AMERICAN MUSEUM Novitates

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024

Number 3119, 86 pp., 46 figures, 8 tables | January 27, 1995

Phylogenetic Systematics of Extant Chimaeroid Fishes (Holocephali, Chimaeroidei)

DOMINIQUE A. DIDIER!

CONTENTS

Abstract: 5 2b) cn cee ans no od cin, Lo ee, ee De Bene Se eek eee ea Lee 2 TMITPOGV CUO he 100 2, SRR eee oe lee hd A ee lets 3 sede eee Og Mees And so oh By 2 PACKNOWIEGSMETIES irs Hey Nn at a nd Sead iaaar a ekul Sadien sehen Bake cdaeatie 5 PRO BPEVTALIOTIS: To Bie B baclnsnecis sco eon nnn SO oe ape hewn ela Stet dg ean. 2 atta gc Nthenaeg ER (SS ete 5, gun gee 5 Maternials:and- Methods: «2.00.66 ove bbe qces wien goteg boy eee pose WIRE Sues a Uw Grode ate en see ple be 7 Review of Holocephalan Systematics ....... 0.0.0... ccc ccc cc eee eee eee eens 11 Taxonomic Review of Chimaeroid Species .......... 0.00... ccc cece eee eee tees 13 Comparative Anatomical Studies ............0 0.00... ccc eee eee eens 16 Gomments on Development — .....ccc i owed be ep eran Dann beeen e ow nee wee ee al 16 EO CASES nn cen he eee ee ae, cg ehee 50, AER, weroxe hes cea PR ae Be a eR Tee 20 Exterial Features-of Adults’... 4.55244 cuss ba ae bce ere ble et aca dc oe vl bok ee eae 22 SKE CAPA TALON Vans eran. Sena Mice a Be ee ted Seine Sealey aliens abdog te easy. cence Audterk A 31 Neurocranium, Jaws, and Labial Cartilages ............0... 0.0.0... cece eee eee 32 Gra ATOMS S fs oa tesa he t § topes pSovonly obi AAS. eae eat enedcadirseg Wa MM tn 4 Ua Soacon ys 4 eters 38 Vertebral Column and Unpaired Fins .............. 0.00... cc cece een eeee 40 Paired Fins-and: Girdles. 8 aE ote ee & bin, Sey Poveda el eels ween daa reac wemee 41 Secondary Sexual Characters ........... 0.0000. e cee eee nen e tenet eneaes 43 TOG BAGG Sime aie beast Nos ieee, et ee AR SE beter es ec eaern art eee Gl Ne oe ner 46 Wesco a (urea 51% esaececsts sn eateteserne: BLA ert ae A ad che SEE oo hy Ry dk, pellta wich ndabae Sich: 53 Character Analysis ce. 5 ob teerh ccd Meat noel deeoria dng OE aid ges Benet Ae ot hay deeb aha ona, wat 67 IDISCUSSION 5... See etn AU, ae eae. BL ye eerie Se Latae altel ena weg ee UA aay 75 TAXONOMIC SUMMALY jee oxi e sodesvcges esi 4 eld ca ve wD en Whee BL oe we eden wna 4 anvie rll Balan es 77 INGIEREM COST fy Beale ie Lee Bart co setta ghia WO PLS ote ee ee Pa 78

' The University of Massachusetts, Department of Zoology, Amherst, MA 01003-0027; Present Affiliation: Assistant Curator, Department of Ichthyology, The Academy of Natural Sciences, 1900 Benjamin Franklin Parkway, Phila- delphia, PA 19103-1195.

2 AMERICAN MUSEUM NOVITATES

NO. 3119

ABSTRACT

This investigation analyzes generic- and famil- ial-level phylogenetic relationships of extant Hol- ocephali using morphological characters. The six genera studied—Callorhinchus, Rhinochimaera, Harriotta, Neoharriotta, Chimaera, and Hydro- lagus—belong to the suborder Chimaeroidei and are the only living representatives of the class Hol- ocephali. Details of relationships among the Re- cent taxa have not been considered in any phy- logenetic hypothesis. This comparative anatomical analysis, combined with developmental infor- mation, offers a new approach. The comparative morphology of the lateral line canals, skeleton, tooth plates, secondary sexual characters, and musculature of all six living genera of chimaeroid

fishes is described and compared with living elas- mobranchs, the nearest Recent outgroup. Devel- opment of the jaws, hyoid arch, and ethmoid canal is briefly described for Callorhinchus milii of the family Callorhynchidae. Callorhinchus is the most primitive living chimaeroid, and the superfamily Chimaeroidea is erected to include the remaining five genera: Rhinochimaera, Harriotta, Neohar- riotta, Chimaera, and Hydrolagus. A phylogeny of higher-level chimaeroid relationships is hy- pothesized on the basis of anatomical characters, and a new classification of chimaeroid fishes is proposed to reflect this phylogenetic interpreta- tion.

INTRODUCTION

The chimaeroid fishes (order Chimaeri- formes in this work) belong to the class Hol- ocephali and are regarded as an obscure lin- eage of mostly deep-water cartilaginous fishes. Holocephali and Elasmobranchii are widely considered to belong to a monophyletic Chondrichthyes, a relationship supported by numerous skeletal and soft tissue characters (Schaeffer, 1981; Maisey, 1984, 1986: table 1). Nevertheless, the early evolution and re- lationships of holocephalans are still poorly understood. My primary objective in this in- troduction is to condense a vast and com- plicated history of systematic research and to briefly discuss the history of holocephalan classification.

Numerous studies have focused on the de- scription and classification of Holocephali (Miller, 1834, 1844; Newberry and Wor- then, 1870; Saint John and Worthen, 1883; Zittel, 1887; Garman, 1901, 1904, 1908, 1911; Dean, 1906, 1909; Woodward, 1889, 1921; Hussakof, 1912; Nielsen, 1932; Moy-Tho- mas, 1936a, b, 1939; Obruchev, 1967; Berg, 1965; Arambourg and Bertin, 1958; Orvig, 1962, 1980; Stensid, 1963; Patterson, 1965, 1968; Saint-Seine, Devillers and Blot, 1969; Lund, 1977, 1986a, b; Zangerl, 1973, 1981; Zangerl and Case, 1973; Maisey, 1984, 1986). Historically, the focus of this research has been on fossil taxa, and there has been no prior systematic study focusing on the rela- tionships of extant holocephalans. My pur-

pose is to examine the interrelationships of living forms through detailed comparative anatomical studies. Once monophyletic groups are established, the classification of fossil taxa, side by side with living forms, in a modern phylogenetic context will be facil- itated.

There are two hypotheses on the origin of Holocephali. The first and most generally ac- cepted scenario is that holocephalans have evolved from some lineage of bradyodont sharks (Woodward, 1921). The second hy- pothesis suggests that holocephalans are most closely related to placoderms (Stensi6, 1925, 1936; Orvig, 1962). Because of the historic significance of this question, these two hy- potheses warrant review.

Woodward (1921) originally erected the order Bradyodonti to include five families of Paleozoic chondrichthyans: Petalodontidae, Psammodontidae, Copodontidae, Cochlio- dontidae, and Edestidae. These families are referred to casually as the “‘bradyodont taxa,” an assemblage of cartilaginous fishes that possessed tooth plates with a crown of hy- permineralized tissue and had slow replace- ment of the teeth (Woodward, 1921). Nielsen (1932) recognized the inconsistency of Woodward’s definition of the Bradyodonti because some bradyodonts had rapid replace- ment of teeth. Despite this problematic def- inition, the Bradyodonti remained in the lit- erature as a convenient taxonomic group. It

1995

DIDIER: CHIMAEROID FISHES 3

TABLE 1 Characters That Support a Monophyletic Chondrichthyes

1. Prismatic calcification of cartilage (Schaeffer and Williams, 1977; Schaeffer, 1981; Zangerl, 1981;

Maisey, 1986)

. Labial cartilages (Schaeffer and Williams, 1977)

2 3. Ceratotrichia in paired and unpaired fins (Schaeffer and Williams, 1977) 4. Dermal skeleton represented by denticles with a particular pattern of enameloid, dentine and basal

tissue (Schaeffer and Williams, 1977)

. Rectal salt gland in living forms (Schaeffer and Williams, 1977) . Mixipterygial claspers in males (Schaeffer and Williams, 1977; Maisey, 1986)

. Efferent pseudobranchial artery over trabeculae (Schaeffer, 1981) . Occipital arch between auditory capsules projecting behind the capsule with separate foramina for

5 6 7. Precerebral fontanelle and fossa (Schaeffer, 1981) 8 9

occipitospinal nerves (Schaeffer, 1981) a

10. Otico-occipital proportions less than ethmo-orbital portion (Schaeffer, 1981) :

11. Scales with neck canals (Maisey, 1986)

12. Posteriorly directed median basibranchial copula (Maisey, 1986)

13. Dorsal fin with basal cartilage (Maisey, 1986)

14. All premetapterygial radials of pectoral fin articulate with basals (Maisey, 1986) 15. Pelvic basipterygium articulates with all but the anteriormost few radials (Maisey, 1986)

16. Metameric suprarenal ganglia (Maisey, 1986)

17. Multiple jointing of pectoral radials (Maisey, 1986) 18. Anterior median sinus between anterior and posterior cardinal veins (Maisey, 1986)

19. Endolymphatic duct (Maisey, 1986)

20. Mesoadenohypophysis with ventral lobe detached from main body of the pituitary (Maisey, 1986) 21. Nidamental gland secretes horny egg case (Maisey, 1986) 22. Replacement tooth rows attached to mesial surface of jaws by basement membrane; linguo-labial

tooth replacement (Maisey, 1986)

23. Teeth with specialized nutritive foramina in basal plate (Maisey, 1986) b

4 May be primitive; see discussion in Maisey (1983).

b secondarily modified in Holocephali (Maisey, 1986).

has since been established that the micro- structure of teeth is not a useful character for systematics (Radinsky, 1961; @rvig, 1967). Asa result, Bradyodonti is no longer accepted as a valid taxonomic group (Romer, 1966; Bendix-Almgreen, 1968; Patterson, 1968; Lund, 1986a) and the relationships among Woodward’s five bradyodont families have remained obscure.

Complete or partial skeletal material of holocephalans is extremely rare and almost all fossil holocephalans are known from teeth, denticles, or spines (e.g., Zittel, 1887, 1932; Newberry and Worthen, 1870; Saint John and Worthen, 1883; Woodward, 1889, 1891; Reis, 1895; Dean, 1909; Hussakof, 1912; Chap- man, 1918; Nielsen, 1932; Patterson, 1965; Obruchev, 1967; Ward, 1973; Ward and Mc- Namara, 1977; Lund, 1982, 1986b; Duffin,

1984). As a result, holocephalans have been classified primarily on the basis of tooth plate characteristics. Because holocephalans have tooth plates that are not shed, with a micro- structure like that of bradyodont teeth, most workers have allied them with at least one of the bradyodont taxa (e.g., Moy-Thomas, 1939; Obruchev, 1967; Berg, 1965; Aram- bourg and Bertin, 1958; Patterson, 1965, 1968; Lund, 1977; Zangerl, 1981).

In 1925, Stensid proposed that holoce- phalans were closely related to placoderms. This suggestion was supported by his obser- vation that some placoderms, particularly the ptyctodontid arthrodires (e.g., Ctenurella), shared certain anatomical similarities with holocephalans, which include: the joint be- tween the head and vertebral column, an operculum, the shape of the pectoral girdle,

4 AMERICAN MUSEUM NOVITATES

an autostylic jaw, tooth plates, gill arches concentrated underneath the neurocranium, a well-developed rostral portion of the neu- rocranium, first dorsal fin with a basal car- tilage and fin spine, and the general body shape. Orvig (1962) discovered several ad- ditional characteristics present in both hol- ocephalans and ptyctodont arthrodires: a synarcual, prepelvic spines, pelvic claspers, the position and shape of the mandible, and paired rostral processes. Detailed compara- tive morphological investigations have prompted others to consider a relationship between holocephalans and arthrodires on the basis of these and other anatomical similar- ities (Holmgren, 1942; Westoll, 1962; Stahl, 1967; Jarvik, 1980). Based on evidence pre- sented by Stensid (1925, 1936) that chon- drichthyans were related to placoderms, the group Elasmobranchiomorphi was erected by Jarvik (1955) to include elasmobranchs, hol- ocephalans and placoderms. The implication of this grouping was that placoderms and chondrichthyans are closely related and may share a common ancestry.

Miles and Young (1977), in a study of pla- coderm relationships, dismissed the anatom- ical evidence for a close relationship to hol- ocephalans on the grounds that placoderm characters have been interpreted based on a holocephalan model. It was therefore con- cluded that the characters presented by Sten- sid (1925) and Orvig (1962) appear to be con- vergent features. Recent studies of the systematics of placoderms support a closer relationship between placoderms and os- teichthyans than between placoderms and chondrichthyan fishes (Forey, 1980; Gardi- ner, 1984; Forey and Gardiner, 1986). This evidence against a relationship between pla- coderms and holocephalans is reflected in a recent redefinition and classification of Elas- mobranchiomorphi (Jarvik, 1980: 324) that excludes holocephalans.

In a thorough analysis of holocephalan re- lationships, Patterson (1965) focused almost exclusively on the fossil chimaeroids, includ- ing squalorajoids, myriacanthoids, and men- aspoids. The goal of Patterson’s (1965) anal- ysis was to examine the evidence for a placoderm-holocephalan relationship as sug- gested by Stensi6 (1925, 1936) and supported by Orvig (1962). Patterson (1965: 213) con-

NO. 3119

cluded that holocephalans are not derived di- rectly from placoderms, but he cautiously suggested that holocephalans may share a common ancestor with arthrodires. This con- clusion is reflected in his grouping of Holo- cephali within the superclass Elasmobran- chiomorphi, which by his definition includes all Selachii and Arthrodira (Patterson, 1965: 105).

As a complement to paleontological in- vestigation it is important to approach the problem of holocephalan evolution through neontological studies. Many workers have contributed to our understanding of holoce- phalan relationships through studies of ex- tant taxa (Vetter, 1878; Garman, 1888, 1904; Dean, 1895, 1904a, b, 1906; Cole, 1896a, b; Jungersen, 1899; Schauinsland, 1903; Cole and Dakin, 1906; Luther, 1909; Burlend, 1910; Reese, 1910; Allis, 1912, 1917, 1926; Shann, 1919, 1924; Leigh-Sharpe, 1922, 1926; Rabinerson, 1925; Kesteven, 1933; de Beer and Moy-Thomas, 1935; Edgeworth, 1935; Holmgren, 1940, 1941, 1942; Patter- son, 1965; Stahl, 1967; Ribbink, 1971; Rai- kow and Swierczewski, 1975; Jarvik, 1980). Embryological material of chimaeroids has historically been difficult to obtain, and only Hydrolagus colliei and Callorhinchus milii inhabit relatively shallow near-shore waters, making it possible to obtain developmental material for these two species. Four studies have focused on aspects of development (Schauinsland, 1903; Dean, 1903, 1906; de Beer and Moy-Thomas, 1935; Kemp, 1984). Of these studies Schauinsland (1903, Callor- hinchus milii) and Dean (1903, 1906, Hy- drolagus colliei) have studied a series of em- bryos. Kemp (1984) described tooth plate development in Callorhinchus milii and de Beer and Moy-Thomas (1935) studied the development of the head in a single 95-mm Callorhinchus embryo.

Despite these efforts to describe and un- derstand the anatomy of living chimaeroid fishes, there have been no comparative stud- ies of all extant genera and none that has attempted to make broad comparisons with the expressed purpose of understanding phy- logenetic relationships. Although the com- parative anatomy of living forms was dis- cussed in Patterson’s (1965) publication, his focus was on fossil chimaeroids. Since then

1995

the advent of cladistics along with new in- formation and techniques have changed our approach to evolutionary morphological re- search and a reexamination of chimaeroid relationships is warranted. The goal of this comparative anatomical analysis, then, is to provide the first complete analysis of the morphology of all genera of living chima- eroids.

ACKNOWLEDGMENTS

Much of this work was done in New Zea- land and I extend my gratitude to the faculty, staff, and students at the Portobello Marine Laboratory for all their help and support, with special thanks to Gerald Stokes and Ken Mil- ler at the University of Otago, Dunedin. In New Zealand use of the scanning electron microscope (SEM) was kindly provided by the University of Otago Dental School. The fishermen of Port Chalmers were essential in helping me collect specimens for research, and without the assistance of Clinton Duffy from the Department of Conservation I would never have found or collected any embryos. In Wellington, Clive Roberts, Andrew Stew- art, and Chris Paulin at the National Muse- um, Larry Paul, Peter MacMillan, and Alan Blacklock at the Ministry of Agriculture and Fisheries (MAF), and Peter Castle at Victoria University provided helpful discussions, ma- terials, and specimens.

The American Museum of Natural History (AMNH), Field Museum of Natural History (FMNH), and United States National Mu- seum (USNM) provided specimens and work space to conduct research; Leslie Knapp at the Smithsonian Oceanographic Sorting Cen- ter helped by coordinating the shipment of specimens from New Zealand; and Nigel Merrett provided specimens from the British Museum (Natural History) (BMNH). A spe- cial thanks to Kathy Lundmark for embryo care at Friday Harbor Laboratories (FHL) and especially Willy Bemis, Tom Griffiths, Dave Klingener, Lance Grande, Rainer Zan- gerl, Anne Kemp, Sandy Whidden, Eric Fin- deis, and Judy Shardo for advice and inspi- ration. I also thank Colin Patterson, Barbara Stahl, John Maisey, and Carl Ferraris for their helpful comments and input on this manu- script.

DIDIER: CHIMAEROID FISHES 5

Funding for this research was provided by the Fulbright Foundation, a Donn E. Rosen grant from the AMNH, and a University Fel- lowship from the University of Massachu- setts; travel funding was provided by the USNM;; and financial assistance for course- work and research was provided by FHL.

ABBREVIATIONS Institutional

Institutional abbreviations follow Leviton et al. (1985), with the following additions:

MAF Ministry of Agriculture and Fish- Fisheries _eries, Wellington, New Zealand UMA University of Massachusetts, Am-

herst Anatomical ACC anterior clasper cartilage of pelvic clasper

AF _ anal fin

M. adductor mandibulae anterior

M. adductor mandibulae posterior AN _ angular lateral line canal

ANC antorbital crest

AP = ampullary pore

angular ampullary pore field

AR _ anterior radial element of pectoral fin basibranchial cartilage

basal cartilage of the first dorsal fin BHY basihyal cartilage

BT _basipterygium of the pelvic fin basipterygial process of the pelvic fin cb Mm. constrictores branchiales

cbr Mm. coracobranchiales

cd M. constrictor operculi dorsalis

cda WM. constrictor operculi dorsalis anterior ch M. coracohyoideus

CHY ceratohyal cartilage

cm WM. coracomandibularis

COR coracoid region of the pectoral girdle cp M. cucullaris profundus

CP chin process of Meckel’s cartilage

cs M. cucullaris superficialis

CT connective tissue

ceratobranchial cartilage

CV M. constrictor operculi ventralis

D2 _ second dorsal fin

DI diencephalon

DL __ descending lamina of tooth plates DR _ distal radial elements of the paired fins E eye

ethmoid ampullary field epibranchial cartilage

EC ethmoid canal

EHY epihyal cartilage

AMERICAN MUSEUM NOVITATES

enameloid layer

M. epaxialis

ethmoid process of the neurocranium frontal tenaculum

growth base of tooth plate

horizontal lateral line canal hypobranchial cartilage

hypermineralized rod (= compact pleromin of Orvig, 1985)

hyomandibular lateral line canal hypermineralized pad (= vascular pler- omin of Orvig, 1985)

cartilaginous hyoid rays supporting oper- cular flap

hypermineralized tissue (= pleromin of Or- vig, 1985)

M. interhyoideus

M. intermandibularis

inferior maxillary cartilage

infraorbital lateral line canal

infraorbital ampullary pore field

jaw joint

dorsal keel of chimaerid egg case tubercles on the snout of Harriotta lymphoid tissue

M. labialis anterior

M. levator anguli oris anterior

M. levator anguli oris posterior

lateral bulges of the yolk sac

lateral flange

M. levator hyoideus

ligamentum labialis

lateral line canal

lateral line canal pores

labial ampullary pore field

lateral rostral rod

lateral web of egg case

Meckel’s cartilage

mesencephalon

mandibular ampullary pore field metapterygium of pectoral fin

middle radial elements of the paired fins medial rostral rod

mesopterygium of pectoral fin mandibular tooth plates

margin of wear (the border between the at- trition surface and unworn portion of the oral surface of the tooth plate)

notochord

nasal lateral line canal

nasal capsule

notochord

nasal ampullary pore field

superficial ophthalmic nerve

cranial nerve VII

occipital lateral line canal

occipital crest

oral ampullary pore field

NO. 3119

otic process

opercular cartilage

ophthalmic foramen

oral lateral line canal

orbital septum

otic lateral line canal

otic capsule

pharyngobranchial cartilage

pulp cavity

posterior clasper cartilage of pelvic clasper M. protractor dorsalis pectoralis pharyngohyal cartilage

prelabial cartilage

premandibular cartilage

premaxillary cartilage

postotic lateral line canal

preorbital ampullary pore field postorbital ridge

prepelvic tenaculum

palatoquadrate cartilage

M. levator anguli oris anterior pars rostri propterygium of the pectoral fin palatine tooth plates

pituitary gland

pelvic girdle

proximal radial elements of the paired fins M. retractor dorsalis pectoralis

roof of the ethmoid canal

rostral foramen

ligamentum rostralis

rostrolateral ampullary pore field

M. retractor latero-ventralis pectoralis M. retractor mesio-ventralis pectoralis retroarticular process

rostral ampullary pore field supraorbital ampullary field

scapular process of pectoral girdle supracaudal lobe of the tail symphysial ridge of mandibular tooth plate superior maxillary cartilage

snout

supraorbital lateral line canal suborbital ridge

spindle of egg case

subrostral lateral line canal

subrostral groove

suprarostral ampullary pore field supratemporal lateral line canal supratemporal ampullary pore field subrostral ampullary pore field subcaudal lobe of the tail

M. superficialis

synarcual

trabecular dentine

telencephalon

tooth plate

transverse ridges

tail sheath of egg case

1995

TUD tubular dentine

VA _ vacuity in the anterior process of the pelvic girdle

VC _ vascular canal

VTP vomerine tooth plates

WS _ oral surface (working surface) of the tooth plates

MATERIALS AND METHODS

The taxa used in this study include all ex- tant genera of chimaeroids belonging to the families Callorhynchidae, Rhinochimaeri- dae, and Chimaeridae. Six genera were stud- ied: Callorhinchus, Rhinochimaera, Harriot- ta, Neoharriotta, Chimaera, and Hydrolagus. At least one species of each genus was chosen for detailed anatomical examination and whenever possible multiple species were studied to ensure that observed anatomical differences among genera were not interspe- cific. Outgroup comparisons were based on previously published studies, particularly Patterson (1965) and Maisey (1986).

Hydrolagus colliei (Pacific ratfish) were col- lected during May 1988 and June—September 1989 at Friday Harbor Laboratories, San Juan Island, Washington. Specimens were cap- tured by the RV “‘Nugget’”’ using a small otter trawl fishing on the bottom at depths ranging from 60 to 140 m (fig. 1A).

While in residence as a visiting scientist at the Portobello Marine Laboratory (October 1989—December 1990) I collected five species of chimaeroids from New Zealand. Callor- hinchus milii (elephant fish) were collected by trawling at 57 m outside Tairoa Head, off the coast of Dunedin. Juvenile Callorhinchus milii were collected from coastal waters at 16—26 m just north and south of Dunedin (fig. 1B). Hydrolagus novaezealandiae (ghost shark) were caught by scallop fishermen at 240-288 m, 100 km off Cape Saunders, Dun- edin, New Zealand (fig. 1B). Harriotta ral- eighana (no common name), Hydrolagus sp. B (pale ghost shark), and Hydrolagus novae- zealandiae were trawled during a deep-sea fishing trip on the MAF Fisheries vessel, RV “James Cook,” May 21-27, 1990. Fishing was done over a range of 300-800 m in depth from Mernoo Bank and south to Dunedin, off the coasts of Canterbury and northern Otago (fig. 1B). Two specimens of Rhinochi-

DIDIER: CHIMAEROID FISHES 7

Fig. 1. Localities where chimaeroid fishes were collected for this study. A, Map of Washington state; inset, demarcated by dashed lines, shows the San Juan Islands. The location of Friday Harbor Laboratories is indicated by a dot. Primary col- lection sites for Hydrolagus colliei in Puget Sound are as follows: 1) San Juan Channel, 2) Lopez Channel, 3) Upright Head, and 4) West Sound. B, Map of the south island of New Zealand. Collec- tion sites are numbered as follows: 1) Dunedin, Tairoa Head, Blueskin Bay, and Cape Saunders; 2) Green Island and Nugget Point; 3) Mernoo Bank; 4) Challenger Plateau; and 5S) Marlborough Sounds.

maera pacifica were trawled from 1034 m on the Challenger Plateau (fig. 1B).

