Habitat selection as a reduction in habitat variance

When organisms select habitat, they may lessen the environmental variance they experience relative to their surroundings. Such discrepancies in variance could reveal organisms’ habitat preferences and perceptual ranges, particularly when examined across spatial and temporal scales. To test whether habitat variance might provide such understanding, we applied geostatistics to the variance in availability and use of habitat by GPS-collared adult female white-tailed deer (Odocoileus virginianus) in central Ontario, Canada. First, to quantify use and availability, we measured vegetation at the locations used by deer during May–June 2022, as well as in the general environment, then applied principal components analysis (PCA) to capture the major gradients in vegetation conditions. Second, to identify habitat selection, we tested for differences between use and availability in both the means and variance of vegetation characteristics. Finally, to depict how variance changed across scales, we constructed spatial and temporal variograms. Based on the first two axes of the PCA, we found that deer selected greater abundance in forage and lower variation in canopy closure. Across space, the selection for reduced variance in canopy appeared largely independent of scale (50–1000 m), implying that the perception capacity of deer may exceed this range. Across time, deer exhibited rising variance in forage abundance at short lags (4–12 h), resembling the periods (6–10 h) when movements were more linear. Deer thus selected for lower variance of habitat without selecting for disproportionate levels of habitat. We propose that selection for diminished variance is a fundamental property of habitat selection, whose scale-dependence might be uncovered with geostatistics.