Design and Analysis of Mountain Lion Track Surveys

1. What they did

In the mid-1980s mountain lions in California were a protected nongame animal, the ban on sport hunting was about to end (January 1986), and people disagreed over how many lions there were. Track surveys are cheaper than radio-telemetry studies and suit secretive animals that live at low densities. In 1984 the authors began mountain lion track surveys in California to build on earlier surveys and see whether they could show local population trends.

This paper sets out how to design and run a track survey in a limited area, and how to analyse the results, as a first step towards standard methods so that future surveys can be compared. It covers:

• Choosing the study area and route, with sampling principles and the arrangement of lion home ranges in mind.
• What to count: the number of individual lions or the number of track sets.
• Vehicles, training and timing of the survey.
• Field procedures: riding the route, rating tracking conditions, tracing, measuring and photographing tracks, and marking them.
• Tracking techniques for finding tracks.
• Data analysis: telling individual lions apart, using judgement, how sensitive the index is, which comparisons are valid, and which statistical tests to use.

They draw on their own survey in the Sierra National Forest, where five lions were being followed by radio, and on earlier work by Koford (1978), Kutilek et al. (1981) and Van Dyke (1983).

2. What they found

The survey route

• A route should be 64 to 96 km of dusty dirt roads (or snow), surveyed again periodically. Gravel and paved roads are unsuitable. Roads closed to traffic do not work either: some traffic is needed to make dust, and grass grows on unused roads. Logging traffic, grading, watering and dust suppression destroy tracks, as does any frequent traffic.
• Lions probably move through their home ranges in a non-random way, and roads are not built at random, so random sampling is usually not possible. It is not needed to compare one year with the next on the same route.
• Where roads cross or switchback, those areas are sampled more than their share. To describe the whole area, aim for proportional sampling.
• To find the most tracks, choose roads along ridges or rivers or crossing saddles. To find the most different lions, a long, spread-out route crossing many home ranges works better than a dense network of roads in a small area.
• Van Dyke (1983) needed 195 km of transect to detect all the lions in his Utah study area under fair tracking conditions. In theory, five days at 64 km a day should find most lions, but near Fresno a 77 km route surveyed for four days failed to detect five radio-collared lions whose home ranges lay mostly within the route area.

Track sets versus individual lions

• The authors propose a standard definition of a track set: a line of tracks made continuously by the same lion, physically connected. Short gaps where a lion leaves the road and comes back count as one set if the observer is certain. Tracks further apart are separate sets, even if they look alike or even if a peculiar feature proves they are the same lion, so that easily recognised animals do not bias the data. Tracks made on different days, or going in different directions on the same stretch, are separate sets. A maze of indistinguishable tracks made the same night at one place is one set.
• Surveys usually detect only 0 to 4 individual lions, too few to show a fall in numbers. Track sets give larger numbers, but may vary with the behaviour of different lions, and whether that extra variation cancels the advantage is unknown.

Vehicles, training and timing

• Off-road motorcycles (125 to 175 cc) are economical and easy to manoeuvre, let the rider circle back with little disturbance to the tracking surface, and give a good view of tracks. All-terrain vehicles (ATVs) are illegal on public roads, unstable and tiring, their wide tyres can wipe out tracks before the rider stops, and new lion tracks are hard to see over their tyre marks; four-wheeled ATVs may be better on mud or snow. Pickup trucks cost the most, need two people each, and keep observers further from the tracks. For safety the observer should ride in the cab, not on the hood.
• Riders need a licence, at least eight hours of dirt-road riding, and training in identifying the tracks of all species likely on the route, using field guides. The authors point out an error in one guide: it shows the lion’s hind foot landing behind the front print, but the hind foot usually lands on top of or in front of it. A field trip with a professional lion trapper helps. All riders should practise tracing together (dog tracks will do) to reduce differences between tracers.
• Survey in the dry season, before hunting and hound training seasons, because hunters’ vehicles destroy tracks, hunters may change lion movements, and dog tracks make lion tracks harder to find.

Field procedures

• Start shortly after sunrise. Two observers each ride about half the route at 4.8 to 8.0 km per hour, swapping halves each day so each checks the other’s identifications.
• Kutilek et al. (1981) found new lions on the fourth or fifth day in four of eight surveys. In the authors’ survey, only one of five radio-tracked lions entered the part of its home range covered by the route in five days, and it was not detected. Leaving one or more days for tracks to build up between survey days may find more lions; traffic often limits this to one or two days.
• Rate tracking conditions with Van Dyke’s (1983) method: walk 10 steps across the road and score each boot print from 1 (not visible) to 4 (complete outline and all details of the sole).
• Trace every track set, even if it seems to be from a lion already found, following Panwar (1979): lay a thread along the line of travel, place plate glass as close over the track as possible (not Plexiglas, which scratches and builds static), and trace on transparent film with a fine permanent pen, keeping your eye directly above the track to avoid parallax error. Solid lines show clear edges, dashed lines unclear edges. Trace several prints of each foot.
• Measure heel pad width and length on the track itself, not the tracing, after lifting the glass. Photograph each set, record the location, habitat, soil surface, road condition, depth of surface layer and other details, and mark the tracks so the other rider can check them and not record them again.
• If logging, rain or other disturbances are excessive, abandon the route, because trend studies assume an equal chance of seeing tracks each year.

