A GeoTiff of a reclassified black bear mahalanobis distance statistic model for Arkansas and Louisiana.

The original model was created using 75% fixed kernel home ranges (n = 52) for bears throughout Arkansas and Louisiana. The four variables that were in the model were water density, distance to roads, distance to urban areas, and a weighted bear land cover variable. We used the cumulative frequency distribution of the pixel values within the bear home ranges to determine binned values of the reclassification. So a pixel with a habitat index value of 1,000 is twice as likely to be similar to the bear home range target "ideal" as a pixel with a habitat index value of 500. It does not mean that that habitat is twice as likely to be selected by a bear as a pixel with a value of 500; that probability is unknown and cannot be estimated with the mahalanobis method. The threshold of sensitivity and specificity is approximately 300. So, any value above 300 could be considered "bear habitat", with the higher values indicating a closer approximation of the bear home range target "ideal".