RESUMO
Anthrax is an economically important livestock disease affecting subsistence farmers and it is of zoonotic importance. Anthrax is endemic in many states of India including Karnataka. Identification of spatial risk factors for occurrence of anthrax and development of predictive risk maps are required for planning adequate vaccination in high-risk areas as well as targeted surveillance activities in animals, humans and environment. In this study, village level anthrax outbreak locations from Karnataka (1997-2016) were geo-referenced and predictive risk map was developed using temporally Fourier Processed remotely sensed variables. A non-linear discriminant analysis approach was used to develop the risk map for Karnataka. Elevation was identified as top predictor variable in the 10 variables selected. The predicted risk map showed good accuracy and validation statistics when evaluated using different metrics (Kappa, sensitivity, specificity, AUC). The predicted risk map also showed good correspondence with past outbreaks. Further, we used Bayesian Penalised spline method to estimate species response curves for top 10 variables selected. The validated risk map can be used in planning vaccination strategy and surveillance in high-risk areas.