Hierarchical Bayesian models of length-specific catchability of research trawl surveys

To estimate absolute abundance from research trawl surveys, the catchability of the fish to the gear must be known or estimated. Using 47 data sets of length-specific catchability, we conducted a hierarchical Bayesian meta-analysis of length-specific catchability for a number of different species groups. It was found that the Bayesian estimates of catchability were seldom near or above 1 for any species or size. This suggests that any assumption that swept-area abundance estimates are in fact absolute abundance would likely underestimate the true abundance. Catchability was higher for haddock (Melanogrammus aeglefinus) than for Atlantic cod (Gadus morhua) of the same size, suggesting behavioural differences between these two species. We found a seasonal difference in catchability with higher catchability for surveys in summer-fall than for those in spring-winter. The results of this study can be applied both for the reconstruction of fish community structure for ecosystem models and as auxiliary (or prior) information for single-species stock assessment where catchability is estimated within the stock assessment procedure.