(...) Birds are the most diverse vertebrates on land, and they have always been central to ideas about the natural world. In 1837, a taxonomist in London told Charles Darwin that the finches he had shot and carelessly lumped together in the Galápagos Islands were, in fact, many different species.
(...) The rise of genome sequencing, at the turn of the twenty-first century, seemed to bring Darwin’s dream within reach.
(...) Hoatzins, which live along oxbow lakes in tropical South America, have blood-red eyes, blue cheeks, and crests of spiky auburn feathers. Their chicks have primitive claws on their tiny wings and respond to danger by plunging into water and then clawing their way back to their nests—a trait that inspired some ornithologists to link them to dinosaurs. Other taxonomists argued that the hoatzin is closely related to pheasants, cuckoos, pigeons, and a group of African birds called turacos. Alejandro Grajal, the director of Seattle’s Woodland Park Zoo, said that the bird looks like a “punk-rock chicken,” and smells like manure because it digests leaves through bacterial fermentation, similar to a cow.
DNA research has not solved the mysteries of the hoatzin; it has deepened them. One 2014 analysis suggested that the bird’s closest living relatives are cranes and shorebirds such as gulls and plovers. Another, in 2020, concluded that this clumsy flier is a sister species to a group that includes tiny, hovering hummingbirds and high-speed swifts. “Frankly, there is no one in the world who knows what hoatzins are,” Cracraft, who is now a member of B10K, said. The hoatzin may be more than a missing piece of the evolutionary puzzle. It may be a sphinx with a riddle that many biologists are reluctant to consider: What if the pattern of evolution is not actually a tree?
Fossils that resemble hoatzins have been found in Europe and Africa, but today the birds can be found only in the river basins of the Amazon and Orinoco of South America.
(...) Hoatzins—“in some respects the most aberrant of birds,” according to one Victorian ornithologist—were a problem from the beginning. Early European naturalists described them as pheasants, and the first major tree for birds, published in 1888 by Max Fürbringer, placed them on the fowl branch. But, by the early nineteen-hundreds, some scientists were comparing hoatzins and cuckoos on the basis of traits such as jaws and feathers, and others were noting similarities between hoatzins and turacos, pigeons, barn owls, and rails. Even the hoatzin’s parasites defied classification: they hosted feather lice found on no other birds.
One crucial problem in phylogeny was convergent evolution. Sometimes natural selection nudges two organisms toward the same trait. Birds and bats independently evolved the ability to fly. Swifts and swallows each evolved into aerodynamic insectivores with nearly identical silhouettes, but traits such as their vocal organs and foot bones reveal that they are only distantly related. Because taxonomists often disagreed about things such as how to distinguish common ancestry from convergent evolution, the literature grew thick with conflicting trees, to the point that some twentieth-century biologists seemed ready to give up. “The construction of phylogenetic trees has opened the door to a wave of uninhibited speculation,” one wrote in 1959. “Science ends where comparative morphology, comparative physiology, comparative ethology have failed us.”
Phylogeny made a comeback in the seventies and eighties, after the German entomologist Willi Hennig developed more rigorous criteria for identifying common ancestry and drawing evolutionary trees. These innovations laid a foundation for a new wave of research that did not rely solely on physical specimens but, rather, on the emerging science of DNA. “Organisms are related to one another by the degree to which they share genetic information,” two ornithologists wrote in the early nineties, adding that genetics could reveal “a different view of the process of evolution and its effects.” The typical bird genome is a string of more than a billion base pairs that mutate randomly over time. Scientists can compare the same parts of the genome across multiple species to estimate their evolutionary closeness. Typically, species that share mutations have a more recent common ancestor, and species that do not are more distantly related.
Early sequencing was expensive and tedious, but, by the beginning of the twenty-first century, a signal was emerging from the noise. The journal Nature published an article about the promise of a single unified tree of life. But its author also identified a complication: each genome contains many different genes, and each one could generate a different evolutionary tree.
In 2001, a paper in the Proceedings of the Royal Society identified a pair of bird siblings as unlikely as Arnold Schwarzenegger and Danny DeVito: the flamingo’s closest relative was a little diving bird called a grebe. “That was probably the single most astounding result that anybody’s ever gotten,” Peter Houde, an avian biologist from New Mexico State University, told me. Ornithologists had always reasoned that grebes were closely related to short-legged loons, whereas tall wading birds such as flamingos, storks, and herons probably had a long-legged common ancestor.
