Snow leopards: turning scattered sightings into a population estimate

Snow leopards are difficult to count. They live at low densities across rugged mountain terrain, move over large areas and are rarely seen directly. In Pakistan, older estimates often came from tracks, interviews, habitat assumptions or expert judgement. A 2026 paper in Ecography offered a more formal baseline: 155 animals, with a 95% confidence interval of 100 to 239, extrapolated across suitable habitat in the country.

The number is based on a decade of camera-trap work from 2010 to 2019, not a survey conducted in 2026. Researchers combined 19 surveys across the Hindu Kush, Himalaya and Pamir-Karakoram ranges. The cameras produced thousands of photographs, and coat patterns allowed the team to identify individuals. Spatial capture-recapture models then used where and how often animals appeared to estimate density and abundance.

From photographs to individual histories

Snow leopards have individually distinctive spot and rosette patterns, so camera photographs can help researchers recognise the same animal at different stations. A record of repeated detections contains more information than a simple number of photographs. It can show how often an identified cat was encountered and how detections relate to camera locations.

The study assembled 828 cameras and 26,540 camera-days across its survey network, producing 4,712 snow leopard photographs from 65 locations. Fifty-three distinct animals were identified in the combined encounter data; many were photographed only once. The model used these records to estimate movement and density, then project across suitable habitat. Because a cat can travel far between cameras, the spacing and coverage of the network matter.

The predicted density was not uniform. The Karakoram had the largest modelled share, while the Hindu Kush and Himalaya estimates were smaller. The study also associated density with elevation and distance from glaciers. These patterns are useful for planning surveys and habitat protection, but they are predictions from the model, not a map of exact territories occupied by known cats.

Why the interval is part of the result

The estimate is reported as 155, with a 95 per cent interval from 100 to 239. That range is not a flaw to hide. It expresses uncertainty from sampling and modelling. A point estimate is a convenient summary, but repeating “155 snow leopards in Pakistan” without the interval can make a probabilistic result sound like an exact census.

The paper’s authors also explain that their estimate averages a decade of camera-trap effort. It cannot tell us whether the number rose or fell between 2010 and 2019, or what the population is doing now. A newer estimate would require repeat surveys with comparable spatial coverage and methods, ideally supported by genetic sampling. Monitoring has to be repeated before trend can be separated from sampling variation.

Why previous estimates were less certain

Past national figures for Pakistan ranged roughly from 250 to 400, but they drew on limited areas or indirect signs and often extended local observations across much wider habitat. Surveying promising valleys alone risks a high estimate because those places may hold more animals than average. Conversely, inaccessible terrain can remain under-sampled. The new work tried to include both stronger and weaker habitat and made its modelling assumptions explicit.

Even a more systematic estimate has limits. The study extrapolates from sampled areas to suitable habitat that was not directly surveyed. It assumes that relationships between density, elevation and glaciers extend into those areas. The authors tested alternative models and described this extrapolation openly. The result is a defensible baseline, not certainty about every mountain range.

Counting cats is only part of conserving them

A robust estimate can help governments decide where to focus monitoring and how to compare future surveys. It does not itself protect snow leopards. The species still faces loss of wild prey, retaliatory killing after livestock losses, illegal trade and changes to mountain habitat. Climate shifts and roads or other infrastructure can add pressure, although their local effects need to be assessed rather than assumed to be identical across the range.

The Global Snow Leopard and Ecosystem Protection Program’s PAWS initiative works with the range countries to improve population assessment. Shared methods matter because national estimates must eventually fit into a wider picture. A new Pakistan result is valuable precisely because it says what it covers and what it leaves uncertain.

The Indian vulture survey also illustrates why method belongs beside a wildlife number: its nest-based estimate describes breeding adults at surveyed sites, not every bird. For snow leopards, the relevant evidence is camera-trap detections and spatial modelling. Different species call for different survey designs, but every estimate becomes more useful when its date, area and uncertainty travel with it.