1. Individual variation in habitat selection and movement behavior is receiving growing attention, but primarily with respect to characterizing behaviors in different contexts as opposed to decomposing structure in behavior within populations. This focus may be limiting advances in understanding the diversity of individual behavior and its influence on population organization. We propose a framework for characterizing variation in space-use behavior with the aim of advancing interpretation of its form and function.

2. Using outputs from integrated Step Selection Analyses of 20 years of telemetry data from African elephants (Loxodonta Africana), we developed four metrics characterizing differentiation in resource selection behavior within a population [specialization (magnitude of the response independent of direction), heterogeneity (inter-individual variation), consistency (temporal shift in response) and reversal (frequency of directional changes in the response)].

3. We contrast insight from the developed metrics relative to the mean population response using an example focused on two covariates. We then expanded this contrast by evaluating if the metrics identify structurally important information on seasonal shifts in resource selection behaviors in addition to that provided by mean selection coefficients through Principal Component Analyses (PCAs) and a random forest classification.

4. The simplified example highlighted that for some covariates focusing on the population average failed to capture complex individual variation in behaviors. The PCAs revealed that the developed metrics provided additional information in explaining the patterns in elephant selection beyond that offered by population average covariate values. For elephants, specialization and heterogeneity were informative, with specialization often being a better descriptor of differences in seasonal resource selection behavior than population average responses. Summarizing these metrics spatially and temporally, we illustrate how these metrics can provide insights on overlooked aspects of animal behavior.

5. Our work offers a new approach in how we conceptualize variation in space-use behavior (i.e., habitat selection and movement) by providing ways of encapsulating variation that enables diagnoses of the drivers of individual level variability in a population. The developed metrics explicitly distill how variation in a behavior is structured among individuals and over time which could facilitate comparative work across time, populations, or strata within populations.