- The use of audio to study, monitor, detect and conserve species has gained popularity in recent years.
- Passive acoustic monitoring has been found to be more efficient than traditional camera traps; however, the use of audio can be data-heavy and laborious to pore through.
- Technological developments such as artificial intelligence have made audio analysis easier, but conservationists say gaps still exist.
Think Sonoma County, and the picturesque valley and vineyards come to mind. But the locale is also home to rich and incredible biodiversity. Soundscapes to Landscapes, a biodiversity monitoring initiative in the county, aims to document just that.
Over the past five years, from mid-spring to late summer here in California wine country, the initiative collected a massive amount of sound data by placing acoustic recorders in 1,300 locations in the county. The project, run by Sonoma State University, conservation NGO Point Blue Conservation Science, and several other partners, armed citizen volunteers with recorders and collaborated with private landowners to amass audio, which was then processed and classified with the help of artificial intelligence technology.
“The idea is either to detect individual species, or to find information that tells you something new about the types of sounds there,” Leonardo Salas, a quantitative ecologist at Point Blue Conservation Science, tells Mongabay in a video interview. “We can characterize entire environments based on that.”
The methodology has been effective in monitoring changes in ecosystems and studying wildlife patterns. Before the 2017 wildfires in California, Soundscapes to Landscapes had placed audio recorders in a park. Upon surveying the data after the fires, the team detected a “preponderance” of lazuli buntings (Passerina amoena), a species of songbird that had never been seen or heard in the park before the fires. Initially, the citizen scientist monitoring the park thought it was an error in the AI models. But later, he deduced that the songbirds preferred burned areas and might have flown in after the fires, helping the team understand how the fires changed the park’s ecosystem.
Audio data have been used for decades to monitor, study and conserve wildlife. In recent years, bioacoustics has gained prominence as a non-invasive way to study wild animals. It can be used to study entire landscapes and detect species, like Salas’s team does, but also to understand the behavioral and communication patterns of animals.
The ability of audio recorders to gather large amounts of data can make them more efficient than traditional camera-trapping and remote-tracking methodologies. A study published in the journal Methods in Ecology and Evolution in 2020 found passive acoustic monitoring to be “a powerful tool for species monitoring” that detected wild chimpanzees (Pan troglodytes) in Tanzania five times faster than visual techniques. Another study, published in the journal Ecological Indicators in 2019, compared acoustic recorders to camera traps, finding the former’s advantage to be its “superior detection areas, which were 100-7,000 times wider than those of camera traps.”
However, larger coverage areas mean larger amounts of data to parse through, making the analysis of sound data labor-intensive. Technological innovations such as artificial intelligence and machine learning have helped make the process easier. But conservationists say there’s still a long way for the technology to go to make the processing of audio data faster and easier.
Salas says the AI models used by Soundscapes to Landscapes often expose these technological gaps. In the past, the models have mistaken the sound of a motorcycle engine for the cooing of a species of dove, and confused the chatter of little girls with the sounds of quails. “There is immense capacity to monitor wildlife using sound data but the technology is not there yet,” he says. “My concern is [whether] it can happen fast enough so that we can start keeping a record of how the planet is changing.”
Related audio from Mongabay’s podcast: Cornell Lab of Ornithology’s bioacoustics lab assistant director Laurel Symes discusses how their program plans to use new funding to address gaps in the field, listen here:
Darren Proppe, who has been using audio data for years to study songbirds in Texas, says he’s “skeptical of AI without any human ground truthing.” Human intervention, he says, is necessary not only to spot errors, but to also raise larger questions that automated analysis can’t deduce.
“If I am just looking for the presence or absence of a bird or a mountain lion or an insect, then vocalizations can confirm that,” Proppe, the director of the Wild Basin Creative Research Center at St. Edward’s University in Texas, tells Mongabay in a video interview. “But the bigger question would be, what are you missing? And humans will have to really do some checking to make sure they are not being misguided.”
Accessibility to inexpensive real-time monitoring and data transfer is another concern when it comes to handling bioacoustics data.
It’s a problem Daniela Hedwig knows too well. As the director of Cornell University’s Elephant Listening Project, she and her team have for years been listening to and recording the African forest elephants (Loxodonta cyclotis) that roam the rainforests of Central Africa. As a keystone species, the elephants play a vital role in maintaining and shaping the structure of the forest. The data collected by the project are passed on to governments, which can use them to identify locations for conservation activities. The project also collects data that help track poaching activities by detecting gunshots in the audio. But the inability to conduct real-time monitoring, combined with inefficiencies in automated detectors, makes the process slow and laborious.
The data are collected from the recorders every four months, following which Hedwig’s team takes almost three weeks to go through and analyze the audio, which can often amount to 8 terabytes — about 1,100 hours of 4K-quality video streamed over Netflix. “The reason is that the detectors are not perfect, and we have to go through each detection, look at it and decide whether it was actually a gunshot or not,” Hedwig tells Mongabay in a video interview.
Conquering these challenges along with the incorporation of real-time monitoring, Hedwig says, will push bioacoustics technology further ahead. Given the immense interest the field has garnered in recent times, she says she’s optimistic.
“Just imagine anti-poaching units sitting in their control room, and they can get information on a poacher in real time and say ‘Hey, we need to send out people and catch them,’” Hedwig says. “That’s going to be the big game changer.”
Related audio from Mongabay’s podcast: Elephant Listening Project research analyst Ana Verahrami explains forest elephants’ role as keystone species for tropical forest survival, and plays acoustic recordings of elephant behavior that inform the project’s work, listen here:
Citations:
Crunchant, A., Borchers, D., Kühl, H., & Piel, A. (2020). Listening and watching: Do camera traps or acoustic sensors more efficiently detect wild chimpanzees in an open habitat? Methods in Ecology and Evolution, 11(4), 542-552. doi:
Enari, H., Enari, H. S., Okuda, K., Maruyama, T., & Okuda, K. N. (2019). An evaluation of the efficiency of passive acoustic monitoring in detecting deer and primates in comparison with camera traps. Ecological Indicators, 98, 753-762. doi:10.1016/j.ecolind.2018.11.062
Abhishyant Kidangoor is the 2022 Sue Palminteri WildTech Reporting Fellow, which honors the memory of Mongabay Wildtech editor Sue Palminteri by providing opportunities for students to gain experience in conservation technology and writing. You can support this program here.