Evaluating the performance of a multispecies statistical catch-at-age model

Predation is a substantial source of mortality that is a function of the abundance of predator and prey species. This source of mortality creates a challenge of incorporating species interactions in statistical catch-at-age models in a way that accounts for the uncertainty in input data, parameters, and results. We developed a statistical, age-structured, multispecies model for three important species in the Georges Bank fish community: Atlantic cod (Gadus morhua), silver hake (Merluccius bilinearis), and Atlantic herring (Clupea harengus). The model was fit to commercial catch, survey, and diet data from 1978 to 2007. The estimated predation rates were high, compared with fishing mortality, and variable with time. The dynamics of the three species can be explained by the interplay between fishing and predation mortality. Monte Carlo simulations were used to evaluate the ability of the model to estimate parameters with known error introduced into each of the data types. The model parameters could be estimated with confidence from input data with error levels similar to those obtained from the model fit to the observed data. This evaluation of model performance should help to move multispecies statistical catch-at-age models from proof of concept to functional tools for ecosystem-based fisheries management.