The Osteology of the Congrid Eel Gorgasia punctata and the Relationships of the Heterocongrinae'

RICHARD H. ROSENBLATT2

ABSTRACT : Th e osteology of Gorgasia punctata is described, figured , and com­ pared with that of other congrids. Gorgasia is clearly referable to the subfamily Heterocongrinae. The heterocongrin es agree with the Congridae in several important featu res, and do not differ in fund amental respects. Th erefore, the group is recog­ nized as a subfamily of the Congridae. Gorgasia is the most primitive hetero­ congrine, and agrees with the anagoine congrids in having a lateral ethmoid process. Because of this and other similarities it is suggested that the Anagoinae and Hete rocongrin ae arose from a common stem. The genus X arifania was erected on the erroneous basis of lack of caudal rays. It is synonymized with Tuenioconger,

THE CONGRID EEL Gorgasia ptl11ctata was placed eastern Pacific heterocongrines, including Gor­ by its describers in the little known apodal gasia punctata. Because of previous uncertainties family Derichthyidae (Meek and H ildebrand, regarding the exact position of Gorgasia in eel 1923) . Bohlke (1951) was the first to point classification, a study of the osteology of this out that the affinities of Gorgasia were with species seemed worthwhile. Heteroconger. Gosline (1952) provisionally placed Heteroconger and Gorgasia in the Con­ ACKNOWLEDGM ENTS gridae. Bohlke ( 1957) described the osteology The figures were drawn under my supervision of the related N ystactichthys halis and placed by E. David Lane. Part of the cost of this the eels allied to Heteroconger in the Congridae, investigation was defrayed by a grant to the but considered them to constitute the distinct Institute of Fisheries, University of British subfamily Heterocongrinae. Columbia, from the H . R. McMillan Expedition­ Bohlke considered Gorgasia to be the most ary Fund. Th e specimen of Tnenioconger hassi primitive genus of the Heterocongrin ae on the used in this study was collected as a part of the basis of its more complete complement of head U. S. Biological Program, Ind ian Ocean Ex­ pores, its uncoalesced upper labial flanges pedition. (called by him the "free edge of lip"), a well­ developed pectoral fin, and its unspecialized MATERIALS AND METHODS maxillary dentition. He considered Gorgasia to be specialized, however, in that the caudal rays The two adults of Gorgasi« pnnctat« were are much reduced and covered by thick skin. taken from a series of 69 (SI062-720-26A, Internal characters were not considered, since Bahia Magdalena, Baja California, Mexico) . the only complete specimen then available was Th ese were bone-stained with alizarin and the holotype. cleared in glycerine. The neurocranium was dis­ Recent collections made by personnel of the sected out of one specimen, and the drawings Scripps Institution of Oceanography have were made from the dried preparation. In amassed rich material of several species of addition I have utilized single-stained and cleared specimens of Taeniocongel' diglleti Chabanaud (SI065-278, Gulf of Californ ia) , 1 Contribution from the Scripps Institution of Oceanography, University of California, San Diego, T. berrei W ade (51061-261, Gulf of Califor­ and the Institute of Fisheries, University of British nia), T . hassi3 (Klausewitz and Eibl-Eibesfeldt) Columbia. Manuscript received December 22, 1965. 2 Scripps Institution of Oceanography, University 3 Taenioconger hassi was originally described in of California , San Diego, La Jolla, California . the genus X arifania, of which it is the type species 91 92 PACIFIC SCIENCE, Vol. XXI, January 1967

(unnumbered, D 'Arros Island, Amirantes Is- CL lands) , T. n.sp. (SI062-42, Bahia Banderas, Mexico) , and A riosoma gilberti (Ogilby) (SI062-77, Sinaloa, Mexico).

