The effect of subsampling when monitoring bycatch in a penaeid trawl fishery

Abstract

Fishery dependent and independent survey programs for multi-species trawl fisheries often employ subsampling to increase efficiency, but this is not usually accompanied by any quantitative analysis to optimise its application. Here, we measure the effect of subsampling on the information generated from trawl surveys, and provide relationships to assist the interpretation of data derived from such surveys, and confidence in associated estimates for future studies. A commercial fishing vessel was chartered to perform normal fishing operations, and various proportional subsamples (by weight) were sorted from each tow. Overall, 50 % of total species diversity in a sample would be detected if only 10 % of the total sample mass was sorted, however abundance estimates of individual species under different subsampling scenarios varied depending on their rarity. For common species, the abundance in a sample was estimated with reasonable accuracy from sorting only 10 % of the biomass, whereas sorting >60 % of the biomass was required to obtain an accurate estimate of abundance for rarer species. The careful consideration of program objectives is important in optimising the sampling approach employed, as subsampling may not be appropriate for detection of rare species. The patterns presented here provide a quantitative basis to support sampling and sorting methodology for multispecies trawl samples, as well quantifying the implications of subsampling on the confidence in estimates of diversity or abundance.