We still haven’t reached peak oil. But peak milk happened in 2004, peak soybeans in 2009, and peak chicken in 2006. Rice peaked in 1988.

A new study published in Ecology and Society explains that 21 key resources that humans rely on—mostly food—have already passed their peak rate of production.

“Peak,” in this case, doesn’t mean that we’re actually producing fewer chickens or less milk yet. Instead, the researchers looked at the fact that the rate of production has plateaued, at the same time that population is increasing.

“We wanted to develop something like early warning signals,” explains Ralf Seppelt, a researcher from the Helmholtz Centre for Environmental Research in Germany, who worked on the study with colleagues from Yale University and Michigan State University.

“We approached the whole thing with an open mind,” he says. “We didn’t want it to be apocalyptic. We tried to seek patterns that give reliable information about how we really harvest the Earth, knowing that no one really wants to experience the time when the whole thing plateaus.”

The researchers analyzed production rates over time for 27 key resources, including some fossil fuels. But while they found that nonrenewable resources like coal, oil, and gas haven’t peaked, most foods have.

“We were actually surprised to find so many peak year signs—surprised that this is such a consistent pattern happening in the last 10-20 years,” Seppelt says.

The production rate is slowing across so many foods at the same time partly because each relies on the same limited resources, like land and water. Some foods rely on each other, like meat, an industry that uses around 70% of the grain grown in the U.S.

While the data could possibly be used to try to predict a final peak—the point at which total production might decline—the researchers say that’s difficult to predict. “To make a prediction, you need to make assumptions about interrelated resources,” Seppelt says. “We said, let’s stick with the data we have and see what it says.”

It’s also hard to predict innovations that might dramatically change our ability to grow food. “We know that we could expect an increase in yields of new varieties by 1-2% per year,” he says. “But nobody knows if there will be a disruptive innovation that brings us much easier production of animal proteins or even more productive plants.”

Since we don’t know if that radical innovation will happen, Seppelt argues that it makes sense to plan based on the assumption that we won’t necessarily be able to dramatically change production.

Still, he’s optimistic about society’s ability to meet the challenge. The world wastes 1.3 billion tons of food a year, and most of that waste can be avoided. In some parts of the world, yields can easily be doubled. And a shift to a more vegetarian diet would also make a huge difference.

Seppelt points to the fact that consumption of meat in the U.S. has started to decline. “I would love to dig into this data and to work with social scientists to understand why this is happening,” he says. “My assumption is that these changes occur because of demographic shifts, education, and lots of social influences. If we understand them and foster and promote them, this might lead to a quicker and very reliable change.”

The key, he says, is to stay focused on solutions. “The major part of the story is that renewable resources aren’t as infinite as we always thought,” he says. “We should carefully use them. I think we can do that.”