Did Ancestry kill their African breakdown? (part 1)

Last week Ancestry finally rolled out the updated version of its Ethnicity Estimates for all its customers.  Sadly the concerns I raised in July have become reality. Many people are now left confused by their revised African breakdown as reported by AncestryDNA.¹ Understandably so given the often drastic and seemingly incoherent changes compared with the previous set-up. In this three-part blog post I will argue that Ancestry’s pioneering analysis of especially West African DNA has been downgraded rather than upgraded!

Establishing the accuracy of AncestryDNA’s African breakdown has been the focal point of my AncestryDNA survey which I started five years ago already. From my experience the best indication of such predictive accuracy is obtained by looking at how Africans themselves are being described when tested by Ancestry.² Which is why I have performed a before & after analysis of 130 AncestryDNA results of African customers from across the continent. Regrettably it turns out that Ancestry’s African breakdown has indeed taken a turn for the worse. In most cases the informational value to be derived from their results is showing a decrease rather than any improvement. See also this overview which contains all their individual results, before and after the update:

• AncestryDNA results of 130 Africans before and after update

***(click to enlarge)

This is only a small selection but it illustrates some wider patterns I have observed for other Africans as well. It is not only southern Nigerians who are getting a bad deal from this update! Also people of Senegambian/Guinean descent, people with Akan or Kru lineage, Northeast Africans and also northern Nigerians will see drastic changes in their results. But mostly not for the better…

***

I am a guy who prefers to see the glass as half full rather than half empty. Which is why I have been surveying AncestryDNA results for so long despite imperfections. After all when Tracing African Roots most people do not have the luxury to be snobbish about admixture analysis. Instead they will want to maximize informational value from any promising source available, again despite shortcomings. Combining research findings in order to achieve complementarity rather than putting all your eggs in just one basket. But right now after this long awaited update my main feeling is that Ancestry has simply broken the glass 😉 . At least in regards to their African “Ethnicity Estimates“. I find it very regrettable to say that currently I do not see much added value in Ancestry’s updated African breakdown and I will not be surveying it any longer. Nonetheless in this three-part blog series I will attempt to point out some redeeming features as well. As I do still also believe that being overly dismissive may deprive you of valuable insights yet to be gained.

Main topics I will cover in this first part:

1. African breakdown for African AncestryDNA testers
2. Implications for Afro-Diasporans
3. Screenshots of updated African AncestryDNA results

Without wanting to give away the cliffhanger ( 😉 ) these will be the topics for part 2 & 3 of this blog series:

• Did Ancestry kill their African breakdown? (part 2)
  • More is not always better: AncestryDNA’s updated Reference Panel
  • What are the ethnic backgrounds of Ancestry’s African samples?
  • New algorithm has issues with describing mixed/complex lineage?
• Did Ancestry kill their African breakdown? (part 3)
  • FAQ /Common Reactions
  • What’s next?

***

 

1) African breakdown for African AncestryDNA testers

Table 1 (click to enlarge)

Click on this link for an expanded & up-to-date version of this table. Also including group averages for other continental scores (West Asia, Asia & Pacific, Europe).

***

Over 100 samples with backgrounds from at least 30 different African countries have been used for my main survey findings featured above.³ This seems like a reasonably robust number and a wide enough array to pick up on some preliminary patterns. Even when for most of the separate nationalities I was only able to obtain a minimal sample size. Obviously these findings are not intended to reflect any fictional national or ethnic averages! The main purpose of this overview is to give an approximate idea of what to expect when wondering about how AncestryDNA’s update has affected the results of African customers. It is admittedly much to take in and therefore I will only focus on the main changes for each part of the continent:

West Africans (Upper Guinea)

• The “Senegal” region is sharply decreasing while the “Mali” region is skyrocketing.

•  A fall of about 20%-30% on average for “Senegal”, even for confirmed Senegalese.

• An increase of about 35-45% on average for the “Mali” region. Implying that this region is not only replacing much of the former “Senegal” scores. But in fact also many of the other regional scores which were either showing up as secondary regions or minor trace regions.

• Due to being neighbouring regions it was to be expected that “Senegal” and “Mali” would show considerable overlap. Still this outcome mostly contradicts known background and historical plausibility for the people involved, aside from the Malian sample.

