Somerton et al.: Whole-gear efficiency of a benthic survey trawl for flatfisfi 



287 



2001). Although the two trawls have bridles that are 

 identical in length and thickness, we found the similar- 

 ity in the k^^ estimates surprising because the bridles 

 on the Poly Nor 'eastern trawl are obscured over their 

 entire length by mud clouds during trawling (Somerton, 

 2003), whereas those on the 83-112 Eastern trawl are 

 mostly unobscured and likely visible to fish. This indi- 

 cates that either the lower bridle, even in a mudcloud, 

 is more visible to a flatfish than it seems to be when 

 viewed with a video camera or that visibility of the 

 lower bridle is not particularly important for flatfish 

 herding — at least for the type of bridles used on AFSC 

 bottom survey trawls. 



For flathead, rex, and Dover sole, 55% (average value 

 of /)) of the individuals within the bridle contact path 

 and 23% of the individuals within the entire bridle path 

 were herded into the net path. Assuming that herded 

 fish have the same probability of being captured as fish 

 originally in the net path, then the herded fish comprise 

 about 32% of the catch. For arrowtooth flounder, herded 

 fish comprise 25% of the catch. This finding indicates 

 that herding contributes substantially to the catch of 

 these species and cannot be ignored when computing 

 swept area estimates of abundance. 



Assumptions with the herding experiment 



The objective of the herding experiment was to change 

 the size of the area experiencing a herding stimulus 

 without altering other aspects of trawl geometry or 

 performance. However, as in our previous experiment 

 (Somerton and Munro, 2001), and those of Ramm and 

 Xiao (1996) and Engas and God© (1989a), both the net 

 width and the bridle angle changed in response to the 

 change in bridle length (Table 2). Because of these unin- 

 tended changes in trawl geometry, the width of the bridle 

 area did not change in proportion to the change in bridle 

 length. Thus, the increase in the width of the bridle 

 path (i.e., W^-W„ ) was 8.1 m between the short length 

 and standard length bridles, but only 6.6 m between the 

 standard length and the long length bridles. A better 

 design for a herding study would be one in which the 

 incremental changes in bridle path width, or, better yet 

 W^„, were approximately equal among bridle lengths. By 

 studying bridle geometry as a function of bridle length, 

 it should be possible to choose the correct experimental 

 bridle lengths to achieve this equality. 



One assumption with the herding model is that flat- 

 fish are stimulated to herd by the bridles only from the 



