Why the US military is listening to shrimp

Whale skeletons stand guard around the coastline of Fuerteventura in the Canary Islands, a stark reminder of the damaging effects of military sonar. Sonar from ships and submarines is thought to be one of the contributing factors to whale strandings, confusing the whales' own sonar and casuing them to beach themselves on the shore.

This whale-unfriendly technology, however, may soon have a rival. Lori Adornato, a project manager at US military research agency Darpa, believes we could detect submarines by paying more attention to natural sound than blasting out pulses of sonar.

"At the moment we treat all this natural sound as background noise, or interference, which we try to remove," says Adornato. "Why don't we take advantage of these sounds, see if we can find a signal?"

Her project, Persistent Aquatic Living Sensors (Pals), eavesdrops on marine animals as a way of detecting underwater threats. Current air-dropped sonar buoys – deployed by the military to detect enemy underwater activity – only work for a few hours over a small area because of limited battery life. The Pals system could instead cover a wide region for months. It could provide a near constant way of monitoring coastlines and underwater channels. Adornato says reef-dwelling species that can be relied on to stay in one place are likely to be the best sentinels.

"You want to make sure your organism always is going to be there," says Adornato.

Pals is sponsoring several teams looking at different approaches using very different reef species.

You might also like:

The record-breaking dive under the Arctic ice

The epic void that peered into the deep

Russia’s ‘slow motion’ Chernobyl under the sea

Laurent Cherubin is the lead researcher of the Grouper Guard team at Florida Atlantic University, working with goliath groupers. These fish, which can weigh up to 300kg (660lb), are common in US waters and produce loud calls to deter intruders.

"It's a loud, low-frequency boom," says Cherubin. "They are territorial and will boom at any intruder on their territory."

A booming grouper can be detected from 800m (2,640ft) away, though not every boom means a contact. As well as specific alert calls for intruders and predators, the grouper repertoire includes courtship sounds to attract mates threatening calls when staking a territory, and other sounds whose purpose remains a mystery.

The team is focusing on alert calls, much like listening out for a guard dog barking at intruders, says Cherubin. Distinguishing these calls from the others is not easy, so the team have set machine-learning algorithms to the task. These have been trained by listening to a catalogue of thousands of recordings until they can distinguish and classify different grouper calls.

The algorithm can then be turned into software which runs on a small but powerful processor built into an underwater microphone or hydrophone. An array of these hydrophones can cover a reef, listening to grouper calls and following them as the cause moves from one grouper territory to another.

Tapping fish conversations may seem outlandish; by contrast the work of Pals team at defence contractor Raytheon looks much more like traditional anti-submarine sonar. It does, however, have a twist.

"We are trying to detect the echoes that are created when shrimp snaps reflect off of the vehicles," says Raytheon scientist Alison Laferriere. "In much the same way that a traditional sonar system detects echoes from the sound that its source generates."

A bed of snapping shrimp emits a steady roar which Laferriere compares to bacon frying

In other words, it works like other normal sonar but using noise produced by shrimp rather than artificial pings. Snapping shrimp, also known as pistol shrimp, have been called the loudest creatures on Earth. They make their distinctive snap by closing their pincers so fast they create a vacuum bubble which collapses in a burst of plasma measuring thousands of degrees. This produces a flash of light and a shockwave powerful enough to stun prey.

Shrimp also snap to communicate with each other. A bed of snapping shrimp emits a steady roar which Laferriere compares to bacon frying.

"The signal created by a pistol shrimp is very short in duration and incredibly broadband," says Laferriere. "A single shrimp snap is much quieter than a traditional sonar source, but there can be thousands of snaps happening per minute."

Laferriere says the sound varies with the time of day and water temperature, but a shrimp colony is never quiet.

"One of the biggest challenges we've faced is dealing with the huge amount of noise created by the shrimp themselves and the reflections of all of those sounds off of the surrounding area," says Laferriere.

Interpreting these reflections is especially challenging, because, unlike traditional sonar, the location of the sound source is unknown. Again, the solution comes with modern software. Laferriere's team have developed smart algorithms to analyse the sound and pick out a single snap, first calculating the location of the shrimp, and then working out the path taken by the reflected sound and finally deducing where it was reflected.

To make sense of the returning sound, Laferriere's team had to create computer models to determine which echoes came off stationary background objects and could be ignored. Subtracting these highlights objects moving through the environment – these might be fish, submarines or unmanned underwater vehicles.

An ecosystem of permanently floating dispersed living sensors is appealing in principle – Sidharth Kaushal

Again, the finished solution will be an array of smart hydrophones with onboard computing, able to process shrimp sounds and determine the location of any targets of interest in the area.

Other Pals teams have ollowed similar approaches. Northrop Grumman's researchers are working on another shrimp-based sonar system and a Navy team is looking at general reef sounds and how intruders disturb them. All promise a cyborg sensor net covering wide areas for extended periods, with most of the hardware conveniently provided by nature. Only the hydrophones would need replacing or repairing.

"Darpa's approach would be a truly major breakthrough – if achieved," says Sidharth Kaushal, a specialist in naval warfare at UK defence thinktank RUSI. "An ecosystem of permanently floating dispersed living sensors is appealing in principle."

In principle, but not necessarily in practice. Kaushal is doubtful because previous projects using marine life to detect submarines were not successful. German U-boats were sometimes spotted by their effect on bioluminescent plankton, which emit a bright glow when disturbed; one in WWI was allegedly even sunk thanks to the effect. But later attempts to use the effect more widely, with special sensors looking for light sources over a wide area, made little progress.

"Cold War efforts by both the Soviets and the Americans to utilise them in a systematic way came to nothing," says Kaushal. "Partially as they had no way of differentiating false positives, such as the reaction from a passing whale, from the real thing."

How well Pals can distinguish a submarine from a shark remains to be seen. Adornato believes the combination of marine organisms and modern smart algorithms will provide a reliable "tripwire warning", to guide more traditional submarine hunters to check out a possible intruder.

Pals has already completed its initial feasibility phase, and the developers are now working on a second stage to demonstrate how well their solutions work in controlled tests this summer. Adornato says technologies developed for Pals could also be used for scientific research, by monitoring reefs and other underwater environments with a handful of sensors.

Tuning in to the sounds made by normal marine life would give researchers a low-cost, environmentally friendly way of tracking the impact of human activities underwater

"These low-impact, observational systems can be deployed to many different environments without disrupting the ecosystem nature has established," says Adornato.

Tuning in to the sounds made by normal marine life, and learning how they change, would give researchers a low-cost, environmentally friendly way of tracking the impact of human activities underwater. This would be useful for projects like off-shore windfarms, oil drilling, and seabed mining. All we have to do is listen to nature.

The project focuses on species which are common in US territorial waters, so it would not necessarily be easy to move it to other regions. However, the technology in general may be more widely applicable.

Pals has completed the first phase, which was a feasibility study for the two different approaches of listening in on how reef species react to intruders, and snapping shrimp sonar. Adornato hopes to carry out field testing in 2023. After that, if successful, the technology would be transferred to users (initially the US Navy) for development into a production system.

After that, we might see a sea-change in submarine detection, with be no more whale casualties caused by sonar. Rather than being a threat to wildlife, submarine-hunters might start working in partnership with the natural world, to the benefit of both.