Iberomaurusians descend from a unique Sub-Saharan African lineage: Ancestral North African (ANA)

  I've been away for some time but one thing that's always stuck with me are the results of the Iberomaurusian (Taforalt) samples from the original Loosdrecht et al. paper [1] that published them and the later results from the Dzudzuana pre-print by Iosif Lazaridis et al. [2]

A two-way admix-ture model, comprising Natufian and a sub-Saharan African population, does not significantly deviate from our data (χ2 p ≥ 0.128) with 63.5% Natufian and 36.5% sub-Saharan African ancestry on average (table S8).

  

As quoted above, the original paper appeared to posit—though with only a modest fit—that the Iberomaurusians were essentially within the range of about 40:60 for Sub-Saharan:Eurasian ancestry by trying to model the Iberomaurusians as a two-way admixture between something Natufian-like and something Sub-Saharan that very intriguingly had broad SSA affinities. Seeming part West-African, East-African and even various sorts of SSA Hunter-Gatherer in its affinities:

These results can only be explained by Taforalt harboring an ancestry that contains additional affinity with South, East and Central African outgroups.

Whereas, upon finding the Dzudzuana samples from the Caucasus, that appeared to be an early form of Anatolian type Hunter-Gatherers and farmers a new model was put forward by Lazaridis et al.—once again with a rather modest fit (z-score of 2.4)—where the Iberomaurusians are instead modeled as close to 50:50 for Sub-Saharan:Eurasian ancestry:

They also, as you can see, shifted the narrative backwards in that it was most likely an Iberomaurusian-like population that contributed ancestry to Natufians rather than the other way around. Their percentages of Sub-Saharan-related and Eurasian ancestry in Iberomaurusians were also struck by another unrelated paper that mostly offhandedly touched upon the Iberomaurusians [3]:

This paper also landed on a more or less 50:50 Sub-Saharan:Eurasian background for the Iberomaurusians as you can see above, and I'm inclined to agree with it and Dzudzuana preprint in that their results are very consistent with how Iberomaurusians cluster on a global PCA between Sub-Saharan and Eurasian populations:

From looking at the above PCA they appear to cluster almost exactly the way Tigrinyas do in terms of distance between "purely" Eurasian populations—such as Western Hunter-Gatherers (Villabruna) and Anatolian Hunter-Gatherers (Pinarbasi)—and mostly "purely" Sub-Saharan groups such as Mbutis, the Gumuz and the Mota ancient DNA sample above. 

Their similarity in positioning to Tigrinyas is important because Pickrell et al. 2014 [4], as old of a paper as it is, did the best job I've ever seen of trying to calculate the total Eurasian ancestry of various populations across Africa such as Tigrinyas and had them at about 50%:

Its results are so good that they even, oddly enough, align somewhat closely with Somalis and Tigrinyas' mtDNA results where the former are at about 35-40% for Eurasian mtDNA lineages and Tigrinyas are at 50-55% which roughly fits even the autosomal variation I've seen among various commercial samples over the years where—barring outside recent admixture—Somalis to me seem 38-40% and Tigrinyas about 50-52%, though I'm not as sure with Tigrinyas since I've seen much less of them.

At any rate, the study did a good job compared to many others before or after because it utilized ALDER, a method that detects how much Eurasian ancestry entered African populations and when, rather than just assuming a simple one-time event. This allowed them to separate different waves of admixture rather than lumping everything together. They also reconstructed the ancestral Eurasian source instead of using modern populations like Sardinians or Europeans, which can distort results. 

By doing this, they provided a more accurate estimate of true Eurasian ancestry in African populations rather than forcing them into comparisons with groups that don’t fully represent their ancient genetic makeup. 

Finally, they recognized that bad reference choices can make ancestry "disappear." Studies before the Dzudzuana preprint failed to detect anything remotely Sub-Saharan related in populations like Natufians or as much Sub-Saharan related ancestry as there likely really is in Iberomaurusians because they compared them to the wrong groups such as Yorubas, Dinkas, the Mbuti or whomever else who may not be relevant to the type of SSA ancestry these populations actually carried. 

Pickrell and his colleagues avoided this sort of pitfall, making their results much more reliable than most and theirs is the only study I've ever seen where the percentages actually almost perfectly match a given population's global PCA position in respect to where they sit between Eurasians and Sub-Saharan Africans:

The results you see above are my own concoction using R and David Wesolowski's Global25 PCA coordinates. I calculated ancestry proportions by measuring the distance of each population to two reference clusters—one Sub-Saharan and one Eurasian—using mean PCA coordinates. The closer a population is to one cluster, the higher its assigned ancestry from that group, creating a proportional estimate of SSA vs. Eurasian ancestry.

