Bezos grant to fund AI innovations to monitor and protect wildlife

A $1.8 million grant from the Bezos Earth Fundwill help the Cornell Lab of Ornithology develop acoustic sensors and artificial intelligence analytics to identify real-time threats to forests from illegal activities and provide insights into ecosystem health.

The Cornell Lab was one of 15 recipients of the Bezos Earth Fund's AI Grand Challenge for Climate and Nature, to scale AI solutions that address the world's most pressing environmental challenges - from biodiversity loss and food insecurity to climate change.

The Cornell Lab, in collaboration with Mongabay; the Wildlife Conservation Society; Chemnitz University of Technology, Germany; and the Programa de Pesquisa em Biodiversidade do Pantanal and the Federal University of Mato Grosso do Sul, Brazil, will develop cutting-edge bioacoustic technology to advance conservation monitoring in the Global South, a missing link in today's biodiversity crisis.

"We are facing the biggest environmental challenge because of the scale of the changes and the speed at which they're happening - it's effectively a mass extinction that's happening within decades rather than millennia," said Ian Owens, director of the Cornell Lab.

"Right now, we simply don't know how wildlife populations are doing, particularly in some of the most biodiverse areas such as in South America, Southeast Asia and Africa. Satellite images help us monitor habitat change, but that's not enough to uncover what's happening across all species and research indicates that drastic declines are underway," Owens said. "We need to figure out what's causing the declines and how we can reverse them. We can't do that using traditional methods and support from the Bezos Earth Fund will help us unlock exactly the kind of efficient, scalable approach we need."

Traditional biodiversity monitoring relies on humans to count or observe changes in wildlife populations, but these methods often fall short at the scale needed for conservation. Traditional methods are also difficult to scale up, since only small portions of the land can be monitored at any given time.

"Technology is key to enable comprehensive and timely understanding of wildlife in these regions, and bioacoustics is the best tool for doing that," Owens said.

Bioacoustics, the study of animal sounds, is one of the "very few technologies that enable us to do what is beyond the scope of human observation," said Holger Klinck, director of the Cornell Lab's K. Lisa Yang Center for Conservation Bioacoustics. "The goal is to develop fast, accurate, scalable and affordable methods to monitor biodiversity using microphones to listen in on the world."

The research team from the Cornell Lab will address real-world conservation challenges using state-of-the-art bioacoustic technology in two highly threatened biodiversity hotspots: Guatemala's Maya Biosphere Reserve and Brazil's Pantanal wetland.

The Maya Biosphere Reserve, the largest remaining Mesoamerican tropical forest, is threatened by deforestation, narco-trafficking, fires and poaching. Traditional monitoring efforts have failed to keep pace with these threats.

"In Guatemala we will be working with a real-time version of this new acoustic recorder that we're developing with Analog Devices that will run and analyze data in real time in the field, and then report back what's happening," Klinck said. "It's this kind of groundbreaking technology that is needed to really protect the ecosystem and the people that work in those areas - no commercial product out there can do what we're proposing."

Farther south, in Brazil's Pantanal tropical wetland - the world's largest - the team will deploy new technology to monitor wildlife. The Pantanal is home to more than 80 species of mammals, 650 species of birds, 50 of reptiles and 300 of fish, as well as numerous threatened species such as the jaguar, the giant armadillo, the giant anteater and the hyacinth macaw. Unsustainable cattle ranching expansion and large-scale fires threaten this highly diverse ecosystem, which is difficult to monitor using traditional methods.

The team will develop the first foundation model for natural sounds, providing a fast and flexible tool for sound classification of multiple species across all types of habitats.

"This bioacoustic model will enable a much more comprehensive assessment of ecosystems, representing a turning point in biodiversity monitoring and conservation," said Larissa Sayuri Moreira Sugai, the interim assistant director at the Yang Center.

The project will be the first biome-wide ecosystem health assessment that has ever been conducted in the Global South using acoustics, Klinck said.

"Conservation cannot advance without a solid understanding of what is present and what is at stake," he said, "and these new tools will allow managers to act in support of biodiversity around the world."

Funding from Bezos Earth Fund will also allow the team to share information with local partners, to deploy the technologies to assist in conservation and decision-making.

"Bioacoustics has been the 'silver bullet' of hope for conservationists for some time, but right now it takes an expert team to do the work at scale," Owens said. "We're going to put together the hardware, software and interfaces to allow conservation decision-makers and researchers to use the technology easily to help reverse biodiversity declines."

Kathi Borgman is the communications manager for the Cornell Lab of Ornithology.