Long before ancient DNA was a technical possibility, scientists were constructing models of human origins from stones and bones. The fossil record, despite its frustrating incompleteness, established certain broad contours that genetics has since both confirmed and complicated.

The genus Homo appears in the African fossil record roughly 2.5 to 3 million years ago, though the exact boundary is debated and somewhat dependent on how you define the genus. Homo erectus, which emerged around 1.9 million years ago, was the first hominin to leave Africa in significant numbers, spreading eventually across Asia and parts of Europe. What happened next — whether H. erectus populations in different regions evolved locally into modern humans, or whether modern humans emerged in Africa and replaced existing populations elsewhere — was the central controversy of twentieth-century paleoanthropology.

This controversy crystallized into two opposing camps. The multiregional hypothesis, associated most prominently with Milford Wolpoff and colleagues, proposed that modern humans evolved more or less simultaneously across the Old World, with gene flow between regional populations preventing the kind of divergence that would produce separate species. The competing Out of Africa hypothesis — sometimes called the Recent African Origin model — argued instead that anatomically modern humans emerged in Africa roughly 200,000 to 300,000 years ago and subsequently spread outward, replacing archaic populations they encountered along the way with little or no interbreeding.

For most of the 1980s and 1990s, the Out of Africa model was winning. Mitochondrial DNA studies, beginning with the landmark work of Rebecca Cann, Mark Stoneking, and Allan Wilson in 1987, traced the maternal ancestry of all living humans back to a single African woman — quickly and somewhat misleadingly dubbed "Mitochondrial Eve" — who lived roughly 200,000 years ago. Y-chromosome studies converged on a similar timeframe for the male lineage. The genetic evidence seemed to overwhelmingly favor African origin and, given how little diversity was found outside Africa, substantial replacement of earlier populations.

But the bones told a more complicated story, and a handful of paleoanthropologists kept insisting that continuity features — skeletal traits that seemed to link archaic and modern populations in specific regions — were too numerous to dismiss. They were not entirely wrong. They were just looking at the evidence through instruments too coarse to resolve what was actually happening.

The moment that changed everything arrived in 2010, when Svante Pääbo and his team at the Max Planck Institute for Evolutionary Anthropology published the first draft sequence of the Neanderthal genome. Neanderthals — Homo neanderthalensis — had lived in Europe and western Asia for at least 300,000 years before disappearing roughly 40,000 years ago, precisely around the time anatomically modern humans arrived in their territory. Whether this disappearance was the result of direct conflict, competition for resources, climate change, or some combination remained deeply unclear. What seemed clear, or at least assumed, was that Neanderthals and modern humans were separate lineages that had not meaningfully exchanged genes.

The 2010 paper demolished that assumption. Comparing the Neanderthal genome to those of modern humans from different parts of the world, Pääbo's team found that people with ancestry outside of sub-Saharan Africa carry roughly 1–4% Neanderthal DNA. Africans carried essentially none. The implication was clear: after modern humans left Africa but before they dispersed across the globe, there was interbreeding with Neanderthal populations. The most likely location for this encounter was somewhere in the Middle East or the Levant, where both populations overlapped geographically.

In 2013, the same group published the high-quality, complete genome of a Neanderthal woman from the Denisova Cave in the Altai Mountains of Siberia — the same woman whose presence opened this article. This genome, sequenced to greater depth than any previous ancient human, revealed extraordinary things. Her parents were related at the level of half-siblings, suggesting that inbreeding among close relatives was common in her recent ancestry. Her population had been genetically small and isolated for a very long time, leaving distinctive patterns of homozygosity — stretches of the genome where both copies of a chromosome carry identical sequences — that serve as a fingerprint of a population with limited diversity.

But the Altai Neanderthal genome also revealed something even more surprising: traces of gene flow from yet another archaic hominin group. The signals pointed toward what researchers would come to call Denisovans.

In 2010, the same cave that would later yield the high-quality Neanderthal genome gave up another secret. A small finger bone, so fragmentary that it could not be identified as Neanderthal or modern human on morphological grounds alone, turned out to carry a genome distinct from both. The owners of this bone — a population now called Denisovans — split from the lineage leading to Neanderthals perhaps 400,000 years ago, suggesting a long, separate evolutionary history across Asia.

