Saturday, May 14, 2016

Using Somalis as a proxy: The second attempt

Over a year ago I made a post where I essentially took Somalis and tried to run-down, based on data from Shriner et al. 2014 and Hodgson et al. 2014,  how much "Somali-like" ancestry (labeled as "Cushitic" back then) is seemingly found in populations such as Tigrinyas, Amharas, Beta Israels and Xamir Agaws. [1] [2]

What I'd like to do now is have a second go at that with less "linguistic" terms so as to not completely 1:1 the relationship between linguistics and genetics but still be true to the genetic relationships between the populations at hand:

The above chart is pretty much based on an interesting notion explored in a prior blog post which is that there seems to possibly be a later West Eurasian ancestry carrying gene flow into Agaw speakers that predates the arrival of Proto-Ethiopian-Semitic, the latter of which modern Habeshas such as Tigrinyas likely owe about 10-15% of their ancestry to.

If there really was such an admixture event then it explains why the Xamir Agaw samples (possibly Afars. Check the comment section) and the Beta Israel samples (former Agaw speakers Ethio-Semitized relatively "recently") have about ~10% less of the later West Eurasian elements in the Horn than Habeshas such as various Tigrinya and Amhara samples do and it also explains some discrepancies in mtDNA lineages.

For example mtDNA N lineages such as X, T, H & J which are found in Tigrinyas and Amharas are, so far, not found in Beta Israels. Could these mark what the Southwestern Arabian migrants brought with them? And for one, there are N lineages in Beta Israels, so far, not found in Somalis such as mtDNA U6a1 & W. Could some of these be marks of the theoretical admixture event that may have hit Northern Highlanders before the arrival of Proto-Ethiopian-Semitic? [3]

For one, a recent study even dated the age of U6a1 in the Horn as being about ~4kya with an origin for it being further north in Northeast Africa (it's also found in modern Sudan and peaks in Copts) [4] which really fits with what I outlined in this former blog post as a theoretical point of origin for this admixture in Agaw speakers that is mostly not present in Somalis:

Link to former blog post

If the above was truly the case then what we have is that whatever this admixture event was caused by; it brought with it about ~10% new West Eurasian elements [note] to the Northern Highlands to inhabitants who would have been more or less identical to modern Somalis save for what could have been some admixture from Ari Blacksmith-like peoples who were likely Omotic speakers at a rate of 5-20%.

You can also adopt upper-thresholds from the data above like the Ari Blacksmith-like ancestry being up-to about ~15% as it can be for some Northern Highlander groups such as Beta Israels (I recall their average being a smidgen above that, in fact) and perhaps even the Copt-like West Eurasian elements going up to about ~15% as well just to have an idea of a lower and upper threshold. [note]

But the point here is that this was then added onto by elements from Southwestern Arabia which, about ~3,000 years ago (give or take a century or two)[5] [6], would have most likely housed people much more similar to Yemenite Jews than many modern Yemenite Muslims who, since the Islamization & Arabization of Yemen, have experienced new gene flows such as extra "African" ancestry.

So this would have taken the Agaw speaking group, which over-time would have become an Ethiopian Semitic speaking group due to this influx, and made them more like what modern Ethiopian & Eritrean Tigrinyas look to be:

The truth is much more right in-between the two above thresholds but I hope that can give one a general idea of what we're dealing with here. [note]

Although, there is a possible, but, in my opinion, less likely state of affairs which is that virtually all of the later West Eurasian elements are owed to Southwestern Arabia. Read through here to understand some of the possible rationale for and/or against such a possibility but this, by some slim chance, may in fact be the real state of affairs:

The "upper threshold" levels could be a percentage or two lower for the Somali-like ancestry as Amharas and Beta Israels in particular can be a few percentage points, on average, over ~15% Ari Blacksmith-like (see here) but all in all; I'd say that should give anyone interested a good idea of the ball-park estimates at hand here.

What makes these models intriguingly possible is that, at least all so far sampled, ethnic Somalis tend to seem rather un-admixed with foreign elements within at least the last ~3,000 years, particularly from an autosomal perspective. So they can, to some extent, be used as a slight proxy "basal" population state for many other groups in the Horn.

I.e. Tigrinyas = ancestors practically identical to Somalis ("Somali-like") + Ari Blacksmith-like peoples + peoples likely from Sudan who brought some Copt-like elements with them + admixture from some Yemenite Jew-like peoples OR you entirely do away with the hypothetical Sudan bunch when observing this model.

This is the same reason why even the Maasai, who seemingly have admixture from Southern & possibly also Eastern Cushitic speaking migrants from the Horn, will cluster between Somalis and populations such as relatively un-admixed Nilotic and Niger-Congo speakers as in the PCA above or the one below:

Now, there are perhaps some nuanced distinctions between their likely South Cushitic speaking ancestors and Somalis (something I've taken note of in the past) but for the most part, remarkably, Somalis still seem very close to the Horn-African Cushitic speaker admixture in Maasais which is remarkable because South Cushitic speakers began departing the Horn of Africa quite a long time ago for Southeast Africa.

The point of this post is; trying to outline the related-ness between Somalis and Ethiopian populations such as Tigrinyas whilst trying to give people an idea of roughly just how much non-Somali-like ancestry is in some of these respective groups whilst mapping out the relative effect various gene flow events we've noticed via genetic testing have had on these populations.

Reference List:


1. I've refrained from adding Oromos because of their heterogeneity and all the extra work it would take for me to do ball-park estimates for all the subgroups but honestly; the percentages presented here are hardly distinct from the ones I shared last year so use a little bit of your imagination and figure out how Oromos would look in this post for yourself.