• Plus it means a step backwards when compared with the previous situation whereby “Senegal” seemed to function as a reasonable proxy for describing origins from coastal Upper Guinea (often but not per se Atlantic speakers). While “Mali” was acting more as a proxy for origins from more inland parts of Upper Guinea (often central Mande speakers but also covering Gur speakers).

• Minor “North African” scores (<5%) are no longer being reported after the update. While the considerable level of around 14% “North Africa” among my Fula samples is being halved! For Cape Verdeans this outcome is probably mostly linked to the new “Portugal” region which now includes these older ancestral affiliations with North Africa. However also the newly updated “Mali” region might draw in such North African genetic markers. Continued discussion in part 2 of this blog series.

West Africans (Lower Guinea)

• “Ivory Coast/Ghana” and even more so “Nigeria” amounts are in steep decline, while “Benin/Togo” is widely expanding.

• “Ivory Coast/Ghana” scores are decreasing on average between 20-30% for confirmed Akan people from both Ivory Coast & Ghana as well as Liberians (mostly Kru but also southwest Mande)

• “Nigeria” amounts are being evaporated with 30-40% on average for Nigerians across the country! Not only affecting Yoruba’s and Igbo’s from southern Nigeria. In fact also Hausa-Fulani from northern Nigeria are now receiving less than half “Nigeria” as before.

• “Benin/Togo” is especially on the increase for southern Nigerians, around 30% on average. Not only for Yorubas but also Igbo’s. But actually “Benin/Togo” is also expanding greatly into Ivory Coast and even Liberia. Despite these countries not being neighboring ones! Even Hausa-Fulani are now receiving minor “Benin/Togo scores. When previously it was usually absent for them and could be used as a proxy marker between southern & northern Nigerians.

• Not much change in the average “Benin/Togo” amount for my Ewe samples from Ghana. However for one person who used to have 97% “Benin/Togo” it actually diminished by about 30%! Also peculiar how the “Ivory Coast/Ghana” group average is showing an increase of about 10%. Even when for practically all other West Africans this region has been on the decrease.

• Likewise for two actual Beninese samples with Gbe background (Ayizo & Goun) the “Benin/Togo” amounts did not change much. Although interestingly for one of them it decreased by 9%. The “Ivory Coast/Ghana” amounts making the biggest increase (10% on average). The sample-size might be small but it is perhaps telling that no 100% “Benin/Togo” scores were achieved.

• The new “Cameroon, Congo, and Southern Bantu Peoples” region is advancing strongly into Nigeria. In particular for Igbo’s it shows an increase of about 25% when compared with the former “Cameroon/Congo” and “Southeastern Bantu” regional scores combined. But actually also Yoruba’s and Hausa-Fulani are receiving more of this genetic cluster than previously (5% on average).

•  It seems apparent that AncestryDNA’s update has not produced any meaningful improvement in describing the regional origins of people in this part of West Africa. On the contrary it has created a greatly muddled framework. Especially since “Mali”  and “Cameroon, Congo, and Southern Bantu Peoples” are now also appearing in substantial amounts. Even when both of these regions are technically speaking from outside of the Lower Guinea area.

Central & Southern Africans

• The newly combined “Cameroon, Congo, and Southern Bantu Peoples” region seems to be very accurate for Central Africans but not totally comprehensive for Southern Africans.

• For my samples from Cameroon, Equatorial Guinea, both Congo’s as well as Angola (Bakongo) a near perfect 100% score is being reported even! Which is exactly in line with their recent origins obviously. And therefore certainly to be counted as an improvement.

• However the broadening of this greatly expanded region did lead to a loss of former “Hunter-Gatherer” scores. In case of Central Africans indicating absorbed Pygmy lineage from possibly centuries or even millennia ago. Similarly diluted but distinctive “Hunter-Gatherer” scores have actually also disappeared from the results of Liberians and Sierra Leoneans. This type of ancestral trace elements from more ancient time periods might not conform with a genealogical time frame but can still be intriguing and useful for some people I suppose.  See also:
  • South-Central Hunter-Gatherers” suggestive of remnant West African Pygmy DNA?