What is remarkable about Denisovans is that we know them almost entirely through DNA. Beyond that finger bone, a few teeth, and a partial skull fragment, there is essentially no fossil record. And yet genetics tells us a great deal. Denisovan DNA persists in the genomes of living people, particularly those with ancestry from Oceania — Aboriginal Australians and Melanesians carry roughly 4–6% Denisovan DNA, the highest level of any population studied. Tibetans carry a Denisovan-derived variant of the gene EPAS1, which encodes a protein that helps regulate the response to low oxygen, and which appears to be a critical adaptation for living at high altitude. That specific variant, when traced backward through the genomic evidence, seems to have entered the modern human lineage through interbreeding with Denisovans, not through the normal processes of mutation and selection within modern humans alone.

The Denisovan story keeps getting stranger. A 2018 paper reported the genome of an individual, nicknamed Denny, whose mother was a Neanderthal and whose father was a Denisovan — the direct offspring of two archaic species that had long been classified as distinct. More recent work suggests there may have been multiple, independent Denisovan introgression events into different modern human populations, meaning that the mixing was not a single ancient incident but a recurring pattern across time and geography.

And then there is the question of the unknown. Analysis of the high-quality Altai Neanderthal genome found evidence of gene flow into the Denisovan lineage from a population not matching any known archaic human. Something else was out there. Whether this represents an extremely deep-diverging lineage of Homo erectus, or some other archaic population as yet unsampled, is genuinely unknown. We are catching shadows in data, outlines of beings whose bones we have not found.

It would be a mistake — a common one — to treat the African chapter of human origins as settled. Africa is where anatomically modern humans evolved, and it is also where human genetic diversity is deepest. But precisely because it is deepest, it is also most difficult to interpret with the tools available.

For decades, models of modern human origins located a single origin point in Africa — somewhere in East Africa, perhaps the Rift Valley region, perhaps around 200,000 years ago. The discovery of fossils at sites like Jebel Irhoud in Morocco, now dated to roughly 300,000 years ago, has pushed that timeline back and complicated the geography. These early modern humans, with their striking combination of modern and archaic features, suggest that Homo sapiens was not born at a single moment in a single place but emerged gradually across a continent-wide range of populations that were intermittently connected by gene flow.

Within Africa, ancient DNA evidence (harder to obtain than in colder climates, because DNA degrades faster in heat and humidity) has revealed layers of complexity that population genetics is still untangling. Some studies suggest that certain living African populations carry traces of introgression from archaic African hominins — populations that diverged from the main line long before modern humans emerged and that we have not yet identified from the fossil record. One analysis published in 2020 proposed that West African populations show signs of approximately 2–19% ancestry from a deeply divergent archaic human population, one that may have split from the lineage leading to modern humans around 500,000 years ago.

The concept of population structure — the way geographically separated groups accumulate genetic differences over time, then merge, then separate again — may be more important to understanding modern human origins than the idea of a single founding population. Some researchers now speak of pan-African multi-regionalism: the idea that modern humans emerged from a network of semi-isolated African populations that exchanged genes periodically across hundreds of thousands of years, producing the mosaic of features we see in both the fossil and genetic records.

This is not the same as the old multiregional hypothesis, which encompassed Homo erectus populations across Eurasia. It is a more circumscribed and better-supported claim. But it shares with the old hypothesis an emphasis on complexity and connection rather than singular origin and replacement.

Around 60,000–70,000 years ago — though some evidence pushes this earlier, and there may have been multiple waves — a group of modern humans left Africa and began spreading across the globe. Their descendants would eventually occupy every habitable landmass on Earth, from the Arctic to Tierra del Fuego, from the Pacific islands to the Australian desert. Understanding the routes, timing, and population dynamics of this dispersal is one of the central projects of population genetics.

The genetic evidence for a single major Out of Africa dispersal rests on patterns of decreasing diversity as you move away from Africa — a phenomenon called the serial founder effect. When a small group leaves a larger population to found a new one, it carries only a subset of the original genetic diversity. If that new group subsequently sends out its own founders, diversity decreases further. The further a population is from Africa, the less genetic diversity it tends to show — a pattern that is broadly consistent with a sequential dispersal model.

But the picture has been complicated by ancient DNA studies suggesting earlier, perhaps separate dispersals. Modern humans were present in the Levant by at least 180,000 years ago (Misliya Cave, Israel), and in China at Fuyan Cave by at least 80,000–120,000 years ago. Whether these early forays represent ancestors of living non-Africans, dead-end populations that left no descendants, or some combination, is debated. One influential view holds that there were multiple dispersal waves, and that earlier waves were largely replaced by the later, more successful one — though leaving traces in some living populations.

