Correcting Bias in Survival Probabilities for Partially Monitored Populations via Integrated Models

We provide an integrated capture–recapture–recovery framework for partially monitored populations. In these studies, live resightings are only observable at a set of monitored locations, so that if an individual leaves these specific locations, they become unavailable for capture. Additional ring-recovery data reduce the corresponding bias obtained in the survival probability estimates from capture–recapture data due to the confounding with colony dispersal. We derive an explicit efficient likelihood expression for the integrated capture–recapture–recovery data, and state the associated sufficient statistics. We demonstrate the significant improvements in the estimation of the survival probabilities using the integrated approach for a colony of guillemots (Uria aalge), where we additionally specify a hierarchical approach to deal with low sample size over the early period of the study.