Conserving and managing habitat for animals requires robust models to predict their space use. The functional response in habitat selection posits that animals adjust their habitat selection according to availability.
Habitat availability can change over short time periods and small spatial distances, and thus failing to account for changes in habitat availability while modelling may not produce reliable predictions in the near–term or future. However, because individuals may respond to habitat availability differently, the functional response is also limited for predicting habitat selection by individuals.
Using a functional response in elk (Cervus canadensis) selection for mixed forest in response to road proximity, we compared habitat selection predictions made by population-level resource selection functions (RSFs) with random effects to incorporate individual differences in selection, to generalized functional response (GFR) RSFs.
We found that since not all individuals followed the road-dependent functional response, the random effects model both predicted the distributions of individuals more accurately (R2 = 0.62 vs. R2 = 0.51) and produced coefficient estimates that matched their selection for mixed forest and distance from roads better than the GFR model (RMSE = 0.25 vs. RMSE = 0.29 and 0.37 vs. 0.46).
Individual habitat selection often varies within populations, and revealing those differences shows how individuals help populations respond to environmental change. We suggest that evaluating individual differences using multiple predictive approaches is necessary to forecast long–term habitat selection.