Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
zoological Journal o f t h e Linnean Society (1990), 100 73-100. With 16 figures The syncranial osteology of the southern eelcod family Muraenolepididae, with comments on its phylogenetic relationships and on the biogeography of sub-Antarctic gadoid fishes G. J. HOWES Department of <oology, British Museum (Natural History), Cromzerell Road, London SW75BD Received July 1989, accepted for publication Januav 1990 The monogeneric gadoid family Muraenolepididae has been neglected both taxonomically and anatomically. Without good evidence it has been the opinion of most authors that the family possesses mainly primitive features. A comparison of the syncranial osteology with that of other taxa shows that this can be refuted and that Muraenolepis is a relatively derived ‘higher’ gadoid, having its relationships close to the Phycidae and Gadidae. The circum-Antarctic distribution of Muraenolepk is unique within gadoids and poses the question of its origin. This is discussed with reference to bipolar distributions exhibited by some other gadoid families. I t is concluded that gadoids have evolved with the Atlantic continental shelves and their distribution (and bipolarity) is a consequence of the geological processes which have formed the Atlantic Ocean. K E Y WORDS:-Gadoidei ~ Muraenolepis - osteology - sub-antarctic biogeography - bipolarity. CONTENTS Introduction . . . . . . . . . . . . . . . Material and methods . . . . . . . . . . . . The syncranium of Muraenolepis and comparisons with other gadoids. Cranium. . . . . . . . . . . . . . Derived cranial features . . . . . . . . . . Suspensorium and palato-pterygoid bones . . . . . . Derived suspensorial features . . . . . . . . . Jaws. . . . . . . . . . . . . . . Discussion, . . . . . . . . . . . . . . . . . . . The phylogenetic position of Muraenolepididae Biogeography . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . 73 . . . . . . 75 . . . . . . 75 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 . 79 . 83 . 84 .87 89 . 89 . 92 . 98 98 INrRODUCTION The genus of southern eel-cods, Muraenolepis was described by Gunther (1880) from two specimens of M . marmoratus Gunther collected by HMS Challenger from Kerguelen Island. At first placed by Gunther in the Gadidae, Regan (1903) later recognized certain unique features which caused him to refer the genus to its + 0024-4082/90/090073 28 $03.00/0 73 0 1990 The Linnean Society ofLondon G . J . HOWES 74 TABLE 1. Numbers of vertebrae, dorsal and anal rays, and gillrakers (on inner side of 1st arch) in Muraenolepis species Species .W rnarnioratus it\pe, .M microp3 * 1 n7 1 A1 rnirrocephalus (type; orangtensis I n 1 1 1:ertrbrae Dorsal rays Anal rays Gillrakers 70 66-70 82 75 140 129-137 I78 171 I04 96-107 I35 I50 I1 9-10 9 9 *Fahay & Markle (3984) give for M .microps 67-69 vertebrae, 127-141 dorsal and 98-1 12 anal fin rays. Pectoral fin rays number 37-38 in all species and are the highest numbt-r of any gadoid. own family. Svetovidov ( 1939) considered those characters sufficiently distinctive to recognize a suborder for the Muraenolepididae, a category which has been acknowledged by subsequent authors (e.g. Cohen, 1984). In Svetovidov’s ( 1 939) opinion, the Muraenolepididae represent the “. . . most generalized family of all Gadiformes”, a view shared by most subsequent authors. For example, Cohen (1984) remarks that “Muraenolepis is not obviously related to any other gadiform and appears to represent an ancient lineage”. In attempting to provide a cladistic framework for gadoid relationships, Howes (1989, 1990) treated the Muraenolepididae as a n advanced member of the Gadoidei. This assumption was made principally on an analysis of myological and arthrological characters (based in part on Howes, 1988). The few published anatomical data for Muraenolepis comprise: Regan’s (1903) brief diagnosis; Svetovidov’s (1939) short description of the skull and pectoral girdle; Fahay & Markle’s (1984) figure of the caudal fin skeleton; and Howes’ (1987, 1988) respective descriptions of the palatine bone and the cranial muscles. During previous studies on gadoid anatomy (Howes, 1987, 1988) certain osteological features of the muraenolepidoid skull were recognized as unique or unusual and its is one purpose of this paper to present a detailed description of them together with a comparative analysis of those features amongst other gadoids. Muraenolepis currently contains four species; M . marmoratus Gunther 1880 (Fig. l ) , the type species of the genus from Kerguelen Island and M . microps Lonnberg, 1905 from the Falklands, South Shetlands, South Georgia and Balleny Islands, and the Ross Sea shelc M . orangiensis Vaillant, 1888 from Patagonia and Kerguelen Is., M . microcephalus Norman, 1937, known only from the type taken at 63” 51’S, 54” 16’E. The distinctions between the species have been given by Norman (1937) and depend on differences in body depth, head length and length of the first dorsal fin ray. There are, however, differences in vertebral and dorsal and anal fin ray numbers which distinguish the species (Table 1). The snout length (Fig. l ) , mouth shape and dentition of M . marmoratus also differ from those of the other three species (Fig. 12G). Virtually nothing is known of the species’ biology. Only incidental observations have been made on habitat and food (Burchett et al., 1983); eggs and larvae have been described by Efremenko ( 1983), Fahay & Markle ( 1984) and Gon ( 1988). Muraenolepis species are apparently inshore fishes occurring over the Continental Shelf in depths recorded between 10-400 m; the type of M . microcephalus was caught in midwater open-ocean nets (Norman, 1937). SYNCRANIAL OSTEOLOGY % - 75 - Figure 1 . A, Muraenolepis microps, drawing o f a specimen 275 mm TL (BMNH 1986.7.10 : 2; South Georgia). B, C, Comparisons between the heads of Muraenolepis microps (B) of 160mm T L and M . marmoratus (C) of 1543 mm TL (scale pattern shown above for M . microps). Note differences in eye-size, snout length and shape, development of sensory papillae at tip of lower jaw and extent of gill-openings. MATERIAL AND METHODS The following descriptions are based almost entirely on dry skeletal preparations of Muraenolepis microps, prepared by dissection. Attempts at preparing cleared and stained material were largely unsuccessful owing to the intensively fatty and tenacious nature of the flesh and connective tissue. For comparison, crania of the following gadoid genera were examined: Bathygadus, Brosme, Ciliata, Rhinonemus, Euclichthys, Gadomus, Gaidropsarus, Gadus, Lepidion, Lota, Macruronus, Merlangius, Merluccius, Molva, Phycis, Physiculus, Pollachius, Raniceps, Trisopterus, Urophycis (all dry preparations). In addition, the material listed in Howes (1988) was utilized. THE SYNCRANIUM OF MURAENOLEPZS AND COMPARISONS WITH OTHER GADOIDS Cranium (Figs 2-5) In its overall dorsal outline, the cranium is roughly an equilateral triangle and, in this respect, is typical of some members of the Gadidae (cf. Eleginus, Pollachius, Trisopterus; see Svetovidov, 1948). There are no prominent crests and the laminae which roof the frontal and pterotic sensory canals lie flush with the cranial surface. Neither is there a deep or elaborate anterior frontal depression (mucosal cavity of Svetovidov, 1948) which is the hall-mark of the Gadidae. 76 G. J. HO\I'ES ~~ ex0 \ -a"a Figure 2. .Ifurtittiolepi.! m r t r o p ~ ,cranium in dorsal view: left, ethrnoid region, right, occipital region. .Abhre\ iations: m a - 'arrcssory' neural arch. bo- basioccipital, epo- rpioccipital, exo- rxorripital, fr- frontal. ic- intercalar. le- lateral ethmoid. pa- parietal, ptr- pterotir, ptt- posttrmporal, rc- rostrodermosupr;rcthni~~id. so- supraoccipital. sp- autosphenotic, vo- vomcr. Scale bar in mm. r. 1here is no prominent supraoccipital crest; any existing crest is confined to the posterior slope of that bone. The cranium is deep, being 30°, of its length (measured through its orbital depth). Since the cranial roof remains virtually horizontal for its entire length, the depth of the ethmoid region is nearly the same as that of the otic. The ethmoid bloc is strongly ossified and has a posterior extension of the mesethmoid which contacts the parasphenoid. The anterior face of the ethmoid is narrow and vertical; a thin layer of ethmoid cartilage extends posteriorly to fill an anterior fissure of the parasphenoid. A prominent process extends from the lateral face of the mesethmoid bloc, to which attaches the palatine ligament. The lateral ethmoids have thin, cartilaginous \rentromedial lamina which are separated from one other in the midline by the ethmoid cartilage. T h e lateral wing of each bone is in the form of a stout strut whose outer process is thick, truncate and indented. The vomer has a broadly flared and exceptionally thick head; i t is edentulous. The jrontals are fused together and form a convex surface. Anteriorly each frontal narrows and is indented where it meets the nasal. Posteromedially, the bones rise to form a slight crest where they join the supraoccipital. The nasal bone is an elongate, open channel, sloped anteroventrad. Its ventral margin lies in the same horizontal plane as the base of the lateral ethmoid (Fig. 3 ) . Posteriorly, the nasal is attached to the mesethmoid and anteriorly to the 1st infraorbital (lachrymal) by strong connective tissue. I n preserved specimens the nasals are filled with a fat-like substance. SYNCRANIAL OSTEOLOGY 77 fr --- I I I mec vo I PS .n..r I vo I io Figur? 