During one of California’s most severe droughts on record, juvenile coho salmon survived in small creeks of the Scott River watershed at rates several times higher than in normal years. The difference came down to a simple intervention: hand-built structures designed to mimic beaver dams.
In the summer of 2015, a team from the Scott River Watershed Council, led by executive director Betsy Stapleton, drove to a small tributary of the Scott River in far northern California. They brought willow branches, wooden posts, and buckets of clay. Over the course of one Saturday, they built two small dams by hand across Sugar Creek, about 500 feet upstream from where it meets the main river.
The structures were not large. The crew hammered posts into the stream bed, wove willow branches between them, and packed gravel, straw, and clay into the upstream side until water began to pool behind them. The materials cost a few thousand dollars. A conventional state-funded concrete pond project would typically cost well over one hundred thousand dollars.
At the time, the idea faced skepticism. State regulators approved only six of the 36 structures the council initially proposed. Local ranchers watched from a distance, ready to sue if water crossed onto their pastures. The main fisheries establishment regarded the project as a cheap trick that could not replace traditional hatcheries and concrete infrastructure.
To understand why these small structures mattered, it helps to know the history of the Scott River. Two centuries ago, fur trappers called this valley “Beaver Valley” because its 50-mile floodplain was a connected network of beaver dams and ponds. Within a decade, trappers had hunted the beavers to near-extinction in the region. The dams decayed, the wetlands disappeared, and with them went the natural systems that once kept summer water cool and slowly released winter snowmelt.
Later, gold mining in the 1850s reshaped stream beds, turning slow, winding wetlands into straight, shallow ditches. Then came hay fields and cattle grazing, along with groundwater pumping that could leave the river disconnected from itself in late summer.
What was lost was not just the beavers themselves but the ecological memory of how the system worked. Tribal communities in the region, including the Shasta and Karuk peoples, had maintained knowledge of how beavers and salmon interacted for thousands of years. Treaties in 1851 disrupted that knowledge. By the time coho salmon in the Scott River were listed as threatened under the Endangered Species Act in 1997, most biologists working on the issue had never seen a river functioning as it had before the fur trade.
Betsy Stapleton was not a federal official. She worked for a small nonprofit. During years of watching the river dry up each summer, she noticed something unusual in one stretch of the main channel that should have been completely dry. There was still water in it, enough to shelter small fish. Upstream, she found a small cluster of natural beaver dams, built by one of the few golden beaver families that had survived a century of trapping.
Around the same time, fisheries scientist Michael Pollock at a federal lab in Seattle had been trying for a decade to draw attention to the same connection. In 2004, he estimated that removing beavers from one Washington river, the Stillaguamish, reduced coho production by 61 percent because of a single missing rodent. Unlike Chinook salmon, which head to sea within months of hatching, coho salmon spend a full year in freshwater. That year makes the loss of beaver ponds particularly damaging for coho.
In the late 2000s, Pollock and a colleague named Chris Jordan built their own beaver dams on a degraded Oregon creek called Bridge Creek. These were not exact replicas. They were designed to be cheap and slightly porous, meant to slow water long enough to attract real beavers. After seven years and 35,000 tagged fish, steelhead juvenile production increased by 175 percent. Real beavers began moving into the fake dams and building their own.
By 2014, state fisheries crews in the Scott and Shasta rivers were manually relocating an estimated 116,000 juvenile coho salmon from pools that were drying up. It was an unprecedented rescue effort, yet neighboring fish populations continued to collapse.
In 2015, the first two beaver dam analogs were installed in the entire basin on Sugar Creek. They performed two functions. They slowed surface flow, creating shallow ponds. Because those ponds sat over porous gravel and mining debris, they also drew cool groundwater up from below, a process ecologists call “upwelling.” In August, upwelled water can be 15 to 20 degrees Fahrenheit cooler than surface water. For a fish the size of a finger that needs water below 70 degrees to breathe, that difference can mean survival.
Convincing local ranchers took time. By the second summer, monitoring wells around Sugar Creek showed groundwater levels had risen by about three feet. Downstream farmers found water returning to their channels earlier and staying longer. The council’s executive director later said she and her colleagues had previously been what she described as “secret beaver supporters.” That era ended. She said ranchers began asking them to build more of the structures.
By 2018, there were ten beaver dam analogs across five sites. Then something unexpected happened. Real beavers began appearing above the fake dams, some maintaining them, some rebuilding them, and some building new dams downstream. None had been imported. They were descendants of the families that had survived trapping all along. The researchers had thought they were buying time for the fish. In fact, they were inviting the animal that should have been living there all along.
