Mozambique is a somewhat overlooked country of origin for many Afro-descendants. To be sure the Trans-Atlantic Slave Trade with Southeast Africa was relatively smaller in scope than with either West or Central Africa. Still according to most recent estimates it involved almost half a million people (see this overview from the TAST database). Especially for Brazilians, Haitians and Cubans Mozambican lineage might be considerable in some cases. Plus also the Indian Ocean Slave trade is to be taken into account. Which most likely resulted in a dispersion of a similar number of Mozambicans (see this overview, taken from this paper). Especially into South Africa and the Indian Ocean Islands: such as Mauritius, Réunion and Madagascar. But also to the Middle East and into South Asia and beyond even. For more details:
Map 1 (click to enlarge
Then again some people might also have Mozambican ancestry due to recent migrations. In the last couple of months I have been given access to DNA results which reveal an already confirmed Mozambique connection for two people of mixed background. With very fascinating outcomes! As well as useful implications for other people wanting to learn more about their own possibly Mozambican lineage.
The first person is living in the Netherlands (like me). Both of his parents were born in Mozambique. But going back two generations his father’s grandparents are all of Portuguese descent. While his mother’s family has been living in Mozambique for as long as can be traced back. But her family tree does contain at least two persons who were born in India and migrated to Mozambique as well. When it was still under Portuguese rule. All in all a very fascinating tale of migration across several continents! This person tested with MyHeritage and he has been featured in a Dutch documentary series called “Identity”. I was actually honoured to also contribute to this documentary which was shot in my birth place Rotterdam. The actual day of filming being the 5th of July which happens to be Cape Verde’s Day of Independence! Highly symbolic for me therefore 😉 1 See also:
Figure 1 (click to enlarge)
The other person is a DNA cousin of mine on Ancestry whose mother was born in São Tomé & Principe while his father is from Mozambique. I am related to him by way of a mutual Cape Verdean relative who migrated to São Tomé & Principe as a contract labourer during the first half of the 1900’s. São Tomé & Principe is a former Portuguese colony just like Cape Verde and Mozambique. But nowadays it is an independent island state located nearby Gabon in the Gulf of Guinea. Just like Angolans and Mozambicans many Cape Verdeans ended up settling in São Tomé & Principe during the Portuguese colonial time. Notoriously being employed under very harsh circumstances on the cocoa plantations of São Tomé & Principe. It therefore still has a sizeable community of Cape Verdean descendants. Their continued longing for their motherland has been made world famous by the song “Sodad” by Cesaria Evora (see this video clip).
This blog post will describe the DNA results of both persons. Seeking to demonstrate in particular how correlating regional admixture analysis with DNA matches can be rewarding and mutually reinforcing in many cases. In addition I will also discuss my experience with MyHeritage. As I have myself not tested with this company. And so this was the first time I had the chance to get acquainted with several of its features. Allowing me to also make some cross-comparisons with Ancestry.
- Mozambican Connection on MyHeritage
- Mozambican Connection on Ancestry
- Comparing Ancestry with MyHeritage
- MyHeritage is distorting Central & Southeast African DNA
- Ancestry offers greater potential to find African matches
- MyHeritage provides more advanced filtering tools
1) Mozambican connection on MyHeritage
Figure 2 (click to enlarge)
Regional admixture estimates
I will start my discussion with the MyHeritage results for the Dutch person with one Mozambican parent and one parent of Portuguese descent. I have never tested myself with MyHeritage. Because I do not have a high opinion of their Ethnicity Estimate tool. Frankly I believe it often produces misleading results for both Afro-descendants and Africans themselves. This is mostly to do with their Reference Panel being inadequate. Especially it seems MyHeritage often gets it wrong when covering Central or Southeast African lineage. But probably also their algorithm is not up to par with either Ancestry or 23andme. I will discuss this in greater detail in the last section of this post.
Generally speaking my approach has always been that regional estimates require correct interpretation. Each separate DNA testing company and each subsequent update of their admixture estimates to be judged on its own merits. Despite inherent imperfections you might still obtain informational value from admixture analysis. Even when it is not the best on the market. So from that perspective I will discuss whichever aspects might be educational from the results featured above.
At first look these results admittedly do look all over the place! And for anyone receiving such results for the first time it will be understandable they will be amazed and confused from all the bewildering variety. Trying to make sense of it by comparing with known family history will often be in vain. Which may then lead to either sceptism and overhasty dismissal or also self-deception when trying to justify “exotic” scores!2
As I have always maintained on my blog so-called “Ethnicity Estimates” in reality provide a regional sketch of your genetic origins. Nothing more but also nothing less! By default it will be an imperfect approximation and will not line up exactly with your genealogical origins. Still if you invest some time and research to make more sense out of it all it can be very useful. Especially when combined with findings from other fields: relevant historical context, genetic genealogy and in particular DNA matches! Focus on the bigger picture and don’t obsess over smaller details or get distracted by obvious shortcomings.
The continental breakdown in admixture results is often said to be most reliable. And this also seems to be the case for MyHeritage’s Ethnicity Estimate. We can see for example that the European total (~49%) is in line with having one parent of fully Portuguese descent. His African total (~45%) would correspond also with having one parent of almost fully African descent. Perhaps most insightful being the Asian estimate of 5,8% “South Asian”. Which aligns well enough with this person having two known great-great grandparents from India! Even when the expected genetic inheritance from several generations ago will vary due to recombination (see this chart).
Zooming in closer into the sub-continental or regional breakdown it is essential to be aware of ancient population histories. And especially not to take the labeling of the ancestral categories too literally! His Portuguese side is mostly described as “South European” and “Iberian” which would be correct. The so-called “Italian” and “Greek” results are very likely to be merely a misreading of actual Portuguese DNA. Even when going back thousands of years possibly some Roman ancestry might be indicated I suppose. The minimal score of 1,9% “Irish, Scottish and Welsh” could like wise merely be a reflection of ancient Celtic migrations into Portugal. See also my discussion of Portuguese AncestryDNA results (before the 2018 update!)
