New Update on 23andme: Ethnic Group Matches within Africa! (part 2)

This is the second part of my blogseries which deals with the update last month on 23andme. Introducing no less than 25 ethno-linguistic groups from Africa to get matched with! In order to learn more about the ethnic origins of some of your African ancestors. In the first part of this series I already covered the question of how accurate these ethnic group matches might be. Follow link below for more details:

• New Update on 23andme: Ethnic Group Matches within Africa! (part 1)

In this post I will be focusing on what to expect. Basing myself mostly on the survey findings for 100 African Americans and 50 Cape Verdeans. But I will also post screenshots of updated results from other parts of the Afro-Diaspora.

***

Figure 1 (click to enlarge)

These are the updated results of a Cape Verdean cousin of mine. It was already very useful how 23andme is able to pinpoint recent ancestry from Cape Verde. However with these new ethnic group matches also historically plausible origins from mainland Africa are confirmed for Cape Verdeans. Although to be sure there are many other ethnic groups from across Upper Guinea which also contributed to Cape Verde’s African Roots (see this website).

***

Just to quickly repeat myself. Last month 23andme released its most recent update which could very well signal the start of a truly game-changing feature for zooming into ethno-linguistic African lineage¹. Of course this new feature on 23andme, based on matching strength, is not perfectly flawless. However despite a few shortcomings I am very excited about this update on 23andme! Because based on more than 50 updated results for Africans (see this overview) I am quite impressed with the accuracy of this tool. Also the results I have seen for 50 Cape Verdeans and 100 African Americans are actually in line with historical plausibility.

This update has already been fully rolled out.² But still the odds of obtaining even only one of the 25 ethno-linguistic groups are rather low for most people. Depending also on your background. It seems for African Americans there’s only about a 20% chance. For many people 23andme is simply not able to detect a sufficient level of matching strength. Which is inevitable given the limitations of 23andme’s current African reference database. Understandably this may come as a big disappointement. However don’t despair because this update is a work in progress and upcoming updates will eventually reach more people! Glass half full mentality also for 2022 😉

***

Figure 2 (click to enlarge)

Take note that the updated results on the right only feature one ethnic group match. Despite this African American woman being over 80% African and regionally speaking from atleast 4 major areas in West & Central Africa. Of course this finding in itself is already incredibly valuable. However this possibly Fula and/or Wolof lineage most likely covers less than 5% of her total ancestry. While for her primary region “Nigerian” not a single ethnic group has appeared yet. In fact, although not being detected she might still also have additional Mandinka ancestors. To be grouped under “Senegambian & Guinean”. Follow-up research is still needed if you want to uncover your entire African lineage!   

***

In the remaining part of this blog post I will discuss the following:

1. Survey findings for 100 African Americans
   • “Igbo”, “Kongo & Mbundu” and “Fula & Wolof” most frequent
2. Survey findings for 50 Cape Verdeans
   • “Fula & Wolof” and “Mandinka” most frequent
3. Updated results from other parts of the Afro-Diaspora 
   • Atlantic Afro-descendants (Brazil, Dominican Republic, Jamaica, Haiti etc.)
   • Indian Oceanic Afro-descendants (Mauritius/Réunion, Seychelles, South African Coloured, UAE)
4. Considerations
   • Recap previous disclaimers
   • Some ethnic group matches appear more frequently than others
   • Each regional admixture category potentially includes ancestors from various ethnic groups

***

Map 1 (click to enlarge)

Source: 23andme. 23andMe’s latest update provides new ancestral connections to 25 African ethnolinguistic groups, or groups of people who share a common language and culture. They form a subset of the already existing Recent Ancestor Locations. See also this overview for a full list of the 200+ Ancestry Composition populations in Africa.

1) Survey findings for 100 African Americans

Table 1.1 (click to enlarge) 

This overview is showing the outcomes for 100 African American updated results which were kindly shared with me. Merely the start of investigating which types of African ethnic origin could be most prevalent among African Americans. But already corresponding with historical expectations! 

 

____________________

“[…] However I strongly suspect that the frequency of DNA matches from a certain place/ethnic group may not always correlate with autosomal contribution, proportionally speaking. In other words just because my Fula and Malagasy survey participants seem to be extra “matchy” with African Americans does not right away imply that the Fula & the Malagasy represent the biggest ancestral components for African Americans (to be verified by admixture analysis).

