Application of Bayesian Statistical Inference and Decision Theory to the Adaptive Management of Threatened Wildlife Species

• Activity Number: 440
• Type: Contributed
• Date/Time: Thursday, August 7, 2003 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics & the Environment
• Author(s): Howard B. Stauffer*+
• Companies: Humboldt State University
• Keywords: Bayesian statistics ; adaptive management ; Bayesian decision theory ; habitat conservation

Abstract

I will describe a decision-making protocol for the monitoring of a wildlife population with adaptive management, based upon Bayesian statistical inference and decision theory. The protocol provides an optimal decision-making process to periodically reassess a parameter, such as a proportion, survival, or fitness, with respect to a threshold level. Frequentist decision-making protocols are usually based upon independent, dichotomous decisions, using estimates, confidence intervals, or hypothesis testing. I will present a more informative approach using Bayesian statistical inference. The approach will assess posterior distributions for the parameter, changes in the posteriors using Bayes factors, and a decision-making protocol based upon the Bayes rule of minimal risk depending upon the 1) posterior distribution of the parameter; 2) losses and risks of the actions of the decision protocol; and 3) sensitivity and specificity of the monitoring survey. Application of the odology will be illustrated for the monitoring of the endangered species, Northern Spotted Owl, providing an adaptive management strategy required for a timber company's Habitat Conservation Plan.