In 2012, on a whim, Vincent Lynch decided to search the genome of the African elephant to see if it had extra anti-cancer genes. Cancers happen when cells build up mutations in their DNA that allow them to grow and divide uncontrollably. Bigger animals, whose bodies comprise more cells, should therefore have a higher risk of cancer. This is true within species: On average, taller humans are more likely to develop tumors than shorter ones, and bigger dogs have a higher cancer risk than smaller ones.

But this trend breaks down when you look across species. Elephants are no more susceptible to tumors than Chihuahuas, and whales are no more likely to develop cancers than humans—if anything, their risk is lower. That’s especially strange because big animals also tend to have longer life spans, giving more opportunities for each of their already abundant cells to become cancerous. They ought to be walking (or swimming) masses of tumors—but clearly they aren’t. For the vast majority of mammals that have been studied, the odds of dying from cancer range from 1 to 10 percent, whether you’re talking about a 50-gram grass mouse or a 5,000-kilogram African elephant.

This puzzling trend is called Peto’s paradox, named after the British epidemiologist Richard Peto, who described it in the 1970s. Since then, biologists have proposed hundreds of hypotheses to explain it. Some note that larger animals have lower metabolic rates; this reduces the rate at which they acquire mutations. Others have suggested that in big animals, tumors need more time to reach a lethal size; during that time, the tumors likely to grow debilitating secondary tumors of their own.

But perhaps the most common hypothesis is that big animals simply have more anti-cancer defenses, including the “tumor suppressor” genes that stop cancers from developing. That idea was on Vincent Lynch’s mind in 2012, when he agreed to give a guest lecture on Peto’s paradox. “I was sitting at my computer, and I thought: Let me just look at the elephant genome and see if they have extra tumor suppressors,” says Lynch, who is an evolutionary biologist at the University of Chicago. “It turns out they have lots. And then I had something to tell the class.”

At first, Lynch looked at specific tumor-suppressor genes like p53. Nicknamed the “guardian of the genome,” p53 plays a pivotal role in responding to damaged DNA, which could ultimately lead to cancer-causing mutations. When such damage is detected, p53 activates other genes that try to fix the problems. It stops cells from growing while those repairs are underway. And if the damage is irreparable, it triggers a self-destruct system that forces the affected cells to commit suicide, before they can turn into tumors.

We have one copy of p53. Lynch discovered that elephants have 20. (A second team independently made the same discovery at the same time.) These extra copies make elephant cells exquisitely sensitive to damaged DNA, and they’ll launch their self-destruct programs at levels of damage that human cells would tolerate. This hair-trigger propensity for cellular suicide helps to explain why elephants are so resistant to cancer for an animal of their size.

But p53 is only part of the story. Lynch’s student Juan Manuel Vazquez found that elephants also have many extra copies of another tumor-suppressor called LIF, as do their closest relatives—the large aquatic manatees, and the tiny rodent-like hyraxes. Millions of years ago, in the common ancestor of all these species, the original LIF gene was accidentally duplicated many times over, creating the extra copies that exist today.

Most of these copies are “pseudogenes”—dead or dormant genes that don’t actually do anything. That’s because the same duplication events that created them failed to copy their promoters—small stretches of DNA that sit in front of genes and allow them to be switched on. Creating a gene without a promoter is like building a car without an ignition switch. You have something that could work, but doesn’t.

There’s one exception. Vazquez found that one of the elephant’s nine extra LIF genes—LIF6—does work. When LIF6 was first created, by fortunate happenstance, the random stretch of DNA just ahead of it was already pretty close to being a promoter. It only took a few small changes, acquired over the course of elephant evolution, for that sequence to evolve into a brand new promoter, allowing the once-dormant LIF6 to whirr into life.

Now, when DNA gets damaged, p53 lands at the promoter and spurs LIF6 into action. LIF6 then triggers that self-destruct sequence that sends damaged cells to their doom. For good reason, Lynch calls it a zombie gene—it used to be dead, it’s now reanimated, and it kills the cells that rouse it.

There are other potential zombie genes lying around—elephants have 8 inactive copies of LIF, while hyraxes have 6 and manatees have 11. Reanimating these genes might help to protect against cancer, but they also come with potential downsides. “It’s like having a trap in your genome,” says Lynch. “If they accidentally get reactivated, they’ll cause cells to [self-destruct], and unless there’s a good reason for that, you don’t want your cells to randomly die.” Perhaps the risks are only worth taking once a species reaches a certain size.

But “LIF is also important in fertility,” says Amy Boddy from the University of California, Santa Barbara, who studies cancer and evolution. The gene allows newly fertilized embryos to implant themselves in their mother’s uterus. So, the reanimation of LIF6 might have something to do with an elephant’s reproductive life rather than its resistance to cancer.

Even if LIF6 does protect elephants against cancer, it doesn’t act alone. When Lynch’s team disabled the gene, they saw that elephant cells were less sensitive to damaged DNA—but only slightly less. “It makes sense for elephants to have multiple mechanisms of cancer resistance since they’re large and so long-lived,” says Lisa Abegglen from the University of Utah, who helped to show that elephants have extra p53 copies.

It’s likely that different animals have all evolved their own solutions to Peto’s paradox. If large animals mitigated the risk posed by their extra cells by evolving extra tumor suppressors, “one might argue that the horse should have more copies than the mouse—and that’s not the case,” says Chi Dang from the Ludwig Institute of Cancer Research and the Wistar Institute. “Is there some threshold of body size when this happens?”

Not quite. Lynch adds that he and his team have tried looking for extra copies of tumor-suppressor genes in whales. “We couldn’t find anything,” he says. “They don’t have extra copies of p53 or LIF. However they’re resolving Peto’s paradox, it’s not the way elephants did it.”