That was the first domino to fall. In 2008, Science published a new avian tree based on DNA. Research led by Shannon Hackett, Rebecca Kimball, and Sushma Reddy, scientists affiliated with the Field Museum and the University of Florida, examined nineteen parts of the genomes of a hundred and sixty-nine avian species. The “root” of their tree resembled trees based on physical specimens: large, flightless birds such as ostriches, emus, and kiwis—known collectively as ratites—were first to diverge from all the others, followed by land fowl and waterfowl. The remaining ninety-five per cent of living birds, from parrots to penguins and pigeons, are known as “modern birds” and descended from a common ancestor, probably around the time that an asteroid hit the earth, sixty-six million years ago, and the dinosaurs went extinct. The youngest order—passerines, which include all songbirds—branched out into a staggering six thousand species in the span of tens of millions of years. The genetic tree for modern birds was decked with relationships that few, if any, taxonomists had guessed from anatomy; key groups such as parrots, owls, woodpeckers, vultures, and cranes shifted places.
Scientists had long assumed, for example, that daytime hunters such as hawks, eagles, and falcons all descended from a single bird of prey. But, in the genetic tree, hawks and eagles shared a branch with vultures, yet falcons turned out to be closer relatives of passerines and parrots. This meant that the peregrine falcon is more closely related to colorful macaws and tiny sparrows than to any hawk or eagle. The traditional explanation for flightlessness in ratites—that a common ancestor diverged into ostriches, emus, rheas, cassowaries, and kiwis after the southern continents split apart—also collapsed. DNA showed that the ratites also included flying birds called tinamous, suggesting that the group evolved flightlessness at least three separate times. “That study revolutionized our understanding of how the major groups of living birds are related to each other,” Daniel J. Field, an avian paleontologist at the University of Cambridge, said. Bird-watching guides had to reorganize their contents to reflect the new relationships.
What the study could not settle was the early evolution of modern birds. It was easy to tell when pheasants and ostriches turned off the highway of avian evolution, but modern birds did not follow a simple off-ramp. They seemed to zoom off in different directions, as though each kind of bird took a different exit from a busy roundabout. From their common ancestor—perhaps a little ground bird that pecked seeds and insects out of the ash that the asteroid left behind—modern birds split quickly into more than half a dozen branches. But the fastest computers of the time weren’t fast enough to disentangle them. All but one of these branches diversified into about ten thousand bird species. The last belonged to the hoatzin alone. The strange bird likely made the journey to the present day all by itself. “The enigmatic Opisthocomus (hoatzin) still cannot be confidently placed,” Hackett’s team wrote.
The first human genome sequences required hundreds of scientists and billions of dollars, but the costs fell quickly as the technology improved. In 2010, Tom Gilbert, an evolutionary biologist at the University of Copenhagen who previously studied mammoth and ancient human DNA, turned his attention to the pigeon genome. “I’m really interested in how regular city pigeons have spread around the world and done so well,” he told me. When he read the Hackett group’s study, he became curious about how the pigeon genome might fit into the larger picture of avian evolution. He wondered, “What if you had the perfect data set—all of the genome and not just parts of it?”
With the neurobiologist Erich Jarvis and the evolutionary geneticist Guojie Zhang, Gilbert assembled a team to pick up where researchers like Hackett had left off. The team, which grew to more than a hundred and twenty researchers, used nine supercomputer processing centers to sequence and analyze the genomes of forty-eight birds. (They got the hoatzin DNA sample from Houde, in New Mexico.) The tree they developed—featured on the cover of Science, along with an image of the hoatzin, in 2014—confirmed many of the Hackett group’s findings, challenged others, revealed new relationships, and used fossils to estimate the dates of divergences. Within fifteen million years of the extinction of dinosaurs, all the major lineages of modern birds emerged. The hoatzin’s long branch connected to the ancestor of cranes and shorebirds. “It kind of is a marshy waterbird,” Jarvis reasoned. But he and the other researchers couldn’t get strong statistical support for the hoatzin branch. He compared the bird’s origins with some of the most difficult questions he has faced in neurobiology. “Studying consciousness or language is the equivalent of figuring out where the hoatzin belongs,” he said.
The next year, the rival journal Nature published yet another tree. The Yale ornithologist Richard Prum argued that forty-eight species were too few, so his team compared a hundred and ninety-eight, sequencing a much smaller portion of their DNA. In this tree, several branches changed places around the time that the dinosaurs went extinct, suggesting new relationships for doves and pigeons; hummingbirds, swifts, and their relatives; and, of course, the hoatzin. Instead of yielding an authoritative tree of life, DNA had entrenched disagreement in the part of the tree most crucial to understanding the diversity of living birds. “There may be no amount of sleuthing or data or analysis that is going to resolve the placement of some of these lineages of birds,” Hackett told me. The conflicting signals in the hoatzin genome may not be analytical errors but biological realities—and they may require a different paradigm than the tree.
The tree is so ingrained in evolutionary biology that scientists encourage “tree thinking.” By learning to think in terms of trees, students can avoid the common fallacy of reading evolution as a ladder in which simpler organisms become more complex, as in the famous image “The Ascent of Man,” which shows a knuckle-walking ape evolving into an upright human. For all its pedagogical value, however, the tree also embeds subtle assumptions about evolution. The tree tends to downplay the genetic variation within species, which can obscure the fact that common ancestors are actually diverse populations that can pass on different versions of a gene to different descendants. It tells a story of endless partition and diversification, with branches that diverge and never reticulate.