• Most Southern Africans (except for South Africans) were formerly described by AncestryDNA as a combination of both “Southeastern Bantu” and “Cameroon/Congo”. Which could be somewhat confusing but given awareness of the wider context (relatively recent migrations originating in southern Congo in the 1400’s-1600’s;  genetic deviation from South African samples in Ancestry’s Reference Panel) still largely made sense. A relatively elevated level of either “Cameroon/Congo” or “Southeast Bantu” (when comparing with test results across the area) served to indicate distinctive ethnic origins at times even (such as Ngoni lineage in Malawi who are relatively recent migrants from South Africa or western Tanzanian Bantu people).

• This potentially insightful distinction has now been lost and instead minor but still considerable “Eastern African” scores are showing up consistently for all Southern Africans in my survey. Peaking among my 3 Malagasy samples but also quite elevated in Malawi and Zimbabwe. And even showing up for my Bantu speaking South African samples (Swazi, Xhosa, Zulu). As shall be explained in part 2 of this blog post these so-called “Eastern African” scores can actually also be made less puzzling when knowing which exact East African samples are being used by AncestryDNA. Still intuitively (just going by the regional labeling) this outcome is probably poised to generate greater confusion than previously.

• The “Hunter-Gatherer” regional scores may have practically disappeared for Central Africans. But for South Africans, in particular South African Coloureds, they are still very pronounced. To a larger degree than previously even. Given the widely known and relatively recent ancestral connections with the Khoi-San people this comes as no surprise. However it remains to be seen if these updated scores do indeed correspond better to their actual genetic legacy.

East & North Africans

• “Eastern Africa” is the only truly new African region within AncestryDNA’s update. In itself this makes for a very useful addition. However the prediction accuracy of “Eastern Africa” is rather disappointing. Only reaching levels of 45-55% on average for my Kenyan, Tanzanian, Ugandan and Tutsi samples. However for the few results I have seen from Somalia, Sudan and South Sudan it is even less predictive.

• Even when misleadingly labeled the former “Southeastern  Bantu” region did seem better equipped to single out Nilotic-(like) DNA markers for Northeast Africans. With this update instead also the “Hunter-Gatherer” region is needed as an additional proxy. Out of the blue! Quite astonishingly reaching levels of nearly 60% even for one Ugandan but also appearing very pronounced for my South Sudanese, Somali and Tutsi samples. It is more variable among Tanzanians and Kenyans, most likely correlating with actual ethnic background. This outcome comes across as quite random and outlandish but is possibly to be explained by the addition of new Hunter-Gatherer samples in AncestryDNA’s Reference Panel.  Presumably from Tanzania, continued discussion in part 2 of this blog series.

• The newly combined “Cameroon, Congo, and Southern Bantu Peoples” region is another major component in the results of Kenyans, Tanzanians and Ugandans. On average in between 20-50% for my survey participants. Individual scores will most likely be correlating with ethnic background. Although perhaps less apparent than in the former upset when “Cameroon/Congo” could be used for example to indicate eastern Congolese ancestry for Swahili Kenyans. Still it is tellingly absent for my Somali sample and quite subdued for my Tutsi samples.

• Minor “North African” scores are on the decrease for Kenyans and Somali. However for actual Maghrebi the “North Africa” region appears to have become more predictive. More so for my 3 Moroccan samples (79%) than for the Algerian ones (44%). However my minimal group averages are somewhat skewed. As it is quite likely that 1 Moroccan person who scored 100% “Northern Africa” after the update has been included as an customer sample into AncestryDNA’s Reference Panel. Causing an overfitting or calculator effect which arguably also occurred on 23andme in the past. Actually this also seems to have happened with two of my Nigerian samples (Yoruba & Igbo).

• Aside from that one perhaps atypical 100% result my other North African survey participants still show greatly blended compositions. Also including West Asian and South European scores. Even if to a lower degree when compared to before the update. Such breakdowns (mostly reflecting ancient geneflow/similarity) can still be misleading therefore, when taken too literally. Especially given the appearance of the new seemingly specific “Portugal” region.

• A regrettable loss of distinction seems to be occurring for the usually minor but still clearly detectable West African admixture being reported for Moroccans and Algerians. Before the update this ancestral portion was consistently being described in either “Senegal” or “Mali” regional terms. However after the update it seems to have been incorporated mostly within the new “Northern Africa” region only leaving room for a very minimal scattering of various West African regions. Historically speaking a mostly Upper Guinean component reflecting West African admixture among Maghrebi’s is much more plausible and therefore more informative.

