A time-lagged effect of conspecific density on habitat selection by snowshoe hare

Ideal free distribution theory predicts that increased conspecific density redistributes individuals to low-density, suboptimal habitat. However, possible lags in response to population density remain poorly documented. Snowshoe hares (Lepus americanus) may exhibit density-dependent habitat selection due to its marked variation in population density. Based on 11 years (2004-2014) of snow tracking in Quebec (Canada), we investigated snowshoe hares' short-term and delayed habitat selection responses to population density. We predicted that at high densities, hare distribution expands into low-density habitat, thus weakening the association between hares and high-density habitat. We surveyed hare tracks along 95 km of transects on average each year and georeferenced 14,240 tracks. We used Generalized Estimating Equations for track count per 100 m transect segment as a function of the proportion of different forest age classes (0-20 y, 20-40 y and 40-80 y) within 50 m of the segments. We used model coefficients for each age class as a measure of habitat preference, and modeled those coefficients as a function of a population density index in current and previous winters. Coefficients for 20- to 40-y-old forests were positive each year, indicating that this habitat was preferred. The association between track counts and 20- to 40-y-old forest significantly declined with density during the previous winter, suggesting that hare spread from preferred forest with a lagged response to density. To our knowledge, no previous empirical studies have documented a lagged habitat selection response to population density. Time lags offer possible explanation for documented deviations from ideal free distribution models.