OSTEOLOGY MX PAS PP K)p SR ""O-",5,,,,m,,--~~_ A N E URO CRANIUM (Fi g. 1) : The skull is truncated posteriorly, except where the exoc­ cipital flanges break the outline. The pre­ "" ~~, ' EO maxillaries, ethmoid, and vomer are fused, with • ~ C --::r- no suggestion of articulations, such as were ~ PAS PTS $PO PRO BO reported by Bohlke (1 957) for N ystactichthys . o.s cm halis. The anterior, triangular tooth-patch may B represent the premaxillary dentition. The dorsal or ethmoid portion of the complex is very thin, and is reduced medially to a septum, so that it is shaped much like an I-beam in cross section. Laterally the ethmoid portion is expanded, and gives rise to two heavy, forward-curving pro­ c cesses. These are very similar to the "lateral ethmoid processes" reported by Asano (1962) for A nago and Alloconger. Below this are two small proj ections from the lateral face of the vomer. These vomerine processes are diffi cult to distinguish from the base of the lateral ethmoid process. They are more evident in o o.e crn Tnenioconger and, judging from Bohlke's FIG. 1. H ead skeleton of Gorgasia punctata. A , figures 3B and 3C, are developed in Nystactich­ Lateral view, includin g pectoral girdle ; B, neu rocran­ thys as well. Posteriorly on the under side of the ium, lateral view ; C, neurocranium, top view ; D, cranium, parts of the prootic and basioccipital neurocranium, bottom view. AA , Articular angular; are expanded to form a prominent auditory BO, basioccipi tal; BR, branchiostegal ray; CL, cleith­ rum ; CO, circumorbital; D, dentary; EO, exoccipital ; bulla, which contains a large otolith (p resum­ EP, epiotic ; FR, frontal; H MD; hyomandibular ; ably the sagitta). The foramen magnum is H OC, hypocoracoid; H RC, hypercoracoid ; l OP, in­ surrounded by exoccipital flanges, which grasp teropercle ; LEP, lateral ethmoid process; MX, maxil­ the first vertebra. Th e supraoccipital is well lary; N, nasal; OBS, orbitosphenoid; OP, opercle ; PAR, parietal; PAS, parasphen oid; PEV, premaxil­ developed, but does not completely separate the lary ethm ovomerine block ; POP, preopercle; PP, epiotics, which are in contact posteriorly. The palatopt erygoid ; PR, pectoral ray; PRO, proo tic ; parietals are sutured in the specimen figured, PT, pterygiophore; PTO, pterotic ; PTS, pterosp hen­ oid ; QU , quadrate; SCL, supracleithrum; S0 , supra­ occipital ; SOP, subopercle; SPO, sphen otic. (Klausewitz and Eibl-Eibesfeldt, 1959 ) . The sole distinctio n of the genus X arijania was the supposed lack of caudal rays. Th e tail-tip of X. hassi is fleshy but in another they are fused for the anterior but flexible and clearly contains well-developed cauda l one-quarter of their lengths . Th e frontals are rays. These are visible und er direct light and are ob­ completely fused, with no sign of a suture or vious when transmitted light is used. In addi tion to median ridge. There are well-developed canals the Arnirantes specimen, I have examined a para type along the lateral margins of the pterotics and of X . hassi (ANSP 94706) through the courtesy of ]. Bohlke. I can lind no other important differences, frontals, with two large foramina anteriorly, either in external morphology or osteology, between but there is no transverse canal across the X . hassi and the species of T'aeniocanger examined. frontals. Th e nominal genus X arijania is consider ed, therefore, to be a synonym of Taenioconger . SUSPENSORIUM AND JAWS (F ig. 1): As Osteology of Gorgasia p1t11ctata- RoSENBLATT 93 might be expected from the short oblique SHOULDER GIRDLE (Fig. 1): The cleithrum mouth, the suspensorium is strongly inclined and supracleithrum are well developed. The ex­ forward. Th e hyomandibular and quadrate are panded head of the supracleithrum is bifurcate massive. The palatopterygoid is developed as a in the specimen illustrated, but not in another. broad lamina, which is attached by a ligament The well-ossified hypercoracoid and hypocora­ to the vomer. The maxillary contacts the neuro­ coid are connected by cartilage. Th e four hour­ cranium at the tip of the snout. The posterior glass-shaped actinosts are small, but well ossi­ end of the maxillary is expanded, but the fied. According to Bohlke N ystactichthys has remainder is a narrow lamina. There is no no actinosts, but the species of Taenioconger pedicel anteriorly. that I have examined are like Gorgasin in this respect. OPERCU LAR SERIES ( Fig. 1) : Th e well-devel­ oped opercular bones are strongly ossified. The VERTEBRAE AND ASSOCIATED BONES (Fig. 3) : preopercle is triangular, like that of Taenio­ In one specimen the vertebrae number 144, of conger, but unlike that of N ystactichthys as which 45 precede the anus. Figure 3A repre­ illustrated by Bohlke (1957). The blocky and sents a cross section at the level of the 17th subtriangular interopercle has a pronounced vertebra ; Figure 3B illustrates the 17th to 19th anterior extension. The crescentic subopercle vertebrae in lateral view. The vertebrae anterior curves upward under the lower angle of the to the dorsal origin bear well-developed, crest­ opercle. The dorsal margin of the broadly like neural spines. The remainder of the ab­ crescentic opercle is deeply concave; its upper­ dominal vertebrae have large neural arches, but rear corner is far above the upper end of the hyomandibular.