3. Muraenolepis microps, ethmovomerine region; above, lateral view of'skeletal elements, below, showing superficial elements and position of nasal organ. Abbreviations: io- infraorbital (lachrymal), me- mesethmoid, mec- mesethmoid cartilage, na- nasal bone, nr- nasal rosette, ps- parasphenoid; others as in previous figure. The pterosphenoid is a small, oblong, lamellate bone curving somewhat mesad where it meets the frontal (Figs 4, 5). The autosphenotic is elongate and dorsally occupies only a small area between the frontal and pterotic (Figs2, 4, 5); its posterolateral portion contains much of the anterior part of the hyomandibular fossa. Both the pterotic and parietal are large and together form the majority of the occipital cranial roof (Figs ZB, 4). The ventrolateral face of the pterotic bears th.posterior part of the hyomandibular fossa. The epioccipital has a rounded posterior margin and slopes outward to meet the intercalar; medially it contacts the exoccipital and supraoccipital (Figs 4,5). The parasphenoid is narrow for most of its orbital length; anterodorsally there extends a thick, medial lamina which sutures with the posterior extension of the mesethmoid (Fig. 3). The ascending process is shallow, with a long, low-angled slope; the posterior part of the parasphenoid is deep and covers the lateral face of the basioccipital (Fig. 4). The prootic is small with a deeply indented anterior margin forming the border of the optic-trigeminal foramen (see below for further discussion); anterodorsally 78 Pro I I , PS ic bo I I 7" Figure 4. Mura~nolepi~mzcrops, occipital and otic cranial regions in lateral (above) and ventral (below) views. Abbreviations: fg- glossophayngeal foramen, pro- prootic, pts- pterosphenoid; others as in Fig. 2 . fr PfS w P'e k IC W , ' A am 6 bo Figure 5 Muroenolepzs mtrrops, cranium in A, anterior and B, posterior (B) views Abbrevmtions. psn- parasphenoid ascending process; others as in previous figures SYNCRANIAL OSTEOLOGY 79 it bears the lower part of the hyomandibular fossa (Fig. 4). The intercalar is extensive and is bordered by the basioccipital, prootic, pterotic, parietal, epioccipital and exoccipital. The glossopharyngeal foramen lies in its posterior aspect, below the attachment of the posttemporal (Figs 4, 5B). The lateral face of the basioccipital is covered for much of its length by the parasphenoid and thus only the posterior part of the bone is visible in lateral aspect; the central articulatory facet is almost circular. The two lateral facets of the exoccipitals are small and ventromedially directed. A prominent flange rises behind the facet and forms the posteroventral extension of a triangular element overlapping the posterior surface of the exoccipital and which appears to be an ‘accessory’ neural arch (ana, Figs 2, 4, 5). Derived cranial features Frontals fused and absence of ventral laminae Other than in Muraenolepis, fused frontals are encountered only amongst Phycidae ( Urophycis, Phycis, Rhinonemus) and Gadidae (e.g. Gadus, Trisopterus, Pollachius) although not occurring in all species of those genera (Svetovidov, 1948; Mujib, 1967). Unlike those in the majority of gadoids the frontals of Muraenolepis lack any ventral laminae, a condition shared with the Melanonidae. Ventral frontal laminae are variously developed, being shallow and widely separated in Bathygadidae, Steindachneriidae and Euclichthyidae; moderately developed and posteriorly convergent in Macruronidae and Phycidae (Phycis, Urophycis); parallel in Merlucciidae and Ranicipitidae, and in Lotidae (Lota) meet the ascending laminae of the parasphenoid to form paired septa. In the lotid Molua, the frontal laminae, together with the pterosphenoids form parallel arches (Fig. 6C), the posterior section of the arch meeting the pterosphenoid and parasphenoid to form a narrow, walled passage leading to the optic foramen. In the Gadidae, front ventral laminae curve mesad to meet almost in the midline, forming an enclosed channel reminiscent of that in some Moridae (cJ Fig. 6A,B). In most gadoid taxa the olfactory tract runs medial to the lateral ethmoid, passes ventrally, attaching (via connective tissue) to the ventral margin of the frontal lamina. The course of the tract then exhibits two conditions: (1) lying along the outer edge of the lamina before turning inward to the optic foramen, where the lamina becomes reduced and meets the parasphenoid; (2) lying medial to the lamina for the remainder of its course and enclosed by a membrane stretching between the laminae. The second condition is present in the Gadidae. The condition in the Moridae is often cited as being one whereby the olfactory tracts are entirely enclosed within an osseous canal formed by medial front laminae (Svetovidov, 1948; Paulin, 1983; Howes, 1990). However, this is not entirely correct since the condition within the Moridae appears to be as variable as that throughout the Gadoidei. For example, in Lepidion eques there is no ossified canal, each olfactory tract passing into the orbital cavity via a foramen between the frontal and lateral ethmoid, the tracts lying freely beneath the frontals; in Halargyreus johnsonii the tracts are partially enclosed by a membranous septum. In Physiculus the tracts each pass through a canal enclosed within the respective frontal and meet beneath the centre of the orbital cavity within a narrow, ventrally closed canal through which they continue into the G. J . HOM'ES 80 :.- . .~ - -. Figure 6 . Degrer of drvelopmrnt of the frontal laminar. A. Trtsopleru.\ luscuj (Gadidar). B, Lepidion fl- fi.onta1 c q w s [Moridaei, both ventral views. C , .%fo/wmohn (Lotidae),lateral view. Abbreviations: ventral lamina. psk- parasphenoid kerl; others as in pre\-ious figures. cranial cavity. I n the two species of Physiculus examined, P. brevisculus and P. argyropastus, there are marked differences in the degree of ossification of the canal, its structure, and the extent of the lamina extending from beneath the canal to form the medial septum. The widespread occurrence of frontal ventral laminae and their association with the olfactory tract appears to be a characteristic (synapomorphy) of the Gadoidei (in Macrouroidei there is a medial septum of, apparently, ossified membrane separating the tracts). Their absence in both Muraenolepididae and Melanonidae is taken to be a derived, rather than a plesiomorphic situation. In both taxa, but particularly the Melanonidae (which also has markedly short frontals), the brain protrudes forward into the orbital cavity and so obviates the need for long olfactory tract supporting structures. Reduced sphenotic Although never a large bone, the sphenotic of gadoids nearly always has a laterally produced dorsal surface. In the phycids Ciliata and Gaidropsarus, however, it is longitudinally extended (see TablesXXII and XXIII in Svetovidov, 1948). In the Moridae and Ranicipitidae there is usually a gap or 81 SYSCRANIAL OSI'EOLOGY I re I vo -.