A hybrid correlative‐mechanistic approach for modeling winter distributions of North American bat species

Abstract

Aim The fungal pathogen Pseudogymnoascus destructans and resultant white‐nose syndrome (WNS) continues to advance across North America, infecting new bat hibernacula. Western North America hosts the highest bat diversity in the United States and Canada, yet little is known about hibernacula and hibernation behaviour in this region. An improved understanding of the distribution of suitable hibernacula is critical for land managers to anticipate conservation needs of WNS‐susceptible species in currently uninfected regions. Location United States, Canada. Taxon Bats. Methods We estimated suitability of potential winter hibernaculum sites across five bat species' ranges. We estimated winter survival capacity from a mechanistic survivorship model based on bat bioenergetics and climate conditions. We then used boosted regression trees to relate these estimates, along with key landscape attributes, to bat occurrence data in a hybrid correlative‐mechanistic approach. Results Winter survival capacity, topography, land cover and access to subterranean features were important predictors of winter hibernaculum selection, but the shape and relative importance of these relationships varied amongst species. This suggests that the occurrence of bat hibernacula can, in part, be predicted from readily mapped above‐ground features, not just below‐ground characteristics for which spatial data are lacking. Furthermore, our mechanistic estimate of winter survivorship was, on average, the third strongest predictor of winter occurrence probability across focal species. Main conclusions Winter distributions of North American bat species were driven by their physiological capacity to survive winter conditions and duration in a given location, as well as selection for topographic and other landscape features but in species‐specific ways. The influence of winter survivorship on several species' distributions, the underlying influence of climate conditions on winter survivorship and the anticipated influence of WNS on bats' hibernation physiology and survivorship together suggest that North American bat distributions may undergo future shifts as these species are exposed not only to WNS but also to climate change.

Predicting habitat suitability for Townsend's big‐eared bats across California in relation to climate change

“…Notably, in our study area, elevation was highly correlated with annual temperature (bioclimatic variable 1). We retained elevation for our final models as this variable has been found to be important predictors of roost selection in previous studies of C. townsendii (Harris et al, 2019 ; McClure et al, 2021 , 2022 ; Sherwin et al, 2000 ). For future climate conditions, we selected three general circulation models (GCMs) based on previous species distribution models of temperate bat species (Razgour et al, 2019 ) [Hadley Centre Global Environment Model version 2 Earth Systems model (HadGEM3‐GC31_LL; Webb, 2019 ), Institut Pierre‐Simon Laplace Coupled Model 6th Assessment Low Resolution (IPSL‐CM6A‐LR; Boucher et al, 2018 ), and Max Planck Institute for Meteorology Earth System Model Low Resolution (MPI‐ESM1‐2‐LR; Brovkin et al, 2019 )] and two contrasting greenhouse concentration trajectories (Shared Socio‐economic Pathways; SSPs): a steady decline pathway with CO 2 concentrations of 360 ppmv (SSP1‐2.6) and an increasing pathway with CO 2 reaching around 2000 ppmv (SSP5‐8.5; Masson‐Delmotte et al, 2021 ).…”

“…Notably, in our study area, elevation was highly correlated with annual temperature (bioclimatic variable 1). We retained elevation for our final models as this variable has been found to be important predictors of roost selection in previous studies of C. townsendii (Harris et al, 2019;McClure et al, 2021McClure et al, , 2022Sherwin et al, 2000). For future climate conditions, we selected three general circulation models (GCMs) based on previous species distribution models of temperate bat species (Razgour et al, 2019)…”

“…We did not include acoustic record or mist net capture as these types of records do not allow us to characterize type of environmental use (e.g., maternity vs. transition roosts), meaning that these types of landscape detections are likely far less ecologically meaningful than the presence of the presumably more limiting roost locations. Excluding foraging locations may underestimate the realized niche, but roost data have successfully produced roost habitat maps for other temperate bat species (McClure et al, 2021(McClure et al, , 2022Smeraldo et al, 2018). Additionally, we did not include roost-habitat covariates (e.g., humidity, size) in our models because we do not have adequate dimensional or microclimate data for all subterranean and anthropogenic features in the study area.…”

Macroevolutionary Constraint and Selection on Anti-Bat Moth Tails

“…We selected model predictors based on other macroecological studies of bats [49][50][51][52]; for example, we used topological ruggedness and roughness as proxies for cave and carst roosting habitats used by bat species [49]. Initial models were fit using 15 candidate predictors from BioClim (BIO1-2,4-6,12-17 [53]), three topographic (elevation, roughness, and terrain roughness index [54]), and one from MODIS data (percent tree cover [55]).…”

Wintering Conditions and Heat Loss during Hibernation in the Brown Long-Eared Bat

“…The interior microclimates of underground environments (caves, adits, bunkers, etc.) where bats hibernate can be highly diverse due to the number of entrances, airflow direction and velocity, depth and other properties of each underground system [10,11]. The field of science that researches and analyses such microclimates is called micrometeorology [12][13][14][15], sometimes referred to as cave meteorology [16].…”