateral commissure ab- sent); hyomandibular trunk and lateral head vein pass below the postorbital process (e.g., Notorynchus).

According to Schaeffer (1981) the hyo- mandibular trunk emerges from the braincase behind the postorbital process in Cladodoi- des and Orthacanthus. This may reflect a more posterior position of the prefacial com- missure in those taxa (similar to the situation described above in Squalus, but with a com- pletely chondrified lateral commissure; Mais- ey, 198327).

Slightly different arrangements are found in other modern elasmobranchs, although in most forms the lateral commissure is un- chondrified or absent and a jugular canal is not developed; for example, the prefacial commissure is absent in galeomorphs and the hyomandibular trunk may lack a discrete fo- ramen altogether (Holmgren, 1941). A ‘“complete”’ postorbital process (with a large jugular canal) occurs in many extinct elas- mobranchs (Schaeffer, 1981) and may rep- resent a primitive elasmobranch condition. Holmgren’s (1941) argument that the lateral head vein did not pass through the canal in

MAISEY: BRAINCASE OF NOTORYNCHUS 45

early sharks does not withstand critical ex- amination (see also Maisey, 1983: 27).

POSTORBITAL PROCESS AND OTIC CAPSULE: In Notorynchus the anterior ampulla of the vestibular region is located in the same trans- verse plane as the postorbital process (fig. 10 H, I). During the development of Squalus, Heterodontus, and Torpedo, the anteriormost part of the otic capsule extends forward as far as the lateral commissure (Holmgren, 1940: figs. 67, 119, 140), and in Etmopterus, Scyliorhinus, and Raja it can extend slightly farther anteriorly (Holmgren, 1940: figs. 86, 103, 109, 114, 128). In modern carcharhini- form sharks, however, the postorbital process is usually positioned on the lateral wall of the capsule, well behind the level of the an- terior ampulla (Compagno, 1988: fig. 6.12). In hybodonts the postorbital process is also located on the lateral wall of the otic capsule, although its anteroposterior extent is variable (e.g., Hybodus, Tribodus; Maisey, 1983; Maisey and Carvalho, 1997).

Orthacanthus and Tamiobatis both have a comparatively long otic region, and their an- terior ampulla is located some distance be- hind the postorbital process (Schaeffer, 1981). The anterior ampulla is also located well behind the postorbital process in the De- vonian stem chondrichthyan Pucapampella (Maisey, 2001b; Maisey and Anderson, 2001) and also in primitive actinopterygians (e.g., Kansasiella; Poplin, 1974: figs. 12, 22, 30), where the otic region is clearly not elon- gated. When compared with Notorynchus, the postorbital process in Orthacanthus is |o- cated anteriorly not only with respect to the otic capsule, but also to the passages for the facial, trigeminal and anterior lateral line nerves within the cranial wall. The mesen- cephalic chamber is proportionately longer in Orthacanthus and Tamiobatis than in Noto- rynchus and Squalus, but endocranial struc- tures of the orbital region also seem to be positioned farther posteriorly in these fossils, so these differences evidently involve more than just the proportions of the otic region. For example, in both Squalus and Notoryn- chus the optic lobe of the mesencephalic chamber has a midorbital position, whereas in Orthacanthus it lies farther posteriorly, in line with the postorbital process, suggesting

46 AMERICAN MUSEUM NOVITATES

a somewhat different position for the embry- onic lateral commissure.

To summarize, although the developmen- tal relationships of the lateral commissure, the prefacial commissure, and the otic cap- sule are highly conserved in gnathostomes, there is clearly some localized variation in their arrangement which may have phyloge- netic significance. Primitively, the otic cap- sule was probably located entirely behind the postorbital process in gnathostomes, but in neoselachians and some extinct sharks the lateral commissure is plastered onto the lat- eral or anterolateral wall of the capsule.

