Read our COVID-19 research and news.

As the result of a restoration project that removed 22,000 tons of colonial-era sediment, Big Spring Run in Pennsylvania now snakes through lush wetlands.


A secret hidden in centuries-old mud reveals a new way to save polluted rivers

BIG SPRING RUN IN PENNSYLVANIA—Centuries ago, parts of the eastern United States were drowned in mud. Now, Robert Walter was dancing in it. The geochemist stood calf deep in this small stream 100 kilometers west of Philadelphia, thick curlicues of chocolate sediment flowing around his legs. Walter did a little jig as his colleague and spouse, geomorphologist Dorothy Merritts, watched. More mud stirred, heading downstream.

Brown water might not hold much interest for many researchers. But a dozen years ago, it catapulted Merritts and Walter to scientific prominence. The pair, professors at Franklin & Marshall College (F&M), showed that Big Spring Run and many other meandering, high-banked streams in the eastern United States look nothing like the low-banked, marshy waterways that existed when European explorers first arrived nearly 500 years ago. The original streams, Merritts and Walter argued in an influential 2008 paper published in Science, are now buried beneath millions of tons of “legacy sediment” that was released by colonial-era farming and logging, and then trapped behind countless dams built to power flour, timber, and textile mills. “We realized,” Walter says, “that the [streams] had been completely manufactured and altered.”

The finding challenged decades of conventional scientific wisdom and sparked pushback from researchers who said the pair had overstated its case. It called into question expensive efforts to restore rivers by using heavy equipment to resculpt them into what practitioners believed had been their natural shapes. And the work raised concerns that a massive, multibillion-dollar effort to clean up the nearby Chesapeake Bay would fail if planners didn’t figure out how to prevent massive slugs of legacy sediment, which also carries harmful nutrients, from sloshing down the bay’s many tributaries. “It was uncomfortable,” Merritts says, “because I knew that my colleagues had other ideas.”

New research is settling many of the debates that Merritts’s and Walter’s paper touched off. Although dams are not solely to blame for legacy sediment, it’s now clear colonial-era erosion did dramatically alter streams in much of the continent’s tectonically quiet eastern half, says Ellen Wohl, a geomorphologist at Colorado State University, Fort Collins. “There’s been an accelerated recognition of how ubiquitous this sediment is,” she says. And that recognition has been driven by Walter and Merritts, says Noah Snyder, a geomorphologist at Boston College. Their study is “one of the most influential papers I’ve seen.”

Now, the duo is hoping to inspire a new approach to stream restoration by turning back the clock at Big Spring Run. By removing centuries of mud, they have returned the stream to its marshy, precolonial glory, and are now demonstrating the environmental payoff such strategies can deliver.

Merritts and Walter weren’t the first to realize that erosion has clogged many U.S. stream valleys with sediment. In 1917, Grove Karl Gilbert, a storied geologist who studied western North America, revealed that gold mining in the late 1800s had caused sediment to fill and reshape deep river valleys in the California Sierra Nevadas. In the 1940s, Stafford Happ, a soil scientist at the U.S. Department of Agriculture, documented how silt eroded over centuries had buried and transformed Wisconsin waterways. In the following decades, two other researchers—geomorphologist Jim Knox and hydrologist Stanley Trimble—documented thick beds of legacy sediments beneath waterways in Georgia and the Upper Midwest.

“Agricultural erosion in parts of this country was far more severe” than many geologists realized, says Trimble, who recently retired from the University of California, Los Angeles. “We are talking about buried farms and villages.” Beaver, a small town in Minnesota, had been smothered by nearly 5 meters of eroded silt from uphill farms that reached the second floors of homes. Port Tobacco, Maryland, once a boomtown, faded after its wharfs silted up. But these pioneering studies never quite persuaded scientists that some waterways had been utterly transformed.

Much of our understanding of how rivers behave and evolve comes from two geomorphologists, Luna Leopold and Gordon “Reds” Wolman. While working together at the U.S. Geological Survey (USGS) in the 1950s, they studied streams in Virginia, Maryland, and Pennsylvania—all easily accessible from USGS headquarters in Washington, D.C. Using quantitative techniques rare at the time, they developed an influential explanation for how rivers form stable, braided, meandering channels and sculpt the land around them. As part of their work, they showed the importance of rivers frequently spilling over their banks during floods and depositing sediment on adjacent floodplains; such overbank deposition, they found, was a fundamental part of a natural, healthy waterway.

Dorothy Merritts and Robert Walter put their research on rivers to a real-world test by helping restore Big Spring Run.


Merritts, 62, says she grew up wanting to tell her own stories about the landscape. Raised in central Pennsylvania, she spent her childhood outdoors, climbing and hiking with packs of kids. Her grandfather, a conductor on the Pennsylvania Railroad, told her of the wonders he saw in the state’s valleys. “That’s what I wanted to do,” she recalls. “I wanted to be able to understand everywhere around me.”

