AMERICAN MUSEUM Novitates PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3181, 18 pp., 9 figures November 22, 1996

Ontogeny of the Ethmoidal Region and Hyopalatine Arch in Macrognathus pancalus (Percomorpha, Mastacembeloidei), with Critical Remarks on Mastacembeloid Inter- and Intrarelationships

RALF BRITZ'

ABSTRACT Ontogenetic changes in head shape and struc- cartilaginous meniscus, which forms the articu- ture during development of Macrognathus pan- lation of the ecto- and entopterygoid with the lat- calus Hamilton, 1822, were investigated. The on- eral ethmoid in juvenile and adult specimens, de- togeny of the skeleton in the ethmoidal and hy- velops ontogenetically from the anterior tip ofthe opalatine complex is described. The following no- pars autopalatina. table features were discovered: pars autopalatina Hypotheses about inter- and intrarelationships chondrifies autogenously and lacks a maxillary of mastacembeloids are reviewed and tested using process at all developmental stages; the unusually ontogenetic data. The phylogenetic position ofSi- situated palatine bone develops without a carti- nobdella among mastacembeloids is reinvestigat- laginous precursor and is considered a dermopa- ed, and the genus is excluded from the Chaudhu- latine; the autopalatine is apparently lacking; the riidae and reassigned to the Mastacembelidae. INTRODUCTION The percomorph suborder Mastacembe- freshwaters of Africa, the Middle East, and loidei consists of the families Mastacembel- Asia (Nelson, 1994). Earlier authors consid- idae and Chaudhuriidae (Travers, 1984b). ered them to be closely related to true eels, The mastacembeloids currently comprise but Gunther (1861, 1880) demonstrated their about 67 species that are restricted to the affinities with acanthopterygian fishes. Nev-

' Postdoctoral Fellow, Department of Herpetology and Ichthyology, American Museum of Natural History.

