23andMe Chromosome Segment Sharing

Since hearing about 23andMe finally bringing back the chromosome browsing feature that allows one to see one's matches with DNA Relatives on the site, we checked it out for our account (and that of our Haitian parent). For us, unfortunately, 23andMe still maxes out one's DNA matches at 5000. This means that the overwhelming majority of matches I have on that site are those sharing my "Hispanic" roots. For African matches, our Haitian parent has more luck. We were able to browse a number of them and see the shared segments with people from Nigeria, Congo (via South Africa), Sierra Leone, Ghana, Liberia, and other places. 

Interestingly, we saw that our Haitian parent shares a segment of DNA on their X chromosome that overlaps with that of a Fulani person (judging from surname) who is likely from Sierra Leone and someone else from Sierra Leone who is possibly Temne (inferring from surname). And through our Haitian parent's account, we saw that we share the Temne person as a match (even though 23andMe doesn't actually list this person in my DNA Relatives). In any case, the shared segments were always small (11.37 cM for the Temne, 11.58 cM for the Fulani). So, through our Haitian parent's side, we appear to have a female ancestor who was possibly from a Temne-like (or adjacent group) background who was trafficked to Saint Domingue. We assume the Fulani shares this Temne-like ancestry with us. But to make things more surprising, our parent did share a segment of DNA with another Fulani on the second chromosome that was assigned as "Senegambian and Guinean."

In terms of the paper trail, the colonial parish registries, runaway slave ads and notarized contracts or plantation inventories list a number of "nations" from what is now Sierra Leone and Liberia. Some examples include the Timbou, Mamou, Susu, and Mende. Cangas, Mesurades, Miserables, Bouriquis, and more from the Liberian and Ivory Coast regions were also present in the colony, according to Moreau de Saint-Mery. 

We will likely keep exploring this new shared segment viewing feature to explore our Haitian parent's matches with Liberians, Nigerians, and Ghanaians. It is just unfortunate that the Historical Matches feature doesn't allow for this. Nor can one see specific shared segments used by 23andMe to assign the African Genetic Group matches that appear in the Ancestry Composition breakdown. Adding these features would make the site more useful and important for genealogical research. 

Qui sont les vrais ancêtres des Haïtiens ?

We found Haiti Inter's video, "Qui sont les vrais ancêtres des Haïtiens?" to be an intriguing video exploring the African ancestry of the Haitian people. Nothing new, but still interesting. 

23andme Sub-Saharan Africa Update

 

Like many others, I was eagerly awaiting the update for sub-Saharan Africa on 23andme. We manage the account of a close relative from Haiti and were very interested in seeing any changes. Overall, the changes were not a surprise for Haitians and match what little I have uncovered with historical/genealogical sources. For example, the only "Very Close" African genetic group was the Igbo. This matches our findings of a consistent "Ibo" presence among African slaves in 18th century Bainet. 

The distant Yoruba people genetic group was also not surprising. Given the well-known historical interactions between peoples of Benin and southwestern Nigeria as well as the Yoruba presence in both regions, we suspect many Haitians may have this genetic group. Indeed, it is also matched by the high number of Yoruba matches for this relative on Ancestry.

The Ewe, Ga, and Ga-Dangme genetic groups are also to be expected as distant. One of the mutual African DNA matches between this person and myself on Ancestry is Ga, suggestive of an identical by descent shared ancestry. The Ga also overlap with Togo and Benin peoples, a known source of the African captives brought to Saint Domingue. As for the Mende, Temne and Kru, our relative has DNA matches from Sierra Leone on Ancestry and 23andme. From the former site, she and I share a match with someone who is half-Sierra Leonean (and half-Nigerian). This relative also has 2 Mende matches on Ancestry. 

Unfortunately, they did not get any African genetic groups for Angolan/Congolese or Senegambian and Guinean. This was disappointing as it would have been interesting to see African genetic groups for each region. Nonetheless, this person has 2 Fulani matches on 23andme and may see genetic groups in the next update.

Gold Coast and "Mina" Captives in Saint Domingue

We have been searching through runaway ads posted in Saint Domingue's newspaper, Affiches américaines. A number of runaways were of the "Arada" or "Mina" nations, often vague terms referring to the Bight of Benin and Gold Coast as places where the captives were purchased and shipped across the Atlantic. We decided to share some of the ads for interesting cases of people from this broad region, particularly those with names, origins, or skills that shed light on the Gold Coast/Slave Coast in the 18th century. Like the above example, a Mina named Carabi. 

