Krieger Distribution and abundance of Sebastes spp. 



95 



STTE STTE 



 OBSERVED t SAMPLED I OBSERVED + SAMPLED 



Figure 5 



Densities of rockfish observed from a submersible and estimated from bottom-trawl 

 catch rates, and correlation coefficients (r) at nine sites in the eastern Gulf of Alaska, 

 August 1989. 



rate than large rockfish, and (2) most small rockfish 

 use rugged habitat, which bottom trawls do not sample 

 as effectively as smooth habitat. 



The 2.2:1 ratio for large rockfish and 1.3:1 for small 

 rockfish should be considered preliminary, because 

 these are based on only nine comparisons and the area 

 sampled by the trawl may be underestimated. If rock- 

 fish were captured during trawl retrieval, the area 

 sampled was underestimated and the ratios would be 

 less. Studies are planned to determine the sampling 

 capabilities of the trawl during retrieval, and to in- 

 crease the number of trawl-to-submersible compari- 

 sons. 



Reliability of fish counts 



Fish densities were determined from the submersible 

 by counting fish within an estimated distance. Esti- 

 mates to within 1 m were possible because of uniform 

 illumination and minimal change in water clarity be- 

 tween sites. Large rockfish were ideal for counting 

 within the illuminated area because they were brightly 

 colored, solitary or loosely grouped, not obstructed by 

 rugged habitat, and usually motionless. The only move- 

 ments were by individuals moving out of the direct 

 path of the submersible, and a few larger groups mov- 

 ing toward the submersible. These fish swam slowly 

 and maintained their spacing and orientation. The pi- 

 lot observed similar rockfish behavior completely 

 around the submersible. The species and size of rock- 



fish captured in the trawls con- 

 firmed that most large rockfish 

 observed from the submersible 

 were adult Pacific ocean perch; 

 81% of the rockfish catch were 

 Pacific ocean perch, of which 92% 

 were >30 cm long. 



Besides large rockfish, short- 

 spine thornyhead were ideal for 

 counting because they were mo- 

 tionless on the bottom and not 

 obstructed by rugged habitat. 

 Counts were biased for other fish 

 species observed from the sub- 

 mersible. Small rockfish and 

 "other" rockfish were under- 

 estimated because some were 

 blocked from view by rugged 

 habitat. Flatfish were underesti- 

 mated because they blended into 

 the bottom and were difficult to 

 see more than about 3 m from the 

 submersible. Walleye pollock and 

 sablefish Anoplopoma fimbria re- 

 acted both positively and negatively to the submers- 

 ible, and the accuracy of their counts could not be 

 determined. 



Application to bottom trawl assessments 



Estimates of rockfish abundance derived from bottom 

 trawl assessments are based on the assumptions that 

 rockfish densities at untrawlable sites are similar to 

 their densities at trawlable sites, and that the seafloor 

 area sampled is determined from the horizontal open- 

 ing of the net. Results from this study indicate these 

 assumptions are incorrect for Pacific ocean perch. High 

 densities of adult Pacific ocean perch were observed 

 only at trawlable sites; hence, extrapolation of catch 

 rates from trawlable substrate to untrawlable substrate 

 would overestimate their abundance. Also, seafloor area 

 sampled may include area swept by the bridles and 

 otter doors, resulting in additional overestimates of 

 abundance. 



Acknowledgments 



I thank the crews of the submersible Delta and the 

 support vessel Wm. A. McGaw for completing safe and 

 successful dives under adverse weather conditions. I 

 thank the crew and scientists aboard the NOAA RV 

 Townsend Cromwell for their detailed sampling that 

 allowed comparisons of bottom-trawl catches to sub- 



