GORDON ET AL.: OPHIDIIFORMES 



319 



group in his revision of carapids. Williams { 1 984) in his synopsis 

 considered it as a subfamily. The common possession of a vex- 

 illifer larva is the most convincing evidence that the genera of 

 Carapidae are monophyletic, thus we recognize one family. 



The genera Pyramodon and Snyderidia were considered closely 

 related by Robins and Nielsen (1970), and Markle and Olney 

 (1981), on the basis of osteological and larval characters, added 

 further support to this presumed relationship. It now appears 

 that many of the character states of these genera are primitive. 

 The pelvic fins, lost in all other carapids, are obviously a prim- 

 itive state since they are widely present in all other ophidiiforms. 

 Similarly, the dorsal origin is over or in advance of the anal fin 

 in all non-carapid ophidiiforms as well as in Pyramodon and 

 Snyderidia. The posterior placements of the first dorsal fin ray 

 or vexillum can therefore be considered advanced states. Thus, 

 the anterior placement of the vexillum relative to first anal ray 

 (a primitive state) in combination with a posteriorly placed first 

 dorsal fin ray (advanced state) appears to define larvae of Car- 

 apiis (Fig. 162) and presumably Encheliophis. The genera pos- 

 sess further derived states such as adult inquiline behavior and 

 parasitism (Trott, 1970). In addition, the tenuis stage, unknown 

 in Pyramodon and Snyderidia. may represent an advanced state, 

 namely retaining larval characters in the early benthic stage. 



Larvae of the genus Echiodon display a wide variety of mor- 

 phology especially in gut configuration, vexillum and first dorsal 

 fin ray position and dorsal pterygiophores (Fig. 161; Maul, 1976; 

 Olney and Markle, 1979; Markle et al., 1983). Williams and 

 Shipp (1982) consider Echiodon to be composed of two species 

 complexes and the gross morphology of larvae seems to support 

 this contention. In addition, the peculiar gut configuration of 

 E. rendahli (Fig. 161B, Robertson 1975b) represents another 

 extreme in morphological variability which suggests the genus 

 (as presently known) is polyphyletic. 



Inter-ordinal relationships. — Based upon anatomical similari- 

 ties shared with the cods, the ophidiiform fishes have been 

 treated as a suborder within Gadiformes (Greenwood et al., 

 1966; Rosen and Patterson, 1969). These similarities include 

 the development of the levator maxillae superioris and the struc- 

 ture of the caudal skeleton. Freihofer (1963, 1970) presented 

 further evidence for this relationship based upon the pattern of 

 the ramus lateralis accessorius nerve. Alternatively, these sim- 

 ilarities may be the result of convergence due to the require- 

 ments of bottom feeding behavior (Gosline, 1968; Fraser 1972b; 

 Marshall and Cohen, 1973). Similarities to the perciform fishes 

 in osteology (Gosline, 1968; Regan, 1912b) and biochemistry 

 (Shaklee and Whitt, 1981) point to a perciform origin of the 

 group. 



The structure and the development of the ophidiiform caudal 

 skeleton support the hypothesis of a closer relationship to the 



gadiform fishes than to the perciform fishes. In Brotula. as in 

 gadiforms, two separate ural centra support hypurals. In the 

 Ophidiinae, a single urostyle, which develops from a single car- 

 tilaginous structure in the larvae, supports two hypurals. This 

 urostyle is apparently homologous to the two ural centra of 

 Brotula. A vestigial neural arch develops on the urostyle, as on 

 the first ural centrum oi Brotula. Also, the last neural and haemal 

 spines in both Brotula and the Ophidiinae are modified. These 

 spines support caudal rays in Brotula and share in the support 

 of the last dorsal and anal rays in the Ophidiinae. In the gadiform 

 caudal skeleton, similarly modified spines on the first preural 

 centrum support caudal rays. In both gadiform fishes (Markle, 

 1982) and ophidiine fishes these spines remain cartilaginous on 

 the distal articular surface. 



The ophidiine caudal skeleton differs from perciform skele- 

 tons in the development of the hypural elements and last two 

 haemal arches. In ophidiine larvae, only two cartilaginous hy- 

 pural elements form, whereas five or more hypural elements are 

 typically present in the skeleton of larval perciforms. The last 

 two haemal arches in perciform fishes remain autogenous; these 

 arches fuse to the corresponding centra in the Ophidiinae. 



Ophidiiform larvae share other developmental features with 

 gadiform larvae. Larvae of both orders develop coiled guts (ex- 

 cept for aphyonid larvae) and larvae of Carapidae and Ma- 

 crouridae have high vertebral numbers resulting in elongate 

 larvae with reduced or absent caudal fins. Another similarity 

 apparent in the orders is the presence, in larvae of some species, 

 of modified anterior dorsal rays. In Ophidiiformes, this char- 

 acter IS present in larval Carapidae. In Gadiformes, somewhat 

 similar structures appear in larvae of Bregmaceros, Enchelyopus 

 and Muraenolepis although comparative studies of the gross 

 and micro-structure of these larval specializations are lacking 

 (Govoni et al., 1984). 



Cohen and Nielsen (1978) consider ophidiiform fishes to be 

 too poorly known to resolve questions of phylogeny. Our as- 

 sessment based on larval data is similar. Further comparative 

 studies focusing on the developmental osteology of such struc- 

 tures as the caudal fin, anterior vertebral column and pectoral 

 girdle, as well as the development of the gut, will allow mean- 

 ingful interpretation of the significance of these structures to 

 phyletic studies. 



(D.J.G.) RosENSTiEL School of Marine and Atmospheric 

 Science, University of Miami, 4600 Rickenbacker 

 Causeway, Miami, Florida 33149; (D.F.M.) The 

 Huntsman Marine Laboratory, St. Andrews, New 

 Brunswick EOG 2X0, Canada; (J.E.O.) Virginia Institute 

 OF Marine Science, College of William and Mary, 

 Gloucester Point, Virginia 23062. 



