Integrated recovery/recapture data analysis

The integration of recovery and recapture data, providing information on the same individuals, is important for the stable fitting of a wide range of stochastic models, resulting in more realistic estimates of survival probabilities of wild animals than when either the recovery or recapture data are used separately. Previous integrated analyses have either concentrated on time-dependent parameters only or, in the age-dependent case, have not provided a flexible framework for model refinement and selection. We derive the likelihood for a general integrated model, allowing both age-and time-dependent parameters. The work of this paper was motivated by a set of recapture histories on shags (Phalacrocorax aristotelis), for which biological knowledge and previous data analysis suggest age dependence in both survival and recapture probabilities. We use these data to illustrate the general method. All model fitting is done by the method of maximum likelihood and the programming is done in MATLAB. For the shags, we found an annual survival probability of 0.87 (standard error 0.01) for birds of breeding age and 0.70 (0.02) for birds in their second and third years. First-year survival varied widely with time about an average of 0.46.