A federal program could fix Washington’s salmon-killing culverts

In the early 1900s, Hardy Creek was throttled by BNSF Railway, the United States’ largest freight railroad network. When the company built its Columbia River line, engineers routed Hardy Creek under the tracks via a culvert — a 2.5-meter-wide arch atop a concrete pad. The culvert, far narrower than Hardy Creek’s natural channel, concentrated the stream like a fire hose and blasted away approaching salmon. Over time, the rushing flow scoured out a deep pool, and the culvert became an impassable cascade disconnected from the stream below — a “perched” culvert, in the jargon of engineers.

“It’s an obvious barrier,” says Peter Barber, manager of the Cowlitz Indian Tribe’s habitat restoration program. “A fish would be hard-pressed to navigate through that culvert.”

The strangulation of Hardy Creek is an archetypal story. Culverts, the unassuming concrete and metal pipes that convey streams beneath human-made infrastructure, are everywhere, undergirding our planet’s sprawling road networks and rail lines. Researchers estimate that more than 200,000 culverts lie beneath state highways in California alone, nearly 100,000 in Germany, and another 60,000 in Great Britain. In Europe, they thwart endangered eels; in Australia, they curtail the movements of Murray cod. In Massachusetts’ Herring River, snapping turtles lurk in culverts to devour passing fish, largely preventing herring from spawning. Taken as a whole, these obstacles are a major reason that three-quarters of the world’s migratory fish species are endangered.

Compared with dams, however, culverts have historically escaped public attention; most people drive over them every day without noticing. “I used to tell people I assess culverts,” recalls Mark Eisenman, a planner at the Alaska Department of Transportation. “They’d say, what the hell’s a culvert?”

In 2022, however, the U.S. Federal Highway Administration launched a $1 billion program to replace culverts that block oceangoing fish on streams like Hardy Creek — among the largest pots of money ever devoted to these humble pipes. Fixing the countless barriers that underlie infrastructure, according to Barber, is “one of the best ways to restore our salmon runs locally.” But given the sheer scale of the culvert crisis, even a billion dollars will only go so far. Can we repair our faulty culverts while there’s still time to save sea-run fish?

Roads have impeded fish since well before the proliferation of automobiles. In 1893, a log drifted down a tributary of British Columbia’s Fraser River and wedged itself in a wagon-road culvert, preventing sockeye salmon from migrating. So many fish crowded against the jammed pipe that they writhed onto the road, obliging an inspector to “engage men and teams to cart the salmon in order to keep the road clear for traffic,” per one reporter. In 1932, engineers in Arlington Heights, Illinois, found a pair of pickerel caught in culverts, and reportedly “enjoyed a good fish dinner.”

But the real problem wasn’t the few fish caught in culverts — it was the millions who couldn’t swim through them. Early engineers hadn’t given fish a moment’s thought; one 19th-century road-making manual suggested that culverts simply be “large enough to admit a boy to enter to clean them out.” As road systems metastasized throughout the 20th century, the so-called boy standard proved disastrous. As in Hardy Creek, culverts cinched flows into torrents too powerful for adult fish to swim against. Cramped and perched culverts also prevented juveniles from flitting up and downstream in search of food and refuge. Even as conservationists fretted about megadams and overfishing, roads and railroads were secretly sapping the vitality of coastal streams.

By mid-century, a few scientists had begun to pay attention. After a 1949 report from the Washington Department of Fisheries observed that culverts had created a vast “lost frontier” of foreclosed fish habitat, scientists in Oregon and Washington began retrofitting culverts. Their first attempt was to deploy baffles, rungs that slowed torrential flows into gentle pools, along with angled ramps called fishways. Soon, though, they realized that replacing culverts altogether was far better than retrofitting them. If most culverts failed fish because they were too small, the solution was simple: swap out narrow pipes with wider ones (or, even better, bridges).

Fish biologists came to tout a design known as the stream simulation model — a culvert wide enough to accommodate a waterway and its banks, even during floods, without altering its flows. “The idea is that the river doesn’t know that it’s going through the pipe,” says Eisenman. A gravel floor, rather than a concrete or metal one, completes the effect. “In theory, the fish don’t know when they’re swimming through, either. It just gets a little dark for a while.”

A few autumns ago, I saw an impressive “stream-sim” culvert in action at Little Skookum Creek near Olympia. In 2001, a coalition of 21 Indigenous tribes had sued the state of Washington, arguing that its many crummy culverts violated their right to harvest fish in traditional places. The tribes eventually won, and the state’s Department of Transportation began the arduous, court-ordered process of replacing hundreds of salmon-impeding culverts—including the one at Little Skookum Creek. Here, the department had torn out a rusted metal pipe and replaced it with a wider, shorter span more akin to a small bridge. The stream ran through it unencumbered.

The new crossing was anything but glamorous; if I hadn’t known the state had recently installed it, I wouldn’t have thought twice about it. “This is not the most complicated project,” admitted Paul Wagner, at the time a biologist (since retired) with the state’s transportation department and my guide for the day.

Simple, perhaps, but effective. As we watched, a dozen chum salmon finned in Little Skookum’s channel, snouts pointed upstream. Their flanks were barred with electric purple-and-green stripes, their fins ragged with decay. One by one, with the remains of their flagging strength, they kicked their tails and skittered through the culvert to spawn and die. Cars rumbled on the road above them. “It’s pretty cool to see the animals actually using this,” Wagner said. Better culverts struck me as a form of ecological justice, reconnecting the rivers that infrastructure had sundered.

For all the virtues of culvert replacement, however, many transportation departments have been slow to pursue it for the usual reason: money. The Little Skookum Creek project seemed straightforward, but it cost more than $2.7 million, and other streams require far larger fixes. That afternoon, Wagner took me to Coffee Creek, a humble trickle that, decades earlier, had been interred under 30 feet of earthen roadbed where it passed beneath a highway. Rather than excavating all that fill, the agency had decided it would be simpler to reroute the whole stream through a new artificial channel. We stood on a bluff and surveyed the unfinished creek, which lay as bare and muddy as a ditch in drought. The undertaking would ultimately cost around $20 million, Wagner said, sounding a touch rueful. Culverts might be inconspicuous, but they didn’t come cheap.

In 2021, the United States’ culvert-funding shortfall caught the belated attention of politicians. That November, Congress passed the Infrastructure Investment and Jobs Act, a $1.2 trillion package that included money for everything from high-speed rail to electric vehicle charging stations to basic highway repairs. Tucked deep in the law’s thousand-odd pages was a section that attracted little media coverage, but had immense consequences for fish: the National Culvert Removal, Replacement, and Restoration Grant Program.