The Amazon rainforest was alive with sound. It was 3 a.m. , and Ari Grele’s eyes were bloodshot from hours of staring at multiple screens, working furiously to ensure his team’s data flowed seamlessly through the pipeline. The Limelight Rainforest team had 24 hours to collect data from 100 hectares of dense, tropical ecosystem and another 48 hours to analyze it. Grele’s task was critical: to ensure the team’s diverse streams of data — image, acoustic and DNA — came together into one cohesive set that could be analyzed quickly. But the DNA samples arrived late in the process.
“We spent 12 hours straight just fixing formatting errors,” Grele recalled. “By the time we finished, it was 5 a.m. The final report was due at noon.”
Grele and the DNA team worked side by side, troubleshooting errors and recalibrating the data. There was no time for rest. As the team began drafting their sections of the final report, Grele processed the last data and sent it off, barely managing to catch 30 minutes of sleep.
“I couldn’t think about anything else,” he said. “I just had to focus on the task at hand.”
At the end of the 48-hour competition against six teams, Limelight Rainforest had identified more than 250 species and 700 unique plant and animal populations from their 24-hour drone deployment. And despite the immense pressure, the team’s efforts paid off. They were named first prize winners of XPRIZE Rainforest, earning global recognition for their technology that could revolutionize biodiversity monitoring.
is a $10 million competition designed to accelerate the development of technologies that could rapidly survey biodiversity in tropical forests. Limelight Rainforest’s solution — a drone-deployed system capable of capturing bioacoustic data, images and insect specimens — included a machine learning system developed by Grele that could quickly identify more species.
From Zoom to XPrize
Grele’s involvement in XPrize began far from the rainforest. In 2022, his advisor, Lora Richards, associate professor of biology and co-director of the Hitchcock Center for Chemical Ecology was in Ecuador with Tom Walla, professor of biology at Colorado Mesa University and the project lead for XPRIZE Rainforest. During their conversation at Richards’s field site near the Ecuadorian Andes, Lora was “bragging about” her grad student and mentioned the system Grele had developed and how it could help Walla’s efforts for the competition.
“You would need Grele to make this work,” she’d told Walla. Within a few weeks, Richards, Grele and Walla met remotely, a seemingly casual conversation.
“That Zoom call was a turning point for me,” Grele said. “It’s surreal how something so small had such a massive impact on my academic career.”
Joining the competition was a chance to apply his expertise in computer vision and data systems on a global stage, using cutting-edge technology to help conserve the world’s most vital ecosystems. And, Grele’s work with XPrize was in addition to his dissertation.
“When we started we said it was a sidequest,” Richards laughs, “but it was a big sidequest. And doing this XPrize work highlights Grele’s skillset in a different way than what published papers can do.”
Grele’s contribution to Limelight was key to the team’s success. He developed a system that automated the collection and analysis of vast datasets. He calls his system Bugnet.
“Imagine each data is a single balloon on a string,” Richards explained. “Grele was the one holding all the strings together. He combined citizen science, museum collections and online databases into a cohesive system that could identify species and assess biodiversity.”
In an extravagant awards ceremony in Brazil last November, the team accepted first prize. Although 22 were in Rio to accept the prize, more than 60 researchers, organized into sub teams, made up the Limelight Rainforest. Their success has far-reaching implications: by developing scalable technology that can survey and monitor biodiversity, the team has paved the way for more effective conservation efforts globally. Their work directly aligns with the United Nations’ Kunming-Montreal Global Biodiversity Framework, which aims to address the accelerating loss of biodiversity.
Among the team members, were other members of the Wolf Pack, including Ankita Shukla, Computer Science & Engineering assistant professor. Shukla developed an artificial intelligence (AI) learning model that can identify different frog species in the Brazilian rainforest by the sounds they make. Read more about Ankita Shukla's work.
Beyond the prize
Grele is excited about what these technologies will mean for biodiversity conservation. “There are lots of unknowns in the rainforest, but this tool gives us a way to start measuring what’s out there,” he says.
The ability to capture data on everything from insect populations to plant diversity is just the beginning. For Grele, this project represents both a career milestone and the launch of future endeavors in biodiversity research.
“I’ve worked with other teams before, but this was my first large-scale collaboration,” he said. “It’s been an incredibly positive experience, and I’ve grown so much as a researcher.”
As he looks ahead, Grele plans to continue developing Bugnet and will be making it open source. He hopes to take his expertise now to a fellowship in Wisconsin.
