COS 82-5
Hierarchical Bayesian analysis to measure differences in stock-recruitment parameters across harvested fish taxa
Understanding the density dependence of fish reproduction is critically important for management and conservation of commercial fisheries. Stock-recruitment relationships describe how reproductive output changes relative to the number of fish in a population. Common models used to fit fisheries data include the Beverton-Holt model, which describes a system where total recruitment levels off at higher spawner densities (called compensation), and the Ricker model, where total recruitment declines at high densities (overcompensation). While large amounts of data allow an appropriate model to be chosen for a species or population, no clear method has been developed to objectively determine the appropriate stock-recruitment relationship for commercial species where limited or no stock-recruitment data exist.
To address this gap, we developed a hierarchical model that uses Bayesian inference to link stock-recruitment parameters among species, taxonomic orders (e.g. Gadiformes, Perciformes, Clupeiformes, Salmoniformes), and trophic levels of commercial fish. We use our hierarchical framework to estimate the parameters of the Shepherd stock-recruitment model, a flexible stock-recruitment model that allows for a variety of hypothesized relationships. We examine how model parameters vary by taxonomic groupings and trophic levels and how these results can inform model selection for data-poor fish populations.
Results/Conclusions
Our results show that for many of the orders well represented in our dataset the posterior parameter distributions are indicative of good fits to a compensatory recruitment relationship (e.g., similar to a Beverton-Holt model). However, in Salmoniformes, the posterior distribution for the parameter governing the degree of compensation leads to stock-recruitment curves that are distinctly overcompensatory (e.g., similar to a Ricker model). In contrast to this strong taxonomic signal, the influence of trophic level on model choice is not clear, as the orders in our dataset do not exhibit a strong relationship between degree of compensation and trophic level.
An advantage to this hierarchical approach is that populations and species with limited data are informed by other populations and species, as they come from a common, order-level distribution. As a result, order-level differences in stock-recruitment models can be detected and closely related species with large amounts of data can be used to inform stock-recruitment model selection for data-poor species.