• 1848. Congrus Richardson (partim), Voy. Ereb. Terr., Fish., p. 109.
• 1856. Congermuraena Kaup (partim), Cat. Apod. Fish., p. 105.
• 1859. Gnathophis Kaup, Abh. Naturw. Hamburg, 4 (2): 7.
• 1925. Rhynchocymba Jordan & Hubbs, Mem. Carneg. Mus., 10 (2): 194.
• 1936. Poutawa Griffin, Trans. roy. Soc. N.Z., 66: 16.

Six species are here referred to the genus Gnathophis Kaup, of which two have been each further subdivided into two subspecies. Three of these species are known from the shallow waters of Japan and China: the type species G. heterognathus (Bleeker, 1859), G. nystromi (Jordan & Snyder, 1901) and G. xenicus (Matsubara & Ochiai, 1951). As indicated earlier in this account, the dorsal fin originating over the middle of the pectoral, the reduced upper labial flange, the three prong-like bones in the upper lip, the relatively short, robust tail, the molariform vomerine teeth and the scroll-like or flap-like anterior nostril are easily recognisable features of these gnathophid eels. Schultz et al. (1953, p. 70) give measurements taken from Bleeker's description of the type specimen of G. heterognathus and these are as follows (in per cent total length) : head 17.0, snout 4.3, eye 2.9, cleft of mouth 5.3, pectoral 4.7, predorsal 20.7, preanal 37.3. Jordan & Snyder (1901, p. 853) give similar proportions for the type specimen of G. nystromi: head 16.1, snout 4.6, eye 3.1, cleft of mouth 5.3, pectoral 5.3, predorsal ca. 18.5, preanal 37.5. The number of vertebrae in the latter species is 114–132 (Matsubara & Ochiai, 1951, p. 6) but unfortunately no vertebral count is available for G. heterognathus. Nevertheless, considering the relative abundance of G. nystromi in Japanese waters and that G. heterognathus is still only known from the one specimen, the very close similarity of the two species in proportional measurements indicates that G. nystromi is very probably a synonym of page 19 G. heterognathus. A complete re-examination of Bleeker's type is necessary to confirm or refute this suggestion.

G. nystromi has been divided into two subspecies, G. nystromi nystromi (Jordan & Snyder, 1901) and G. nystromi ginanago (Asano, 1958). The former has fewer lateral line pores before the vent, fewer vertebrae, shorter trunk, smaller eye, sharper snout, longer vomerine band and dark pigmented alimentary canal. The third Japanese species, G. xenicus, which Matsubara & Ochiai (1951, p. 8) originally described as a subspecies of G. nystromi is distinguished from G. heterognathus-G. nystromi in having a shorter head and a much greater number of vertebrae, 152–154.

One other species has been described from the North Pacific; this is G. catalinensis (Wade, 1946) known from one individual collected off Santa Catalina Island, California. G. catalinensis has the many features characteristic of gnathophid eels, but Wade's illustrations (pl. 26, figs. 1–3, p. 209) show a simple tubular anterior nostril, an inconsistent feature which should be checked in view of the close similarity of this species in other characters to Japanese species. Wade gives the following proportions for the type of G. catalinensis: head 16.8, snout 4.6, eye 3.2, cleft of mouth 5.3, pectoral 6.0, predorsal ca. 18.0, preanal 37.0, vertebrae 132. Wade distinguishes his species from G. nystromi in having "a longer, more pointed vomerine band of teeth, smaller mouth, shorter eye, less projecting snout and a more anteriorly inserted dorsal" but these figures, compared with those given from Jordan & Snyder's account of G. nystromi do not support the distinction. In addition, the vomerine band of teeth in G. nystromi is subject to such variation (Matsubara & Ochiai, 1951, fig. 1, A-E) as to include the condition shown for G. catalinensis. The relationships of G. heterognathus-G. nystromi-G. catalinensis are clearly in need of further clarification.

Two species of Gnathophis are present in the Australasian region of which the only one known previous to this account is G. habenatus (Richardson, 1848) having 120–129 vertebrae, scroll-like anterior nostrils, and a small, triangular premaxillary-ethmoid patch of teeth which does not extend conspicuously in front of the maxillary patches. G. habenatus inhabits the shallow waters of harbours and river mouths from New Zealand to Lord Howe Island, New South Wales, Victoria, Tasmania, West Australia and possibly St. Paul's Island, South Indian Ocean, and is redescribed in this paper.

