196 ELOPIFORM FISHES 



neural arch associated with the first preural centrum is poorly developed (the neural 

 arch of the first ural centrum is not known). The epurals of these two species are 

 of unequal length and form a graded series. 



Significantly, the elopiform Anaethalion vidali shows the development of laminar 

 bone and moderately expanded neural arches associated with the first preural and 

 ural centra, but the epurals form a graded series (Text-fig. 17). 



Anaethalion cannot be regarded as ancestral to both the elopiforms and euteleo- 

 steans (assuming the latter to be monophyletic) since there are euteleosts such as 

 alepocephaloids and the osmerid-like Humbertia more primitive than Anaethalion 

 in possessing a basipterygoid process (for alepocephaloids see Gosline 1969 ; for 

 Humbertia see Patterson 1970b). 



Considering the similarity between the caudal skeleton of primitive elopiforms 

 and euteleosts, there are two conclusions to be drawn. Either the development of 

 laminar bone has little significance in indicating relationship, having been developed 

 in parallel, or a true relationship exists, beneath the Anaethalion level, from an an- 

 cestor very similar to Anaethalion in which there was a basipterygoid process. 



In summary, the Elopiformes show no relationship with either the Clupeomorpha 

 or the Osteoglossomorpha above the pholidophorid level. A relationship with the 

 euteleosts is suggested by the caudal anatomy of Anaethalion but definite evidence is 

 lacking. 



The order Elopiformes is constituted by two radically different suborders, the 

 Elopoidei and the Albuloidei. Each suborder shows trends not seen in the other, 

 and it is difficult to discern any trend common to both. The elopoids retain the 

 greatest number of primitive characters and appear in the fossil record well before 

 the albuloids. On these criteria (and see below) the albuloids are considered to have 

 evolved from the elopoids. 



Albuloids may be traced back to the Albian, where they are represented by 

 Osmeroides lewesiensis, a form that exhibits an essentially albuloid neurocranium 

 but resembles elopids rather than other albuloids in many primitive features (e.g. 

 48, 49, 50, 51, 55, 56, 58, 59, 60, 62, 63, 68 and possibly 69 from the list on p. 189). 

 The dichotomy between the albuloids and the elopoids must therefore have taken 

 place at some time prior to the mid-Albian. Within the elopoids only the primitive 

 Elopidae could have been ancestral to the albuloids. The other elopoid family, the 

 Megalopidae, shows trends very different from those seen in albuloids, such as deepen- 

 ing of the head and body, enlarged scales, enlarged post-temporal fossae and the 

 development of an otophysic connection. 



(c) Suborder Elopoidei 



The suborder Elopoidei consists of the conservative Elopidae and the more 

 specialized Megalopidae. Opinion is divided as to whether all elopoids should be 

 grouped in a single family or whether to recognize two families, as is done here. 

 Many authors, writing from a primarily neontological standpoint, recognize a single 

 family, the Elopidae (Boulenger 1910 ; Barnard 1925 ; Fowler 1936 ; Bertin & 

 Arambourg 1958 ; Okada i960 ; Hildebrand 1963). Hildebrand (1963) speaks 

 against separate family status for Elops on the one hand and Megalops (and Tarpon) 



