Adams et al.: Population estimates of Pacific coast groundfishes 



449 



Figure 2 



Sampling dimensions of a remotely operated vehicle (ROV) video transect and a trawl. Draw- 

 ings of gear and sampling volumes are made to scale. 



(Rose 2 ). On deck, catches were handled with stan- 

 dard protocol (Smith and Bakkala, 1982) and were 

 sorted to species, weighed, and measured. Lengths 

 were measured for either a subsample of 100 fish or 

 the entire catch, if it comprised less than 100 fish. 

 When the catch was greater than 100 fish for a spe- 

 cies, total species number was extrapolated from to- 

 tal species weight by using the average weight per 

 fish from the subsample weight and numbers. Num- 

 bers of species per hectare were expanded by divid- 

 ing the total species number by the area covered (dis- 

 tance traveled by a tow multiplied by the average 

 width of the trawl). 



Differences in ROV and trawl abundance estimates 

 for individual species at each depth were tested by 

 using £-tests for differences of unpaired log-trans- 

 formed means. A sign test was used to determine 

 whether the number of times that an ROV abundance 

 estimate (or coefficient of variation) was higher than 

 the trawl was greater than would have been expected 

 randomly (i.e. a 50/50 ratio). 



Statistical power is defined as 1-/J, where P is the 

 probability of failing to reject a hypothesis when it is 

 false, and therefore power is the probability of cor- 

 rectly rejecting a false hypothesis (Peterman, 1990). 



2 Rose, C. National Marine Fisheries Service, Alaska Fisheries 

 Science Center, 7600 Sand Point Way N.E., BIN C15700, Se- 

 attle, WA, 98115-0700. Personal commun., 1992. 



The power of the ROV and trawl abundance esti- 

 mates was evaluated by calculating the required 

 sample size to detect a 50% reduction from the log- 

 transformed mean abundance at a fixed level of a 

 (0.05) and at a high level of power (1-/3, 0.80). Com- 

 parisons were made for the commercially important 

 species (Dover sole, Microstomas pacificus; thorny- 

 heads, Sebastolobus spp.; and sablefish, Anoplopoma 

 fimbria) and for a group of other abundant taxa 

 (catsharks, Scyliorhinidae; skates, Rajidae; and eel- 

 pouts, Zoarcidae) at the 400-m depth (the only depth 

 where all of these taxa occurred in both the ROV 

 and the trawl surveys). 



Results 



More fish per hectare were observed from the ROV 

 than were captured in the trawl at the 400-m and 

 600-m depths, whereas more fish were captured in 

 the trawl at the 200-m depth (Fig. 3). However, the 

 differences in the log-transformed total ROV and 

 trawl estimates were only significant at the 400-m 

 and 600-m depths (400 m: <=5.50, P<0.001; 600 m: 

 <=3.28, P=0.011) and not at the 200-m depth (200 m: 

 <=0.32, P=0.713). Numbers captured in the trawl at 

 the 200-m depth were higher owing to a single large 

 catch of two species of rockfish in one trawl. The ROV 

 estimates offish numbers were higher for most indi- 

 vidual species or taxonomic groups that occurred in 