Ramsay & Ogilby (1888, p. 1022) described a small eel taken from the Parramatta River, Sydney, as Congromuraena longicauda which has been accepted of recent years to be a subspecies of G. habenatus. Dr J. C. Yaldwyn, of the Australian Museum, Sydney, has kindly re-examined the type specimen of G. habenatus longicaudatus (Aust. Mus. regd. no. 1.1618) and informs me that, although the specimen is in poor condition with the caudal tip detached, a total of not more than 123 vertebrae are present. Two more specimens, Aust. Mus. regd. nos. IA.4984 (Karuah River Mouth, N.S.W., 14/6/1913) and IA.3630 (Lord Howe Island) have vertebrae numbering 128 and ca. 129 respectively. These numbers agree well with the 120–127 of New Zealand specimens. The single character which has been used to separate the Australian subspecies is the greater length of the tail, so that the preanal length is about 38.5% of total length in the type specimen compared with an average of 42.9% in 23 specimens of G. habenatus from New Zealand. The Australian subspecies thus appears to be valid, although there is some little evidence to support the belief that differing conditions or duration of larval development in G. habenatus may affect the ultimate length of the tail in the adult. This matter will be discussed in greater detail below.

During the preliminary sorting of over 200 gnathophid leptocephali described elsewhere in this paper two groups of larvae emerged from the sorting, one having 116–131 myomeres, the other with 134–150 myomeres, both of which were otherwise closely similar. The group with the lower number of segments were readily identified with Gnathophis habenatus (120–129 vertebrae), but the second group was referable to no known adult. The abundance of larvae in this group suggested that the adult had been previously confused with G. habenatus. This subsequently proved to be the case as a close examination of the available New Zealand specimens of G. habenatus revealed the second species with vertebrae numbering 139–147, a longer tail, minute epidermal papillae, anterior nostril with a simple free flap and a round premaxillary-ethmoid patch of teeth extending conspicuously in advance of the maxillary patches. This species is known from rather deeper water than is G. habenatus, from New Zealand as far north as the Kermadec Islands.

The number of myomeres in leptocephali identified with Congermuraena mystax from the Atlantic (132–147) is almost identical with the known range for the second species in New Zealand waters (134–150). Having regard only to this and to the proposal that C. mystax is a Gnathophis would suggest that C. mystax may prove to be a species of the widest distribution, including New Zealand. Until the generic relationships of the Atlantic species are clarified, the present author feels justified in separating the second New Zealand species of Gnathophis from Congermuraena mystax.

Gnathophis habenatus habenatus (Richardson, 1848). Text-fig. 1, A-K.

• 1848. Congrus habenatus Richardson, Voy. Ereb. Terr., Fish., p. 109, pl. 50, figs. 1–5.
• 1856. Congermuraena habenata (Richardson). Kaup, Cat. Apod. Fish., p. 105.
• 1870. Congromuraena habenata (Richardson). Gunther, Cat. Fish. Brit. Mus., 8: 42.
• 1872. Congromuraena habentata (Richardson). Hutton, Cat. Fish. N.Z., p. 66.
• 1898. Congermuraena habenata (Richardson). Ogilby, Proc. Linn. Soc. N.S.W., 23: 285.
• 1901. Congermuraena habenata (Richardson). Jordan & Snyder, Proc. U.S. nat. Mus., 23: 851.
• 1911. Congermuraena habenata (Richardson). Waite, Rec. Canterbury (N.Z.) Mus., 1 (3): 163–164.
• 1927. Gnathophis habenata (Richardson). Phillipps, N.Z. Mar. Dept. Fish. Bull., 1: 17.
• 1936. Poutawa habenata (Richardson). Griffin, Trans, roy. Soc. N.Z., 66: 16, pl. 7, fig. 2, text-figs. 3-4.
• 1956. Poutawa habenata (Richardson). Whitley, in Graham, Treas. N.Z. Fish., app., p. 401.
• 1961. Gnathophis habenata (Richardson). Moreland, in Doogue & Moreland, N.Z. Sea Anglers' Guide, p. 199, fig.

