Ghosts in the Sediment: Finding Ancient People Who Left No Bones Behind

A human signal in an empty layer

At Denisova Cave in southern Siberia, the imbalance is striking. Excavation has produced a small collection of hominin fossils, yet hundreds of sediment samples contain traces of hominin DNA. In a cave occupied intermittently over hundreds of thousands of years, much of the human history is therefore recorded not in skeletons but in the soil surrounding them.

The distinction is more than technical. Archaeologists have always inferred absent people from what they left behind: a hearth, a retouched flake, butchered bone, a footprint, an activity surface. Sedimentary ancient DNA, usually abbreviated sedaDNA, introduces another kind of evidence. Human biological material can survive in deposits even when no recognisable human bone or tooth has survived, or has yet been excavated.

That possibility became difficult to ignore in 2017, when Viviane Slon and colleagues reported mitochondrial DNA from Pleistocene cave sediments at sites where Neanderthals and Denisovans had lived. They analysed 85 sediment samples from seven archaeological sites, ranging in age from roughly 14,000 years to at least 550,000 years. Neanderthal mitochondrial DNA was recovered from several archaeological layers at four caves; at Denisova Cave, Denisovan DNA was also detected. Crucially, some positive samples came from layers in which no hominin skeletal remains had been found.

This did not mean that archaeologists could extract a complete human genome from any handful of cave earth. The earliest applications relied heavily on mitochondrial DNA, or mtDNA. Each cell contains many copies of the mitochondrial genome, making it much easier to recover from degraded material than the nuclear genome. It is extremely useful for identifying broad genetic lineages and detecting the presence of hominins, but it represents only the maternally inherited component of ancestry. Population history reconstructed from mtDNA alone can therefore be misleading if treated as though it describes an entire population.

Nor is ancient DNA simply dissolved evenly through a sediment layer. It may derive from decomposed tissue, microscopic fragments of bone, faecal material or other biological residues; some extracellular DNA can also bind to mineral surfaces. Its survival depends on chemistry, temperature, moisture, microbial activity and the history of the deposit after burial. The archaeological problem begins before the sample reaches a genetics laboratory.

The 2017 result was nevertheless consequential because it changed the scale of possible evidence. A rare fossil represents one individual at one location. Sediment can be sampled systematically across a sequence. In principle, the occupation history of a cave can be investigated even through periods from which no human fossil has survived.

Denisova without the Denisovans

Denisova Cave provided the clearest test of what this might mean in practice. The site already occupied an unusual place in human evolutionary research. A tiny finger-bone fragment excavated there yielded the genome of a previously unknown human population, now called Denisovans; subsequent genetic and fossil discoveries showed that Neanderthals also used the cave. Yet the number of hominin bones remained minute compared with the depth and complexity of the archaeological sequence.

In 2021 Elena Zavala and colleagues reported an unusually intensive sediment-DNA study of the cave. They analysed 728 samples, many collected on a grid at intervals of roughly 10–15 centimetres through exposed stratigraphic profiles. Ancient faunal mitochondrial DNA was recovered from 685 samples and hominin mtDNA from 175. The number of sediment samples carrying hominin DNA exceeded the number of known hominin fossils from the sequence by more than an order of magnitude.

The resulting history is far more detailed than the skeletal record alone would allow. The earliest hominin mitochondrial DNA detected in the sequence was Denisovan. In deposits associated with the early Middle Palaeolithic, approximately 250,000–170,000 years ago, Denisovan signals predominate, although Neanderthal DNA appears toward the end of this interval. Later parts of the sequence contain both Denisovan and Neanderthal mtDNA, with an interval in which Neanderthal signals dominate, followed by renewed Denisovan evidence. Modern human mtDNA occurs in Initial Upper Palaeolithic deposits dating to at least about 45,000 years ago.

Here the distinction between observation and interpretation is important.

The observation is that DNA assignable to particular hominin mitochondrial lineages occurs in sediments whose stratigraphic positions and approximate ages can be established.

The interpretation is that those genetic signals represent occupation of the cave by Denisovans, Neanderthals or modern humans during the formation of those deposits.

A further inference is that particular archaeological assemblages were made by whichever population is genetically represented in the same part of the sequence. That may be reasonable, and at Denisova it is sometimes the best explanation available, but it is not logically equivalent to finding a tool in the hand of its maker. Deposits accumulate over time; material can move; two populations might use a cave within the chronological resolution of a single layer.