OTIC CAPSULE MORPHOLOGY: The anterior part of the medial capsular wall in Notoryn- chus contributes to the acustico-trigemino-fa- cialis recess (sensu Allis, 1914, 1923) within the endocranial cavity. According to Schaef- fer (1981: fig. 14), Orthacanthus lacks an acustico-trigemino-facialis recess. Such ab- sence could be explained by the incomplete chondrification of the medial capsular wall in Orthacanthus, as this cartilage helps define the wall of the recess in modern elasmo- branchs. In Cladodoides, however, CT scan- ning suggests that the prefacial commisure was chondrified and formed a wall behind the main exits of the facial and trigeminal nerves in the expected position of an acus- tico-trigemino-facialis recess. It is therefore entirely possible that in Orthacanthus an acustico-trigemino-facialis recess was pre- sent although its walls may not have chon- drified. The medial capsular wall is chondri- fied in hybodonts (e.g., Hybodus; Maisey, 1983: 32), and CT scanning suggests that an acustico-trigemino-facialis recess was also well developed.

Several specializations for low-frequency semidirectional phonoreception have been identified in modern elasmobranchs, many of which are reflected by the architecture of the Skeletal labyrinth cavity. Features include separation of the posterior semicircular canal, which describes an almost complete circle and is connected to the vestibular region by a single duct; separation of the utricular and saccular chambers; perilymphatic fenestrae; and a deep parietal fossa (Maisey, 2001a). In addition, the macula neglecta of modern elas- mobranchs is greatly enlarged, and is con- fined to an area within or immediately adja-

NO. 3429

cent to the duct connecting the posterior semicircular canal to the vestibular region, but this feature cannot be determined in CT scans of the braincase.

Orthacanthus and some other Paleozoic sharks lack most of these features, although a parietal fossa is usually present. The car- tilaginous lateral walls of the parietal fossa in Orthacanthus extend deep inside the braincase, however, and probably no direct communication existed between the macula neglecta and the perilymphatic canals as in modern elasmobranchs. As restored by Schaeffer (1981), the floor of the saccular chamber in Orthacanthus seems to be level with that of the medullary region, rather than extending below it. This arrangement char- acterizes modern elasmobranchs and certain hybodonts (e.g., Hybodus, Tribodus). By contrast, in some Paleozoic sharks the sac- cular chambers seem to have extended much farther ventrally (as in osteichthyans), al- though in fossils where the floor of the sac- culus was unchondrified the glossopharyn- geal canal may sometimes appear confluent with the saccular space, creating the illusion of a deeper saccular chamber.

PARIETAL (ENDOLYMPHATIC) FOSSA AND POSTERIOR DORSAL FONTANELLE: There is some developmental and paleontological ev- idence (reviewed in Maisey and Anderson, 2001) that the parietal fossa is homologous with the posterior dorsal fontanelle of os- teichthyans, and that both these openings represent parts of a primitive otico-occipital fissure. One plausible evolutionary scenario is that primitively the fontanelle was contin- uous with the otico-occipital fissure (as a posterior dorsal fontanelle; e.g., Pucapam- pella; Maisey, 2001b), and that more ad- vanced elasmobranchs acquired a posterior tectum which isolated the parietal fossa from the otico-occipital fissure (which is primi- tively retained in a number of Paleozoic sharks). According to this scenario, the pa- rietal fossa of modern elasmobranchs can be regarded as a posterior dorsal fontanelle that is separated from the primitive cranial fissure by a posterior tectum.

BASICRANIAL CIRCLE: In gnathostomes, the dorsal aorta generally divides on either side of the notochord, and the paired vessels (lat- eral aortae, internal carotids) anterior to the

2004

notochord subsequently converge and form a cephalic circle (Bertin, 1958), which is al- most entirely confined to the posterior (para- chordal) region of the basicranium. Differ- ences in the basicranial circle mostly involve the shape of the cephalic “‘circle’’, and the extent to which the arteries lie beneath or within the parachordal cartilage (in fact, con- siderably greater variation of both these var- iables has been noted than is at present doc- umented in the literature). In Notorynchus, most of the vessels forming the circle are sit- uated below (1.e., external to) the basicrani- um, and the internal carotids trace a bell- shaped path. A similar arrangement is also found in other modern elasmobranchs, and this was considered a derived condition by Schaeffer (1981). In some extinct elasmo- branchs, the lateral aortae are essentially Straight or divergent anteriorly, but do not adopt an outward curve (e.g., Hybodus, Cla- doselache, Tamiobatis, Orthacanthus). The cephalic circle was largely external to the braincase in hybodonts (Maisey, 1982, 1983), but in many Paleozoic sharks the lat- eral aortae and internal carotids are partly buried within the basicranial cartilage (Schaeffer, 1981; Williams, 1998). Thus, No- torynchus shares two important derived fea- tures of its cephalic circle with other neose- lachians; the outward swing of its lateral aor- tae, and the extensive exposure of the lateral aortae and internal carotids.

