A statistical censoring approach accounts for hook competition in abundance indices from longline surveys

Abstract

Fishery-independent longline surveys provide valuable data to monitor fish populations. However, competition for bait on the finite number of hooks leads to biased estimates of relative abundance when using simple catch-per-unit-effort methods. Numerous bias-correcting instantaneous-catch-rate methods have been proposed, modelling the bait removal times as independent random variables. However, experiments have cast doubts on the many assumptions required for these to accurately infer relative abundance. We develop a new approach by treating some observations as right-censored, acknowledging that observed catch counts are lower bounds of what they would have been in the absence of hook competition. Through simulation experiments we confirm that our approach consistently outperforms previous methods. We demonstrate performance of all methods on longline survey data of eleven species. Accounting for hook competition leads to large differences in relative indices (often -50% to +100%), with effects of hook competition varying between species (unlike other methods). Our method can be applied using existing statistical packages and can include environmental influences, making it a general and reliable method for analyzing longline survey data.