pooled and used to calculate the number of samples required to detect a 

 specified ratio R of the geometric means (corresponding to a difference 

 of In transformed means) at certain a and 3 levels (Table 8). 



Historically, nonpar ametric analyses of variance based on ranks 

 were used to test for differences between yearly, bimonthly, and station 

 means. However, since appropriate transformations allowed the use of 

 normal theory, and because limits of detectability were difficult to 

 determine for nonparametric tests, this approach was not pursued further. 



To determine if significant changes occurred in the demersal percent 

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

 proportions was applied to the data. Since this procedure does not 

 involve a CPUE calculation, data collected prior to October 1977 were 

 included. Also data from some stations were grouped into regions. The 

 results, presented in Table 9, indicated that the composition was signi- 

 ficantly affected by sample period, region and the interactions of 

 sample period with year and sample period with region. 



Finally, the time series approach to the data produced models that, 

 to some extent, described the patterns of abundance of the demersal 

 fish. Additionally, the models produced forecasts that could be compared 

 to actual data. The results from the time series investigations are 

 presented in Table 10. 



DISCUSSION 

 In order to address the first objective, i.e., describe the occurrence, 

 distribution and relative abundance of potentially impacted demersal 

 finfish in the Greater Millstone Bight, 'demersal' and 'potentially 

 impacted' were defined. The demersal finfish were considered to be 



27 



