The first Denisovan forearm bone came out of a 22-sample shortlist
Fu Qiaomei's team could not get DNA out of the Bianfu Cave bones, so protein variants had to carry the identification, and a morphological pre-screen had to make the mass spectrometry affordable at a site holding more than 64,000 fragments.
The Scientist · Science desk
What happened
- A 2019 rescue excavation at Bianfu Cave in Yunnan cut three trenches to 6.6 metres through 18 sediment layers, ten of which held hominin artefacts, and one of which yielded four intact teeth.
- Hand sorting of more than 64,000 bone fragments picked 22 candidates for mass spectrometry, and three passed: two parietal fragments and a proximal piece of radius, the first Denisovan forearm bone on record.
- Those three bones and two teeth from Layer 7 all carried the Denisovan-specific COL1A2 R996K variant, and none carried protein variants specific to modern humans or Neanderthals.
- DNA extraction from the three bones failed, which Fu Qiaomei attributes to the region's warm climate and the age of the fossils.
- The hominin fossils date to about 167,000 to 134,000 years ago, while the cave's cultural sequence of tools and butchered bone runs from about 190,000 to 70,000 years ago.
Why it matters
- What's now possible capability With a radius in hand instead of a finger bone and a rib, Denisovan arm biomechanics becomes a question that can be asked of an actual long bone, even if the answer currently rests on a single proximal fragment.
- What it limits constraint When DNA will not survive, identity, sex and phylogeny have to be read off a few thousand amino acid residues and a handful of diagnostic positions, which cannot deliver admixture history or relatedness between individuals.
- What doesn't add up contradiction Coverage settles on five confirmed Denisovan specimens, but New Scientist reports the excavators concluded the remains represent at least four individuals because the bones and teeth differ in age, so "five Denisovans" overstates the headcount.
- Who has to decide decision Institutions sitting on boxes of unidentifiable fragments now have a published cost profile for screening them, which changes what a sampling proposal has to justify and to whom.
Next story · Science · 5h ago
Baker lab's from-scratch zinc proteases cut their chosen target in 36% of 135 designsConfirmed2 publishers
The arithmetic behind the triage is the part other sites can copy. Sending 22 fragments to the mass spectrometer rather than the full assemblage is roughly 2,900 times fewer runs [35], and three of the 22 came back hominin, a hit rate near 14 percent [33]. Blind ZooMS screening at Denisova Cave and Baishiya Karst Cave has recovered hominin bone at about 0.1 percent of specimens [26], so on those two numbers the pre-screen improved yield per sample by something like a hundredfold [34]. The comparison flatters the new method, though: 0.1 percent is a site-level recovery rate across everything screened, while 14 percent is a hit rate inside a subset chosen precisely because it already looked like hominin bone [30].
Fragments that did not look diagnostic were never tested. Three hominin bones is therefore a floor on what the assemblage holds rather than a measurement of it, and nothing in the design constrains how many hominin pieces stayed in the discard trays.
Before this excavation, everything known of the Denisovan body below the skull was a finger bone and a rib fragment [6]. The Bianfu radius is a proximal piece, and its head, where the forearm meets the humerus at the elbow, is closer to Neanderthal form, while the bone's overall shape sits nearer modern humans [8]. Fu told Live Science that the human-like form "likely provided functional advantages for lightweight, precise manipulation" [10]. One proximal fragment from one individual: the shape is data, the function is inference, and there is no population range around it yet.
Each specimen gave between six and 16 endogenous proteins, covering 1,972 to 4,076 amino acid residues [1], and Bayesian trees built from those consensus sequences reproduced the topology of nuclear-genome studies, with every Bianfu sample falling into a clade with Denisova 3 [4].
The dated fossils span about 33,000 years inside a cultural sequence of about 120,000, near 28 percent of the occupation [28], and the Nature paper describes that record as the one most parsimoniously attributed to Denisovans [21]. The layers with no hominin bone in them are assigned by continuity of technology, not by protein.
