To count the tigers hiding in India’s forests, researchers set out tens of thousands of hidden cameras and waited, and out of nearly 35 million photographs the machines took, the animal they were looking for appeared in only a tiny fraction, each tiger finally known not by a number but by the one thing no two of them share, the pattern of its stripes.

There are, on the current best count, about three thousand wild tigers left in India. Which is roughly seventy per cent of the entire global population of wild tigers still alive on Earth. Everything the species does or doesn’t do in the coming decades will, on any honest reading of the population geography, be decided largely by how well or badly things go in Indian forests. Which is why the Indian government has, since 2006, been conducting the largest recurring wildlife census anywhere in the world, at a scale that dwarfs any comparable effort by any other tiger range country.

What that looks like in practice is worth sitting with for a moment. Because tigers, being the ambush predators they are, spend most of their lives being deliberately invisible. They walk quietly. They rest in dense undergrowth. They hunt at dusk and dawn and after dark. They will spend weeks in a valley without any human visitor to that valley knowing they are there. And the sheer difficulty of getting a scientifically defensible count of an animal that has evolved for two million years to avoid being seen is what led the Indian government, along with the Wildlife Institute of India and the National Tiger Conservation Authority, to eventually stop trying to see the tigers directly and instead let the tigers count themselves.

How you set the trap for a tiger you never actually meet

According to Down to Earth magazine’s detailed reporting on the fourth All India Tiger Estimation, published in 2019 following the release of the census results and drawing on the methodology paper describing the 2018 field season, the 2018 iteration of the census involved a coordinated effort by roughly 44,000 forest staff across twenty tiger-occupied Indian states. The work took two years. Habitat surveys were conducted at 318,000 locations across 381,400 square kilometres, meaning most of the forested land in the country where tigers might plausibly still live. The scale is difficult to visualise. It’s larger than the entire land area of Japan. It required a distributed field operation of a size that no wildlife survey anywhere in the world had previously attempted.

Related: a short film from Space Daily on the invisible walls that shape elephant movement across their remaining ranges — another version of the same tension between what humans can see of a landscape and what the animal living in it actually perceives.

And that was the ground-based part, meaning the walking-around, looking-for-signs-of-tigers part. The actual counting was done by cameras. The census team placed 26,838 motion-triggered camera traps in the field, arranged in a systematic grid across 141 separate study sites. Each camera was designed to fire when a warm-bodied animal walked in front of its sensor. Each was set up in pairs, meaning that two cameras faced each other across the same trail so that any tiger passing through would have both its flanks photographed simultaneously. Each pair was placed roughly two square kilometres apart from the next, dense enough to catch any tiger moving through the sampled area but sparse enough that the total budget could cover the full landscape.

Then the team went home and left them running. For between twenty-five and forty-five days per location. Over the whole of 2018 and into 2019, without any human intervention beyond the periodic collection of memory cards, the cameras did their work.

The total number of photographs the cameras captured, on the final Guinness-verified tally, was 34,858,623. Of those, 76,651 contained tigers. Which is, in percentage terms, roughly two-tenths of one per cent. The other 34.8 million photographs were mostly of leopards, wild pigs, sambar deer, chital, langurs, mongoose, jungle cats, forest workers, occasional herders, occasional poachers, and the vast quantity of ordinary forest animals that walk in front of a motion-triggered camera in an Indian jungle over the course of a month. The tigers were, quantitatively, the rare event. Which is exactly what the census had been designed to catch.

What the pictures showed once someone actually looked

The seventy-six thousand tiger photographs then had to be sorted. According to reporting by the Indian news site Scroll drawing on direct interviews with scientists at the Wildlife Institute of India’s Tiger Cell, including former dean Yadvendradev Jhala and long-serving census scientist Qamar Qureshi, the sorting depended on a specific fact about tigers that turns out to have made the whole census possible in the first place. Every tiger’s coat pattern is different. The number, thickness, spacing, curvature, and arrangement of the black stripes on the orange fur of any given tiger are, on the accumulated genetic and developmental evidence, unique to that individual animal. No two tigers on Earth have ever had the same stripes. It’s the animal-kingdom equivalent of a human fingerprint.

Which meant that the 76,651 photographs could, at least in principle, be sorted into unique individual tigers by comparing the stripe patterns visible on the flanks and shoulders and faces of the animals in the images. The team wrote software to do most of it automatically. A tool called the Camera Trap Data Repository and Analysis Tool, known by the acronym CaTRAT, took each tiger photograph, extracted the visible stripe pattern, converted the pattern into a mathematical coordinate representation, and then compared the coordinate representation against a national database containing the stripe patterns of every previously identified Indian tiger. Where the system found a match, the photograph was assigned to a known individual. Where no match was found, a new individual was registered in the database, with its stripe pattern indexed for future comparison.

The 2018 census produced 2,461 individually identified tigers over the age of one year, catalogued in the national database not by numbers but by their stripe coordinates. Statistical modelling extrapolated from that catalogued population to give a total nationwide estimate of 2,967 tigers, plus or minus twelve per cent. The 2022 iteration, released three years later, added another five hundred to the number. The 2024 iteration, currently in the field, will most likely add more.

None of which is a particularly reassuring number in absolute terms. Three thousand wild tigers spread across a country of 1.4 billion people is a fragile population, and the census exists precisely because tracking the movement of that number, over the coming decades, is the only way anyone will know whether Indian tigers are recovering or quietly declining. What the 2018 census did establish, and what all subsequent iterations continue to establish on the same methodology, is that the counting itself is now possible at a scale that would have been unimaginable a generation ago. The cameras don’t miss the tigers. The stripe-matching software doesn’t confuse individuals. The numbers, whatever they end up being over the coming decades, will be numbers that can be trusted.

Which is a small piece of good news buried in an otherwise sobering conservation story. The tigers are still endangered. They’re still confined to a fraction of their historical range. They’re still under pressure from habitat loss, poaching, human-wildlife conflict, and the slow attritional collapse of the ecosystems they depend on. But at least, for the first time in the history of the species, someone knows exactly how many of them there are. And which specific one is which. And where each one lives. On the accumulated evidence of 34.86 million photographs and the specific stripe pattern of each tiger who was, briefly, caught in front of a camera she almost certainly never noticed, they can be counted, tracked, and, if the political will exists to do it, protected.