Denisovans

What happened

1. 1980A monk finds a hominin mandible at Baishiya Karst Cave, Xiahe, on the Tibetan Plateau, China, later called the Xiahe mandible.
2. 2008A hominin finger-bone fragment, later named Denisova 3, is excavated at Denisova Cave.
3. 2010Krause et al. sequence its complete mitochondrial genome and report a lineage unlike Neanderthals or modern humans.
4. 2010Reich et al. sequence the nuclear genome, name the group 'Denisovans', and detect ~4-6% Denisovan ancestry in present-day Melanesians.
5. 2017Slon et al. recover Neanderthal and Denisovan DNA directly from cave sediment with no skeletal remains, a method later applied at Baishiya.
6. 2018Slon et al. sequence 'Denisova 11', a first-generation daughter of a Neanderthal mother and a Denisovan father.
7. 2019Chen et al. identify the Xiahe mandible as Denisovan by ancient protein analysis, the first Denisovan fossil outside Denisova Cave.
8. 2019Douka, Jacobs et al. use Bayesian-modelled radiocarbon and optical dating to show hominin occupation at Denisova Cave extends back at least 200,000 years.
9. 2020Zhang et al. detect Denisovan mtDNA in Baishiya sediment, corroborating Denisovan presence there; authors overlap with the Denisova Cave team.
10. 2021Ji, Wu et al. describe the Harbin cranium as a new species, 'Homo longi', with no molecular link to Denisovans.
11. 2025Fu et al. identify the Harbin cranium as Denisovan via mtDNA from dental calculus and, in a companion paper, ancient proteins; both led by the same team.
12. 2025Tsutaya et al. identify the Penghu 1 mandible from Taiwan as a male Denisovan by ancient protein analysis, a fourth Denisovan locality.

Objections 1 of 5 proved wrong

• ✓
  The finger bone's original 48,000-30,000 year age came from its stratigraphic layer, not a direct measurement, in a karst cave where deposits can be disturbed by burrowing, subsidence or freeze-thaw.

  Dating or samples · Krause et al. (2010); Dennell (2019) · 2010

  Held up — Douka et al. (2019) modelled the youngest directly-dated Denisovan remains to 52,000-76,000 years, older than the original layer-based estimate; 48,000-30,000 years is no longer used for this bone.
• ✓
  Krause et al.'s reading of the mtDNA split as evidence for a distinct hominin migration out of Africa was an over-interpretation; the divergence need not imply a separate migration.

  Other · Martinón-Torres, Dennell & Bermúdez de Castro (2011) · 2011

  Held up — Reich et al.'s nuclear genome, published later the same year, placed Denisovans as sister to Neanderthals rather than a separate deep migration, dropping the reading this objection challenged.
• ?
  Denisovans were named from molecular data alone with no morphological holotype of their own; only a handful of fragmentary remains from one site were attributed to them for their first decade.

  Natural, not human · General taxonomic caution in paleoanthropology · 2013

  Unresolved — Xiahe, Penghu and Harbin fossils now corroborate the lineage by protein or DNA evidence. Homo longi (2021) and Homo juluensis (2024) have been proposed as binomials, but neither is in consensus use.
• ✕
  Denisovan and Neanderthal genetic affinities with living humans could reflect shared ancestral variation retained from a common ancestor, not introgression from real interbreeding.

  Other · Lowery, Uribe, Jimenez, Weiss et al. (2013) · 2013

  Proved wrong — A directly sequenced first-generation Neanderthal-Denisovan hybrid (Denisova 11, 2018) shows these lineages did interbreed; no later paper defending the shared-ancestral-polymorphism reading was found.
• ✓
  The Xiahe mandible was a surface find handed to researchers by a monk in 1980 with no excavation record; its ~160,000-year age comes from U-series dating of a carbonate crust on the bone, not the bone itself.

  Layers or association · Acknowledged limitation of the find · 2019

  Held up — The protein-based Denisovan identification of the mandible is not disputed, but its own age remains an inference from the adhering crust rather than a direct measurement; no subsequent direct date was found.