Eight egg cases were collected from captive females at Friday Harbor Laboratories in May 1988. Each egg case was tagged and held in

AMERICAN MUSEUM NOVITATES

NO. 3119

TABLE 2 List of Specimens Dissected for Study of Muscles :

Callorhynchidae Callorhinchus Callorhinchus milii

Rhinochimaeridae Rhinochimaera Rhinochimaera pacifica Harriotta Harriotta raleighana Neoharriotta Neoharriotta pinnata

Chimaeridae

Hydrolagus Hydrolagus colliei Hydrolagus novaezealandiae

Hydrolagus sp. B

Chimaera Chimaera sp. C Chimaera monstrosa

CM34, CM35, CM36

AMNH 96939, AMNH 96940 AMNH 96931, AMNH 96935

USNM 204778

DAD36, HC18, HC5 HNZ1, HNZ2 AMNH 96941, HspB1, HspB6

NMNZ P19694 USNM 222708

4 Each number in the table refers to a single individual; data on these specimens available upon

request and also on file at the AMNH.

a free-flowing seawater table. Two egg cases were opened each month and the embryos were removed and fixed in 4% glutaraldehyde in seawater. Of the eight egg cases, four em- bryos were collected. In June 1989, four ad- ditional egg cases were collected using similar procedures. Two of these egg cases contained embryos. An additional egg case containing an embryo was collected during July 1989 by the RV “Nugget” during a bottom trawl. Egg cases of Callorhinchus milii were col- lected by SCUBA diving in the Marlborough Sounds, New Zealand, on June 5—6, 1990. A total of 44 egg cases was collected from 5 to 20 m at four sites (fig. 1B). Of the 44 egg cases collected, 38 were fertile and the remaining 6 were empty. Fourteen embryos, with their yolks intact, were fixed in the field in 10% formalin in seawater. The remaining 24 vi- able egg cases were transported in buckets of seawater to the Ministry of Agriculture and Fisheries laboratory in Wellington where they were held in two seawater tanks. All egg cases were opened within one week of collection and 20 embryos were fixed in 5—10% for-

malin and seawater. Four embryos were cho- sen randomly and fixed in 4% glutaraldehyde in seawater. Two additional embryos of Cal- lorhinchus milii were collected from egg cases trawled off the coast of Dunedin and both were fixed in 10% formalin in elasmobranch Ringer’s solution.

All material collected has been donated to the USNM, AMNH, FMNH, or UMA. Data for each specimen, including collection notes, locality, detailed measurements, and gut con- tents, are kept on file at the AMNH.

Gross ANATOMICAL METHODS

At least one representative species of each genus was dissected for gross anatomical study of the musculature and skeleton (table 2). Species of the genus Neoharriotta are es- pecially difficult to obtain for dissection, and I was unable to complete a detailed dissection of the musculature of any Neoharriotta; how- ever, I was permitted to superficially dissect the musculature of Neoharriotta pinnata.

Adult specimens to be dissected were anes-

1995

thetized in MS-222, fixed in 10% buffered formalin in seawater for a minimum of 2 weeks, and then transferred into 70% ethanol for long-term storage. Point-to-point mea- surements were taken on the left side of each specimen and gut contents were analyzed be- fore any anatomical procedures were done. All dissections were done on the left side ex- cept for borrowed museum specimens, which were dissected on the right side. Separate in- dividuals were dissected for study of lateral and ventral musculature to ensure that the lateral dissection did not affect ventral mus- culature. The heads of three Hydrolagus col- liei and one Callorhinchus milii were also studied in sagittal section by cutting pre- served specimens with a hacksaw.

Both wet and dry skeletons were prepared from fresh or fresh-frozen material. All skel- etal material was prepared by immersing the specimen in hot water for 1-5 minutes, then handpicking the flesh from the cartilage. Wet skeletons were immersed and kept in 70% ethanol. Dry skeletal preparations were either left to air dry for several days or dried in a lyophilizer.

The procedure used for clearing and dou- ble-staining follows Hanken’s method (slight modification from Hanken and Wassersug, 1981), with some modifications adopted from Potthoff (1984). Table 3 lists specimens that were cleared and stained for comparative study of the developing skeleton and calcified tissues. Small juveniles were prepared whole, without skinning, so that the dermal denticles would not be lost. In order to study the cal- cified rings of the lateral line canals, pieces of skin were peeled from the heads of Cal- lorhinchus milii (CM35, 890 mm TL), Rhin- ochimaera pacifica (AMNH 96940, 950 mm TL), Harriotta raleighana (AMNH 96935, 800 mm TL), Hydrolagus colliei (AMNH 96933, 470 mm TL), Hydrolagus novaezeal- andiae (HNZ12, 660 mm TL) and Chimaera sp. C(NMNZ P12921, 990 mm TL). Sections of the vertebral column, 3—5 cm long, were dissected out of the trunk or caudal region to study the vertebral calcifications in the fol- lowing taxa: Rhinochimaera pacifica (AMNH 96939, 1100 mm TL), Harriotta raleighana (AMNH 96935), Hydrolagus colliei (HC8, 500 mm TL), and Chimaera sp. C (NMNZ P12921).

DIDIER: CHIMAEROID FISHES 9

TABLE 3 Specimens Prepared by Clearing and Double Staining?

Callorhinchus mili CM 17 juvenile male, 175 mm TL CM 18 juvenile male, 215 mm TL CM 19 juvenile female, 205 mm TL CM20 juvenile female, 208 mm TL CM73 juvenile female, 205 mm TL CM45 embryo, 72 mm TL CM48 embryo, 75 mm TL CM49 embryo, 80 mm TL CM54 embryo, 87 mm TL CMS53 embryo, 89 mm TL (sagittal preparation)

Rhinochimaera pacifica ex NMNZ P25812 juvenile female, 227 mm TL

Neoharriotta pinnata USNM 217445 juvenile female, 167 mm TL

Harriotta raleighana NMNZ P21442 juvenile male, 253 mm TL

Hydrolagus colliei DAD26 (UMA) juvenile male, 113 mm TL

Hydrolagus novaezealandiae HNZI11 juvenile male, 110 mm TL

Hydrolagus mirabilis BM(NH) 1990.9.11:6 juvenile male, 232 mm TL

Chimaera monstrosa BM(NH) 1990.9.11:1 juvenile male, 172 mm TL

aData on these specimens are available upon request and are also on file at the AMNH.

HISTOLOGICAL METHODS

To study the histology of chimaeroid tooth plates the right mandibular tooth plates of adult Rhinochimaera pacifica (AMNH 96940, 950 mm TL), Harriotta raleighana (AMNH 96931, 850 mm TL), Chimaera sp. C (NMNZ 19694, 944 mm TL), and a juvenile Hydro- lagus colliei (AMNH 96933, 470 mm TL) were removed from formalin-fixed speci- mens. The tooth plates averaged a maximum labiolingual width of 2 cm, measured along the posterior edge, and a mesiodistal length of 2—3 cm along the symphysial edge. Tooth plates were decalcified in formic acid A (10 ml formalin and 5 ml formic acid in 85 ml water; Humason, 1972) for up to 1 month. Dehydration, infiltration, and embedding in low viscosity nitrocellulose (LVN) followed the methods of Thomas (1980). The tooth plates were sectioned transversely at 30 um thickness on a Reichert sliding microtome (vintage 1920) and stained with Ehrlich’s he-

10 AMERICAN MUSEUM NOVITATES NO. 3119 TABLE 4 Histological Preparations of Embryos of Callorhinchus milii Embryo mm TL Section Stain CM68 234 transverse hematoxylin and eosin CM70 px) transverse hematoxylin and eosin CM72 24 transverse hematoxylin and eosin CM69 26 transverse hematoxylin and eosin CM47 57 transverse hematoxylin and eosin CM46 61 transverse hematoxylin and eosin CM43 62 frontal hematoxylin and eosin CMS50 67 transverse hematoxylin and eosin CM44 70 frontal hematoxylin and eosin CM42 74 transverse Mallory’s trichrome CM40 79 transverse Mallory’s trichrome CMS51 85 Sagittal Mallory’s trichrome CM52 89 transverse Mallory’s trichrome

@ Estimated measurement for this specimen due to loss of the tail.

matoxylin and picro-ponceau (Humason, 1972). The tooth plates of one juvenile Cal- lorhinchus milii (228 mm TL) were removed from a fixed specimen and prepared using standard paraffin techniques (prepared by Gerald Stokes, University of Otago, Dune- din, New Zealand).

To study the developmental morphology of the head, 13 embryos and 1 juvenile Cai- lorhinchus milii (166 mm TL) were transect- ed behind the pectoral girdle with a razor blade. The anterior half of each embryo was prepared using paraffin techniques and serial sectioned at 8—10 um thickness. Sections were stained with Lillie-Mayer haemalum and eo- sin or Mallory’s trichrome collagen stain, Hughesdon’s modification (Carleton and Leach, 1947; prepared by Gerald Stokes). Ta- ble 4 lists the data for each embryo of Cal- lorhinchus milii that was prepared.

In addition, the entire head of one Hydro- lagus colliei embryo (39 mm TL) was em- bedded in LVN and transversely sectioned at 30 um thickness. Sections were stained with Ehrlich’s hematoxylin and picro-ponceau (Humason, 1972).

Scanning electron microscopy was used to examine hard tissues and whole embryos of Hydrolagus colliei and Callorhinchus milii. Material to be studied with SEM was fixed in either 4% glutaraldehyde or 10% formalin in Gomori’s phosphate buffer or elasmo-

branch Ringer’s solution. Tooth plates were removed from the jaws of fixed adult speci- mens and smashed with a hammer into 5-mm/? pieces. The tooth plates of one ju- venile Callorhinchus milii (234 mm TL) were removed intact from the jaws and studied. The fin spine was also removed from this specimen and cut into three pieces, each ap- proximately 4 mm long, for study of the in- ternal anatomy in transverse and longitudi- nal view. To study the calcifications of the lateral line canals, fresh pieces of skin peeled from the heads of a juvenile Callorhinchus milli (190 mm TL), a juvenile Harriotta ral- eighana (AMNH 96944, 887 mm TL), and an adult Hydrolagus sp. B (AMNH 96945, 824 mm TL) were air-dried and sputter-coat- ed before study under the SEM. The heads of two embryos were also studied using SEM (Hydrolagus colliei, 44 mm TL; Callorhin- chus milii, 40 mm TL).

Two procedures for SEM preparation were followed. At Friday Harbor Laboratories specimens were postfixed in 1% OsO,, de- hydrated and dried in a Samdri-790 critical point dryer, sputter-coated with gold palla- dium (300-400 A thick), and examined un- der a JEOL JSM 35 scanning electron mi- croscope.

At the University of Otago School of Den- tistry in Dunedin, New Zealand, a Cambridge Stereoscan 360 scanning electron microscope

1995

was used. Specimens were dried in a Polaron E3000 critical point dryer and coated using a Polaron E5100 sputter coater.

REVIEW OF HOLOCEPHALAN SYSTEMATICS

As shown in figure 2, relationships among the extant taxa are an unresolved trichotomy. This cladogram (from Maisey, 1986) is my working hypothesis of holocephalan relation- ships, and this study focuses on the unre- solved trichotomy of the extant families of chimaeroid fishes. In this work the chimae- roid fishes are ‘“‘crown-group”’ holocephalans. These include all Recent holocephalans and their immediate fossil relatives, whose rela- tionships I have not attempted to resolve in the diagram. The chimaeriform fishes include all chimaeroids plus certain stem-group hol- ocephalans.

Several recent studies have examined hol- ocephalan relationships with an emphasis on modern phylogenetic concepts (Zangerl and Case, 1973; Lund, 1977, 1986a, b; Zangerl, 1981; Maisey 1984, 1986). Some of these studies are problematic; nevertheless, the perspectives warrant discussion.

The recent discovery of a new order of Pennsylvanian Chondrichthyes, the Iniop- terygia, which was interpreted to have an au- tostylic jaw suspension like that of Holoce- phali with the upper jaw fused to the neurocranium (Zangerl and Case, 1973), led to new interpretations of the origin of Hol- ocephali. Based on comparative anatomical studies of iniopterygians and other chondri- chthyan fishes, Zangerl and Case (1973: 64) determined that the iniopterygians represent structural intermediates between elasmo- branchs and chimaeroids, and it was con- cluded that Holocephali and Iniopterygii are sister groups that evolved from a common chondrichthyan ancestor. The subclass Sub- terbranchialia (Zangerl, 1979) was erected to include all Chondrichthyes with an autostylic jaw suspension and gill arches concentrated underneath the neurocranium. It was later discovered that autostyly is not found in all Iniopterygii (Stahl, 1980). Although Subter- branchialia is not a monophyletic group be- cause it is united by a symplesiomorphy (Zangerl, 1981: 39), it remains in the litera-

DIDIER: CHIMAEROID FISHES 11

& ~ te & ~ 2 ¥ Cd S = 7 ‘ YY, 4 (e) x 0 S eo ¢ g ra) ry » + of ) e < o é oe? e = < § BY) Ve we + + ¥ oe & ¢

J 2-448, 5,6,23,24

Fig. 2. Phylogenetic interpretation of Maisey (1986). In this first cladistic analysis of holoce- phalan relationships only the best known fossil forms are indicated and the living forms, repre- sented by Callorhynchus (= Callorhinchus), Chi- maera, and Rhinochimaera, are shown as the ter- minal branch of this cladogram united by char- acters J7, J20, and J22. Characters J1—J24 are re- numbered in this study and listed in table 3 (re- produced from Maisey, 1986: 220).

ture today as a “‘wastebasket”’ group of chon- drichthyan fishes that excludes elasmo- branchs.

A second perspective (Lund, 1977) resur- rected the subclass Bradyodonti to include Holocephali and was defined as the sister group to Elasmobranchii. Lund (1977) united the Bradyodonti on the basis of slow tooth replacement and an operculate gill chamber. With the discovery of new cochliodont fossils from the Mississippian, Lund (1986a, b) re- vised his previous (1977) interpretation of holocephalan relationships and dismissed Bradyodonti as an invalid taxonomic group. Instead, he suggested that holocephalans are probably closely related to the cochliodonts (Lund, 1986a, b).

In the most recent analysis of holocephalan relationships, Maisey (1986) completed a cla- distic analysis of the following fossil and Re- cent taxa: Helodus, Menaspoidei, My- acanthoidei, Squaloraja, Ischyodus, Callorhinchus, Chimaera, and Rhinochi- maera. The resulting cladogram (fig. 2) was supported by 24 morphological characters of both the skeleton and soft tissues (Maisey, 1986: 220, table 5). Based on this cladistic analysis, Ischyodus was determined to be the immediate sister taxon to the living chima-

12 AMERICAN MUSEUM NOVITATES

NO. 3119

TABLE 5 Holocephalan Characters Summarized by Maisey (1986)

24. (J2) Holostylic jaw suspension 25. (J3) Branchial arches located below braincase 26. (J4a) Synarcual articulates with dorsal basal

27. (J5) Dibasal pectoral (propterygium and metapterygium)

28. (J6) Erectile dorsal spine and associated cartilage

29. (J23) Few intermediate segments between pelvic basipterygium and clasper

30. (J24) Basipterygium spans entire pelvic fin 31. (J9a) Tooth plates

32. (J4b) Synarcual shorter than neurocranium 33. (J8) Rostral cartilages

34. (J10) Tritoral areas on tooth plates

35. (J11) Polyspondylous notochordal rings

36. (J12) Sensory canals in open grooves lined by crescent-shaped scales 37. (J13a) Tenacula (prepelvic clasper) present only in males

38. (J14a) Frontal clasper

39. (J9b) Two upper pairs of tooth plates and one lower pair of tooth plates (more in some fossils)

40. (J14b) Frontal clasper only in males 41. (J15) Enlarged labial cartilages

42. (J16a) Reduced squamation (nongrowing denticles)

43. (J17) Characteristic sensory canal arrangement on head

44, (J21a) Process on pelvic girdle for tenaculum

45. (J21b) Process on pelvic girdle for tenaculum is jointed and movable 46. (J16b) Denticles only at dorsal midline and along sensory canals 47. (J1) Ethmoid region encloses superficial ophthalmic nerves in an ethmoid canal; no ethmoid keel

48. (J19) Interorbital septum

49. (J7) Lateral walls of fin-spine lack trabecular osteodentine (= trabecular dentine)

50. (J20) Ventral fusion of scapulocoracoids

51. (J22) Closure of adult hypophyseal foramen; isolation of ventral lobe of pituitary external to

cranium in roof of oral cavity*

4 It is noted by Maisey (1986) that this feature may have evolved independently in Recent

elasmobranchs and chimaeroids.

eroids, and Helodus was interpreted as rep- resenting the primitive holocephalan condi- tion (Maisey, 1986: fig. 2).

Figure 3 is my summary diagram of the cladistic analysis of Maisey (1986). Although Maisey (1986) provided the most rigorous cladistic analysis of holocephalan relation- ships to date, he did not attempt to resolve the relationships of the extant chimaeroids. Nevertheless, I use Maisey’s (1986) analysis for two reasons: first, this work is the most recent rigorous cladistic analysis of holoce- phalan relationships; and second, it is the only study with the specific goal of understanding the phylogeny of Recent taxa and therefore uses soft tissue characters and embryological evidence. This working hypothesis of holo-

cephalan relationships is my framework for phylogenetic study of the living forms.

I have renumbered the chondrichthyan and holocephalan characters of Schaeffer (1981) and Maisey (1984, 1986: tables 1, 5) accord- ing to my synapomorphy scheme (fig. 3; char- acters 1-23). It is accepted that Holocephali and Elasmobranchii are Chondrichthyes and that they are sister groups. Holocephali is a monophyletic group based on Maisey’s (1986) characters: J2—4a, 5, 6, 23, 24 (24-30 in fig. 3). Characters at these levels will not be treat- ed further in this study.

Because my goal is to understand the re- lationships of living forms, I have not at- tempted to resolve relationships among fossil taxa, whose relationships remain problem-

1995

DIDIER: CHIMAEROID FISHES 13

Elasmobranchii Holocephali

Chimaeriformes

Chimaeroidei Callorhynchidae Rhinochimaeridae Chimaeridae

t Menaspoidei

t Myriacanthoidei Neoharriotta

t Squaloraja Harriotta Chimaera Squalus tHelodus t Ischyodus Callorhinchus Rhinochimaera Hydrolagus

49-5| 31-48 24-30

| -23

Fig. 3. Summary of phylogenetic interpretations. This summary diagram, based primarily on the cladistic interpretation of Maisey (1986), shows the three families of living chimaeroids as an unresolved trichotomy. Although some fossil taxa are here represented outside Chimaeroidei (e.g., Ischyodus), there is evidence to suggest that this is an oversimplification; however, the purpose of this analysis is to resolve relationships among the living forms and a thorough discussion of relationships among fossil taxa is beyond the scope of this study. Chondrichthyan characters (numbers 1-23) are taken from Schaeffer (1981) and Maisey (1986). The remaining characters (numbers 24—48) are from Maisey (1986). See text and table 1 for discussion of this interpretation and the characters.

atic. The fossil holocephalan taxa used by Maisey (1986) are grouped as a single lineage at the level of Chimaeriformes (fig. 3), all of which may or may not share characters 31- 48. Because I do not deal with these char- acters individually in this analysis, I use them to artificially unite a diverse assemblage of fossil chimaeriforms; however, it should be understood that these taxa are not accepted as a single lineage (as indicated in fig. 3).

The relationships among Recent forms are still not well understood nor have the char- acters for defining monophyletic groups (e.g., at the familial level) among the extant forms been analyzed. This study focuses exclusively on relationships among Chimaeroidei.

TAXONOMIC REVIEW OF CHIMAEROID SPECIES

Species-level relationships of extant chi- maeroids have yet to be examined and many species are still undescribed (Last, personal commun.; Stehmann, personal commun.; personal obs.). Table 6 lists all all 34 known extant species of chimaeroids (Didier, 1993).

Three families of Recent chimaeroids are recognized (Nelson, 1984): Callorhynchidae, Rhinochimaeridae, and Chimaeridae. Gill (1898) published a classification in which the subfamily Harriottinae was erected, and Dean (1904b: 20) considered that Harriotta might more properly belong to its own family, Har-

14 AMERICAN MUSEUM NOVITATES

TABLE 6 List of Recent Taxa and Synonymy

Family Callorhynchidae Garman, 1901 Callorhinchus Lacépéde, 1798 (ex Gronovius, 1763) Callorhinchus callorhynchus Linnaeus, 1758 =_Callorhynchus antarcticus Lay and Bennett, 1839 = _Callorhynchus peronii Duméril, 1865 = Callorhynchus argenteus Philippi, 1892 = Callorhynchus elephantinus Bory de St. Vincent, 1923 = Callorhinchus smythii Lay and Bennett, 1839 = Callorhinchus tritoris Garman, 1904 Callorhinchus milii Bory de St. Vincent, 1823 = Callorhynchus australis Owen, 1854 = _Callorhynchus tasmanius Richardson, 1840 = Callorhynchus dasycaudatus Colenso, 1878 Callorhinchus capensis Dumeéril, 1865 Family Rhinochimaeridae Garman, 1901 Rhinochimaera Garman, 1901 Rhinochimaera pacifica (Mitsukurii, 1895) Rhinochimaera atlantica Holt and Byrne, 1909 Rhinochimaera africana Compagno, Stehmann, and Ebert, 1990 Harriotta Goode and Bean, 1895 = Anteliochimaera Tanaka, 1909 Harriotta raleighana Goode and Bean, 1895 = _Anteliochimaera chaetirhamphus Tanaka, 1909 = Harriotta curtis-jamesi Townsend and Nichols, 1925 = Harriotta opisthoptera Deng, Xiong, and Zhan, 1983 Harriotta haeckeli Karrer, 1972 Neoharriotta Bigelow and Schroeder, 1950 Neoharriotta pinnata (Schnakenbeck, 1929) Neoharriotta carri Bullis and Carpenter, 1966 Family Chimaeridae Bonaparte, 1831 Chimaera Linnaeus, 1758 unavailable snynonyms: Callorhynchus americanus Gronovius, 1772 2 Chimaera praecisa Walbaum, 1792 a Callorhynchus centrina Gronow in Gray, 1854 ¢ Chimaera vaillanti Dean, 1906 Chimaera monstrosa Linnaeus, 1758 = Chimaera argentea Ascanius, 1772 = Chimaera borealis Shaw, 1804 = Chimaera mediterranea Risso, 1826 = Chimaera cristata Faber, 1829 = Chimaera arctica Gistel, 1848 = Chimaera dubia Osorio, 1909 Chimaera cubana Howell-Rivero, 1936 Chimaera owstoni Tanaka, 1905 Chimaera jordani Tanaka, 1905 Chimaera phantasma Jordan and Snyder, 1900 Chimaera pseudomonstrosa Fang and Wang, 1932 Hydrolagus Gill, 1862 = _Bathyalopex Collett, 1904 = _psychichthys Fowler, 1902

NO. 3119

eee ee ee ee eee ee eee

1995

DIDIER: CHIMAEROID FISHES

TABLE 6—(Continued )

= _phasmichthyes Jordan and Hubbs, 1925 Hydrolagus mitsukurii (Dean, 1904a)

= Chimaera mitsukuni Dean, 1904a Hydrolagus affinis (Capello, 1868)

=_Chimaera plumbea Gill, 1878

= Chimaera abbreviata Gill, 1883 Hydrolagus africanus (Gilchrist, 1922) Hydrolagus alberti Bigelow and Schroeder, 1951 Hydrolagus barbouri (Garman, 1908)

=_Chimaera spilota Tanaka, 1908 Hydrolagus colliei (Lay and Bennett, 1839)

= Chimaera colliei Lay and Bennett, 1839

= Chimaera neglecta Ogilby, 1888 Hydrolagus deani (Smith and Radcliffe, 1912) Hydrolagus eidolon (Jordan and Hubbs, 1925)

= _Chimaera purpurescens Jordan and Snyder, 1904 Hydrolagus lemures Whitley, 1939 Hydrolagus macrophthalmus Buen, 1959 Hydrolagus media (Garman, 1911) Hydrolagus mirabilis (Collett, 1904)

15

Hydrolagus novaezealandiae (Fowler, 1910) = Chimaera australis Hector, 1902

Hydrolagus ogilbyi (Waite, 1898)

=_Hydrolagus tsengi (Fang and Wang, 1932)

Hydrolagus purpurescens (Gilbert, 1905) =_Chimaera gilberti (Garman, 1911)

Hydrolagus waitei Fowler, 1908

Hydrolagus pallidus (Hardy and Stehmann, 1990)

4 It is possible that these are synonyms of Chimaera monstrosa or perhaps Hydrolagus colliei because all lack an elongate rostrum and the respective localities are listed as "American Ocean" Gronovius (1772), Gronow in Gray (1854), and "American Pacific Ocean" Walbaum (1792); however, it is impossible to know exactly which species to refer these synonyms to and I have listed them as unavailable names under the genus. Dean (1906) suggested that the vague description of Callorhynchus centrina was a reference to Harriotta based on the locality, but the description does not

fit that of a rhinochimaerid.

b This species exists aS a name only, no description of a specimen or figure was ever published.

riottidae. Unfortunately Dean (1904b) did not provide a formal analysis of this familial dis- tinction. Based on my analysis of the mor- phological features of rhinochimaerids, it seems that Harriotta and Neoharriotta might more properly be classified separately from Rhinochimaera.