For the Igbo such a case could arguably already be made. Based on both historical and genetic evidence. But especially for the Malagasy a more subdued & diluted overall ancestral share is to be expected (even if still significant and clearly detectable). Certainly lower than for example the Congolese or Angolan input in the average African breakdown for African Americans.” (Fonte Felipe, 2018)

____________________

This section features the preliminary findings of my ongoing research into African DNA matching patterns for African Americans. I actually already started doing this type of research in 2015 when 23andme’s former Countries of Ancestry (CoA) tool was still available. See also quotes above and this blogpost. The high frequency of especially Igbo and Fula matches is consistent up till now.

One of the main differences is the much greater frequency of Central African matching (“Kongo & Mbundu”) with this new update. Which certainly is in line with historical plausibility! Regrettably 23andme does not yet feature any Malagasy reference population. Because of sharing profiles with Malagasy customers I was however already able to establish in 2015 that African Americans tend to have a considerable matching strength with Madagascar as well.

Despite limitations of my samplesize it is also interesting to compare with 23andme’s findings. These were featured in their announcement of this update. Although it should be stressed that 23andme’s data is based on all 23andme customers with a certain degree of African DNA. Hence it is also including data of African, Latin American and West Indian migrants. As well as people who are not multi-generational African Americans (on all lines). This was actually also an issue for 23andme’s study from 2020 (see this review, footnotes 2 & 13). I do think the data will mostly cover African Americans though as they are testing in greatest numbers. And in fact the top 3 ranking order of my own findings corresponds quite nicely with those from 23andme. More discussion in section 4. See also:

• Most common ethnolinguistic group connections among 23andMe customers (23andme, 2022)

Just as a heads up: this year I intend to publish a new series on the African DNA matches being reported by Ancestry for survey groups across the Diaspora. I collected the data already in August 2020 when Ancestry deleted the smaller matches (<8 cM) (see this blogpost). It’s still a lot of work to sort it all out. However hopefully in the next months I will get things started. First post will be dealing with the African DNA matches for Anglo-Caribbeans as well as a few Surinamese actually. Followed by separate posts for Haitians, Brazilians and Hispanics. The last post in that series will cover my findings for 50 African American survey participants. Which should be quite interesting as I will also attempt to take into account the state origins of my survey participants as well as their genetic communities.

• African DNA matches reported by Ancestry for 50 African Americans (under preparation)

For illustrationial purposes I will post the 21 updated results for the African Americans in my survey below. Lots of variation on display. Take note especially that there is (hardly) any correlation with getting matched with any particular ethnic group and the absolute amount of associated regional admixture! This makes sense when you keep in mind that this update is about matching strength with 23andme’s database of African reference populations. This database will inherently be faced with limitations. But hopefully 23andme will expand on it soon in future updates!

On a final note Table 1.1. above also shows that only 21 out of 100 of my African American survey participants received any ethnic group match (a few received more than 1). Hence the odds of receiving this update (~20%) are not that high. But still it is already a clear improvement when compared with the odds of receiving a match for an African country by way of 23andme’s Recent Ancestor Location tool. Based on a previous survey of mine this was only 3% for African Americans (see this overview). Among the 21 results below only 1 person received an additional match for a country (Nigeria). 

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

Two ethnic group matches grouped under the same regional admixture category. This is quite rare from what I have seen. However realistically speaking of course most people might have several ethnic origins conflated in one regional admixture category. In fact the “Senegambian & Guinean” ancestors of this person could even be of other additional ethnic origins as well. Such as Sereer, Balanta, Jola etc.. Just because these groups are undetected by 23andme’s database doesn’t mean it couldn’t be there!

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN (Texas 4gp)

***

Remarkable to see Nigeria showing up as country match but not ethnically specified. I suppose this suggests that 23andme’s reference samples for Nigeria as a Recent Ancestor Location are different or only partially overlapping with the ethnic group categories for “Igbo”, Yoruba”, “Edo & Ijaw” etc..

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN

***

***

AFRICAN AMERICAN 

***

This person has one of the highest “Angolan & Congolese” scores I have seen for African Americans. And it has been appropriately specified by 23andme! This person will also be featured in my upcoming surveyfindings for African DNA matches on Ancestry. Ironically none of his 25 African matches on Ancestry were Central African. The majority being Nigerian/Igbo (see this overview). Remarkably not detected by 23andme. Which goes to show how the composition of the database for each company (incl. also customers) can cause disproportional outcomes when confirming African lineage. The addition of Central African samples by 23andme therefore is a huge boost!

