Some Weird Looking Wires Showed Up on a Road in England. They Didn’t Do What They Were Meant To. They Were a Sign of a Larger Problem.

This essay is adapted from  Beyond Belief: How Evidence Shows What Really Works , by Helen Pearson . Copyright © 2026 by the author and reprinted with permission of Princeton University Press.

Near the border of Norfolk in England, some weird-looking wire bridges span a busy road. They consist of pairs of poles on each side of the road, with a strip of wire mesh strung between and running down to the ground on either side. These are bat bridges, and they are a good example of a well-meaning conservation idea gone wrong.

The bridges appeared in 2014, when an extension of the A11 road was being built. The new section cut a path through Thetford Forest, a vast and precious woodland which is home to several bat species. Major roads are bad news for bats: They can disrupt feeding, breeding, and flying routes. Many bat species fly close to the ground or trees and will avoid crossing a road. If they do cross, they’re likely to do so at traffic height and are often killed by vehicles. Bats are protected by law in the U.K.—some species are at risk of extinction—and so developers have to assess the environmental impacts of a new road on bats and take measures to mitigate them.

The authorities behind the A11 extension decided to build a series of bridges to help bats cross the road. Bats navigate by emitting high-frequency sounds and detecting their echo, and the theory was that bats speeding toward the new road would detect and follow the line of the wire bridge upward and over the road, just as they follow the line of a hedge. Several mainland European countries had installed similar bridges and a handful already spanned British roads further north. In August 2014, part of the new A11 was closed for five nights so a suite of bat bridges could be installed at an eventual cost of more than half a million dollars.

Was there any evidence that bat bridges would conserve bats? The best person to answer this question was John Altringham, an ecologist with a specialty in the creatures. Altringham started examining bat bridges in 2007, and was skeptical about them from the outset. While it was reasonable to think that a bat bridge could work like a substitute hedge, he could see that the bridge’s exposed high wires didn’t look like a hedge, didn’t sound like a hedge to a bat’s echolocation, and didn’t feel like a hedge because it was elevated and failed to offer the shelter of a low-lying hedge. If Altringham, a bat expert, had wanted to design something that might help bats cross the road, then a bat bridge would have none of the necessary properties. In fact, he could never understand where the idea had come from.

In 2012, before the A11 bat bridges were built, Altringham and fellow ecologist Anna Berthinussen had tried to assess whether bat crossings in northern England were being used. They monitored the behavior of bats at foraging routes around the bridges and compared it with routes without a crossing. The scientists found no evidence that bats were more likely to cross the road using bridges or that they were fooled into thinking that they were hedges. Instead, most bats flew like an arrow directly across the road at a height at which they would collide with vehicles. A second study found that bats consistently ignored the pricey wire bridges on the A11 too. Highways England (now called National Highways) defended the crossings at the time and said they were part of a larger package of programs to support bat populations, including landscaping and roosts. National Highways continued to monitor bat species and said a survey in 2018 indicated that bat wires were helping to guide over 40 percent of monitored bats safely across the carriageway.

There are potentially better ways to protect bats. These include underpasses beneath roads and “green bridges,” covered with vegetation, that are more like a hedge. After studying a few of these, the ecologists concluded that these crossings were more likely to be used by bats if they were well designed and positioned along existing flight routes. They also cost more.

The findings about the bat bridges were demoralizing for Altringham, who during his career appeared at several public inquiries into the impact of road-building projects on bats. He found himself pitted against consultants who had been hired by road developers, and realized that there was no interest in discussing what evidence showed about protecting species. Altringham felt his integrity and credentials were under attack, and eventually he lost faith in public inquiries.

Even more depressing was the fact that this situation was not unique to Britain, or to bats. Critical decisions about how to save species were being made all over the world with no regard for evidence at all.

Yes, sometimes in conservation efforts, common sense is enough. It’s generally better to create a nature reserve and try to protect species than to let them be destroyed. But Bill Sutherland, a biologist at the University of Cambridge, has found that some dogma in conservation is flat-out wrong when put to the test. For example, in the early 1990s, he and his colleagues examined standard recommendations for conserving reed beds—important habitats for birds that become overgrown with trees over time. Burning reed beds is a quick way to restore them, but everyone knew it should be avoided because it killed soil invertebrates, such as snails and earthworms. Yet Sutherland’s careful experiments suggested the opposite: Burning did little to harm invertebrates, whereas flooding—which was recommended—decimated them.

In India, populations of tigers were for decades monitored using what’s known as the “pugmark census method.” (A pugmark is a wild mammal’s footprint.) An army of people would spend a week or two searching for tiger tracks and making tracings or plaster casts of the left hind foot. The footprints were supposed to identify individual tigers like fingerprints and were used to estimate the total number of wild animals. In 2003, a group of conservationists showed that this technique was as unreliable as it sounds , that three decades of censuses had produced deeply flawed tiger counts, and it was unclear if protection measures had worked.

In North America, it’s been common for nearly a century to engineer streams so that salmon and trout can journey upstream and spawn. These fish are important in river ecosystems and commercially. Stream engineering aims to improve habitats and water flow, for example by building dams or placing boulders in a stream, and these efforts have cost hundreds of millions of dollars over the years. And yet a systematic review showed there was almost no evidence that they benefit salmon and trout populations.

