Hierarchical Bayesian approach for population dynamics modelling of fish complexes without species-specific data

Abstract

Modelling the population dynamics of fish complexes is challenging, and many species have been assessed and managed as a complex that was treated as a single species. Two Bayesian state-space surplus production models with multilevel priors (hierarchical models) were developed to simulate variability in population growth rates of species in a complex, using the hammerhead shark complex (Sphyrna spp.) of the Atlantic and Gulf of Mexico coasts of the US as an example. The complex consists of three species: scalloped (Sphyrna lewini), great (Sphyrna mokarran), and smooth hammerhead (Sphyrna zygaena). Bayesian state-space surplus production models with multilevel priors fitted the hammerhead data better than a model based on single-level priors. The hierarchical Bayesian approach represents an intermediate strategy between traditional models that do not include variability among species, and highly parameterized models that assign an estimate of parameters to each species. By ignoring the variability among species, confidence intervals of the estimates of stock status indicators can be unrealistically narrow, possibly leading to high-risk management strategies being adopted. Use of multilevel priors in a hierarchical Bayesian approach is suggested for future hammerhead shark stock assessments and for modelling fish complexes lacking species-specific data. © 2008 International Council for the Exploration of the Sea. Published by Oxford Journals. All rights reserved.