Substack cometh, and lo it is good. (Pricing)

Tracing the paths of Noah’s sons

 
The above admixture graph is from a new preprint, Paleolithic DNA from the Caucasus reveals core of West Eurasian ancestry. To be honest, if you read the supplementary text there’s almost no point in reading the main preprint, as it is far more in depth when it comes to the methodology as well as spotlighting a variety of particular results. It’s hard to know where to begin with such a preprint so I want to highlight the “this is a simplified model” portion in the figure above. That’s actually the truth. Remember, no admixture graph is the Truth, it is an attempt by humans to capture concisely and informatively the major features of our species’ population history dynamics. The reality was never as clear and distinct as stylized graphical representations would have you think, and the researchers are aware of this.

In any case, if you want to really get at how they arrived at the conclusions they did, really read the supplementary section SI 2, “An admixture graph model of Upper Paleolithic West Eurasians.” The authors have so many potential combinations of ancestral populations that they can’t simply manually and intuitively posit admixtures. Rather, they have to explore a huge number of combinations (trees/graphs)…at which point they run into computational limits. This section explicitly lays out computationally efficient ways to automatically traverse the possibility space, and arrive at the best fitting set of models, within reason.

The title of the preprint says it all, but let me quote the abstract in full:

The earliest ancient DNA data of modern humans from Europe dates to ~40 thousand years ago, but that from the Caucasus and the Near East to only ~14 thousand years ago, from populations who lived long after the Last Glacial Maximum (LGM) ~26.5-19 thousand years ago. To address this imbalance and to better understand the relationship of Europeans and Near Easterners, we report genome-wide data from two ~26 thousand year old individuals from Dzudzuana Cave in Georgia in the Caucasus from around the beginning of the LGM. Surprisingly, the Dzudzuana population was more closely related to early agriculturalists from western Anatolia ~8 thousand years ago than to the hunter-gatherers of the Caucasus from the same region of western Georgia of ~13-10 thousand years ago. Most of the Dzudzuana population’s ancestry was deeply related to the post-glacial western European hunter-gatherers of the ‘Villabruna cluster’, but it also had ancestry from a lineage that had separated from the great majority of non-African populations before they separated from each other, proving that such ‘Basal Eurasians’ were present in West Eurasia twice as early as previously recorded. We document major population turnover in the Near East after the time of Dzudzuana, showing that the highly differentiated Holocene populations of the region were formed by ‘Ancient North Eurasian’ admixture into the Caucasus and Iran and North African admixture into the Natufians of the Levant. We finally show that the Dzudzuana population contributed the majority of the ancestry of post-Ice Age people in the Near East, North Africa, and even parts of Europe, thereby becoming the largest single contributor of ancestry of all present-day West Eurasians.

Ancestry from Dzudzuana

Longtime readers know that I hate the American racial term “Caucasians.” It’s pretentious when you could just say “white European,” because that’s what people really mean, judging by the fact that the real people from the Caucasus are marginally Caucasian in the eyes of many Americans. The genealogical origin of the term goes back to Johann Friedrich Blumenbach. And yet this paper takes these two samples, and finds that a lot of the ancestry of modern groups can be attributed to them! (also, a religion interpretation of the results is in the title of the post)

To be fair, they caution that these ancient Caucasian samples are representative of a particular thread of human heritage, not that the center of this thread was necessarily in the Caucasus. This does make me wonder about ascertainment bias in the Near East toward samples from mountainous areas which were colder. But, at the granularity they are attempting to understand human population history, it’s probably not that big of a deal. Ultimately, they conclude that this Paleo-Caucasian population contributes “~46-88% of the ancestry” of modern Europeans, Near Easterners, and North Africans. That’s kind of a big deal.

There are so many results in this preprint, so I think we need to back to the “beginning” of the non-African branch. The Paleo-Caucasian sample is of note in part because it is from before the Last Glacial Maximum, and, about halfway back to the massive diversification of most non-African populations around 55,000 years ago.  Using the Paleo-Caucasian samples’ affinities this preprint reinterprets results from last spring on ancient DNA from Northwest Africa. In that paper, the authors conclude that Paleolithic North Africans were a mix between an unspecific Sub-Saharan population and Natufians. Here though the authors suggest that the Natufians and Yoruba both received gene flow from Paleolithic North Africans. And, these Paleolithic North Africans were themselves mixed between something similar to the Paleo-Caucasians (a mix between an ancient West Eurasian ancestry and “Basal Eurasian”), and a “Deep” ancestry which diverged from other non-Sub-Saharan Africans before the Basal Eurasians did.

