Adaptive response of Siberian roe deer (Capreolus pygargus) to climate and altitude in the temperate forests of South Korea

Understanding climatic effect on wildlife is essential to prediction and management of climate change’s impact on the ecosystem. The climatic effect can interact with other environmental factors. This study aimed to determine effects of climate and altitude on Siberian roe deer (Capreolus pygargus) activity in temperate forests of South Korea. We conducted camera trapping to investigate roe deer’s activity level from spring to fall. Logistic regressions were used to determine effects of diel period, temperature, rain, and altitude on the activity level. A negative relationship was noted between temperature and the activity level due to thermoregulatory costs. Roe deer activity exhibited nocturnal and crepuscular patterns during summer and the other seasons, respectively, possibly due to heat stress in summer. In addition, the effect of temperature differed between high- and low-altitude areas. In low-altitude areas, temperature affected negatively the activity level throughout the study period. Conversely, in high-altitude areas, temperature affected activity levels only in summer and early fall. Lower temperatures in higher altitudes favoured roe deer activity, resulting in roe deer’s preference towards higher altitude areas. However, roe deer’s movement toward lower altitudes was observed in summer. Reduced heat stress by changing activity pattern allowed them to access lower altitude areas with greater resource availability during summer. This study revealed how roe deer activity varied across seasons and altitudes, considering the interactions among weather, microclimate and resource availability. It provides insight into how montane species adapt to various climatic conditions, and this could have important implications for wildlife management and conservation efforts.

Methods
The camera trapping method was used to observe temporal variations in roe deer capture (sampling days: September to October 2021 and April to August 2022). In the study area, a 5 × 6 grid design (interval = 600 m) was established, and one trail camera (Spec Ops Elite HP4; Browning Co., USA) was deployed corresponding to each cell of the grid. The study period was divided into five seasons, and further analyses were performed for each season: spring (15 April to 15 May, 960 trap-days), early summer (16 May to 30 June, 1380 trap-days), summer (1 July to 31 August, 1860 trap-days), early fall (September, 900 trap-days) and fall (October, 810 trap-days). The camera-plot altitudes were categorised into four classes: 600 (600–800 m asl, n = 3), 800 (800–1,000 m asl, n = 10), 1,000 (1,000–1,200 m asl, n = 11) and 1,200 (1,200–1,400 m asl, n = 6). We created a roedeer variable as presence/absence of observation per 2-h in each altitude class. In order to account for sampling effort depending on altitude, a weight variable was created as the total number of plots/the number of plots at the altitude. A temp variable is a predicted value of temperature at an altitude of 1,000 m, based on temperature data from a weather station, located 560 m asl. A rainy variable was established based on precipitation data from the weather station; it determined whether precipitation of sampling day was 0 mm or not. The diel variable explains diel period, such as dawn, day, dusk and night. Dawn and dusk were designated as the time within 2 h before and after sunrise and sunset, respectively.