Holmgren (1942) noted that in modern elasmobranchs the internal carotid foramina may lie a considerable distance behind the hypophyseal foramen (from which they are separated by the precarotid commissure; De Beer, 1931). According to Gross (1937), in **Cladodus”’ wildungensis the internal carot- ids entered the braincase via the hypophyseal foramen, and CT scanning suggests that the precarotid commissure was absent (in prep- aration). The commissure may also be absent in the stethacanthid Akmonistion (Coates and Sequeira, 1998) and possibly in Tamiobatis vetustus, but in other Paleozoic sharks there is an extensive commissure separating the in- ternal carotid foramina and hypophyseal fe- nestra (e.g., Orthacanthus, ““Tamiobatis sp.”’, Cladoselache, Hybodus: Schaeffer, 1981; Maisey, 1983; Williams, 1998). The area around the hypophysis is among the last to

MAISEY: BRAINCASE OF NOTORYNCHUS 47

chondrify in the braincase of modern elas- mobranchs (De Beer, 1931), and the presence or absence of the precarotid commissure in extinct sharks may reflect variation in the timing of chondrification as well as actual differences in the basicranial aortic circle. OccIPITAL ARCH: Schaeffer (1981) pre- sented a convincing argument that having the occipital arch wedged to a greater or lesser degree between the otic capsules is a derived condition in elasmobranchs, not seen in other gnathostomes apart from a few teleosts (e.g., Clupea, mormyroids and some pholidophor- ids). In his phylogenetic discussion, he rec- ognized two states: (1) occipital arch pro- jecting behind capsules, with separate foram- ina for spino-occipital nerves, and (2) arch not projecting behind capsules, with most spino-occipital nerves leaving braincase through the vagal canal. The first of these states was said to characterize all chondri- chthyans; the second supposedly represented a synapomorphy of Hybodus and neosela- chians. However, in chimaeroids there is no evidence that the occipital block extends an- teriorly between the otic capsules to any ap- preciable extent, a condition which is clearly different from that found in some early sharks such as Orthacanthus, Tamiobatis, Akmonistion, and Cladodoides, where the oc- cipital arch does extend at least some dis- tance between the capsules (Schaeffer, 1981; Coates and Sequeira, 1998; Maisey, 2001C). Furthermore, the occiput projects appreciably behind the posterior wall of the otic capsule in some modern elasmobranchs (e.g., Noto- rynchus, Hexanchus). Schaeffer’s (1981: 59) proposal that “‘the relatively forward position of the occipital arch in neoselachians ac- counts for the fact that the occipitospinal nerves leave the braincase through the vagal canal or behind the condyles ... rather than separately along the sides of the occipital segment’? does not hold true for all neose- lachians or hybodonts, because in Notoryn- chus and Hybodus the spino-occipital nerves are located on the lateral surface of the oc- cipital block rather than entirely within the vagal canal. Collectively, these observations suggest that Schaeffer’s (1981) proposals re- garding occipital arch arrangement (especial- ly that the ‘“‘short’’ occipital block is a syn- apomorphy of neoselachians and hybodont

48 AMERICAN MUSEUM NOVITATES

sharks) require some reappraisal and refine- ment. There is certainly a tendency toward shortening the occipital region in neoselachi- ans and hybodonts, but no clear-cut phylo- genetic pattern has yet emerged.

In gnathostomes generally, the parachord- als project posteriorly and give rise to a var- iable number of neural arches (= occipital and preoccipital arches; Goodrich, 1930). In modern elasmobranchs (and most lower ver- tebrates) these become fused with the synotic tectum and posterior walls of the otic cap- sules. However this fusion evidently has a complex evolutionary history and seems to have occurred independently in osteichth- yans and chondrichthyans (perhaps several times; Maisey, 2001b). In many Paleozoic sharks the occipital region is separated from the otic region dorsally by the persistent oti- co-occipital fissure, but in modern elasmo- branchs and hybodonts this fissure is closed and the occipital arches are fused with the otic region.