Although the classification in table 6 is cur- rent, there are three notable taxonomic prob- lems. First, note that the family name, Cal- lorhynchidae, differs in spelling from the generic name, Callorhinchus. Although the generic name Callorhynchus is often found in the literature it was published in a work

that has been rejected by the International Commission on Zoological Nomenclature (Opinion 261, 1954; in Eschmeyer, 1990), and the emended generic name Callorhinchus is accepted in this work. Although correction in the generic name requires an emended family name (International Code of Zoolog- ical Nomenclature, 1985: art. 35d), the fam- ily name Callorhynchidae is maintained in this work for the sake of stability as being of “general acceptance” (ibid.: art. 40b, ii) until the International Commission on Zoological Nomenclature can address this problem. The second taxonomic problem concerns

16 AMERICAN MUSEUM NOVITATES

the suggestion that species of Callorhinchus may actually be variants of a single species (Norman, 1937; Bigelow and Schroeder, 1953: 562; Krefft, 1990: 117; Hardy, 1990: 88). The morphology of tritor pads on the tooth plates and the length of the pectoral fins are used to distinguish species of Callor- hinchus (see Fowler [1941] for a key to spe- cies). Norman (1937) suggested that the pec- toral fin and tooth plate characteristics used to distinguish species of Callorhinchus are highly variable and their usefulness as spe- cies-defining characteristics is questionable. This conclusion received support from Big- elow and Schroeder (1953: 562); neverthe- less, until a revision of Callorhinchus is com- pleted, all three species listed in table 6 will be accepted.

The third noteworthy point concerns the validity of Chimaera and Hydrolagus as dis- tinct genera (Hardy and Stehmann, 1990). The only feature that distinguishes these two genera is the presence of an anal notch in the subcaudal fin of Chimaera that is absent in Hydrolagus. All known undescribed species of chimaeroids belong to the family Chi- maeridae, but generic determination is dif- ficult because the delicate fin webs are easily torn, making it impossible to determine if an anal notch is present. These undescribed spe- cies will likely remain so until the significance of the anal notch as a generic character is better understood and more reliable specific characters are defined.

COMPARATIVE ANATOMICAL STUDIES

In subsequent descriptions, general com- ments on anatomical features precede family- by-family treatments, with Callorhynchidae always first, Rhinochimaeridae second, and Chimaeridae third. This order reflects the systematic conclusions of this work. I first provide a brief discussion of some aspects of development including a description of the morphology of the egg capsules. All other an- atomical descriptions focus on adult mor- phology with an analysis of external features, comparative anatomy of the lateral line ca- nals, skeletal anatomy including descriptions of the neurocranium, gill arches, vertebral column, paired fins, and girdles, and gross

NO. 3119

morphology and histology of the tooth plates. Lastly, the musculature is described for all six genera.

A list of all specimens studied in detail is provided in table 7. Adults are defined as sexually mature or maturing individuals. Relatively smaller individuals that do not ex- hibit any features of sexual maturity (such as eggs maturing in the ovaries of females or secondary sexual characteristics in males) are defined as juveniles. Embryos include all stages inside the egg case up to hatching. The number of skeletons is the total of wet and dry skeletal preparations of both adults and juveniles.

COMMENTS ON DEVELOPMENT

All chimaeroid fishes are oviparous. Be- cause these fishes generally live at great depths, the only obstacle to collecting devel- opmental material has been locating the eggs and spawning sites. Females have large yolky eggs that are fertilized inside the oviduct and enclosed within a keratinous egg case pro- duced by the shell gland located at the an- terior end of each oviduct. Two egg cases are deposited simultaneously, one from each oviduct. A single embryo will develop within the enlarged portion of the egg case. Inside the egg case the embryos always lie on their left side with the head facing anteriorly and the tail extending posteriorly into the tail sheath. Like elasmobranchs, these embryos beat their tails to facilitate circulation of wa- ter inside the egg case. Dean (1903) thought that constant movement of the tail is used to circulate water through the egg case for ex- change of gases. This behavior has been ob- served for elasmobranchs, such as skates, that develop inside similar egg cases (Pelster and Bemis, 1992).

Although only two egg cases are deposited at a time, the ovaries of gravid females have developing eggs of all stages and a single fe- male may deposit several pairs of eggs during the breeding season (Dean, 1906; Gorman, 1963; Stanley, 1963; Sathyanesan, 1966). It has been suggested that females may store sperm (as do other chondrichthyans) and this would facilitate multiple spawning events in chimaeroids (Stanley, 1963).

Due to the method of bulk collecting eggs

1995 DIDIER: CHIMAEROID FISHES 17 TABLE 7 Summary of Specimens Examined in Comparative Morphological Analysis® adult juvenile embryo TOTAL _ skeletons Family Callorhinchidae Callorhinchus milii 12 29 37 78 23 Family Rhinochimaeridae Rhinochimaera pacifica 2 3 - 5 1 Harriotta raleighana 3 5 - 8 5 Neoharriotta pinnata 1 1 - 2 - Family Chimaeridae Hydrolagus colliei 33 53 7 95 32 Hydrolagus novaezealandiae r 5 12 7 Hydrolagus sp. B 5 6 - 11 9 Hydrolagus mirabilis - 1 - 1 - Chimaera monstrosa 1 2 - 3 - Chimaera sp. C 2 - - 2 -

4 For each specimen, collection data, locality, detailed measurements, and gut contents were recorded and are available from the author; data and collection numbers are also on file at the AMNH.

of Callorhinchus in the field, there was no reliable way to determine the age of embryos. Because there is no complete staging scheme for any chimaeroid embryos, the only way to determine relative stage of development was by comparing total lengths (TL). This is not a particularly good indicator of the actual state of development because temperature may have affected growth in subtle ways and al- lometric growth of the body relative to the length of the tail will lead one to mistakenly interpret total length measurements. Until a scheme for staging chimaeroid embryos is de- veloped the total length measurement will serve as an indication of relative embryonic stage. A representative series of embryos of Callorhinchus milii is shown in figure 4.

In earliest embryonic stages the yolk sac (YS; fig. 4) is ovoid. As the embryo develops, the yolk sac changes shape, hardens, and forms characteristic bulges (LB). As the em- bryo grows, the head lies within the pocket formed by these bulges. This occurs in Hy- drolagus as well as in Callorhinchus, and the shape of the yolk sac is the same in both species. The embryo is attached to its yolk sac by a yolk stalk just anterior to the pectoral girdle. Most of the yolk mass lies anterior to the embryo, and as it is resorbed the yolk sac becomes more ventral in position.

Early embryos have an anterior bulbous process, the rostral bulb (RB; fig. 4). This elongate rostral bulb is a hollow vesicle lined by a layer of columnar epithelium. It extends from the anterior end of the developing head and curves dorsally to lie directly in front of the head. The rostral bulb is at its maximum size in younger embryos 20-40 mm TL, and gradually becomes reduced in size until it is no longer evident. The fate of this structure is unknown; however, Allis (1917, 1926) sug- gested that the rostral bulb plays a role in shaping the neurocranium in its early stages of development.

External gill filaments (GF; fig. 4) first be- gin to develop when embryos are about 50 mm TL. Tiny, developing gill filaments can be seen as small loops from the external edges of the gill arches in even the very smallest embryos. As the embryo grows the external gill filaments lengthen and appear bright red as they are infused with blood. Each gill fil- ament consists of a single, looped blood ves- sel, an outgrowth of the aortic arch, which extends from the back of the developing gill arch. These blood vessels are looped by fold- ing back on themselves and a thin membrane of tissue holds the loop to itself. The external gill filaments reach their maximum length of 20-30 mm in embryos of 75-80 mm TL. As

18 AMERICAN MUSEUM NOVITATES NO. 3119

Fig. 4. Series of embryos of Callorhinchus milii. These six embryos of Callorhinchus milii range in size from 53 mm TL (at top) to 122 mm TL (bottom). In all chimaeroid embryos studied the yolk sac (YS) develops a unique bulged shape (LB). Embryos develop long external gill filaments (GF) that are gradually resorbed as the functional gills and operculum develop. The tail, which in this species is heterocercal, does not develop its characteristic morphology until the embryo is near hatching. Based on the relatively straight tail in this embryo, it appears that some posthatching development and mod- ification of the tail occurs before it is fully heterocercal. The fin spine (FS) remains relatively soft until after hatching, when it soon becomes mineralized. Scale bar = 2 cm.

1995

the gill arches and opercular flap become fully developed the external gill filaments are grad- ually resorbed by the embryo until they are no longer visible externally.

The eyes (E) are extremely large and well developed in early embryonic stages and the pigment of the retina is evident in embryos that are 50 mm TL. The large eyes play a key role in shaping the neurocranium, as is evi- dent when examining the neurocranial struc- ture of adults, especially species with ex- tremely large orbits. It has been hypothesized (Dean, 1906; Allis, 1917, 1926; de Beer and Moy-Thomas, 1935; Holmgren, 1942) that the formation of a connective tissue inter- orbital septum as well as the unique brain formation in which the brain lies behind the eye and the olfactory tracts run ventral to the eye are the result of development of such large eyes.

Pigmentation is first visible as a series of dark patches on the dorsal surface of Callor- hinchus embryos that are 75 mm TL. The dorsal surface of the head becomes darkly pigmented shortly thereafter. Older embryos (100 mm TL) are fully pigmented, showing the adult color patterns with black saddle- like bands on the dorsal surface and black patches along the trunk. The paired and me- dian fins become darkly pigmented at their distal ends.

The dorsal fin spine (FS; fig. 4) begins to form in association with the first dorsal fin in embryos of about 50 mm TL. In these early stages the fin spine is soft and is bent to fit inside the egg case.

Embryos and newly hatched juveniles pos- sess denticles on top of the head and along the dorsal surface of the body. The early de- velopment of these denticles was first de- scribed by Schauinsland (1903). The head denticles lie medial to the supraorbital lateral line canal and form a U-shaped pattern atop the head. Occasionally a few denticles may develop lateral to the supraorbital lateral line canal. Along the dorsal surface there are two rows of denticles in the interspaces between the first and second dorsal fins and between the second dorsal and caudal fins. These den- ticles are absent in adults.

The palatoquadrate becomes indistin- guishably fused to the trabecular cartilages at a very early stage in development. In the Cal-

DIDIER: CHIMAEROID FISHES 19

lorhinchus embryos that I studied, the pala- toquadrate had already fused to the neuro- cranium in embryos of about 75 mm TL. Procartilaginous rudiments are evident in embryos of 60-70 mm TL. At this stage the trabecular, polar, and parachordal cartilages are beginning to form the base of the neu- rocranium. The roof of the neurocranium has not yet begun to form and the ethmoid and otic regions are in early stages of formation. Extending from the lateral edges of the tra- becular cartilages in the anterior region of the developing neurocranium are two small tri- angular cartilages. In a cleared and stained embryo (72 mm TL) there is a small space between these triangular cartilages and the trabeculae. In all other stages these small car- tilages are attached to the lateral edge of the trabecular cartilage and extend ventrally. It appears that these cartilages may be involved in the formation of the quadrate process of the upper jaw. The fusion of these cartilages to the base of the neurocranium can be seen in a histological section (PQ; fig. 5). I am unable to determine the exact boundaries of the palatoquadrate cartilage, but I interpret it to be, at least in part, the tiny triangular cartilage that fuses to the trabeculae very ear- ly in development (Didier, 1990). Further study of earlier stages should confirm the ex- act formation of the autostylic palatoquad- rate. Schauinsland (1903: fig. 130) illustrated this tiny cartilage in his reconstruction of the neurocranium of an embryonic Callorhin- chus (60 mm TL), and this beautiful figure has been cited and refigured many times in the literature (e.g., de Beer and Moy-Thomas, 1935: fig. 16). Unfortunately, Schauinsland (1903: fig. 130, md) has misinterpreted or mislabelled this cartilage as Meckel’s carti- lage and in his text states that Meckel’s car- tilage is not yet developed. In my study of Callorhinchus embryos a well-developed Meckel’s cartilage is evident below the de- veloping neurocranium in early stages, so I must disagree with Schauinsland’s (1903) in- terpretation.

A spiracle is present in early developmen- tal stages of Callorhinchus and becomes obliterated as the opercular flap develops. The spiracle has also been reported in embryonic Hydrolagus (Dean, 1906). Before skeletal el- ements are beginning to form (in embryos of

20 AMERICAN MUSEUM NOVITATES NO. 3119

Fig. 5. Development of the autostylic jaw in Callorhinchus milli. In this cross section of the head of an embryo of Callorhinchus milii (CM42, 74 mm TL), the fusion of the palatoquadrate to the lateral edges of the trabeculae (TR) is evident. The efferent pseudobranchial artery (EPA) is en- closed in a short canal formed by the connection of the palatoquadrate to the neurocranium. The EPA is visible in more posterior sections dorsal to the trabeculae for a very short distance before entering the orbit. The dorsal surface of the em- bryo is at the top of the figure where the tip of the notochord (NO), surrounded by developing par- achordal cartilages, lies dorsal to the diencephalon (DI) and ventral to the mesencephalon (MES). Scale bar = 0.5 mm.

20-30 mm TL), the hyoid arch is displaced anteriorly and lies in close proximity to the developing jaws. Anterior to the hyoid arch the spiracle can be seen as a small, dorsally located opening (SP; fig. 6). Posterior to the hyoid arch is an enlarged first gill opening. The opercular flap develops as a sheet of tis- sue extending from the posterior edge of the hyoid arch and grows caudally to cover the developing gill arches. It is almost completely formed in embryos of 100 mm TL.

Ri id

Fig. 6. Development and closure of the spi- racle. A close-up of the spiracle is shown in this scanning electron micrograph of an early embryo of Callorhinchus milii (CM74, 40 mm TL). Dor- sally, near the top of the figure, the spiracle is evident as a tiny opening anterior to the hyoid arch. From the posterior edge of the hyoid arch the opercular flap (OP) and external gill filaments (GF) are beginning to develop. As the hyoid arch and opercular flap develop, the spiracle closes and is not present in adults. Scale bar = 1 mm.

EGG CASES

Chimaeroid egg cases are composed of a flexible keratinous material that hardens as it ages in water. The egg cases are of three distinct types, each characteristic of one of the three families. It is difficult to associate chimaeroid egg cases with a particular genus or species unless an embryo is found within the egg case itself or only one species is known from the locality; therefore, most egg cases should be attributed to a family and not toa genus or species.

Callorhynchids have extremely large ovoid egg cases. A representative egg case of Cal-

1995

A | an,

DIDIER: CHIMAEROID FISHES 21

B a

Fig. 7. Egg cases of two chimaeroid fishes. A, An egg case of Callorhinchus milti. The dorsal surface is shown with the anterior end of the egg case at the top. The embryo will develop within the central spindle (SPD), with its long tail extending into the tail sheath (TS). The egg cases of callorhynchids have well-developed, ridged lateral webs (LW). Scale bar = 1 cm. B, The egg cases of an undetermined species of Hydrolagus shown in lateral (left) and ventral (right) view. The flattened anterior end of the spindle (SPD; top of the page) is equipped with perforated lateral edges to allow easy separation of the egg case during hatching. On the dorsal surface of the spindle is a raised keel (K). Scale bar = 1 cm.

lorhinchus milii is 270 mm long and 130 mm wide (fig. 7A). The central spindle (SPD) con- sists of an enlarged oval capsule that tapers to a point at each end; the elongate posterior end forms the tail sheath (TS), which houses the tail of the developing embryo. The thick walls of the central spindle form a sturdy pro- tective shell for the developing embryo. A wide-ribbed lateral web (LW) extends around the edges of the central spindle. The web is quite thin and nearly transparent when held to the light; it is easily ripped and often tat- tered in older egg cases. The dorsal surface of the egg case is convex and is covered by

fine filamentous hairs that trap sediment and camouflage the egg case on the bottom. The smooth ventral side of the egg case is con- cave. At the anterior end of the egg case are two slits at the edge of the central spindle. These slits will gradually open to allow flow of water through the egg case and provide an opening from which the embryo hatches. The egg cases of rhinochimaerids (not fig- ured) are similar to those of callorhynchids in that they have a ribbed lateral web, but the central spindle is constricted giving the egg case a pearlike shape. The dorsal surface is convex and the ventral side is concave. The

22 AMERICAN MUSEUM NOVITATES

egg cases of rhinochimaerids are smaller than callorhynchid egg cases. A typical egg case is about 145 mm long and 60 mm wide. Along the posterior edges of the central spindle 1s a series of small pores. These pores are tightly closed in newly laid egg cases and will grad- ually open to allow flow of water through the egg cases. The anterior end of the spindle has slits that open for release of the embryo at hatching.

Egg cases of Hydrolagus and Chimaera have a relatively smooth teardrop shape with a small lateral web (fig. 7B). The dorsal surface of the egg case is identified by the presence of a raised keel (K) along the midline. A rep- resentative egg case is 170 mm long and 25 mm at its maximum width. The anterior end of the spindle is bulbous and tapers posteri- orly to an elongate tail sheath (TS). Along each side of the tail sheath is a series of small pores that open during later stages of embry- onic development (Dean, 1906). A raised lip extends around the anterior edge of the egg capsule. This lip demarcates the opening through which the embryo is hatched from the egg case. The lip is tightly sealed during the early stages of development and gradually loosens along its edge during development.

EXTERNAL FEATURES OF ADULTS

All chimaeroid fishes have large heads and long, tapering bodies. They rarely exceed 1 m in length, but will grow to masssive sizes by increasing their overall bulk. Each of the three families of chimaeroids is distinguished by a unique snout, which is specialized to house numerous electroreceptive ampullary sense organs (Fields and Lange, 1980; Fields et al., 1993). Internally, the snout is sup- ported by three rostral rods, a dense jelly like material, and connective tissue. Externally, the snout is characterized by groups of am- pullary sense organs that appear as clusters of open pores on the surface. The pores vary in number and size but are located in a con- sistent pattern next to the cranial lateral line canals in all species examined.

The notochord extends to the tip of the tail which often terminates in an extremely long, thin whiplike extension. In callorhynchids, all of which have a heterocercal tail, the no-

NO. 3119

tochord is bent upward and a small hypo- cercal lobe supported by a few small carti- laginous elements is present ventral to the notochord. All other chimaeroids have a lep- tocercal tail in which the notochord is straight. The leptocercal tail is characterized by dorsal and ventral fin webs (the supracaudal and subcaudal lobes of the tail, respectively), which lie opposite each other on the dorsal and ventral side of the notochord and are almost the same size and shape. The fin webs of the supracaudal and subcaudal lobes of the tail are supported solely by ceratotrichia.

Pigmentation of these fishes ranges from pale white, gray, or brown to a deep purple- black. Some species (e.g., Hydrolagus colliei, Hydrolagus novaezealandiae, and Chimaera monstrosa) are patterned with spots and lines. The scaleless skin is smooth with an almost rubbery texture. Small denticles are present on the head and dorsal surface of embryonic and juvenile chimaeroids but are lacking in adults.

One of the most distinctive external fea- tures of chimaeroids is the unusual lateral line canal system that forms a pattern of grooves or tubes on the surface of the skin. In all chimaeroids the canals are supported by C-shaped calcified rings that are open to the surface. These have been regarded as modified scales (Schauinsland, 1903: 13; Pat- terson, 1965: 198). The primitive condition is undoubtedly to have tubular canals, as in elasmobranchs and primitive osteichthyans. This condition is retained in callorhynchids; however, in rhinochimaerids and chimaerids the canals are modified as open grooves.

Although many workers have examined the lateral line canals, there have been few at- tempts to compare them among chimaeroids. The first comparative study (Garman, 1888) examined the pattern of lateral line canals in Chimaera monstrosa and Callorhinchus an- tarcticus. An extensive study of the lateral lines of Chimaera colliei was done by Reese (1910). The lateral line canals of rhinochi- maerids were described and figured by Gar- man (1904), Bullis and Carpenter (1966), Karrer (1972), and Compagno, Stehmann, and Ebert (1990). Bigelow and Schroeder (1953) provided detailed descriptions of the pattern of lateral line canals in a variety of chimaeroid species, including Chimaera cu-

1995

bana, Hydrolagus affinis, Hydrolagus alberti, Harriotta raleighana, and Callorhinchus mil- ii.

The overall pattern of lateral line canals in chimaeroids is unique and difficult to ho- mologize among vertebrates (Patterson, 1965; Northcutt, 1989). The pattern of lateral line canals is poorly understood in chimaeroid fishes and further study of their development is essential for determining their exact ho- mologies and evolution. Until these issues are worked out, any scheme of nomenclature will be incomplete. For my descriptive pur- poses, I use a simplified nomenclature based on morphological landmarks rather than ner- vous innervation (although Fields et al., [1993] provide new information which should result in changes to the interpretation pre- sented here). The terminology I use is based on the system first devised by Garman (1888); with modern terminological modifications from Northcutt (1989), Compagno et al. (1990), and Fields et al. (1993). Occasionally the pattern of lateral line canals differs slight- ly between the right and left sides of the an- imal (Bigelow and Schroeder, 1954; personal obs.). Significant variations are noted in the text.

The pattern of cranial lateral line canals is similar for all chimaeroids, and this descrip- tion begins at the otic region and follows the canals anteriorly and ventrally. Figure 8 il- lustrates the general lateral line canal pattern in chimaerids, although there are some sig- nificant differences that are noted in the in- dividual family descriptions. The supratem- poral canal (ST) forms a commissure across the top of the head just anterior to the fin spine. At the dorsal midline there may be a tiny caudally directed extension of this canal toward the fin spine. This short extension varies among individuals from a straight line to a tiny curl (Bigelow and Schroeder, 1954). From the supratemporal canal the supraor- bital canal (SO) passes anteriorly above the eye. At this junction the occipital canal (OC) continues ventrally and intersects the lateral line canal of the trunk (LLC), which extends posteriorly along the lateral surface of the body to the tip of the tail. In all specimens studied the lateral line canal runs along the lateral body wall and makes a dramatic ven- tral dip at the origin of the supracaudal fin

DIDIER: CHIMAEROID FISHES 23

POP ST

Fig. 8. Comparative cranial lateral line canals and ampullary pore fields of chimaerids. A com- parison of the lateral view of the head is shown for A, Hydrolagus sp. B; B, Chimaera sp. C (NMNZ P19694); and C, Chimaera monstrosa (USNM 222708). Varying branching patterns of the preo- percular (HOC) and oral (OR) canals are shown, and this feature may be species specific. The am- pullary pore fields are also shown for comparison, although the labial and mandibular ampullary fields were not observed in all specimens. See text for a more detailed description of the lateral line canals and ampullary fields of chimaerids. Scale bars = 2 cm.

24 AMERICAN MUSEUM NOVITATES

to run along the ventral edge of the caudal fin to its tip (fig. 8B).

The otic canal (OT) runs posterior to the eye and ventral to the branch point of the lateral line canal. Just below the eye the preo- percular canal (=hyomandibular canal [HOC], Fields et al., 1993) extends ventrally over the opercular flap. This canal ends abruptly and continues around to the ventral midline as a series of short canals. Anterior to the preopercular canal a short canal runs horizontally, and in many species extends ventrally from the infraorbital canal. This horizontal canal (H) soon branches into the angular canal (AN) and the oral canal (OR). The angular canal runs above the mouth and the oral canal extends ventrally below the lower jaw where it breaks up into a series of short canals. The infraorbital canal (IO) runs underneath the orbit and extends rostrally onto the snout.

Associated with the lateral line canals of the head and snout are groups of pores that open to canals housing the ampullary sense organs. These clusters of pores, here termed ampullary fields, were first figured for Hy- drolagus colliei by Allis (1916) and are ho- mologous to the ampullae of Lorenzini of elasmobranchs (Fields et al., 1993). In Allis’ unpublished drawings, three rostral ampul- lary fields and four cranial ampullary fields were identified on the basis of internal anat- omy and innervation. Allis (1916, unpubl.; 1917) also identified a group of tiny, closely packed pores on the snout lying just above the nares in Hydrolagus colliei that he de- scribed as the openings of tiny epidermal gland cells. In keeping with the nomenclature for ampullary fields already established, I adopt the terminology of Fields et al. (1993).

Thirteen ampullary pore fields have been identified for Hydrolagus colliei on the basis of their innervation and pore location (Fields et al., 1993). These ampullary pore fields can be easily identified externally in all chima- eroids, although the relative location of the pore fields and the size and number of pores may vary between species. For a general il- lustration of the ampullary pore fields see fig- ure 8; significant differences are noted in the text.

Eight discrete ampullary pore fields are ev- ident on the lateral surface of the head. The

NO. 3119

supratemporal ampullary pore field (STP) lies just posterior to the orbit along the anterior edge of the otic canal. Just above and slightly anterior to the eye is the preorbital ampullary pore field (POP). Ventral to the supraorbital canal, lying anterior to the preorbital am- pullary pore field and bounded anteriorly by the junction of the supraorbital and infra- orbital canals, is the suprarostral ampullary pore field (SRP). The rostrolateral ampullary pore field (RLP) lies below the infraorbital canal anterior to the eye. In chimaeroids this field lies within the anterior loop of the in- fraorbital canal. At the junction of the infra- orbital and oral canals is the infraorbital am- pullary pore field (IOP), which is bounded caudally by the horizontal lateral line canal. Along the anterior edge of the oral lateral line canal is the oral ampullary pore field (OF), which usually can be seen continuing ante- riorly along the dorsal edge of the lip. Along the dorsal edge of the angular canal is the angular ampullary pore field (APF). These pores sometimes appear to be part of the more anterior rostral ampullary pore field (RPP), which is located ventral to the infraorbital canal and dorsal to the subrostral canal near their junction on the snout. A ninth, spirac- ular ampullary pore field shown as a small cluster of pores posterior to the preopercular canal was described by Fields et al. (1993). I have not found such a group of ampullary pores; however, I have observed a few indi- vidual ampullary pores (not shown) posterior to the preopercular canal in some chimaerid species only. These may be part of the spi- racular ampullary pore field described by Fields et al. (1993).