After earning a doctorate in geomorphology from the University of Arizona in 1987, Merritts joined the F&M faculty. She became a field junkie, spending much of her career deciphering how plate tectonics had reshaped landscapes around the Pacific Rim by looking at how rivers had shifted over time. The work often took her to hazardous spots, including East Timor, where she needed a bodyguard because of a civil war, and Humboldt county in California, where she was threatened at gunpoint by cannabis growers. Until recently, Merritts carried two life insurance policies.

These hazards prompted her, in 2002, to look for a safer project closer to home. She had heard concerns about silt eroding from the banks of rivers flowing through farms in Pennsylvania, so she and her students began to survey local waterways. They appeared to behave in the ways that Leopold and Wolman had laid out. But the traditional model of river evolution couldn’t fully explain a picture that a student showed Merritts one day; it displayed a nearly vertical, 6-meter-high wall of layered sediments along the Little Conestoga River.

As it happened, Walter, who had recently arrived at F&M, was visiting as Merritts and the student discussed the photo. Now 69, Walter was born and raised in Lancaster, Pennsylvania, where F&M is located, and had spent days fishing nearby streams, but landscape evolution was not his focus. A specialist in the chemistry of volcanic rocks, he began his career dating the terrain surrounding the skeleton known as Lucy, the famed human ancestor discovered in Ethiopia. Still, one look at the student’s photo was enough to persuade him that the layers of sediment it showed had been deposited in still—not moving—water. “There has to be a dam there,” he said. There’s only one way to get that kind of deposit, Merritts adds. “A lake.”

Curious, the next day the two researchers journeyed to the Little Conestoga. Sure enough, just downstream from the towering bank of finely laminated mud they found the remains of a colonial-era milldam. That’s when Walter made a leap. “These are everywhere,” he said. “I bet all these streams come from these old dams.”

Merritts was doubtful. “I thought it was kind of crazy that you could [make that] leap from one outcrop,” she recalls. But subsequent trips to Lancaster’s historical society, along with reviews of other records, confirmed the dams had, indeed, been seemingly everywhere. On some rivers, settlers had built one every few kilometers. “It was,” Merritt says, “just astonishing.”

It was also disconcerting. The ubiquitous dams could mean many of the rivers that Leopold and Wolman had used to draw their conclusions had this unrecognized backstory, and so sat atop far more anthropogenic sediment than realized. It suggested that efforts to restore streams to meandering, high-banked single channels were misguided. And it implied that massive blankets of stored sediment could be a major source of nutrient pollution that would run downriver for decades to come.

Meters of mud had buried the rich, black soil (bottom layer) that typified Big Spring Run before Europeans arrived. The black soil contained seeds that revealed what plants had once grown along the stream.


The duo spent the next several years building its case, driving to dam sites and documenting and dating sediments. The collaboration also became a courtship, as the two scientists found they made both a scientific and personal match. They were married next to an old mill in 2004.

Together, Merritts and Walter make a formidable team, say those who know them. Merritts is meticulous, Wohl says, “just thorough and detailed and comprehensive.” Walter is more of a provocateur and discipline jumper. Their qualities are complementary, says Kathy Boomer, a river scientist at the Foundation for Food and Agriculture Research. “They’re the most collaborative and open-minded scientists I know.”

In January 2008, Merritts and Walter unveiled their ideas in Science. “The modern, incised, meandering stream is an artifact of the rise and fall of mid-Atlantic streams in response to human manipulation of stream valleys for water power,” they wrote. Ultimately, they concluded, the findings “imply the need to reconsider current procedures for stream restoration” that rest on “the assumption that eroding channel banks are natural and replenishable.” The paper quickly became the most influential of their careers, with some 750 citations.

Not all the attention was positive. “What surprised me was the resistance they met,” Wohl says. “People really had a hard time accepting this.”

In critiques later published in Science and elsewhere, some researchers faulted Walter and Merritts for implying that their findings, based largely on rivers in eastern Pennsylvania, where colonial mill dams were common, could be applied widely throughout the eastern United States. “I thought the conclusions far exceeded the evidence,” Trimble recalls. Other research, he and others noted, had found that legacy sediments had piled up even along river reaches that didn’t have dams. But until the couple’s paper came out, those studies had failed to gain broad traction.

Other scientists were irked by the suggestion that Leopold and Wolman’s iconic theoretical framework was flawed. “To say channel morphology is dependent on historic milldams is incorrect,” says Martin Doyle, a river ecologist at Duke University. “The classic understanding of how rivers work is still true.”

The real-world implications raised the stakes. River restoration specialists risked wasting heaps of cash on projects that might be quickly undone if floods pushed piles of old sediment into newly carved streams. State and federal agencies had to decide how to account for legacy sediments as they set water quality guidelines and environmental cleanup goals. And efforts to curb the supply of silt washing into the Chesapeake Bay might have to contend with far more of it than planners had counted on. “This is the 900-pound gorilla for how we restore our streams,” says Gregory Noe, a USGS ecologist who studies mid-Atlantic streams.