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

AFRICAN AMERICAN 

***

***

 

2) Survey findings for 50 Cape Verdeans

Table 2.1 (click to enlarge)

This overview is showing the outcomes for 50 Cape Verdean updated results which were kindly shared with me. Certainly in line with historical expectations and hence job well done by 23andme! However do keep in mind that Upper Guinea has many more ethnic groups besides Fula, Mandinga and Wolof. All of them potentially ancestral to Cape Verdeans as well.

***

____________________

“The high number of Fula matches is quite striking. But this could very well reflect a greater popularity of DNA testing among Fula people when compared with people from for example Guiné Bissau who are greatly underrepresented in Ancestry’s customer database.” 

“We should then however make allowance for the possibility that especially many of these >8cM matches may be indicative of relatively late African lineage from the 1700’s & 1800’s.” […]

“If indeed many of these bigger matches are to be traced rather to the 1800’s or late 1700’s this could be very helpful actually. As it could increase the odds of zooming into particular family lines whose generational distance with mainland African MRCA’s might be relatively small. On the other hand it might obscure the ethnic origins of African ancestors to be traced back to the earliest period (1500’s/1600’s).”(Fonte Felipe, 2018)

____________________

This section features the findings of my ongoing research into African DNA matching patterns for Cape Verdeans. I actually already started doing this type of research nearly 10 years ago when 23andme’s former Countries of Ancestry (CoA) tool was still available. See also last section of this webpage. Back then it was still very uncommon to receive any mainland African DNA matches. Really incredible how much progress has been made eventually!

In 2018 I published a comprehensive blogpost on African DNA matches reported for 50 Cape Verdeans on AncestryDNA. It is therefore very useful to compare with my previous surveyfindings on Ancestry. See also quotations above. I have a hunch that the “Fula and Wolof” category on 23andme is mostly due to matching with Fula samples. Hence the high frequency of Fula matches is consistent with my older Ancestry findings. And therefore not that surprising. However I find it very evocative to also see “Mandinga” lineage being confirmed.³ Because due to a lack of actual Guinean ancestry testers (non-Fula) such DNA matches are still quite rare on Ancestry. See also:

• DNA matches reported for 50 Cape Verdeans on AncestryDNA (2018)

____________________

“TOP 3 BREAKDOWN OF AFRICAN BORN SLAVES IN 1856

Mandinga (Upper Guinea) 34 – 26% of African specified
Fula (Upper Guinea) 19 – 15% of African specified
Bijago (Guiné Bissau) 18 – 14% of African specified”

“Because of their history of early creolization and sharply decreasing enslavement since the late 1600’s most Cape Verdeans will however have to trace back much further than 1856 to find African born ancestors whose ethnic composition is naturally going to show overlap with the data shown above but proportionally things could very well be different.” (Fonte Felipe, 2015)

____________________

The implied time frame of these ethnic group matches on 23andme is: “within the last 250 to 300 years.”  Which is why it should also be useful to compare with ethnic origins mentioned in Cape Verde’s Slave Census of 1856. Although at that time actually the vast majority of Cape Verdeans had been free of slavery for many generations. Also being locally born and “native” to Cape Verde for many generations.

Still very striking to see Mandinga and Fula appear in the top 3! It could very well be however that in previous centuries, especially the 1500/1600’s, the ethnic composition of mainland Africans in Cape Verde would have been quite different. Showing moreso a shift to the coastal ethnic groups of Guiné Bissau. And the earliest nucleus of Cape Verdean population (1400’s/1500’s) most likely consisted of a majority Wolof/Sereer background.

Of course this update by 23andme is already very valuable for confirming especially Fula and Mandinga lineage. However do not forget that according to Cape Verde’s well respected historian António Carreira there are at least 27 historically attested African ethnic groups for Cape Verde! This historical evidence is mostly dating from the 1500’s/1600’s and overwhelmingly referring to ethnic groups from all over Upper Guinea. Not just the area of “Senegambia & Guinea”, but also extending into northern Sierra Leone in fact!

In this last regard it is regrettable that sofar I have not seen any ethnic group matches yet for Sierra Leone. Aside from possible database issues this might also merely reflect that these connections go back further in time than the 300 years implied by 23andme’s new feature. Going by regional admixture Cape Verdeans are almost exclusively “Senegambian & Guinean”, on their African side. Especially after the 2020 update. However in 2019 “Ghanaian, Liberian & Sierra Leonean” scores were still quite substantial at around 10% (scaled). And also consistently appearing (see this page). Most likely describing an additional part of Upper Guinean lineage (incl. Temne) for Cape Verdeans. 

