By comparing the DNA to that of modern populations, scientists found evidence that these "Denisovans" from more than 30,000 years ago ranged all across Asia. They apparently interbred with the ancestors of people now living in Melanesia, a group of islands northeast of Australia.
There's no sign that Denisovans mingled with the ancestors of people now living in Eurasia, which made the connection between Siberia and distant Melanesia quite a shock.
As for the Denisovans, the new work is probably just the start of what can be learned from their genome, said one expert familiar with the research. Eventually, it should provide clues to traits like eye and skin color, said Todd Disotell of New York University.
"We're going to be able to piece these people together in the next few years from this genome," he said.
The existence of a new human relative was first revealed just nine months ago from a sampling of DNA recovered from a finger bone discovered in the Denisova Cave in southern Siberia. Researchers proposed the informal name Denisovans for them in Thursday's issue of the journal Nature, where they report the new results.
There's not enough evidence to determine whether Denisovans are a distinct species, the researchers said.
The genome, recovered from the finger bone, showed that Denisovans are more closely related to Neanderthals than to modern humans. That indicates that both they and Neanderthals sprang from a common ancestor on a different branch of the evolutionary family tree than the one leading to modern humans.
Scientists have no idea what Denisovans looked like, said David Reich, a Harvard University researcher and an author of the new paper.
Apart from the genome, the researchers reported finding a Denisovan upper molar in the cave. Its large size and features differ from teeth of Neanderthals or early modern humans, both of which lived in the same area at about the same time as the Denisovans.
Neither the finger bone nor the tooth can be dated directly, but tests of animal bones found nearby show the Denisovan remains are at least 30,000 years old, and maybe more than 50,000 years old, Reich said.
Scientists found evidence that in the genomes of people now living in Melanesia, about 5 percent of their DNA can be traced to Denisovans, a sign of ancient interbreeding that took researchers by surprise.
"We thought it was a mistake when we first saw it," Reich said. "But it's real."
And that suggests Denisovans once ranged widely across Asia, he said. Somehow, they or their ancestors had to encounter anatomically modern humans who started leaving Africa some 55,000 years ago and reached New Guinea by some 45,000 years ago.
It seems implausible that this journey took a detour through southern Siberia without leaving a genetic legacy in other Eurasian populations, Reich said. It makes more sense that this encounter happened much farther south, indicating Denisovans ranged throughout Asia, over thousands of miles and different climate zones, he said.
Yet, archaeologists have reported virtually no sign of the Denisovans, no tools or other indications of how they lived. Maybe that's because sites in Asia haven't been studied as systematically as Neanderthal sites in Europe, he said.
Disotell said he and colleagues were "blown away" by the unexpected Melanesia finding, with its implication for where Denisovans lived.
"Clearly they had to have been very widespread in Asia," and DNA sampling of isolated Asian populations might turn up more of their genetic legacy, he said.
Still, they may not be a new species, because they might represent a creature already known from fossils but which didn't leave any DNA to compare, such as a late-surviving Homo heidelbergensis, he said.
Potts also said the Melanesia finding could mean that the Melanesians and the Denisovans didn't intermix, but simply happened to retain ancestral DNA sequences that had been lost in other populations sampled in the study. But he stressed he doesn't know if that's a better explanation than the one offered by the authors.
"I am excited about this paper (because) it just throws so much out there for contemplation that is testable," Potts said. "And that's good science."