Copyright K) American Museum of Natural History 1996 ISSN 0003-0082 / Price $2.40 2 AMERICAN MUSEUM NOVITATES NO. 3181 ertheless, their relationships within Acan- Melanie Stiassny, Monica Toledo-Piza, Mar- thopterygii have remained obscure. Gosline celo Carvalho, and Maurice Kottelat also (1983) hypothesized that mastacembeloids commented on the manuscript, and the paper are the closest relatives of the swamp-eel was further improved by the thorough re- family Synbranchidae, and this view subse- views of Dave Johnson and Colin Patterson. quently was supported by Travers (1984a, This work was supported by a Kalbfleisch 1984b) and Johnson and Patterson (1993). Postdoctoral Fellowship in the Department Mastacembelids possess a peculiar snout, ofHerpetology and Ichthyology ofthe Amer- a feature noted by early ichthyologists (e.g., ican Museum ofNatural History, New York. Bloch, 1786). It consists of an unpaired median trunklike tentacle flanked by a pair of prominent tubular nostrils. The skeleton of MATERIAL, METHODS, AND TERMINOLOGY the ethmoidal region, also remarkable in I follow Patterson (1977: 79) in the defi- structure and shape, has been described by nition of dermal bone, cartilage bone, and Regan (1912), Sufi (1956), Bhargava (1963), membrane bone, and the terminology I use Roberts (1980), Gosline (1983), and Travers for the different parts ofthe chondrocranium (1984a, 1984b). Although these authors com- is that of Gaupp (1906) and de Beer (1937). mented on the unusual shape and position of Pars autopalatina, pars quadrata, and pars the mastacembelid palatine bone, no studies metapterygoidea refer to parts of the carti- have been conducted to reveal its develop- laginous palatoquadrate, but the terms au- ment. Bhargava's (1958) ontogenetic inves- topalatine, quadrate, and metapterygoid are tigation of the cranium of Mastacembelus restricted to the cartilage bones of the pala- mastacembelus was restricted to the cartilag- toquadrate, as they are in Arratia and Schultze inous skull only. (1991). Autopalatine and dermopalatine re- This paper provides a detailed account of fer to a cartilage bone and a dermal bone, the ontogeny ofthe hyopalatine arch and eth- respectively. The term palatine is used when moidal region in Macrognathus pancalus, an dermo- and autopalatine have fused during Asian representative of the family Masta- ontogeny or when a distinction between the cembelidae. This ontogenetic information is two components cannot be made. then used to evaluate the homology of the Seven wild-caught specimens of Macrog- palatine bone in mastacembelids. Since the nathus pancalus Hamilton, 1822, were ob- study of ontogeny can also provide new in- tained from an aquarium fish importer in sights into the phylogenetic relationships be- Frankfurt, Germany. They were kept and bred tween taxa, Travers's (1984b) hypotheses in a 40-liter tank and fed mainly on aquatic about the intra- and interrelationships of insect larvae. When spawning first began- Mastacembeloidei then were tested with these usually in May or June-the eggs were re- new developmental data. moved, and the hatched larvae were reared in another tank. In the first few days after the beginning of external feeding, larvae fed ACKNOWLEDGMENTS mainly on Artemia nauplii. Their diet was I thank Wolfgang Maier for his support and later changed to Cyclops, Daphnia, and the for the opportunity to maintain the Macrog- larvae ofephemeropterans, plecopterans, and nathus at the Lehrstuhl fur Spezielle Zoolo- dipterans in accordance with their size. gie, Universitiit Tiibingen, and Horst Specimens were fixed in buffered formalin Schoppmann, Universitiit Tiubingen, for his at regular intervals. All measurements refer expert skills in SEM. I am also grateful to to total length. Kai-Erik Witte, who made the specimen of Five fixed specimens of M. pancalus Chaudhuria sp. available for my study, and (AMNH 217413: 8, 12, 17.8, 33, and 107 to Maurice Kottelat, who provided a copy of mm) were chosen for investigation ofthe gross his unpublished manuscript on chaudhuriid ontogenetic changes associated with head osteology. I extend special thanks to Jim Atz shape. They were stained for 10 to 40 seconds for his valuable criticism on the style and in a concentrated solution ofmethylene blue, contents of the manuscript. Gareth Nelson, then excess stain was removed by rinsing in 1 996 BRITZ: ONTOGENY OF MACROGNATHUS 3 water for one to two minutes. In the study of mtp metapterygoid skeletal development, 49 specimens of M. ns nasal septum pancalus between 3.3 mm (newly hatched) pap pars autopalatina and 36 mm (- 3 months) were cleared and pmt+pq pars metapterygoidea and In pars quadrata stained for cartilage and bone. addition, ppq posterior process of quadrate eight specimens ranging from 5.7 to 26 mm Ps parasphenoid were stained with alizarin only. Clearing and q quadrate staining followed the procedure outlined by sy symplectic Dingerkus and Uhler (1977), including the tmarg taenia marginalis modifications for larval material described tr trabecula by Taylor and vanDyke (1985). The eight trcom trabecula communis double-stained specimens ofM. pancalus fig- trh trabecular horn ured in this paper were deposited under v vomer AMNH 217414. For comparison, the following cleared and RESULTS double-stained material also was studied: MASTACEMBELOIDEI: Chaudhuria sp., 42 mm DEVELOPMENT OF HEAD SHAPE IN (AMNH 217415); Macrognathus maculatus, MACROGNATHUS PANCALUS 22, 32 mm (AMNH 217416); Macrognathus With the exception ofJob's (1941) account circumcinctus, 19 mm (AMNH 217412); Si- and some incidental remarks by Polder nobdella sinensis, 53 mm (AMNH 10259) (1963), the development of the highly pecu- and 113, 114, 117 mm (all AMNH 11078, liar snout structure of the mastacembelids caudal fins missing); "Mastacembelus pan- has received little attention, and no detailed calus," 15 mm head length (AMNH 3447); information is available. Mastacembelus armatus, 11 1, 130 mm (both The snout of the newly hatched embryo of AMNH 10274, caudal fins missing); Afro- Macrognathuspancalus is rounded and much mastacembelus congicus, 53,62 mm (AMNH different from that of the adult. Figure 1A 6157). SYNBRANCHOIDEI: Ophisternon aenig- shows the head of a larva of 8 mm in which maticum, 72 mm (AMNH 31573); Synbran- the snout is still rounded and the anterior and chus marmoratus, 47 mm (AMNH 74541), posterior openings ofthe nasal cavity are small 184 mm (AMNH 215280); Monopterus al- apertures in the skin. The lipfolds of the up- bus, 41 mm (AMNH 41579), 164 mm per and lower jaw are well developed. (AMNH 41579). At 12 mm, the snout region is slightly elon- All drawings were made with the aid of a gated (fig. IB). The anterior nasal openings camera lucida attached to a stereomicroscope are now located at the end of ventrally di- (Zeiss SV8). One 9-mm specimen of Ma- rected tubes. crognathus pancalus was prepared for scan- At 17.8 mm, the snout region has elongated ning electron microscopy (SEM) according to still further (fig. 1 C). Its anterior end projects the procedure described by Britz et al. (1995). as a median proboscis-like fleshy appendage with paired anterior nasal tubes on either side. ABBREVLATIONS At this stage the postorbital region ofthe head artmen articulatory meniscus also has undergone significant elongation. cartpr cartilaginous process on lateral At 33 mm (fig. 1 D), the snout region closely ethmoid resembles that of the adult. The entire eth- dp dermopalatine moidal region has continued to elongate and ectp ectopterygoid now projects in the long fleshy appendage entp entopterygoid typical ofmastacembelids. The lipfolds ofthe ethpl ethmoidal plate upperjaw are less conspicuous than in earlier f frontal at the hy hyomandibula stages, and the anterior nostrils open hys hyosymplectic cartilage end oflong nasal tubes. The rim ofeach tube leth lateral ethmoid exhibits two flaps. Medial to the tubes lies lo lamina orbitonasalis the tapering central nasal tentacle. meth mesethmoid The snout region ofadult M. pancalus was 4 AMERICAN MUSEUM NOVITATES NO. 3181