There are four ampullary pore fields visible on the ventral surface of the snout and around the mouth. The nasal ampullary field (NP) lies below the nasal canal and above the nar- es. Dorsal to the nasal canal, in the region bounded by the subrostral canal, is the sub- rostral ampullary pore field (SUB). Surround- ing the mouth are two ampullary fields con- sisting of small groups of tiny ampullary pores. The mandibular ampullary pore field (MP) consists of groups of tiny pores at the oral edge of the lower lip, near the corners of the mouth, with some larger pores located near the short, ventral portions of the oral canal. The labial ampulary pore field (LP) is located

1995

DIDIER: CHIMAEROID FISHES 25

Fig. 9. External morphology of Callorhinchus milii. A, This specimen is an adult female showing the distinctive external features of callorhynchids, including a plowlike snout (SN), stout fin spine (FS), prominent second dorsal fin (D2), and anal fin (AF). Scale bar = 1 cm. (photograph courtesy of Clive Roberts, NMNZ) B, Heterocercal tail of Callorhinchus milii. The tail figured here shows the posterior dip of the trunk lateral line canal (LLC) and a short distal tail filament (TF). In all chimaeroids the trunk lateral line canal lies dorsal to the horizontal septum until it reaches the caudal fin. At the origin of the supracaudal lobe the trunk lateral line canal turns downward and runs along the ventral edge of the tail.

Scale bar = 2 cm.

on the skin surrounding the labial cartilages at the corners of the mouth. The mandibular and ampullary pore fields could not be pos- itively identified in all specimens. This may indicate their absence in some species or may be due to the small size of the pores and the shrinkage that occurs in preserved speci- mens.

Along the dorsal surface of the snout is what I term the ethmoid ampullary pore field (EAF), which lies between the supraorbital canals. This includes ampullary pores dorsal

to the supraorbital canal and above the junc- tion of the supraorbital and infraorbital ca- nals. This ampullary pore field is described from external observations only, and a de- tailed internal examination of these ampul- lary organs has not been completed. There is no reference to this group of ampullary pores in Fields et al. (1993) and it may be that these ampullary organs are not consistently present in all species or are unrelated to the ampullary organs described in their work. CALLORHYNCHIDAE: Callorhynchids are

26 AMERICAN MUSEUM NOVITATES

NO. 3119

Fig. 10. Cranial lateral line canals and ampullary pore fields of Callorhinchus milii. Lateral line canals and ampullary fields are shown in A, lateral and B, dorsal view; C, ventral surface of the fleshy rostral flap. See text for detailed description of lateral line canals and ampullary pore fields. Scale bar

= 2 cm.

characterized by an elongate snout with a fleshy plow-shaped flap that has no cartilag- inous support (fig. 9A). The ventral surface of the fleshy flap is covered with dense con- centrations of tiny pores that house electro- receptive organs (fig. 10). Although the canal pattern on the ventral surface of the snout flap differs from that of other chimaeroids in that the canals are not continuous across the midline, a nasal, subrostral, and ethmoid am- pullary pore field can be recognized. The re- maining ampullary fields in Callorhinchus are not as extensive as those found in other chi- maeroids. The pores are much smaller and fewer in number. This difference in the am- pullary fields may be related to the presence of a closed lateral line canal in callorhyn- chids.

All callorhynchid species have a hetero- cercal tail (fig. 9B) with a hypocercal lobe in which the ceratotrichia are supported by a series of small cartilaginous basal elements (usually five in number). The tail remains straight throughout development and only after hatching from the egg case does it achieve its heterocercal shape. A short tail filament

extends from the distal end of the epicercal lobe of the tail. Callorhynchids also have a distinct anal fin (AF) that is supported by a single basal element embedded in the ventral musculature of the tail.

The basic pattern of cranial lateral line ca- nals in Callorhinchus milii is shown in figure 10. Callorhinchus is the only chimaeroid in which the oral (OR) and angular (AN) canals branch separately from the infraorbital canal. The oral canal ends at the margin of the mouth and continues around the lower jaw as a series of short canals. The angular canal runs above the mouth and ends abruptly on the ventral surface of the snout. Both the supraorbital and infraorbital canals extend onto the flap of the snout where they end. The supraorbital and infraorbital canals are joined near the tip of the rostrum by a short canal.

The lateral line canals are tubular, as op- posed to the open grooves found in other chimaeroid families (fig. 11A). Although the canals appear as enclosed tubes in callorhyn- chids, they are supported by the calcified rings characteristic of all other chimaeroids. These calcified rings are relatively tiny and closely

1995 DIDIER: CHIMAEROID FISHES

spaced with about 53 per centimeter. The canal is almost entirely enclosed within these calcified rings and the tissue covering the lat- eral line canal is invested with numerous tiny calcifications.

RHINOCHIMAERIDAE: All members of this family are characterized by a fleshy, elongate snout that tapers to a point anteriorly (fig. 12A, B). Sexually mature males of the genus Harriotta are interesting in that they develop a series of rounded tubercles on the tip of the snout. The caudal fin is leptocercal, and basal cartilages are lacking in both the supracaudal and subcaudal fins in all rhinochimaerids. The tail ends in a long whiplike extension that extends posteriorly beyond the supracaudal and subcaudal lobes of the tail. This tail fil- ament may be extremely long, almost equal to body length in some species. In Rhino- chimaera and Neoharriotta the subcaudal lobe has long ceratotrichia, whereas the ceratotri- chia of the supracaudal lobe are short, giving the tail an externally heterocercal shape. In Rhinochimaera the supracaudal lobe of males develops a series of paired tubercles along the distal margin of the fin web. Neoharriotta is distinguished among rhinochimaerids by the presence of an anal fin. The anal fin in Neo- harriotta consists of a web of ceratotrichia that is supported by three tiny cartilages that form the fin base. Callorhynchids are the only other chimaeroids known to have an anal fin.

~

Fig. 11. Comparison of the lateral line canals among the three families of chimaeroids. A, Scan- ning electron micrograph of the closed lateral line canals of Callorhinchus milii at the junction of the infraorbital (IO) and oral (OR) canals. Lateral line openings (LLP) occur along the length of the canal. Scale bar = 0.5 mm. B, Scanning electron micro- graph of the open lateral line canals of Harriotta raleighana (AMNH 96944) at the junction of the infraorbital (IO) and oral (OR) canals. The canals are open grooves supported by C-shaped calcified rings (R). Scale bar = 1 mm. C, Scanning electron micrograph of the open lateral line canals of Hy- drolagus sp. B at the junction of the occipital (OC) and otic (OT) canals. A large ampullary pore lies lateral to the grooved canal, which is supported by calcified rings (R). The open, C-shaped rings are branched at their dorsal ends (often seen in chimaeroids), giving the edge of the canal a fringed appearance. Scale bar = 0.5 mm.

28 AMERICAN MUSEUM NOVITATES NO. 3119

Cc amzco Mek COMP

Fig. 12. External morphology of rhinochimaerid and chimaerid fishes. A, Photograph of Rhinochi- maera pacifica (AMNH 96940). This immature male specimen, from the Challenger Plateau (fig. 1B, #4), measures 950 mm TL with a snout length of almost 250 mm. Characteristic of this genus is the wide, fleshy snout (SN), small eyes (E), and whiplike tail with elongate supracaudal (SCL) and subcaudal (SUL) lobes, the subcaudal lobe being much deeper. The fin spine (FS) in this specimen is broken. B, Photograph of Harriotta raleighana from Mernoo Bank (fig. 1B, #3). The snout (SN) of this genus is more firm than fleshy and tapers to a fine point anteriorly. Unlike Rhinochimaera, the eye is relatively

1995

DIDIER: CHIMAEROID FISHES 29

Fig. 13. Cranial lateral line canals and ampullary pore fields of Rhinochimaera pacifica (NMNZ P24198). Both the hyomandibular and oral lateral line canals terminate in a ventral series of short “broken” canals, illustrated dashed lines. The ampullary pore fields are shown in their approximate locations as groups of open circles. The size and number of pores shown are indicative of their actual size and density. See text for a detailed description and terminology. Scale bar = 2 cm.

All rhinochimaerids have open lateral line canals supported by calcified rings (fig. 1 1B). The canals of Rhinochimaera pacifica are supported by 39 calcified rings per centime- ter. In Harriotta raleighana there are 18 rings per centimeter.

In rhinochimaerids the angular (AN) and oral (OR) canals share a common trunk, the horizontal canal (H), which curves antero- ventrally from the infraorbital canal (figs. 13- 15). Although the rostrum is elongate, the general pattern of lateral line canals is con- served (fig. 16). The supraorbital canal runs along the dorsal surface of the snout, turning under at the anteriormost tip, where it con- tinues on the ventral side of the snout to join the infraorbital canal. Posterior to this junc- tion this canal continues along the lateral edge of the snout as the subrostral canal (SR), which joins the angular canal in Harriotta and Neo- _harriotta. Rhinochimaera lacks a connection

,

between the angular and subrostral canal (Garman, 1904; Compagno, Stehmann and Ebert, 1990). This connection between the angular and subrostral canals is not a syna- pomorphy of Harriotta and Neoharriotta be- cause a connection is lacking in Neoharriotta carri (observed but not figured; Bullis and Carpenter, 1966).

The angular canal runs above the mouth and dips down onto the ventral surface of the snout where it loops above the nostrils. The nasal canal (N) is distinguished as that por- tion of the angular canal anterior to the junc- tion of the subrostral canal. In Rhinochi- maera, which lacks such a connection, the nasal canal is distinguished as that portion of the angular canal looped above the nostrils.

CHIMAERIDAE: Members of this family are characterized by a blunt fleshy snout (fig. 12C). The cartilaginous rostral rods that support the snout are greatly reduced. The snout is

large and the supracaudal and subcaudal lobes of the leptocercal tail are not elongate and are almost equal in size. C, Photograph of Hydrolagus sp. B. This species of Hydrolagus is common from Mernoo Bank, New Zealand, at 500-800 m depth. Evident in this photograph is the filamentous whip of the tail (TF) found in many chimaeroid species, an elongate second dorsal fin (D2), and snub-nosed snout (SN) characteristic of chimaerids. Scale bar = 5 cm; a 15-cm ruler is shown for scale in all three photographs.

30 | AMERICAN MUSEUM NOVITATES

NO. 3119

Fig. 14. Cranial lateral line canals and ampullary pore fields of Harriotta raleighana. The pattern of lateral line canals and location of ampullary pore fields is shown in A, lateral and B, dorsal view. The tubercles on the tip of the rostrum (KN) are shown as solid black dots, and in dorsal view the frontal tenaculum (FT) is shown atop the head. The oral (OR) and hyomandibular (HOC) canals end in a series of short “broken” canals, illustrated as dashed lines. The size and density of open pores reflect their actual size and density; the labial and mandibular ampullary pore fields are often small and were not observed in this specimen. See text for more detailed description and terminology. Scale bar = 2 cm.

supported by dense connective tissue and is almost entirely filled with a firm jellylike ma- trix.

Fig. 15. Cranial lateral line canals and am- pullary pore fields of Neoharriotta pinnata (UF 23839). Only the more prominent ampullary pore fields are shown. The general pattern of lateral line canals in Harriotta and Neoharriotta is similar, although in this figure the subrostral canal is not evident in lateral view. See text for more detailed description and terminology. Scale bar = 2 cm.

The tail is leptocercal with small supra- caudal and subcaudal lobes. Distally, the tail extends to a long, whiplike tail filament (TF). There is no anal fin present; however, the genus Chimaera is defined by the presence of a notch in the subcaudal fin that separates the anal fin web from the subcaudal fin web.

The lateral line canals are open grooves supported by calcified rings (fig. 11C). In all species of chimaeroids with open lateral line canals the calcified rings tend to split at the ends, giving the edge of the canal a fringed appearance. The narrow portions of the ca- nals are supported by numerous calcified rings ranging from 28 per centimeter in Chimaera sp. C to 53 per centimeter in Hydrolagus no- vaezealandiae. The enlarged snout canals are supported by more widely spaced rings (e.g., five rings between each dilation in Hydrola- gus novaezealandiae and five to seven rings between each large dilation in Hydrolagus colliei).

1995

DIDIER: CHIMAEROID FISHES 31

Fig. 16. Ventral view of snouts of rhinochimaerids. The snouts of A, Rhinochimaera pacifica (NMNZ P24198), B, Harriotta raleighana, and C, Neoharriotta pinnata (UF 23839) are shown in ventral view to illustrate the different branching patterns of the subrostral (SR) and nasal (NA) canals. Also shown are the ampullary pore fields to allow comparison of the relative size and abundance of ampullary pores among these species. See text for a detailed discussion of the different canal patterns. Scale bars = 2 cm.

The basic pattern of cranial lateral line ca- nals in chimaerids follows that described above for callorhynchids and rhinochimaer- ids. The supratemporal, occipital, otic, hyo-

mandibular, supraorbital, and infraorbital |

canals share the same branching pattern in all taxa of chimaerids studied (fig. 17). The horizontal canal in most chimaerids branches off the infraorbital canal separate from the preopercular canal; however, in some species the preopercular canal may also share a com- mon trunk with the horizontal canal (fig. 8). On the basis of the comparisons made in this study this feature appears to be species-spe- cific within Chimaeridae.

In the snout region the supraorbital, an- gular, subrostral, nasal, and infraorbital ca- nals of chimaerids become enlarged and are interrupted by dilations at regular intervals along their length. The supraorbital and in- fraorbital canals, which are narrow along most of their length, begin to widen and open into large pores as they extend into the snout re- gion, where they join at the anterior tip of

the snout. From this junction a short canal forms a Y-shaped commissure across the snout. At its ventral end the right and left subrostral canals split and join the angular canals on the ventral surface of the snout. The nasal canal extends from the junction of the subrostral and angular canals and forms an inverted “V”’ on the ventral surface of the snout above the nostrils.

SKELETAL ANATOMY

Skeletal features have long been held to have special phylogenetic importance be- cause these characters can be studied in fos- sils as well as in living forms. Many workers have described aspects of the skeletal anat- omy of chimaeroid fishes (e.g., Hubrecht, 1876; Schauinsland, 1903; Dean, 1906; Allis, 1917, 1926; Rabinerson, 1925; de Beer and Moy-Thomas, 1935; de Beer, 1937; Patter- son, 1965; Stahl, 1967; Ribbink, 1971; Rai- kow and Swierczewski, 1975; Jarvik, 1980). This description is the first comprehensive

32 AMERICAN MUSEUM NOVITATES

Fig. 17. Cranial lateral line canals and am- pullary pore fields of Hydrolagus novaezealandiae. The lateral line canals and ampullary fields are shown in A, lateral, B, dorsal and C, ventral view. All of the ampullary pore fields described in the text are shown in this figure, including the labial ampullary pore field (shown in C, not labelled), which appears as a group of small pores on the fleshy lips surrounding the mouth. The relative abundance and size of pores are to scale. See text for more detailed description and terminology. Scale bar = 2 cm.

study of the skeleton of all three families of chimaeroid fishes and is based on detailed examination of six species in five different genera (table 7). |

NEUROCRANIUM, JAWS, AND LABIAL CARTILAGES

The neurocranium is characterized by sev- eral specializations, particularly the autos-

NO. 3119

tylic jaw. As in other autostylic gnathos- tomes, the hyoid arch does not participate in supporting or suspending the lower jaw. The class name, Holocephali (Muller, 1834), de- rives from the fact that there is complete fu- sion of the palatoquadrate to the neurocran- ium.

Some features of the neurocranium can be generalized among all chimaeroids (refer to fig. 18A). The orbit is anterior to the brain and dorsal to the paired olfactory tracts, which pass ventral to the orbit into the ethmoid region of the skull. Anteriorly, the orbit is bounded by a sheet of dense connective tissue from which the adductor mandibulae mus- cles originate. Dorsally, at the anterior edge of the orbit, the antorbital crest (ANC) forms a small lateral projection from which origi- nates a caudally directed sheet of connective tissue that forms the top of the orbit. This connective tissue sheet inserts onto the post- orbital ridge (POR), which forms the poste- rior boundary of the orbit. The orbit is bounded ventrally by the suborbital ridge (SOR). This shelf of cartilage originates just posterior to the jaw joint and ends at the otic process (OP). The orbits are separated by a thin wall of connective tissue, which forms the interorbital septum.

The ethmoid region of the neurocranium is a broad area anterior to the orbit. The nasal capsules (NC) bulge laterally from the an- teroventral portion of the ethmoid region. They are open posteriorly for the passage of the olfactory tracts into the olfactory bulbs. At the external opening of each nasal capsule is a scroll of cartilage that divides the nostril into incurrent and excurrent openings. The nasal openings lie within the fleshy tissue of the lips and are continuous with the mouth. This connection between the mouth and nos- trils was described by Jarvik (1980) as a pseu- dochoana. Extending anteriorly from the eth- moid region are paired lateral rostral rods (LRR) that lie medial to the nasal capsules. Dorsal to the lateral rostral rods is a single median rostral rod (MRR).

The ethmoid canal and olfactory tracts ex- tend into the ethmoid region. The ethmoid canal is an enclosed tube dorsal to the nasal capsules. The ophthalmic foramen (OPF) at the anterior edge of the orbit forms the en-

1995 DIDIER: CHIMAEROID FISHES 33

PMD

Fig. 18. Neurocranium of Callorhinchus milii. A, In this illustration of the neurocranium the gill arches and pectoral girdle are shown in their natural position in relation to the neurocranium. The labial cartilages are partially concealed by connective tissue (dashed lines). B, Detail of the synarcual and articulation of the dorsal fin spine (FS) and basal cartilage of the first dorsal fin (BD1). C, Lateral view of the six labial cartilages of Callorhinchus milii illustrated as they articulate within the connective tissue surrounding the mouth. See text for a detailed description of neurocranial anatomy and terminology.

Scale bar = 2 cm.

trance to this canal, which exits from the ros- tral foramen (RF) at the base of the median rostral rod. The canal contains arteries pass- ing to the snout, a mass of lymphoid tissue, and nerves for the skin and special sense or- gans of the snout. The nerves that traverse the ethmoid canal are the superficial oph- thalmic branch of the anterior lateral line nerve and the profundal nerve (Cole, 1896a; Holmgren, 1942; Northcutt, 1989; Song and Northcutt, 1991). The anterior cerebral vein also runs within the ethmoid canal along part of its length (Holmgren, 1942: 200). Based on the study of a single embryo it was sug- gested that this canal forms as a secondarily roofed-over passage (de Beer and Moy-Tho-

mas, 1935). Holmgren (1942) disagreed with this interpretation and suggested that this ca- nal is a remnant of the cranial cavity. My study of the skeleton of Rhinochimaera shows that the ethmoid canal is entirely enclosed within the ethmoid region and studies of the embryonic development of the ethmoid ca- nal in Callorhinchus show it to be formed as an extracranial space, which is roofed over relatively late in development (fig. 19). Posterior to the orbit in all chimaeroids is the otic region. The semicircular canals are visible as external bulges on the dorsal sur- face of the neurocranium. An important de- rived feature of chimaeroids is that the otic capsule has no internal wall. Posteriorly, the

34 AMERICAN MUSEUM NOVITATES

NO. 3119

Fig. 19. Development of the ethmoid region in Callorhinchus milii. In this sagittal section of the head of an embryo (CM51, 85 mm TL) the developing ethmoid canal (EC) lies above the telencephalon (TEL) and is separated from the brain by a cartilaginous floor. Within the ethmoid canal is a branch of the Vth cranial nerve (NV). The roof of the canal develops later and is almost fully formed in this embryo. The median rostral rod (MRR) extends into the snout (SN) and develops dorsal to the nasal capsule (NC). The vomerine tooth plate (VTP) is evident directly below the nasal capsule. Scale bar =

1 mm.

otic region extends to a prominent dorsal pro- cess, the occipital crest (OCC). From this crest a sheet of dense connective tissue inserts onto the anterior edge of the synarcual (SYN) and tightly binds the neurocranium to the syn- arcual. The synarcual is a cartilaginous plate formed by the fusion of the first 10 vertebral segments. On the dorsal edge of the synarcual is an articulation surface for the basal carti- lage of the first dorsal fin and the fin spine (fig. 18B). At the back of the neurocranium are two condyles that articulate with the base of the synarcual. There is only a loose con- nection between the neurocranium and syn- arcual at this joint. The otic process (OP) is the posterior edge of the suborbital shelf ven- tral to the otic capsule. I use the term otic process for convenience only, with no refer- ence to its embryological origin.

The ventrally directed jaws are located an- terior to the orbit. The jaw articulation of all Recent chimaeroids is anterior to the eye, and all the jaw closing muscles originate anterior to the eye. Meckel’s cartilage (M) forms the lower jaw and is fused at the symphysis form- ing a U-shaped element that is suspended

below the neurocranium. This lower jaw car- tilage is short and deep with the line of the symphysis extending to a prominent chin process (CP) at the posteroventral edge of the lower jaw. All chimaeroids have a double- articulating jaw joint in which the quadrate process is equipped with a lateral process and medial fossa that interlock with the lateral fossa and medial process of Meckel’s cartilage (Didier, 1988). This type of jaw joint was interpreted to be primitive for gnathostomes by Hotton (1952) and Miles (1964). Along the posterior edge of the lower jaw is the re- troarticular process (RP).

Labial cartilages are found in selachians and batoids as well as chimaeroids; however, in no other extant chondrichthyans are the labial cartilages as complex as among the liv- ing chimaeroids. Large, complex labial car- tilages, similar to those found in chimaeroids, are described for Hybodus basanus (Maisey, 1983). However, the homology of these labial cartilages is difficult to determine and will require further study, including comparative examination of the embryonic development of labial cartilages in chimaeroids and mod-

1995

ern sharks. With the exception of Callorhin- chus, which has five pairs of upper labial car- tilages, chimaeroids have four pairs of upper labial cartilages and one pair of lower labial cartilages. The terminology used to describe the labial cartilages is based primarily on Holmgren (1942: 242-243).

CALLORHYNCHIDAE: In Callorhinchus milti the neurocranium is dorsoventrally com- pressed, with an elongate antorbital and otic regions giving the neurocranium a long, low profile when compared with other chima- eroids (fig. 18A). The prominent ethmoid process (EP) extends from the dorsal midline of the ethmoid region. Likewise, the occipital crest (OCC) is a prominent posterior process extending from the dorsal midline of the otic region. In Callorhinchus the orbit and inter- orbital septum are small.

The plow-shaped snout of Callorhinchus is supported by two lateral rostral rods and a single median rostral rod, all three being of approximately equal length. A posteriorly di- rected flap of tissue is suspended from the distal ends of these rods. In Callorhinchus the median rostral rod is located just dorsal to the nasal capsules, in close proximity to the origin of the lateral rostral rods.

There are five paired labial cartilages in Callorhinchus (fig. 18C). The prelabial car- tilage (PL) is a dorsally directed S-shaped car- tilage that lies anterior to the nasal capsule and lateral to the rostral rods. A ligament extends from the dorsal tip of the prelabial cartilage and inserts onto the lateral rostral rod. Another ligament originates from a tiny dorsal process of the nasal capsule and inserts onto the prelabial cartilage. At its ventral end, the prelabial cartilage articulates with the pe- dicular cartilage (PED). The pedicular carti- lage is directed medially and lies at the an- teroventral edge of the nasal opening. Articulating with the pedicular cartilage is the premaxillary cartilage (PMX), which is en- sheathed in dense connective tissue and hangs ventrally to support the upper lip. The elon- gate superior maxillary cartilage (SMX) also articulates with the pedicular cartilage and lies ventral to the nasal capsule. The superior maxillary cartilage is deeply curved along its dorsal edge and has a dorsal extension at its posterior end. Articulating at the ventral edge of the superior maxillary cartilage is the in-

DIDIER: CHIMAEROID FISHES 35

ferior maxillary cartilage (IMX). This carti- lage lies lateral to the mouth and is enveloped in a fold of connective tissue that forms the side of the mouth. The premandibular car- tilage (PMD) is the only labial cartilage as- sociated with the lower jaw. These large paired elements lie lateral to Meckel’s cartilage with- in the thick tissue of the lower lip.

RHINOCHIMAERIDAE: Rhinochimaera has a robust neurocranium with an elongate, ta- pering ethmoid region (fig. 20A). There is no prominent ethmoid process. The elongate median rostral rod articulates with the neu- rocranium dorsal to the nasal capsules and extends to the tip of the snout, becoming deeper and laterally flattened midway along its length (MRR). The lateral rostral rods are angled upward and support only the base of the rostrum. The otic region is long with a rounded occipital crest. The neurocranium of Rhinochimaerais \ess angular and box-shaped than is that of Callorhinchus and instead ap- pears as though it is stretched in a cranio- caudal direction. As in Callorhinchus, the or- bit and interorbital septum are small.

In Harriotta and Neoharriotta the neuro- cranium is deep dorsoventrally with short ethmoid and otic regions (fig. 21A). The or- bits are large with an extensive interorbital septum composed of connective tissue. The ethmoid and otic regions are almost equal in size, and in lateral view the neurocranium appears to be very short and tall.

