Thursday, December 3, 2015

The Reer Xamar are a substantially mixed population?

A while back I encountered a young man of Reer Xamar origins (the "X" is pronounced like this) on both sides (paternal & maternal) who'd been genotyped via AncestryDNA & I ended up helping him figure out his origins at Anthrogenica via services like Gedmatch (running him through various calculators) for quite some time:

Above is a general gist of what we discovered about this individual's ancestry by running him through various analyses aimed at analyzing autosomal DNA. [note] He looks to be a mixture between Somalis (perhaps with some other Horn African input such as from Oromos?), West Asians (seem to mostly look Arabian or Levantine Arab related but some Iranian is plausible), South Asians & Southeast African Bantu speakers.

Of course, a great amount of emphasis was placed on this ancestry being old on his part. F.e. None of  his grandparents are full-blown ethnic Somalis neither any of great grandparents from what he's alluded to me. None are full-blown Arabs and so on. Instead his family is quite incessant that all of this ancestry including the very strange substantial South Asian ancestry (which is very real) is centuries old.

Honestly, I was expecting him to be somewhat "mixed" from the get-go as the historical data on Reer Xamars sometimes referred to more generally as "Benadiris" (they are speakers of Benadiri Somali) is that they are of somewhat mixed origins as even Wikipedia will simplistically outline:

"Although the Benadiri are sometimes described as the founders of Mogadishu (hence, their colloquial name Reer Xamar or "People of Mogadishu", though the city itself is postulated to be a successor of ancient Sarapion), their members actually trace their origins to diverse groups. The latter include Arab, Persian and Somali people."

Some sources will also outline that they have some ancestry from Bantu speakers of mostly or seemingly Southeast African origins given their strong involvement in Mogadishu's medieval and early modern slave trade.

So to my mind seeing some Somali & some Southeastern African Bantu speaker related ancestry as well as some Arabian and Iranian ancestry was plausible and expected but the real deal is quite shocking if this individual is representative (and I somewhat think he is as there is another person from a different region who is quite similar to him, though they are related) of Reer Xamars.

However what was ultimately shocking is how substantial his Somali-related ancestry is. As a matter of fact, he clearly draws the greatest proportion of his ancestry from the Horn of Africa / Somalis (could be some Oromo-related stuff in there I'm missing as there were some Oromo or Oromo-esque tribals in Southern Somalia during the Middle Ages and such), numbering up to ~50% of the chap's ancestry.

The most surprising element of his ancestry however is the South Asian element which seems substantial as it consistently shows up as ~20% of his ancestry, in fact even AncestrDNA's equivalent of 23andme's ancestry composition had him pegged as ~20% South Asian as does nearly every decent Gedmatch admixture calculator I've run him through.

This ancestry is further corroborated by how he constantly turns up with South Asian or South-Central Asian (Pakistan & Afghanistan) populations in various admixture calculators' Oracle-4 results where a person is modeled as a clean mix between 4 populations (each making up 25% of his or her ancestry) they best fit as a mixture of based on their ADMIXTURE proportions

He consistently turns up as something like "Somali or Horn African + South Asian or South-Central Asian + West Asian or North African + any population with substantial West-Central African / "Niger-Congo" related ancestry" (whether they be "Sudanese/ South Sudanese" or peoples like the Hema):

At this point his various ancestries are backed up by too many lines of evidence and analyses not to be real so it's quite perplexing where his ancestors could've acquired South Asian ancestry.

Contact between the Medieval and Early Modern Somali coast and South Asia was quite strong to a point where the Somali language itself has a plethora of Indo-Aryan loan-words (some via proxy through Arabic or Persian though) [2], and traders from the Somali coast as well as South Asian traders supposedly made stops on each other's lands.