Fig. 1. Macrognathus pancalus. Ontogeny of the snout region in lateral view. Arrow points to the anterior nasal opening. A, 8 mm; B, 12 mm; C, 17.8 mm; D, 33 mm. Scale bar 1 mm. 1996 BRITZ: ONTOGENY OF MACROGNATHUS 5 figured by Travers (1 984b: fig. 13b). The main anteriorly directed membranous lamina (fig. difference from the 33-mm stage is that in 3A). The symplectic forms as a thin peri- the adult the terminal rim ofthe anterior na- chondral ossification around the symplectic sal tubes consists of six flaps of skin instead process. In the palatoquadrate, only the of two. The 107-mm specimen I examined quadrate is present, bearing a posteroventral confirms this observation, but details of the process that consists only ofmembrane bone, transition from two to six flaps could not be a synapomorphy for teleosts (Arratia and determined. Schultze, 1991). The pars autopalatina approaches the ventrolateral part ofthe lamina DEVELOPMENT OF THE HYOPALATINE orbitonasalis (figs. 3A, 7B). There is no trace ARCH AND ETHMOID REGION OF of a maxillary process on the pars autopala- MACROGNATHUS PANCALUS tina. In the space between the orbitonasal A radical skeletal transformation accom- lamina and the trabecular horn, there is a panies the developmental changes in head small, thin splint ofbone without any contact shape. When the larvae hatch, no traces of to cartilage (fig. 3A); this is the developing the cartilaginous skeleton can be found. dermopalatine (see discussion below). When Although the cartilaginous paired trabec- seen in ventral view (fig. 7B), the dermopa- ulae are separate from each other in the 4.25- latine is situated in direct anterior prolon- mm larva, their rostral parts have already gation of the tip of pars autopalatina. The started to fuse along the midline by 4.75 mm nasal septum and the trabecular horns are (fig. 7A). Their anterior tips are slightly curved pronounced. The narrow vomer underlies the outward to form trabecular horns. The sym- broad ethmoidal plate. The postorbital pro- plectic process ofthe hyosymplectic cartilage cess of the auditory capsule has fused with approaches the palatoquadrate closely (fig. the lamina orbitonasalis, thus forming a com- 2A), and the palatoquadrate consists only of plete taenia marginalis or orbital cartilage (fig. the pars metapterygoidea and pars quadrata; 3A). the pars autopalatina has not yet formed. By 7.4 mm, the ossifications ofthe hyopa- By 5.6 mm, the anterior parts of the tra- latine arch cover large parts of the cartilage beculae have completely fused along the mid- (fig. 3B). The metapterygoid has formed as a line, forming the ethmoidal plate. This solid perichondral ossification around the postero- block of cartilage bears five processes (figs. dorsal part ofthe palatoquadrate, but its most 2B, 7B): originating in the midline is the nasal distal tip remains cartilaginous. The pars au- septum; anteriorly, the trabecular horns are topalatina is a narrow strip of cartilage still pronounced and form a pair of lateral pro- continuous with the pars quadrata, and its cesses; and posterior to them and directed anterior tip now articulates with the ventral dorsolaterally are the paired developing ru- part of the lamina orbitonasalis. Above and diments of the laminae orbitonasales. The below the pars autopalatina, the entoptery- hyosymplectic cartilage has changed only in goid and the ectopterygoid, respectively, have size. The pars autopalatina has chondrified formed (fig. 3B). Both are narrow, splintlike as a bar of cartilage, separate from pars me- dermal bones that ossify without cartilagi- tapterygoidea and pars quadrata, with its an- nous precursors. The dermopalatine is situ- terior end projecting toward the orbitonasal ated in a gap between the orbitonasal lamina lamina of the ethmoidal plate. and the trabecular horn, which is now only By 5.9 mm, the pars autopalatina and pars a small cartilaginous nodule lying lateral to quadrata have fused (fig. 2C), and the tra- the cartilage ofthe ethmoidal plate but with- beculae have separated from the basal plate. out any connection to it (fig. 3B). Part of the The postorbital process projects rostrally from orbital cartilage has been resorbed. The tae- the auditory capsule. A dorsally directed pro- nia marginalis is thus separated into an an- cess has formed on the lamina orbitonasalis, terior and a posterior part, producing a con- and the nasal septum has become larger. dition that resembles the 6-mm stage (cf. figs. At the 6.1-mm stage, ossification of the 2C and 3B). The enlarged nasal septum has hyopalatine arch has begun. The hyoman- fused with the orbitonasal lamina to form the dibula is a perichondral bone with a small olfactory foramen, the passage of the olfac- 6 AMERICAN MUSEUM NOVITATES NO. 3181

.tr- A t t+pq ~~trh hys t..... pm A

tr r ns

trh

hys pap B

ns trcoms

hys pap

C Fig. 2. Macrognathus pancalus. Ontogeny of the hyopalatine arch and ethmoidal complex in lateral view. Rostral cartilage not shown; lacrimal and nasal removed where developed; cartilage stippled. A, 4.75 mm; B, 5.6 mm; C, 5.9 mm. Scale bar 0.5 mm. tory nerve. The most anterior tip ofthe nasal By 9 mm, the hyopalatine arch is similar septum is surrounded by the perichondral os- to that ofthe preceding stage, but the cartilage sification of the mesethmoid. It is still sep- bones have encroached on larger areas ofcar- arated from the vomer and has already de- tilage (fig. 4). The ecto- and entopterygoid veloped a long, pointed posterodorsal process still consist ofnarrow splints ofbone, but the of membrane bone. former has broadened its ventrocaudal part. 1 996 BRITZ: ONTOGENY OF MACROGNA THUS 7

trh

ppq

Ps ,V trh hy

sy B q Fig. 3. Macrognathuspancalus. Ontogeny ofthe hyopalatine arch and ethmoidal complex, continued. A, 6.1 mm; B, 7.4 mm. Scale bar 0.5 mm.

The dermopalatine bone is stronger, has lamina orbitonasalis is covered laterally by shifted closer to the midline, and now exhib- the perichondral ossification of the lateral its a number ofconspicuous fanglike teeth on ethmoid (fig. 4). The mesethmoid fully sur- its ventral surface (figs. 4, 7C, 8). The rudi- rounds the dorsal and lateral sides of the an- ment ofthe trabecular horn is fully resorbed. terior part ofthe ethmoidal plate, and its dor- A well-developed frontal bone overlies the socaudally directed membranous process lies anterior part ofthe taenia marginalis, and the between the anterior tips ofthe frontals. The 8 AMERICAN MUSEUM NOVITATES NO. 3181

leth meth

V hy

sy ectp q Fig. 4. Macrognathuspancalus. Ontogeny ofthe hyopalatine arch and ethmoidal complex, continued. 9 mm. Scale bar 0.5 mm. anterior tips of the mesethmoid and vomer terygoid, still articulating with the ventral part have fused. of the lamina orbitonasalis. The entoptery- By 12.8 mm, the cartilaginous connection goid and, especially, the ectopterygoid are well between the pars quadrata and pars autopa- developed. The dermopalatine still bears a latina (figs. 2C, 3A,B, 4) has been resorbed number of teeth. The ossification of the lat- (fig. 5). Only a small anterodorsally directed eral ethmoid covers more of the lamina or- process is present on the cartilage ofthe pars bitonasalis, and the mesethmoid has elon- quadrata, but the pars autopalatina persists gated considerably. as an elongate nodule ofcartilage (fig. 5, pap) The most advanced stage investigated in between the anterior tips ofecto- and entop- this study measured 36 mm. All of the car-

leth meth

Ps, hy. v

-ectp

q Fig. 5. Macrognathuspancalus. Ontogeny ofthe hyopalatine arch and ethmoidal complex, continued. 12.8 mm. Scale bar 0.5 mm. 1 996 BRITZ: ONTOGENY OF MACROGNA THUS 9