The labial cartilages of Rhinochimaera are large (fig. 20B). The prelabial cartilage (PL) is a nearly straight rod and does not have a ligamentous connection either to the rostral rods or to the nasal capsule. This prelabial cartilage curves below the nasal capsule and extends medially to articulate with the neu- rocranium. There is no pedicular cartilage. The long, thin premaxillary cartilage (PMX) articulates with the ventral edge of the pre- labial cartilage and hangs ventrally within the skin of the upper lip. The blocklike superior maxillary cartilage (SMX) articulates at the posterior edge of the prelabial cartilage. Ar- ticulating with the superior maxillary carti- lage is a large, ventrally directed cartilage, the inferior maxillary (IMX), that lies lateral to the mouth opening and supports the edge of the lip. Associated with the lower jaw are the paired premandibular cartilages (PMD).

36 AMERICAN MUSEUM NOVITATES NO. 3119

MRR ANC opr POR OCC SYN

Fig. 20. Neurocranium of Rhinochimaera pacifica (AMNH 96939). A, The pectoral girdle has been removed to reveal the gill arches which have been displaced ventrally from their normal position to illustrate their structure in lateral view. All of the labial cartilages except the prelabial (PL) and pre- maxillary (PMX) have been removed. B, Detail of the labial cartilages of Rhinochimaera pacifica showing all five labial cartilages in approximate relationship to each other. The premandibular cartilages (PMD) lie within connective tissue ventral to the lower jaw and are illustrated as though the connective tissue were cut and peeled away from Meckel’s cartilage. Scale bar = 2 cm.

These thin, flat cartilages are joined by con- harriotta are similar in their arrangement and nective tissue at the symphysis and lie within articulations to those of Rhinochimaera (fig. the connective tissue of the lower lip. 21B). The premaxillary cartilage (PMX) is

The labial cartilages of Harriotta and Neo- tiny and located entirely within the upper lip;

POR OCC

Fig. 21. Neurocranium of Harriotta raleighana. A, The gill arches and pectoral girdle have been ventrally displaced in this illustration to expose details of their anatorny in lateral view. Of the labial cartilages only the prelabial cartilage (PL) is illustrated. B, Detail of the four upper labial cartilages of Harriotta raleighana (AMNH 96935). The mandibular cartilages are small 2lements buried in connective tissue ventral to the lower jaw and are not shown here. Scale bar = 2 cm.

1995

DIDIER: CHIMAEROID FISHES 37

Fig. 22. Neurocranium of Hydrolagus novaezealandiae. A, In this lateral view the pectoral girdle and gill arches have been displaced ventrally. Only the prelabial (PL) and inferior maxillary (IMX) labial cartilages are shown. B, Detail of the four upper labial cartilages of Hydrolagus novaezealandiae. The mandibular labial cartilages, not shown, are reduced to two very tiny bits of cartilage ventral to the

lower jaw. Scale bar = 2 cm.

it may or may not articulate with the prelabial cartilage. The superior maxillary and inferior maxillary cartilages are similar to those de- scribed for Callorhinchus and Rhinochi- maera, but are much smaller. There is a sin- gle flat premandibular cartilage embedded in dense connective tissue at the symphysis of the lower jaw.

CHIMAERIDAE: The chimaerids Hydrolagus and Chimaera have tall neurocrania with large orbits (fig. 22A). The ethmoid and otic regions are short and almost equal in length. In these respects, the neurocrania of Chimaera, Hy- drolagus, Neoharriotta, and Harriotta are similar; however, this is probably a correlat- ed set of characters. Anteriorly, there is a slightly rounded ethmoid process (EP). The prominent occipital crest extends from the dorsal midline of the otic region and differs morphologically from the rounded occipital crest (OCC) of rhinochimaerids.

The small lateral rostral rods lie along the medial edge of the nasal capsule and are di- rected upward. A strand of connective tissue extends from the tips of the lateral rostral rods to a sheet of connective tissue that sup- ports the rostrum and its associated ampul- lary organs. The small median rostral rod articulates just ventral to the ethmoid process and is embedded in the connective tissue of the fleshy snout.

The labial cartilages are small in all chi- maerids (fig. 22B). The prelabial cartilage (PL) of Hydrolagus has a short dorsally directed process and is elongate posteriorly. The pre- maxillary cartilage (PMX) articulates with the anterior edge of the prelabial cartilage and this flat, crescent-shaped cartilage lies within the upper lip. The superior maxillary (SMX) and inferior maxillary cartilages (IMX) retain the same positional relationships as de- scribed above for rhinochimaerids. In some

38 AMERICAN MUSEUM NOVITATES

specimens a tiny cartilage or fibrocartilagi- nous mass lies between the prelabial and su- perior maxillary cartilage. The paired pre- mandibular cartilages are minute fibrocartilage masses embedded in connec- tive tissue at the symphysis of the lower jaw (not shown).

GILL ARCHES

All holocephalans have a morphologically complete hyoid arch plus five typical gill arches. The branchial skeletal elements are concentrated underneath the neurocranium and covered laterally by a fleshy opercular flap. There is a single gill opening on each side that opens ventrally just anterior to the pectoral fin. There is a single demibranch on the hyoid arch and four holobranchs, one as- sociated with each subsequent gill arch. The fifth gill arch is reduced and lacks respiratory specializations.

Holocephalans are the only gnathostomes that have a hyoid arch in which a pharyn- gohyal element is present. This has been in- terpreted as a retention of the plesiomorphic condition for gnathostomes (de Beer and Moy-Thomas, 1935) on the basis of a study of a single embryo of Callorhinchus milii. However, Maisey (1984) argued that the hy- oid arch of Recent chimaeroids is not an un- modified primitive hyoid arch on the basis of five anatomical features that do not sup- port the hypothesis of de Beer and Moy-Tho- mas (1935): (1) the pharyngohyal of chimae- roid fishes lies lateral to the efferent hyoidean artery; (2) the embryonic spiracle is small and does not represent a complete gill slit; (3) the epihyal does not bear endoskeletal rays; (4) the hyoid arch of Recent chimaeroids does not have branchial adductor muscles; and (5) the interpharyngobranchial muscles do not insert on the pharyngohyal. Furthermore, Maisey (1989) suggests that there is no evi- dence that the hyoid arch of gnathostomes was ever unmodified in the sense that it ex- hibited the characteristic morphology of a typical branchial gill arch.

In my observations of some earlier devel- opmental stages a discrete cartilage anlage for each hyoid arch element is present. Each an- lage appears to develop into a single element with no contribution to or from any nearby

NO. 3119

elements, suggesting that this hyoid arch does develop in an unmodified form; however, based on this developmental evidence alone, it is not clear that this hyoid arch represents the primitive state. Instead, the development of a complete hyoid arch may be a second- arily derived feature associated with the evo- lution of autostyly in this lineage. This seems a reasonable hypothesis based on the conclu- sion of Maisey (1980) that an amphistylic jaw suspension (Maisey’s hyostyly) is the primi- tive condition for gnathostomes. Outgroup comparison would then dictate that autostyly is derived in Holocephali, and it is therefore more parsimonious to accept the complete hyoid arch as a secondary derivation.

The hyoid arch in chimaeroids consists of pharyngohyal, epihyal, ceratohyal, and ba- sihyal elements (fig. 23A, B). The epihyal ar- ticulates at the dorsal end of the ceratohyal. At its posterior edge is a facet for the oper- cular cartilage that articulates at the junction of the epihyal and ceratohyal. This cartilage extends posteriorly to a thin filament. Con- tinuous with the opercular cartilage are nu- merous filamentous processes, the hyoid rays, which support the opercular flap. Several hy- oid rays also articulate independently with the posterior edge of the ceratohyal. The cer- atohyal is the largest of the hyoid arch ele- ments. It is angled with a short dorsally di- rected caudal portion and a cranial portion that extends anteriorly to articulate with the basihyal. A single basihyal supports the tongue pad and in many species there are additional paired cartilages or fibrocartilaginous masses just posterior to the basihyal element.

The remaining gill arches are almost iden- tical among all chimaeroid genera, and the condition in Callorhinchus serves to illustrate the general pattern (fig. 23A, B). Ventrally, there are four unpaired basibranchial ele- ments, the anterior three of which exist as fibrocartilage lumps embedded within the connective tissue of the branchial basket (BBR). In chimaerids there are accessory paired fibrocartilages associated with the first three basibranchials. These are figured by Garman (1904). The fourth basibranchial is a flat, posteriorly directed cartilage that in- serts into a connective tissue sheet attached to the medial surface of the coracoid. Basi- branchial 1! articulates directly with hypo-

1995

branchial 1; basibranchial 2 lies between but does not articulate with hypobranchials 2 and 3; and basibranchial 3 lies between but does not articulate with hypobranchials 3 and 4. These basibranchial elements are intercon- nected to the hypobranchials and to each oth- er by a surrounding loose connective tissue fascia.

There are four separate hypobranchial ele- ments (HBR). Each articulates with one cer- atobranchial except for the most caudal one, which articulates with both ceratobranchials 4 and 5. I consider that this hypobranchial element may have arisen by fusion of origi- nally separate hypobranchials 4 and 5 and therefore term it HBR 4—5. Each of the three anterior hypobranchial elements articulates with its corresponding ceratobranchial ele- ment.

The five ceratobranchials are long dorso- ventrally curved elements (CTB). Cerato- branchials 1-3 have posteriorly directed pro- cesses at their ventral articulation surfaces that articulate with the hypobranchial ele- ments. This ceratobranchial process is most evident in the first ceratobranchial and be- comes smaller in ceratobranchials 2 and 3.

Dorsal to the ceratobranchials are three epibranchial elements (EBR) that articulate with ceratobranchials 1, 2, and 3. The first epibranchial is concave along its anterior edge with a dorsal anterior process that articulates at the junction of the pharyngohyal and epih- yal. There are no epibranchial elements as- sociated with ceratobranchials 4 and 5.

There are two elongate, bladelike pharyn- gobranchial elements (PBR) that are angled dorsoventrally and medially. The first pha- ryngobranchial articulates at the junction of epibranchials 1 and 2, and the second pha- ryngobranchial articulates at the junction of epibranchials 2 and 3. The third pharyngo- branchial is a large element and is probably formed by the fusion of several cartilages, including epibranchials 4 and 5; thus, I label this element PBR 3-5. This element lies di- rectly behind pharyngobranchial 2 and artic- ulates with epibranchial 3 anteriorly. Its an- terodorsal edge continues as an elongate dorsal process that is bound to the ventral side of the notochord by dense connective tissue. Ceratobranchials 4 and 5 articulate at the ventral edge of the body of this element. The

DIDIER: CHIMAEROID FISHES 39

Fig. 23. Gill arches of Callorhinchus milii (AMNH 96954). A, Lateral view of the gill arches shown in this figure as representative of the gill arch morphology for all chimaeroids. B, Gill arch- es shown in ventral view. The solid lines represent loose connective tissue that binds the basibran- chial elements (BBR) to the hypobranchial ele- ments (HBR). The ceratobranchial elements (CTB) are shown in cutaway view. See text for specific details of gill arch morphology. Scale bar = 2 cm.

element extends caudally as a posterior pro- cess that is tightly bound to the medial side of the scapular process.

CALLORHYNCHIDAE: The very tiny phar- yngohyal articulates at the dorsal surface of the epihyal and is roughly rectangular in Cal- lorhinchus (PHY; fig. 23A). The basihyal in Callorhinchus is quite small and is the only cartilage observed in the tongue pad (BHY).

The opercular cartilage (OPC) is a roughly square blocklike cartilage that supports about

40 AMERICAN MUSEUM NOVITATES

five hyoid rays. An additional five or six hy- oid rays are attached to the posterior edge of the ceratohyal. It is unlikely that the number of rays is constant among individuals, be- cause the hyoid rays often split and fuse.

RHINOCHIMAERIDAE: The pharyngohyal is a tiny boomerang-shaped element in Rhin- ochimaera (PHY; fig. 20A). The basihyal is composed of two elements; the anteriormost is roughly triangular in shape and extends to a point at its tip. This cartilage is followed by a second smaller element. In Rhinochi- maera the basihyal element is robust, where- as in Harriotta and Neoharriotta the basihyal is more slender and pointed. The epibran- chial cartilages figured for Harriotta (unla- beled; fig. 21A) are small; however, this is a young individual and my observations of other taxa indicate that larger adults have better developed branchial cartilages and this may be the case for Harriotta as well. In Har- riotta a tiny cartilage element has been ob- served dorsal to epibranchial 1 and just an- terior to the first pharyngobranchial (unlabeled; fig. 21A). This element has also been observed in species of Hydrolagus and may be a transient element.

The elongate opercular cartilage in species of Rhinochimaeridae is largest at its articu- lation with the hyoid arch and curves to a thin filament posteriorly (figs. 20A, 214A). Rhinochimaera has roughly 7 hyoid rays ar- ticulating with the opercular cartilage and 12 hyoid rays that articulate at the posterior edge of the ceratohyal. There are about 13 hyoid rays that originate from the opercular carti- lage in Harriotta and only 5 hyoid rays as- sociated with the ceratohyal. Neoharriotta has a total of 15 hyoid rays, 6 of which articulate at the posterior edge of the ceratohyal.

CHIMAERIDAE: The pharyngohyal in Hy- drolagus resembles that of rhinochimaerids (PHY; fig. 22A). The basihyal (BHY) is a large, squared element that does not extend to a prominent point anteriorly.

The opercular cartilage in species of Hy- drolagus and Chimaera is a thin, lunate car- tilage that supports numerous hyoid rays. It is difficult to distinguish hyoid rays that ar- ticulate along the ceratohyal from those that extend from the opercular cartilage. The number of hyoid rays is variable. For ex- ample, Hydrolagus colliei has a total of 25-

NO. 3119

27 hyoid rays, Hydrolagus mirabilis has 21, and there are 27 in Chimaera monstrosa. This increase in number of hyoid rays may be characteristic of the family Chimaeridae; however, a more thorough survey of this character is needed.

VERTEBRAL COLUMN AND UNPAIRED FINS

The vertebral column of all chimaeroids lacks true centra. The sheath of the notochord in chimaeroids is invested with numerous calcified rings, although Callorhinchus is an important exception to this. These noto- chordal rings are not segmentally organized and there are typically many rings per seg- ment. External to the calcified rings, along the dorsal and ventral surfaces of the noto- chord, is a series of small cartilages that are unrelated to the pattern of calcified rings. In all species examined these cartilages are well developed anteriorly with triangular-shaped dorsal cartilages, minute interdorsal carti- lages, and small ventral cartilages. In the an- terior half of the body, anterior to the pelvic girdle, these cartilages seem to reflect a seg- mental pattern, although the dorsal and ven- tral series do not necessarily exhibit a one- to-one correspondence. In the posterior half of the body, and into the tail, the cartilages are formed into irregular rectangular shapes and no longer exhibit any recognizable pat- tern. These cartilages are probably homolo- gous to the basidorsal and basiventral series in the vertebral column of elasmobranchs and other fishes.

All chimaeroids have a vertical bladelike synarcual (SYN; fig. 18B) that supports the first dorsal fin and fin spine. The synarcual forms the anterior portion of the vertebral column and encloses the notochord. The base of the synarcual is penetrated by 10 foramina through which emerge the anterior spinal nerves.

The first dorsal fin is supported by a single basal element and is preceded by a stout fin spine that is attached to the basal of the first dorsal fin and together they articulate on the dorsal process of the synarcual (fig. 18B). The fin spine and first dorsal fin can collapse into a fleshy groove along the dorsal midline.

The fin spine is roughly triangular in ex- ternal shape and is thickest at the base and

1995

tapers to a needle-sharp point. The anterior edge of the spine may be smooth or deeply keeled. The posterior face is concave with a series of recurved hooks along the lateral edg- es. The fin spine is considered to be ven- omous and the tissue within the posterior concavity of the fin spine of Hydrolagus col- liei contains venom glands (Halstead and Bunker, 1952).

The second dorsal fin is supported by nu- merous small radial cartilages embedded in connective tissue along the dorsal midline. These radial elements support the ceratotri- chia, which form the fin web. The number of radials varies depending on the length of the fin.

CALLORHYNCHIDAE: Species of Callorhin- chus do not have calcified rings in the sheath of the notochord, but they do possess a series of small basidorsal and basiventral cartilages external to the notochordal sheath.

The fin spine is smooth along its anterior edge and extends well beyond the distal edge of the first dorsal fin (FS; fig. 9). The second dorsal fin (D2) is roughly triangular in shape, tall anteriorly, and tapers posteriorly. The tail is heterocercal with an anal fin that is sup- ported by cartilage.

RHINOCHIMAERIDAE: In all rhinochimaer- ids the notochord is surrounded by numerous calcified rings that are closely packed togeth- er. In an adult Rhinochimaera there are ap- proximately 11 rings per centimeter. In adults these calcified rings completely ensheath the notochord, which extends to the tip of the tail. Juveniles usually lack these calcified rings and the number of rings seems to increase with age.

The fin spine in Rhinochimaera is shorter than the first dorsal fin (FS; fig. 12A). Har- riotta and Neoharriotta have a fin spine that is slightly longer than the first dorsal fin (FS; fig. 12B). The elongate second dorsal fin in rhinochimaerids has an even height along its length (D2). Rhinochimaerids all possess a leptocercal tail. Neoharriotta is the only neo- chimaeroid to possess an anal fin that is sup- ported by cartilage.

CHIMAERIDAE: Calcified rings are present in the notochordal sheath of all adult chi- maerids. I have counted 12 calcified rings per centimeter in the notochordal sheath of an adult Hydrolagus colliei.

DIDIER: CHIMAEROID FISHES 41

The fin spine in chimaerids may be smooth anteriorly or with a prominent keel along the anterior edge (FS; fig. 12C). The fin spine may be slightly longer or shorter than the first dor- sal fin. The second dorsal fin has an even height along its length. A leptocercal tail is present in all chimaerids. The anal fin web is separated from the subcaudal fin in Chimae- ra by a notch, but there is no cartilaginous support for this anal fin, and by definition I do not consider it a true anal fin.

PAIRED FINS AND GIRDLES

The pectoral girdle lies just posterior to the neurocranium (e.g., figs. 18A, 21A,22A). This large element is fused at the symphysis and can be divided into two regions based on the location of the pectoral fin articulation at the glenoid fossa (GL). Ventral to the glenoid fos- sa is the coracoid region, which forms the ventralmost portion of the pectoral girdle (COR). The anterior surface has two large depressions where the m. coracobranchialis complex originates. Passing through the cor- acoid is a canal for the passage of blood ves- sels and nerves to the pectoral fin. This canal originates on the medial face of the scapular process, just above the pectoral fin, and exits ventrally at the anterior end of the coracoid (CF). Spanning the pectoral girdle is a dense sheet of connective tissue that runs trans- versely along the dorsal edge of the coracoid region to form the posterior pericardial wall.

The elongate scapular process (SCA) is an- gled anteriorly at the pectoral fin articulation where it joins the coracoid and extends dor- sally to lie lateral to the base of the synarcual. At its dorsal end the scapular process be- comes thin and flat with an elongate carti- laginous filament extending posteriorly from the tip. It is embedded in a complex mus- culature except at its dorsal end, which is exposed above the epaxial muscles.

The pectoral fins of Callorhinchus milti and Harriotta raleighana are shown in figure 24. They articulate at the glenoid fossa on the posterior edge of the coracoid by a small propterygial element. This articulation de- marcates the coracoid from the scapular pro- cess. Articulating at the anterior edge of the propterygial element is a single cartilaginous element, which I interpret to be formed by

42 AMERICAN MUSEUM NOVITATES

PXR

NO. 3119

Fig. 24. Skeletal anatomy of the paired fins of Callorhinchus milii and Harriotta raleighana. A, Left pectoral fin of Callorhinchus milii (CM77). B, Left pectoral fin of Harriotta raleighana (AMNH 96946). The pectoral fins are dibasal, possessing only a propterygial (PT) and metapterygial (MPT) element. A single anterior radial element (AR) articulates with the propterygium, and the remaining radials, which are usually segmented into proximal (PXR) and middle (MR) radials, articulate with the metapterygium. The tiny distal radials (DR) support the ceratotrichia (not shown) of the fin web. C, Left pelvic fin of Callorhinchus milii. D) Left pelvic fin of Harriotta raleighana. The single large basipterygial element (BT) is characterized by a lateral extension, the basipterygial process (BTP). All of the radials articulate with the basipterygium and may be segmented into proximal and middle radials or remain unsegmented. The distal radials support the certatotrichia (not shown) of the pelvic fin. Scale bars = 2 cm.

the fusion of the anterior radials of the pec- toral fin (AR). Future developmental studies will confirm the precise formation and ho- mology of this anterior radial element in the pectoral fin. The remaining radials articulate with the metapterygial element.

The propterygium and metapterygium support the radials of the pectoral fin; thus it is defined as dibasal. A tribasal fin in which there are three basal elements (propterygium, mesopterygium, and metapterygium) that bear radials is a derived feature among elas- mobranchs (Rosen et al., 1981). In juveniles

all of the radials are singly jointed, but in adults there is a tendency for the proximal elements (PXR) to fuse to the longer middle radials (MR). The long middle radials de- velop central regions of prismatic calcifica- tion. At the medial edge of the pectoral fin there are several small uncalcified cartilages that are not in parallel with the middle ra- dials. The number of radials varies; for ex- ample, in three juveniles there were 22 mid- dle radials, whereas adults had 21-27. Articulating at the distal ends of the middle radials is a series of small triangular carti-

1995

lages. These small distal radial cartilages are variable in number, but there are roughly two per radial element. One lies ventral and the other lies dorsal to each radial, forming a ridge of small distal radials that support the ceratotrichia of the large, winglike pectoral fins.

The pelvic girdle lies embedded within the ventral body musculature and forms no ar- ticulation with any other skeletal elements. In Callorhinchus the pelvic girdle is separated at the symphysis and the two halves are con- nected by a broad ligament. In all other chi- maeroids the two halves of the pectoral girdle are tightly joined at the symphysis. Laterally, the pelvic girdle tapers to an elongate process, which extends dorsally and posteriorly into the muscle of the body wall. On either side of the symphysis the pelvic girdle is gently rounded anteriorly and is penetrated by one or two foramina for the passage of nerves and blood vessels to the pelvic fin. In Callorhin- chus these rounded anterior edges of the pel- vic girdle are greatly enlarged and each has a central vacuity that is closed off by a con- nective tissue membrane. This enlarged an- terior process of the pelvic girdle in callo- rhynchids supports the large prepelvic tenacula in males.

The pelvic fins lie along the ventral surface of the body and are almost identical in all taxa studied (fig. 24). The single basiptery- gium (BT) of the pelvic fin articulates at a raised acetabulum located on the posterior edge of the pelvic girdle. The basipterygium of the pelvic fin is a flat ovoid cartilage. The first few radials fuse to the basipterygium and form an elongate basipterygial process, which extends along the anterior margin of the fin (BTP). Articulating with the basal cartilage is a series of long radial elements (MR), which lie parallel to the basipterygial process, and like the pectoral fin radials, these radials have a region of calcification in the center. Pelvic radials were counted in at least one member of each genus and usually 12-15 radials were found in all chimaeroids examined. In some specimens as few as 8 or 10 have been count- ed, but it is possible that some of the radial elements were lost in preparation. Like the pectoral fins, the pelvic fins possess a distal series of small triangular cartilages that sup- port the ceratotrichia.

DIDIER: CHIMAEROID FISHES 43

SECONDARY SEXUAL CHARACTERS

Pelvic claspers specialized for internal fer- tilization are characteristic of all Chondrich- thyes, but prepelvic and frontal tenacula are unique to Chimaeriformes. Only in males are these secondary sexual characteristics fully developed at the onset of sexual maturity; however, cartilage rudiments of frontal te- nacula have been observed in females of all taxa examined and some even possess tiny cartilage rudiments of pelvic claspers, partic- ularly Callorhinchus. The claspers and their musculature in Callorhinchus and Chimaera were described by Jungersen (1899). Leigh- Sharpe (1922, 1926) described the morphol- ogy of these secondary sexual structures in detail for Callorhinchus, Rhinochimaera, Chimaera, and Hydrolagus. The morphology of the urogenital system has been described for Chimaera monstrosa (Burlend, 1910) and more recently for Hydrolagus colliei (Stanley, 1963) and will not be discussed here.

In addition to the pelvic claspers, all male chimaeroids have paired organs extending anteriorly from the pelvic girdle known as prepelvic tenacula. The term tenaculum means “‘a holder,” and the prepelvic tenacula derive their name from the interpretation that they are specialized organs to hold females during copulation, although this has never been observed. The prepelvic tenacula are formed by a single cartilaginous element that articulates at the anterior edge of the pelvic girdle. These tenacula bear denticles in all chimaeroids and are housed within fleshy prepelvic pouches that are found on the ven- tral side of the trunk just anterior to the pelvic girdle in all males.

In sexually mature males a single median frontal tenaculum is found atop the head. The tenaculum is composed of dense fibrocarti- lage with a cluster of sharp denticles at its tip (Patterson, 1965; Raikow and Swierczewski, 1975). The anterior end of the frontal tenac- ulum rests in a fleshy pouch that may be lined with placoid denticles (e.g., Hydrolagus no- vaezealandiae). The morphology of the fron- tal tenaculum is variable and examples of the frontal tenacula of chimaeroids are illustrated in figure 25. The frontal tenaculum is at- tached to the neurocranium by two ligaments that run from the supraorbital crest to the

44 AMERICAN MUSEUM NOVITATES

Vie A B

G2) E "IN OF

Fig. 25. Comparative anatomy of the frontal tenacula. The frontal tenaculum is shown for A, Callorhinchus milii, B, Rhinochimaera pacifica, C, Harriotta raleighana, D, Chimaera sp. C, E, Hydrolagus colliei, and F, Hydrolagus sp. B. The tenacula exhibit a diverse morphology across taxa. Although the exact number of tenacular denticles is not shown, the figures are intended to represent the approximate size and shape of the denticles for each species. Scale bar = 2 cm.

concave tenacular base (not illustrated). The frontal tenaculum can be raised and lowered and it shares a complex musculature with the labial cartilages (Raikow and Swierczewski, 1975).