The small “East European” and “Ashkenazi Jewish” scores are possibly suggesting distant Jewish ancestry. Which certainly could be historically plausible. Although in Portugal’s case it would be Sephardi Jewish rather. But frankly I greatly suspect this is just an artefact of MyHeritage’s imprecise analysis. I find that both Ancestry and 23andme are much better equipped at detecting Jewish admixture. One might also want to seek corroboration by Jewish DNA matches. However I hardly found any for this person.3
Moving onwards to the African breakdown. This part of the results again looks quite messy on first sight. Mainly due to MyHeritage’s poor choice of African reference samples. These are used to compare with your own DNA for genetic similarity. All according to a certain algorithm and in order to arrive at their estimates. But only a handful of African countries seem to be represented in MyHeritage’s database of reference samples. This limited set of samples forms the basis for an African breakdown of 10 regions. Which looks decent but actually is heavily skewed to North and Northeast Africa (6 regions!) and therefore often leads to distorted outcomes. More details in section 3.
Still when wanting to focus on the positive aspects it is noteworthy that “East African” is the main component within the African portion. As might be expected for a person of Southeast African descent. However in addition also “West Africa” shows up rather strongly with 14.8%. Even when a central-south pull would be much more fitting. But “Central Africa” is quite minimal (2,4%). While “North Africa” is actually most likely to be added to the European breakdown as forming part of Portuguese genetics (as can be verified also by looking into the results of Portuguese MyHeritage testers who tend to score around 5-10% for it).
Further specification is not intended to be taken as gospel! The mentioning of only “Kenya” and “Nigeria” does make it clear though that obviously Africa is composed of more than 50 countries. But like many other African countries Mozambique is not represented with its own category. Because currently MyHeritage does not have any reference samples for it. Given the limitations of the regional set-up it is only natural that a person of Mozambique will be described in proxies and best fitting equivalents. Being aware of the Bantu migrations originating in the borderland between Nigeria and Cameroon and also moving by way of Kenya into Mozambique does help to better understand this outcome. Although really this has zero genealogical relevance because these migrations took place hundreds if not thousands of years ago…
DNA Matching Patterns
Table 1 (click to enlarge)
Regional admixture results are inherently imperfect but can still be informative with proper interpretation. Then again they can only take you that far. My advise therefore is to also look into your DNA matches for greater combined insight. Aside from historical plausibility and just plain genetic genealogy. This is exactly what I have done also for this Dutch person of mixed Mozambican and Portuguese descent. Because his total number of DNA matches was quite small (n=422) I have performed a manual scan of his matches. Checking each and every profile of his DNA matches for relevant details to establish their plausible background. Obviously 100% accuracy is not guaranteed as this was done on a best effort basis in order to get a general idea.4
There are always several disclaimers to take into consideration when dealing with DNA matches. First of all the possibility of so-called false positives or matches not being IBD (Identical by Descent). Also especially for mixed people there might always be several ancestral scenarios which might explain a DNA match. The shared MRCA (Most Recent Common Ancestor) might not always be of the same nationality or background as the DNA match himself due to migrations and inter-ethnic unions in previous generations. I discuss these considerations in greater detail in this blog post (section 1):
The overview above contains a lot of data so I will just focus on the main outcomes. The closest matches (shared DNA >20 cM) will be most important for learning about recent family origins. As they are after all the result of relatively recent ancestral ties and will also be most reliable. We can see that these close matches are from a more restricted range of countries. Pretty much already confirming this person’s Portuguese side.5 I did not find any strictly South Asian DNA matches to corroborate his partial Indian side. But his 3 Indo-Portuguese DNA matches might also qualify. Even when also shared Portuguese ancestry could be relevant in their case. Having a closer look at the ethnicity of the shared DNA segments will clarify things.
Only 1 out of 422 matches turned out to be Mozambican. Highlighting the difficulty of receiving appropriate matches for people of less common backgrounds. Judging from the admixture results shown for this match she is very much of mixed descent (called mestiço in Mozambique). Actually mostly Portuguese and Indian but also to a minor degree some African admixture.6 Her family is presumably from Pemba. Which is a port city in northern Mozambique. Intriguingly also home to a Swahili community. So again several ancestral scenarios are possible. Still very telling that this single match from Mozambique also turned out to be by far the closest match! With shared DNA being 81 cM while the maximum amount of shared DNA among Portuguese and/or Brazilian matches was 32-33 cM.
Reviewing table 1 it will be obvious that Portuguese related matches are the overwhelming majority. Not only going by fully Portuguese matches but also those of partial Portuguese descent. The very wide ranging migrations of Portuguese people to be kept in mind of course. Not only restricted to Brazil and other parts of the former Portuguese colonial empire. But also including the Hispanic Americas in fact (especially during the Iberian Union).7 As well as Portuguese/Madeiran contract labourers settling in the West Indies during the 1800’s.
Aside from the strictly Portuguese DNA matches (n=94), the Brazilian (n=91) and Hispanic DNA matches (n=95) were most numerous. Although I suspect for a greater part most of these matches will indeed be Portuguese or rather Iberian related. Still intriguingly a Mozambican connection might be possible as well for some of the Latin American matches! Whenever they also happen to show African admixture of course. I have not explored this any further. However this could be established by focusing on shared ethnicity, especially when being labeled “East African”. And also by using MyHeritage’s chromosome browser.
