In a previous post, I speculated that that Denisovans might stem from from Homo erectus, at least in part, rather than being a sister group to Neanderthals as suggested in the paper by Reich and Patterson. It now looks as if this may the case (the in-part bit) : David Reich has said that the Denisovan sample had admixture from Neanderthals and also from an unknown, ultra-archaic population, which might well be erectus.
Although my conclusion was approximately correct, some of my data was shaky. I was considering that Ann Gibbons said – “She was not a modern human, but a descendant of Homo erectus, an ancestral species that left Africa almost 2 million years ago. “ Which is mostly wrong, but it might have been the product of misunderstanding something that Reich actually said – maybe his group had already seen signs of that ultra-archaic fraction. Which is not to slam Gibbons, who is a pretty good reporter.
Other relevant info was more solid. There are a couple of examples, in Melanesians, of archaic alleles of genes in the innate immune system that are very divergent, old enough to have originated in erectus. Next, we already knew that Denisovan mtDNA had diverged a long, long time ago, way before the likely date of the split between neanderthals and homo sap. Lastly, we already knew that modern humans did the deed with Neanderthals, Denisovans, and some archaic group in Africa – surely the ancestors of of the Denisovans got around too. In other words, admixture with the previous owners was the default hypothesis, to be assumed until proved otherwise.