The development of this structure has nev- er been described in detail, but it is interesting to note that the frontal tenaculum is not fully developed until late in life, during the tran- sition from juveniles to sexually mature adults. In juvenile males a small white streak along the dorsal midline of the head, anterior to the orbits, marks the location where the frontal tenaculum will develop. Females do not have this structure externally; however, mature females possess a tiny fibrocartilage rudiment underneath the skin at the site where the frontal tenaculum develops in males.

The exact function of the frontal tenaculum

NO. 3119

is unknown; however, because it is sexually dimorphic, it has been assumed to be im- portant for courtship or copulation. Dean (1906: 24) examined the pattern of scratches on the dorsal surface of females and con- cluded that the frontal tenaculum was used to grasp the female during copulation.

CALLORHYNCHIDAE: In Callorhinchus the pelvic claspers are in the form of smooth hol- low tubes formed by a thin sheet of cartilage wrapped around itself in scroll-like fashion (fig. 26A). The pelvic clasper is composed of two cartilages: the anterior clasper cartilage (ACC) articulates with the basal cartilage of the pelvic fin at its posteromedial edge, and the posterior clasper cartilage (PCC) forms the more distal tubular portion of the clasper. At its articulation with the posterior clasper cartilage the anterior clasper cartilage has a small fold of cartilage that forms an opening to the interior of the posterior clasper carti- lage. The clasper tapers distally to a rounded tip with an external opening that is partially covered by a fleshy tissue flap.

The prepelvic tenacula of Callorhinchus are extremely complex, unlike those found in any other chimaeroids (fig. 26A). The prepelvic tenacula articulate at the anterior edge of the pelvic girdle, which is greatly enlarged in Cal- lorhinchus. The cartilaginous skeleton of the prepelvic tenaculum (PPT) has two distinct parts. The larger part is a flat, wide cartilage blade covered by tiny denticles with scal- loped edges. Lying deep to this flat cartilage is a rolled tube of cartilage, open at its pos- terior end. Attached to the flat cartilaginous blade is a ruffled flap of dense tissue. Within the prepelvic pouch is an elongate sac and associated gland (not shown). This glandular component is only found in callorhinchids.

At rest, this entire complex is housed with- in the prepelvic pouch. It can be extended anteriorly out of the pouch to lie flat along the ventral body surface. It does not seem possible that the prepelvic tenacula in Cal- lorhinchus could be used for grasping because the denticles are extremely tiny. The presence of a glandular component suggests that the prepelvic tenacula of Callorhinchus play an accessory role in fertilization (Leigh-Sharpe, 1922, 1926).

In female Callorhinchus there are small prepelvic slits that house tiny cartilaginous

1995 DIDIER: CHIMAEROID FISHES 45

Fig. 26. Skeletal anatomy of the pelvic claspers and prepelvic tenacula of Callorhinchus milii, Rhin- ochimaera pacifica, and Hydrolagus novaezealandiae. A, Ventral view of the left side of the pelvic girdle (PVG) of Callorhinchus milii showing the pelvic clasper and prepelvic tenaculum in their normal ana- tomical relationship as they would appear at rest. The pelvic clasper, consisting of an anterior clasper cartilage (ACC) and posterior clasper cartilage (PCC), articulates with the medial side of the basipterygium (BT) of the pelvic fin. At the anterior edge of the pelvic girdle the prepelvic tenaculum (PPT) overlies a large vacuity (VA) within which lies a sheet of connective tissue (not shown). B, Left half of the pelvic girdle of Rhinochimaera pacifica shown in dorsal view. The pelvic clasper, consisting of an anterior clasper cartilage and posterior clasper cartilage, is shown as it would appear externally, extending pos- teriorly from the basipterygium of the pelvic fin. The prepelvic tenaculum is armed with denticles and is shown extended anteriorly as it would be when extended from the prepelvic pouch. C, Ventral view of the left side of the pelvic girdle of Hydrolagus novaezealandiae. The pelvic clasper, with anterior clasper cartilage and posterior clasper cartilage, and the prepelvic tenaculum are shown in their normal anatomical relationships as they would appear at rest. Scale bars = 2 cm.

46 AMERICAN MUSEUM NOVITATES

rudiments of prepelvic tenacula. This is the only occurrence of this secondary sexual character among female chimaeroids.

The frontal tenaculum, present only in males, is flat with a broad denticulate tip in Callorhinchus. The denticles are small.

RHINOCHIMAERIDAE: The pelvic claspers and prepelvic tenacula of Harriotta raleigh- ana are shown in figure 26B as representative of Rhinochimaeridae. All rhinochimaerids possess simple, rodlike pelvic claspers con- sisting of two elements. The anterior clasper cartilage articulates with the pelvic fin at the medial edge of the basal cartilage and the posterior clasper cartilage is a slender, almost solid rod. At its tip is a fleshy bulb, composed of a loose tissue with cavernous vascular spaces, which is covered with placoid den- ticles.

The prepelvic tenacula of rhinochimaerids are simple blades of cartilage, narrow at the proximal end and expanded distally (fig. 26B). At the distal end is a notch that may be a deep groove or a shallow depression. Along the medial edge is a single row of placoid denticles. The prepelvic tenacula vary among the genera and species of Rhinochimaeridae in the shape of the distal end and the number of denticles, but the range of variation has not been thoroughly examined.

The frontal tenaculum in Rhinochimaera is robust with a broad base and flat dorsal surface, whereas Harriotta has a more slender frontal tenaculum with a distinct dorsal curve (fig. 25). In all species the bulbous tip has numerous recurved denticles on its ventral surface.

CHIMAERIDAE: The greatest variation and complexity of pelvic claspers among chi- maeroids is found within the family Chi- maeridae. The anterior clasper cartilage (ACC; fig. 26C) articulates with the basal car- tilage of the pelvic fin at its posteromedial edge. This cartilage is forked at its anterior end and a thin flap of cartilage forms a chan- nel along the medial edge. Articulating with the anterior clasper cartilage is the posterior clasper cartilage (PCC), which divides into two or three distal rods. The distribution of this character does not seem to be of system- atic relevance at the generic level. All species of Chimaera have trifid pelvic claspers, as do

NO. 3119

some species of Hydrolagus (e.g., Hydrolagus affinis and Hydrolagus pallidus).

In species with bifurcate claspers (e.g., Hy- drolagus colliei) the two arms of the posterior clasper cartilage are enveloped by fleshy tis- sue along most of their length. Externally this fleshy tissue is covered with a shagreen of tiny denticles and internally it consists of cavern- ous vascular spaces. Each arm of the clasper has a central canal that is a continuation of the channel formed by the anterior clasper cartilage. In species that possess trifid pelvic claspers, the medial arm is a smooth straight rod with only a small patch of denticulate tissue at the tip (e.g., Chimaera sp. C). This third arm is equal in length to or sometimes slightly longer than the two lateral fleshy rods.

There are large paired swellings on the ven- tral midline posterior to the pelvic fins. This postanal pad may be found only in females (e.g., Hydrolagus colliei Dean, 1906, and Hy- drolagus sp. B) or in both males and females (e.g., Hydrolagus novaezealandiae). The postanal pads are composed primarily of dense connective tissue and their function is unknown (McCutcheon, 1980).

The prepelvic tenacula of chimaerids are simple, bladelike cartilages with a distinct notch at the expanded distal end (fig. 26C). The lateral edge is armed with a single row of four to six recurved denticles. The frontal tenaculum is slender and deeply curved dor- sally with numerous recurved denticles at the tip. The frontal tenaculum varies in shape among species of chimaerids (fig. 25).

TOOTH PLATES

Hypermineralized tooth plates are one of the most important synapomorphies of Hol- ocephali, and the presence of one pair of tooth plates in the lower jaw and two pairs in the upper jaw is a synapomorphy of Chimaeri- formes. The tooth plates are assumed to be related to durophagy, yet gut content analysis shows that living chimaeroids include a va- riety of soft- and hard-bodied prey in their diet (Graham, 1939; Gorman, 1963; Didier, personal obs.; Stehmann, personal com- mun.).

Because tooth plates are one of the few mineralized tissues in holocephalans, they

1995

also fossilize well and most fossil holoce- phalans are known only from tooth plates. As a result, a large effort has been dedicated solely to their description and analysis (Eger- ton, 1843; Newberry and Worthen, 1870; Saint John and Worthen, 1883; Zittel, 1887, 1932; Woodward, 1892, 1921; Schauinsland, 1903; Dean, 1906; Hussakof, 1912; Chap- man, 1918; Nielsen, 1932; Bargmann, 1933, 1941; Moy-Thomas, 1936a, b, 1939; Ber- man, 1967; Peyer, 1968; Ward, 1973; Ward and McNamara, 1977; Lund, 1977, 1986a, b; Orvig, 1980, 1985; Zangerl, 1981; Kemp, 1984; Duffin, 1984; Ward and Duffin, 1989; Ward and Grande, 1991).

All chimaeroids have six tooth plates: two pairs of tooth plates in the upper jaw and one pair in the lower jaw (refer to fig. 27). In the upper jaw is an anterior pair of small vo- merine tooth plates (VTP) followed by a pair of palatine tooth plates (PTP). In the lower jaw are the two large mandibular tooth plates (MTP). The oral surfaces (or working sur- faces, WS) of the tooth plates are easily iden- tified by their rough texture, which appears to be riddled with tiny pores. The oral surface of the tooth plates is characterized by regions of hypermineralized tissue commonly called tritors. Distal to the working surface a cov- ering of oral epithelium prevents the surface tissue from wearing away and the tooth plate is smooth and shiny. The margin of wear (MW) demarcates the worn and unworn por- tions of the oral surface of the tooth plate. The aboral surface is characterized by the presence of a distinct ridge, the descending lamina (DL; Patterson, 1992). This descend- ing lamina is well developed in all the tooth plates of Callorhinchus, but has been ob- served to be variously developed in only the mandibular and palatine tooth plates of other chimaeroids and may be absent or extremely reduced in Rhinochimaera (personal obs.).

Recent lungfishes also possess six hyper- mineralized tooth plates as well as an autos- tylic jaw suspension and have often been compared to holocephalans (see Jarvik [1980] for a recent review); however, on the basis of characters other than the jaws and tooth plates the lungfishes are clearly sarcopterygians, a well-established monophyletic lineage that does not include holocephalans (e.g., Rosen

DIDIER: CHIMAEROID FISHES 47

et al., 1981; Lauder and Liem, 1983). Al- though the tooth plates of lungfishes and hol- ocephalans possess the same tissues by def- inition (Kemp, 1984), they develop differently. The tooth plates in lungfishes de- velop by fusion of several individual tooth primordia (Kemp, 1977, 1984, Neocerato- dus; Bemis, 1984, Protopterus), whereas each of the six tooth plates of chimaeroids has been observed to develop from an individual tooth plate primordium (Schauinsland, 1903; Kemp, 1984).

Recently, the tooth plates of adult Callor- hinchus milii and Chimaera monstrosa have been observed to have a compound structure; therefore, their development may be more complex than reported previously (R. Zan- gerl, personal commun.). A new analysis of tooth plate development in Callorhinchus milii, with a comparative microanatomical study of tooth plates in all other genera, in- dicates that living chimaeroids possess tooth plates of a compound structure that repre- sents the fusion of adjacent tooth-forming territories. This type of tooth plate devel- opment is reminiscent of what would be ex- pected by the fusion and reduction of two elements of a typical chondrichthyan tooth family (Didier et al., 1994).

I have confirmed these basic points in a preliminary histological study of embryos of Callorhinchus milii. The primordia of the tooth plates are first evident as an infolding of the dental epithelium, which forms a cap over the entire tooth plate field in Callorhin- chus milii embryos of 70-80 mm TL (DE; fig. 28A). In later stages this DE extends ven- trally and a second, aboral tooth-forming ter- ritory begins to form. This aboral territory is related to the DL, which demarcates the re- gion of aboral tooth plate territory growth (Didier, 1993; Didier et al., 1994, in press). The development of the tooth plates occurs rapidly and in embryos of 80-85 mm TL mineralized tissues are beginning to form and the tooth plates are already developing their characteristic structure (fig. 28B). Hypermi- neralized tissue begins to form in distinct patches after the initial framework of trabec- ular dentine has been established (Kemp, 1984). I have confirmed this early formation of hypermineralized tissue in embryos of 103-

48 AMERICAN MUSEUM NOVITATES NO. 3119

Fig. 27. Tooth plates of Callorhinchus milii (AMNH 96954). A, Ventral view ofa dried neurocranium showing the palatine (PTP) and vomerine (VTP) tooth plates in the upper jaw of Callorhinchus milii. The large hypermineralized pad (HP) is evident in the center of the oral surface (WS) of the palatine

1995

Fig. 28. Development of tooth plates of Cal- lorhinchus milii. A, Transverse section through the head of an embryo (CM40, 79 mm TL) show- ing the early development of the mandibular tooth plates (TP), which begin to form as mesenchyme cells, aggregate below the infolded dental epithe-

—_—

DIDIER: CHIMAEROID FISHES 49

109 mm TL. The trabecular dentine matrix continues to develop around the areas of hy- permineralization until these two tissue types appear in direct continuity (fig. 29). Future research on the ontogeny of tooth plates of chimaeroids should focus on the develop- ment of hypermineralized tissues in order to clarify the morphological and histological distinctions.

In all chimaeroids trabecular dentine com- prises the bulk of the tooth plate and forms a mineralized network of tissue penetrated by blood vessels (TD; fig. 29). At the oral surface of the tooth plate the trabecular den- tine forms a solid layer and the vascular ca- nals become aligned in a parallel arrange- ment. This surface layer of trabecular dentine in which the vascular canals are parallel to each other is also known in the literature as tubular dentine (Denison, 1974), a mislead- ing identification that is no longer accepted. The trabecular dentine matrix is less dense near the base of the tooth plate where large vascular spaces form the pulp cavity (PC).

In direct contact with the trabecular den- tine is the hypermineralized tissue (HT; = tritors), which is penetrated by dentine tu- bules. The hypermineralized tissue occurs in two forms that appear on the oral surface of the tooth plates as either a single large tritor or a series of small beadlike tritors, the

—

lium (DE). The lower jaw (M), mesencephalon (MES), notochord (N), and otic capsule (OTC) are shown as morphological landmarks. Scale bar = 1 mm. B, Transverse section of the head of an older embryo (CM52, 89 mm TL) showing the later development of the mandibular tooth plates. Underneath the oral epithelium (EPI) lies a layer of mineralized tissue that has been variously (and perhaps incorrectly) named “enameloid” (EN) or “‘vitrodentine.” Internally, the tooth plate consists of trabecular dentine (TD) and vascular canals (VC). There are no hypermineralized tissues at this early stage. Scale bar = 0.5 mm.

tooth plate. The margin of wear (MW) demarcates the worn, oral surface of the tooth plate from the more distal, unworn portion. B, Lower jaw of Callorhinchus milii showing the mandibular tooth plates (MTP) with a large hypermineralized pad (HP). Scale bar = 1 cm.

50 AMERICAN MUSEUM NOVITATES

NO. 3119

eo Pas:

Fig. 29. Microscopic anatomy of the tooth plates of Callorhinchus milii. A sagittal section of the mandibular tooth plate illustrates the tissues of the tooth plate and their relationship to each other. Trabecular dentine (TD) is a dark-staining tissue forming a thick layer at the oral surface (WS). The hypermineralized tissue (HT) stains lightly and is in direct contact with, and surrounded by, trabecular dentine. Discrete regions of hypermineralized tissue appear to be fusing to form a large, hypermineralized pad. Scale bar = 0.5 mm.

‘“‘pearlstrings” of Bargmann (1933). Orvig has defined these two types of tritors as vascular pleromin and compact pleromin, respective- ly. Vascular pleromin is distinguished by the enclosure of nearly parallel vascular canals

whereas compact pleromin lacks vascular ca- —

nals and is formed into columns (@rvig, 1985). The term pleromin has sparked con- troversy because by definition it implies a developmental mechanism whereby the tis- sue develops to fill in spaces and is deposited by a population of cells that Orvig has iden- tified as pleromoblasts. In an effort to clarify the nature of the hypermineralized tissue it- self, the term orthotrabeculine has recently been proposed by Zanger] et al. (1993). I refer to Orvig’s compact pleromin as hyperminer- alized rods (HD) and vascular pleromin as hypermineralized pads (HP), respectively. I have defined these as hypermineralized rods and hypermineralized pads in order to make the morphological distinction clear while making no assumptions about the histology of these tissues.

There are potential problems with some tooth plate characters. It appears that there is a developmental continuity between the

two forms of hypermineralized tissue. This continuity of hypermineralized tissues was first suggested by Garman (1904). Compar- ative microscopic anatomical study of the mandibular tooth plates of Callorhinchus, Harriotta, Chimaera, and Hydrolagus shows that the distinction between hypermineral- ized rods and hypermineralized pads is blurred and developmentally they seem to be continuous (Didier, 1991). These results are only preliminary, but the implication may be of systematic importance because the pattern of hypermineralized tissue on the surface of the tooth plates is used to identify species, particularly fossil taxa (e.g., Ward, 1973, Ward and McNamara, 1977; Duffin, 1984).

Unlike the teeth in other chondrichthyan fishes, which are regularly shed and replaced, the tooth plates of holocephalans are never replaced. The tooth plates have an active growing region at the base (GB) and under- neath the descending lamina. As the tooth plate is abraded, the trabecular dentine wears away, leaving a relief of hypermineralized tis- sue. The pattern of hypermineralized tissue on the oral surface of the tooth plates may not be a good character because there is no

1995

way to assess the affect of wear on shaping the hypermineralized tissues.

CALLORHYNCHIDAE: The tooth plates of Callorhinchus are robust, crushing plates (fig. 27A,B). On the oral surface of the tooth plates the hypermineralized tissue is in the form of large tritoral pads. A well-developed de- scending lamina is present on the aboral sur- face of all tooth plates. The vomerine tooth plates are firmly attached to the ethmoid re- gion of the neurocranium and, except for the tips, are obscured by the palatine tooth plates. There is no obvious hypermineralized tissue exposed on the surface, but histological sec- tions show evidence of a tiny hyperminer- alized pad in the center of the tooth plate.

The triangular palatine tooth plates cover the roof of the mouth. In the center of each tooth plate is a bifurcate hypermineralized pad (HP).

In the lower jaw the mandibular tooth plates are tightly bound to Meckel’s cartilage by a dense connective tissue pad. The labial edge of these tooth plates gently curves medially, with the curve continuing around to the an- terior tip of the tooth plate to a nearly straight symphysial edge where left and right tooth plates meet. In the center of each mandibular tooth plate is a large hypermineralized pad.

RHINOCHIMAERIDAE: All six tooth plates of Rhinochimaera are triangular, very thin, and bladelike with smooth surfaces (fig. 30). The lateral edges of these tooth plates are darkly pigmented, fading to a light gray-brown near the center. Externally the tooth plates appear as a sharp beak where the anterior tips of the vomerine and mandibular tooth plates meet. The only visible sign of wear on these tooth plates is at the edges. There is no hypermi- neralized tissue exposed at the surface of the tooth plates of Rhinochimaera, and in trans- verse sections through a mandibular tooth plate of an adult Rhinochimaera pacifica there is no evidence of hypermineralized tissue (fig. 31). At the surface is a layer of dentine and the interior of the tooth plate consists of a very loose trabecular dentine with numerous large vascular spaces (PC). A descending lam- ina has not been observed on the aboral sur- face of any of the tooth plates.

The tooth plates of Harriotta and Neohar- riotta have hypermineralized tritors in the form of pads and rods (fig. 32). Although these

DIDIER: CHIMAEROID FISHES 31

Fig. 30. Tooth plates of Rhinochimaera paci- fica. The three pairs of tooth plates are oriented as they would be in a wide-open mouth with vo- merine (VTP) and palatine (PTP) tooth plates at the top of the figure and mandibular tooth plates (MTP) facing downward. The margin of wear (MW), demarcating the worn oral surface (WS) from the unworn portions of the tooth plate, is difficult to distinguish because of the dark pig- mentation of these tooth plates. Scale bar = 1 cm.

tooth plates are robust, the edges are blade- like, forming a sharp cutting edge. On the aboral surface of the palatine and mandibular tooth plates, at the symphysial edge, is a de- scending lamina. The descending lamina is reduced in these species and is especially small in the mandibular tooth plates.

In both Harriotta and Neocharriotta the vo- merine tooth plates are ventrally directed, in- cisiform blades. There are five to seven hy-

52 AMERICAN MUSEUM NOVITATES

NO. 3119

Fig. 31.

Microscopic anatomy of the tooth plates of Rhinochimaera pacifica (AMNH 96940). This

transverse section of the mandibular tooth plate of an adult shows the loose network of trabecular dentine (TD) and pulp (PC) in the interior of the tooth plates. No hypermineralized tissue has been observed in the tooth plates of Rhinochimaera. Scale bar = 0.5 mm.

permineralized rods that appear as a row of beadlike dots on the surface of the tooth plate.

The palatine tooth plates of Harriotta and Neoharriotta are triangular in shape. In the center of each palatine tooth plate is a large hypermineralized pad that extends almost to the labial edge of the tooth plate. Along the labial edge of the palatine tooth plate is a series of hypermineralized rods.

The mandibular tooth plates of Harriotta and Neoharriotta have a raised hyperminer- alized pad located just lateral of the center. A series of hypermineralized rods is visible along the labial edge of the tooth plate (HR).

CHIMAERIDAE: The tooth plates of chi- maerids are relatively thin and the sharp an- terior edges form nipping blades (fig. 33). A small descending lamina is present along the symphysial edge of the palatine and mandib- ular tooth plates, but no descending lamina has been observed in the vomerine tooth plates. The pigmentation of these tooth plates is species specific and ranges from a dark brown or black in Hydrolagus sp. B to white in Hydrolagus colliei and Hydrolagus novae- zealandiae.

The vomerine tooth plates of Hydrolagus are similar to those of Harriotta and Neo- harriotta. These incisorlike blades are di-

rected ventrally and meet the mesial edge of the mandibular tooth plates near the sym- physis. There are five hypermineralized rods visible at the cutting edge of the vomerine tooth plates. On the posterior face of the oral surface of the vomerine tooth plates is a series of transverse ridges (TR). Although the vo- merine tooth plates of chimaerids are similar in external morphology to those of Harriotta and Neoharriotta, these transverse ridges are found only in the vomerine tooth plates of chimaerids.

The palatine tooth plates of Hydrolagus are roughly triangular with a prominent lateral flange (LF). As in all other chimaeroids these tooth plates lie flat on the roof of the mouth directly behind the vomerine tooth plates. At the mesial edge is a prominent hyperminer- alized rod and extending along the cutting edge of the tooth plate are numerous smaller hypermineralized rods, usually about 10.

The mandibular tooth plates of Hydrola- gus and Chimaera have a series of hyper- mineralized rods along the cutting edge of the tooth plate. There is no obvious hypermi- neralized tritor pad although some species, especially Chimaera monstrosa, appear to have a small, centrally located pad at the margin of wear, and microanatomical obser-

1995

Fig. 32. Tooth plates of Harriotta raleighana. The tooth plates are oriented as they would be in the mouth if it were open wide. Exposed on the oral surfaces of these tooth plates are tritors in the form of hypermineralized pads (HP) and rods (HD). At the distal ends of the mandibular (MTP) and vomerine (VTP) tooth plates a region of growth (GB) is visible. Scale bar = 1 cm.

vations confirm the presence of a small hy- permineralized tritor pad in the mandibular tooth plates of both Hydrolagus and Chi- maera.

MUSCULATURE

The comparative musculature of all extant genera of chimaeroid fishes is described be- low, including descriptions of the muscula- ture of the head, gill arches, pectoral girdle, and trunk. Although soft characters cannot

DIDIER: CHIMAEROID FISHES 53

Fig. 33. Tooth plates of Hydrolagus novae- zealandiae. The tooth plates are oriented as they would be in the mouth if it were open wide. The hypermineralized rods (HD) are difficult to distin- guish in these white tooth plates, but can be seen at the edges as prominent knobs. The palatine tooth plates (PTP) have a prominent lateral flange (LF; broken in the right tooth plate). Transverse ridges (TR) are characteristic of the vomerine tooth plates (VTP) of chimaerids. Scale bar = 1 cm.

help resolve relationships between fossil and Recent taxa, muscle characters can be used in outgroup comparison with Recent elas- mobranchs.

Nomenclature for the muscles follows Vet- ter (1878) on the basis of primacy. In cases where developmental evidence refutes Vet- ter’s (1878) work, the nomenclature of Edge- worth (1935) takes precedence. Shann (1919) provided the most complete description of the muscles of the pectoral girdle for chi- maeroids. I confirm his descriptions for Chi-

54 AMERICAN MUSEUM NOVITATES

y laoa Ay

sup re WEE

NO. 3119

Fig. 34. Superficial muscles of the head and pectoral region of Callorhinchus milii. The skin, con- nective tissue, and ampullary canals have been removed to reveal the underlying muscles. See text for a detailed description of the musculature. Scale bar = 2 cm.

maera and Callorhinchus and find them ap- propriate; therefore, I follow his terminology.