leth meth

hy v

sy ectp

q Fig. 6. Macrognathuspancalus. Ontogeny ofthe hyopalatine arch and ethmoidal complex, continued. 36 mm. Scale bar 0.5 mm. tilage bones in the palatoquadrate have de- the ectopterygoid a boomerang-like shape (fig. veloped considerable amounts of membrane 9B). Anteriorly the ectopterygoid approaches bone (fig. 6). Cartilage in the palatoquadrate the vomer very closely, running parallel to it, is restricted to the articulations of the hyo- thus occupying the same position as the pal- mandibula, small strips between hyomandib- atine in mastacembelids. An articulation be- ula and symplectic and between metaptery- tween the hyopalatine arch and the lateral goid and quadrate, and a small nodule ofcar- ethmoid via a cartilaginous meniscus, seen tilage that forms the articulatory meniscus in mastacembelids, is not present in Chau- (fig. 6, artmen) of the hyopalatine arch with dhuria. The large mesethmoid projects as a the ethmoidal region. The complete eth- bony septum between the olfactory organs. moidal region has further elongated, especial- Its pointed posterior part separates the two ly the mesethmoid and dermopalatine (figs. frontals. Its anterior tip is fused to the vomer. 6, 7D). The latter has lost all of its teeth and The mesethmoid and lateral ethmoid are is attached to the vomer very closely for most connected by cartilage that is a rudiment of of its length (fig. 7D). the nasal septum ofthe chondrocranium. The lateral ethmoid ossification is small com- THE HYOPALATINE ARCH AND pared to that of M. pancalus. Its dorsal part ETHMOIDAL SKULL IN CHAUDHURIA is firmly bound to the nasal, and it has a laterally directed ridge. A long cartilaginous The skeleton ofa specimen of Chaudhuria process projects from the ventral part of the caudata was described by Travers (1984a). lateral ethmoid ventrolaterally (fig. 9B, As he pointed out, the Chaudhuriidae lack cartpr). The medial side of this cartilaginous an entopterygoid as well as a palatine bone process articulates with the lateral side ofthe (also Kottelat, personal commun.). However, ectopterygoid, and the lateral tip of the car- further details are needed before meaningful tilage is tightly bound to the posterior end of comparisons can be made between the Chau- the lacrimal by connective tissue. dhuriidae and the Mastacembelidae. In Chaudhuria sp. (AMNH 217415), the DISCUSSION ectopterygoid extends rostrally from the THE quadrate to the anterior third of the meseth- HOMOLOGY OF THE PALATINE moid (fig. 9B). It articulates syndesmotically BoNE IN MASTACEMBELIDS with the medial side of the quadrate, from Regan (1912: 219) first reported on the pe- which it is dorsally directed toward the lateral culiar shape and position ofthe palatine bone ethmoid. From here, its anterior body is in- in the spiny eels. He described it as "an elon- clined toward the posterior at an angle, giving gate narrow lamina which is firmly united to 10 AMERICAN MUSEUM NOVITATES NO. 3181

tr trh

v -ethpl S-ns

pap trh pap d B

Ps meth

ectp leth

meth dp ethpl D Fig. 7. Macrognathus pancalus. Ontogeny of the anterior hyopalatine arch and ethmoidal complex in ventral view. Same specimens as in figures 2A, 3A, 4, 6. Cartilage stippled. A, 4.75 mm; B, 6.1 mm; C, 9 mm; D, 36 mm. Scale bar 0.5 mm. the lower surface of the vomer and also to The present study of the ontogeny of the the mesethmoid, lateral ethmoid, and para- hyopalatine arch in Macrognathus pancalus sphenoid." However, the exact nature of the reveals that the palatine ossifies without a palatine in mastacembelids has remained un- cartilaginous precursor, and at no time during known for lack of developmental informa- ontogeny does it have any connection with tion. According to Travers (1984a: 21), "der- the cartilage of the palatoquadrate. Even in mal and endochondral components cannot later development, it never fuses to or is be distinguished" in Mastacembelus masta- closely connected with any bone of the pal- cembelus. atoquadrate. This can be taken as unequiv- 1 996 BRITZ: ONTOGENY OF MACROGNATHUS I1I

ocal evidence ofan exclusively dermal origin of the palatine in M. pancalus. Accordingly, this bone is a dermopalatine with no autopalatine component. Given the homogeneity of palatine morphology in all mastacembelids so far studied, it seems reasonable to consider this interpretation valid for all representatives ofthe family. Moreover, this study has demonstrated that the dermopalatine does not achieve its peculiar position close to the vomer until late in development, gradually shifting from its original lateral position to the final medial position (cf. fig. 7B,C,D). Another remarkable and unexpected feature of this Asian mastacembelid is the presence of teeth on the palatine of M. pancalus during some ontogenetic stages. According to Sufi (1956: 95), the palatines of mastacembelids are "narrow flakes ofbone immovably united to mesethmoid, vomer, and parasphenoid, bearing teeth in some African species, toothless in other African and all Asiatic Fig. 8. Macrognathus pancalusX9-mm larva species." This observation was confirmed by SEM photograph ofthe roofofthe mouth showing Travers (1984a: 84): "All Asian, and the ma- palatal dentition. Arrow points to a tooth on the jority ofAfrican species lack a palatine, tooth right palatine. Scale bar 0. i mm. bearing, alveolar surface (dermopalatine)." In contrast, my observations demonstrate that the palatine of M. pancalus bears teeth by a comparative analysis of the food spec- during a significant phase of development. trum at different ontogenetic stages. The times of first appearance and disappear- The presence of teeth on the palatine of ance ofpalatine teeth in M. pancalus are vari- Asian mastacembeloids was confirmed in able. In juveniles of 6.5 mm, teeth on the three additional species. In one ofthree adult palatine can be present; these teeth can persist specimens of Sinobdella sinensis (AMNH in specimens up to 23.5 mm in length, yet 11078), two teeth are present at the anterior they can be lost in specimens 22 mm long. tip of the left palatine. In a smaller specimen The fact that palatine teeth cannot be found of 5t3mm, the left palatine bears six teeth in specimens of M. pancalus larger than 24 and the right bears two (fig. 9A). A 22-mm mm demonstrates that they are lost during specimen of Macrognathus maculatus ontogeny, and this is certainly the main rea- (AMNH 217416) showed one tooth on the son for the notion that Asian mastacembelids left palatine, whereas both palatines of a 32- lack palatine teeth. When present in the mm specimen (AMNH 217416) of the same smaller juveniles of M. pancalus, palatine species were toothless. Furthermore, a 19- teeth are always fanglike (figs. 4, 5, 8) and mm specimen of Macrognathus circumcinc- probably serve to grasp and hold prey. Ob- tus (AMNH 217412) had one tooth on each servations of larvae and juveniles feeding in palatine, although this bone is toothless in captivity show that they are able to engulf adults (Sufi, 1956). These three examples can prey very much larger in relation to their body be taken as additional evidence for a change size than are adults. Thus, the presence of of the palatal dentition during ontogeny and palatine teeth in earlier ontogenetic stages in a probable concomitant change in feeding M. pancalus may be correlated with a differ- mode in at least some Asian mastacembelids. ent feeding mode of these larvae and juve- Further studies must test whether this as- niles-a hypothesis easily testable in the field sumption holds for all Asian spiny eels. 12 AMERICAN MUSEUM NOVITATES NO. 3181

