Community Abundance 



Mean quarterly abundance (no. /core) of subtidal communities from March 1979 - March 1986 

 ranged from 60 to 393 at EF, 107 to 360 at GN, 9 to 303 at IN and 106 to 633 at JC, (values are 

 exponentials of those presented in Fig. 9). At EF, GN and .?C, abundance was generally highest in 

 June or September with declines evident during colder months of December and March. This seasonal 

 pattern was evident at IN until June 1984 after which peaks in density occurred in September and December. 



Regression analysis using abiotic factors as explanatory variables (i.e., covariates) removed 46%, 

 48%, 55% and 57% of the temporal variation in community abundance at EF, GN, IN, and JC, 

 respectively. After removing abiotic sources of variation, no significant long-term increases or decreases 

 in annual mean abundance (for 1980 - 1985 only) were evident. In fact, overall abundance at all stations 

 was relatively stable over the sampling period (Fig. 10). Inter-annual differences did occur; however, 

 and paired-t tests of adjusted annual mean revealed significant a < 0.05 differences at all stations. Except 

 at IN, these differences were generally due to the very high abundances that occurred in past years. For 

 example, the high density at JC in 1984, was significantly different from 1981 to 1983 and 1985. 

 Similarly, the high 1984 density at EF was significantly different from all previous years. In contrast, 

 the low density at IN in 1984 was significantly different from every other year except 1981. I^wer 

 density during 1981, at GN also resulted in significant differences with 1980 and 1984. In addition, the 

 low 1983 density was significantly different from densities in 1980, 1982, 1984 and 1985. 



Numbers of Species 



The mean quarterly species numbers (no. /core) at subtidal stations ranged from 5-23 at IN, 19-40 

 at J(^ 12-40 at GN and 13-46 at EF (Fig. 11). Species numbers were highest at EF and lowest at IN. 

 Seasonal patterns generally followed those of density (higher in June or September at F^F, GN and .IC 

 and December at IN). 



Multiple regressions of species numbers accounted for 75% (EF), 63% (JC), 61% (GN) and 47% 

 (IN) of the temporal variation observed since 1980. At EF the model included a significant increasing 

 trend which persisted after removal of quantifiable sources of variation (Fig. 12). Results of paired t-tests 

 showed that significant inter-annual differences occurred at all stations and were frequently in years in 



31 