Another interesting case is a maroon named Codiot, perhaps the name Kwadwo used by Akan peoples. 

Another Akan name that appears among Africans of the "Mina" nation is Cofy, likely the Akan Koffi. 

Akan names like Coichy, probably Kwasi, can similarly be found among "Mina" Africans in colonial Haiti.

Apia, possibly the Akan name Appiah, was another Akan-derived name found among "Mina" blacks. Or he was perhaps named after the Apia of Greece...

An "Aguia" black named Couaminan might be a Kwamina, another Akan-derived name. 

Another Koffi, or Coffi, was a maroon of the Cramanti-Adia background. Cramanti was the Coromantee of the British West Indies, and Adia may be a reference to the Aja, or Adja, peoples.

An "Arada" named Kwaku from 1785. If "Arada" captives could have Akan-derived names, one wonders if this one was of Ewe origin.  

Sometimes, the "Mina" were listed as Amine. Here we have another Coffi, who apparently spoke the language of Curacao, Papiamento.

Another "Mine" was specified as a "Quimbou-Mine" in the colony. There is a street in Accra called Kinbu, perhaps "Quimbou" being a reference to an area of what is now Accra?

A Bandia African who was only in Saint Domingue for three months, was said to speak the "Mine" language. 

As if things weren't complicated enough, the Mallais, Malez, or Male nation were present in the colony. Since the word appears to be derived from the word for Muslims used in Allada and other parts of today's Benin, they were presumably Muslims or captives supplied by Muslim traders from the North. Since the French slave traders sometimes used the word Gold Coast for parts of the Slave Coast, it is probable that the latter coast was intended here. 

Yet another ambiguous case is Bayard, of the Otam nation. Since the "Gold Coast" in this context could have included parts of the Slave Coast, Otam might have nothing to do with Ghana. Indeed, the Otam name was associated with a people in the Calabar region. Yet, Oldendorp wrote of a "nation" of people called Attem who were neighbors of the Kassenti, or Tjemba.

Last, the case of Marie. Possibly Ibo or Cramanti, she was said to speak the two languages. If she was actually "Ibo" in origin, then one wonders if she learned the "Cramanti" language in West Africa or in Saint Domingue.

Mondongues of Saint-Domingue

Whilst reading John Thornton's history of West Central Africa, we have been revisiting the runaway advertisements in colonial Haiti for any and every allusion to Congos, Mondongues and other likely Central African captives. We have always believed the Mondongue to be a particular people or "nation" and not a category entirely concocted by slave traders and slaveholders in Saint-Domigue (and elsewhere in the Caribbean and Latin America). Since Oldendorp met "Mondongues" who told them they lived far away from the coast, spoke related languages, and were of 3 different groups, we hoped to find evidence of that in Saint Domingue's "Mondongue" Africans. Sadly, the runaway ads usually not too detailed about the African origins of captives in the colony. Sometimes, they weren't even sure who was a "Congo" or "Mondongue" among runaways.

Some of the runaway ads for Mondongues in the colony do refer to them by their native names. Thus, one, named Moyala, may be an instance of a Mondongue name. Comparing this name with personal names in Central Africa might reveal clues about the ethnic origins of the Mondongues.

Another Mondongue was named Petiton in his country. This name doesn't sound particularly Central African at all, although Goulau might be.

Gouala, or Favori, another Mondongue runaway, bears an African-sounding name. 

More ambiguity on the differences between Mondongues and Congos can be seen in the case of Azor. If he only spoke Congo, does this mean the Mondongue were speakers of Kikongo or a closely related language? 

Another Mondongue with an African name, Ougoulou dit Charlemagne, was "owned" by someone in Bainet. The name sounds African, but to what extent it could be matched with a particular area or language of Central Africa remains unknown. 

Updated 23andme African DNA Matches

With all of the problems at 23andme, we wanted to do a more thorough search of our Haitian parent's African matches as soon as possible. Thus, the table able represents our attempt to tally the number of matches she has based on a search from two days ago. We suspect more matches ("DNA Relatives") will disappear since they are deleting their data from the site. However, we were able to find 16 African matches. Fortunately, some even included basic data about the birthplace of their grandparents, which helped us to clarify their possible ethnic origins. Unsurprisingly, they are overwhelmingly West Africans, with obvious Nigerians more than half of the total. 