Evidence at the time within ~5 years of the claim

• Unambiguous evidenceYes — A complete mitochondrial genome, then a nuclear genome, matched no known human, Neanderthal or ape sequence; contamination ran up to ~0.4% (mtDNA ~0.1-0.3%, nuclear to 0.43%), too low to explain it.
• Documented contextPartly — The finger bone (Denisova 3) came from an excavated layer at Denisova Cave; its age came from that layer, not a direct date on the bone. Karst deposits can be locally disturbed.
• Independent dating methodsNo — At the time of the claim, the age rested on the radiocarbon-dated stratigraphic sequence alone. Direct multi-method dating of hominin fragments followed only in 2019.
• More than one labNo — Both 2010 papers sequenced DNA at the same Leipzig (MPI-EVA) facility. Nuclear-genome statistics involved a Harvard co-author (Reich) already part of that consortium paper, not an independent lab.
• Professional recoveryYes — Excavated by the long-running Russian Academy of Sciences (Siberian Branch) project at Denisova Cave, led by Derevianko, Shunkov and colleagues.
• Peer-reviewed reportYes — Both founding results appeared as full papers in Nature within the same year (March and December 2010).
• Open to criticsYes — Sequence data were deposited in public repositories, and specimens curated in Novosibirsk have since been restudied by other groups, e.g. by collagen fingerprinting of further bone fragments.
• Independent teamsPartly — Recovery and sequencing stayed within the Leipzig consortium; but Skoglund & Jakobsson (Uppsala) independently reanalysed the genome within a year, finding further Denisovan affinity in East Asians.
• Outside stakeNo — Academic consortium (Max Planck Institute for Evolutionary Anthropology; Harvard Medical School); no commercial, religious or nationalist motive documented in the sources reviewed.

What would settle it for a field partner

1. Recover nuclear DNA or a proteome from additional Xiahe-, Penghu- or Harbin-associated skeletal material to build a second high-coverage Denisovan genome outside Denisova Cave.Medium cost
   Would show: Whether the Tibetan, Taiwanese and Northeast Asian Denisovan populations were genetically distinct from the Siberian lineage, and would support formally naming the group.

What settled it DNA, Morphology, Expert review

What would change this record

Independent reanalysis showing the Denisova 3 sequences were contaminated or artefactual, or repeated failure to replicate Denisovan DNA or proteins in the Siberian, Tibetan, Taiwanese or Harbin material, would overturn this status. No such failure is reported.

Notes

• Status. A genetically distinct Denisovan lineage is accepted across paleoanthropology, attested at Denisova Cave, Baishiya, Penghu and Harbin. The 2010 result was accepted quickly, not independently replicated; rank and naming remain debated.
• Accepted view then. Eastern Asian archaic hominins were debated from fragmentary fossils; no archaic Eurasian lineage besides Neanderthals was genetically characterised.
• Now understood. ~52,000-76,000 years (Bayesian model)
• Dating basis. Age of the finger bone (Denisova 3) as originally reported, inferred from its stratigraphic layer at Denisova Cave, not a direct date on the bone itself.
• Correction. The widely cited 48,000-30,000 year age for Denisova 3 was a stratigraphic estimate, not a direct measurement; Douka et al. (2019) modelled the youngest Denisovan remains to 52,000-76,000 years, older than its layer implied.
• This record treats 2010-2025 as one case: later specimens (Xiahe, Penghu, Harbin) corroborate the 2010 claim but share investigators with the Leipzig consortium, so they are corroboration, not independent replication. Puzzle: 2010 mtDNA implied a split twice as deep as the Neanderthal-human split; the nuclear genome, months later, placed Denisovans as Neanderthals sister, so mtDNA reflected lineage history, not divergence time. No binomial is in consensus use: Homo longi (2021) and Homo juluensis (2024) have been proposed for populations that may include Denisovans. No sustained rebuttal of the lineage existence was found: fast acceptance of a surprise discovery, not a contested anomaly.

People and evidence

• Proponents. Svante Pääbo, Johannes Krause, David Reich, Bence Viola, Fahu Chen, Qiaomei Fu
• Critics. María Martinón-Torres, Robin Dennell, Ryan K. Lowery
• Evidence types. dna, proteins, morphology, human remains, stratigraphy, u series, radiocarbon, osl · 3 independent lines
• Record confidence. medium — Founding, hybrid, Xiahe, Baishiya, Penghu and Harbin-proteome papers read as full text or full abstracts. S7/S9/S10 are paywalled beyond abstracts ('exists'); two 2011-13 critiques are metadata-only.