A BIOMASS-BASED ASSESSMENT MODEL FOR NORTHERN ANCHOVY, ENGRAULIS-MORDAX

We developed a relatively simple and parsimonious (SMPAR) biomass dynamics model for estimating abundance of northern anchovy, Engraulis mordax, off southern California and Baja California, Mexico, during the 1963 to 1991 fishing seasons. The SMPAR model was a compromise between simple surplus production and complex age-structured models. It was designed to give more precise biomass estimates for management of northern anchovy for which there are no age-composition data and only noisy abundance index data. We evaluated consistent bias in biomass and recruitment estimates, bias in recruitment estimates due to log transformation, and retrospective bias. Simple corrections based on bootstrap procedures were used to remove consistent bias and log transformation bias. Retrospective bias was not a significant problem. Results indicate that the SMPAR model estimates stock biomass more reliably than recruitment because abundance indices for northern anchovy contain little information about interannual recruitment variability. Asymptotic variance estimates calculated by inverting the Hessian matrix averaged 20% smaller than variances calculated by bootstrapping. Outliers in abundance data were the biggest source of uncertainty in biomass estimates. Simulation results indicate that our approach could be useful in a variety of situations.