 

2) Implications for Afro-Diasporans

____________________

“Please note that genetic ethnicity estimates are based on individuals living in this region today. While a prediction of genetic ethnicity from this region suggests a connection to the groups occupying this location, it is not conclusive evidence of membership to any particular tribe or ethnic group.” (source: Ancestry)

____________________

Ancestry fortunately continues to provide useful background information and careful disclaimers on how to interpret their “Ethnicity Estimates” and the various regions which are shown in it. Customers should really invest more time in informing themselves properly about this AncestryDNA update. In order to avoid being left confused or mislead by their new results.  A good start is: pay close attention to the regional maps integrated within your Ethnicity Estimate!

Compared with other DNA testing companies I still find that Ancestry does a rather good job at providing helpful sections/pages offering guidance and context.⁴ Although it seems the information they are offering is less extensive than it used to be. For example they appear to have discontinued the very helpful “genetic diversity” tabs which included detailed statistical information about the predictive accuracy of each single region. Also the regional descriptions appear to have remained the same, even when there have been such drastic changes! However very crucially Ancestry did update their “also found in...” details mentioned within the regional maps. Essential reading for those who insist on taking the country name labeling on Ancestry as gospel 😉

For further reading see:

• Frequently asked questions about our next-generation ethnicity estimate. (Ancestry)
• DNA Story Help & Tips (Ancestry)
• Ethnicity FAQ (Ancestry)
• AncestryDNA Regions (before update) (Tracing African Roots, 2015)

In the past I have attempted to extract general lessons to be learnt from the AncestryDNA results being reported for Africans. Wishing to improve correct interpretation of AncestryDNA’s African breakdown for Afro-Diasporans as well as Africans themselves. Sadly I do not think such an effort will be very fruitful any more. At least not based on the data currently available to me. Given the general decline of informational value as well as the great degree of incoherency for many of the results I discussed above. It is to be feared that the randomness factor will be even greater for Afro-Diasporans who will usually have a much more complex background than Africans themselves. Multiple ancestral scenarios might be valid now more than ever. None to be ruled out in advance. Decreasing the odds of any clearcut interpretation based on deductive logic.

I will keep it brief therefore. As always this overview firstmost reflects my own reasoning and naturally other interpretations remain possible as well.  You will need to perform your own follow-up research (relevant historical context, DNA matches, genealogy etc.) in order to find out more specifics. Although again, personally I think such an attempt will often be futile and not worth the effort. In many cases it might be best to stick with your previous results…Continued discussion in part 3 of this blogseries.

1) “Senegal” has become less predictive of  actual Senegambian/Guinean DNA

***click to enlarge)

Source: Ancestry. Text in red added by myself based on my survey findings and informed speculation. Ironically the geographical spread shown in this map corresponds very well with my survey findings before the update. But less so afterwards. Follow this link for the pre-update version of this map.

***

2) “Mali” scores are likely to be inflated. Not only replacing former “Senegal” scores but also often over-extending into other West African countries (not per se neighbouring ones even!).

***click to enlarge)

Source: Ancestry. Text in red added by myself based on my survey findings. Take note of the vast territory covered now by this region. Which probably still is an underestimation of its true extent into also Ghana and Nigeria. Follow this link for the pre-update version of this map.

***

3) “Ivory Coast/Ghana” has become less predictive of actual Ivorian/Ghanaian Akan lineage as well as less predictive of Liberian and Sierra Leonean origins.

***click to enlarge)

Source: Ancestry. Text in red added by myself based on informed speculation. Ironically the geographical spread shown in this map corresponds very well with my survey findings before the update. But less so afterwards. Follow this link for the pre-update version of this map.

***

4) “Benin/Togo” scores are likely to be inflated. Not only replacing former “Nigeria” and “Ivory Coast/Ghana” scores but also often over-extending into other West African countries (not per se neighbouring ones even!).

***click to enlarge)

Source: Ancestry. Text in red added by myself based on my survey findings. Astounding coverage of nearly all West Africa and even including Gabon! It begs the question why AncestryDNA still bothers with a seemingly exact country name labeling. As it only sets up for major confusion… Follow this link for the pre-update version of this map.

***

5) “Nigeria” has become less predictive of actual Nigerian DNA. Both southern and northern Nigerian ancestry is now very poorly covered by “Nigeria”. Any remaining “Nigeria” will not per se be linkable to any given ethnic group or subregion within Nigeria.

