Listening to the Reef: The Evolution of Ecoacoustics in Coral Reef Conservation
Underwater microphones and AI-assisted monitoring will empower protection of reefs from world's most destructive fishing method, reveals ZSL pilot
In a first-of-its-kind study, lead author Dr. Ben Williams, a long-standing Mars Sustainable Solutions (MSS) research collaborator, showcases how innovative marine monitoring tools are being advanced to aid conservation strategies and protect our precious coral reefs. Delivered as part of a long-term initiative through the MSS Thematic Research Programme - and in close collaboration with key scientific partners at University College London (UCL), the Zoological Society of London (ZSL), Hasanuddin University (UNHAS), Lancaster University, and Bogor Agricultural University (IPB) - this study is the latest milestone in a five-year research journey.
This ongoing effort is dedicated to increasing the efficiency, standardisation, and accessibility of high-quality data to accurately assess the ecological performance of coral reef restoration. By developing these automated methods, the programme addresses a critical barrier in marine conservation: the need to scale up reef assessment so it no longer depends solely on expensive diver time or highly specialised expert observation alone. This collaborative work has successfully evolved marine ecoacoustics into a highly scalable, AI-powered tool, utilising acoustic data to inform actionable conservation strategies.
As a prime example of this technology in action, this newly published research details how these acoustic systems can now be used to accurately detect and quantify illegal blast fishing. This capability is a genuine game-changer, providing the hard evidence needed to track and understand destructive practices that have long devastated reef ecosystems around the world. Sharing these insights reflects our absolute commitment to scientific transparency, ensuring we openly report on all findings from our restoration programmes and the science that underpins our work.
To understand how the team arrived at this real-time monitoring breakthrough, it is useful to look at the five-year scientific journey that made it possible. Turning marine ecoacoustics from an experimental concept into a scalable, real-world tool required overcoming three critical barriers: expensive hardware, overwhelming data volume, and the reliance on manual expert analysis.
Here is how our collaborative research systematically solved these challenges across four key phases:
1. The Acoustic Blueprint: Determining What a Healthy Reef Sounds Like
The initial step in this collaborative programme was to decode the biological orchestra of a thriving marine habitat. Healthy coral reefs are surprisingly loud, filled with the crackle of snapping shrimp and the low-frequency grunts of vocal fish.
In the early phases, the team led by Dr. Tim Lamont set out to study whether these soundscapes could serve as a reliable metric to monitor restoration performance—shifting ways of working away from labour-intensive visual dive surveys. By demonstrating that restored reefs recover their natural acoustic complexity within just a few years, the research proved that active ecological restoration has a rapid, detectable impact on marine community structures (Lamont et al., 2021).
Crucially, because soundscapes reflect the collective presence of diverse marine life, passive acoustic monitoring became a highly objective, non-invasive tool to track long-term restoration performance. However, actually scaling this breakthrough was limited by the high cost of commercial audio recorders, which restricted our data quantity and study replicability. To unlock the true potential of ecoacoustics, the team had to find a way to make the hardware affordable.
2. Making Ecoacoustics Accessible: The Low-Cost Hardware Evolution
To scale ecological monitoring and overcome this challenge, the team partnered with Open Acoustic Devices to co-develop and test a breakthrough solution capable of low-cost deployments across multiple locations simultaneously to capture the rich, variable sounds of the reefscape: the HydroMoth was born.
The HydroMoth, an open-source, cost-effective underwater acoustic recorder, was designed to overcome these barriers and is 20 times cheaper than other commercially available hydrophones. By testing this prototype in diverse marine environments, the team proved that high-quality acoustic data could be gathered using hardware that costs a fraction of standard commercial recorders (Lamont et al., 2022). This hardware evolution suddenly made continuous, localised acoustic monitoring accessible to community-led restoration programmes globally.
3. Scale Through AI: From Months of Audio to Minutes of Analysis
With low-cost hydrophones deployed across multiple restoration sites, researchers faced a new challenge: an overwhelming quantity of audio data. Analysing tens of thousands of hours of underwater audio to identify fish calls, shrimp snaps, or anthropogenic activity manually would take years of expert analysis.
To solve this, the team turned to machine learning (Williams et al., 2022). Instead of relying on manual classification, the team pioneered the use of unsupervised machine learning and pretrained neural networks. By teaching AI models to identify patterns in marine soundscapes without pre-existing labels, the team proved that unsupervised algorithms could process vast acoustic datasets with high degrees of accuracy Williams et al., 2025).
This software breakthrough is what allowed the automated detection of blast fishing in this latest research. What would have taken months of human auditing was processed rapidly by AI, turning raw noise into clear, actionable, and objective ecological insights.
4. Ecoacoustics in Action: Monitoring, Detecting, and Informing Conservation Strategies
The study led by Dr. Ben Williams represents the latest chapter in this scientific evolution, demonstrating how low-cost acoustic technology can be successfully paired with artificial intelligence to expose highly destructive marine activities at a regional scale.
To quantify what has historically been a highly underreported threat, researchers developed BombDetect—a machine learning system trained to recognise the distinct acoustic signatures of underwater explosions in Spermonde. From over 3,650 hours of recorded audio, the AI successfully detected and verified 3,567 blast fishing incidents—translating to an average of 23.14 blasts per day.
While these numbers underscore the urgent pressure facing marine ecosystems, this research offers objective, indisputable evidence of the true scale and frequency of blast fishing. This data can support how we design conservation strategies, providing regional coastal managers and policymakers with the precise information needed to implement targeted, high-impact protection.
Bomb fishing threatens dozens of reefs off the coast of Indonesia, and is thought to be the leading cause of reef loss on the Spermonde Archipelago. Over the last 20 years of working in the Archipelago, we’ve seen the difference that restoration can make – but restoration efforts struggle to keep up with such rapid rates of destruction caused by bombing. Immediate work to tackle bomb fishing is needed to future-proof these reefs, and our work shows how technology can ensure action is as rapid and effective as possible.
Professor Jamaluddin Jompa
renowned coral reef scientist and Rector at Hasanuddin University, Indonesia and co-author of the paper
This five-year journey is just one part of the broader Mars Sustainable Solutions (MSS) research programme. Working alongside our global network of research practitioners, our mission has always been bigger than any single technology. We are working to make coral reef restoration science-backed, community-driven, and standardised—providing the practical tools needed to safeguard our coral reefs at a global scale.
