Modeling the risk for a new invasive forest disease in the United States: An evaluation of five environmental niche models

Maggi Kelly,Qinghua Guo,Desheng Liu,David Shaari

DOI: https://doi.org/10.1016/j.compenvurbsys.2006.10.002

2007-01-01

Abstract:Efforts to model the potential habitat and risk for spread of invasive diseases such as Sudden Oak Death (SOD) are important for disease regulation and management. However, spatially referenced risk models using identical data can have differing results, making decision-making based on the mapped results problematic. We examined the results from five spatial risk models generated from common input parameters, and investigated model agreement for mapping risk for the causal pathogen for SOD, Phytophthora ramorum across the conterminous United States. We examined five models: Expert-driven Rule-based, Logistic Regression, Classification and Regression Trees, Genetic Algorithms, and Support Vector Machines. All models were consistent in their prediction of some SOD risk in coastal California, Oregon and Washington states, and in the northern foothills of the Sierra Nevada Mountains in California, and in an east–west oriented band including eastern Oklahoma, central Arkansas, Tennessee, Kentucky, northern Mississippi, Alabama, Georgia and South Carolina, parts of central North Carolina, and eastern Virginia, Delaware and Maryland states. The SVM model was the most accurate model, and had several advantages over the other models. Although theoretical in nature, this paper presents results that have practical, applied value for managers and regulators of this disease, and discusses common challenges in modeling invasive species niches over large scales.