The study published in 2026 in Frontiers in Ecology and Evolution, involving Pollock and colleagues, was designed for normal conditions. That is not what happened. By the time of the first installation on Sugar Creek, California was breaking modern drought records. Tree ring data from 1,200-year-old blue oaks, analyzed by climate scientists Daniel Griffin and Kevin Anchukaitis, showed that the 2012–2016 drought was the worst combined heat and dryness in California in at least a millennium. For juvenile coho salmon, this was extinction-level weather.
The results stood out clearly. Sugar Creek, which had dried up nearly every summer before restoration, did not dry up in 2015. Elsewhere in the sub-basin, surface temperatures exceeded lethal thresholds for weeks. Juvenile coho survival in French Creek averaged about 8 percent before restoration. After restoration, it reached 60 percent. In Sugar Creek, it reached 55 percent in years when neighboring streams lost everything.
The two Sugar Creek dams supported an estimated 6,700 juvenile salmon in about three-quarters of an acre of water. The smaller French Creek pond held another 1,700 fish. The team ran Bayesian statistical analyses using snowpack as a drought indicator. Restored streams outperformed unrestored streams in every model, and the gap widened in the worst drought years.
The Scott River is one of the last coho rivers in California where the fish are almost entirely wild. In the neighboring Trinity River, more than nine out of ten adults come from hatcheries. The standard tools of recovery—trucks, concrete, reservoirs—were not what kept the Scott River going.
The salmon were not the only thing that returned. Willows and sedges flourished along the banks. Waterfowl returned to the ponds. Parts of the simplified river began behaving like small wetlands. The structures did not just support salmon. They brought back the beavers, and the beavers brought back everything else.
Adult coho returns to the Scott River fluctuated in the following years: about 1,766 in 2020, 858 in 2021, and 238 in 2022. Salmon runs naturally vary. What changed was the lower bound.
Today, at the confluence of Sugar Creek and the Scott River, there is a visible dividing line. Downstream, in most late summers, the main stem of the Scott becomes a thin, warm thread of water. Upstream, at the structures installed eleven years ago, there is a small pool connected by willows and posts. Beavers have since taken over maintenance of the human-built structures, and every site with a beaver dam analog in the Scott basin now hosts at least one family of wild beavers. None were introduced artificially.
In recent years, Sugar Creek, with its three-quarters of an acre of standing water, has contained more than two-thirds of the entire sub-basin’s juvenile coho production. There are now about a dozen beaver dam analog sites across five sub-basins. Permits that once took years now take months. Of the four monitored coho tributaries in the Klamath basin, the Scott was the only one that held through the drought.
Pollock was direct about what the data shows. These sites are now producing coho salmon in old mining sediment areas that have not seen a beaver in over a century. The team is careful to note that the structures coincided with these responses rather than necessarily caused them. Ocean conditions and snowpack matter. The key point is the ratio. In a river that had lost almost all its ecological memory, a small number of posts, some willow branches, and about 40 pounds of clay made the difference between having a functioning salmon population and having none.
The story did not end at the Scott River. In 2022, California established its first beaver restoration program under wildlife biologist Molly Alves. In October 2023, the state released its first live beavers in nearly 75 years, a family of seven on Maidu tribal land in the northern Sierra Nevada. Another 28 animals followed at five sites with the Tule River Tribe and the Maidu Summit Consortium. Two litters of kits and three new dams have been recorded. At the most productive site, surface water expanded by 23 percent in one season.
None of those relocated beavers went to the Scott River. The river did not need them. Its beavers returned on their own from a population that was never completely eliminated. In 2024, Assembly Bill 2196 made the program permanent. The Scott River was the pilot project that made all of it possible.
Between 2023 and 2024, the last four hydroelectric dams on the lower Klamath River were removed, the largest dam removal project in U.S. history. About 420 miles of salmon habitat were reconnected. Within weeks, Chinook salmon began spawning in tributaries that had not seen them for generations. Upstream of the former Copco 1 dam, built in 1918, the Shasta Indian Nation regained 2,800 acres of its ancestral land.
Researchers on the Scott River are honest about the limits. A dozen sites across an entire basin. Groundwater rights on hundreds of farms remain unchanged. Chinook salmon still cannot always move upstream in the worst years. Climate models from NASA’s Benjamin Cook put the probability of a medieval-scale megadrought in the American Southwest above 80 percent by the end of this century. The fish are not safe. The Scott River has not been saved. It is still holding on.
What has changed is the sense of what is possible. A small watershed council, federal scientists, and a few ranching families convinced a fish that evolved over millions of years to come back. Not with a hatchery, not with a dam, but with willow branches, clay, and a rodent that returned to a river that finally had what it needed.
