Stanley et al.: Diel vertical migration by Sebastes flavtdus 



'ill 



Table 5 



Results of Bartlett's test for significant difference between 

 variance of diurnal and nocturnal observations (X" for 

 P=0.01 is 2.706; x^ for P=0.05 is 3.841 ). 



Short section 



Long section 



2.45 



2.23 



geneity of variance between the two diel periods 

 (Table 6). These data included the observations from 

 the comparisons above; therefore, the data sets and 

 tests are not independent of the main ANOVA. 



Results for the short segments for B5 and B6 also 

 showed no significant differences between mean noc- 

 turnal and diurnal biomass. However, unlike the re- 

 sults of the expanded B5 data set and the original 

 all-transect test, diurnal variance for B6 was signifi- 

 cantly higher than nocturnal variance for this 

 transect. 



Missing observation 



The impact of using an average value (69 t) for the 

 missing nocturnal observation was examined by con- 

 ducting the randomization tests with minimum and 

 maximum values of 56 t and 76 t for the nocturnal 

 observations from B4. The alternate values had a 

 negligible impact on the results. 



Nocturnal dispersion beyond the shallow end of the 

 transect 



During one night series, the echograms indicated that 

 the cliff aggregations dispersed beyond our empiri- 

 cally chosen shallow endpoint in transects B5-B9. 

 We therefore slightly underestimated the nocturnal 

 abundance of the cliff aggregations this one night. 

 We re-analyzed the biomass for these transects with 

 the new endpoints and then repeated the statistical 

 analyses for the B2-B9 and B5-B6 series only. The 

 changes to the one night's observations did not af- 

 fect the results significantly. 



Table 6 



Results of comparison of expanded observation set from 

 transects B5 and B6 ( * indicates > 5% level of significance). 

 Percentages indicate the % of 4999 random combinations 

 of the observed data which resulted in a difference between 

 treatment means greater than the observed difference in 

 means. 



% greater than observed 



Long 

 section 



22.54 

 15.90 



10.08 

 1.60* 



Variance in transect and total biomass estimates 



Independent total biomass estimates for the study 

 area were estimated as the sum of the transect bio- 

 mass estimates from each series (Table 7) . The over- 

 all average was 1152 t with a CV of 13.9*7^ among 

 series for the short sections. For the long sections, 

 the average was 1329 t with a CV of 15.4%. The 

 within-transect CVs ranged from 16.6% to 65.8% for 

 the eight transects (short sections). 



Aglen (1983) derived an empirical formula for pre- 

 dicting the CV based on the ratio of linear distance 

 surveyed (L) over the square root of the surveyed area 

 (A), which he called the "degree of coverage" or DOC: 



Predicted CV = 0.5 



L 

 A 



-0 41 



The variance among the eight series replicates of the 

 area were significantly less than the predicted value 



