igo ELOPIFORM FISHES 



Features 43-46 are advanced to some degree over the basic teleost condition 

 but may be seen in a large number of distantly related teleosts and have probably 

 arisen independently in a number of lineages. They are therefore of limited use in 

 indicating relationship. 



Only three characters (nos. 40, 41 and 42 2 ) indicate that the Elopiformes represent 

 an interrelated assemblage of fishes distinct from other ' lower ' teleosts. Elopiforms 

 show the specialized character combination of a leptocephalus, rostral ossicles and 

 pectoral splint bone. The leptocephalus larva has been used by Greenwood et al. 

 (1966) as a feature relating the Elopiformes, Anguilliformes and Notacanthiformes 

 as a distinct, natural lineage. 



A note of caution must be introduced here since it is not certain whether the 

 development of the leptocephalus is a specialized condition or if it is simply another 

 feature retained from halecostome ancestors. Harrisson (1966) and Romer (1966) 

 both imply that the leptocephalus is a retentional feature, but give no evidence for 

 this. 



A leptocephalus occurs in three orders (also linked by other intergroup charac- 

 teristics, Greenwood et al. 1966) which are anatomically, ecologically and biologically 

 diverse. Yet it is present in these three orders only. The absence of a leptocephalus, 

 or anything approaching such an ontogenetic stage, in clupeomorphs, osteoglosso- 

 morphs or protacanthopterygians is perhaps the strongest argument for assuming 

 that this larva evolved only within the elopiforms. Furthermore, the peculiarities 

 of the leptocephalus suggest that it arose but once and must have been present in 

 early Cretaceous times, before the elopoid-albuloid dichotomy. In short, circum- 

 stantial evidence is in favour of the leptocephalus being an elopiform specialization 

 subsequently inherited by the eels and notacanths (+ halosaurs and lipogenyids). 



If, in the future, it should be demonstrated that the leptocephalus is a primitive 

 or retained feature then its use as an indicator of interordinal relationship will no 

 longer be valid and it will have to be added to the already long list of elopiform charac- 

 ters that have been inherited from halecostome ancestors. 



A second feature that is here assumed to be an elopiform development is the 

 possession of rostral ossicles. The snout of teleosts appears to be very labile, 

 subject to many modifications related to feeding mechanisms. The sensory canals 

 and associated ossifications are no less labile and may be seen to vary considerably 

 both between and within each family. Within the Elopiformes, variation in the 

 sensory canals of the snout has involved the movement, loss, or even multiplication 

 of rostral ossicles, the existence of which is unique to the Elopiformes and Nota- 

 canthiformes. 



The origin of rostral ossicles as seen in Elops has been discussed by Nybelin (1956, 

 1967a), who suggested that they arose by fragmentation of the dermethmoid (rostral 



2 (Added in proof.) Jessen (1972) has since published an excellent study of the pectoral girdle and fin 

 of actinopterygians. Jessen suggests that the pectoral splint is a fin-ray derivative and is a primitive 

 character, being found (in a more primitive state) in Lepisosteus and Amia. Jessen 's conclusions are 

 accepted here. The presence of a pectoral splint in elopiforms must therefore be regarded as of little 

 use in linking elopoids with albuloids. The pectoral splint is nevertheless diagnostic for elopoids and 

 albuloids among teleostean fishes. The present author knows of no other teleost with a pectoral splint, 

 implying that neither elopoids nor albuloids could have been derived from known teleosteans. 



