Somerton and Otto: Net efficiency of a survey trawl for Ch/onoecetes opilio and C, bairdi 



619 



additional feature of the auxiliary net was that the 

 footrope was not attached anywhere along the length 

 of the trawl footrope but instead joined it only at the 

 wing tips. The auxiliary net footrope was intended 

 to be positioned 2-3 m behind the trawl footrope 

 along the center line of the trawl, but this position 

 was not verified in the field. 



eled the relationship between net efficiency and cara- 

 pace width as a function of two width-dependent pro- 

 cesses: the probability of entering the net at the 

 footrope (P,) and the probability of entering the net 

 through the belly mesh after escaping under the 

 footrope (Pj,). Algebraically this relationship can be 

 expressed as 



Description of the experiment 



The net efficiency experiment was conducted aboard 

 a chartered 40-m stern trawler (YV Arcturus), from 

 20 July 1997 to 8 August 1997, immediately follow- 

 ing completion of the annual AFSC trawl survey of 

 the eastern Bering Sea. The experimental site was 

 located east of the Pribilof Islands (57°11.4'N, 

 166°26.5'W) at depths between 75 and 90 m on a 

 smooth bottom. 



All hauls were made by using the same trawling 

 procedure and towing speed (3 knots) as those used 

 during the AFSC survey but tow duration ( 15 min) 

 was only one half of the standard. Trawl net width, 

 from wing-tip to wing-tip, was measured on all hauls 

 with a SCANMAR net mensuration system. 



The experiment began with test hauls of both the 

 standard trawl and the trawl with the attached aux- 

 iliary net (experimental trawl) to determine if trawl 

 performance was altered by the auxiliary net. All test 

 hauls were conducted with the codends left open and 

 with a video camera attached so that the trawl 

 footrope near its center could be viewed. After three 

 hauls of the experimental trawl, it became evident 

 that the additional drag of the auxiliary net resulted 

 in a reduction in the opening width of the trawl net. 

 To help correct this problem, the bridles were short- 

 ened from the standard length of 55 m to 28 m. The 

 shorter bridles were used for the remainder or the 

 experiment. 



All crab were removed from each catch and sorted 

 into two groups; mature females, which were recog- 

 nized by the presence of an enlarged abdomen, and 

 all other crab. For simplicity, we will refer to the group 

 "all other crab" as "mixed sexes." Carapace width, to 

 the nearest 1 mm, was measured with vernier calipers 

 on all individuals. On some hauls, randomly selected 

 male Tanner crab were subsequently removed from the 

 mixed sexes baskets and measured for both carapace 

 width and chela height (right hand side) to allow de- 

 termination of morphometric maturity. 



Estimating net efficiency 



Net efficiency (E^^) typically is an increasing func- 

 tion of body size (Engas and Godo, 1989; Walsh, 

 1992). To provide more flexibility, however, we mod- 



E„=Pf^(^-P,^P,- 



(1) 



The probability of entering the trawl at the footrope, 

 was described with a three-parameter logistic function: 



where a, b, and c 



w 



1 + be'-"" 



parameters; and 

 carapace width. 



(2) 



This function was chosen because it provides pre- 

 dicted values of P, that increase with w asymptoti- 

 cally to a value of a and permits cases where effi- 

 ciency is less than unity even at the largest body 

 sizes. The probability of entering the trawl through 

 the belly mesh was described by a two-parameter 

 decreasing logistic function: 



1 + rfe'-^"' 



where d and f = parameters. 



This function was chosen because it provides pre- 

 dicted values of P/, that decrease with w asymptoti- 

 cally to a value of zero. 



The model resulting from substituting Equations 

 2 and 3 into Equation 1 was fitted to values of net 

 efficiency for each 1-mm increment of carapace width 

 by using a maximum likelihood procedure where the 

 number of crab captured in the trawl net at each 

 carapace width was described by a binomial random 

 variable (Millar, 1992). The parameters of the model 

 were estimated by minimizing the negative logarithm 

 of the likelihood function: 



L = -Y^{n,.\ogE„,,.+(N,,-nJloga- £„.„,)), (4) 



where, at each value of carapace width w, 



n = the number of crab captured in the 

 trawl net; 



TV = the total number of crab in the trawl 



If 



and auxiliary nets combined; and 

 E = the net efficiency. 