Ineffective conservation practices are a massive problem, given that the stakes are so high. In 2019, a pivotal United Nations–backed report found that nature was being destroyed faster than at any time in human history and around 1 million plant and animal species faced extinction. People had significantly altered 75 percent of land and 66 percent of ocean areas through agriculture, fishing, climate change, pollution, and other activities. Halting this biodiversity crisis will require radical action. It is vital that mitigation measures work. And yet the bat bridges, tiger counts, reed-bed flooding, stream engineering, and many other examples all point to the uncomfortable fact that many standard conservation measures are ineffective or even harmful.

But there is a fundamental problem in conservation: rigorous field experiments, such as randomized controlled trials, are difficult and rare. Researchers talk about the “units” in their experiments; in medical trials, the units are humans and hundreds of humans can be randomly assigned to treatment and control groups. But in conservation, animals can’t be randomly assigned to groups in the wild and told how to behave. Instead, the experimental units can be plants, plots of land, fields, moors, forests, or oceans. When the units are small and the problem well defined, it may be possible to assign reasonably large numbers to different groups. For example, researchers can randomly assign plots of forest to receive nest boxes in order to test whether they help songbird populations grow.

Experiments get harder the bigger the units become. It’s unfeasible to randomly assign hundreds of forests or oceans to become protected areas or a control group. It would be unethical to experimentally introduce invasive species to randomly selected islands or to randomly spark wildfires to test their impact. It is difficult to study rare or camouflaged species when scientists might only glimpse one per year. What’s more, conservation experiments can take decades to produce results, because many species reproduce slowly and ecosystems are slow to change. Researchers and practitioners also struggle to get funding to run well-designed experiments when conservation is so cash-strapped.

Often, evidence-based conservation requires looking at the best available data, and combining it with the knowledge of experts and local people. And the effort worth expending on collecting evidence depends on the importance of the decision—something Sutherland rates on a scale of 1 to 7. A Level 1 decision is a no-brainer—for example, putting up a nest box on a garden tree. A Level 4 decision, such as how to save a population of rare toads, might involve using the Conservation Evidence database that Sutherland created showing what works to protect species and habitats. The most extensive assessment of evidence is reserved for Level 7 decisions, involving high stakes and unclear or conflicting evidence.

One of the most remarkable efforts at evidence synthesis I’ve come across was a Level 7 decision about a vulnerable population of woodland caribou in Canada’s Rocky Mountains. Few species are more iconic than caribou, known as reindeer in other parts of the world. For Indigenous communities in Canada, they are culturally important and a source of food. For forest ecosystems, they are ecologically valuable, including as prey. But the destruction of forests means that woodland caribou are threatened or endangered, and the population in Jasper National Park is so small that it is heading toward extinction.

In 2021, Parks Canada, which oversees the national park, was working out whether to try to save them by investing some $24 million in a major caribou breeding program. This would involve capturing wild animals, breeding them in captivity, and then releasing juveniles back into the wild. It was an expensive and high-risk venture, so Parks Canada turned to a nonprofit organization called Foundations of Success, which was co-founded by Nick Salafsky. The group specializes in improving conservation projects.

Parks Canada and Foundations of Success wanted to make sure that the caribou decision was informed by the best evidence available. But there were no evaluations or experiments showing whether captive breeding programs are effective for caribou—and even if there were, it would be dangerous to assume that the results would generalize to the unique ecosystem of Jasper National Park.

Instead, the team broke down their unanswerable question—will the breeding program work—into four smaller, more tractable assumptions:

If all four of these assumptions were correct, there would be a good case that the breeding program would help.

Next, the team searched for evidence that might support or refute each assumption. They looked for data and research that had been collected locally—about wolf abundance in Jasper National Park, for example. They added evidence that had been collected globally, such as whether wild caribou survive the process of being captured. Then the team assessed how strongly the evidence supported or refuted each assumption.

Salafsky calls the type of evidence synthesis the group was doing “aluminium standard,” because it’s cheaper and quicker than a gold-standard systematic review but still good enough for the job. He is firmly in the pragmatic camp when it comes to evidence synthesis, believing that investing at least some effort in collecting evidence still dramatically ups the chances of making a good choice.

The caribou group added another step to increase their confidence in the evidence. They invited over 30 experts from around the world who cared about caribou—including Indigenous peoples, academics, zookeepers, and reindeer experts from Finland—to independently review the evidence for each assumption. Then everyone got together in a big room to hammer out their disagreements.

By the end of the process, the group had agreed there was enough evidence to be confident in all four assumptions, with some caveats. They couldn’t be completely sure that other threats to the caribou had been mitigated, because the long-term effects of climate change were unknown, and it would be important to monitor the herd once they were released from captivity. The group concluded that there was a strong case to green-light the breeding program. After inviting public feedback on the proposal, Parks Canada decided to proceed. In March 2025, it moved the first wild caribou to its new breeding center—and by June, some adorable calves had been born.

If this process sounds complicated and intensive, that’s because it is. The project took over 700 person-hours. The summary report on this one decision spans 57 detailed slides of spreadsheets and flow charts. But it was worthwhile because it was a high-stakes, Level 7 decision. If the caribou breeding program fails, one of the last remaining populations of these iconic creatures would be lost, along with millions of dollars. There is only one shot at saving them, and everyone wants to get it right.