A Bayesian Method to Assess the Fit of N-Mixture Models Used in the Estimation of Animal Abundance

• Activity Number: 356
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Sherwin Toribio*+ and Jason Rubbert
• Companies: University of Wisconsin at La Crosse and University of Wisconsin at La Crosse
• Keywords: N-Mixture ; Abundance Estimation ; Bayesian ; PPMC ; MCMC

Abstract

Estimating the abundance (or population size) of animals is a major concern within wildlife statistics. One convenient sampling design used to gather information about animal abundance is the simple count method. In this sampling procedure, observers visit randomly selected sites several times and record the number of animals of a certain species per visit. Although this sampling method is more convenient than the capture-recapture method, the statistical methods needed to obtain the estimates of animal abundance is much more complicated. In 2004, Andrew Royle proposed a statistical method, using N-mixture models, to obtain abundance estimates from this kind of data. However, this method is only good if the model assumptions are satisfied. If some of the assumptions are not true, we have discovered that the estimates from this procedure can be very biased. In this talk, a Bayesian method is presented that can effectively detect violations of some of the model assumptions in the data.