Much of the behavioural evidence is independent of the molecular work. There are 1,654 identifiable animal bones, dominated by deer and bovids, with rhinoceros, hyena and bamboo rat also present; many carry butchery marks and few show non-human gnawing [23]. The stone kit runs to 1,366 artefacts of flakes, cores, scrapers and choppers, simple next to contemporaneous Neanderthal assemblages [22], used alongside pervasive unmodified bone in a conifer-dominated forest or forest-steppe [24]. Bence Viola of the University of Toronto, who was not involved in the work, calls Bianfu only the third Denisovan site "where we have everything," after Denisova Cave and Baishiya Karst Cave [25]. That sets the size of the comparative sample any of this feeds: two sites with full context before, three now [29].
What to watch
- Whether anyone runs the pre-screen on an assemblage with independently known hominin fragments, which would finally measure its false-negative rate.
- Whether a dated hominin fossil ever turns up in the layers older than 167,000 or younger than 134,000 years, converting the long occupation from parsimony to observation.
- Whether other warm-climate Asian sites yield the COL1A2 R996K call, or produce specimens with no diagnostic variant at all.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence80
- Adoption
- Not scored
- Hype gap+20
- Incentives35
- Confidence72
Perspective Coverage
7 publishers- Builder
- Builder 48%
- Operator
- Operator 45%
- Investor
- Investor 7%
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Six to 16 endogenous proteins were identified from each of the five specimens, covering 1,972 to 4,076 amino acid residues.
- [2]
All five samples carried the Denisovan-specific variant COL1A2 R996K, indicating their Denisovan identity.
- [3]
None of the five specimens had protein variants specific to H. sapiens or Neanderthals.
- [4]
Bayesian phylogenetic trees built from consensus sequences of endogenous proteins had topologies highly consistent with nuclear genome studies, with each Bianfu Cave sample forming a monophyletic clade with Denisova 3.
- [5]
Bianfu Cave is a Paleolithic site in Yunnan province, southwestern China.
- [6]
A tiny finger bone and a rib fragment were all that had been found of Denisovan bodies before the Bianfu Cave excavation; the new radius fragment is the only Denisovan radius found to date.
- [7]
The area around Bianfu Cave has been heavily quarried, destroying much of the cave including the entrance; in 2019 local workers spotted animal fossils and alerted authorities, leading Hao Li and colleagues to conduct a rescue excavation that summer.
- [8]
The newly discovered radius has a large head, where it connects to the humerus at the elbow, more similar to Neanderthals than to modern humans, but its overall shape is more similar to modern-human radii.
- [9]
The team dug three trenches to a depth of 6.6 metres, revealing 18 layers of sediment, of which 10 contained artefacts made by hominins; one layer yielded four intact hominin teeth.
- [10]
Study co-author Qiaomei Fu, director of the Ancient DNA Laboratory at the IVPP, Chinese Academy of Sciences, told Live Science by email that the human-like radius shape "likely provided functional advantages for lightweight, precise manipulation".
- [11]
The excavation unearthed over 64,000 fragments of bone.
- [12]
Five human fossils dating to between 134,000 and 167,000 years ago confirmed that the people who used the cave were Denisovans; paleoproteomic analysis confirmed five hominin remains as Denisovans.
- [13]
ZooMS analysis confirmed that two parietal bone fragments (BFD767 and BFD769) and one proximal radial fragment (BFD771) came from hominins.
- [14]
BFD767 and BFD771 came from Layer 7, dated to approximately 148,000-134,000 years ago; BFD769 came from the older Layer 9, dated to approximately 167,000-150,000 years ago.
- [15]
The team also analysed two hominin teeth from Layer 7: a lower fourth premolar (YHB3518) and a lower second molar (YHB3075).
- [16]
Abundant amelogenin Y isoform peptides in the enamel of the two teeth indicate both teeth belonged to male individuals.
- [17]
The team tried to extract DNA from the three bones without success; Fu says the region's warm climate and the age of the fossils were unsuitable for DNA preservation.
- [18]
Two studies on Bianfu Cave were published on Wednesday, Sept. 9, in the journal Nature.
- [19]
The proteomic study was conducted by a research team led by Professor Fu Qiaomei of the Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, with multiple collaborating institutions.