The Sub-Saharan cluster in this case was represented by Mota, the Gumuz, and Mbutis, whereas the Eurasian cluster was represented by the previously mentioned WHG and Anatolian HG samples (AHG). I find it quite remarkable how I was able to so closely match Pickrell's results. In my opinion, much of the difference is attributable to the minor Eurasian ancestry in the Gumuz. I would not be shocked to see an almost 1:1 match between my results and Pickrell's if we ever have a "pure" AEA population to include among the SSA cluster.

It’s always encouraging when distinct methodologies yield similar results and this to me suggests Iberomaurusians were indeed roughly 50:50 Sub-Saharan:Eurasian with a slight skew toward the latter, as indicated by the Dzudzuana preprint, the Shum Laka paper, and the PCA positionings.

But moving on from that, you might wonder why I refer to Ancestral North African (ANA) ancestry as "Sub-Saharan" admixture. The Dzudzuana preprint and Shum Laka paper appear to theorize that the ANA lineage—if it existed— may have originated and spent much of its history no farther south than the Sahel, making "SSA" somewhat of a misnomer. While this might be true geographically, genetically, it is very much a Sub-Saharan population:

Above are a public simulated ANA component I encountered being used around the anthro-sphere from an unknown source (possibly made using Genoplot) and my own concoction in R. 

Funnily enough, I initially attempted to reconstruct ANA following Loosdrecht et al.'s ~65% Natufian-like and ~35% Sub-Saharan model whilst trying to explain Iberomaurusians' clustering but instead noticed this population—albeit with a poor fit—looked more like it contributed ancestry to both Natufians and Iberomaurusians with AHGs looking more like they fit the bill for the other side of Iberomaurusians' roots:

Generated using a Monte Carlo Simulation technique via the VahaduoJS tool

 That led me to try a different approach: instead of assuming Iberomaurusians were a fixed proportion of different ancestries, I used linear algebra to infer the missing population. I considered their PCA clustering position, the clustering position of AHGs, and calculated where another population must have been positioned to explain Iberomaurusians' pull away from AHGs:

ANA = Iberomaurusian + λ(Iberomaurusians − AHG)

Solve for that using something like R or Python and you should be able to get my final simulated ANA population's coordinates (it is the one used in the PCAs above):

My_Simulated_ANA,-0.494675,-0.00507719999999998,-0.0769332,-0.091086,-0.018835,-0.0725678,-0.1372456,0.0376134,0.2645316,-0.0786172,0.0414416,-0.0735542,0.1736056,-0.0870324,0.1799378,-0.067753,-0.027589,-0.1310216,-0.2921226,0.0829396,-0.064087,-0.2577662,0.1478732,-0.0187496,0.0461032

 

And what's incredibly interesting about this simulated sample I managed to put together is that it plainly clusters like a Sub-Saharan population and—like the 63.5% Natufian and 36.5% ANA based simulation and the public ANA simulation—it shifts Loosdrecht et al.'s model around in that both Iberomaurusians and Natufians look more like varying mixtures between something Anatolian Hunter-Gatherer related and this component rather than Iberomaurusians looking Natufian-like admixed:

Despite the poor fit for Natufians and the overdone fit for Iberomaurusians—we will surely need more appropriate samples such as Dzudzuana, real ANAs and ancient DNA from Egypt—quite interesting results because I effectively came to the same conclusion as the Dzudzuana pre-print before rereading and refreshing my memory on their findings and did so via a totally different methodology.

From the supplemental materials of the Dzudzuana pre-print

Again, always encouraging when differing methodologies point in the same direction like this. In both my PCA clustering of it and in the Dzudzuana pre-print it clusters plainly like a Sub-Saharan component, albeit like one that seems to perhaps have elevated affinities to Eurasians in a manner similar to but greater than Mota in displaying something of a pull toward Eurasians mostly along the y-axis of the following PCA:

 I strongly suspect that this is because—as the Dzudzuana pre-print's qpGraph essentially depicts—it is a cousin to the AEA cluster often discussed on this blog. A component that makes up most of the ancestry in Mota as well as the modern Dinka and Gumuz and even more so than this component ANA might have remained related to the Proto-Eurasians longer.

