Winter Habitat Use of Sichuan Sika Deer Cervus nippon sichuanicus in the Tiebu Sika Deer Nature Reserve

1Yibin Key Laboratory of Zoological Diversity and Ecological Conservation, Yibin University, Yibin 644007, P. R. China. 2Faculty of Agriculture, Forestry and Food Engineering, Yibin University, Yibin 644007, P. R. China. 3Sichuan Ruoergai Forestry Bureau, Aba 624500, P. R. China. Article Information Received 16 July 2021 Revised 18 September 2021 Accepted 01 October 2021 Available online 03 March 2022 (early access)


INTRODUCTION
S ichuan sika deer Cervus nippon sichuanicus is a subspecies of sika deer Cervus nippon, which belongs to the Cervinae, Cervidae, Artiodactyla (Corbet and Hill, 1991). Sichuan sika deer is distributed in a narrow range along the eastern edge of the Qinghai-Tibet Plateau. This deer is threatened by habitat loss and human disturbance (poaching, grazing, road building) (Zhao et al., 2014a). The wild population is about 1000 individuals (Qi et al., 2014). Accordingly, it is classified as a category i protected wild animal species in China, and listed as critically Endangered in the Red List of China's Vertebrates (Jiang et al., 2016).
Habitat is crucial for wildlife survival, and understanding the habitat use of target species is useful for designing an animal protection plan (Khadka and James, 2016). Spatial temporal variation affects wildlife habitat use . It has been shown that the key factors for habitat use by wild animals varied over different seasons (Chen et al., 2012;Rahman et al., 2017;Dargan et al., 2019), as the environmental pressure altered seasonally. Guo (2002) pointed that due to food availability, winter was the critical season affecting survival of Sichuan sika deer. However, till now, little research was focused on the winter habitat use by Sichuan sika deer, which severely restricts the development of Sichuan sika deer conservation.
To understand the relationship between the winter environmental pressure and Sichuan sika deer habitat use and to assist managers to design more effective protection plans, we conducted the research on the winter habitat use of Sichuan sika deer in Tiebu Sika Deer Nature Reserve. Our goals were to determine which factors had a greater influence on the habitat use of Sichuan sika deer in winter, to detect the habitat characteristics which Sichuan sika deer preferred in winter, and to improve protection plans for Sichuan sika deer conservation.

Study area
The study area is in the Tiebu Sika Deer Nature Reserve (abbreviation, TNR), China, which is located on the eastern edge of the Qinghai-Tibet Plateau. Elevation ranges from 2,450 m to 3,800 m above sea level. The annual precipitation is about 650 mm, with only ten percent of rainfall appearing from November to March. The mean O n l i n e

F i r s t A r t i c l e
annual temperature is 6.7 o C and extreme low temperature recorded is -18.9 o C. The TNR covers about 260 km 2 (Yang et al., 2012), which is a mosaic of forest, bush, and meadow. Forest is distributed mainly in shady slopes, bush distributed mainly on sunny slopes, and meadow occurs mainly on gentle areas (Zhao et al., 2014b).

Sampling method and variable design
We established transects along altitudinal gradients in the TNR, ensuring the representative vegetation types could be sampled with a transect (Morrison et al., 1992). Sampling plots were centered on signs of sika deer, were treated as used plots, with a distance not less than 100 m between them along transects. Control plots were established along transects at every 100 m change in elevation.
After each sampling point location was established, the vegetation type was determined. Then, seventeen biological and physical variables (Table I) were measured following the method of Wei et al. (2000), which was used in former sika deer research (Zhao et al., 2014b). We sampled 132 used plots and 62 control plots from January to February 2019 and 2020 (Fig. 1).

Data analyses
For categorical variables, exact goodness-of-fit tests were used to detect the difference between used and control plots, then post-hoc tests were conducted. Since we could not confirm that every variable in our research was normally distributed, both independent sample T tests and Mann-Whitney U tests were used to find the difference between used and control plots.

Variable Definition and description Vegetation type
Five categories: coniferous forest, broad-leaved forest, bush, meadow, cultivated field Slope aspect Aspect of each 400m 2 plot, defined by five categories: northern slope (315-45°), eastern slope (45-135°), southern slope (135-225°), western slope (225-315°), and no slope aspect (the plot in a flat land) Slope (°) The slope measured at the center of the 400 m 2 plot Elevation (m) The elevation measured at the center of the 400 m 2 plot Concealing condition Mean greatest distance looking eastward, southward, westward, and northward at a height of 1.0m at the center of the 400 m 2 plot, four categories: >40 m, 20-40 m, 10-20 m, and <10 m (as the concealing condition from low to high) Tree height (m) Average height of trees in 400 m 2 plot Tree density (n) Number of trees in 400 m 2 plot Tree size (cm) Average diameter at breast height of the trees in 400 m 2 plot Shrub height (m) Average height of shrubs in two 20 m 2 rectangular subplots Shrub density (n) Average number of shrubs in two 20 m 2 rectangular subplots Shrub cover (%) Average percent shrub cover in two 20 m 2 rectangular subplots Herb height (cm) Average height of herbs in five 1.0 m 2 subplots Herb cover (%) Average percent herbs cover in five 1.0 m 2 subplots Distance to forest edge (m) Estimated straight-line distance from the sampling plot to the nearest forest edge, if the plot was in the forest, record as +x, otherwise, record as -x Distance to water (m) Straight-line distance from the sampling plot centre to the nearest water source Distance to roads (m) Straight-line distance from the sampling plot centre to the nearest road Distance to habitation (m) Straight-line distance from the sampling plot centre to the nearest residential area O n l i n e

