This isn’t to say that advances in agriculture efficiency over the past few decades aren’t valuable, but a push to more uniform systems can sometimes box out techniques that we might not understand. In inland West Africa, scientists have figured out how one of those techniques works.

Parts of inland West Africa are dry, with relatively little greenery, prone to drought. That can make growing crops difficult; farmers have to select certain crops and certain varieties of them that can handle the climate and soil. Water is incredibly precious, as it is in most of the world.

Richard Dick, a soil scientist currently at Ohio State University, noticed a certain woody shrub growing in a particular thirsty part of Senegal. Few other plants survived near the millet farmers had planted there, but this shrub did. It turns out that farmers had been allowing these shrubs—Guiera senegalensis, to be exact—to grow near crops for for centuries. Dick wondered how they could survive there, and why they had been allowed to.

It turns out that Guiera is a pretty remarkable plant; it practices something called “hydraulic redistribution.” Guiera has very very long roots, tapping water deep beneath the ground, as deep as 30 to 40 feet. Many plants intake water through their roots during the day, and excess water is let out through the leaves at night, through a process called guttation. But Gueira actually lets out excess water through pores in its roots, right near the earth’s surface. They are, basically, irrigation systems unto themselves.

Dick constructed a pattern of planting crops near Guiera plants to take advantage of this extra irrigation. By vastly increasing the amount of Guiera in a given plot of land—Dick says it’s pretty easy to grow—millet crops could increase by up to 900 percent, along with other benefits like quicker growth rates, healthier soil (which needs less fertilizer), and improved protection against erosion. It’s a pretty amazing project and plant, and who knows how many other strategies are out there, lodged in human memory?