498 



ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 



to its centrum, spine absent; endopterygoid absent; quadrate 

 with a lateral shelf; palatine and upper jaw bones distinctively 

 modified; postcleithra absent; supracleithrum extremely re- 

 duced; medial tabular bones absent; posttemporal modified in 

 shape and angle of articulation with supracleithrum; pelvic gir- 

 dle broad and short, with two distinct anterior processes; caudal 

 skeleton with a full neural spine on the second preural centrum; 

 branchial skeleton with main arm of first epibranchial reduced 

 to a nubbin, uncinate process enlarged and articulating directly 

 with second pharyngobranchial, and interarcual cartilage absent. 

 None of these extreme modifications (those of the caudal and 

 branchial skeletons being unique among percoids) are even fore- 

 shadowed in the skeleton of Rachycentron, which is instead 

 remarkably similar to that oi Coryphaena. except in the anterior 

 portion of the dorsal fin and the neurocranium. 



In Coryphaena, the dorsal fin is elaborated anteriorly and 

 extended into the first intemeural space (second in Rachycen- 

 tron) and there is an extreme supraoccipito-frontal crest on the 

 neurocranium. The dorsal fin modification is autapomorphic 

 for Coryphaena. but the median cranial crest is probably prim- 

 itive for echeneoids since it is variously developed in all caran- 

 gids and well-developed in Nematislius. The absence of this 

 crest in Rachycentron, associated with a slight flattening of the 

 neurocranium, is the only specialization shared with the Eche- 

 neididae. Here again, however, there is little similarity between 

 the slightly flattened neurocranium of Rachycentron and the 

 extremely flattened and restructured neurocranium of the eche- 

 neidids, in which, for instance, the supraethmoid and vomer 

 have become flat plates and the orbit is completely occluded by 

 enlargement and anterior extension of the pterosphenoids. This 

 extreme restructuring of most cranial bones is evident even in 

 larval echeneidids at the earliest development of the neuro- 

 cranium, whereas the neurocrania of Rachycentron and Cory- 

 phaena exhibit a generalized development similar to that of 

 carangids. Prior to development of the median crest in Cory- 

 phaena(> 100 mm), the neurocrania of cobia and dolphin differ 

 mainly in relative depth. Echeneidids also have an exceptionally 

 modified adductor mandibulae in which A, is absent and A, 

 and A„ are distinctively subdivided. Coryphaena and Rachy- 

 centron share a relatively generalized adductor mandibulae, spe- 

 cialized with respect to the primitive carangids (see section on 

 Carangidae) in having A, somewhat reduced and inserting nar- 

 rowly on the maxillo-mandibular ligament. 



The pronounced similarities between Coryphaena and Rachy- 

 centron in the adductor manidbulae and most osteological fea- 

 tures merely serve to reiterate the lack of evidence for the fre- 

 quently proposed sister-group relationship between Rachycentron 

 and the echeneidids. Further comparison with character states 

 throughout the Carangidae will be required to define these adult 

 similarities as primitive or derived features. The most com- 

 pelling evidence for a sister-group relationship between Cory- 

 phaena and Rachycentron is found in the morphology of their 

 larvae. As noted above all echeneoid larvae have a similar body 

 form and pattern of development, but the elongate, flattened 

 head of larval echeneidids lacks ornamentation. In contrast, 

 larval dolphin and cobia share identical patterns of head spi- 

 nation: a small posttemporal spine; several spines on the pos- 

 terior and lateral margin of the preopercle, including one en- 

 larged spine on either side of its angle; and a very large, 

 posterolaterally directed spine on the supraorbital ridge of each 

 frontal bone. Another obvious feature is the presence of laterally 

 swollen pterotics, previously described in Coryphaena as blunt 



sphenotic spines (Gibbs and Collette, 1959). This specific pat- 

 tern of head spines is distinctive, but similar features occur in 

 various combinations among carangid larvae, and it is pre- 

 mature to interpret this configuration as synapomorphic for 

 Coryphaena and Rachycentron until detailed comparisons with 

 carangids have been made. 



A specialization clearly unique to the larvae of dolphin and 

 cobia, however, is a modified epithelial cuticle in which are 

 borne minute crown-shaped spicules (Figs. 263A-D, 264A, B). 

 The cuticle itself is composed of large, multinucleate "cells," 

 40-100 m in diameter, that appear to continually produce and 

 slough-off" the thorny spicules. Each epithelial "cell" produces 

 one spicule, so that these extraordinary structures cover all ex- 

 posed body surfaces, excluding the pupil of the eye, giving the 

 integument a bristly appearance under magnification (Fig. 264A). 

 They first appear at about 8mm and are present in some indi- 

 viduals as large as 100 mm. Further histological work and elec- 

 tron microscopy will be necessary to determine the composition 

 of the spicules, which may be keratinous. It is clear, however, 

 that they are neither bony nor cartilaginous. Their function is 

 unknown, but as with spinous scales, it seems likely that they 

 are defensive. 



The surface and cellular composition of the epithelium of 

 larval echeneidids appear normal, but some modification of the 

 larval epithelium may actually be a primitive feature of car- 

 angoids. In larvae of trachinotine and naucratine carangids ex- 

 amined thus far (Trachinotus, Naucrates, Seriola) the epithelial 

 cells are of normal size ( ~ 8- 1 2 m), but their surfaces bear clusters 

 of bumplike structures, seemingly the result of keratinization 

 (Fig. 264C, D). Absence of these modified epithelial cells in 

 larvae of carangine carangids is parsimoniously interpreted as 

 secondary (see Laroche et al., this volume). Their presence in 

 the larvae of Neinalistius (curtently unknown) would corrobo- 

 rate the hypothesis that modified larval epithelium is primitive 

 for carangoids and thus also for echeneioids, suggesting that it 

 has been lost in carangines and echeneidids. 



The multinucleate epithelial cells and enlarged, thorny spic- 

 ules of larval Coryphaena and Rachycentron represent a com- 

 plex, shared specialization, unique among percoids. The phy- 

 logenetic significance of this synapomorphy is lessened only by 

 the unlikely possibility that loss of a modified epithelium in 

 echeneidids occurted after development of multinucleate cells 

 and spicules. Available evidence strongly points to a Cory- 

 phaena-Rachycentron sister-group relationship, and it should 

 be clear that further investigations testing this hypothesis must 

 integrate larval, adult and developmental characters. 



In conclusion, the study of early life history stages of fishes 

 has traditionally been treated as a discipline somewhat removed 

 from the mainstream of systematic ichthyology. As a result, 

 larval morphology has rarely beeen incorporated into studies of 

 evolutionary relationships of fishes. It is evident that the larvae 

 of percoid fishes exhibit a prodigious array of complexity and 

 diversity that offers exceptional potential applicability to phy- 

 logenetic studies. Recognition and application of this potential 

 will be an important step in understanding the complex evo- 

 lutionary history of the Percoidei. 



South Carolina Wildlife and Marine Resources De- 

 partment, Post Office Box 12559, Charleston, South 

 Carolina 29412. Present Address: Fish Division, Na- 

 tional Museum of Natural History, Washington, 

 District of Columbia 20560. 