leth meth mtp hy dp v A q

leth meth

Ps hy

Sy B

f leth

v p ectp q C Fig. 9. Hyopalatine arch and ethmoidal complex in ventral view. Rostral cartilage not shown; lacrimal and nasal removed; cartilage stippled. A, Sinobdella sinensis, 53 mm; B, Chaudhuria sp., 42 mm; C, Synbranchus marmoratus, 47 mm. Scale bar 0.5 mm. 1 996 BRITZ: ONTOGENY OF MACROGNATHUS 13

FURTHER UNUSUAL FEATURES IN THE The two bones develop from separate ossi- ONTOGENY OF THE HYOPALATINE fication centers and remain separate in lower ARCH IN MACROGNATHUS PANCALUS teleosts but are considered to be fused in Clu- peocephala (Arratia and Schultze, 1991), al- As described above, the pars autopalatina though being separate in Alepocephalus (Gos- ofM. pancalus develops as a chondrification line, 1969) and some osmerids (Johnson, per- separate from all other parts of the palato- sonal commun.). The ontogenetic fusion quadrate (fig. 2A). In later developmental seems to have been demonstrated in only two stages, the pars autopalatina fuses with pars clupeocephalans: Salmo (de Beer, 1937) and quadrata and metapterygoidea (fig. 2C), es- Plecoglossus (Howes and Sanford, 1987). The tablishing a condition that is found in most in other teleosts from the beginning (e.g., Swin- dermopalatine is lost several teleostean nerton, 1902; de Beer, 1937; Daget, 1964; lineages, e.g., the Ostariophysi (according to Arratia and Schultze, 1991). The separate Fink and Fink, 1981). Teeth on the palatine chondrification and later fusion were de- occur in a wide range ofteleosts. Usually the scribed and illustrated by Bhargava (1958) possession ofpalatal teeth in the adult is taken for Mastacembelus armatus. Among perco- as evidence for the presence of a dermopa- morphs, this seems to be an unusual char- latine. However, detailed studies of the on- acter, but comparative data are lacking at the togeny of this region that incorporate a well- moment. In teleosts, a separately chondri- graded series of developmental stages are fying pars autopalatina has been reported for lacking for representatives of percomorphs some clupeids (Wells, 1923; Shardo, 1995), with palatine teeth. siluroids (Kindred, 1919; Arratia, 1990; Ar- As my ontogenetic studies have revealed, ratia and Schultze, 1991), and syngnathids palatine teeth can also be lost during ontog- (Kindred, 1921; Kadam, 1961). All these eny, a fact hitherto reported only rarely groups are distantly related to mastacembel- (Howes and Sanford, 1987, for Plecoglossus) ids. and one that must influence the evaluation A second unusual feature in the skeletal of the homology of the palatine bone in tel- ontogeny of Macrognathus pancalus is the eosts. Lack of palatine teeth in adult teleosts lack of a maxillary process of the pars auto- cannot be taken as evidence that a dermo- palatina (cf. figs. 2C, 3-6). Mastacembelus palatine is lacking in these fish, as, for ex- armatus also lacks a maxillary process (Bhar- ample, Travers (1 984a) has done. On the oth- gava, 1958: figs. 11, 12, 17, 18). Unlike the er hand, the existence ofteeth on the palatine situation in most other teleosts, in these mas- does not necessarily indicate that a dermo- tacembelids, the anterior end of the pars au- and autopalatine are present, inasmuch as in topalatina stops at the level of the lamina mastacembelids, only the dermopalatine may orbitonasalis. Separate chondrification ofthe be developed. This demonstrates that onto- autopalatine, as well as lack of a maxillary genetic studies are imperative in order to process, can be hypothesized as synapomor- evaluate the homology of the palatine if this phies for mastacembelids or even mastacem- feature is to be used in discussions of phy- beloids. However, ontogenetic data on chau- logeny. dhuriids and synbranchids are needed to con- firm the hypotheses. INTERRELATIONSHIPS OF MASTACEMBELOIDEI REMARKS ON THE HOMOLOGY OF THE The hypothesis ofGosline (1983) that mas- PALATINE iN TELEOSTS tacembeloids and synbranchids are closely related was substantiated by Travers (1 984b), The palatine in teleosts primitively con- who cited characters shared sists of two different bones: the autopalatine, six derived by which forms around the anterior tip of the both groups: palatoquadrate, and the dermopalatine, which 1. Loss or reduction in size of the posttem- ossifies ventral to the autopalatine without a poral bone, accompanied by loss of the cartilaginous precursor and bears teeth connection to the pectoral girdle. (Gaupp, 1906; Arratia and Schultze, 1991). 2. Extension of the dentary posteroventrally 14 AMERICAN MUSEUM NOVITATES NO. 3181

