- A forest can appear intact from above while losing part of its animal community below the canopy. Satellite images and carbon accounting can miss these changes, making bioacoustics a useful way to detect whether a forest’s living rhythms remain intact.
- The Soundscape Baselines Project, described by Zuzana Buřivalová and colleagues, is building acoustic reference points for intact forests before those baselines disappear. Its pilot sites span Brunei, Ecuador, Gabon, Germany, Peru, and the United States, using continuous recordings managed with local teams.
- Acoustic monitoring can reveal changes that averages and visual measures obscure. In Gabon, logged forests could appear similar to baseline forests in coarse daily measures, but the timing and shape of dawn and dusk choruses showed important differences.
- Bioacoustics has both promise and limits. Tools such as acoustic indices and BirdNET can expand conservation monitoring, but they require careful calibration, local knowledge, and transparent treatment of uncertainty if they are to support credible claims about biodiversity protection or recovery.
From above, an intact forest can look reassuringly complete. A satellite image may show an unbroken canopy, a block of green still standing amid plantations, roads or logged land. For many conservation programs, that view has become the starting point for measurement. If the canopy remains, the forest is often treated as if much of its ecological value remains as well.
The forest itself may tell a more complicated story. Birds, insects, frogs and primates divide the day among them. Some call at dawn, others at night. Some occupy narrow frequency bands; others fill the background with a steady rasp. A forest that looks intact can still lose part of this living structure. The canopy may close after logging. Carbon may remain on a balance sheet. The animal community may not return in the same form.
A new paper in Global Change Biology, by Zuzana Buřivalová and colleagues, examines that problem through sound. The study describes the Soundscape Baselines Project, an effort to record the acoustic signatures of some of the world’s remaining intact forests before those reference points become harder to find.
The idea is straightforward. To know whether a forest has changed, one needs to know what it sounded like before the change. That baseline is not only a technical convenience. It is a guard against a familiar problem in conservation: each generation tends to accept the nature it first encountered as normal. Daniel Pauly called this shifting baseline syndrome in fisheries. The same pattern applies on land. A forest heard for the first time after decades of hunting, logging or climate stress may seem full of life, even if much of its former community has already disappeared.
This is not a new concern. Bernie Krause, a musician turned bioacoustician and founder of Wild Sanctuary, has spent more than five decades recording and archiving natural soundscapes around the world. His recordings have helped popularize the idea that ecosystems have acoustic structures, with species occupying different times and frequencies. They also show the cost of delay. Some places were recorded before they changed. Many others were not. The Soundscape Baselines Project is now trying to bring that lesson into a more systematic conservation framework.
The paper notes that Intact Forest Landscapes declined by 1.5 million square kilometers, or 12% of their remaining area, between 2000 and 2020. Even forests with little direct industrial disturbance are changing as temperatures rise and species ranges shift. Reference sites are becoming scarcer. Waiting to collect baseline data means accepting a progressively altered world as the comparison point.
Bioacoustics offers one way to shorten that delay. Passive recorders can sit in a forest for months, collecting the sounds of animals, weather and human activity. They do not replace field biologists, camera traps, satellite data or Indigenous and local knowledge. Their value is different: they create a continuous record of ecological activity at times and scales that human observers cannot match.
The Soundscape Baselines Project began with six pilot locations: Brunei, Ecuador, Gabon, Germany, Peru and the United States. Each location included six recording sites, chosen with local teams and spaced at least one kilometer apart to reduce overlap. The recorders were placed about 1.5 meters above the ground and programmed to record continuously for at least a year. At 44.1 kilohertz, each recorder generated about 5.85 gigabytes of data a day. A six-recorder network produced roughly 1.5 terabytes a month.
Those details reveal both the promise and the burden of the method. Acoustic monitoring is often described as cheap and scalable, which is partly true. The equipment can collect enormous amounts of data without a specialist standing beside it. The harder work comes later: maintaining the devices, managing storage, processing files, deciding which signals matter and making sure the data remain useful to people beyond the original research team.
