Clouded leopards, the secretive top-carnivore of South-East Asian rainforests: their distribution, status and conservation needs in Sabah, Malaysia

1. What they did

Clouded leopards are the largest cat on Borneo, weighing 11 to 25 kg. They are secretive, partly nocturnal, live at low density in dense forest, and are very hard to count. The only earlier density figure for Sabah, 25 animals per 100 km², was a rough guess from three observations that later became a “quoted fact”. The authors wanted a reliable, low-cost way to count clouded leopards, and a first estimate of how many live in Sabah’s reserves.

• Track surveys. From March to August 2005 they walked eight transects in Tabin Wildlife Reserve, each 20 times: two along a gravel road, one along an old logging road, two along jungle trails and three along streams, about 35 km in total.
• Recording tracks. A track set was one or more connected footprints from the same clouded leopard. Each footprint in good condition, with clear edges and on flat ground, was photographed from directly above with a scale on two sides, and its location taken with the Global Positioning System (GPS).
• Telling individuals apart. On the photographs they took 14 length and 5 area measurements of each footprint, and used principal component analysis (PCA) to see which track sets grouped together, so that each group could be treated as one animal.
• Counting. They treated each new track set like a camera-trap “capture” and ran capture-recapture models to estimate how many animals were in the area, then divided by the area surveyed (each transect plus a buffer based on how far clouded leopards move) to get density.
• Scaling up. They applied that density to Sabah’s protected and commercial forest reserves, using reserve size, connectivity, protection status and whether clouded leopards had been recorded there.

2. What they found

Tracks

• Six usable track sets were found, with 1, 4, 8, 10, 13 and 14 footprints. All were on roads or on a mud volcano that many animals visit as a salt lick.
• None were found on the jungle trail transects, perhaps because leaves covered the ground. Eight track sets along streams were too poor to measure or to assign to clouded leopards for certain, and were left out.
• One single footprint was found right after a sighting of an adult male, so it could be assigned to him.
• In three large track sets the front left foot was always smaller than the front right. Rear left and right feet did not differ, so they could be combined.

Individuals and density

• The PCA grouped the track sets into four different clouded leopards. The three large sets with uneven front feet clustered together and were almost certainly one animal.
• The capture-recapture model estimated five animals (plus or minus 2.26) in the 56 km² research area.
• That gives about nine clouded leopards per 100 km², most likely between 8 and 17, far below the earlier figure of 25.

Sabah as a whole

• Clouded leopards are confirmed in about 25% of Sabah and may occur in about 37%.
• Only four totally protected reserves, covering 5% of Sabah, are large and connected enough (over 350 km²) for a long-term population of more than 50 animals. Together they may hold only 275 to 585 clouded leopards.
• Across all suitable reserves the very rough total is 1500 to 3200 animals, so most live in commercial forest reserves, where selective logging and licensed hunting are allowed.

3. What they concluded

• The earlier estimate of 25 per 100 km² most likely overestimated clouded leopard density.
• Identifying individuals from careful footprint measurements, combined with capture-recapture, is a feasible way to study even secretive cats in tropical rainforest, as long as the limits are respected.
• Limits of the method: it works only in small study areas with fewer than about 10 animals to separate; two animals can have very similar tracks, so it always gives a minimum count; and good tracks may be hard to find in dense vegetation. Substrate did not appear to bias the results, because track sizes varied more within a track set than between track sets.
• All the numbers are rough and meant as a first working hypothesis. The Sabah-wide figures probably overestimate numbers in logged forest, where densities are likely lower.
• Higher priority should go to sustainable management of commercial forest reserves.

4. How tracking was or was not used

Tracking was the core method. Observers searched roads, trails and streams for clouded leopard footprints, judged which were clear enough to use, and told front from rear and left from right feet. The footprints were photographed and measured to identify individual animals, and each track set became a “capture” in the population model. No trailing was reported. The authors note that clouded leopards like to travel along water courses, so stream beds and sand banks could be good places to find tracks elsewhere, although the stream tracks in this study were of poor quality.

5. Tracker or observer certification

Not reported. The paper does not say who found and identified the tracks, or describe their experience. A field assistant, Arthur, is thanked for his significant contribution, and co-author Soffian Abu Bakar of the Sabah Wildlife Department took part in data collection. The method relies on recognising good-quality clouded leopard footprints and telling each foot apart, but these judgements are not tested; individuals were separated by the measurements and PCA.

6. Other technologies, skills or innovations

• Digital photographs of footprints (4-megapixel camera on a monopod held at right angles to the track, an umbrella to even out the light, and a scale on two sides), measured afterwards in Adobe Acrobat 7.0 Professional.
• 14 length and 5 area measurements per footprint; angle measurements were dropped because they varied too much within one animal.
• PCA rather than discriminant analysis, because it does not need the number of animals to be known in advance.
• A new formula for the buffer around transects, combining core home range and daily movement, with a narrower 300 m buffer next to the oil palm plantations.
• GPS points every 250 m along transects and mapping in ArcGIS 9.1.

7. Implications for conservation (from the paper)

• Areas smaller than about 350 km² that are not connected to other forest are probably too small for a stable clouded leopard population.
• Two of the four core refuges, Maliau Basin and Danum Valley, may hold only 50 to 107 and 37 to 78 animals, just above the minimum viable population size.
• Commercial forest reserves, if protected and sustainably managed, could act as refuges and allow genetic exchange through dispersing animals. Harvest of natural resources there should be limited and controlled.
• Clouded leopards seem less able to adapt to people than leopards or mountain lions: they rarely take domestic animals and avoid settlements (unpublished data). Logging and illegal hunting of their prey may reduce their numbers.
• More surveys are needed across Sabah, and if the Bornean clouded leopard proves to be a separate species, its conservation becomes even more important.