Chapter 7 - Estimation of Survival from Capture–Recapture Data Using the Cormack–Jolly–Seber Model

Abstract

Survival is a key quantity in the dynamics of populations. The Cormack–Jolly–Seber (CJS) model estimates survival and recapture probability from capture–recapture data of individually recognizable animals or plants. We introduce two different approaches to fit the CJS model (state-space and multinomial likelihood) and highlight advantages and disadvantages. Much room is devoted to explain how survival and recapture is modeled using GLM formulations. Modeling can be performed along the time axis (e.g., year effects), along the individual axis (e.g., group effects), and using an interaction of both (e.g., age effects). The specified effects can either be random or fixed, and they may be categorical or continuous. We also introduce goodness-of-fit testing using posterior predictive checks(Bayesian p -values).