A Bayesian Gear Change: estimating and using a correction factor for gear catchability

No abstracts are to be cited without prior reference to the author.

A research trawl survey, carried out since 1988, contributes to the assessment of the stock of Pandalus borealis on the West Greenland shelf. The Skjervøy 3000 trawl used since 1988 has been replaced by a Cosmos 2000. In order to be able to compare old data with new, calibration experiments were carried out in 2004 and 2005 by trawling twice consecutively on the same track, using either the same gear twice or the two different gears in one order or the other. Data were analysed with likelihood models and by using Bayesian methods to fit model parameters to data. Disturbance effects, i.e. reduction in second catch relative to the first, were found to depend on density, second catches being a smaller proportion of first catches when densities were high. The Cosmos trawl was estimated to fish with about 85–86% (s.e. 6%) of the catchability of the Skjervøy trawl. The assessment of northern shrimp in West Greenland waters is carried out using a stock-production model, including a term for predation by cod. The model is fitted to the data series by Bayesian methods. The results of the present study include a probability distribution for the catchability change. This facilitates using the result in the assessment model, as the gear change can be written into the model as a shift of catchability parameter, and the probability distribution of the catchability ratio can be specified as a Bayesian prior in the assessment model.