6-8 



H' . is defined as the lowest possible diversity (H') value for a 

 population with a given number of species. This occurs when all species 

 but one are represented by a single individual and is calculated as: 



1 N' 



min N ^ (N-S+1) ! 



Normal (Q-mode) and inverse (R-mode) cluster analyses were 

 performed on data from each sampling period separately and for all 

 sampling periods combined. In all cases the mean value of all repli- 

 cates was used for each species. All classification analyses were 

 computed using the program SPSTCL at the WHOI computing center. No 

 transformations were applied to the data. The similarity coefficient 

 used was percent similarity (standardized Bray-Curtis) and the cluster- 

 ing strategy was group average sorting. 



Normal cluster analyses were conducted on the results for 

 individual sampling events from the R & M study from March 1974 to 

 January 1978. NAI samples that had been collected within one month of 

 the R St M samples were included in the analyses . Norinal and inverse 

 analyses were also computed on the annual data-set for each of the four 

 years. In each inverse analysis only the 40 most abundant species for 

 the particular year were considered. Nodal analyses, in which the 

 normal and inverse classification results are compared in a single 

 matrix, were prepared for 1974 and 1977. Because no consistent station 

 groupings had been identified at the time the nodal analyses were com- 

 puted, constancy values were calculated for individual stations and 

 species. The nodal analyses did not add any new information and, 

 hence, are not presented in this report. 



Means of various population parameters (species richness, 

 faunal density, diversity) were tested for significant differences over 

 space and time using standard paired and unpaired Student's t proce- 

 dures. In all cases two-tailed tests were used with a rejection cri- 

 terion of 0.05. Spearman's coefficient of rank correlation was used to 