The African DNA matches form less than 3% of the total (12/422). And actually when focusing on those who show “100% African” admixture it is only 2 matches (1 from Kenya with a score of 97,9% “East African” and one from South Africa showing 40% “East African” aside from other African scores). Naturally this outcome is determined by the composition of MyHeritage’s customer database. Although MyHeritage is actually based in Israel its customers are overwhelmingly American and European. Luckily many migrants residing there do also take DNA tests. But the fact remains that the chances of finding Portuguese related matches are much higher than finding African or even Asian matches.
Like I said already the greater part of these African matches are actually of mixed background. Often also including “South Asian” admixture to some degree. So I suppose a few of them might perhaps be related through a mutual South Asian ancestor! This goes especially for the 5 South African matches who appear to be either Coloured a.k.a. Kleurling or Afrikaaner. Still a Mozambican connection for them would also be in line with historical expectations. As it is known that the Dutch imported Mozambican captives to their Cape Colony, especially during the late 1700’s. This connection by way of Indian Ocean slave trade is even more likely for the match from Réunion. A former French sugar colony which had a high proportion of Mozambicans in its slave population. Continued also in the 1800’s actually. Tellingly this match showed a rather high shared DNA segment (16,1 cM). Follow these links for more details:
- The Early Cape Slave Trade (South African History Online)
- The French Slave Trade in East Africa (1721-1810) (E. Alpers, 1970)
- In the place of slave labor: the trafficking of engaged workers from Mozambique to Reunion Island in the post-abolition of French colonial slavery (P. Gonçalves, 2019)
Swahili Coast Connection
Map 2 (click to enlarge)
I find it very fascinating to see how DNA matching patterns do not only reflect your personal family history but also reveal wider histories of population migrations and intermingling. Usually going back much further back in time than your own family knowledge may allow for. Embedding your own family-identity in a much broader context. Often spanning several parts of the globe. A nice illustration of the Six Degrees of Separation theory. Although again one must be careful to analyze the direction of geneflow in each separate case.
Within the family traditions of his Mozambican mother there is no known direct relation with the Swahili people, beyond acquaintances. But his mother’s family in fact is Muslim. And also her family has been living in several coastal cities of Mozambique with a rich Swahili past. Including Inhambane in the south as well as Quelimane further north. Although nowadays Swahili culture is mostly centered in Kenya and Tanzania. It does also extend into northern Mozambique. And formerly Swahili trading posts were scattered all over the Mozambican coastal line. Sofala probably being most famous for also providing access to the gold trade with Great Zimbabwe in the interior.
I find it very intriguing that probably two DNA matches from Kenya and Comoros are Swahili related. Judging from their surnames as well as their admixture results (which also feature minor amounts of South Asian and Middle East) I believe the odds are high that these matches would self-identify as Swahili. Such a finding is a great way to connect with Mozambique’s past and historical dealings with its neighbouring countries. We already saw that in fact his closest single DNA match from Mozambique is also most likely from a town with a Swahili community (Pemba). Although because of her mixed background the ancestral possibilities are multiple. And this will be true for these matches as well given the cosmopolitan origins of the Swahili.
His two Indian great-great grandparents who settled in Mozambique during the late 1800’s are further testimony of the rich commercial history of Mozambique. One of them being from the former Portuguese colony of Goa. All interconnected within the greater Indian Ocean area. This international trade and associated movement of people was already flourishing during the Swahili era. And also continued after the arrival of the Portuguese in the 1500’s. Although their effective rule was rather limited for the most part (restricted to coastal areas) it is still noteworthy that the Portuguese also used to have a fort in Mombasa, Kenya. Which was kept till 1698 (see map above). Therefore mutual relations between Mozambique and the Swahili coast up north have always been common. Especially for very mobile people such as the commercially orientated Swahili and Indian traders. Suggested reading:
- Port Cities and Intruders: The Swahili Coast, India, and Portugal in the Early Modern Era (M.N. Pearson, 1998)
- Indians of Mozambique and Eastern Africa (Lusotopie 2008)
2) Mozambican connection on Ancestry
Map 3 (click to enlarge)
Regional admixture estimates
Map 4 (click to enlarge)
The AncestryDNA results for my DNA cousin whose father is from Mozambique and whose mother is from São Tome & Principe have been shown already in Figure 1. His Southeast African DNA is clearly indicated by his main regions. Principally 72% “Cameroon, Congo & Southern Batu Peoples” but also 12% “Eastern Bantu”! This covers the expected 50% genetic inheritance from his father in a rather straightforward fashion. As shown in the screenshot above Mozambique is specifically mentioned as being included within “Cameroon, Congo & Southern Bantu Peoples”. Not mentioned by Ancestry itself but “Eastern Bantu” often appears as secondary region for Southeast Africans (10-20%). As I have already observed during my survey among Zambians, Zimbabweans, South Africans etc. (see this chart). “Eastern Bantu” scores are quite uncommon among Afro-Diasporans from what I have seen (see this chart). But when it is being reported – even as low as a trace amount of 1% – it could like wise also be traced back to Mozambique I imagine!
His mother is from São Tomé & Principe which has rather divers African origins. But Central African (Angola, Congo etc.) lineage is also very plausible. Which most likely contributed to the 72% “Cameroon, Congo & Southern Bantu” score. Another intriguing aspect is that she also has either a Cape Verdean-born mother or grandmother. Who migrated to São Tomé as contract labourer in the early 1900’s. This Cape Verde connection is revealed by a very distinctive score of 6% “Senegal”. Probably also to be combined with the “Portugal” & “Spain” scores. Highlighting that regional admixture DOES matter!
