Multispecies estimation of Bayesian priors for catchability trends and density dependence in the US Gulf of Mexico

Abstract

Fishery-dependent catch-per-unit-effort (CPUE) derived indices of stock abundance are commonly used in fishery stock assessment models and may be significantly biased due to changes in catchability over time. Factors causing time-varying catchability include density-dependent habitat selection and technology improvements such as global positioning systems. In this study, we develop a novel multispecies method to estimate Bayesian priors for catchability functional parameters. This method uses the deviance information criterion to select a parsimonious functional model for catchability among 10 hierarchical and measurement error models. The parsimonious model is then applied to multispecies data, while excluding one species at a time, to develop Bayesian priors that can be used for each excluded species. We use this method to estimate catchability trends and density dependence for seven stocks and four gears in the Gulf of Mexico by comparing CPUE-derived index data with abundance estimates from virtual population analysis calibrated with fishery-independent indices. Catchability density dependence estimates mean that CPUE indices are hyperstable, implying that stock rebuilding in the Gulf may be progressing faster than previously estimated. This method for estimating Bayesian priors can provide a parsimonious method to compensate for time-varying catchability and uses multispecies fishery data in a novel manner.