Sunday, January 31, 2016

Some Horn African phenotypes: The Supposed Controversy

This is an odd controversy I've seen crop-up even extremely recently in the comment section of this video I posted months ago. The controversy seems to be what the West Eurasian ancestors of various modern Horn Africans looked like. Some even suggest they were dark-skinned and attempt to downplay just how similar they would have looked to modern West Eurasians.

The problem with the dark-skinned argument in particular is strewn out above. Somalis and other Horn Africans like the Beta Israel tend to have a frequency of about ~60% for the derived A allele of the SLC24A5 gene. You can check this through even your own raw genetic data by seeing if you're either "AA", "AG" or "GG" for the rs1426654 SNP (if you're AA or AG; you carry its derived rs1426654-A allele).

Why is this relevant? Well, it's because this gene is strongly associated with the pigmentation variation observed among modern South Asians, Europeans, West Asians and North Africans [1] and seems to have had a significant part in why individuals like the Algerian, Swedish and Pashtun man below are capable of being more de-pigmented than Africans like Dinkas or some South Asians like the Juang. [note]

It accounts for roughly 1/3 of the pigmentation difference between Europeans and "Africans", the latter of whom mostly don't have the derived allele but the ancestral "G" one (being "GG").

It also has a part in why we assume for now that Western European Hunter-Gatherers (a model of which is observable above with that blue eyed & brown-skinned man who's supposed to be La-Brana 1) were predominantly brown to dark skinned. It's because they, unlike modern Northern Europeans who are rich in WHG-related ancestry, and unlike Early European Farmers mostly from Anatolia; were mostly not derived for SLC24A5. [2]

They were seemingly "GG" or at times "AG" for the rs1426654 SNP (essentially so far lacking or having little of the rs1426654-A allele) whilst Early European Farmers who have their closest modern proxy in modern Europe being Sardinians- :

-were mostly "AA" much like most modern Europeans. It was they and the later expanding Steppe pastoralists who were seemingly carrying EHG and CHG related ancestry whom modern Europeans mostly owe their current de-pigmentation / "light skin" to.

There are of course other genes responsible for modern Europeans' pigmentation but the problem I am highlighting is that if some think the West Eurasian ancestors of Somalis and the earliest West Eurasian ancestors of Habeshas were essentially just like the Western European Hunter-Gatherers we've so far sampled in regards to skin-tone; they're already wrong as Somalis have an allele concerned with the development of light-skin that WHGs looked to, for now, mostly lack.

Again, the problem with the "dark-skinned West Eurasian ancestors" idea for Horn Africans' like Somalis' West Eurasian ancestors in this regard is rather simple... Many Somalis tend to be "AG" [4]  (such as myself) and some such as three of my own distant relatives can even be "AA" like Early European Farmers:

This suggests that our West Eurasian ancestors unlike some of the probably notably pigmented (darker-skinned) Western Hunter-Gatherers of Europe; would have been rather rich in regards to being "AA" with no doubt some "AG" individuals. Then there's modern Somalis and Habeshas' skin-tone variation that, while I dislike going into phenotypic traits, should be taken into account:

Somalis for example can vary between a near ink-black to an almost olive-ish skin-tone. The former or something somewhat lighter is more common back in Somalia itself (less common among the diaspora, I've found) where people spend much of their time outside basting in the sun but on average I'd say most people tend to be like those two women in the middle in the above image.

Though at the end of the day, all ridiculous biases and agendas aside, most will likely encounter someone in their own immediate and/or extended family within any of the above skin tone ranges (some maybe darker or lighter than the two extremes by a bit even). 

This to me suggests that Somalis are most likely the progeny, as their genomes somewhat suggest, of people who on the one end would possibly be as dark-skinned as modern Dinkas and people on the other end who would possibly be as light-skinned as some West Asians and North Africans like Peninsular Arabians.

The only real controversy here, in my humble opinion, is the one going on in people who think otherwise' heads because as I've noticed via experience; there seems to be some absurd "butt-hurt" reaction to clearly being, in some part, the phenotypic result of admixture which is nonsensical because most modern peoples' phenotypes only really developed within the last 5,000 to 25,000 years in some great part due to admixture.

East-Central Asians such as the Kazakhs above's phenotypes no doubt developed as a result of them being an inter-mixture between East Asian-related peoples and West Eurasians along with subsequent selection of course. 

Blonde or red-ish hair among some East-Central Asians? Most likely derived from their West Eurasian ancestors. Why do many East-Central Asians in terms of facial features tend to resemble various East Asians to some extent? Most likely due to their East Asian-related ancestors.

