Sex Determination of Amur Tigers (Panthera tigris altaica) From Footprints in Snow

1. What they did

Amur tigers once ranged widely across northeastern China, but only an estimated 20 remain there, spread thinly over a large area. China hopes tigers will recolonise from the larger population in the Russian Far East, but monitoring so few animals is hard. Camera traps only work in small areas, transect surveys can cover hundreds of kilometres without finding a single sign, and droppings for DNA are scarce. Footprints in snow are much easier to find. Russian censuses have long used the width of the front pad to tell males from females, but this cannot reliably separate adult females from young males.

The authors set out to build a simple, cheap way to tell a tiger’s sex from photographs of their footprints in snow, using the Footprint Identification Technique (FIT):

• Captive tigers of known sex. At a breeding centre, they photographed 523 left-hind footprints from 40 tigers (19 females, 21 males, aged 3 to 13). There was little natural snow that year, so they spread and sieved snow over the enclosures, and found that 3 to 5 cm (1 to 2 cm of firm snow under 2 to 3 cm of loose snow) gave the clearest prints. Keepers used food and calls to get the tigers walking normally.
• A photo protocol for snow. Each print was photographed from directly above, 30 to 50 cm high, with a folding ruler and a label. They solved new problems along the way: paper labels got wet, so they fixed a plastic card slot to the ruler and wrote on it with a dry-erase pen; snow is so bright that prints lack contrast, so they photographed early and late in the day and used a flashlight in overcast light.
• Measuring the prints. In JMP software they marked 25 points on each print and extracted 128 measurements (lengths, angles and areas). A stepwise method picked the 10 measurements that best separated the sexes.
• Testing the model. They checked accuracy by leaving out each footprint in turn (jackknife) and by five random 50/50 splits into training and test sets.
• Wild tigers. Villagers and local patrollers reported tiger trails, and the team went straight to each one to follow it and photograph several clear left-hind prints. They applied the model to 83 prints from 8 trails.

2. What they found

Captive tigers

• Accuracy levelled off at 8 to 10 measurements, so they used 10.
• With all the data, 11 of 523 footprints were misclassified (7 female prints called male, 4 male prints called female).
• The jackknife test misclassified 13 prints, an accuracy of 97.5%. In the 50/50 splits, test sets scored 97.4% to 98.5%.
• The authors think the few errors came from slightly distorted prints, perhaps from a sudden change of gait or ground.

Wild tigers

• The model classed 5 trails as female and 3 as male.
• Within each trail, 71% to 100% of prints agreed with the trail’s final sex; across all trails it was 92%.

Footprints in snow

• Clear prints were not hard to find. Tigers preferred level ground such as frozen rivers, roads, ridges and farmland.
• In deep snow (over 5 cm), loose snow hid the true outline and the print looked smaller, because the ruler sat further from the outline; sometimes they had to clear snow so the ruler and print were level. In very shallow snow (under 3 cm), the print held too little detail.
• Fresh snowfall was the most common cause of prints degrading in January and February, when prints were otherwise fairly stable. Prints made in March degraded faster because of warm weather and wind (citing Hayward et al. 2002).
• Field monitors were confident with the cameras after a short training period.

3. What they concluded

• Footprints in snow are the best way of detecting Amur tigers in northeastern China, and FIT gives a robust, non-invasive and cheap way to tell their sex, including the adult female versus young male problem that front pad width cannot solve.
• Their approach is more rigorous than earlier work: more footprints per animal and more animals, a wide age range with no age difference between the sexes, a strict photo quality protocol, measurements taken from photographs rather than tracings, and proper validation.
• The 10 measurements chosen can change depending on which data go into the model, as other researchers have also found.
• They recommend recording prints in snow about 3 to 5 cm deep, with the ruler level with the print outline.
• The method may help count resident females, which they see as an indicator of local recovery because females move less and disperse less far than males.
• They are now working on identifying individuals and ages from footprints, and hope Russian colleagues will adopt the same method for the shared cross-border population.

4. How tracking was or was not used

The whole study rests on footprints. Villagers and local patrollers reported trails, and the team then followed each trail in the snow and chose clear left-hind prints with the outlines of all 4 toes and the main pad visible. They numbered each unbroken trail separately (a broken or obscured trail got a new number) and recorded Global Positioning System (GPS) locations and habitat for each trail. Telling left-hind from other feet and judging which prints were clear enough is tracks and signs identification. The sex itself was then worked out by the computer model from the photographs, not by eye.

5. Tracker or observer certification

Not reported. The paper does not say who followed the trails or chose the prints, or how they were trained beyond a short training period with the cameras. The villagers and patrollers who reported trails are not described. This matters because the method depends on picking the right foot and the clearest prints.

6. Other technologies, skills or innovations

• A low-cost kit: a compact digital camera, a carpenter’s folding ruler and a label, with a home-made plastic card slot on the ruler and a dry-erase pen so labels survive the snow.
• A photo quality scale for snow prints (good, poorer but usable, and unusable) and tricks for low contrast: shooting early or late in the day and adding a flashlight in overcast weather.
• Sieved snow in captive enclosures to imitate natural snowfall when there was too little.
• The FIT add-in for JMP software, with 25 landmark points, 128 measurements and a validation platform, which the authors describe as one user-friendly program for image processing and analysis.

7. Implications for conservation (from the paper)

• With so few tigers in China, conventional methods struggle. FIT on snow footprints could be a valuable tool for monitoring the tigers’ recovery across whole landscapes, and the new Chinese tiger monitoring network needs a more accurate method.
• Footprint collection does not disturb the animals, and the authors cite evidence that non-invasive approaches give better scientific results.
• Local people in snowy regions, and citizen scientists, could take part in collecting the data.
• Counting resident females could show whether the local population is recovering.