A Rigorous Technique for Identifying Individual Mountain Lions Felis concolor by Their Tracks

1. What they did

Mountain lions are sparse, secretive and hard to study, and managers wanted better ways to follow their numbers. Track surveys were already used to follow population trends, but nobody had tested whether individual mountain lions could be told apart by their tracks. The authors set out to build a measurable, objective way of doing this.

• A test of the eye. At the Third Mountain Lion Workshop in Arizona (December 1988), they showed 52 participants six left rear tracks, three from one mountain lion and three from three others, and asked them to pick the three from the same animal.
• Collecting tracks. In summer and fall 1987, K. Shawn Smallwood searched dusty roads in three mountain areas of northern California. When he found mountain lion tracks, he followed the track set along the road and traced each track outline onto acetate sheets (about 5 minutes per track). He traced two to six tracks from each rear foot of nine mountain lions, on different slopes and dust depths. To avoid recording one animal twice, each track set was at least 17.6 km in a straight line from any other, and all tracks in the Georgetown Divide area were recorded within one day, and in the Mendocino National Forest within two days.
• Measuring. On each tracing they took ten simple measurements of the toes and heel pad, such as heel width, the spread of the outer toes and the width of the heel pad along a midline.
• Measuring error. Smallwood traced one track eight times from different angles and postures, and six different tracks from the same foot on level ground, uphill and downhill, to separate tracing error from the effect of slope.
• Analysis. They used multiple-group discriminant analysis to see how well the measurements sorted each track to the right animal, with separate models for the left and right rear feet. They then repeated the analysis 72 times on random subsets of the lions (bootstrapping) to see which measurements were consistently useful.

2. What they found

People matching tracks by eye

• None of the 52 respondents picked all three tracks from the same mountain lion, and 75% picked tracks from three different animals.
• Only 23% picked two of the three. This did not change with tracking experience: 26% of those with mountain lion tracking experience and 27% of those without.

Recording error and slope

• Changes in slope caused about twice as much variation as tracing error: on average 193% of the recording error for the length measurements, and 210% for the angle between the toes.
• Slope affected the angle between the toes, the outer toes spread and heel width most.
• The middle lobe width was so hard to trace consistently that its tracing error was larger than all the variation between tracks, so the authors caution against using it.

Telling the lions apart

• No single measurement was enough: on its own, each sorted only 21% to 52% of tracks correctly, because values overlapped a great deal between animals.
• Combined, the measurements worked. With four measurements from the left rear foot, or six from the right, all nine lions were identified by more than 75% of their tracks.
• With eight measurements, 100% of the left rear tracks and 92% of the right rear tracks were grouped with the right animal.
• In the 72 bootstrap analyses, success ranged from 77.8% to 100%, and 40 of them were 100% correct.
• The most consistently useful measurements were heel width, midline width of the heel pad and the spread of the outer toes.
• Front tracks looked more distinctive between individuals than rear tracks, but too few were recorded to analyse (11 left front tracks from five lions and 10 right front tracks from four).

3. What they concluded

• Individual mountain lions can be identified by measuring their tracks, even with few tracks per animal and tracks from a wide range of slopes and dust conditions. A purely visual comparison usually will not work.
• More research on recording error and track variation is needed. Discrimination may drop with larger samples where more track sets look alike.
• Practical advice for a study site: set aside track sets already known to differ (for example by location); trace or measure at least four tracks per track set; try different measurements; and record tracks from the same angle and posture. Recording error will increase if more than one person traces tracks.
• They describe a step-by-step way of comparing new track sets two at a time, suggesting 75% correct grouping as a cut-off for small samples (fewer than 6 tracks) and 65% for larger ones. A new track set is more convincingly assigned with ten tracks than with three.
• They do not recommend using the method on tracks in mud that later dried, deep snow or loose sand, because their size and shape change over time. Fresh, clear tracks in mud or light snow may be usable.
• With more research, the method may suit other large animals with complex tracks (large cats, dogs, bears, deer, rhinoceroses), possibly with standardized photography or plaster casts for smaller or deep tracks.

4. How tracking was or was not used

Tracking was the basis of the study. Smallwood searched dusty roads for mountain lion tracks, followed each track set along the road so that only tracks from one animal were recorded, and traced individual footprints onto acetate for measurement. The workshop test also asked people to match tracks to individuals by eye.

5. Tracker or observer certification

Not reported. All tracks were found, followed and traced by K. Shawn Smallwood; his tracking experience is not described. The workshop respondents’ experience was self-reported (39 were biologists, 4 houndsmen, 31 had mountain lion tracking experience) and not validated beyond the test itself.

6. Other technologies, skills or innovations

• Acetate track tracings, measured for ten features of the toes and heel pad.
• Multiple-group discriminant analysis with stepwise selection of measurements, and bootstrap analyses to test which measurements were consistently useful.
• A method for separating tracing error from the effect of slope on track shape.

7. Implications for conservation (from the paper)

• Identifying individuals by their tracks could improve regional population estimates, especially when track sets on a study site are calibrated against another method such as radio-telemetry and then extended to other areas with track counts.
• Capture-recapture analysis could be done without disturbing animals by capturing and marking them.
• Animals visiting several scent stations could be detected, improving those surveys, and individual movements could be better monitored.
• The method may also help in police work when tracks need to be identified.