Giacomo Delgado has lost his microphone. As we peer into the cloud forest of Costa Rica’s Barva volcano, a GPS device tells him it is somewhere nearby, but the only note he left when he placed the mic here is that it is attached to a moss-covered tree. That would be helpful, except it describes nearly every tree in the old forest. He squints. “I hate it when I don’t give myself good notes.”
For the past two months, Delgado, a doctoral researcher at ETH Zurich in Switzerland, and two other teams from Costa Rican universities have bounded across the country recording the soundscapes of its forests. The survey involves positioning microphones at more than 600 sites in every type of forest ecosystem in Costa Rica, making it, by some measures, the largest such “ecoacoustic” survey ever.
Advertisement
The survey is part of a shift in the way we monitor ecosystems. Cheaper audio recorders and improved methods of analysing complex acoustic data using machine learning have led to a boom in this field. And researchers are increasingly listening in to ecosystems to hear how they change and monitor their health.
In Costa Rica, by comparing the soundscapes of regenerating forests with those of forests with their biodiversity intact, the researchers aim to take the measure of ecosystems coming back to life across the country, in what could be some of the first evidence that large-scale forest restoration brings back the full diversity of ecosystems. But this has the potential to reach far beyond Costa Rica. The effort could support plans to change the way we monitor biodiversity across the entire planet.
What is bioacoustics?
The study of animal sounds in nature, known as bioacoustics, largely got its start underwater, where sound reaches further than light. Many marine animals leverage this quality to communicate. Likewise, researchers have long used sounds to study their lives and habitats, from tracking how songs are passed between different groups of humpback whales to documenting how a .
Traditionally, researchers hoping to gauge biodiversity in terrestrial ecosystems would use old-fashioned field surveys to count species or set up automatically triggered camera traps that capture images of animals moving past. These days, an increasingly common method is to sequence the ambient environmental DNA found in air, water or soil and take a species census that way.
Now, though, bioacoustics on land is “transforming how we can monitor and understand the environment”, says at the Queensland University of Technology in Australia, where he runs a network of more than 300 permanent acoustic sensors distributed across the continent. Some researchers use sound to track the behaviour of a particular species. A might reveal a hidden pair in the jungle. The rate of has been linked to changes in vulture populations. Conservation organisations .
Roe has used his acoustic observatory to study the ecological effects of the invasive species and bushfires that are becoming more frequent with global warming. This method, which captures the dynamics not only of individual species but also of entire ecosystems, is called ecoacoustics.
Microphones offer an affordable tool to record ecosystems across large areas, revealing subtle changes to their biodiversity. Last June, for instance, , then at Columbia University, New York, and his colleagues studied the soundscapes of restored forests in the mountains of south-western India. They found that insects like katydids, also known as bush crickets, that make high-frequency sounds were . In 2022, at the University of Wisconsin-Madison and her team found that tree plantations and forests used for logging in Borneo common to pristine forests.
The Costa Rican survey
Delgado’s survey is looking for similar gaps, but at a much larger scale and in a country many see as a beacon of hope in the fight against biodiversity loss. At its lowest point, in the 1980s, following rapid deforestation due to cattle ranching and agriculture, only about a quarter of Costa Rica was covered in forest. But the forest started to rebound as the price of beef fell and people abandoned agricultural land. In 1996, the country introduced a programme called Payment for Environmental Services (PES), in which the government pays landowners to preserve and restore forest on their property.
Today, trees cover about 60 per cent of Costa Rica. If the survey finds this has also led to the return of biodiversity, it would be some of the for successful large-scale ecosystem restoration anywhere. “A country turning around the state of its nature is really unheard of,” says Delgado. He wants to hear it.
By the time I catch up with at the end of March, he has already logged more than 300 kilometres on foot and thousands more by car. Two months of sleeping in bare-bones ranger stations hasn’t been restful. But other than some lost data and an incident with several angry spider monkeys, the survey is going as planned. “The hardest parts of my day are when nature is unpredictable,” he says.
Our first visit is to an area of tropical wet forest in the mountains west of Costa Rica’s central valley, called the Alberto Manuel Brenes Biological Reserve. We bounce up a rutted road accompanied for the day by , who administers the reserve for the University of Costa Rica. Vega explains that the strictly protected area can only be visited to conduct research. This pristine forest will serve as a baseline for what an intact ecosystem should sound like (listen to an intact forest below).
Cued by Delgado’s GPS, we hop out of the truck to collect several microphones he placed eight days earlier at 1-kilometre intervals, far enough apart to ensure the mics wouldn’t capture the same sounds. The mics are a short walk from the road, but Vega still recommends we take care because the area is known to have a “healthy” number of pit vipers.
We press into the forest, hearing the ring of a few cicadas and some intermittent birdsong, including the weird quack of a and the bassoon hoot of an . An iridescent blue morpho butterfly flits by. Any other animals are quiet in the late morning heat. But Vega assures us that the forest can be full of sound. “At the beginning of the rainy season, the forest becomes so loud it’s difficult to hear,” he says. Indeed, the researchers are racing to beat the weather: the survey must be completed by May to avoid the rain, which will make it impossible to compare the soundscapes of different sites.
Other microphones are more difficult to reach, requiring a steep hike up a ridge across from the reserve’s research station. A few minutes up the trail, Delgado halts as a large group of boar-like peccaries snuffle past. Other than our crashing, which surely scares silent whatever wildlife is around, the loudest sound is the whine of black flies eating our necks.
