A statistical evaluation of the design and precision of the shrimp trawl survey off West Greenland

Stocks of Pandalus borealis off West Greenland have been assessed using a r esearch trawl survey since 1988. The survey has used a design of randomly p laced stations, stratified ton depth data where available, using small bloc ks elsewhere), with sampling effort proportional to stratum area. In some y ears, a two-stage adaptive sampling scheme was used to place more stations into strata with large first-stage variation in catches. The design of the survey was reviewed in 1998. Modifications in survey design suggested were to shorten tow duration, to pool strata so that effort could be allocated m ore efficiently, to put a higher proportion of stations in high-density are as and to abandon two-stage sampling. All these changes were implemented fo r the 1998 survey, except that tow duration was reduced to 30 min at 25% of the stations. To analyze the-efficiency of the present survey design, vari ous statistical techniques were used to estimate two indices of shrimp abun dance and their precision, and to determine the effective sample sizes for estimates of length-frequency distributions. It is concluded that the surve ys produce a fairly precise abundance index, and that given the relatively small effective sample size, reducing tow duration to 15 min would increase overall survey precision. An unexpected outcome of the analysis is that th e density of shrimp appears to have been fairly stable over the last 11 yea rs. (C) 2000 Elsevier Science B.V. All rights reserved.