- '\J pc I P O Figure 7. Rhinonnus cimbriur, antericir part of the cranium in latcral vicw. Abbreviations as in prcvious figures. indentation between the frontal and sphenotic and the sphenotic projects beyond the lateral border of the pterotic as it does in other phycids (Phycis and Urophycis). In Merlucciidae and Gadidae, the sphenotic is relatively small, largely overlapped by the frontal, and has its lateral margin confluent with that of the pterotic. In Muraenolepis the dorsal surface of the sphenotic is reduced to such an extent that only a small area is exposed and its lateral margin is confluent with that of the frontal (Fig. 2B). In its overall morphology, disposition and size it resembles the sphenotic of the Gadidae more than that of any other family. The parasphenoid In Muraenoleflzs a thick median lamina rises from the orbital part of the parasphenoid to meet the posterior extension of the mesethmoid (Fig. 3A). Many gadoid taxa have a median ridge along the parasphenoid but it is never produced to the extent it is in Muramolepis, although in some specimens of Trisopterus luscur there is a prominent eminence of the medial ridge. Rhinonemus is exceptional in having a well-developed medial process (Fig. 7) which antcrodorsally meets the posteriorly extended medial walls of the lateral ethmoids and in this respect is unlike the condition in Muraenolepis. The median septum in Rhinonemus consists mostly of lateral ethmoid with the parasphenoid contributing to only a third of its length. Furthermore, the parasphenoid part of the septum is broad and appears to consist of two lateral laminae which have become joined in the midline. Paired, parallel and outwardly directed parasphenoid laminae which contact the bases of the lateral ethmoids are common to most gadoids. I n Bathygadidae and Steindachneriidae (and macrouroids) the lateral laminae are elevated anteriorly, a feature which Okamura (1989) had incorrectly interpreted as the same condition as in Muramolepis (see above). In the majority of taxa the parasphenoid is broad and flat ventrally. Exceptions are Manuronus (Macruronidae) and P h y i s (Phycidae) in which the keel of the parasphenoid is narrow and near-circular in crosssection, the parallel laminae having curved toward the midline to form a tubelike structure. 82 G. J . HOiYES In Muraenolepis the medial walls of the lateral ethmoids are separated from one another across the midline by the ethmoid cartilage, and contact with the parasphenoid is with that bone’s anterior margin. I n the majority of gadoids, the lateral ethmoids are also separated across the midline by ethmoid cartilage but contact with the parasphenoid is between their medial walls and the parasphenoid lateral laminae. T h e extent of this contact is variable, being narrow in Melanonidae, Steindachneriidae, Moridae, Macruronidae and more extensive in Phycidae, Lotidae, Ranicipitidae and Gadidae. I n the latter, there is often considerable posterior extension of the medial wall. I n the phycids Phycis and Urophycis the lateral ethmoids contact one another across the midline. I n Rhinonemus, as described above, although the lateral ethmoids extend posteriorly along the parasphenoid laminae and closely approach one another, they are at no point in contact. Supraoccipital crest According to Svetovidov ( 1939) Muraenolepis has a prominent supraoccipital crest. This is not so; there is only a slight median ridge which posteriorly slopes below the level of the cranial roof (Fig. 4A). The majority of gadoids possess a well-developed supraoccipital crest but in Ranicipitidae and Lotidae (Lota, Molva, Brosme) the crest is reduced to a posterior eminence, level with the cranial roof. I n the physids Rhinonemus and Gaidropsarus the crest is represented by a slight ridge, while the Bregmacerotidae lack any form of crest. Optic fenestra Gadoid fishes, in common with macrouroids, possess a single common opening for the passage of rectus muscles, cranial nerves and vessels which pierce the membranous covering, T h e width of the opening is variable, as is the disposition of the bones surrounding it. Muraenolepis has an extensive optic foramen which deeply indents the border of the prootic, occupying half its length (Fig. 4A). I n the Moridae, Euclichthyidae, Melanonidae, Bathygadidae and Macruronidae the anterior part of the prootic is directed medially so that the border of the foramen is orientated transversely (Fig. 8A-D) . Morids also have a transversely directed parasphenoid ascending process which, together with a descending pterosphenoid lamina provides two openings, the outer allowing the passage of the trigeminal trunk, the medial for the optic nerve and rectus muscles (Fig. 8A & Howes, 1989). Among Phycidae (Phycis, Urophycis), and in Ranicipitidae, the notch lies entirely in the sagittal plane, although the prootic wall is thickened posterior to the notch (Fig. 8E-F). Other ‘higher’ gadoids (including all other phycids) have a thin-walled prootic with the trigeminal notch more deeply incised into its anterior border, as in Muraenolepis (Fig. 8G-I). I n some of these taxa there occurs a further derived condition in which a ventral lamina extends from the pterosphenoid to contact the ascending process of the parasphenoid so forming, as in the Moridae (see above), two separate foramina for the optic and trigeminal nerves. The taxa having this condition are Merlucciidae, Lotidae (Molva, Fig. 6C, and Brosme; in Lota the pterosphenoid lamina remains separated from the parasphenoid) and Gadidae ( Trisopterus, Fig. 8, and Micromesistius), a3 SYNCRANIAL OSTEOLOGY IB F D E F G H I A Figure 8. The size and shape of the optic-trigerninal opening and the dispositions of the surrounding bones. A, Moridae (Lepidion). B, Euclichthyidae. C, Bathygadidae (Bathygadur). D, Macruronidae. E, Phycidae (Phycis). F, Ranicipitidae. G , Gadidae (IriSopterus). H, Gadidae (Gadus). I, Muraenolepididae. Each divided diagram shows lateral (left) and anterior (right) views. Pterosphenoid- shaded; autosphenotic- hatched; optic-trigeminal opening- black; mt- membranous tissue; other abbreviations as in previous figures. Suspensorium and palato-pteryp.oid bones (Fig. 9) The hyomandibular is longer than broad and is orientated vertically; its anterodorsal portion is formed into a large cranial articulatory condyle, its posteroventral margin contacts the preoperculum and anteroventrally the metapterygoid and symplectic. The lateral face of the hyomandibular stem is partially covered by the large, triangular ‘symplectic process’ of the preoperculum (spp, Fig. 9C). The fan-shaped body of the quadrate is widely separated from its ventroposterior spine which lies along the lower limb of the preoperculum. The quadrate condyle is broad and anteroposteriorly expanded. The symplectic is a large, broadly triangular bone with a concave posterior margin, widely separated from the hyomandibular stem and preopercular limb. Dorsally, the symplectic bears a medial process which lies inside the anteroventral lamina of the hyomandibular; the posterior margin of the symplectic contacts the vertical limb of the preoperculum. The metapterygoid is small and oblong and in the specimen illustrated is G . J. HOWES 84 hY0 Figurc 9. .Lfum~nolepismil-rops. suspensorial and palaroptcrygoid bones. In A, medial, B, anterior and C. lateral views. Abbrcviations: ecp- rctopterygoid, rnp- entopterygoid, hyo- hyomandibular, met- metapterygoid, pal- palatine, po- preopercular, qu- quadrate, spp- symplectic process of prcopercular. sy- synplectic. fragmented into two parts (Fig. 9A). The anterior border of the bone is convex and widely separated from the entoplevgoid; ventrally, contact is with both the quadrate and symplectic. The entopterygoid is shallow and is in the form of a thin lamina along the posterodorsal border of the ectoptevgoid which is a wellossified plank-like bone. The palatine is a hammer-head shaped bone, its medial head being ligamentously attached to the medial part of the lateral ethmoid; the area between the two heads is tied by connective tissue to the lateral ethmoid wing, and the lateral head o\,erlies the maxilla (Howes, 1987: fig. 4). The palatine sits in a right-angled terminal recess of the ectopterygoid (Fig. 9B). Dericed suspen rorial features The suspensorium and palatopterygoid series of Muraenolepis are distinctly modified in many respects when compared with those of other gadoids, viz: palatine double-headed; entopterygoid reduced; symplectic enlarged; quadrate with large space between its body and posterior spine; large interosseous space between symplectic and quadrate-preopercular. The uniqueness of the muraenolepidid palatine morphology has been mentioned elsewhere I Howes, 1987). According to Dunn ( 1989), a rounded lower process of the gadoid hyomandibular is the derived condition, compared Mith the plesiomorphic pointed process. .