"ASI"s or frankly any component of South Asians' ancestry isn't truly well understood at this point as you can see here. We are in desperate need of ancient DNA from the region but ASI was thought by Reich and company to be related to the likes of the indigenous Andamanese or have contributed most strongly to South Indian groups such as the Paniya man and woman here

Regardless of ASI's validity as a full-blown component, Tamils' phenotypic traits are probably, in some part, to be the result of intermingling between peoples who looked at least vaguely similar to that Paniya man and woman, whom themselves have some notable West Eurasian ancestry, and peoples migrating into the subcontinent from West Eurasia among some contributions from other groups.

The two European politicians above (one Swedish and the other German) owe their phenotypes to selection and intermixing within the last 8,000 or so years. [2] Their phenotypes are the direct result of Western European Hunter-Gatherers, Early European Farmers and Pontic-Caspian Steppe Pastoralists intermixing.

Lighter eyes? Mostly owed to the indigenous Hunter-Gatherers of Europe. Light-skin? Looks to be mostly owed to the Anatolian Neolithic Farmers, Eastern European Hunter-Gatherers and Caucasus Hunter-Gatherer related peoples who contributed to modern Europeans' genesis.

Horn Africans like Somalis or the Tigrinya people above are likely no special exception. Our genomes clearly paint out a pre-historic inter-mingling between both "African" & "West Eurasian" peoples who would've been close to much of the ancestry in modern Dinkas in regards to the African side and much of the ancestry in Anatolian Neolithic Farmers or Bedouins (more so with the early farmers) in regards to the West Eurasian side.

So in line with this my opinion is simply: 

Looser curly hair among Horn Africans? Mostly Likely a result of nappy haired and straight to wavy haired people intermixing. No surprise; both nappy and straight to wavy hair can be found in modern Horn Africans. Horn Africans' skin-tones varying from ink-black to very light-brown? Most likely an intermingling between very dark-skinned and de-pigmented pre-historic peoples. And I highly doubt more ancient DNA will strongly contradict this.

A great many of the phenotypes you see today in the modern world are the result of intermixing between peoples who existed before these modern populations or even peoples who still exist in some cases, various modern Horn Africans are unlikely to be some sort of magical exception.

Horn of Africa

Although I must note something... Some people like in that video's comment section keep jumping on my saying that our West Eurasian ancestors would have looked at least similar to modern and even contemporary West Eurasians and think this means we don't truly have any inkling based on mere logic and current genomic evidence what they looked like.

This is false. What I mean when I always, as per usual, remain cautious and use the term "similar" is... We don't have ancient remains clearly representing our pre-historic ancestors on "either side"

We haven't sampled these hypothetical remains & we haven't tried reconstructing their facial features so of course we won't know exactly (down to the last hair follicle) how they looked until this is done but the notion because of this that they'd have magically looked obscenely distinct from modern and contemporary Eastern Africans (like Dinkas) & West Eurasians is a fantasy.

The only reason, in my humble opinion, this is a controversy to some is because again; there's something "insulting" or "troubling" to some people about their precious looks being the result of admixture in conjunction with selection.
Reference List:

4. Ethiopian Genetic Diversity Reveals Linguistic Stratification and Complex Influences on the Ethiopian Gene Pool, Pagani et al.


1. The first image was made by a Wikipedia member and was based on data from the Yale School of Medicine's Allele Frequency Database.

2. I'm really not one for posts like this... I like to steer clear of discussing the development of phenotypic traits like pigmentation, hair texture, facial morphology and so on because discussing this kind of stuff eventually, when dealing with various groups, devolves into some retarded exchange about which non-European population is more "White" or "Black" or "Negroid" & "Caucasoid" (scientifically obsolete nonsense I tend to have a dislike for...) and seems to carry a lot of unnecessary emotional vitriol with it. However, this particular "issue" was getting slightly annoying and I was getting tired of repeating myself to some people on this matter so I made this post mostly to just link future encounters to.

3. Also, modern West Asians and North Africans like Egyptians do have the derived allele for SLC45A2's rs1891982 SNP as well (From what I know; this other gene is responsible for a decent part of what remains when dealing with the pigmentation difference between Europeans and "Africans").

Arabians seem to be 15-20% derived for that SLC45A2 SNP and about 100% derived for the SLC24A5 SNP. Horn Africans like Habeshas, Somalis and Beta Israels do not look to carry the derived allele for this SLC45A2 SNP though whilst Neolithic Europeans & Neolithic Western Anatolians carry it at somewhat low frequencies.

Recommended Reads:

The Genetic Architecture & Natural History of Pigmentation

Wednesday, January 27, 2016

The Mota Mistake

It seems there was indeed good reason to remain cautious about Llorente et al.'s findings that all Africans are part Eurasian or West Eurasian as they've now retracted their old results after getting help from David Reich and company. 

