Researchers analyzing Denisovan fossils from Paleolithic sites in China have revealed a distinct facial reconstruction for the extinct human group. The findings, published in September and October 2026, combine ancient protein data with newly identified forearm and skull remains to link multiple enigmatic East Asian hominin fossils into a single lineage.
Xujiayao Fossils Reveal Distinct Denisovan Facial Features
Excavated in northern China in the late 1970s, the Xujiayao hominin remains were originally dated to between 138,000 and 199,000 years ago, and were originally discovered in 1974 in Xujiayao, northeastern China. A new study published in the journal Nature applied paleoproteomics to 20 remains from the site, identifying all of them as Denisovans through specific bone and tooth protein signatures.
That identification allowed researchers to examine skull shapes closely and figure out what Denisovans might have looked like. Because the fossil collection included substantial portions of individuals’ skulls, jaws, and dentition, the researchers were capable of forming a distinct understanding of Denisovan appearance. Ancient humans known as Denisovans inhabited Asia before dying out approximately 30,000 years ago. This human relative was first identified in 2010 from the DNA of a finger bone discovered in a cave in Siberia.
This analysis provides a comprehensive view of Denisovan features, showing both traits reminiscent of Homo erectus… and those characteristic of modern humans.
Study authors
Bianfu Cave Yields Denisovan Forearm Bone
Further evidence emerged from Bianfu Cave in China’s Yunnan Province. A research team searched through bone fragments there to uncover a Denisovan radius. That forearm bone offers rare clues to how these ancient human relatives were built. Participating in the study was Qiaomei Fu, a scientist at the Chinese Academy of Sciences’ (CAS) Institute of Vertebrate Paleontology and Paleoanthropology (IVPP).
That specimen represents a Denisovan radius. It provides a rare look at a forearm bone below the skull. Its substantial measurements at that extremity match those of Neanderthal radii. They also match the probable inward direction of the radial tuberosity—the bone prominence where the biceps tendon connects.
Its general outer form aligns more closely with that of present-day humans. It matches the structural profile of a cross-section taken across the center of its neck. That is the tapered segment situated near the elbow joint. Associated with the Marine Isotope Stage 6 glacial period, the fossils indicate that the mild environment and plentiful nourishment of the Yunnan-Guizhou Plateau could have offered an ideal home for Denisovans.
Connecting Fossils Across East Asia
By comparing the Xujiayao and Bianfu Cave specimens with other regional finds, researchers identified shared cranial traits linking them to the Denisovan lineage.
Extended analysis identifies a set of shared morphological traits between Denisovan fossils and other East Asian late Middle Pleistocene archaic Homo, indicating that they all belong to the Denisovan lineage, conclude the researchers.
We know from DNA that Denisovans were a very varied group genetically, which split early into different populations,” Chris Stringer, a paleoanthropologist at the Natural History Museum in London who was not involved in the research, told Live Science in an email. “These could have developed a lot of morphological variation, which may well surprise us as the sample of fossils grows.
Chris Stringer, paleoanthropologist at the Natural History Museum in London
Genetic information indicates an early divergence into distinct groups. Qiaomei Fu—a co-author of the study and head of the Ancient DNA Laboratory within the Chinese Academy of Sciences’ Institute of Vertebrate Paleontology and Paleoanthropology—pointed out that additional lineages may still have populated East Asia concurrently with the documented presence of Denisovans.
Which other distinct lineages may have populated East Asia concurrently with the documented presence of Denisovans remains unknown?
Related reading