For the earliest Middle Palaeolithic levels, for example, Denisovan DNA is much more frequent than Neanderthal DNA, making Denisovans strong candidates for the principal occupants associated with those assemblages. The evidence does not require that every artefact in those levels was made by a Denisovan. The chronology matters here, as does the physical integrity of each deposit.

The same caution applies to an apparently suggestive ecological pattern. Denisova’s sediment samples also produced abundant animal DNA. The sequence records substantial changes in faunal communities, and some of these broadly coincide with changes in the hominin genetic record. Environmental change may have influenced who occupied the Altai and when. But temporal coincidence is not, by itself, evidence that climate or changing animal communities caused a human population turnover. The sediment provides the association; causation requires a larger argument.

From mitochondrial traces to population history

The first sediment-DNA studies could answer a relatively simple but archaeologically valuable question: was a particular hominin lineage present? The more difficult ambition was to recover enough nuclear DNA to ask questions normally addressed with human fossils: how closely related were these populations, were they genetically replaced, and how did they fit into wider population histories?

In 2021 Benjamin Vernot and colleagues demonstrated that nuclear hominin DNA could be recovered and analysed from Pleistocene cave sediments. Working with deposits in Spain and Siberia, they obtained enough chromosomal DNA to examine Neanderthal population relationships. At Estatuas Cave in northern Spain, the genetic record indicated a population replacement around 100,000 years ago, broadly corresponding to a change also visible in mitochondrial lineages. Sediment had moved from identifying a visitor to documenting demographic history.

The methodological importance of the Siberian material was that sediment results could be compared with genomes from actual bones recovered at Denisova and Chagyrskaya caves. The genetic affinities of sediment-derived DNA were consistent with locally recovered skeletal genomes, an important validation of the method.

An Upper Palaeolithic example from Satsurblia Cave in Georgia went further. Pere Gelabert and colleagues recovered genome-scale human DNA directly from a roughly 25,000-year-old sediment sample, together with wolf and bison DNA. The human sequence was sufficiently informative to contribute to discussion of ancestry among populations in the Caucasus, including a component related to the deep Eurasian population history usually described as Basal Eurasian ancestry. No human skeleton was required for that genetic observation.

Sediments have also widened the known geographical record of groups represented by very few fossils. At Baishiya Karst Cave on the Tibetan Plateau, Dongju Zhang and colleagues recovered Denisovan mitochondrial DNA from Late Pleistocene deposits, with signals around 100,000 and 60,000 years ago and possibly later. This evidence sits alongside the Xiahe mandible from the region, identified as Denisovan through palaeoproteomic analysis rather than ancient DNA. Together, independent classes of evidence establish a Denisovan history well beyond the Altai and show repeated occupation of a demanding high-altitude environment.

It is tempting to turn such results immediately into stories about adaptation, migration or technology. The genetic evidence is usually narrower. Sedimentary mtDNA can demonstrate a population's presence within the limits imposed by stratigraphy and dating. It cannot by itself reveal why those people were there, how many there were, how long they stayed or which particular subsistence strategies they practised. Those questions remain archaeological ones, requiring artefacts, fauna, environmental evidence, spatial patterning and chronology.

The molecule still has a stratigraphy

The seductive phrase “DNA from dirt” makes the method sound easier than it is. Archaeologists would be rightly suspicious of a diagnostic artefact whose precise provenience was unknown. Genetic material deserves the same suspicion.

Ancient-DNA sampling therefore begins as a field procedure. At Denisova, profiles were cleaned before sampling and sediments were collected with sterile tools while researchers wore protective equipment intended to reduce modern contamination. Laboratory extraction and library preparation take place under dedicated ancient-DNA conditions, and researchers assess characteristic patterns of molecular degradation alongside sequence identity.

Modern contamination is only one problem. Ancient DNA can itself be out of place.

Sediments are not sealed archival drawers. Water moves through them. Animals burrow. People dig pits. Sediment slumps or is redeposited. Freeze-thaw action and other geological processes deform layers. A sample may also average material accumulated over centuries or millennia. Denisova’s investigators explicitly noted gaps in the sampled sequence, evidence of disturbance in some deposits and limits imposed by the precision of optical dating.

