Could fish sounds help scientists monitor ocean health?

As the day fades to evening over the coastline of La Jolla, Calif., a chorus begins under the ocean surface. In the kelp forests just off of the rocky cliffs, a symphony of fish sounds—from grunts and chirps to buzzes and burps that are heard during the day—reaches its crescendo just after dark.
 
It is well known that whales and dolphins communicate through sound. But the soundscape under the sea is far richer than commonly acknowledged, according to a recent research project funded by California Sea Grant. This research suggests that underwater sound recordings could prove a powerful new tool in efforts to monitor the health of our ocean ecosystems.  
 
“There’s a whole soundscape of different noises made by marine life,” said Ana Širović, who led the study as a researcher at Scripps Institution of Oceanography at the University of California San Diego. Širović started out her research career working on whale song, but soon discovered an untapped field of research in the sounds made by fish and invertebrate species.
 
It turns out that the acoustic signals of fishes may be as varied and important in their lives as for other species more known to be vocal. Yet until recently, researchers knew very little about what sounds which fish made, or when and why they made them. “We’re just starting to understand the importance of sound to these animals,” Širović says.

The ballad of the bocaccio

In one of the first long-term studies of fish sounds, Širović and a team of Scripps researchers including study collaborators Jules Jaffe, Ed Parnell, postdoctoral researcher Jack Butler, and PhD student Camille Pagniello, set up an autonomous system to record sound and images in the La Jolla kelp forest.
 
While acoustic recorders are readily available, autonomous underwater cameras were another story. Pagniello, who was funded as a Sea Grant trainee, took on the task of modifying a commercially available camera into a system that could be left alone with the acoustic recorders in the kelp forest.
 
With advice and support from Jaffe and his team of ocean engineers, Pagniello built a camera housing and battery pack that could be deployed underwater for up to two weeks at a time, and connected the camera to a tiny computer to set it to automatically take photos at intervals from dawn to dusk.
 
“I hacked the camera so that it would take photos from 5 a.m. to 9 p.m. each night,” said Pagniello. “The limiting factor was the power.” 

Pagniello ended up building a 3-foot-long battery pack housed in PVC piping to power the system for up to two weeks.

The researchers were surprised at how well the cameras worked in the low light conditions when the fish chorus reached its peak.

“This is really important because the water clarity is not very good even on the best of days,” said Jaffe. “We were astounded that even during dawn and dusk when there’s a lot of activity, the cameras worked pretty well and you could identify fish even meters away.”

One fish, two fish

The system snapped more than 224,500 photos in the summers of 2017 and 2018 and recorded 1,950 hours of four-channel acoustic data.

The next phase was to match sounds to fish, a complex exercise in data analysis given the massive volume of data. The bulk of the work was done by Pagniello and undergraduate interns who are still combing through the acoustic and visual data.

“In one period of 13 days, we got over 17,000 usable images, and were able to identify 9,700 fish of 16 species, as well as 1,243 fish that we couldn’t identify,” said Pagniello.

The researchers then made connections between the photographs and sound, building up a repository of observations to establish clear links between fish species and specific sounds. With multiple hydrophones, the team can tell where the sounds are coming from.

So far, the study identified sounds from at least six different species. Several of the species identified have not been previously documented to make sounds. The researchers are currently working on research publications to officially document these new discoveries.

Other sounds that the team identified were already known, such as garibaldi, which makes a thumping sound, bocaccio, which makes a purring sound, and plain midshipman, which makes both humming and grunting sounds. The recorders also detected sounds made by invertebrates, like snapping shrimp, and even the sounds of boats passing by. And the long-term data showed that the acoustic recorders could detect small differences between locations and populations.

The researchers also found that the soundscape changes from month to month and that sounds were likely primarily linked to spawning. Most of the sounds appeared to be mating calls.

Putting sound to work

The new study comes at a time when hydrophone technology and big data analytical methods are opening up a new realm of possibilities for passive monitoring of ocean life.

Researchers say that fish songs may prove to be not just a fascinating new area for marine biology but also a powerful tool for marine resources management. The La Jolla kelp forest is one of a statewide network of marine protected areas (MPAs) that the state of California implemented in 2012. But to track the effectiveness of these areas in protecting fish populations, researchers and managers need reliable, cost-effective tools, and methods to analyze large-scale data.

Acoustic monitoring could be one such tool.

“If you could simply deploy a network of hydrophones and leave them down for weeks at a time, you could gather continuous data in contrast to occasional visits by underwater observers,” said Parnell. “The continuous nature of sound observations means they could be more powerful to detect seasonality and longer-term trends. Such information would help us identify areas where fish are spawning, and improve resource management.”

The research team is currently working to create a catalog of common sounds and information on fish species found in the kelp forest off La Jolla. Once the fish chorus is better documented, acoustic monitoring could be used on its own to monitor kelp forest ecosystems.

“Kelp forests produce sounds like a symphony orchestra. You can hear which instruments are playing at any one time or even if the orchestra is a complete one. In a similar manner, we are trying to understand how the different sounds of the kelp forest reflect their state in time and space all in an effort to develop a unique perspective of the functioning and health of these diverse communities,” says Parnell.