Finding tracks

• The best light on clear days is from shortly after sunrise to mid-morning, when the low sun casts shadows in the tracks. Late afternoon light is poorer because of dust and glare; at midday tracks are hard to see.
• Shadows show best when the track is between the observer and the sun, so riders sometimes ride on the shoulder, look back over their shoulder, or weave across the road.
• Other cues are flattening (compressed soil with a different texture) and colour change (darker, moister soil exposed, or dew settling differently).
• Lions sometimes cut corners, so tracks are often on the insides of bends. Check ridge crests, drainage bottoms and saddles. Slide marks on road-cut banks may show where a lion crossed.

Telling individual lions apart

• No standard method exists yet, but tracks can be compared by shape, size, angles, patterns, gait (from stride measurements or prints that consistently stray from the line of travel), the angle of the paw to the line of travel, heel pad shape and size, toe placement relative to the heel pad, and on-site evidence (for example, one track set crossing another).
• Van Dyke (1983) estimated that a lion crosses at least one unimproved dirt road in their home range every 5 to 12 days, though this depends on road density.
• A decision matrix, marking each pair of track sets as “equal”, “not equal” or “questionable”, helps group track sets into individual lions.
• Two approaches: a conservative one (used by Kutilek et al. 1981) that separates lions only when tracings are obviously different, which is repeatable but gives a “minimum” estimate; and a judgement-based one, in which trained, objective researchers weigh all the evidence. The authors’ opinion is that heel pad width from different tracks of the same lion, traced by different trained trackers, should not vary by more than about 6 mm.
• In the Sierra National Forest, five radio-collared lions had home ranges overlapping at one point in the route area. Four stayed outside the route area during the five days, and the fifth was present but not detected. The team found eight track sets: two lions by the conservative method and three by their best judgement, from two good tracking days (rain stopped work on one day and spoiled tracking on two others), while five other lions were known to use the area.

Analysis and comparisons

• Counts per kilometre assume that the chance of finding different lions depends on the number of home ranges the route crosses, but home range shapes and positions are usually unknown and may change over time.
• Van Dyke (1983) stated that track frequency could compare relative populations in different areas. The authors disagree: without more testing, comparing different areas is invalid, and even comparing one route between years assumes an unproven equal chance of detecting the lions present.
• Each survey period (five days in their work) is one sample, and several years on one route can be analysed as paired samples with nonparametric tests.

3. What they concluded

• Track surveys need standard methods so results can be compared. Different goals (most tracks, most lions, one known lion) call for different route designs.
• The number of track sets may be more useful than the number of lions for statistical comparisons, though this is untested.
• Even the judgement-based count of individual lions is probably an underestimate, because not all lions’ tracks can be told apart and not every lion whose home range crosses the route will leave tracks on it during the survey.
• With the low numbers usually found, the index may not detect a drop in the population.
• Results from different areas should not be compared, because routes and home range patterns give different chances of sampling lions.
• Combining radio-tracking data with intensive track surveys could probably estimate local population size.
• Research needs: an objective model of observer error in tracings and natural variation in lion tracks in different soils (the authors had started a pilot study), and track surveys in areas where lions are being radio-tracked, to measure how variable the surveys are.

4. How tracking was or was not used

Tracking is the whole subject of the paper. Observers search dirt roads for mountain lion tracks from slow-moving motorcycles, stopping or circling back to check faint marks. They identify the tracks, trace them on plate glass and film, measure the heel pad, photograph them and rate the tracking conditions. Track sets are followed along the road to see where they begin and end and whether gaps are short enough to count as one set. Individual lions are then told apart from track features, gait and on-site evidence. The paper also describes how to read tracks in the field: using low sunlight and shadow, flattening and colour change of the soil, and likely lion travel routes.

5. Tracker or observer certification

Not reported. The paper recommends training riders in track identification and tracing, practising tracing together, and if possible a field trip with a professional lion trapper, and it calls observers “trained trackers”. It does not say who did the authors’ own surveys or how their skill was tested. The authors note that observer differences in tracings have not been measured and call for research on them.

6. Other technologies, skills or innovations

• Off-road motorcycles (125 to 175 cc) fitted with trail tyres, odometer, spark arrestor and luggage rack, with a clear plastic map holder taped to the tank.
• Track tracing on 4.8 mm plate glass with transparent film printed as a field form, a thread to mark the line of travel, and fine permanent-ink pens.
• A draftsman’s compass for measuring heel pads.
• Van Dyke’s (1983) boot-print scoring of tracking conditions.
• A decision matrix for grouping track sets into individual lions.
• Radio telemetry of collared lions as a check on what the track survey missed.

7. Implications for conservation (from the paper)

• Track surveys may be especially useful for large, secretive predators at low densities, and are cheaper than radio-telemetry studies for following population trends.
• Because counts of individual lions are low, a track index may not show whether a population has fallen significantly.
• Comparing track survey results between different areas is likely invalid without more testing.