To do that, they tried something unusual. Researchers placed small recorders along the base of the cliffs. Then they used the same algorithms the big tech companies developed for Internet searches to sort through weeks of sound files for the distinctive call of the Puaiohi.
“This is just an incredibly rugged, difficult area,’’ said Chris Farmer, a conservation biologist with the Plains, Va.-based American Bird Conservancy, a partner in the project that has pinpointed the birds’ locations. “These new techniques are critical.’’
Bioacoustic research uses sound to study land and marine environments, often focusing on animals in ecosystems that are hard to reach. What began decades ago as a way to listen to ocean sounds now uses artificial intelligence to analyze massive amounts of sound data from such environments as distant islands, remote jungles and massive tracts of land.
Bioacoustics has been used to track whales, bats and frogs, and even monitor submarines and measure urban noise. But the practice has turned out to be particularly effective for studying birds. That’s because their songs are so clear and consistent that recordings now can be mined for powerful scientific data.
The new way of listening to nature lets scientists track how fires affect bird populations in California, how shifting climate alters breeding for more and more species across the United States, and how habitat loss affects rare species in Peru. One experiment seeks to catch poachers in Africa by listening for the crack of gunshots.
“I’m well aware that you can talk to any scientist from any time period, and they’ll say we’re right on cusp of something extraordinary,’’ said Justin Kitzes, an assistant professor and a computational ecologist with the University of Pittsburgh. “But there’s a convergence of a lot of things right now that in the next 20 years are going to be very important.’’
Holger Klinck, director of the Center for Conservation Bioacoustics at the Cornell Lab of Ornithology, said that research using sound had long been cumbersome and inexact. But that changed when Google, Facebook and Microsoft shared their artificial intelligence solutions that can be trained to recognize the myriad bird songs, calls, hoots and screeches.
“That really unleashed the power of these recordings,’’ he said.
Now a new generation of ornithologists has launched experiments around the world at the same time as they work on the remaining obstacles: perfecting listening devices; refining the algorithms to recognize a growing number of spectrographs of bird sounds; and handling monstrous files created by collecting months and even years of recordings.
Connor Wood, a PhD candidate at the University of Wisconsin at Madison, joined a grant to study the endangered spotted owl across California’s 38,000 square-mile Sierra Nevada range. He was stumped by the assignment until he heard about bioacoustics.
“I was wracking my brain looking at the size of the landscape,’’ he said. “So I thought, ‘I’m going to look into this idea.’ ”
Over two years, he and his crew set up a network of hundreds of recording devices the size of lunchboxes, moving them gradually across the range to capture “soundscapes” of the forest. They indeed picked up the hoots of the owls as they searched the data. “Now we’re beginning to get the population trends we can match up over time,’’ he said.
Reid Rumelt, a computational ornithologist, just returned to his D.C. home from the Andes Mountains, where he recorded 3,600 hours of the endangered undulated tinamou. The chicken-sized birds are so camouflaged you can walk right by and still miss them, he said.
His recordings of the unique, mournful song of the tinamou will form the baseline count for a long-running study that couldn’t have been done by mere observation.
“Even if you were to send human observers out to look for them, the amount of effort it would take, for long periods of time, it’s impossible,’’ he said.
At Huntley Meadows Park in Alexandria, Va., wildlife managers are using bioacoustics to compare different water levels with the number of visits from species such as American bitterns, Virginia rails and pied-billed grebes. Dave Lawlor, the park’s resource manager, said he hopes to discover the wetland depths that will coax the birds to breed again in this Northern Virginia urban oasis.
Few terrains are as difficult as the mountains of Hawaii, the front lines of the struggle to save rare birds. A third of the United States’ endangered birds can be found on the islands, where a combination of predators, such as rats and pigs, and the avian malaria and nonnative plants have been lethal for many species.
As the bird species slowly disappeared, a group of state and nonprofit agencies turned to bioacoustics to find the remaining 500 or so Puaiohi. Now they’re putting up fences, rat traps and mosquito treatments they hope will reverse the bird’s decline.
Lisa “Cali” Crampton, project leader of the Kauai Forest Bird Recovery Project, said she often is asked why it is important to rescue a bird that people rarely see, in places you can hardly even reach. She said the thrushes, the island’s primary fruit eater that spreads seeds over the forest, are an example of how every species has a role to play.
