NO. 5 BARNARD: AMPHIPODA 15 



is 47.7. The "standard" index of association computed by the conven- 

 tional method is only 21.8 because of swamping of Maldane by other 

 species in 4851. 



Station 4851 is also compared with 5674, a sample dominated by 

 Heteromastus and with a conventional index of association of 6.5. In the 

 alternate method the index is 45.6, indicating the importance of 

 Nothria iridescens, Praxillella pacifica and Heteromastus in both sam- 

 ples and suggesting the possibility that 4851 represents a bottom area on 

 which the Maldane community and its associates are mixed with a 

 Nothria iridescens subcommunity. Indeed, 4851 also has a significant 

 number of Pectinaria, more specimens than in some other Pectinaria- 

 dominated samples. There is no way to account for such mixed samples 

 as 4851, or for species-impoverished samples such as 5674, or samples 

 with impoverishment of the dominant species, except to assume that 

 large sampling devices do collect closely contiguous but independent as- 

 semblages. 



As a comparison, samples from the well-explored Amphiodia urtica 

 community off southern California were interrelated using the trellis- 

 presentation (Graph 9). Out of 67 samples, 20 were selected as having 

 180 or more specimens of Amphiodia urtica (extending from 180 to 

 more than 600). Although some sample pairs had rather low indices of 

 association, all were clearly dominated by Amphiodia urtica. As in the 

 canyon samples, some A. urtica samples were swamped with numerous 

 individuals of other species. In some cases a few species existed in high 

 frequency and in other cases samples had nearly twice as many species 

 with low numbers of specimens, but both reduced the prominence of 

 the Amphiodia as a numerical dominant. Again, the overlaying of sub- 

 community matrices on the background of the basic community must be 

 considered, for within the Amphiodia community may be seen numer- 

 ous subcommunities that become prominent in certain samples, even 

 though the basic Amphiodia structure is maintained. Indices of associa- 

 tion extend from 24 to 75 with a median of 51. The principal domi- 

 nants of the Amphiodia community are summed as to their percent- 

 age compositions in the left column of Graph 9, with a range of 27% 

 to 81%. These samples are a good representation of the classical Peter- 

 sen marine community (Thorson, 1957), but their strong variation per- 

 mits wide latitude in the assignment of samples to a community nucleus. 

 Those samples with highest coefficient interrelationships are also those 

 samples generally having the most individuals of Amphiodia (Graph 

 9). Summation of the total coefficient values for each sample shows a 

 range of nearly 50% of the highest value. 



