Using citizen journalists and AI to preserve WETLANDS
Climate change is making it harder for researchers to monitor and help protect wetlands.
Our research group at McMaster University is meeting that challenge with the help of citizen scientists and leading AI experts.
Our Wetlands Ecosystem Research Lab has been monitoring the health of wetlands here in Hamilton and in Ontario’s Georgian Bay for decades. One of the ways we do our work is by casting nets and counting fish. In 2019, our nets came up empty in Georgian Bay. Unprecedented and extreme swings in water levels have drastically changed Georgian Bay’s wetlands in fundamental ways. The wetlands now have a new zone of vegetation made up of flooded dead trees and shrubs. This zone blocks fish from moving into traditional nursery habitats within the wetlands. The fish are now staying in deeper water that our nets can’t reach. We needed to rethink our research methods if we were to continue our work.
We’ve come up with three potentially game-changing ideas for environmental scientists conducting field research in critically important ecosystems like wetlands.
Instead of catching fish in nets, could we capture them on camera?
We custom-built camera stands and dropped them into 15 locations along the southeastern edge of Georgian Bay. Half the cameras on our stands faced out into the bay while the other half faced the shore. We collected nearly 1,000 hours of underwater video.
Already, the footage from our cameras is paying dividends. Video clips were used as evidence during a hearing of the Ontario Land Tribunal. The footage helped protect a fragile coastal wetland from development.
All of that footage inspired our second big idea.
Our lab has a long history of engaging students and collaborating with community members living along the shores of Georgian Bay in our research. Sorting through 42 days worth of video to spot and identify fish would take a record number of volunteers. This led us to Zooniverse, the world’s largest and most popular platform for people-powered research.
On Jan. 22, our "Where’s Walleye?" project went live. Citizen scientists from around the world are supporting our wetlands research by watching 10-second clips and noting what appears on video. On just the first day of launch, more than 300 volunteers classified 4,000 subjects, for a total of 19,000 classifications in a 24-hour period.
Citizen scientists will help us build a database that’s led to our third big idea. Could we use all that data to train AI to spot and identify fish in future video footage?
Just as cameras eliminated the need for nets, could AI replace the need for a global brigade of volunteers? We couldn’t find an easy-touse open access machine learning platform available to biologists. So we have reached out to leading AI researchers at Stanford University to see if they can build one to support this project. They have began using a few videos to test their machine learning model with promising results.
Our project underscores how advances in technology and knowledge-sharing are transforming how problems can be solved by researchers. All it takes is a little creativity and a willingness on the part of researchers to ask for help.
We are biologists who have cast a wider net of collaborators, welcoming coders, engineers and citizen scientists to help advance environmental science and protect wetlands close to home and around the world.