ON A HOT Saturday morning in the Jordan Valley, Abu Ahmed stands on his date plantation, overlooking 5,000 palm trees aligned at regular intervals, like soldiers in formation. “All the palms are healthy,” he says. “Everything is fine.” He is relieved, because not too long ago, everything was not fine: an infestation of weevils had bore its way into his palms, destroying them from within. Typically, by the time the farmers discover the pests, it is too late. They are forced to cut down and burn one infested palm after another.

Around four years ago, Ahmed had lost five percent of his plantation to the red palm weevil, about 250 trees, and with them, five percent of his crop. He had put six years or more of work into each tree, the amount of time it takes for a palm to bear fruit. By then each tree was worth about $1,000, but the worst part for him was not the economic damage. “When a date palm dies, it feels like losing your little child,” he says. “You raised it, cared for it, fed it.”

Ahmed is by no means alone. The red palm weevil, or Asian palm weevil, has become a global problem. It has destroyed coconut palms in Southeast Asia, date plantations in the Middle East, and ornamental palms in southern Europe and California. In many cases, farmers do not discover the red palm weevil until it is too late. All they can do is cut down the palm and destroy it along with the pest, as Ahmed has had to do all too often. That may all change, though, thanks to machine learning and a new app for palm growers.

The app is called Palmear, and its inventor is Zeid Sinokrot, a 37-year-old engineer and entrepreneur from Jordan. Abu Ahmed was one of his first customers. On a recent morning, Sinokrot left Jordan’s capital, Amman, to visit the farm, the Jordan Valley stretching before him like an oasis. Some 90 percent of Jordan’s date palms grow here. “The Jordan Valley is the region’s breadbasket and vegetable garden,” Sinokrot says. Water scarcity is increasingly becoming an issue here, which makes controlling the red palm weevil all the more critical.

According to the Food and Agriculture Organization of the United Nations (FAO), the red palm weevil causes around half a billion euros of damage each year in the Mediterranean region alone. In countries like Jordan, the beetle threatens the livelihoods of date farmers and thousands of workers. The Jordanian date industry is comparatively small, with an annual yield of about 26,000 tons. But the crop is vital to this arid country: About half its date harvest is exported abroad, bringing important revenue into the country. The other half, meanwhile, contributes to Jordan’s food security, which has been under threat for years.

Sinokrot knows the farmers’ concerns all too well. His parents were in the date business. As a boy he spent a lot of time on farms in Jericho, on the west bank of the Jordan River. He used to watch as men held stethoscopes to palm trees, listening to their insides like a doctor with a patient. Every now and then, this method successfully detected the pest. But farmers were unable to contain the infestation in the long term. “Few people have trained hearing enough to detect the beetle larvae early enough,” Sinokrot says. “They couldn’t possibly listen to all the date farms in the region.”

SINOKROT STUDIED ENGINEERING in the United Kingdom. He later worked as an industrial engineer in Iraq, for eight years, manufacturing bricks. But the image of men with stethoscopes listening to the palms never left him. How, he wondered, could this method be translated into the 21st century and made accessible to the masses? “I wanted to develop a device that would be easy for anyone to use and that would give fast, reliable results,” he says.

When the Covid pandemic turned the world upside down, Sinokrot put all his eggs in one basket. He founded Palmear, turning a few rooms on a farm in the Jordan Valley into a research lab. He and his team wanted to teach an AI to recognize the sounds of red palm weevils, so they built soundproof audio booths, grew date palms, and injected larvae into them. Sinokrot and his team then recorded the sounds inside the palms.

“Training the AI was the hardest part,” Sinokrot says. They had to prepare it for all possible scenarios: What does a healthy palm sound like? What sounds do the larvae of the red palm weevil make? What other sounds might be confusing to the machine? It was only after three years of trial and error that they were able to bring Palmear to market. “Now the app is right about nine times out of ten,” Sinokrot says. “The more data we collect, the more accurate it gets.”

Sinokrot now wants to bring Palmear to the world at large. He and his team have already tested the app in Saudi Arabia, the United Arab Emirates, Qatar, Egypt and Malaysia. A few months ago, he met with foresters in Sweden who are battling the bark beetle. “Our method can be applied to any pest that bores into the inside of trees,” he says. To achieve this, they need to retrain the AI using audio data from bark beetles or other pests. “If we can do that, we can help millions of people around the world.”

ON HIS FARM, Abu Ahmed attaches his cell phone to the Palmear device, a slim box not much bigger than the phone. A cable connects it to a plastic wedge containing a needle the size of a matchstick. With two or three gentle pushes, Ahmed taps the tip of the needle into the palm. “The procedure is minimally invasive,” says Sinokrot, who stands nearby. “We want to protect the palm, not expose it to additional risk.”

The device connects to a pair of headphones. When you put them on, you are immersed in the palm’s acoustic interior. If the tree is healthy, a soothing murmur flows into the headphones. If a palm is infested, the audio signals reveal larvae moving and feeding through the plant; sometimes you can hear them smacking their lips.

Ahmed’s eyes are glued to the countdown on his phone screen. After 50 seconds, the app delivers a result: A green icon lights up: No sign of the red palm weevil. If the pest is detected early enough, farmers can target the plant with pesticides.

With the device in hand, Ahmed moves to a line of palm trees that worries him, which has has already treated with pesticide after the app gave him a warning. Again, he jabs the needle into the palm. Sinokrot puts on his headphones and listens. “The sounds are faint, but they’re there,” he says. When the countdown is over, the app displays an orange icon: “Suspicious.” For the neighboring palm, the app lights up red. He’ll have to treat the palm again. “I’d rather do it this way than have to destroy it,” he says.

“Ideally, we find the larvae two weeks after they hatch,” Sinokrot says. “That’s a safe window.” By the time the larvae have grown into beetles, it’s too late. The adults will crawl out of the palms, mate, and lay their eggs a little later in the shelter of the next tree.

To break the cycle, date farmers must check each palm every 45 days. With 5,000 trees, that’s an enormous effort. In other countries, plantations are even larger. Sinokrot sees this as an opportunity, not an obstacle. “Unemployment is high in the country. With our app, we can create new jobs.” Sinokrot’s customers include private farmers as well as the Jordanian Ministry of Agriculture. With each scan, users like Abu Ahmed enter new data that is automatically sent to decision-makers in Amman. The result is a map of the country’s red palm weevil hotspots, to which farmers at the small scale and the government at the large can respond quickly.

Sinokrot’s startup is not the only one taking up the fight against the red palm weevil. Farmers and experts around the world are relying on a wide range of tools, from thermal and satellite imagery to specially trained sniffer dogs.

The idea of sensors inside palm trees is not new either, says entomologist Ibrahim Al-Jboory, of the Arab Society for Plant Conservation. But a portable early detection device that anyone can use hasn’t existed before, he says. Al-Jboory sees potential in Sinokrot’s invention: “The method is quick, easy, and relatively inexpensive,” he says, adding that the accuracy of the app still needs to be tested. “But even if we could detect 75 to 80 percent of cases early, that would be a big step forward.”