Spatiotemporal index standardization improves the stock assessment of northern shrimp in the Gulf of Maine

Abstract

Estimated trends in relative stock abundance are a primary input to fish stock assessments. Accurate and precise estimates are essential for successful conservation and management. Scientifically designed data collection ensures that estimates of relative abundance are unbiased. However, the statistical efficiency of a design-based estimator may be low under certain circumstances. We apply a recently developed spatiotemporal model that incorporates habitat variables to estimate a model-based abundance index for northern shrimp (Pandalus borealis) in the Gulf of Maine. We contrast this spatiotemporal index with a classical design-based index and evaluate the impacts of differences between the two abundance indices on the stock assessment. We show that using the spatiotemporal index in the assessment model greatly alters the estimates of recruitment and spawning stock biomass and the determination of stock status. Also, incorporating the spatiotemporal index leads to less retrospective bias and outperforms the model with design-based index in terms of predictive performance through a retrospective cross-validation test. Our results suggest that temporal variability of population abundance could be exaggerated by the design-based estimator, and such imprecision may greatly affect the performance of a stock assessment and subsequent development of management decisions.