Here's a quote from their erratum:

"The results presented in the Report “Ancient Ethiopian genome reveals extensive Eurasian admixture throughout the African continent“ were affected by a bioinformatics error. A script necessary to convert the input produced by samtools v0.1.19 to be compatible with PLINK was not run when merging the ancient genome, Mota, with the contemporary populations SNP panel, leading to homozygote positions to the human reference genome being dropped as missing data (the analysis of admixture with Neanderthals and Denisovans was not affected). When those positions were included, 255,922 SNP out of 256,540 from the contemporary reference panel could be called in Mota. The conclusion of a large migration into East Africa from Western Eurasia, and more precisely from a source genetically close to the early Neolithic farmers, is not affected. However, the geographic extent of the genetic impact of this migration was overestimated: the Western Eurasian backflow mostly affected East Africa and only a few Sub-Saharan populations; the Yoruba and Mbuti do not show higher levels of Western Eurasian ancestry compared to Mota. 
We thank Pontus Skoglund and David Reich for letting us know about this problem."

Funnily enough, despite this, there's still certain issues with their results. For one, their results for various Eastern Africans, now with these adjusted results, make no sense. For one, observe the below results using Yorubas relative to Early European Farmers (LBK):

At first I observed the lowered by 10% results when Mota is used relative to LBK and thought that they'd found out something interesting here (that maybe Horn Africans like Tigrinyas and Somalis are much less West Eurasian than previously thought) but you can see there's already something funky going on with these improved stats, at least for various Eastern Africans, in that the 10% drop occurs even when Yorubas & LBKs are used as well.

Problem is... We've had formal stats like this run before... Where Somalis, Tigrinyas, Amharas, Oromos and Beta Israels have been run against or with more or less non-Eurasian admixed African populations (i.e. Dinkas, Yorubas, Anuaks etc.) and more or less fully West Eurasian populations (i.e. Sardinians) and those weren't the results we got at all:


The above results are much more consistent with results we see in ADMIXTURE and also in regards to where Somalis & Habeshas cluster in global PCAs if compared to Anuaks for example:

Or when Yorubas are used as the African anchor point:

All prior analyses clearly imply that Somalis are mixed at a rate of something like ~40% West Eurasian ("Near Eastern" related ancestry) and Habeshas like Tigrinyas are at a level of about ~50% ("Near Eastern" related ancestry + some possible CHG-related ancestry). In fact, these two populations often look even more West Eurasian than that when Yorubas are used as the African population to compare them to.

 So those ~30% for Somalis and ~40% for Tigrinyas results are about as suspect as Llorente et al.'s old claim that African populations we'd always thought had very little or no Eurasian/Out-of-Africa input were all roughly ~7% Eurasian:

So I personally wouldn't latch on too much to the minuscule details of these new results beyond the fact that they're most likely quite correct about Yoruba or Mbuti folk definitely not being ~7% Eurasian, that always seemed a little odd to me hence why I made this post.

Reference List:

1. Ancient Ethiopian genome reveals extensive Eurasian admixture throughout the African continent, Llorente et al. (the study's text is pay-walled but the supplementary information which contains most of what you'll need anyway is free)


1. I may have to do some edits to this blog post of mine... While I consider it quite on point as to how "African" and West Eurasian peoples like Somalis and Habeshas are, and whilst the Llorente et al. data used in the post seems accurate at least for various Horn Africans, it's a bit tough to just leave up some of that data now. I may edit the post or honestly leave it the way it is because it's to the point enough and there's plenty of other evidence shared beyond the Mota study's data.

2. These chaps were, as a friend of mine put it, a little too eager to get their work published for their own good...

3.  Kurd is using the Lazaridis et al. Kenyan Somali samples in his analyses there and they're just a smidgen less West Eurasian than most Somalian and Ethiopian Somalis who've so far been sampled and some might be Borana admixed. The "Afro-Eurasian" component has a small amount of West Eurasian ancestry in it as Hadzas are used as the basis for it, hence Horn Africans' dipped West Eurasian scores in that run.

4. What could be causing those odd results may also just be the poor choice of using LBKs/Neolithic Europeans as a proxy for Horn Africans like Somalis' West Eurasian admixture. 

Sunday, January 24, 2016

Roman genome from West Asia

Well, we got some ancient genomes from Britain but what's of interest to me personally is that one of the Roman Period individuals sampled was seemingly of West Asian (particularly Southern Levantine looking) origins.

You can see him sitting in the "Middle East" cluster above as the yellow Roman sample, he and the other Roman Period samples looked to have been something along the lines of soldiers or gladiators but honestly; David Wesolowski does a better job of going into details than I'm going to bother doing so read his blog post on this matter.

