Predicting the Geographic Distribution of Two Invasive Termite Species in Florida Using a Bayesian Logistic Model for Presence-Only Data

• Activity Number: 195
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Francesco Tonini*+ and Fabio Divino and Giovanna Jona Lasinio and Hartwig Hochmair and Rudolf H. Scheffrahn
• Companies: and University of Molise and University of Rome, La Sapienza and University of Florida and University of Florida
• Keywords: MCMC ; Bayesian logistic modeling ; presence-only data ; data augmentation ; subterranean termite ; invasive species

Abstract

The ability to predict potential geographic distribution of species is a tenet in ecology, biogeography, and conservation. The past decade has seen a marked increase in the development of models and approaches addressing the difficulties of predicting biotic distributions. Sites of known occurrence (presence data) and, sometimes, non-occurrence (absence data) are related with predictor variables recorded over the whole study region. When absence data is either lacking or not reliable, classical generalized linear models cannot be used without ad-hoc modifications. In this work, a Bayesian logistic model adapted to presence-only data is used to predict the potential geographic distribution of two invasive termite species, Coptotermes gestroi (Wasmann) and Coptotermes formosanus Shiraki, over a 2 km grid for the State of Florida. For both species, we use their recorded occurrences and a set of relevant climatic/environmental predictor variables. Within our model framework, a MCMC algorithm with a data augmentation step is used for the estimation of logistic model parameters.