24 



currences per species. Another 22 species formed combinations of not less 

 than three species with those species having between five and seven 

 occurrences. Since almost every index species listed for one group may also 

 be part of several other groups, in actuality, only 10 or 12 valid distinct 

 groupings of species resulted. Most of these groups occurred on the conti- 

 nental shelf in from 11 to 126 meters. 



The next step in producing stations and faunas which characterize 

 particular environments, was to perform a separate analysis by personal 

 inspection of the index groups with the largest number of associated 

 species. This was done by listing all species in each group, and then listing 

 all station occurrences at which each species was taken, living only, 

 although the computor program was based on both living and dead 

 occurrences. It was then possible to determine only those living species 

 which were important components of the station groups. Those stations 

 which proved to have a large number of commonly associated living species 

 were then examined as to location and common physical factors. For- 

 tunately for this study, most of the resultant station groupings were very 

 distinct, both as to geographical location and the same range of physical 

 factors. Therefore, instead of attempting to determine what species oc- 

 curred in any one set of physical factors, the matrix actually provided 

 sets of stations with common physical attributes, bound together by index 

 or important living species. Finally, all living species taken at these 

 characteristic stations were listed, and those with the most station oc- 

 currences within the group of stations were listed for various environments 

 as given in Table II. In some cases, particularly environments I, II, IV 

 and V, lists of index species as given directly by the computor program 

 are given in the text, in order to show that there were certain species 

 occurrences which tied these stations together. More often than not, 

 however, these lists did not represent either the most characteristic species 

 nor the common living ones, and the somewhat more subjective analysis 

 which produced the lists in Table II, are more reliable indicators of 

 environment. The computor method itself appears to be an excellent tool 

 for defining environmental and faunal boundaries as pointed out by 

 Fager (1963). Unfortunately, the sampling as produced by this project 

 was inadequate to do justice to the problem. As can be seen in Table III 

 (the lists of sampling devices for each environment), a strong bias will be 

 produced as to the kinds of animals which will predominate in each 

 environment. Certain areas were sampled primarily by shell dredges, 

 others by large trawls, and one by grab samplers (fig. 2b). Although most 

 of the environments were sampled by a variety of instruments, there were 



