The first time that Terry Vandenbos watched a bear run from a drone, on a spring day two years ago, he was chasing the animal himself. After he saw the grizzly cross a road near his property, the Montana rancher hopped on his all-terrain vehicle, planning to scare it away from his cattle if needed.

But the bear began sprinting away when he was still far from it, looking over its shoulder as it ran, and Vandenbos looked up too; overhead, a small drone was following the bear, its four propellers emitting a high-pitched whine as it sent the animal towards a nearby lake.

“I don’t think I need to be here,” Vandenbos remembers thinking. He drove back home. The bear never touched his cows.

On the other end of that drone was Wesley Sarmento, a former grizzly bear management specialist for Montana’s department of fish, wildlife and parks (MFWP) who has spent the last six years testing different non-lethal methods for scaring bears away from human habitation, a practice commonly referred to as “hazing”. In research forthcoming in the journal Frontiers of Conservation Science, Sarmento – now a PhD student at the University of Montana – shows that aerial drones outperformed all other hazing methods tested in his experiments. They provide a way to move grizzly bears away from humans that is safe for humans and animals alike.

“The drone’s become a tool where I can’t see doing the job without it now,” Sarmento said. “It’s just that handy.”

Increases in human-wildlife conflict

For nearly two centuries, prairies like those around the Vandenbos farm in north-eastern Montana have been nearly free of large predators. As humans converted native grasslands to farms, they also waged a highly successful campaign to shoot, poison and drive away animals like grizzlies, wolves, coyotes and mountain lions. But in the last 50 or so years, that has changed. Thanks to laws like the Endangered Species Act and a growing awareness of predators’ important roles in the ecosystem, predator populations have regrown.

“The really good news is that we’ve done a good job of recovering some of our large carnivores,” said Julie Young, a wildlife biologist at Utah State University studying how to reduce human-wildlife conflict. “At the same time, the human population increased when carnivores’ were at their lowest. We didn’t think about how to live with them because we didn’t have to.”

As returning predators find their former habitats occupied, conflicts are increasing. Between 2013 and 2021, cattle lost to grizzly depredation in Montana increased from roughly 20 a year to more than 140 a year, according to the most recent statistics available from MFWP. Grizzly populations are growing in Idaho, Wyoming and parts of Washington state.

Seeing what sticks

MFWP hired Sarmento in 2017 to help deal with these growing conflicts. Locals in the agricultural hub of Conrad were told to call the grizzly management specialist if a bear entered their property to scare the animal away. In his early years on the job, Sarmento focused on using his truck to scare bears off, driving toward the animals and honking his horn, as well as firing non-lethal firearms like rubber bullets, loud cracker shells and paintballs. He also persuaded the US Fish and Wildlife Service – which manages grizzly conservation, due to their endangered status – to give residents permission to use paintball guns and their own vehicles to scare bears away themselves if his team couldn’t be there in time.

Sarmento found that these techniques had their limits. Projectiles usually drove bears off but required getting close to the animals; rubber bullets also posed a risk of injuring, and even killing, bears if not used correctly. Cracker shells were a mixed bag; some bears seemed unfazed by them, perhaps from becoming habituated to the sound of gunshots. Vehicles were consistently effective at hazing bears but could only drive so far. Sarmento’s team often received calls about bears bayed up – wildlife ranger parlance for taking shelter – in the windbreaks planted around people’s houses. And even if they were able to follow a bear into a nearby field, they would need to stop if he hit a creek or a patch of trees, or the sticky morass of “gumbo” left in Montana’s clayey soils after rainfall.

In 2019, Sarmento added two new members to the bear management team: Huckleberry and Gum, a pair of redline Airedale terriers. Large and curly-haired, with scruffy steel wool-like coats, they came from a line known for chasing off wildlife, and had been trained on feral hogs. But the dogs were inconsistent at bear hazing. “Basically, they just chase the first thing they see,” Sarmento said. Both dogs visited the vet multiple times to remove hundreds of quills after focusing on a porcupine rather than a bear. Another time, Sarmento released the dogs on a grizzly only about 150ft (45 metres) away, and both dogs decided to chase a feral cat that was nearer.

‘They flee pretty quickly’

The MFWP approved the purchase of a drone, Sarmento obtained his Federal Aviation Administration remote pilot certification, and he started flying out in 2021. It was a bright orange Autel EVO II, equipped with a high-definition camera and, at only 2.5lb (1.13kg), lightweight enough to fly for a full 40 minutes. Drones first rose in popularity in wildlife management for hazing birds, particularly around airports and in agricultural fields. And as decreasing prices made these aircraft more accessible – the cost of an EVO II starts at just over $2,000 – researchers like Sarmento began wondering whether they might have an effect on larger animals.