Nevertheless, I never expected real inter-mixture to have ever occurred. The majority of ethnic Somalis so far tested seem quite homogeneous and while "mixed" are more of an ancient mixture that seems to have (for the most part anyway) remained endogamous through periods such as the Middle Ages, avoiding notable Arabian, Iranian or Ethiopian Semitic speaker or Oromo-related input. [note]

At any rate, strong history between the Horn of Africa's coastline and South to South-Central Asia or not; I was not expecting this chap to be so substantially South to South-Central Asian admixed. And at the same time so low on West Asian ancestry (either Arabian or Iranian) when Reer Xamars are quite often characterized as Peninsular Arabian (Yemenite et al.) & Iranian Plateau migrants who intermixed with the likes of Somalis to some extent.

Further details on this Reer Xamar's results are available at Anthrogenica where I ran him through a plethora of analyses and such in order to corroborate the kinds of inferences I've shared in this blog post

The most important point now would be to see the results of other Reer Xamars or Reer Xamar-related peoples and see if their results at all resemble this individual I encountered's results and I'm surprised to say that we (myself and this individual) encountered another chap who has extremely similar but somewhat different results to his:

This individual who is a relative of the Reer Xamar I originally encountered and helped out is of paternally Barawi / "Bravanese" descent & maternally Reer Xamar descent (of the "Shanshiyo" tribe like the full-blown Reer Xamar).

The Bravanese are basically Swahili speaking coastal inhabitants of Southern Somalia associated with the town of Barawa much like Reer Xamars are associated with Mogadishu (another name for Mogadishu is "Xamar/Ḥamar"). They too are thought to be some kind of Arabian-Iranian-Bantu-"Somali" mixture with, to my knowledge, no real mention of South to South-Central Asian related ancestry.

At any rate, this individual's results are staggering as while a bit different from that apparently full-blown Reer Xamar he's extremely similar to him and fundamentally of more or less the exact same mixture.

Old picture of Barawa / Baraawe / Brava

This individual basically shows up as a very similar mixture to the full-blown Reer Xamar although he does have (as you can see from that modified Eurogenes K=36 based pie chart I made) more West Asian ancestry than the individual who is fully Reer Xamar also resulting in lower levels of Somali or Somali/Horn African-related ancestry.

One cool line of data this new sample affords is Haplogroup data. You see, the reason I never touched on that Reer Xamar's Y-DNA & mtDNA (which would've been very interesting to touch upon) is because he got tested at AncestryDNA which does not test for Haplogroups anymore but simply autosomal DNA. This other chap who is paternally Bravanese however got himself genotyped by 23andme meaning we have Y-DNA & mtDNA results.

His Y-DNA Haplogroup is L-M20 (of its L1 subclade):

His mtDNA is L0f:

His Y-DNA given his substantial South to South-Central Asian Asian ancestry is likely owed to South to South-Central Asian ancestry especially since his subclade (L1) is supposedly most often found in India however as a colleague more knowledgeable about this Haplogroup pointed out- :

-an Iranian origin of this Y-DNA is possible because as I noted myself; this guy does likely have some Iranian (I'm speaking of the Iranian plateau here & not Iranic speaker related ancestry as a whole) ancestry.

His mtDNA L0f, funnily enough, tends to peak in South Cushitic speakers like the Iraqw (whose language I've uploaded recordings of by the way, as just a side-note) and substantially South (to perhaps East?) Cushitic speaker admixed populations in Southeastern Africa. [1] 

It is most likely to be owed to his Southeast African Bantu related ancestry as various Bantu speaking populations in Southeast Africa have some South Cushitic speaker related admixture (even if it's barely even noticeable via analyses such as ADMIXTURE but instead rears its head at times via Haplogroup frequencies). Although this Haplogroup, as a friend notes, is originally owed to African Hunter-Gatherer groups ("Khoisan" related and so on):

The hunters in Southern Somalia he's talking about are the likes of the Eyle who may indeed have contributed just a bit to some ethnic Somalis in Southern Somalia whilst being assimilated decades ago, and perhaps that's where he's getting his mtDNA but I personally find the SE African origin more plausible.