What's of particular interest to me is that we seem to have confirmed via ADMXITURE and analyses like qpAdm that this individual, whilst non-negligibly "African", carries no Yoruba/West-Central African/Niger-Congo-speaker-related ancestry but seems to instead, much like Yemenite Jews below, be showing "East African" related admixture:

You can see what I mean with the below ADMIXTURE results for him (qpAdm doesn't actually tell us that he lacks such admixture from what I can see):

North_Sea: 0.02
Atlantic: 4.06
Baltic: 0
Eastern_Euro: 0
West_Med: 11.24
West_Asian: 10.99
East_Med: 46.16
Red_Sea: 20.98
South_Asian: 0
Southeast_Asian: 0
Siberian: 0
Amerindian: 0
Oceanian: 0
Northeast_African: 6.54
Sub-Saharan: 0.02

Pretty much nothing from the Sub-Saharan/West African/Yoruba peaking cluster but notable amounts from the Northeast African and Red Sea clusters which contain East African-related ancestry from what I can tell.

The samples top matching population in terms of pairwise Identical-by-State (IBS) allele sharing are present-day Saudis

These results are interesting as they confirm something I stated just this month in a prior post... Which is that West-Central/Niger-Congo-speaker-related ancestry in West Asian Arab populations looks to be something more recent in origin (perhaps owed to the Arab Slave Trade) but the East African cluster-related ancestry in the region looks to have a more ancient root.

This is often made obvious by how such admixture is present in the Muslim majorities but, in many cases, not their religious minorities, who may have not been as immersed in such trade and influences, such as Yemenite Jews lacking such input that is present in Yemenite Muslims or Coptic Egyptians (a North African population) seemingly lacking such input unlike Egyptian Muslims.

Despite these religious minorities still carrying non-negligible amounts of "African" ancestry as a whole but more East African/Nilo-Saharan-speaker-esque ancestry instead.

At any rate, I can't say much else that has been said in this Anthrogenica thread or at Eurogenes and will leave it at that...

Reference List:

2. Ancient human genomes suggest three ancestral populations for present-day Europeans, Lazaridis et al. (where the ADMIXTURE run including Yemenite Jews and co. comes from)

The Later West Eurasian element in Agaws

One thing that's really been of interest to me is the "extra" West Eurasian/West Asian related element in Agaws (Xamir/Afar samples & Beta Israels) and its origins.

Beta Israels and the Xamir (who may perhaps be Afars, though I doubt it. Check this comment section) have less of the later West Eurasian elements in the Horn that seem to mostly or entirely be lacking in all so far sampled Somalis but why? [note]

I think it takes some special pleading to claim the lesser influences in Agaws were brought to the Horn entirely by Proto-Ethio-Semitic speakers because these influences also seem to exist in Wolaytas who are a Southwestern Ethiopian and Omotic speaking population:

Wolaytas, for most of their history, have been surrounded by other Omotic speakers, Lowland East Cushitic speakers and Highland East Cushitic speakers with only a much later influx from Southern-Ethiopian-Semitic speakers such as Eastern Gurage speakers.

What's interesting about the Wolayta is if you look at their "mixture proportions" in various ADMIXTURE runs; they work quite well as a hybridization between an Agaw-Habesha type population and Ari Cultivators:

This suggests that they may seriously be the result of such a mixture OR they're more likely a mixture between Highland East to Lowland East Cushitic speakers in the region and an Ari-like population. This coincides a bit with how we once got the genomic results of a fellow who was apparently paternally Gurage and maternally Sidamo yet he didn't look too off from Northern Highlanders such as Agaws and Northern Habeshas in his results.

This may suggest that the influence in Agaws, that perhaps predates the Proto-Ethio-Semitic arrival, is in fact a pretty old one that might have made its way across much of the Ethiopian Highlands? It'd certainly be a treat to get more genomes from Southwestern Ethiopia.

Even when all of that is taken into account, one must grasp the history of "Agaw" populations. Beta Israels were only Ethio-Semitized within the last few centuries, and as you can see, maintain a profile you may expect from pre-Ethio-Semitic Northern Highlanders; lower new West Eurasian admixture scores and less West Eurasian ancestry than the likes of Tigray-Tigrinyas.

Then we have the Xamir Agaw samples, and whether or not they are Afars, it once again takes special pleading to argue that they derive 10-15% of their ancestry from the Southwestern Arabians who left their mark on Habeshas when neither Xamirs nor the Afar have been Ethio-Semitized and I doubt trade contact and such would cause them to incur substantial admixture from surrounding Habeshas without even being Ethio-Semitized.

