The Jolly-Seber-Tag-Loss model with group heterogeneity

Abstract

Mark-recapture experiments are performed to estimate population parameters such as survival probabilities. Animals are captured, tagged, released, and recaptured at subsequent time periods in order to obtain parameter information. The Jolly-Seber-Tag-Loss (JSTL) model (Cowen & Schwarz, 2006) requires some individuals to be double tagged in order to account for the possibility of animals losing their tags. The Jolly-Seber-Tag-Loss model does not, however, consider the possibility of parameters being different among different groups of individuals, that is, group heterogeneity (for example, males may have higher capture probabilities than females). Our research extends the Jolly-Seber-Tag-Loss model to account for this possibility of group heterogeneity among parameters. We use a Newton-Raphson method to obtain maximum likelihood estimators and R software to create a program that estimates population parameters from tag histories. Our simulation study concludes that when group heterogeneity exists, accounting for this group heterogeneity results in more accurate parameter estimates than the original JSTL model. We present the group heterogeneous JSTL (g-hJSTL) for this purpose.