“Out of Africa” in 2019

The figure to the left is from Paleolithic DNA from the Caucasus reveals core of West Eurasian ancestry. It is a graph which captures general features of human population historical relationships as we understand them today. Or at least the model fits the data (remember, many models may fit the data!). The graph is complex…but even within the text of the preprint, the author admits that it is characterized by simplifying assumptions, which nevertheless are informative of some general dynamics and processes (e.g., pulse admixtures).

To some extent, the whole last generation or so has been characterized by the victory of a simplifying assumption that captures general truths about the past, with the accumulation of modifications on the margins as more nuanced results enter the picture. The simplifying assumption I am talking about here is the “out of Africa” 50,000 years ago with a total replacement of all other human lineages framework.

By the last quarter of the 20th century, a combination of archaeological and genetic evidence pointed to the likelihood of a massive bottleneck and expansion of humans outside of Africa in the relatively recent past. In the pre-genomic era, the tools were coarse, from uniparental lineages, classical markers, microsatellites, morphometric analyses, as well as archaeological surveys. But, they strongly pointed to massive expansion and population turnover ~50,000 years ago. This, combined with a line of thinking which suggested that Neanderthals were “evolutionary dead-ends” led to the thesis that there was a total replacement.

To a great extent, this model seems to hold up in the broad sketch. But not to an absolute and total degree. Some paleoanthropologists and geneticists were pointing out for decades that the tools we had could not exclude the possibility of admixture at lower fractions with earlier lineages in Eurasia on purely statistical grounds. These scholars were correct, as it turns out. There is now high confidence that in the range of 1-5% of the ancestry of non-Africans derives from highly diverged “archaic” lineages, Neanderthals and Denisovans. The fraction is low enough that more coarse methods did not definitively pick them up, and without ancient genomes, the “game of inference” was not dispositive in either direction. This, despite the fact that these Eurasian hominins’ ancestors seem to have diverged from those of modern humans ~750,000 years ago. Ultimately, scientists needed a physical ancient genome which they could compare to modern populations to come to this conclusion (before the Denisovan result, scientists had been noticing anomalies in Oceanian data for a decade or so but generally ignored it as beneath comment…a presentation was given an anthropology conference on archaic admixture in Oceania right before the Denisova cave paper).

The second major issue is that the massive expansion and bottleneck that occurred ~50,000 years ago may not explain all of the remaining ancestry that is not “archaic.” That is, there were many modern human lineages present 50,000 years ago. The major lacunae in the current model is a huge one: populations within Sub-Saharan Africa maintained larger population sizes throughout this event. And, anatomically modern humans predate this expansion by hundreds of thousands of years. From an archaeological perspective, a lower limit is 200,000 years ago, and an upper limit probably exceeds 300,000 years ago. Additionally, there are “deep lineages” within Africa which clearly predate the expansion 50,000 years ago. There is a strong consensus that the Khoisan people have at least some substantial ancestry that diverged more than 150,000 years ago from other humans, and tentative suggestions from several different research groups suggest that there are even more “basal” (deep divergence) lineages in parts of West Africa that the component within the Khoisan.

This does not even address the likelihood that some “archaic” ancestry persists within Sub-Saharan Africa just as it does outside of Sub-Saharan Africa.

The third issue are diverged anatomically modern humans in eastern Eurasia. By this, I mean lineages which are in a clade with anatomically modern humans rather than Neanderthals and Denisovans, but, split off from other non-Africans before the massive expansion of ~50,000 years before the present. There are two major points here. First, the circumstantial evidence that these people existed is very strong now. We know this from archaeological sites in Southeast Asia and possibly China. And, there is genomic evidence that a lineage closer to modern humans contributed ancestry to Altai Neanderthals ~100,000 years ago. Second, there is some suggestive, though highly disputed, evidence of low levels of earlier-than-50,000-years-ago modern human ancestry in Papuans. The fractions are low, just as they are with Neanderthals. A major problem with detecting diverged lineages at low fractions without ancient DNA is that the statistical power is not there to be definitive. These people are far closer to the dominant “50K expansion” group than Neanderthals and Denisovans.