The picture for first Americans is particularly complex. The prevailing model holds that the ancestors of Native Americans crossed from Siberia to Alaska via a land bridge (Beringia) that was exposed during the glacial maximum, when sea levels were lower. Genetic evidence largely supports this, pointing to East Asian ancestry for most Native American populations. But there are anomalous signals: some Amazonian populations show affinities with Australasian populations that are difficult to explain under a simple single-wave model, suggesting either an earlier, separate migration or a very complex pattern of ancestry within Asia before the crossing.

The Pacific presents its own puzzle. The settlement of Remote Oceania — the islands of the central and eastern Pacific — was one of the most remarkable feats of human navigation in prehistory, accomplished within the last 3,000 years by the ancestors of Polynesian peoples. Genetic and archaeological evidence traces this expansion to Taiwan, through the Philippines, into the Pacific islands, and eventually to Hawaii, Easter Island, and New Zealand. But those Melanesian and Aboriginal Australian populations with high Denisovan ancestry sit at the edge of this story, raising questions about layered histories and encounters that the genetic record is slowly, painstakingly beginning to resolve.

Knowing that modern humans carry Neanderthal and Denisovan DNA is intellectually fascinating, but the question with immediate stakes is: does it matter for our biology today? The answer, increasingly, appears to be yes — in some cases surprisingly so.

Adaptive introgression — the process by which genes flowing in from another population are favored by natural selection because they provide some advantage — seems to have been a significant feature of the encounter between modern humans and archaic populations. Consider the challenge facing early modern humans as they moved into environments occupied by Neanderthals and Denisovans: those archaic populations had lived in those environments for hundreds of thousands of years. They had accumulated local adaptations — to climate, to pathogens, to altitude — that modern humans lacked. Acquiring some of those adaptations through interbreeding would have been, in evolutionary terms, an enormous shortcut.

The Tibetan EPAS1 example is perhaps the clearest case of Denisovan-derived adaptive introgression. More broadly, several immune system genes — particularly those in the HLA complex (human leukocyte antigen), which helps the immune system distinguish self from non-self and plays a role in fighting pathogens — appear to have been inherited at least in part from Neanderthals and Denisovans. The pathogens that Neanderthals had long coexisted with would have been, to newly arrived modern humans, novel and potentially devastating. Inheriting immune recognition machinery already calibrated to those pathogens may have been a lifesaving advantage.

But not all Neanderthal inheritance appears beneficial. Research published in recent years has linked Neanderthal-derived variants to increased risk for certain conditions in modern populations, including depression, blood coagulation, and aspects of immune response that can go wrong. A striking finding from a 2020 study identified a major genetic risk factor for severe COVID-19 outcomes that was derived from Neanderthal ancestry — carried at relatively high frequency in South Asian populations and at low frequency in European populations, but absent in East Asian and African populations. The same Neanderthal-derived segment of chromosome 3 also appears to be associated with reduced risk of HIV infection. Ancient gifts can have complicated legacies.

It is worth noting that the distribution of Neanderthal ancestry in the modern human genome is not random. Certain regions of the genome are virtually devoid of Neanderthal sequences — particularly regions involved in spermatogenesis and regions on the X chromosome. This pattern suggests that some Neanderthal-modern human hybrid individuals had reduced fertility, particularly males, and that selection gradually purged certain Neanderthal variants from the gene pool over the generations that followed interbreeding. This is consistent with what evolutionary theory would predict for the offspring of populations that had been separated for hundreds of thousands of years: reproductive compatibility, but not perfect compatibility.

The methods that have made this revolution possible deserve their own examination, not because most readers need to understand them technically, but because appreciating what they can and cannot do helps calibrate appropriate confidence in the results.

Paleogenomics — the sequencing and analysis of ancient genomes — is technically demanding in ways that modern genomics is not. Ancient DNA is degraded, fragmented, and contaminated. In a bone that has lain in the ground for 50,000 years, the original DNA is in short, broken pieces, chemically modified at the ends by a process called deamination, and mixed with vast quantities of microbial DNA from the soil. Early ancient DNA work, in the 1980s and 1990s, was plagued by contamination — researchers were sequencing their own DNA, or the DNA of microbes in their labs, without realizing it. The field developed stringent authentication protocols, and the advent of next-generation sequencing technology in the 2000s changed everything: it became possible to sequence millions of short, degraded fragments simultaneously and use the characteristic damage patterns of ancient DNA as a verification signature.