Ethnicity estimates are often dismissed as being just entertainment. MyHeritage certainly seems to play this angle when providing their own take on so-called reveal videos. Complete with fittingly dramatic music soundtracks even 😉 But when performed adequately and also given proper understanding as well as enhanced tools regional admixture can be serious business! Just imagine for example if this person had been adopted then surely these results would be very illuminating! Even more so when combined with his DNA matches (see map 3 and below). As it happens this person is already aware of his main origins, going back 2 or 3 generations. But he was not sure yet about the exact details. Seeing his regional breakdown really helps to put things in perspective. Distinguishing major parts of his African roots from minor ones. Even when those smaller scores such as “Senegal” as well as “Eastern Bantu” can still be very useful for pointing towards distinctive lineage! See also:
- Ancestry’s 2019 Update: Back on Track Again?
- 100 Cape Verdean AncestryDNA results (2013-2018)
- Cesária Évora “Sâo Tomé na Equador” (Youtube video)
DNA Matching Patterns
Table 2 (click to enlarge)
Because my DNA cousin’s total number of DNA matches on Ancestry is quite large (n=3632) I was of course not able to perform a manual scan of all his matches. Instead I applied my automated scanning & filtering method to zoom into African profiles. Which I then checked mainly for plausible surnames and associated regional admixture. As well as any other useful clues provided by their profile details. The other groups of matches were singled out by various text filters. And afterwards again counter-checked for plausible admixture & surnames. Obviously 100% accuracy is not guaranteed as this was done on a best effort basis in order to get a general idea.8 See also:
In the overview above as well as in Map 3 a dazzling array of African countries is being shown. But when you focus on the bigger and therefore closer matches more clarity is obtained. A true stand-out feature being the amazingly high number of Cape Verdean DNA matches! No less than 63 ones are close matches (shared DNA >20cM). I have actually taken a very conservative approach in my guesstimate of at least 539 Cape Verdean descended matches. Because I would not be surprised if he actually has more than a thousand Cape Verde-related DNA matches! If my DNA cousin had not known in advance about his Cape Verdean connection from either 2 or 3 generations ago than this outcome would certainly have confirmed it for him. With his regional admixture results providing additional perspective on ancestral proportions. Intriguingly I also found one quite likely Fula match from possibly Guinea Conakry which is probably to be associated with his Cape Verdean side as well.
My DNA cousin only obtained 4 Mozambican DNA matches. And not a single one that I could detect from São Tomé & Principe itself. This seems in contradiction of the fact that actually Mozambique is the biggest source of his African roots. Because that is where his father is from. This outcome is however merely caused by a skewed database effect. DNA testers from both Mozambique and São Tomé & Principe still being very rare within Ancestry’s customer database. Probably because Luso-phone Africans (except Cape Verdeans!) do not have a strong migrant presence in the West (except Portugal). Unlike English speaking Africans who are quite numerous in the USA (see this graph) and also in Canada and the UK. Then again it is very interesting to see the two smaller matches from Cameroon and also one from Nigeria. As most likely these are inherited by way of his strictly São Tomense side.
The number of Portuguese matches is also greatly disproportionate (n=181). Probably the second most numerous group of matches, going by nationality. Although tellingly no close matches among them. Backed up by only a 8% combined score for “Portugal” and “Spain”. I greatly suspect that this genetic component is inherited by way of his Cape Verdean connection and not due to any recent Portuguese ancestry. As was the case for the Dutch guy on MyHeritage. The Latin American matches are even more numerous but also spread across several countries. Surprisingly the biggest match among them (21cM) is with a Mexican and not a Brazilian as you might expect. Usually shared Iberian ancestry will have lead to these matches but several other ancestral scenarios might apply as well.
Indian Ocean slave trade connections are very apparent too. In particular the 23 South African Coloured matches. As well as matches from most likely Madagascar, Mauritius, Réunion and the Seychelles. Very special also the rather big match (13cM!) from Saint Helena (4 grandparents confirmed). As this island is known to have harboured so-called Liberated Africans during the mid 1800’s. Illegal slaveships departing from Mozambique were frequently rerouted to this island after having been intercepted by the British fleet. Intriguingly DNA testing is shedding ever more light on this episode. Furthermore also a Mozambican connection among Middle Easterners seems to be revealed. Through at least 19 DNA matches from various countries across the region (possibly incl. not only Gulf Arab states but also Palestine, Iran, Egypt and even Cyprus!). Compare also with the one single Turkish match I found on MyHeritage. This match was ethnically Turkish but still had a detectable amount of 0,8% African admixture (see table 1).
- Slavery on Saint Helena
- What DNA reveals about St Helena’s freed slaves
- Slave trade and slavery on the Swahili coast (1500-1750) (T. Vernet, 2009)
- Makoa (or Masombika) an ethnic group in Madagascar (Wikipedia)
Migrations across Southeast Africa & Population matches (IBP)
Map 5 (click to enlarge)
Reviewing only the relevant Southeast African DNA matches in table 2 and map 3 one needs to be aware of both ancient Bantu population movements as well as more recent historical migrations across the region. This will help make greater sense of the matching patterns on display. Aside from the skewed customer data base effect which explains the relative scarcity of Mozambican matches. It can be seen that South Africa Bantu speakers and Zimbabweans are providing most of the matches. But smaller matches actually are to be found as far north as Uganda and also to the west into Zambia and Botswana.
Given historical plausibility I greatly suspect that these smaller matches (6-7cM) might mostly be false positives (IBS) or also population matches (IBP, see this chart). Basically ancient shared origins lead to a great degree of genetic overlap in neighbouring and/or ethno-linguistically related populations (such as Bantu speakers). Which can result in identical DNA segments being detected. This circumstance operates to create DNA matches from unexpected and seemingly exotic places! Even when technically speaking you are not related to such IBP matches in a genealogically meaningful time frame (let’s say 500 years).