As the debate swirled, Merritts and Walter decided to put their ideas into practice. During their research, they had met Joe Sweeney, a farmer who owned land that encompassed Big Spring Run, and Ward Oberholtzer, an engineer at LandStudies, a river restoration firm. Sweeney had hired Oberholtzer to examine why trees he had planted on Big Spring Run’s high banks to prevent erosion were dying. The conclusion: Their roots couldn’t reach the groundwater; trenches dug by Merritts, Walter, and their students suggested several meters of legacy sediment caked over the site. To restore such connections, the team proposed re-creating the kind of languid wetland that Walter and Merritts believed had once existed on the spot. But first they would monitor it for several years, to establish a baseline that could be used to evaluate any postrestoration changes.

In 2011, after more than 2 years of planning and assistance from the Pennsylvania Department of Environmental Protection, the National Science Foundation, the Environmental Protection Agency (EPA), USGS, and others, bulldozers began to remove 22,000 tons of legacy sediment along 4 square kilometers of the valley. (The silt was trucked to F&M and used as fill beneath a new building.) A layer of rich, black, precolonial soil emerged from beneath the legacy sediment. In it, researchers found seeds that provided an archive of the wetland plants that had once grown along the stream. Although federal regulations required the restoration team to carve a single new channel, they built low banks and installed stumps and other obstacles that would encourage high waters to jump the banks, transforming the stream into a multithreaded wetland.

It took heavy machines to remove the thousands of tons of legacy sediment that had buried Big Spring Run. The sediment was ultimately used as fill beneath a new building.


Within 1 year, the banks bloomed with sedges like a Chia pet. Today, bog turtles scuttle and geese nest in thick native vegetation that has put down roots that hold sediment in place. There’s room for floodwaters to slow down and spread out, instead of sweeping away bankside trees and plants. “The biology does not have to re-establish itself” after every severe storm, Oberholtzer says.

Monitoring shows the restoration has also altered the stream’s biogeochemistry. Storage of organic carbon tripled in the restored area and levels of nitrate, a key pollutant, dropped sharply, soaked up by the wetland plants. The load of sediment swept downstream from the restored area declined drastically, by 85%, according to a USGS report published this year. Polluting phosphorus, which hitches a ride on silt particles, dropped 79%. Ken Forshay, a research ecologist with EPA based in Ada, Oklahoma, says he was skeptical he’d see such improvements. But the data have “turned a nonbeliever into a believer,” he says.

Even before all the results were in, the Big Spring Run project prompted similar restorations in Pennsylvania and Maryland, with 20 now completed and 10 more underway. It’s simple to see why: Though the project would have cost $1 million in today’s dollars to restore its 800 meters, it was at least 16 times more cost effective at reducing pollution than other techniques, found Patrick Fleming, an agricultural economist at F&M. “This practice blew the other ones away.”

Twelve years after their Science paper appeared, a clearer picture is emerging of how far beyond Big Spring Run the ideas floated by Merritts and Walter can be applied. Evidence that precolonial streams often resembled wetlands has popped up in more places—in Kentucky, for example, says Arthur Parola, a stream scientist at the University of Louisville. “The more we look, the more we’re finding,” he says. “These wetland systems were maybe the common types of streams in the eastern United States.”

In New England, however, Merritts and Walter found a different picture when they surveyed streams with Snyder. Although colonial dams did trap sediment, they found, the glaciated landscapes provided far less grist than those farther south. The thick beds of legacy sediment seen in the mid-Atlantic are “not going to be seen everywhere,” Merritts says. And “not every place had that many milldams.”

Noe found similar variation in a massive study of 68 river sites in the mid-Atlantic, now nearing publication. “There’s more nuance now,” Noe says. “Milldams are very important” in understanding sediment in some watersheds, he says, “but they’re not necessarily the causative factor everywhere.”

Noe’s study will also provide the first detailed, large-scale accounting of sediment sources and sinks for the region. The good news is that, at nearly all the rivers his team studied, the floodplains downstream were capturing as much sediment as was eroding upstream, potentially curbing pollution. The floodplains are acting as kidneys, he says, and are “water quality superheroes.”

But Noe adds that if those floodplains weren’t busy capturing colonial silt, they could instead be a greater sink for the sediment runoff from farms and cities. And the further removal of dams, as many states are pursuing, will only free up new slugs of mud. So legacy sediment problems aren’t going away, says Karl Wegmann, a geomorphologist at North Carolina State University. “It’s like Chernobyl. We’re going to be living with it for centuries.”

The question now is what to do about it. The Chesapeake Bay Commission, which leads the cleanup of the bay, is evaluating how to credit legacy sediment restorations for their pollution reductions, based on long-term data from project like Big Spring Run. It’s “been tremendously valuable,” says David Wood of the Chesapeake Stormwater Network, a nonprofit that coordinates restoration practices. This is “the type of research that is needed elsewhere across the watershed,” he says.

For their part, Merritts and Walter are pragmatic, not environmental romantics. They may have revealed a precolonial baseline for many waterways, but they know change is a constant of geology. Many rivers are so drowned in silt that they cannot be redeemed. The world is not pristine. But only by acknowledging and accounting for the legacy of the past, they say, can we take a first step toward solving the problems of today.

*Correction, 24 August, 5:15 p.m.: USGS headquarters in the 1960s was in Washington, D.C., not Reston, Virginia, as a previous version of this story stated.