+furaenolepis has an almost intermediate condition in that the process is elongate and truncate. Another feature of the h) omandibular, which Dunn regarded as derived, is a horizontal preopercular process, a f'eature lacking in LCluraenolejis. Such a process is present in the SYNCRANIAL OSTEOLOGY 85 Figure 10. Suspensorial and palatopterygoid bones. A, Lotidae (Lotu), medial view. B, Ranicipitidae (Ranzceps), lateral view. C, Phycidae (Rhinonamus), medial view. D, Gadidae (Gudur), medial view. Abbreviations: hpo- hyomandibular preopercular process; others as in previous figures. Entopterygoid is shaded, metapterygoid vertically hatched. Bregmacerotidae, Phycidae and Lotidae, but in the Ranicipitidae, Gadidae and Merlucciidae it is ventrally directed. A small entopterygoid is not unique amongst gadoids and occurs in the lotids Lota and Brosme (Fig. 10A); in the phycids Rhinonemus, where it is a shallow lamina running the length of the ectopterygoid (Fig. lOC), and Gaidropsarus, where it is a small oblong lamina lying above and separated from the central part of the ectopterygoid; in Merlucciidae and most Gadidae it is shallow and confined above the anterior half of the ectopterygoid (Figs IOD, 11D). In the Bregmacerotidae there is only a single pterygoid element, which from its disposition between the quadrate border and palatine, identifies it as an ectopterygoid (Fig. 11B); the entopterygoid thus appears to be lost. The gadoid symplectic is variable in shape. In Bathygadidae, Melanonidae, Steindachneriidae, Moridae, Euclichthyidae, Bregmacerotidae and Macruronidae, it has a thin, narrow stem and a broad triangular head. In other taxa the bone varies from triangular to oblong and, apart from Muraenolepis, is deepest in the phycid Gaidropsarus (where there is no space between it and the preopercular limb). A large symplectic-preopercular space occurs in the Macruronidae, Lotidae, Merlucciidae and Gadidae; such is not the case in other families. The ectopterygoid of most gadoids is of similar shape to that in Muraenolepis although in most taxa the posteroventral tip of the bone extends down the anterior border of the quadrate, a feature absent in Muraenolepis. The largest ecto- and entopterygoids occur in Bathygadidae, Melanonidae, Moridae, Steindachneriidae, Euclichthyidae (Fig. 1 1A) and Macruronidae, taxa which also possess the largest metapterygoids. Although in comparison to these taxa, the metapterygoid of Muraenolepis is small it is equally proportioned to that found G . ,J. HOM’ES 86 Figure 1 I . Suspcnsorial and palatopterygoid bones. A, Euclichthyidae (Euclichthys) medial view. B, Bregmauerotidae (Rregmarerm) medial view. C. Phycidae (Phyris) lateral view. D, Merlucciidae Merluccius~ lateral view. in the Phycidae, Lotidae, Ranicipitidae, Gadidae and Merlucciidae. Only in the Bregmacerotidae is it so reduced that i t fails to contact either the hyomandibular or symplectic (Fig. 1 1B ) , Dunn ( 1989) considers the sharply pointed, triangular metapterygoid of Gadus (Fig. 1OD) to represent the plesiomorphic gadoid condition. My out-group comparisons suggest the contrary, that this morphotype is derived and can be considered as synapomorphic for the Gadidae. The gadoid quadrate usually bears a posteroventral spine but it is not separated, or only slightly, from the body of the bone; Muraenolepis, with the Lotidae and Gadidae, exhibits the widest separation (cf. Figs 9, 10A,D). Dunn (1989) considered the length of the posterior process of the quadrate but not its degree of separation from the body of the bone. His analysis suggested that the more elongate spine is the more derived. In this respect Muraenolepis has the plesiomorphic state. T h e anteroposterior expansion of the quadrate condyle in Muraenolepis is doubtless correlated with the medially extended articular surface on the anguloarticular. Contact between the hyomandibular and preopercular is variable amongst gadoids. The central part of the preopercular extends anterodorsally and somewhat medially and is referred to as the ‘symplectic process’ since it normally articulates with the head of the symplectic or with the symplectic cartilage. I n the Bathygadidae, Steindachneriidae, Melanonidae and most of the Moridae, the symplectic process contacts the symplectic head and the ventral border of the hyomandibular stem; the upright limb of the preopercular lies close to or contacts the posterior edge of the hyomandibular. This condition is also present SYNCRANIAL OSTEOLOGY a7 in the Lotidae and Ranicipitidae except that the upright preopercular limb is widely separated from the posterior margin of the hyomandibular and contacts a posteroventral process of that bone (hyomandibular-preopercular process, Fig. lOA,B). I n both the Macruronidae and Merlucciidae the symplectic process contacts the symplectic cartilage but in the former the preopercular margin overlaps the posterior border of the hyomandibular, whereas in the latter the bones are widely separated (Fig. 11D & Howes, 1989). I n the Bregmacerotidae the preapercular lacks a symplectic process and contacts the entire posterior margin of the hyomandibular (Fig. 1 1B). Surface-to-surface contact between the preopercular symplectic process and the lateral face of the hyomandibular occurs in some of the Moridae, Phycidae and all of the Gadidae. As in the Muraenolepididae the symplectic process covers the posteroventral face of the hyomandibular (Figs 10D, 11C). The dissociation of the preopercular symplectic process with the symplectic cartilage and its repositioning against the lateral face of the hyomandibular mark this as a derived feature. Dunn (1989) considered the shape and length of the symplectic process; rounded and low he identified as plesiomorphic, and pointed and high as derived; Muraenolepis has the latter. Gill (1890) drew attention to the modifications of the suspensorium of Raniceps (one of the set of features which caused him to establish a separate family for the genus). H e noted particularly the forward inclination of the hyomandibular, that its articulation was on a line in front of the quadrate, and that the ". . . relation of the hyomandibular, quadrate, symplectic and metapterygoid to each other and the neighbouring bones" differed from that of other gadoids known to him (Fig. 10B). In fact the relationships of these bones, one to the other, is no different from that in any other gadoid (as noted above). It is, however, unique for the hyomandibular to have a forward inclination, which as Gill surmised, is probably a correlate of the short cranium of Raniceps. Whereas in other gadoids the contact zone between the quadrate and metapterygoid is angled posteroventrally, that in Muraenolepis is horizontal; furthermore, the synchondrosis is so weak that it is in the nature of a hinge-joint allowing the pterygoid bones to be directed medially (Fig. 9C). Such a possibility is not available to other gadoids where the posteroventral stem of the ectopterygoid abuts against the quadrate margin and offers resistance to any medial flexure. It is apparent that the wide, posterior bifurcation of the quadrate has been an essential attribute in bringing about the level metapterygoidquadrate junction. Jaws (Fig. 12) The premaxilla has high ascending and articular processes, the latter being somewhat shorter than the former (Fig. 12A). The postmaxillary process is shallow and the distal tip of the maxilla is blunt and outwardly curved. There are three rows of conical teeth anteriorly, those in the outer row being the larger; posteriorly, the outer row teeth diminish in size and the row terminates prior to the tip of the jaw so that the inner rows take on the function of the outer row (Fig. 12B). The maxilla has a slight concavity just posterior to its head; this is the 88 G. J. HOM’ES Figure 12. Jaw bones of Muraenolepis microps. A-F, Premaxilla in A, lateral and B, ventral views. C, Maxilla in lateral view. Lower jaw in D, dorsal, E, medial and F, ventral views. G, dentigerous area of dentary of M . marmorafus (drawn from alcohol preserved specimen, BMNH 1986.7.10: I . ) . .4bbreviations: aa- anguloarticular, cm- coronomeckelian bone, de- dentary, pma- premaxillary articular process, pmp- postmaxillary process of the premaxilla, pms- ascending process of the premaxilla, ra- retroarticular. insertion point of adductor muscle A1 b. Posteriorly, the bone is deep, with a slightly convex dorsal margin and straight posterior margin (Fig. 12C). The dentary is deep with a blunt anterior tip, and has a narrow dentigerous surface; ventrally it has 7-8 large openings to the sensory canal (Fig. 12D). There is an inner row of small, thin, conical teeth, interspersed with a few larger teeth. An outer row of small, straight unicuspid teeth ends halfway along the dentigerous surface; anteriorly, near the symphysis, the two rows become irregular and broaden into a patch of moderately-sized, recurved teeth (Fig. 12D). Muraenolepis marmoratus has a similar arrangement of dentary teeth to that of M . microps except that there is a regular arrangement of large teeth in the inner row and the toothed area is broader (Fig. 12G). The anguloarticular is long with a blunt posterior process and a relatively short articular surface (for the quadrate) which extends onto its medial face (Fig. 12D,E). The retroarticular is an elongate element confined to the ventral surface of the anguloarticular; the coronomeckelian bone is large