There’s a conflict taking place at game reserves across Africa. On one side are [[ link url=“https://www.theatlantic.com/science/archive/2018/09/a-clever-way-of-exposing-elephant-killing-cartels/570691/” text=“poachers” ]], who trespass on private and public land to hunt and harvest the world’s most threatened wildlife. On the other are anti-poaching teams run by the reserves, government agencies, and paramilitary subcontractors.
While poaching threatens a number of species, rhinos face a particularly uncertain future. Loss of habitat and an insatiable demand for their horns—particularly from China and Vietnam—have put immense pressure to survive on the once plentiful animals. Made of keratin and possessing no medicinal properties, the horns are falsely believed to be an aphrodisiac and to cure cancer, along with a host of other medical problems; they can command higher per-kilogram prices than gold. As supplies have declined over the last decade, poachers have become increasingly aggressive.
The question for François Spruyt, the chairman of the Welgevonden Game Reserve in South Africa, and other conservationists is, how best to stop them?
At Welgevonden, a team of experts from the reserve, Wageningen University in the Netherlands, South African telecommunications company MTN, and IBM is working to develop a unique solution, one that uses mobile communications, machine learning, predictive analytics, and the Internet of Things (IoT) to reduce the incidence of poaching.
Across industries, IoT technology—networks of connected, sensor-equipped physical devices ranging from cars and home appliances to custom-built urban traffic and air-quality monitoring pods—can act as a force multiplier, allowing organizations to be more productive without having to expand staff or expend additional resources.
“When it comes to rhino protection, the need to do more with the same amount of resources is particularly acute,” says Hamilton Ratshefola, country general manager for IBM South Africa. “Given the vast open spaces that the patrols must protect, these technologies augment and expand what they can do many times over.”
South Africa is home to most of the world’s remaining rhinos. Of the five species of the animal, only two—the white rhino and the greater one-horned rhino —are not critically endangered. But International Rhino Foundation Executive Director Dr. Susie Ellis warns that may change, and soon.
According to the IUCN African Rhino Specialist Group, rhino poaching in South Africa increased from 13 animals killed in 2007 to 1,215 killed in 2014 (the all-time high), a jump of 9,000 percent. While government officials reported that the death toll decreased slightly to 1,028 in 2017, Ellis says rhinos are on the verge of a population tipping point where live births may no longer reach replacement levels. “It’ll be in the next five years, if we’re not there right now,” she says.
Meanwhile, traditional anti-poaching is a difficult, expensive, and high-risk business that mostly consists of sending armed humans on patrol in the hope of intercepting poachers. In recent years, an entire industry has developed around training and deploying anti-poaching teams. The task is immense.
Located about 150 miles north of Johannesburg, Welgevonden is a private reserve that contains all of Africa’s “big five” animals: lion, elephant, leopard, Cape buffalo, and rhino. Sprawling across 90,000 acres, the reserve is roughly the size of the Caribbean island of Grenada, with a varied terrain of bushveld, mesas, and buttes. It’s just one of more than 1,000 protected reserves and national parks in Africa.
For Spruyt and his peers, that’s a lot of ground to cover, even at a reserve like Welgevonden, which has a large team and an enviable track record of conserving rhino. Others are not so well resourced. Too often at some reserves, the first sign that poachers are in an area is the discovery of a dead, hornless rhino or tuskless elephant.
“Everyone’s at the point now where we’re willing to entertain any innovation to try and address the problem,” Ellis says.
That’s where the Welgevonden project comes in. While conservationists at other reserves have used drones to watch over rhinos and have experimented with implanting global positioning tracking chips directly into the animals’ horn, neither approach is ideal. By the time poachers are spotted from above, it’s often too late: The tracking chips primarily make it easier for police to prosecute poachers and horn traffickers only after animals have been killed.
The team at Welgevonden has taken a novel and potentially revolutionary approach. Rather than track the rhinos themselves, they track the other animals on the sprawling reserve—the better to predict when poachers are preparing to attack.
Biologists have long known that animals have varied reactions to threats and other stimuli. An impala behaves differently around a safari vehicle than it does when confronted by a predator like a lion—and differently still around, say, an unknown human carrying a rifle, prowling through the bush under a full moon. The data scientists on the Wageningen University team believe they can use those reactions to their rhino-protecting advantage by monitoring the behavioral changes in the reserve’s animals in aggregate, creating an early warning system that will help them find and catch poachers before they strike.
Over the past several months, the team has outfitted 32 impala, 33 wildebeest, 42 zebra, and 24 eland with special collars that collect information about location, temperature, and movement speed and direction. All four plains animals are more abundant than rhinos, and poachers are likely to encounter any of them before finding what they’re looking for.
The information from the collars is transmitted over a custom-built MTN network and integrated with other information, including locations of park employees and safari vehicles. With every new data point, Wageningen University—using IBM machine learning technology—is better able to determine whether the animals are reacting to known park personnel or unauthorized human trespassers such as poachers.
The project draws heavily on the research of Dutch ecologist Dr. Henjo de Knegt, whose original work had nothing to do with rhinos. His interest in using animal sensors and IoT began in Europe, where he tried to help farmers better monitor cow pastures. By analyzing information gathered in sensors he had placed in the collar around each cow’s neck, De Knegt was able to “measure indirectly the quality of the pasture” and when it was time to move the cattle somewhere greener.
The idea, De Knegt says, is to “use animals and how they behave as our way of looking at the environment.” He adds, “They are very clever. And they respond in intricate and sometimes very complex ways to how they perceive the environment.”
Professor Herbert Prins, who leads De Knegt’s department at Wageningen University, realized that the technique could be applied to wild animal populations as well as domestic ones and brought the project to Welgevonden.
While Welgevonden staffers can dart and collar animals “in our sleep,” according to Spruyt, designing the special animal collars containing the IoT components presented considerable challenges. The collars themselves needed enough battery power to last multiple years and had to be sturdy enough to withstand their natural environment—including the swift kick of a zebra’s hind legs. The team also constructed a reliable, long-range, low-power communication network that receives the data from the collars and transmits it onward for storage and processing.
Simon Hodgkinson, the founder and CEO of the Smart Earth Network—a United Kingdom-based nonprofit dedicated to bringing technological solutions to park managers in Africa, Asia, and Europe—says that many of Africa’s parks have largely failed to leverage even the most basic network technology and instead still run on pen and paper.
“If you ask a park ranger what tools they want, they’ll always say drones because they want toys,” Hodgkinson says. “But the first problem in any protected area is communications and creating a network for working—and then deploying devices in the field.”
For now, the Welgevonden project is very much a work in progress. It will take several more months of gathering and analyzing animal-movement data and refining algorithms to determine just how well the collared animals can help identify and predict human trespassers. Early results are positive, De Knegt says, suggesting that the ultimate goal of using animals and IoT technology to protect rhinos from poachers—a solution that could one day help endangered species across the planet—is within reach. “If we can tap into information from individuals through their movements, through their behavior,” he says, “then we make a big step forward.”