ENDOCRANIAL FEATURES AS PHYLOGENETIC CHARACTERS

Only some of the morphological features noted in Notorynchus have been included in previous phylogenetic analyses of elasmo- branchs. The following list combines fea- tures which have previously been considered phylogenetically informative (.e., they have appeared at one time or another in previously published phylogenetic analyses), as well as many other whose phylogenetic potential is largely untested. Characters listed in paren- theses may not occur primitively within the group or within all extant representatives. An asterisk denotes an internal cranial feature.

A. Craniate Characters chondral neurocranium surrounding the brain chambered, platybasic endocranial cavity with hypophyseal opening* trabecular and parachordal cartilages forming basis cranii* otic capsule with ampullae, maculae, and semicircular canals* olfactory capsules B. Gnathostome Characters trabeculae with a rostral extension

NO. 3429

dorsum sellae and deep hypophyseal chamber* articulations for palatoquadrate and epi- hyal optic pedicel separate superficial ophthalmic division of trigemino-facialis complex lateral commissure endolymphatic ducts external semicircular canal* macula neglecta* (utricular recess)* orbitonasal canal ectethmoid chamber? C. Chondrichthyan Characters prismatically calcified cartilage, no chon- dral centers of ossification (single-unit braincase; not in certain stem chondrichthyans) D. Elasmobranch Characters precerebral fontanelle parietal fossa posterior position of hyomandibular ar- ticulation (specializations in otic region for low-fre- quency phonoreception; see Maisey, 2001a)* (Floor of saccular chamber level with medullary region)* (perilymphatic canals)* (ectethmoid process) (otic capsule protrudes between postor- bital processes)* (otic capsule with chondrified medial wall)* (glossopharyngeal canal)* (dorsal otic ridge lacks horizontal crests; see Coates and Sequeira, 1998) (posterior tectum present) (vestibular process) (postotic process) (occipital region wedged between otic capsules posteriorly) E. Neoselachian (= crown elasmobranch) Characters flat internasal plate (no ethmoidal keel) occipital hemicentrum (postorbital process ventrally unchondri- fied) (hypophyseal duct closed externally) E Orbitostylic Shark Characters orbital articulation extends into orbit

2004

(basal angle) (hypophyseal duct associated with basal angle)* G. Hexanchiform Characters postorbital articulation braincase poorly calcified

This list is clearly dominated by external features except at very generalized levels (e.g., craniates, gnathostomes). The only sig- nificant exception to this is in the ear region, where 16 apomorphic features (7 of which are associated with low-frequency semidirec- tional phonoreception) have been identified in the elasmobranch labyrinth (Maisey, 2001a; not listed separately above). One ad- ditional character of the elasmobranch ear re- gion is noted here (floor of saccular chamber level with that of the medullary region). This bias is undoubtedly artificial, since the ear region has been investigated in much more detail, and many features of the endocranial cavity will probably be discovered as inves- tigations proceed. Historically, endocranial morphology has not been widely utilized in higher-level systematic studies of elasmo- branchs (or other vertebrate groups), perhaps reflecting the difficulties in obtaining such data. Hopefully, the present study has dem- onstrated that endocranial morphology can be a rich source of phylogenetically infor- mative characters at many levels. It certainly deserves greater attention by future investi- gators using modern noninvasive scanning and imaging technologies.

A detailed discussion of all the characters listed above is beyond the scope of this work. The most fundamental craniate and gnatho- stome characters undoubtedly reflect phylo- genetically deep-rooted and relatively im- mutable patterns of gnathostome structure and development within major evolutionary clades. For example, there are fundamental differences in the skeletal labyrinth mor- phology of gnathostomes and agnathan cra- niates and between modern elasmobranchs and other gnathostomes (Maisey, 2001a). Similarly, in all gnathostomes the cranium supports or articulates with parts of the vis- ceral skeleton (e.g., palatoquadrate, epihyal), and each type of articulation has a distinct and unique relationship with parts of the braincase that have their own distinct and

MAISEY: BRAINCASE OF NOTORYNCHUS 49

unique ontogenetic origins. The hyomandib- ular articulation with the otic capsule is an- other conservative gnathostome feature, al- though its posterior position on the capsular wall is apparently a derived feature of elas- mobranchs.

