In this two-part blogseries I will analyze the DNA matches being reported by AncestryDNA for 50 of my Cape Verdean survey participants. A follow-up to my previous blog post about 100 Cape Verdean AncestryDNA results (see this link). Because I was kindly given access to their profiles I was able to use my scanning and filtering method of DNA matches in Excel (see this link). Aside from matches with mainland Africans I am also including matches with people of (presumably) fully Portuguese, Jewish, West Asian and South Asian descent.1 Below a statistical overview of my main findings. Going by group averages. For the individual results which do display greater variation follow this link:
Table 1 (click to enlarge)
Table 2 (click to enlarge)
This project was merely intended as an exploratory exercise. Of course my research findings have limitations in several regards. And therefore they should be interpreted carefully in order not to jump to premature or even misleading conclusions. Still I do believe they can reveal relevant tendencies in DNA matching for Cape Verdeans in general. These outcomes may also provide valuable insight into the various ancestral components found within Cape Verdean DNA. In particular when aiming for complementarity by also taking in to account admixture analysis, genealogy and relevant historical context.
Below an overview of the topics I will cover in this blog post:
- Considerations when dealing with DNA matches
- Upper Guinean matches: as expected African matches (south of Sahara) were overwhelmingly from Upper Guinea (Senegal-Sierra Leone): 88% of the total. In line with the 92% Upper Guinean admixture proportion (“Senegal” + “Mali” / total African) I found for my survey group.
- North African matches: fairly consistent despite minimal shared DNA
- Other African matches: unexpected & uncommon. Higher odds of false positives but in some cases to be corroborated by additional clues, such as AncestryDNA’s ethnicity estimates?
- Methodology: how I filtered the African DNA matches as well as the decision rules I applied when determining a plausible background for each DNA match.
Part 2 of this blogseries will have the following topics:
- Portuguese matches: omnipresent and clearly most numerous as well as often hinting at relatively recent ancestral ties (1800’s-1900’s).
- Jewish matches: Sephardi matches more likely to be truly genealogical than Ashkenazi matches?
- West Asian matches: quite rare, possibly indicating that West Asian admixture among Cape Verdeans is generally indicative of actual North African or Sephardi lineage.
- South Asian matches: also rare, but on a hit and miss basis still sometimes already seemingly validating trace amounts of South Asian admixture.
- Inter-island matching patterns: illustrated by the distribution of the shared DNA segments between myself and my 100 Cape Verdean survey participants.
- Methodology: how I filtered the non-African DNA matches as well as the decision rules I applied when determining a plausible background for each DNA match.
Dedicated to all my Cape Verdean primos and primas participating in this survey.2 And special dedication to my newly born nephew Max!
Whenever possible DNA matches should of course always be placed in their proper genealogical & historical context. The banner displayed above belongs to a pioneering blogger specializing in this field for Cape Verdeans: the Creola Genealogist. She recently organized an excellent conference on Cape Verde DNA and Genealogy (see this link). For helpful resources on Cape Verdean genealogy & history see also:
- Barros & Brito Online Family Tree (searchable and very extensive!)
- Cape Verde, Catholic Church Records, 1787-1957 (FamilySearch)
- Cape Verde DNA, Inc (Facebook: the biggest online community of Cape Verdean Genealogy & DNA enthusiasts!)
- Cape Verde DNA, Inc (website)
- Cape Verdean Genealogical Resources (by James J. Lopes)
- Cape Verdean Historical Resources (Cabo Verde Raizes na Africa)
- Cape Verde History Unearthed (blog by one of my closest DNA cousins)
- Genealogia da Familia Barbosa (by way of Web Archive)
- Historical visits to the Cape Verde Islands + Chronology (Heuijerjans.net)
- The Creola Genealogist (blog)
While analyzing the DNA matches of my Cape Verdean survey participants I have always taken a cautious stance when deciding on their most likely background. Going by any clues given but in particular: plausible regional admixture combinations and plausible surnames. As well as actual birth locations and other relevant profile details (whenever available). My methodology is described in more detail in the last section of this blog post. But already I would like to underline that my categorization of DNA matches being African, Portuguese, Jewish etc. is not intended to be waterproof. Obviously I did not have perfect and complete information at hand for each single DNA match. While the number of DNA matches to be analyzed often remained quite high even after automated filtering and sorting in Excel. This goes especially for the Portuguese and Jewish matches. Instead of aiming for 100% accuracy I therefore processed the data on a best effort basis to get a general idea.
Other considerations to keep in mind when dealing with DNA matches3:
Table 3 (click to enlarge)
“Identical by Descent (IBD) refers to a segment that you share with someone from a common ancestor within the genealogical timeline.
Identical by State (IBS) is used to describe a variety of scenarios where the segment cannot be tracked to a common ancestor, including IBC and IBP.
Identical by Chance (IBC) describes a segment where an accident of recombination means that a segment matches, but this is not a genuine match.
Identical by Population (IBP) refers to segments that are widely shared among populations, to the extent that they are not genealogically useful – e.g. in endogamous populations. ” (source: DNA Painter Glossary)
The main thing to consider for anyone who is reviewing their DNA matches is establishing whether your DNA matches are indeed genuine or “Identical By Descent” (IBD) and not just random “Identical By State” (IBS) matches. Getting your parents tested or any other relatives of older generations will increase your ability to be more certain about the genetic inheritance of any given reported DNA match. As several of my survey participants also shared the results of their parents I was actually able to perform such a IBD check in a few cases (see section 4, Table 11). Socalled population matches (IBP) might be most tricky to distinguish. But also the size of the shared DNA segment will be indicative to find out if your DNA match could in fact be a socalled false positive. It is well advised to carefully read the Predicted Relationship Info given by Ancestry and especially the confidence score they will assign to your DNA matches.
As shown above in Table 3 increased chances of false positives exist in particular for smaller matches with a shared DNA amount of in between 6 cM – 16 cM. Other DNA testing companies tend to apply similar criteria. Although with some differences in actual thresholds. For example on 23andme the cut-off for DNA matches being reported starts at 7 cM and not 6 cM. Also a shared DNA amount of 10 cM and greater is often considered quite reliable already. With an estimated IBD probability higher than 50%, unlike what is stated in table 3. During my survey I mostly came across smaller DNA matches however. While DNA matches of around 10 cM let alone greater than 20 cM (“close matches” in table 1) were much less frequent. These latter matches would naturally be indicative of the most robust matching patterns.
Despite a lower confidence score I do believe that these smaller matches may still also be informative. Even when I suspect that especially matches of in between 6 cM – 7 cM are usually population matches (IBP)4 or just random matches (IBC). And again such matches do call for very careful scrutiny. In many cases smaller matches may indeed not be genealogically meaningful. But they can provide extra historical perspective when reported in distinctive frequencies! Cape Verde has been settled since at least the late 1400’s and a blending of people from various backgrounds took place right from the start. Resulting in the early formation of a new ethnic group of locally born (=crioulo) Cape Verdeans. Therefore the genetic inheritance of distinctive ancestors from Upper Guinea, Portugal and elsewhere is bound to be greatly diluted and fragmented for the most part due to recombination. Which would be in line with smaller sized DNA matches. The larger sized matches (>10cM) being suggestive rather of relatively recent admixture taking place in the 1800’s/1900’s. For more discussion see:
- Identical by Descent Matches (ISOGG)
- How Phasing Works and Determining IBD Versus IBS Matches (DNAeXplained)
- The Danger of Distant Matches (The Genetic Genealogist)
Ethnic composition of Ancestry’s customer database matters!
Another aspect to keep in mind is that my survey findings are based on the scanning of DNA matches I performed in July-August 2018. Therefore the matching patterns I am describing in this blog post are representing a snapshot of Ancestry’s customer database in the summer of 2018. Because of ever increasing popularity of DNA testing it is inevitable that when performing the exact same exercise in 2019 or 2020 my survey findings will look quite different. Especially quantitatively speaking. Then again I imagine that quality-wise most of the matching patterns would still broadly be the same.
According to Ancestry’s own information a staggering 10 million DNA tested people are included in their customer database already! Easily making Ancestry the largest DNA testing company in the world (see this link or also this one). In itself this provides an unprecedented platform to establish statistically meaningful matching patterns for my survey group. Basically speaking 50 Cape Verdeans are being compared for shared DNA segments with 10 million other DNA tested people! Even though individual variation will still be relevant and not all Cape Verdean island origins are equally represented in my survey group the wider implications of such an undertaking are already quite significant I would say.
Then again the composition of Ancestry’s customer database did greatly influence the outcomes. Because Ancestry is an USA based company it is only natural its customers are also overwhelmingly American and mostly of European descent. Luckily the USA is home to many 1st or 2nd generation migrants from all over the world who also take DNA tests. Which still enables a fairly global representation within its database. Plus lately Ancestry has also expanded in to foreign markets where likewise also people of migrant descent are taking DNA tests. Nonetheless because of this skewed database factor for Afro-Diasporans in general there has always been an inherent bias towards being matched with white Americans, rather than actual Africans. This goes not only for Ancestry but also for other DNA testing companies. Especially in the early days (2010-2017) when DNA testing was not yet as popular as it now is among minorities. While in particular African DNA testers were very rare. For more details see:
- The ethnic breakdown of 23andMe customers in 2011 (Gene Expression)
- 2017 was the year consumer DNA testing blew up (Technology Review)
- DNA databases are too white. This man aims to fix that (Technology Review)
Specifically for Cape Verdeans therefore the chances of finding Portuguese and Jewish matches are much higher than finding African and Asian matches.5 Which is clearly reflected in the main overview of my findings in Table 1. Still while performing my survey I was pleasantly surprised to see that the number of African matches is steadily on the increase when compared with previous years. As ever more Africans are also starting to take DNA tests, incl. even a few persons from Guiné Bissau!
Your matches may be related to you in unexpected or intricate ways
Map 1 (click to enlarge)
Perhaps redundant but it might still be worth reminding that the MRCA‘s (most recent common ancestors) shared between my survey participants and their matches will not per se have been of the same background as their matches themselves. Especially due to ongoing migrations there are usually several possible ancestral scenario’s to consider when you get “matched” with someone. Assumptions about the direction of gene flow may be proven wrong after follow-up research. Context is everything and historical plausibility combined with solid genealogical research should be leading instead of wishful thinking.
For Cape Verdeans in particular the practically global Portuguese presence throughout modern history (1500’s and onwards) and the resulting gene flow among local populations in many parts of the world is a factor of paramount importance. This circumstance may very well explain most of the Latin American as well as West Indian DNA matches being reported for Cape Verdeans. Shared Portuguese MRCA’s then being very likely because of resp. Portuguese colonial settlers (also in the Hispanic Americas!) and Portuguese/Madeiran contract labourers.
The Portuguese in conjunction with the Spaniards however also set in motion the Trans Atlantic Slave Trade. In which Cape Verde initially played a major role as one of the first transit hubs during the 1500’s & early 1600’s (see this link). This arguably has lead to a major Upper Guinean founding effect for many Hispanic populations (see this link). Therefore shared Upper Guinean roots may of course also be a valid possibility. The Trans Atlantic Slave Trade did result in the transfer of numerous Upper Guinean people into North America and the (English/French speaking) West Indies as well (even though without any direct Portuguese involvement). So again such a scenario of shared Upper Guinean lineage could be relevant.6
Also in other ways the Portuguese colonial empire caused either voluntary or involuntary movements of people (sailors, traders, soldiers, officials, personal servants, slaves destined for Portugal/Spain etc.) of various backgrounds from all corners of the world, incl. Asia and Brazil. Often also including people of mixed Portuguese background. Because of Cape Verde’s strategic location in trans-oceanic shipping routes, many of these individuals could have made a temporary stop-over or remained for a longer time/permanently on one of the Cape Verde islands. Engaging in either fleeting or more lasting relationships with native Cape Verdeans which lead to offspring.
Obviously the worldwide Jewish migrations and their highly endogamous genetics are to be kept in mind as well when contemplating the occurrence of either Ashkenazi or Sephardi matches. As will be discussed in greater detail in the second part of this blogseries. And in fact (free) Cape Verdeans themselves have had an impressive Diaspora too! Dating from several centuries ago and leaving mixed offspring in various places of the world. See also these previous blog posts of mine:
- Cape Verdean Diaspora or Afro-Diapora?
- Shared Upper Guinean roots between Cape Verdeans and Latin Americans
- Do Cape Verdeans and African Americans share African ethnic roots?
Combine DNA matches with admixture analysis for more insight
Table 4 (click to enlarge)
It is often said that admixture analysis should be taken with a grain of salt.7 Rightfully so as it can only provide (informed) estimates about your ancestral make-up. 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!). This is why I have chosen to also take admixture levels into consideration during this survey of DNA matches.8
Going by the correlation coefficients shown in Table 4 a fair degree of correlation exists between the number of Portuguese, Jewish and South Asian matches with levels of associated admixture. But less so for African matches/admixture as well as North African or West Asian matches/admixture. In other words if for example one of my Cape Verdean survey participants happened to have an above average degree of African admixture this did not per translate into also having an above average number of African matches. But people with above average Jewish admixture did often clearly receive a greater number of Jewish matches. I will discuss these outcomes in greater detail in following sections as well as part two of this blogseries. But based on my current findings I can already say that admixture results and DNA matches can be of mutually reinforcing value indeed. For example during my survey I was often able to corroborate trace amounts of admixture by finding associated DNA matches. In particular for Jewish, North African, West & South Asian ancestry.9
2) Upper Guinean matches
Table 5 (click to enlarge)
- Spreadsheet of all African matches
- Spreadsheet of individual results sorted on number of African matches
The Upper Guinean matches I was able to find for my Cape Verdean survey participants are featured in full detail in table 5 as well as within the online spreadsheets linked above. The total number (227) and scope of these Upper Guinean matches surpassed my prior expectations! Before I started this project I was assuming that the odds of finding Upper Guinean matches in particular would be quite small. First of all because of a lack of DNA testers from this particular part of Africa (see this page for maps). Luckily the increasing popularity of DNA testing among African migrants in North America, Europe and even Australia is clearly noticeable now. And it is having a really beneficial impact!10 Only for two survey participants I was not able to find any apparent African (south of the Sahara) matches. Although both of these persons (CV11 & CV50) did actually have North African matches.
The second reason for my reservations beforehand had to do with my assumption that most mainland African ancestors for Cape Verdeans are likely to be traced back quite far back in time: the 1500’s & 1600’s. In line with Cape Verde’s distinctive demographic history which includes having a population overwhelmingly consisting of black/mulatto people with freed status already in the early 1700’s if not late 1600’s. Slavery did continue up till 1878 for a minor part of the population. However the resulting gene flow from mainland Africa must have been much more subdued in later time periods, on average. Given that the enslaved portion of Cape Verde’s population was below 10% throughout the 1800’s and below 20% throughout the 1700’s. For more details see:
- From Creole to African (Tracing African Roots)
- Historical demography of Cape Verde (Cabo Verde Raizes na Africa)
I suspected that this circumstance would generally lead to heavily diluted shared DNA segments with any Upper Guinean matches. Which might then prevent a detection above a certain minimum threshold. However from my actual findings the average DNA amount shared with Upper Guinean matches is rather robust already (8.04 cM). It holds up quite nicely in fact with the average I calculated for the Portuguese matches (8.22 cM, see table 1). While for the other matches (Jewish/North African/Asian) a more subdued average of around 7 cM was obtained. Increasing the odds of either false positives or population matches.
The share of matches greater than 8 cM might also be an indicator of more robust matching patterns. In table 5 we can see that this proportion of larger amounts of shared DNA is especially pronounced for Upper Guinean related matches (80/227 = 35%). But much less so for North African matches (23/180=13%). While also for other parts of Africa matches greater than 8 cM were unusual (5/30 = 17%). Of course my findings are only preliminary but perhaps this outcome is already indicating how the relatively greater match strength for Upper Guinea is in line with historical plausibility. Take note also how the only close African match (>20cM) is being reported for a most likely Senegalese person!
We should then however make allowance for the possibility that especially many of these >8cM matches may be indicative of relatively late African lineage from the 1700’s & 1800’s. Cape Verdeans will have many different ancestors originating in mainland Africa (hundreds if not thousands when tracing back to the 1500’s! see this link). Also including some relatively recent connections. Even when again I still suspect that on average Cape Verde’s mainland African connections are mostly to be traced back to the 1500’s/1600’s. If indeed many of these bigger matches are to be traced rather to the 1800’s or late 1700’s this could be very helpful actually. As it could increase the odds of zooming into particular family lines whose generational distance with mainland African MRCA’s might be relatively small. On the other hand it might obscure the ethnic origins of African ancestors to be traced back to the earliest period (1500’s/1600’s). Although possibly some of the smallest matches I found are also referring to that period.
It is quite noteworthy also that the largest number of African DNA matches (20) was found for someone from Brava. In line with the generally subdued level of African admixture on that island this person (CV08) had a total amount of 40% African (see this spreadsheet).11 Surprising perhaps for those expecting that persons with above average African admixture would have received the largest numbers of African matches. Especially persons from the island of Santiago where African retention (both cultural and genetic) is often said to be the greatest. While also the relative importance of continued slavery into the 1800’s on Santiago was more pronounced than for other islands (along with Fogo and to a lesser degree also Boavista and Maio, see this chart based on the 1856 census).
To be sure the correlation I calculated between African admixture and number of African matches is positive, but still rather weak (0.31, see table 4). Possibly with a greater inclusion of Santiago participants in my survey group a different outcome would have been obtained. And to be sure among the 6 survey participants with 10 or more African matches there were two with Santiago island roots (CV05 & CV27). However it should also be kept in mind that probably the greater part of Santiago’s mainland African lineage is likewise to be traced back to the 1500’s/1600’s. As after all Santiago was the first Cape Verdean island to be settled. While also the first runaway slave communities (Badiu) go back to that early period! Possibly a weak correlation between African admixture/matches is suggesting that other variables are also to be taken into account. In particular genealogy leading to relatively recent mainland African ancestors from the 1800’s.
