Infusing food web ecology into analyses of geographic distributions of species

Background/Question/Methods

Community ecology involves studying the interdependence of species with each other and their environment to predict their geographic distribution. Species distribution models (SDMs) characterize species-environment dependency quite well, but offer only crude approximations of interdependency. Typically the dependency between species is characterized using other species’ occurrences as spatial covariates to constrain the focal species’ predicted range, thus implicitly assuming homogeneous interdependency across space. We introduce a framework, known as food web module, which integrates principles from food web analyses, resource selection theory, and SDMs that spatially refined understanding of species distribution consequent to species resource selection and strength of species interactions within food webs across space. Our approach is illustrated using a case study involving Canada lynx (Lynx canadensis) and snowshoe hare (SSH; Lepus americanus) interactions with each other and their environment in Colorado, USA. After systematically produced a series of SDMs to predict the probability of presence for both lynx and SSH based only on biophysical variables alone, we estimated the probability of lynx distribution, given the biophysical variables and geospatial presence (i.e., availability) of SSH. We used a trophic interaction distribution model (TIDM) using spatial locations of known lynx-snowshoe hare encounters and interactions (i.e., accessibility).

Results/Conclusions

The food web module predicting the availability of lynx and SSH indicates indirect associations among species by overlapping occurrence within similar environmental conditions (proximity to intact and dense subalpine fir forest). In contrast, the food web module created with lynx and SSH encounters indicates strong direct interactions arise predominantly in subalpine fir, perhaps because forest patches are refuges from predation. These different perspectives show the kind of skew in landscape-scale fitness that may arise consequent to characterizing the geospatial locations of interactions based on prey accessibility. Integrating the spatial variation in accessibility of SSH into the lynx distribution model reveals a more nuanced structure within the geographic niche space where only 60% of the range where SSH are available to lynx has environmental conditions that facilitate greater accessibility of SSH to lynx. Distinguishing between potential resource availability and actual resource accessibility is the fundamental difference between characterizing species geospatial niches in terms of SDM versus TIDM. This approach not only represents a way of overcoming some of the hurdles in advancing niche theory with geospatial analyses of species distributions but also can serve as a predictive framework for conservation and wildlife management in the face of global environmental change.