FAU researchers use AI to identify prey from predator crunching sounds

Interactions between hard-shelled molluscs such as clams and snails and their predators play an important but largely invisible role in shaping coastal ecosystems. These organisms stabilize coastlines, filter water, support biodiversity, and are the foundation of coastal health. But they are increasingly threatened by ocean acidification and increasing populations of predators that move and crush shells.

What makes these interactions particularly difficult to study is not only where they occur, but also how rapidly they unfold. Many predators, including highly migratory rays, forage in subtidal environments where direct observation is limited. As a result, the consumption of molluscs by predators, an important ecological process, remains difficult to quantify in nature, even though its importance has been recognized for decades.

Fortunately, these interactions are not silent. Crushed clams and crushed snail shells all create unique acoustic signatures. This is a short but informative sound that can be recorded underwater. Passive acoustic monitoring and autonomous recording systems allow researchers to “hear” these feeding events occur in real time. However, the challenge is how to reliably extract data from vast and noisy underwater records.

Researchers at Florida Atlantic University have developed a machine learning framework to improve the detection and classification of shell impact sounds in underwater recordings. The researchers used controlled aquarium experiments with white-tailed rays (Aetobatus nalinalii), a highly mobile predator known for cracking hard shells, to train the system to more accurately identify these feeding events amidst ocean noise.

The system uses a multi-step approach rather than relying on a single method. It first scans a large dataset to flag potential shell crunching sounds based on acoustic patterns, then applies a second layer of machine learning to reduce false positives by separating actual feeding events from background noise.

Once validated, the system will further classify the type of prey ingested using both traditional and deep learning techniques, such as random forests, long short-term memory networks, and convolutional neural networks (CNNs), which are trained to recognize subtle patterns in acoustic structure.

A key finding of the study, published in Ecological Informatics, was that you don’t necessarily need a highly complex AI model to achieve good performance. A simpler method using gammatone-based features was almost as effective as advanced deep learning systems at detecting shell-crushing sounds, yet required much less computational power. Our results suggest that these streamlined approaches may make long-term underwater monitoring in real marine environments more practical, scalable, and cost-effective.

“The sound of shell crunching contains an incredible amount of ecological information about predator-prey interactions and feeding behavior,” said corresponding author and research professor Laurent Cherbin, Ph.D., of the FAU Harbor Branch Marine Laboratory. “This study shows how passive acoustic monitoring can be used not only to detect these events, but also to better understand how marine predators interact with their environment in places that are otherwise difficult to observe.”

This approach brings scientists closer to remotely measuring shellfish predation rates in natural marine environments by detecting and classifying the sounds that predators make as they eat different types of prey.

“From an ecological perspective, this technology opens the door to quantifying predator impacts in ways that have not been possible before,” said senior author Matt Ajemian, Ph.D., associate research professor and director of the Fisheries Ecology Conservation Laboratory (FEC) at the FAU Ports Branch. “Being able to remotely detect and classify feeding events means we can go beyond isolated observations and begin to measure predation pressure on mollusk populations at the ecosystem scale. This represents a major step forward for coastal ecology and conservation.”

Importantly, this system is not only effective in controlled aquarium conditions, but also demonstrated strong performance in field settings using both animal-borne acoustic tags and fixed underwater recorders. Even when trained only on aquarium data, the model was able to reliably detect feeding events in natural environments and identify associated prey types.

“Beyond simple detection, our approach also provides insight into the predator behavior itself,” Ajemian said. “Acoustic patterns reflect not only prey type, but also coping strategies and processing times, raising the possibility that researchers will eventually be able to distinguish between individual feeding behaviors and even prey size classes based on these sounds.”

As shellfish aquaculture and coastal restoration expand, understanding interactions between predators and mollusk populations becomes increasingly important for conservation and management. Because the prey studied ranged from buried filter feeders to more mobile species, this system could be useful for tracking mollusk mortality across a wide range of coastal habitats.

“Our findings point a clear path toward scalable, real-time acoustic monitoring of marine ecosystems,” said first author Dr. Ali Ibrahim, assistant professor in FAU’s School of Engineering and Computer Science. “The computational efficiency of GTCC-based models makes them particularly suitable for autonomous underwater platforms with limited power and processing power, enabling long-term monitoring in remote marine environments where high-performance computing is not practical.”

Co-author of the study is Dr. Cecilia M. Hampton. student in the FEC laboratory at FAU Harbor Branch. Brianna C. DeGroot, Florida State University; Dr. Hanqi Zhaung is associate dean and professor in the Department of Electrical Engineering and Computer Science at FAU.

This research was supported by a Special License Plate Fund managed by the Harbor Branch Oceanographic Institute Foundation and a National Science Foundation grant.

– Fau –

About Florida Atlantic University:

Florida Atlantic University has more than 32,000 undergraduate and graduate students on six campuses along the southeast Florida coast. Recognized as one of only 13 institutions in the nation to earn three Carnegie Foundation designations – R1: Very high research costs and production of Ph.D.,””Opportunity Colleges and Universities,” and Carnegie Community Engagement Classification – FAU is located at the intersection of academic excellence and social mobility. Ranked among the top 100 public universities by U.S. News & World Report, FAU is also recognized nationally as a Top 25 Best-in-Class University and cited by Washington Monthly as “one of the nation’s most effective drivers of upward mobility.” If you would like to learn more please visit: www.fau.edu.