The analysis side presents its own challenges. Comparing ancient and modern genomes requires sophisticated statistical models that account for the complex population histories, demographic events, and selection pressures that have shaped genetic diversity over time. Methods like D-statistics (also called ABBA-BABA tests) allow researchers to detect gene flow between populations by looking at patterns of shared derived variants. Admixture modeling tries to decompose modern genomes into proportions of ancestry from different source populations. Coalescent theory provides frameworks for inferring population sizes and divergence times from patterns of genetic variation. These are powerful tools, but they rest on assumptions — about mutation rates, generation times, the structure of ancestral populations — that introduce uncertainty at every step.

This uncertainty is worth honoring. The confidence intervals around dates and admixture proportions in ancient DNA papers are often wide. Models that fit the data well are not necessarily unique — different population histories can sometimes produce similar genetic signatures. When researchers report that a population diverged from another "approximately 400,000 years ago," the word "approximately" is doing heavy lifting. These numbers are best understood as order-of-magnitude estimates embedded in ongoing revision, not as established facts with the precision of carbon dating.

For all that ancient DNA has revealed, the horizon has not cleared — it has expanded. Here are some of the most consequential questions that remain genuinely open.

Who were the archaic Africans? Multiple lines of genetic evidence now suggest that one or more deeply divergent archaic hominin populations contributed genes to the ancestors of living Africans, but we have not yet identified the fossil remains of these populations. Were they a late-surviving form of Homo heidelbergensis? A regional variant of Homo naledi, whose remains were found in South Africa and dated to as recently as 230,000 years ago? We do not know. Africa's ancient DNA record is the most important and the most incomplete, because tropical conditions destroy DNA more rapidly. Better preservation — in caves, in arid regions — may eventually yield genomes from these populations, but for now they remain known only as shadows in modern genomes.

Was there a single Out of Africa dispersal, or many? The genetic evidence for a major dispersal around 60,000–70,000 years ago is strong, but earlier evidence of modern humans in Arabia, the Levant, and East Asia complicates the picture. Were these earlier populations completely replaced, or did they leave genetic traces in some living populations? The anomalous Australasian-like genetic signal in some Amazonian populations has not been fully explained. Multiple dispersals with complex interactions between waves remain a real possibility.

How did the encounter with Neanderthals and Denisovans actually unfold, culturally and ecologically? Genetics can tell us that interbreeding happened, but it cannot easily tell us whether these encounters were peaceful or violent, frequent or rare, whether they involved extended cohabitation or brief contacts. Archaeological evidence of behavioral modernity — art, complex tools, symbolic objects — in both Neanderthal and modern human contexts raises the possibility that the cognitive gulf between these populations was smaller than once assumed. Neanderthals made jewelry, created pigments, perhaps made music. What does it mean that we carry their DNA while they are gone?

What does Neanderthal and Denisovan ancestry actually do to living human phenotypes? The examples of adaptive introgression we have so far — altitude adaptation in Tibet, certain immune variants, the COVID-19 risk locus — are almost certainly the tip of an iceberg. Systematic studies of how archaic ancestry variants affect complex traits like cognition, metabolism, disease resistance, and sensory perception are still in early stages. The picture will become much clearer over the next decade, but the full inventory of what we inherited, and what it costs or confers, remains unmapped.

What was the cognitive and behavioral profile of Denisovans? Because Denisovans are known almost entirely from DNA, we have essentially no behavioral or physical profile for them beyond scattered dental and skeletal fragments. A fragment of skull found in China, tentatively associated with Denisovans, suggests a population quite different in cranial morphology from both Neanderthals and modern humans. What tools did they make? What did they eat? How widely did they range across Asia? The discovery of more Denisovan fossil material with associated archaeology would transform our understanding — but so far, the genome is outrunning the bones.

The Siberian woman in her cave left no account of her life. She did not know that her genome would one day be read by the descendants of people she never encountered, in a world she could not have imagined. And yet the text she left behind — four billion letters of molecular code, degraded but legible — has done more to reshape our understanding of human origins than a century of fossil hunting. What it reveals is not a clean, linear story of one species arising and replacing another, but something messier and more generous: a web of encounters, exchanges, and minglings that produced, among many possibilities, us. The boundaries we draw around Homo sapiens turn out to be lines drawn in water. We are, all of us, the offspring of meetings we forgot.