Space use is a hierarchical process by which animals place their home ranges within the population range and selectively use habitat therein according to its availability, often producing a functional response pattern. This combination of scale-specific processes means that as the context of the home range within the population range changes, so too does the functional response within the home range. Modelling habitat selection either as a hierarchical process or incorporating the functional response at the home range scale improves the ability of habitat selection models to predict future habitat selection. To integrate these approaches, we present a scale-specific functional response model we call the ordered functional response (OFR). Within the established resource selection function framework, we model the functional response as an interaction between home range habitat selection and the difference between habitat available at the home range and population scales. We first simulate changes in habitat selection when individual home range placement differs, showing that the OFR both detects and models scale-specific functional responses that are often obscured by existing approaches. We then empirically test the OFR predictions by comparing how elk (Cervus canadensis) change their selection for conspecific density as population density changes across years.
Results/Conclusions
Our simulations showed that modelling the functional response at both the population and home range scales improved the predictive performance of habitat selection models. In simulations, the OFR predicted an increasingly stronger functional response at the home range scale the more individuals selected for simulated habitat at the population scale. We also confirmed empirically that consistent with a functional response, elk selection for density declined as density increased in their home ranges. Within a single year where there was no change in population density, both the OFR and home range functional response models provided a good fit to individual selection coefficients as density increased in their home ranges (RMSE = 1.23). However, when population density changed between years, the home range functional response model provided a comparatively poorer fit to individual selection coefficients for density (RMSE = 2.39) while the OFR retained its predictive accuracy across both years. Our work provides evidence that habitat selection is influenced by the context of habitat availability at multiple scales. Cross-scale functional responses can be employed to predict changes in animal space use with environmental variation, which is necessary as we make management recommendations for future populations.