Volstad et al Estimating dredge catching efficiency for Cal/inectes sapidus 



415 



priate if the density of depletion experiments was 

 higher for some strata than for others. 



Results 



Catching efTiciency 



Estimates of catching efficiency for the standard Virgi- 

 nia crab dredge from individual depletion experiments 

 conducted during the winters between November 

 1992 and March 1995 are presented in Tables 1 

 through 3. There was large variation in catchability 

 estimates among individual experiments (from -0.13 

 to 0.45); however, the difference among average an- 

 nual catchability coefficients for the entire survey 

 area was small; q varied from 0.13 to 0.18 for model 

 1 and model 2 (Fig. 3). For both depletion models, 

 an analysis of variance revealed that yearly differen- 

 ces in q are not significant (ANOVA, P>0.3; we used 

 a significance level of 5% for all tests in our study). 

 Experiments from all years, therefore, were pooled 

 to estimate an overall catchability coefficient for the 

 dredge survey. The weighted mean catchability coef- 

 ficients were 0.16 (SE=0.01) for model 1 (Eq. 1), and 

 0.15 (SE=0.02) for model 2 (Eq. 2). The means and 

 standard errors were estimated by using the jackk- 

 nife method; catchability coefficients from individual 

 experiments were assumed to be independent. The 

 two methods for estimating a mean catchability coef- 

 ficient for model 2 produced identical estimates. The 

 difference in estimates from the two models was not 

 significant. 



Mean carapace width of crabs in each removal for 

 all years combined, which is plotted in Figure 4, 

 did not show any significant trend (nonparametric 

 Mann-Kendall test (Gilbert, 1987), n = 10, P=0.19). 

 Further, the size-frequency distributions of crabs 

 for individual removals were similar (Fig. 5). These 

 results support our assumption that probability of 

 capture is independent of carapace width. 



Our test for gear saturation effects resulted in an 

 average catch per minute of 3.7 (SE=0.8) for the half- 

 minute tows, and of 2.3 (SE=0.7) for the 1-minute 

 tows. The higher CPUE for half-minute tows, although 

 not significant, suggests saturation effects and might 

 be explained by the period of time that the dredge con- 

 tinues to be towed along the bottom after it has begun 

 to be hauled back. Such a delay would affect shorter 

 tows more than longer tows. 



For the experiments on the effect of the dredge 

 liner, estimates of mean q for the dredge lined with 

 chicken wire were 0.17 (SE=0.02) based on model 

 1, and 0.15 (SE=0.03) for model 2. For the nylon 

 liner the estimates were 0.22 (SE=0.03) based on 



model 1, and 0.21 (SE=0.03) for model 2. Means and 

 standard errors were estimated by using the jack- 

 knife method. The results indicated that nylon has 

 a higher catching efficiency, but the difference in 

 catchability between the two liners was not signi- 

 ficant (ANOVA, P>0.54). Results from all 88 expe- 

 riments combined showed no significant difference 

 in catching efficiency between liners. The jackknife 

 estimates of mean catchability across years for the 