The project gives considerable attention to that problem. The paper describes a modular system designed to be expanded and combined with other technologies. Local teams of scientists, conservation practitioners and community members manage the work at each site. Copies of the data are stored within the country of each baseline site, as well as at the University of Wisconsin-Madison and in cloud storage. The approach is not only about efficiency. If biodiversity data are to guide decisions about forests, the people closest to those forests need a role in producing and using them.
Once recorded, a forest soundscape can be examined in several ways. One approach is to treat the whole soundscape as a signal rather than trying to identify every species. The paper focuses on acoustic indices, including Power Minus Noise and Soundscape Saturation. Power Minus Noise estimates acoustic activity by subtracting background noise from each frequency band. Soundscape Saturation measures the proportion of frequency bands that are active in a given minute.
These measures simplify an immense amount of sound into patterns that can be compared across places and times. They also avoid a common limitation of species-by-species monitoring. Many tropical animals are hard to identify by sound, and many have not been well represented in digital sound libraries. An index does not need to know which katydid or frog is calling to detect that the nighttime soundscape has changed.
The initial results show that intact forests have daily acoustic rhythms. In the tropical sites, activity often rose sharply around dawn, dipped later in the day and increased again at other times. Nighttime sound in many tropical forests was sustained by insects and amphibians. Temperate sites followed different rhythms. The point is not that one pattern is better than another, but that each forest has a measurable acoustic signature. Once that signature is known, departures from it become easier to detect.
The Gabon comparison is especially useful. Researchers compared baseline sites with selectively logged forests in the same broad region, including non-certified logging concessions and Forest Stewardship Council certified concessions. A coarse 24-hour average could make the FSC-certified sites appear similar to the baseline forest in Soundscape Saturation. The time series told a different story. The timing and shape of the dawn and dusk choruses shifted in the logged forests.
That distinction is important for conservation measurement. Averages are convenient, but they can hide ecological change. The relevant question may not be whether a forest is generally loud or generally green. It may be whether its daily pattern still resembles the pattern of an intact system. A chorus delayed or flattened at key hours may point to a change in community structure that a satellite image cannot show.
This is where bioacoustics becomes useful for forest management, carbon projects and emerging biodiversity-credit systems. Many nature-based climate projects still rely heavily on forest cover or carbon storage. Those are necessary measures, but they do not show whether a forest still supports its animal communities. A project can retain trees while losing part of its fauna. Acoustic baselines give managers another test: whether the living rhythm of the forest is being maintained, altered or restored.
The paper also shows how sound can be used to identify individual species, with more caution than the public discussion of artificial intelligence sometimes suggests. The researchers used BirdNET, a machine-learning system for detecting bird vocalizations, to examine recordings from Gabon and Peru. In Peru, six months of recordings across a montane gradient produced detections of 329 bird species. The results showed substantial turnover with elevation, with higher sites containing communities that were not simply reduced versions of the lowland sites.
That finding is useful. So is the difficulty behind it. In preliminary work, BirdNET’s location model supplied species that do not occur in Cuzco province or occur above the treeline. The team had to use a custom species list, draw on regional surveys, review sample detections manually and convert confidence scores into species-specific probabilities through validation. Automated detection can expand monitoring, but it still depends on local ecological knowledge and careful calibration.
This point is reinforced by separate guidance from Connor Wood and Stefan Kahl on BirdNET scores. A score that looks like a probability is not necessarily one. It is a unitless output from a model, and its meaning can vary by species, location, equipment and recording conditions. For conservation, that distinction is important. A biodiversity system built on automated detections will only be credible if it treats uncertainty as part of the method rather than as an inconvenience to be hidden.
