PARENT SESSION
Oral Session # 13: Statistical Ecology.
Presiding: R Stevens
Monday, August 4. 8:00 AM to 11:30 AM, SITCC Meeting Room 205.
Space-time modeling of a discrete ecological response: Characterization of Mallard nesting in Minnesota.
Das, Abhik*,1, Riggs, Michael1, Zicus, Michael2, Rave, David2, 1 Research Triangle Institute, Rockville, MD, U.S.A.2 Minnesota Department of Natural Resources, Bemidji, MN, U.S.A.
ABSTRACT- Discrete responses in the form of binary, categorical or count data, distributed over space and time, are common in ecology. For example, the U.S. Prairie Pothole Joint Venture seeks to increase waterfowl populations by reducing nest loss to predators. Mallards nesting in constructed nest structures often experience higher nest success than those nesting on the ground. Occupancy of these structures can be thought of as a binary (yes/no) outcome. Identifying environmental features (such as land use and cover attractiveness) responsible for the geographic distribution of nests over time is crucial to understanding the ecological mechanism affecting mallard nest distribution. Using this knowledge to predict occupancy of structures would be particularly useful for waterfowl management, since it would optimize deployment of structures where probability of use was greatest. In this study, we developed a generalized linear mixed model (GLMM) to characterize the spatio-temporal distribution of Mallard nests in 2 types of nest structures. Our approach used logistic regression, which is natural for binary data, while extending its scope to accommodate both spatial structure and temporal trends. Moreover, in order to identify the size of the area surrounding the nesting structure that had the most influence on nest occupancy, we present a likelihood-based procedure for model selection in a GLMM. The results show that, even after adjusting for spatio-temporal effects, ecological features such as nesting cover attractiveness type of nesting structure, and size of open-water area in deployment wetlands were significantly associated with nest occupancy. Our approach can accommodate space-time modeling for any discrete outcome. Thus, given the prevalence of such data in ecological studies, we believe these methods are broadly applicable to a variety of ecological research questions.
Key words: Space-time modeling, Discrete response, GLMM