The differing relationships between endo- cranial features associated with the mid- and hindbrain in modern and extinct elasmo- branchs probably involve many developmen- tal factors that are as yet poorly understood. The observations discussed briefly in the pre- vious section have broader implications for the role of fossils in future developmental studies, since they show that some patterns of structural and topographic relationship may not be revealed by modern ontogenetic studies and exist only in extinct taxa. This is not an entirely new revelation, but it certain- ly underscores the increasing relevance of paleontology to modern evolutionary-devel- opmental investigations.

Unfortunately, many of the features iden- tified in this work cannot be incorporated into phylogenetic analyses of craniates or gnathostomes until many more taxa are sam- pled. Such features therefore offer a poten- tially rich but still largely untapped source of characters at many phylogenetic levels, yet they show great promise in resolving prob- lematic phylogenetic issues. A comprehen- sive program of scanning and imaging en- docranial features in Recent and fossil elas- mobranchs (indeed, in all vertebrate groups) should provide many more characters for in- clusion in future phylogenetic analyses, and may perhaps improve our understanding of several crucial problems in lower vertebrate evolution.

ACKNOWLEDGMENTS

I thank Tim Rowe, Ritchard Ketchum and Matthew Colbert (University of Texas at Austin) for making the CT scans and prelim- inary image processing of the Notorynchus braincase used in this work. I also thank the California Academy of Sciences for making the specimen available for investigation. Several people at the American Museum of Natural History provided technical support to help bring this work to fruition, including David Reddy, Angela Klaus, Lorraine Meek-

50 AMERICAN MUSEUM NOVITATES

er and Frank Ippolito. I am grateful to Gavin Naylor, Michael Williams, Michael Gott- fried, Philippe Janvier, Gavin Young, Mar- celo de Carvalho, and the late Les Marcus for many constructive discussions about scanning, imaging, morphology, and elas- mobranch phylogeny, and I especially thank R. Glenn Northcutt and Michael Coates for their insightful reviews of this paper. I ac- knowledge the fine work on Notorynchus by Katherine Wolfram and Robert Hunt (Uni- versity of Nebraska), and am grateful for be- ing able to use some of Kathy’s unpublished observations here. This research was con- ducted at the American Museum and was supported by the Herbert and Evelyn Axel- rod Research Chair in Vertebrate Paleontol-

Ogy.

REFERENCES

Allis, E.P. 1913. The homologies of the ethmoidal region of the selachian skull. Anatomischer An- zeiger 44: 579-589.

Allis, E.P. 1914. The pituitary fossa and trigemi- no-facialis chamber in selachians. Anatomisch- er Anzeiger 46: 225-253.

Allis, E.P. 1923. The cranial anatomy of Chla- mydoselachus anguineus. Acta Zoologica 4: 123-221.

Bertin, L. 1958. Appareil circulatoire. In PP. Grassé (editor), Traité de zoologie, vol.13, fasc. 2: 1399-1458.Paris: Masson.

Blot, J. 1969. Systématique. Jn J. Pivetau (editor), Traité de paléontologie, IV (2), Gnathostomes, acanthodiens, placodermes, élasmobranches: 702—776.Paris: Masson.

Carvalho, M.R. 1996. Higher-level elasmobranch phylogeny, basal squaleans, and paraphyly. Jn M.L.J. Stiassny, L.R. Parenti, and G.D. Johnson (editors), Interrelationships of fishes: 35-— 62.London: Academic Press.

Carvalho, M.R., and J.G. Maisey 1996. Phyloge- netic relationships of the Late Jurassic shark Protospinax Woodward 1919 (Chondrichthyes: Elasmobranchii). Jn G. Arratia and G. Viohl (editors), Mesozoic fishes—systematics and pa- leoecology: 9—46. Munich: Pfeil.

Coates, M.I., and S.E.K. Sequeira. 1998. The braincase of a primitive shark. Transactions of the Royal Society of Edinburgh Earth Sciences 89: 63-85.

Compagno, L.J.V. 1977. Phyletic relationships of living sharks and rays. American Zoologist 17: 303-322.

