There is a new paper out in Science that analyzes the genome of a man (K14) that lived and died about 37,000 years ago, in Russia. They found that this individual came from a population that had shared ancestry with A. Basal Eurasians, that mysterious population, sister to the main Out-of-Africa expansion and B. a population ancestral to both western hunter gatherers (WHG) and Ancient North Eurasians ( ANE, Sibermen). In other words, a mix something like modern Europeans, but long, long ago.
Willerslev concludes that there was widespread intermingling back in the stone age: Western Asia was a metapopulation with repeated, possibly continuous gene flow, instead of a few discrete migration events.
The problem is that none of the other evidence agrees with this theory. We’ve looked at the DNA of Mesolithic hunter-gatherers in Western hunter-gatherers – funny, they don’t have any Basal Eurasian mixture, or any ANE, either. The Neolithic farmers of Europe have Basal Eurasian, but they don’t have any ANE, and although they have a component that is similar to the WHG, their uniparental lineages have almost no overlap with those of the western hunter-gatherers. The 24,000-year-old Mal’ta skeleton from Siberia is a good fit to ANE, but has no Basal Eurasian admixture. Moreover, the Mesolithic and Neolithic samples are ~20,000 years later than K14. If they were contemporaneous with K14, you could maybe argue that gene flow was common but just hadn’t reached them yet. But they weren’t, and you can’t.
The thing is , you can mix two populations with distinct drift histories, and later detect the admixture event and say something about the original populations – but you can’t unmix them – at least not by natural selection or drift, the only available processes in nature.
So Willerslev’s conclusion is wrong. Moreover, we have plenty of evidence of serious migrations in Europe ( and elsewhere): over time, the EEF expand and the Mesolithic hunters shrink. Later, all the villages in the Balkans are burned down, and all the houses in Germany and Poland disappear: sure looks as if someone came knocking.
What then is the real story? Well, I see two possible explanations. One is that there was was a small mixing event that produced a population with all three components – a population that wasn’t a major source of ancestry for later populations. In the same way, if the daughter of a European Mesolithic hunter and a Basal Eurasian had married the son of an Ancient North Eurasian dude and a proto-Chinese chick in 10,000 BC, , they would have produced the world’s first Mexican, even if this complicated love-story happened thousands of years before Columbus. Still, one synthetic Mexican wouldn’t have mattered much – and it wouldn’t have meant that real, live Mexicans were descended from that early synthetic Mexican. For that matter, although most of the people in Chile and Mexico are a mix of European and Amerindian, those two populations originated in separate admixture events.
The other possible explanation is error: they’ve made a mistake. The genome quality is not all that good (2.42 X), and the fraction of genome recovered is not large – this might have made things more difficult.