Finding enslaved ancestors within one’s family tree tends to be quite difficult for Cape Verdeans. Because as mentioned less than 10% of Cape Verde’s population was enslaved in the 1800’s (see this link). Still the Cape Verdeans who remained enslaved during the 1800’s are fairly well documented. In particular because of a slave census held in 1856. Which specified among other things whether a person was born in Cape Verde or rather on the mainland (usually only a generic Guiné was given as origin but sometimes also a more specific ethnonym!). Also after manumission a socalled “Liberto” (liberated) status was often mentioned in official records. For more details see:
- Cape Verde Slave Census of 1856 (part 1) (Tracing African Roots)
- Cape Verde Slave Census of 1856 (part 2) (Tracing African Roots)
- Ending Slavery in Cabo Verde: Between Manumission and Emancipation, 1856-1876 (L. Shabaka, 2015)
- Slavery, Freedom and the “Libertos” of Cabo Verde (The Creola Genealogist)
- The question of slavery in Cape Verde (The Creola Genealogist)
Historical Cape Verde-Upper Guinea trading nexus: Guiné de Cabo Verde
Map 2 (click to enlarge)
Other ancestral scenario’s besides slavery are however also possible to account for some of the Upper Guinean matches I found for my Cape Verdean survey participants. One should not forget that Cape Verdeans have been voluntarily migrating/travelling to Upper Guinea as well as leaving offspring there for many centuries! Some of these people remained on the mainland permanently but others also eventually returned back to Cape Verde with local wives, children and servants. Either way resulting in abundant possibilities of reversed geneflow (from Cape Verde into Upper Guinea). In most recent time periods (1800’s/1900’s) Cape Verdeans often went to live in Guiné Bissau taking jobs in Portuguese colonial administration. While also Dakar in Senegal became home to many Cape Verdeans in the 20th century. However much earlier already many socalled Lançado (often Cape Verde-born) and their Luso-African descendants were settling in official Portuguese trading posts on the mainland but actually not exclusively so. As they often preferred to operate outside of Portuguese control. Originally to be found all along the whole Upper Guinean coastline from the Senegal river in the north to presentday Freetown in Sierra Leone, as shown in the map above.12
Upper Guinean predominance in admixture results also mirrored in DNA matches?
Table 6 (click to enlarge)
In my previous blog post about 100 Cape Verdean AncestryDNA results I have already discussed the consistent predominance of regional scores for “Senegal” and “Mali”. Both indicating Upper Guinean lineage for Cape Verdeans. After Ancestry’s latest update the combined share of these two Upper Guinean regions within the total African breakdown (scaled to 100%) has even increased for Cape Verdeans! It used be around 75% as shown in table 6 and also in this table. But among my 50 Cape Verdean survey participants I have now calculated an updated Upper Guinean share (“Senegal” + “Mali”) of 92% (see this spreadsheet). Which arguably is in greater alignment with Cape Verde’s overwhelmingly Upper Guinean documented roots. This outcome may serve as an illustration of how sub-continental admixture results can be of great informational value. On condition of correct interpretation! In this case facilitated by the usage of a macro-regional format.
Looking into the geographic/ethnic distribution of all the 437 African DNA matches I found for my survey participants (see table 2) one might get the wrong impression when taking it too literally. And especially without being aware of the underlying context. By applying a more suitable categorization we can establish that the proportion of Upper Guinean related matches is 88% of all African matches (south of the Sahara). That proportion being equal to 227/257. Excluding the 180 North African matches from the total to arrive at a subtotal of 257.13 And combining the 182 strictly Upper Guinean matches with the 45 presumably Hausa-Fulani matches from Nigeria/Niger who are most likely sharing Upper Guinean DNA with Cape Verdeans by way of mutual Fula ancestors.14 When seen from that perspective the African DNA matching patterns for my Cape Verdean survey group are indeed in line with their predominant historically documented Upper Guinean lineage. Which was already corroborated by admixture analysis as well! See also:
- Specifying the African ethnic origins for Cape Verdeans (Cabo Verde Raizes na Africa)
- 100 Cape Verdean AncestryDNA results (Tracing African Roots)
Fula matches over-represented due to skewed customer database?
Table 7 (click to enlarge)
I have demonstrated above that when applying a more basic macro-regional resolution (i.e. Upper Guinea) you can already derive much informational value from both admixture analysis as well as DNA matches. However when going by seemingly more specific country/ethnic labeling people might still get the wrong impression about Cape Verde’s Upper Guinean roots. Which in reality are much more coastal, incl. Guiné Bissau. Rather than from the deep interior as suggested for example by the currently much inflated “Mali” scores.15 As shown in table 7 these regional labels on Ancestry are not intended to be taken too literally. After all genetics does not care about man-made borders 😉 Prior to Ancestry’s recent update socalled “Senegal” used to be indicative of DNA from not only Senegal and Gambia but also Guiné Bissau, Guinea Conakry, western Mali and Sierra Leone! In other words serving as a reasonably proxy for ancestral connections throughout the historical area of Upper Guinea!
One of the main benefits of combining DNA matches with admixture results is the greater degree of specification you might obtain. However at least four considerations should be taken into account:
- The ethnic composition of Ancestry’s customer database is often disproportional.
- Nationality does not equate ethnicity. As many African countries are in fact multi-ethnic.
- Ethnic identity in Africa (as well as elsewhere in the world) is often fluid. Especially across the generations people will tend to have ancestors who originally came from different ethnic groups.
- The mutual ancestors shared between you and your matches may therefore be of a different ethnic background than your matches.
The high number of Fula matches (95) I came across during this project is quite striking (see table 5). These Fula matches are bound to be from various countries as the originally nomadic Fula people are pretty much the epitome of border-crossing ethnic groups. In a way they can be said to be pan-Upper Guinean. But the matches I found were mostly from Senegal and Guinea Conakry as far as I was able to verify. These Fula matches make up nearly half of all the Upper Guinean matches I analyzed for my survey participants! Even more so when adding the 45 Hausa-Fulani matches who as an eastern offshoot of the Fula people are very likely to share mutual Fula ancestors with Cape Verdeans.16
In light of Cape Verde’s varied Upper Guinean lineage this predominant frequency of Fula matches might however be seen as a bit misleading or disproportional rather. I highly suspect that this outcome is first of all reflecting a greater popularity of DNA testing among Fula people when compared with (non-Fula) people from for example Guiné Bissau who are greatly underrepresented in Ancestry’s customer database. Imagine for example that Ancestry had been an African-based company with millions of African customers! Surely you would see many more Senegambian and Guinean matches showing up then and also of various ethnic background. In line with historical plausibility and cultural retention! One must therefore be very careful not to jump to conclusions based on this preliminary overview. It has to be kept in mind that the matching patterns I am describing in this blog post are first most representing a snapshot of Ancestry’s customer database during the summer of 2018.17
The mere fact that Hausa-Fulani matches from most likely northern Nigeria are more numerous than (non-Fula) matches from actual Upper Guinean countries is already quite telling. This outcome highlights how wideranging migrations & inter-ethnic unions from the past will impact your current day matching patterns. In fact the Fula people did not only intermingle with Hausa people in Nigeria. Also in their original homeland Senegal and other parts of Upper Guinea they have mixed with other ethnic groups. Such as the Wolof, Sereer, Mandinga, Bambara etc.. Usually by way of their maternal line. Despite some degree of endogamy among more traditional and still nomadic subgroups this is especially true for sedentary subgroups such as the Halpulaar or Toucouleur from Senegal and the Fulakunda or Fula Preto from Guiné Bissau. In this way receiving a Fula match does not automatically imply that your common ancestor was also Fula! Also Mandinga, Wolof or Sereer etc. MRCA’s can very well be possible.
Despite the quite likely skewed database factor in favour of Fula matches the average shared DNA amount was highest for (non-Fula) matches from Senegal, Gambia and Sierra Leone. Especially for Senegalese matches this average was quite high (9.1 cM, see table 5). While also the only closest match (>20cM) was most likely a Senegalese person (non-Fula going by both admixture & surname). Ironically in light of the recent update and the currently inflated “Mali” amounts I only found 1 possibly Malian match. Although quite likely this is again caused by under-representation in Ancestry’s customer database in the first place. As undoubtedly also is true for the “only” 3 matches from Guiné Bissau. The rather prominent reporting of matches from Sierra Leone might surprise some people. However (northern) Sierra Leone and especially the Temne people are very much part of Cape Verde’s historically documented ancestral as well as linguistic heritage (see this link).
Having said all that of course the high frequency of Fula matches among my 50 Cape Verdean survey participants is a significant finding in itself! And in fact there is plentiful historical evidence to corroborate considerable Fula ancestry among Cape Verdeans. This goes back to the earliest settlement period of Cape Verde (1500’s) when a few Fula persons are even said to have come voluntarily to the islands. These particular ancestral connections are however not likely to be detected yet due to a built-in bias (because of the matching threshold) against ancestry which is to be traced back that far in time. Rather it seems that it is the relatively late time period of the mid 1700’s/1800’s we should be looking at. For example indicated by the maximum shared DNA obtained by Fula and Hausa-Fulani matches (15.9 cM & 15.1 cM, see table 5). And to be sure precisely in this time period the warfare between the Fula people from Futa Djallon and Fuladu and the Mandinga from the Kaabu kingdom intensified greatly. Which resulted in many captives on both sides. According to Cape Verde’s slave census of 1856 such Fula captives might have been among the most numerous at that time.
For more discussion and background information18:
- Detailed overview of various Fula subgroups across Africa (Jamtan)
- Cape Verde’s Slave census of 1856 showing Fula as most numerous together with Mandinga (Tracing African Roots)
- Maps showing geographical distribution of the Fula people (Tracing African Roots)
- Top 20 Ethnic Roots for Cape Verdeans (Cabo Verde Raizes na Africa)
3) North African matches
Table 8 (click to enlarge)
- Spreadsheet of all African matches
- Spreadsheet of individual results sorted on number North African matches
The North African matches I was able to find for my Cape Verdean survey participants are featured in full detail in table 8 as well as within the online spreadsheets linked above. Overwhelmingly these matches seem to be from the most western part of North Africa, also known as the Maghreb. The ones possibly being from either Libya or Egypt were much less common. Which would be in line with historical plausibility. They are mostly smaller matches. With 64% (116/180) of them being smaller than 7 cM. Which is a considerably higher share than the 37% (85/227) obtained for my Upper Guinean findings (see table 8). Minimal shared DNA amounts will increase the odds of such matches being false positives or rather population matches. Nonetheless the frequency of these matches is quite steady. North African matches were reported for 80% (40/50) of my survey participants. So clearly not just some fluke but rather consistent already. The highest number of North African matches being reported for one person was 16. While the biggest amount of shared DNA was 17.86 cM, reported for a person with a known Moroccan Jewish ancestor from the late 1800’s.
The total number of North African matches (180) is not that far removed from the total number of Upper Guinean matches (227). Resulting in equal group averages (4) when rounded up (see table 1). Some of my survey participants even received more North African than Upper Guinean matches! This relatively high number of North African matches is clearly in contradiction with actual ancestral proportions. Reviewing the associated admixture levels (46% for “Senegal” +”Mali” versus 0.5% for “Northern Africa”) is particularly useful in this regard to get a proper sense of proportions. I therefore suspect that North African DNA testers may be over-represented within Ancestry’s customer database when compared with Upper Guinean DNA testers. Similar to how Fula DNA testers are probably over-represented when compared with non-Fula Upper Guinean DNA testers.
On the other hand the occurrence of these North African matches is certainly not unexpected for Cape Verde! In fact several ancestral scenarios might apply. Mostly to be traced back rather far back in time. Which would be in line with the generally subdued amount of shared DNA. Most people will already be aware of the possibilities of a Mourisco connection by way of Portugal/Spain. The Moorish occupation of the Iberian Peninsula leading to some degree of North African gene flow among native Portuguese. And therefore many North African matches might actually have been inherited through Portuguese ancestors. However a direct North African connection might also be possible due to deportation of former Muslims from Portugal. As well as early Iberian slave trade along the Northwest African coast (Morocco & Mauritania) (1400’s/1500’s) (see this map).
More intricate ways of being connected to North African matches might arise because of exiled Sephardi Jews having descendants in both North Africa and Cape Verde.19 This goes for exiled Muslims too. Especially the socalled Andalusian community still being distinctive in countries such as Morocco and Tunisia. To add to complexity these people could very well have native Iberian lineage as well! Furthermore many North Africans actually also have minor but still clearly detectable amounts of Upper Guinean admixture (as described by “Senegal” and Mali” prior to the update). Mostly due to the Trans-Saharan Slave trade which partially involved some of the same people who also ended up in Cape Verde. While the Fula people may also have intermingled with the Moors of Mauritania to some degree. For more details see:
- Moroccan Jewish migration to Cape Verde (Cape Verde Jewish Heritage Project)
- “North African” percentages via Portugal, the Fula or directly from North Africa? (scroll down to section 4)
Sharp decrease in North African admixture due to Ancestry’s update
Figure 1 (click to enlarge)
During my survey Ancestry carried out an update of its Ethnicity Estimates in September 2018. Among other things this had a great impact on the reporting of minor so-called trace regions.9 Especially former “Africa North” amounts have generally been greatly reduced. Before the update the group average for “Africa North” was 2.8%. This has now decreased to 0.5% “Northern Africa”. Many people loosing their former “Africa North” scores. See also this spreadsheet which features the before & after amounts for North African admixture. The introduction of the new “Portugal” region is most likely responsible for this change. Quite likely that former “Africa North” scores are now mostly incorporated in that region.20
Either way my survey findings for North African matches do seem to highlight that trace amounts of admixture can be indicative of distinctive lineage. In my opinion dismissing small amounts of admixture as “noise” beforehand is not conducive to an open minded research strategy. Close scrutiny and independent verification are naturally always called for. However each case is to be judged on it own merits. Otherwise you run the risk of loosing out on potentially very useful discoveries about your ancestry!
As described above for Cape Verdeans a plausible historical context already exists to explain the reporting of North African matches. The rather high frequency of such matches I established during this survey can be further corroborated with associated admixture scores and vice versa. Although it should be kept in mind that admixture estimates will by default be variable and do not represent any exact measure of pedigree! Not only due to updates but also when comparing with other DNA testing companies which each use different algorithms and each have different databases of reference populations. Genealogy naturally would be the surest way to identify North African ancestors. However this option will not always be available. North African admixture may have been largely concealed for Cape Verdeans after Ancestry’s latest update. But this is not the case when testing with other companies. See for example:
- 23andme results for a Cape Verdean with 10.8% North African DNA (confirmed great-grandparent from Morocco)
- MyHeritage results for Cape Verdean with 14.6% North African DNA
The correlation I found between North African matches and North African admixture certainly is not perfect (0.22%, see table 4). In fact it is the lowest when compared with other types of admixture/matches. At first sight it might also be deemed peculiar that many of my survey participants received North African matches without having any “North African” admixture! In most cases they did however have “Africa North” scores before the update. Still in one case (CV16) neither before or after the update any North African admixture was reported. Even when I did find 6 North African matches for this person! I highly suspect that a minor West Asian trace amount (usually “Middle East”) could however still account for the connection. This also goes for many other survey participants who may have lost their former “Africa North” amounts but still retain some minor West Asian trace amounts. This is why I have also included West Asian admixture amounts in this spreadsheet.
One must remember that North African DNA is being described by Ancestry as a combination of not only “Africa North” but also “Middle East’ and even “Portugal” after the update. So I suppose some North African matches may now even be associated with “Portugal” admixture! Again the labeling is not to be taken too literally but rather to be seen as reflecting genetic overlap between North Africa and the Iberian Peninsula. Going back to ancient times, before the Moorish period even. Which greatly complicates establishing the exact direction of the gene flow and also its approximate timing.
In this matter a chromosome browser to establish the ethnic/regional origin of the shared DNA amounts with your matches would be of tremendous help! Also to distinguish for example such cases whereby a shared Upper Guinean ancestor might be relevant. A blog post has been published recently about the very intriguing findings by a person of Fula descent who shares Northwest African DNA segments with some of his Puerto Rican matches (brought to light by analysis on Gedmatch). I imagine this could also happen for Cape Verdean matches in selected cases. Once more illustrating how figuring out the exact way you are related to your matches can be quite challenging. But still admixture analysis can often be indicative in this effort! For more discussion:
- North African AncestryDNA results (Tracing African Roots)
- Major Guanche genetic influence in Puerto Ricans (For what they were… we are; guest article by Thierno)
North African founding effects for Sotavento?
Table 9 (click to enlarge)
I would like to stress that my survey findings are naturally to be seen as preliminary! Given the limitations of sample size for each particular island and relative lack of proportional island representation. Even if actually aside from Santiago most other islands are fairly well represented. And an overall sample size of 50 is already quite robust. It should also be noted that some of my survey participants were close family members. This goes especially for persons with island origins from São Nicolau & Santo Antão (= Barlavento islands, also incl. São Vicente). While also at least 3 persons are aware of relatively recent Moroccan Jewish ancestors 21.
Nonetheless a potentially insightful finding is depicted in table 9 above. Although the differences may seem to be quite slight already it seems that people with island origins from Brava & Fogo might expect more North African matches, on average (5 versus 2 for Barlavento). When reviewing the admixture scores (prior to update) again these two islands stand out somewhat, together with Santiago (all 3 islands combined also known as Sotavento). In fact I have previously also established such a pattern based on 23andme results (see MENA %’s in this spreadsheet).