Compagno, L.J.V. 1988. Sharks of the order Car-

NO. 3429

charhiniformes. Princeton, NJ: Princeton Uni- versity Press.

Corwin, J.T. 1989. Functional anatomy of the au- ditory system in sharks and rays. Journal of Ex- perimental Zoology, 2(suppl.): 62—74.

Daniel, J.E 1915. The anatomy of Heterodontus francisci. U. The endoskeleton. Journal of Mor- phology 26(3): 447-493.

Daniel, J.E 1934. The elasmobranch fishes. Berkeley: University of California.

Dean, B. 1909. Studies on fossil fishes (sharks, chimaeroids and arthrodires). Memoirs of the American Museum of Natural History 9: 209-— 287.

De Beer, G.R. 1924. Contributions to the study of the development of the head in Heterodontus. Quarterly Journal of Microscopical Science 68: 39-65.

De Beer G.R. 1931. The development of the skull in Scyllium (Scyliorhinus) canicula L. Quarterly Journal of Microscopic Science, n. ser. 74: 591-652.

De Beer, G.R. 1937. The development of the ver- tebrate skull. Oxford: Clarendon Press.

Devillers, C. 1958. Le crane des poissons. Jn PP. Grassé (editor), Traité de zoologie, vol.13, fasc. 1: 551—687.Paris: Masson.

Dick, J.R.E 1978. On the Carboniferous shark Tristychius arcuatus Agassiz from Scotland. Transactions of the Royal Society of Edinburgh 70: 63-109.

Douady, C.J., M. Dosay, M. Shivji, and M.J. Stan- hope. 2003. Molecular and phylogenetic evi- dence refuting the hypothesis of Batoidea (rays and skates) as derived sharks. Molecular Phy- logenetics and Evolution 26: 215-221.

El-Toubi, M.R. 1949. The development of the chondrocranium of the spiny dogfish, Acanthias vulgaris (Squalus acanthias). Part 1. Neurocra- nium, mandibular and hyoid arches. Journal of Morphology 84: 227-280.

Gadow, H. 1888. On the modifications of the first and second visceral arches, with especial ref- erence to the homologies of the auditory ossi- cles. Philosophical Transactions of the Royal Society of London, Series B Biological Scienc- es 179: 451-485.

Gardiner, B.G. 1984. The relationships of the pa- laeoniscid fishes, a review based on new spec- imens of Mimia and Moythomasia from the Up- per Devonian of Western Australia. Bulletin of the British Museum (Natural History), Geology 37(4): 173-428.

Garman, S. 1913. Plagiostomia. Memoirs of the Harvard Museum of Comparative Zoology 36: 1-528.

Gaupp, E. 1906. Die Entwickelung des Kopfske- lettes. In O. Hertwig (editor), Handbuch der

2004

vergleichenden und entwickelungen der Wir- belthiere: 573-874. Jena: Fischer.

Gegenbaur, C. 1872. Untersuchungen zur verglei- chenden Anatomie der Wirbelthiere. III. Das Kopfskelet der Selachier, ein Beitrag zur Er- kenntniss der Genese des kopfskeletes der Wir- belthiere. Leipzig: Engelmann.

Goodey, T. 1910. A contribution to the skeletal anatomy of the frilled shark, Chlamydoselachus anguineus Garman. Proceedings of the Zoolog- ical Society of London 1910: 540-571.

Goodrich, E.S. 1930. Studies of the structure and development of vertebrates. London: Macmil- lan.

Gross, W. 1937. Das Kopfskelett von Cladodus wildungensis Jaekel. 1. Theil. Endocranium und Palatoquadratum. Senckenbergiana 19: 80—107.

Holmgren, N. 1940. Studies on the head in fishes. Part I. Development of the skull in sharks and rays. Acta Zoologica (Stockholm) 21: 51-267.

Holmgren, N. 1941. Studies on the head in fishes. Part Il. Comparative anatomy of the adult se- lachian skull with remarks on the dorsal fins in sharks. Acta Zoologica (Stockholm) 22: 1—100.

Holmgren, N. 1942. Studies on the head in fishes. Part III. The phylogeny of elasmobranch fishes. Acta Zoologica (Stockholm) 23: 129-261.

Hotton, N. 1952. Jaws and teeth of American xen- acanth sharks. Journal of Paleontology 26: 489-500.