On a final note Table 2.1. above also shows that 24 out of 50 of my Cape Verdean survey participants received an ethnic group match (a few received more than 1). Hence the odds of receiving this update (~50%) are quite good. Especially when compared with African Americans for whom there only seems to be a 20% chance from my findings. This outcome, although again preliminary, is still interesting. As it seems to suggest that mainland African lineage for Cape Verdeans is at times still quite recent (1700’s-1800’s).

To be kept in mind is that this does not say anything per se about the overall proportion of this “Fula & Wolof” or “Mandinka” lineage. Further research is needed to establish how much African lineage for Cape Verdeans is tracing back to earlier centuries and also from other ethnic groups (such as Papel, Biafada, Jola etc.). Because right now such lineage is undetected by 23andme’s matching algorithm and not covered by 23andme’s database. See also:

• Cabo Verde Raizes Na Africa: Top 20 Ethnic Roots (2014)
• Cape Verde Slave Census of 1856 (part 1) (2015)
• Cape Verde Slave Census of 1856 (part 2) (2015)
• From Creole to African (2015)

 

***

CABO VERDE (Santiago)

***

Splendid specification of “Senegambian & Guinean”. Also the pinpointing of Cape Verde is correct. However notice that island specification (Brava) is still misleading. As it is also for most people below who do not have any actual Brava origins. This may serve as a warning sign that also the ethnic group matches are not always to be taken too literally. In fact when you read the description of “Mandinka” it says: “Common reported ethnicities or languages spoken within this genetic group: Mandinka, Bassa, Soninke“. And probably other additional ones as well.

***

CABO VERDE 

***

***

CABO VERDE (Santo Antão)

***

***

CABO VERDE 

***

***

CABO VERDE 

***

***

CABO VERDE 

***

Two ethnic group matches grouped under the same regional admixture category. This is quite rare from what I have seen. However realistically speaking practically all Cape Verdeans should have both Mandinga and Fula/Wolof lineage, due to endogamy and shared African roots. In fact the “Senegambian & Guinean” ancestors of this person and other Cape Verdeans should also include other additional ethnic origins as well. Such as Sereer, Balanta, Jola etc.. Just because these groups are undetected or not covered yet by 23andme’s database doesn’t mean it couldn’t be there!

***

CABO VERDE (Fogo)

***

***

CABO VERDE 

***

***

CABO VERDE 

***

***

CABO VERDE (1/2) & Nigeria (1/2: Igbo) 

***

Amazing how 23andme got this right! This person has one Nigerian parent (confirmed Igbo) and also 1 Cape Verdean parent. Notice that all this info is captured in the results above, incl. the country locations. However the additional “Mandinka” lineage was not known yet as it most likely traces back to at least the early 1800’s.

***

 

3) Updated results from other parts of the Afro-Diaspora

I intend to collect more results from other parts of the Afro-Diaspora but this small selection should already be useful to illustrate some of the main patterns. More analysis in upcoming blogposts. But one thing I find quite interesting is how these results are often highlighting the shared African ethnic roots across the Diaspora. Even when the actual proportional shares might still be more differentiated. But from historical sources it was of course already known that for example Central African heritage is widespread across the Americas. And certainly not exclusive to any particular country or area. Although most likely when based on bigger samplesize it will turn out that for example “Kongo & Mbundu” will be more frequently seen for certain nationalities than others.

Just to quickly share some preliminary observations:

• Sofar I have hardly seen any “Yoruba” matches yet. Also not for Brazilians from Bahia for whom recent and substantial ancestral ties to Yorubaland are a given.
• Sofar I have hardly seen any ethnic specification for the “Ghanaian, Liberian and Sierra Leonean” category. Also not for Anglo-Caribbeans and Surinamese for whom this category is often prevailing.

***

HAITI

***

***

HAITI 

***

***

HAITI 

***

***

HAITI 

***

The previous 3 Haitian results may have conveyed the impression that “Kongo & Mbundu” results are quite frequent among Haitians. Which is most likely correct. Serving as a corrroboration of regional admixture patterns and also historical plausibility. However as this outcome shows the African roots of Haitians are multi-ethnic and multi regional. And so other ethnic group matches are also very much to be expected! Even if they are associated with relatively minor regional admixture categories.

***

HAITI 

***

Very special outcome! Only time I have seen a Cameroonian match sofar for someone from the Afro-Diaspora. Take note this ethnic group match is grouped under “Nigerian”. Somewhat confusing at first sight perhaps. But historically speaking it does make sense because it is most likely referring to Bight of Biafra origins, which also includes Igbo lineage. Which should also be quite considerable for Haitians. 