HY OID ARCH (Fig. 2): Thi s arch consists of the unpaired glossohyal and urohyal, and paired upper hypohyal, ceratohyal, and epihyal. Th e interhyal is absent. All the branchiostegal rays are inserted on the lateral surfaces of the arch, one on the ceratohyal and seven on the epihyal. In N ystactichthys and Tnenioconger, in contrast, two branchiostegals are inserted on the cerate­ PL CE hyal. The urohyal is needle-like, with an ex­ panded and flattened anterior end. Th e dorsal ,1mm. I surface of the glossohyal is grooved. A NA 1M EN I I

GH P PL

B

FIG. 3. Vertebrae and associated bones of G Of­ FIG. 2. H yoid apparatus of Gorgasia punctata gasia punctata: A, Front view of 17th vertebra; B, from below. BR, Branchiostegal ray ; CH, ceratohyal; side view of 17th to 19th vertebrae. CE, Centr um; EH, epihyal ; GH , glossohyal; HH, hypohyal ; UH , EN, epicentral ; 1M, intermuscular; NA, neural arch; urohyal. P, parapoph ysis; PL, pleura l rib. 94 PACIFIC SCIENCE, Vol. XXI, January 1967

no neural spines. The first 4 vertebrae bear NA strongly-developed, winglike transverse pro­ cesses.These curve out and back, and each has on its posterior margin a long, thin, backward­ directed process that seems to represent a fused NA

