In 1898, scientists in Uruguay noticed that some of their laboratory rabbits were dying from a mysterious illness, their skin riddled with tumors and weeping wounds. The researchers named the disease myxomatosis. They showed that it was caused by a new virus. And they argued that this myxoma virus—highly lethal, specific to rabbits, and spread by mosquito bites—was exactly what the Australian government was looking for.

Europeans had introduced rabbits to Australia at the end of the 18th century, whereupon the fuzzy critters started breeding like, well, y’know. A century later, they had become a serious problem for both the nation’s wildlife and its farmers. Perhaps a disease could control the bunny blight?

In 1950, after some resistance and much cajoling, government scientists finally released myxoma-infected rabbits into the Murray Valley of southeastern Australia. That summer, the virus blazed brightly, but its spark appeared to peter out. Then, by year’s end, it rekindled into an almighty conflagration that swept through southern Australia, killing millions of rabbits. “Thus, inadvertently, began one of the great experiments in natural selection, conducted on a continental scale,” wrote Australian scientist Peter Kerr.

The myxoma virus quickly evolved. The strain that had initially been used was almost inescapably lethal, killing virtually every rabbit it infected. But virologist Frank Fenner discovered that, within a few years, this strain had been replaced with milder ones, which killed less rapidly and frequently.

These events provided an unprecedented view of how viruses evolve in the wild. They’ve also permeated into the popular consciousness, creating an intuitive sense that lethal viruses eventually evolve into milder forms, which are less likely to completely wipe out their hosts. But “the notion that everything’s heading toward a state of long-term co-existence and happiness is not always the case,” says Andrew Read, an evolutionary biologist based at Pennsylvania State University. “There are plenty of examples where the virus has got nastier over time.”

And as it happens, myxoma is one such example. “It went from exceptionally nasty to just nasty, and now has turned round and cranked up the nastiness again,” Read says.

The virus was never entirely defanged. After its release in 1950, it went from killing more than 99 percent of rabbits to killing around 75 percent of them, or under 50 percent in some cases. In response, the rabbits evolved resistance, shrugging off strains that would once have finished them off. And that relaunched the arms race between myxoma and rabbits, prompting the virus to evolve its own countermeasures, which it still deploys today.

Read worked out how it responded by teaming up with Peter Kerr, who had collected and stored myxoma samples from the last several decades. By exposing lab rabbits to these archived strains, the team showed that by the 1990s, the virus had gained a new ability: It could completely shut down a rabbit’s immune system. This stops the animals from effectively clearing the virus. Inadvertently, it also means the bacteria that normally live peacefully in the rabbits’ bodies run amok, spreading through their internal organs and causing septic shock. These rabbits never develop the skin tumors or any of the classic symptoms of myxomatosis. Instead, they die from massive and sudden infections. Their lungs fill with fluid and they start bleeding uncontrollably.

These immune-suppressing strains might have emerged as early as the 1970s, and they’re circulating broadly now. Still, their effects are hard to spot in wild rabbits, which still die from the same kinds of symptoms as they used to. That’s because their genetic resistance partly counteracts the virus’s new ability, which only becomes clear when it infects lab animals that have no history of coevolving with this virus. The wild rabbits started to resist the virus, the virus started to kill them in a new way, and neither side gained any ground. “It’s like a duck in a stream, paddling like crazy under the water and not getting anywhere,” says Read.

“Laboratory experiments using bacteria and their viruses have shown that when hosts evolve resistance against infections, viruses can rapidly overcome host immunity,” says Lotta-Riina Sundberg, from the University of Jyväskylä. “But monitoring these long-term coevolutionary arms races in natural settings with such accuracy is challenging.” That’s why the myxoma story is so important, she adds.

The same dynamics played out in Europe, where a different strain of myxoma was used to control rabbits, following the Australian success. There, too, the virus evolved into milder forms. And there, too, new immunosuppressive strains have emerged. No one knows what will happen in the future. In South America, myxoma’s birthplace, the virus causes an innocuous disease in the local cottontails. But there’s no indication that the Australian or European strains are heading in that direction.

“The broad lesson is that there’s a variety of evolutionary trajectories that pathogens can take,” says Read. “There are situations, no question, where virulence can go quite low. Sexually transmitted diseases, for example, require hosts to be sexually active and that requires that they stay alive for some time. But there’s no reason to think that the average long-term state will be coexistence, and that’s a mistake that’s permeated the public.”

Consider rabbit hemorrhagic disease—another infection that Australia considered as a way of controlling rabbits, and that escaped from a quarantine facility in 1995. The virus behind the disease is transmitted by corpse flies, which are attracted to cadavers, so this virus actually benefits by killing its hosts in spectacular fashion. It’s present in huge numbers at the time of death. As such, it started off lethal and has only become more so with time. In the United States, West Nile virus has become more virulent in house sparrows, in response to the birds evolving resistance. And Marek’s disease—an illness of fowl—became fouler after farmers treated birds with a “leaky” vaccine, which stops them from developing the disease, but not from becoming infected or spreading the virus.

These consequences are relevant to various companies and researchers who are trying to make farmed animals more resistant to diseases. Some are doing it by traditional breeding. Others are looking to genetic engineering. Whatever the route, the myxoma example shows that such measures could drive the evolution of more potent viruses. These may not be a problem for the resistant animals, just as immunosuppressive myxoma strains aren’t especially deadly to wild rabbits. But if the viruses spread to naive animals, they would suffer.

“If you had a bunch of companies in one river system, and one is creating more resistant fish, causing pathogens to become more virulent, what does that do to the wildlife and the fish belonging to other companies?” says Read. “You have to ask about the long-term consequences. Maybe there are some types of resistance that are less likely to provoke this arms race than others. We need to understand that.”