Identification of individual tigers (Panthera tigris) from their pugmarks

1. What they did

In India, tigers have long been counted by tracing their pugmarks (footprints) and having experienced people pick out each individual. Critics say this is subjective and has never been tested on known wild tigers, since a pugmark changes with the soil and with the person tracing it. The authors set out to find an objective, measurable way to tell individual tigers apart from their tracks.

• Collecting pugmarks of known tigers. They gathered sets of hind-foot pugmarks from tigers whose identity was known, in four tiger reserves and parks and one zoo. A set was taken only when more than five pugmarks of the same tiger were found in a fresh trail, mostly where the tiger had walked in a normal gait, judged from a steady stride and footfall pattern. Different sets in one reserve were confirmed as different tigers mainly by seeing the tigers (10 tigers), or by distance and time apart.
• Recording. Pugmarks on dirt roads with 0.5 to 1.0 cm of fine soil were traced onto acetate sheets and photographed from a fixed height on a stand. Stride, straddle and step were measured 5 to 10 times for each set.
• Measuring. The tracings and photographs were scanned, and 93 measurements of each pugmark’s shape and size were taken on the computer.
• Choosing the best measurements. They kept measurements that varied little within one tiger but a lot between tigers, were the same on left and right feet, and were not closely correlated with each other.
• Testing. The 23 pugmark-sets from 19 tigers were split into 15 sets for building the model and eight for testing it. They also checked the effect of soil depth, of three different people tracing the same pugmarks, of tracings versus photographs, and of how many pugmarks a set needs. Finally they ran a mock population count to see whether the model could tell new tigers from ones already in the dataset.

2. What they found

• Eleven measurements did the job: nine from the pugmark (such as toe area, toe spacing and pugmark width) and two from the gait (stride and straddle). With these, the model correctly assigned all 15 model-building sets to the right tiger.
• Testing: in all five test runs, the test pugmarks were assigned to the correct tiger. On average a pugmark was assigned to its correct set with a probability of 0.92.
• Soil depth matters: pugmarks of the same tigress in very shallow soil (under 0.5 cm) and in deep soil (2 cm) were scattered and mixed with other tigers’ pugmarks. In soil 0.5 to 1.0 cm deep they formed a distinct group.
• Different tracers: pugmarks of two zoo tigers traced by each of the three authors still grouped correctly as two tigers.
• Tracings versus photographs: the analysis could not tell tracings and photographs of the same tiger apart; both were grouped as one tiger.
• How many pugmarks: accuracy rose with more pugmarks per set. Ten pugmarks gave 96.2% correct classification; 12 gave 100% for a sample of four sets.
• Mock population count: all eight test sets were correctly sorted, either as one of four new tigers or as a tiger already known. New tigers were picked out without ambiguity. When a set came from a tiger already in the data, its pugmarks mixed with that tiger’s earlier set, with an average combined probability of 0.71 (lower bound 0.64).
• Unique features: 11 of the 19 tigers had obvious individual quirks such as creases in the pads, irregular toe placement or distinctive toe shape. These were not used in the analysis.

3. What they concluded

• Tigers can be identified individually from their hind pugmarks and gait with high accuracy, though the dataset was small (23 sets from 19 tigers) and the study needs repeating with other tigers.
• Identification of individuals is the first step. Rather than total counts, which are only feasible in very small reserves, pugmark-sets could feed mark-recapture models for population estimates.
• Practical guidance from their data: record at least 10 pugmarks per set; collect from soil 0.5 to 1.0 cm deep (soil deeper than 1 cm was likely to give imprecise results); use a training set of at least five to eight known tigers of both sexes and different sizes; treat a new set as the same tiger when its combined mixing probability with an existing set is above 0.64. Photographs are preferred to tracings, to avoid poor data from untrained tracers.
• Accuracy may fall as more sets are compared, but the authors expect that comparisons would rarely exceed 35 sets.
• The method works only where the ground takes good pugmarks, as in central and western India.

4. How tracking was or was not used

The whole study rests on tracks and signs identification. Tiger pugmarks (footprints) were located in fresh trails on dirt roads, and the trail was read for consistent stride length and footfall pattern to check that the tiger was walking normally. Pugmarks were then traced and photographed, and stride, straddle and step were measured along the trail. Individual tigers were identified from measurements of the hind pugmarks plus gait. The paper does not describe following trails to find tigers; known tigers were confirmed mainly by sighting them.

5. Tracker or observer certification

Not reported. The paper does not say who found and recorded the pugmark-sets in the field. The three authors traced the same 28 pugmarks of two zoo tigers to test differences between tracers, and their tracings still sorted correctly, but their tracking experience is not described. The paper notes that the traditional census depends on “experienced persons” and recommends photographs over tracings to avoid poor data from untrained tracers.

6. Other technologies, skills or innovations

• Standard pugmark tracing on acetate sheets with an indelible ink pen, following the “tiger tracer” method introduced by S. R. Choudhury.
• A pugmark-photography stand for taking photographs from a fixed height.
• Scanning tracings and photographs (with a 5 cm scale line for calibration) and measuring them with Arc Info, Arc View and Sigma Scan Pro software.
• Combining footprint shape with gait measurements (stride and straddle) in one model.

7. Implications for conservation (from the paper)

• Pugmark monitoring is cost-effective, non-invasive, rapid and practical, and fits the traditional tiger census done by wildlife managers, so it is likely to be accepted and could fill a gap in objective tiger monitoring in central and western India.
• Because tigers use trails, pugmark-sets are fairly easy to collect, even in low-density areas with intensive searching, whereas camera traps sample only fixed sites and are costly and at risk of theft.
• The method could be used for other large carnivores (cats, dogs and bears), for studying the general ranging of individual tigers where radio-telemetry is not feasible, for sex ratios and population structure, and for identifying problem tigers and resolving conflicts.