Table 8 is a complete list of all the muscles described in this study with a list of the syn- onymous names and remarks regarding iden- tification or description of the muscle. The descriptions of the musculature are based on detailed study of Callorhinchus milii and rep- resentatives of the remaining five extant gen- era of chimaeroids that were dissected (see table 2). Significant differences among genera are noted in the remarks and where no re- marks are made, it is understood that I have found the muscles to be identical in all genera examined.

MANDIBULAR GROUP

The following description is of muscles de- rived from the mandibular arch with Vth cra- nial nerve innervation.

M. superficialis, sup (fig. 34).

Origin: From connective tissue of the pre- orbital fascia.

Insertion (not shown): Into connective tis- sue of the rostral flap.

Remarks: Among chimaeroids this muscle is only found in Callorhinchus. It is a very

thin, almost indistinguishable sheet of mus- cle superficial to the M. adductor mandibulae anterior. Many of its fibers insert into the connective tissue sheet covering the antor- bital portion of the head while a tiny bundle of muscle extends into the rostrum and in- serts into the connective tissue of the flap of the rostrum. This muscle is probably derived from the M. adductor mandibulae anterior and is innervated by the Vth cranial nerve.

M. levator anguli oris anterior, laoa (figs. 34, 35, 38, 41, 43)

Origin: Via a connective tissue band from the antorbital crest; in males at the articu- lation of the frontal tenaculum.

Insertion: After passing under the labial ligament this muscle inserts on the ventro- medial face of the superior maxillary cartilage with some fibers to the posterior end of this cartilage.

Remarks: In Callorhinchus a small bundle of anterior fibers inserts on the lateral rostral rod, and in Rhinochimaera and Harriotta a small anterior muscle bundle inserts on the inner angle of the prelabial cartilage. Both Luther (1909) and Kesteven (1933) distin- guished this as a separate muscle in Callor- hinchus. Raikow and Swierczewski (1975) in-

1995

———

DIDIER: CHIMAEROID FISHES 55

Fig. 35. Deep muscles of the head and pectoral region of Callorhinchus milii. Superficial muscles and opercular flap have been removed to reveal deep muscles of the jaws and labial cartilages. See text for a detailed description of the musculature. Scale bar = 2 cm.

dicated some anterior fibers to the upper lip in female Chimaera and I follow Kesteven (1933) and Raikow and Swierczewski (1975) in calling this portion of the M. levator anguli oris anterior the M. levator anguli oris ante- rior pars rostralis because of its rostral inser-

tion in Callorhinchus and its proximity to the rostrum in rhinochimaerids. In males this portion of the M. levator anguli oris anterior runs to the base of the frontal tenaculum with some fibers inserting onto the tenaculum by dense connective tissue fibers.

rdp

Fig. 36. Deep muscles of the head and pectoral region of Callorhinchus milii. Labial cartilages and gill arches have been removed to expose deep muscles of the snout and branchial region. See text for a detailed description of the musculature. Scale bar = 2 cm.

56

AMERICAN MUSEUM NOVITATES

TABLE 8 Synonymy of Muscles of Chimaeroids

M. superficialis, sup This muscle is possibly part of Kesteven’s (1933) M. depressor rostri et veli, but I can find

no description of this thin muscle sheet in any previous works. Kesteven (1933: 465) indicated that the many layers he described for the deep ventral constrictors may be portions of the same muscle. I believe this to be true as I cannot easily separate out the layers he described.

M. levator anguli oris anterior pars rostralis, pr

M. levator cartilaginous prelabialis, Luther (1909) M. levator rostri, Kesteven (1933)

M._ levator anguli oris pars rostralis, Raikow and Swierczewski (1975)

According to Edgeworth (1935), all of the superficial muscles that insert on the labial cartilages and lips share a common origin and are distinguished as separate muscles only at their insertions. Only Luther (1909) and Kesteven (1933) distinguished the anterior part of the M. levator anguli oris (below) as a separate muscle.

M. levator anguli oris anterior, laoa M. levator anguli oris 1, Vetter (1878) M. levator anguli oris anterior, Luther (1909) M. levator labii superioris, Kesteven (1933)

M. levator anguli oris posterior, Edgeworth (1935) M. levator anguli oris pars rostralis, Raikow and Swierczewski (1975)

M. levator anguli oris posterior, laop M. levator anguli oris 2, Vetter (1878) M. levator anguli oris posterior, Luther (1909) M. levator labii inferioris, Kesteven (1933)

M. levator anguli oris posterior, Edgeworth (1935)

M. labialis anterior, la M. labialis anterior, Vetter (1878) M. labialis anterior, Luther (1909)

M. protractor labii superioris, Kesteven (1933) M. labialis, Edgeworth (1935)

M. intermandibularis I, im M. labialis posterior, Vetter (1878) M. labialis inferioris, Luther (1909)

M. protractor superior labii inferioris, Kesteven (1933) M. intermandibularis anterior, Edgeworth (1935)

M. intermandibularis II, im M. labialis posterior, Luther (1909) M. protractor inferior labii inferioris, Kesteven (1933)

M. intermandibularis posterior, Edgeworth (1935) This two-part muscle is well developed in Callorhinchus where it runs from the posterior

edge of the superior maxillary cartilage to the posterior two-thirds of the premandibular cartilage along its dorsal edge. Descriptions by Luther (1909) and Edgeworth (1935) as well as Vetter’s (1878) illustrations indicate that in Chimaera a muscle runs ventrally along the lower jaw and inserts on the minute premandibular cartilage. Vetter (1878) also illustrated a second tiny muscle from the premandibular

NO. 3119

a er eee ee

1995

DIDIER: CHIMAEROID FISHES

TABLE 8—(Continued )

cartilage to the lower jaw, but did not indicate that it is a separate muscle and I

assume that he interpreted this as a continuation of his M. labialis posterior. Edgeworth (1935: 40) claimed to follow Luther in his description of the muscles but in the text he reversed the names and muscles as they are shown by Luther (1909: 39). I agree with Vetter’s (1878) interpretation and believe these two muscles to be parts of the M. intermandibularis on the basis of a common innervation.

M. adductor mandibulae anterior, ama

M. adductor mandibulae, Vetter (1878)

M. preorbitalis, Luther (1909)

M. adductor mandibulae anterior head, Kesteven (1933)

M. levator mandibulae anterior, Edgeworth (1935)

Vetter (1878: 463) did not distinguish anterior and posterior portions of this muscle in his illustrations, but he described two portions of this muscle for Callorhinchus in his text. Edgeworth (1935: 255) synonomized this anterior portion with Vetter’s (1878) M. levator labii superior, but I find no mention of this muscle in Vetter’s work.

M. adductor mandibulae posterior, amp

M. adductor mandibulae, Vetter (1878)

M. adductor mandibulae, Luther (1909)

M. adductor mandibulae posterior head, Kesteven (1933) M. levator mandibulae posterior, Edgeworth (1935)

M. constrictor operculi dorsalis, cd

M. constrictor superficialis, Vetter (1878)

M. levator operculi, Kesteven (1933)

M. constrictor hyoideus, Edgeworth (1935)

Vetter (1878) illustrated this muscle as several parts in an attempt to homologize this muscle sheet with the dorsal and ventral constrictors of sharks. Vetter’s (1878) Cs5 appears to be the part of the constrictor muscle sheet that is continuous with the M. coracomandibularis and extends into the connective tissue sheet of the rostrum. Kesteven (1933) indicated that this muscle is innervated by both the Vth and VIIth cranial nerves. Edgeworth (1935: 97-98) described the development of the muscles extending into the connective tissue of the rostrum from the hyoid constrictor muscle sheet; therefore, it seems reasonable to accept a VIIth nerve innervation of these muscles.

M. constrictor operculi ventralis, cv

M. constrictor superficialis, Vetter (1878)

M. superficial ventral constrictor, Kesteven (1933)

M. constrictor hyoideus, Edgeworth (1935)

According to Kesteven (1933), this muscle is innervated by only the Vth nerve.

M. constrictor operculi dorsalis anterior, cda

M. depressor mandibulae superior, Kesteven (1933) This muscle is unique to Callorhinchus.

M. levator hyoideus, Ih

M. hyoideus superior, Vetter (1878) M. levator hyomandibulae, Edgeworth (1935)

57

58 AMERICAN MUSEUM NOVITATES NO. 3119

TABLE 8—(Continued )

M. interhyoideus, ih M. hyoideus inferior, Vetter (1878) M. geniohyoideus, Kesteven (1933)

M. interhyoideus, Edgeworth (1935)

Kesteven (1933) suggested that this muscle is a homologue to the M. intermandibularis or M. retractor hyoideus of elasmobranchs, and also indicated, as does Vetter (1878), that it is innervated by the VIIth nerve. According to Edgeworth (1935), both this muscle and the M. levator hyoideus originate from the hyoid constrictor muscle sheet.

Mm. constrictores branchiales, cb Mm. interbranchiales, Vetter (1878) Mm. constrictores branchiales, Edgeworth (1935) I interpret Kesteven (1933) to have described these muscles as part of the Mm. transversi ventrales series. Edgeworth (1935) indicated that elasmobranchs lack Mm. transversi ventrales. My observations confirm Edgeworth (1935).

Mm. adductores arcuum branchialium (not illustrated) Mm. adductores arcuum branchialium, Vetter (1878) Mm. adductores arcuum branchialium, Edgeworth (1935)

Mm. interbranchiales (not present) Edgeworth (1935) claimed that these muscles are not present in Holocephali. My interpretation of Kesteven’s (1933) description is that he indicated they are present. I have not found these muscles in my dissections.

M. cucullaris superficialis, cs M. trapezius superficialis, Vetter (1878) M. levator pectoralis, Shann (1919) M. cucullaris, Kesteven (1933)

M. cucullaris superficialis, Edgeworth (1935)

M. protractor dorsalis pectoralis, pdp M. protractor dorsalis pectoralis, Shann (1919)

M. cucullaris profundus, cp M. cucullaris profundus, Vetter (1878) M. levator arcuum branchialium posterior, Kesteven (1933) M. cucullaris profundus, Edgeworth (1935)

M. subspinalis (not illustrated) M. protractor arcuum branchialium, Vetter (1878) M. levator arcuum branchialium anterior, Kesteven (1933) M. subspinalis, Edgeworth (1935) Edgeworth also described a M. interpharyngobranchialis from pharyngobranchial 2 to pharyngobranchial 3, which I have not found.

M. coracomandibularis, cm M. coracomandibularis, Vetter (1878) M. coracomandibularis, Shann (1919) M. coracomandibularis, Kesteven (1933) M. geniocoracoideus, Edgeworth (1935) ee 2 a eee ee ee

1995

DIDIER: CHIMAEROID FISHES 59

TABLE 8—(Continued )

M. coracohyoideus, ch M. coracohyoideus Vetter (1878) M. coracohyoideus Shann (1919) M. coracohyoideus Kesteven (1933) M. rectus cervicus, Edgeworth (1935)

Mm. coracobranchiales, cbr Mm. coracobranchiales Vetter (1878) Mm. coracobranchiales Shann (1919) Mm. coracobranchiales Kesteven (1933)

Mm. coracobranchiales Edgeworth (1935)

M. epaxialis, ep

M. retractor dorsalis pectoralis, Shann (1919) Shann (1919) described the M. retractor dorsalis pectoralis as separate from the epaxial

musculature. Although its fibers intermingle with the M. epaxialis posteriorly, its origin from the dorsal posterior edge of the scapular process seems to indicate a

separate embryonic origin.

M. retractor latero-ventralis pectoralis, rlvp

M. retractor latero-ventralis pectoralis, Shann (1919)

This muscle is described as having two points of origin. Shann (1919) distinguished these two heads as an external and internal portion of this muscle.

M. retractor mesio-ventralis pectoralis, rmivp

M. retractor mesio-ventralis pectoralis, Shann (1919) This muscle lies underneath the M. retractor latero-ventralis pectoralis and originates from the otic process of the cranium with the fibers running under the scapular

process where they are joined by fibers from the posterior rim and medial face of the

scapular process.

M. levator anguli oris posterior, laop (figs. 35, 38, 41, 43)

Origin: From the antorbital crest with the M. levator anguli oris anterior. In rhinochi- maerids the posterior fibers take origin from the preorbital fascia and in chimaerids the origin is only from the preorbital fascia.

Insertion: Onto the posterior end of the superior maxillary cartilage with a small ten- don inserting medially onto the connective tissue of the lip between the inferior maxil- lary cartilage and the premandibular carti- lage.

Remarks: In rhinochimaerids and chi- maerids the muscle insertion extends over the dorsal edge of the superior and inferior maxillary cartilages.

M. labialis anterior, la (figs. 35, 38, 41, 43) Origin: From the anterior tip of the pre-

labial cartilage. In chimaerids this muscle originates from the prelabial portion of the premaxillary cartilage.

Insertion: Onto the posterior end of the superior maxillary cartilage.

M. intermandibularis, im (figs. 34, 35, 37, 41, 45)

Origin: From the posterior edge of the su- perior maxillary cartilage in Callorhinchus and from the medial side of the inferior max- illary cartilage in all other genera.

Insertion: Onto dense connective tissue surrounding the premandibular cartilages. In Callorhinchus this muscle inserts superficial to the interhyoideus at the anteroventral edge of the lower jaw.

Remarks: This muscle is interrupted by an insertion onto the premandibular cartilage and has been described as two separate mus-

60 AMERICAN MUSEUM NOVITATES

NO. 3119

Fig. 37. Ventral muscles of Callorhinchus milii. A, The skin has been removed to expose superficial musculature. B, Ventral constrictor muscles have been cut to expose deep hypobranchial muscles. See text for a detailed description of the musculature. Scale bar = 2 cm.

cles (Luther, 1909; Kesteven, 1933; Edge- worth, 1935). I can find no distinct muscular connection between the premandibular car- tilage and the lower jaw in Rhinochimaera, Harriotta, or Hydrolagus even though it is figured and described for Hydrolagus (Vetter, 1878; Luther, 1909; Kesteven, 1933; Edge- worth, 1935). My observations indicate that the so-called M. labialis posterior is actually the M. intermandibularis and the two parts described are one muscle that is interrupted by the premandibular cartilage. Dense con- nective tissue binds the ventral premandi- bular cartilages to the lower jaw and it is likely that the second part. of the M. intermandi- bularis running from the premandibular car- tilage to the lower jaw has been reduced and lost with dense connective tissue taking its place. If Luther (1909) and Vetter (1878) are correct in their figures for Chimaera, it seems

reasonable to assume that the M. interman- dibularis is a single muscle interrupted by the premandibular cartilage.

M. adductor mandibulae anterior, ama (figs. 34, 36, 39, 41, 42, 44)

Origin: From the preorbital lamina with posterior fibers also taking origin from the preorbital fascia.

Insertion: Onto the lower jaw by a stout tendon that wraps over the lateral surface of Meckel’s cartilage, forming a dense connec- tive tissue sling around the lower jaw.

Remarks: This muscle is unusual in that its entire mass lies anterior to the orbit. The muscle fibers insert in a pinnate fashion with a thin strip of connective tissue dividing this muscle into two parts. In males the frontal tenaculum, jaws, and labial cartilages are functionally linked (Raikow and Swierczews-

1995

DIDIER: CHIMAEROID FISHES 61

laoa_ laop |

Fig. 38. Superficial muscles of the head and pectoral region of Rhinochimaera pacifica (AMNH 96939). Skin and connective tissue of the head and snout have been removed to expose the underlying muscles. See text for a detailed description of the musculature. Scale bar = 2 cm.

ki, 1975) and coordinated movement of the jaws and frontal tenaculum is possible. The tendon of the M. adductor mandibulae an- terior has a dorsal connection to the frontal tenaculum with some fibers inserting onto the skin of the pocket in which the tenaculum rests. This dorsal insertion occurs via con- nective tissue that runs from the supraorbital crest to skin of the pocket of the frontal te- naculum. There is also a muscular connec- tion between the labial cartilages and the frontal tenaculum (see description for M. le- vator anguli oris anterior above).

M. adductor mandibulae posterior, amp (figs. 36, 39, 41, 44)

Origin: From the preorbital fascia and the suborbital ridge.

Insertion: Onto the lower jaw by a stout tendon.

Remarks: In Callorhinchus this muscle originates from the preorbital fascia and par- tially overlies the M. adductor mandibulae anterior with some fibers inserting into the central connective tissue ventrally. In rhin- ochimaerids this muscle originates solely from the suborbital ridge and in chimaerids the M.

SM MX. ih

Fig. 39. Deep muscles of the head and pectoral region of Rhinochimaera pacifica (AMNH 96939). Superficial muscles of the jaws, labial cartilages, and opercular flap have been removed, exposing deep muscles underneath. See text for a detailed description of the musculature. Scale bar = 2 cm.

62 AMERICAN MUSEUM NOVITATES

NO. 3119

cm CV

Fig. 40. Superficial muscles of the head and pectoral region of Harriotta raleighana (AMNH 96931). Skin and connective tissue of the head and snout have been removed to expose underlying musculature. See text for a detailed description of the musculature. Scale bar = 2 cm.

adductor mandibulae posterior is reduced to a small patch of muscle fibers almost com- pletely obscured by the posteroventral edge of the M7. adductor mandibulae anterior.

HyYorb GROUP

The following is a description of muscles derived from the hyoid arch with VIIth cra- nial nerve innervation.

M. constrictor operculi dorsalis, cd (figs. 34, 35, 38, 40, 41, 43)

Origin: The anterior fibers originate via connective tissue from the ventral side of the notochord with posterior fibers originating from the anterior edge of the scapular process just dorsal to the articulation of the pectoral fin.

Insertion: Onto the connective tissue of the opercular flap.

Remarks: In all species examined this mus- cle is the same.

M. constrictor operculi ventralis, cv (figs. 34, 36-38, 40, 41, 43, 45)

Origin: From connective tissue of the oper- cular cover. Some posterior fibers are con- tinuous with the M. constrictor operculi dor- salis.

Insertion: A ventral band of fascia joins this muscle to its antimere.

Remarks: Some fibers fused to the under- side of this muscle sheet extend ventrally and fuse to the M. interhyoideus at its tendon. This is what Kesteven (1933) described as the posterior deep constrictor and I believe it to be a portion of the M. interhyoideus.

In all species examined there is a ventrally directed layer of muscle fibers fused to the

Fig. 41. Deep muscles of the head and pectoral region of Harriotta raleighana (AMNH 96931). Superficial musculature and connective tissue of the head and snout have been removed to reveal deep muscles underneath. See text for a detailed description of the musculature. Scale bar = 2 cm.

1995

DIDIER: CHIMAEROID FISHES 63

\ uN

\

ee 4 =r ~~ F — : a a a,

Zs er ‘

=

Fig. 42. Deep muscles of the head and pectoral region of Harriotta raleighana (AMNH 96931). Musculature of the labial cartilages and the opercular flap has been removed to expose deep muscles of the jaws and gill arches. See text for a detailed description of the musculature. Scale bar = 2 cm.

underside of this ventral constrictor muscle. In Rhinochimaera pacifica, Harriotta ral- eighana, and Hydrolagus novaezealandiae these deep fibers insert onto the posterior edge of the lower jaw where the M. coracoman- dibularis inserts. There is also a sheet of an- teriorly directed muscle fibers that extends over the preorbital portion of the head and rostrum. This muscle sheet forms as anteri- orly directed fibers from the M. constrictor operculi ventralis fuse with muscle fibers that

RL

originate at the insertion of the M. coraco- mandibularis. This sheet of muscle inserts into the dense connective tissue just under- neath the skin and ampullary canals that overlie the labial muscles and M. adductor mandibulae anterior. Kesteven (1933) distin- guished this muscle in Callorhinchus milii as the M. depressor rostri et veli along with a deeper layer, the anterior deep constrictor. He later suggested that these layers may be part of a single muscle sheet (1933: 465). I

Fig. 43. Superficial muscles of the head and pectoral region of Hydrolagus novaezealandiae. The skin has been removed and connective tissue and ampullary organs of the snout have been cut away to reveal superficial muscles of the head. See text for a detailed description of the musculature. Scale bar

= 2 cm.

64 AMERICAN MUSEUM NOVITATES

NO. 3119

DAD‘49

Fig. 44. Deep muscles of the head and pectoral region of Hydrolagus novaezealandiae. Superficial muscles of the head and opercular flap have been removed. See text for a detailed description of the

musculature. Scale bar = 2 cm.

cannot separate this muscle sheet into dis- tinct layers and therefore consider it to be a single muscle. Edgeworth (1935: 97-98) de- scribed this anterior extension of muscle fi- bers as derived from the hyoid constrictor sheet and its innervation by the VIIth cranial nerve confirms this hypothesis.

M. constrictor operculi dorsalis anterior, cda (figs. 34, 35)

Origin: From the lateral edge of the sub- orbital shelf.

Insertion: Via a tendon that extends ven- trally and splits to insert anteriorly into the connective tissue sheet that covers the ros- trum and posteriorly wraps around the jaw joint.

Remarks: I have only found this muscle in Callorhinchus milii. This muscle seems to be derived as part of the M. constrictor operculi dorsalis, yet it is distinctly separate except at its very dorsal end where some muscle fibers intermingle with the M. constrictor operculi dorsalis. Kesteven (1933: 457) called this muscle a part of the M. depressor mandibuli superior (D.m.s., Kesteven, 1933: 458, fig. 6) and described its insertion onto the lower jaw and floor of the mouth. He also stated that it is innervated by the VIIth cranial nerve. Nowhere does he describe the unique tendon

associated with this muscle and Edgeworth (1935) does not mention this muscle.

M. levator hyoideus, lh (figs. 35, 39, 42, 44)

Origin: From the underside of the neuro- cranium ventral to the orbit.

Insertion: Onto the dorsal edge of the epih- yal.

Remarks: Edgeworth (1935) described this muscle as being derived from the hyoid con- strictor muscle sheet.

M. interhyoideus, ih (figs. 35, 39, 42, 44)

Origin: From the posterior edge of the low- er jaw at the symphysis, deep to the insertion of the M. coracomandibularis.

Insertion: Onto the lateral face of the cer- atohyal at its ventral angle.

Remarks: This muscle is also derived from the hyoid constrictor muscle sheet (Edge- worth, 1935). Based on positional informa- tion only, this muscle may be functionally analogous to the mandibular depressor mus- cle of lungfishes and correlated with the evo- lution of autostyly in both of these lineages (Bemis, 1987a, b; Didier, 1988).

BRANCHIAL MUSCLES

The following is a description of all bran- chial and epibranchial muscles innervated by nerves IX and X.

1995

pao

Fig. 45.

DIDIER: CHIMAEROID FISHES 65

Ventral muscles of Hydrolagus novaezealandiae. A, The skin has been removed to reveal

the superficial musculature. B, Ventral constrictor muscles have been removed to expose deep hypo- branchial muscles. See text for a detailed description of the musculature. Scale bar = 2 cm.

Mm. constrictores branchiales, cb (fig. 36) Origin: From epibranchials 1-3. Insertion: Onto ceratobranchials 1-3. Remarks: I found only three branchial con-

strictors. Edgeworth (1935: 137) described a

fourth branchial constrictor that lies between

the fifth gill cleft and the M. coracobranchial- is. In Rhinochimaera pacifica 1 found only the first branchial constrictor, but it is as- sumed that the pattern follows that of Cal-

lorhinchus. For Chimaera Edgeworth (1935:

137) described these muscles as extending

from ceratobranchials to epibranchials of the

following arches. This may explain why Vet- ter (1878) named these muscles Mm. inter- branchiales.

Mm. adductores arcuum branchialium, aac (not shown)

Origin: From epibranchials 1-3 and the pharyngobranchial complex of branchial arches 4 and 5.

Insertion: Onto ceratobranchials 1-4.

Remarks: These small muscles lie on the medial side of the branchial arches and four are always present. Edgeworth (1935: 137, 166) claimed that the most caudal adductor is developed into a M. constrictor oesophagi that originates in chimaeroids from the third pharyngobranchial and inserts onto the pen- ultimate basibranchial. I have found this

muscle and it originates from the pharyn- gobranchial complex and inserts onto the fourth basibranchial. I follow Edgeworth (1935) in calling this last branchial constric- tor the M. constrictor oesophagi.

M. cucullaris superficialis, cs (figs. 34, 38, 40, 42, 43)

Origin: From the postorbital crest and pos- teriorly from a connective tissue sheet that overlies the epaxial muscles.

Insertion: Onto the lateral face of the scap- ular process.

Remarks: This muscle is present in all taxa examined. It angles ventrally from its origin to its insertion. There is often a raised ridge or protuberance at the site of insertion and it is especially prominent in Harriotta. In Hy- drolagus novaezealandiae some muscle fibers run under the scapular process and insert onto its ventral side.

M. protractor dorsalis pectoralis, pdp (figs. 35, 38, 42, 44)

Origin: From the postorbital ridge and the otic capsule deep to the M. cucullaris super- ficialis.

Insertion: Onto the scapular process along the anterior edge and medial surface.

Remarks: This muscle was first described by Shann (1919); however, Edgeworth (1935)

66 AMERICAN MUSEUM NOVITATES

made no reference to this muscle as part of the cucullaris complex in Holocephali.

M. cucullaris profundus, cp (figs. 36, 38, 44) Origin: From the underside of the otic shelf. Insertion: Onto the posterior end of the

pharyngobranchial complex of branchial

arches 4 and 5.

EPIBRANCHIAL SPINAL MUSCLES

The following is a description of epibran- chial musculature with spinal nerve inner- vation.

