A Snow-tracking Protocol Used to Delineate Local Lynx, Lynx canadensis, Distributions

1. What they did

Canada lynx are listed as threatened in the contiguous United States, so land managers need to know where they live. Hair-snagging pads can confirm lynx with DNA, but they cannot tell a resident population from a single animal passing through. The authors set out to build a snow-track survey for lynx that covers the whole landscape fairly and deals with the risk of misidentified tracks.

• Computer simulation. They back-tracked 15 radio-collared lynx and mapped their tracks with a Global Positioning System (GPS) unit. From these, they measured how winding (tortuous) real lynx tracks are. They then simulated lynx travelling 1 to 4 km a day inside a 100 km² home range, with an 8 km survey transect running through it, to see how likely a survey would be to cross a track after 1, 3 or 7 days since the last snowfall, and with more visits.
• The survey protocol. Using a Geographic Information System (GIS), they laid an 8 x 8 km grid over each study area. Units in moist subalpine fir forest (which lynx use most) were high priority and surveyed at least twice a winter; drier, open forest was low priority and surveyed once; open country and high alpine areas were not surveyed. Observers searched 10 km per unit, mostly by snowmobile at 15 to 20 km per hour on roads and trails, or on snowshoes, choosing routes through forest with dense cover (good snowshoe hare habitat).
• Recording and confirming tracks. Every lynx track (and other carnivore tracks) was located by GPS. Rare carnivore tracks were measured (stride, straddle, length, width, depth) and photographed. In Wyoming, they followed presumed lynx tracks to look for scats and daybeds, and picked hairs frozen into the beds for DNA testing.
• Testing against telemetry. In the Garnet Range, Montana, where they believed most lynx were radio-collared, they compared the area where tracks were found with the lynx distribution from radio telemetry (a 95% kernel home range of all locations). They also surveyed the Wyoming and Salt River ranges, where lynx were thought to be present but scarce.

2. What they found

Simulations

• With one visit, the chance of detecting lynx tracks rose from about 0.23 one day after snowfall to 0.78 seven days after.
• With three visits, it rose from 0.58 one day after snowfall to about 0.95 seven days after.
• Two visits found tracks about 80% of the time once more than 3 days had passed since snowfall.

Garnet Range, Montana

• They surveyed 242 km of roads and trails in winter 2001 to 2002 and 438 km in 2002 to 2003.
• They recorded 37 lynx track detections in 4 of 12 survey units the first winter, and 37 in 4 of 16 units the second.
• They saw a further 115 and 53 lynx tracks outside the formal surveys, during trapping in areas of high lynx use.
• Lynx tracks were at an average elevation of 1839 m (range 1443 to 2012 m).
• Seven lynx were collared (five in the first winter, two in the second), which the authors believed was a local census, though they could not test this.
• The telemetry home range (96 locations from 7 lynx) contained 97% of the tracks found on the surveys. The two methods gave the same general distribution.

Wyoming and Salt River ranges

• They surveyed 1055 km in 32 units (1999 to 2000), 1103 km in 37 units (2000 to 2001) and 1080 km in 34 units (2001 to 2002).
• They found two lynx tracks in the first winter, five in the second, and none in the third. Wyoming Game and Fish Department biologists also saw tracks of a family group.
• Two tracks from the first winter looked like lynx but were poorly defined: wind had scoured the snow, leaving each print as a raised pedestal. The team followed them for about 200 m to daybeds and collected five hairs; two had enough DNA, which confirmed lynx.
• Five of the seven detections fell in the same area of the northeastern Wyoming Range where two lynx trapped by state biologists had spent the winter of 1996 to 1997.

3. What they concluded

• Snow-track surveys spread fairly across all available habitat can at least coarsely map the local winter distribution of lynx, not just presence or absence.
• In Wyoming, local managers believed lynx were spread through both ranges, but the surveys showed they were rare and confined to a small area in the north of the Wyoming Range.
• Where lynx are rare, a single misidentified track can have serious management consequences. The authors suggest treating the track as a “collection device” for genetic samples rather than the main means of identification, and recommend genetic confirmation whenever correct identification of a set of tracks is critical.
• Surveys are usually done a day or two after fresh snow, when tracks are clearest, but detection is much higher several days later. Confirming older tracks with DNA lets surveys run when detection is high, even in poor tracking conditions.
• The key elements of their surveys were: sample units spread across the landscape but focused on good habitat; a defined, consistent search effort per unit; search intensity matched to the simulated detection rates; well-trained technicians; carefully documented, GPS-referenced track identifications; and searching lynx tracks for genetic samples.
• The method is labour intensive. For whole regions or states, the authors suggest adaptive cluster sampling.

4. How tracking was or was not used

Tracking is the core of the study. Observers searched for lynx tracks in snow from snowmobiles and on snowshoes, recorded each one by GPS, and measured and photographed rare carnivore tracks. Crossings of the same animal less than 100 m apart counted as one detection. The researchers back-tracked radio-collared lynx to map real travel routes for the simulation, and in Wyoming they trailed presumed lynx tracks about 200 m to daybeds to collect hairs for DNA confirmation of poorly defined, wind-scoured tracks.

5. Tracker or observer certification

Reported but not validated. Observers were trained by back-tracking radio-collared lynx in varied forest, snow and terrain at the study area near Seeley Lake, Montana, and after this training the authors “assumed that all observers could recognize lynx tracks”. Their skill was not tested. The authors list well-trained technicians as a key element of the protocol, and they relied on DNA, not observer skill, to confirm the rare Wyoming tracks.

6. Other technologies, skills or innovations

• Trimble GEO Explorer III GPS units, used to map back-tracked lynx routes (a point every 2 seconds) and to record track locations.
• Computer simulation of lynx travel based on real track tortuosity.
• An 8 x 8 km survey grid drawn in GIS, with habitat-based priorities.
• Collecting hairs with tweezers from daybeds after clearing loose snow down to the frozen bed layer; DNA analysis by the Rocky Mountain Research Station’s Wildlife Genetics Laboratory.
• Box traps with carrion, beaver castor and visual lures (pie plates and wings); very high frequency (VHF) radio collars; mostly aerial telemetry.

7. Implications for conservation (from the paper)

• Managers need reliable distribution data for lynx at the southern edge of their range, where they are threatened.
• False positives can overstate where a rare species lives, and misidentification can have serious management impacts for species listed under the U.S. Endangered Species Act; genetic confirmation guards against this.
• Snow-track surveys can pick up changes in lynx occupancy quickly across large areas, and even a coarse map of local distribution is useful to habitat managers.