Increasingly, climate change is pushing species to the limits of their thermal tolerance, with cascading effects across ecosystems. Animals use behavior to prevent these harmful physiological states, but their need and ability to do so varies with their traits. Within species, traits like sex and reproductive status affect heat sensitivity, perhaps eliciting differences in behavioral responses to thermal extremes. We evaluated whether sex and reproductive status affected thermoregulatory behavior and its efficacy in moose (Alces alces), a heat-sensitive endotherm that relies on thermal refuge. We expected traits associated with elevated heat load would be linked to heightened selection for thermal refuge and that differences in selection would successfully alleviate differing risks of overheating. Thus, reproductive females and males, who are more heat-sensitive, would have stronger selection for thermal refuge than non-reproductive females. We assessed selection of thermal refuge at bed sites and generated biophysical models to evaluate if selection mitigated risk of overheating. Reproductive status did not elicit differences in selection by females. The sexes, however, differed in selection of the tradeoff between solar cover and cooling from wind. Females selected refuge with canopy cover and avoided wind. Males did not select cover and had weaker avoidance of wind than females. Yet, both sexes were more likely to overheat in areas of low cover, even if wind speeds were high. Hence, males had weaker selection of refuge than females despite being more likely to overheat, and life history tradeoffs failed to explain the sub-optimal thermoregulatory behavior. We identify sex-specific thermoregulatory tradeoffs, highlighting the disproportionate effects of climate change on certain demographic groups. Moreover, we emphasize the relevance of trait-based approaches for studying changing ecosystems.

Capture and Handling

We used a helicopter to chemically immobilize adult moose (≥ 1 year) each year in March (2020–2023; Levine et al. 2022a). For the duration of the study, we aimed to maintain a sample size of 15 collared animals per sex. Males were captured once, and females were recaptured annually. We removed the right, incisiform canine (Swift et al. 2002) and determined age via cementum annuli (Matson’s Laboratory, Milltown, MT). We estimated body mass using morphometrics (Hundertmark and Schwartz 1998). To assess nutritional condition, we measured the maximum depth of rump fat using a portable ultrasound device (Ibex Pro, E.I. Medical Imaging, Loveland, Colorado) with a 5-MHz linear-array transducer (Stephenson et al. 1998) and accompanied ultrasound with palpation to determine a body condition score (Levine et al. 2022b). We derived percent ingesta-free body fat (hereafter nutritional condition) from scaled rump fat (Cook et al. 2010; Stephenson et al. 1998) and body condition score (Levine et al. 2022b). Nutritional condition of cervids in late winter is highly correlated with condition throughout summer (Monteith et al. 2013). We checked pregnancy status with transrectal ultrasonography (Stephenson et al. 1995). We fitted pregnant females with vaginal implant transmitters and deployed GPS collars with hourly fix rates on all moose (VERTEX PLUS Collar; Vectronic Aerospace GmBH, Berlin, Germany).

Field Observations

We assessed reproductive status of females by monitoring juvenile presence through observation (10 May–15 August). Within 48 h of vaginal implant expulsion, we observed females to determine neonate presence (detected or not detected), status (dead or alive in birth site), and number. We located females every 2-weeks throughout the summer to search for the juvenile. If a juvenile was not detected, we performed two additional searches to confirm the non-detection represented juvenile loss. Females were considered reproductive while the juvenile was present and non-reproductive when a juvenile was not observed for 2 consecutive visits Thus, following juvenile mortality, a female could switch in classification from reproductive to non-reproductive.

Bed Sites

During the summers of 2021 and 2022, we identified bed sites used by collared moose between parturition (mid-May) and growth of winter pelage (September) using mean location of clustered (≤15 m apart) GPS-collar fixes (Verzuh et al. 2021). We censored GPS data to 3-D or 2-D fixes and dilution of precision <2 (D’Eon and Delparte 2005; Moen et al. 1997). Sites were limited to beds used for 1+ h during daylight hours (i.e., sunrise to sunset at the site). We visited bed sites within 2 weeks of use. We confirmed bed sites by searching a 15m radius for signs of use: matted vegetation, muddy depressions, faecal matter, or moose hair. For each confirmed bed site (i.e., used site), we assessed three available sites at a randomly generated azimuth (0° to 359°) and distance (15 to 330m; Verzuh et al. 2021) from the used site.

To collect microclimate data at used and available sites, we placed weather stations at the centre of the bed (Kestrel Instruments 5500 Weather Meter, Boothwyn, PA). The weather stations were on tripods 30 cm off the ground with a rotating mount, wind vane, and shaded temperature logger. The stations collected temperature, wind speed, and relative humidity at 5 min intervals. We deployed weather stations in new locations each day for the hours of the day the bed site was used by the moose. We assessed other attributes relevant to thermal conditions at all sites including canopy cover, landcover type, vegetation density, and soil moisture (Verzuh et al. 2021). We measured canopy cover at 25% increments with a densitometer (Graphic Resource Solutions, Arcata, CA, U.S.A.) at the four corners of the bed site and averaged values to get mean percent canopy cover. We selected landcover type based on predominant habitat: conifer, riparian, sagebrush, meadow, agricultural, alpine, aspen, and talus. We later combined rare landcover classes (alpine, talus, and agricultural) and the common sagebrush class into a single ‘dry open’ classification to be used as a reference category. To assess soil moisture, we qualitatively classified dry (0%), moist (25%), or wet soil (75%), and used 2.5 cm as the lower limit for standing water (100%; NRCS 1998). We estimated vegetation density (i.e., horizontal cover) up to the height of a bedded moose (0.75 m) in the immediate vicinity of the bed as either sparse (0–10% cover), open (11–60% cover), or dense (61–100% cover). Soil moisture and vegetation density were modelled as continuous for ease of interpretation.