Saturday, 5 May 2018

Unpacking our African ancestry

EDIT 9 May 2018: My theory of Haitian ancestry is proven wrong. Our African DNA comes from a different line! This blog post, however, analyzes my great-uncle's African segment without regard to my former theory, and the information given here is therefore still accurate.

My paternal grandmother's mother father has distant African ancestry and a trace of African admixture. I discovered it when I noticed a rather large (West) African segment in my paternal grandmother's brother's chromosome paintings on GEDmatch. Further detailed comparison with Haitian matches on that segment confirmed shared European, African and Asian heritage. The European and African is consistent with Haiti's history of colonization and slavery; the Asian could be Malagasy, as suggested to me by the highly respected genealogist and blogger TL Dixon.

My family's African trace is so small that some might say it makes little sense to try to analyze it further. My curiosity about my family history, however, never wavers. More importantly, remembering our enslaved African ancestors as individual human beings with their own cultures, languages and histories is an important way of honouring them and acknowledging their existence and humanity.

Gorée Island off the western tip of Senegal. The island was the largest slave-trading centre on the African coast from the 15th to the 19th centuries. As genetic and historical evidence indicates that we have ancestors from Senegal, it is very likely that some of our ancestors were shipped to the New World via Gorée. Photo: Justin D. Pyle (Source; public domain)

West Africa

What is quite clear is that my grandmother's brother's significant African segment on chromosome 4 is primarily West African in origin. Julie Mulroy has calculated the segment to be around 9 cM long. West Africa, of course, is a very good fit with what we know about the history of slavery in the Caribbean. The West African origin of the segment is shown most clearly in the HarappaWorld chromosome painting, where it is painted in black (the accompanying Asian DNA, probably Malagasy, is shown here as red "S-Indian", and the grey is "Pygmy"):


Because both HarappaWorld and puntDNAL K15 finds* West African (the latter picking it up at 0.60%), TL Dixon wrote to me that he believes it is real: "if both are [picking it up] it is likely to be of West African origin". However, he also believes it might be far back, e.g. from the 1500s or 1600s. (My current theory is that our closest fully African-descended ancestors were born in the first half of the 1700s).

I discovered this West African segment - and the Haitian matches sharing it - before FTDNA's last upgrade of myOrigins. Before the upgrade, my grandmother's brother showed no African at all in myOrigins, but after the upgrade, he got assigned a <1% trace of North African! The GEDmatch calculators all seem to agree that his African admixture is Sub-Saharan,** not North African. I have been advised that this probably indicates that our ancestors came from the northernmost parts of Western Africa, perhaps the Senegambia region. This fits well with the historical fact that many Senegalese people were taken to Saint-Domingue as slaves, notably people of Wolof ethnicity (a heritage that still shows traces in the vocabulary of the Haitian Kreyòl language today). I believe it is quite likely that some of our ancestors were Wolof.

The way to truly prove this, of course, would be to find Wolof cousin matches living in Senegal today (or with very recent Senegalese ancestry). We do not have any such matches yet, but with a bit of luck, they might show up in the future.


The pie chart above is my grandmother's brother's puntDNAL K13 Global result. It is almost identical with his FTDNA myOrigins result in every respect (including the Native American and South Asian), except that the majority of his African admixture is read as West African rather than North African. This supports my conclusion that our African ancestry comes from the borderland between North and West Africa.

Interestingly, puntDNAL K13 and other calculators give my grandmother's brother not only West African percentages, but also percentages from other areas of Africa. In the chart above, for instance, there is 0.34% South Africa. This might be noise, or it might be related to our Malagasy side, or it might be something very ancient and unknown. In the following section, I will be diving deeper into the various regional contributions to our African admixture, looking at the kits of my whole family.

The "just for fun" part: Digging into the traces

When I run the autosomal raw data of myself, my father and my grandmother's brother through the Africa-centric Ethiohelix and puntDNAL calculators at GEDmatch, several of them find traces of various African populations in our admixture. My paternal grandmother's phased kit (made from a comparison between my father's and my grandfather's kits) shows an even more detailed African breakdown. Of course, this phased kit only contains 50% of my grandmother's autosomal DNA, and admixture estimates on such kits should be interpreted with extreme caution. My grandmother's results, however, seem to match the rest of us very well.

