# Winter bed-site selection by red deer Cervus elaphus xanthopygus and roe deer Capreolus capreolus bedfordi in forests of northeastern China

> Where does a deer sleep when the forest drops to minus 40? Following trails in the snow in northeastern China, researchers found that roe deer scraped every bed and chose thick conifer cover at night, while the larger red deer never scraped and changed their choices less between day and night. By day, both favoured south-facing slopes, and roe deer hid in denser ground cover.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/1999-chen-li-winter-bed-site-selection/

## Citation and access

- **Authors:** Huapeng Chen, Feng Li, Liyang Luo, Huai Wang, Jianzhang Ma and Fei Li
- **Year published:** 1999
- **Journal or source:** Acta Theriologica
- **Full citation:** Chen, H., Li, F., Luo, L., Wang, H., Ma, J. and Li, F. 1999. Winter bed-site selection by red deer Cervus elaphus xanthopygus and roe deer Capreolus capreolus bedfordi in forests of northeastern China. Acta Theriologica 44(2): 195-206.
- **DOI:** 10.4098/at.arch.99-17
- **Access status:** Hosted (freely available elsewhere)
- **PDF:** https://www.originalwisdom.com/wp-content/uploads/2026/11/1999_Chen_and_Li_etal_winter-bed-site-selection-by-red-deer-and-roe-deer.pdf

## Tags

- **Species:** Mammals > Red deer; Mammals > Siberian roe deer
- **Region:** Asia > China
- **Tracking:** Beds; Feeding signs; Scat; Snow tracks; Spoor age estimation; Tracking; Tracks; Tracks and signs identification; Trailing
- **Behaviour and ecology:** Behaviour; Energy budget; Habitat selection; Human disturbance; Predator avoidance
- **Methods and technology:** Chi-square test; Cover boards
- **Data type:** New field data
- **Observer qualification:** Reported but not validated
- **Discipline:** Behaviour and ecology; Wildlife tracking
- **Year published:** 1999

## Detailed keywords

- **Species:** Studied: red deer (Cervus elaphus xanthopygus) and roe deer (Capreolus capreolus bedfordi), as named in the paper. Other ungulates in the study area: wild boar (Sus scrofa), common; musk deer (Moschus moschiferus) and common goral (Naemorhedus goral), rare. A 1986 survey estimated 79 to 175 red deer, 136 to 350 roe deer and 43 to 93 wild boar on the hunting farm. Carnivores present at very low numbers: black bear (Selenarctos thibetanus), brown bear (Ursus arctos), yellow-throated marten (Martes flavigula), sable (Martes zibellina) and leopard cat (Felis bengalensis). Comparison species from other studies: white-tailed deer (Odocoileus virginianus), mule deer (Odocoileus hemionus), elk, moose.
- **Location and habitat:** Tonghe Hunting Farm (30 000 ha), southern slope of the Less Xingan Mountains (45°43’N, 128°40’E), northeastern China. Low, gently rising mountains at 400 to 900 m. Continental climate with long, cold winters (temperatures from about -40.4°C to 36.7°C); snow lies from late November to the end of April, average depth 43 cm. Mixed coniferous and broad-leaved forest (main trees Manchurian ash Fraxinus mandshurica, Korean pine Pinus koraiensis, Tilia amurensis, Phellodendron amurense, elms Ulmus spp., Acer mono) and oak (Quercus mongolica) forest. Five sampling areas of 50 to 200 ha each.
- **Topics:** winter bed-site selection; day beds versus night beds; thermal cover and hiding (concealment) cover; canopy closure; conifer cover; ground cover; slope aspect (north-facing night beds, south-facing day beds); bed scraping by roe deer; bed size; energy conservation in winter; body size and energy needs; antipredator behaviour; human hunting as the main predation pressure on roe deer; browsing at bed sites; seasonal change across early, mid and late winter; backtracking and forward tracking in snow.
- **Statistics or models:** Principal component analysis (PCA); discriminant function analysis (DFA: canonical discriminant analysis and stepwise discriminant analysis); Fisher’s protected least significant difference (FPLSD) for comparing each habitat variable; chi-square tests (slope aspect, browsing); point-quarter sampling of trees and shrubs; canopy closure from vertical canopy photographs scored on a 100-dot grid; cover density board read from 15 m; computed in SAS. Inconsistencies in the paper: (1) sample sizes differ between Table 1 (bed groups: red deer 23 day and 26 night, roe deer 36 and 36), Table 2 (bedding sites: red deer 22 and 26, roe deer 41 and 38) and Figure 4 (labels n = 21, 26, 38 and 40); the text says 75 red deer beds and 128 roe deer beds, which matches the Table 1 bed counts for size. (2) The Discussion says Armstrong et al.’s finding that night beds were closer to trees than day beds is “opposite to our results for roe deer”, but the Results, Table 2 and Figure 4 show no significant difference for roe deer (3.7 m day, 3.1 m night); it was red deer whose night beds were closer to trees. (3) The abstract lists conifer composition among the variables that separate day and night beds, but the stepwise analysis selected canopy closure, distance to the nearest tree, ground cover above 1 m, tree height and snow depth; conifer composition appears only as correlated with the first canonical variable. (4) The stepwise results mix “R” and “R²” labels. (5) In Figure 2 the error bar for roe deer night beds in early winter runs above 100% canopy closure.
- **Data type:** New field data: beds located by tracking in snow from late November to late March, 1991 to 1992; forest survey in winter 1991. Population figures from an unpublished 1986 survey.

## Synopsis

### 1. What they did

Red deer and roe deer share the forests of northeastern China, where winters are long and very cold and winter food is at its poorest. Where a deer chooses to lie down can help them save energy, so the researchers compared the winter beds of the two species. They expected that the smaller roe deer, which have higher energy needs for their size and digest winter browse less well, would choose more sheltered beds than red deer. They also expected night beds to be under denser canopy than day beds, and canopy cover over beds to be greatest in the coldest part of winter.