***

JAMAICA

***

No ethnic group matches detected for this person. Despite being nearly 100% African! From what I have seen sofar the odds of receiving this update are quite low for West Indians, same as for African Americans. Regrettably also not including any specification of “Ghanaian, Liberian and Sierra Leonean”. Even when this person shows a remarkably high level of this category! Interestingly I am also sharing with this person on Ancestry. And therefore she was already featured (JAM09) in my Jamaican survey findings for African DNA matches on Ancestry. She received 31 African matches on Ancestry, corroborating lineage from Nigeria, Ghana and Benin (see this overview). Which goes to show that it’s always best to cast your net wide and search among several pools of potential African DNA matches!

***

JAMAICA 

***

***

JAMAICA 

***

***

JAMAICA (1/2) & GAMBIA (1/2) 

***

Very interesting results. The “Mandinka” match is obviously reflecting this person’s recent Gambian side. Regular Jamaicans may also have additional Gambian lineage. But usually it will be quite subdued. As measured by “Senegambian & Guinean”. I suspect that such “Mandinka” matches will also not be very frequent. Although I did already see “Fula  Wolof” appear for some Anglo-Caribbeans.

***

BARBADOS

***

***

BRAZIL (Minas Gerais)

***

***

BRAZIL (Minas Gerais)

***

***

BRAZIL (3/4: Rio de Janeiro) & CAPE VERDE (1/4)

***

Phenomenal results! Very special to see this person’s Southeast African lineage being confirmed as well. In this case so-called “Shona & Nguni” is most likely due to actual Mozambican ancestry. Do notice as well how this person’s partial Cape Verdean side is specified with “Fula and Wolof”. Something which I suspect will be quite rare for regular Brazilians, safe for northern Brazilians (Maranhão, Amazonas, Pará etc.)

 

***

BRAZIL (Rio de Janeiro)

***

***

DOMINICAN REPUBLIC 

***

***

DOMINICAN REPUBLIC 

***

***

COLOMBIA

***

 

 

Indian Oceanic Afro-Diaspora

____________________

” Looking into their results as well is greatly beneficial for evaluating the overall coherence of 23andme’s African breakdown in my opinion. Plus it also enables a broader conceptualization of the Afro-Diaspora. Trans-Atlantic and Indian Ocean slave trading circuits overlapping precisely in Southeast Africa.” (Fonte Felipe, 2021)

____________________

 

***

MAURITIUS or RÉUNION (Creole)

***

Never mind the possibly misleading labeling of both regional admixture and ethnic group match. The main idea is clear enough: Bantu origins from Southeast Africa. Which is of course perfectly in line with historical plausibility for the Indian Oceanic Diaspora!. In this case so-called “Shona & Nguni” is most likely due to actual Mozambican ancestry. See also: The Mozambique connection on Ancestry & MyHeritage (2019)

 

***

SEYCHELLES (Creole)

***

Again “Shona & Ngoni” appearing. Which is very likely to be the most frequent ethnic group match from mainland Africa for Indian Oceanic Creoles. Keep in mind though that currently 23andme does not yet have a separate category for ethnic Malagasy lineage in place. Which should actually also be greatly beneficial for African Americans! However 23andme does already have a recent ancestor location for Madagascar. Which is also showing up for this person!

***

SOUTH AFRICA (Coloured)

***

Quite surprising! The mainland African roots of South African Coloureds is actually more so from Mozambique as well as local Khoisan. However early slave trade between Angola and the Dutch Cape Colony has been documented in fact! See this article.

***

UNITED ARAB EMIRATES (4gp)

***

Amazingly 3 ethnic group matches! This makes perfect sense given the scope of the Indian Oceanic Slave trade: Eastern Africa and into Central Africa as well due to relatively late (1800’s) Swahili slave trading. This person is multi-generationally Gulf-Arab. At least going back to his 4 grandparents being born in the United Arab Emirates. Which is correctly pinpointed by the recent ancestor location for Abu Dhabi as well.

