Ten years ago the story of how modern humans expanded across the face of the world would have been a relatively simple one. The story generally recounted for popular consumption was most forcefully articulated in Richard Klein’s The Dawn of Human Culture. Around ~50 thousand years ago a small group of Africans resident in the east of the continent changed in some sense. The mostly likely cause for this change was presumably a mutation which conferred upon this group the ability to engage in fully fluent language, and therefore engendered cultural flexibility previously unseen in our lineage.* The rest, as they say, was prehistory. These Africans expanded across the whole Afro-Eurasian zone, replacing archaic hominins, such as the Neandertals, in totality. They pushed the frontier of human habitation into Oceania and the New World.
That model is in some important aspects very wrong. First, the expanding African population mixed with other groups, including likely within Africa. Though less than ten percent of the ancestry outside of Africa, some of it was functionally very important. Second, the demographic expansion ~50-100 thousand years ago more accurately captures the dynamics of non-Africans, than Africans, who were characterized by much larger ancestral populations when the 100 to 1,000 ancestors of modern non-Africans left the continent. Third, modern populations seem to be to a large extent the product of fusions across the Pleistocene branches of humanity, brought together during the Holocene by rapid demographic expansion triggered by cultural innovations such as agriculture. Finally, these dynamics were not limited to populations outside of Africa.
For far too long Africa was conceived of as a blackbox in genetic terms, eternally useful as an outgroup, basal to the rest of humanity. Yes, there were platitudes about how most human genetic diversity was localized in African populations due to the bottleneck during the Out of Africa event, but beyond that maps of human migration implicitly left one the impression that after the ancestors of modern humans left little occurred within Africa. Yes, everyone could agree that there were back migrations along the periphery of the continent, the great swath north of the Sahara around to the Horn of Africa, but Sub-Saharan Africa was neglected in these treatments. No more.
A new paper in Nature is a major step forward in bringing together a lot of the elements of new findings we’ve seen in other work, The African Genome Variation Project shapes medical genetics in Africa. The medical genetics part is important downstream. It doesn’t loom large in the paper itself. I recommend you check out the supplementary info and supplementary data, it has a lot of the meat of the paper. For me the big topline result is in the figure above, where you see the collection of results suggestiong admixture into African populations from Eurasia and also between agriculturalists and hunter-gatherers.
On the Eurasian admixture, the authors confirm what we always knew about Ethiopia and the Horn of Africa, that it was the scene of a relatively recent admixture event between an Afro-Asiatic people, and a group related to modern Nilotic peoples. What is more interesting is that they observed Eurasian admixture within Yoruba people. This admixture has been suggested by others, as the Yoruba have traces of Neandertal ancestry. This group dates the admixture back to nearly 10,000 years ago, so it as likely associated with goings on that were trans-Saharan. If that is the case these were almost certain quasi-Eurasian hunter-gatherers, and their ancestry might have been diminished in current North African groups subject to waves of farmers issuing from the east during the Neolithic. But there is also admixture with Eurasians further east in Uganda among Bantu groups. Reading the details of the supplements there is a chance that this was mediated through admixture of Eurasians with hunter-gatherer populations, and then the absorbtion of this hybrid group into the expanding wave of Bantu farmers. Speaking of which, this issue is solved, it is clear that the Bantu expansion was a major demographic transformation of eastern and southern Africa. The genes speak loudly and clearly. Additionally, the Sub-Saharan African admixture of Ethiopians is more closely related to that of the Nilotic people than the Bantus. The paper didn’t tease out the details archaeological and historically, but if you look at the dates all this was going on in eastern Africa during the rise and fall of ancient Egypt. In other words, within historical memory the whole demographic landscape of Africa was reshaped. Contrary to the idea that Africa was static, there are indications here of massive transformations.
Though the Eurasian admixture story among these populations is fascinating, there is also nuance in the input of hunter-gatherer ancestry within West African and Bantu populations. First, I suspect that these estimates are low bounds, because they don’t have exact reference populations. Some of the hunter-gatherers mixed into the Igbo and Bantu groups may have been more like agriculturalists than the extant hunter-gatherer groups within Africa. One of the peculiarities of the genetics is that it looks as if the hunter-gatherers of Sub-Saharan Africa, the Khoisan in the south and the Pygmies in the center, share more recent common ancestors than they do with the agriculturalists. This may simply be due to the fact that the agriculturalists went through rapid expansion, and this whole constellation of peoples derive from a group which was an outgroup to extant hunter-gatherers. The only complicating issue is that of Eurasian admixture; it seems likely that for very old admixture events we’re seeing underestimates, or they aren’t picked up. In other words, the “reference” Sub-Saharan Africans themselves are compounds of people who remained within Africa, and Out of Africa. The eastern Pygmies may be the only people in the world without much Out of Africa input (recall that the Khoisan have some level of Out of Africa input mediated by East African pastoralists).
A second interesting aspect of the paper is about selection within African populations. As I said above you can find much in the supplements, so I won’t review that laundry list. But, it is interesting that many of the signatures disappeared once Eurasian ancestry was “masked.” That is, within the genomes of individuals you have a mosaic of ancestries, and high genetic distances between populations at particular loci turn out often to be simply due to historical demography. Once you remove this confound you pick out signals of selection which might be due to local adaptation (though some of the Eurasian alleles might also have been subject to selection, so in some ways the filtering might be too stringent). But the masking of Eurasian ancestry also highlighted something important: the genetic variation across African populations once you remove Eurasian ancestry is not that high. This is curious in light of the truism that most genetic variation in humans is found within Africa, but as Nick Patterson pointed out to me years ago: this applies to variation within populations, not across them. Since most variation is not partitioned across populations that explains why Africans can be so genetically varied despite exhibiting not too high between population variation. After masking Eurasian ancestry the mean pairwise Fst was ~0.015. To give a sense of perspective, the Fst between Northern Italians and Lithuanians is 0.01. The Fst between the Ethiopian African ancestry (so Eurasian segments are masked) and other African populations is still 0.027, on average (the distance between Lithuanians and Southern Italians is 0.015). This reinforces the fact that the African ancestors of Ethiopians are somewhat atypical (further confirmed by the relative inaccuracy of imputation from public data sets).
The result that the Igbo seem to have ancestry from a hunter-gatherer group genetically closer to the Khoisan than the Mbuti Pygmies makes a lot more sense when you accept that much of the genetic population structure within African disappeared with the rise of agriculturalist groups which demographically swamped them. It seems plausible that the preexistent variation can be reconstructed to some extent by analyzing patterns within agriculturalists, as they likely absorbed hunter-gatherer groups over time. Within the paper the authors suggest that whole genome sequencing of more populations should be high on the priority list, and I agree. The future is going to be interesting.
* Klein appealed to some of Stephen Jay Gould’s more macromutationist/saltationist speculations.