***click to enlarge)

Source: Ancestry. Text in red added by myself based on informed speculation. As previously “Nigeria” amounts were also reported for people from South Sudan. Possibly indicating some Chadic-Nilo-Saharan connections. However, a northern shift of the “Nigeria” region has not been confirmed by my survey findings. Follow this link for the pre-update version of this map.

***

6) “Cameroon, Congo, and Southern Bantu Peoples” scores are likely to be inflated. Replacing former “Nigeria” scores in particular. Very lamentable that AncestryDNA has only aggravated the already existing issue of overlap between Bight of Biafra and Central African origins! Also the removal of the former “Southeastern Bantu” region is a step backwards in achieving finer regional resolution for Afro-Diasporans.

***click to enlarge)

Source: Ancestry. The regional span of the combined “Cameroon/Congo” and Southeastern Bantu” regions covers an enormous area, including more than 20 countries and not just Cameroon & Congo! But also for example Angola, Mozambique, Madagascar and Nigeria! Together with “Benin/Togo” epitomizing the loss of focus in AncestryDNA’s updated African breakdown. Follow this link for the pre-update version of this map.

***

7) “Eastern Africa” is not a perfect measure of Northeast African DNA. It is also predictive of Southeast African ancestors. This latter ancestral scenario will usually be much more relevant for Trans-Atlantic Afro-Diasporans due to historical plausibility.

***click to enlarge)

Source: Ancestry. Text in red added by myself based on my survey findings. This new region is  to be found as far south as South Africa and also in countries such as Madagascar, Malawi, Zimbabwe and Zambia.

***

8) “South-Central Hunter-Gathers” has become less predictive of actual Pygmy DNA. And is now also to be found in Northeast Africa!

***click to enlarge)

Source: Ancestry.  Although not mentioned in the regional description text it seems very apparent that new Hunter-Gatherer samples have been added within AncestryDNA’s Reference Panel. Most likely from Tanzania (either Hadza or Sandawe). The map does clearly show this unexpected (and unhelpful) expansion into Northeast Africa. Follow this link for the pre-update version of this map.

***

9) “Northern Africa” has become less predictive of more ancient affiliations with North Africa. 

***click to enlarge)

Source: Ancestry. Text in red added by myself based on my survey findings. Before the update “Africa North” scores were more pronounced among Cape Verdeans, Northeast Africans and Fula & Hausa-Fulani people. Most likely either absorbed in the new “Portugal” and “Eastern Africa” regions or also partially in the “Mali” region due to the addition of new samples.

***

Macro-regional framework still useful?

Table 2 (click to enlarge)

This table is based on my survey findings for 5 Cape Verdean AncestryDNA results. The African breakdown being scaled to 100% and corrected for North African scores. This is however not necessary if you want to compare your own before & after results using this macro regional framework. Simply use your original and updated African amounts and contrast your former “Cameroon/Congo” and “Southeastern Bantu” scores with the newly combined “Cameroon, Congo, and Southern Bantu Peoples” regional score. Also add in any “Eastern African” or “North African” score whenever present.

***(click to enlarge)

Combining various levels of regional resolution: on the left the 4 main regions of provenance for Afro-Diasporans (based on slave trade records, see also this link). African ethno-linguistic groups according to the most recent classification are depicted in the middle (see also this link). The map on the right features the current 9 African AncestryDNA regions (source: Ancestry).

***

To make more sense of my Afro-Diasporan survey findings I have been using an additional more basic macro-regional framework divided into: Upper Guinea, Lower Guinea and Central & Southeast Africa (see this table for my latest findings). Basically combining interrelated & neighbouring AncestryDNA regions on a macro-level. Such a grouping also being based on historical and ethno-linguistical considerations, aside from genetic ones (see maps directly above). This research approach may still have some added value when trying to determine how any newly updated African breakdown compares with the previous one. Because it allows certain regional patterns to show up more clearly. Drastic proportional changes between macro-regions perhaps being less likely than any relative shifting between neighbouring regions within either Upper- or Lower Guinea or Central & Southeast Africa.

