Based on these considerations, a sample size of eight replicate cores 

 was chosen. The sample size contained 76 percent of the number of species 

 in 28 cores, and only about 20 of these species were important numerically; 

 i.e., they contained 87 percent of the total individuals. The confidence 

 limits for the 20 species indicate how accurately each species' abundance 

 can be estimated with eight cores as compared to more or less; 8 appears to 

 be the optimum number in most cases. In addition, the number of species 

 per core is estimated well with only four cores; eight is satisfactory for 

 the number of individuals per major group. Figure A-3 indicates that eight 

 replicates also provide a good estimate of species diversity and evenness. 



IV. SAMPLE DISTRIBUTION 



It was beyond the scope of this study to examine the detailed patterns 

 of distribution of the species sampled. However, to construct a quantita- 

 tive sampling plan it is necessary to have some measure of the gross 

 patchiness of the fauna. Sampling was stratified once by locating stations 

 at different depths, but it was unknown whether the magnitude of spatial 

 variation of the infauna warranted further stratification of sampling with- 

 in a depth contour. 



To determine the gross patchiness of the infauna, four sets of repli-. 

 cate cores were taken randomly from progressively larger areas: 10 from 

 each of the two 2-meter squares, and 8 from a 10-meter square and a 1.6- 

 kilometer swath approximately 200 meters wide. The distribution of the 

 samples at the 20-meter control station is shown in Figure A-l- If any of 

 the major parameters changed significantly from one area to the next, it 

 might be possible to adjust the sampling plan to produce a more accurate 

 representation of the fauna. 



Means of individuals per core of the dominant species were calculated 

 for each area and tested with the Kruskal-Wallis test and Wilcoxon-Mann- 

 Whitney a posteriori test. Tlie species are listed by abundance in Table 

 A-2; significant differences of means, species, diversity, and evenness 

 are also indicated. Table A-3 is a matrix of the number of significant 

 differences between areas for each species, summed from Table A-2. 



Similarity coefficients (Bray and Curtis, 1957) were computed for all 

 the possible pairings of the means of the dominant species (two individuals 

 per core or more) for each of the four areas (Table A-4) . The index has 

 no statistical basis so that differences cannot be tested. It equals the 

 sum of the lower relative abundance values for all the species common to 

 both of the samples being compared. 



Differences or similarities between the four areas may also be char- 

 acterized by changes in the rank order of the dominant species as listed 

 in Table A-2. Notable differences between each area and the remaining 

 three are: (a) In area 2M^-a, T. modesta and M. saoQulata exchanged rank, 

 and the abundance of P. oirrifera and H. calif omiensis was higher; (b) 

 in area 2M2-b, the abundance of W . elegans was the lowest; (c) in the 1.6- 

 kilometer swath, M. saooulata and P. pygmaeus exchanged rank, and 



71 