Again I am not sure if these matching patterns qualify as statistically meaningful or would be replicated with greater sample size. However it may be worthwhile to compare my findings with previous research done on North African lineage for Cape Verde. In particular two studies based on Y-DNA (male haplogroups):
- Y-chromosome lineages in Cabo Verde Islands witness the diverse geographic origin of its first male settlers.(Gonçalves et al., 2003)
- The Admixture Structure and Genetic Variation of the Archipelago of Cape Verde and Its Implications for Admixture Mapping Studies (Beleza et al., 2012)
In both studies an ancestral contribution from possibly North African (as well as Jewish) males was detected. More so in Gonçalves (2003) than in Beleza (2012). However the interesting thing is that both studies mention differentiation according to island origins.22 With the northern islands (Barlavento) being distinguished from the southern (Sotavento) islands because of founding effects and earlier settlement in Santiago & Fogo. I was able to also include Brava samples in my own survey but such samples seem to not have been used in either study. However in this context Brava’s genetics may be considered a subset of its neighbouring and closely interrelated sister island Fogo.
Although these studies are not completely inter-comparable with my own survey (which is based on autosomal IBD matches rather than more restricted haplogroup analysis). It is still interesting to see the same kind of seemingly founding effects showing up according to island origin. This would be historically plausible and caused by particular settlers who may have been more numerous in certain islands than others. For example exiled Mourisco’s or Conversos. Despite differentiation it must still be stressed that North African matches were reported for survey participants from all islands. And in this sense such lineage will not be exclusive to any island! Despite relative isolation of course inter-island migration is also relevant in this regard. In the second part of this blogseries I will attempt to explore such island variation again. For more details see:
- “Africa North” scores for 100 Capeverdean AncestryDNA testers (scroll to section 6)
- Estimated admixture proportions (incl. Sephardic Jews & North Africa) of Y-chromosome lineages from Cape Verde(Beleza et al., 2012, p.6)
- Spreadsheet with 23andme results for 32 Cape Verdeans, featuring North African admixture (MENA)
4) Other African matches
Table 10 (click to enlarge)
The 30 remaining African matches I was able to find for my Cape Verdean survey participants are featured in full detail in table 10 as well as within the online spreadsheet linked above. This type of African matches was least common. Not surprising given that they fall outside of the expected areas of Upper Guinea & North Africa. Usually reported with small amounts of shared DNA. Increasing the odds of either being false positives (IBC) or population matches (IBP) (see section 1 or also this chart). The share of matches smaller than 7 cM was 57% (17/30). Although intriguingly a few bigger matches were still also being reported. Three times even with shared DNA greater than 10 cM!
Strictly relying on historical plausibility many of these widely dispersed matches seem rather puzzling. After all Upper Guinea is by far the predominant source of African origins for Cape Verde. However 9 matches shown above appear to be from Northeast Africa (Ethiopia, Sudan, Kenya, Tanzania). Six matches appear to be from Central or Southern Africa (Angola, Congo, Zambia, Malawi, Zimbabwe, South Africa). While 15 matches seem to be from the remaining parts of West Africa (a.k.a. Lower Guinea). Mostly from southern Nigeria but also from Liberia, Ivory Coast, Ghana, either Benin or Togo as well as Cameroon and possibly Niger or southern Libya.
In section 2 it has already been explained why I have chosen to group presumably Hausa-Fulani matches under Upper Guinea rather than with Nigeria. In fact several other matches mentioned in table 10 may still also be Upper Guinean-related. This goes especially for the 3 matches from Liberia (who all had “Senegal” scores of >20%, prior to the update) and the Ivorian match (who seems to be a Malinké with partial Fula lineage and scored 29% “Senegal”, prior to the update). But quite possibly also for the Tuareg match from either Niger or Libya. As well as one of the southern Nigerian matches, who judging by surname & rather high “Mali” as well as “Senegal” scores could be a Muslim Yoruba with partial Fulani lineage.23
This serves as a reminder of the considerations I discussed in section 1. These somewhat outlandish matches are not per se indicative of genetic influence from these countries into the Cape Verdean gene pool. And certainly not conclusively so! Several ancestral scenarios may apply as always. Including reversed gene flow originating from Cape Verde/Upper Guinea or by way of a third country/ethnicity. All of which should be critically investigated on a case by case basis. Especially for these unexpected and (largely) undocumented sources of African origins for Cape Verdeans. For example the disproportional representation of various African countries within Ancestry’s customer database probably explains the elevated frequency of Nigerian matches. While also the matter of IBD or IBS matches remains even more crucial to avoid misleading misinterpretation.
No IBD matches from Central, Southern and Eastern Africa?
Table 11 (click to enlarge)
Seven of my survey participants also shared the AncestryDNA results from one of their parents. While one survey participant (CV04) shared the results from both of his parents.24 Enabling me to determine if African matches were indeed inherited by way of one of the parents. In other words this allowed me to investigate their share of IBD (Identical By Descent) matches. I have already discussed in section 1 how such an exercise can be very useful! Because if any African match is indeed also reported for one of my survey participants’ parents then the possibility of a false positive or IBC (Identical By Chance) match may practically be ruled out. And the odds of a genealogically meaningful match will be greatly increased. Even if there will still also be a chance that you are dealing with an IBP (Identical by Population) match. As these are a subset of IBD matches. Given correct interpretation these IBP matches may at times also be useful. However they will generally trace back to a common ancestor from ancient times. Outside of a genealogically meaningful time frame (~500 years). And in this way they can potentially be misleading when you are not aware of the broader context.25 See also:
- Concepts – Identical by…Descent, State, Population and Chance (DNAeXplained, 2016)
Of course my findings shown in table 11 and also contained within my spreadsheet are limited to a small portion of my entire survey group. While also obviously a more robust outcome would have been obtained if I had access to the results of both parents for everyone. Then again even when only preliminary these matching patterns could still be rather insightful already. As they seem to indicate that “exotic” and unexpected African matches may often turn out to be false positives. As was also the case for the 8 survey participants for whom I could actually determine this (mostly based on a single parental phasing basis). After all I could not verify that matches from Cameroon, Ethiopia, Kenya, Tanzania, Zambia, South Africa, Sudan were inherited by way of the parents of my survey participants. Also North African matches from Libya and Egypt were not being confirmed. Making it more likely that these were just random IBC matches.26
I suppose two opposing views may exist in regards to African matches. Either one believes all of them are the real deal and therefore IBD. To be fully embraced from the moment of discovery. But others may be more cautious and assume African matches are possibly false positives or IBS until proven otherwise. These are rather extreme positions of course. But it seems to me that especially in the absence of a historically plausible context a more conservative approach is called for. Understandably the eagerness for African matches among Afro-Diasporans is very high. But I feel it is wise to resist the urge of finding a fitting ancestral scenario right away whenever a more surprising African match is reported. As this could be a form of self-serving confirmation bias. I have already outlined some crucial considerations to be kept in mind in section 1. Still I suppose it is also a matter of personal preference whether to focus on historically plausible matches in line with the majority of your regional African admixture. Or rather to go on a wild goose chase of African matches from more eccentric and unexpected places.
To be sure direct ancestral connections with Central & Southern Africa as well as West Africa (beyond Upper Guinea) are not entirely impossible for Cape Verdeans. As the presence of a few captives from these areas has actually been documented for Cape Verde. Although very sporadically so and with minimal numbers (see this link for an overview).27 Another possibility might also occur by way of detour. Not directly related to slavery but rather via voluntary migrations. Through gene flow from either other parts of the Afro-Diaspora or from other former Portuguese colonies in Africa (see next section).
Direct ancestral ties with Northeast Africa however seem least likely to be supported by any documented evidence I am aware of. Granted: absence of evidence is not evidence of absence. However historical plausibility does greatly impact probability. Therefore I remain sceptical of any supposedly East African DNA results being reported. In regards to both DNA matches as well as admixture. Not only for Cape Verdeans but also other (Trans-Atlantic) Afro-Diasporans. To be frank I think the odds of any genuine genetic inheritance from especially Northeast Africa (within a genealogical time frame) are going to be very slim. And currently it is certainly being overstated in DNA testing. Then again in individual cases it is not be ruled out. But solid follow-up research and extra scrutiny is always required to corroborate such findings.28 See also:
Regional admixture as corroboration for African matches outside of Upper Guinea?
Figure 2 (click to enlarge)
Figure 3 (click to enlarge)
In section 1 it has already been mentioned that combining admixture analysis with DNA matches may lead to more insight as well as mutual corroboration. The screenshots shown above are provided as two striking examples. Actually I found that such an approach can be fruitful for my other survey participants as well.29 The profiles shown in figures 2 & 3 stand out for featuring distinctive regional scores. But they also happen to have received the largest number of African matches outside of the expected areas of Upper Guinea & North Africa: no less than four each! Taken together this accounts for more than a quarter (8/30) of all unexpected African matches for my survey group!
I do not think this outcome was caused by any coincidence. But rather it seems to me that a singular type of foreign lineage from more than 2 generations ago (possibly Brazilian) is being expressed both in admixture results as well as DNA matches. This would definitely help to explain unexpected matches from places like Nigeria and Central/Southeast Africa being reported for Cape Verdeans. As after all these places are major sources of African origins for Brazilians (see this table)! Also other Trans-Atlantic Afro-Diasporans are bound to be much more varied in their regional African origins than Cape Verdeans (see this blog post). Let alone people from other Portuguese African colonies. Even when overall historical plausibility must be leading the genetic impact of individuals migrating across place & time should not be underestimated! I do not have any certainty about such an ancestral scenario however and obviously more follow-research is required on a case by case basis.
Initial clues may be provided in the first place by unexpected and relatively pronounced regional admixture from within Africa. In other words regional scores from beyond Upper Guinea (“Senegal” and “Mali”) being reported above trace level. Unrealistic expectations should be avoided though. As Ancestry’s ethnicity estimates obviously will not be conforming exactly with known pedigree or be 100% precise in their regional labeling. Focusing on apparent deviations from group averages will be more helpful. In my 2015 discussion of Cape Verdean AncestryDNA results I already mentioned the somewhat unexpected high level of African regions not directly to be associated with Upper Guinea. That is “Benin/Togo”, “Nigeria”, “Cameroon/Congo” and “Southeastern Bantu”. I continued this discussion in my latest blog post about 100 Cape Verdean AncestryDNA results. See also:
- Beyond Upper Guinea: valid outcomes or misreading by AncestryDNA? (scroll to section 4)
As foreseen in that article the recent update by Ancestry has eliminated many of these unexpected regional scores, seemingly indicating origins from either Lower Guinea or Bantu speaking Africa. However not entirely! As can be seen in this table 6 and also this spreadsheet. The Upper Guinean proportion of the African breakdown for my Cape Verdean survey participants has increased considerably to above 90%. But still it is not fully 100%. Especially the new “Cameroon, Congo, Southern Bantu” region at times attains a rather high level. The maximum score I observed being 9% as shown in figure 2. But also scores of 7%, 6% and 5% were obtained (see footnote 9 for screenshots). Because Ancestry’s algorithm is said to have become more thorough I suppose the persistence of these unexpected regional scores may be seen as providing more solid clues than before. Even more so when actual DNA matches from such regions are appearing at the same time! 30
Regional admixture scores from outside Africa might also provide useful hints, as shown in figure 3. In particular partial Brazilian lineage might be hinted at by trace amounts of “Native American” DNA. In my survey of 100 Cape Verdean AncestryDNA results “Native America” was only reported for 5 of my 100 survey participants with a maximum score of 1% (see this table). Quite uncommon therefore but still very noteworthy as Cape Verde is an African island group! Generally speaking trace region reporting is of course to be taken with a grain of salt. I would however argue against complete dismissal in all cases because plentiful historical evidence can be found for ancestral connections between Cape Verde & Brazil. And to a lesser degree also between Cape Verde and other places in the Americas, which might likewise involve the transferal of minor Native American admixture as well as African DNA from beyond Upper Guinea.31
This occurrence of unusual African matches certainly is intriguing and can at times also be corroborated by additional clues. For CV41 & CV43 it seems that an ancestral scenario involving recent foreign (Brazilian?) lineage could be quite likely. An IBD check (by testing at least one of your parents) is highly recommended to gain more certainty and rule out any false positives. Nonetheless these findings should not distract from the fact that Upper Guinea unequivocally remains the principal source of African origins for Cape Verdeans! Nearly exclusively so, even if not quite. One must be very careful therefore to make proper distinctions while reviewing these 30 “Other African” matches, I was able to find during my survey.
In order to learn more about Cape Verde’s African origins in a generalized manner it seems reasonable to me to focus on unambiguous matches. In this light the 8 possibly Brazilian linked matches from CV41 and CV43 may be taken out of consideration, as well as the 6 possibly Upper Guinean linked matches (from Liberia, Ivory Coast, Nigeria and Niger/Libya), and also the 1 verified IBC match from Sudan. So that only a subtotal of 15 relevant matches remains. Which gives a proportion of around 5% of all African matches south of the Sahara (15/257), which genuinely seem to indicate African connections beyond Upper Guinea. With further research this number may actually diminish even further. Either way confirming once more how Cape Verdeans are overwhelmingly Upper Guinean when it comes to their African roots (south of Sahara).
Figure 4 (click to enlarge)
While analyzing the DNA matches of my Cape Verdean survey participants I have always taken a cautious research approach when deciding on their most likely background. Going by any clues given but in particular:
- plausible regional admixture
- plausible surnames.
Whenever available I also took into consideration actual birth locations and other relevant profile details. Unfortunately such information was usually not provided. Because most people on Ancestry do not tend to have public family trees. In this section I will describe the methodology I applied in more detail. But first of all I would like to repeat that my categorization of African DNA matches 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 therefore processed the data on a best effort basis to get a general idea. In order to avoid any false assignment I actually preferred to leave out several potential African matches from this survey whenever I was in doubt.
Finding African DNA matches can often seem like a daunting task. Almost like finding a needle in a haystack. However the filtering method I blogged about in 2017 makes it quite easy and time-saving to single out distinctive ethnic profiles. Because my Cape Verdean survey participants kindly agreed to share their Ancestry profiles with me I was able to zoom in to their most likely African DNA matches. By first scanning all DNA matches with DNAGedcom and then filtering in Excel for profiles containing only African or compatible ethnic regions (“Middle East” etc.). This gave me an initial overview of potentially African DNA matches which however needed to be scrutinized further to remove Afro-Diasporan profiles. For each of my 50 survey participants I generated an Excel file featuring all of their DNA matches. As well as their filtered African, Asian, Jewish and Portuguese matches in separate tabs. Finally all of this data was then combined for statistical analysis in my online spreadsheet.32
For my step-by-step tutorial on how to detect African DNA matches see:
- How to find those elusive African DNA matches on Ancestry (Tracing African Roots)
- Advanced Excel Filter Settings (link to my googlesheet containing the filter criteria I applied)
- DNAGedcom Client (software I used for scanning all of the DNA matches of my survey participants)
For other useful resources on how to find and interpret African DNA matches see:
- African DNA Cousins reported for people across the Diaspora (Tracing African Roots)
- DNA matches reported for 75 Africans on 23andme (Tracing African Roots)
- Confirming African Matches: Abuelo’s Peul (Fula) Relatives (Dominican Roots)
- Finding Your Wakanda Africa with DNA Testing (Roots & Recombinant DNA)
Plausible regional admixture
Figure 5 (click to enlarge)
Regional admixture is one of the two key pillars in my analysis of African DNA matches. More specifically the African breakdown as it was reported by AncestryDNA before the recent update.33 Ancestry provides a very useful summary of the ethnic regions for each one of your DNA matches on their profile page. And these details are also being scanned by DNAGedcom. Which made it possible for me to devise my filtering method (see figure 4). Although DNAGedcom does usually apply a a somewhat different labeling of the ethnic regions on Ancestry. For example “AfricaN” = “Africa North” (for a complete listing see step 4 on this page). As of October 2018 I was also able to use the very helpful compare ethnicity tool on Ancestry in order to verify the exact regional percentages of the African DNA matches I had filtered! Which is something I checked for each separate African profile to decrease the odds of mis-identification.
In previous blog posts I have demonstrated how the regional (a.k.a. subcontinental) categories in DNA testing CAN be indicative of distinctive lineage. This goes not only for African DNA but also European and Asian DNA in fact.7 Despite inevitable overlap regional admixture is certainly not meaningless or totally random as is sometimes assumed. Some regions on the far end of the spectrum may even be said to be almost mutually exclusive. Such as “Senegal” and “Cameroon/Congo” being proxies for resp. Upper Guinean and Central African lineage. Going by my observations over the years the previous African breakdown on AncestryDNA has been very insightful to gain a greater understanding of the regional African roots for people across the Afro-Diaspora as well as actual Africans themselves. Despite several shortcomings as well as the continued need for correct interpretation. My survey findings on a group level have still been reasonably in line with either historical plausibility or actual verifiable genealogy. For more details see also:
- African AncestryDNA results (screenshots and context)
- African AncestryDNA Results (spreadsheet with individual results)
- African AncestryDNA Results (group averages, incl. ethnic groups)
In this current endeavor of analyzing the African DNA matches for my Cape Verdean survey participants these regional admixture patterns turned out to be particularly helpful to establish a plausible background within Africa. Obviously not pinpointing anything precise but already zooming into specific areas, countries or even ethnic groups at times. I mostly relied on the following decision rules:
- Upper Guinean matches (excl. Fula): “Senegal” and/or “Mali” clearly primary and predominant + “Africa North” and/or “Middle East” either absent or only showing up as trace region. People of mixed or diluted Fula lineage could still be included in some cases therefore. But not very likely for the most part. Whenever the main regional combination also included “Ivory Coast/Ghana” aside from “Senegal” and/or “Mali” I tried looking for other clues (esp. surnames) to make a Sierra Leonean or Guinean (Conakry) assignment more plausible.