Then again the recent regional history of Southeast Africa does offer plentiful opportunity which may also explain these border crossing DNA matches. Especially the bigger matches as reported at times for South Africans and Zimbabweans. But also 1 Kenyan and 1 person from Lesotho in fact (see column “above 10cM” in table 2). People living within the modernday state of Mozambique have been migrating to neighbouring countries and vice-versa for many centuries. Often eventually intermingling with local people. The above map shows the impact of the so-called Mfecane migrations which lead to the Gaza kingdom in southern Mozambique. But several other historical migrations have been documented as well, associated with:
- Swahili trade up the coast into Tanzania and Kenya
- Maravi kingdom, centered in presentday Malawi, but stretching across northern Mozambique and also into Zambia
- Portuguese gold fairs in Shonaland during the 1600’s/1700’s (formerly Monomotapa kingdom)
- Mozambican labourers in South African gold mines, already during the late 1800’s!
I do not have any further ethnic details about my DNA match’s father. But I do know he was born in Maputo, the capital of Mozambique. Located all the way south. Practically at the border of South Africa. And so especially the matches to the southwest of Mozambique are not surprising at all. But rather to be expected. As always one does very carefully need to assess the actual direction of geneflow in each separate case. Because the shared ancestry might be due to either outgoing or incoming migrations and subsequent inter-ethnic unions! In many cases you might not be able to find the actual MRCA (Most Recent Common Ancestor). Still then it will be useful to focus on the main matching patterns (indicated both by frequency and closer matches). Rather than being distracted by “exotic” or puzzling findings. Suggested reading on Mozambique’s recent history:
- A History of Mozambique (M. Newitt, 1995)
- Southeast African section (featuring maps taken from Newitt 1995)
3) Comparing MyHeritage with Ancestry
This will just be a short overview of my first experiences with MyHeritage and how I think it compares with Ancestry. Obviously my impressions may change with further exposure. At this moment I am seeing three main implications for Afro-Diasporans:
- MyHeritage is distorting Central & Southeast African DNA
- Ancestry offers greater potential to find African matches
- MyHeritage provides more advanced filtering tools for DNA matches + a chromosome browser
Central & Southeast African DNA is not well-described on MyHeritage
Figure 3 (click to enlarge)
Figure 4 (click to enlarge)
What is about to follow is based on my assessment of MyHeritage’s current Ethnicity Estimate as reported in 2019. To be fair MyHeritage has recently announced a major update of this tool. So improvement may very well be on its way soon! My comparison also being based on the currently updated 2019 version of AncestryDNA. See also:
In order to judge the accuracy of any particular DNA test I personally always find it best to look at the results of people whose origins can be verified independently. Furthermore it will also be insightful to look at which Reference Populations are being used to compare your own DNA against. And whenever possible also learn more about the algorithm used to calculate their ethnicity estimates. I have been aware for quite some time already how MyHeritage tends to report greatly inflated East African scores for Afro-descendants. And also in other aspects their African breakdown looks quite messy to me. Going against all historical plausibility. And so frankly up till now I did not find it worth my effort to get deeper involved.9
The two results shown above are illustrating my point about Central & Southeast African African DNA being greatly misrepresented on MyHeritage. Notice how “Central African” is very low for a confirmed Congolese person (from Kasai). It is more noticeable for the South African match of the Dutch guy on MyHeritage. However even for this South African Bantu speaker “West African” is showing up with a higher amount. While the so-called “East African” category is very prominent for both the Congolese and the South African. Misleadingly also specified into “Kenya”. Without any further clues given you would never have guessed their real background!
As always the labeling of ancestral categories is not to be taken too literally. But these flawed outcomes are mainly due to the fact that currently MyHeritage provides an incoherent and poorly designed African breakdown. If they get it so wrong for people of confirmed Central & Southeast African background it should not be that hard to imagine how reliable their estimates will be for Afro-descendants with more complex backgrounds! This goes also for the Dutch guy with one Mozambican parent.
I have not been able to find out which exact African reference samples are currently used by MyHeritage. Also I am not aware of any detailed White Paper dealing with the methodology used for their Ethnicity Estimate. However their website does offer some helpful clues. First of all currently there are 10 African regions on offer. Which compares favourably with the current 11 regions on offer on Ancestry. But in contrast with Ancestry’s finer resolution for West & Central Africa (K=6) MyHeritage’s breakdown is heavily skewed to North and Northeast Africa with no less than 6 regions! See also this link on MyHeritage’s website:
- North African
- Sephardic Jewish – North African
- Ethiopian Jewish
- West African
- Sierra Leonean
- Central African
This circumstance is probably the main culprit for all the unexpected “East African” scores reported for Trans-Atlantic Afro-descendants. In practically all cases these will merely be a reflection of Bantu lineage from either Central or Southeast Africa. Which is not properly detected by MyHeritage because of lacking reference samples. Such as from Congo or even Cameroon which produce much more appropriate and useful results on Ancestry (see this page). MyHeritage does have a separate “Central African” region in place. But it seems to me that most likely they are using Pygmy and/or Khoi-San samples for it. In other words “Central African” looks like an equivalent of the “Hunter-Gatherer” regions on both Ancestry and 23andme.
Again I have no confirmation for this because MyHeritage is not very transparent about its Reference Panel. But judging from the way an actual Congolese and South African are being described I do think this is most likely. MyHeritage in fact also specifically mentions “Central African pygmies” in their regional description. As I have said before I personally do not see much added value in such categories based on marginal Hunter-Gatherers samples. They are generally speaking not useful to understand the main genealogical African roots for Afro-descended people (South Africans excepted). And especially not when this is combined with misleading labeling and misconstrued so-called “East African” scores…Quite comparable actually to the situation on 23andme before its 2019 update:
Ancestry is the place to be for finding African DNA matches
Table 4 (click to enlarge)
As described in this blog post I have performed extensive scanning for African DNA matches for two persons who each have 1 Mozambican parent. One of whom tested on MyHeritage and the other tested on Ancestry. The outcomes were quite different as I found 97 African DNA matches on Ancestry versus 12 on MyHeritage. This is excluding a far greater number of Cape Verdean DNA matches on Ancestry (at least around 500). While actually on MyHeritage most of the 12 matches were racially mixed and only 2 were “100% African”. On Ancestry the number of genetically “100% African” matches was 67. And in fact I did also find a few recently mixed African DNA matches but I chose not to include them in my overview.
