How studying nature's symphony can help scientists determine the fate of rainforests
Improvements in audio recorders, artificial intelligence, and data storage make sound a more powerful tool for ecology than ever before.
Listen 10:12Every morning, come daybreak, ecosystems around the world break out in song.
These are dawn choruses, explosions of sound that happen at the edge of night and day, for reasons scientists don’t fully understand. In terrestrial, freshwater and marine environments at every latitude, insects, fish, shrimp, frogs, birds and mammals join in with squawks, warbles and chirps.
In the clamor, you get a sense of the rich tapestry that is an ecosystem — a community of many players and layers, worlds within worlds.
“You have so many different species calling at the same time, and the sun is coming up, and it’s amazing,” Marconi Campos, an ornithologist who works in Puerto Rico, said.
Dawn chorus erupts on a farm in Hertfordshire, England. (Video courtesy of ESL and Popular Culture via YouTube)
For Campos, sound has always had a special appeal. He’s struggled with sight since he was young — he’s colorblind and acutely nearsighted. That made his job hard, he said, because to identify birds, “you need to rely on colors and patterns, and I’m not good at that.”
Campos got his start in research studying birds in the Amazon rainforest, where the dense jungle made it even harder to see things. So instead, he trained his ears on the sounds of the forest.
Rigging and deploying his own portable recorders, Campos started to do acoustic monitoring at his field sites.
In 2012, he met Mitch Aide, a professor at the University of Puerto Rico. Aide recalled being in the field together, sleeping in hammocks, and realizing just how tuned in Campos was to the sounds of the rainforest.
“I didn’t sleep that much that night, because there was just so much noise in the forest,” Aide said. When they got up in the morning, he remarked to Campos, “Boy, there were a lot of birds calling last night, or early this morning.”
In reply, Marconi said, “Yes — there were 32 species calling.”
By that point, Campos had hundreds of recorders in the field, and terabytes of data collected.
Afterward, he’d spend days, painstakingly listening to the recordings, trying to identify different bird species to figure out where they were spending their time. It took forever.
“I was using just 5% or less of my recordings because I couldn’t listen to [them] all,” Campos said.
But Aide had been training computers to do that work.
He started a company called Sieve Analytics, which developed a software called Arbimon. Using a form of artificial intelligence called machine learning, Arbimon automates the Sisyphean task of sifting through acoustic data.
ARBIMON works by analyzing visual representations of sound called spectrograms. Here, spectrograms illustrate the calls of a Puerto Rican screech owl. (Video courtesy of Marconi Campos/ ARBIMON)
According to Aide, improvements in technologies like artificial intelligence, data storage, and audio recorders have converged to make sound a powerful tool for ecological research.
Scientists are “taking advantage of acoustics in a different way than people have done in the past,” he said.
Acoustic surveys, he added, are a relatively easy way to gather a wealth of information about a given site — and they provide a permanent record that others can use and revisit.
Some tropical ecologists have even said that audio data is low-hanging fruit that everyone in the field should consider collecting.
A new field of study
Enter, the discipline of ecoacoustics. In 2014, a French ecologist, Jerome Sueur, and an Italian ecologist, Almo Farina, led an effort to define and declare ecoacoustics its own branch of study.
In ecoacoustics, scientists use animal sounds as signals of ecosystem health, or cues and proxies for biodiversity.
There are two big forces at play here, ecologists hypothesize. The first, called the acoustic niche hypothesis, suggests that sound diversity might reflect biodiversity.
“Every organism tries to have an independent signal from the other species, in order to avoid competition,” Farina said.
In other words, animals sharing an ecosystem evolve to chatter across different frequencies — like radio stations occupying different channels. Look for which frequencies are missing, and that might point you toward species that are no longer there.
Then, there’s the acoustic adaptation hypothesis. It’s thought that animals evolve distinct calls that best travel across their habitat — across a river valley, a dense jungle, or whatever it may be.
That gives each type of habitat “its own acoustic footprint,” Sueur said.
When a habitat is modified — like a forest is logged — that acoustic signature can be used to assess how much change has occurred, as some soniferous creatures are pushed out, and others come in.
