Standardization of landing efficiency for opilio crab in the western Bering Sea by using of generalized additive models

Abstract

Generalized additive models are applied for standardization of daily landing per unit effort (LPUE) for opilio crab using the data of fishery statistics for the West Bering Sea fishery zone in 2003–2020. A set of 12 models with various combinations of predictors was examined and the best model with the smallest value of Akaike criterion was selected (information criterion Akaike 21743, explained variance 58.6 %). The selected model reflects the effect of depth, distance from the coast, daily effort and tensor product of geographic coordinates and day of the year. LPUE was standardized using the selected model by substituting median values of nominal predictors and modal values of categorical predictors. Then the crab stock was estimated using the state-space form of Deriso-Schnute delay-difference model. The estimates based on both standardized and nominal indices are compared and a significant difference between them is found: the stock is assessed as 23,040 t with nominal indices but as 17,070 t using the standardized indices.