Multi-scale habitat selection modeling: a review and outlook

In their recent review, McGarigal et al. (2016) summarize the origins and types of multi-scale models in studies of habitat selection and perform a meta-analysis to examine the most common methods of multi-scale modeling in the field. At the outset, the authors take care to define what they mean by a “multi-scale” model, specifically by distinguishing them from “multi-level” models; according to this definition, a model that considers multiple levels of organization (e.g. species, populations, and individuals) is a multi-level model, whereas one that considers multiple spatial or temporal scales is a multi-scale model. This definition applies most clearly to statistical (rather than mechanistic) models, where each environmental covariate that may explain habitat selection could be used on a different spatial or temporal scale. McGarigal et al. argue that multi-level and multi-scale models – multi-level models being implicitly multi-scale – are relevant to habitat selection because habitat selection is a process that occurs on multiple spatial and temporal scales. For instance, habitat selection can produce a biogeographical species range, a home range of a social group, patch selection of an individual within that home range, and even resource selection within a habitat patch. Habitat selection can occur at a different spatial and temporal scale at each of these levels, meaning that a multi-scale model could potentially address the processes and drivers at multiple levels.

In their meta-analysis, the authors distinguish between two main groups of multi-level and multi-scale models: those that select a single scale of analysis and those that use multiple scales. A study on a single scale would select a priori a single (spatial/temporal) scale on which each predictor variable is relevant, though these scales may be different for each variable. In contrast, a multiple scale model would use multiple (spatial/temporal) scales for a single predictor variable.  After analyzing methodology from 173 papers, the authors conclude that only 20% of these studies of habitat selection use multi-level and multi-scale models. Within this 20%, the models fall into two main groups: hierarchical modelers, who analyze multiple levels of habitat selection (e.g. resource selection and home ranges), and scale modelers, who examine just one level but identify a potentially different scale of habitat selection for each predictor variable. For both of these groups, it was more common that studies examined multiple spatial scales than multiple temporal scales, likely because of the practicality of collecting temporal data. The paper concludes with a set of guidelines for multi-scale modeling of habitat selection, including that scales should be selected based on biological relevance and that, when possible, it is most effective to consider multiple scales even within one organizational level.

McGarigal, K., Wan, H. Y., Zeller, K. A., Timm, B. C., & Cushman, S. A. (2016). Multi-scale habitat selection modeling: a review and outlook. Landscape Ecology, 1-15.