Both Xamirs and Afars live in rather isolating environments (Xamirs are concentrated in a mountainous region (it's the Highlands of course...) with rivers around as well, if I recall correctly) and Afars live in one of the world's most unforgiving deserts.

Danakil Desert
Then one might want to take the uniparental distinctions between Beta Israels and the likes of Tigrinyas and Amharas into account [3]:

Those Ethiopian Jew/Beta Israel Haplogroup frequencies are almost identical to the results from another study and in both cases there are certain Eurasian markers found in Ethiopian Semitic speakers that are lacking in Beta Israels such as T, J, H & X which are also, so far, missing in Somalis.

Someone more knowledgeable than myself in regards to these Haplogroups and what to associate with them may want to chime in but I do wonder if it's possible that the Southwestern Arabian settlers in the Highlands brought these lineages to the region although mtDNA J & T seemingly exist among the Northeastern Sudan residing & Cushitic speaking Beja which is intriguing.

So what brought this influence to the Highlands but was avoided by Southern Cushitic speakers who migrated south and Lowland East Cushitic speakers like Somalis whose ancestors eventually migrated east into the northern part of the Somali Peninsula?

I often assumed the above... That peoples more north in Northeastern Africa migrated southwards and were possibly carrying more northerly and Copt-like or related ancestry and for some reason managed to leave a noticeable impact on the Northern Highlands and eventually, perhaps via proxy, regions more south.

If this had happened early enough then many of the peoples in Northern Sudan might have been genetically very similar to the Cushitic speaking Horn Africans of the time, evident by how there definitely seems to be some substantial pre-historic shared ancestry between even Somalis and Sudanese Arabs and how Sudan was likely inhabited by Cushitic speaking peoples (discounting Bejas) before the eventual dominance of Nilo-Saharan and then Arabic.

For all we know there's actually more "outside" admixture in Agaw populations and Highlanders as a whole than Omotic-related admixture and this later West Eurasian admixture indicate. If their early ancestors mixed with a people who were say "70-80% Somali-like" but just had some later West Eurasian influences then the influence of these people in Highlanders, even if they were literally a hybridization between these migrants and earlier Highland inhabitants, might be difficult to notice without ancient DNA.

Ties with Southwestern Arabia/Yemen cannot be fully ruled out as well. There's evidence of cultural ties between the two sides of the Red Sea dating back to even 2,000 BCE [4] but in this case the influence looks to have clearly come from the African side of the coast and not the opposite end so perhaps this is a poor contender.

So yes... This looks like quite the unsolved debacle. Is it owed to Ethio-Semitization? Doubtful. Is it owed to some migrations from further north perhaps bringing certain farming techniques or what have you? Who knows... We need more modern and ancient DNA. Is it owed to pre-historic exchanges with Southwestern Arabia? Again... Can't be sure but perhaps more doubtful in my personal opinion.

We can only be sure that it exists and that it seems to be of possibly pre-500 BCE to 1,000 BCE origins (predating Proto-Ethio-Semitic's advent) and may in fact be present in the likes of Highland East Cushitic speakers, many of whom have been assimilated into the Oromo ethnic group, and not just Northern Highlanders.

Reference List:

Sunday, January 10, 2016

IBS for Tigrinyas and Xamir Agaws

Thanks to David Wesolowski we now have IBS data for Tigrinya and Xamtanga speaking people:


I thought something like this would be an interesting exercise and I suppose shows in ways beyond ADMIXTURE that Habeshas like the Tigrinya are closest to fellow Horn African populations who speak Afro-Asiatic (Cushitic & Ethiopian Semitic) languages. 

Using formal stats (f-stats, D-stats etc.) might've been rather uninformative in this regard in that all Horn Africans like Somalis due to their West Eurasian admixture would also share a lot of genetic drift with West Asians like Yemenite Jews or Yemenite Muslims who themselves even have African ancestry hence IBS seemed like a good way to display "closeness" in this case.

Although, the shared "mixed" nature of these particular Horn Africans probably had some part in amplifying why they seem so close in regards to IBS.

Recommended reads:

Friday, January 8, 2016

African ancestry in West Asian & North African Arabs

I've gotten some questions about this at forums like Anthrogenica via private messages and I suppose it would be something some people would generally be curious about so this post will be dedicated to pointing out just how much "African" ancestry is roughly present in West Asian and North African "Arab" populations or whether or not they have any.

As you can see in the global PCA above utilizing autosomal DNA and basically 166,000 SNPs; West Asian and North African "Arab" populations such as Yemenite Muslims, Saudis, Palestinians, Moroccans, Egyptians and so on; all display a clear northward pull towards populations in Africa, away from from the orange cluster whom Southern Europeans such as Sardinians are within.

