FISHERY BULLETIN: VOL. 78, NO. 2 



Table 2. — Comparison of within-site and between-site means ( ±\ standard deviation) of similarity (/) and rank correlation (tau) 

 between all pairs of species-abundance arrays from yearly cinetransect samples filmed during 1971-74 at mainland and Santa Cruz 

 Island Study sites ofTSanta Barbara, southern California. Within-site means are of values for all pairs (1971 vs. 1972, 1971 vs. 1973, 

 . . ., 1973 vs. 1974), both members of which were filmed in canopy or bottom habitats either at the mainland or island site; between-site 

 means are of values for all such habitat-year pairs, one member of which was from an island sample, the other from a mamland 

 sample; and mean ratio is the between-site value/mean within-site value (mainland and island). (Figure 2 is a cluster diagram of 

 yearly samples, computed from all values of/.) 



'Difference of wittiin- and between-site means significant at (f-test) "P 05:"P 0.01, ""P 0,001 



Table 4 occurred because the relation was re- 

 versed in 1973 as indicated in Table 3.) Diversity 

 and evenness (Table 3, h and J) of canopy as- 

 semblages, however, did not differ significantly 

 between sites. 



However, sporadically abundant endemic is- 

 land seaperch were not included in these compari- 

 sons. Adding this species to 1974 counts of the 

 island-canopy samples increased the fish total 

 from 1,722 to 3,084 individuals, median fish 

 counts per transect from 31 to 64 individuals, and 

 median number of species per transect from^.5 to 

 6.0. The slight increase in species diversity (A) was 



not significant (Table 3). Island seaperch were sel- 

 dom observed in bottom transects. 



In contrast to the canopy assemblages, main- 

 land and island bottom assemblages were easily 

 distinguishable. Because between-site re- 

 semblance was comparatively small for arrays of 

 bottom species (Table 2), mainland and island 

 clusters of samples were sharply defined (Figure 

 2). Island samples contained relatively more 

 California sheephead, garibaldi, and opaleye, 

 Girella nigricans, and included rock wrasse, 

 Halichoeres semicinctus , which were not recorded 

 from the mainland (Table 1). Mainland-site sam- 



TABLE 3. — Yearly abundance and species diversity of canopy and bottom assemblages of kelp-bed fishes m cinetransect samples from 

 mainland and Santa Cruz Island study sites off Santa Barbara, southern California. Columns include: n, no. of transects in sample; 

 Geom. X, the antilog of the mean transformed fish count with 95% confidence limits; S, total species in sample; n{h), sample size to 

 compute /i; /i, the mean of successively pooled transect estimates of diversity per individual (see text); and J, evenness of distribution 

 of individuals among species in the sample. Contrasts among means that were shown to differ significantly by analysis of variance 

 (Table 4) were by the Student-Nevraian-Keuls procedure (Sokal and Rohlf 1969:239); means making up homogeneous subsets are 

 indicated by X's in the same column. 



' Including counts of Cymalogasler gracilis 

 ^Difference between unweighted means of 



mainland and island values significant at P = 0,05 



368 



