More than 60,000 bone fragments from a cave in southwestern China have yielded three previously unidentified Denisovan bones, including the first Denisovan radius known so far, along with two teeth that also belong to the enigmatic ancient human group. The remains place Denisovans in Yunnan between about 167,000 and 134,000 years ago and provide new evidence about the shape of their skeletons.
The discoveries came from Bianfu Cave in Yunnan, a region at the southeastern edge of the Tibetan Plateau and a geographic crossroads between East Asia, South Asia and Southeast Asia. Excavations at the cave have produced thousands of stone tools and more than 64,000 mammalian fossils, with cultural deposits dating from roughly 190,000 to 70,000 years ago.
The researchers were looking specifically for hominin remains among the site’s many fragmentary bones. Because most of the fragments were too incomplete to identify by appearance alone, they first examined more than 60,000 pieces for features such as size, cortical thickness and surface morphology. That prescreening produced 22 possible hominin fragments.
The team then used ZooMS, short for zooarchaeology by mass spectrometry. The method identifies species from characteristic collagen peptides and can screen fragmentary bones that have few or no recognizable anatomical features.
Three of the 22 candidate fragments produced collagen signatures associated with Homo. They were two pieces of parietal bone, one from layer 7 and another from layer 9, and a fragment from the upper part of a radius, a forearm bone, from layer 7.
Attempts to recover ancient DNA from all three bones failed. Instead, the researchers turned to proteins preserved inside the fossils.
The clue hidden in ancient collagen
Protein sequences provided the crucial identification. The researchers recovered ancient collagen and other proteins from the three bones and from two teeth excavated from layer 7, a lower fourth premolar and a lower second molar.
Ancient proteins can preserve amino-acid differences that distinguish closely related hominin groups. In this case, all five Bianfu Cave specimens carried the same diagnostic change in the collagen protein COL1A2: an arginine-to-lysine substitution at position 996, known as R996K.
The variant was found in every identified Denisovan individual used for comparison, including Denisova 3 from Siberia, as well as Denisovan remains from Harbin, Penghu and Xiahe. The Bianfu Cave samples also lacked confidently identified amino-acid variants derived from Neanderthals or modern humans.
The evidence was checked in several ways. The researchers required sufficient peptide support for informative variants, manually examined the mass-spectrometry spectra around those variants and removed sequences that could have come from microbes, contaminants or other genes. They also searched the data with multiple protein-analysis programs.
The result was consistent across the five specimens. Each carried the Denisovan collagen marker.
The researchers then constructed protein-based phylogenetic trees using the recovered sequences. In every case, the Bianfu Cave specimen grouped with Denisova 3, with a posterior probability of 100 percent. The tree also reproduced the expected broader relationship in which modern humans form a sister group to the Neanderthal-Denisovan lineage.
The two teeth provided additional evidence. Their enamel contained variants in the proteins ameloblastin and amelogenin Y that were consistent with Denisovan affinity. Peptides specific to the AMELY protein also indicated that both teeth came from males.
Denisovans turn up in a new corner of Asia
The fossils fill a gap in the Denisovan map.
Before the Bianfu Cave discoveries, molecularly identified Denisovan fossils had been reported from sites stretching from Siberia to the Tibetan Plateau, northeastern China and the Taiwan Strait. Southwestern China remained a conspicuous gap.
Bianfu Cave extends that known distribution into Yunnan.
The site also adds Denisovan remains from two archaeological layers. The six Denisovan specimens from layer 7 include the two teeth and the radius and parietal fragment identified in this study, while the second parietal fragment comes from layer 9, where no hominin fossils had been identified during excavation.
The researchers place the Denisovan remains within a period of roughly 167,000 to 134,000 years ago. Direct uranium-series measurements on the bones produced younger minimum ages than the ages assigned to their archaeological layers. Because bone can exchange uranium after burial, the researchers treated those bone ages as minimum estimates and relied on the site’s stratigraphic chronology as more accurate.
During this interval, Bianfu Cave was within Marine Isotope Stage 6, a glacial period. The authors note that the local environment was relatively warm, with conifer-dominated forest or forest-steppe and abundant medium- to large-sized herbivores. They propose that these conditions could have supported hominin occupation, but the available genomic and proteomic evidence is not sufficient to determine the particular Denisovan population or lineage represented at Bianfu Cave.
One forearm bone changes the picture
The most unusual fossil in the collection is BFD771, a roughly 45-millimeter-long piece of the proximal radius.
Micro-CT imaging showed a robust internal structure and a fully fused proximal epiphysis, indicating that the individual was probably an adult or near-adult. The preserved anatomy allowed the researchers to identify it as most likely a right radius with a medially oriented radial tuberosity.
The radial tuberosity is the roughened projection of the radius where the biceps brachii attaches. Its orientation and the shape of the nearby bone are among the features that can be used to study how forces were transmitted through the forearm.
The Bianfu Cave radius had a large radial head. Its estimated head diameter was 23.2 millimeters, while the edge of the fovea, the depression that articulates with the humerus, measured 17.2 millimeters. The researchers found these dimensions more similar to those of early and late Neanderthals than to modern humans.
Its radial neck was relatively short, however. The measured neck length was 22.5 millimeters, within the range of late Neanderthals and the upper portion of the modern-human range, but shorter than that of early Neanderthals.
Other aspects of the radius produced a different comparison. Its mid-neck had greater total area and torsional rigidity than the Neanderthal, Homo cf. erectus and several australopith and Paranthropus specimens examined by the researchers, while falling within the range of modern humans and living great apes.
A separate analysis of the external shape also placed the Bianfu Cave radius closer to modern humans and Homo cf. erectus than to Neanderthals. The difference was driven in part by the relatively short radial neck compared with the size of the radial head.
The researchers therefore describe the radius as having a mosaic of characteristics. Some proximal dimensions and the probable orientation of its radial tuberosity resemble Neanderthals, while its overall external shape is closer to modern humans in the comparison.
Two pieces of skull, and another layer of evidence
The two parietal fragments provide a second view of Denisovan anatomy.
One fragment, BFD767, came from the anterior portion of the right parietal. The other, BFD769, came from a region near the lambdoid suture. At their centers, the fragments measured 7.6 and 10.0 millimeters thick, respectively.
Those measurements fall within the broad ranges seen in several Middle Pleistocene Homo groups. The first fragment had a thickness pattern resembling some specimens attributed to Homo heidelbergensis, Neanderthals and East Asian late Middle Pleistocene archaic Homo. The second was substantially thicker and more similar to Asian late Homo erectus, Homo heidelbergensis and some East Asian archaic Homo specimens.
Taken together, the two fragments had a cranial-vault thickness pattern that the researchers considered most similar to Homo heidelbergensis and East Asian late Middle Pleistocene archaic Homo.
The comparison does not by itself establish a distinctive Denisovan skull shape. Instead, it adds anatomical measurements to a fossil record that has previously been dominated by cranial and dental remains and only a small number of postcranial fossils.
The researchers emphasize that the radius is particularly useful because Denisovan postcranial remains are extremely scarce. Before Bianfu Cave, the known Denisovan postcranial fossil record included the Denisova 3 finger phalanx and a rib from Xiahe. The Bianfu radius therefore provides a new anatomical element with which to examine Denisovan skeletal variation.
The protein data establish the taxonomic identity of the fossils, while the bone’s three-dimensional morphology provides the physical evidence for its anatomical comparisons. Together, the two approaches identify Bianfu Cave as a Denisovan locality and add a long bone to the small collection of Denisovan skeletal remains.
The study was published in Nature.