Nifty but perhaps simplified and not wholly accurate illustration of the Out-of-Africa migrations from Wikipedia

I'm sure those of you well-versed in population genetics are well-aware that all the populations outside of Africa including Native Americans, Pacific Islanders and Australasians appear to descend from a single population that expanded out of Africa sometime between 50,000 to 75,000 years ago: the "Proto-Eurasians".

This bottleneck may have formed in Africa itself before they left and I ultimately suspect, whatever ANA is, it probably represents a population that may have remained a sister population to these Proto-Eurasians right up until the last second. In a way Basal Eurasian before Basal Eurasian, but since it doesn't seem to have participated in the Eurasian bottleneck unlike Basal Eurasian it remains a firmly "Sub-Saharan" component in appearance:

And I would say it is possibly the source of most of the SSA ancestry we've always seen across the Middle-East, North-Africa and wider Mediterranean prior to the ancient DNA revolution when we were heavily dependent on global and regional ADMIXTURE runs. You will quickly notice that when ANA or a seemingly ANA admixed population such as Natufians or Iberomaurusians are introduced to a modeling of various MENA and Mediterranean peoples' ancestries their previous AEA and West-African (WA) affinities greatly depress:

I would venture to say that some populations, as you can see above, possibly never had AEA or WA admixture. It was all perhaps just ANA related ancestry being misattributed.

Then on the other side of things we now find populations such as Natufians, Neolithic Levantines and Iron-Age Egyptians who previously couldn't be modeled as part SSA being able to now show such ancestry because we found the unique SSA population we needed in the form of these hypothesized ANAs and they finally explain why groups such as Natufians and those Egyptians have a plain as day SSA pull in global PCAs:

I would strongly argue that—barring the interesting but unrelated things going on with East pulling groups such as Mal'ta boy, Ust-Ishim and the Han—you simply cannot pull away toward SSA populations on the x-axis and away from WHGs and AHGs in the manner above without possessing some sort of SSA ancestry and I suspect that, whatever ANA is, it may have existed along some sort of admixture continuum with AHGs from the southern Levant to what is now Morocco in different forms. Time and ancient DNA results will tell.

If and when this ANA group is found I personally don't doubt it will be shown to have originated and spent much of its time no more south than the Sahel. In fact, the same is likely true for AEA which only appears more south than modern Sudan in an admixed form such as in the case of Mota. But, in the end, both ANA and AEA clearly cluster like populations that did not participate in the Eurasian bottleneck just like any other "SSA" group.

References

1. Loosdrecht MV, Bouzouggar A, Humphrey L, Posth C, Barton N, Aximu-Petri A, et al. Pleistocene North African genomes link Near Eastern and sub-Saharan African human populations. Science. 2018;360(6387):548-552. doi:10.1126/science.aar8380. Available from: https://doi.org/10.1126/science.aar8380

2. Lazaridis I, Belfer-Cohen A, Mallick S, Patterson N, Cheronet O, Rohland N, et al. Paleolithic DNA from the Caucasus reveals core of West Eurasian ancestry [preprint]. Cold Spring Harbor (NY): bioRxiv; 2018. doi:10.1101/423079. Available from: https://doi.org/10.1101/423079

3. Lipson M, Ribot I, Mallick S, Rohland N, Olalde I, Adamski N, et al. Ancient West African foragers in the context of African population history. Nature. 2020;577(7792):665-670. doi:10.1038/s41586-020-1929-1. Available from: https://doi.org/10.1038/s41586-020-1929-1

4. Pickrell JK, Patterson N, Loh P-R, Lipson M, Berger B, Stoneking M, et al. Ancient west Eurasian ancestry in southern and eastern Africa. Proc Natl Acad Sci U S A. 2014;111(7):2632-2637. doi:10.1073/pnas.1313787111. Available from: https://doi.org/10.1073/pnas.1313787111

R Scripts and other files for the PCAs, charts and ANA simulations: https://github.com/Awale-Abdi/Anthromadness_ANA_post

Simplified and shortened for complete laymen:

Based on the findings of a peer-reviewed paper, a groundbreaking yet-to-be-peer-reviewed paper, and my own data analysis that corroborates and attempts to build upon their findings, there appears to have been an ancient population from over 15,000 years ago in North Africa that genetically clustered with present-day "Black" Africans. This group likely contributed ancestry to prehistoric populations such as the Natufians, and further influenced the genetic makeup of  modern Middle Easterners, North Africans, and potentially other populations with Middle Eastern ancestral ties such as those in Central Asia, South Asia, and at least Southern Europe. More on this in future posts and with future ancient DNA that rolls in.