along the ventral edge of the angulo-ar- Strictly speaking, there is no direct articula- ticular. tion between the ectopterygoid and the lateral 3. Wide anterolateral face of the ectoptery- ethmoid in mastacembelids. goid and its direct articulation with the An articulatory cartilage is apparently lateral ethmoid. missing in synbranchids, but there are no de- 4. Palatine sutured along the posterolateral tailed data available concerning the ontogeny face of the vomerine shaft. of the head skeleton of synbranchids. Ac- 5. Insertion of the levator operculi muscle cording to my own observations on juvenile on the dorsolateral face ofthe operculum. Ophisternon, Synbranchus, and Monopterus, 6. Dorsolateral expansion of the hyohyoidei both the ectopterygoid and the palatine con- adductores muscle, sealing the operculum tribute to the articulation with the lateral eth- to the body wall and causing a restricted moid (fig. 9C). opercular opening. The whole picture is further complicated when we consider the condition in chaudhu- Doubts concerning the validity of some of riids. As previously indicated there is no ar- these characters as synapomorphies of syn- ticulatory cartilage between the ectopterygoid branchids and mastacembeloids arose during and lateral ethmoid in Chaudhuria. The eth- this study. moidal articulation of the ectopterygoid in Loss or reduction in size oftheposttemporal this taxon is significantly different from that bone, accompanied by a loss ofconnection to in mastacembelids (cf. fig. 5A and B). Un- the pectoral girdle (character 1): The ambi- fortunately, there are no ontogenetic data for guity of this character as a synapomorphy of any chaudhuriid. It may well be that the synbranchids and mastacembelids can be seen chaudhuriid condition evolved from the by Travers's (1984b: 96) own remarks: "A mastacembelid condition and that, accord- well developed, forked, posttemporal con- ingly, an articulatory cartilage was present in necting the basicranium to the pectoral girdle the last common ancestor of Chaudhuriidae ... occurs in Monopterus albus and Ophis- and Mastacembelidae. On the other hand, it ternon... ." Thus, disconnection ofshoulder is at least as likely that the meniscus is a girdle and skull and reduction in size of the synapomorphy ofthe Mastacembelidae. Ad- posttemporal have occurred only in some ditional evidence is needed before a choice synbranchids and cannot be used as a syna- can be made between the two hypotheses. pomorphy to align synbranchids with mas- However, the question of possible similari- tacembelids. To do so is to assume reversal ties between synbranchids and mastacem- in those genera of synbranchids in which a beloids in the articulation in the hyopalatine connection exists between the shoulder girdle arch with the lateral ethmoid still remains. and the skull. According to the present hy- Despite the weakness of some of Travers's pothesis ofthe intrarelationships ofsynbran- (1984b) characters, the hypothesized sister- chids (Rosen and Greenwood, 1976: figs. 66, group relationship of synbranchids and mas- 67), this is a less parsimonious assumption. tacembeloids still seems to be convincing Wide anterolateralface ofthe ectopterygoid considering the additional synapomorphies and its direct articulation with the lateral eth- cited by Johnson and Patterson (1993). moid (character 3): As Travers (1984a) cor- rectly pointed out and as has been confirmed in this study, a small cartilaginous meniscus REMARKS ON THE PHYLOGENETIC situated at the anterior ends of the ecto- and POSITION OF SINOBDELLA SINENSIS entopterygoid bones forms the articulatory The spiny eel Sinobdella (= Rhynchobdella surface between the hyopalatine arch and the in Travers, 1984a, 1984b) sinensis was con- ethmoidal region in adult mastacembelids sidered to be a member of the family Mas- (and in Sinobdella sinensis; see below and fig. tacembelidae until Travers (1984b: 132) in- 9A, artmen). This meniscus is a rudiment of cluded it in the family Chaudhuriidae, as the the anterior tip of the cartilaginous palato- sister group of all other chaudhuriids, on the quadrate, the pars autopalatina (figs. 2-6). basis of shared similarities between Sinob- 1 996 BRITZ: ONTOGENY OF MACROGNA THUS 15 della, Chaudhuria, and Pillaia. These simi- the exoccipitals meeting in the midline and larities are: not separated by the supraoccipital. In addition, two specimens of Mastacembelus ar- 1. Long, narrow posterior parasphenoid matus (AMNH 10274) showed exoccipitals processes. separated by the supraoccipital. These ob- 2. Dorsomedial processes ofthe exoccipital servations are limited, but clearly demon- separated by the supraoccipital. strate variation in this character and thus its 3. Dorsal surface of the exoccipital perfo- ambiguity as a synapomorphy for Sinobdella rated. and chaudhuriids. 4. One or two inner rows of premaxillary Dorsal surface ofthe exoccipital perforated teeth. (similarity 3): In the four specimens of Si- 5. Loss of entopterygoid. nobdella investigated in this study, I was un- 6. Ectopterygoid with narrow lateral face able to recognize the "dense perforations" and very long anterodorsal process. that Travers (1984b: 112) described. No 7. Palatine reduced in size. striking differences could be discerned be- 8. Interdigitating processes of anterior and tween the perforations of the exoccipital in posterior ceratohyals reduced or lost. Sinobdella, Chaudhuria, and the investigated 9. Loss of ventral processes on basibran- mastacembelids. Furthermore, there seems chial 2. to be large intraspecific variation that may 10. Loss ofepipleural ribs and three or fewer have been overlooked by Travers (1984a, epicentral ribs. 1984b), who, according to his list of study material (1984a: table 3), investigated only Kottelat (1991) disagreed with Travers's one cleared-and-stained specimen of