Abstract
Wildlife population dynamics are influenced by a complex array of intrinsic and extrinsic factors, including competition, predation, human disturbance, and nutrition. Nutrition is among the most fundamentally important of these factors, and sound population management requires careful consideration of the potential role of nutrition in limiting population growth. Early spring is a critical time for ungulates to access nutritious forage following winter, and in many ecosystems newly emergent vegetation gradually moves up the elevational gradient in a process known as the green wave. This pattern is commonly quantified using remotely sensed greenness indices (RSGIs) like the Normalized Difference Vegetation Index. How well such indices capture variation in forage quality and abundance in forested environments, however, is uncertain. We evaluated relationships among NDVI, biomass and nutritional quality of forage, and movements of white-tailed deer (Odocoileus virginianus), elk (Cervus canadensis), and moose (Alces alces) in northern Idaho. We used generalized additive modeling to evaluate the utility of remotely sensed greenness indices versus other environmental covariates (e.g., climate and topography) for predicting spatiotemporal variation in (1) total forage biomass, and (2) biomass of high-quality forage in the Panhandle and Clearwater Regions of northern Idaho, USA. We used the best of those models to generate dynamic foodscape (i.e., total or high-quality forage biomass) maps for the study area. We then fit resource selection functions (RSFs) for each herbivore species to evaluate the relative importance of total biomass, high-quality biomass, canopy cover (percent tree canopy cover), elevation, vegetation greenness (NDVI) and rate of green-up (IRG), habitat type, and years since fire as drivers of movement and space use. Both total biomass and high-quality forage biomass were more strongly related to canopy cover than to RSGIs. In addition, total biomass (but not high-quality biomass) was more strongly related to elevation than to RSGIs. Overall, the inclusion of RSGIs in our foodscape models yielded limited additional predictive power. All three herbivore species showed strong selection for high-quality forage biomass, and indeed, forage biomass was the top continuous predictor of selection, with the exception of selection for high elevations by moose. In contrast, RSGIs such as NDVI and days from peak IRG were among the least useful predictors of selection by all three species. Moreover, for elk and moose, coefficients for NDVI were negative, indicating that those species were simultaneously showing strong selection for forage while avoiding ‘greenness’. Our results indicate that RSGIs are of limited value for predicting variation in forage resources or their use by large herbivores in forested environments. Despite their intuitive appeal and ease of accessibility, in forested systems RSGIs cannot be readily substituted for on-the-ground measurements of the plant traits that ultimately govern nutrient intake by herbivores.