Like I said earlier I will no longer be continuing my Ancestry survey for Afro-Diasporans. However I did perform such an exercise for the 5 updated Cape Verdean results in my survey, as shown in table 2. The consistent replacement of “Senegal” scores by “Mali” scores for Cape Verdeans is rather misleading and represents a decrease in informational value in my opinion. However both “Senegal” and “Mali” regions are ultimately indicative of Upper Guinean lineage. And it might be best therefore to simply combine both scores to get an approximate idea. Seen from that perspective the African macro-regional breakdown for my Cape Verdean survey participants arguably has improved. Take note especially how most of the unexpected non-Upper Guinean scores before the update have been diminished considerably! On the other hand due to the country name labeling people might get the wrong impression about Cape Verde’s Upper Guinean roots really being much more coastal, incl. Guiné Bissau, rather than interior. So this outcome still goes against historical context.⁵ See also:

• Cabo Verde Raizes Na Africa
• 100 Cape Verdean AncestryDNA results

Although possibly still useful in individual cases I am doubtful this framework can be used in a meaningful way on a group-level when wanting to contrast with historical slave trade data. Given the far greater degree of regional overlap the distinction between Upper & Lower Guinea and between Lower Guinea and Central/Southeast Africa has become much more blurry than before. According to Ancestry’s own information “Benin/Togo” is now to be found as far south as Gabon (!) which technically speaking would be Central Africa. On the other hand the new “Cameroon, Congo, and Southern Bantu Peoples” region is also advancing across Nigeria. And probably also reported in subdued amounts for Beninese themselves. It already used to be quite difficult to separate Bight of Biafra (to be included in Lower Guinea) from Central African ancestral connections. But making the distinction has now become nearly impossible! While “Mali” is now showing up with substantial scores for people native to not only Liberia but also across southern parts of Ghana and the Ivory Coast and even into Nigeria! Undoing its signalling function for primarily Upper Guinean lineage.

During my AncestryDNA survey I also kept score of the frequency of top ranking regions within the African breakdown (see this chart for my latest findings). Again I think such an exercise, which used to be quite insightful, will no longer be meaningful when wanting to contrast with historical slave trade data. Just going by the many updated results I have seen already (mainly for African Americans and Hispanics) I highly suspect that “Benin/Togo” as well as “Cameroon, Congo, and Southern Bantu Peoples” will overwhelmingly and often misleadingly be reported as primary region henceforth. For many parts of the Afro-Diaspora, incl. also West Indians and not just Haitians and Brazilians for whom a top ranking frequency of both regions is indeed historically plausible (on a group level, so leaving aside individual variation).

In one of my previous blog posts (before the update had been rolled out completely) I asked the following question:

____________________

“It remains to be seen how well my present findings will correspond with any newly calculated AncestryDNA results. Will they be rendered completely obsolete or may they still contain lasting insights about the approximate composition of African regional origins for my survey groups? “

____________________

Without wanting to come across as pretentious (I do realize I am a mere layman researcher 😉 ) I think it is already apparent that the updated African breakdown on AncestryDNA provides a worse basis for research purposes than the previous one. Obviously AncestryDNA’s African breakdown which was provided between 2013-2018 also had its shortcomings. But still it represented a major advance at that time and offered unrivaled West African regional specificity when compared with other commercial DNA testing companies during that same time period as well as any third-party analysis such as available on Ged-Match, DNA Land etc..

Personally I consider it to be a true loss that Ancestry’s pioneering analysis of especially West African DNA seems to be going down the drain with this update… In some ways I even find that the prevailing and over-extended regional trio of “Mali”, “Benin/Togo” and “Cameroon, Congo, and Southern Bantu Peoples” is reminiscent of the simplistic three-way model (based on Mandenka, Yoruba and Bantu samples) used in DNA studies from 2009/2010! Almost making it seem like Ancestry turned back the clock on its African breakdown with practically ten years!

Eventhough I have argued in this blog post that AncestryDNA’s update has generally not been worthwhile for the African breakdown. I still do want to commend them for at least giving their customers ample opportunity to get used to their new results and providing feedback. When being informed about the update you are actually given the choice to not accept the new results right away (see this screenshot). In other words you are allowed to hold on to your old results, at least for a while. Not exactly a return to sender option as I understand it but still quite considerate given the often drastic changes 😉 If you are not happy with your updated results let Ancestry know about it!!! Also forward them this link (when in agreement of course):

• Suggestions for improving the African breakdown on AncestryDNA

 

3) Screenshots of updated African AncestryDNA results

I will only be posting a limited selection of screenshots. But this array does cover all main parts of Africa. Obviously there might be greater individual variation within each country or even within a particular ethnic group. Still this should already be quite illustrative of the main patterns I have described above. As far as I was able to verify all of the following screenshots below are from persons with four grandparents from said nationality/ethnicity, unless specified otherwise. But naturally I did not have absolute certainty in all cases. Practically all results have been shared with me by the DNA testers themselves.  I like to thank all my survey participants for having tested on AncestryDNA and sharing their results online so that it may benefit other people as well!