- Fula matches: “Senegal” and/or “Mali” clearly primary and predominant + “Africa North” and/or “Middle East” also showing up as main regions. People of mixed or diluted non-Fula lineage could still be included obviously given frequent inter-ethnic unions in the past. However combining with Fula-specific surnames often made this categorization quite solid.
- Hausa-Fulani matches: a predominant regional combination consisting of at least “Nigeria” and “Senegal” + “Africa North” and/or “Middle East” also showing up as main regions. Combining with surnames specific to northern Nigeria often made such identification even more plausible.
- North African matches (all): Because of the consistent appearance of South European regions (best to be seen as a proxy for ancient shared Mediterranean DNA) I actually had to adjust the filtering method I applied for North African matches. A simple text filtering or rather sorting by ethnic region and then zooming into DNA matches with either “Africa North” or “Middle East” in first place proved to be very effective already. Especially also when combining with Arab/Berber surnames.
- North African matches (Morocco, Algeria, Tunisia): a predominant regional combination consisting of at least “Africa North” and “Middle East” + “Senegal” and/or “Mali” being reported either above or below trace level + “Iberian Peninsula” and/or “Europe South” being reported either above or below trace level. Based on confirmed Maghrebi results as well as their surnames I have made the assumptions such profiles were either Moroccan, Algerian or Tunisian.
- North African matches (Libya & Egypt): I was able to distinguish a few Libyan and Egyptian matches by comparing with the results of confirmed Egyptians/Libyans. Who typically tend to have “Middle East” firmly in first place rather than “Africa North” as well as often show minor amounts of “Southeast Bantu” rather than either “Senegal” or “Mali”. Also their surnames added to plausibility.
- Other African matches (Lower Guinea, Central, Southern and Eastern Africa): plausibility being dependent on whichever regional combination I was already aware of based on the regional group averages (bench marks) for African countries and ethnic groups in my ongoing African survey of AncestryDNA results. Again also taking other clues into consideration: in particular surnames.
Plausible African surnames
Figure 6 (click to enlarge)
- Overview of all African DNA matches (incl. surnames in column D)
Regional admixture provided the first step in my identification of a plausible background for African DNA matches. However I nearly always continued my plausibility check by also focusing on African surnames (and at times also African first names). In fact I have tended to exclude possibly African matches from my survey out of precaution. Whenever they did not feature an African surname on their profile page. And also in absence of any other corroborating evidence.
For each possibly African surname, first name or even nickname I have performed a search on not only Google but also Facebook to find out more details. The forebears.io website has been a particularly helpful resource in my efforts (see figure 6). As it provides more solid statistics for surname incidence and frequency. Often enabling me to zoom into most likely nationality or even ethnic group. Although many African surnames are in fact widespread across several countries. Especially Muslim names tend to be quite generic.
Furthermore of course also ethnic-specific surnames are often only reflecting one single family line on either the paternal or maternal side. Inter-ethnic unions are never to be ruled out however! Both recently or further down the line. Also one needs to be careful because sometimes Afro-Diasporans adopt African names (due to conversion to Islam, marriage or as an Afro-pride statement). There is also the issue of different spelling variations. Which can however also be used advantageously (see figure 6). But sometimes seemingly similar surnames will be used in various countries for unrelated reasons but sheer coincidence. Naturally one needs to be aware of such limitations and not jump to conclusions. Still by combining with regional admixture and other clues the odds of finding a plausible background can often be greatly enhanced!
Throughout my various research projects I have always made great efforts to safeguard the privacy of my survey participants. Leaving out any name details, except for acronyms when approval was given. As in fact I have also done this time for my Cape Verdean survey participants who are only identifiable by CV01, CV02 etc.. However in regards to their African DNA matches I have decided to include their surnames in my spreadsheet. As I believe these African surnames can have great educational purpose. In order to prevent full identification I have still not displayed their full names. If anyone recognizes themselves in these matches and should not be be comfortable with this display please send me a PM and I will act accordingly. For a small selection of useful resources in regards to Upper Guinean surnames:
- Forebears.io (surname search)
- Fula common names (Afropedea)
- Fula surnames (Revolvy)
- Wolof Patronymes (Wikipedia)
- Serer maternal clans (Wikipedia)
- Noms et prénoms du Sénégal (Planète Sénégal)
- Nomes proprios da Guiné-Bissau (CART 3494)
Figure 7 (click to enlarge)
I consider the various survey findings I have presented on this page to be the culmination of my efforts across the years to dig into the specific African origins of Cape Verdeans. Although of course there is plenty room for improvement and follow-up research! As I have maintained from the very start of my blogging “career” in 2014 I am convinced that new insights are often generated by just putting two and two together. In other words I think a multi-disciplinary approach often works better than just limiting yourself to a one-sided view. Combining genealogy not only with various aspects of genetics (admixture analysis, autosomal DNA matches, haplogroups etc.) but also with history, ethnography, linguistics and so forth. Throughout this blog post I have already highlighted several times how correlating (regional) admixture analysis with DNA matches can be rewarding and reinforcing in many cases.
Both aspects of genetic genealogy obviously have their own shortcomings. But 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!
I have also stated from the beginning that I am a firm believer of democratizing knowledge. Intriguingly several of my Cape Verdean survey participants received the very same African DNA match. One quite likely Fula match was reported for no less than 8 of my survey participants (see A. Mohamed in this sheet)! All of them unrelated within the last 2 generations as far as I am aware. I suspect such outcomes may partially be explained by high endogamy among Cape Verdeans as well as recurring population (IBP) matches. However it may in some cases also be indicative of a mutual African ancestor within a genealogically meaningful timeframe! I imagine most likely from a relatively late time period (1800’s or late 1700’s).
It could be very beneficial therefore to focus on shared African matches within my survey group as one avenue of promising follow-up research. Using advanced tools such as triangulation or chromosome mapping to identify the associated shared DNA segments may then lead to fitting African DNA matches along specific family lines. Mapping African MRCA’s as far back as possible by looking into shared matches. Exploring how you relate to your African DNA matches may often involve mutual learning and joint-effort research. Such collaborations are not only limited to your direct African connections but also triangulated ones! This type of advanced and innovative research is something I aim to engage with myself in 2019!
The ultimate goal of trying to fit your African matches (MRCA’s) into your family tree will usually be very tricky though given scanty information.34 And it will also require a lot of patience. But it might still be worthwhile for eventually zooming in closer to your African origins along a certain family line or even actually identifying an African ancestor! Breaking down those brick walls based on paper trails! Having other close relatives also DNA tested will of course be greatly helpful. Especially if they share the same African matches with you. However any shared matches for your African match might do the trick I suppose. As long as you can figure out how these shared matches are to be placed in your family tree. Although I have no personal experience with it right now triangulation and DNA Painter seem poised to give you more insight. For more details see:
- Benefits of Triangulation (Segmentology, 2015)
- Using DNA Painter to reconstruct ancestral DNA (Ultimate Family Historians, 2018)
- Pinpointing the Origin of Family’s Igbo Ancestry with DNA (Roots Revealed, 2017)
- Using DNA Painter to Verify Igbo Origins (Roots Revealed, 2018)
1) While filtering the DNA matches of my Cape Verdean survey participants for non-European profiles I actually also very frequently came across matches from Hawaii and various other parts of the Pacific. The ethnicity estimates for these matches (as mentioned on their profile page) were entirely consisting of Asian and/or Pacific regions. Safe for some trace amounts of African and /or European admixture at times. However in practically all cases these matches do not seem to correspond with trace amounts of Asian/Pacific admixture found among Cape Verdeans. But rather these matches reflect mutual Cape Verdean and/or Portuguese ancestors who had offspring and often also settled in this remote part of the world so far removed from Cape Verde. Literally oceans away!
A very intriguing outcome which is to be explained by the participation of Cape Verdean seamen in the American whaling enterprise during the 1800’s and early 1900’s. An amazing saga which unfortunately is not widely known despite being so evocative. In a next follow-up blog post I will discuss this Cape Verdean connection with the Pacific in greater detail. For a quick introduction see also:
- Cape Verdean Maritime Exhibit (New Bedford Whaling Museum)
- Cape Verdeans to Hawaii: A Pre-Sugar Plantation Era Migration (Your Island Routes)
2) I am greatly appreciative of all my survey participants! Without their kind willingness to share their results with me my research would not have been possible! As a minor token of my gratitude I have sent each one of them the Excel file containing all their DNA matches. Aside from giving me access to their results many people also shared useful and very interesting details about their family history with me in PM’s. As far as I was able to verify through such contact or also through public family trees and other publicly available information all my survey participants are of fully Cape Verdean descent. That is going back on all lines for at least 2 generations. In other words all 4 grandparents being Cape Verde-born.
For a vast majority of my survey participants I received confirmation of their Cape Verdean background by way of PM. In a few cases I was informed by my survey participants that they were aware of migrant ancestors in their family tree, going beyond my 4 Cape Verde-born grandparents criterium. For example 1 Portuguese great grandparent or 1 Moroccan Jewish great-great grandparent. As such lineage is generally in line with Cape Verde’s history of incoming migrants I have included these results in my survey. Whenever these ancestral details were available to me I have mentioned them in column AK in this spread sheet.
Aside from PM’s and public family trees I also tried to verify the background of my survey participants in other ways. Naturally to the best of my ability. Going by any clues given but in particular: plausible surnames and plausible regional combinations as well as the “migrations” mentioned in AncestryDNA’s Ethnicity Estimates. Taking a cautious approach and preferring to leave out possible survey participants when in doubt.
3) In 2017 I discovered my first West African (mainland) DNA match for myself. I have not yet been in meaningful contact with him. But quite likely this match is from Gambia going by plausible regional admixture and plausible surname (Jatta). I have applied the considerations to be taken into account for myself as well in this comment:
4) Naturally even within the 6cM-7cM range some genuine IBD matches might appear. But you will have to judge the plausibility on a case by case basis. Then again I suspect most of these matches are likely to be false positives. Also socalled population matches (IBP) are more common than many people might expect. Basically ancient shared origins lead to a great degree of genetic overlap in neighbouring and/or ethno-linguistically related populations. Which can result in identical DNA segments being detected.
This circumstance operates to create DNA matches from all over! Even when technically speaking you are not related to such DNA matches in a genealogically meaningful time frame (let’s say 500 years). I have observed this myself for both Europeans and Africans. Also several DNA studies have already brought such generic matching patterns to light. Greater awareness of migrations and ethno-linguistic relatedness within Europe and Africa is therefore required for proper interpretation of these smaller matches. See also:
- Dispersals and genetic adaptation of Bantu-speaking populations in Africa and North America. (Patin et al., 2017)
- How Phasing Works and Determining IBD Versus IBS Matches (DNAeXplained, 2015)
- Study Suggests Europeans Are Closely Related (SciTechDaily, 2013)
5) When it comes to finding Cape Verdean descended DNA matches it is a whole different ball game due to the already existing popularity of DNA testing among Cape Verdean-Americans. I have not kept an exact score of it during this survey. But I would estimate that for my survey participants at least a plurality of all DNA matches could be (partially) Cape Verdean related. When only counting close matches (>20cM) they are certainly a majority (overwhelmingly so). The number of people (over 6,000 most recently) being assigned to the socalled “Portuguese Islander” migration is a telling indicator.
In fact also Latin American matches are very frequently reported for Cape Verdeans. Probably second in number behind Cape Verdean related matches. African Americans (without any recent Cape Verdean connection), white Americans (of Northwest European descent) and West Indians also show up as matches quite often but to a lesser degree from what I have seen.
6) A chromosome browser would be perfect to establish whether the shared DNA segments between Cape Verdeans and their Latin American and West Indian DNA matches happen to be European or African. Helping to make the distinction between a possibly Portuguese (or Galician/Spanish/Canarian) MRCA scenario versus an Upper Guinean MRCA scenario. It should be mentioned also that according to some sources aside from Portuguese/Madeirans also Cape Verdean contract labourers might have settled in some parts of the West Indies during the late 1800’s. In particular Guyana. See this link:
- The Arrival Of The Portuguese (Guyana Story)
Generally speaking when it comes to African American matches for Cape Verdeans and vice versa I tend to think there are even more possible ancestral scenario’s (in case of IBD matches). None of them to be ruled out in advance. A mutual European ancestor (not only Portuguese/Spanish but also English, French, Dutch etc.) who left descendants in both Cape Verde and the US could be possible again. But probably with lesser odds than for Latin America and the West Indies. In this case there would not be any direct Cape Verdean connection. Again using a chromosome-browser to establish if the shared segment is either African or European would be ideal of course. But unfortunately not always possible.
Looking at the size of the shared segment also can be helpful. All things being the same you would expect larger amounts of shared DNA (>10 cM) to suggest a relatively recent connection with Cape Verdean-Americans who started arriving in the US during the early 1800’s. They were the first Africans to voluntarily migrate to the Americas! At times even on self-owned ships! They are primarily located in the Massachusetts and Rhode Island area but there has also been an early presence in California. Despite relative endogamy they have often also intermarried with African Americans across the generations. African Americans who have recent family ties to those states might therefore have greater chances of finding Cape Verdeans matches I suppose. And then it would be a direct Cape Verdean connection. But not due to slave trade.
The MRCA’s (most recent common ancestor) for African-American-Cape Verdean matches could also have been from anywhere in the Upper Guinea region (Senegal-Sierra Leone). This person could have had one relative taken as captive to Cape Verde while another was taken across the Atlantic by slave traders. In this scenario the African American match doesn’t actually have Cape Verdean ancestry but does share the same Upper Guinean ancestor with his Cape Verdean match.
From what has been documented relatively very few Cape Verdeans were taken as captives to the USA, about 400 according to the slave voyages database. And strictly speaking they might not even have been Cape Verdean-born but just passing through as captives from the Upper Guinean mainland. This is a negligible number if you compare it with the estimated almost 400.000 Africans to have disembarked in the US. Therefore I personally believe the odds of having a match by way of this scenario are the slimmest (~0,1%), but it’s still a possibility I suppose. See this blog for more details:
7) For an overview of blog articles discussing the usefulness of admixture analysis, see heading “Blog Posts” in this ISOGG entry: Admixture Analysis.
Sometimes it almost seems that admixture analysis is being considered mere guessology by its fiercest critics. Or only fit for cocktail parties as the saying goes 😉 This has however not been my experience. I do agree that performance among the various DNA testing companies and third party websites is very variable. I have myself only tested with 23andme and Ancestry and I found that with correct interpretation and knowledge of their methodology you can derive useful information from their ethnicity results. Which were of course not just pulled out of a hat!
I know of many people who made important discoveries about their genetic ancestry by using ethnicity estimates. For example I have heard several stories by West Indians who had unexpected Asian admixture, minor but still substantial (say >10%). And this information was really useful to them as it lead them to previously unknown Asian contract labourer ancestors. There are plenty of other ancestral scenarios for Afro-Diasporans which can be illuminated by way of the continental breakdown which is usually quite accurate.
The regional or subcontinental percentages are indeed not to be taken all too literally. But again I know several persons who relied on distinctive regional admixture to make a breakthrough in their ancestral quest. For example I have been told about at least three instances of NPE being confirmed whereby the father turned out to be East African instead of African American or West Indian. In one case indicated by the very predictive “East African” category on 23andme but also by a singular combination of “Southeastern Bantu” and “Middle Eastern” regional scores on AncestryDNA. Many times I have also seen how unexpected partial Cape Verdean lineage could quite reliably be corroborated by “Senegal” scores on AncestryDNA. Not only for African Americans, but also for Hawaiians (due to whaling connections, see upcoming blogpost)!
In my AncestryDNA survey for Africans & Afro-Diasporans I have also made good use of a macro-regional format. Which is still sub-continental. Basically combining interrelated & neighbouring AncestryDNA 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 and slave trade patterns. Which I indeed established for the most part. But also for European DNA I found that this approach works quite well. For example by making a distinction between Northwest European DNA ((“Great Britain”, “Ireland”, “Europe West” and “Scandinavia”) versus Southwest European DNA (“Iberian Peninsula” and “Europe South”) and East European DNA (“Europe East”, “Finland/Northwest Russia”, “European Jewish”). For more details see:
- Afro-Diaspora AncestryDNA results: A Comparison
- Update: Afro-Diasporan AncestryDNA Survey (part 1)
- Update: Afro-Diasporan AncestryDNA Survey (part 2)
8) Imagine not being aware of the average degree of African admixture (50%, of which 46% = “Senegal” + “Mali”, see table 1) for my Cape Verdean survey participants. And instead you were only informed about the relatively subdued average number of African DNA matches (5, of which 4 from Upper Guinea). You might then be mislead in thinking that the Portuguese and Jewish ancestry for my survey participants would be far more significant. Again when only going by their number of Portuguese and Jewish matches (resp. 89 and 30) and not also the associated admixture levels for “Portugal” and “European Jewish” (resp. 35% & 0.4%). While also the average number of North African matches (4) based on merely 0.5% “Northern Africa” admixture on average is clearly disproportional when compared with the average level of West African matches/admixture. Combining both aspects of DNA testing therefore leads to greater overall informational value. Rather than only focusing on DNA matches.
9) When I started out with this survey AncestryDNA’s Ethnicity Estimates had not been updated yet. Compared with the previous version (current between 2013-2018) AncestryDNA’s update has caused quite some changes:
- predominant “Senegal” amounts (within the African breakdown) have consistently been replaced by predominant “Mali” amounts. (see this blog post for more discussion)
- the Upper Guinean share (“Senegal” + “Mali”) within the African breakdown increased considerably.
