Paper No. 10-04
C Mazzetta, BJT Morgan and T Coulson
A state-space modelling approach to population size estimation
Abstract: We consider populations of wild animals that are closely monitored over time, by being recaptured on multiple occasions, until finally recovered dead or lost to follow up. We propose a state-space formulation that enables us to estimate, simultaneously: time-varying size, demographic composition and geographical dispersal of an open population. Simulations show that our method is robust to low proportions of monitored individuals. Parameters are estimated with MCMC methods within a fully Bayesian approach and our model is applied to a real population for which we are able to provide new results.
Keywords: Bayesian estimation; MCMC; Population ecology; Recapture; Recovery; Seasonality; Soay sheep; Spatio-Temporal; State-Space.