Bayesian Analysis and Matching Errors in Closed Population Capture Recapture Models

Abstract

Capture-Recapture models are used to estimate the unknown sizes of animal populations. When the population is closed, with constant size, during the study, eight standard models exist for estimating population size. These models allow for variation in animal capture probabilities due to time effects, heterogeneity among animals, and behavioral effects after the first capture. Our research focuses on two areas:

1. Using Bayesian statistical modeling, we present versions of these eight models. We explore the use of Akaike's Information Criterion (AIC), and the Deviance Information Criterion (DIC) as tools for selecting the appropriate model for a given dataset. Through simulation, we show that AIC performs well in model selection.

2. A new, non-invasive method of capturing animals is to substitute captures of DNA profiles, through sources such as hair samples, for live animal captures. However, DNA profiles of close relatives may not be distinguishable from each other, and some animals in the population may not be uniquely identifiable. This problem leads to negative bias in estimating population size. We present a hierarchical statistical model which accounts for this type of matching error, leading to more accurate estimation of population size.