326 



Fishery Bulletin 97(2), 1999 



sions of the transect, suggesting some added contami- 

 nation; however most of the backscatter was still con- 

 centrated near the cliffs. 



A second aggregation type was represented by the 

 schools observed at the shelf break, 10-20 m deeper 

 than the base of the cliffs (Fig. 3). These schools, al- 

 though exhibiting diel behavior similar to that of the 

 first group, remained closer to the bottom during 

 daytime. The greater depth of the aggregations ( 160- 

 180 m), their closer proximity to the shelf edge, and 

 their stronger bottom affinity suggested to us, as well 

 as to various fishermen with whom we consulted, that 

 these aggi-egations probably consisted of a combina- 

 tion of Pacific ocean perch (S. alutiis). canary (S. 

 pinniger), redstripe (S. proriger), sharpchin (S. zacen- 

 trus). and silvergray iS. breuispinis) rockfish (Leaman 

 and Nagtegaal, 1982, 1986; Leaman et al., 1990). 



The third and fourth target types included small 

 schools or individual targets within the shallow end of 

 the transects and scattered distributions over much 

 deeper water beyond our transects and the shelf edge. 

 The shallower signals were thought to be plankton or 

 small fishes. Fishermen reported that attempts to fish 

 on these aggregations were unsuccessful and suggested 

 that the source of the signals was too small to be re- 

 tained by their gear. No in situ target strength mea- 

 surements were conducted. Signals in deeper water, 

 off the edge of the shelf break, have been shown in other 

 studies to be a deep plankton layer or hake (Merluccius 

 productiis) (SaundersM. Signals close to the slope off 

 the edge could also be generated by side echoes. 



Analysis of variance 



The randomization tests provide the percent of the 

 random re-orderings that exceed the value of the 



Table 4 



Results of three-factor randomization ANOVA (df=degrees 



of freedom; **>19c significance level). Percentages indicate 

 the 'i of 4999 random combinations of the observed data 

 which resulted in a difference between treatments greater 

 than the observed difference. 



% greater than observed 



Factor 



df Short section Long section 



47.40 



**0.08 



66.89 



' Saunders, M. 1991. Pacific Biological Station, Nanaimo.B. C, 

 Canada V9R 5K6. Personal conimun. 



observed response. Statistical significance is assumed 

 when fewer than 5% of the re-orderings exceed 

 the observed response. The three-factor ANOVA 

 indicated for both short and long transect sections 

 (Table 4) that 



1 there was no significant difference in mean bio- 

 mass between night and day; 



2 there was a highly significant difference in mean 

 biomass among transects; and 



3 there was no significant difference in mean bio- 

 mass among different series (over time). 



Bartlett's test for homogeneity of variance indicated 

 no significant difference between the variance of di- 

 urnal and nocturnal observations, although the trend 

 was towards greater diurnal variance (Table 5). 



We repeated the same basic analysis using the 

 larger but nonpaired set of observations for transects 

 B5 and B6 (Table 3). Each transect was analyzed 

 separately to test for significant differences between 

 diurnal and nocturnal mean biomass and for homo- 



