318 



ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 



is found in Echiodon and pyramodontines while an expanded 

 third rib is found in Carapus (Olney and Markle, 1979; Markle 

 and Olney, 1980; Williams and Shipp, 1982) and Oimxodon 

 (Courtenay and McKittrick, 1 970). The sexually dimorphic and 

 interspecific differences in swimbladder morphology of ophi- 

 diines appear only in juveniles and adults and are not useful in 

 distinguishing larvae. 



The visceral cradle, formed from the criss-crossing elongate 

 proximal pterygiophores of the anterior anal fin rays, is a unique 

 specialization of Pyramodon (Markle and Olney, 1980). Its pre- 

 sumed precursor, non-crossing elongate proximal pterygio- 

 phores, is found in larval Snydehdia. The elongate proximal 

 pterygiophores found in pyramodontines are conspicuous in 

 larvae. 



The pectoral fin of carapids is a variable structure and po- 

 tentially useful in the study of relationships as well as for iden- 

 tification. Adults of some species of Encheliophis completely 

 lack a pectoral fin while pyramodontines have a well-developed 

 fin with up to 29 rays. Most cleared and stained carapid and 

 ophidiine larvae have an elongate, cartilaginous, ventral process 

 of the coracoid (VPC). In the carapid "exterilium" larvae (Fig. 

 16 IB, see also Robertson, 1975b) the development of the VPC 

 has been carried to an extreme. The hanging or trailing gut of 

 this larva is supported by a skeleton of the two VPC's which 

 intertwine with the intestine. Support of a trailing gut by VPC's 

 is not unique since we have also seen it in the ophidioid "ex- 

 terilium" (Fraser and Smith, 1974; Moser, 1981) and Symphii- 

 rus minor (unpublished data). 



The dentition of carapids is useful for adult identification 

 purposes (Arnold, 1956) and enlarged canines as well as the 

 dentary diastema have been used to separate Carapus and 

 Echiodon larvae (Olney and Markle, 1979). 



Relationships 



Intra-ordinal relationships.— T\\t classification of Ophidi- 

 iformes proposed by Cohen and Nielsen (1978) differs most 

 significantly from earlier classifications in the use of mode of 

 reproduction as a subordinal character. Previous classifications 

 recognized the highly specialized carapids as either one or two 

 families (Carapidae and Pyramadontidae) and, based on the 

 position of the pelvic fins, divided the remaining ophidiiforms 

 into two groups, the ophidiids (ophidiines, pelvics mental) and 

 the brotulids (pelvics absent or jugular). 



Relationships within the Bythitoidei remain unclear. The 

 aphyonids share a number of neotenic characters serving to 

 define the family. This may be a polyphyletic group, however, 

 with common character states reflecting convergent trends (Co- 

 hen and Nielsen, 1 978). Comparisons ofembryonic adaptations, 

 such as trophotaeniae (Wourms and Cohen, 1975), may prove 

 useful in resolving systematic problems within Bythitidae. Two 

 subfamilies (Bythitinae and Brosmophycinae) are presently de- 

 fined on the basis of confluence of anal and dorsal fins with the 

 caudal fin, though neither subfamily has been adequately stud- 

 ied. 



Ophidioidei is defined by the presence of oviparity and the 

 anterior nostril (in most genera) well above the upper lip. The 

 relationships of the ophidioid subfamilies are also uncertain and 

 the suborder may be paraphyletic. Carapidae and subfamily 

 Ophidiinae each seem to form natural groupings based upon 

 well-defined synapomorphies. Further study of the neobythi- 

 tines may reveal several natural groupings (Cohen and Nielsen, 

 1978). The relationships of Brotulotaeniinae and Brotulinae are 

 unknown. 



Two tribes of Ophidiinae can be defined on the basis of squa- 

 mation and the presence of pyloric caecae. Lepophidiini (im- 

 bricate scales; pyloric caecae present) contains three genera: the 

 monotypic Cherublemma emmelas. Genypierus, and Lepophi- 

 diiim. Lea (1980) has proposed the elevation of Genypterus to 

 the level of tribe. The Ophidiini (anguilloid squamation; pyloric 

 caeca absent) contains the genera Ophidian, Otophidium. Chi- 

 lara, Raneya and Parophidion. These genera, established on the 

 basis of meristics. morphometries, swimbladder morphology 

 and squamation, are not well-defined and require further study. 



A comparative study of the development of ophidiine larvae 

 of three nominal genera, Ophidion. Otophidium and Lepophi- 

 dium, suggests that body shape, development of the caudal fin 

 and pigmentation can provide useful taxonomic characters 

 (Gordon, 1982). The body shape and development of Lepo- 

 phidium larvae may represent the primitive state for the subfam- 

 ily. Otophidium omostigmum and most Ophidion species retain 

 this morphology, as does Parophidion (Fig. 1 59F; Padoa, 1 956i). 

 The morphology and development of O. selcnops and O. no- 

 comis. however, differ markedly from that of other ophidiine 

 larvae. The possession of an elongate larva is a derived character 

 uniting these two species. 



Robins and Bohike (1959) recognized the close relationship 

 between O. selenops and O. nocomis based upon the shared 

 possession of a well-developed rostral spine, similar to that 

 found in Lepophidium, and the tendency for the dorsal fin to 

 originate relatively far back on the body. The larvae of Chilara 

 taylori are slightly more elongate than typical ophidiine larvae, 

 but bear no close similarity to the larvae of O. selenops. 



A character shared by all Lepophidium larvae examined by 

 Gordon ( 1 982) is the presence of an elongate cartilaginous epural 

 which ossifies by 15 mm SL. All larvae of the tribe Ophidiini 

 develop a short cartilaginous epural by 10 mm, but the epural 

 never ossifies and is not visible by 15 mm SL. The presence of 

 an epural in the caudal skeleton of the adults is presumably the 

 primitive character state for the subfamily. 



The shared pigmentation pattern oi Lepophidium larvae unites 

 these species. This character may not extend throughout the 

 tribe, however. Brownell (1979) illustrates a Genypterus larva 

 (day eight) that has a pigmentation pattern similar to that of 

 Lepophidium. Other stages resembled Ophidion, however, with 

 stellate melanophores scattered laterally on the body. The pos- 

 session of similarly pigmented larvae by closely related species 

 in Ophidion argues for the validity of pigmentation as a character 

 to show phyletic relationships. The larval pigmentation of O. 

 selenops and O. nocomis shows only slight differences as does 

 larval pigmentation of O. welshi and O. marginatum. If the 

 proposed identities of Ophidion Type I and Type 2 and Le- 

 pophidiumType 1 are correct (Gordon, 1982), then species which 

 these types represent are presumably closely related. 



Adult carapids are morphologically conservative and present 

 some difficulty in identification and elucidation of phylogenetic 

 relationships. Larvae, on the other hand, are reasonably well- 

 known for all genera, fall into fairly distinct morphological groups 

 and provide morphological diversity which is potentially useful 

 in understanding intra-familial relationships (Olney and Mar- 

 kle, 1979; Markle and Olney, 1 980). Robins and Nielsen ( 1 970) 

 and Cohen and Nielsen (1978) recognized a single family, Carap- 

 idae, consisting of two subfamilies: the Pyramodontinae with 

 two genera, Pyramodon and Snyderidia: and the Carapinae with 

 four genera, Carapus, Echiodon, Encheliophis and Onu.xodon. 

 However, Gosline ( 1 960b) and Trott (1981) considered the Pyr- 

 amodontidae a separate family while Arnold (1956) ignored this 