Material Examined. One specimen: gravid female, 275.5mm total length; Shelly Bay, Wellington Harbour; beach seine; 12/10/52; Dominion Museum No. 1160. Three specimens: 329.9mm, 333.0mm and 338.0mm (female); York Bay, Wellington Harbour; beach seine; 5/3/53; Dom. Mus. No. 1241. One specimen: 396.0mm; Lowry Bay, Wellington Harbour; beach seine; 5/4/53; Dom. Mus. No. 1262. Three specimens: 429.5mm (gravid female), 403.7mm (gravid female) and 382.8mm (mature male); York Bay, Wellington Harbour; beach seine; 16/4/53; Dom. Mus. Nos. 1270(a), 1270(b) and 1270(c) respectively. One specimen: 184.0mm; Shelly Bay, Wellington Harbour; beach seine; 28/6/53; Dom. Mus. No. 1335. One specimen: gravid female, 361.2mm; York Bay, Wellington Harbour; beach seine; 2/3/54; Dom. Mus. No. 1484. One specimen: female, 278.5mm; power house intake, Evans Bay, Wellington Harbour; 12/8/54; Dom. Mus. No. 1561. One specimen: 389.0mm; Shelly Bay, Wellington Harbour; 7/1/56; Dom. Mus. No, 1860. One specimen; 142.7mm; off Wanganui page 21

Text-fig. 1.—Gnathophis habenatus habenatus (Richardson, 1848), 429.5mm total length, female, Dom. Mus. No. 1270 (a), York Bay, Wellington Harbour, 16/4/53. Fig. A—Lateral view, arrows indicate origin of dorsal, anterior margin of vent and origin of anal. Fig. B—Dorsal view. Fig. C—Lateral view of head to indicate throat folds and prominent myomeres. Fig. D—Dorsal view of head. Fig. E—Ventral view of snout to show anterior nostrils and pores. Fig. F—Pattern of upper (left) and lower (right) teeth. Fig. G—Otolith, internal surface (upper) and dorsal surface (lower). Fig. H—Left preorbital bones, lateral view. Fig. I—Urohyal bone, left lateral view (upper) and ventral view (lower). Fig. J—Glossohyal bone, left lateral view (upper) and ventral view (lower). Fig. K—Caudal skeleton (184.0mm, t.1., Dom. Mus. No. 1335).

page 22 in 40 fathoms by trawler Admiral; May, 1957; Dom. Mus. No. 2327. One specimen: gravid female, 269.0mm; off Oamaru by trawler Orion; 1960; VUW Zoo. Dept. Coll. One specimen: gravid female, 268.5mm; Shelly Bay, Wellington Harbour; beach seine; April, 1961; VUW Zoo. Dept. Coll. Eight specimens: five gravid females and three mature males, 289.8mm-373.9mm; Shelly Bay, Wellington Harbour; beach seine; 26/6/62; VUW Zoo. Dept. Coll. One skull prepared from a specimen seined from York Bay, Wellington Harbour; February, 1954; Dom. Mus. No. 2495.

Description. Proportional measurements (in per cent of total length) and counts from the above 23 specimens: total length 142.7mm-429.5mm, standard 98.4–99.2, head 14.8–19.3, snout 3.5–5.1, eye 2.5–4.8, interorbital 1.9–3.4, cleft of mouth 4.6–6.9, postorbital 8.5–9.5, branchial aperture 1.9–2.9, branchial interspace 3.3–6.4, pectoral 3.5–6.9, snout-vent 40.9–45.5, preanal 41.6–46.6, predorsal 17.2–20.0, depth just before eye 3.7–5.1, depth at pectoral origin 5.3–7.3, depth at anal origin 4.9–6.5, depth at midpoint of caudal region 3.8–4.7. Pectoral rays 12–13, dorsal rays before level of vent 43–55, dorsal rays 180–224, anal rays 129–160, caudal rays 5+4, lateral line pores before level of vent 33–38, vertebrae 120–127.

A small but robust eel with a body which is rounded in section especially along the trunk, with prominent myomeres and with the vent placed only a little in advance of the midpoint of the body. The head is depressed, with a sharp snout, underslung mouth, large eye, and the throat is thrown into many folds of skin. The fins are conspicuous and the caudal is noticeably rounded and abbreviated. Live specimens are very active animals, escaping easily from all but the finest mesh, and are strikingly coloured sliver to silver-bronze on the anterior part of the body with a brilliant silver iris.