and boot-shaped (Fig. 12E). In comparison with other gadoids the upper jaw is rather short and the maxilla has a truncate rather than an indented or produced margin. The premaxilla lacks a posterior indentation, that is ‘gadoid notch’ of Rosen & Patterson (1969: 401). Howes (1988) drew attention to the various types of upper jaw morphology and noted the correlation between the development of the labial ligament and the stepped form of the premaxilla. According to Dunn SYNCRANIAL OSTEOLOGY 89 ( 1989) a rounded postmaxillary process is the plesiomorphic gadoid condition whereas the triangular one (of Gadidae) represents the derived condition. Likewise, Dunn considers a truncate to bifurcate posterior maxillary border as plesiomorphic and recognizes as secondarily derived the truncate condition exemplified by Gadiculus. The shortened form of the muraenolepidid maxilla appears to be secondarily derived. The posterior convex dorsal elevation of the bone appears to correspond with the anterior or central elevation in the Gadidae (Fig. 12C). DISCUSSION The phylogenetic position of Muraenolepididae The ‘higher gadoids’ is a group recognized by Howes (1988, 1989, 1990) as comprising the Macruronidae, Bregmacerotidae, Phycidae, Lotidae, Ranicipitidae, Gadidae, Merlucciidae and Muraenolepididae, and is diagnosed primarily on the basis of fused upper hypurals; complex division and medial shift of the adductor mandibulae musculature; well-developed facet on the interoperculum for the articulation of the interhyal, and a firm articulation between the 1st infraorbital (lachrymal) and the posterior or ventral surface of the lateral ethmoid wing (in Muraenolepis articulation is within the ventral surface). The Muraenolepididae clearly belongs to this category from its possession of these characters, but its sister-group relationship is ambiguous and it was placed with the Ranicipitidae, Phycidae, Gadidae and Merlucciidae as part of an unresolved polychotomy (Fig. 13; Howes, 1990). Muraenolepis has several derived syncranial osteological features, four of which are autapomorphic, namely, absence of frontal ventral laminae; median parasphenoid septum contacting a posterior extension of the mesethmoid; cranial part of the parasphenoid dorsally extended, and a double-headed palatine. Two unique non-osteological features are the anguillid-like scales (Fig. 2B) and high number of pectoral fin rays (viz 36-38). Four derived cranial characters are shared with a broad spectrum of taxa, namely reduced supraoccipital crest, with the Bregmacerotidae, Ranicipitidae, Lotidae and Phycidae (Rhinonemus and Gaidropsarus) ; reduced sphenotic, with the Gadidae and Merlucciidae; thin-walled prootic with deeply indented optictrigeminal notch, with the Phycidae (Rhinonemus, Ciliata, Gadiropsarus, ? Motella), Merlucciidae and Gadidae; and fused frontals with the Phycidae (Phycis, Urophycis, Rhinonemus) and some of the Gadidae. A general reduction in size of pterygoid bones is a feature shared by all ‘higher’ gadoids apart from the Macruronidae and a further reduction in entopterygoid size occurs in the Lotidae, Phycidae, Gadidae and Merlucciidae. Muraenolepis, Raniceps, Phycis and Urophycis share a direct attachment to the urohyal of the rectus communis muscle, a feature interpreted as derived by Howes (1988: 47). I had reported earlier (Howes, 1988) the shared specialization between Muraenolepis, Phycis, Urophycis and Lota of an opercular insertion of part of the epaxialis musculature. My interpretation of the epaxial insertion in Muraenolepis is, however, somewhat flawed since examination of subsequent specimens has shown that the so-called epaxialis could well be a levator operculi with a shifted origin. Of two specimens, one has the levator operculi as a single 90 Figurr 13. (:ladogram olgadoid phylogenctir relationships (after Howcs, 1990). In addition to those c.haracters diagnosing 'higher' gadoids, \.iz: fused upper h!-purals; medial shift of adductor mondihulap musc ulature; interopcrcula~facet for iiiwrhyal; 1st infiaorhital articulation with posteroventral area of'lateral rthmoid wins. should be added reduction in sizc of pterygoid bones and, long preopercular procc'\s of' the hyomandibular text -1'Iir reinterpretation oC the epaxial muscle-operculum <liars( tcr p. 89: and the presence of an extensive optic-trigrminai notrh in the prootic repositions t l i c lluraeriolcpididat.~~~~pididae b e t w r n the 'Phycidae' r Rhtnonrtnus. Goidrop.rarn.t. Cilioto, Moidlnr-Lotidae, t. and l l r t l u c & l a e+ Gadidaz. element whereas in the other, it is derived, the outer part stemming partly from the posttemporal. Concerning Phycis, Uropfvcis and Lota, however, I believe my earlier interpretation of opercular insertion of epaxialis to be correct. Rosen & Patterson i1969: 428) noted that muraenolepidids have a "normal SYNCRANIAL OSTEOLOGY 91 gadoid skull roof and occipital region and that the caudal skeleton, although reduced, has the characteristic gadoid dorsoventral asymmetry”. According to these authors, the jaw muscles “closely resemble those of ophidioids and some macrouroids”. Although the cranial morphology is characteristic of ‘higher’ gadoids as is that of the caudal skeleton, I would disagree that there is any resemblance between the jaw muscles of Muraenolepis and ophidioids and macrouroids. Indeed, the musculature of Muraenolepis is typical of a ‘higher’ gadoid (Howes, 1988). The caudal fin skeleton ofMuraenolepis is simple in that hypurals 3-5 are fused with the terminal centrum; hypurals 1 and 2 are also fused together and there is a single large (possibly two fused) epural(s). Fahay & Markle (1984) identified the muraenolepidid caudal skeleton as the ‘primitive state’, but it more likely represents a derived condition in the continuity of its dorsal and anal rays with the caudal rays. Indeed, the caudal skeleton does not differ significantly from that of Macruronus, or for that matter any other ‘higher’ gadoid (see Howes, 1990 for discussion). The muraenolepidids resemble the phycids in having a small dorsal fin. In Muraenolepis the fin comprises a single ray supported by a feeble radial but with a full complement of arrector and depressor muscles and innervated by a spinal nerve branch. I n contrast, the first dorsal fin of the phycids Mutella, Ciliata, Rhinonemus and Gaidropsarus is a specialized ‘vibratile’ structure lying in a sensory groove and innervated by the facial nerve (see Whitear & Kotrschal, 1988 for description and references). Markle ( 1989) relates the Muraenolepididae to the Bregmacerotidae on the basis of loss of ligamentous connection between the 1st and 2nd epibranchials and in turn, this ‘clade’ to the Phycidae through the possession of an elongate first dorsal fin. From these descriptions, however, it is clear that there is a marked distinction in dorsal fin structure between muraenolepidids and phycids. The Muraenolepididae possess a simple swimbladder lacking the anterior diverticula of phycids and gadids. There is no intimate connection between the swimbladder wall and the skull as in Phycis and Urophycis where the swimbladder lies adnate to an intercalar-exoccipital fenestra (similar to the condition in the Moridae). Dunn (1989) has presented a series of osteological characters in the Gadidae; his polarity assignments are derived through out-group comparisons with the Ranicipitidae, Phycidae (Rhinonemus), Lotidae and Merlucciidae. Of the 28 characters presented, Muraenolepis scores 16 as plesiomorphic (state l ) , and 9 as derived (8, state2; 1, state3). Two characters (14 & 15, concerning the preopercular process of the hyomandibular) are not applicable. The remaining character (No. 17, shape of the metapterygoid), is scored in Muraenulepis as derived. However, I would question Dunn’s interpretation of this character polarity since I believe the rounded or truncated metapterygoid is plesiomorphic and the narrow, triangular element, confined to the Gadidae, is synapomorphic for that family (see discussion above, p. 89). Using Dunn’s character analysis, Muraenolepis scores higher than Merluccius (i.e. is more derived). The irregular distribution of the above detailed apomorphies amongst ‘higher’ gadoids does not provide a succinct dichotomous pattern of taxa. Which of these features are homoplastic has yet to be determined through further analysis of other character suites. Nonetheless, it is clear that the Muraenolepididae shares a 92 G . J . HOWES higher number of apomorphies with the Gadidae, Merlucciidae and some of the Phycidae than with other taxa, and may stand a5 the sister group to this assemblage. Howes ( 1990) sought to demonstrate the sister-group relationship of the Gadidae with