Immediately, one is struck by the number of matches to southeastern Nigerian groups (Igbo and possibly Ibibio) and the paucity of Yoruba. Unlike her Ancestry DNA matches, our Haitian parent only has 1 unambiguous Yoruba match here (someone whose name might indicate an Islamic background, too). Despite sharing more DNA with the lone Yoruba than most of the others, it is quite interesting to see so few of this ethnic group on 23andme among her matches. We are not sure if that is due to the lack of a Benin & Togo category in which Yoruba groups overlap more frequently with Haitians through shared ancestry in the Bight of Benin? Either way, 23andme is picking up on closer genetic ties to the Igbo and groups in southeastern Nigeria, including what appears to be Ibibio or Efik (our parent's largest matches, at 0.54% shared DNA). This ancestral tie to southeastern Nigeria is also suggested by her 7 matches to Igbos on Ancestry as well as her shared 29 cM of DNA with an Igbo on that site. Indeed, the only ethnolinguistic group specified in her 23andme Ancestry Composition report was also the Igbo. Seeing a closer match to the Ibibio or Efik, however, suggests ties to other groups in southeastern Nigeria, too. 

Besides the obvious Igbo and southeastern Nigeria-related ancestry implied here, one can also see Central African represented by two individuals with roots in Congo. According to my limited understanding of Congolese surnames, these two individuals may be of Luba or Lunda origins. So far, we have yet to uncover any particularly close matches to Central Africans. The more obvious suspects for Congolese and Central African ancestry in Haiti would be groups from or near the Kikongo-speaking peoples, but they are likely not represented well among customers of 23andme and Ancestry. A quick perusal of the varieties of Central African captives imported into colonial Saint Domingue, however, would likely include groups from inner regions of Central Africa, too.  

Moving on to Ghana, one finds here yet another Ga match! Our parent had 2 Ga DNA matches on Ancestry, although both of those had elevated levels of "Benin & Togo" estimates. Here, the Ga match is someone with all 4 grandparents from the Accra. Their "Nigeria" score was only 17.4%, but that may be the algorithm's attempts to capture the ancestry of people from eastern Ghana to Benin. The Ga-Adangbe were likely represented among the captives in Saint Domingue, too. Perhaps they were under the label of "Mina" ("Mine" in some of the colonial Haitian documentation) or mixed with other groups from the Bight of Benin and Gold Coast. One of our few shared DNA matches was actually with another Ga person on Ancestry, making us think that a meaningful link with this area is possible. 

As for her two Fulani matches, these are almost certainly due to shared, non-Fulani ancestry in Upper Guinea .The two Fulanis, who both possess typical Fulani surnames, also have elevated Senegambian & Guinean ancestry (as well as minor North African) that clearly establish their origin. We suspect they are from Mali, Guinea, Sierra Leone or, in one case, possibly from Nigeria (an assumption based on their 7.6% Nigerian estimate). Since our parent had one DNA match with a Mandinka from Sierra Leone on Ancestry, we would not be surprised if the matches with the two Fulanis might stem from a Mandinka-related shared ancestry (or, perhaps, Senegambian or Malian, which could be inferred from the high numbers of "Bambara" and "Mandingues" in colonial Haiti). Of course we are merely speculating here, but we suspect that shared ancestry with the Fulani via actual Fulani ancestry would have led to more matches from this group. After all, they seem to be more heavily represented among West Africans using DNA testing services like Ancestry or 23andme. 

Last, but certainly not least, the two Liberian matches. One, based on surnames, is at least partly Vai in origin with some grandparents from the Cape Mesurado area. Our parent shares the most DNA with this Liberian (besides Igbo or southeastern Nigerians) who appears to be of Vai ancestry. In Saint Domingue, groups from parts of Liberia were known as Miserables, Canga and other terms. We would love to find more matches from Liberia and Sierra Leone to fully explore the possible ancestral ties to this area.

Overall, the 23andme matches are consistent with the largely Lower Guinea ancestry of our parent. Unfortunately, the ancestral ties to the Bight of Benin are not so easily seen here in her matches, but a link to the slave trade from the Bight of Biafra is very clear. As to be expected, the lower numbers of African customers from Benin, Central Africa, and Upper Guinea limits the data and analysis, but it is largely consistent with her 23andme Ancestry Composition. It also, by and large, is in accord with the general pattern of her Ancestry matches.

