Tutsis are genetically very similar to Masai

Many years ago, before I used ggplot, I did a little analysis of the genetics of the Tutsi. Actually, it was the genetics of a single Tutsi, or more precisely, someone who was 75% Tutsi ancestry (3 out of 4 grandparents).

I found that the Tutsi individual seemed quite distinct from the Bantu peoples in nearby Kenya. I suggested that it was likely that the Tutsi were then genetically distinct from the Hutu people amongst whom they lived. For many years this was part of the genetics section of the Wikipedia entry on the Tutsi, but recently the reference was removed and the page seems to have been re-edited.

That’s fine. I’m just a random blogger who had one sample. But as it happened recently about a dozen Diasporic Tutsis reached out to me. Over the last decade, the number of people who have been genotyped has increased greatly. So it wasn’t that difficult for interested parties to find these genotypes.

The mission they put before me is simple: “tell us about our genetics”. Over the next few weeks, I’ll do that. As there is no IRB, this won’t be published in a peer-reviewed journal (I am open to putting any researcher in contact with these Tutsis who reached out to me). I’m just going to put what I find out there so that Tutsis who do personalized genetic testing can make sense of what they’re finding out.

I received these genotypes today. A quick merge of samples I have reduced it down to 50,000 markers. I will work on creating a merge with a larger number of markers. But, I’ll report what I have found out so far as a first pass.

As you can see on the PCA plot above the Tutsi overlap almost perfectly with the Masai. Not with the Kenyan Bantu, or the Luo, who are more “African” shifted. But with the Masai. But, they are not as “Eurasian shifted” as the Somali.

Treemix confirms this:

The Tutsi and Masai are right next to each other. Between the Somalis, and other Sub-Saharan Africans.

Running 5 migration edges, you see gene flow into the Tutsi from Cushitic populations.

I ran Admixture at K = 6. You can see the raw results, but the chart makes it clear, the Tutsi are like the Masai:

The same for K = 7:

I think that’s all for now. More later!

16 thoughts on “Tutsis are genetically very similar to Masai

  1. Thank you so much for this truly eye-opening post. This is a new beginning to genetic study of our people. Whenever we look up online, nothing comes up. But from your courage, we will understand ourselves better. Thank you Razib, you are our hero for doing what many failed to do.

  2. Nice finding, Razib. However, can you use more East African samples? In particular Nilotes (South Sudanese), Hadza, and Omotic (Ari, Chabu) populations so more East African-specific clusters form that can tell us more.

    Sudanese samples are available here: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006976

    Omotic samples: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1005397

    pan-Ethiopian samples: https://www.cell.com/ajhg/fulltext/S0002-9297(12)00271-6

    Perhaps also add more Yemeni populations, some new studies out there with a diverse pool of Yemeni Arabs: https://onlinelibrary.wiley.com/doi/abs/10.1002/ajpa.23312

    Also, there is a new study out focused on ‘minority’ East African populations: https://www.pnas.org/content/116/10/4166


  3. Does it mean that the prevailing view that the Masai have been historically recently admixed with the Cushitic populations during their Southward expansion is incorrect, and the elements of Cushitic DNA there are considerably older and more Northerly in origin?

  4. Can you also add east African hunter-gathers like that Hadza and Tanzania Pemba 1400bp. Also southern cushitic and South-eastern cushites groups like Iraqw, Tanzania Luxmanda 3100 bp, Borana, Rendilles. Even cushiti c admixed nilotes like Datooga and Samburu would be great reference, since, phenotypically are very similar to Tutsis.

  5. The available Y-DNA data on the Tutsi and Maasai does suggest that their Bantu and Nilotic ancestors absorbed some Cushitic males, so the ancient gene flow inferred above could be legit.

    If we consult the peer-reviewed genetic data, Luis et al. (2004) observed that their Rwandan Tutsi samples largely belong to the same paternal haplogroups as their Hutu neighbors. Most of these lineages consist of the E3a/E1b1a clade, like the majority of other Bantu speakers.

    Luis et al. (2004):



    However, Trombetta et al. (2015) report that 22% of their Burundian Tutsi sample, like 24% of their Kenyan Maasai sample, carry the M293 subclade of the E1b1b haplogroup. The majority of the Tutsi and the Maasai in the study bore instead other haplogroups, but the moderate occurrence of M293 in these populations is still indicative of gene flow because that subclade is closely associated with Southern Cushitic speakers. Henn (2008) found the E1b1b-M293 subhaplogroup at highest frequencies among the Wafiome/Ufiomi/Gorowa, who speak a remnant Southern Cushitic language (100% frequency).

    Trombetta et al. (2015):


    https://bit.ly/2Ifjuzb (Supplementary Table 7)

    Henn et al. (2008): https://www.pnas.org/content/pnas/105/31/10693.full.pdf

    On the whole, this does indicate that the ancestors of the Tutsi and Maasai likely assimilated some Cushitic males. Specifically, some Southern Cushitic-speaking men, such as the ancestral Iraqw and Wafiome, since the latter mainly belong to the M293 subclade of E1b1b. Eastern and Northern Cushitic speakers, such as the Somali, Oromo, Beja, Afar and Saho, primarily belong instead to the V32, V6 and V22 subclades of E1b1b, so their Cushitic ancestors probably did not directly contribute to the Tutsi’s and Maasai’s gene pool (see Trombetta link).

