The ACFs of the total CPUE and selected species catch at JC are typical 

 of all stations and are presented in Figure 4. It is apparent that 

 significant autocorrelations do exist among the data points. Glass et 

 al. (1975) have concluded that the effect of autocorrelations on proba- 

 bility statements cannot be designated. 



The last aspects of the data to be examined were the variance 

 components. Based on the analysis of variance model presented earlier, 

 estimates of the variance components, Ss, St and Se for month, station 

 and error respectively, were calculated and are presented in Table 8. 

 When these variance components were divided by 8 (number of sampling 

 months per year) , 6 (number of stations sampled per sampling month) , and 

 144 (total number of samples collected per year) respectively, the 

 resulting variance terms were used to calculate the number of stations 

 to sample and the number of times per year to sample to minimize the 

 variance given the current cost of the seining program (Table 9) . The 

 error variances were used to calculate the number of samples required to 

 detect a specified difference of the means of the log transformed catch 

 data (d) , corresponding to the indicated ratio of the geometric means 

 (R) at certain a and 6 levels (Table 10). 



All of the characteristics of the data base indicated that the use 

 of standard analysis of variance tests might not be appropriate. Thus a 

 nonparametric analysis of variance based on ranks was used to test for 

 differences between yearly, bimonthly, and station means. The results 

 are presented in Table 11. 



To determine if significant changes occurred in the shore-zone 

 percent species composition, the method of Krumbein and Tukey (1956) of 

 analysing proportions was applied to the data. The results, presented 



57 