- Finding beds. In five sampling areas on Tonghe Hunting Farm, from late November to late March 1991 to 1992, a team of three walked in parallel about 100 m apart, looking for tracks in the snow. When they found a fresh trail in the early morning, they backtracked it to find the night bed and followed it forward to find the day bed. An experienced hunter judged whether tracks were fresh from the looseness and puffiness of the snow.

- Telling the species apart. Beds were identified as red deer or roe deer from the tracks, the size of the beds, whether the bed had been scraped, and the pellets.

- Measuring the site. Beds within 20 m of the first bed found counted as one bedding site, and one bed was chosen at random to represent it. They counted and measured the beds, and recorded 14 habitat features, including snow depth, distance to human disturbance, slope, canopy closure (from photographs taken straight up from each bed), ground cover (with a cover board read from 15 m), conifer composition, the nearest trees and shrubs, and browsing on nearby shrubs.

### 2. What they found

The beds

- All 128 roe deer beds had been scraped. None of the 75 red deer beds had.

- Red deer beds were larger and deeper than roe deer beds. Red deer night beds were longer than their day beds (about 133 cm versus 124 cm), and roe deer night beds were longer and wider than their day beds.

- The number of beds at a site (about 1.4 to 1.8) did not differ between species or between day and night.

Canopy and conifer cover

- For both species, night beds had more canopy closure than day beds: red deer 49.6% at night versus 23.5% by day, roe deer 78.5% versus 21.6%.

- At night, roe deer chose denser canopy than red deer.

- Roe deer night beds had far more conifers around them than their day beds (32.2% versus 4.3%). Red deer showed no such difference.

- Roe deer night beds had the most canopy closure in the cold mid winter (about 92%, read from the figure), more than in early or late winter. Their day beds did not change. Red deer showed no difference between mid and late winter.

Ground cover, slope and snow

- Roe deer day beds had better concealment (ground cover) than their night beds and than red deer day beds.

- Red deer bedded closer to trees at night (4.0 m) than by day (7.1 m). Roe deer showed no difference.

- For both species, most night beds were on north-facing slopes (61.6% red deer, 62.5% roe deer) and most day beds on south-facing slopes (60.8% and 57.5%).

- Red deer bedded in deeper snow than roe deer (about 21 cm versus 14 to 16 cm).

- Distance to human disturbance (about 820 to 890 m) did not differ between species or bed types.

Browsing

- Red deer browsed less around night beds (7.7% of sites) than day beds (31.8%). Roe deer showed no difference.

The analyses

- Six principal components explained 75.3% of the variation in the bed sites.

- Canopy closure, distance to the nearest tree, ground cover above 1 m, tree height and snow depth best separated day and night beds. Red deer and roe deer chose different bed sites, and roe deer differed much more between day and night beds than red deer did.

### 3. What they concluded

- Red deer and roe deer differ in where they bed in winter. Roe deer chose night beds with denser canopy and more conifers, and day beds with better concealment, than red deer.

- These differences reflect thermal cover and hiding cover, which may be closely linked to the different energy needs of the large-bodied red deer and the small-bodied roe deer.

- North-facing slopes usually carry the heaviest forest cover, which may explain why night beds were found there.

- Human hunting is the main predation pressure on roe deer in the area, so the dense ground cover at their day beds may be an antipredator strategy. Red deer may be more used to people, which may explain why their day and night beds did not differ in concealment.

- Bed scraping by roe deer may be a comfort and energy-saving behaviour.

- The authors caution that it is very difficult to tell whether tracks are fresh and whether a bed was used by day or by night, and that some conclusions need care because of some confusion in identifying tracks.

### 4. How tracking was or was not used

Tracking in snow was the core method. A team of three searched for tracks walking about 100 m apart. From a fresh trail found in the early morning, they backtracked to the night bed and trailed forward to the day bed. Freshness was judged by the looseness and puffiness of the snow in the tracks. Beds were assigned to red deer or roe deer from the tracks, the size of the bed, signs of scraping, and the pellets. Browsing on shrubs around each bed was also recorded.

### 5. Tracker or observer certification

Reported but not validated. An experienced hunter (described in the Discussion as having over 30 years of hunting experience) judged the freshness of tracks and helped decide whether beds were used by day or night. The hunter’s skill was not tested. Who identified the species from the beds is not stated. The authors note “some confusion in the identification of tracks”, which matters because the day and night comparison depends on these judgements.

### 6. Other technologies, skills or innovations

- Canopy closure measured from black and white photographs taken straight up from 0.5 m above the centre of each bed, scored on a 100-dot grid.

- A cover density board (0.3 by 2.5 m, marked in 0.5 m bands) read from 15 m in each of the four compass directions.

- Point-quarter sampling of the nearest trees and shrubs around each bed.

### 8. Implications for tracking (Original Wisdom's observations)

_These are Original Wisdom's observations, not the authors'._

- The study shows how much a trail in snow can reveal: by following one fresh trail backwards and forwards, the team found both where a deer spent the night and where they lay up after feeding.

- The beds themselves helped tell the species apart: roe deer scraped every bed, red deer never did, and red deer beds were larger.

- The authors’ own caution is a reminder that judging track freshness, and so whether a bed was used by day or by night, is a skill the results depend on.

## Cite this summary

Lawrence, K. 2026. Summary of: Winter bed-site selection by red deer Cervus elaphus xanthopygus and roe deer Capreolus capreolus bedfordi in forests of northeastern China (Chen and Li et al., 1999). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/1999-chen-li-winter-bed-site-selection/