Estimating larger carnivore numbers from track counts and measurements

1. What they did

Knowing how many large carnivores live in an area matters for their conservation, but counting them directly is difficult, expensive and slow. The authors set out to compare three indirect ways of estimating carnivore numbers from tracks, because earlier work had called for several independent surveys to reach a reliable figure.

They worked on Shayamanzi Game Ranch, a 1500 hectare game ranch in the Waterberg region of Limpopo Province, South Africa. Over 6 months in 2001 they recorded all the fresh tracks of cats, hyaenas and members of the dog family that they could find, identifying them with the field guide by Stuart and Stuart (1998). Most tracks were on sand roads, the only substrate where tracks could be identified and measured, and the roads were not prepared beforehand. For each set of tracks, the same person recorded the location and measured the total length, total width and main pad height of the front and hind feet, as well as straddle and stride.

They then estimated numbers in three ways:

• Qualitative estimate. Individuals were told apart subjectively by comparing track measurements, before any statistics were run.
• Regression method (leopards only). They used the straight-line relationship between track density and true leopard density found in Namibia by Stander (1998), and multiplied the result by 1.3 because radio-tracking had shown leopard density in the Waterberg to be higher than in Stander’s study area.
• ANOVA method. A statistical test (repeated measures analysis of variance, ANOVA, with follow-up comparisons) on the foot measurements counted how many track sets differed significantly from each other. Stride and straddle were left out because they can change with slope and with what the animal is doing.

2. What they found

• They found and identified 283 tracks of six larger carnivores: leopard, caracal, serval, African wildcat, brown hyaena and black-backed jackal. Of these, 138 could be measured.
• Tracks of every species were found throughout the 6 months and all over the ranch.
• Caracal and African wildcat often placed the hind foot exactly over the front foot, so many of their tracks gave measurements for one foot only.
• Average track sizes matched those given by Stuart and Stuart (1998).
• Leopard track density was 4.7 tracks per 100 km of sand road, from about 300 km of road surveyed.
• The ANOVA found significant differences between track sets for all six species, so individual track sets could be separated.

Estimates (qualitative, regression, ANOVA)

• Leopard: 4; 3.2; 3
• Caracal: 6; not applicable; 5
• Serval: 3; not applicable; 3
• African wildcat: 5; not applicable; 5
• Brown hyaena: 13; not applicable; 12
• Black-backed jackal: 10; not applicable; 7

The regression figure for leopards (3.2 per 100 km²) is higher than the 1.9 per 100 km² found earlier in the Waterberg with radio-tracking. The authors suggest this may be due to less persecution and more prey since game ranching began, or to the difference in methods.

3. What they concluded

• All three methods gave similar results that seem reasonable given the animals’ social and spatial organisation, and may reflect the real number of larger carnivores living in or passing through the ranch.
• The estimates could not be checked against a direct count, which the authors call a major drawback. Sample sizes were small, and the ranch is smaller than the home ranges of all these species.
• Stander’s method needs only counting, not measuring, which saves time: measuring one complete track set took about 10 minutes. More studies should test it against direct counts and develop equations for other species and habitats. Its drawback is that individuals cannot be identified, unless experienced local trackers who can tell individual tracks apart without measuring them are available.
• The subjective approach also seemed fairly accurate, although it usually gave higher numbers than the statistical methods. This differs from earlier work on cougars, which found that subjective identification of individuals does not usually work.
• Future studies should record the substrate and slope where each track is measured.
• Overall, tracking data can give a reliable and cost-effective indirect estimate of larger carnivore numbers.

4. How tracking was or was not used

The whole study rests on tracks and signs identification and track measurement. Fresh tracks were found mainly on sand roads, identified to species with a field guide, and measured to tell individuals apart. Tracks were counted per 100 km of road for the leopard regression method. No trailing was reported, and tracking conditions were not assessed.

5. Tracker or observer certification

Not reported. The paper says only that the measurements were always taken by the same person, and that tracks were identified with a field guide. Nothing is said about that person’s tracking experience, which matters here because the qualitative estimate depends on a subjective judgement of which tracks belong to which individual.

6. Other technologies, skills or innovations

• Car odometer readings to estimate the distance of sand road surveyed.
• Statistical software (InStat V2.04, GraphPad Software) for the repeated measures ANOVA and Tukey-Kramer post-tests.

7. Implications for conservation (from the paper)

• Indirect estimates from tracking data appear reliable and cost-effective, which is promising for research and conservation of larger carnivores where direct counts are not possible for practical or financial reasons.
• All the larger carnivores expected in the area were present, which the authors take as a sign of good habitat for them. They point out the presence of the brown hyaena, whose role as a scavenger makes them an integral part of the ecosystem.