Optimizing energy acquisition and expenditure is a fundamental trade-off for consumers, strikingly reflected in how mobile organisms use space. Several studies have established that home range size decreases as resource density increases, but the balance of costs and benefits associated with exploiting a given resource density is unclear. We evaluate how the ability of consumers to exploit their resources through movement (termed “resource exploitation”) interacts with resource density to influence home range size. We then contrast two hypotheses to evaluate how resource exploitation influences home range size across a vast gradient of productivity and density of human-created linear features (roads and seismic lines) that are known to facilitate animal movements. Under the Diffusion Facilitation Hypothesis, linear features are predicted to lead to more diffuse space use and larger home ranges. Under the Exploitation Efficiency Hypothesis, linear features are predicted to increase foraging efficiency, resulting in less space being required to meet energetic demands and therefore smaller home ranges. Using GPS telemetry data from 142 wolves (Canis lupus) distributed over more than 500,000 km², we found that wolf home range size was influenced by the interaction between resource density and exploitation efficiency. Home range size decreased as linear feature density increased, supporting the Exploitation Efficiency Hypothesis. However, the effect of linear features on home range size diminished in more productive areas, suggesting that exploitation efficiency is of greater importance when resource density is low. These results suggest that smaller home ranges will occur where both linear feature density and primary productivity are higher, thereby increasing regional wolf density.

GPS telemetry data were collected between 2011 and 2017 from 36 individuals in 15 packs (BC), 34 individuals from 10 packs (AB N), 44 individuals from 16 packs (AB NE), and 28 individuals from 19 packs (SK). All animals were captured and handled using approved Animal Care protocols. Collars were programmed to retrieve GPS locations at varying intervals, from 5 minutes to 3 hours. We visually checked for location errors, and rarified data to the longest fix rate of 3-hour locations. BWe divided data into two seasons, roughly corresponding to the snow and snow-free seasons. To quantify home range area we created 95% autocorrelated Kernel Density Estimators (aKDEs; Fleming et al., 2015; Noonan et al., 2019) for each individual in each season using the ctmm package in R (Calabrese et al., 2016). Prior to creating aKDEs, we removed any individual season combination with less than 50 locations and visually assessed and removed individual seasons with extra-territorial forays. As suggested by Calabrese et al (2016), we used Maximum Likelihood to fit independent identically distributed, Brownian motion, Ornstein–Uhlenbeck, integrated Ornstein–Uhlenbeck, and Ornstein–Uhlenbeck Foraging models, and selected the model with the lowest Akaike Information Criterion (AIC) score (Akaike, 1974). For additional details see https://github.com/MelanieDickie/WolfHomeRange.