The sequencing of ancient DNA is generating dramatic results. The sequence from a bone fragment has revealed the existence of an unknown type of extinct human ancestor that lived in Asia 40,000 years ago.
Forty thousand years ago, the planet was more crowded than we thought. For decades we believed that there were just two members of the genus Homo living at that time. First, Neanderthals, which occupied large tracts of Europe and north Asia, although their days were numbered and within 15,000 years they became extinct. Second, early modern humans, members of our own species, who spread across Eurasia from the African homelands that they had left some 10,000 to 20,000 years earlier. This picture changed in 2003 when a third human species, the tiny 'Hobbit', Homo floresiensis, was discovered in Indonesia, the most recent of these fossils being only 13,000 years old.
Now, in a paper on page 894, Krause et al.1 force a rethink of the recent human occupation of Eurasia by describing an unknown hominin — an extinct species of human ancestor — present in Asia around 40,000 years ago. Their discovery is remarkable not just for the insight it gives into the human past. For the first time a hominin has been described, not from the morphology of its fossilized bones, but from the sequence of its DNA.
The DNA comes from a piece of finger bone discovered in Denisova Cave in the Altai Mountains of southern Siberia (Fig. 1). This cave was intermittently occupied by humans for 125,000 years. It is rich in stone tools and bone implements, but has yielded few human bones, most of them isolated finds such as that used in this study. With such incomplete specimens, the morphological information needed to identify the species to which a human bone belongs is impossible.
The finger bone came from a layer dated to between 48,000 and 30,000 years ago. DNA sequencing from fossils of this age is still quite an achievement, and would not have been possible without technical innovations developed by the same group in their work with Neanderthal DNA. These include methods for assessing whether a DNA fragment is genuinely ancient on the basis of its length and chemical damage, thereby solving the great conundrum for studies of ancient DNA by ensuring that the sequence obtained is not a contaminant from a researcher.
Krause et al.1 focused on the DNA found in cellular organelles called mitochondria, as there are about 8,000 copies of this mtDNA per cell, compared with just two of the DNA in the nucleus, giving a greater chance of finding mtDNA in an ancient specimen. Using the next-generation sequencing methods that have been applied to Neanderthals2, mammoths3 and, most recently, the 4,000-year-old remains of an ancient Eskimo4, a complete mtDNA sequence was assembled from the Denisova finger, with each nucleotide being read, on average, 156 times, so ensuring a high degree of accuracy. The uniqueness of the sequence was revealed when it was compared with that of modern humans and Neanderthals. It matched neither, even though both species were living in the Altai Mountains 40,000 years ago.
If not a modern human or a Neanderthal, then to what species does the owner of the Denisova finger belong? Candidates are thin on the ground. Homo erectus was the first hominin to move from Africa to Eurasia, some 1.9 million years ago. There is fossil evidence that H. erectus survived in Indonesia as recently as 100,000 years ago, but nothing to suggest that the species was present in mainland Asia at anything approaching the time of the Denisova sample. The DNA sequence also argues against this individual being H. erectus.
By comparing the sequence of the Denisova mtDNA with those of Neanderthals and modern humans, Krause et al.1 were able to draw a family tree showing the three species' evolutionary relationships (Fig. 2). The tree revealed that the common ancestor of the species dates to about 1 million years ago. If modern humans evolved in Africa, then this ancestor must also have been in Africa, making it impossible for the Denisova hominin to be descended from the H. erectus populations that moved to Europe 900,000 years earlier. The Denisova sequence is also distinct from that of the immediate ancestors of Neanderthals, which split away from the lineage leading to modern humans some 450,000 years ago, much later than the branch leading to Denisova (Fig. 2). Left with few alternatives, Krause et al. make the logical deduction that the Denisova DNA sequence represents an unknown type of hominin that left Africa in a previously unsuspected migration about 1 million years ago, and that survived in at least some parts of Eurasia until 40,000 years ago or later.
What next? The relationship between the Denisova sample and Neanderthals and modern humans will become clearer when nuclear DNA is obtained. And if less-fragmentary remains with the same mtDNA can be identified, their morphological examination might place the Denisova hominin in the broader scheme of human evolution.
A final question is whether there are other big surprises around the corner. Krause et al. point out that it is unlikely that ancient DNA will ever be recovered from extinct hominins from warmer parts of the world, because under these conditions DNA does not survive for more than a few thousand years. But there are many other interesting archaeological sites in the cooler latitudes. The demonstration that a bone fragment can provide evidence for an unknown hominin will surely prompt more studies of this kind and, possibly, increase the crowd of ancestors that early modern humans met when they travelled into Eurasia.