MODELING REGIONAL WATERFOWL HARVEST RATES USING MARKOV CHAIN MONTE CARLO

Abstract

We developed models for simultaneous inference on movement and harvest rates, and of factors influencing harvest rates, using band-recovery data and Markov chain Monte Carlo (MCMC) modeling. We modeled variation in harvest rates for American black ducks (Anas rubripes) during 1971–1994 using recoveries of ducks banded in 3 breeding regions and recovered in 6 harvest regions in Canada and the United States. Models based on season length or bag limit together with season length, and incorporating a random year- and area-specific effect, were superior to other models as gauged by information criteria, fit statistics, and cross-validation. We used these models to generate posterior predictive distributions for harvest rates as a function of harvest regulations, for application to adaptive harvest management.