Spatially Modeling Occupancy for Detection Data Collected Using Alternating Survey Methods

• Activity Number: 254
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Katherine St. Clair*+ and Jun Young Park and Brian Gray
• Companies: Carleton College and Carleton College and U.S. Geological Survey
• Keywords: occupancy modeling ; hierarchical model ; Bayesian ; detection probability

Abstract

We constructed a hierarchical single-species occupancy model that incorporates heterogeneity of conditional detection probabilities and addresses correlation of within-site occupancy status using a spatial CAR model, and applied this model to submersed aquatic vegetation (SAV) detection data from 34 sites from the Connecticut River, USA. Each site was divided into16 plots and detection data collected using a visual survey method at eight plots and using a more accurate in-water method at the remaining plots; survey methods were allocated using a checkerboard design. We used Bayesian inference methods to compare the detection accuracy of the two survey methods and to estimate occupancy across the study site. The in-water survey method was found to be more accurate than the visual method, though we found that its accuracy was species dependent.