Tutsis are less Nilotic than the Maasai

Last spring I was approached by some Tutsis in the Diaspora to analyze their genomes. The first thing that jumped out at me is how genetically similar Tutsis were to the Maasai.

Since that time readers have suggested other datasets. One particular dataset was very rich in Sudanese ethnic groups. Generating a new PCA you can see that the Tutsi above are now distinct from the Maasai. They are genetically similar, but there are subtle differences. From what I can tell, the major difference is that the Tutsi are less Nilotic and more Bantu that the Maasai.

The admixture plot sequence below should make that clear (also see the f3 stats at the bottom of this post):





8 thoughts on “Tutsis are less Nilotic than the Maasai

  1. They seem to be slightly more West Eurasian than Maasai too and I guess they show significant individual variation. Would be interesting to see if f.e. Kagame and his daughter show a genetic difference to the average or not. So how much admixture predicts the phenotypes.

    The same applies to the Hutu even more, if I think about the case of a leaders son being killed in the genocide because “he looked Tutsi”.

    Did you compare with Hutu and Twa and how the gene flow looks like between these populations?

  2. thanks for this razib, been waiting on it. very interesting, i did assume that the tutsi would be significantly more bantu than the maasai as they’ve had some sorta linguo genesis picking up bantu for what can only be assumed was a southern cushitic language…funnytht the tutsi would assume the language of the hutu even though they were in serfdom….

    what i find intersting is that at this [macro level?] how similar luhya and yoruba are even with the nilotic-omotic(do i see some cushite?) admixture in luhya. did you use the LWK sample? they are mostly bukusu who are known to be the most admixed luhya (some consider em not to even be true luhyas) so if there was a luhya proper sample, i’d bet they would be close to 95% niger-congo-esque…do u think you could ever get round to adding in the baganda sample from hollfelder and the basoga sample from here: https://www.biorxiv.org/content/10.1101/186700v2.full

    i still find it surprising how similar or ‘close’ east nigercongo speakers are with west-benue-congo speakers…that paper apparently has basoga not being too distant from ivorians…all with a pincha salt i know, but still interesting.

  3. @Obs
    In general, admixture has a significant impact on phenotype. The influence might vary from person to person though.
    Kagame and his daughter look Ethiopian or Somali, although in the genetic distance they would fall in between horners and Yoruba. Somalis got their look by 45% western Eurasian, but Kagame clan got the same look with 25%. The fact is genotype doesn’t equal phenotype.

  4. @Kevin: Of course, since we have random recombination and selection at work. The typical Tutsi is obviously a result of significant selection towards this phenotype. Recombine the same ancestral components by chance you get a different and much wider variation for sure. Tutsis would be an interesting case for checking for physical traits selection, especially elongated limbs and body height.

    What would be interesting, though, would be comparisons inside of a specific group of people like Tutsi. Obviously a Somali can look less Ethiopian with more WEA percentage than the more WEA looking Tutsi. But how about the same variation in phenotype within a set group of closely related individuals? How close correspond phenotypic percentages to “looks” within a closely related group which went through the same processes of drift and selection.

  5. The Masai are basically the Nilotic Kenya Iron Age pastoralists (~35-65 Nilotic/S. Cushitic) but have elevated Nilotic ancestry, whereas Tutsis and Great Lakes pastoralist ethnic groups can be modelled as Kenya Iron Age pastoralists + around a quarter Bantu/West African, which Great Lakes pastoralist ydna/mtdna reflects. The Pastoral Iron Age samples are around 35% Nilotic, and Masais are at the 50% area, so the Masai have a 15% increase in Nilotic ancestry, this could probably be explained by Maa Nilotes migrating from eastern Equatoria much later than the ancestors of the Kalenjin and Datooga (who represented some of the Iron Age groups) and concurrently with some Luo groups. Tutsis are somewhere from 10-20% Bantu/West African admixed.

    Tutsis and Banyamulenge (Congo) I’ve seen are 45-60% South Cushitic in ancestry, and 20 – 30% Nilotic/Nilo-Saharan and 10-20 Bantu/West African. The Nilotic-looking ancestry in Tutsis and Great Lakes pastoralist groups could be due to the assimilation of incoming Nilotic Luos, and or it can be from Central Sudanics in Uganda who have anciently inhabited much of northern Uganda and neighboring areas in Ituri province. Some of the Bantu-looking ancestry can be from West African admixture that is elevated in Central Sudanics. Great Lakes pastoralists have a history of intermarriage and assimilating the Nilotic but now heavily Bantu admixed Luos, as the early Luo families who migrated into the Great Lakes now look no different than Tutsis and Great Lakes pastoralist groups. And the many Luo subgroups in Uganda and Kenya all have significant Bantu/West African ancestry. All around Uganda from the Kenya border to the Rwenzori mountians and Ituri province, I see alot of interaction between Luos and Central Sudanics on one hand, and Great Lakes pastoralist groups on the other.

