Estimating population projection matrices from multi-stage mark-recapture data

Multi-stage mark-recapture (MSMR) statistics provide the best method for estimating the transition probabilities in matrix population models when individual capture history data are available. In this paper, we improve the method in four major ways. We use a Markov chain formulation of the life cycle to express the likelihood functions in matrix form, which makes numerical calculations simpler. We introduce a method to incorporate capture histories with uncertain stage and sex identifications, which allows the use of capture history data with incomplete information. We introduce a simple function that allows multinomial transition probabilities to be written as functions of covariates (time or environmental factors): Finally, we show how to convert transition probabilities estimated by the MSMR method into a matrix population model. These methods are applied to data on the North Atlantic right whale (Eubalaena glacialis).