Remember, without ancient genomes, many geneticists would probably be skeptical to this day about Neanderthal admixture, though there had been suggestions from the genome-wide data of archaic admixture by the middle years of the 2000s. The Neanderthal genome obviously changed our priors, but without ancient genomes, we don’t have as much ability to differentiate possible models which fit the data we have in a manner which resolves disagreements from what I can tell when we are talking about low levels of admixture.

Finally, backing up to the figure at the top of this post: notice that there is now a model with two nestings of “basal” populations in relation to the major non-African ancestry component that expanded ~50,000 years ago. That is, there are famous “Basal Eurasians”, but Lazaridis et al. are proposing a “Basal North African” clade, which diverged from Basal Eurasians & the major Eurasian components before the latter two split apart. Lazaridis et al. also propose ~10 percent of Yoruba ancestry is from the Basal North African clade. There has been a lot of talk for years that Yoruba was an unfortunate choice for “unmixed Sub-Saharan African”, because there are clearly some Neanderthal alleles in this population, indicative of some “back-migration.” But in fact, from modern populations, it would probably be impossible to pick an “unmixed Sub-Saharan African” because there aren’t any.

As I make clear in my previous post, there is a fair amount of evidence that modern Sub-Saharan Africans have been impacted by gene flow from outside of Sub-Saharan Africa. Or at least a population which is somehow closely connected to the ancestors of non-Africans. A substantial proportion of this is probably due to the major modern human expansion that dates to 50,000 years ago. But, some if it probably predates this period, and, some of it postdates this period. If Lazaridis et al. are correct that there is Basal North African admixture in West Africa, then it may actually predate the migration 50,000 years ago because this lineage separated in the period between 50 and 100,000 years ago (of course the admixture could have been later) from other North African-West Asians. And, we happen to know of at least once Holocene era migration into this region of Africa that was prehistoric.

Circa 2010 we had a simple story. The Omo find in Ethiopia was the earliest anatomically modern human, dated to 200,000 years ago. Around 50,000 years ago anatomically modern humans left East Africa and replaced everyone else, in Africa, and outside of Africa. Recent human evolution proceeded from Africa, to the Near East, and then outward in all directions.

I believe this story is not fundamentally flawed in a broad sense. If we ever discover ancient DNA that dates to before 50,000 years ago in East Africa, I think we will find that most modern East Africans share substantial, perhaps even preponderant, ancestry from the expansion that dates to 50,000 years ago. But, the high diversity of extant Sub-Saharan Africans and the suggestions of deeply basal lineages through statistical inference indicate that “deep structure” of modern humans within Africa was not erased by migration from the north or east, but that the combination of the two led to the emergence of African lineages as part of the dynamic process of admixture ~50,000 years ago.

If Basal Eurasians and Basal North Africans, are discovered, they too are part of the story of “deep structure.” And, the earlier modern lineages that were present in eastern Eurasia may also be part of the “deep structure,” though population turnover may also have erased that imprint by the present.

Twenty years ago many scholars conceived of a model where a very small founding population in East Africa expanded and replaced all other humans in “blitzkrieg” fashion. There was massive radiation into Eurasia and Oceania ~50,000 years ago from a small ancestral population, while sister modern lineages in Africa were also replaced. This is still evident in indications of a massive bottleneck, and shared Neanderthal ancestry, in groups from Northwest Eurasia to the New World to Australia. But, there was also a lot of older “deep structure” that was absorbed and integrated. Some of this is easy to pick up because it was so different. Neanderthals. Denisovans. But some of them, like the Basal Eurasians and Basal North Africans, were likely part of the broad family of modern human lineages that were developing in concert for hundreds of thousands of years, a great fanlike phylogenetic tree descended from common ancestors. Further south in Africa there were probably other modern groups, whose phylogenetic relatedness would be a function of their distance to the proto-North African-West Asians. Out of the broad radiation of “modern humans” one group contributed disproportionately to the ancestry of most humans today. More in East Asia than in Europe, and more in Europe than in the Near East, and more in the Near East than Africa.

As we are in a stage of greater complexification errors and mistakes will be made. The evaluation of the models are only as good as the data we have. More data will come.

Addendum: I have not specified where the dominant modern signal is coming from. I think the candidates are probably the Levant, Arabia, all of Northern Africa, and Eastern Africa. Without far more ancient DNA in these regions we may never know.