Krieger and Sigler: Catchability coefficient for rockfish 



285 



floor area sampled, but were not adjusted for the 

 larger water column sampled from the submersible; 

 submersible personnel surveyed to 10 m above the 

 seafloor, whereas the net sampled to 1.8 m above the 

 seafloor. A catchability coefficient was estimated by 

 comparing the average density from trawl catch rates 

 to the average density from submersible counts. The 

 bootstrap resampling method (Efron, 1982; Efron and 

 Tibshirani, 1986 ) was used to test the statistical sig- 

 nificance of the difference and to estimate a confi- 

 dence interval for the catchability coefficient. A boot- 

 strap sample was created by sampling 1,000 times 

 with replacement. The statistical significance of the 

 difference (P) between submersible and trawl com- 

 parisons was computed by tallying how many of the 

 1,000 bootstrap replicates were greater than zero; P 

 is for a two-sided test. 



Results 



Sampling during trawl retrieval 



During trawl retrieval, the net remained in contact 

 with the seafloor an additional 7-9 min. The speed 

 of the net along the seafloor during retrieval was 

 4-5 km/h — a combination of the vessel moving slowly 

 forward and retrieval of the cable. Catch rates for 

 Pacific ocean perch averaged 34.9/1,000 m 2 during 

 "standard" trawling and 29.8/1,000 m 2 during "re- 

 trieval" trawling (Table 1 ) and were not significantly 

 different (2-sided f-test, P>0.5). Apparently, the net 

 continued to sample effectively during retrieval; 

 therefore, trawl catch rates included the seafloor area 

 swept during trawl retrieval. 



Detection probability from submersible 



Observed densities of sea stars began decreasing 3.4 

 m from the submersible; therefore, the expected den- 

 sity was the average density of targets within the 

 3.4-m full-detection range. Assuming DP = 1.0 within 

 3.4 m, we calculated a DP of 0.77 for the entire 5.3-m 

 range of visibility, which implies that about 3/4 of 

 the sea stars were counted within the range of vis- 

 ibility. Sea stars were more difficult to detect than 

 were rockfish because of differences in body shape; 

 the flat bodies of sea stars blended into the seafloor, 

 but the fusiform bodies of rockfish extended above 

 the seafloor. Therefore, the 0.77 DP for sea stars is 

 considered a minimal DP for rockfish and catchability 

 coefficients are calculated for both a 0.77 DP and a 

 1.0 DP. 



Comparison of submersible counts with 

 trawling catch rates 



Sixteen dive sites were successfully sampled by bot- 

 tom trawling (Table 2). The average rockfish densi- 

 ties were 21.7/1,000 m 2 from trawling, and 22.4/1,000 

 m 2 (DP=0.77) and 17.1/1,000 m 2 (DP=1.0) from sub- 

 mersible counts, resulting in catchability coefficients 

 of 0.97 (95% confidence interval: 0.70-1.32) and 1.27 

 (95% confidence interval: 0.91-1.73) for the area 

 swept between the trawl wingtips. No significant 

 difference was detected between density estimates 

 from the submersible and trawl for either the 0.77 

 DP (P=0.95) or the 1.0 DP (P= 0.13). 



Discussion 



The 0.97 and 1.27 catchability coefficients are con- 

 siderably less than the 2.1 (SE=0.4) catchability co- 

 efficient reported by Krieger ( 1993), mainly because 

 they include the area sampled during trawl retrieval. 

 The 0.77 DP from the submersible and the increased 

 estimate of net width also contributed to lower esti- 

 mates of catchability coefficients. 



If rockfish did not react to the trawl gear, density 

 estimates from submersible counts would have ex- 

 ceeded those from trawl catch rates because approxi- 