The same caution applies to acoustic indices. Soundscapes include biophony, the sounds made by organisms; geophony, such as rain and wind; and anthrophony, the sounds of human activity. A storm, a road, a river or a malfunctioning microphone can shape the data. Earlier work on acoustic indices has emphasized the need for consistent sampling and analysis, because different indices can produce different patterns. Bioacoustics is powerful, but it is not self-interpreting.
Its larger value may be as a form of ecological memory. A recording made today can be analyzed with methods that do not yet exist. Species that current algorithms miss may later become identifiable. Patterns that now seem obscure may become meaningful when paired with camera traps, satellite data, field surveys or community observations. In that sense, the raw archive is as important as the first analysis.
That distinction is becoming more important as conservation moves toward more formal claims about impact. Governments, companies and funders increasingly ask whether an intervention protected biodiversity, improved ecosystem condition or produced benefits beyond carbon. Those claims require baselines. They also require a counterfactual, an estimate of what would have happened without the intervention. Sound alone cannot provide all of that, but it can make the biological side of the question less abstract.
A forest may keep its canopy and lose part of its animal life. It may recover structure before it recovers function. It may be managed well enough to retain some acoustic patterns and not others. These differences are hard to see from above. They are easier to examine when a forest has been listened to over time.
The conclusion is not that every conservation project needs a year of continuous recordings at every site. The authors themselves note that such dense sampling may not be necessary for all uses. The more modest lesson is that baselines should be collected before they vanish, and that measurement should be designed around the systems it is meant to describe.
Forests are not only carbon stocks or blocks of canopy. They are communities moving through time. Some of that movement can be heard. Recording it will not, by itself, protect a forest. It can make loss harder to miss, recovery easier to test and conservation claims harder to make without evidence.
Banner image: Sathya Chandra Sagar (left) and Maia Persche (right) at the Soundscape Baselines Site in Baraboo Hills, Wisconsin, USA, includes view of bioacoustic recorder and camera trap. Photo by Caleb Alvarado
Citations:
- Buřivalová, Z., S.Perea, L. M.Berman, et al. (2026). “Bioacoustic Baselines for Intact Forests.” Global Change Biology32, no. 5: e70917. https://doi.org/10.1111/gcb.70917.
- Wood, C.M., Kahl, S (2024). Guidelines for appropriate use of BirdNET scores and other detector outputs. J Ornithol 165, 777–782. https://doi.org/10.1007/s10336-024-02144-5.
- Kent H. Redford (1992). The Empty Forest: Many large animals are already ecologically extinct in vast areas of neotropical forest where the vegetation still appears intact, BioScience, Volume 42, Issue 6, June 1992, Pages 412–422, https://doi.org/10.2307/1311860
- Pauly D. (1995). Anecdotes and the shifting baseline syndrome of fisheries. Trends in ecology & evolution, 10(10), 430. https://doi.org/10.1016/s0169-5347(00)89171-5.
- Pijanowski, B.C., Farina, A., Gage, S.H. et al. What is soundscape ecology? An introduction and overview of an emerging new science. Landscape Ecol 26, 1213–1232 (2011). https://doi.org/10.1007/s10980-011-9600-8.
- Rappaport, D. I., Swain, Anshuman, Fagan, W. F., Dubayah, R., & Morton, D. C. (2022). Animal soundscapes reveal key markers of Amazon forest degradation from fire and logging. Proceedings of the National Academy of Sciences, 119(18), e2102878119. https://doi.org/10.1073/pnas.2102878119
- Sethi, S. S., Ewers, R. M., Jones, N. S., Orme, C. D. L., Picinali, L., & Woodward, G. (2022). Soundscapes predict species occurrence in tropical forests. Oikos, 2022(5), e08525. https://doi.org/10.1111/oik.08525
Can listening to a forest reveal whether it is ecologically healthy?
The more degraded a forest, the quieter its wildlife, new study shows
Cite this article
Rhett Ayers Butler (2026). Why conservation urgently needs acoustic baselines. Mongabay Conservation news. DOI: https://doi.org/10.66709/news-320333Credits
