A Bayesian mark–recapture model for multiple-recapture data in a catch-and-release fishery

Abstract

This paper extends a state–space Bayesian mark–recapture framework to multiple-recapture data to estimate fishery-specific capture and mortality rates and seasonal movement rates for fish in different length classes. The methodology is applied to tag recapture data for white sturgeon ( Acipenser transmontanus ) collected in the recreational fishery and the Canadian Department of Fisheries and Ocean’s test fishery at Albion in the lower Fraser River. Significant differences were found between some estimated movement rates by season and length class, supporting the notion of there being marked differences in seasonal movement patterns between different life history stages of A. transmontanus in the lower Fraser River. Uncertainty in the tag reporting rate parameter, quantified using a recreational creel sampling program, is summarized by a prior distribution. The utility of recreational fishing effort as a model covariate in accounting for seasonal and spatial variation in recapture rates is addressed using Bayesian model evaluation criteria. The data provide strong support in favour of models that include fishing effort as a covariate. The appropriate level of stratification for the recreational catchability parameter q is assessed using Bayesian model evaluation criteria; models in which q is estimated by season and length class have the highest posterior probabilities.