Social Network


How a Simple Statistical Error Killed 463 Wolves

A study showed that culling wolves could save caribou. But a second group of researchers saw a flaw in that conclusion.


This winter, 463 wolves died in British Columbia. Their deaths were not due to a freak accident or a natural disaster, but a government-sponsored cull meant to save endangered mountain caribou. Killing wolves is often controversial, and in this case their deaths may have been in vain: A group of scientists says the decision to cull the wolves rested on a statistical error.

In the spring of 2019, British Columbia’s government was embroiled in a series of high-profile community-feedback sessions concerning the conservation of mountain caribou. The endangered ungulates depend on old forests targeted for logging that also happen to grow on top of highly coveted oil-and-gas deposits. The sessions were heated, to say the least.

Amid the clash between industry advocates and conservationists, a group of researchers led by Robert Serrouya, a caribou biologist at the University of Alberta, published a paper that offered a practical solution: Killing wolves, which are predators of caribou, and penning pregnant caribou could help save the vanishing population. The study made a considerable splash in Canadian media. “It was quite a story,” says Chris Darimont, a conservation scientist at the University of Victoria, who worked on the new paper. “It’s a pretty desperate time for caribou, and people from policy makers to the public were very interested in the prospect of a solution.”

In September, the government came to a decision. It would not designate any new caribou-protection areas for the deep-snow caribou. This spelled disaster for the caribou, which have experienced one of the steepest population declines of any caribou population in the world, as their forest habitat has been razed by clear-cutting or fragmented by roads. The deep-snow caribou once roamed as far south as Montana’s Bitterroot Mountains, but almost all populations had disappeared from the United States by 1980. By the early 2000s in Canada, the deep-snow caribou had lost as much as 45 percent of their population in just 27 years. But the government had an alternate plan: a new wolf-cull program that cited Serrouya’s study as proof of why killing wolves works.

Now a new group of scientists has reassessed the statistical findings of that paper. Their rebuttal, published today in Biodiversity and Conservation, shows that a simple oversight doomed those 463 wolves.

“It all started with the statistics,” says Viktoria Wagner, a plant ecologist and statistician at the University of Alberta and an author of the rebuttal. When Wagner first saw Serrouya’s paper shared on Twitter, the results seemed exciting—a rare glimmer of hope for the besieged caribou. But when she examined its findings more closely, she noticed that the study considered only 12 populations (and six control populations), a surprisingly low sample size. When she dug deeper, she found even more irregularities.

In wildlife ecology, researchers rely on models that can potentially explain how the world works—for example, why a population of caribou could have increased or decreased in a given year. The 2019 paper included models examining how solutions such as killing wolves and penning pregnant caribou could stabilize or increase populations. The most glaring error in the paper, though, was its omission of a null model, the new paper argues. “A null model represents this scenario of uncertainty, the possibility that none of these predictors would be important,” Wagner says. When the researchers in the rebuttal study ran a null model, they found that it performed just as well as the models examining wolf culls and caribou pens. In other words, the solutions proposed in the 2019 study, which subsequently spiraled into policy, had no statistical support.

Wolves have long been a charismatic scapegoat for caribou decline. “It’s kind of seductively compelling to believe that if one species eats another species, if we remove many of the first species, the second species is going to do better,” Darimont says. But wolf culls are a blunt instrument that have no real track record of protecting caribou. They do, however, offer a way for the oil-and-gas industry to fulfill caribou-conservation requirements.

But to Serrouya, the lead author of the 2019 paper, the solutions proposed, including the wolf cull, make logical sense. In some of the caribou populations he examined, wolf culling or maternal penning did lead to population growth. “The main point of the [rebuttal] relies on a statistical argument, whereas ours relies more so on logic,” Serrouya told me in an email, adding that both papers share the mission of understanding how to to save mountain caribou.

Even though the authors of the new paper found no statistical support for penning pregnant caribou as an effective way to stabilize all of British Columbia’s caribou populations, Darimont does not dismiss the strategy entirely. “Maternal penning is a strategic investment by local people for many reasons that could very well pay off,” Darimont told me in an email. The practice is spearheaded by certain First Nations communities who traditionally harvested caribou on their ancestral lands. But the rebuttal argues that the original study’s small sample size and lack of focus on ecotypes—specific populations, such as the deep-snow caribou, that depend on unique environments—mean that a solution that works in one caribou population might not work in all of them.

Perhaps the most important finding in the reassessment is that the best indicators of the mountain caribou’s overall decline are the species’ different ecotypes. The most prominent example is the steep decline of the deep-snow ecotype, a population that depends on arboreal hair lichens present only in forests more than 80 years old, says Toby Spribille, a lichenologist at the University of Alberta and an author of the rebuttal. The deep-snow caribou are behaviorally unique from any other ecotype, able to paw through snowpack 13 feet deep to feast on lichens. “I’d like to see the deep-snow caribou not go extinct,” Spribille says, adding that wolves are only the fourth-most-prominent predator of the deep-snow caribou.

And while the 463 dead wolves represent needless tragedy, the damage from the 2019 study has and will most significantly affect the caribou, the researchers say in the rebuttal. From 2014 to 2019, 350 square miles of crucial deep-snow-caribou habitat has been lost to logging, according to a study in Conservation Science and Practice. “That’s irreplaceable on the timescale that we have left to save the caribou,” Darimont told me. Though the rebuttal comes a year late, the researchers hope that it will influence future policy regarding caribou management.

Wagner, one of the authors of the rebuttal, studies grassland ecology, not caribou. But when she saw the 2019 study, she felt no other choice but to set the statistical record straight. “I just felt a sense of responsibility as a scientist,” she says. Now that the rebuttal is published, she can get back to her research studying a rare type of meadow in Montana, near mountains where caribou once roamed.