A recent paper showed up in Nature Genetics, Inferring human population size and separation history from multiple genome sequences, which I had also seen on bioRxiv. I’ll gloss over the technical details, except to say this sort of method (of which there are now several) is an incredible extension over PSMC. Basically, rather than having lots of genetic data from many individuals you are looking at whole genomes from a few individuals. From the patterns of genetic variation within these individuals, at the finest grain possible, you can infer the demographic history of whole populations. A standard issue with this sort of thing is the ability for the method to be executed with contemporary computational resources, and it looks like their approximations can do it. The simulations are broadly persuasive to me, and the authors peg some population separations and bottlenecks perfectly. In particular the case of Native Americans seems spot on.
That being said, what about the reality that most modern populations are admixed? They had a sample from a Gujarati, an individual who derives from a population which is a compound of West and East Eurasian. The initial results make sense in light of this fact, but their explanation is not clear to me: “These results suggest that the GIH ancestors remained in close contact with the CEU ancestors until about 10,000 years ago but received some historic admixture component from East Asian populations, part of which is old enough to have occurred before the split with the MXL ancestors.” The non-West Eurasian ancestors of Indians probably diverged from East Asians on the order of 30 thousand years ago. Are the authors implying multiple admixture events in deep time (i.e., pre-Holocene?).
The most interesting, though not entirely surprising, result is the complexity of the “Out of Africa” event which resulted in the worldwide domination of modern humans. There are fewer independent checks on these inferences than the ones more recent, so all must be taken with a grain of salt. But using their method the authors find that the separation between non-Africans and Africans seems rather gradual up until ~50,000 years ago. This suggests that there was a lot of population structure and gene flow within Africa before the expansion of Eurasians. Not surprising, but it is another nail in the coffin of the idea that modern humans emerged in a punctuated fashion and exploded from a singular tribe in eastern or southern Africa. Additionally, it seems that the authors detect the likelihood that the dominant ancestral element of the Masai diverged later from non-Africans’ ancestors than that of the Yoruba. Totally expected, but clarifying because of the scarcity of archaeology in some regions. Finally, the authors report that “As expected, the oldest split among out-of-Africa populations was between European and East Asian (CHB and MXL) populations, most of which occurred between 20,000 and 40,000 years ago (Fig. 4b). Intriguingly, there might be a small component (10% or less) of this separation extending much further back toward 100,000 years ago, which is not compatible with a single out-of-Africa event around 50,000 years ago.” I have a hard time not wondering if this element is related to Basal Eurasian.