    The Maasai mtDNA analysed by Castri et al. (2008) is similar to that of other Nilotes in the Great Lakes region, mainly consisting of L derivatives. It is quite unlike that of the Afroasiatic speakers to the north in the Horn region, as observed by Non (2010) (50+ percent of the Horn individuals carried instead haplogroup M and N derivatives). This confirms that the gene flow from Afroasiatic speakers into the ancestral Maasai’s gene pool was largely driven by Cushitic males rather than females.

    Castri et al. (2008):


    Non (2010):




    Although there’s no published mtDNA data on the Tutsi, we can infer from Castri et al. (2009) that they probably have a similar maternal profile as the Maasai. This is implied by the fact that the Hutu, who have over the centuries unidirectionally married Tutsi females, almost entirely bear L derivatives like the Maasai. Only 0.4% of Hutus were found to carry non-L haplogroups, whereas the M1 clade is perhaps the most common maternal haplogroup among Cushitic speakers.

    Castri et al. (2009): https://docdro.id/Tc1kcF1 (Table 1)

    We know from Henn (2008) that among Nilotes and Bantus, gene flow from Cushitic males is maximized among the Datog. However, Non (2010) reports that the mtDNA of the Datog is similar to that of other Nilotic and Bantu populations in Eastern Africa, whereas the mtDNA profiles of the Afroasiatic speakers in the Horn region cluster with that of other Afroasiatic-speaking groups in the Arabian peninsula. This means that the Datog’s Nilotic ancestors, like the Tutsi and Maasai, also primarily received their Afroasiatic admixture from Cushitic males rather than females.

    Had the Maasai and Tutsi been examined in the Tishkoff lab’s latest meta-analysis of Afroasiatic, Nilo-Saharan, Niger-Congo and Khoisan-speaking populations in Africa, these two groups at K=2 probably would have shown significant Afroasiatic admixture, though not quite as much as the Datog. The Datog sample therein appears almost 50% like the Mozabite Berber sample, whereas the examined Cushitic and Semitic-speaking Horn populations are predominantly Mozabite-like. This is a considerable improvement over the lab’s previous meta-analysis from a decade ago since it is more consistent with the aforenoted uniparental marker data/non-inferred data.

    Scheinfeldt et al. (2019):



  6. @Ryan

    If you crossed a tetrachromatic woman with a honeybee who was also an artist, maybe, but probably not.

  7. Modern South Cushites and those who are admixed with them do not show as much mtDNA M/N as Horn groups, but have other signature lineages like L0f, which peaks in South Cushites but is also present in East African HGs. It presumably originated among the latter group, as it was also found in a Holocene Malawian HG, and eventually became frequent in South Cushites due to sex-biased gene flow and genetic drift. The Luxmanda (likely South Cushitic) 3100 YBP sample already shows signs of East African HG admixture.

    L0f is rather frequent in Hutus, as well as these Tutsis. In addition, the Tutsis tested privately so far apparently have 15% M/N, which is fairly significant. If anything their maternal lineages look more (non-Bantu) East African than the Y-DNA.

  8. Razib could you please add the Baganda and Burundi samples? I know from Hollfelder et al that Baganda have both Nilotic and Cushitic/Eurasian ancestry but no where near as significant as Rwandans, Kikuyus and Tanzanians which I guess is reflected phenotypically. Would like to see how West-African Baganda really are, would be dope especially since Uganda is so incredibly mtdna diverse.

  9. Is it conceptually wrong to imagine the XVI-XVIII c. populations movements in Eastern / SE Africa, including the documented spread of the Masai and of Luba societies, the expansion of the Zulus and their ancestors, and the hypothesized Tutsi conquest (if this is what shapes up from the DNA data) as a rough parallel of the domino-effects of the Iron Age migrations in the Eurasian Steppe, shifting the populations of the savanna belt in the general Southern direction in fits and starts? With the remnants of the past flow lingering longer in the highlands and woodlands surrounding the “main thoroughfare” of the migrations?

  10. 1) everyone asking about samples, the snp overlap is the reason i’m taking so long.

    2) to everyone asking about a hutu sample. i have one. lands right around the luhya from kenya. so that’s a good proxy

  11. I know I’m late but Luo/Luhya have always fascinated me cuz of the mix between Nilotic+West African [bantu?] Now of course in the Luhya they admixed with Nilotic populations…but in the case of the Luo a part of me thinks they were already admixed Nilotic+West African [mende] before coming into kenya/uganda and mixing even more so with Bantu peoples [Luyha etc]….to this day I study that old admixture analysis of yours where you separated Ks Bantu modal and Mande modal…whew…even Yorubas were a surprise when you did that but geographically makes perfect sense. Do you think you could see if the west-african in Luos is both Bantu and Mande?

Comments are closed.