The reason for this pull these populations demonstrate is owed to what is noticeable above [note] based on the results from the old Eurogenes K=8 ADMIXTURE analysis or because of similar notable levels of "African" ancestry in these populations as noted by Hodgson et al. 2014:

Yemenite Muslims tend to be at levels comparable to those of North Africans such as Egyptians and Morrocans or even Mozabites (varying between ~15-25% on average from what I've noticed) most likely due to the Arab Slave Trade which resulted in them incurring later West-Central African-related input seemingly from peoples like Southeast African Bantu speakers which is noticeable in the K=14 admixture results from Lazaridis et al. 2013-2014 below:

Basically, various Muslim West Asian "Arab" populations, as opposed to some of their non-Muslim counterparts (Lebanese Christians, most so far tested Egyptian Copts, Yemenite Jews etc.), show signs of West-Central African-esque admixture probably acquired via the Arab Slave Trade. [note]

The more "ancient" looking East African cluster-related "African" ancestry (pink cluster that forms much of Dinkas' ancestry above) seems markedly older and is present in their non-Muslim minorities at notable levels (~5-15% in Yemenite Jews) who in the eyes of some academics seem like a bit of a throwback to some of these regions' pre-Islamic genetic profiles [3].

At the end of the day, these "African" contributions in Arabian and later Arabized populations of the Levant & North Africa are undeniable and can even have effects on formal stat based analyses like d-stats where they've actually caused some trouble for geneticists such as Iosif Lazaridis when trying to gauge just how Basal Eurasian some of these populations were:

These "African" contributions are also noticeable via uniparental data, most notably in regards to maternal markers / mtDNA like with the below Hadhramauti Yemenites who practically show as many Macro-Haplogroup L (discounting M or N ones) / mostly post Out-of-Africa migration "African" derived lineages as various Horn African populations:
Though they don't seem entirely representative of all the regions of Yemen as other studies with  Yemenite samples show lower but still substantial frequencies of non-M & N (which are more or less "Eurasian") Macro-Haplogroup L lineages among Yemenites:

In the end it seems quite clear that "African" ancestry similar and related to much of the ancestry in modern Yorubas or Dinkas or Anuaks and such is found all over the so called "Arab world" and is something that can be noticed via both autosomal DNA & Haplogroups whilst being backed up by various analyses from ADMIXTURE, PCAs to formal stats.

Reference List:


1. Some degree of variation is to be expected. Do not expect all members of every population to conform to these estimates perfectly (it's often why I use ballpark type estimates like 15-25%) as some variation is to expected but relatively all the members of any of these given populations do have at least some or notable "African" ancestry, this has been analysed too thoroughly at this point not to be the case.

2. I've generally placed the term "Arab" in quotation marks like "African" because it too is a rather broad term spread out across a rather ridiculously wide region due to Arabization, Pan-Arabism and so on. Many peoples who consider themselves "Arab" or whom are considered as such don't in fact have much or sometimes any Peninsular Arabian ancestry. I.e. The average Moroccan is closer to a Mozabite Berber than to a Saudi Arabian and I'd wager most Moroccans trace the overwhelming majority of their ancestry to the original Berber speakers of their region who were Arabized (and it certainly does seem to be so based on ADMIXTURE runs like the ones I've shared so far) .

Just how "African" & "West Eurasian" are Horn Africans like Somalis and Tigrinyas?

Alright, some of you who maybe revisiting this post will likely notice that it's nothing like it used to be. I've decided to pseudo-delete the old post by editing in new and more concise info. Partly because the way this post used to be felt a little too convoluted for its simplistic purpose.

Instead, I found that I managed to explain things much more comprehensively but also concisely to an acquaintance who was rather persistently inquiring about the exact West Eurasian admixture levels among Horn Africans such as Somalis and Tigrinyas:

That message above should serve as a good enough substitute to the majority of the meanderings found in the prior version of this post. It's much more "West Eurasian" centric than this post was originally meant to be but; you can just take the West Eurasian estimates discussed above and see the "African" elements as the rest of the ancestry, of course.

