Thursday, December 25, 2014

Lazaridis Somalis from Kenya not all representative... Some are Borana admixed?

A Somali I correspond with who has an account on 23andme whilst being tested via the site brought to my attention that the Kenyan Somalis from Lazaridis et al.'s supplemental [1] reminded him of a Gharre Somali who was half Ogadeni. The Gharre apparently clustered between him and Borana Oromos in a PCA plot (Principal Component Analysis).

I was curious to see if this would be the case for the Kenyan Somalis and approached the author of Eurogenes; asking him to throw them into a PCA plot with Pagani et al. Somalis [2] who are essentially 100% representative of ethnic Somalis, coming out no different from over 90% to almost all of the full Somalis on ancestry services like 23andme and also coming out relatively identical to their Ethiopian Somali counterparts who were also tested in Pagani et al. with the only quantifiable difference being the low levels of Omotic admixture in many of the Ethiopian Somali samples (2%~ average).

Sure enough, about 6 to 8 of the Kenyan Somali samples came out as outliers among the Somalis in an "intra-African" plot integrating other Horners:

Link to Plot

 As you can see; 6 to 8 of them break off from other Somalis and cluster as intermediates between us and Oromos who are not of the Agaw ~ Ḥabesha-like Highlander group and not of the particular group that tends to overlap with Somalis in PCA plots-> essentially the southernmost Oromos in plots/ Borana Oromos who are passed around between studies and runs & tend to break off on their own.

Ultimately the one flaw with these plots is that they do not measure (as you can see) which group pulls more towards West Asians & which group pulls more towards Nilo-Saharans (Anuaks, South Sudanese et al.) and so on so I asked him to integrate these Kenyan Somalis with Pagani Somalis of both the Northern Somali & Ethiopian Somali variety:

Link to Plot

 I asked him to do so because Borana Oromos tend to come out less Middle Eastern (ME) admixed than other Horners and they break off often with Wolaytas who are known to be a little less ME admixed than Somalis. [3] They are essentially the southernmost Oromos in that PCA plot and also the Oromos who break off with Wolaytas in Hodgson et al.'s PCA plot. [4]

What this ultimately reveals is that 6 to 8 of these Kenyan Somalis look to be intermediates between representative Somalis & Borana Oromos. They do not at all overlap with the Borana samples but merely straddle between them and the Somali samples from Northern Somalia & Ethiopia who are quite representative of their ethnic group. Also, one must note that about 5 or so of these Kenyan Somalis perfectly overlap with the Ethiopian and "Somalian" Somalis from Northern Somalia so one musn't fall into the error of thinking that these 6 to 8 samples are at all representative of Kenyan Somalis at large. 

This also means they are a little less ME admixed than other Somalis. Another individual from the Horn, an Ethiopian of Oromo & Harari descent told me that at Lazaridis' K=20; these samples showed no West Asian (Caucasus) like admixture [1] which is a component Somalis often show at low levels however different studies and ADMIXTURE runs have different ways of identifying these components and I would not make too much out of such results but Borana are known to lack the more "Mediterranean" & "Caucasus" like admixed from the Middle East Somalis, Ḥabeshas & Xamir Agaws alongside other Oromos and even Wolaytas show most of the time (especially on the "Mediterranean" count) so perhaps this is the result of lowered levels of such admixture via an influence from Boranas.

This is consequently the running theory now as these Kenyan Somalis seemingly don't show Niger-Congo admixture (IIRCC as per Lazaridis' supplemental) so that is not at all what is causing the greater pull towards Nilo-Saharan & Niger-Congo speakers. Instead I would wager that they are possibly Borana admixed.

The Somalis in Kenya such as the Ajuran clan (associated with the old Ajuran Sultanate and its founders) often interact with Borana Oromos, in fact they interact to a point where many Ajuran are noted to speak Afaan Oromo as their first language and Somali as their second; they have also come to live amongst the Borana and adapt facets of their culture as is described here:

"Among the Kenyan Ajuran people, the majority speak the Borana language as their first language while others speak the Somali language as their first language especially those from Wajir North District in the areas of Wakhe and Garren. It is vital to note that since Somali is the language of wider communication in Northeastern Province, even the Ajuran who speak Borana as their first language learn the language. The link between the Garreh and Ajuran is their primary language which is Borana and not Somali."

The text comes from a missionary website [-] which utilizes sources like the CIA world factbook [-] and like to gather its info and the information is quite legitimate, it would seem. I have heard of their situation myself through a few relatives who've been to Kenya and encountered Ajuran Somalis & of course; the Borana are found in Kenya as well.

The other Somalis in Kenya tend to be Ogaden clan members however other clans are present... I ultimately assumed that these 6 to 8 peculiar Somalis might be a case of mixing between Somalis & Boranas given their intermediate clustering between us and them and in the Intra-African plot which is much more detailed than the world plot in noticing variation (consult my notes); one can see that a single one of them actually overlaps with one Borana Oromo.

Ultimately though, I would say that this plot Lazaridis et al. shared in their supplemental is perhaps not representative of Somalis:

More representative Somalis would pull closer towards EEFs like Stuttgart by some degree however it is interesting how the Lazaridis plot's Somalis look very close-knit despite their being clear "heterogeneity" present (perhaps they "treated for outliers"? I don't know). Ḥabeshas & Xamir Agaws alongside Beta Israels would cluster where the lonesome African-American (who looks to be 50%~ European based on his clustering) is sitting or perhaps just a bit closer to EEFs than he is.

On another note; the Kenyan Somalis look to be from Garissa.  All 13 of them were taken from the repertoire of a chap named George Ayodo whom Lazaridis et al. cites as the source of their samples:


The information for various samples is shared here. It would seem that four of the Kenyan Somali samples are Female while the rest are Male. It's interesting since Garissa is not exactly smack-dab where the Ajuran are known to be  (albeit these Somalis don't need to Ajurans to be Borana admixed) and I am unaware as to whether some Borana may be there or even if these Somalis are permanent residents. Either way though; this is all interesting to say the least.

Reference List:


 1. The world plot is hampered somewhat due to utilizing low quality SNPs/ a lower number of SNPs than the creator was used to. I.e. 100,000 SNPs were used instead of 200,000 mostly because the Kenyan Somalis and the Pagani Somalis (IIRCC) were tested using a different chip set hence only 100,000 or so of their SNPs overlapped. What this ultimately does is that groups demonstrate a little less variation than they usually usually would in such a plot however the data is relatively the same.

2. Borana tend to be about 20%~ Omotic admixed (OromoB samples, run created by another ethnic Somali using Pagani et al. samples) so I wonder if we can test these Kenyan Somalis for the component and see how they come out.

Happy Holidays! A Merry Christmas to any Christians (I'm not one) reading this or anyone who celebrates the occasion (I don't)!

1 comment:

  1. I know this is off topic here but do you know where did the Amhara and Tigray samples came from. which province did they come from?