Denisovans had wide skulls, elongated faces, protruding jaws, and unusually large teeth, according to a 2019 reconstruction that used ancient DNA methylation patterns to predict skeletal anatomy. Because only a handful of bone fragments and teeth have ever been confidently identified as Denisovan, almost everything we know about their appearance comes from this single genetic technique and from a small but growing list of fossil candidates scattered across Asia. The picture is far from complete, but it is surprisingly detailed for a human group that was unknown to science until 2010.
How Scientists Built a Face from Chemistry, Not Bone
The central challenge with Denisovans is that they left behind almost no recognizable skeletal remains. The group was first identified from a pinky bone fragment found in a Siberian cave, from which researchers sequenced a genome at roughly two-fold coverage. That genome revealed a population distinct from both modern humans and Neanderthals, but a finger bone tells you nothing about what someone’s face looked like.
In 2019, a team led by David Gokhman at the Hebrew University of Jerusalem published a method that sidestepped the fossil shortage entirely. Instead of looking at the DNA sequence itself, they looked at DNA methylation, the chemical tags that sit on top of genes and regulate whether those genes are turned on or off. Certain methylation changes are known to cause specific skeletal abnormalities when they occur in living humans. The researchers reasoned that if a gene governing, say, brow ridge size showed a distinctive methylation pattern in Denisovans compared to modern humans and Neanderthals, that gene was probably regulated differently, and the anatomy it controlled probably looked different too.1Cell. Reconstructing Denisovan Anatomy Using DNA Methylation Maps
To check whether this actually worked, the team first tested it on Neanderthals and chimpanzees, two groups whose skeletons are well known. The method correctly identified divergent skeletal traits with better than 85% precision.2PubMed Central. Candidate Denisovan fossils identified through gene regulatory phenotyping That gave the researchers enough confidence to apply it to Denisovans. The result was a list of 32 skeletal features predicted to differ between Denisovans and their closest relatives. The predictions are qualitative rather than quantitative: they tell you the direction of a change (wider, longer, more robust) but not the exact measurement.
What the Reconstruction Predicts
The methylation analysis found that most predicted Denisovan traits overlapped with Neanderthal features, which makes sense given that the two groups shared a common ancestor. Twenty-one out of 32 predicted traits were shared with Neanderthals, including robust jaws, a low and elongated cranium, a low forehead, thick tooth enamel, a wide pelvis, large joints in the thigh bones, broad fingertips, and a large ribcage.3Cell. Reconstructing Denisovan Anatomy Using DNA Methylation Maps In broad strokes, Denisovans probably had a stocky, powerfully built body not unlike a Neanderthal’s.
The remaining traits, however, set Denisovans apart. The reconstruction flagged an increased dental arch, meaning the jaw was wider from side to side than in either Neanderthals or modern humans. It also predicted lateral cranial expansion: the skull was broader, flaring outward at the sides. The face was long, the pelvis was wide even by Neanderthal standards, and the overall cranial shape was distinct enough that the researchers suggested a trained anatomist could tell a Denisovan skull from a Neanderthal one.4Cell. Reconstructing Denisovan Anatomy Using DNA Methylation Maps – Section: Results When the team compared their predictions against the Xuchang skulls from China, which some researchers had already proposed might be Denisovan, the match was encouraging.
The Teeth Tell Their Own Story
Teeth preserve far better than other bones, and Denisovan molars are among the few physical specimens scientists can study directly. The pattern is consistent: Denisovan teeth are big. A molar recovered from a cave in the Annamite mountains of Laos, dated to the Middle Pleistocene, had crown dimensions that statistically differed from the smaller teeth of modern humans and European Middle Pleistocene groups, while falling within the range of Asian archaic humans and Neanderthals.5Nature Communications. A Middle Pleistocene Denisovan molar from the Annamite Chain of northern Laos
The methylation reconstruction did not specifically predict larger molars, partly because the medical database it relied on did not include phenotypes specific to molar size. It did, however, predict a longer dental arch, which the researchers noted could be anatomically linked to accommodating larger teeth.6Cell. Reconstructing Denisovan Anatomy Using DNA Methylation Maps – Section: Discussion The molars from Denisova Cave itself are significantly broader than those of modern humans and mostly outside the Neanderthal range as well. In other words, the physical fossils and the methylation predictions point in the same direction, even where they do not perfectly overlap.
A mandible from Taiwan known as Penghu 1, recently confirmed as Denisovan through ancient protein analysis, reinforces this picture. The jaw is distinctly robust, with large teeth, and the researchers who identified it noted that the “robust dentognathic traits” of confirmed Denisovan specimens markedly contrast with those of Neanderthals.7PubMed. A male Denisovan mandible from Pleistocene Taiwan So while Denisovans shared much of the Neanderthal body plan, their jaws and teeth were their own.
