picture of a Tibetan mastiff with blurry ghost wolves behind it
tibetan mastiff with ghostly wolves in the background

The Extinct Species Within

The genomes of living animals are littered with DNA from long-gone relatives, providing a lens on evolution, past extinctions, and perhaps even solutions to agricultural problems.

christie wilcox buehler
Christie Wilcox

Christie joined The Scientist's team as Newsletter Editor in 2021, after more than a decade of science writing. She has a PhD in cell and molecular biology, and her debut book Venomous: How Earth’s Deadliest Creatures Mastered Biochemistry, received widespread acclaim.

View full profile.

Learn about our editorial policies.

Aug 6, 2021

ABOVE: Snippets of DNA from extinct species linger in the genomes of many animals, including the Tibetan mastiff depicted above.

Nestled in the middle of the Himalayas is the Tibetan plateau—a large, flat, largely grassy expanse with an average elevation of over 4,500 meters. At such heights, the air is thin, and because of the surrounding mountains, the region receives little rain. It’s a cold, harsh environment—one that many animals simply aren’t cut out for.

Homo sapiens managed to settle in this unforgiving landscape around 30,000 to 40,000 years ago, and around 10,000 years ago, they brought their dogs. While that might suggest our species is especially rugged or adaptable, we now know that neither the people nor their pets toughed it out alone—both cribbed DNA notes from other species in order to adapt. Either before, during, or shortly after their migration to the plateau, H. sapiens got friendly with Denisovans, while their domesticated dogs interbred with Tibetan wolves. And from those hybridizations, both picked up adaptive variants of the EPAS1 gene, which encode version of the protein that help their bodies, and especially their blood, cope with lower levels of oxygen. “You have the exact same [phenomenon] between dogs and wolves as you have between humans and Denisovans,” explains Rasmus Nielsen, a geneticist with the University of California, Berkeley. “It’s so cool.”

But it turns out that, for the canines, that’s not the whole story. Nielsen and his colleagues discovered that before wolves passed EPAS1 along to dogs, the wild canids obtained the helpful EPAS1 variant by breeding with another canine species—one that, to this day, remains unknown.

Researchers refer to these extinct species, whose genes linger in the genomes of living animals, as “ghost” lineages, and we now know they’re everywhere in the tree of life—they simply remained obscured until recently, when advances in sequencing technology and genomic analyses began to reveal them.

The more genomes that have been sequenced from the more different lineages and species and places in the world, the more we see that when things interact with each other in space and can interbreed, they do.

—Beth Shapiro, University of California, Santa Cruz

For example, while scientists have known for more than a decade that modern humans carry sequences from ancient hybridizations with Neanderthals and Denisovans, more recent analyses suggest there are other ancestors haunting our genomes. Discovering when and where species of humans interbred with and interacted with each other will tell the hidden stories of our past and help us understand why H. sapiens is the only hominin species left alive today. And we’re just one case—“There are instances where these kinds of events are even more profound, even more dramatic, in other species,” says Sriram Sankararaman, a computational biologist at the University of California, Los Angeles, who has studied ancient hybridizations in humans.

The more genomes that are sequenced, the more researchers are finding that these ancient genetic whispers have many secrets to tell about all kinds of animals.

Ghostly gifts

Wolves are not alone, of course, in their penchant for mating with more distant kin. Thanks to similar genetic echoes of hybridization, scientists know brown bears cozied up to cave bears before the latter went extinct (and they continue to romp with polar bears), elephant species interbred frequently back in the time of mammoths, and cats apparently fornicate with other felines at almost any opportunity. “The more genomes that have been sequenced from the more different lineages and species and places in the world, the more we see that when things interact with each other in space and can interbreed, they do,” says Beth Shapiro, an evolutionary biologist at the University of California, Santa Cruz, who worked with Nielsen on the Tibetan dogs paper.

Now, scientists are realizing those matings aren’t just fruitful in the sense that they produce surviving offspring. “I think there’s a growing sense that this could be a way for a population or a species to quickly adapt as it moves into new environments,” says Sankararaman. In fact, this kind of adaptation via hybridization, or what’s often referred to as adaptive introgression, seems to happen all the time.

