622 



Fishery Bulletin 97(3), 1999 



hood ratio=4.27, P=0.014), indicating that net efficiency 

 increased significantly over the narrow width range of 

 mature females (66-107 mm). Estimated parameter 

 values for the fitted models are shown in Table 1. 



For male Tanner crab, net efficiency for mature 

 males (0.82, n-99) was not significantly different 

 (analysis of deviance, P=0.47) than for immature 

 males (0.81, n=39) at the center of the width inter- 

 val spanned by the largest immature and the small- 

 est mature individuals. This indicates that mature 

 males are not preferentially selected by the survey 

 trawl. For mafure female Tanner crab, net efficiency 

 (0.47, n=578) was significantly different (analysis of 

 deviance, P<0.001) than the value for the mixed-sex 

 group (0.72, 77=441) at the midpoint of the mature 

 female size range. This finding indicates that net 

 efficiency was substantially less for mature females 

 than for the mixed-sexes category (combined sexes 

 excluding mature females). 



Tanner crab 



Mixed sexes 



Snow crab 



Mixed sexes 



50 



100 



150 



50 



Mature females 



50 100 150 



Carapace width (mm) 



Figure 3 



Values of net efficiency by 1-mm intervals of carapace width (filled circles) and the 

 fitted model (dashed line). Also shown for the mixed sexes categories of both spe- 

 cies are the two model components, P. (solid line ascending to the right) and P^ 

 (solid line decending to the right). 



Discussion 



Experimental trawl performance 



Net efficiency estimates from auxiliary net experi- 

 ments are potentially subject to two sources of bias. 

 First, bias may result if the efficiency of the experi- 

 mental trawl differs from that of the standard trawl. 

 Of particular concern is any distortion of trawl ge- 

 ometry caused by the auxiliary net that changes the 

 position of the footrope and fishing line in relation to 

 the bottom. In our case, trawl net width was reduced 

 by the increased drag of the auxiliary net. We at- 

 tempted to compensate for this increased drag by 

 reducing the bridles to one half of their standard 

 length, but the resulting trawl net width was still 

 approximately one meter less than the standard. The 

 decreased trawl net width, however, apparently did 

 not alter the proximity of the footrope to the bottom 

 because our video observations 

 did not reveal any conspicuous 

 differences between the standard 

 and experimental trawls. Al- 

 though these observations were 

 confined to the center of the 

 footrope, previous video observa- 

 tions of the 83-112 trawl indi- 

 cated that the center of the 

 footrope tends to have lighter bot- 

 tom contact than the wings and 

 is therefore the area most likely 

 to be impacted by the auxiliary 

 net. 



Second, bias may result if the 

 experimental trawl cannot sample 

 all habitats sampled by the stan- 

 dard trawl. To be effective, the 

 footrope of an auxiliary net must 

 fish harder on the bottom than 

 the footrope of a trawl net. Con- 

 sequently, the auxiliary net is 

 more prone to damage from snag- 

 ging on rocks and from filling 

 with epibenthos and debris. Al- 

 though, in our case, the chain 

 footrope on the auxiliary net may 

 have been effective at excavating 

 buried crab, it also resulted in 

 large catches of epibenthic fauna 

 which, in turn, aggravated the 

 problem of narrowed trawl width. 

 We reduced the narrowing by 

 moving the experiment to a loca- 

 tion lacking high quantities of 

 epibenthos, but by doing so we 



100 



150 