Candidate Fossils Across Asia
For years, the only confirmed Denisovan fossils came from a single cave in Siberia’s Altai Mountains. That has changed. The Penghu mandible from Taiwan expanded the confirmed range into a subtropical island environment, demonstrating that Denisovans lived in warm and humid regions, not only in cold montane settings.7PubMed. A male Denisovan mandible from Pleistocene Taiwan Ancient proteins, rather than DNA, provided the identification in that case, opening a new avenue for matching fossils to lineages in environments where DNA does not survive well.
Beyond confirmed specimens, a growing roster of Chinese fossils has been proposed as possibly Denisovan based on morphological similarities. A review of the anthropological evidence found that molars from the Chinese sites of Xujiayao and Hexian showed the greatest similarity to the Altai Denisovans. Skulls from Dali and Jinniushan, as well as the massive Harbin cranium, are also considered related forms. The postcranial remains from Jinniushan suggest specific skeletal proportions that may reflect genetic adaptation to cold temperatures, and the overall build of these candidate specimens is described as distinctly massive.8Russian Archaeology. DENISOVANS BEYOND ALTAI (A REVIEW BASED ON ANTHROPOLOGICAL EVIDENCE)
The geographic pattern these candidates trace is telling. Rather than being spread evenly across the continent, the proposed Denisovan habitat based on fossil locations shows a “compact” pattern concentrated in predominantly mountainous landscapes of eastern China during the second half of the Middle Pleistocene. That concentration is consistent with a population adapted to rugged, varied terrain, though the Penghu and Laos finds show that Denisovans were not limited to highlands.
The Dragon Man Question
One of the most contentious debates in paleoanthropology right now is whether the Harbin cranium, popularly known as “Dragon Man,” represents a Denisovan. The Harbin skull is enormous, with a braincase roughly as large as a modern human’s, a wide mid-face, and heavy brow ridges. In 2021, a team formally named it Homo longi and argued it was a new species more closely related to modern humans than Neanderthals are.
A 2025 study examining the Yunxian cranium, another massive Chinese fossil, placed it as an early member of the Asian H. longi clade through morphometric and phylogenetic analyses and suggested that this clade includes the Denisovans and forms the main part of the sister group to the modern human clade.9PubMed. The phylogenetic position of the Yunxian cranium elucidates the origin of Homo longi and the Denisovans If that analysis holds up, Dragon Man may not be a separate species from Denisovans at all but rather a member of the same lineage. The review of candidate fossils supports this, noting that the Harbin skull and the Dali skull are similar to the methylation-based Denisovan predictions, and that the relative similarity between Harbin, Xuchang, Dali, and Jinniushan specimens expands the range of what a Denisovan might look like.8Russian Archaeology. DENISOVANS BEYOND ALTAI (A REVIEW BASED ON ANTHROPOLOGICAL EVIDENCE)
This matters for the reconstruction because, if Dragon Man is a Denisovan, we suddenly have a nearly complete skull to work with. It would mean Denisovans had a face that was recognizably human in its proportions, with a brain as large as ours, but with heavier brow ridges and a broader, more massive overall build. The debate is far from settled, though. DNA has not yet been extracted from the Harbin cranium, so the connection remains based on shape comparisons and phylogenetic modeling rather than molecular confirmation.
What the Method Cannot Tell Us
The methylation reconstruction is a genuinely novel approach, but it has clear limits that are worth understanding. The predictions describe bone, not soft tissue. Skin color, hair texture, eye shape, fat distribution, ear form: none of these leave a methylation signature in the kind of data available from ancient specimens. Any illustration that shows a Denisovan with particular skin tone or hair is an artist’s interpretation layered on top of what the science actually provides.
The method also works only in one direction. It can tell you that the Denisovan skull was wider than a modern human’s, but it cannot tell you how much wider. A 2025 study applying the same approach to identify candidate fossils reiterated that the phenotypic predictions are qualitative rather than quantitative, providing the direction of change but not its precise magnitude.2PubMed Central. Candidate Denisovan fossils identified through gene regulatory phenotyping A skull that is 5% wider and one that is 20% wider would both match the prediction equally well.
Validation of the method’s precision has been extended beyond skeletal tissue. Researchers testing the algorithm on non-skeletal tissues in extant species obtained a precision of 0.85 for predicting reference-derived changes, compared with roughly 0.22 in randomized control data, and 0.77 for outgroup-specific changes versus 0.40 in controls.10Nature Ecology & Evolution. Inferring DNA methylation in non-skeletal tissues of ancient specimens Those numbers are far above chance, but they also mean that roughly one in five to one in four predictions could be wrong. For any individual trait on the list of 32, there is a real possibility the prediction is off.