In the end his Haplogroups corroborate what autosomal DNA based analyses like admixture analyses or Oracle-4 are telling us which is that this individual is of mixed origins (carrying a South to South-Central Asian to perhaps even West Asian Iranian Y-DNA & either a Southeast African or Southern Somali mtDNA).

Old picture of Mogadishu before the Somali Civil War

Now, one thing that I'd like corroborated via more than one line of evidence is just how "Somali" these two seem (40-50%, that's an incredible amount of Somali to Somali-related ancestry, I must say) via analyses like Eurogenes K=36 which affords us components like "Northeast African" & "East African" (I've commented on this run's usefulness for noticing West Eurasian admixed Horn African ancestry here).
Now, ANE K=7 is a rather outdated run (the old ANE + WHG + ENF model is mostly dead now even more so with the recent finding of "CHG" although his ANE-related ancestry corroborates being substantially South Asian as they tended to show high-levels of such ancestry before CHG's appearance) but in the end it proves very useful in this context because it allows us to gauge just how "Eurasian" (Out-of-Africa ancestry) & "African" (non-Eurasian or substantially non-Eurasian admixed African ancestry) a person is.

 Various analyses that afford for a sort of "African" (green) Vs. "Eurasian (blue)" dichotomy in their results display relatively the exact same results where the full-blown Reer Xamar is about 35-40% "African" & the half Bravanese and half Reer Xamar chap is 30-35% "African". This does interestingly go in line with their estimated scores of Somali-related ancestry...

How so? Ethnic Somalis as I've outlined in the past are like many of their close relatives in the Horn of Africa; substantially West Eurasian admixed (possessing as a result a strong amount of likely West Eurasian Hunter-Gatherer & definitely Basal Eurasian ancestry).

In such admixture analyses Somalis tend to turn up as 55 to 65% "African" (most often ~60%) so suppose I take the Reer Xamar's result in the Jtest where he's ~38% "African" (essentially ~40%) and account for the fact that as various calculators show; about 5-10% of this is owed to West-Central African-related (most likely SE African Bantu-derived) ancestry. [note]

What do I have left? About ~30% "African" ancestry (though a tiny amount of this could be owed to West Asian Arab ancestry as they all have some "African" ancestry at a rate of 5-15%); roughly half the amount ethnic Somalis such as myself tend to show which corroborates that this individual could indeed roughly-to-nearly half ethnic Somali in ancestry.

South-Central Asia
I mean there's no doubt at this point that these two have substantial Somali or Somali-related (again; perhaps there's something Oromo about some of their ancestry but I haven't seen overt evidence of this) but I needed to be sure about these incredible levels implied by runs such as Eurogenes K=36 and I'd say their levels of "African" ancestry do corroborate such levels.

Although the problem here with the chap who is paternally Bravanese but maternally Reer Xamar and our Reer Xamar is that they're quite notably related (2nd to 4th cousin range related; though they were not aware of one another before matching on Gedmatch):

This relation can really skew things in that it would always be more ideal to have two totally unrelated samples & it could explain why they're so similar, though, somehow, I have a feeling that most Reer Xamars and such will turn out interrelated within the last few centuries to some extent due to how small their population is.

But one telling detail is that one of them is of a different group paternally yet both of them look very similar in ancestry anyway (implying perhaps that both of his parents; either Bravanese or Reer Xamar are not very distinct), this may imply that both groups could be well-represented by the type of mixture that characterizes these two individuals but for the time being; this is too small a sample size to be sure of anything.

This is all very very intriguing stuff but it's obvious what's needed here; more samples & more study... Won't be sure before but for the time being if these two individuals are at all representative of coastal people like Reer Xamars and Barawis then it can only be said that these populations are incredibly mixed and have quite the "cosmopolitan genetic profile". [note]

Reference List:


1. The fully Reer Xamar guy has a cousin who was tested at 23andme meaning he got some Y-DNA, some information about the supposed origins of his clan and his cousin's Y-DNA which could be shared with him (I'll have to ask soon if this cousin shares  a paternal line with him): [-]

2. Original K=36 results

3. I urge anyone of either Reer Xamar or Barawi / Bravanese descent who's had their genome sequenced to contact me via email ( and share their raw data or their Gedmatch kit number. The more samples we have the better!