- the new “Portugal” region is being reported in higher amounts than the former “Iberian Peninsula” region.
- former “Italy/Greece” or “Europe South” amounts have decreased quite sharply when comparing with the combined scores of the new regions “Italy”, “Sardinia” and “Greece & the Balkans”.
- Trace region reporting has largely been smoothed out due to Ancestry’s new algorithm (see this blog post for more discussion). Resulting in the loss especially of many former “Africa North” scores (now incorporated in “Portugal”). But also the group averages of “Middle East ” and “European Jewish” have decreased as a consequence.
For an overview of the pre-update ethnicity estimates of my Cape Verdean survey participants see:
- 100 Cape Verdean AncestryDNA results
- Table 1 (pre-update)
- Table 4 (pre-update)
Below the updated results of a subset of my survey participants are being shown. Not all 50 individuals are included because not all my survey participants have chosen to update their results yet. I still managed to obtain the updated results of these persons by way of this trick.
Screenshots of updated Cape Verdean results (rightclick and open in new tab to enlarge)
10) Based on my ongoing survey of African AncestryDNA results I would estimate that Ancestry’s customer database already contains hundreds if not several thousands of Africans (incl. 1st and 2nd generation migrants living in North America, Europe or Australia)! I have personally seen results from practically all African countries already! However some countries tend to be overrepresented (such as Nigeria & Morocco) while others are underrepresented (such as Angola & Guiné Bissau). For Cape Verdeans in particular this situation creates more chances of being matched with for example their Moroccan lineage but does not invalidate that they might still also have Guinean ancestry in addition! This skewed customer database factor is mostly due to African migrant presence in the US/UK. For example follow this link below:
A few years ago I also kept score of how many West African matches were being reported for Cape Verdeans on 23andme. When 23andme’s Countries of Ancestry Tool was still in place. Back then (2014) naturally the number of African DNA testers was much fewer. Out of 20-30 Cape Verdean profiles shared with me only 4 showed any West African match. These matches furthermore tended to be smaller than 7cM. See this link for an overview (bottom of page):
- DNA Evidence (Cabo Verde Raizes na Africa, 2014)
11) Extensive family tree research has been done for the person (CV08) from Brava with the highest number of African matches (20) in my survey. No relatively recent mainland African connection is known. Intriguingly a rather African sounding surname has been passed down though: “Dandan“. Recorded as fas back as the mid 1700’s! Which is quite unique for Cape Verdeans who tend to have Portuguese surnames (although this surname also has some frequency in Brazil when spelled as Dandão). However this may just have been a coincidence.
I can only speculate but still the possibility of a relatively recent mainland African ancestor along atleast one family line does seem realistic. Quite likely this person being a Fula because CV08 seems to have about 14 Fula matches! The shared segment size at times being quite high with a maximum of 15.88 cM. Quite extraordinary from what I have seen. Eventhough Fula DNA testers might be overrepresented in Ancestry’s customer database. At the same time Fula captives residing in Cape Verde may also have been relatively frequent especially in the 1800’s. Another unique aspect of CV08’s African matches is the occurrence of one most likely southern Nigerian DNA match. Judging from his name this match appears to be Igbo. And the shared segment size is quite considerable too, around 10 cM!
12) Aside from Cape Verdeans freely migrating and settling in Upper Guinea throughout history the opposite also took place. That is mainland Upper Guineans voluntarily settling in Cape Verde. Several historical reports exist of this phenomenon. Although based on what I have read this flow of people was most likely small in number and also probably mostly limited to the earliest time periods (1400’s-1500’s).
Another ancestral scenario to keep in mind is that it cannot be ruled out that some of the Upper Guinean matches being reported might be due to shared European lineage. Although I mostly filtered for “100% African” profiles I did make an exception for Fula matches or partially Fula matches who tend to show a considerable North African(-like) component in their DNA. Because of the imperfections of AncestryDNA’s analysis this component is described not only in terms of “Africa North” and “Middle East” but also “Iberian Peninsula”, “Europe South” and even “European Jewish” at times. Generally only in trace amounts.
I personally believe that most of these seemingly South European scores are simply mislabeled artefacts to be added to the North African(-like) DNA for Fula people. As many North African results I have seen on Ancestry also tend to score considerable amounts for these seemingly European regions. Genetic similarity and (ancient) shared origins between North Africa and the Iberian Peninsula most likely being the main culprit. I have observed the same on 23andme:
- North African AncestryDNA results
- Fula AncestryDNA results (spreadsheet)
Nonetheless genuine diluted European admixture for Fula as well as other people from the Upper Guinea area could still be a real possibility for selected individuals. Before the recent update Ancestry’s reporting of trace amounts of admixture tended to be not very accurate. I suppose the possibility of very diluted trace amounts of European admixture among some of the Upper Guinean matches I filtered is therefore not to be ruled out. Going back to the slave trade period (1400’s-1800’s) especially the intermingling and absorption of mixed-race descendants of European traders by local populations has been widely documented. For references see:
- “Landlords and Strangers: Ecology, Society, and Trade in Western Africa, 1000-1630″, George E. Brooks, (1993).
- “Eurafricans In Western Africa: Commerce Social Status Gender & Religious Observance”, George E. Brooks, (2003).
- “Western Africa and Cabo Verde, 1790s-1830s: Symbiosis of Slave and Legitimate Trades”, George E. Brooks, (2010).
13) I find it most insightful to separate the North African matches (180) from the total of 437 African matches I detected for my survey participants. Because these North African matches are unlikely to be directly associated with the main origins of West African admixture for Cape Verdeans. Although not per se! As in fact minor Upper Guinean lineage is also very common for North Africans. And therefore Cape Verdeans might also be receiving some of their North African matches due to shared Upper Guinean ancestors. See section 3 for continued discussion.
14) The Fula people go by several names, usually borrowed from other languages. Illustrating their wideranging geographic distribution across the Sahel region. From Upper Guinea into Sudan. The ethnonym used by Fula people themselves would be Fulbe or Pullo (singular). I am choosing however to use Fula. Because historically speaking this was the main term they have been known by in Cape Verde. This name originally being derived from the Mandinga language, which has been greatly influential on Cape Verdean Crioulo (see this link).
In other languages other ethnonyms are more current. For example Peul in French and Fulani in English. These names are likewise historically derived from neighbouring ethnic groups (resp. the Wolof in Senegal and the Hausa in Nigeria). Whenever I am using the word Fulani instead of Fula I am therefore specifically referring to the eastern offshoot of the Fula people. Nowadays residing in places such as Niger and northern parts of Nigeria and Cameroon. Often intermingling with local Hausa people etc. but originally arriving from Upper Guinea. And more specifically the Senegal river area being their presumed homeland. For more details see also:
- Detailed overview of various Fula subgroups across Africa (Jamtan)
- Maps showing geographical distribution of the Fula people (Tracing African Roots)
- The Difference Between Hausas and Fulanis (A Fulani-Nigerian in England)
- Wikipedia article on the Fula people (Wikipedia)
15) 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”. For more discussion see also:
- Did Ancestry kill their African breakdown? (part 1)
- Did Ancestry kill their African breakdown? (part 2)
- Did Ancestry kill their African breakdown? (part 3)
By the way I only managed to find 1 Malian DNA match for my 50 Cape Verdean survey participants (see table 5). Versus 35 from Senegal and 10 from Gambia! To be sure this might again reflect a skewed composition of Ancestry’s customer database. As is also the case for matches from Guiné Bissau. However because of historical plausibility I am fairly certain that even with greater numbers of Malian customers their matches with Cape Verdeans will be relatively subdued when compared with matches from Senegambia or Guiné Bissau.
16) Fula & Hausa-Fulani matches can be distinguished quite easily due to their distinctive genetics. Which is expressed on AncestryDNA by the reporting of “Africa North” and “Middle East” in minor but still considerable amounts. Whenever I have grouped Upper Guinean matches under a particular country (in tables 2, 5 and 7), the implicit assumption will be that these are not Fula matches but rather matches belonging to other ethnic groups. See also section 5 for more details.
At first sight the high number of presumably Hausa-Fulani matches (45) from either northern Nigeria or Niger seems pretty astonishing. And even perplexing when one is not aware of the complete context. After all Nigeria is quite far removed from Cape Verde. While practically no documented records exist of slave trade between both places. Still these Hausa-Fulani matches are second in place after Fula matches in my survey and more numerous than for example Senegalese matches!
I am pretty sure that these Hausa-Fulani matches are ultimately caused by way of shared Upper Guinean DNA. As I have seen several Hausa-Fulani AncestryDNA results and they all consistently showed a considerable “Senegal” score (prior to the update) in addition of also primary “Nigeria” scores. I have also seen the DNA matches being reported for two Hausa-Fulani. And unsurprisingly they often tend to be closely related with Fula DNA testers from Upper Guinea. However only a chromosome browser might confirm I suppose or perhaps also triangulation with Fula matches. See these links for actual Hausa-Fulani results:
A typical ancestral scenario might involve one Fula man residing in for example Senegambia or Guinea in the mid 1700’s. Due to local warfare he ends up being deported as a captive to Cape Verde. His brother however decides to migrate eastwards as many Fula people had been doing then for quite some time already. He finally settles down in northern Nigeria (Sokoto empire). Where his descendants eventually intermingled with the native Hausa people. Resulting in a presentday DNA match with a Cape Verdean!
17) I have always been fascinated by the potential individual African DNA matches may have to construct some generalized matching patterns between Africans and Afro-Diasporans. Being very anxious to learn if these patterns roughly corroborate what we know from historical sources and cultural retention (see this link). And I do believe that my findings sofar for Cape Verdeans in particular have been very valuable and insightful in that regard.
However I strongly suspect that the frequency of DNA matches from a certain place/ethnic group may not always correlate with autosomal contribution, proportionally speaking. In other words just because my Cape Verdean survey participants seem to be extra “matchy” with Fula people does not right away imply that the Fula represent the biggest ancestral component for Cape Verdeans. Although arguably already a case can be made that Fula ancestry from especially the late 1700’s and 1800’s among Cape Verdeans has been very considerable. As in fact this may also be corroborated by historical evidence. However earlier lineage from different ethnic groups may very well not be showing up in adequate proportions yet. When relying only on DNA matching patterns.
Based on my review of historical sources I am personally inclined to assume that a majority of Upper Guinean ancestors for Cape Verdeans would be hailing from at least 20 different ethnic groups. With possibly the Wolof, Mandinga, Biafada, Papel and Bainuk standing out somewhat. Although of course we cannot yet put any exact numbers to such speculations. Regionally speaking I am guessing that most people would have hailed from the presentday territory of Guiné Bissau and Casamance combined. However from historical documentation, cultural retention (language etc.) and now also by way of DNA matches we can be quite sure already that also many ancestors came by way of Senegal, Gambia, Guinea Conakry and Sierra Leone. Again it is unfortunate that not that many people from Guiné Bissau have tested with Ancestry. Otherwise we would see many more matches with them as well. See also:
- Top 20 Ethnic Roots for Cape Verdeans (Cabo Verde Raizes na Africa)
18) It is perhaps also noteworthy that similarities in physical appearance between Cape Verdeans and Fula people are often remarked upon. I have personally heard such observations many times, incl. also by Fula people themselves. For example see this blog post:
- Fula Connection ?! (The Creola Genealogist)
During slavery the Portuguese even applied a separate colour category named “Cor Fula” referring to enslaved Cape Verdeans with medium brown skin tone (see p.121 of this paper). Finding resemblances between Afro-Diasporans and actual African persons is always fascinating. It is something that I have been intrigued by for a long time actually. And I do think something is to be learnt from it. But more so going by generalized trends and not so much on a person to person basis as the individual variation among Africans themselves will be too great. While the multiple African origins of Afro-Diasporans are too complex to be reduced to just one certain “ethnic look”.
As they say phenotype does not always follow genotype (although it usually is correlated even when not perfectly so!). And from that perspective appearances can indeed just be skin deep and even misleading. If you have a good eye for it Cape Verdeans can be said to have certain recurrent & distinctive features. However this cannot be seen apart from the variation in their generally admixed genetics which causes them to still be quite diverse looking across the range. In case of the Fula people it is known they have a minor yet considerable ancestral component in common with North Africa. Which is quite likely affecting their phenotype as well. It is described by AncestryDNA as “Africa North” combined with “Middle East”. These components are also present among Cape Verdeans. But to a lesser degree and quite likely also (partially) inherited by way of Portuguese ancestors. See this table. For a fascinating study on the interplay between genetics & phenotype among Cape Verdeans see:
- Genetic Architecture of Skin and Eye Color in an African-European Admixed Population (Beleza et al., 2013)
19) I made a conscious effort to make a distinction between “mainstream” North African matches and North African Jewish matches (which I grouped under Sephardi matches). I tried to achieve this by looking at plausible admixture combinations as well as filtering for Jewish and Arab/Berber surnames. For more details see also section 5 and the second part of this blogseries. I am fairly confident therefore that few if any “full fledged” North African Jewish matches have been included in my North African tally. However I cannot rule this out naturally for North African matches of partial Jewish descent, especially involving one single distant Jewish ancestor.
20) The “culprit” is most likely Ancestry’s new algorithm which has a greater focus on relatively recent ancestry. Seemingly causing the new “Portugal” region to absorb these older ancestral affiliations with North Africa. Under different settings such North African DNA markers have many times been demonstrated to exist in the Portuguese gene pool. While prior to the update Portuguese customers actually received higher “Africa North” scores than Cape Verdeans (see Iberian AncestryDNA results).
An ancient and complex history of mutual gene flow exists between North Africa and the Iberian Peninsula. Which actually goes back thousands of years rather than being restricted to the Moorish occupation which everyone automatically has in mind. Because of widespread genetic similarities therefore North Africans are usually described by Ancestry as having a substantial southern European component. From what I have seen after the update this is now usually described as “Portugal”. In fact also for Fula people. It is essential to be aware that such scores usually are not referring to any genealogical time frame! Rather they are to be seen as proxies of ancient shared DNA. For more details see also:
- Ancient genomes from North Africa evidence Neolithic migrations to the Maghreb (Fregel et al., 2018)
- Iberian genetic clusters (For what they were… we are)
- Updated North African AncestryDNA result
- Updated Portuguese AncestryDNA result, (second one), (third one)
21) Whenever my survey participants were close family members of each other I have indicated this within my spreadsheet. It should also be noted that at least three of my survey participants are aware of relatively recent Moroccan Jewish ancestry (late 1800’s). This concerns CV03, CV04 (son of CV03) and CV06. Moroccan Jews are genetically speaking quite distinct from “mainstream” Muslim Moroccans. Which is why going by admixture scores “Africa North” will not be the most adequate indicator. I will discuss this in greater detail in the second part of this blogseries. However when going by DNA matching patterns it seems this lineage is also reflected in North African matches and not only measurable by Sephardi matches. The highest amount of shared DNA with a North African match (17.86 cM) was found for CV06. While also the number of matches for CV03 & CV04 was clearly above average, as well as average shared DNA for all three.
I tried to verify the island origins of my survey participants to the best of my ability. All the overviews specifying island origins on this page are merely an approximation though and not based on a 4 grand parents criterium per se. As many people were not fully aware. A greater part of my survey participants had mixed island origins already within the last 2 generations. Let alone going back further in time. Barlavento is referring to people with island origins from São Vicente, Santo Antão & São Nicolau, incl. mixed between those 3 islands. The same goes for Brava & Fogo, which also includes people with combined Brava & Fogo island origins. Regrettably only very few persons had island origins from Boavista and Sal but they were mixed with other island origins as well.
22) Starting with the earliest study Gonçalves (2003): (CVS=Sotavento & CVN=Barlavento)
“A comparison of two African sources with the Cabo Verde islanders indicates that only haplogroup E3b is shared by the Guinean sample and north Africans (Bosch et al. 2001). The relatively high homogeneity of the west Africans here is unlikely to be a consequence of restricted sampling, because the 276 Guineans represent an extremely diverse assortment of ethnic groups, some of them, interestingly, with known historical records of trade with north African people. The two populations of Cabo Verde (CVS and CVN) differ significantly, in an opposite way, in their proportion of sub-Saharan (A and E3a) and north African (E3b) haplogroups, whereas they have a similar proportion of west Eurasian lineages, totalling about 60%. Although the populations of the separate islands of the windward group differ considerably in haplogroup E3b frequencies (data not shown), the difference between the windward and leeward groups probably reflects the effect of drift rather than differences in the source of the settlers. More than one half of the samples of the leeward (CVS) group were taken from its major island Santiago, which showed the lowest frequency of E3b (6%)” , (Gonçalves et al. , 2003, p.471)
Whereas Beleza (2012) has Fogo having the highest North African ancestral contribution (10.4% based on Y-DNA) see this table. Although it does state that:
“To formally assess the paternal contribution of different populations, we performed an admixture analysis, using the approach implemented in Admix 2.0  and treating the Cape Verdean population as a result of admixture of four parental populations: Iberian Peninsula, West Africa, North Africa and the Sephardic Jewish population (Table 1). This 4-way admixture analysis was prompted by previous suggestions, based on genetic and historical data, that enslaved North Africans and Iberian Jews represented non-negligible fractions of African and European parental groups, respectively , , .
