Age-structured multispecies model of the Georges Bank fish community

Georges Bank has exhibited marked changes in ecosystem structure over the last half century. The community once dominated by groundfish and other finfish is now primarily composed of pelagic and elasmobranch species. Community dynamics are profoundly influenced by species interactions such as predation. The focus of this dissertation is to estimate the trophic interactions among fish species on Georges Bank. These species interactions are estimated by developing a multispecies, statistical catch-at-age model of the Georges Bank fish community that explicitly quantifies the mortality due to predation. Manuscript I: A submodel was developed for three important fish species on Georges Bank: Atlantic cod, silver hake, and Atlantic herring. The model was fit to time series of commercial catch, fishery-independent survey catch, and predator diet composition data from 1978–2007. Estimated predation rates were high, compared with fishing mortality, and variable with time. Monte Carlo simulations were used to evaluate the ability of the model to statistically estimate parameters with known error introduced into each of the three input datasets. The model parameters and derived indices could be estimated with confidence from input data with error levels similar to those obtained from the model fit to the observed data. This manuscript helps to elevate multispecies statistical catch-at-age models to the level of statistical rigor expected of fishery dynamics models. Manuscript II: The submodel developed in the first chapter was expanded by applying the model to nine fish species within the Georges Bank community. While sensitivity to dataset weights and initial parameter estimates was apparent, both the magnitude and temporal trends in the predation mortality rates experienced by prey species were statistically estimated. Predation mortality rates again varied both temporally and ontogenetically. Mackerel, herring and silver hake experienced the greatest mortalities due to predation, with maximum predation mortality rates of 0.62 for mackerel, 1.01 for herring and 1.58 for silver hake. For these species, losses due to predation generally exceeded annual landings. Goosefish was the most dominant predator, followed by cod and silver hake, and consumption of modeled fish was related to prey abundance. These results demonstrate the strong impact of predation on Georges Bank fish community dynamics. Manuscript III: The multispecies statistical catch-at-age model developed in the first two manuscripts was used to develop stochastic forward projections. These projections were used to explore the consequences of different fishing scenarios while accounting for predation. Stochastic projections indicated strong interactions between modeled species, though the interactions were not always direct. Examples of indirect interactions included the effect of goosefish on herring and similarly, the impact of goosefish on white hake. Consequently, population responses to fishing were a function of not only the rates of fishing, but also of these direct and indirect interactions among species. Collectively, this dissertation demonstrates the strong impact of predation on Georges Bank fish community dynamics and provides a tool for statistically estimating the mortality due to predation. Through the development of a multispecies statistical catch-at-age model, we have expanded the multispecies modeling approaches implemented for Georges Bank and the Northeast U.S. Continental Shelf