epicentral (these structures lie on the same plane CE as the epicentrals associated with the more pos­ terior vertebrae; furthermore, epicentrals are otherwise lacking on the first 4 vertebrae). The 5th through 9th vertebrae bear epicentrals, and rr---HY weak transverse processes without backward prolongations. Bohlke mentioned no such pecu­ liar condition in N ystactichthys balis, nor can I find transverse processes on the anterior verte­ HY? brae in Tuenioconger digueti or T. berri. In T. HA hassi, however, weak transverse processes are developed on the first few vertebrae. In Gor­ gasia the transverse processes are more weakly FIG . 4. Last 3 vertebrae of Gorgasia punctata, developed posterior to the 4th vertebra, and are lateral view. AR, Anal ray ; CE, centrum ; BAS, basal not noticeable posterior to the 10th vertebra. element of pterygiophore ; CR, caudal ray; DR, dorsal ray; HY, hypural; NA, neural arch; RAD, radial The abdominal vertebrae bear strong para­ element of pterygiophore. pophyses to which, posterior to the 6th vertebra, are articulated strong pleural ribs. There is a Dorsal and ventral intramuscular bones are strong median vertical ridge on each parapo­ well developed, and associated with all but the physis. The first haemal spine appears 15 ver­ first 12 and last 4 vertebrae. Most of the intra­ tebrae behind the anal origin, and the pleural musculars are simple; one of those illustrated in ribs are present to this point. Figure 3 happens to be bifurcate. The caudal vertebrae bear transverse pro­ cesses, commencing 6 vertebrae behind the anal CAUDAL AND ASSOCIATED STRUCTURES (Fig. origin. The transverse processes regress toward 4): The tail-tip of Gorgasia is hard and the tail-tip and are no longer apparent on the pointed, with the fin rays concealed. However, 10th vertebra before the caudal. The neural the caudal skeleton is well developed and com­ arches of the caudal vertebrae are smooth until plex. As mentioned above, the neural and about the 60th postanal vertebra, which bears haemal arches and associated spines become the first neural spine, in the form of a small expanded posteriorly, and the haemal and, to projection. The neural spines persist as low a lesser extent, the neural arches become open. conical projections until 17 vertebrae from the According to the terminology of Nybelin tail-tip, behind which they become increasingly (1963), there is but one ural centrum. Fused higher and more bladelike until they assume to it is a large hypural plate, probably consisting the shape shown in Figure 3. The haemal of several fused hypurals. The structure labeled spines are also small and inconspicuous anterior HY? in Figure 4 is somewhat problematical. It to the 17th vertebra from the tail-tip. Thereafter, has a basal-lessfin ray associated with it and thus like the neural spines, they become increasingly might be termed a hypural, but from its posi­ higher and more bladelike. Shortly before the tion it is difficult to determine whether it is tail-tip the haemal spines become divided, so itself associated with the last preural centrum that the haemal arches are again open, as on or the ural centrum. Likewise, the nature of the the precaudal vertebrae (Fig. 4). dorsal element labeled NA is somewhat ambig­ Epicentrals are associated with all vertebrae uous. It might be termed an epural, but since it is except the last 10. Epipleurals appear 6 verte­ fused to the centrum, and divided anteriorly like brae behind the anus and persist until 15 verte­ the preceding neural arch, I term it a neural arch, brae before the tail-tip. despite the circumstance that it bears two "prin- Osteology of Gorgasia pm1etata-RoSE NBLATT 95 cipal" caudal rays. A strong process curves to which the palatopterygoid is ligamentously forwar d and downward from the neural arch attached-features that appear to be lacking in element. It probably serves as a muscle attach­ the Ophichthidae. ment and may be associated with tail-first dig­ The superficial similarities between Gorgasia ging in sand. An essentially similar caudal and the ophichthids are certainly parallel adap­ skeleton has been figured by Bohlke for N ystae­ tations to a similar mode of life, and the minor tiehth ys balis, and I have found th e caudal osteological similarities may be adaptations as skeleton of Taenio conger digtteti, T. berrei, T. well. hassi, and T. n.sp. to be basically the same. Although recognizing the close relationship between the two groups, Klausewitz and Eibl­ CIRCUMORBITALS AND LATERAL LINE CA­ Eibesfeldt (19 59) maintain ed the family Het­ N ALS: The circumorbital ring is complete, con­ erocongridae as distinct from the Congridae. sisting of at least five weakly-ossified and roof­ Their action was based on behavioral differ­ less bones. Apparently the small "supraorbital" ences and on bone reduction and "Fensterbil­ illustrated by Asano (1 962) for several Japa­ dung" (fenestration ?) in the skeleton. How­ nese congrids is absent. The temporal canal is ever, their illustration