Sinob- (1984b) taxonomic changes and, in addition, della. Johnson and Patterson (1993: 586) found Loss of entopterygoid; ectopterygoid with "problems with Travers's characterization of narrow lateralface and very long anterodorsal the Chaudhuriidae." The latter authors process; and palatine reduced in size (simi- pointed out that at least some of the above- larities 5, 6, and 7): Inasmuch as these three cited similarities (nos. 1, 2, 4, 9) may be ple- characters seem to be correlated, they are siomorphies and others (nos. 8, 10) may not treated together in this paragraph. In Sinob- exist. Recently, Kottelat and Lim (1994) re- della, Chaudhuria, and Pillaia (Travers, expressed their doubt about the phylogenetic 1984a, 1984b), as well as in other chaudhu- placement ofSinobdella sinensis. Some ofthe riids (Kottelat, personal commun.), an en- characters thought by Travers (1984b) to align topterygoid is lacking. However, the articu- Sinobdella with Chaudhuriidae will be dis- lations of the ectopterygoid with the lateral cussed by Kottelat in a forthcoming paper on ethmoid in Sinobdella and Chaudhuria are the osteology ofchaudhuriids and will not be quite different from each other. The articu- addressed in this work. In addition to the lation in Sinobdella more closely resembles criticisms by Johnson and Patterson (1993) the condition in all other mastacembelids: and Kottelat (personal commun.), results of the ectopterygoid articulates with the lateral my studies further weaken Travers's (1 984b) ethmoid via a cartilaginous meniscus (fig. 9A). hypothesis ofa close relationship between Si- This meniscus is absent in Chaudhuria (fig. nobdella and the chaudhuriids. These points 9B). Furthermore, the shape of the ectopter- will be discussed here in detail. ygoid in Sinobdella is not significantly dif- Dorsomedial processes of the exoccipitals ferent from the shape of the ectopterygoid in separated by the supraoccipital (similarity 2; other mastacembelids, especially if we con- cited by Johnson and Patterson [1993: 586] sider developmental stages, and is not spe- as "presence of a spina occipitalis"): My in- cifically similar to that in Chaudhuria (cf. fig. vestigation of four specimens of Sinobdella 9A with figs. 5, 6 and fig. 9B). The dermo- sinensis revealed that in only three of them palatine of Sinobdella cannot be considered did the supraoccipital separate the exoccip- reduced in size, inasmuch as it is not different itals. The specimen of Chaudhuria sp. has in shape from the dermopalatines in other 16 AMERICAN MUSEUM NOVITATES NO. 3181 mastacembelids and even bears teeth in ear- niscus (a remnant of pars autopalatina; see lier developmental stages (fig. 9A). above) in juveniles and adults. This specific Loss of epipleural ribs and three or fewer articulation has not been found in chaudhu- epicentral ribs (similarity 10): As already not- riids or synbranchids or any other teleost. ed by Johnson and Patterson (1993), Travers Sinobdella shares with mastacembelids the (1984b) confused intermuscular bones with peculiar elongated shape and dorsally shifted pleural ribs. Johnson and Patterson (1993) position of the coronomeckelian bone that found four intermuscular bones (their epi- does not occur in chaudhuriids or synbran- neurals) in Sinobdella, not three as reported chids or any other teleost. Travers (1 984b: by Travers (1984a, 1984b), and these origi- 115) was aware that chaudhuriids lack an es- nate on the first four centra. Their observa- pecially elongated coronomeckelian, and that tion was confirmed by the four specimens of it more "resembles the plesiomorphic con- S. sinensis examined in the present study. dition seen in other groups as the percoids," Two of them had four intermuscular bones but he interpreted the chaudhuriid condition on both sides one five on the left and four on as "secondarily reduced." This interpretation the right side, and one five on both sides. rests on the hypothesis that Sinobdella and Pleural ribs are present on all abdominal ver- the chaudhuriids are sister groups, an as- tebrae, from the third or fourth toward the sumption that lacks support at present. rear. The number of intermuscular bones Based on these shared characters between varies among chaudhuriids between zero and Sinobdella and the mastacembelids, and on two (Kottelat, personal commun). These re- the weakness ofa number ofTravers's (1984b) marks invalidate the above-cited character as similarities, I concur with the recommen- a synapomorphy at this level because Sinob- dation of Kottelat and Lim (1994) that Si- della does not differ in these characters from nobdella should be retained in the family other mastacembelids like M. mastacembe- Mastacembelidae, in which it may form the lus (see Travers, 1984a: fig. 12). sister group ofall remaining mastacembelids. According to the information presently The present study has demonstrated that available, there are no significant similarities investigations of skeletal ontogeny can be of between Sinobdella and Chaudhuriidae, ex- outstanding importance in anatomical and cept for the lost entopterygoid, that would phylogenetic studies. It yields new sources of support the inclusion of S. sinensis in this characters with phylogenetic significance and family. On the contrary, there are at least two is a powerful tool in evaluating the homology characters shared only by Sinobdella and ofosteological characters. A limitation to the mastacembelids that could be interpreted as utilization ofthese characters in phylogenetic synapomorphies for that group. discussions at the moment seems to be a lack Sinobdella shares with mastacembelids the of such ontogenetic data in the majority of special articulation ofthe ectopterygoid with teleostean groups-a gap in our knowledge the lateral ethmoid via a cartilaginous me- that hopefully will be filled in the future.