The reason that the Paleo-Caucasian sample is so important is that it allowed the researchers to see that the early Holocene Near East, where Anatolian and Iranian farmers, as well as Natufians in the Levant, were ancestral to many later groups, was subject to many genetic changes from before the Last Glacial Maximum. The Natufians seem to be well modeled as having ancestry from the Paleolithic North Africans as one of the major ways they are distinctive from the Paleo-Caucasians. This presents us with a reasonable model for the west to east movement of haplogroup E, and, the Afro-Asiatic languages. The gene flow of Paleolithic North African also explains the non-trivial level of Neanderthal admixture which is found in the Yoruba population. This is mediated through the presumed back migration of Paleo-Caucasians from the Near East at some point in the Pleistocene, contributing some Neanderthal ancestry to the genetic background of Paleolithic North Africans.

Additionally, the distinction between western (Anatolian/Levant) and eastern (Iran) farmers during the early Holocene can now be understood as a product of later admixture into eastern proto-farmers of basic Paleo-Caucasian stock. The relative closeness of Anatolian farmers to the Paleo-Caucasian samples is indicative of the fact that there was an “Ancestral North Eurasian” (ANE) admixture cline into the Near East during the Pleistocene, which meant that some populations to the east became rather different from the pre-LGM samples. Probably after the Last Glacial Maximum proto-Siberian ancestry became prominent in the zone between the Caucasus and Iran (additionally, some of the models imply there was eastern Eurasian ancestry). This is in keeping with the fact that ANE ancestry does seem to have been found in places like Khorasan before the expansion south of steppe populations after 2,000 BC.

As noted in the abstract, Paleo-Caucasians had Basal Eurasian ancestry ~30,000 years ago. This increases the likelihood that Basal Eurasians weren’t recent migrants from deep inside Africa. Additionally, for various reasons, the authors are now positing a Deep ancestry which diverged even further into the past. Both Basal Eurasians and Deep populations seem to lack Neanderthal admixture. The authors also repeatedly suggest that Basal Eurasians were part of the Out of Africa bottleneck event. In Who We Are and How We Got Here David Reich presents the model that this bottleneck population had a low effective population size for a long time. This seems plausible because the genetic homogeneity that you see in non-Africans is pretty striking vis-a-vis Sub-Saharan Africans. On the other hand, this work confirms earlier results that imply that Basal Eurasians did not admix with Neanderthals, and also indicates that the divergence has to be greater than 60,000 years before the present from other non-Africans, who diversified more recently.

In contrast, the Deep ancestry group, which nevertheless forms a clade with the new Eurasian lineages (Basal and non-Basal), does not clearly seem to have undergone the bottleneck event according to this preprint. It’s more a matter of what they don’t say, rather than what they say in this case.

The big picture needs to be integrated I think with the new “modern humans emerged through a multi-regional process” within Africa. If you think of modern humans as emerging across an African range which shifted in the Near East based on oscillating climatic conditions, the ancestors of the “non-African” lineages can be thought of as one of the main deeply rooted lineages, probably in the northeast of the continent. During the Pleistocene, the Sahara was even more brutal than today during many periods, so it is not implausible that some of these marginal populations on the edge of Africa were subject to long periods of very small effective population sizes. Most of them presumably went extinct. But one population was probably far enough north and east that it had a little more margin to play with. This population was probably connected along the Mediterranean littoral at some point with the Deep component in North Africa, which had higher effective population sizes because the mountainous terrain of the Atlas region was always going to remain more clement through dry phases.

At some point one a group of the bottlenecked population mixed with some Neanderthals, and began to break out of containment in southwest Asia. If I had to bet money, I suspect there were already other related groups, probably somewhat admixed with local hominin lineages, further east. That is, I believe the archaeological results in Southeast Asia, and think that those in Australia are credible. But these groups were probably small in number, and totally absorbed by the later migration wave.

Also, the timing of the separation of Africans and “non-Africans” is such that I wouldn’t be surprised Qafzeh-Skull people were somehow ancestral to, or closely related to, the ancestors of non-Africans.

