Detecting declines in catch rates of diverse trawl bycatch species, and implications for monitoring

Trawl fisheries around the world are being pressured to demonstrate that their impacts on both the target and bycatch species are sustainable in the long-term. However, the effectiveness of sampling catches to assess the viability of populations of non-targeted species is rarely examined. We estimated the levels of fishery-dependent sampling effort required to detect declines in catch rates of prawn trawl bycatch from 95 commercial trawls in two regions of Australia's Northern Prawn Fishery. We explore a range of possible monitoring options including combining different sampling intensities, time-frames and levels of statistical power.Poisson and negative binomial models were used to determine the number of trawls required to detect a range of declines. We found that between 15,536 and 24,933 trawls, depending on the region, would be required to detect a 20% decline in catch rates of the rarest taxa (<0.1 individuals h−1), assuming a power of 90% and a level of significance of 5%. Assuming a lower detection power (70%), trawl numbers would drop to 9126 and 14,646, respectively.Using a model of a constant decline in relative abundance (over 5 years), data accumulated from modest-sized annual surveys (52 and 43 trawls in two regions) would provide increasing power to detect changes in catch rates. After 3 years, surveys of this size (and power of 70%), could detect declines of 99.9% in 72–81% of taxa, declines of 50% in 34–43% of taxa, and declines of 20% in 20–34% of taxa, depending on the region. After 5 years, the power to detect declines of 50% had increased to cover 43–72% of taxa, and declines of 20% in 34–43% of taxa. Our results indicate that the power to detect even quite large declines in catch rates of rarely caught species would only be possible after some years of modest-sized annual surveys.