Vera Aldeias and Mareike Stahlschmidt have argued that sediment DNA must therefore be treated within the same geoarchaeological and taphonomic framework applied to other archaeological evidence. The nanoscale character of DNA does not exempt it from site-formation processes. If sediment has been transported, mixed or redeposited, the DNA it carries may have travelled with it. Micromorphology and detailed study of depositional history can consequently be as important to interpretation as sequencing depth.

Research on intact blocks of archaeological sediment has made this physical relationship more tangible. Massilani and colleagues examined resin-impregnated sediment blocks collected from Pleistocene sites and found ancient mammalian and hominin DNA concentrated in microscopic “hot spots”. Some were associated with tiny fragments of bone or faecal material. In reasonably intact sediments, DNA could remain spatially localised at the microscale rather than being distributed indiscriminately through an entire layer.

That finding is reassuring, but only conditionally. It suggests that sedimentary DNA can preserve fine archaeological associations when the deposit itself is intact. It also means that sampling matters enormously. Two subsamples taken centimetres apart need not contain the same biological history.

This makes absence particularly difficult to interpret. Failure to detect Neanderthal DNA in a sample does not establish that Neanderthals were absent from the cave at that time. They may have contributed little DNA at that point; preservation may have been poor; the sampled patch may simply have missed the microscopic material in which DNA survived. Conversely, the amount of DNA recovered is not a census. Different activities, bodily sources and preservation histories affect how much genetic material enters the sediment.

A recent study at El Mirón Cave in northern Spain illustrates both the power and the sampling problem. Thirty-two sediment samples spanning Late Pleistocene occupations produced DNA from humans and numerous animal taxa, including taxa not represented in the excavated zooarchaeological assemblage. Yet the sampled area represented only a small part of the cave vestibule. Genetic presence can supplement the bone record very effectively; relative abundance and apparent absence are more treacherous.

Presence is not authorship

The frontier is now moving beyond cave floors.

In June 2026, Alba Bossoms Mesa and colleagues reported experiments recovering ancient human DNA from cave walls and rock-art contexts in Spain and Portugal. Among 24 panels in eleven caves, a pigmented calcite crust at Escoural Cave yielded ancient human mitochondrial and nuclear DNA without corresponding faunal DNA, a pattern consistent with direct human contact. Other wall samples produced mixtures of human and animal material more plausibly explained by indirect transfer, perhaps involving sediment.

The attractive interpretation is obvious: DNA on a painted surface might identify the people who made the art.

At present, that goes beyond the evidence. The chronology of the genetic material cannot necessarily be tied closely enough to the application of pigment. A person might have touched the wall long before or long after a motif was created. The investigators themselves stress that the recovered DNA cannot yet conclusively be assigned to the act of painting.

This is precisely the sort of distinction that will determine whether sedimentary genetics becomes merely a source of spectacular claims or a mature archaeological method. A molecular trace can establish biological presence. Behaviour requires contextual linkage. Authorship requires a tighter linkage still.

The issue is familiar from other branches of archaeology. Residue inside a vessel does not automatically identify who consumed it. A fingerprint on a pot does not necessarily identify the potter. A burial assemblage does not transparently reproduce the social identity of the deceased. DNA is unusually powerful evidence, but it does not abolish inference.

What a ghost can prove

The phrase “people who left no bones behind” is slightly misleading. Some may have left bones that decayed; others may be represented by microscopic skeletal particles too small to recognise during excavation; still others may have left bodily material whose DNA became bound within the sediment. What is new is not that humans leave biological traces wherever they live, but that archaeology can now recover and identify traces that previously belonged to the invisible fraction of the record.

The safest conclusions are already substantial.

Sedimentary ancient DNA can demonstrate that hominin genetic material is present in archaeological deposits even where no identifiable human fossil has been recovered. Mitochondrial DNA can distinguish major lineages such as Neanderthals, Denisovans and modern humans. Dense sampling through long stratigraphic sequences can reveal changes in occupation invisible in the skeletal record. Under favourable conditions, nuclear DNA can reconstruct population relationships and even contribute genome-scale ancestry information. When human and animal DNA are recovered from the same sequence, archaeologists gain a parallel record of occupants and aspects of their biological environment.

What it cannot routinely do is equally important. A positive sample does not automatically identify the maker of an artefact assemblage. DNA quantity does not translate simply into population size. A negative sample does not prove absence. Genetic proximity within a layer does not guarantee exact contemporaneity with a hearth, tool or painting. No sequencing method can rescue an interpretation from poorly understood stratigraphy.

