Tiger census: Role of quantification

1. What they did

In India, tiger numbers were estimated from tracings of pug marks (footprints). During a set period, foresters searched the study area and traced fresh pug marks. Two pug marks made on the same day were taken to be from different tigers if they were farther apart than a tiger normally travels in a day. Pug marks found close together were compared by eye to decide whether they could be from the same animal, and the tracing was also used to judge the animal’s sex. The census figures were often disputed: critics argued that prints of animals of the same sex and size cannot usually be told apart, and that the method had never been tested on a known population.

The authors, statisticians and ecologists working with Project Tiger at Melghat, set out to make the method more objective and repeatable:

• Measuring the tracing. Instead of judging shape by eye, they defined a set of measurements that can be taken by hand or with a digitiser: areas of the toes and pad, distances and angles between their centres, maximum widths and lengths (36 variables in all).
• Testing what affects the measurements. Six foresters traced the same three pug marks from one trail at Melghat. In a second pilot, prints of two animals were traced in fine soil, sand and wet mud.
• Telling the sexes apart. Two male and three female tigers were led to walk on specially prepared ground, and around 100 of their prints were traced and used to build a sex identification rule.
• Comparing two reserves. They used 90 tracings from Kanha and 90 from Melghat, half to build a rule and half to test it.
• Counting tigers. They outlined a statistical method for estimating how many individual tigers left a set of pug marks.

2. What they found

What affects the measurements

• With six people tracing the same prints, 28 of the 36 variables showed no significant difference between the people tracing or between the prints. Six variables differed significantly between people, and two differed between prints.
• Across the three substrates, 32 variables were unaffected and four showed a significant effect.
• Together, 10 variables were sensitive to one of these factors, leaving 26 of 36 that seem potentially useful for telling animals apart.

Sex from pug marks

• Most measurements were of little use for telling males from females. Three were useful: the distances from the pad centre to the centres of the first and second toes (indicating the print’s length), and the distance between the centres of the first and fourth toes (indicating its breadth).
• The rule: prints with a probability of being female below 0.25 were classed as male, above 0.75 as female, and in between were left unclassified.
• Of 30 male prints, 23 were classed correctly, 1 as female and 6 left unclassified. Of 67 female prints, 59 were classed correctly, 7 as male and 1 left unclassified.

Two reserves

• A rule built on 10 distance variables placed prints in the right reserve about 90% of the time: 40 of 45 Kanha prints and 41 of 45 Melghat prints.

Counting tigers

• The proposed method starts by testing whether the n pug marks could come from n minus 1 tigers, then n minus 2, and so on, until that idea is rejected. Prints from one tiger should vary little among themselves. The number of ways to group the prints can be huge, so a large enough sample of groupings could be used instead of checking them all.

3. What they concluded

• The size and shape of a tiger pug mark can be quantified and measured objectively.
• These measurements have some potential for telling the sex of an animal; the three length and breadth distances above seem to play the major role.
• Most of the measurements seem robust: substrate and the person tracing may not affect them significantly.
• Prints from Kanha and Melghat appear to differ significantly, which the authors suggest may be a new dimension revealed by this analysis.
• The findings seem promising enough to warrant a much larger study. In particular, a substantial database of prints of known animals is needed to set up numerical decision rules and test them.
• Using these and similar techniques to identify individual animals needs to be explored.

4. How tracking was or was not used

The whole study is about tiger pug marks. Foresters searched for fresh prints and traced them, and the tracings were measured. The authors tested how much the person tracing, the substrate and repeat prints in a trail change the measurements, and whether prints reveal sex, home reserve and individual identity. No trailing is reported, apart from three prints taken “from a trail (i.e. the same animal)”.

5. Tracker or observer certification

Not reported. Six foresters traced the test prints, but nothing is said about their experience or training. The study measured how much their tracings differed from each other: six of 36 measurements differed significantly between them.

6. Other technologies, skills or innovations

• Measuring pug mark tracings by hand or with a digitiser, using simple programs that the authors say can easily automate the measurements.
• A logistic regression formula that gives the probability that a print is from a female, with an “unclassified” zone for uncertain prints.
• A discriminant function to tell which reserve a print came from.
• An outline algorithm for estimating the number of tigers from a set of measured prints.

7. Implications for conservation (from the paper)

• Tiger census figures based on pug marks are contested, so a procedure that is objective and based on sound statistics is needed.
• Before the methods can be recommended for field use, a large database of prints from animals of known identity is needed to build and test decision rules.