A Bayesian spatiotemporal approach to inform management unit appropriateness

Abstract

One prerequisite for sustainable fisheries management is to match management actions with biological processes. Stocks are fundamental units for fisheries management. Understanding the spatial structure of fish stocks is critical for conducting defensible stock assessments, applying efficient management strategies, and ensuring the sustainability of fish stocks. Yellow perch (Perca flavescens) is an important fishery in the Great Lakes. The appropriateness of its management units (MUs) has been identified as of high concern by the Great Lakes Fisheries Commission. Here we established integrated nested Laplace approximations and stochastic partial differential equations as two powerful tools for modeling spatiotemporal patterns of fish relative biomass. These fast computational approaches were applied to fit a Bayesian hierarchical hurdle model to occurrence and positive mass of yellow perch caught in gill-net surveys. Yellow perch relative biomass index has clear temporal variation and spatial heterogeneity, with the two middle MUs for yellow perch within Lake Erie merging together. The method explicitly models the spatiotemporal correlation structure inherent in biomass survey data at a reasonable computational cost, and the estimated spatiotemporal correlation informs stock structure.