Can the abundance of tigers be assessed from their signs?

1. What they did

Counting tigers directly with camera traps is accurate but slow and expensive, so it is rarely done across whole landscapes. Counts of tracks and signs are much cheaper, but they are often criticised because they are seldom checked against true numbers. The authors asked whether simple sign counts could reliably estimate how many tigers live in an area.

They used “double sampling”: at the same sites and at the same time, they measured both tiger density and tiger signs, then worked out how closely the two were related.

• Tiger density. At 21 sites in Central and North India (2006 to 2007), they placed one double-sided camera unit in every 2 × 2 km grid cell, for 20 to 96 days. Individual tigers were identified from their stripe patterns, and numbers were estimated with capture-recapture models. At two sites (Kuno and Phen) no tiger was photographed, so the team searched a larger area for pugmarks and scats and conservatively estimated one and two tigers.
• Tiger signs. A pilot study (2002 to 2004) showed that encounter rates of pugmarks and scats settled after a search of 4 to 5 km, so each survey was a 5 km search. Surveys followed places likely to have tiger sign, such as dirt roads, dry watercourses and animal trails. Three surveys were made in different parts of each forest beat, with the local beat guard on the team. Thirty biologists were trained for 15 days to find and identify tiger sign consistently, and every search path was walked by two observers, measured with a handheld Global Positioning System (GPS) unit.
• Calibration. They related tiger density to pugmark track sets and scats found per km, alone and together, weighting each site by how precise its density estimate was. They tested how well the models predicted density by leaving out each site in turn (a jackknife).
• Detecting change. They ran a power analysis to see how large a change in tiger numbers camera trapping and the sign indices could detect between two surveys.

2. What they found

Tiger density

• Camera-trap densities ranged from 0.55 tigers per 100 km² (Dhaulkhand, a single tigress photographed twice) to 19 per 100 km² (Corbett, where 102 individual tigers were photographed). Including the two sign-based estimates, the range was 0.25 to 19.
• Average search effort was 147 km per site.

Signs as a measure of tiger numbers

• Pugmark encounter rates alone explained 84% of the differences in tiger density between sites.
• Scats had a curved relationship with density, which the authors suggest is because scats last longer in the environment than pugmarks. Corbett had few scats for its number of tigers. Without Corbett, the square root of scat encounter rate explained 82% of the variation.
• Pugmarks and scats together explained about 95% of the variation, and this model predicted density very well (PRESS R² = 0.99). Densities predicted by the jackknife correlated closely with camera-trap densities (R = 0.91).
• Leaving out Kuno and Phen did not change the results.

Detecting change

• Camera trapping could detect a 12% change in density with 80% power.
• Pugmark rates could detect a 25% decline, and scat rates a 35% decline.
• The best sign model had 50% to 85% power to detect a 30% decline. It reached the authors’ target of 80% power only where there were more than 10 tigers per 100 km².

Cost

• Sign surveys cost about US$ 1240 and 220 man-days per site; camera trapping cost about US$ 17,000 and 720 man-days. Sign surveys therefore cost about 7% of camera trapping, and took about a third of the time.

3. What they concluded

• Sign indices, collected across varied habitats and a wide range of tiger densities, can reliably estimate tiger abundance once they have been calibrated against camera-trap estimates.
• The model should only be used when sign data are collected in the same way, and within the range of densities used to build it. It applies across about 300,000 km² of forest in Central and North India; similar relationships could be built for other regions and species.
• Sign indices are weak at detecting short-term declines, especially at low densities. Precision could be improved with spatially explicit capture-recapture models, more search effort, and trends over several surveys rather than comparing two.
• Source populations (reserves with breeding tigers) should be monitored every year by camera trapping, and all areas with tigers should be surveyed for signs every 2 years.
• The methods form part of the protocol used by the Indian Government to assess the status of tigers.

4. How tracking was or was not used

Tracks and signs were the core of the study. Teams walked 5 km search paths and recorded every tiger pugmark track set and scat, telling them apart from those of other carnivores using published criteria and field training. The number of signs per km walked became the index of tiger abundance. At Kuno and Phen, where cameras caught no tigers, pugmark track sets and scats from a wider search were used to estimate how many tigers were present. No trailing was reported.

5. Tracker or observer certification

Reported but not validated. A team of 30 biologists was trained for 15 days in identifying and searching for tiger sign, and for consistency between observers, and each path was walked by two observers with the local beat guard. No test of their identification skill or of agreement between observers is reported.

6. Other technologies, skills or innovations

• Double-sided camera units (Trailmaster and DeerCam), with individual tigers identified from their stripes.
• Handheld GPS (Garmin 72) to measure each search path.
• Use of India’s forest beats, and the beat guards’ knowledge of them, to spread sign surveys across large forests where a grid would be hard to lay out.
• Statistical power analysis to set a management standard: 80% power to detect a 30% change between two surveys.

7. Implications for conservation (from the paper)

• Calibrated sign surveys allow rapid, low-cost assessments of tiger abundance across whole landscapes, which matters for conservation investment, habitat management, planning development projects, policy and law enforcement.
• India’s official tiger count, which used experts to identify individual tigers from their pugmarks, had been strongly criticised and kept reporting tigers in Sariska and Panna after poaching had wiped them out there. A credible status assessment was needed.
• Low-density areas contribute few tigers, but may serve as dispersal corridors or hold breeding populations in future.
• A continuing programme combining camera trapping and sign surveys would greatly improve the ability to detect trends in tiger numbers.