|HOME SCHEDULE AUTHOR INDEX SUBJECT INDEX|
Predicting the vulnerability of endangered vertebrate species to fire ant impacts using landscape level predictors.
Parris, Leslie 1,2, Allen, Craig1,2, Schmidt, Elise3, Horton, Mac1, 1 2 3
ABSTRACT- Fire ants are capable of impacting populations of rare and endangered vertebrate species, yet risks have been quantified for few species. To investigate individual species would be time consuming and costly. We developed a predictive landscape-scale model of fire ant distribution for the state of South Carolina and conducted a spatial risk assessment of endangered species to fire ant impacts. We developed our landscape-scale model to predict fire ant occurrence using multivariate logistic regression based on 400 samples collected across South Carolina. We stratified sampling by physiogeographic region and satellite-derived landcover class, with multiple replicates of each site. Independent variables included percent tree canopy and understory, aspect, slope, and landcover type. Results of the logistic regression analysis identified 37.5% (6 of 16) of the land cover classes investigated as significant predictors of fire ant presence (P<0.10). We overlayed the model of predicted fire ant distribution with spatial models of rare and endangered vertebrate distribution and found that spatial overlap between fire ants and endangered species ranged 10% to 100%. Species with life history traits such as ovipary, ground nests, and altricial offspring, and that have distributions and habitat entirely shared with fire ants, are especially at risk to fire ant impacts. Species identified by these techniques to be at high risk, such as the least tern, are candidates for future manipulative research that will thoroughly quantify the level of risk.
KEY WORDS: fire ant, endangered species, GIS , spatial modeling