Predictive Modeling of Kudzu (<i>Pueraria montana</i>) Habitat in the Great Lakes Basin of the United States

Abstract

Kudzu (Pueraria montana [Lour.] Merr. var. lobata [Willd.] Maesen & S.M. Almeida ex Sanjappa & Predeep) is an invasive woody vine widespread throughout much of the southeastern United States. New occurrences and recent studies using climatic parameters suggest that the Midwestern region of the United States is at the greatest risk of kudzu invasion. As there are already multiple reports of kudzu within the Great Lakes basin and no previous landscape models exist specifically for the basin, we developed probability models from existing spatial data (forest type, geology, land cover, precipitation, temperature, and known kudzu locations) by using maximum entropy methods at the national, regional, and basin scales. All three models had relatively high accuracy and strong positive correlation between predicted and observed values. Based on evaluation of the models using a testing data set, we determined a presence threshold and categorized areas within each model as suitable or unsuitable habitat. We pooled the models and calculated mean habitat suitability within the Great Lakes basin. Much of the southern half of the basin was suitable for kudzu. Continuing management and further monitoring of kudzu spread are likely necessary to limit further introduction and mitigate spread of kudzu within the Great Lakes region.

Keywords

• species distribution model
• SDM
• kudzu
• invasive
• maximum entropy
• great lakes