The reason for this relative abundance of African DNA matches on Ancestry is quite simple. Ancestry has a far greater customer database (more than 15 million people). Ancestry is based in the USA which already has a big and diverse African migrant community (see this link). But from personal observation I know that also African migrants residing in the UK, Canada and even Australia are taking Ancestry tests which nowadays can be ordered from many countries. MyHeritage is actually based in Israel still its customers are overwhelmingly American and European. But it seems sofar fewer African migrants have taken the MyHeritage test.
The potential for obtaining African DNA matches therefore seems clearly greater on Ancestry. Still it is noteworthy that both on MyHeritage as well as on Ancestry the number of Mozambican DNA matches was relatively smaller than for either South Africa or also Zimbabwe and Kenya. So we can definitely conclude that the ethnic composition of DNA testing databases matters! Whether it be for reference samples to be used for admixture analysis or also for customers to get matched with. This issue should be well known by now and also under development for both companies. I should stress as well that the DNA matching patterns I found are merely a snapshot of their customer databases in 2019.
One aspect to keep in mind also is that MyHeritage applies a more stringent threshold for their DNA matches. The minimum amount of shared DNA being 8cM versus 6cM on Ancestry. Aside from their smaller customer database this will also have contributed to the relatively small total number of DNA matches on MyHeritage (n=422) versus 3632 DNA matches on Ancestry! On the one hand one might say this results in more robust and reliable DNA matches. As after all smaller DNA matches (in particular within the 6-7cM range, see this link) are often false positives or population matches. Still I find that with proper interpretation such smaller matches might still be informative. Although one must be aware they will not always be genealogically meaningful!
It is often said that admixture analysis should be taken with a grain of salt. Rightfully so as it can only provide (informed) estimates about your ancestral make-up.10 Then again valuable insights are still to be gained when these estimates are interpreted correctly. However we might also say that neither DNA matches nor admixture analysis will be 100% accurate. As in fact DNA matches can be misleading as well when there is no confirmed paper trail or when there is a lack of relevant context or when you are dealing with smaller matches (<10cM) which possibly might be false positives.
I personally always aim for maximizing informational value despite imperfections. From my experience combining your admixture results with your DNA matches will often lead to enhanced insights and complementarity. A major shortcoming of admixture analysis: pinpointing exact ethnicity, might be compensated by finding relevant DNA matches. On the other hand 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 customer database. 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!).
I hope to have demonstrated the benefits of such a multi-faceted approach for these two persons with mixed Mozambican heritage. Also including historical plausibility within my analysis. Their regional admixture results were quite ambivalent on MyHeritage but fairly accurate and useful already on Ancestry. The additional DNA matching patterns I have described for them featured a disproportionate number of Portuguese and Cape Verdean DNA matches. Correctly indicating recent ancestral ties with both Portugal and Cape Verde. But without having their regional admixture results available it would be very easy to underestimate their Mozambican connection. Then again I did find some associated DNA matches also for Mozambique. But far fewer in number. While actually African DNA matches from surrounding countries especially South Africa were more numerous overall.
Advanced filtering of DNA matches on MyHeritage
Figure 5 (click to enlarge)
One very pleasant discovery for me on MyHeritage was its advanced filtering & sorting tools for DNA matches. Useful to zoom into shared ethnicity especially! But also family tree details, segment size etc. Furthermore MyHeritage also offers a chromosome browser. And there are several other helpful tools and options as well. Such as their innovative Autocluster tool. Most of these tools are lacking on Ancestry! And therefore certainly giving MyHeritage an edge on this terrain. Just to give a quick overview what I personally found striking:
- MyHeritage allows for filtering matches on ethnicity while Ancestry seems to have actually discontinued its former option of filtering on “migrations”! (see this link)
- MyHeritage enables a download of all your DNA matches in an Excel file. While on Ancestry I am obliged to use DNAGEDcom for a full scan and download of all DNA matches. And only afterwards am I able to perform any filtering or sorting on ethnicity or admixture (see this link for my tutorial).
- Both Ancestry & MyHeritage do now provide the ethnicity estimates of your DNA matches. Which can be very useful to ascertain their background (when not given). However on Ancestry these overviews are not always complete due to optional privacy settings.
- Unlike Ancestry MyHeritage also provides the exact spot of your shared DNA segment(s) with your DNA match. And this can also be verified in their chromosome browser. Although presently the actual admixture of these shared segments is not given if I am not mistaken. Which would be quite crucial especially for mixed people to determine on which side a DNA match might be related.
I have not really experimented that much with these tools aside from filtering on ethnicity. Which again I found very useful! For anyone who has tested with MyHeritage it is much recommended to make more use of these tools! Below some links to get more acquainted with MyHeritage’s advanced tools:
- Learn about MyHeritage products
- New Filtering System for DNA Matches
- How to Use Chromosome Browsers for Genealogy
In conclusion it is essential to be fully informed about both strengths and weaknesses for each separate aspect of DNA testing. As well as to judge each DNA testing company on its own merits. Correlating (regional) admixture analysis with DNA matches can be rewarding and reinforcing in many cases. I find it very encouraging that MyHeritage is also adopting this approach. Both aspects of genetic genealogy obviously have their own shortcomings. This certainly is true for the two Mozambique related cases I have discussed in this blog post. Exploring a Mozambique connection by way of DNA testing is quite tricky for people who are already aware of having confirmed Mozambican lineage. Let alone for Afro-descendants who can only speculate about it!