The idea that hybridization plays a significant role in evolution is old hat to botanists, but fairly new for zoologists—“from within the last five years,” says Nielsen. The prevailing view, thanks to influential 20th century biologists such as Ernst Mayr, had been that the comingling of distant relatives was rare and of little importance, especially in mammals.

See “Hybrid Animals Are Not Nature’s Misfits

One of the earliest pieces of evidence for adaptive introgression in mammals came from a 2015 study on domesticated pigs (Sus scrofa domesticus) where a ghost Sus lineage was uncovered. Researchers in China were looking for genomic signatures of adaptation to northern latitudes (and, therefore, genes that may confer cold tolerance) in 11 domesticated breeds when they spotted something strange: A 14-megabase region of the X chromosome that not only differed between northern and southern breeds, it appeared that the northern version emerged some 3.5 million years before the entire Sus scrofa species split from from other wild pigs.

A 2015 study in Nature Genetics found that pig breeds from northern China, such as the Meishan pig above, possess DNA from an unknown, extinct Sus species. 

When the team created an evolutionary tree for that chunk of the genome, which included their domesticated pig breeds as well as Chinese and European wild boars (S. scrofa) and four other pigs (genus Sus), they found that the southern breeds clustered with the other Sus species, as one might expect, but the northern breeds formed their own distinctive group with European wild boars—a pattern which suggested they both received the 14-Mb chunk from an unknown and likely extinct pig species. Two years later, the same research group found another gene—this time, one that may have been involved in domestication—that also appears to have entered the domestic pig genome through hybridization with another, as-yet-unidentified Sus species.

Whether people intentionally bred their pigs with other swine species or just happened to select for genes from a natural hybridization event is unknown. Either way, these findings are far from the only documented examples of adaptive introgression. In addition to the EPAS1 examples in humans and dogs, genetic research has confirmed that western European house mice (Mus musculus domesticus) obtained a gene conferring resistance to the rodenticide warfarin from the Algerian mouse (Mus spretus) and gulf killifish (Fundulus grandis) can tolerate heavily polluted waters thanks to genes garnered from Atlantic killifish (F. heteroclitus). These and numerous other instances of adaptive introgression from recent years have bolstered the idea that hybridization is a key mechanism for evolution. “It’s a new way of thinking about evolution, that really species . . . [are] not isolated—they’re connected to other species,” says Nielsen, “and when the environment changes, they can pick up DNA to adapt to new environmental conditions.”

See “Neanderthal DNA in Modern Human Genomes Is Not Silent

If that’s broadly true, then looking for ghost sequences could be a way to find useful genes, argue Yunnan University’s Yan Li and the Kunming Institute of Zoology’s Dong-Dong Wu in a July Journal of Genetics and Genomics review paper. “[T]he search for a genetic legacy of unknown species, particularly adaptive introgressed variants, in the genomes of extant livestock and crops will provide new sources of genetic variation for breeding and therefore help solve a pressing issue for humans,” they write.

Nielsen agrees with that premise. “Those genes that have been jumping from one species to another species and so on, they’re probably the important genes for that environment,” he says, and could be used to grant species desirable traits. Humans have long been trying to capture such traits through hybridization. Some 10,000 years ago, people bred Chinese pigs with European ones, passing along key fertility and immunity traits to the latter. Similarly, research has revealed that domesticated cattle in China were bred with yak and banteng, a species of cattle endemic to Southeast Asia, to help them survive high altitudes and tropical environments, respectively.

Modern genetic science enables a more surgical approach: using gene editing to insert specific genes or variants into animals’ genomes, rather than producing hybrid offspring with a mishmash of genes from different species and using selective breeding to fine-tune the traits of future generations. This type of gene editing is already being done with plants and is being explored in livestock, so in the future, ghost DNA could be targeted to confer desired traits to the plants and animals we cultivate.

