Online Resource 1 – Model Generation
We generated a candidate set of models for each thermoregulatory and foraging response variable (Table 1). Each model represented a hypothesis and incorporated factors reported in literature to influence torpor use in mammal and avian species or foraging behavior in aerial insectivore bats. We included each factor separately as a univariate model as well as in various combinations in multivariate models. The thermoregulatory global model included minimum daily Ta, roost type, reproductive condition, and previous night foraging duration. Minimum daytime Ta was included because energy use during thermoregulation is greatest at this point (e.g., Scholander 1955; Willis et al. 2005). We included reproductive condition due to the differences in physiological costs that likely occur during torpor use (delayed parturition or reduced milk quality and quantity) as well as energetic requirements in each condition. Because bats roosted in different structures at East site, we included roost type because microclimatic conditions typically vary between different structures (e.g., Lausen and Barclay 2006; Law and Chidel 2007). Foraging duration of the previous night was included because it is a relative indicator of energy availability: the amount of time spent flying was assumed to equate the amount of energy consumed. The model for roost type and minimum Ta was included because roosts likely buffer ambient conditions differently, thus affecting the temperature bats experience during the day. We constructed a model including roost type and foraging duration because it is conceivable that where a bat day roosted influences the amount of time foraging. Roost type and reproductive condition were combined because roost type selection varied by reproductive condition. Finally, foraging duration and reproductive condition were modeled together because of the energetic requirements of each reproductive condition (lactation requires more energy than pregnancy) that would need to be met through foraging.
A similar process was used for the foraging models. The global model included number of foraging bouts (Barclay 1989; Lučan and Radil 2010), reproductive condition (e.g., Brigham 1991; Dietz and Kalko 2007; Encarnação and Dietz 2006), daily torpor duration, minimum nightly temperature and average nightly wind speed (km/h; Arbuthnott and Brigham 2007; Barclay 1991; Kusch et al. 2004). Individual bat and night were included as random effects because there were many nights of data for individual bats and often multiple bats tracked on the same night. The models included combinations of the above variables as follows: 1) foraging bouts and reproductive condition because lactating bats are more likely to take more foraging trips than pregnant because they need to return to feed their pups. Foraging trips, daily torpor duration, and reproductive condition were included because of the energy balance in each condition through torpor use and the number of foraging trips taken. We included the interaction between minimum nightly temperature and average wind speed as a measure of relative insect abundance and reproductive condition because energy requirements vary between the conditions. An interaction model of temperature and wind speed to determine if insect abundance affects foraging duration.