Built for Thin Air and Cold
One thing we do know about Denisovans with high confidence is that at least some populations lived in extreme environments. The clearest evidence is the EPAS1 gene carried by modern Tibetans, which regulates the body’s response to low oxygen. This gene variant shows strong signatures of both positive selection and introgression from Denisovans, meaning Tibetans inherited it through ancient interbreeding and it helped them thrive at altitudes above 4,000 meters.11PubMed Central. The history and evolution of the Denisovan-EPAS1 haplotype in Tibetans
For Denisovans to have evolved a gene that confers high-altitude advantage, some portion of the population must have lived at elevation long enough for natural selection to act. This fits with the fossil geography: the Xiahe mandible, one of the few confirmed Denisovan specimens outside Siberia, was found on the Tibetan Plateau at about 3,280 meters. The candidate fossils from Jinniushan, with their skeletal proportions suggesting cold adaptation, add another dimension. The emerging picture is of a population that ranged across a variety of Asian environments, from Siberian caves to tropical lowlands to the roof of the world, and that different Denisovan populations may have looked quite different from each other as a result of local adaptation.
What Denisovan DNA Does in Living People
The reconstruction is not just an exercise in paleontology. Denisovan DNA persists in modern human genomes, and it has measurable effects on health and appearance today. Populations in East and Southeast Asia, as well as Melanesian and Aboriginal Australian groups, carry the highest proportions of Denisovan ancestry, typically between 2% and 5% depending on the population and the method of measurement.
A genome-wide study of archaic introgression found that in East Asian populations, Denisovan-derived genetic variants were enriched for heritability related to coronary artery disease, with about a 1.7-fold enrichment. The same study identified a specific Denisovan-introgressed variant in the major histocompatibility complex region associated with changes in albumin-to-globulin ratio, a marker of immune and liver function.12PubMed Central. Denisovan and Neanderthal archaic introgression differentially impacted the genetics of complex traits in modern populations In other words, Denisovan ancestry does not just tell us about the past; it shapes immune responses and disease risk in the present.
The high-altitude EPAS1 variant is the most dramatic example, but researchers have linked Denisovan-introgressed DNA to traits ranging from fat metabolism to immune cell signaling. Intriguingly, the effects of Denisovan and Neanderthal introgression are not interchangeable. The same study found that the two archaic lineages differentially impacted the genetics of complex traits, suggesting that Denisovan biology was distinct enough from Neanderthal biology to leave a different signature in the people who inherited it.
How Denisovans Might Have Sounded
One question the methylation reconstruction cannot address is what Denisovans sounded like. Research on the vocal tract and auditory anatomy of archaic humans has focused primarily on Neanderthals, for whom hyoid bones and ear ossicles have been directly preserved. Functional analyses of Neanderthal hearing structures suggest their audition was very similar to that of modern humans and clearly different from that of chimpanzees.13PubMed Central. The vocal tract as a time machine: inferences about past speech and language from the anatomy of the speech organs No equivalent structures have been preserved from Denisovans, so any statement about Denisovan speech or hearing is speculative.
Given how closely related Denisovans and Neanderthals were, sharing a common ancestor roughly 400,000 to 500,000 years ago, it is reasonable to guess that their vocal and auditory anatomy was broadly similar. But guessing is all we can do for now. The Denisova Cave finger bone, the Xiahe mandible, the Penghu jaw, and a handful of teeth are not the parts of the body that tell you about speech. Until someone finds a Denisovan hyoid or temporal bone, the question of whether they could speak, sing, or vocalize in ways we might recognize remains open.
Why the Reconstruction Keeps Changing
The Denisovan reconstruction is not a fixed portrait. Every new fossil and every methodological refinement adjusts the picture. In the six years since the original methylation study, the field has gained the Penghu mandible, a confirmed Denisovan specimen from a completely different environment with robust jaw features that “markedly contrast” with Neanderthals.7PubMed. A male Denisovan mandible from Pleistocene Taiwan The same group that developed the methylation method has extended it to identify candidate fossils by scoring whether unattributed specimens match the predicted Denisovan phenotypic profile, turning the reconstruction into a forensic tool rather than just an illustration.
What makes this field unusual is the degree to which identity and appearance are being worked out in reverse order. For every other hominin species, morphology came first: someone found a skull, described it, and named a species. For Denisovans, genetics came first. We knew their DNA before we had any real idea what their bones looked like. The reconstruction is essentially an effort to run the tape backward, from genome to phenotype, using a tool that did not exist ten years ago. That means the portrait we have today is provisional in ways that a Neanderthal or Homo erectus reconstruction is not. But it also means the rate of discovery is unusually fast. Ancient protein analysis is now identifying Denisovan specimens that DNA alone could never have reached. Methylation mapping is being extended to new tissue types. And new fossil candidates continue to surface across eastern Asia. The next decade will almost certainly revise the picture substantially, but for now, we can say with reasonable confidence that Denisovans were broad-skulled, heavy-jawed, large-toothed people whose bodies were built for endurance in some of the harshest landscapes on Earth.