702 



Fishery Bulletin 94(4), 1996 



length, all again showing the divergence of O. 

 atlanticum , but this time defining some structure 

 to the other species. The 50% majority-rule consen- 

 sus tree is shown in Figure 4. The most significant 

 difference in this analysis from the two phenetic 

 analyses is the reversal of the positions of the two 

 Allocyttus species. In the cladistic analysis A. niger 

 is separated from the other species, whereas A. 

 verrucosus was grouped with either P. maculatus and 

 N. sp. or the other two Neocyttus species. Applying a 

 topological constraint to the search (enforcing pre- 

 determined groupings and keeping only those trees 



that satisfy the constraints) for the three Neocyttus 

 species or the two Allocyttus species produced the 

 shortest trees only one step longer than the most par- 

 simonious under no-constraint searches. A search for 

 near-optimal trees with lengths of 53 units produced 

 56 trees, with 98% confirming the branch separa- 

 tion of O. atlanticum, and 71% supporting the sepa- 

 ration of N. rhomboidalis and N. helgae from the 

 other four species, among which the branching points 

 could not be resolved. 



The zeids appear to be the most likely sister fam- 

 ily to the oreosomatids after the cladistic analyses. 