According to the admixture analysis, the majority of male contributions to Cape Verde were derived from the Iberian Peninsula (0.68). The second most important contribution (0.27) came from West Africa, while contributions from Northern Africa and Sephardic Jews seem to have been residual (∼0.03 each, with wide confidence intervals).” (Beleza et al., 2012, p.5)
23) I was able to verify the ethnicity estimates of these possibly Upper Guinean related matches because of the Compare Ethnicity tool which Ancestry made available to all of its customers in October 2018. Actually also the two matches from Cameroon might be Upper Guinean related. Given the Fulani presence in northern Cameroon as well as the minor reporting of “Mali” amounts for these matches. But I suspect that this scenario is much less likely for these two persons. Especially since no North African trace amounts were detected for them.
The possibly Yoruba/Fulani match was reported for my survey participant CV40. Adding to plausibility he also received 3 most likely Hausa-Fulani matches as well as 1 Bambara match from Mali.
The possibly Tuareg match was reported for my survey participant CV27. Again adding to plausibility I found 3 additional Hausa-Fulani matches for her as well. Plus I was able to verify that this possibly Tuareg person also is being matched with one Fula and 1 Hausa-Fulani profile which have been shared with me. Implying an eastward migration by mutual Fula ancestors who may also somehow have intermingled with Tuareg people.
Screenshots of possibly Upper Guinean related matches (rightclick and open in new tab to enlarge)
24) This IBD check (see table 11) was of course most complete for survey participant CV04 because I had access to the DNA matches of both his parents (CV03 & CV47) as well. Actually his mother (CV03) has a known North African Jewish ancestor from the late 1800’s. Which most likely explains the rather elevated level of shared DNA (8.23 cM) among the 3 North African matches I found for CV04. Two of which also showed up for his mother but one of them was actually inherited by way of his father (see also this spreadsheet). The overall IBD share of his 5 African matches is quite impressive (80%). However when performing a similar exercise among his more numerous Portuguese matches (74) or Jewish matches (35) this IBD share might very well be lower (see part 2 of this blogseries).
For seven survey participants I could only use the DNA matches from one single parent when performing the IBD check. So this exercise was not fully complete but still already useful. Obviously much more data is needed for any generalizing inferences. But reviewing table 11 it is perhaps already telling that the IBD share among the 21 Upper Guinean matches was higher (29%) than among the 15 North African matches (13%). While for other parts of Africa I did not come across any IBD matches at all. Also interesting to see that despite a minimal shared DNA average (6.32 cM) some North African matches did turn out to be IBD! I suppose the possibility of population matches is still very much present. But otherwise this finding will be in line with a generally diluted inheritance of this type of DNA. Either way illustrating how small matches are not per se to be disregarded but can be quite useful for corroborating equally small amounts of admixture.
25) The frequency of population matches (IBP) is more common than many people might expect. For a clearer understanding these following statements can be useful:
“These segments truly are IBD, but since they exist in a large population, you may see matches on these segments from multiple ancestors. Typically these are small because they have been passed within a population for a very long time, although based on the Anzick ancient DNA matches, they are not always small. Often, in population genetics, these would or could be called AIMS or Ancestry Informative Markers, meaning that they show up in a particular population at higher levels than elsewhere. “
“Are these useful to genealogy? It depends on what you are looking for and the frequency at which they are found in any given population. They wouldn’t be terribly useful in terms of European genealogy, if you’re primarily European, but if you have minority admixture, finding one of these IBS by population segments would be extremely informative.”
“Identical by population means that a large portion of a population group shares a particular segment of DNA. Some people feel IBP segments are not useful and want all of these segments to be stripped away by population (or academic) based phasing software.”
“If you are from a highly endogamous population, you will have many IBP matches, generally on smaller segments that have been chopped up over time, and you will want to use a higher matching threshold, perhaps up to 10cM, for genealogical matching, or higher.
If you have endogamous lines in your tree, but are not entirely endogamous, IBP segments may actually be beneficial because you may be able to attribute matches to a specific line, even if not the specific ancestor in that line.”
- How Phasing Works and Determining IBD Versus IBS Matches (DNAeXplained, 2015)
- Concepts – Identical by…Descent, State, Population and Chance (DNAeXplained, 2016)
Many people misunderstand these type of matches as implying that they have identifiable ancestors from unexpected places. Aside from being caused by endogamy I suppose population matches might also be a reflection of (pre)historical migration patterns. Often not verifiable by genealogy but more so correlating with rather ancient population movements across time and space.
This circumstance may hold some far-reaching implications. As highlighted above IBP matches are not to be dismissed in all cases. However they do require careful interpretation. In the African context countless migrations from the past may result in IBP matches. Perhaps the most impactful ones have been the Fula migrations across the Sahel corridor, from Upper Guinea into Sudan. As well as the Bantu migrations from southeast Nigeria/Cameroon into Central, East and Southern Africa.
My African DNA matches findings for Cape Verdeans certainly seem to testify to the significance of the Fula migrations. The elevated frequency of presumably Hausa-Fulani matches from Nigeria perhaps being most evocative in this regard! However when performing a similar survey for other parts of the Afro-Diaspora I am quite certain that also the significance of the Bantu migrations will become apparent in matching patterns. At times confusingly so. After all I have myself observed on 23andme how Kenyans are able to receive unexpected Zimbabwean DNA matches. Seemingly due to the genetic legacy of the Bantu Expansion from many centuries or even several millennia ago. Likewise Afro-Diasporans might receive Kenyan matches due to shared Bantu origins from Central Africa (Angola/Congo). In a recently published research paper it has been revealed that Angolans were having IBD matches with people from South Africa and also from Kenya/Uganda! See:
- Countries of Ancestry Tool on 23andme (African DNA matching patterns)
- Number of IBD segments among Bantu speakers (Patin et al, 2017)
26) Due do its Timber filter Ancestry may however not always be consistent in how it reports smaller matches. Possibly resulting in leaving out an African match for one of your parents while this match is being reported for yourself. In case your African match is also on Gedmatch you might want to verify again for yourself whether the match is indeed a false positive or perhaps yet an IBD match.
27) In my reading of Cape Verdean history I did actually at times also come across specific historical references (“Arda”, “Mina”, “Congo”, “Angola”) used for African captives outside of the expected Upper Guinea area being present in Cape Verde. Already during the 1500’s/1600’s but also in the 1856 Slave census. Always in very small numbers and seemingly referring to isolated individuals. Their ancestral legacy if it exists will be direct and not by detour via Brazil, São Tomé & Principe or elsewhere. I suppose such lineage has been greatly diluted by now but might still also produce unexpected African matches. I will try to eventually create a new section on my Cabo Verde Raizes Na Africa website to provide a more detailed overview of such references. For now see:
28) Generally speaking I believe that any DNA test result indicating East African ancestry should always be critically scrutinized. I tend to be very sceptical about the degree or frequency of “East African” DNA results reported for Cape Verdeans as well as for other Afro-Diasporans from the Americas. Because this does not fit well with historical plausibility nor cultural retention. Of course one must remain open minded and within it self this topic of any possible East African connection for the Atlantic Afro-Diaspora is quite intriguing. Hypothetically speaking in rare and individual cases I suppose it might be possible to have one single East African ancestor. However I am assuming it would be greatly diluted then in most cases.
But certainly such atypical cases do not justify the currently much inflated level of so-called East African DNA results being reported by various DNA testing companies, across the board. Either haplogroups, admixture scores or DNA matches suggesting such connections. The tricky thing is that DNA testing of course is no exact science, due to faulty algorithms, lack of reference populations etc.. On all fronts, also including DNA matches! Which can very well be IBS or false positives, in particular smaller matches (see section 1 or also this chart). Especially the implied time framing is often unclear. I highly suspect DNA test results suggesting East African ancestry are often merely a consequence of VERY ancient population migrations across the continent (going back millennia instead of centuries). Something which would also be detected among actual West or Central Africans. Irrelevant therefore from a genealogical perspective (last 500 years or so).
I think it helps to be as specific as possible when outlining such results. Making a distinction between various parts within East Africa. Many people may already be aware of the legitimate Southeast African connection (mostly Madagascar & Mozambique). Which might actually also be relevant for Cape Verde because of the known transshipment of Mozambican captives into Lisbon/Portugal in the 1700’s (see section 5 of this page). Strictly speaking East Africa for me would be the Swahili speaking countries: Tanzania, Kenya, Uganda and possibly also eastern DRC, Ruanda/Burundi. While Northeast Africa would be Somalia, Ethiopia, Eritrea, Djibouti and Sudan.
Now when it comes to historically documented ancestral connections between these latter areas and Cape Verde as well as the Americas it is far less apparent than for Southeast Africa. Although it is known that some atypical Trans-Atlantic slave voyages did depart from the Swahili Coast: Mombasa (which used to be ruled by the Portuguese!) and also Zanzibar. But going by actual numbers as well proportionally this flow of people was quite minuscule. Going by documented slave voyages the East African share in Trans-Atlantic Slave Trade might be less than 0.1% (=6,324/6,709,327; see also note 1 of this page).
Actually the Portuguese presence in this area used to be quite substantial during the 1500’s-1600’s, reaching even into Ethiopia and Somalia. But these early historical forays were not slave trade orientated as far as I know. Starting from the first decades of the 1700’s this presence was also practically discontinued. To be frank I think the odds of any meaningful genetic inheritance from these areas (Swahili Coast and further north) are going to be very slim and currently it is greatly overstated in DNA testing. Then again in individual cases it is not be ruled out. But solid follow-up research is required to corroborate such findings.
DNA matches might very well provide more validation. But personally I again feel it is best to be cautious to rule out any false positives. As IBS or very generic population matches might be more rampant than many people realize. Also the possibility & extent of inter-ethnic unions taking place within Africa is often not taken into consideration. Implying that the MRCA for any given match might actually not be of the same background as your DNA match. Just to name one possibility: Swahili traders incorporating Central African DNA (eastern DRC and Zambia) and subsequently intermixing with other East African ethnic groups might result in an East African match at times I suppose.
Specifically for my Cape Verdean survey group it turned out that most East African matches (4/9, see this spreadsheet) did not pass the IBD test. Even if usually only determined by way of one parent. Still it remains fascinating that one Ethiopian match showed a rather robust shared DNA amount of 12 cM. Although actually that particular survey participant did not show any admixture score for the new “Eastern Africa” region on AncestryDNA. Also she has not tested her parents yet, so I could not perform any IBD check. Nor did I find any other possibly supporting clues. Which leaves this finding still somewhat tenuous.
29) A few other persons in my survey also would have made good examples to demonstrate the interplay between unexpected African matches and atypical admixture results. Such as CV25 and CV29 who each had 2 African matches outside of the expected range. CV25 possibly having one match from Ghana/Benin/Togo as well as one rather big match (12,7 cM) from southern Nigeria. CV29 also received one match from possibly southern Nigeria. As well as one from the Ivory Coast (see this spreadsheet).
These survey participants showed distinctive regional admixture scores both before and after the update. CV29 used to have one of the highest “Nigeria” scores (11%) among 100 Cape Verdean AncestryDNA results I have surveyed. This amount completely disappeared after the update. To be replaced however by 1% of “Benin/Togo” as well as 2% of “Cameroon, Congo, Southern Bantu”. The Nigerian match I found for her still seems to be in line with her regional admixture therefore. Also her 8% “Ivory Coast/Ghana” score stood out before the update. Although again this regional score has completely evaporated (probably absorbed by “Mali” right now).
CV25 used to have a singular top ranking “Mali” score (instead of “Senegal”). Also as one of the few (5/100) he received a trace amount of <1% “Native America”. Which was however not maintained after the update. Right now his new “Cameroon, Congo, Southern Bantu” score of 5% stands out. Especially when contrasted against his relatively low overall African admixture of 31%. Resulting in the lowest Upper Guinean proportion among my survey group (81%, see this spreadsheet). All of which suggesting that possibly along one family line some foreign ancestry (Brazil perhaps?) may exist. Introducing these atypical regional admixture components as well as causing the unexpected African matches from outside of Upper Guinea.
- Screenshot of CV25’s AncestryDNA results (before update)
- Screenshot of CV25’s AncestryDNA results (after update)
- Screenshot of CV29’s AncestryDNA results (before update)
30) The minor but still detectable “Cameroon, Congo, Southern Bantu” scores reported for Cape Verdeans are all the more distinctive because arguably this region is least likely to show overlap with “Senegal” and “Mali” within Ancestry’s current African breakdown. It should be noted though that often former “Nigeria” or even “Benin/Togo” scores may have been incorporated in this newly created category. Therefore such scores could still imply Lower Guinean lineage too, aside from strictly Central/Southeastern African origins.
I find it quite useful also to compare with Angolans of recently mixed Cape Verdean descent. I happen to have a few DNA cousins myself with such a background (confirmed also by public family trees). In their case the “Cameroon, Congo, Southern Bantu” scores will be much more striking and even predominant. But still the distinctive combination with “Mali” and “Senegal” also betrays their partial Cape Verdean lineage.
The highest non-Upper Guinean regional admixture scores I observed among my Cape Verdean survey group are perhaps to be seen as atypical outliers (see this compilation). Whenever first confronted with such unusual breakdowns I have always made an extra effort to verify if these survey participants were indeed of fully Cape Verdean descent. Not only to their own knowledge but also by looking into possibly other clues such as DNA matches and migrations they were assigned to on AncestryDNA. I performed such a quality check for all my 50 Cape Verdean survey participants. Including also CV41 & CV43, both of whom confirmed to me they had 2 Cape Verde-born parents as well as 4 Cape Verde-born grandparents.
However as also described in the main text I still cannot rule out the possibility that perhaps such scores are in some selected cases also caused by atypical family histories. Going back more than two generations ago and involving return or back & forth migrations to and from fellow former Portuguese colonies such as Angola, Brazil, São Tomé & Principe and Mozambique. And possibly also other migrant destinations (the USA being least likely because of the quality check I performed).
Such ancestral scenario’s become even more credible when knowing that Cape Verdean migrants during the 1800’s and early 1900’s at times chose to return to Cape Verde bringing along spouse and/or children from foreign countries. Although again I did not come across any apparent clue for this and neither were the survey participants in question aware of these possibilities. Such foreign lineage from more than 2 or 3 generations ago may however not always be recollected in family lore. But it does provide more solid ground not only in regards to diluted Native American scores but also any accompanying non-Upper Guinean African admixture and unexpected African DNA matches.
31) Trace amounts of “Native American” DNA reported for Cape Verdeans could very well be suggesting Brazilian lineage. Because Native American DNA is widely spread among (mixed-race) Brazilians generally speaking. In my survey of 45 Brazilians I found a group average of 8.8% “Native American” (see this blog post). Such an amount could easily have been diluted to about 1% after 3 generations I suppose. In addition there is also plentiful historical scope of Brazilian-Cape Verdean unions leading to offspring. Taking place in either Cape Verde itself or in Brazil. Or perhaps even in Portugal. Sailing routes between Brazil and Portugal often included a stop-over in Cape Verde during early colonial times. While during the 1800’s also a direct salt trade between Brazil & Cape Verde became quite active. Furthermore during a short period of time (1755-1778) northern Brazil & Cape Verde were even falling under the same trading monopoly by the Companhia Geral do Grão-Pará e Maranhão. For more details:
- Atlantic Islands: Madeira, the Azores, and the Cape Verdes in Seventeenth-Century Commerce and Navigation (T. Bentley Duncan, 1972).
- Cape Verdean sailor working in Brazil during the 1800’s: Marinheiro Simão Salvador (Esquina do Tempo, 2014)
- Das Relações Históricas Cabo Verde/Brasil (D. Perreira, 2011)
Minor Native American admixture found among Cape Verdeans may also be explained otherwise however! An intriguing early reference may already be found from the 1500’s when enslaved Native Americans were moved around across various parts of the Americas and even across oceans at times. At one time apparently even involving Cape Verde as a destination for deported Beothuk Indians from Labrador/Canada! (see Loewen, p.58). Also early trading connections with the Hispanic Caribbean may have introduced some Native American DNA within the Cape Verdean gene pool I suppose. It should be kept in mind though that due to the so called wash-out effect any genetic inheritance from these people may no longer be detectable. Generally speaking due to recombination any distinctive autosomal DNA legacy from one single person will only last about 7-8 generations. To be traced back at most to the 1700’s but certainly not to the early 1600’s or 1500’s.
A more plausible source of diluted Native American DNA among Cape Verdeans would be the USA. And more specifically New England where Cape Verdean sailors have been residing ever since they were recruited by whaling ships from that area during the 1800’s and late 1700’s even. It is known that some of the local Native American groups intermarried with Cape Verdean men quite frequently. In particular the Wampanoags. Amazingly such unions have been documented from as early as 1836 already (Shoemaker, p.166)! I imagine that because of occasional return migrations to Cape Verde mixed offspring or even Native American spouses might have ended up in Cape Verde in some exceptional cases. In fact some Native American men may also have stopped by Cape Verde as it appears that they were themselves also recruited as crew members by New England whaling ships already starting from the mid 1700’s (Nicholas, 2002)! For sources and further reading:
- Lies My Teacher Told Me: Everything Your American History Textbook Got Wrong
(J.W. Loewen, 1995)
- Indian Harvest: The Rise of the Indigenous Slave Trade and Diaspora from Española to the Circum-Caribbean, 1492-1542 (E. Woodruff-Stone, 2014) (Chapter 6)
- Mashpee Wampanoags of Cape Cod, the Whalefishery, and Seafaring’s Impact on Community Development (M.A. Nicholas, 2002)
- Native American Whalemen and the World: Indigenous Encounters and the Contingency of Race (N. Shoemaker, 2015).
32) A more detailed description of my methodology for finding Portuguese, Jewish and Asian matches for my Cape Verdean survey participants will be provided in the second part of this blogseries. Although basically it follows the same format as applied for zooming into African DNA matches. First scanning & filtering all DNA matches and then focusing on plausible regional admixture and plausible surnames.