Windows on the past

Not all of these spectral sequences are adaptive. Still, even genomic ghosts that have persisted through chance could prove invaluable to researchers, as they may reveal novel insights about evolution and the ecology of bygone ecosystems.

Studies on ancient introgression in felines have noted that interspecies dalliances have a marked impact on our ability to accurately reconstruct evolutionary relationships, so they’re important for evolutionary biologists to consider when reconstructing the tree of life. These sequences aren’t mere noise, though—on the contrary, analyzing them is “like another way of looking into the fossil record,” says Shapiro, “but rather than having fossils that are actual bones, we have tiny little snippets of the genomes of these extinct species that tell us that they existed.”

Take those pigs, for example. The identified ancient genes suggest that the origin story of domesticated pigs is more complex than previously thought and point to gaps in our knowledge of the ecosystems where domestication occurred. After all, the genomic findings imply that there are two species of pigs we’ve never sequenced that were common enough in the past to leave a genetic footprint on our swine.   

Technical and other challenges

The investigation of genomic ghosts for ecological and evolutionary purposes is still in its infancy, as cutting-edge statistical methods for detecting ancient introgression have only recently been developed. Plus, these methods have mostly been designed to delve deeper into the hybridizations that occurred in hominins, says Martin Kuhlwilm, an evolutionary biologist at the University of Vienna. Because of that, they may not work as well in other species.

For instance, many of these methods require full genomes from the ancient relatives in question—something we have for Neanderthals and Denisovans, but which are rare for other extinct animals. Still, the field of ancient DNA is exploding, so it’s not hard to imagine a future where scientists can employ tools developed for human ancestry studies on any animal species, yielding information that could help explain why they, and not their ghost kin, are the ones still around today.

See “Remnants of Extinct Hominin Species Found in West African Genomes

Analytical tools aren’t the only challenge to this kind of work. For some species, especially endangered ones, simply obtaining specimens can be onerous or expensive. And when the science doesn’t directly teach us more about human health, Kuhlwilm says, “it’s difficult to convince someone to pay for all of that.”

The ghost in the bonobo

Kuhlwilm maintains that such research is worth the investment, because “beyond being cool,” the data they provide is invaluable and often impossible to obtain through other methods. His work on chimpanzees and bonobos (Pan troglodytes and P. paniscus), our closest living relatives, is a perfect example. These animals are similar in many ways, but they differ markedly in behavior—with bonobos often noted for being less aggressive and more sexual than chimps—and have subtler variations in their physiology and ecology. Many researchers are interested in understanding the origin of these differences—unfortunately, the fossil record for great apes is particularly spotty, so there is little to draw on when seeking answers about their evolutionary history.

Bonobos received DNA from a now-extinct ape approximately 500,000 years ago. 
MARTIN KUHLWILM; Modified from KUHLWILM ET AL. 2019, Figure 3

So Kuhlwilm and his colleagues looked to the apes’ genomes for clues instead. Introgression analyses run on 69 chimpanzee and bonobo genomes revealed that the two species had hybridized in the past, but even more surprisingly, 0.9–4.2 percent of the bonobo genome was made up of DNA from an otherwise unknown ape. These segments contained genes related to immunity, physiology, and behavior, all of which suggests some of the notable differences between bonobos and chimpanzees may stem in part from the former’s hybridization with another species.  

So far, we don’t know much about this ghost ape that likely shaped bonobos into the gentler of our great ape cousins. The researchers were able to reconstruct an estimated 4.8 percent of its genome from their samples, but widespread sequencing of bonobos could reveal much more, allowing researchers to dig into questions about the ape’s physiology—answers that could provide novel insights.  

Indeed, Sankararaman notes that with enough data, ghost sequences could bring the past to life in an unprecedented way. “We might be able to use [reconstructed archaic] genomes to say something about the phenotypes of these extinct populations,” he says. From such reconstructions, researchers could garner even more information about extinct animals’ biology and ecology, as some things are just easier to glean from a visual.

Such inferences are still a long way off. Connecting mutations to anatomy or behaviors is “an incredibly hard problem,” he notes, and “even in humans, our ability to go from genome to a phenotype or trait is pretty limited.” Still, such work would be “really exciting,” he says.

