Estimating predation mortality in the Georges Bank fish community

Multispecies virtual population analysis (MSVPA) is one of the most successful methods of including predation in fishery models. By applying MSVPA to nine important fish species on Georges Bank, we estimated predation mortality of prey species, fishing mortality, and population abundance from 1978 to 1992. One of the inputs to the MSVPA, relative stomach content, was estimated by fitting gamma distributions to the logarithmic predator-to-prey size ratios. Chi-square tests indicated that the gamma distributions fit the observed ratios well. Predation mortality was highest at ages 0 and 1. Total biomass of all species remained relatively constant with decreasing predator biomass and increasing prey biomass. MSVPA requires extensive input data, and the uncertainty in the inputs will propagate into the model output. The sensitiv- ity of MSVPA to perturbations in the inputs was assessed with a two-level fractional factorial design. Results of the sensi - tivity test indicated that MSVPA outputs were most sensitive to predator consumption rates and terminal fishing mortalities. With ±25% perturbations to the input parameters, MSVPA outputs varied within ±10% of the levels from the base run. Therefore, MSVPA appears to be relatively robust to uncertainty in the input data.