M. subspinalis (not shown)

Origin: From the underside of the otic shelf medial to the M. cucullaris profundus.

Insertion: Onto pharyngobranchials 1 and 2:

Remarks: A single M. interpharyngobran- chialis is found in all taxa examined. It runs from pharyngobranchial 2 to pharyngo- branchial 3 (Edgeworth, 1935).

HYPOBRANCHIAL SPINAL MUSCLES

The following description includes ventral branchial muscles of myotomic origin with spinal innervation.

M. coracomandibularis, cm (figs. 35-37, 39, 40, 42, 44, 45)

Origin: From a central depression on the anterior face of the coracoid with some bun- dles of muscle fibers taking origin lateral and dorsal to this depression.

Insertion: Onto the posterior edge of the lower jaw lateral to the symphysis.

Remarks: This large muscle mass origi- nates on the ventral surface of the coracoid. Just cranial to the origin the muscle is divided by a V-shaped septum of connective tissue at which point the mass divides into several muscle bundles. Two bundles insert on the lower jaw laterally, ventral to the jaw joint, and a central sheet of muscle inserts along the posterior rim of the lower jaw. This mus- cle pattern is found in all chimaeroids I have examined and is described by Shann (1919), but the V-shaped septum is not present in Callorhinchus. In Callorhinchus a central bundle of muscle heads anteriorly and ends in a long tendon that splits as it reaches the anterior edge of the lower jaw; the tendon

NO. 3119

inserts onto the right and left sides of the lower jaw.

M. coracohyoideus, ch (fig. 39)

Origin: From the connective tissue septum of the M. coracomandibularis complex.

Insertion: Onto the ventral surface of the basihyal cartilage.

Remarks: This paired muscle runs deep to the central portion of the MZ. coracomandi- bularis. Further studies are needed to verify the presence or absence of the V-shaped sep- tum in Callorhinchus and Rhinochimaera and will establish if this muscle does indeed take origin from the septum or the coracoid.

VENTRAL BRANCHIAL MUSCLES

The following muscle complex is devel- oped from the branchial muscle plate and in all chondrichthyans has spinal nerve inner- vation that is considered to be secondarily derived (Edgeworth, 1935).

Mm. coracobranchiales, cbr (figs. 37, 42, 45) Origin: As separate slips of muscle from the ventral edges of the ceratobranchial car- tilages. Insertion: Onto the anterior edge of the coracoid lateral to the origin of the M. cor- acomandibularis.

TRUNK MUSCLES

The following includes a description of the epaxial musculature as well as myotomic trunk muscles.

M. epaxialis, ep (figs. 34-36, 38-40, 42, 43) Origin: From the top of head dorsal to the orbit. Insertion: Continues posteriorly into the myomeres of the dorsal body musculature.

M. retractor dorsalis pectoralis, rdp (figs. 34, 36, 38, 40, 44)

Origin: From the dorsal end of the scapular process along its posterior edge.

Insertion: Continues into myomeres of the dorsal body musculature.

M. retractor latero-ventralis pectoralis, rlvp (figs. 34, 38, 40, 42, 44)

Origin: As two heads from the lateral and medial face of the scapular process ventral to the M. protractor dorsalis pectoralis.

1995

Insertion: Continues deep to the M. re- tractor mesio-ventralis pectoralis as a large muscle bundle that forms the dorsal muscle mass of the body cavity.

Remarks: This muscle was described by Shann (1919) as having two portions, an ex- ternal and internal division, corresponding to the two heads of origin I have described. It is found in all taxa studied.

M. retractor mesio-ventralis pectoralis, rmvp (figs. 34-36, 39, 40, 43, 44)

Origin: Originates as a sheet of muscle from the posterior edge of the pectoral girdle, ven- tral to the origin of the M. retractor dorsalis pectoralis, and extends around the coracoid ventrally.

Insertion: This muscle sheet fans out to form the lateral and ventral body wall.

Remarks: The description given by Shann (1919) indicates a superficial, medial, and in- ferior division of this muscle. The superficial portion originates just dorsal to the articu- lation of the pectoral fin and fans out over the trunk to insert into connective tissue of the transverse septum. In rhinochimaerids this muscle is indistinguishable. I find that the medial portion consists of some deep fi- bers that originate as a discrete bundle via a tendon from the posterior edge of the otic process. These fibers continue under the scapular process where they join fibers orig- inating from the posterior edge of the scap- ular process before fanning out to form the lateral musculature of the body wall. The in- ferior portion originates from the posterior edge of the coracoid, just dorsal to the artic- ulation of the pectoral fin, and extends pos- teriorly, forming the ventral body muscula- ture.

LIGAMENTS

There are two large ligaments in the snout. These have only been observed and described for Callorhinchus, but comparable ligaments may be found in other chimaeroids when more detailed investigation of the snout is completed.

ligamentum labialis, ll (figs. 18A, 35) Origin: From the top of the nasal capsule. Insertion: Into the connective tissue of the

DIDIER: CHIMAEROID FISHES 67

upper lip at the base of the snout just ventral to the prelabial cartilage.

Remarks: I have not found this ligament in any chimaerids or rhinochimaerids. Kes- teven (1933: 459) calied this the Jigamentum radicis rostri.

ligamentum rostralis, ri (fig. 18A)

Origin: From the anterior tip of the pre- labial cartilage.

Insertion: Onto the lateral rostral rod.

Remarks: I have found this ligament only in Callorhinchus. Kesteven (1933: 458) de- scribed the lateral ligament of the rostrum as originating from the nasal capsule and in- serting on the rostral spine (median rostral rod); however, my observations indicate that this ligament originates from the anterior tip of the prelabial cartilage and inserts onto the lateral rostral rod. In Hydrolagus I have found a comparable ligament, but it originates from the dorsal surface of the nasal capsule and inserts on the small lateral rostral rods. I have not found a /igamentum rostralis in the snouts of rhinochimaerids and chimaerids.

CHARACTER ANALYSIS

The phylogenetic conclusions about the re- lationships of living chimaeroids are shown in figure 46. My purpose is to resolve rela- tionships among extant chimaeroids, and the following discussion will assess shared de- rived characters (synapomorphies) only as they relate to living forms. It is understood that many taxa known only from fossils (e.g., Ischyodus) probably belong to Chimaeroidei, but the systematic position of fossil forms is beyond the scope of this study.

The following character scheme is orga- nized around the families Callorhynchidae, Rhinochimaeridae, and Chimaeridae and their genera. Many of the characters listed are new or are redefined from previous ac- counts. The synapomorphies of Chondrich- thyes, Holocephali, and Chimaeriformes are listed in tables 1 and 5 and will not be treated further in this analysis.

SYNAPOMORPHIES OF CHIMAEROIDEI

49. REDUCTION OF TRABECULAR DENTINE IN THE LATERAL WALLS OF THE FIN SPINE. Pat-

68 AMERICAN MUSEUM NOVITATES

NO. 3119

Elasmobranchii Holocephali

Chimaeriformes

Chimaeroidei

Callorhynchoidea

Chimaeroidea

Callorhynchidae Rhinochimaeridae Chimaeridae Rhinochimaerinae —_ Harriottinae tMenaspoidei tMyriacanthoidei t Squaloraja Neoharriotta Chimaera Squalus tHelodus t ischyodus Callorhinchus Rhinochimaera Harriotta Hydrolagus 95-97 92-94 98-103 73-82 83-91 49-72 31-48 24-30

1-23

Fig. 46. Phylogenetic relationships of extant chimaeroids. In this first cladistic interpretation of relationships among the living forms, callorhynchids represent the most primitive living chimaeroids and the chimaerids, Chimaera and Hydrolagus, are derived forms. The rhinochimaerids are illustrated as a polychotomy with both lineages sharing characters 92-94. There are three apomorphies of Rhin- ochimaera (characters 95-97) and no synapomorphies uniting Harriotta and Neoharriotta. Although these problematic relationships could also have been represented as a Y-shaped branching, I have here used a polychotomy to emphasize the distinction between Rhinochimaera and the Harriottines (Harriotta plus Neoharriotta). Based on characters 92-94 alone, it is not clear that Rhinochimaeridae is necessarily a monophyletic group. See text for further discussion and description of characters.

terson (1965: 197) described a narrow zone of tissue, which he defined as osteodentine, in the fin spines of Recent chimaeroids and hypothesized that the thick layer of osteo- dentine found in extinct chimaeriforms was replaced by lamellar tissue in the modern forms. Because of this histological difference, Patterson (1965) interpreted the fin spine of Recent chimaeroids to be a recently evolved feature from a spineless form, not homolo- gous to the fin spine of fossil forms such as Helodus. According to Maisey (1986), the lack

of trabecular dentine (“‘osteodentine’’) is a synapomorphy of living chimaeroids (see ta- ble 5). A comparative investigation of the fin spines of Recent and fossil holocephalans in- dicates that they are homologous on the basis of morphological and developmental simi- larities and Recent forms are characterized not by the absence of trabecular dentine in the fin spine, but by a reduction in the amount of trabecular dentine present (Maisey, un- publ. Ph.D. diss., 1974; personal commun.).

50. SCAPULOCORACOIDS ARE FUSED

1995

VENTRALLY (MAISEY, 1986: J20). This char- acter also occurs in some Recent elasmo- branchs and is considered by Compagno (1973) to be a derived character of neosela- chians. Because this feature is not present in hexanchoids, Chlamydoselachus, and some fossil holocephalans, its occurrence in elas- mobranchs and holocephalans is convergent (Maisey, 1984, 1986).

51. VENTRAL LOBE OF THE PITUITARY IS ISOLATED EXTERNAL TO THE CRANIUM (Mais- ey, 1986: J22). According to Maisey (1986) this is probably a convergent feature of elas- mobranchs and chimaeroids. A similar mod- ification of the pituitary is also found in coe- lacanths (Lagios, 1979). Most workers agree that it has been independently derived in all of these lineages; however, more study of this character is needed.

52. ALL OF THE TOOTH PLATES ARE COMPOSED OF TRABECULAR DENTINE AND HAVE HYPERMINERALIZED REGIONS (TRITORS) IN LARGE DISCONTINUOUS PATCHES. Aspects of this character have been discussed by many workers (e.g., Nielsen, 1932; Bargmann, 1933, 1941; Peyer, 1968; Orvig, 1985); however, this character has historically been problem- atic because of the complex and confusing terminology related to the tooth plates and their tissues (e.g., Kemp [1984] for a review). Fossil holocephalans (e.g., Helodus, cochlio- donts) have tooth plates in which the entire crown is covered with hypermineralized tra- - becular dentine; however, other fossil taxa (e.g., Myriacanthus) exhibit a reduction in the amount of hypermineralized tissue in the crown of only the upper anterior tooth plates that have characteristic patches of hypermi- neralized tissue. Therefore, I interpret tooth plates in which the hypermineralized tissue does not cover the entire crown as derived for Chimaeriformes and the presence of dis- crete patches of hypermineralized tissue in all tooth plates as derived for Chimaeroidei.

53. A DESCENDING LAMINA IS PRESENT ON THE ABORAL SURFACE OF THE TOOTH PLATES. The descending lamina of chimaeroid tooth plates has been described by Patterson (1992). A descending lamina is present, at the very least, in the mandibular and palatine tooth plates of many fossil forms, including [schy- odus and Myriacanthus, and is well devel- oped in all tooth plates of Callorhinchus

DIDIER: CHIMAEROID FISHES 69

milii. I agree with Patterson (1992) that a well-developed descending lamina is primi- tive for chimaeroids. This character certainly applies to more than just the living forms; however, a closer examination of this char- acter in all tooth plates of the fossil forms is beyond the scope of this analysis and the dis- tribution of this character among fossil taxa is not indicated on my cladogram.

54. A MORPHOLOGICALLY COMPLETE HYOID ARCH THAT INCLUDES A PHARYNGOHYAL ELE- MENT IS PRESENT. Chimaeroids are the only living vertebrates with this character. Be- cause the pharyngohyal is cartilaginous and small in size, it is difficult (or impossible) to assess in fossils. Outgoup comparison and embryological and paleontological evidence suggest that this is a derived feature related to the evolution of autostyly in this lineage.

55. JAW JOINT IS ANTERIOR TO THE EYE WITH JAW MUSCLES ORIGINATING ANTERIOR TO THE EYE. This character is not known in fossils and is correlated with the evolution of au- tostyly. The adductor muscles of Heterodon- tus also lie anterior to the eye and superfi- cially they resemble chimaeroid fishes in this respect. I interpret this as a convergent fea- ture of heterodontids and chimaeroids.

56. FUSED PHARYNGO-EPIBRANCHIAL PLATE ASSOCIATED WITH THE THIRD, FOURTH, AND FIFTH BRANCHIAL ARCHES. This element ar- ticulates with both the notochord and pec- toral girdle via a strong ligament, and also articulates with the third epibranchial and fourth and fifth ceratobranchials. Although this feature has been noted by many workers (e.g., figured by Garman, 1904), its phylo- genetic significance has never been consid- ered. I have found this skeletal element in all genera of extant chimaeroids and consider it to be a synapomorphy of Chimaeroidei. I have found no reference to this character in any other vertebrate taxa.

57. THE FIRST EPIBRANCHIAL ARTICULATES WITH THE HYOID ARCH. An articulation be- tween the first epibranchial and the hyoid arch is found in all extant chimaeroids. This character is probably related to the presence of a complete hyoid arch in holocephalans. It is likely that this is a synapomorphy of Holocephali, although this feature is not known for any fossil taxa due to poor pres- ervation of the gill arches. It is considered a

70 AMERICAN MUSEUM NOVITATES

synapomorphy of Chimaeroidei based on this study.

58. PRESENCE OF A FLESHY OPERCULUM THAT IS FORMED BY THE DORSAL AND VENTRAL CONSTRICTOR MUSCLES AND SUPPORTED BY AN OPERCULAR CARTILAGE AND HYOID RAYS. Among Recent Chondrichthyes, this char- acter appears to be unique to chimaeroid fish- es. The opercular cartilage, formed by the fusion of dorsal hyoid rays, articulates at the joint between the epihyal and ceratohyal el- ements of the hyoid arch and supports a series of filamentous hyoid rays ventrally. A series of hyoid rays is also present along the pos- terior edge of the ceratohyal. These hyoid rays lie directly underneath (and support) the fleshy opercular flap that is formed by the dorsal and ventral constrictor muscles. Holmgren (1942: 207) claimed that hyoid rays were present in galeoid sharks and rays; how- ever, among living chondrichthyans, only chimaeroids possess a fleshy operculum sup- ported by hyoid rays. It has been suggested that the presence of a fleshy operculum is a convergent feature among Chondrichthyes (Dean, 1909; Lund, 1977; Maisey, 1986), and the presence of long hyoid rays in some xen- acanths and symmoriids indicates that a fleshy operculum may have been present in some fossil forms.

59. THE M. LEVATOR HYOIDEUS ORIGINATES FROM THE SUBORBITAL SHELF ANTERIOR TO THE OTIC CAPSULE. This muscle is not homologous to the M. levator hyoideus of elasmobranchs, which originates from the otic capsule (Holmgren, 1942: 210). According to Edge- worth (1935), the M. levator hyoideus in hol- ocephalans is derived from the constrictor sheet and is homologous to the levator in dipnoans. This is probably a convergent fea- ture in dipnoans and chimaeroids related to the independent evolution of autostyly and the presence of a nonsuspensory hyomandib- ula 1n both of these lineages. This character cannot be assessed in the fossil Holocephali; however, it may be a synapomorphy at that level since it is associated with autostyly.

60. THE PRESENCE OF A HYOID ARCH MUSCLE THAT EXTENDS ANTERIOR TO THE ORBIT. Por- tions of the dorsal constrictor sheet, inner- vated by the VIIth nerve (Allis, 1916, un- publ.; Edgeworth, 1935), extend into connective tissue of the snout in all chim-

NO. 3119

aeroids. Its origin varies among chimaeroids and I interpret the development of a second- ary origin from the lower jaw to be derived among Chimaeroidei. This character has not previously been described for chimaeroids and is a unique feature of this lineage. This muscle may play a functional role in move- ment of the large snout in these fishes.

61. SIX PAIRS OF LABIAL CARTILAGES ARE PRESENT. Elasmobranchs also possess labial cartilages, but never more than three pairs are present in modern forms. The complexity and degree of development of the labial car- tilages is unique to chimaeroids among living chondrichthyans. The presence of well-de- veloped, complex labial cartilages in Hybod- us basanus (Maisey, 1983) is intriguing in that is suggests that complex labial cartilages may be a primitive feature shared by chi- maeroids and elasmobranchs; however, fur- ther study will be required to test this hy- pothesis. For this analysis I accept the interpretation of Holmgren (1942: 246) that the presence of six labial cartilages is the primitive condition in chimaeroids.

62. PREPELVIC TENACULA WITH INDEP- ENDENT CARTILAGINOUS SKELETON PRESENT IN BOTH MALES AND FEMALES. The prepelvic te- nacula articulate with the anterior edge of the pelvic girdle and are housed in small pouches anterior to the pelvic fins. This character is unique to chimaeroid fishes and has generally been considered to be a sexually dimorphic characteristic; however, there are rudimen- tary prepelvic tenacula and prepelvic pouch- es in all female callorhynchids.

63. PRESENCE OF A FRONTAL TENACULUM. A frontal tenaculum is unique to chimaeroid fishes and in all male chimaeroids this is a median, unpaired structure armed with den- ticles at the bulbous tip and housed in a small groove atop the head just anterior to the an- torbital crest. This structure is present but remains undeveloped (beneath the skin) in females. The evolution and significance of the frontal tenaculum has been studied by several workers (e.g., Reis, 1895; Dean, 1906; Pat- terson, 1965; Raikow and Swierczewski, 1975). It has been hypothesized that the un- paired frontal tenaculum in chimaeroids evolved from paired structures (Reis, 1895; Patterson, 1965).

64. FUSED ANTERIOR RADIALS ARTICULATE

1995

WITH THE PROPTERYGIUM OF THE PECTORAL FIN. This feature characterizes the pectoral fins of all extant chimaeroids and 1s probably related to the presence of a dibasal fin.

65. THE FIRST TWO OR THREE RADIALS OF THE PELVIC FIN ARE FUSED WITH THE BASIPTERYGIUM. The basipterygium of the pelvic fin in all extant chimaeroids has a char- acteristic process along the anterior margin formed by the fusion of the first few pelvic radials. Present in all extant chimaeroids, this feature is considered to be derived in this lineage.

66. THE OTIC CAPSULES HAVE A MEM- BRANOUS MEDIAN WALL. The otic capsules of chimaeroids have only a membranous inner wall (Holmgren 1942: 195), whereas in elas- mobranchs the inner wall of the otic capsules is cartilaginous. A membranous inner wall of the otic capsule is considered to be a derived feature of chimaeroids.

67. THE SPIRACLE IS ABSENT IN ADULTS DUE TO ONTOGENETIC Loss. A spiracle is lacking in all chimaeroids, although a small spiracle is present in early developmental stages (the only taxa surveyed to date are Callorhinchus and Hydrolagus). The early ontogenetic clo- sure of the spiracle is probably related to the development of the operculum. Loss of the spiracle is derived in Chimaeroidei, although it is a convergent feature found in other lin- eages (e.g., dipnoans, scaphyrhinchine stur- geons, and neopterygians).

68. TWO LATERAL LINE CANALS ARE PRESENT ABOVE THE MOUTH. Lateral line canals have been described and discussed by several workers, in particular Garman (1888) for Chimaera monstrosa and Callorhinchus an- tarcticus; Garman (1904) for Rhinochimaera pacifica; Reese (1910) for Chimaera mon- strosa; Bigelow and Schroeder (1953) for North Atlantic species; as well as Bullis and Carpenter (1966) and Compagno et al. (1990) for rhinochimaerids. The lateral line canals are compared among all genera of chima- eroids for the first time in this study. The pattern of lateral line canals is unknown in fossil forms. This feature is present in all ex- tant forms and is thus considered here as a synapomorphy of Chimaeroidei.

69. THE FIRST THREE BASIBRANCHIAL CAR- TILAGES ARE REDUCED TO LUMPS OF FIBRO- CARTILAGE. In all extant chimaeroids these

DIDIER: CHIMAEROID FISHES 71

are unpaired ventral elements that do not articulate directly with the corresponding hy- pobranchial cartilages, but are instead inter- connected to each other and to the hypo- branchial elements by connective tissue. These three cartilages remain uncalcified even in the largest and most mature fish.

70. LARGE EGG CASES WITH A BROAD, RIBBED LATERAL WEB EXTENDING AROUND THE BULBOUS CENTRAL SPINDLE. This type of egg case is common to all callorhynchids and rhinochimaerids, and egg cases with a broad lateral web are identified here as a plesiom- orphic character for all Recent chimaeroids. The reduction of the lateral web in egg cases of several species of chimaerids had been documented by Dean (1912) and I agree with his interpretation that a large lateral web is primitive. Although it is impossible to dis- criminate species based solely on egg cases, the discovery of fossil egg capsules resem- bling those of callorhynchids (Brown, 1946) is intriguing evidence for the potential use- fulness of this character for higher level phy- logenetic analysis when considered in con- junction with other fossil evidence, embryological studies, and when compared with egg cases of living forms.

71. AT LEAST THIRTEEN DISTINCT AMPULLARY PORE FIELDS ARE PRESENT ON THE HEAD AND SNOUT. The distribution of the pores of the ampullary organs with respect to the lateral line canals appears to be very un- usual in chimaeroids in comparison to those gnathostomes in which the ampullary organs have been carefully studied (e.g., Ambysto- ma, Northcutt, 1990, Northcutt et al., 1994; elasmobranchs, Bodznick and Boord, 1986; Polyodon, Bemis and Northcutt, personal commun.). Specifically, it appears that the Openings in a particular ampullary field are located only on one side of the adjacent lat- eral line canal rather than flanking it as in these other taxa.

72. ORBITS LIE DORSAL TO THE TELENCEPH- ALON AND ARE SEPARATED BY A MEMBRANOUS INTERORBITAL SEPTUM. Chimaeroids are un- usual in that the orbit is located dorsal to the telencephalon, which extends anteriorly as the olfactory tracts run ventral to the eyes into the ethmoid region. Embryological studies indicate that the skull is shaped by devel- opment of the eyes, and early development

72 AMERICAN MUSEUM NOVITATES

of large eyes in these fishes plays an important role in the development of a membranous interorbital septum that is present only in this lineage.

SYNAPOMORPHIES OF CALLORHYNCHOIDEA

73. CALCIFIED RINGS ARE NOT PRESENT IN THE NOTOCHORDAL SHEATH. All juveniles lack calcified rings in the notochordal sheath, but callorhynchids are the only chimaeroids that do not develop notochordal calcifications as adults. Notochordal calcifications are almost certainly a synapomorphy of Holocephali as this feature is present in all Recent chima- eroids (except Callorhinchus) as well as sev- eral fossil taxa (e.g., myriacanthoids and Squaloraja). The absence of notochordal cal- cifications in callorhynchids is interpreted as a synapomorphy of this lineage, although some extinct genera with calcified notochord- al rings have been classified as callorhynchids (Stahl, personal commun.).

74. ANGULAR AND ORAL CANALS BRANCH SEPARATELY FROM THE INFRAORBITAL CANAL. In all other chimaeroids the angular and oral canals share a common branch (the horizon- tal canal) that is lacking in callorhynchids.

75. PELVIC CLASPERS ARE IN THE FORM OF CARTILAGINOUS SCROLLS THAT LACK DENTI- CLES. Scrolled claspers are also present in many Recent elasmobranchs (Compagno, 1973, 1977); however, this type of clasper morphology is probably independently de- rived in callorhynchids and elasmobranchs based on comparisons with fossil chondri- chthyan taxa, which have claspers in the form of several jointed cartilages (e.g., Zangerl, 1981). The rodlike pelvic claspers of rhino- chimaerids are morphologically very differ- ent from the scrolled rods of callorhynchids and the presence of denticles on rhinochi- maerid claspers distinguishes them from those of callorhynchids. Pelvic claspers are known in a few fossil forms, but their morphology is not well known. I consider these smooth, scroll-like claspers to be primitive for chi- maeroids, although it is possible that dentic- ulated pelvic claspers are primitive for chi- maeroids, and this feature is an apomorphy of callorhynchids.

76. COMPLEX PREPELVIC TENACULA. The prepelvic tenacula of callorhynchids consist

NO. 3119

of a large spatulate blade of cartilage armed with flat scalloped denticles and a scrolled cartilaginous tube. Two fleshy, frilled lobes and a gland are also associated with the pre- pelvic tenacula, and the entire complex is housed within prepelvic pouches. Correlated with the presence of complex prepelvic te- nacula are large anterior processes of the pel- vic girdle with open vacuities. Among living fishes only chimaeroids have prepelvic te- nacula. Based on the scanty fossil evidence, it appears that retractable prepelvic tenacula and prepelvic pouches are recent specializa- tions of chimaeroids. Patterson (1965) con- sidered the articulated, skeletally supported prepelvic tenacula of Recent chimaeroids to be derived from groups of enlarged scales with no skeletal support. I consider the presence of complex prepelvic tenacula in male cal- lorhynchids and the presence of rudimentary prepelvic cartilages in females to be primi- tive. It is equally likely that this feature is an apomorphy of callorhynchids and simple prepelvic tenacula are primitive, but this can- not be resolved until further paleontological and embryological studies are completed.

77. THE ROSTRUM IS FORMED INTO A PLOW-SHAPE. The three rostral rods are al- most equal in length and support