DIRECT ESTIMATES OF GEAR SELECTIVITY FROM MULTIPLE TAGGING EXPERIMENTS

A new method is introduced for estimating selectivity of fishing gear from tagging data in which data from many experiments are combined. Se lectivity is modeled as a multiplicative function of length and experi ment effects using a generalized linear model with a log link function and a binomial error structure. We apply this method to 137 tagging e xperiments on Atlantic cod (Gadus morhua) conducted from 1954 to 1991. We show that the selectivity of otter trawls changed from the 1960s t o the 1980s; during the earlier period the maximum probability of capt ure occurred at 55 cm and declined for longer fish, whereas in recent years the maximum probability is at approximately 60 cm and remains co nstant for longer fish. We discuss how selectivity estimates can be us ed to improve stock assessments.