In the sections below, I will use the Ethiohelix K10+French calculator as my point of departure. The "French" reference is a proxy for European, which makes this calculator useful for people of mixed European and African descent.

Before anyone berates me for relying on a single calculator, or for putting too much stock in very small admixture traces: This section is (mostly) just for fun. But what's the point in doing genealogy if there's no fun in it? Indeed, it is possible to be serious about something and have fun at the same time.

Note that the admixture figures are probably all a bit inflated; if my theory is correct, my grandmother and her brother have only slightly more than 1/64 (1.56%) African ancestry in total according to the paper trail. However, they might have inherited considerably more or considerably less African DNA than that, due to the random nature of autosomal recombination. To me, it is not the amount that is most interesting, but the category breakdown.

K10+French gives all of us Khoi-San admixture. I have 1.11%, my father has 0.46%, and my grandmother's brother has 0.15%. My grandmother's phased kit gets 1.31% Khoi-San, which could, in reality, be as low as half (i.e. 0.66%). Even in that case, it seems that my grandmother has more Khoi-San than her brother, which explains why my father and I also have more than him. I have no idea why my percentage is as high as 1.11%; it seems unrealistically high, and some of it may well be noise (my mother has 0% Khoi-San).

The Khoi and the San are the indigenous people of southern Africa, including Cape Town, where I am currently living. It's fascinating that I have a genetic connection to the original inhabitants of this place, even though that connection might be thousands of years old (and possibly in a completely different location). TL Dixon has suggested that the Khoi-San admixture could come through Malagasy ancestors, since their Bantu ancestors would have mixed with indigenous peoples of southeastern Africa before sailing to Madagascar. In that case, the Khoi-San admixture is in fact real, though very old.

All of us also have Hadza admixture in K10+French. I have 0.63%, my father has 0.93%, and my grandmother's brother has 1.30%. My grandmother's phased kit gets 1.78% Hadza, which could, in reality, be as low as half (i.e. 0.89%). It seems that my grandmother and her brother have more or less the same amount of Hadza admixture.

The Hadza or Hadzabe are an ethnic group in Tanzania, whose language has traditionally been grouped together with the Khoi-San languages. It seems, however, that the Hadza language is an isolate, i.e. a language unrelated to any other. The Hadza have traditionally lived as hunters and gatherers, just like the San in southern Africa, and considering the geography (Tanzania being a neighbour of Mozambique), our Hadza admixture might perhaps come from the same source as our Khoi-San admixture. Provided it's not all noise, of course.

My grandmother's brother also has 0.53% Mbuti-Pygmy. The Mbuti are an indigenous people living in the Ituri forest in the Democratic Republic of the Congo. I'm not sure if this admixture is connected to our East African/Malagasy side or to our West African side.

Additionally, my grandmother's phased kit shows a very small trace of Eastern Bantu admixture (0.21%). This admixture - if it is real - might well be connected to our Malagasy side, since the Malagasy people are a mix of Austronesians from Borneo and Bantu people from today's Mozambique. (It is also possible, but perhaps less likely, that the Bantu admixture might be related to our West African side, since the Bantu migrations began in West-Central Africa.)

We also get percentages of other African categories such as North African, Nilo-Saharan and Omotic. However, North African admixture in this calculator seems to be ubiquitous among Europeans, and the other categories are also geographically closer to Europe and therefore more likely to be noise or misreadings of European or West Asian admixture.

Strangely, none of us get West African in this particular calculator, but the West African connection might be expressed through one of the other Sub-Saharan** categories. It might also be hiding in the "North Africa" category, just like at FTDNA.

My grandmother's brother's results from the K10+French calculator


My grandmother's phased kit, showing quite a bit of African admixture

* The segment on chromosome 4 is picked up by four datasets: Eurogenes, HarappaWorld, Dodecad and puntDNAL.
** I know that some people find this term offensive, but I could think of no better term to express the "north/south of the Sahara" divide.

No comments:

Post a Comment