Pre-hibernation fat and body mass data
The data required to directly assess the spatial variation in body fat for M. lucifugus prior to hibernation do not exist, so we relied upon scaling fat mass with spatially-varying body mass. We used body composition datasets obtained by quantitative magnetic resonance (QMR; McGuire and Guglielmo 2010) measurements that provided both pre-hibernation fat mass and body mass information from multipleM. lucifugus populations, including New York, Vermont (McGuire et al. 2018) and Montana (Haase et al. 2019). We tested for differences in fat among locations using a linear model with fat as a response variable. Given that there was no significant difference among locations, we then fit a linear model with fat as the response and body mass as the predictor variable to predict fat mass from body mass (French 1985). This fitted relationship then allowed us to predict fat mass at locations where we only had body mass measurements. We gathered additional body mass data from the literature and VertNet (vertnet.org), in which records were filtered to include only those with geographic location, body mass, and recorded between September and December (Table S2, Figure S1).