The bigger issue in restoration, though, isn't single channel versus multiple threads; most rivers left alone will display sections of both. Rather, it's the question of strict human control over the intricate and gradual patterns of natural processes. Over-controlled rivers stop functioning, some scientists say, becoming what Colorado State University geomorphologist Ellen Wohl calls "virtual rivers," because floods and crucial processes like beaver damming and log jamming are thwarted.

The Clark Fork restoration plan aims for a delicate balance of control: Hold the banks in place until plants grow in, about 15 years, then allow the river to migrate slowly across the reconstructed floodplain. Moore and fellow University of Montana geologist Andrew Wilcox worry that the rigid banks will prevent spring flooding. While the reconstructed channel  may look natural, Wilcox says that it won't act like a natural river. Without floods, cottonwoods won't establish and weeds could choke out native shrubs; the diverse ecosystem that nurtures trout and migratory songbirds might never appear. However, if even a minor flood strikes before plants stabilize the banks, Martin says the river could scour out thousands of tons of arsenic and copper-contaminated mud. Toxic sediment released in a 2006 reservoir drawdown caused massive fish kills.

Silver Bow Creek, the Clark Fork's most contaminated tributary, presents a study in miniature of what Milltown might look like some years down the line. In 2000, Brooks Priest helped start a 25-mile, 11-year restoration project on Silver Bow. Workers excavated mine tailings, dumped hilltop soil on the floodplain, and carved a gracefully curved channel. Pockets of tailings remain, marked by aqua or rust-orange crystals. Even so, last year's restoration reach looks pretty good. Bunchgrasses, currants and sagebrush cover most of the soil. A few fish are back, and killdeer fly between ponds. Downstream, islands of bright green trees erupt from gray clay. State officials consider the Silver Bow Creek restoration a resounding success -- "almost a miracle," according to Greg Mullen, the restoration coordinator for the state's natural resources damage program.

But other restoration experts see plenty of room for improvement in the attempt to return Silver Bow Creek to full health. Karin Boyd, a geomorphologist and consultant on the project, wants to incorporate branching sections, to mimic the way beaver dams once split the creek's flow. In the project area, actual beaver are trapped and removed to protect new trees. Along the Silver Bow, restoration leader Priest surveys this year's restoration reach, with its fabric-armored creek banks and bulldozers stuck in the mire. "This will look how most people think a river should look ... but it's not a dynamic ecosystem." Dynamic rivers, with their wandering channels and complex floodplains, seem chaotic to people, but require less long-term maintenance and support more species diversity.

With a fleet of Bobcats and ATVs at hand, Priest speaks wistfully of the humble beaver, which evolved with North American rivers and is largely responsible for their pre-European settlement flows. Beavers by instinct follow the prime directive of process-based restoration: You can't make an instant river. What you can do is provide ample detritus and room for the river to wander and recreate its own stability and health. In the late 1990s, the Zuni Fish and Wildlife Department in New Mexico relocated 23 beaver to a reservation stream. Within three years, the beavers built a series of dams that raised the water table, flooding out invasive tamarisk and regenerating the willow forest. On Utah's Provo River and California's Cosumnes River, flood-control levees cut off the rivers from their floodplains. Over the past decade, state agencies have purchased floodplain land, then breached the levees. On the Cosumnes, native fish populations rebounded, flood risk declined, and willow and cottonwood survival has improved. Breaches and multi-thread channel reconstruction on the Provo show similar improvements.

But despite their success, such long-term strategies are generally out of sync with government funding cycles. Finding money to buy floodplains is difficult, so single-channel reconstruction "quick fixes" predominate.

Our predecessors streamlined crooked streams with dynamite and built thousands of dams. Now, some scientists and restoration practitioners envision a new kind of stewardship wherein we modify infrastructure to survive wildfires, droughts, and floods, then let natural processes take their course. But on most Western rivers, the dams, railroads, highways, and floodplain development that constrict river floodplains will remain. River restoration design will simply have to adapt. Ecological restoration experts wonder just how much we've learned from our earlier meddling.

"We might do a $10 million remedy at Milltown, then realize that's not where the river means to be," says Pat Munday, professor of history at Montana Tech University in Butte. "The question is, will we have the money to re-do it?"

Cleo Woelfle-Erskine has worked on river restoration projects in New Mexico, California, and Montana, where he now lives. He co-edited the anthology Dam Nation: Dispatches from the Water Underground, published last year by Soft Skull.


This article was made possible with support from the William C. Kenney Watershed Protection Foundation and the Jay Kenney Foundation.