Manipulative, multi-standard test of a white-tailed deer habitat suitability model

Despite widespread use of habitat suitability index (HSI) models, testing has been limited in 2 areas that are of critical importance. First, while HSI models are commonly employed to predict the effects of hypothetical landscape manipulations, their predictive ability before and after a landscape manipulation is rarely evaluated. Second, model performance typically is compared to a single standard (e.g., density). If this standard does not reflect every way in which the animals use habitat, then management based on models calibrated according to the single standard may be flawed. I compared the predictions of an HSI model (summer, winter, and total) for white-tailed deer (Odocoileus virginianus) against 2 standards (foraging intensity and pellet-group count) before and after a timber harvest and in the presence and absence of hunting traffic. HSI (total model) values were significantly positively correlated to foraging intensity before, but not after, the timber harvest. The winter model was always significantly positively correlated with pellet-group count (r s ≥ 0.195). However, the total model was significantly (but negatively) correlated (r s ≤ -0.132) with pellet-group count in all conditions except in the presence of hunting traffic. The unreliable predictive ability of the model tested in this study emphasizes the caution with which HSI models should be applied and the need for further multi-standard, manipulative studies.