A-26. 



However, three major faunal zones superimposed on 

 several minor ones were identified by the rotated principal 

 components analysis. The shelf and upper slope (100-350 m) 

 samples can be mainly explained by the presence of four 

 species. The middle slope (800-1300 m) samples were domi- 

 nated by fish, while the deeper slope ( 1500 m) samples were 

 dominated by three species. The samples from the canyons, 

 on the whole, tend to be composed of a more complex mixture 

 of species, with each species or pair of species being rather 

 independent of the others. Also there are non-consistent 

 variations of assemblages with depth on the canyon walls, 

 which tend to indicate a very patchy distribution. 



Two of the dominant taxa, fish and a species of sponge 

 from the genus Cladorhiza , were then excluded from the second 

 run; the rationale being that these taxa were either exceed- 

 ingly dominant or had not been separated into distinct 

 species. These taxa were omitted from the next analysis to 

 ascertain whether they might be obscuring fine scale varia- 

 tion in some of the other components. The first three end- 

 members of this run were monospecific, again meaning that 

 the distribution of the dominant taxa are probably being 

 controlled by different parameters. Sixty-eight percent of 

 the sample variance could be accounted for by the first 10 

 end-members, with 35% of the variance being explained by 

 the first three monospecific end-members. 



