Artificial intelligence probably wasn’t at the forefront of mind when the cranberry sauce was passed around the table this Thanksgiving. But chances are, AI might one day play a role in bringing one of America’s favorite meals to the holiday table.

“We’re focused on developing new strains of cranberries that have better productivity, or resistant to disease, or climate resilience,” says Jeff Neyhart with USDA’s Agricultural Research Service in New Jersey, who is using AI to streamline his research.

Within a changing climate, Neyhart says cranberries — which are produced domestically from Maine to West Virginia, and from Wisconsin to New Jersey — are increasingly facing environmental stressors.

“Cranberries are very sensitive to environmental extremes — heat stress, cold stress and even flooding. When you see cranberry bogs underwater, that’s not how they usually grow,” he says.

Based on future climate projections, Neyhart says environments where cranberries grow are expected to become hotter. There’s concern that crops won’t survive.

“We’re trying to head off that problem by developing and screening through new candidates of cranberry strains that are most tolerate to extreme heat. Breeding is a numbers game. We’re trying to screen through hundreds or thousands of new candidate strains. That takes a long time, and it’s very labor-intensive,” he says.

Related:Introducing Sunflare, Washington’s newest apple

That’s where AI comes in.

Streamlining research

Before AI, researchers would crawl on hands and knees through test plots, which are too small for mechanical harvest, to hand-harvest and count cranberries. It would usually take a crew of 12 people several hours to work through one plot.

These days, Neyhart’s crew sends a wheeled gantry cart outfitted with cameras through the bog. It rolls along taking photographs, which are fed into an AI model that’s been trained to estimate the number of berries and their size.

“Now, we can get through a few hundred plots, and do that in an hour and a half. It’s rapidly increased the scale of data we can collect,” he says. More than easing physical labor, the AI model lets researchers “react a bit more nimbly to new challenges.”

It typically takes about 20 years for a new cross to make it to a farmer’s field. AI can speed up that process.

Other research applications

AI is also used during the hotter months of the year, while the berries are growing. Canopy heat measurements are captured using a thermal imager during extreme heat events. The AI interprets which strains are cooling themselves better than others based on the photographs.

Related:Scientists shift from defense to offense to fight citrus greening

“We found genetic differences,” Neyhart says about the findings. Strains developed in the mid-Atlantic, farther south, are “able to better withstand heat than those from Maine or Michigan.”

While intuitive, Neyhart says findings like this empower growers and researchers with data so they can make better decisions.

And while research service’s AI is investigating cranberries now, it can be applied elsewhere. Researchers are working to apply the same practice to blueberries and, after that, generalize it to work on many other crops.

Neyhart says the AI was able to count blueberries accurately without having seen blueberries before. “Ideally, this research will be routine. We’ll use this technology either to augment or replace more manual ways to collect data in the breeding program.”