F i r s t A r t i c l e
To identify the key factors affecting sika deer habitat use, information-theoretic approach (Burnham and Anderson, 2002) was used. This investigation has conducted a correlation analysis to test independence between variables firstly. For those variables with a correlation coefficient (Pearson) above 0.5, only one variable remained in the subsequent analysis in order to reduce multicollinearity (Zhang et al., 2011;Torres et al., 2012). Then, GLM models with binomial distribution and logit-link were built. We calculated the Akaike information criterion (corrected for sample size, AIC c ) to evaluate model fit, using the differences in AIC c values (ΔAIC c ) to assess a candidate set of models of all of the possible combinations of variables. For models with ΔAIC c <2, we used model averaging to obtain estimates of each regression coefficient and confidence limits to identify the key factors (Creel et al., 2014).

RESULTS
For categorical variables, both vegetation type and slope aspect showed significant differences between habitat use plots and control plots (Table II). The outcome of post-hoc tests indicated that sika deer preferred bush and south slope, avoiding broad-leaved forest (Table II). For continuous variables, elevation, distance to forest edge, shrub height, shrub cover, shrub density, and herb cover showed significant differences between habitat use plots and control plots (Table III).
The top ten models contained 7 variables, herb cover, shrub cover, elevation, shrub height, concealed condition, distance to forest edge, distance to water respectively, of which, herb cover and elevation appeared in all the ten models (Table IV). The models averaged coefficients showed that, Sichuan sika deer preferred habitat with higher herb cover and lower elevation (Table V).

DISCUSSION
Diet and safety are the basic needs of wild animals (Rozhnov et al., 2015;Fattebert et al., 2019). Food availability varies over time, and winter is the diet shortage season for Sichuan sika deer (Guo, 2002). Previous research indicated that meadow was the main foraging patch of Sichuan sika deer habitat (Zhao et al., 2014b). Our research showed that herb cover was the key factor affecting the habitat use of Sichuan sika deer in winter (Table V), which indicated the importance of the diet need in winter. The safety needs caused Sichuan sika deer to utilize habitat with better concealing condition (Table III). The preference of shrub vegetation (Table II) was due to the fact that shrub could not only provide food but also acted as a shelter patch (Zhao et al., 2014b).
It has been demonstrated that human disturbance is an important factor affecting wildlife habitat use (Filla et al., 2017;Fattebert et al., 2019), while the influence of roads and habitations of Sichuan sika deer habitat use were not found in our results (Tables III and V). This may have been caused by the food availability stress, which forced the deer to utilize the habitat close to roads and habitations, and the protection of Sichuan sika deer by local Tibetans leads to O n l i n e

F i r s t A r t i c l e
Winter Habitat Use of Sika Deer 5 the harmonious coexistence between man and deer. Compared with summer, winter habitat use of Sichuan sika deer appeared different. In winter, Sichuan sika deer preferred habitat at lower elevation and south slope aspect, which lacked in summer (Zhao et al., 2020). These two factors may be highly related to temperature. Previous study demonstrated that temperature could affect habitat use behaviour in many kinds of large herbivores, such as alpine ibex Capra ibex (Aublet et al., 2009), black-tailed deer Odocoileus hemionus columbianus (Bowyer and Kie, 2009), mouflon Ovis gmelini musimon (Bourgoin et al., 2011), and moose Alces alces (van Beest et al., 2012). The lower elevation and south slope aspect may mainly be caused by warmer habitat selection of Sichuan sika deer. In addition, distance to water is an important factor affecting Sichuan sika deer habitat use in the summer (Zhao et al., 2020), which had little to contribute on habitat use of Sichuan sika deer in winter (Tables III, V). The snow scattered on the mountains served as a water source.

CONCLUSIONS
Our results clearly showed that the meadow was important as a food patch for Sichuan sika deer in winter, while water was not a restricting factor. Our results also indirectly demonstrated the influence of temperature on habitat use of Sichuan sika deer. Even though road and habitations had little influence on the habitat use of Sichuan sika deer in this research, considering the huge group of free-ranging livestock and intensive firewood cutting, we suggest the higher protection of the shrub and meadow vegetation especially in winter.