Iselst6ger, H. 1937. Das neurokranium von Rhina squatina und einige bemerkungen tiber ihre systematische stellung. Zoologische Jahrbiicher Abteilung fiir Anatomie und Ontogenie der Ti- ere 62: 349-394.

Jarvik, E. 1942. On the structure of the snout of crossopterygians and lower gnathostomes in general. Zoologiska Bidrag fran Uppsala 21: 235-675.

Jarvik, E. 1980. Basic structure and evolution of vertebrates. 2 vols.; 575 and 337 pp. London: Academic Press.

Jollie, M. 1971. Some developmental aspects of the head skeleton of the 35-37 Squalus acan- thias foetus. Journal of Morphology 133: 17— AO.

Luther, A. 1908. Untersuchungen tiber die vom N. trigeminus innervierte Muskulatur der selachier (Haie und Rochen). Acta Societatis Scientiarum Fennicae 36: 1-168.

Maisey, J.G. 1980. An evaluation of jaw suspen- sion in sharks. American Museum Novitates 2706: 1-17.

Maisey, J.G. 1982. The anatomy and interrela- tionships of Mesozoic hybodont sharks. Amer- ican Museum Novitates 2724: 1-48.

Maisey, J.G. 1983. Cranial anatomy of Hybodus basanus Egerton from the Lower Cretaceous of

MAISEY: BRAINCASE OF NOTORYNCHUS oH

England. American Museum Novitates 2758: 1-64.

Maisey, J.G. 1984. Chondrichthyan phylogeny: a look at the evidence. Journal of Vertebrate Pa- leontology 4(3): 359-371.

Maisey, J.G. 1985. Cranial morphology of the fos- sil elasmobranch Synechodus dubrisiensis. American Museum Novitates 2804: 1-28.

Maisey, J.G. 1986. The Upper Jurassic hexan- choid elasmobranch Notidanoides n.g. Neues Jahrbuch fiir Paliontologie, Abhandlungen 172: 83-106.

Maisey, J.G. 1989. Visceral skeleton and muscu- lature of a late Devonian shark. Journal of Ver- tebrate Paleontology 9(2): 174-190.

Maisey, J.G. 2001a. Remarks on the inner ear of elasmobranchs and its interpretation from skel- etal labyrinth morphology. Journal of Mor- phology 250: 236-264.

Maisey, J.G. 2001b. A primitive chondrichthyan braincase from the Middle Devonian of Boliv- ia. In PE. Ahlberg (editor), Major events in ear- ly vertebrate evolution: paleontology, phyloge- ny, genetics, and development: 263—288. New York: Taylor and Francis.

Maisey, J.G. 2001c. CT scan reveals new cranial features in Devonian chondrichthyan ‘‘Clado- dus’’ wildungensis. Journal of Vertebrate Pale- ontology 21(4): 807-810.

Maisey, J.G., and M.E. Anderson. 2001. A prim- itive chondrichthyan braincase from the early Devonian of South Africa. Journal of Verte- brate Paleontology 21(4): 702-713.

Maisey, J.G., and M.R. Carvalho. 1997. A new look at old sharks. Nature 385: 779-780.

Maisey, J.G., G.J.P. Naylor, and D.J. Ward. In press. Mesozoic elasmobranchs, neoselachian phylogeny and the rise of modern elasmo- branch diversity. Jn G. Arratia and A. Tintori (editors), Mesozoic fishes, vol. 3. Munich: Pfeil.

Marinelli, W., and A. Strenger. 1959. Vergleichen- de Anatomie und Morphologie der Wirbelthi- ere. III. Lieferung, pt. 2, Squalus acanthias: 179-308. Vienna: Franz Deuticke.

Morsli, H., D. Choo, A. Ryan, R. Johnson, and D.K. Wu. 1998. Development of the mouse in- ner ear region and origins of its sensory organs. Journal of Neuroscience 18: 3327-3335.

Moy-Thomas, J., and R.S. Miles. 1971. Palaeo- zoic fishes. Philadelphia: Saunders.

Norris, H.W. 1929. The parietal fossa and related structures in the plagiostome fishes. Journal of Morphology 48: 543-561.