“I would hate to see us have to save only one bird, but it would be the Puaiohi,’’ she said. “We have no forest without the Puaiohi. Without forests, we have no flood control. We have no drinking water. The forest is the backbone of these islands.’’
Most of these projects are still in the initial data collection stage, so it will take time to compare results from one season to the next to come up with the long-term population, migration and breeding trends.
“We’re still very much in the trenches,’’ said Mary Clapp, an avian ecologist at the University of California at Davis.
But the potential is growing, not just for science but also for commercial applications, researchers say.
Experiments beyond bird studies are exploring how to monitor the sound of machines in factories and on farms for malfunctions, help the deaf navigate their environments, and listen for the sounds of illegal poaching and lumbering in national parks.
One long sought-after invention in beta testing at Cornell Lab of Ornithologyis the creation of a cellphone app that can instantly recognize birds as well as the popular app Shazam identifies music.
“If I had a dollar for every time somebody asked me when we’re going to have a Shazam for birds, I’d be able to fund all this research with that alone,’’ said John Fitzpatrick, the lab’s director.
Once the use of bioacoustics is widespread, scientists envision building a library of nature’s sounds from around the world that will grow steadily more useful.
“These data will be the museum specimens of our century,’’ Kitzes said. “These are things we should be archiving. Future generations will return to them, with better methods, better equipment, new questions we can’t even think of today. They are going to be very thankful to us for preserving this.’’
Anders Gyllenhaal is a journalist, a veteran birdwatcher and the co-publisher of the birding website, FlyingLessons.US.
Various projects and studies using bioacoustics
Bioacoustics, which have been especially effective for researching birds, is helping wildlife managers and scientists answer a long list of questions. They include how to manage land for wildlife and how different species are doing amid habitat loss and climate change.
Here are some of the projects in the works:
Impacts of wildfires: Researchers are using bioacoustics to study how prescribed fires in California affect bird populations over time. The study, sponsored by federal agencies and nonprofit groups, is documenting the pace of recovery by measuring bird songs as an indicator of the health of the forest.
Sea birds: Pacific island wildlife managers have found that solar-powered recording devices left for long periods on the remote islands are the most effective way of monitoring populations of sea birds and studying how to protect them. Birds such as shearwaters and petrels are almost impossible to track otherwise because they are most active at night and nest in underground burrows.
Coffee farming practices: The Cornell Lab of Ornithology is studying which coffee growing practices in Central and South America are most effective for farmers and best for birds. The project records the soundscapes at dozens of farms to see how different practices, such as the use of pesticides and the amount of shade, affect nearby bird populations.
Impacts of military: At the Avon Park Air Force Range in Central Florida, a long-running study is using bioacoustics to study how several of Florida’s endangered species, including the Red-cockaded woodpecker, the Florida grasshopper sparrow and the Florida scrub jay, are faring near the bombing and gunnery range.
Flight calls: At the Powdermill Nature Reserve near Pittsburgh, researchers are building a library of the calls that birds make in flight to study migration, long a challenging research topic since the flights take place high in the sky and often at night.
Studying power line collisions: Another Pacific project uses bioacoustics to determine the number of birds crashing into high-tension wires by recording the distinctive sound of the collisions. Federal managers are looking for specifics of how and when the deaths are occurring in hopes of finding remedies.
Building a Shazam for birds: Cornell research has built a cellphone app that records birdsongs and instantly identifies the species. An Android version, available in the United States and Europe, has been downloaded 300,000 times and shows good promise in a field in which predecessors struggled with accuracy. The iOS version is expected in 2020.
Anti-poaching experiments: Other projects extend beyond the birding kingdom. In Congo Republic’s Nouabale-Ndoki National Park, a network of recorders is in place to track the population of elephants. Along with the sounds of the elephants, the project is tuned to pick up gunshots in what researchers hope will become an anti-poaching measure.
Creating a recording grid: At the University of Pittsburgh, researchers are taking the concept of recorded sound a step further. They are developing a grid of listening devices that can determine the exact species of birds — and their locations — based on the timing and relative volume of birdsongs picked up on different recorders. They hope such grids will enable computers to convert the data into precise maps of where birds are in any given area.