of the head skeleton of encased in bone in G. pm1etata only, among Xarifania h. hassi shows a well-developed skull the species examined. The lateral-line canal and well-integrated jaws, suspensorium, and along the body is contained in a series of ossi­ opercular series. On the basis of the present cles (l ateral-line scales?) . Anteriorly these are investigation there are no grounds (except that developed as unconnected but closely opposed the circumorbital series is less well developed ) tubes, which posteriorly gradually become less for the contention that the skeleton has under­ strongly ossified, so that along the midbody gone important reduction in comparison with there is an open trough consisting of a series that of the Congridae. of ossified half-rings. In the species of Taenio ­ There is, in fact, nothing in the osteology of conger examined, the lateral-line ossicles are the heterocongrines I have examined that would developed as short, widely-spaced, troughlike preclude the inclusion of the group in the Con­ ossifications. gridae. The heterocongrines possess the basic congri d characters of ankylosed frontals, for­ RELATIO NSHIPS OF THE H ETE ROCONGRINAE ward-inclined suspensorium, few and nonover­ The heterocongrines resemble the Ophichthi­ lapping branchiostegals, maxillary-ethmoid ar­ dae in several respects. In both groups the ribs ticulation near tip of snout, caudal vertebrae are laminar, and the neural spines reduced with transverse processes, skull truncate poste­ (vestigial in the ophichthids) , as are the cir­ riorly, parapophyses divided by a vertical ridge, cumorbitals. In Gorgasia and in some species and a lateral process on the vomer. of Taenioconger the caudal fin is short and the The chief osteological differences are: neural tail-tip fleshy. In all, the body is elongate and spines absent on most abdominal vertebrae (and circular in cross section. The pectoral fin, as in on most caudal vertebrae as well in Gorgasia ) ; many ophichthids, is reduced (varying from neural and haemal arches becoming high and small in Gorgasia to minu te in Nystaetiehthys bladelike near the tail-tip; urostylar vertebra and Tuenioconger to absent in H eteroeonger) . better developed, and supporting structure of In addition to these structural characters, both caudal more complex; epineurals and epipleu­ groups share the sand-dwelling habit. However, rals lost 10- 15 vertebrae before tail-tip ; cir­ the characters given by Gosline (1 951) to cumorbital series reduced and less ossified. separate the Congridae and the Ophi chthidae Stronger divergences from the basic congrid (except that it is now known that many con­ type are found in nonosteological characters. grids have an auditory bulla) serve to distin­ In most of the Congr idae the muzzle is elon­ guish Gorgasia and its allies from the ophich­ gate, and the olfactory organ is well developed, thids as well. In addition, it may be noted that with numerous lamellae. In the heterocongrines the Congri dae have the parapophyses divided the muzzle is short, the eye is relatively large, by a vertical ridge and have a vomerine process, and the olfactory rosette is much smaller, with 96 PACIFIC SCIENCE, Vol. XXI, January 1967 few lamellae (ca. 20) . Th is distinction is no gave reasons for considering Gorgasia to be in doubt correlated with a change in food habits most respects the most primitive of the Hetero­ (Klausewitz and Eibl-Eibesfeldt, 1959). Also congrinae.His conclusions are borne out in this the habit of living colonially in sand tubes is study, except for the discovery in Gorgasi« of unknown in other congrids. peculiar, expanded transverse processes on the These differences and similarities seem to anterior vertebrae, and the loss of an anterior bear out Bohlke's (1957) contention that the maxillary pedicel. These specializations proba­ Heterocongrinae should be regarded as a well­ bly preclude Gorgasia as an ancestor, but they defined subfamily within the Congridae. That do not militate against the hypoth esis that there are profo und differences in behavior and Gorgasia is more generalized over-all, and prob­ ecology is undoubted , but these have not in­ ably was an earlier offshoot of the heterocon­ volved any fundamental changes in the basic grine line. congrid body plan. The retention in Gorgasia of a lateral ethmoid Until recently, little information has been process indicates relationship with the anagoine available on the osteology of the family Con­ line. It seems unlikely that the agreement rep­ gridae. However, Asano (1962) has presented resents convergence. Eels have evolved a num­ detailed information on the anatomy of 10 ber of structures bracing the maxillary, cor­ genera and 14 species of Japanese congrids. related with elongation of the gape and with the On the basis of his study, Asano recognized use of the jaws in biting and crushing (Gos ­ two subfamilies, the Anagoinae and the Con­ line, 1951; Asano, 1962). However, the trend grinae (the Heterocongrinae were not