33) AncestryDNA’s ethnicity estimates have been updated as of September 2018. Because Ancestry now provides 43 global regions instead of 26 global regions in the previous version this has impacted my filtering methodology. I have expanded my advanced filter criteria in my online spreadsheet to reflect the new situation (only for DNAGedcom). My main survey findings are however based on the scanning & filtering of DNA matches I performed in the summer of 2018. So therefore only exclusively pertaining to the AncestryDNA version 2 which was current between September 2013 and September 2018.
In my opinion the new AncestryDNA version regrettably has been a downgrade rather than providing meaningful improvement in regards to the African breakdown. This means it has become less easy to distinguish African DNA matches based solely on their regional admixture scores. Another complicating factor is that Ancestry has discontinued the usage of trace regions a.k.a. “low confidence” regions. For my scanning/filtering method this implies many finer distinctions can no longer be made. At least not for any profiles which have been updated already. For more details on the update see:
- Did Ancestry kill their African breakdown? (part 1)
- Did Ancestry kill their African breakdown? (part 2)
- Did Ancestry kill their African breakdown? (part 3)
34) From what I have seen many Africans on Ancestry tend to not have public family trees. And if they do it is usually not going back that many generations. Not beyond the 1900’s at least. One also has to keep in mind that traditionally speaking many African cultures keep track only of the direct paternal and/or direct maternal line. All other family lines, starting for example with one’s paternal grandmother or maternal grandfather and all their predecessors are often not recorded going back more than 3 generations.
Of course for most Afro-Diasporans there will also be a great deal of missing information going beyond the 1900’s. So generally speaking the chances of finding a mutual ancestor by way of comparing family trees are quite slim. Nonetheless I do think that several advanced genetic genealogy tools/techniques might be fruitful in learning more about which possible family line the common ancestor with your African match is to be associated with.
Depending on the amount of shared DNA with your match you can already get an idea of how solid the match could be. However by testing close relatives, preferably your parents or older generations, it makes it much more easier to establish if your match is indeed a socalled IBD match (Identical By Descent). And also on which family line your common ancestor may be placed.
Although I have no personal experience with it triangulation and DNA Painter may also give you more insight. This will be very tricky given scanty information and will also require a lot of patience. But it might still be worthwhile for eventually zooming in closer to your West African origins along a certain family line or even actually identifying a West African ancestor! For a very inspirational blog post see:
- Using DNA Painter to Verify Igbo Origins (Roots Revealed, 2018)
Wow! You’ve done a lot of work and analysis. My family had Mali as their highest African DNA before the Ancestry update. I’m seeing a pattern with my matches that do not seem to have Cape Verdean ancestors, but they do have Madeira ancestors. In at least a couple cases, Mali is their only African region. My 2nd great grandfather was from Madeira, I’m wondering if he had some African Ancestry when he arrived in Cape Verde. Have you done any analysis on African Ancestry in the Madeira Islands?
LikeLiked by 2 people
Thanks a lot for your comment Carol! I have seen several Madeiran results on Ancestry but I have not performed any research specific to their African regions yet. Generally speaking I would assume Madeira received a similar mix of Upper Guinean captives as Cape Verde. Although apparently they had a large share of Guanche slaves from the Canaries in the very beginning of Madeira’s settlement as well. The bloodlines of all these people have been greatly diluted within Madeira’s genepool, but often can still be detected on a subdued level (from what I have seen ~1% SSA). I suspect that after the latest update “Mali” scores have largely replaced “Senegal” scores for Madeirans, as also occurred for Cape Verdeans.
What I found interesting about your uncle’s results (CV25) is that despite having an exceptional “Mali” score before the update I did not find any apparent African match from the Upper Guinea region for him. This could just be because of an underrepresentation within Ancestry’s customer database though. Instead, quite strikingly, I only found two African matches from possibly Benin and Nigeria. As I discuss in more detail in section 4 as well as footnote 29 I am speculating he might have some Brazilian lineage to account for these unusual African matches. Also because of his previous Native American trace region. The back & forth migration of Cape Verdeans apparently sometimes also included their mixed children. So this indeed makes for an intriguing ancestral scenario to explore.
Another fascinating aspect about your uncles’ African matches btw is the rather pronounced number of North African matches (9x) I managed to find for him. Only three other survey participants had a higher number. Also your uncle had the highest “Africa North” admixture score (7%) before the update! See this spreadsheet:
Does your family have any known Moroccan Jewish lineage from the 1800’s? Otherwise I suppose it may be correlated with his above average Portuguese admixture (his 60% “Portugal” score actually was the highest in this survey!). But possibly again there might (partially) be a connection with Madeira and its Guanche heritage as well.
My mother’s maternal great grandfather was from Madeira and some of my family have indicated that they were of Jewish Ancestry. I have no evidence other than that and my immediate family has never identified as being of Jewish descent. My mother’s paternal grandmother’s family had the surname Galina. The first appearance of the surname that I’ve found was in Boa Vista. I connected with some other Galina descendants and asked if they knew where the family came from and a couple of them said Spain but no one is really sure. Maybe the Galina family were Morrocan. My parents, grandparents, great grandparents were all born in Cape Verde. For my uncle to appear to have Brazilian would mean that an ancestor from Brazil moved to Cape Verde. The Native American doesn’t make sense either unless a Native American went to Cape Verde. So other than my recent Madeira ancestor and my mystery Galina ancestor, I think the rest of my ancestors were from Cape Verde and should have a similar African mix to other Cape Verdeans.
LikeLiked by 1 person
I think so too. And to be sure this is also reflected in the predominant Upper Guinean scores in your uncle’s African breakdown (18% “Mali” + 7% “Senegal” out of 31% total African = 81%). However what I find intriguing is how relatively speaking this Upper Guinean share is somewhat decreased for your uncle and actually the lowest when compared with my other survey participants. Your uncle having 81% (25/31) versus 92% as a group average (see this sheet). This may just be a case of individual variation. And again it is clearly predominant. But on the other hand combined with his atypical African DNA matches his 5% “Cameroon, Congo, South Bantu” score might also be suggestive of something else on one specific family line. Perhaps to be traced back further than the 1800’s even?
Wow, amazing post as always!
LikeLiked by 1 person
It’s interesting that you were able to surpass your initial expectations on the presumed Upper Guinean roots of Cape Verdeans, by collecting AncestryDNA results from 50 participants. As you mentioned repeatedly in your blog there were many clues which certainly indicated, overwhelmingly Upper Guinean roots.
This term “Upper Guinean”, in the scope of commercial DNA tests, in general, including African ethnicity results, certainly helps to refine the geographic area and to identify as closely as the designed test allows us to, the selected populations of the dataset.
However, as you pointed in your post, this type of admixture analysis limits our capacity to identify more precisely even the local ethnic groups of an already restricted “macro-region” such as Upper Guinea.
We have already discussed this issue privately and I also went to some extent over it in my guest article that was posted on Maju’s blog, in the comment section.
Considering the steps that I have taken, personally, in order to gain more insight and accurate information on my autosomal DNA results, for the sake of clarity for your readers, I will briefly go over some of the issues of commercial DNA tests.
For those interested in more details, I have addressed them in my replies, in the comment section.
I want to start by saying that I understand that for many Afro-descendants, commercial DNA tests are an opportunity to get more information on the origins of their enslaved ancestors. So, I wouldn’t want to be overly fixated on the shortcomings of these tests. As Fonte suggests, despite inherent imperfections, some helpful insights can still be gained.
Nonetheless, I think that it is still important to mention those “technical” mistakes, hoping that eventually, they can be improved for the benefit of DNA testers.
There are principally 2 issues with DNA tests.
The method which are used to run people’s data and frequently the populations that are selected as part the dataset.
Individual samples are analyzed in categories that have been “arbitrarily” selected. They don’t let each individual sample form itself. So ALL the existing individual variations of the dataset are ignored. This element, I suspect, increases standard errors, and produces overall more bias.
Secondly, populations such as the Mozabites, isolated Berber Tunisian communities, Jewish samples, Fulani samples have morphed into respective genetic homogenized/single component, when they have been identified by various genetic studies, as a result of founder effects. They tend to cause problems and distort the outputs at higher K values when the dataset is not properly designed and/or depending on the algorithms of the software. It’s just not safe to use them as reference population, in this context.
In the study from George BJ Busby et al., “Admixture into and within sub-Saharan Africa. ELife 2016.”, even in the most western parts of West Africa, they reported different admixture events and signals, and importantly, different types of admixture for ethnic groups of the same region or even the same country.
check this link: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4915815/
In this case even a so-called Upper Guinean category which most likely groups a variety of already genetically differentiated ethnic groups, doesn’t inform beyond the macro-regional scale. I think that it is a mistake to have so-called admixture categories based on modern-day African countries (incl. Senegal, Mali etc…) or regions. There is simply too much existing genetic diversity (even despite occasional inter-mixing) that’s inevitably ignored in these analyses which ultimately produces inaccurate results.
I have addressed a similar issue in the guest article, following the claims from a Dominican gentleman, in search of his roots, of seeing a “Fulani-specific” component in the new 23andme Senegambian category.
“Your description of “fulani-specific” component in the context of the so-called “Senegambian” category from 23andme, is very incorrect.
The category called “Senegambian & Guinean”, according to their own description, groups people from Senegal, Gambia, Guinea-Bissau and Guinea. Needless to say, that it has many different ethnic groups that are found all across the region. In addition, the method used by commercial DNA companies produces imprecise results. […] The Fulani-specific component which was identified in the Henn’s study at k=10, by Maju, as well as in my work, is produced by using the unsupervised mode of ADMIXTURE, where individual members of specific “ethnic categories” are distinctively identified.
It’s completely different from the approach that consists of selecting people on the basis of recently established borders/countries or regions which aren’t reflecting the genetic or even the linguistic diversity. This is why, I suggested to you and others to become more familiar with the history of the region. Commercial DNA test can differentiate between continentally and regionally (in some cases) separated populations. With endogamous or ancient populations, the true ancestral components won’t be detected and identified.”[…] Based on the information that you provided, you have no evidence to claim a fulani-specific component. If you don’t properly compare your data using the unsupervised mode of ADMIXTURE with the right populations, there is no way to know.”
For admixture analyses, this is what I replied:
“After ADMIXTURE, I experimented with many types of software that are used in the recent studies on autosomal DNA by current geneticists. Naturally, each of them has its pros and cons, and ultimately its own limitations, depending on a myriad of factors and what’s expected from the user. For present-day modern populations, with those peculiarities, ADMIXTURE works fine, if “the sampling strategy” makes sense.
For ancient populations, qpAdm, f4/D stats might be better adapted than ADMIXTURE, just to give an example. Checking for standard errors and adjusting to it, is also the key.”
LikeLiked by 1 person
Thanks so much for this detailed comment. Clarifying the shortcomings of admixture analysis, which certainly do exist! We have indeed had a fruitful exchange of ideas in private about these issues, also afterwards. Just for brevity’s sake and for any one else who might be interested I will leave a link below where my own stand on admixture analysis is described (in particular see reaction 7 and footnotes 4-8)
Also at the time of writing this comment you & I were not yet aware of how the 23andme update worked out for people of Fula descent and other populations from the Upper Guinea region. These updated results can been seen over here:
I have been most impressed by the way 23andme is now able to indicate the Upper Guinean lineage among Hausa-Fulani and even one Sudanese Fulani by way of their new “Senegambian & Guinean” category. And also how they separate that from “Nigerian” and “Sudanese” admixture. Not corresponding with true genealogical proportions naturally but still quite insightful I would say. The Sudanese Fulani sample very interestingly also received a socalled ancestral location for both Guinea and Nigeria! This last feature is based on having shared IBD segments for atleast 5 people and above a certain threshold within 23andme’s database (see this link).
He probably was already aware of course but to me this suggests this feature is potentially very informative also for Afro-descendants. Although they do not seem to receive such African “ancestral locations” yet (Cape Verdeans do very consistently receive “Cabo Verde”). Possibly due to failing match strength and the thresholds being set up too high. Probably also 23andme needs to expand their African sample dataset. Ironically Afro-descendants almost always do get European ancestral locations (in line with their colonial history). Given my findings in this post on the African matching patterns for Cape Verdeans on Ancestry I expect that “Guinea” will eventually also appear as additional ancestral location on 23andme for Cape Verdeans, indicating their match strength with that area.
LikeLiked by 1 person
Thanks for taking the time to reply, although you had already addressed this topic in the links which you shared with me, eventually later.
Indeed, this new update has improved the results of the samples that you shared with me. It seems that a better selection of samples and a refinement of macro-regional categories more in phase with historical records had positive effects on it.
Following your suggestion to use IBD segments, I used PLINK 1.07 pairwise IBD estimation and shared IBD. I didn’t know what guidelines or methods to follow at first, but I checked ISOGG wiki and was able to understand some notions. I used the dataset with the Fula samples from Nigeria. I set a minimum of 5 cM and 100 SNPs for the detection of segments. It reported that I share a total of 773 cM across 88 segments with those Fula samples, which is likely the result of endogamy. In contrast, I only matched with 2 Hausa samples for 24 cM over 3 segments.
LikeLiked by 1 person
Nice corroboration! I wonder if any of those Hausa samples might still yet have some distant Fula lineage they may not be aware of. I have been reading the Fula migrations into northern Nigeria already started in the 1600’s or even 1500’s? Even when it culminated later I imagine especially during the Sokoto Caliphate of the 1800’s.
Just out of curiosity how are the eastern Fulani from Nigeria or even beyond seen among the western Fula communities in Guinea and Senegal? As distant relatives or markedly different despite commonalities?
LikeLiked by 1 person
Based on what I know from the elders of my family, and also from the ones of surrounding communities, there is a cultural continuity and in some cases, recent common ancestral origins. The language has morphed into different dialects, but there is still a common foundation relating to semantics.
I had mentioned back migrations from eastward bound Fula and so far, admixture analyses and matching patterns over the months seem to corroborate it. My data and the one of my mother share more DNA with eastern Fulani people whose recent ancestry is from Sudan, around the Red Sea, Cameroon and Nigeria than it does with Fulani from Senegal and Guinea. The shared cM values per individual are usually higher and more frequent.
My mother’s data also shares matching segments with many different ethnic groups from Nigeria and Cameroon which corroborates her admixture analyses. It reported that she shares a single segment of 19.9 cM with a Bamoun sample and another one of 19.8 cM with a Bulala sample. It’s possible that maternal lineages in their complexity may have been underreported or forgotten in some cases.
Another pattern which seems to emerge, is that those eastern Fulani with reported Y-DNA, have E1b1b-M81, R and G haplogroups which aren’t typically found in West Africa. They have, however, mtDNA L(xM,N) lineages that are common in West Africa. It seems that perhaps the absorption of Fula lineages was manifested via their maternal line. Quite similar to the findings from Begona Dobon et al., (2015).
LikeLiked by 1 person
Your findings on the affinity between Cape Verdeans and Fulani communities of West Africa are interesting. I agree with you that Fulani samples may indeed be over represented in the customer database of commercial DNA. Primarily, because many of us have heard so many unverifiable hypotheses on the so-called origins of Fulbe that we are more likely to be interested in it, in comparison to other West African communities.
It’s still approximate, in my opinion, for the reasons I have mentioned already. While there are strong correlations and patterns between your surveys and various historical accounts in the region, it could also be that the MRCA between the samples belong to other ethnic groups, considering the chaotic patterns of the Slave Trade, given internal historical developments and also the patriarchal structure of traditional Fulbe communities.
If we are making hypotheses based on non-commercial DNA sources, but current academic evidence, a single component has been identified in many genetic studies and independent experiments, for Fulani people.
This is what I wrote on the subject in the guest article.
“In 2012, a genetic study was published (Henn et al. 2012):
Here is a link https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1002397
or see the link of the post on this study that was made by Maju on this blog:
They have highlighted the different ancestral categories of North African and other tropical African populations by including the Yoruba, the Bulala, the Fulani, the Maasai and Luhya. It covered a wide range of North African populations from the Western Sahara to Egypt. Of course European Basques and Middle Eastern Qatari were included for “control” purposes.” […] At k=8, it is clear that the peculiarity of the Fulani; in addition to their West African ancestry, is what looks like an ancestral component similar to the ones of Berber ancestry. At k=10 (see section “Supporting Information”: Figure_S1.tif) , the Fulanis form a “Fula” specific component restricted to them only, indicating, an old trans-Saharan admixture event, which after enough generations of endogamous patterns led to a distinctive genetic population in tropical Africa. Intriguingly, Maju had pretty much ran similar exercises which corroborated the findings of Henn, just weeks earlier.
See here: http://forwhattheywereweare.blogspot.com/2011/12/north-african-genetics-through-prism-of.html
At k=8, he had noticed the same “Fula-specific” of the Henn’s study.
His description: “A Fulani-specific component shows up. Intriguingly it is almost equidistant by Fst measure from the Mandenka and the Sahrawi components (0.105 and 0.115 respectively). All the North African specific components are much closer to West Eurasian ones than to the Mandenka component, so this might suggest a very old kind of trans-Saharan admixture, then homogenized in a single component.
[…] there is systematically a common old trans-Saharan component.
If there are differences, I would say that it is mostly due to inter-mixing with local populations of the geographic area where they live.
There are several studies other than the ones that I have already mentioned to you with links.
This study, “Begona Dobon et al., The genetics of East African populations: a Nilo-Saharan component in the African genetic landscape. Nature – Scientific Reports 2015″
Here is the link: http://www.nature.com/articles/srep09996
Maju made a post on it some years ago.
