This page was first published on 15 February 2015. Please keep in mind that 23andme’s Ancestry Composition has been updated several times now! On this page I am dealing exclusively with results being reported in the period 2013-2018. All matters being discussed on this page are therefore not pertaining to updated results (2018 up till now). For more details see:
- 23andme’s new African breakdown put to the test (Feb. 2019)
- East African 23andme results (after 2018/2019 update)
Knowing which African populations score the highest for each of the 23andme “Sub-Saharan African” (SSA) categories can be a good indication for their predictive ability. Below screenshots are all taken from people who have kindly agreed to share their results with me. For which i am very grateful! They were either born in the African country highlighted or have both parents from that country. These are obviously first of all individual results and very limited in number because there’s only very few Africans yet who have tested with 23andme. I’m posting them for illustrative purposes, mainly to get a very rough idea what to expect. Undoubtedly with more African 23andme test results you might see different or additional patterns. Still i think in most cases these screenshots below would be representative to some degree for how other people from their nationality or ethnic group would score hypothetically speaking. I will provide a brief overview of the main patterns i’m able to pick up on. Of course it merely shows my personal opinions & thoughts and is not meant to be conclusive in any way 😉
It will be insightful to also compare with the AncestryDNA results of East Africans which can be seen via this page:
p.s. I’m only showing screenshots of the African breakdown. You’ll notice it will often not add up to 100%. In most cases this is because of a well known “bug” in the current version of Ancestry Composition causing people of 100% “Sub-Saharan African” (SSA) descent to show trace levels of non-SSA admixture or “unassigned” ancestry, this can generally be considered “noise”, i.e. reflecting an artefact of the DNA test. Hopefully it will be fixed with the next update. In some other cases though the individuals will have genuine additional non-SSA ancestry, which might however be “native” to Africa still if it’s labeled as “North African”, otherwise it might reflect historical geneflow from outside of Africa within the last 500 years or even earlier.
- the “East African” category on 23andme or more properly said the “Northeast African” category probably peaks among Somali. Most of my 7 Somali examples being nearly 100% “East African”. Interestingly the last two being exceptions. Most likely because of some additional minor ancestry from the Middle East and South Asia dating from within the last 500 years. The “West African” is still minimal or even absent. This peaking of the “East African” category among Somali is likely caused by “overfitting” or a “calculator effect” because 12 Somali customers from the 23andme database were used as samples (see screenshot above).
- Otherwise it is Ethiopians who seem to score most consistently for “East Africa”, they were also among the samples being used. However unlike the 7 Somali results they are also shown as having a significant “North African” component. This should not be taken as suggestive of recent ancestry from that area but rather indicative of their DNA markers in part being similar to the ones for the samples being used to define the North African category.
- The most notable difference between the South Sudanese sample and the northern Sudanese is precisely this socalled “North African” component. Besides also a different degree of “West African”. Their “East African” percentages seem to be in a similar range.
- All the Ethiopians, the Eritrean and also the Somali hardly show any “West African” % beyond trace level. This seems to be a robust sign of mutual exclusivity implying that the “East Africa” category seems to be very predictive of specifically Northeast African ancestral origins or affinity because there’s no overlap with the “West African” category.
- Most of the other East African examples below however do show a significant “West African” component. Again this should not be taken as recent ancestry from that area but rather indicative of their DNA markers in part being similar to the ones for the samples being used to define the “West African” category, i.e. “Niger-Congo-like”. As discussed on the previous page Kenyan Bantu samples were used among others to define the “West African” category so it’s not that surprising that this similarity is showing up.
- Among Rwandans and Kenyans there are some high outliers, higher even than the Ethiopian ones! The ones for Rwanda approaching the Somali results of over 90%. But there’s also some lower “East African” percentages for especially the Kenyans, suggesting a greater variation according to ethnic background. Tanzanians also show considerable scores, with the Sudan and Ugandan results somewhere in between. I have some results from Mozambique and Zimbabwe as well and they show either zero or very miniscule “East African” %’s of below 1%. For two Madagascar results it’s also very minor but still at detectable level of in between 1-3% (see “Central & South Africa” for their screenshots). Indicative of this ancestral cluster rapidly decreasing when going south.