This is why the most productive future for sedaDNA is likely to be less spectacular than the phrase “DNA in dirt” suggests and more archaeological. Sampling can be designed together with micromorphology, sedimentology, dating, spatial analysis, artefact study and zooarchaeology rather than added after excavation as an independent laboratory technique. The 2025 work integrating renewed chronology, archaeology, faunal evidence and sediment DNA across Denisova Cave points in precisely that direction.

Dyani Lewis’s 2026 account in Nature captures a field that has moved quickly since the first Pleistocene hominin detections of 2017. The central problem is no longer whether ancient human DNA can survive in cave sediments. It plainly can. The harder question is how closely an ancient molecule can be tied to a particular depositional event, occupation episode or human action.

If that relationship can be established at increasingly fine resolution, archaeology will possess something it has rarely had: biological evidence for populations that passed through a place without leaving an excavated body behind. But those people cease to be archaeological “ghosts” only to the extent that their molecular traces can be securely placed within the material history of the site. On that point, genetics and stratigraphy have to tell the same story.

Selected Sources and Further Reading

1. Aldeias, Vera & Mareike C. Stahlschmidt. 2024. “Sediment DNA can revolutionize archaeology—if it is used the right way.” Proceedings of the National Academy of Sciences 121(26): e2317042121. DOI: 10.1073/pnas.2317042121.
2. Bossoms Mesa, Alba, Elena Essel, Louisa Jáuregui et al. 2026. “Investigating ancient human DNA preservation on cave walls and in rock art.” Nature Communications 17: 5561. DOI: 10.1038/s41467-026-74234-2.
3. Gelabert, Pere, Susanna Sawyer, Anders Bergström et al. 2021. “Genome-scale sequencing and analysis of human, wolf, and bison DNA from 25,000-year-old sediment.” Current Biology 31(16): 3564–3574.e9. DOI: 10.1016/j.cub.2021.06.023.
4. Gelabert, Pere et al. 2025. “A sedimentary ancient DNA perspective on human and carnivore persistence through the Late Pleistocene in El Mirón Cave, Spain.” Nature Communications 16: 107. DOI: 10.1038/s41467-024-55740-7.
5. Jacobs, Zenobia, Elena I. Zavala, Bo Li et al. 2025. “Pleistocene chronology and history of hominins and fauna at Denisova Cave.” Nature Communications 16: 4738. DOI: 10.1038/s41467-025-60140-6.
6. Lewis, Dyani. 2026. “How DNA in dirt is shaking up the study of human origins.” Nature 651: 868–870, 24 March 2026. DOI: 10.1038/d41586-026-00879-0.
7. Massilani, Diyendo et al. 2022. “Microstratigraphic preservation of ancient faunal and hominin DNA in Pleistocene cave sediments.” Proceedings of the National Academy of Sciences 119(1): e2113666118. DOI: 10.1073/pnas.2113666118.
8. Slon, Viviane et al. 2017. “Neandertal and Denisovan DNA from Pleistocene sediments.” Science 356: 605–608. DOI: 10.1126/science.aam9695.
9. Vernot, Benjamin et al. 2021. “Unearthing Neanderthal population history using nuclear and mitochondrial DNA from cave sediments.” Science 372: eabf1667. DOI: 10.1126/science.abf1667.
10. Zavala, Elena I. et al. 2021. “Pleistocene sediment DNA reveals hominin and faunal turnovers at Denisova Cave.” Nature 595: 399–403. DOI: 10.1038/s41586-021-03675-0.
11. Zhang, Dongju et al. 2020. “Denisovan DNA in Late Pleistocene sediments from Baishiya Karst Cave on the Tibetan Plateau.” Science 370: 584–587. DOI: 10.1126/science.abb6320.

Suggested keywords

sedimentary ancient DNA; sedaDNA; ancient DNA; Denisova Cave; Denisovans; Neanderthals; Pleistocene archaeology; archaeogenetics; geoarchaeology; taphonomy; cave sediments; human evolution

Editorial summary

Sedimentary ancient DNA has matured from a technique for detecting otherwise invisible hominin presence into a tool capable of addressing population history, making 2026 an appropriate moment to ask not simply what DNA in archaeological sediment can reveal, but how securely those genetic signals can be connected to layers, artefacts and human behaviour.