Head sharply conical, depressed anteriorly, relatively long, contained about 6.5 times in total length but poorly differentiated from the trunk, the throat swollen in some individuals but nearly always thrown into narrow longitudinal folds which curve up around branchial region; snout sharp, bony, its length contained about 3.5 times in head, always projecting in advance of tip of lower jaw by just less than diameter of pupil; lower jaw rounded in ventral view, rather shovel-like, slender; mouth subterminal, oblique, with cleft of mouth extending to level of anterior margin of pupil and contained about 3.0 in head; upper lip weak, but limits of maxilla clearly distinguishable on side of snout; lower lip thick, fleshy, overlapping side of lower jaw; tongue well-developed, pointed, extending forwards to a point level with anterior margin of posterior nostril.

Teeth comparatively small, but larger on the shaft of the vomer than on the other dentigerous bones, sharply conical and closely packed except those on the shaft of the vomer where they are rounded and more widely spaced. Maxillary teeth in a broad band of about four longitudinal rows anteriorly of about 38 teeth each with the teeth slightly smaller anteriorly than posteriorly and more rounded medially than laterally. Premaxillary-ethmoid teeth small, about 20 in number, in an inconspicuous, broadly triangular patch which does not project much in advance of maxillary bands. Vomerine teeth clearly separated from those of the premaxillary-ethmoid patch and in a long, cigar-shaped band extending to a level two-thirds of the way along the maxillary bands; the teeth small and acute on the head of the vomer but larger, rounded and molariform on the shaft and in about five longitudinal rows of about 15 teeth at the broadest point of the patch. Teeth on the dentary similar in number, size and distribution to those of the maxillary bands.

Anterior nostril small, subterminal, placed on the ventral surface of snout just in advance of premaxillary-ethmoid patch of teeth. On superficial examination the nostril appears to be subtubular with the tube directed anteroventrally to ventrally; on closer examination the tube is found to have a scroll-like rim with an anterior free flap (see text-fig. 1, E). Posterior nostril inconspicuous, slit-like, placed just in advance of eye, level with horizontal diameter and with a simple, raised rim but no external tube. Eye subcircular to oval, large, contained about three times in postorbital; fleshy interorbital narrow, less than eye. Branchial aperture lateral, having a concave free edge, oblique, with lower extremity a little posterior to upper, which is slightly below middle of pectoral base. Vent usually protruding into profile, especially in gravid females.

Dorsal and anal fins delicate in life, although fleshy in preservative; caudal fin with the rays relatively short and the tip of fin rounded so that the fin forms a burrowing tip to the body. Pectoral fin elongate-oval in shape with its posterior margin rounded, originating clearly above midlateral level with the base oblique and the fin directed more or less posteriorly; a little longer than snout.

Lateral line conspicuous but scarcely raised above surface of trunk, originating very high on head, descending to meet midlateral level just posterior to level of vent but fading out at extreme posterior tip of tail. Cephalic sensory pores few, restricted to one occipital pore which occasionally has a small pore lateral to it, two or three postorbital pores, two inconspicuous but larger pores on the ventral aspect of opercular region, four to five minute pores along dentary and two pairs of small pores on ventral aspect of snout; of these the first pore lies medially to the base of the anterior nostril, the second in advance of it on the extreme ventral tip of snout. A pair of inconspicuous slit-like openings of the mucous cavities of the snout are present on the anterodorsal tip of snout. There are no discernible minute papillae on the head or above lateral line.

Colour in life olive-grey posterior to vent and above lateral line from level of pectoral base to vent; side of trunk conspicuously silvery-olive but becoming creamy-white ventrally; opercular region bright silver to silver-bronze; branchiostegal region pink; iris very brightly silver tinged with gold with crescentic dorsal and ventral black patches; dorsal and posterior part of anal fins with a relatively narrow black edging which becomes wider around caudal fin; caudal pink in some specimens; fin-rays pigmented with minute black spots. The bright silver colour and the prominent myomeres showing through the skin are unmistakeable features of this eel.

Remarks. Griffin (1936, p. 17) gives comparative measurements of the preanal length in "four specimens, three of which came from the Manukau Harbour, Auckland, and one from Sydney, N.S. Wales". These preanal lengths are 36%, 38%, 39% and 42% of total length, but unfortunately Griffin does not say which, of these belonged to the Sydney specimen. In view of the observed preanal length in other Australian examples it is unlikely to be the last-mentioned, and this is undoubtedly a true Gnathophis habenatus habenatus. Although Griffin gives a vertebral count of 44 + 78 for the New Zealand species this has probably been taken from Waite (1911, p. 164) so that the identity of the three other specimens from Manukau and Sydney cannot easily be determined. The shorter preanal length in these specimens suggests that (a) they belong to the Australian subspecies, G. habenatus longicaudatus, most unlikely for the Manukau examples, or (b) they have been confused with the new species of Gnathophis described below in which the vertebrae number 139–147, but in which the preanal length is short, ranging from 37.8% to 41.0% of total length, more likely for the Manukau examples especially since the new species is more abundant in the northern waters.

