If a tree falls in a forest and no one is around to hear it, does it make a sound? Today, we know the answer is a resounding yes -- thanks to a bio-acoustic monitoring system that uses artificial intelligence (AI) to identify sounds in the wild. This technology is helping conservationists across the globe reduce rainforest destruction, improve biodiversity, and boost conservation efforts.
Putting in a Group Effort
The Mammal Society in the United Kingdom along with Huawei, the nonprofit technology startup Rainforest Connection (RFCx), and researchers from the U.K.’s University of Bristol is launching a project this autumn that is designed to monitor red and grey squirrel populations in the country. The program is crucial to the conservation efforts of endangered red squirrels. The red squirrel population in the U.K. is estimated between 120,000 to 160,000, according to the Woodland Trust, down from 3.5 million. The population in England alone is especially dire, with numbers as low as 15,000.
This isn’t RFCx’s first foray into bio-acoustic monitoring, saysChrissy Durkin, the nonprofit’s director of international expansion. It began its work back in 2014, using sensors to identify and thwart illegal logging in the Brazilian rainforests. The proprietary system called Guardian, is “basically a mini-computer in a weatherproof box with solar panels, an antenna, and a microphone,” explains Durkin.
“Guardian records the soundscape continuously, and streams data collected to the cloud via the cell network. Then we run artificial intelligence models, identifying specific sounds and sending alerts to people on the ground” she said.
Originally, the Guardian system was listening for chainsaws. With the U.K. project it will listen for squirrel sounds to identify where the different species populations are located. The project will use Guardian systems along with dozens of less expensive offline devices that will simply listen for and record forest sounds. The benefit of using both streaming and non-streaming technologies is the ability to add additional sampling sites that can be spread out extensively across the ecosystem. It also gives researchers a better idea of where the squirrels are concentrated so that the live, connected devices can be brought to those areas.
Red squirrels are native to the U.K. and have been around for more than 10,000 years. Grey squirrels have a much shorter tenure. They were introduced to the country in the late 1800s. The two species don’t co-exist well, with the greys tending to edge out their smaller, red cousins, but the problem is more complex than a battle for territory, said Dr. Stephanie Wray, the chair of the Mammal Society, a nonprofit focused on science-based conservation.
“That’s why we are excited about using AI in this project. Being able to capture so much data will allow the researchers to analyse the complex environmental factors that allow red squirrels to persist and help us to conserve them,” she said. The technology has applications for other species as well. “Extinction rates are increasing at a speed we’ve never seen before, and there are more than a million species considered at risk of extinction in the world at the moment,” Wray added.
The Guardian technology and others like it can help researchers gather the data they need to rein in extinction. Once researchers know where the red squirrels are and their numbers in relation to grey squirrels,effortscan be made to help boost the red squirrel population. Strategies include breeding programs, creating safe havens, trapping grey squirrels, educating the public, and reintroduction of red squirrels to areas where populations have thinned out.
“As conservationists, we need to restore ecosystems and reestablish the natural processes that keep the natural world in balance. We can use technology to create measures of biodiversity based on sound. Basically, we filter out all the noise made by humans, then what’s left – the noise made by nature – is a measure of biodiversity.
This approach will help us to understand what works and where our interventions are being successful, not just for red squirrels but for whole ecosystems,” Wray said. “Such proxy measures that relate to how well we’re restoring nature -- how biodiverse a habitat is -- are going to be hugely important for conservation and restoration work over the next decade.”
Changing the Pace – and the Face of Conservation
Although bio-acoustic monitoring systems have been around for about a decade, it’s only recently that there is enough computing power, cloud storage, and interest to bring them to a wider audience, said Ronald Schmelzer, a managing partner and principal analyst at Cognilytica.
Schmelzer said that the quality and quantity of data that is used to train what’s called supervised learning models – a machine learning task that trains a computer based on input data – has improved significantly, paving the way for researchers and conservationists. “All these innovations that we’re seeing are mostly innovations around getting access to these data sets and doing a better job of cleaning them up, and of training these systems,” he said.
RFCx’s Durkin said the ability to do pattern matching and the power of having cloud storage is opening up new avenues for conservation. “We’ve worked with Huawei on projects around the world with all sorts of different use cases. Just in the last year, we did projects in Chile, Malaysia, Greece and the Philippines…we can detect sounds of things like gunshots… to alert Rangers on the ground to potential poaching. There’s a ton of different applications of the technology,” she said. “Our goal is to build this unified global network of in-depth understanding of ecosystems via these regional AI models for biodiversity.”
It’s this potential for real change – and good – that’s drawn Huawei to supporting this work, said Karl Song, the company’s VP of Corporate Communication. “Huawei believes in the power of technology to help protect the planet and we’re committed to supporting conservation projects in many ecosystems, including oceans, rainforests, and mountains,” he said. “Combining our technologies with the expertise of our partners can address biodiversity challenges and transform how we protect vulnerable species.”