But I find it important to stay positive and focus on what ever informational value you can obtain despite imperfections. Instead of taking an overtly dismissive stance. Admixture results can reveal insightful ancestral connections even when DNA matches are not (yet) available and vice versa. Regional admixture may also enable the identification of a plausible background of ones DNA matches. Your admixture results can be helpful as well to gauge how your DNA matches relate with actual ancestral proportions. While DNA matches may serve to corroborate distinctive ethnic/regional lineage detected by your admixture results. As always it pays to use scrutiny and discretion in stead of jumping to conclusions or putting all your eggs in just one basket!
1) There is also an additional reason why this Mozambican connection has proven to be quite symbolic to me. I got my first DNA test results almost ten years ago. In 2010 through 23andme. Initially I was merely told that I had an x amount of African ancestry. Without further specification. Of course I already knew I had African DNA as I am of Cape Verdean descent with Cape Verde being a West African island group (see also “What Tribe Am I?” ). I was also told that my maternal lineage (L3e4) originated in Mozambique out of all places! An African country which is probably furthest removed from Cape Verde geographically speaking. Located right at the other end of the continent. Thousands of miles away from Cape Verde. With little known documented ancestral connections between the two countries besides both having been part of the Portuguese colonial empire. Safe to say I was confused and clueless about this one single concrete hint provided about my African heritage!
Since then I have luckily discovered that information provided by DNA testing companies should always be scrutinized carefully. Correct interpretation is everything 😉 From reading peer-reviewed scientific studies I learnt that L3e4 is actually most frequently reported for people across the wider Upper Guinea region (Senegal-Sierra Leone). Which makes perfect sense as this is the practically exclusive region of African provenance for Cape Verdeans (see this website). And accordingly L3e4 is therefore also very common among Cape Verdeans themselves (see this paper). Furthermore by educating myself about the basics of haplogroups I learnt that this part of our DNA is actually very minimal. And that ancestral origins which go back thousands of years are usually not genealogically relevant at all. Sadly there are still some widely held misconceptions about haplogroups. For more info:
2) According to some people only continental admixture is to be taken seriously in DNA testing. Sub-continental admixture, a.k.a. ethnicity estimates, a.k.a. regional admixture only being fit for entertainment purposes. I myself have never taken this stance. Preferring to judge each case on its own merits. Attempting to maximize informational value despite imperfections and avoiding source snobbery. Which is why I have conducted my AncestryDNA surveys among Africans and Afro-descendants in the past. Applying an additional macro-regional framework for extra insight (Upper Guinea, Lower Guinea, Central/Southeast Africa).
Basically combining interrelated & neighbouring genetic regions in order to allow certain regional patterns to show up more clearly. This turned out to be particularly helpful when wanting to explore any rough correlation between aggregated AncestryDNA results for Afro-Diasporans and slave trade patterns. Which I indeed established for the most part. See this post below for a summary of how my Afro-Diasporan findings (2013-2018) more or less fall in line with historical plausibility.
The Ethnicity Estimate currently provided by MyHeritage has many irreconcilable flaws. And frankly I think it will remain misleading and confusing even after having taken into account wider regional patterns and ancient population migrations. Still a macro-regional approach does help to see things more clearly. For example within the European breakdown it is obvious that “South Europe” is predominant while “Northwest Europe” is minimal. Which is in line with Portuguese genetics. The African breakdown is more messy and frankly beyond redemption. But still it is noteworthy that “East African” is the main component.
3) I actually did find one single match from France who appeared to be partially Jewish. Her “Sephardi Jewish” score being 28,5%. However this category was not a shared ethnicity. Furthermore from my observation scanning other people’s DNA matches who have some minor Jewish admixture you should practically be guaranteed of receiving a multitude of Jewish matches. Depending on which company, dozens, hundreds or even thousands. Even when only backed up by 1% “Jewish” admixture. This is because of the known distinctiveness of Jewish genetics. Characterized by strong endogamy which results in many shared DNA segments, even when not closely related. And also because Jewish DNA testers tend to be overrepresented in the customer databases of DNA testing companies.
4) MyHeritage enables a download of all your DNA matches in an Excel file. I used this as my starting point and also to make all the needed calculations to produce table 1. While analyzing the DNA matches I have always taken a cautious research approach when deciding on their most likely background. Going by any clues given but in particular: 1) plausible regional admixture and 2) plausible surnames. Whenever available I also took into consideration actual birth location, public family trees and other relevant profile details. Such details are actually abundantly available on MyHeritage. Making such an endeavour much easier. Still I would like to underline that my categorization of DNA matches being African, Portuguese, Brazilian, Latin American etc. is not intended to be waterproof. Obviously I did not have perfect and complete information at hand for each single DNA match. Instead of aiming for 100% accuracy I processed the data on a best effort basis to get a general idea.
I mostly relied on the following decision rules:
- African matches: either 100% African admixture or distinctive amounts for “East Africa”. To be combined with birth locations in Africa and/or associated surnames.
- Portuguese matches: predominant “South European” & “Iberian” scores; absence of any “Native American”. To be combined with birth locations in Portugal and/or associated surnames.
- Brazilian matches: predominant “South European” & “Iberian” scores within the European breakdown; Also containing African and/or Native American admixture to any degree. To be combined with birth locations in Brazil and/or associated surnames.
- Hispanic matches: predominant “South European” & “Iberian” scores within the European breakdown. Also containing African and/or Native American admixture to any degree. To be combined with birth locations in Latin America and associated surnames being distinctly Spanish and not Portuguese.