“Immediately, it became clear that it was the best thing,” Sarmento says. He’s still not sure why bears so dislike drones; he’s theorized that the loud buzz of the rotors could sound like an approaching swarm of bees, or remind bears of being dive-bombed by birds when they raid nests for eggs. Conversely, it could be that bears have no experience to compare a drone to.

“It’s like, if we were to see a UFO, we would probably get pretty scared,” Sarmento said. “But immediately, they get vigilant. And then as you approach the bears, they flee pretty quickly.”

Data corroborated the responses he and his team were seeing: over the course of 163 encounters between the management team and bears – 35 with the drone, 52 with a vehicle, 30 with dogs and 46 with projectiles – Sarmento’s drone successfully hazed bears away from human habitations 91% of the time. Dogs succeeded only 57% of the time. (Huck and Gum had a happy ending, though. After they retired from bear-chasing, Sarmento adopted them as pets.)

The drones’ success rate was within the margin of error for vehicles (85%) and projectiles (74%), meaning that in a statistical sense, the quadcopters may not be significantly better at hazing bears. But in a qualitative sense, the drone was the clear winner. It allowed Sarmento to chase a bear across topographical barriers, such as roads or streams, as well as legal ones; while flying, he didn’t need landowners’ permission to bodily enter their property.

“I could precisely run a bear exactly where I wanted it to go. It’s just so maneuverable,” he says.

Drone deterrence with other predators

Wildlife managers are seeing a similar effect when drones are used on wolves, a major source of conflict as their populations recover and spread across the American west. Young, the Utah State biologist, is supervising a master’s student studying how drones can reduce livestock depredation by wolves in Oregon, which had promising results during its first field season this summer.

Dustin Ranglack, the predator project leader and Utah field station leader for the National Wildlife Research Center, was a collaborator on a 2022 project that showed drones projecting the sounds of human voices reduced the number of cows killed by wolves in Oregon from one almost every other night to only two over 85 nights – a decrease of 95%.

“It’s really effective, but we still have a lot of questions as to how well it is going to work, what makes it effective, and how quickly will wolves habituate – because with most non-lethal tools, they do habituate,” he says.

Habituation is one of Young’s concerns, too. Unlike with bears, researchers have found that wolves aren’t as scared of drones on their own; some have even shown play behavior when a drone approaches. This lack of fear could mean that wolves could grow accustomed to drones more quickly.

Back in Montana, Sarmento didn’t see signs that grizzlies were getting used to his drone. In fact, it appeared that drones could potentially teach bears to stay away from humans long-term. Sarmento usually needed to haze the highest number of bears in the spring, soon after they woke from hibernation and sought food; typically, those calls would drop off by July as the bears turned to wild berries rather than farms. But even in drought years, with poor berry crops, Sarmento found that bears he had hazed didn’t tend to return to human food sources. In September, after frost killed most berries, hazed bears likewise didn’t return even while bulking up for winter hibernation. Young bears, which had not yet learned to avoid humans and their flying toys, tended to be involved in the most hazing interactions.

Hazing would be less necessary, researchers say, if people utilized practices that prevent conflict like removing spilled grain and carcasses that attract predators. Even as the team in Conrad continues to use drones for hazing, the bear management team spends as much time on bear safety education, building electric fences, cleaning up bear attractants, and giving out airhorns, bear spray and bear-proof garbage cans, acknowledging that they can’t be everywhere at once.

Limitations and drawbacks

The researchers emphasized that drones aren’t a silver bullet. They can’t fly in high winds or inclement weather, and there’s currently no system available outside of the military that can automate drone flights; a trained human pilot always needs to always be on the other side of the controller, making drones a time-consuming strategy. Researchers are actively working on algorithms that can do this, including recognizing the shape of approaching predators, but these programs are still in development.

Additionally, because wolves and grizzlies are still on the endangered species list in most places in the US, only researchers with special permits are allowed to harass them. Locals are still asking about the technology, though; Montana recently changed its licensing system to allow ranchers to use drones to haze non-endangered species such as elk out of their fields. Drones are now part of an existing FAA license that previously covered hazing with helicopters.

Researchers regard these drawbacks to drones as real but surmountable, especially as enthusiasm for the aircrafts’ new uses spreads and more people try them.

“It seems like every other week I’m getting a different phone call or email from somebody who has heard what we’re doing with drones,” said Ranglack in Utah. He noted that even conservation groups, which often clash with federal wildlife services due to predator policies, have gotten in touch to express their support. “It’s one of those unique tools that’s really uniting people around this purpose, because it can be so effective.”

This story was made possible in part by a grant from the Institute for Journalism and Natural Resources.

• This article was amended on 18 November 2024. A previous version said that Wesley Sarmento was a current, not former, grizzly bear management specialist for Montana’s department of fish, wildlife and parks.