If anyone is interested in the ball-park admixture levels of several of the other Horn African samples we have nowadays then, to give you a simple idea of how admixed many of them are, here is the general cline in regards to West Eurasian ancestry levels in the Horn region among the groups whom we have samples from currently:

Anuaks<Gumuz<Aris<Borana Oromos<Wolaytas<Somalis~Certain Oromos<Xamir Agaws ("Afar")<Beta Israels~Certain Oromos<Amharas<Tigrinyas<Tigres~Bejas

I've thrown in Tigres because, despite them not appearing in any studies with autosomal DNA data, a friend managed to get his hands on one or two Tigre samples via 23andme and they seemed slightly more West Eurasian than Tigrinyas, much like the Beja samples from Dobon et al. did. So, the Tigre insertion above is just a guess based on what he showed me and the fact that Tigres are lowland semi-nomadic pastoralists (unlike other Ethiopian Semitic speakers) who are known to intermingle with Bejas whilst their language seems to have a Beja/North-Cushitic substratum. [1]

Nevertheless, whether or not Tigres are, on average, more West Eurasian than Tigrinyas will need to be corroborated by more samples in the future or a decent peer-reviewed study where the ethnic group is included.

Reference List:


1. The reference list is strangely empty this time around because most of the data I referenced in that message I linked to can be found all over this blog or in the sources shared in the old version of this blog post which you now have a picture of. So... Help yourself.

2. You can use methods like "masking" in some f-stats, from what I recall, but even so... All three methods are ultimately not perfect without good reference samples. 

Sunday, January 3, 2016

Habeshas are more or less acculturated Agaws

Inspired by an old post of Razib Khan's on Anatolian Turks and their genetic relationship to Armenians (and in truth also Pontic Greeks). I decided to make a similar point about the genetic nature of Habeshas, namely Tigrinyas and Amharas.

You can see in the above results from a global PCA (principal component analysis) using autosomal DNA data (166,000 SNPs) that Tigrinyas, Beta Israels, Amharas & Xamir Agaws are seemingly extremely close, particularly in terms of their levels of West Eurasian ancestry which tend to vary between ~45-55% collectively. 

In another Intra-East African PCA the extremely close relationship between both populations (Agaws; Beta Israels & the Xamir & Habeshas; Tigrinya and the Amhara) generally remains:

Habeshas like Amharas and Tigrinyas are essentially Ethiopian Semitic speakers whose languages have a Central-Cushitic (Agaw) sub-stratum and like with many linguistic shifts throughout history; some genetic contribution from the people whose language they adopted is to be expected. Linguistic shifts do tend to leave a genetic impression, however small or massive.

They tend to be massive when there's clear evidence of large scale population movements (i.e. via archaeology and/or historical records) and borderline population replacements where large-scale intermixing occurs. A good example of this would be the Indo-Europeanization of Peninsular Europe from the Pontic-Caspian Steppe:

In this case, we know through various peer-reviewed studies on population genetics where ancient genomes have been sampled, linguistic study, and archaeology; that pastoral nomadic peoples originally from the Steppe moved west-ward into the rest of Europe and essentially began intermixing with the local non-Indo-European speaking Farmer and Hunter-Gatherer populations and, in time, their languages (the Indo-European languages) would come to dominate Europe as they do today and mostly all Europeans trace a substantial portion of their ancestry to them. [1] [2] [3]

A more "small" contribution which still led to a linguistic shift and clearly not massive population changes would be the Turkification of Anatolia which was a gradual process, so gradual in fact that decent portions of Anatolia as late in history as the early 1900s was still Indo-European or ultimately non-Turkic speaking:

Populations like Pontic Greeks, Armenians among some other groups formerly quite numerous in Anatolia/Turkey are scarce in Anatolia now due to terrible events such as the Armenian genocide

However, Anatolian Turks are clearly, from a genetic standpoint, mainly derived from peoples very much like Armenians (and various Anatolian Greek peoples such as Pontic Greeks, if I'm not mistaken) as Razib Khan outlined in his own article on this matter:

The overwhelming majority of the ancestry in Anatolian Turks is clearly derived from native peoples from Anatolia itself with only something along the lines of ~10-15% of their ancestry being derived from the Central Asians who started the shift of their region toward Oghuz Turkic. This is not something Razib and Razib alone realized but something studied & backed up by a plethora of peer-reviewed genomic studies. [4] [5]

This ultimately non-trivial but not substantial contribution in Anatolian Turks' case is a good example of a "small" genetic contribution caused by a linguistic shift and the question is ultimately: 

From a genetic point of view, just how substantive does the Southwestern Arabian contribution in Habeshas seem to be, and just how much of their ancestry is clearly derived from the Central-Cushitic speaking inhabitants of the Horn who themselves seem to have some Omotic-related and later West Eurasian admixture? - note

Keep in mind that the "Afar" are Xamirs

If we observe the above ADMIXTURE analysis from Hodgson et al. 2014 and the below analysis of the same sort with similar results from Shriner et al. 2014- :

-it becomes quite evident that the overwhelming majority of Habeshas' ancestry is shared with their Central-Cushitic speaking neighbors.

