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.