Forage quantity, quality and depletion as scale-dependent mechanisms driving habitat selection of a large browsing herbivore.

1. Mechanisms that affect the spatial distribution of animals are typically scale-dependent and may involve forage distribution. Forage quality and quantity are often inversely correlated and a much discussed trade-off is whether or not to select for high-quality forage at the expense of forage abundance. This discussion has rarely involved scale-dependence or been applied to Northern browsing herbivores. At small spatial scales, browsers are assumed to select for the best quality forage. But, as high-quality forage resources are often scarce and may become depleted, coarse-scale habitat selection is assumed to be driven by forage availability. 2. To evaluate if moose selection for forage quantity and quality is scale-dependent we modelled summer and winter habitat selection of 32 GPS-marked female moose (Alces alces) at two spatial scales (landscape-scale vs. within-home range-scale). We used mixed-effects resource selection functions (RSFs) and landscape-scale forage availability models of six tree species of varying quality for moose. We considered silver birch (Betula pendula), downy birch (Betula pubescens.), Scots pine (Pinus sylvestris) as low quality browse species and rowan (Sorbus aucuparia), aspen (Populus tremula), willow (Salix spp.) as high-quality species. 3. As expected, the overall selection patterns for available browse biomass and quality varied across spatiotemporal scales. At the landscape-scale, moose selected for habitat with high available browse biomass of low quality species while at the within-home range-scale moose selected for sites with the highest quality browse species available. Furthermore, selection patterns during summer remained fairly stable, while during winter, selection at the within-home range-scale switched from sites with high quality to sites with lower quality browse species which suggests depletion of high-quality species. Consistent with expectations from seasonal resource depletion, site fidelity (bimonthly home range overlap) was much lower in winter than in summer. 4. Coarse-scale habitat selection by moose as a function of forage variability revealed a scale-dependent trade-off between available browse quantity and browse quality. Moreover, resource depletion changed the winter selection criteria of free-ranging moose and we demonstrate how the behavioural response to such a dynamic process can be inferred from RSFs.