Russell Mapes was trying to determine how fast age-old Amazon River sediments travel from the Andes Mountains to the Atlantic Ocean when he uncovered a bizarre truth — the Amazon once flowed in the opposite direction.

Mapes found ancient zircon grains in sedimentary deposits in the center of South America that could only have come from the eastern part of the continent. If the Amazon had always flowed west to east, as it does today, Mapes would’ve found much younger minerals in the sediments from the Andes.

“But we didn’t see that,” he says. “All along the basin, the ages of the mineral grains pointed to very specific locations in central and eastern South America.”

How could this happen?

Mapes’ advisor Drew Coleman says geologists suspected that sections of the Amazon had flowed westward but lacked the data to prove it. “But the Amazon is so flat that a tilt in either direction can change things drastically,” Coleman says.

Mapes says that during the Cretaceous Period — between 65 and 145 million years ago — what are now South America and Africa separated and formed a highland area in northeastern South America. The highlands caused river sediment to flow westward before the Andes were formed.

Also during the Cretaceous Period, a ridge called the Purus Arch rose in the middle of the continent. As the eastern highlands cooled and eroded over millions of years, water began flowing from the arch to the Atlantic. Meanwhile, water on the west side of the arch began flowing toward the Pacific. But as intense seismic activity caused the Andes to grow, water and river sediment began filling the basin between the mountains and the arch and formed a huge lake.

The Andes continued to grow and send sediment into the basin over the course of tens of millions of years. Eventually the water and sediment breached the arch, resulting in the Amazon’s current flow pattern from deep within the Andes to the Atlantic.

“We think this last change happened within the past five to ten million years, which is really fast, geologically speaking,” Mapes says. “This shows how transient the surface of the earth is.”

As Mapes continued to analyze mineral ages, he found another nugget of information. Sediment at the Amazon’s Atlantic headwaters contained zircon from the entire river basin, not just from the Andes, debunking assumptions about how ancient landscapes evolve and river basins erode.

For example, geologists studying zircon samples in Arizona and New Mexico hypothesized that almost all of the zircon there came from the Appalachian Mountains some 160 million years ago.

“I just didn’t think that made sense,” Coleman says. “And it’s easy to test. We can go to a big river and see what a big river does.”

And that’s what sent Coleman and Mapes to the Amazon to begin with.

“Some geologists think that 90 to 95 percent of the sediment coming out at the end of the Amazon is from the Andes,” Mapes says. “But they’ve been studying the suspended sediment, which is the stuff that makes the river muddy. And most of that does come from the Andes. But a much higher percentage of the stuff moving along the river’s bottom comes from the old flat part of the continent — the stuff that’s more than a billion years old.”

Russell Mapes is a geology doctoral candidate. Drew Coleman is an associate professor of geology in the College of Arts and Sciences. They received funding from the American Chemical Society.