If you observe the Fulani samples from Sudan, at K=3, they shared ancestry with local Sudanese ethnic groups. At k=5, however, they display an additional unique specific component of the color pink. The authors have suggested:
“A population that shows signals of recent admixture is the Fulani. Fulani are nomadic pastoralists who speak a Niger-Kordofanian (Niger-Congo) language and occupy a large area in Africa’s Sahel. Their origin is still controversial, as mitochondrial DNA indicates a West African and traces of North African origin23, whereas Y-chromosome studies showed shared ancestry with Afro-Asiatic and Nilo-Saharan Sudanese populations8. This shared ancestry with East African populations can be seen in Fig. 3 (k = 3), suggesting that they have admixed with local populations.”
Secondly, the paper Gurdasani et al.2014 showed in figure 7 b. an exercise which included Fula from Gambia, dating an eurasian admixture around 100 generations ago and a second one btw 320-780 ago.
Lastly, the paper Busby et al., 2015, Figure 6-figure supplement 1 shows that the penetration of IBS and TSI gene flow in Africa before 0CE was via Northeast Africa.
The authors stated that that they had identified some southern European ancestry in the Fulani which suggested that it might have entered West Africa via North East Africa.
This is a passage from the paper: “The Fulani, a nomadic pastoralist group found across West Africa, were sampled in The Gambia, at the very western edge of their current range, and have previously reported genetic affinities with Niger-Congo speaking, Sudanic, Saharan, and Eurasian populations [Tishkoff et al., 2009; Henn et al., 2012], consistent with the results of our mixture model analysis (Figure 4A). Admixture in the Fulani differs from other populations from this region, with sources containing greater amounts of Eurasian and Afroasiatic ancestry, but appears to have occurred during roughly the same period (c. 0CE; Figure 5).”
Check this link here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4915815/
or this one https://elifesciences.org/articles/15266
On this blog, Maju made a quick post on it.
This is likely the same source of admixture. So already, there are Fulani samples from Nigeria, Sudan, Gambia and my anecdotal data for what it’s worth, since my grand parents were from the Fouta Djallon region of present-day Guinea. The Y-DNA study from the paper of D’Atanasio 2018, suggests a strong correlation between haplogroup E-M2_Z15939 and Fulani communities from various areas in the Sahel.”
A common mistake, in the case of Fulani samples, is to confuse the North African component of their single/homogenized trans-Saharan admixture with modern-day, regular North African/Berber ancestry. They are obviously different.
In my personal case, I merged my private data with publicly available datasets from the same academic studies, by including samples from Fula communities, North Africa. and of course West Africa. I used the unsupervised mode of ADMIXTURE and based on the outputs, I was able to distinguish which ancestral components indicate trans-Saharan/Fulani, North African/Berber, and just regular West African ancestry. I also used malder and alder which detects admixture signal events, using 2-ref weighted LD curve.
LikeLiked by 1 person
Good points and thanks for providing those insightful links!
” I agree with you that Fulani samples may indeed be over represented in the customer database of commercial DNA. Primarily, because many of us have heard so many unverifiable hypotheses on the so-called origins of Fulbe that we are more likely to be interested in it, in comparison to other West African communities.”
Yes quite understandable. I was intrigued by these comments you left on Maju’s blog:
In particular I was wondering if you could perhaps expand on your family’s more recent ancestral ties “up North” as well as the fascinating Arma people and what is known about them in local traditions.
LikeLiked by 1 person
Yes, of course. As you know the Fouta Djalon is located in present-day Guinea where my grandparents are from. However, my grandfather’s recent lineage is from the Massina area (=Empire du Macina of the 19th century). So my lineage just few generations ago was not from present-day Fouta Djallon. Going back in time, his family has claimed an Upper-Nile origin. Genealogical records which only list the names of men in his family going back several generations have been kept and translated. The names are commonly found in the Central Sahel.
The Massina area in Mali, is part of the Sahelian corridor that’s located up north of the Futa Djallon. Historically, it’s also an area which is at the border between the Sahara and the Sahel.
I also used IBD tools, as you suggested, this time with North African samples of the dataset and I share IBD segments with 9 of them. Saharawis, South Moroccan and Algerians. Intriguingly, 1 segment of 32 cM was reported. I have the outputs if you interested.
Haha, the Arma people. It’s been one of many pervasive narratives that I have heard in various communities, not just Fulani. The “Morianais” from Guinea also.
Any in case, it has nothing to do with my known or recent family history that I know of.
LikeLiked by 1 person
I really think that there are 2 approaches to the study of population genetics, in general.
Academic/peer reviewed studies investigate admixture more deeply with a much bigger time span than the average DNA test.
Commercial DNA deals more with recent genealogy, I would say. Even for Afro-descendants, they will tend to get more information or clues on their genealogically related ancestors than they will on more distant or ancient admixture.
Which is precisely what’s needed in order to make the distinction between older populations and/or groups with peculiarities from other communities of the same macro-regional area. Fulani people are a perfect example. And there are many other ethnic groups in this case, in Africa.
Consequently, the admixture tools, the strategies, or methods that are used by academic geneticists and DNA companies won’t be the same. They serve different purposes that will ultimately produce different outcomes.
Here is a quick list of various recent academic studies on the autosomal DNA of different African populations.
Systematically, unsupervised ADMIXTURE is used. It certainly has its limitations, and should be used with good “sampling strategies”, but for this type of autosomal analyses, it is really helpful.
Link for ADMIXTURE: http://software.genetics.ucla.edu/admixture/
LikeLiked by 1 person
Indeed, as I always mention I favour a multidisciplinary approach to tackle the challenge of Tracing African Roots for Afro-Diasporans. I have in the past reviewed a couple of relevant academic studies (see this section). And they often provide helpful additional insight to make better sense of commercial DNA test results. However I do find that even with peer reviewed studies there might still be a chance an article might be lacking rigour in some aspects. I am therefore always wary of bias of any kind (sampling or otherwise). This will often occur unintentionally of course.
DNA studies are naturally focused on the technical aspects of whatever they are researching. However genetics is embedded in the social sciences as well. Especially the history of population migrations. I often find that DNA studies can be faulty in providing an accurately detailed historical framework and sometimes the historical details being mentioned can even be plain incorrect! Which often results in awkwardly formulated conclusions. This is something I strive to avoid whenever possible, by keeping thing open-ended, contextualizing as well as explicitly pointing out any limitations.
In this particular blog post I chose to combine admixture analysis with DNA matches. Which I believe can be rewarding and reinforcing in many cases. Both aspects of genetic genealogy obviously do have their own shortcomings. But admixture results can reveal insightful ancestral connections even when DNA matches are not (yet) available and vice versa.
LikeLiked by 1 person
Indeed! That’s why we need more scholars like you LOL.
In scientific communities, too often their methodologies and tools are considered to be the only mediums for discovery. And of course there are other heuristic processes that can also be helpful. Frankly, I rarely read conclusions from papers. The data that’s analyzed is what I look for and try to make sense of.
In the case of ADMIXTURE. To paraphrase my follow-up on the guest post, the estimates are not meant to be interpreted literally. The clusters are not “real” populations. They serve to evaluate variation between sampled individuals.
So the FST values can be useful to detect bottlenecks in samples, ranges where limited variation would be obtained.
I prefer to think of it as a tool which evaluates “distance” between clusters or populations and not as a system of precise measurement for mixtures.
Ideally, it works well for recently admixed groups like African Americans or Hispanic Americans because they derive their ancestries from populations that are continentally separated with higher FST values.
The total complex and chaotic patterns of ancient human migrations couldn’t reasonably be captured by admixture tools, anyhow. In spite of it, with correct interpretation some interesting insight can still be gained.
LikeLiked by 1 person
I agree with you that there are valuable elements of information that can be useful from surnames, in the cases of Fulbe communities relative to other groups, but also possibly from within the same groups.
Having said that, I would say that it deals more with fairly recent common geographic locations and/or shared linguistics, in this case the pulaar language. I am unsure that surnames systematically translate into common genetic ancestry for several reasons.
Considering my own anecdotal family story, for what it’s worth of course, my direct great grand father had a different last name/surname than me or my father, for example.
In many Fula communities of the Fouta Djallon region, the identification of paternal lineages isn’t necessarily based or traceable with the use of last names or surnames. In many towns that have been established prior the colonial era, there are as many complex family histories and lineages as there are towns or distinct territories, just to give an idea. Usually the elders, or the descendants of local rulers have either kept records of their genealogy, or through oral history this sort of information has been passed. However, the reliability of such stories isn’t always certain.
In those times, up until the early to mid 20th century, the social stratification with the notion of “caste” and nobility within these aristocracies was still strong. Communities which had been made captives or that had statuses similar to what servitude would be considered, would be given similar surnames of the social categories that they were immediately under or subject of. The other aspect to consider, is that many Fula aristocrats inter-mixed with other populations while changing or keeping surnames for reasons that depended on the local contexts of those eras.
Only the descendants of so-called Fula aristocrats who inherited land properties and have had some deep and strong presence in the traditional local activities of the geographic areas via theology for those who were Muslim scholars or simply rulers, really have some credibility to make claims.
Naturally, I am not condoning these practices or associating morality to them, but simply trying to explain the limitations of relying excessively on so-called surnames, when considering how these societies were organized.
LikeLiked by 1 person
“Communities which had been made captives or that had statuses similar to what servitude would be considered, would be given similar surnames of the social categories that they were immediately under or subject of. The other aspect to consider, is that many Fula aristocrats inter-mixed with other populations while changing or keeping surnames for reasons that depended on the local contexts of those eras.”
Thank you for this insightful comment Thierno, much appreciated! Especially the part I quoted above. I find that such a frank insider’s perspective is often lacking in discussions of Afro-Diasporans wanting to trace back to specific lineage. Sometimes it can be a bit sobering. But I imagine for those seeking the truth it would be preferable over wishful thinking.
LikeLiked by 1 person
The North African connection with Cape Verdeans is intriguing. I think that you have explored all the possible historical scenarios.
Recent study by Marieke van de Loosdrecht and his colleagues, “Pleistocene North African genomes link Near Eastern and sub-Saharan African human populations “seems to suggest strong affinities with West Asia, based on the samples from Taforalt in Morocco, which are associated with the Iberomaurusian culture. The authors have rejected the European Upper Paleolithic hypothesis.
It seems that the most recent scenarios could be more applicable than more ancient hypotheses which are now rejected.
I am very skeptical of the reported ethnic percentages for the North African category. Of course, the use of Mozabite or isolated Tunisian Berber samples doesn’t provide good representative samples for the whole population of Berbers who are located in Northwest Africa. They form their own respective components (as shown in many academic studies on the autosomal DNA of Northwest African populations) which genetically diverge from other regular Northwest populations such as the Saharawi or Moroccan samples.
And that’s in addition to the obstacles that I described already with the “arbitrarily” selected clusters of commercial DNA tests.
So, it’s just difficult, in my opinion, to rely on these ethnic percentages, even though, there are some elements of information that can be extracted for sure.
For Afro-descendants who may have West African origins and/or at the same time possible North African ancestry, the possibility to identify very precisely their reported North African ethnic percentages, whether it belongs to the old trans-Saharan Fula component, so in this case, in addition to West African admixture, or an actual Berber ancestry, is very challenging. A major obstacle that DNA tests, at this point, can’t elucidate.
LikeLiked by 1 person
It is very challenging indeed. However as argued in this blog post DNA matching patterns might very well offer additional clues, or even corroboration (on a population level). It will be very interesting to see if any of my Cape Verdean survey participants also share “North African” DNA segments with any of their Fula matches. This is something I hope to investigate eventually. Although the absence of a chromosome browser on Ancestry does unnecessarily complicate things.
LikeLiked by 1 person
Intriguingly, the new 23andme update has a category called “North African & Arabian”. It has people of Algerian, Bahrani, Bedouin, Egyptian, Jordanian, Kuwaiti, Moroccan, Mozabite, Palestinian, Saudi Arabian, Tunisian, Emirati, and Yemeni descent.
When you look at FST values of Intra-European/Mediterranean autosomal DNA, they fall around 0.01. Even less between Palestinian and Italians/Greeks. In contrast, between Northwest Africans and West Eurasians, it is around 0.06. There are gradual variations between Northwest Africa and Egypt with greater affinity to populations from the Levant, as we move East.
By grouping Northwest Africans with East Mediterraneans and other populations from the Arabian Peninsula, especially in a supervised mode of analysis, the allele frequencies of the category don’t just reflect Northwest Africans but also the ones of other groups such as Palestinians, Jordanians etc… Such groups actually share closer affinity to other West Mediterraneans than Northwest Africans do.
It becomes a comparison between East Meds and West Meds for samples who derived much, if not the majority of their ancestry from Iberians. I ran into the same problem, initially, in the analysis of Hispanic Caribbeans. It’s very difficult to compare and pinpoint clusters or populations with FST values that are around 0.01 or below. For this reason, we tried to restrict the samples to Moroccans and the Saharawis.
In the case of the Senegambian category, the inclusion of Fula samples doesn’t seem to cause similar issues probably due to the higher degree of genetic variation among groups of the nearby area, in general. Just FYI, FST values between Fula and other West Africans (Yoruba and Mandenka samples) is btw 0.08 and 0.10. Btw Fula samples from Nigeria and North Africans, 0.12.
The access to the data and the methods that are employed during those processes allow to have an informed opinion on studies, even when the reality is that few findings from studies in various fields replicate over time. It’s a benefit that companies, where there is more opacity, don’t offer.
The combination of your surveys with shared IBD segments is certainly very useful in order to help folks zero in a particular line of their ancestry.
There is a distinction between what I referred to as “unverifiable hypotheses”, in the case of narratives that are commonly found among Fula communities and actual recent ancestry which are not the same and are unique to each individual’s family history.
Many communities can have mythologies or stories that are attached to their ethnogenesis, and it may be part of their culture, due to very old pilgrimages and nomadic lifestyles, or else. They should be taken with a “grain of salt” and not interpreted for evidence of any ancestry.
At the same time, they may also reveal elements of information about the communities in other fields such as anthropology, especially with regards to pastoralism, but also linguistics and theology.
LikeLiked by 1 person
It is indeed not an ideal grouping! If I’m not mistaken 23andme already applied such a grouping before the current update. I am sharing profiles with about a dozen North Africans on 23andme and I have seen several other ones posted elsewhere as well. Generally speaking the socalled “North African & Arabian” region will not be fully covering their Maghrebi background. In addition almost always a substantial amount of “Southern European” affinity is picked up. As you know this also shows up in lesser %’s (but probably for the same reason) for many Fula profiles on both Ancestry and 23andme. Undoubtedly in most cases caused by a (very) ancient nexus of overlapping North African & Iberian genetics. Although other ancestral scenarios might also apply in particular cases I suppose.
Interestingly this breakdown below shows a Moroccan with a slightly elevated SSA proportion. And in addition to the expected “Senegambian & Guinean” score also a socalled “Sudanese” score is appearing! I have seen other North African profiles as well with minor “Sudanese” scores. I imagine it might represent a mislabeled and ancient ancestral component which still would indicate an ultimately Northeastern African origin. But I am not sure. Then again I suppose the “Sudanese” category might also suggest relatively recent ancestral connections with Chad and northeastern Nigeria.
By any chance have you ever read this paper?
Genomic Ancestry of North Africans Supports Back-to-Africa Migrations
The “Senegambian & Guinean” scores seem much more straightforward and to be explained by historical geneflow caused by Trans-Saharan trading networks in the last 1000 years or so. And also to be corroborated by IBD matches. Which I actually have done by looking into the DNA matches of a few North Africans I am sharing profiles with on Ancestry.
LikeLiked by 1 person
Several sources have shown that haplogroup E-M78 and E-M81 are the most frequent haplogroup in North Africa. With an origin of E1b1b that’s ultimately from around NE Africa, it may reveal a complex story of old admixture events that are still manifested somehow.
“By any chance have you ever read this paper?
Genomic Ancestry of North Africans Supports Back-to-Africa Migrations”
I read about the findings of Henn and her colleagues. I believe that I have shared this source with you privately, in direct relation with admixture analyses of Fula samples, many times, already. In the sampled data, Luhya and Maasai samples were used. They warned of the interpretation of any direct migration from Kenya to Morocco.
Did you know this paper? “Ancient Egyptian mummy genomes suggest an increase of Sub-Saharan African ancestry in post-Roman periods”
The authors have presented data showing that present-day Egyptians can be modeled as mixture between ancient Egyptians and Tropical Africans. Fig. 5C display f3-statistics with the most likely Z-scores which indicate the most likely source of populations. Mandenka samples have one the highest scores among African groups. They also used ALDER with 2-ref decay for Yoruba and French (generations) and estimated admixture event 24 generations ago (700 years ago), corroborating recent estimates from Henn et al (2012).
This connection due to the extended trans-Saharan trade, but also to other related commercial activities, had been corroborated. I also remember sharing with you some admixture analyses with North Africans who displayed affinity with Mandenkas, in the presence of Fulbe samples.
There is another important element which came to light during the sequencing of trans-Saharan patrilineages among 8,000 subjects from 145 African, Eurasian and African American populations in D’Atanasio et al., (2018). The data revealed that there is a dicrepancy in the reported clades of African Y-DNA. They are geographically restricted to different macro-regions which suggests that the presence of North African specific clades, dates back to ancient times during the Green Sahara period.
So a trans-Saharan trade involving a sex-biased SSA contribution to the North African gene pool via mtDNA and reflected in autosomal ancestry. It corroborates the findings of other sources.
“Extensive female-mediated gene flow from sub-Saharan Africa into near eastern Arab population.” From Richards M, Rengo C, Cruciani F, Gratrix F, Wilson JF, Scozzari R, et al.
Wright J. The trans-Saharan slave trade. London and New York: Routledge; 2007.
LikeLiked by 1 person