The challenge of monitoring elusive large carnivores: An accurate and cost-effective tool to identify and sex pumas (Puma concolor) from footprints

1. What they did

Pumas are solitary, elusive and mostly active at night, so they are hard to count. Capturing and collaring them is expensive and can harm the animals, and pumas have no spots or stripes for camera traps to tell individuals apart. The authors set out to show that the Footprint Identification Technique (FIT), first developed for black rhino, can tell individual pumas apart, and tell males from females, from photographs of their footprints.

• Collecting footprints. At 10 captive facilities across the United States, they raked builder’s sand into paths about 1 cm deep and 2 to 3 m wide and had pumas walk over them at a relaxed pace. They photographed left hind footprints only, from directly overhead, with a ruler and a photo slip in each picture. Only clear footprints were used.
• The reference set. 535 footprints from 35 pumas: 19 females (300 footprints) and 16 males (235 footprints), aged 1 to 20 years, in 2 to 3 trails per animal on average.
• Measuring. In the FIT software (an add-in to JMP statistics software), 25 landmark points were placed by hand on each footprint image and 15 more were derived from them, giving well over a hundred distances, angles and areas per footprint.
• Identifying individuals. Trails were compared two at a time. If the confidence ellipses around two trails overlapped, they were judged to be the same animal. A clustering step then grouped trails into individuals. Accuracy was tested by holding back some animals as “unknowns” and classifying them against the rest.
• Sexing. A separate discriminant analysis was built to classify single footprints as male or female.
• Free-ranging test. Footprints from two wild pumas, one in Big Bend National Park, Texas, and one in Los Padres National Forest, California, were run through both models.

2. What they found

Telling individuals apart

• When test sets were small relative to the training set (4 test animals against 31), accuracy was highest. Even with 28 test animals against only 7 in the training set, accuracy was 93.2%.
• In the clustering, 3 of 77 trails were placed with the wrong animal, giving 96.11% correct trail placement.
• The software’s likelihood slider gave 73.7% likelihood for 34 pumas and 76.9% for 36, when the true number was 35.
• The two wild pumas were correctly separated as two individuals, distinct from each other and from all captive animals.

Telling males from females

• About 20 measurements gave the best result. Sex was classified correctly 99% of the time from a single footprint.
• The wild puma from Los Padres National Forest was classified as male, and the one from Big Bend as female. The paper does not say whether these were confirmed.

What makes a good footprint

The clarity of the heel pad and toe outlines decides quality. Hind feet often land on the front footprints, hind footprints are longer, narrower and more symmetrical than fronts, the third toe is the longest, and the first toe does not show.

3. What they concluded

• FIT is an accurate, non-invasive and inexpensive way to identify and sex individual pumas from footprints, and could be used for other large cats.
• Footprints could be collected systematically in a small closed population for an absolute count, opportunistically over a larger area for a minimum known alive estimate, or used as the “mark” in capture-recapture estimates.
• FIT could be combined with camera traps, for example by preparing the ground in front of a camera so whole-animal images can be matched to footprints.
• Limitations: the method needs good impressions in a suitable substrate, and the person collecting must be able to recognise a puma left hind footprint. Where substrate is poor, sand “footprint traps” in travel corridors may help.
• Next steps: wider field validation with free-ranging pumas on different substrates, and on snow.

4. How tracking was or was not used

The whole method rests on tracks and signs identification. The authors say the photographer (or an assistant, who might be an expert tracker) must be able to recognise a puma left hind footprint on the ground and follow the trail of footprints forwards or backwards along the line of travel. They describe how to tell hind from front footprints and left from right. Footprints from free-ranging pumas were also collected, one trail per area, so each trail came from a different animal. The authors note that even experienced biologists can struggle to identify footprints, but in their experience most observers could identify puma left hind footprints after a day of training. Because the footprints are photographed, a skilled observer can check them later.

5. Tracker or observer certification

Not reported. The paper does not say who collected or identified the footprints, or test their skill. Several people are thanked for contributing puma footprint images.

6. Other technologies, skills or innovations

• FIT software as an add-in to JMP data visualization software, with a single graphical interface and built-in validation.
• A standard photography protocol: camera directly overhead, ruler along two edges, photo slip in the frame, footprint shaded to avoid partial shadows.
• Prepared sand paths for collecting reference footprints from captive animals.
• The authors’ earlier video guide to FIT using cheetahs is included as supporting material.

7. Implications for conservation (from the paper)

• Capture with dogs or snares can injure or kill pumas, and repeated capture, immobilization and collaring can harm health, behaviour and breeding. A non-invasive method avoids these costs and the risk that invasive methods bias the data.
• Footprints are the most common animal sign, and collecting them is simple and cheap, so expert trackers, conservation biologists and citizen scientists could all contribute.