***

 

4) Considerations

____________________

“Since cultural identity is so much more than DNA, though, keep in mind that these results alone do not say that a customer does belong to an ethnic group today. ” (23andme, 2022)

“When a customer receives a match, it means that they definitely share some of their ancestors with that population within the last 250 to 300 years. And because of the populations’ roots in these areas for generations, those ancestors likely lived in the same area and shared their communities’ cultures. ” (23andme, 2022)

“As always it will be important to also independently verify any claims made by 23andme. Contrasting with historical plausibility as well any other relevant factors. Remaining critical but also open-minded and careful not to be dismissive when informational value can still be obtained! Afterall it is not just about the results but also about the correct interpretation of the results.” (Fonte Felipe 2021)

“When you click on your ethnic match group don’t get lost in the excitement right away. But also make sure to read the portion which says: Common reported ethnicities or languages spoken within this genetic group:..” (Fonte Felipe 2022)

” I would also encourage everyone to have a good look at the additional maps for grasping the broader implications. Which often go beyond one particular ethnic group in fact. The incorporation of such maps (similar as on Ancestry) is something I have been making a plea for a long time already (see this blogpost). I am relieved that apparently 23andme has now taken the cue. To be sure I did already detect some errors in the ethnic description box/maps.^5″ (Fonte Felipe 2022)

____________________

Short recap of disclaimers

1. Not intended to be conclusive! Ethnic identities are fluid and contextdependent.
2. Ethnic group matches can be considered as a subcategory of Recent Ancestor Locations.
   • Common Questions (23andme)
   • Recent Ancestor Locations: Pinpointing African lineage and Cross-Diaspora connections (scroll down to section 4)
3. Keep in mind implied timeframe.
   • From African to Creole (2015)
   • Why even small African matches matter! (2020)
   • Trans-Atlantic & Intra-American Slave Trade Patterns (Slave Voyages Database)
4. Focus on complementarity of DNA matches and regional admixture
   • DNA matches reported for 50 Cape Verdeans on AncestryDNA (2018)
   • African DNA matches reported for 30 Jamaicans on Ancestry (2020)
5. Educate yourself on historically plausible ethnic origins
   • Africa
     • African Maps (Ethno-Linguistic, Slave Trade, Lower Guinea, Upper Guinea, Sahel/Interior, West-Central Africa, Southeast Africa)
     • General History of Africa by Unesco (all volumes online available)
   • Afro-Diaspora
     • Ethnic/Regional Origins of the Afro-Diaspora (Africa (incl. Cape Verde and Sierra Leone), Anglo-Caribbean, Brazil, Dutch-Caribbean, Surinam & South Africa, Franco-Caribbean, Hispanic Americas, USA)
   • DNA testing
     • 23andme’s new African breakdown put to the test (2019)
     • Are African Americans really mostly “Nigerian”? (review of Micheletti et al., 2020)
     • 23andme’s African breakdown put to the test: Afro Diaspora edition! (2021)

For a more detailed discussion of these disclaimers see section 1 of my previous blogpost in this series:

• New Update on 23andme: Ethnic Group Matches within Africa! (part 1)

 

Higher odds for certain ethnic group matches?

Figure 4.1 (click to enlarge) 

Source: 23andme. Most common ethnolinguistic group connections among 23andMe customers. Keep in mind that therefore this is also including data of African, Latin American and West Indian migrants. As well as people who are not multi-generational African Americans (on all lines). Still the top 3 especially should be reflective of patterns among African Americans firstmost. Compare also with my own findings (see Table 1.1)

____________________

“Of 23andMe customers who receive a match to a new group, the most common matches are with the Igbo people and Kongo and Mbundu peoples. Both of these groups are matches for around 30 percent of customers. The next most common group is the Fula and Wolof peoples, which is a match for around 13 percent of customers.

These proportions likely reflect the impact of the transatlantic slave trade in which the majority of the 12.5 million enslaved Africans were taken from ethnolinguistic groups residing in the mid to southern Atlantic coast of Africa and forced into the Americas. Each of the remaining ethnolinguistic groups are matches for fewer than 5 percent of customers.” (23andme, 2022)

____________________

I find it very interesting that some of 23andme’s ethnic group categories appear more frequently than others. Undoubtedly this is correlated with historical plausibility. My Cape Verdean survey results in section 2 make a perfect illustration. But I also find it quite intriguing for example that “Igbo” and “Kongo & Mbundu” are much more frequent than “Bamileke and Kom” from Cameroon. This seems to vindicate my earlier blogposts about the issue of how common true Cameroonian lineage might really be. See also:

• The Igbo Connection for Virginia & Virginia-Descendants (2015)
• “Cameroon/Congo” = moreso Angola/Congo for Diasporans? (2017)
• Degree of Cameroonian lineage among African Americans (2019)

However otherwise I find that it is too soon to jump to conclusions! I intend to revisit this theme in future blogposts when I will follow-up with more details on the implications of these ethnic group matches within Africa. In fact I might also edit this blogpost itself with more information in the upcoming weeks (especially footnotes). So you might want to check that out too by revisiting this blogpost.

