Somerton and Otto: Net efficiency of a survey trawl for Chionoecetes opilio and C. bairdi 



623 



excluded from consideration a type of habitat occu- 

 pied by snow and Tanner crabs. Thus the net effi- 

 ciency estimates we obtained may not be represen- 

 tative of the entire area covered by the survey. 



Model form 



The form of the model that we chose to describe the 

 relation between net efficiency and carapace width 

 included a term (P.) that increases with width to ac- 

 count for entry into the trawl over the footrope and 

 another term (Pj,) that decreases with width to ac- 

 count for entry of crabs through the belly meshes of 

 the trawl after they have passed under the footrope. 

 The model with both P, and P^ terms fitted the mixed- 

 sexes data of both species significantly better than the 

 model with only a P, term, confirming the significance 

 of the apparent increase in net efficiency with decreas- 

 ing size at carapace widths less than 50 mm. However, 

 we have been unable to verify that the increased effi- 

 ciency is indeed due to the entry of small crab through 

 the belly meshes. Regardless of the mechanism, the 

 important question is whether the increase in net effi- 

 ciency at small size is a normal property of the 83-112 

 trawl or an artifact of the experiment. We attempted 

 to answer this question by comparing the width distri- 

 butions of crab captured in the trawl net of the experi- 

 mental trawl with those captured in the standard trawl 

 at the same or nearby stations during the survey pre- 

 ceding the experiment, but patchiness and small sample 

 sizes made the results equivocal. 



In light of this uncertainty, the composite form of 

 the model is advantageous because it allows separa- 

 tion of the P. and P^ effects. For example, if further 

 research reveals that the increase in efficiency at 

 small size is associated with the attachment of the 

 auxiliary trawl, then an estimate of net efficiency 

 without this effect can be obtained by deleting the 

 P^ term and by evaluating the reduced model with 

 the parameters from the fit of the complete model 



(Table 1). For this reason, the bootstrapped confi- 

 dence intervals we provide (Fig. 4) are computed two 

 ways, by evaluating the full model and by fitting the 

 full model but evaluating only the P, term. However, 

 because the P^ term decreases rapidly with carapace 

 width, the full and reduced models predict nearly 

 identical values of net efficiency at carapace widths 

 above 50-60 mm, which includes the mature and 

 commercial width ranges for both species. 



Efficiency differences between biological subgroups 



The survey trawl is less efficient for mature female 

 Tanner crab than for the mixed-sexes group over the 

 same range of carapace width. Since the mixed-sexes 

 group is primarily male at these sizes, the difference 

 in efficiency is likely due to between sex differences 

 in both morphological features and behavior. Male 

 Tanner crab have longer legs, in relation to their cara- 

 pace width, than do mature females. Not only does 

 this increase their effective size in relation to the 

 trawl mesh but also may result in their bodies being 

 held higher off the bottom in relation to the footrope. 

 In addition, male Tanner crab may be caught more 

 efficiently because they are less sedentary than fe- 

 males and less prone to bury themselves.^ 



The survey trawl efficiency does not differ between 

 mature and immature male Tanner crab. We were 

 concerned that mature males, which have larger 

 claws and are perhaps more active than immature 

 males, might be preferentially selected. If such se- 

 lection did occur, it would lead to an underestimate 

 of the carapace width at maturity which is used in 

 establishing minimum width limits for the commer- 

 cial fishery. Apparently such preferential selection 

 of mature males does not occur. 



In the foregoing, we estimated values of net effi- 

 ciency or the capture probability of crab that occur 

 between the wing-tips of the trawl net. For stock 

 assessment purposes, however, it may be more con- 

 venient to consider these values expressed as trawl 

 efficiency or the capture probability of crab that oc- 

 cur between the trawl doors. In situations where the 

 bridles herd animals into the path of the trawl, fur- 

 ther experiments to estimate sweep efficiency would 

 be required to estimate trawl efficiency (Dickson, 

 1993a, 1993b). However, experiments on herding 

 snow crab indicate that this species is not herded by 

 the bridles of the 83-112 eastern trawl. ^ Thus, trawl 

 efficiency can be estimated simply by multiplying net 

 efficiency by the quotient of the net width divided by 

 the door width (i.e. 0.30).2 



■* Stevens, B. 1998. Kodiak Laboratory, National Marine Fish- 

 eries, P.O. Box 1638, Kodiak, AK. Personnal commun. 