    Both Great Lakes pastoralists and Masais are around 40-50% E-M35, but Masais have alot of E-M78 such as E-V22 and E-V32. Tutsis are only E-M293 it seems (with a few having low res E-M35/E-M215 clades leading to E-M293). Both groups also have A-M13 and E2, but it is unknown if Great Lakes pastoralist A-M13 is of the Nilo-Saharan A-M13 variety or of the divergent Cushitic A-Y23655 A-M13 subclade. I think both, as the South Cushitic ancient samples had a few A-M13 as well.

    In general, Tutsis are significantly more Cushitic than Masais, and have significant Bantu/West African admixture and significantly less Nilotic ancestry than the Masai do. The Bantu and West African admixture in Tutsis and Great Lakes pastoralists to a large degree come from a Nilotic and or a Central Sudanic group in Uganda/NE Congo who the Great Lakes pastoralists ancestors assimilated, and can explain the combo of significant Nilotic-like + Bantu/West African. Groups known to have high West African or Bantu admixture but who are Nilotic and Central Sudanic do exist in Uganda, and most cluster like Luos from Kenya since they all can be modelled as Dinka + Bantu/West African. Some are more West African/Bantu admixed like Luos.

    Razib – a heads up, but I hope you’re also aware that there are several distinct groups like the Tutsi who span Congo, Uganda, Rwanda, Burundi, and Tanzania. They all are very similar genetically, but are distinct ethnicities that are wrongly conflated as all being Tutsi.


    Kagame and his daughter are very average for Tutsis and other Great Lakes pastoralists groups in Congo, Uganda, Tanzania, and Burundi. That tall, sharp featured phenotype is very common, and i’d say it’s the most common phenotype from the Tutsis I’ve seen irl. Amongst Cushitic-Bantu admixed groups like Kikuyus and Kamba, they look much more Ethiosemitic and Highlands Ethiopian-looking, whereas Nilotic-Cushitic mixed look more similar to Somalis. I’d say generally, the Tutsis and Masai I’ve seen had stern features like this Tutsi officer (https://www.africom.mil/Img/10998/Orig/rwandan-attending-us-air-force-academy-visits-africom) or these Somali clansmen (http://www.geeskaafrika.com/wp-content/uploads/2016/07/troops.jpg) or the first Somali defense minister (https://upload.wikimedia.org/wikipedia/commons/thumb/5/5f/Daud_Abdulle_Hirsi.jpg/270px-Daud_Abdulle_Hirsi.jpg).

    – I always wondered if later East Cushitic groups received any South Cushitic admixture, well here it is:

    “fit”: 1.3176,
    “Somali”: 71.67,
    “KEN_Pastoral_IA”: 15,
    “KEN_Pastoral_N_Elmenteitan”: 10.83,
    “ETH_4500BP”: 2.5,

    The Rendille are the closest linguistic relatives of the Somali, and they are around 25% South Cushitic, Rendille men are 19% E-M293.

  6. @Vytautus of Aukstaitija,

    The Somali data on G25 is a combination of Kenyan Somali (sampled by Ayodo in Garissa) and Somalia Somali data. The Kenyan Somali ones are admixed with either Rendille-like or with Borana Oromo-like populations (not fully ethnic Somali) and change the Somali average. If only Somalia Somalis (those from Pagani/Hodgson) are used likely that calculation of yours will change.

  7. Big up Razib!
    Next session, please touch on Cushitic and/or Eurasian admixture. Analyze which Eurasian population(s) Tutsis owe their Eurasian admixture. I would be surprised if there is anything other than Natufians. Also if the cushitic admixture is South Cushitic( Iraqw or other ancient south Cushites of the Pastoral Neolithic time) east or Northern Cushitic.

  8. @Vytautus of Aukstaitija
    Thanks for your take. I don’t know where people got this thing of saying Kagame’s features are an exception in Tutsis. That tall slender with sharp features are the most common phenotypes in Banyamulenge and Tutsis in general. I think our features can be misleading. Phenotypically we are far closer to cushitic groups than we are genetically.

Comments are closed.