- Again much of the variation will also be depending on ethnic background and deep ancestry aside from just nationality or geography. Nilotic speaking groups (like the Maasai samples) or Cushitic speaking groups (like the Somali) being more likely to score highest for “East Africa”. This might actually also include a few ethnic groups who might have undergone a language shift and are now speaking a Bantu or Afro-Asiatic (Amhara etc.) language instead. Or Bantu/Afro- Asiatic speaking etnicities who absorbed formerly Nilotic/Cushtic speakers within their ranks through intermarriage. It is often underestimated IMO how ancient & complex histories of ethnogenesis are still able to impact on admixture tests that are supposedly in 23andme’s case looking back only “500 years”.
**Highest scores among East Africans**
Kenya 6 (Luo)
**Highest scores outside East Africa**
- Outside of East Africa proper from what i’ve personally observed this category only seems to appear among selected ethnicities, i.e. South African Coloureds and West African Fula, or West Africans with likely partial Fula ancestry (posted near bottom of page). For one individual of confirmed Fula background “East African” reaches almost 20%! For the others it’s much more reduced but still detectable. It’s important to note that this doesn’t imply any recent ancestral connections with Northeast Africa, it’s rather pointing towards genetical similarity/affinity and possibly very ancient shared ancestry with presentday Northeast Africans but not per se originating from that area. It might be testament to very ancient Nilo-Saharan connections across the interior of West to East Africa, obscured nowadays because of language shift and ethnogenesis. Specifically for the South African Coloured results it might be an intriguing testimony of an ancient link between the Khoisan and Northeast Africa, see also this paper from 2012, and this one from 2014.
Guinea Conakry (confirmed Fula)
Fula (unknown country)
Senegal (possibly Fula or partially so)
Gambia (possibly Fula or partially so)
Nigeria (possibly Northern)
Nigeria (Hausa & Fula)
South Africa (Coloured)
**Highest scores among Afro-Diaspora**
- Among the hundreds results i’ve personally seen for Afro-descended individuals (both among my sharing list and also posted on various online forums) practically all score “East African” below trace level of 1%. The highest score i have seen myself being 1,6%, which is also the only one above 1% (see screenshots below). For most Afro-Diasporans it seems “East African” is simply inexistent and only for a few it’s in between 0,5%-1%. Obviously this would represent a very minor part of their overall African origins but it can still provide a valuable indication in some cases for identifying 1 single African ancestor.
- This outcome being in line with documented slave trade to the Americas not involving Northeast Africa but reaching its outer limit in Southeast Africa or Mozambique. So in that sense it could be seen as a confirmation for Afro-Diasporans generally not having any Northeast African origins. Of course individual exceptions should not be ruled out at this preliminary stage but the odds seem to be very small indeed. A small % of “East African” is more likely to be indicative of ethnic West African roots instead for most Afro-diasporeans. The very dilluted “East African” ancestral component being inherited from specifically the Fula who are the only West Africans i’ve personally seen to score minor yet very detectable levels of “East African”. Scenario’s involving other ethnicities from Southeast Africa also still being possible though.
- A notable exception however being Afro-descendants located in the Middle East and the Indian Ocean islands. Unlike the Trans Atlantic Afro-diasporans, these people often will have substantial East African origins resulting from the socalled Indian Ocean slave trade and/or the Trans Saharan slave trade carried out mostly by Muslims. This slave trading circuit was largely separate from the Trans Atlantic one and is known to also have targeted areas within Northeast Africa. The two very last screenshots posted below are testament to that legacy, showing “East African” admixture levels which seem to be very distinct and elevated compared with Trans Atlantic Afro-diasporeans. Making this a recognizable “tell tale” marker in a way to distinguish both groups of Afro-diasporeans. Much of their socalled “West African” component might actually also hail from Bantu speaking areas within East Africa. Hopefully a future update will bring more clarity.
United Arab Emirates