2. The "methodology" for how the ball-park estimates in the chart were made were explained quite well in the old post from last year but to put it simply: it's clear that mostly all of the non-Ari Blacksmith-like & later West Eurasian elements in Northern Highlanders, like Beta Israels and Tigrinyas, are shared with Somalis based on data from Hodgson et al. & Shriner et al.'s ADMIXTURE analyses as well as runs from other studies and even what is implied via uniparentals like mtDNA lineages so I took everything that's mostly not found in Somalis (Ari Blacksmith-like elements, later West Eurasian influences discussed here and here) and marked them out as different from the remaining clearly "Somali-like" ancestry.

3. I'm working on conducting analyses on my own nowadays and would like to try out something similar to these exercises with formal stats someday but this should do for now.

4. The PCAs are owed to David Wesolowski who's the author of the Eurogenes genome blog and project.

5. Somalis themselves may very well, as some ADMIXTURE analyses (like the ones from Hodgson et al. and Shriner et al. posit) show, have some Ari Blacksmith-like ancestry so the Somali-like ancestry, I suppose, would go up to some extent in all the charts in this post if Somalis in fact a total "dead-zone" for such ancestry as posited in this old post.


  1. Interesting. I thought that the extra west asian shared by most other horners were somehow arabian like related, rather than coptic like. Which means Habeshas have less arabian than I expected.

    So most horners are essentially
    Nilo Saharan like (without west-central african admixture) + Copt like + minor arabian and ari blacksmith like (for non Somalis).

    Slightly off topic question
    How long ago did most of the (episodic?) admixture that caused and shared by all horners occur? I was surprised to learn that it's newer than the admixture that caused south asians.
    Was it estimated to be 10k years ago?

  2. Yeah, after noticing lineages like U6a1 among Beta Israels; I'd say the idea of Copt-like gene flow is pretty likely. My guess is that North-Cushitic speakers or really just any sort of predominantly Somali-like population in Sudan over 3,000 years ago passed on such admixture to Central Cushitic speaking Northern Highlanders and then the Proto-Ethiopian Semitic speaking community added onto this with likely Yemenite Jew-like admixture later on through the manner Christopher Ehret theorizes. I could be wrong and most of it might turn out to be Yemenite Jew-like, though I doubt it. Hopefully, ancient DNA might someday seal the issue.

    And yeah, your "low-down" on the likely ancestral structure of non-Somali groups like Habeshas is probably the case. "Nilo-Saharan-like (minus the West-Central African admixture) + Copt-like + Yemenite Jew-like + Ari Blacksmith-like OR there was little to no Copt-like admixture and almost all the newer West Eurasian elements are Arabian/Yemenite Jew-like which I doubt for reasons outlined in a prior post I'm sure you've read.

    As for your off-topic question (:-))... I really can't say regarding the early admixture events that brought Horners into being. Would really need to confirm that with aDNA someday but it seems to me that it was likely within the last 10,000 years. Cushitic speakers were, if we're to 1:1 linguistics and ancestry here and that's not always advised, seemingly north of the Horn in the Red Sea Hills ~ Nile Valley region around lets say ~6,000-8,000ybp when "Cushitic unity", from a linguistic perspective, likely existed and I have a feeling a lot of the admixture didn't go on too long before this. Again; further study and aDNA is definitely needed here and take what I've said regarding this with a bag of salt.

  3. I keep forgetting that there can be intermediates who pass on genes to a group. Like how south cushitics indirectly passed on west eurasian to south east africans without them ever getting into contact with eurasians.

    I see, thanks for answering. Man I hope we find ancient dna, the mota man gave me some hope lol. Do you suppose if somehow they find a west eursian ancient dna and the ancient east african dna that perfect fits in with our ancestry that our ssa/west eurasian percentages would go up or down?

    1. "I see, thanks for answering. Man I hope we find ancient dna, the mota man gave me some hope lol. Do you suppose if somehow they find a west eursian ancient dna and the ancient east african dna that perfect fits in with our ancestry that our ssa/west eurasian percentages would go up or down?"

      I don't think the percentages would vary very much. By a percentage or two or three? Sure. By 10-20% or something; no. Not unless the same thing happens to every other population that has similar "SSA" + "West Asian" roots. I.e. Egyptians hop from ~20% to ~30% or drop from ~20% to ~10% alongside our drops and hops in percentages.

      As for ancient DNA... We have some exciting prospects on the way, actually. There are some Natufian genomes on the way and these, if I recall correctly, ~10,000 year old "Kenyans" whom, I've read, the archaeology ties to originally coming from areas more northwards like Sudan. I for one am really hoping the Natufians prove real useful for our West Eurasian side and the "Kenyans" prove real useful for our "African" side.

      "I keep forgetting that there can be intermediates who pass on genes to a group. Like how south cushitics indirectly passed on west eurasian to south east africans without them ever getting into contact with eurasians."

      Heheh, it's important not to forget this. I.e. Copt-like admixture in Northern Highlanders is more likely to have come from people living in Sudan (pre-historic North-Cushitic speakers, for example?) than from legit "Egyptians" running on down to the Northern Highlands and/or the Eritrean coast and wholesale intermixing with the locals only to leave behind like ~10-15% of themselves over-time.

    2. Thanks for answering this one as well Awale.

  4. I thought Copts were a religious group, not an ethnicity ?

    1. They're not truly a separate "ethnicity" from Muslim Egyptians but there's been clear evidence for a while now that SOME of the religious minorities in "The Middle East" (Yemenite Jews & Copts, for example) are genetically distinct from their Muslim counterparts, on average. This is because they look to have avoided later admixtures their Muslim counterparts incurred. I've done various posts on this in the past and you'll have to do some of the digging for yourself but here're two that allude to what I'm saying:

      Hope this helps.