- Northwest European-American matches: predominant “Northwest European” scores; absence of any non-European admixture and also any “Iberian”. To be combined with birth locations in the USA and surnames being in line with a Northwest European background. I specifically looked for any Portuguese surnames in family trees when available to rule out any partial Portuguese descent
- Anglo-Caribbean & African American matches: predominant “Northwest European” scores within the European breakdown; Otherwise mainly of African descent. To be combined with birth locations in the West Indies or USA and associated surnames. I specifically looked for any Portuguese surnames in family trees when available to rule out any partial Portuguese descent. Enabling me to single out 4 mixed Anglo-Caribbean/Portuguese matches.
5) Both Portugal and Brazil produced 6 close matches. Four other close matches were of confirmed partial Portuguese descent. Reviewing the public family trees of the Brazilian matches many of them actually also had very recent Portuguese ancestry. Going back one or two generations only. The one single close match from France might actually also be partially Portuguese. Although I did not find any direct clues for it on his profile. Which did mention France as birth location and also only French surnames to be found within his family tree. Still Portuguese migrants are very numerous in France and have been residing there for several generations already. For the one single close match with a confirmed Hispanic background (4 grandparents) I suppose perhaps a Galician connection might be possible as well. As they are known to have migrated to Latin America in great numbers and are closely related to northern Portuguese. And the Portuguese family from this DNA tester on MyHeritage is actually from northern Portugal.
6) Going by an associated social media profile this Mozambican DNA match might actually be living in the Netherlands right now for studies! A striking example of how your DNA matches can really be surprising. They also provide an unending source of material for serendipity lovers 😉 Her admixture results mentioned on her MyHeritage profile page were as follows: 45% “South European”; 42% “South Asian”; 12% African; 6.6% “East African/Kenyan”.
7) The Hispanic American DNA matches I found might be related in several ways. Aside from a shared Portuguese or even Mozambican ancestor in fact also a shared Spanish ancestor could still be possible as well. Also depending on the actual shared DNA amount such matches may not always be IBD (Identical by Descent). Due to being close neighbours and having similar population histories Spanish DNA is genetically very similar to Portuguese DNA. And therefore I suppose this circumstance might also result in so-called population matches (IBP, see this chart). Especially Galician origins from northern Spain and perhaps also Andalusian origins from southern Spain might often be mistaken for Portuguese and vice versa. A very extensive genetic study of the Iberian Peninsula was recently published which may be quite relevant in this regard:
- Patterns of genetic differentiation and the footprints of historical migrations in the Iberian Peninsula (2018)
8) See also foot note 4 for my methodology. A clear advantage on Ancestry being that it offers more reliable admixture indications. Because the total number of matches was much greater than on MyHeritage I only checked each and every profile of the 97 African DNA matches (excl. Cape Verde). Otherwise again I was not aiming for 100% accuracy but instead I processed the data on a best effort basis to get a general idea.
The mainland African matches (with “100% African” admixture) were obtained by way of my scanning & filtering method. I mostly relied on text filtering for the other groups of DNA matches. See also these links:
9) As I already mentioned I have not tested with MyHeritage myself because I find their current Ethnicity Estimate to be misleading for both Afro-descendants and Africans. And actually also other types of ancestry are usually not well described from what I have seen. Notice for example how confirmed Portuguese lineage is being described for the Dutch person on MyHeritage in several decomposed categories. While Ancestry has a much more straightforward and also reasonably predictive “Portugal” region on offer. Even when at times to be combined with “Spain” and even “France”.
I have seen quite a few MyHeritage results for people of all kinds of backgrounds. But generally speaking their results looked very “off” compared with their true or to be expected origins. Or also in comparison with results for the same persons on either Ancestry or 23andme. So that is why I have not bothered doing any survey of MyHeritage results so far. Frankly I also have not been impressed by the regional descriptions provided by MyHeritage. Or their additional guidance on how to interpret their results. Although I do like their additional macro-regional breakdown.
Aside from a poorly designed Reference Panel I do also suspect their algorithm up till now has been quite shaky. In their recent announcement they mention an increase of their Reference Panel (mostly based on customer samples it seems). Plus they also aim to improve on their “fancy math”. Hopefully this will also indeed result in more useful Ethnicity Estimates for Afro-descendants and Africans. But we’ll have to wait and see 😉
10) Understandably many people desire to have the most specific degree of resolution when searching for their African roots. They want to be able to pinpoint their exact ethnic origins and preferably also know the exact location of their ancestral village. In a way following in the footsteps of the still very influential ROOTS author Alex Haley. Unfortunately these are rather unrealistic expectations to have in regards to DNA testing (at least in regards to admixture analysis.). Not only given current scientific possibilities. But also because such expectations rest on widely spread misconceptions about ethnicity, genetics, genealogy as well as Afro-Diasporan history.
Too often people ignore how the melting pot concept is really nothing new but has always existed! Also in Africa where inter-ethnic mixing has usually been frequent! Throughout (pre) history and maybe even more so in the last 50 years or so. Generally speaking ethnicity is a fluid concept which is constantly being redefined across time and place.
Too often people fail to take into consideration how due to genetic recombination our DNA will never be a perfect reflection of our family tree but might actually also at times suggest very ancient migrations.
- Everyone Has Two Family Trees – A Genealogical Tree and a Genetic Tree (Genetic Genealogist, 2009)
Too often people underestimate the actual number of relocated African-born ancestors they might have (dozens or even hundreds!). As well as the inevitable ethnic blending which must have taken place across the generations.
Too often people are still not informing themselves properly about Africa itself and the documented origins of the Afro-Diaspora. Many specific details may have been lost forever but there is a wealth of solid and unbiased sources available which can help you see both the greater picture as well as zoom in more closely to your own relevant context. See also:
- Fictional Family Tree incl. African Born Ancestors (Tracing African Roots)
- DNA studies for Africans and Afro-Diasporans (Tracing African Roots)
- Documented ethnic/regional origins of the Afro-Diaspora (Tracing African Roots)
- Maps (ethnolinguistic, slave trade, various parts of Africa) (Tracing African Roots)