Just to expand somewhat on my previous considerations I would like point out the following aspects:

1. The data-collection performed by 23andme covers all of their customers with African DNA, regardless of actual background. Which does not allow for straightforward interpretation. However especially the top 3 should be strongly reflective of patterns among African Americans. Because my survey findings in section 1 are greatly consistent.
2. Low matching strength with groups from Ghana, Liberia and Sierra Leone is conspicious. In fact from what I have seen also “Yoruba” matches are much less frequent than expected for groups such as Brazilians and Haitians.
3. A crucial aspect concerns the sample size of each of the 25 ethno-linguistic groups, featured in this new update. As this might impact the odds of getting matched in the first place. But possibly also explaining why some of these groups seem to be more frequent among Afro-Diasporans than others.
4. Any over-reliance on DNA matches might lead to a disproportional outlook on your complete ancestry. Some ancestral lines being more “matchy” than others as a result of an ethnically skewed database (either customers or reference samples). And this may then be corrected by your admixture results. Which may be imprecise but still should enable you to (roughly) distinguish between major and minor sources of ancestral origins (both continental and subcontinental!).
5. The genetic impact of Trans-Atlantic Slave Trade should not be easily generalized! This impact is in fact highly differentiated according to places of destinations within the Americas (and also within the USA). And also according to timeperiod (see this overview). Furthermore domestic slave trade, inter-colonial slave trade and post-slavery migrations are also part of the equation. Despite their well-intentioned study of last year 23andme’s researchteam regrettably did not fully account for all these factors. Which resulted in questionable historical reasoning (see this review).

Regional admixture can include ancestors from various ethnic groups, all at the same time!

Figure 4.2 (click to enlarge) 

Source 23andme (tab “Scientific Details” in Ancestry Composition). Overview shown for an African American who received ethnic group matches for both “Igbo people” and “Kongo & Mbundu”. However just because 23andme leaves other ethnic groups undetected does not rule out that for example this person could also have additional Ijaw lineage from Nigeria or Luba from DRC Congo. 

***

____________________

“Nigeria” is not an exclusive marker for any given ethnic group “

Your “Nigeria” amount is likely to be traced back to numerous family lines and not a single one (unless you happen to have relatively recent Nigerian ancestry). Just as an example: a 25% score “Nigeria” for a typical Jamaican might be due to the genetic contributions of in between 15 to 50 different African-born ancestors. On average the DNA contribution of an ancestor living in the mid 1700’s could be around 0.5%-1.5%. See also: DNA inheritance from 1 single ancestor

Realize that therefore your “Nigeria” score could include Nigerian ancestors from various ethnic groups, all at the same time. For example a 50% score “Nigeria” for a typical Haitian might possibly be traced back to 25 Yoruba ancestors, 20 Igbo ancestors and 5 Hausa ancestors (leaving out the possibility of other Nigerian ethnic lineage). Just to mention one possible combination out of many others depending on your individual family history.” (Fonte Felipe, 2019)

“When tracing your African roots expect to find ancestors from multiple African regions and also from multiple African ethnic groups.” (Fonte Felipe, 2014)

____________________

Each regional admixture category on 23andme already features several ethnic subcategories.  As shown above. However from what I have seen sofar it is still quite uncommon to receive two or more ethnic group matches belonging to the same regional admixture category. Despite the progress being made this will still be an underestimation of how truly multi-ethnic the African DNA of Afro-descendants can be. Because each regional admixture category potentially includes ancestors from various ethnic groups.

This essential piece of knowledge has often been obscured by history for many Afro-descendants. Some finding it easier to imagine just one single place in Africa to identify as one’s ancestral homeland. Or just one single African “tribe” they wish to be matched with. In almost all cases this will however prove to be a big misconception. An intricate mix and unique blending of various ancestral lineages from throughout Western, Central  and Southeastern Africa being much more likely.

