U23 pro v2

Hourly temperature and relative humidity at 1.5 m.a.g.l were obtained from an automatic battery-powered recording device (HOBO U23 Pro v2; Onset Computer Corp., Cape Cod, MA, USA) fitted with a custom-made sun shield, which had been set up some 100 m from the study site in September, 2018. The downloaded data allowed characterization of temperature conditions and relative humidity on the site as of snow melt and for the specific days and hours of the pollinator observations each year. Instantaneous wind speed was recorded at 1.5 m.a.g.l. in two of the pollinator observation plots in Yr-1. In Yr-2, given that the elevation of the plots showed little variation, a single anemometer was set up. At the beginning and end of each 15 min observation period, wind speed (kph) was recorded to give six and three pairs of values per hour in 2019 and 2020, respectively. Cloud cover was recorded according to three categories (cloudless, partially cloudy, heavy clouds) during each 15 min observation.

The vaginal temperature was recorded using a Thermochron iButton 1922L data logger (Embedded Data Systems, Lawrenceburg, KY, USA) affixed to a blank CIDR using heat shrink tubing (Burdick et al., 2012 (link)). Beyond Burdick, paraffin wax was placed beneath where iButton was to be positioned to create a wax seal when heat was applied to shrink tubing. Paraffin wax was also inserted in the open spaces, on top and bottom of shrink tubing, where tubing contacted the blank CIDR. Hot glue was then placed along the top and bottom edge of the shrink tubing to create a final seal. The vaginal temperature was recorded every 5 min (0.065 °C resolution) beginning 4 to 12 h after iButton insertion and continued until iButton removal (11.9 ± 0.1 h after heifer was first observed to stand to be mounted by another). No iButton failures occurred. Onsite ambient temperature and humidity were recorded hourly using the HOBO U23 Pro v2 data logger (Onset Computer Corporation, Bourne, MA, USA) from PGF_2α administration to the time that the last iButton was removed. Temperature humidity index (THI) was calculated per Abbott et al. (2018) (link).

We measured the understory ambient temperature with temperature-data loggers (HOBO U23 Pro v2, Onset Computer Corp, Bourne, USA) placed ~30 cm above ground and recording temperature at 15 or 30 min intervals. Because our study species are exclusively diurnal, we calculated daytime temperature by averaging all measures from sunrise to sunset. At the O. sylvatica enclosure site, we manually measured temperature three times per day at the start, middle, and end of each tracking session with a handheld electronic thermometer (GFTH 95, GHM Messtechnik, Regenstauf, Germany) held slightly above the ground. We averaged the three measurements to obtain daytime temperature. We also calculated mean daytime temperatures for each frog during the navigation experiments by averaging daytime temperatures over the tracking period of each frog (also see Supplementary file 1, Table 1l). We did not use the rainfall data because they were strongly correlated with daytime temperature and missing for some tracking periods.

Temperature and relative humidity (RH) (Hobo U23 Pro v2, −40 °C to 70 °C, ± 0.2 °C, and 0 to 100%, ± 2.5%, Onset Computer Corporation, Bourne, MA, USA) were recorded continuously at 25 locations throughout the trailer with T recorded at additional 20 locations (DS1921L Thermochron, ± 0.5 °C, Maxim Integrated, Sunnyvale, CA, USA). Air speed were measured by six anemometers (Kestrel 4000; accuracy ± 0.1 m s⁻¹ from 0.4 to 40 m s⁻¹, Minneapolis, MN, USA) on two modules. Before each scheduled monitoring campaign, data loggers were installed at pre-determined module locations on empty trailers using cable ties at the processing plant (Figure 3a,b). Nine thermal loggers occupied each of the five vertical cross-sections (Figure 4). Temperature, humidity and wind were measured every 30 s. Times of departure from farm to the processing plant were recorded using a GPS unit (eTrex 20, Garmin, Olathe, KS, USA) that logged waypoints every second, downloaded after each trip.

Temperature was measured at hourly intervals for the following dates:
We did not include detailed surface measurements because the reduction of variation in caves relative to the surface is well known. Our focus was on variation within caves and the persistence of cyclical patterns. Onset Computer Corporation HOBO™ U23 Pro v2 data loggers were used to measure temperature and readings were accurate to ±0.21°C with a resolution of 0.02°C.