Unfortunately, research into the ghosts in animal genomes is racing against the clock, says Kuhlwilm, because it relies on sequencing many genomes from extant species. “I think the main obstacle right now is the speed at which these species disappear. Finding enough genomes from wild individuals and getting them sequenced is becoming a challenge . . . and that is very sad.”

Genomic exorcism: the key to what makes us “human”

All people alive today carry ghost sequences from other human species. According to a Science Advances paper published July 16 that Shapiro coauthored, about half of the human genome can contain sequences from “introgression” events, during which DNA flowed in by mating with Neanderthals, Denisovans, and potentially other as yet uncharacterized hominin species—even though each individual’s proportion of DNA from other species is only about 2 to 4 percent.

Shapiro and her colleagues largely focused their study on the parts of the genome without these introgressions. Their reasoning, says Shapiro, is that those segments make us, well, human. “It’s probably in there, in that little, tiny portion of the genome where nobody has any archaic DNA . . . where we really need to look hard for those genes that make us unique,” she says.

Shapiro explains that even if you sequenced the genome of every person on the planet today and pieced together all the ancient bits of DNA that exist within them, you wouldn’t be able construct a complete Neanderthal or Denisovan genome. Parts of their genomes simply don’t exist in modern humans.

See “Genetics Steps In to Help Tell the Story of Human Origins

It’s probable that some, if not most, of these missing pieces dropped out of the genomes of modern humans by chance, but Shapiro says that for others, “you just couldn’t have the Neanderthal or Denisovan version and still be a human,” so those segments of archaic DNA were eliminated from modern humans through negative selection.

Ed Green, a biomolecular engineer at University of California, Santa Cruz, Shapiro, and their graduate student at the time, Nathan Schaefer (now a postdoc at University of California, San Francisco), went in search of those incompatible regions.

First, they developed a method for detecting archaic introgression that they called “SARGE” because it’s based on what’s known as an ancestral recombination graph (ARG). Essentially, Green says, it creates an evolutionary tree for every locus, which gives it the power needed to separate archaic introgression from genes shared with other species because of ancestry, as well as the ability to detect what Green and his colleagues refer to as “archaic deserts” that genetically separate us from our kin.

Using SARGE, the team examined 279 modern human genomes from the Simons Genome Diversity Project, which sampled from populations all over the world, as well as two Neanderthal genomes and one Denisovan genome. The analysis suggested at least one major wave of breeding between Neanderthals and modern humans and several smaller mixing events with Denisovans. The algorithm also detected other archaic genes—genetic variation retained from the common ancestors that gave rise to us and our closest kin (what geneticists refer to as incomplete lineage sorting). A mere 7 percent of our genomes lacked any trace of archaic DNA.

Within these archaic deserts, the team zeroed in on regions that also had high-frequency mutations totally unique to modern humans, reasoning that these changes occurred after Homo sapiens split from our kin and then spread through much of the human population. They’re where any H. sapiens-specific novelty lies, says Green.

The team estimates that such “human-specific regions” make up roughly 1.5 percent of our genomes. Per base, these regions have more genes, coding regions of genes, and regulatory element binding sites than other parts of the genome—another clue that they’re especially important to us. In addition, that 1.5 percent turned out to be “highly enriched in genes that have to do with nervous system function,” says Green.

Other groups have reached similar conclusions, especially with regard to the uniqueness of human nervous system genes, notes University of Vienna evolutionary biologist Martin Kuhlwilm, who did not participate in the study, but the new work provides higher resolution than past studies. “That’s the most valuable contribution of their new method,” he says, adding that “with their methods, they can basically go down to a handful of genes,” which can be the target of further studies to uncover the functional consequences of human-specific variants.

He adds that he hopes to see this kind of detailed work occur in nonhuman animals, especially primates. For instance, analyzing what the human-specific parts of the genome look like in chimpanzees could reveal whether those regions are broadly species-defining, or if they’re only special in us. Such analyses could “further refine what makes us really human,” he says.