the Merlucciidae, which were aligned in a polychotomy with the Lotidae and Phycidae. One of the features used to unite this assemblage as the marked attrition of the prootic border, but this also occurs in Muraenolepididae (see above). The Phycidae were recognized there as paraphyletic, a view upheld by this study. T o determine more precisely the phylogenetic relationships of the Muraenolepididae it will first be necessary to identify those monophyletic lineages within the 'Phycidae' and determine their relationships within this complex of gadoids. Biogeography Those localities sampled indicate that the distribution of the Muraenolepididae is circum-Antarctic, a possibility considered by De Witt (1971). Captures of Muraenolepis are from the Falklands Plateau, South Georgia and South Sandwich Islands (i.e. between the subtropical and Antarctic convergences) ; Kerguelen Island and the Antarctic shelf area of Balleny Islands, Ross Sea and off Enderby Land, and also from Indian Ocean sub-Antarctic islands and seamounts; recorded by Duhamel & Hureau (1982)) Tomo & Hureau (1985), Hulley et al. (1989) and Gon & Klages (1988) (Fig. 14). Among gadoids, the Muraenolepididae is unique in that it is the only family endemic to the Antarctic zone, i.e. south of the Antarctic convergence. Only one other is endemic to the Southern Ocean (the New Zealand area, or Campbell Plateau), namely, the Euclichthyidae. Other families of gadoids represented in the southern temperate regions are the Moridae, Macruronidae, Phycidae, Gadidae and Merlucciidae, all of which have a bipolar or bitropical distribution. The Moridae contain eight bipolar genera (Mora, Halargyreus, Lotella, Physiculus, Laemonema, Letidion, Anlimora and Gadella), a pattern also exhibited at species level by Antimora rostrata, Mora moro and Halargyreus johnsonii. The Phycidae has one (bipolar) genus, Urophycis, with two or three species on the Brazilian shelf, between 35"-40" S. The Gadidae has two (bipolar) genera: (1) Gaidropsarus, with three species, G. nouaezelandiae (Campbell Plateau), G . capensis (South African Cape) and G. insularurn (S.W. African coast, Tristan d a Cunha, Gough and St Paul Islands; see Penrith, 1967 for details of species of their distribution); and (2) Micromesistius with a single species, M . australis, recognized by Inada & Nakumura (1975) as comprising two subspecies respectively on the Campbell Plateau and Patagonian shelf (further specimens reported by these authors were collected off Chile). The single merlucciid genus Merluccius has four species which occur throughout the tropics (Inada, 1981). The Macruronidae has three species of Macruronus which occur off Patagonia, South African Cape and the Campbell Plateau (Howes, 1990). Before considering in more detail the distribution of these Southern Ocean families, a brief synopsis of gadoid distribution follows. The Bathygadidae, like the Bregmacerotidae, have a tropical-subtropical distribution; the former between 50"N and 50"s (in the Atlantic), the latter within 40"N and 40"s.T h e Moridae has a cosmopolitan distribution; of its eighteen genera, seven are confined to the Southern Ocean. The Melanonidae SYNCRANIAL OSTEOLOGY 93 Figurc 14. Distribution of Muracnolcpididac; bascd on data from Dc Wit1 (1971), Duhamel & Hurcau (1982), Tom0 & Hurcau (1985) and Hullcy rr of. (1989). are recorded as cosmopolitan, occurring antiequatorially in the Atlantic with one species (Melanonu gmcilzs) being circum-Antarctic. Species also occur in the western Pacific (recorded by Parin, 1978, from off the Solomon Islands) and eastern Pacific (off California and British Columbia; Berry & Perkins, 1966; Peden, 1974). Restricted distributions are exhibited by the Steindachneriidae (tropical western Atlantic) and Ranicipitidae (European coasts). The Lotidae are restricted to the temperate Northern Hemisphere and have an amphiAtlantic distribution. Svetovidov (1948) attempted to reconcile the distribution of the Gadidae (his concept of which included the Phycidae, Lotidae, Merlucciidae and Ranicipitidae, thus corresponding to my ‘higher’ gadoids) with the zoogeographical subregions as proposed by Ortman (1896) and later modified by Eckman (1953). Like Eckman (1953: 144), he was unable to recognize precise boundaries of such regions, only ‘transitional areas’. For example, Svetovidov could not recognize a Mediterranean gadoid group since most species (all but two) occur (and spawn) in the Atlantic. He noted, however, that unlike most Atlantic-Mediterranean fishes, gadoids show no relationship with the Indo-west Pacific. Svetovidov (1953) noted “The peculiar distribution of the Gadidae is contrary to the distribution of other (cold-water) fishes.. .”. The pecularity of this distribution is that the majority of gadoid families are more speciose in the north 94 L J HOI\ tS Atlantic than in the north Pacific. T h e absence of a n Atlantic-Indo-west Pacific link in ‘higher’ gadoids appeared to Svetovidov (1953) to be correlated with climatic factors, but it seems also, and perhaps primarily, to be linked to their association with the continental shelf and upper part of the continental slope. There are no continental shelf gadoids in the Indian or western Pacific oceans i the ‘lower’ gadoids in these oceans, ZYC bathygadids and morids are associated with the upper and lower continental slopes). Inada (1981) explains the absence of merlucciids (the onlv shelf gadoids crossing the tropics) in the Indian Ocean as being due to ecological factors, i.e. high temperature of the Mozambique current and he proposes an elaborate dispersal pattern to account for their present-da) distribution). Inada’s concept of dispersal appears to follow the classical ideas of Darlington (1957) whereby the assumed advanced species lie at the centre of origin, and furthermore, he has viewed this dispersal against the background of present-day ecological factors. Hocutt i 1987) has argued that vicariant events responsible for Indian Ocean endemism bere the drift of India and its eventual amalgamation with the Asian marqin. Likewise, corrclated e\ents such as the closure of the Tethys sea in the Palaeocenr and the opening of the Atlantic might correlate with the phylogeny of the shelf-dwelling (‘higher’) gadoids in that region. Fedotov & Bannikov i 1989) point to the origin of distinctive gadoid genera in late Oligocene/early Miocene, re5ulting from “hydrological changes in the seas of the later Tethys”. 1 hc complcx geomorphologies resulting from the opening of the ‘Atlantic corridor’ (various papers in Soug) & Rodgers, 1988) undoubtedly produced continental slope and shelf areas offering a wide variety of new habitats (I’alentine 8r Moores, 1970, 1972). Antarctica became the remaining segment of Gondwanaland after the others had nioked northward ‘misted by rapid sea-floor spreading of the Indian Ocean ‘Norton & Sclater, 19791. AuTtralia broke awa) from Antarctica in the Eocene, followed by South America during the Oligocene (30 Myr B P ) . Any groups of fishes associated with Antarctica during this period and which have survived to the present would be expected to reflect in their distribution the disposition of these continental associations. ‘Those gadoids with a sub-Antarctic distribution certainl) do so. The GaidropJarzrJ species occur around the eastern rim of Ytarctica, outside the con\ ergence; the .Merluccius, Mzcromesislius and Macruronus species indicate a former range linhing the Campbell Plateau and Patagonia. There $\as certainl) a continuous link between South America and the Antarctic Peninsula in the late Slesozoic (which incorporated the Burdwood Bank, South Georgia and the South Shetlands, but not the Falkland Islands) and which was disrupted during the early Tertiar) b) the eastward bending of the Scotia Arc, displacing South Georgia to its present position (Dalziel & Elliot, 1973). T h e link or track from Patagonia to the Campbell Plateau is one evidenced by other groups of organisms (Fig. 15; see also Humphries et al., 1986). The presence of the same Gaidroprarus species on southern Atlantic and Indian Ocean islands and the South Afi-ican we5t coast needs confirmation, nevertheless, the presence of the same genus suggests a former]) close geographic association of these areas. Tristan and Cough islands lie at the end of a chain with Tristan occupying the location of a hot-spot (Heezen el al., 1972; Morgan, 1981). Likewise, Prince Edward and St Pauls Islands occupy a hot-spot track. ,4lthough some might explain the distribution of Gaidropsarus in these Southern r . ‘1 SYNCRANIAL OSTEOLOGY 95 Figure 15. Distribution of ‘higher’ gadoid taxa in the Southerh Ocean. Hatched area off Atlantic South America- Urop/ycis (Phycidae); dots- G ~ i d r o p s ~ r u s(Gadidae); squares- Micromesittius (Gadidae); shaded areas- Macrumnus (Macruronidae); dashed lines- extent of Mel-luccius (Merlucciidae). The connecting lines indicate the number of shared genera. Map drawn on Hammer transverse elliptical equal-area projection. ~ Ocean localities as the result of dispersal, it apptars that there may be a more general explanation involving island tracking. Rather than viewing distributions of gadoids on a map of the present-day world it seems more promising to view them c)n maps showing the assumed dispositions of the continents during the Palaeoclene-Neogene periods when the Atlantic Ocean was forming. O n this basis the Merlucciidae, for example, are seen to have dispersed through vicariant processed of continental rifting (Fig. 16), rather than via the elaborate routes postulated qy Inada (1981). The present-day circum-Antarctic distributioln of the Muraenolepididae is most likely a reflection of a former Gondwqnic coastal distribution. The Muraenolepididae is an example of one of tHose groups (others being the Nototheniiformes and some Ophidiiformes) which have accommodated relatively rapidly to changing conditions of 1 the progressive isolation of Antarctica. As pointed out by Grande & Eqstman (1986), the increasing distance, from South America, Africa and Austrblia, depth and coldness of the Southern Ocean and development of the circumtAntarctic circulation resulting in the Antarctic Convergence made coloniza$on from the north virtually impossible. The only migration route would have been the Scotia Ridge G . J. HOWES 96 I Figure 16. Hypothesized phylogeny of the Merlucciidae in the Southern Ocean (hatching) with respect to the separation of oceanic margins. A, Arrangement of the southern continents in early Palaeogene. B, In late Palaeogene. C, In mid-Neogene. Abbreviations: Af- Africa, Ant- Antarctic, Aus- Australia, Ker- Kerguelen Plateau, Mel- Melanesia, NZ- New Zealand. Jagged lines between the continents indicate rift-transform systems. Base map modified from Heezen et al. (1972). connecting Patagonia with the western Antarctic Peninsula. De Witt (1971), on the other hand, has proposed that the Scotia Ridge has provided an eastward dispersal route and notes that Muraenolepis possibly had a n Antarctic origin and extended its range to South America. However, in the same paper, he contradicts Andriashev’s ( 1965) opinion that Muraenolepis is an element of the original Antarctic fish fauna and proposes that it is a “more recent invader” since it occurs primarily on the continental slope and not on the inner regions of 97 SYNCRANIAL OSTEOLOGY the continental shelf. The occurrence of Muraenolepis around su b-Antarctic islands does not in itself lend support to the idea of invasion since these islands, for the most part, appear to have had past continental associations with Antarctica (Morgan, 1981). Miller’s (1987) view of Antarctic fish distribution combines some of the factors discussed above, together with others, namely, passive dispersal by tectonic plate movements; ocean currents and modification to habitats by glaciation and sealevel changes. My views concerning the ‘origin’ of Antarctic fishes are more in accord with those of Regan (1914) and Andriashev (1965) both of whom recognized ‘original’ Antarctic faunal elements. However, ‘Antarctica’ must now be viewed as part of Gondwanaland. Contrary to Regan’s (1914) opinion that the distinctiveness of the Antarctic ichthyofauna provides no evidence for a former continuity of continents, as noted by Miller (1987) it is that very distinctiveness that does so since the evolution of that fauna has been ‘advanced’ by the Continent’s rapid isolation. At present no more an explicit sister-group relationship for the Muraenolepidid can be established than the ‘Phycidae’-Gadidae Merlucciidae, groups which are predominantly northern in their distribution. The hierarchical pattern of this relationship (i.e. family level) suggests an earlier bipolar dichotomy than those dichotomies exhibited only by genera among the other families. On the other hand, the bipolarity might have been contemporaneous in all the groups and thus the morphological divergence exhibited between Muraenolepididae and other taxa could be a reflection of its extreme isolation. The bipolarity exhibited by the gadoid taxa detailed above suggests a formerly continuous latitudinal distribution (the Merlucciidae of course still retain such a range). Eckman’s (1953) now classic discussions of bipolarity posit that the majority of marine bipolar distributions occurred during the greater part of the Tertiary period. Eckman (1953: 257-8) assumes the cause of bipolarity to have been climatic change but, at least for higher taxonomic categories, concedes to “geographical changes” which “influenced the course of the cold ocean currents or there may have been other causes”. T o Eckman, bipolarity was a modification of Theel’s (1885) and Ortman’s (1896) dispersal theories which declared that from centres of origin at high latitudes, taxa crossed the equatorial region wherein they later became extinct, leaving bipolar relicts. In the more recent literature, Briggs (1987) supports Theel’s theory whilst arguing against a vicariance approach to bipolarity (advocated by Nelson, 1985 and White, 1986). Briggs, however, maintains that the youngest genera are found in the tropics, whilst the “higher the latitude, the more ancient they become”. Nelson (1985) considers that bipolarity, at least as manifest in the Pacific, could be the result of earth expansion or fragmentation of a former southern continent and dispersal of those fragments to the now Pacific margins. Concerning Atlantic bipolarity (of engraulids), Nelson has no firm explanation other than those “possibilities for dispersal of Pacific fishes into the Atlantic by way of a northern route”. Few gadoid taxa occur in the tropics but those that do such as Melanonidae, Steindachneriidae, Bregmacerotidae and Merlucciidae contradict Briggs’s (1987) notion that high latitude equals primitive taxa since the former two families are considered relatively plesiomorphic (Howes, 1989, 1990). Further contradiction comes from macrouroid distribution, which is latitudinally wide + G J HOiVES 98 for several genera. T h e absence of gadids and phycids in the tropical Atlantic is probably due more to the lack of suitable broad continental shelves which presumably once existed in that region during the time of continental apposition but have now vanished owing the geomorphological processes which have shaped the Ocean. T h e present eastern Pacific coastal distribution of gadids and merlucciids appears to me to be part of a once continuous distribution around the respective southern and northern coasts of the North and South American continents. In summary, the circum-Antarctic distribution of the Muraenolepididae is thought not to be the result of dispersal from South America but the retention of a distribution along the continental shelves of the formerly closely apposed South America and Antarctica. T h e bipolarity exhibited by some gadoid taxa is seen as the result of equatorial disruption of formerly latitudinally continuous distributions. T h e extant Southern Ocean distributions of Macruronidae, Merlucciidae, some Phycidae and Gadidae reflect former continuous continental margins linking the Campbell Plateau with Patagonia. .~C:RSO\VLEL)GE:Xl~s IS I am most grateful t o my colleagues, P. H. Greenwood, N. R. Merrett, C. Patterson and A, C. IYheeler, and an anonymous referee for their critical comments and helpful suggestions for improving the manuscript. My thanks also to P. Campbell, 0.A. Crimmen arid hl. Holloway for their assistance in preparing and radiographing specimens. REFEREX ANDRIASHE\7. .A. P.. 1965. .A gctirral re\ ic\v 01' thr .Antarctic lish l i u n a . .Lfunographine Bzolqirae, 15: 19 1-5.50. BERRY. F. H. & PERKISS. H. C.. 1966. S u n c y of pelagic fishes of thc California currcmt arca. FiJhery Bulidin, 1 'nited States Fish and I I .i/dliJe .Sur BRIGGS. J. C . . 1987. Antitropicality and \ fir' <OoO/O$p. .?6 1-31; 206-207. BLRCHE'I'I', M .S., SAYERS. P. J.. YOR'I'H, .A. i V . & IVHI'I'E, M.G.,1983. Some biological aspects ofthe nearshore fish populations ;it South Gcorgia. Hrrlish h t n r r t i r Surr,pv Bulletin, 59: 63--74. COHEN. D. 51.. 1984. Gadiformes: o\crview. I n H. G. .\.loser. el a / . iEds). Ontngeny and Sjufernalics of Fi,rhe~: 259-265. Spet.ial puhlication So. 1 of the American Society of Ichthyologists and Herpetologists. DALLIEI,. 1. \V, D. & EI,LI(YI. 1). H.. 1973. 'l'he Scoria Arc and Antarctic margin. In A. F,. M. Nairn & F. G . Strhli , E d s , 7 h e Orenn Hnsim nnd .