Appendix 1: Winter survivorship model
Full details describing the hibernation energetic model structure and parameterization are described elsewhere (Haase et al. 2019, Hranac et al. accepted); here we provide a brief description and details regarding spatial application of the model across the study extent. The model uses the hypothesized energetic requirements of bats in torpor to dynamically model torpor bouts for the duration of a predicted winter under specified hibernation conditions. Specifically, ambient temperature and water loss are drivers of hibernating bats’ arousal frequency over the course of the winter, which subsequently drives energy expenditure and fat loss. Likelihood of winter survivorship can be estimated based on the predicted fat mass remaining at the end of winter. Key parameters include bat morphometrics and metabolic rates, whether bats are infected with P. destructans, and hibernaculum climate conditions.
The model (as described in Haase et al. 2019) was applied for bats assumed to be uninfected with P. destructans using species-specific metabolic and morphometric parameter defaults contained with the batwintor R package (in development: github.com/cReedHranac/batwintor). We then applied the resulting model at each 1-km2 grid cell across the study extent using spatially explicit estimates of mean ambient temperatures and winter duration and a fixed relative humidity value of 95%.
Ambient temperature. Hibernating bats are understood to prefer and select particular temperatures from the range of temperatures that may be available within a given hibernaculum. As subterranean temperatures in caves and mines are known to deviate from mean annual surface temperature (MAST) due to a variety of factors (Perry 2013), we developed a model to predict the availability of suitable hibernacula temperatures (McClure et al. 2020).
We estimated the mean ambient temperature likely to be experienced over the course of hibernation in any given location across North America should a suitable hibernaculum exist, for each of our five focal species. We first identified the mean ambient temperature at which each species has been observed during hibernation from the available published literature (Table S1), and made the assumption that this mean represents the species’ preferred hibernation temperature. We then used a spatially explicit model of subterranean winter temperatures to estimate the closest available temperature to this preferred temperature at any given location (McClure et al. 2020). The model estimates subterranean winter temperature based on MAST, distance from the site entrance, site type (cave or mine), and several less influential predictors representing topography, land cover, and presence of water. The model predicts an increase in subterranean temperature with increasing MAST and distance from the site entrance, and predicts higher temperatures in mines than in caves.
We bracketed the conditions expected to be available at a given site by assuming, based on field observations, that bats would hibernate between 50 and 100 meters from the site entrance (except C. townsendii, which were assumed to hibernate between 10 and 100 meters from the site entrance) (C. Lausen, personal communication). To estimate the warmest temperature potentially available at a given site, we predicted subterranean temperatures at 100 m from entrances of mines. To estimate the coldest temperature potentially available, we predicted subterranean temperatures at 50 m (10 m for C. townsendii) from entrances of caves. We then conditionally selected the best available temperature (i.e., the closest to the species-specific preferred temperature) for each raster cell across North America. Each cell was assigned the species’ preferred temperature if this temperature was bounded by the coldest and warmest temperature predicted to be available for that cell. If it was not, the cell was assigned the closest temperature to the preferred temperature that was predicted to be available. If the available range was too cold, the warmest available temperature (i.e., the predicted temperature for mines at 100 m) was assigned, and if the available range was too warm, the coldest available temperature (i.e., the predicted temperature for caves at 50 m/10 m) was assigned. For cells where the best available temperature differed from the parameterized optimal hibernacula temperature, the metabolic rate was scaled through the q10 relationship, as fully described in Haase et al. (2019) and Hayman et al. (2016). In all cases, deviation away from the parameterized temperature will increase the rate of metabolic expenditure. At the warmest of locations, bats may be unable to fully enter torpor and therefore require vastly more energy than a torpid individual to survive a hibernation duration of the same length.
This approach best captured our assumption that bats will select microsites within hibernacula that offer their preferred temperature when possible, but will likely tolerate warmer or cooler temperatures when necessary, especially at the margins of their ranges.
Table A1. Published literature containing observed ambient hibernaculum temperatures for focal species C. townsendii, M. californicus, M. lucifugus, M. velifer, and P. subflavus that were used to estimate preferred hibernation temperatures.