4. I basically have the raw data of both these individuals and have uploaded their kits to Gedmatch where I ran them through a plethora of calculators there; most of the inferences made in this blog post are based on that. You can rummage about with their kits yourself here: A653627 (Reer Xamar) , M044752 (Half Bravanese & half Reer Xamar)

5. If you're wondering about the East Asian-related stuff they both seem to show; it's most likely owed to some of their other ancestries as various South to South-Central Asians as well as West Asian Iranians show such ancestry at relatively low levels. Similarly the affinities they show for African Hunter-Gatherer groups like Pygmies and San are probably owed to other ancestries in my humble opinion (perhaps their Southeast African derived ancestry).

6. I'll possibly be sharing newer and somewhat more accurate ancestry estimates with a future post. For now I'd say, for the three samples I so far have, the estimates are more like 35-45% "Somali/Horn African", 20-25% South Asian, 5-15% SE African Bantu-related ancestry with the rest being mostly West Asian ancestry.


  1. Fantastic job sir
    Do you have any whether people from the north differ from the south autosomally ?

    1. None that I've noticed... The south needs to be sampled more but the few southerners I've heard of or seen don't seem at all different from North-Central Somalis. Heck, I myself am partially "southern" (2/8) and fit quite naturally with people from the North.

  2. What is your explanation for the haplogroup T overrepresentation in northerners and does that indicate a different migration source?

    1. "Northerners" is too broad a term to use because it varies based on the ones we're talking about. Most members of the Darod clan from the north are E-V32 (my paternal line is entirely "northern" (Sanaag) and I'm E-V32) and not T-M184 whilst some Isaaq subclans are also predominantly E-V32 so "the north" is hardly some monolithic T-M184 entity but this Y-DNA marker does indeed dominate many Isaaq and Dir clan males.

      Honestly, the reason E-V32 and T-M184 dominate Somali males is likely just due to either bottlenecks and/or founder effects that just by chance led to these two lineages among various others our ancestors probably had (A-M13, some non V32 E-M78 lineages, J1 etc.) becoming dominant. It's likely that our "original diversity" was much like other Horn African ethnic groups that we're closely related to or at least more balanced (in a manner similar to our mtDNA frequencies). This phenomenon of over-representing one or two paternal lineages is interestingly common for people with pastoral nomadic roots and a history of patriarchal tribal systems (f.e. Arabians, Pashtuns, Somalis, Boranas etc.)

      Are unique migrations and events in pre-history or antiquity or the medieval era responsible for this? Definitely... But I wouldn't say there's some "South to North" migration dichotomy here specifically because "Northern" groups like the Dir clan who are predominantly T-M184 still have significant frequencies of E-V32 (it reaches a rate of 20-30% like in the recent Issa clan members tested in Djibouti or those old Dire Dawan Dir clan members) and then any group of Somalis like various Darod clan members or people from the south who are predominantly E-V32 tend to be 10-30% T-M184 (Darod Somalis sampled in Jijiga are one example here... I'd use an example of people from the south but no study comes to mind now) so there's clearly some shared paternal origins here...

      If there is some migration dichotomy here it might have more to do with clans and/or just certain groups rather than something as simplistic as "North Vs. South". Things look more complex than that...

      Further study (perhaps sampling remains from throughout the region's history someday when things have gotten stable & developed enough?) is definitely needed to perfectly understand the history of T-M184 and E-V32 among Somalis but for now that's most of what I can say.

    2. I think the reason the south doesn't have much of haplogroup t is cause
      Djoubiti,Somaliland and eastern Ethiopia("the north) are closer geographically to the source of haplogroup T which is Via Egypt.

