Imagine if a word in a book—say, bubble—had the ability to magically copy itself, and paste those copies elsewhere in the text. Eventually, you might bubble end up bubble bubble with bubble bubble bubble sentences bubble bubble bubble bubble like these.

This is exactly what happens in our genomes. There are genes known as retrotransposons that can copy themselves and paste the duplicates in other parts of our DNA, creating large tracts of repetitive gobbledygook. Around half of the human genome consists of these jumping genes. And a quarter of a cow’s DNA consists of one particular jumping gene, known as BovB. It, and its descendants, have bloated out the cow genome with thousands of copies of themselves.

This jumping gene seems to have entered the cow genome from the unlikeliest of sources: snakes and lizards.

Retrotransposons typically jump around within a single genome, but sometimes they can travel further afield. Through means that scientists still don’t fully understand, they can leave the DNA of one species and enter that of another. And so it is with BovB. No one knows the animal in which it originated. But from that mystery source, it has jumped into the DNA of snakes and cows, elephants and butterflies, ants and rhinos.

David Adelson, from the University of Adelaide, charted the gene’s travels in 2013 by comparing the subtly different versions of BovB in dozens of animals. That was when his team showed that BovB in cows and other cud-chewing mammals is most similar to the versions in pythons and vipers—and likely descended from them. Now, Adelson’s colleague Atma Ivancevic has extended the search for BovB to more than 500 animal species. And her results show that the gene’s travels are even more erratic than anyone thought.

Genes change over time, and closely related species have more similar versions than distantly related ones. So, if you compare different versions of the same gene across a range of animals, you can usually create a family tree that shows how they’re all related. But that only works for genes that are inherited in the usual vertical way, from parent to child.

When Ivancevic did the same exercise for BovB, which jumps between species horizontally, she got one of the weirdest family trees I’ve ever seen. It’s like a window into a bizarre parallel universe where sheep are more closely related to cobras than they are to elephants, where kangaroos have more in common with bedbugs than with horses, and where pythons, zebrafish, leeches, scorpions, and sea urchins all belong to the same tight-knit family. “It was just bizarre,” Ivancevic says.

She estimated that BovB has repeatedly jumped between the genomes of distantly related animals on at least 11 occasions, and likely many more. In some cases, when it arrived in a new lineage, it ran amok: Cows, sheep, and elephants all have thousands of copies in their DNA. In other groups, like bats and horses, it kept its head down, and produced just a few dozen copies.

No one knows how BovB travels between species, but Ivancevic and Adelson suspect that it might spread via blood-sucking parasites. They have found strong similarities between the BovB versions in leeches and zebrafish, bedbugs and snakes, ticks and lizards. By biting different hosts, parasites might help jumping genes to vault over the species barrier.

This idea makes sense in principle, but it’s hard to imagine in practice. What parasite, for example, could possibly have transferred BovB between, say, a sea urchin and a rattlesnake?

We should be careful before making too much of similarities between BovB in different species, argues Sarah Schaack, from Reed College. “Let’s say some worm DNA was picked up by a virus, that virus began infecting a bird, and the DNA was transferred,” she says. “Subsequently, a tick that specialized on the bird picked up some of the viral DNA. If you only sequenced the worm and the tick, you would infer the worm got the DNA from the tick or vice versa, which might be appealing because ticks are parasites, but would be strange because ticks don’t bite worms.” When genes go on the move, it can be hard to track their footprints accurately.

Ivancevic agrees, and wants to sequence BovB in many more species, parasites included. “I feel like the [patterns we have] look really sporadic because we don’t have all the intermediate species,” she says. By looking at more animals, it may become easier to trace the gene’s history.

That was certainly the case when Ivancevic turned her attention to a different retrotransposon called L1. This jumping gene is more directly relevant to humans because it fills up 17 percent of our genome. Most copies are now broken and stationary, but some have kept their ability to move around—and their presence has been linked to diseases like schizophrenia and cancer.

L1 exists in the genomes of almost every mammal, and was presumably around in the DNA of our common ancestor. And while it can jump around any given genome, scientists believed that, unlike BovB, it doesn’t move between species.

Not so, says Ivancevic. She identified at least three possible cases in which L1 seems to have jumped between major animal groups, typically those that live in water. And she showed that it’s completely absent from the DNA of platypuses and echidnas, which means that it must have jumped into mammalian genomes after these oddities had split off from the main dynasty, between 160 and 191 million years ago.

“There’s not yet a smoking gun for this transfer, like the identification of the donor species,” says Edward Chuong, soon to be at the University of Colorado Boulder. Still, “their failure to find any trace of L1 in [platypuses and echidnas] is fairly strong evidence supporting their claim.”

“These [jumping genes] really seem to be good at moving between genomes,” says Alexander Suh, from Uppsala University. But why, he asks, are some genes like BovB more likely to do so than L1, or at least more likely to establish themselves in a new setting? Possibly, Ivancevic says, it’s because BovB is about half the size of L1 and so is easier to move. Maybe it’s because BovB hitches a ride on parasites, and L1 doesn’t.

Whatever the reason, both genes—and others like them—must surely have influenced the evolution of their hosts. Jumping genes were first discovered by the pioneering scientist Barbara McClintock in the 1940s, but it took decades for people to realize how common they are—and how influential. They could potentially cause diseases by jumping into the wrong place and disrupting vital genes. But they could also provide raw material for evolution, by reshaping or rewiring existing genes. In this way, they spurred the evolution of the placenta, and supercharged our immune systems.

“It’s mind-boggling to think about how just a few [jumps] have fundamentally altered the course of our evolution,” says Chuong. Evolutionary biologists like to wonder what would happen if we replayed the tape of life—if we went back to some earlier point in history and let evolution run its course again. Would history repeat itself? Chuong thinks not. These jumps are so unpredictable, but so potentially important when they happen, that it’s hard to imagine the events unfolding twice.

And what would happen if we ran the tape forward? Where will genes like BovB end up next, and how will they shape the destinies of their hosts? “It would be interesting to see what BovB looks like in a few million years, but I’ll probably not get a chance to do that,” says Ivancevic.