Griffin's figures of the tooth pattern in his specimens are not helpful, since the number of teeth on the vomer is far greater than ever observed by the present author; in addition, the toothed area on the dentary as shown is much wider than usual and there is an abnormal amount of variation in the size of teeth on various parts of the dentigerous surfaces.

G. habenatus habenatus is occasionally captured in shallow trawls along the New Zealand coast, mainly in Cook Strait and south of it, but its more frequent locality is Wellington Harbour. Indeed, although Richardson gives as the type locality "Cook's Strait" it is likely that his specimen came from Wellington Harbour where H.M.S. Erebus and H.M.S. Terror spent some days. From March to June the species is abundant in Wellington Harbour although it can only be captured using a very fine mesh seine net and sufficient hauling rope to enable the net to work over the green-grey mud some distance from the shore in which the eels are possibly buried during the day. These eels are more easily captured in the early or late evenings when the tide is at its lowest. An indication of the abundance of this species in Wellington Harbour during the early winter is shown by one haul on June 26, 1962, when about 150–200 individuals appeared in the seine net. Stomachs of specimens of G. habenatus habenatus collected from this area contained crustaceans and polychaetes but these were always too far digested to be identified.

The majority of specimens collected from Wellington Harbour over a period of years appear to be close to spawning, the females having relatively large eggs of about 1.0mm or more diameter, the males having well-developed testes. Some of the females on capture were so gravid that the abdomen was greatly swollen and slight pressure caused large numbers of ova to be extruded (pers. comm. by Mr J. M. Moreland, Dominion Museum, Wellington). In eight specimens collected in June, 1962, five were gravid females (total lengths 310.0mm-373.9mm) and three were mature males (289.8mm-320.0mm) while in the total number of 23 specimens examined 18 could be sexed, of which 12 were females and six were males. Males of this species tend to be a little smaller and more slender than females in the same catch but there are otherwise no discernible differences in morphological characters between the sexes, although the males are a little brighter in colour than the females. However, if a long series of individuals were examined morphological differences would possibly show up, as has already been shown for Conger myriaster (Brevoort) and Muraenesox cinereus (Forskål) by Takai (1959, p. 547).

The abundance of near-spawning individuals of both sexes of G. habenatus habenatus in Wellington Harbour may indicate that the eels are assembling preparatory to breeding. The species appears to be absent from the harbour during middle spring to late summer. No young stages of any eels have been collected from Wellington Harbour plankton despite regular weekly collections over two years from an area adjacent to Evans Bay where many of the adult eels have been seined. It is now known from the study of a large collection of leptocephali of G. habenatus from eastern Australian waters that this species must spawn some distance off the New South Wales coast over the continental shelf with the adults living in the estuaries and the shallow coastal waters. This pattern may also be true of the New Zealand population and indeed the fully-gravid condition of the specimens collected in Wellington Harbour suggests that the spawning area cannot be far off the New Zealand coast.

Only five leptocephali of G. habenatus have been taken from New Zealand, one on the west coast of the South Island, the remainder from the Cook Strait area. All of these are advanced in development and three appear to be under-going page 25 regressive metamorphosis. They are all large leptocephali and thus may have travelled some distance from their spawning place. As yet, no small leptocephali are known from New Zealand waters and thus it is not possible at the present state of knowledge to consider the location of a spawning area for this species in our waters. However, a large number of leptocephali of G. habenatus have been made available to the author in this study and an account of the larval development of this species is given below. As indicated above, G. habenatus is spawning off the New South Wales coast and the leptocephali are transported to the Tasmanian and New Caledonia areas. It is possible that the larvae of this species are transported further to the New Zealand region giving rise to what is known as G. habenatus habenatus. The shorter tail in this subspecies may be due to a prolonged pelagic life of the leptocephali. There is always the remote possibility that spawning of New Zealand individuals does not take place, or if it does the young perish at an early age due to inadequate or unfavourable environmental conditions.