consid­ in heterocongrine evolution has been in the ered) . The Anagoinae were said to differ from other direction, toward shortening of the gape the Congrinae in that there is a forward and and development of a jaw structure and denti­ laterally directed process on the ethmoid, the tion suitable for snapping at planktonic prey. supraoccipital is absent, there are only four It may be that the retention of the lateral eth­ suborbitals, the abdominal and caudal vertebrae moid process in Gorgasia has allowed the loss are about equal in number, the gas bladder is of the maxillary pedicel. attached to the parapophyses, the tail-tip is hard, It seems plausible to hypothesize that the the caudal rays are short, the fin rays are un­ Heterocongrinae and Anagoinae arose from a segmented, and the lateral-line scales are well common ancestor which had a lateral ethmoid developed. process, a supraoccipital, unsegmented fin rays, Asano assigned two genera, A nago and AUo­ and well-developed lateral-line scales. It seems conger, to the Anagoinae. I can confirm that likely that the sand-burrowing habit (known A riosoma belongs here, as does the recently for Anago) had already been developed. The described Paraconger Kanazawa 1961. two groups have diverged sharply, however. The heterocongrines share characters with The development of the plankton-feeding habit both the Anagoinae and the Congrinae. They in the heterocongrines has been accompanied agree with the congrines in that the supraoc­ by important changes in the head. The mouth cipital is present, there are many more caudal has become short and oblique, and the denti­ than abdominal vertebrae, and the gas bladder tion specialized. The development of a short is free from the parapophyses. They agree with oblique mouth as an adaptation to snapping at the anagoines in that the fin rays are unseg­ plankton or small prey has taken place in a mented, the caudal is reduced, and the lateral­ number of fishes. Compare, for example, the line scales are well ossified (corr esponding to serranid genus Epinepbelus, which feeds on Asano's "Allago type"). I have been unable to relatively large prey, with the plankton-feeding determine with certainty the number of sub­ Parantbias. A similar phenomenon can be seen orbitals in the heterocongrines. if the bottom-feeding embiotocid genus Micro­ Gorgasia alone agrees with the Anagoinae metres is compared with the closely related in having a lateral ethmoid process. In this genus Bracbyistius, which feeds in midwater connection it is important to establish the evo­ (Hubbs and Hubbs, 1954). Walter A. Starck II lutionary position of Gorgasia. Bohlke (1957) has pointed out to me that the shortening of the Osteology of Gorgasia punctata- RoSENBLATT 97 muzzle in these fishes results in the placement - - - 1957. On the occurrence of garden eels of the eye close to the tip of the snout, and thus in the western Atlantic, with a synopsis of allows for close-up binocular vision. Thus, the Heterocongrinae. Proe. Acad. Nat. Sci. vision has become more important in prey find­ Philadelphia 109 :59-79. ing in heterocongrines, and the eye is much GOSLINE, W . A. 1951. The osteology and clas­ enlarged and the olfactory organ much re­ sification of the ophichthid eels of the Ha­ duced. The lateral-line system on the head has waiian Islands. Pacific Sci. 5(4) :298- 320. likewise become reduced, again probably cor­ --- 1952. Notes on the systematic status related with the increased dependence on vision. of four eel families. J. W ashington Acad. The great elongation of the slender body would Sci. 42 (4) :130-135. seem to be an adaptation to getting the head HUBBS, C. L., and L. C. HUBBS. 1954. Data on well off the bottom, and yet maintaining con­ the life history, variation, ecology and rela­ tact with the sand tube which is used for cover. tionships of the Kelpperch, Brachyistius fre­ (The normal posture of a heterocongrine is natus, an embiotocid fish of the Californias. vertical, with the anterior one-half to two-thirds Calif. Fish and Game 40(2) :183- 198. of the body out of the sand tube.) KANAZAWA, R. 1961. Paraconger, a new genus On the other hand, the anagoines, except in with three new species of eels (family Con­ the loss of the supraoccipital, have diverged gridae) . Proe. U. S. N atl. Mus. 113(3450): much less from the basic congrid type, either in 1-14. structure or in behavior. KLAUSEWITZ, W. , and I. EIBL-EIBESFELDT. 1959. Neue Rohrenaale von den Maldiven und Nikob aren (Pisces, Apodes, Heterocon­ REFERENCES gridae) . Senck. BioI. 40(3/4) :135- 153. ASANo, H. 1962. Studies on the congrid eels MEEK, S. E., and S. F. HILDEBRAND. 1923. of Japan. Bull. Misaki Mar. BioI. Inst. 1 :1­ The marine fishes of Panama, Pt. I. Field 143. Mus. Nat. Hist. Zool. Ser, 15. 330 pp. BOHLKE, J. 1951. A new eel of the genus Tae­ NYB ELIN, O. 1963. Zur Morphologie und Ter­ nioconger from the Philippines. Copeia minologie des Schwanzskelettes der Actinop­ 1951 (1) :32-35. terygier. Ark. Zool. (2) 15 (35) :485-516.