Whilst Beta Israels were essentially Ethio-Semitized (from a linguistic standpoint) within the last few hundred years, they've maintained a similar autosomal pattern to the Xamir and seem to lack and/or show less of the "newer" West Eurasian elements found in Habeshas that're no doubt, in some part, owed to their Southwestern Arabian/Semitic speaking ancestors:


Basically, there's ~8-10% extra West Eurasian ancestry of the "Southwest Asian/Arabian" & "Caucasian" clusters in Habeshas that isn't present in Agaws such as the "recently" Ethio-Semitized Beta Israel and the still Agaw / Central-Cushitic speaking Xamir. [note]

Though it can be assumed that there is a little bit more "extra" West Eurasian ancestry in Habeshas because both these studies neglected to add an adequate "Mediterranean" cluster which Horn Africans such as Somalis, Oromos, Habeshas, Agaws and so on do show in various runs. Then one must take into account the fact that Ancient Southwestern Arabians likely had some "African/SSA" ancestry related to the East African cluster.

As the above cluster / global PCA implies; Yemenite Muslims and Jews definitely have a notable amount of African admixture (5-15% in Yemenite Jews and 15-25% in Yemenite Muslims). They also show signs of such admixture in ADMIXTURE runs such as David Wesolowski's K=8 or the run below from Behar et al. 2012:

So if we take the results from Hodgson et al. where Tigrinyas are ~4% more "Arabian" than Xamirs at K=12 and ostensibly ~4% more "Caucasian" and then just assume that if a proper "Mediterranean" cluster were present it would also perhaps amount ~4% or less, then the extra West Eurasian elements in Habeshas over the Xamir would add-up to about ~12% and if we add a small amount of "African/SSA" ancestry (~1-2%) likely derived from Ancient Southwestern Arabians then we might get something similar to this:

This would make Habeshas, very much like Anatolian Turks, only about ~10-15% "non-native" though the values once we have ancient genomes and even more detailed testing could go up or down, so perhaps ~8-25% should be more of a ball-park to account for any possible increases or decreases. Nevertheless, even if we move away from the Central-Cushitic speaking inhabitants of the Horn so far sampled for their autosomal DNA and simply look at the likes of Somalis- :
-Habeshas such as Amharas and Tigrinyas look to share far more of their ancestry with this population than with Yemenite Muslims, Yemenite Jews or modern West Asians in general and are on a more fundamental scale ("African" Vs. "West Eurasian" ancestry) vastly closer to Somalis than they are to any population native to West Asia:

This persists even in regards to what looks for now to be  more "recent" shared ancestry between the two population sets. Similarly, in terms of uniparentals, such as mtDNA & Y-DNA frequencies, one can see that Habeshas are seemingly of largely "native Horn African" origin, sharing the majority of their lineages seemingly with both the "Central Cushitic" & "East Cushitic" speaking peoples of the Horn. [note]

Horn of Africa

 So ultimately, Habeshas are, from a genetic point of view, of mostly "native" Horn African origins similar to how Anatolian Turks are mostly of "Northern West Asian" (Anatolia-Transcaucasia-Iran) descent with clearly some East-Central Asian derived admixture. [note]

Similar to how Turkification was gradual, Ethio-Semitization at the level it is at now was also something that was built up to at  a gradual pace. During the Medieval Period; Central-Cushitic / Agaw speakers were much more widespread in what is now the Amhara region where, sometime after Aksum's collapse, an Agaw dynasty even came to rule parts of the region and is, in some part, responsible for Lalibela's magnificent rock-hewn churches:

In fact, many "Amharas" or predominantly Amharic speakers in the Amhara region were seemingly Central-Cushitic speakers until not too long ago such as the Qemant people whom Beta Israels are linguistically a part of:

The apparently original scope of the Beta Israel

But of course, mass assimilation over the last several centuries has done much to change the demographics of the Northern Ethiopian Highlands and Agaw speakers are mostly a fragmented minority at this point. 

Nevertheless, assimilating them has certainly led to some level of heterogeneity among Amharas in particular (and perhaps also some Tigrinyas?) given how heterogeneous and sometimes Xamir-Agaw or Beta Israel-like some Amharas can seem:

In the end, it seems quite delusional to view Amharas & Tigrinyas as transplants from Southern Arabia or even as a 50-50 split between "natives" and Southern Arabians, they most certainly are not and are seemingly about as "Southwestern Arabian" as Anatolian Turks are "East-Central Asian". It seems obvious from an autosomal and uniparental point of view that the overwhelming majority of their ancestry is drawn from the Cushitic speaking inhabitants of the Horn, most notably so from Central-Cushitic speakers whose language group they were once a part of.

Reference List: