704 



Fishery Bulletin 94(4). 1996 



Elliott and Ward (1992) for the orange roughy, the 

 high heterozygosity shown by these deepwater (500- 

 1200 m) species may reflect their large (prior to ex- 

 ploitation) population sizes and (assuming that 

 deepwater species have been less severely affected 

 by glaciations than shallow water species) a lack of 

 severe bottlenecks in their recent evolutionary past. 



As adults, Pseudocyttus is the most morphologi- 

 cally distinct oreosomatid genus (James et al., 1988). 

 Its distinguishing characters include the first dor- 

 sal-fin spine being longer than the second (vice versa 

 in other species), a pelvic fin with only five rays (usu- 

 ally six or seven in other species), and 40-43 verte- 

 brae (34-41 in other species). However, as juveniles, 

 the genus Oreosoma is the most distinctive with 

 prominent cones over the body. Other juvenile 

 oreosomatids have "warts" or protuberances (such 

 structures are absent in at least N. rhomboidalis) , 

 but none are quite so pronounced or bizarre as in O. 

 atlanticum. Our genetic study confirms the unique- 

 ness of O. atlanticum, which has a very low genetic 

 identity (0.371) with the other oreosomatid species — 

 substantially less than the corresponding mean iden- 

 tity (0.650) of P. maculatus with other oreosomatids. 

 Morphologically, O. atlanticum can be distinguished 

 as an adult by a very large eye (eye diameter 52- 

 60% of head length) and by a prominent horizontal 

 ridge on the operculum. 



James et al. (1988) suggested that further study 

 may synonymize the Northern Hemisphere N. helgae 

 with the Southern Hemisphere iV. rhomboidalis. Our 

 allozyme data suggest that, although these two taxa 

 are indeed very closely related (genetic identity 

 1=0.973 for 26 loci and 1=0.966 for 27 loci including 

 GPI-1*), their distinctive muscle protein patterns, 

 not included in the genetic identity values, are con- 

 sistent with their being separate species. Of the four 

 non-CK-A* protein bands, two appear to be fixed dif- 

 ferently for the two species. However, the amount of 

 genetic differentiation between these two species is 

 only a little greater than that between samples of//. 

 atlanticus taken from the same two areas (North 

 Atlantic and off southern Australia) (Elliott et al., 

 1994). Eleven polymorphic loci were screened in the 

 H. atlanticus comparison and just three loci showed 

 significant heterogeneity and gave a genetic iden- 

 tity of 0.990 (N.G.E.'s unpubl. data). Thus the genetic 

 data do not unequivocally validate the recognition 

 of N. rhomboidalis and N. helgae as distinct species. 

 Morphologically they are also very similar, although 

 there are some differences (Yearsley and Last 2 ). 



As mentioned earlier, there is depth-related varia- 

 tion in the sSOD* polymorphism in N. rhomboidalis, 

 with samples derived from deeper water having a 

 high frequency of sSOD* 140. We are uncertain as 



yet whether this indicates reproductive isolation of 

 two forms or selection acting on sSOD, although the 

 lack of detectable mitochondrial DNA differentiation 

 (Grewe, Innes, and Evans 4 ) suggests that if repro- 

 ductive isolation is responsible, it is likely to be re- 

 cent in origin. These data and analyses will be pre- 

 sented in full elsewhere. 



The new species N. sp., infrequently captured with 

 P. maculatus and A. niger in southern Australian 

 waters and morphologically similar to N. rhom- 

 boidalis, showed quite a high degree of genetic simi- 

 larity to the other two Neocyttus species ( 1=0.903 with 

 N. rhomboidalis and 0.884 with N. helgae). However, 

 it was genetically distinct from them at several loci 

 (Table 6), and numerous meristic and morphological 

 characters (Yearsley and Last 2 ) confirm that it is a 

 separate species. Although it clustered with A. niger 

 (1=0.903) in the two phenetic trees constructed from 

 the genetic distance data, in the cladistic analyses it 

 grouped more often with P. maculatus. However, clas- 

 sical taxonomic techniques suggest a close association 

 with Neocyttus species, particularly the western Indian 

 Ocean N. acanthorhynchus (Yearsley and Last 2 ). 



The two Allocyttus species were found to be ge- 

 netically quite distinct from one another (1=0.695); 

 there was no evidence from either phenetic or cla- 

 distic analyses that they made up an exclusive mono- 

 phyletic group. James et al. ( 1988) gave no justifica- 

 tion for placing A. niger in Allocyttus. However, they 

 drew attention to problems with generic diagnoses 

 of the oreosomatids. Allocyttus, as it currently 

 stands, but excluding A. niger (i.e. A. verrucosus, A. 

 guineensis, and A. folletti), may be a natural group- 

 ing, with A. niger more akin to, but probably not con- 

 generic with, Neocyttus. Ongoing morphological work 

 should elucidate these problems. 



Whereas the branch node for O. atlanticum is 

 clearly resolved to be ancestral to the remaining 

 oreosomatids, the phenetic and cladistic analyses 

 could not resolve unambiguously all the internal 

 nodes for the remaining species. There is strong evi- 

 dence for a branch node separating TV. rhomboidalis 

 and N. helgae from the other four species, but there 

 is no evidence supporting two species in Allocyttus. 



Several problems in the family Oreosomatidae re- 

 main to be resolved. They include the formal descrip- 

 tion of the rough oreo, N. sp. (Yearsley and Last 2 ), 

 a reassessment of the generic affinities of A. niger 

 and A. verrucosus, and further examination of the 

 two depth-related sSOD* forms of N. rhomboidalis 

 and their relationship to N. helgae. As intimated by 



4 Grewe, P. M., B. H. Innes, and B. S. Evans. 1995. CSIRO 

 Division of Fisheries, Castray Esplanade, Hobart, Tasmania 

 7000, Australia. 