The first phenomenon, the niche hypothesis, says that animals in an ecosystem diversify their sounds. While the second, the adaptation hypothesis, suggests that animals’ calls in an ecosystem converge. From this push and pull arises the soundtrack of an ecosystem.
Ecoacoustics in practice
Right now, Campos is studying the soundtrack of a place called Barro Colorado Island, in Panama.
He hopes to compare the current distribution of bird species with the historical distribution. “In that island, studies with birds started in the 1920s so we have a complete knowledge of how species were going extinct,” he said.
A couple years ago, Campos covered the island with recorders, programmed to record one minute of sound every 10 minutes. After two weeks, he ended up with nearly 130,000 recordings.
Campos captured the piercing song of the crested guan, a bird that seems to be “increasing its abundance, because there is no more hunting there, and there is a lack of top predator,” he said.
Call of the crested guan. (Audio courtesy of Marconi Campos/ ARBIMON)
He also recorded the forest falcon, a raptor that specializes in hunting other birds, and usually sings only at dawn. It has a “very human-like voice,” Campos said, “so it seems like someone is screaming in the forest.”
Call of the forest falcon. (Audio courtesy of Marconi Campos/ ARBIMON)
Then comes the hard work of filtering the recordings. Arbimon, Aide’s software, analyzes diagrams of sound, called spectrograms.
Let’s say, Campos wants to train the program to identify all forest falcon calls in his recordings. He would draw a box around several forest falcon sound waves. From those samples, the software would create a composite template and scan through his mounds of data for matching patterns.
“He can pass the template over his 100,000 recordings in less than two minutes,” Aide said, which might spit out hundreds of detections of the species Campos is focusing on.
Then, Campos would review the results and select the ones that are correct. This step is important, because the algorithms are not error-free.
“So you’re always going to need the ornithologists, the biologists, that know their species, that know their organisms,” Aide said. “The algorithms that we’ve developed are good, but they’re never perfect.”
If these rainforests could talk
Aide and Campos have been doing acoustic monitoring in Puerto Rican rainforests for years. They now have recorders at more than 700 sites, which have amassed about a million recordings — and those recorders catch everything.
“It’s one of the things that’s fascinating doing this, is all the sounds that you don’t expect,” Aide added. “You can hear lightning, rain and wind … there’s a lot of information that’s in a one-minute audio recording.”
The ecologists have captured the sounds of these rainforests during different stress periods — for example, Hurricane Maria.
“It was my first hurricane here in Puerto Rico, well, in my life,” Campos said. “And I was pretty worried when I saw the forest. I just thought, ‘Oh my god, the birds and the frogs are gone.’”
Surprisingly, the hurricane didn’t really seem to affect the birds and frogs — at least, not where they were, or their species distribution. How many animals there were, species abundance, is a bit harder to pin down with just sound.
But what the researchers did find is that a big drought in 2015 had a more sizable impact, causing a decline in species distribution for pretty much all bird species.
“The global climate models predict that Puerto Rico will be much drier, and droughts will be much more frequent,” Aide said.
He and Campos are also collaborating with scientists like Jessica Deichmann, an ecologist with the Smithsonian Conservation Biology Institute, to study the impact of human development on wildlife.
Using ecoacoustics, they have found that natural-gas extraction in the Peruvian Amazon has varying effects on animals. While some insects and frogs are attracted to the lights and temporary pools of water that come with drilling, other creatures, such as birds, move farther away from the construction, or grow quiet in fear.
According to Deichmann, these changes in species distribution can have ripple effects that are destabilizing.
“Pulling a thread — without knowing how that might make everything unravel — is really dangerous,” she said.
As for how rainforests of the future will sound, Deichmann fears they may follow the fate of the world’s temperate forests, whose “sounds have dwindled enormously.”
With humans moving deeper and deeper into formerly isolated places, like rainforests, she said, there is the potential to “change wildlife populations in a way that will reduce the soundscape.”
The sounds that Deichmann, Campos and Aide have collected, as a baseline — and the sounds they hope to continue collecting, over the long term — will be important in understanding the effects of climate change and development on tropical ecosystems.
They hold out hope that future generations will still be able to close their eyes, open their ears and tune in — to the symphony of the rainforest.
WHYY is your source for fact-based, in-depth journalism and information. As a nonprofit organization, we rely on financial support from readers like you. Please give today.