REFERENCES Arratia, G. um ofMastacembelus armatus (Cuv. & 1990. Development and diversity of the sus- Val.). J. Morphol. 102: 401-426. pensorium oftrichomycterids and com- 1963. The cranial osteology ofMastacembelus parison with loricarioids (Teleostei: Sil- armatus. Morphol. Jahrb. 105: 1-25. uriformes). J. Morphol. 205: 193-218. Bloch, M. E. Arratia, G., and H.-P. Schultze 1786. Naturgeschichte der auslandischen 1991. Palatoquadrate and its ossifications: de- Fische, vol. 2. Berlin. velopment and homology within os- Britz, R., M. Kokoscha, and R. Riehl teichthyans. J. Morphol. 208: 1-81. 1995. The anabantoid genera Ctenops, Lucio- Bhargava, H. N. cephalus, Parasphaerichthys, and 1958. The development of the chondrocrani- Sphaerichthys (Teleostei: Perciformes) 1996 BRITZ: ONTOGENY OF MACROGNA THUS 17

as a monophyletic group: evidence from Kindred, J. E. egg surface structure and reproductive 1919. The skull ofAmiurus. Ill. Biol. Monogr. behaviour. Jpn. J. Ichthyol. 42: 71-79. 5: 1-105. Daget, J. 1921. The chondrocranium of Syngnathus 1964. Le crane des teleosteens. Mem. Mus. fuscus. J. Morphol. 35: 425-456. Nat. Hist. Nat. Ser. A Zool. 31: 163- Kottelat, M. 340. 1991. Notes on the taxonomy and distribution de Beer, G. R. of some western Indonesian freshwater 1937. The development ofthe vertebrate skull. fishes, with diagnoses of a new genus [Reprint, 1971]. Chicago and London: and six new species (Pisces: Cyprinidae, Univ. of Chicago Press. Belontiidae, and Chaudhuriidae). Ich- Dingerkus, G., and D. L. Uhler thyol. Explor. Freshwaters 2: 273-287. 1977. Enzyme clearing ofAlcian Blue stained Kottelat, M., and K. K. P. Lim whole small vertebrates for demonstra- 1994. Diagnoses of two new genera and three tion ofcartilage. Stain Technol. 32: 229- new species of earthworm eels from the 231. Malay Peninsula and Borneo (Teleostei: Fink, S. V., and W. L. Fink Chaudhuriidae). Ichthyol. Explor. 1981. Interrelationships of the ostariophysan Freshwaters 5: 181-190. fishes. Zool. J. Linn. Soc. 72: 297-353. Nelson, J. S. Gaupp, E. 1994. Fishes of the world, 3rd ed. New York: 1906. Die Entwickelung des Kopfskelettes. In Wiley. 0. Hertwig (ed.), Handbuch der ver- Patterson, C. gleichenden und experimentellen Ent- 1977. Cartilage bones, dermal bones and wickelungslehre der Wirbeltiere 3(2. membrane bones, or the exoskeleton Teil), 6. Kapitel: 573-890. versus the endoskeleton. Linn. Soc. Gosline, W. A. Symp. Ser. 4: 77-12 1. 1969. The morphology and systematic posi- Polder, W. N. tion of the alepocephaloid fishes. Bull. 1963. Mastocembeluspancalus. Aquarien- und Br. Mus. Nat. Hist. (Zool.) 18: 185-218. Terrarienzeitschrift 16: 353-357. 1983. The relationships ofthe mastacembelid Regan, C. T. and synbranchid fishes. Jpn. J. Ichthyol. 1912. The osteology ofthe teleostean fishes of 29: 323-328. the order Opisthomi. Ann. Mag. Nat. Gunther, A. C. L. G. Hist. Ser. 8, 9: 217-219. 1861. Catalogue of the acanthopterygian fish- Roberts, T. R. es in the British Museum, vol. 3. Lon- 1980. A revision ofthe Asian mastacembeloid don: Taylor and Francis. genus Macrognathus. Copeia 1980: 385- 1880. An introduction to the study of fishes. 391. Edinburgh: Adam and Charles Black. Rosen, D. E., and P. H. Greenwood Howes, G. J., and C. P. J. Sanford 1976. A fourth neotropical species of syn- 1987. Oral ontogeny of the ayu, Plecoglossus branchid eel and the phylogeny and sys- altivelis and comparisons with the jaws tematics ofsynbranchiform fishes. Bull. of other salmoniform fishes. Zool. J. Am. Mus. Nat. Hist. 157: 1-70. Linn. Soc. 89: 133-169. Shardo, J. D. Job, T. J. 1995. Comparative embryology of teleostean 1941. Life-history and bionomics ofthe spiny fishes: I. Development and staging of eel, Mastacembelus pancalus (Hamil- the American shad, Alosa sapidissima ton), with notes on the systematics of (Welson, 1811). J. Morphol. 225: 125- the Mastacembelidae. Rec. Indian Mus. 167. 42: 121-135. Sufi, S. M. K. Johnson, G. D., and C. Patterson 1956. Revision of the oriental fishes of the 1993. Percomorph phylogeny: a survey of family Mastacembelidae. Bull. Raffles acanthomorphs and a new proposal. Mus. 27: 94-146. Bull. Mar. Sci. 52: 554-626. Swinnerton, H. H. Kadam, K. M. 1902. A contribution to the morphology ofthe 1961. The development of the skull in Nero- teleostean head skeleton, based upon a phis. Acta Zool. 42: 257-298. study ofthe developing skull ofthe three- 18 AMERICAN MUSEUM NOVITATES NO. 3181

spined stickleback (Gasterosteus acu- 1984b. A review of the Mastacembeloidei, a leatus). Q. J. Microsc. Sci. 45: 503-593. suborder ofsynbranchiform teleost fish- Taylor, W. R., and G. G. VanDyke es. Part II: phylogenetic analysis. Bull. 1985. Revised procedures for staining and Br. Mus. Nat. Hist. (Zool.) 47: 83-150. clearing small fishes and other verte- Wells, F. R. brates for bone and cartilage study. Cy- 1923. On the morphology of the chondrocra- bium 8: 1-7. nium of the larval herring (Clupea har- Travers, R. A. engus). Proc. Zool. Soc. Lond. 1922: 1984a. A review of the Mastacembeloidei, a 1213-1229. suborder ofsynbranchiform fishes. Part I: anatomical descriptions. Bull. Br. Mus. Nat. Hist. (Zool.) 46: 1-133.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: [email protected]

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