A novel Bayesian method for making the most of spatial fishery catch and effort data

Current fisheries monitoring practices in many regions of the world include precise measures of fishing location. However, spatial information is ignored in most current stock assessments, which assume instead that fishery catch per unit effort (CPUE) observations are independent replicate measurements of average stock density. This aspatial approach misleads estimates of stock status and productivity because: (1) harvesters are not random samplers of stock density, and (2) CPUE observations may be spatially autocorrelated. This paper introduces a hierarchical Bayesian method describing the spatial distribution of fishery CPUE. The spatial method is applied to British Columbia sablefish (Anoplopoma fimbria) and compared to traditional aspatial approaches in a stock assessment context. I show that spatial assessments offer less optimistic estimates of stock status and productivity compared to traditional aspatial assessments, and that the area occupied by the commercially exploitable stock is estimated to have declined by 62 percent from 1990 to 2005.