As we climb, Delgado explains what he hopes this survey will reveal. By the time he is rained out, he aims to have recorded audio at around 1200 sites, or as many as he can reach in time, capturing the equivalent of more than a decade’s worth of audio data covering all the major forest types in the country. As well as providing information on how successful the reforestation has been at bringing back biodiversity, data from the survey will be uploaded to an  used to track thousands of ecosystem restoration projects all over the world.
The platform, Restor, built by ecologist Thomas Crowther’s lab, where Delgado works in Zurich, currently enables people involved in ecosystem restoration projects to tap into a sophisticated array of satellite data on forests. The acoustic data will add a direct measure of what is happening “underneath the canopy”, says Delgado. The hope is that people will eventually be able to upload acoustic data of their own. This could be a step towards a global network of acoustic sensors to monitor the planet, allowing researchers to observe changes in biodiversity almost in real time.
Plantations and pastures
The several forests I visited with Delgado are just half of the story. Two other teams of researchers roam the country recording the sounds of forest fragments on ranches and farms, as well as on tree plantations and cow pastures. The moos and motors of those soundscapes will contrast with the sounds of well-protected forests. The big question is what other aspects of their biodiversity will differ as well.
One of the teams exploring these sites is made up of Justin Jiménez Céspedes at the National University of Costa Rica and Daniel Vega Cháves at the National Technical University of Costa Rica. They might not have to hike as much as Delgado to reach their sites, but working on private property comes with its own set of challenges: on one occasion, they were chased out of a forest by ranchers wielding shotguns. Fortunately, on the day I spent with them in the rural Alajuela province, a few kilometres from the border with Nicaragua, we only encountered forest owners happy to have us around.
One is Oscar Rodriguez. He seems suspicious at first, but after Céspedes explains the survey’s purpose, he enthusiastically gives us a tour of his property: a 700-hectare (1700-acre) plantation of hardwood trees and patchy regenerating forest, which his family has owned for decades.
As Cháves places a mic in a small stand of trees growing in an open area, Rodriguez points out a few saplings he planted. He receives around $55 a year through the PES scheme for each hectare of land he replants with trees, and a bit more for each hectare he leaves covered in forest.
While Rodriguez is in favour of the programme, he says the money isn’t enough, and he worries about people continuing to clear forest. The initiative, which is mainly funded by a tax on gasoline, has long been oversubscribed and is chronically cash-strapped. Another hope Delgado has for his survey is for it to help the programme tap into new international sources of funding such as biodiversity credits, which offer a way for people to offset destruction in biodiversity elsewhere. To make ends meet, Rodriguez is considering building tree houses to attract tourists as his forest regrows. He is interested in getting a microphone of his own.
The full survey results won’t be available for at least a year. But there are already indications that ecosystems are returning. Two years ago, as a pilot study for this larger project, Delgado carried out an acoustic survey of the Nicoya peninsula in Costa Rica’s verdant north-west (listen to recordings from the survey below). This involved 119 sites in just one region, but it asked many of the same questions as the country-wide survey. Preliminary results show that the soundscapes of regenerating forests in Nicoya generally resemble those of intact forests much more closely than those of pastures or plantations. Delgado says this suggests Costa Rica’s restoration is working for the whole ecosystem, not just the trees.
By comparing soundscapes in more detail at different frequency ranges and times of day, the researchers have also started spotting missing sounds – and unwanted ones. For example, certain high-frequency sounds that dominate pristine forests at night are absent in restored forests. Delgado thinks this could be due to missing amphibian diversity or other nocturnal species that don’t do well in human-dominated landscapes. During the day, pastures and tree plantations show a spike in lower frequencies. Sounds from human activity, like motors and mooing cows, are the likely culprits there, says Delgado.
There are larger questions about how to reliably use sound as a measure of biodiversity. But research elsewhere suggests the chorus of vocalising species can say something about the ones that are mute. In October last year, at the University of Würzburg in Germany and his colleagues used different measures of acoustic complexity to . They used a series of “acoustic indices” that provide a generic way to compare the complexity of soundscapes without having to identify the individual species making sounds, as well as a machine-learning model that could identify some species of birds. They compared these audio indicators with genetic diversity measured among insects that were physically captured in the different forests.
The researchers found that the combination of acoustic indices could predict several aspects of biodiversity in the different forests, and that these correspond to what Müller calls an “acoustic signature” of an intact forest. The machine-learning model provided an even more reliable prediction of biodiversity in the different forests, despite it identifying only about a quarter of the bird species present. In other words, the sounds of a limited number of singing birds were enough to reliably predict the presence of silent insect species.
It remains to be seen whether these signatures of restoration hold true across many different types of ecosystem, says Müller. Determining that will require surveys like the one in Costa Rica. It will also mean collecting and labelling enough training data to build machine-learning models that can identify a greater diversity of species across all regions. “It’s not a scientific question any more,” says Müller. “It’s a question of training data.”
But Delgado’s survey might also shed light on more theoretical questions about the sounds of ecosystems. For instance, Delgado, who has been known to spend weekends back home in Zurich in his hammock listening to forest sounds, thinks the soundscapes of pristine ecosystems may have qualities that people find more pleasing than degraded ones. He thinks these might even correlate with patterns we enjoy in music. “We’re drawn to these places with sounds that are pleasing to us,” he says. “And they’re pleasing to us because they’re good habitats.”
Here in Costa Rica, however, the requirements of the survey leave him little time to luxuriate in forest sounds. “The microphones do my contemplation in the woods for me,” says Delgado, as we head down yet another bumpy road. “I wish I could take their place sometimes.”
James Dinneen is an environment reporter at żěè¶ĚĘÓƵ