It is highly advisable by the way to make a screenshot of your old results for future reference. Ancestry provides a useful option for visually comparing your updated results with your old results. Just follow steps below (I have encircled wherever you need to click):

Compare your old & updated results

***(rightclick and open in new tab to enlarge)

This slideshow requires JavaScript.

 

***

SENEGAL (Fula)

***

***

SENEGAL & GUINEA (Fula)

***

***

GUINEA  (Mandinka)

***

***

SIERRA LEONE (Mende)

***

***

LIBERIA (Kru)

***

***

IVORY COAST (Akan: Agni/Aowin)

***

***

IVORY COAST (7/8 Akan & 1/8 Krio (Sierra Leone))

***

***

IVORY COAST (3/4 Akan & 1/4 Krio (Sierra Leone))

***

***

GHANA (Akan & Ga)

***

***

BENIN (Gun/Goun)

***

***

BENIN (Ayizo)

***

 

NIGERIA (Hausa-Fulani)

***

***

NIGERIA (Bini, Itsekiri, Urhobo & Isoko)

***

***

NIGERIA (Yoruba)

***

***

NIGERIA (Yoruba)

***

***

NIGERIA (Yoruba)

***

***

NIGERIA (Yoruba)

***

Quite atypical result as the “Nigeria” score is (spectacularly) increasing rather than decreasing. It is quite likely however that this person has been included as an customer sample into AncestryDNA’s Reference Panel. Causing an overfitting or calculator effect which arguably also occurred on 23andme in the past.

***

NIGERIA (Igbo)

***

Quite atypical result as the “Nigeria” score is increasing (it used to be 88% which is already very high) rather than decreasing. It is quite likely however that this person has been included as an customer sample into AncestryDNA’s Reference Panel. Causing an overfitting or calculator effect which arguably also occurred on 23andme in the past. I do not have any other Igbo screenshots available at the moment. However see also my spreadsheet for more updated Igbo results.

***

CAMEROON (?)

***

***

EQUATORIAL GUINEA (Fang)

***

***

CONGO BRAZZAVILLE (Bakongo)

***

***

CONGO (DRC) (Kasai: Ndundu, Bangala and Tetela)

***

***

CONGO (DRC) (?)

***

***

MALAWI (1/2 Chewa & 1/2 Yao)

***

***

MALAWI (1/2 Ngoni, 1/4 Yao &1/4 Ngonde)

***

***

SOUTH AFRICA (Coloured)

***

***

MADAGASCAR (Northeast)

***

***

UGANDA (Nilo-Saharan: Aringa, Kakwa)

***

***

UGANDA (Nilo-Saharan: Acholi)

***

***

UGANDA (?)

***

***

South Sudan? (?)

***

***

SUDAN (Arab)

***

***

TUNISIA (southern)

***

***

MOROCCO (?)

***

See also this Youtube video

***

CAPE VERDEANS (5x)

***

This slideshow requires JavaScript.

 

___________________________________________________________________________

Notes

1) It might be a different story for the European and Asian breakdowns. I have actually seen quite encouraging updated results in this regard. And generally speaking they could be an improvement indeed. Although there are also still some remaining issues. The non-African regional breakdowns are however not a topic of discussion in this blog post.

2) In order to investigate the predictive accuracy of AncestryDNA’s African breakdown I have been calculating provisional benchmarks (group averages, see this overview) based on African customer results. But I have also looked into historical plausibility in regards to the documented African roots of the Afro-Diaspora. Based mostly on slave trade patterns and historical references to specific African ethnic origins for the Afro-Diaspora (see this overview).

