Coral reefs make a lot of noise. There is a low, constant hum of biological activity all around you, like underwater static made up of pops, clicks, thumps, and snaps. Anyone who has dived one knows this. It’s mostly made from shrimp. It comes from fish in some cases. And until not long ago, almost none of it could be identified in a scientifically useful way.
Things are beginning to change. Scientists from Cornell University, Aalto University in Finland, and the conservation technology nonprofit FishEye Collaborative have made a tool that can tell in real time which fish is making which sound in the wild. The name for it is UPAC-360, which stands for “Omnidirectional Underwater Passive Acoustic Camera.” It’s made up of spatial audio hydrophones and a 360-degree camera. It’s the same kind of setup that’s used to make VR content, but it’s now been modified to work underwater for the first time.
The results are very interesting. The team was able to connect sounds to 46 different types of fish on the coral reefs off the coast of Curaçao in the Caribbean using this system. It was known that more than half of those species never made any noise. Now, with these recordings, there is the largest collection of natural fish sounds that has ever been made public.
Take a moment to think about what that really means. For years, ecologists have been dropping sound recorders from the water onto reefs. They gather a huge amount of audio data—hours and hours of it. But figuring out what that data means has always been the hard part. Researchers think that more than 700 species of fish make sounds just in the Caribbean. There are hundreds of people from dozens of species living on the same stretch of reef, making it nearly impossible to figure out who’s doing what. This problem can be solved by the UPAC-360, which lets researchers see which fish made which noise by superimposing sound onto a 360-degree image.
A comparison is made to bird identification by Dr. Marc Dantzker, who is the lead author and executive director of FishEye Collaborative. The Cornell Lab of Ornithology’s Merlin app can now tell the difference between bird species just by listening to them. Ten years ago, this would have sounded like something out of science fiction. Along the same lines, create a big enough sound library and teach a machine learning model on it. Eventually, you’ll have a tool that can identify reef species by sound alone, without a diver or boat nearby. “We are a long way from being able to build ‘Merlin’ for the oceans,” Dantzker acknowledged, “but the sounds are useful for scientists and conservationists right away.”
It seems like the timing is very important here. Only 0.1% of the ocean floor is made up of shallow tropical coral reefs, but they are home to about a quarter of all marine species. They are also seriously declining because the water is warming up, there is pollution, and people are fishing too much. Governments and conservation groups are spending billions to protect and restore reefs, which are important for nearly a billion people who depend on them for food and income. Dantzker says that the problem is that it’s hard to tell if these interventions are really working without good monitoring tools.
That hole could be filled by acoustic monitoring. Sound is a reliable, continuous, and non-intrusive way to keep an eye on the health of a reef, but only if you know what it means. One of the study’s authors, Ph.D. student Matt Duggan, said that loud species like dolphins, whales, and snapping shrimp have long blocked out smaller biological signals. Most of the time, the smaller, quieter voices on a reef have not been looked into. That’s what this tool is starting to change.
The research group is already getting bigger. They are adding to the library in the Caribbean and are planning to send it to Hawaii and Indonesia in the next few months. It’s still not clear how quickly the machine learning part will be ready or if the method can handle the complexity of all the reef ecosystems in the world. But now there is a real way to figure out what a reef sounds like, and more importantly, what it sounds like when something goes wrong.
