Volstad et al Estimating dredge catching efficiency for Calltnectes sapidus 



419 



the identical estimates of q from the two alternative 

 methods indicate that such bias is small. 



Obtaining reasonable estimates of catching effici- 

 ency from catch-effort techniques requires depleting a 

 substantial proportion (usually more than 30%) of the 

 population (see Gould and Pollock, 1997). Also, esti- 

 mates of the regression slopes, and hence q, are likely 

 to be more accurate for both models if the independent 

 variables (i.e. cumulative catch, /C^, or coverage, /) have 

 a wider range. For the depletion experiments with 

 ten removals conducted in Maryland waters, estima- 

 tes of g based on the first five removals were generally 

 higher than estimates of q based on all ten removals. 

 Thus, fewer removals per experiment can bias the 

 estimates of gear efficiency just as vessel effects can. 

 In experiments conducted from two similar vessels in 

 Maryland waters, we did not detect significant diffe- 

 rences in catchability between vessels. Environmental 

 factors probably were the principal cause of variability 

 in catchability estimates from different experiments 

 because sediment type, bottom topography, intensity 

 and direction of the current, towing speed, and crab 

 distribution influence dredge performance. 



When our overall estimates of catching efficiency 

 are used to adjust CPUE from the baywide winter 

 survey, plausible estimates of absolute abundance 

 of blue crab in the Chesapeake Bay are obtained, 

 resulting in an estimated exploitation rate of about 

 45% in 1995. The catchability coefficient estimate 

 (g=0.26j presented in Zhang et al. (1993), in con- 

 trast, would result in an estimated exploitation rate 

 of 75%. Their method eliminated removal experi- 

 ments with low coefficients of determination (r~) for 

 the regressions, or with negative catchability esti- 

 mates, which could result in a positive bias in the 

 estimated overall catchability. 



We have not accounted for landings from the rec- 

 reational fishery or for natural mortality in our exam- 

 ple; therefore the above exploitation rates were pro- 

 bably underestimates. Recreational harvest data are 

 very scarce, but limited surveys conducted by the 

 Maryland Department of Natural Resources in 1983, 

 1988, and 1990 have indicated that recreational har- 

 vest represented approximately 79%, 50% , and 26% of 

 the reported commercial harvest ( Rugolo et al. 1998b ). 

 The 1990 survey is considered the most reliable of the 

 three. Assuming that the recreational harvest in 1995 

 represented 26% of the commercial harvest, then the 

 corrected estimate of total blue crab exploitation rate 

 is 56.7%. A bias in the opposite direction would result 

 if the exploitable part of the stock were underestima- 

 ted because of recruitment during the fishing season. 



The results of this study demonstrate that impro- 

 ved estimates of catching efficiency can substantially 

 increase the accuracy of estimates of absolute abun- 



dance. In our example based on survey data from 

 the winter 1994-1995, the variance in the absolute 

 abundance estimate for the l-i- age group is driven 

 by the variance in q . Size of sampling locations in 

 the Chesapeake Bay winter dredge survey for blue 

 crabs have ranged between 877 and 1412 stations 

 per year, and relative precision (k) of the estimated 

 CPUE (i.e. its standard error divided by the mean) 

 has been around 10% for most years. Because of 

 the asymptotic properties of ^, it would be prohibi- 

 tively expensive to significantly increase the preci- 

 sion of absolute abundance estimates by increasing 

 the sample size in the survey. We believe a more 

 cost-effective way to increase the precision of estima- 

 tes of absolute abundance would be to improve the 

 estimate of catchability by increasing the number 

 of depletion experiments. Our results show that an 

 estimate of catching efficiency based on a single 

 experiment or on a few removal experiments would 

 not represent the entire bay accurately because cat- 

 chability is highly variable among sites owing to dif- 

 ferences in bottom conditions and other factors. We 

 included all the depletion estimates of catchability 

 in our study, including those less than zero. We assu- 

 med that point estimates of catchability are estima- 

 ted with a random error that is normally distributed 

 around a mean: estimates in the tails of the distribu- 

 tion could be substantially higher or lower than the 

 true mean catchability. An accurate estimate of cat- 

 ching efficiency, applicable to the entire Chesapeake 

 Bay, requires conducting a large number of depletion 

 experiments at representative locations. Although no 

 statistical differences were found among the annual 

 catchability coefficient estimates for the years analy- 

 zed in our study, we caution that interannual varia- 

 tion in q is likely. For example, an increase in water 

 temperature during mild winters may affect crab 

 behavior (they would cease hibernation) and hence 

 probability of capture. We recommend, therefore, that 

 depletion experiments be conducted yearly as part of 

 an annual winter survey of blue crab population. 



The conclusions of our study can be generalized 

 and extended to similar resource assessment sur- 

 veys of other slowly moving or sedentary bottom 

 dwelling species with patchy distribution, such as 

 scallops or clams. In surveys of target species, such 

 as these, attention must be paid to variability in 

 capture efficiency of the gear with respect to sedi- 

 ment type, depth, towing speed, and other factors 

 (such as animal interactions) that affect the gear. 

 Sediment or bottom type are often selected as stra- 

 tification criteria in bottom surveys of benthic orga- 

 nisms. Properly designed removal experiments can 

 provide reliable estimates of catchability coefficients 

 for each sampling stratum, allowing adjustment of 