\Jnryin.\, 1. The South .4tlan/ii: 1 7 1 -246. New York & London: 10s.P. J.. 1957. <oo,qeographr: H. H.. 1971. L'unstrzl /he Geogiaphirol Dis/ributiotr u/.lnirnals. New York: iL'ilry. and D e e p - I l b f e i Rerithii. FishuJ oj'the .Intarc-iic. Antarctic map hlio serirs No. 15. Amrrican Gcoqraphical Socirty 10 p p + 15 pls. DI.7H.%XlEL. G . & HURE.AC. J.-C.. 1982. Donni.cs coniplhenraires siir I'ichtyofaunc des lles Australes qaises. Cybiurn. 6 ( 1 1 : 65-80. J . R.. 1089. A provisional phylogeny ofgadid lishcs bawd on adult arid early life-history characters. In 1). 11. Cohcn Ed.:, Papers O N /he Srstrrnnliis nf (;nd~forr I.'rshe.rt 109-23.5. Science series No. 32. Natural History Museum of Los Angelrs County. the Sea. London: Siciswick & Jacksoti. EC:KML1AN, S., 1953. <oogeogmaphy E F R E M E S R O . V. N., 1983. Description of rggs and la]-\ ar 01' .tfirmeno/epi.r rnicropJ Lonnberg Muraenolepididaci from the Scoria Sea. ]oirrnnl oJIrhthrolnp. -?? i I ) : 139-142. F.XH.IY. X l . P. & XIARKL.E, D. F.. 1981. Gadiformes: Drvelopment and relationships. In H. C. Mosrr ut a/. : Eds Onlogei!y and . ~ w / m a t i ~oJ. s Fi.iheJ: '265-283. Special Publication No. I , Amcric;rn Society of lchtliyulogists and Herpetologists. FELXYfOV. V. F. & BANSIKO\'. A. F.. 1989. O n phylogenrtic relationships of bssil Gadidar. I n D. M. Cohcn I Ed.'. Popen on the .Sv,r/enin/irs/ I / GaAi/brni Fishr.1: 187- 195. Science scrirs No. 32. Natural History Musrurn of Los Angeles County. G I L L T.. 1890. On the family R;iuicipitidae. Proseeding, (?/the Iirrted Slates .\htiona/ .\fuseurn, 13: 235-238. I ~ SYNCRANIAL OSTEOLOGY 99 GON, O., 1988. The fishes collected during the South African SIBEX I & I1 expeditions to the Indian Ocean sector of the Southern Ocean (60-66 S, 48-64 E). South African Journal of Antarctic Research, 18 ( 2 ) : 55-70. GON, 0. & KLAGES, N. ’I’ W., . 1988. The marine fish fauna of the sub-Antarrtic Prince Edward Island. South African Journal of Antarctic Research, I8 (2): 32-54. GRANDE, L. & EASTMAN, J. T., 1986. A review of Antarctic ichthyofauna in the light of new fossil discoveries. Palaeontology, 29 ( I ) : 113-137. GUNTHER, A,, 1880. A report on the shore fishes procured during the voyage of HMS ‘Challenger’ in thr years 1873-1876. Challenger Reports, I (6): 82 pp. HEEZEN, B. C., THARP, M. & BENTLEY, C. R., 1972. Morphology of the Earth in the Antarctic and mbAntarctic. Antarctic map folio series No. 16, American Geographical Society. 16 p p f 8 pls. , C. H., 1987. Evolution of the Indian Ocean and the drift of India: a vicariant event. Hydrobiologia, 150: 203-223. HOWES, G. J., 1987. The palatine bone and its associations in gadoid fishes. Journal of Fish Biologv, 31: 625-637. HOWES, G . J., 1988. The cranial muscles and ligaments of macrouroid and gadoid fishes; funrtional, ecological and phylogenetic inferences. Bulletin of the British Museum (Natural History), zoology, 54 ( I ) : 1-62. HOWES, G. J., 1989. Phylogenetic relationships of macrouroid and gadoid fishes based on cranial myology and arthrology. In D. M. Cohen (Ed.) Papers on the Systematics of Gadiform Fishes: 113-128. Science series No. 32, Natural History Museum of Los Angeles County. HOWES, G. J., 1990. Anatomy, phylogeny and taxonomy of the gadoid fish genus Macruronus Giinther, 1873. Bulletin of the British Museum (Natural History), ~ o o l o g y ,in press. HULLEY, P. A,, CAM-US, P. & DUHAMEL, G., 1989. Ichthyological results of cruise MDL42/SlBEX 11; Part I. Fishes from RMT-8 stations, with additional records of lanternfishes (Myctophidae: Ostcichthvcs) from the Indian and Atlantic sectors of the Southern Ocean. Cybium, 13 ( l j : 83-99. HUMPHRIES, C. J., COX, J. M. & NIELSEN, E. S., 1986. Nothojagus and its parasites: a rladistir approach to coevolution. In A. R. Stone & D. L. Hawksworth (Eds), Coevolulion and Systematics: 55-76. Oxford: Clarendon Press. INADA, T., 1981. Studies on the merlucciid fishes. Bulletin ofthe Far Seas Fisheries Research Laboratory, 16: 1-1 72. INADA, T. & NAKUMURA, I., 1975. A comparative study of two populations of the gadoid fishes Micromesistius australis from the New Zealand and Patagonian-Falklands regions. Bulletin of the Far ,Seas Fisheries Research Laboratory, 13: 1-26. MARKLE, D. F., 1989. Aspects of character homology and phylogeny of the Gadiformes. I n D. M. Cohen (Ed.), Papers on the Qstematics of Gadform Fishes: 59-88. Science series No. 32. Natural History Muscum of Los Angeles County. MILLER, R. G., 1987. Origins and pathways possible for the fishes of the Antarctic Ocean. Proceedings V Congress European Ichthyology, Stockholm 1985: 373-380. MORGAN, W. J., 1981. Hotspot tracks and the opening of the Atlantic and Indian Oceans. In C. Emiliani (Ed.). ,, The Oceanic Lithosbhere. The Sea. 1: 443487. MUJIB, K. A,, 1967. The cranial osteology of the Gadidae. Journal of the Fisheries Research Board of Canada, 24 ( S j : 1315-1375. NELSON, G., 1985. A decade of challenge, the future of biogeography. I n A. E. Leviton & M. L. Aldrich (Eds), Plate Tectonics and Biogeography. Special issue of Earth Sciences History. Journal of the History of the Earth Sciences Society, 4 (2): 187-196. NORMAN, J. R., 1937. Fishes. Reports B.A.N.2. Antarctic Research Expedition, Ser. B., I ( 2 ) : 49-88. NORTON, I. 0. & SCLATER, J. G., 1979. A model for the evolution of the Indian Ocean and the break-up of Gondwanaland. Journal of Geophysical Research, 84: 6803-6830. OKAMURA, O., 1989. Relationships of the suborder Macrouroidei and related groups, with comments on Merlucciidae and Steindachneria. I n D. M. Cohen (Ed.), Papers on the Systematics of Gadiform Fishes: 129-142. Science series No. 32, Natural History Museum of Los Angeles County. ORTMAN, A,, 1896. Grundzuge der marinen Tiergeographie. Jena. PARIN, N. V., 1978. New records of midwater fishes from off New Guinea and Tonga islands with descriptions of new species of Eustomias (family Melanostomidaej and Benthodesmus (family Trirhiuridar). Trudy Instituta Okeanologii, I l l : 15C168. PAULIN, C. D., 1983. A revision of the family Moridae (Pisces: Anacanthinij within the New Zedland region. Records of the National Museum of N e w zealand, 2 ( 9 ) : 81-126. PEDEN, A. C., 1974. Rare fishes including first records of thirteen species from British Columbia. Syesis, 7; 47-62. PENRITH, M. J., 1967. The fishes of Tristan d a Cunha, Gough Island and the Vema seamount. Annals ofthe South African Museum, 48 (22j: 523-548. REGAN, C. T., 1903. O n the systematic position and classification of the gadoid or anacanthine fishes. Annals and Magazine of Natural History, (7j 11: 459-466. KEGAN, C. T., 1914. The Antarctic fishes of the Scottish National Antarrtic Expedition. Transactions o f t h e Royal Society of Edinburgh, 59 ( 2 ) , 2: 22S292. ROSEN, D. E. & PATTERSON, C., 1969. The structure and relationships of the paracanthopterygian fishes. Bulletin of the American Museum of Natural History, 14 ( 3 ) : 357-474. \ 100 G . J. HOWES SOGGY, J. & RODGERS, J. (Eds), 1988. The west African connection: evolution of the central Atlantic ocean and its continental margins. Journal of African Earth Sciences, 7 (2): 516pp. SVETOVIDOV, A. N., 1939. Contribution to the comparative anatomy of the genus Mumenolepis (Gadiformes). Compte Rendus de I'AradPmie des Sciences de I'lJ.R.S.S., 23 (6): 583-585. SVETOVIDOV, A. N., 1948. Treskoobraznye (Gadiformes), Fauna SSSR. <oologicheskii Instztut Akademzi Nauk SSSR ( n . 8 . ) 34. Ryby (Fishes) 9, 4: 1-222. [English translation, 1962. Jerusalem: Israel Program for Scientific Translations, 304 pp.] SVETOVIDOV, A. N., 1953 (1956). Morphological principles of the classification of the Gadidae. Proceedings XIV International Congress of <oologv, Copenhagen: 5355540. THEEL, H., 1885. Report on the Holothuroidea Pt. 11. Challenger Reports, <oology, 14: 1-290. TOMO, A. & HUREAU, J. C., Muraenolepididae. I n W. Fischer & J.-C. Hureau (Eds), F A 0 Species Identijcation Sheets for Fishery Purposes. Southern Ocean (Fishing Areas 48, 58 and 881 [CCAMLR Convention Area), 2: 306-3 15. VALENTINE, J . W. & MOORES, E. M., 1970. Plate tectonic regulation of faunal diversity and sea level: a model. Nature, 228: 657-659. VALENTINE, J. U'. & MOORES, E. M.,1972. Global tectonics and the fossil record. Journal of'Geology, 80: 167-1 84. WHITE, B. W.,1986. The isthmian link, antitropicality and American biogeography: Distributional history of the Atherinopsinae (Pisces: Atherinidae). Systematic /=0010gy, 35: 176-194. WHITEAR, M. & KOTRSCHAL, K., 1988. The chemosensory anterior dorsal fin in rocklings (Gazdropsarus and Ciliala, Teleostei, Gadidae): activity, fine structure and innervation. Journal of <oology, London, 216 ( 2 ) : 339-366.