Camera traps count six predator species in South Africa

Africa’s large carnivores do not occupy the landscape in equal ways. Until recently, nobody had counted all of them simultaneously using one camera-trap grid.

Lions range over the broadest areas, while leopards follow distinct corridors. By contrast, the cat-sized genet may spend its entire life in a territory only a small fraction as large.

A research group in South Africa designed a camera grid to address that challenge. Their work has produced the first publicly available dataset that identifies individual animals from six predator species in a single survey.

Counting animals in photographs

The approach, known as spatial capture-recapture, has a cumbersome title but a straightforward principle. Researchers place cameras across an area and use each animal’s markings to link photographs to the same individual.

When an individual appears at enough camera sites, those records can estimate both the local population and the distance it travels. In practice, however, the technique has generally worked well for just one species at a time.

Camera grids are usually designed around a single animal, leaving every other species as incidental bycatch. Laura C. Gigliotti, a wildlife biologist at West Virginia University, set out to create a grid capable of counting several predators at once.

Why a single camera grid falls short

The problem comes down to the amount of space required by each species. Leopards roam across extensive territories, meaning cameras intended to record one of these cats need to be widely spaced.

A large-spotted genet, however, can live within an area only a fraction of that size. If cameras are separated widely enough for leopards, most genets will trigger only one camera, at most-far too few records for reliable counting.

As a result, one grid normally suits either the wide-ranging species or the smaller one, but not both. Previous multi-species surveys largely accepted this compromise, producing robust estimates for the main target and approximate figures for the remaining animals.

Designing a smarter camera-trap grid

Gigliotti’s team created the grid in two stages at the 285-square-kilometre Munywana Conservancy in eastern South Africa.

They initially deployed 60 cameras for leopards, the reserve’s most wide-ranging cat. They then placed the remaining 40 cameras in the spaces between them, targeting the small, short-ranging species that the first layer was likely to overlook.

Combined, these two layers formed a grid unlike a standard survey design. The cameras were positioned about 800 metres apart, with uneven clusters reflecting the different distances travelled by each species.

The arrangement drew on a separate approach developed for species of varying sizes. Paired cameras were installed 30 centimetres above the ground beside the reserve’s roads, photographing animals from both sides.

Carnivores recorded by camera traps

Between September 2021 and January 2022, the cameras identified 438 animals from six predator species. The contrast was pronounced: there were 21 lions and six cheetahs at one extreme, compared with more than 300 genets at the other.

The remaining species were leopards, spotted hyenas and servals, a slender medium-sized cat. Software initially suggested matches, after which the researchers visually verified them using the spots and rosettes on each animal’s coat.

Because lions have no spots to compare, the team identified them using whisker spots and old scars. These features were checked against a reserve catalogue maintained since the 1990s.

Identification rates were high overall, exceeding nine in ten lion photographs and reaching every cheetah. Accuracy was lower, however, for hyenas and genets.

Using cameras on roads has a recognised drawback. One study of comparable systems found that animals which avoid roads can distort population estimates, and cautious cheetahs probably passed the cameras more often than their recorded total of six indicates.

A first-of-its-kind public dataset

No public camera-trap dataset had previously achieved this for every species in a survey at the same time. Earlier projects recorded several species together, but none provided dependable results for all of them while also making the data publicly available for other teams.

This makes it possible to investigate questions beyond the reach of a single-species survey. By placing every predator on one grid, researchers can examine how the animals divide space and whether larger cats push smaller hunters aside.

Recent research describes this community-wide perspective as essential to ecosystem management. It also offers a practical benefit.

A reserve that previously conducted separate surveys for each predator could combine the work, reducing costs and labour for teams with limited resources. The dataset also supplies statisticians with real-world figures for evaluating new methods.

Interpreting the shared predator map

The dataset now provides a ready-made record of where six predators travelled through a single African reserve. It contains enough detail to count every species, with each animal linked to the camera that recorded it.

The findings could change how reserves monitor their predators. Instead of selecting one animal for study each season, a single camera network could observe the whole carnivore community.

For the first time, every predator in one reserve has been counted together, and the method is available to any team prepared to apply it in its own landscape.

Any reserve seeking a complete count of its carnivores now has a tested blueprint.