It is often not fully realized that going by historical plausibility and applying a genealogical timeframe (see this useful overview) most Afro-descendants in the Americas could have atleast dozens and potentially hundreds of separate ancestral lines to be traced back to Africa. It is practically impossible that all these ancestors would belong to a single or even a mere handful of ethnic groups. Even for Cape Verdeans who have a uniquely regionally focused African breakdown (“Senegambia & Guinean”) it is highly likely that they descend from atleast a dozen distinct ethnic groups from all over Upper Guinea. For more discussion:

• What Tribe Am I? (2014)
• Fictional Family Tree incl. African Born Ancestors (2015)
• Is it possible to find a plausible ethnic background ? (ROOTS.NL, part 2, Searching for Gold)
• Fula, Wolof or Temne? Upper Guinean AncestryDNA results 2013-2018
• Akan or Ewe? West African AncestryDNA results 2013-2018
• Igbo, Yoruba or Hausa-Fulani? Nigerian AncestryDNA results 2013-2018

 

 

___________________________________________________________________________

Notes

1)  As far as I am concerned 23andme really blows all competition out of the water with this update. Of course there are still plenty shortcomings as highlighted in this blogseries. However based on more than 50 updated results for African 23andme testers it is undeniable that these ethnic group matches are pretty accurate. Furthermore also for 100 African Americans and 50 Cape Verdeans this update certainly passed the historical plausibility test. From what I have seen most other DNA testing companies tend to fail tremendously on both aspects: getting it right for verifiable backgrounds of African testers and also for historically documented origins of the Afro-Diaspora.

Anyone who’s saying this is just a guessing game doesn’t know what they’re talking about. This feature can really push your research into specific African lineage to the next level! Especially when 23andme expands their African database of reference samples. The specifics of how 23andme is able to come up with these ethnic group matches might not be fully known yet. However the principle of autosomal matching strength is a solid enough basis for me already.

Although I tend to be critical in my assessment of 23andme there is no doubt in my mind that this company is currently best equipped to deliver trustworthy details on African origins in a robust manner. Going by regional admixture and haplogroup assignment 23andme’s trackrecord is already very strong. For many pundits genotyping by 23andme is simply the best around. Especially after the 2018 and subsequent updates it seems 23andme is taking a decisive lead on its competitors. Meanwhile Ancestry has only been putting out lacklustre updates in the last few years. The absence of any new genetic communities for West & Central Africa is really telling…

Given the similar ambition to predict ethnic lineage from Africa it is understandable that some people might also wonder about African Ancestry (the DNA testing company). I have myself not tested with African Ancestry. I found out about my maternal haplogroup (L3E4) on 23andme, which is a much cheaper option and most likely also much more accurate due to higher resolution. My direct maternal lineage is very important to me but I do not have the need to have it matched with a specific ethnic group. Because I realize that Atlantic Afro-descendants are almost by default descended from dozens of ethnic groups, due to the ethnic intermingling across the generations.

It’s good to keep in mind that one’s direct maternal or paternal line represents a very small portion of your overall ancestry if you go back to your first African-born (mainland) ancestor. Around 1% only or even less. This is the reason I personally have always preferred to get tested for my autosomal admixture test results, to get insight into my ENTIRE ancestry. On 23andme you get to receive both admixture analysis, haplogroups and now also ethnic group matches for the same price.

I admire African Ancestry’s pioneering role in providing DNA testing services catered to Afro-descendants. But I regret that they have not made much progress in updating or expanding their services. Also they are known not to be transparent about their methodology and even misleading with their claims (see this article or also this one). It would be really useful to have them do a study about the frequency of matrical/patriclan results among their customers. But they do not seem as interested as 23andme or Ancestry in publishing scientific studies.

Unless this changes in the near future I won’t be blogging about African Ancestry any time soon. However a well-respected African American genetic genealogist published the following article about them:

• Just Say No: African Ancestry’s DNA Tests

2) From my understanding this update has been provided to everyone who tested with 23andme, regardless of chip version. Which is in line with 23andme’s stated intentions in the past. On Facebook 23andme has also confirmed that this update has now been completely rolled out (see this screenshot). I myself am still on V2 and regrettably my Ancestry Composition has not been enriched with the appearance of either “Mandinka” or “Fula & Wolof”. As would be the expectation for people of Cape Verdean descent. However I know for a fact that many people who are likewise on V2 have seen their results updated already.

3) When building their familytree and tracing back into the early to mid-1800’s Cape Verdeans might actually also stumble on relatively recent ancestors carrying the surname “Mandinga”! In fact also other surnames in this period, documented in local church records, appear to be referring to African ethnic groups. I intend to do a more detailed blogpost about this highly fascinating topic eventually.

___________________________________________________________________________

P.s. Congratulations to Senegal for winning the Africa Cup for the first time in history!!! Even when their earlier game against Cape Verde really needs a rematch on equal footing this time 😉