Using snow-track surveys to determine deer winter distribution and habitat

1. What they did

Forest managers in the Slocan Valley of British Columbia needed to know where deer spend the winter, at a level of detail useful for day-to-day planning. Counting deer from the air does not work there, because the dense conifer canopy hides the animals. So the author tested a ground method: reading deer tracks in the snow.

• Sampling design. The study area (26,800 ha, below 1,500 m) was divided into 24 strata by elevation, terrain (ridge, riparian or slope face) and aspect. Transect starting points were chosen at random in proportion to each stratum’s area, and each transect ran straight up or down the slope, 300 to 1,000 m long.
• Surveys. In the first winter (1996 to 1997), field crews walked 64 transects twice, in early winter and midwinter. In the next two winters they walked a subset of 28 low and mid-elevation transects once each in midwinter, when snow is deepest.
• Tracks. Crews recorded every deer track that crossed the transect line. White-tailed deer and mule deer tracks cannot be told apart reliably in the field, so they were combined as “deer”.
• Habitat. Every 50 m, crews measured slope, elevation, aspect, forest type, tree height and diameter, canopy closure, tree lichen, hiding cover, shrub cover, snow depth, and whether each shrub had been browsed by ungulates.
• Analysis. Each 50 m segment was scored for deer tracks present or absent and compared with its habitat. Because snow depth kept coming out as the best predictor, the author built a snow-depth model from mapped data (forest type, elevation, canopy closure, slope and aspect) and turned it into a map of potential midwinter deer habitat.

2. What they found

Tracks and distribution

• Crews recorded 41 to 136 deer tracks per survey period, or 0.17 to 0.57 tracks per 100 m.
• When snow was relatively shallow (56% and 60% of the long-term February average), tracks were common and spread widely: on 16 and 10 of the 28 transects.
• When snow was deep (138% and 113%), tracks were found on only 5 transects, clustered in one south-facing area.

Habitat use in midwinter

• In all three years, deer tracks were found where snow was shallower, elevation lower and slopes steeper than where there were no tracks. In 1997, for example, mean snow depth was 86 cm at track sites versus 135 cm elsewhere.
• More tracks than expected were found in Douglas-fir and ponderosa pine forest, and on southeast-facing slopes (in 1999, southwest slopes too). Mean track aspect was south to southeast.
• Tree height, tree diameter, canopy closure, shrub cover and lichen showed no consistent differences.
• In 1998, Saskatoon berry, redstem ceanothus, Douglas maple, Nootka rose, ocean spray, wild rose and ninebark were browsed more than expected; western hemlock and beaked hazelnut less than expected; white birch and willow in proportion to their availability.

Models

• Logistic regression picked out snow depth, elevation and slope in both years, plus browse species in 1998 and aspect and tree diameter in 1999. The author doubted these models were reliable because there were so few tracks.
• A snow-depth model using only mapped variables explained 58.8% of the variation in snow depth. Split at 92.7 cm (the average 90th percentile snow depth where tracks were found), it classified 77.6% of the first winter’s plots correctly as shallow or deep snow.

3. What they concluded

• Relative snow depth largely explained where deer spent the winter. In shallow-snow periods deer were spread widely; in deep-snow periods they gathered in one small area.
• Snow-track surveys show deer presence over a specific time, unlike spring pellet counts, which cannot separate early, mid and late winter. They also measure where the population is, rather than extrapolating from a few radio-collared animals, and snow depth can be measured on the spot together with the habitat.
• Limits: winter fieldwork takes a lot of resources, especially with few roads; the method only works where snow cover is complete and lasting; tracks show where deer travel, not necessarily where they feed or bed; and deer move less in deep snow, so deep-snow areas may have fewer tracks even with the same number of deer.
• The 50 m segments along continuous transects were not fully independent (possible pseudoreplication). Where access allows, many shorter transects, each within one habitat, would be better.

4. How tracking was or was not used

Tracking was the core method. Field crews walked straight-line transects and recorded each deer track crossing the line in snow, identifying them by the characteristics in Murie’s (1974) field guide. White-tailed and mule deer tracks were combined because they cannot be told apart reliably. Tracks were scored as present or absent in each 50 m segment rather than counted, because there had been no fresh snowfall during the surveys (so track counts could not be standardised to the time since snowfall) and there were too few tracks. Crews also checked shrubs in a 10 m radius at each plot for ungulate browsing. No trailing was reported.

5. Tracker or observer certification

Not reported. The tracks were recorded by field crews, four of whom are thanked by name; no tracking experience, training or test of their identifications is described. The whole analysis rests on crews detecting and identifying deer tracks correctly.

6. Other technologies, skills or innovations

• Habitat plots measured with a clinometer, compass, hip chain, spherical densiometer, cover pole and a graduated snow pole (three snow readings per plot).
• A snow-depth model built from 1:20,000 forest cover and digital elevation data, mapped as 25 m pixels in a Geographic Information System (GIS), producing shallow, deep and very deep snow zones.

7. Implications for conservation (from the paper)

• With enough data, snow-depth zones can be linked directly to deer survival and to forest management prescriptions.
• Snow-track surveys give managers a practical way to map deer winter range in dense forests where aerial surveys fail.