      I also always wondered what happened to the foreigners who used to live in Saylac and berbera,seems southerners are more hospitable and more open then us

    3. What? The T-M184 in Somalis did not come from medieval inhabitants and all this stuff you're maybe thinking of... It's pretty obvious that it's a very old marker in the Horn (that has little to do with being "closer to the North") in that it's also found in South Cushites at a high rate:

      It's then found in lower frequencies in various Horn African populations that we're closely related to (Habeshas, Oromos, Agaws, Sidamic speakers et al.). It's obvious what's happened here; some Somalis just bottlenecked and or founder effect-ed (or whatever explanation we might see someday with further study) to make what was originally a present but likely obscure-ish Y-DNA dominant. Also, your "closer to Egypt" logic is pretty foolish... Habeshas and Agaws are closer to Egypt and tend to have a 5% T-M184 frequency and it's hardly even present in the likes of Nubians and the Beja:

      "South" & "East Cushitic" (Somali is a part of East Cushitic) are given a pretty close relationship by most linguists to a point where it is believed that they share an ancestor with one another that they don't share with Central (Agaw) and Northern (Beja) Cushitic, one might dub this shared node "East-South Cushitic" (hell, some linguists theorize that South Cushitic could be an off-shoot of Lowland East Cushitic, long story...) so for the descendants of very old migrants from the Horn (such as the Iraqw)- :

      -to have T-M184 the same way Somalis do says a lot about the origins of this HG among them & Somalis... Top that off with its presence in various other Horn groups we are closely related to & it's pretty obviously an old Y-DNA lineage that was present in some of the earliest Cushitic speakers in the region that Somalis by mere chance managed to make prominent among them.

      No fanciful theories required... And as for what happened to some of the non-Somali inhabitants of places like Zeila; we shall see with future study into the matter.

  3. one last question regarding the latest study(mota) if it actually true that up to 25% of Eurasian dna is the result of migration which dated to 1000BC how come Somalis lack any substantial Eurasian y haplogroup (J,,R,G Etc) ?
    How come there no historical documentation of such massive event?

    1. I wrote a rely to your questions but it was too long to fit into this comment section so here's a google doc link to it:

      On another note; gonna have to ask that you guys cease and desist with the off-topic subject matter (Somali Y-DNA) and questioning for the time being as this is not the place for it (email me if you want a discussion :-)) and my replies to your questions and/or posts are getting too long, partly because what you're posting requires such replies.

  4. This is very interesting. Thanks for sharing this information.

    I think it might be even the first time Benadiris have been analyzed like this.

    1. No prob and yeah, it seems to be. There are others out there according to the Reer Xamar whose raw data I have and one Barawi contacted me about getting tested so lets hope we have more and more samples soon.

  5. so reer xamar having 48% Somali are no different from otha Somalis who have close to 50% Eurasian dna.we all in the same boat, they no more or less Somali than the rest of us.

    1. Well, they're not necessarily like ethnic Somalis and I plan to come out soon with some newer and more accurate estimates. I'd say their mixture levels are more like 30-50% Somali-like ancestry (at least the samples I have) and the rest of their ancestry seems to be 20-25% South Asian, 5-15% West-Central African-related ancestry seemingly owed to Bantu speakers and much of the rest is seemingly West Asian ancestry (Iranian or "Arab" or both and such). In the long-run; they do carry a lot of ancestries Somalis don't (like their West-Central African-related admixture and their South Asian admixture) but sure; they're certainly not totally foreign to Somalia, at least the 3-6 samples I've seen aren't.

    2. they're obviously way less somali than real ethnic somalis what on earth are you talking about? they're either half or quarter somali while the average ethnic somali is 100%

    3. And where did I say at any point that they're as "Somali" as ethnic Somalis? What are YOU talking about?

  6. I when you say reer hamar do you mean shanshiya or a any of the somali in that area my dad a pure shanshia and n i want ti have him tested he doesn't look like a somali at all