In the latest summary of my main findings I concluded that based on 1,377 surveyed results from 34 nationalities AncestryDNA’s African breakdown has usually been in alignment (broadly) with the verifiable or historically plausible background of both Afro-Diasporans and Africans. Despite obvious shortcomings such an outcome is not something to carelessly brush aside when wanting to Trace African Roots. It is a true shame therefore that Ancestry’s update does not build upon this promising and already quite solid foundation but instead seems to have taken a step backwards. See also:

• Update: Afro-Diasporan AncestryDNA Survey (part 1)

3) Practically all African AncestryDNA results have been shared with me by the DNA testers themselves. Although a few times I also obtained them by way of social media. Naturally I verified the background of each sample to the best of my capabilities but I did not have absolute certainty in all cases. I like to thank all my African survey participants for having tested on AncestryDNA and sharing their results with me so that it may benefit other people as well!

When I first started out in 2013 with my AncestryDNA survey among Afro-descendants I had to wait for a quite while to also include a few African test results. Even when back then I already fully realized their corroborating potential. In the last two years or so DNA testing has fortunately become increasingly popular among Africans. And therefore I was able to compile the data based on 130 African testresults in this spreadsheet. Of which 114 results were used in table 1. Actually I have collected many more African AncestryDNA results over the years. Around 500 right now. Which can be seen in these online spreadsheets:

• African AncestryDNA results (n=299)
• Cape Verdean AncestryDNA results (n=100)
• Fula AncestryDNA results (n=44)
• Nigerian AncestryDNA results (n=77)

4) Generally speaking the African regions on AncestryDNA have become more generic and less specific. Which certainly does not count as an improvement in my book! The country name labeling is still potentially very misleading when taken at face value. Even more so after the update. After all so-called “Benin/Togo” is now also to be found in Sierra Leone and even in Gabon! The new region called “Cameroon, Congo, and Southern Bantu Peoples” is even including over 20 countries! With such a wide area covered,  it really begs the question why this seemingly exact country name labeling is to be maintained…

On the other hand ancestral categories referring to ethnic groups might be just as deceptive or even more so. As many people will again tend to take them too literally. Underestimating not only the sheer number of ethnic groups existing in Africa (thousands!) but also the complexity of interplay between fluid ethnicity, overlapping genetics and shifting political borders. An intermediate solution might be ancestral regions which are referring to either non-political geography or ethno-linguistic groupings. But I fear there will always be some degree of blurriness involved and exact delineation might be impossible to achieve in many cases.

However to their credit Ancestry is not entirely trying to cover up this tricky issue. In their short but still reasonably informative regional descriptions they do mention ethnic diversity existing for each region explicitly. For example for Nigeria:

____________________

“Nigeria as a country is a concept and product of colonialism, bringing together more than 250 ethnic groups within fairly arbitrary borders.” (source: Ancestry)

____________________

Also from Ancestry’s FAQ section it can be learnt how their regional maps are to be interpreted. They are again not to be taken too literally as Ancestry obviously does not possess perfect information about the genetic background of all Africans! And this should also not be expected from them. But these maps can still be very useful to give you an approximate idea of how widely extended each region might be. As long as you keep in mind these maps are made on a best-effort basis. From my previous survey findings I found that Ancestry tended to underestimate how widespread some of their regions could be. For example “Ivory Coast/Ghana” extending into Liberia and Sierra Leone. They have mostly corrected this in their present update however there may still be some omissions.

____________________

“A region map shows where we find the greatest concentrations of people assigned to a particular region. The maps are made using people born in the area, and where possible we use people with known deep roots in that area. The darker the color, the more people we find who have this region in their estimate. The maps are approximate, and people outside the highlighted shape may still have that region in their estimate. Similarly, not everyone living within the highlighted area will have the region in their ethnicity estimate.” (Source: Ancestry)

____________________

5) There is actually solid evidence pointing towards genuine Malian lineage for Cape Verdeans (see section 3 of this blog page). However to a much lesser degree than these updated AncestryDNA results are suggesting. Certainly not predominating other types of Upper Guinean lineage, hailing from Senegambia, Guiné Bissau, Guinea Conakry and Sierra Leone!  I am quite certain that these inflated “Mali” scores are mostly an artefact of Ancestry’s new algorithm as well as their new selection of reference samples (it now has 169 samples from Mali versus only 31 from Senegal and ZERO from Guiné Bissau, see this link). Hence why Cape Verdean’s African DNA now gravitates towards “Mali” rather than to “Senegal”. I find that the new “Senegambian & Guinean” category on 23andme has a far more fitting proxy labeling for Cape Verde’s Upper Guinean lineage than either “Mali” or “Senegal”.