30, 



The species similarity measure relies on the degree of faunal 

 overlap, in terms of species shared, between the samples. 

 The clustering of the samples is based on their entire faunal 

 complement, and thus tends to group samples of similar taxo- 

 nomic composition. Conversely, principal components analysis 

 attempts to group species which have similar distributions, 

 such that the maximum amount of variance in the data set is 

 accounted for by the fewest species or faunal assemblages. 

 The taxonomic composition of each sample is then explained 

 in terms of these species. Both of these methodologies gave 

 similar results. 



Community analysis showed some marked differences in 

 faunal assemblages between the three canyons studied. While 

 a few basic patterns common to these canyons may be identified 

 local variations within the individual canyons, such as 

 substrate differences, tend to enhance faunal variability. 

 Baltimore Canyon exhibits the least amount of substrate 

 variability of any of the three canyons, thus the fauna of this 

 canyon tends to resemble that found on the continental slope. 



Species similarity measures between the areas surveyed in 

 Baltimore Canyon identify a diverse shallow-water assemblage 

 with a somewhat more homogeneous deeper one (Fig. 12) . The 

 dominant species in the shallower parts of Baltimore Canyon are 

 the galatheid crab Munida valida^ the starfish Astevias vulgaris ^ 

 an unidentified fluffy white sea pen, the jonah crab Canaer 