- [20]
The hominin fossils are dated to about 167,000-134,000 years ago, whereas the cultural sequence at the cave spans about 190,000 to 70,000 years ago.
- [21]
The Nature paper describes the Bianfu record as one of the longest cultural records most parsimoniously attributed to Denisovans, identified through enamel-dentine junction morphology and palaeoproteomic analyses.
- [22]
The excavators found 1,366 stone artefacts, including 135 flakes and the cores they came from plus simple tools such as scrapers and choppers, relatively simple compared with those made by Neanderthals elsewhere in Eurasia at the same time.
- [23]
The team found 1,654 identifiable animal bones, including large numbers of deer and bovids plus rhinoceroses, hyenas and bamboo rats; many bones had marks suggesting butchery by hominins and few showed gnawing by non-human animals.
- [24]
Faunal and pollen records indicate a conifer-dominated forest or forest-steppe environment across Marine Isotope Stages 6 to 4, and the inhabitants practised specialized hunting of medium- to large-bodied prey with expedient core reduction alongside pervasive use of unmodified bones.
- [25]
Bence Viola of the University of Toronto, who was not involved in the study, says the new site is only the third Denisovan locale "where we have everything" - human remains, lithics, fauna and environmental context - the other two being Denisova Cave in Siberia and Baishiya Karst Cave on the Tibetan plateau.
- [26]
The recovery rate of hominin fossils from ZooMS screening has been extremely low, only 0.1% at sites such as Denisova Cave and Baishiya Karst Cave.
- [27]
Denisovans are known from only a handful of fragmentary fossils scattered from Siberia to Laos, and no complete Denisovan skeleton has ever been found.
- [28]
The dated hominin fossils cover about 33,000 years, roughly 28 percent of the cave's approximately 120,000-year cultural sequence.
- [29]
Adding Bianfu Cave takes the number of Denisovan sites with remains, stone tools, fauna and environmental context together from two to three.
- [30]
Researchers first went through the bone fragments by hand and identified 22 potential hominin remains, which they then subjected to molecular analysis.
- [31]
The researchers developed a "morphological pre-screening + ZooMS identification" strategy because running ZooMS on all of Bianfu Cave's fragments would be costly and inefficient, and they present it as a replicable framework for similar Paleolithic sites with abundant bone fragments.
- [32]
Because the bones and teeth appear to be different ages, Hao Li's team concluded that they come from at least four individuals.
- [33]
Three hominin identifications from 22 fragments submitted to ZooMS is a hit rate of about 13.6 percent.
- [34]
A 13.6 percent hit rate within the pre-screened subset is roughly 136 times the 0.1 percent hominin recovery rate reported for blind ZooMS screening at Denisova Cave and Baishiya Karst Cave.
- [35]
Testing 22 fragments instead of more than 64,000 is about 2,900 times fewer mass-spectrometry runs.
Sources
7 independent publishers whose own reporting we read for this story.
- discovermagazine.comThe First Known Denisovan Forearm Bone Emerges From a Search Through 60,000 Bone Fragments
1 article · September 9, 2026
- livescience.comEnigmatic Denisovans lived by cave in what is now China for 100,000 years, rare fossils and handcrafted tools reveal
1 article · September 9, 2026
- nature.comDenisovans from southwestern China and their subsistence strategies
3 articles · September 9, 2026
- newscientist.comDenisovan tools from cave in China shows how mysterious humans lived
1 article · September 9, 2026
- phys.orgFirst Denisovan forearm bone discovered in southwest China
2 articles · September 10, 2026
- sciencenews.orgNewfound fossils are spilling Denisovan secrets
1 article · September 9, 2026
- scientificamerican.comScientists unlock new clues to what Denisovans, a mysterious ancient human ancestor, looked like
1 article · September 9, 2026
Topics and entities
Topics
- Paleoanthropology
- Palaeoproteomics
Entities
- Zooarchaeology by Mass Spectrometry
- Institute of Vertebrate Paleontology and Paleoanthropology
- Baishiya Karst Cave
- Nature