While preparing their paper, Gilbert and his team had fiddled with their data set to understand the differences between gene trees. When they told their tree-building software to focus only on regions of the genome that Prum’s team used, it produced a tree that looked like Prum’s. When they shifted focus to other regions, a very different tree emerged. When they divided their bird genomes into thousands of different parts and ran each through their software, they got thousands of different trees, and not one completely matched the “species tree” they had constructed from large portions of genomes. “Different parts of the genome have different stories,” Gilbert realized. Genes do not stay in the lanes of common ancestry but can move much more unpredictably, like zigzagging pieces on a Plinko board. Scientists call this kind of genetic scrambling “incomplete lineage sorting,” and it is especially common during rapid bursts of evolution, such as the one that gave rise to modern birds.
In 2016, Alexander Suh, a biologist on the forty-eight-genome team, superimposed all the different gene trees they had generated. The resulting image of the early evolution of modern birds, around the time the dinosaurs went extinct, was not a tidy series of diverging branches but a kind of web or fishnet, whose contours constantly crossed paths. In a paper, Suh urged his colleagues to consider other patterns of evolution—to argue “less about which species tree is ‘correct,’ and more about if there is such a thing” as a traditional tree of life for modern birds.
(,,,) Relatives often shacked up, braiding their separate lineages back together. Something similar happens in nature when one species mates with another, producing a hybrid. Although tree-thinking biologists used to think that hybridization was extremely rare, genetic studies have shown that it actually happens all the time. Human DNA indicates that early Homo sapiens interbred with Neanderthals and other extinct hominins. Conservative estimates suggest that at least ten per cent of birds hybridize; among South America’s largest group of birds, that number is thirty-eight per cent, according to one recent study.
Hybridization may have been rampant in the aftermath of the asteroid strike, when modern-bird lineages first emerged. Interbreeding would have passed genes from one branch of the tree to another, adding another layer of complexity on top of incomplete lineage sorting. “Lineages that split and never talk to each other again—that’s not how biology works,” Stiller said. Still, she remains hopeful that one day we may build an authoritative diagram of the past. “Our models are still comparatively simple,” she told me. “We should be able to reconstruct evolutionary history if we have the right models and the right data.”
The outlines of animal evolution still look a lot like a tree in many places, which is why scientists continue to spend so much time developing and debating different branches. But, if tree thinking taught biologists that everything is connected, genes are suggesting that the connections can run even deeper than a tree can capture. To gain a more complete picture—and to answer questions like how such an unusual mix of traits came together in the hoatzin—scientists may need to think outside the tree. B10K grew out of the forty-eight-genome group and now includes computer scientists who specialize not in trees but in networks; they try to track the movement of genes between branches, and they often find that even supercomputers aren’t yet up to the job.
In B10K’s preliminary analyses, the hoatzin again winds up closely aligned with cranes and shorebirds, but the conclusion lacks a hundred per cent statistical support. “There’s still a lot of conflict in the data,” Stiller said. “Depending on how you analyze it, you will get different placements.” After B10K finishes its tree for three hundred and sixty-three birds, it’ll move on to the more than two thousand avian genuses, and eventually to every species of bird. These genomes will create a much more complete portrait—but, even then, they may not be able to solve the mystery of the hoatzin, or reconstruct every crook in the early evolution of modern birds.
“If the evolutionary history of the hoatzin conformed to processes we already understand well, then we’d probably have already figured out what it is most closely related to,” Houde wrote via e-mail. “The fact that we don’t know its nearest relative suggests that there were processes involved that we still do not understand.” He indicated that the hoatzin could have more than one set of closest relatives—which he called “an unsettling prospect in the context of existing classification and in the minds of many contemporary biologists.”
This strange-sounding state of affairs is not unique to the hoatzin; we see it in our own DNA. Human beings share their most recent common ancestor with chimpanzees and bonobos, but more than ten per cent of the human genome is actually more closely related to the gorilla genome. Another tiny fraction of the human genome also seems to be most closely shared with an even more distant relative: the orangutan. “This implies that there is no such thing as a unique evolutionary history of the human genome,” a team of molecular biologists wrote in 2007. “Rather, it resembles a patchwork of individual regions following their own genealogy.”
Darwin ended “On the Origin of Species” with a famous description of “an entangled bank, clothed with many plants of many kinds, with birds singing on the bushes, with various insects flitting about, and with worms crawling through the damp earth.” Molecular biologists hoped that genes would reveal the true and final shape of Darwin’s tree. Instead, they found a new kind of entangled bank, in which species are connected in unexpected ways. “There is grandeur in this view of life,” Darwin wrote of his scene. There is grandeur, too, in the view of life that is encoded in DNA.
