Researchers are using AI technology to capture wolves' howls and their locations in Yellowstone National Park, to help prevent them from being shot.
MAMMOTH, Wyo. -- At 1:23 a.m. on a recent summer night, the matriarch of the Rescue Creek wolf pack, known to scientists as 1490F, bellowed out two long, melodic howls that reverberated across the landscape.
Soon, her seven pups joined in, creating a cacophony that enveloped their mother's baseline howl. With this performance taking place in the middle of the night, near the wolves' den, researchers would usually have missed it. But the scientists of the National Park Service's Yellowstone Wolf Project, which has been monitoring wolves across the Yellowstone landscape since they were reintroduced in 1995, have a new tool in their arsenal: Drilled to a tree near the wolf den, a recording unit runs 24/7, eavesdropping on the wolves' conversations.
After 30 years of using traditional boots-on-the-ground observations, the researchers are collecting wolf calls in hopes of honing population estimates, tracking wolves without using radio collars and reducing conflict between the animals and ranchers.
"We're integrating modern cutting-edge scientific tools with old-school natural history," said Dan Stahler, the senior wolf biologist in Yellowstone National Park.
Without these recordings, researchers would miss out on several hours of the day when they can't observe the wolves -- and on what the pups' playful yips, moans and growls say about wolf behavior.
Learning more about how wolves communicate from a young age can shed light on pack dynamics that wildlife managers can use to better track and protect the species.
Stahler -- along with Jeff Reed, a software engineer for Grizzly Systems, the company that designs the visual and audio monitoring devices tracking wolves' howls -- are building an acoustic library of the park. It also includes the burbling of a brook, the songs of chorus frogs, the rustling of leaves and even the sound of airplanes flying overhead. This data helps the scientists better understand how different animals are interacting with each other and the landscape, and how the landscape is changing over time.
"Ecosystems are under a lot of threats right now," Stahler said. "Looking at the soundscape of a place like Yellowstone and understanding how it is now and how it might change in the future will be an important metric to really monitor these places."
Tracking wolves
Wolves, which were hunted and trapped to near extinction in the United States by 1930, were reintroduced to Yellowstone three decades ago, when the Yellowstone Wolf Project released an initial 14 wolves into the wild. Their numbers have grown steadily under the care of the scientists: Today, anywhere between 110 and 120 wolves call the area home.
The project -- which is funded by Yellowstone Forever, the park's official fundraising partner -- keeps track of wolves in the park by outfitting them with collars. Now, scientists are using the recording devices -- essentially camera traps but for sound -- to help monitor wolf populations. They hope that will reduce the need to collar the animals, which is expensive and invasive.
Last November, one recording unit picked up the sound of an elk herd fleeing from a pack of wolves, followed by 30 minutes straight of plaintive howl after plaintive howl.
Jeremy SunderRaj, a biological science technician with the Yellowstone Wolf Project, observed this scene, identifying the howling wolf as 907F, the one-eyed matriarch of the Junction Butte pack, one of the 11 that roam the park.
Without matching the recorded howls to individual wolves, the researchers only know that wolves are on the land, not which pack, how many, where they are going or which wolves they are. They hope 907F will be the first of many wolves they can identify by howl, improving their ability to track the park's wolf population.
"We think we can get there eventually, being able to understand unique pack signatures," Stahler said, adding that artificial intelligence can help identify specific wolf howls from thousands of hours of audio data.
The researchers are one year into the three-year audio collecting study, dubbed the Cry Wolf Project. It aims to collect as much wolf howl data as possible to help train the AI model, which would help the project get more accurate population estimates in the future.
How it works
After researchers retrieve the data from roughly a dozen recorders scattered across the northern range of the park, Reed uses AI to pull howls from the tens of thousands of hours of recordings. Sitting at a desk lined with feathers, the cast of a wolf skull, and other trinkets from his years gallivanting across the greater Yellowstone region, his dual monitors light up with the purple, red and orange hues. Those sound visualizations, or spectrograms, reflect individual howls.
The recording devices can pick up a howl from over four miles away, giving scientists an idea of the wolf's location in addition to the date and time it happened. Without the help of AI, it would be nearly impossible for the scientists to sift through the 30,000 hours of audio data that has been collected so far.
Using spectrograms, Reed is able to visualize each wolf howl, though 907F is the only one he can identify by name so far.
Reed grew up in Paradise Valley at the northern entrance to Yellowstone National Park, where most people make a living through ranching and farming. When they were first reintroduced, the wolves made their way out of the park, killing livestock. That's rare today -- grizzly bears are more of a concern these days. Still, Reed wants to do everything he can to reduce animosity toward the Yellowstone wolves.
He hopes to use sound to create artificial boundaries that could deter wolves from entering ranchers' lands, by playing chorus howls and wolf vocalizations on speakers that sound like a large, intimidating wolf. His research is still experimental, though. "They're smart animals," Reed said. "If it's just a bluff, they're eventually going to counter it."
Conservation converts
Even though the Cry Wolf Project is focused largely on wolves, all of the sounds it collects may help in a broader conservation effort throughout the Yellowstone ecosystem. The data allows researchers to assess which species were on the landscape at a given time, and how those species are changing. "It's a really powerful tool to monitor ecosystems," Stahler said.
The sounds may also recruit conservation advocates. In 1970, when Songs of the Humpback Whale was released by Roger Payne, an American biologist, the general public's interest in whale conservation changed the species' fate. The discovery of whale songs, and the release of the album, led to the 1972 Marine Mammal Protection Act and ended large-scale whaling in the United States. The Cry Wolf Project hopes to inspire a similar movement.
Remembering the first time he heard a howl, Stahler described it as pure magic. "Having listened to wolves for several decades now, you don't lose that connection and how it makes you feel," Stahler said. "That is the power of a wolf howl. It will always resonate with you."
Wolf communication can also highlight just how much the animals resemble humans. They are communal animals, deeply familial and devoted to providing food for themselves and their pack.
As the summer ends in Yellowstone National Park, wolf pups will begin to leave the safety of their dens. As they continue to learn their pack's language, so are the scientists of the Yellowstone Wolf Project. The data collected from the autonomous recording units this fall will allow researchers to study each pup's vocalizations from the moment they are born.
"Natural sounds are something worth protecting," Stahler said.