Binomial Mixture Models for Urban Ecological Monitoring Studies Using American Community Survey Demographic Covariates

• Activity Number: 645
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Guohui Wu*+ and Scott H. Holan and Charles Nilon and Christopher K. Wikle
• Companies: University of Missouri and University of Missouri and University of Missouri and University of Missouri
• Keywords: Bayesian ; Count data ; Conway-Maxwell Poisson ; Negative-Binomial ; Unbalanced design ; Variable selection

Abstract

Studies of abundance constitute and important component of ecological monitoring and facilitate a deeper understanding of various species. In particular, one of the primary objectives of these studies is to understand how certain species are distributed across the study area. To achieve this goal, typically, sampling designs are put in place to collect spatially replicated count data capable of simultaneously assessing the probability of detection. The resulting data are often unbalanced and not equidispersed. Motivated by the Baltimore Ecosystem Study, we propose a class of hierarchical Bayesian Binomial mixture models that includes habitat and environmental covariates, many of which come from the American Community Survey, and accommodates different levels of over- and underdispersion. To increase the flexibility of our models we conduct variable selection and grouping through reversible jump Markov chain Monte Carlo methodology. The effectiveness of our approach is demonstrated through the estimation European starling abundance in the Baltimore Ecosystem Study.