Hibernation duration data
We compiled hibernation duration data from literature, publicly available datasets, and our own field data. We also solicited data-based estimates of hibernation duration from local bat researchers across North America where few records were available from other sources (Table S1, Figure S1). We restricted the study region for this analysis to temperate North America (above the Tropic of Cancer and below the Arctic Circle), although the published range of M. lucifugus extends into the Arctic (Fenton and Barclay 1980). Spatially explicit records of immergence and emergence collected from the literature generally reported the average day of entrance or emergence, although some data were presented only as the duration of hibernation. Where sex-specific dates were given for a location, dates were averaged as insufficient data existed to complete a sex-specific analysis.
Acoustic bat recorders (Songmeter SM2+BAT; Wildlife Acoustics, MA, USA) were deployed by Wildlife Conservation Society Canada (WCS-C) to record bat activity across western Canada between 2008 and 2016. Microphones (either SMX-US or SMX-U1) were placed on 12-18-foot-tall telescoping poles above likely hibernation or commuting areas (e.g. riverbanks, cliff ridge tops) starting as early as mid-August. Acoustic activity was typically recorded throughout the duration of the winter and data were retrieved between mid-March and mid-May. Recordings were manually analysed using AnalookW software (Titley Scientific, Inc.) and customized noise-filters were used to pull files containing bat pulses. The occurrence of at least 2 bat echolocation pulses in a file was required to identify the recording as a ‘bat pass’, and the number of passes were summed nightly. The start of the hibernation period was defined by the last 3-night window between August 15 and December 31, in which ≥10 passes were identified, while the end of the hibernation period was defined by the first 3-night window between March 1 and May 15 in which ≥10 passes were recorded (Lausen and Barclay 2006). In locations where bat activity was consistently low throughout the year (spring and fall nightly activity often failing to exceed 10 passes), hibernation start and end points were defined by date of the last and first bat recordings, respectively. For sites with multiple years of data, we took the mean immergence and emergence dates.