Finally, let’s remember that the authors were focusing on North Africa and Western Eurasia in this preprint. Things will get more complicated as East Asia and Africa come “online” in terms of these analyses. Of course, we are going to be helped by the reality that human genetic variation is not arbitrarily and randomly distributed, but reflects real constraints in our evolutionary history and the forces of geography as well as contingency. The non-African story is made simpler in part because of the great bottleneck, and especially the common descent of most peoples from the population that mixed with Neanderthals. The modeling of effective population size changes over time in Sub-Saharan groups does not lead us to believe that it will be so simple in that continent.

Related papers: The genetic history of Ice Age Europe, Tales of Human Migration, Admixture, and Selection in Africa, and Genomic insights into the origin of farming in the ancient Near East.

8 thoughts on “Tracing the paths of Noah’s sons

  1. The Fig 3a from the paper labelled “Shared drift with Dzudzuana” used in this post actually doesn’t describe the shared drift with Dzudzuana as such but is model proportions from their model. I thought it was an f3 outgroup shared drift at first (since that’s normally how that is presented) but is not!

    (The model itself I find clearly questionable is its assumptions, which are deliberately insensitive to divergences between different strands with splits deeper than the splits between ANE and ENA and the various UP Euro populations, and within the Near East).

    Razib: “In contrast, the Deep ancestry group, which nevertheless forms a clade with the new Eurasian lineages (Basal and non-Basal), does not clearly seem to have undergone the bottleneck event according to this preprint. It’s more a matter of what they don’t say, rather than what they say in this case.”

    I wonder what estimates of within population diversity, as a direct measure in the Taforalt and Natufian populations (supposedly with “Deep” ancestry), tell us about this? Relative to populations such as Iran Neolithic, who only have Basal Eurasian ancestry. If “Deep” did not undergo *the* bottleneck, did they undergo *a* bottleneck?

  2. I’ve been reading for a while, but don’t recall your dislike of the term “caucasian”. I would have thought what you read was focused enough on science that most people would be using the more scientific reading. In contrast, non-scientific writing about the binary between “people of color” and people without typically includes non-European caucasians in the former category without actually using the aforementioned term.

  3. They estimate the divergence of Basal Eurasian at >57.5 kya, so seemingly not that long before the diversification of other early Eurasians. From the supplement:

    “Assuming a conventional Ne=10,000 and taking FST≈0.022 (the mean of the f-statistic and qpGraph estimates of 0.021 and 0.023) of the we obtain t=445 generations; with a generation length of 28.1 years, this corresponds to ~12.5ky. Given the age of ~45kya for Ust’Ishim, it implies that Basal Eurasians split from other non-Africans >57.5kya and also prior to the Neandertal admixture into Ust’Ishim.”

  4. They estimate the divergence of Basal Eurasian at >57.5 kya, so seemingly not that long before the diversification of other early Eurasians. From the supplement:

    estimate very sensitive to parameter assumptions though, as they say. a 60-100 K interval seems reasonable, though the distribution is probably left-skewed toward the low end?

  5. So ANE is admixed after all? Could this mean that (ancestors of) Amerinds weren’t mixed with ANE but the other way around?

    How about EHG? Could it partly be a Vestonice remnant?

    And if CHG derives from Dzudzuana with some Siberian or East Eurasian (am I understanding this correctly?), how do they have such high Basal admixture?

  6. So ANE is admixed after all? Could this mean that (ancestors of) Amerinds weren’t mixed with ANE but the other way around?

    everyone is admixed! probably depends on ANE sample for how much. but remember that with current data multiple phylogenies can match….

  7. estimate very sensitive to parameter assumptions though, as they say. a 60-100 K interval seems reasonable, though the distribution is probably left-skewed toward the low end?

    Yeah, but the relative ages should be somewhat right. Ust’-Ishim does not appear too far removed from the proto-Eurasians or the initial admixture event with Neanderthals, so the divergence of BE should not be very old relative to other Eurasians if their phylogeny is somewhat accurate.

    However, if the phylogeny from the other paper with a deeper Basal Eurasian branch contributing lower levels of admixture in the Near East is closer to the truth, that would be consistent with a fairly old BE divergence.

Comments are closed.