Northcutt, R.G. 1978. Brain organization in the cartilaginous fishes. Jn E.S. Hodgson and R.F Mathewson (editors), Sensory biology of sharks, skates and rays: 117-193. Arlington,

ups AMERICAN MUSEUM NOVITATES

VA: Office of Naval Research, Department of the Navy.

Northcutt, R.G., and W.E. Bemis. 1993. Cranial nerves of the coelacanth Latimeria (Ostei- chthyes: Sarcopterygii: Actinistia) and compar- isons with other Craniata. Brain Behavior and Evolution 42 Suppl. 1: 1-76.

Patterson, C. 1982. Morphological characters and homology. Jn K.A. Joysey and A.E. Friday (ed- itors), Problems of phylogenetic reconstruction: 21—74,London: Academic Press.

Poplin, C. 1974. Etude de quelques paléoniscidés Pennsylvaniens du Kansas. Paris: Cahiers de Paléontologie, Editions du Centre National de la Reserche Scientifique.

Romer, A.S. 1966. Vertebrate paleontology. Chi- cago: University of Chicago Press.

Rowe, T., J. Kappelman, W.D. Carson, R.A. Ket- cham, and C. Denison. 1997. High-resolution computed tomography: a breakthrough for earth scientists. Geotimes 1997 (Sept.): 23-27.

Schaeffer, B. 1981. The xenacanth shark neuro- cranium, with comments on elasmobranch monophyly. Bulletin of the American Museum of Natural History 169: 3—66.

Sewertzoff, 1899. Die entwickelung des selachier- schadels. Ein Beitrag zur Theorie der korrela- tiven Entwickelung. Festschrift Carl von Kuppfer: 281-320. Jena: Fischer.

Shirai, S. 1992. Squalean phylogeny: a new framework of ‘squaloid’ sharks and related taxa. Sapporo: Hokkaido University Press.

Shirai, S. 1996. Phylogenetic interrelationships of neoselachians (Chondrichthyes: Euselachii). Jn M.L.J. Stiassny, L.R. Parenti and G.D. Johnson (editors), Interrelationships of fishes: 9-34. London: Academic Press.

NO. 3429

Shute, C.C.D. 1972. The composition of the ver- tebrae and the occipital region of the skull. Jn K.A. Joysey and T:S. Kemp (editors), studies in Vertebrate evolution: 21—34. Edinburgh, Oliver and Boyd.

Stensi6, E. 1963. The brain and the cranial nerves in fossil, lower craniate vertebrates. Skrifter ut- gitt av det norske Videnskaps-Akademi i Oslo 1 Ny Serie 13: 5—120.

Turner, S., and G.C. Young. 1987. Shark teeth from the Early-Middle Devonian Cravens Peak Beds, Georgina Basin, Queensland. Alcheringa 11: 233-244.

Watson, D.M.S. 1937. The acanthodian fishes. Philosophical Transactions of the Royal Society of London, Series B_ Biological Sciences 228(549): 49-146.

Wells, G.A. 1917. The skull of Acanthias vulgar- is. Journal of Morphology 28: 417-443.

Wijhe, J. W. van. 1922. Friihe Entwickelungsstu- dien des Kopf- und Rumpfskeletts von Acan- thias vulgaris. Bijdragen tot de Dierkunde 22: 271-298.

Wilga, C.D. 2002. A functional analysis of jaw suspension in elasmobranchs. Biological Jour- nal of the Linnean Society 75(4): 483-502.

Williams, M.E. 1998. A new specimen of Tam- iobatis vetustus (Chondrichthyes, Ctenacanthoi- dea) from the late Devonian Cleveland Shale of Ohio. Journal of Vertebrate Paleontology 18(2): 251-260.

Wolfram, K.E. 1984. The functional anatomy of the jaw suspension of Notorynchus cepedianus (Chondrichthyes, Selachii), with application to fossil forms. Unpublished M.Sc. thesis, Uni- versity of Nebraska. 249 pp.

Young, J.Z. 1962. The life of vertebrates. 2"4 ed. Oxford: Oxford University Press.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates and Bulletin published during the last five years are available at World Wide Web site

http://library.amnh.org. Or address mail orders to: American Museum of Natural History Library, Central Park West at 79th St., New York, NY 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: scipubs @amnh.org

This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).