Estimating the Nupe, Hausa, Bariba in Saint Domingue

Another activity we have engaged in is an attempt at tracking runaway slaves and maroons identified in the colonial newspaper, Affiches américaines, already collected in the excellent Le marronage dans le monde atlantique project. We focused on the Nupe, Hausa and Bariba populations as groups who, living to the north of the Oyo state, were probably the bulk of the captives from the "Central Sudan" who were sold into slavery in Saint-Domingue. Unfortunately, we could not include the Fulani or Poulard in our calculations, since it is impossible to determine how many of the Fulani in Saint-Domingue reached the Caribbean from the Bight of Benin. However, it is likely that at least some of the Fulani or "Poulards" in Saint Domingue arrived from the north of Oyo, too. 

In addition, we also included the "Malais" or "Malez" and "Bernon" captives who appear in the runaway notices. The latter group, hailing from Borno, to the northeast of Oyo, were "infinitely rare" in the colony of Saint Domingue. Nonetheless, a few did reach the hell of colonial Haiti, likely via the Bight of Benin. The former group, however, present more difficulties. According to early 18th century sources such as the journal of the Chevalier des Marchais, "Malais" traders had already been traveling to Ouidah by 1704. His sometimes vague understanding of their origins suggests they were Muslim traders from a distant land northeast of Allada, possibly Hausaland or Borno. Since they allegedly followed certain Islamic dietary restrictions, used the Arabic language, wore robes of blue or white, and favored long beards, we are guessing the "Malais" traders hailed from Hausaland or a nearby region. The captives they sold on the Slave Coast, however, were probably from neighboring groups and other nearby regions, which means the "Malais" captives in Saint Domingue were likely some northern group. Alternatively, the "Malais" could also have been Muslim Yoruba, but that seems less likely for the 18th century. 

Overall, we found about 134 unambiguous cases of slaves from areas north of Oyo who engaged in marronage in Saint-Domingue during the years 1766-1790. While the total population of Bariba, Hausa, Nupe, Borno, and "Malais" captives imported in the colony is unknown, we believe Geggus's data on slave ethnicities in the period 1727-1797 may be illustrative of general patterns. In his survey, which is not meant to be a complete total for the various nations, he calculated 532 Nupe, Bariba and Hausa in Saint-Domingue. This number, 532, is out of a total sample of 13,334 slaves, suggesting Nupe, Hausa and Bariba represented nearly 4% of the enslaved population. Naturally, we need further data, particularly in light of evidence for the early appearance of Bariba and Nupe captives in Saint Domingue in the 1710s and 1730s, and, perhaps, Hausa before the 1760s. But, if we use 4% as a very rough estimate for the number of captives from the lands north of Oyo, perhaps 4% of total slave population on the eve of the Haitian Revolution were from the "Central Sudan" and nearby regions north of Oyo. This would give astronomically high figures, but suggest around 20,000 captives were possibly from that aforementioned region, over half Hausa. 

Another possible approach, though still highly flawed, is to use rough estimates for total population of a "nation" by using the general rate of marronage. If, as Eddins indicates, 12,857 individual slave runaways appeared in the colonial newspaper, one could divide that by a rough estimate for the total slave population in 1789 or 1790. Alternatively, one could take that rough rate of marronage and apply it to the total period of colonial rule, dividing it by the estimated 800,000 captives brought to Saint-Domingue. Of course, one would have to take into consideration slave mortality rates and total imports for the period from 1766-1790, which we lack sufficient sources and data to do. Some slaves may have also appeared more frequently as maroons based on their occupation and other factors, too, which can skew results. Consequently, the estimates are likely too high or too low in some cases. Nonetheless, it is probably a more accurate estimate than our attempt to use the numbers from Geggus's survey covering the years 1727-1797. Using the marronage escape rate, calculated by dividing 12,857 by a rough estimate of a total of 500,000 slaves in 1790/91, then about 2.57% of all slaves could be runaways. If, say, the Hausa runaways, which we enumerated to be 83 (possibly 84) individuals from 1766-1790, is our base, then we can estimate a total Hausa population based on the rough estimate that 2.57% of Hausas imported were maroons. Of course, it could have been higher or lower depending on specific conditions that shape the chances of marronage, but if we assume that 83 Hausas were 2.57% of the total Hausa population imported into the colony, based on the rough figure of 500,000 slaves in 1790, then about 3230 Hausas were imported into the colony. If one uses the figure of 800,000 as a rough estimate for total slave imports in Saint Domingue, and assume the Hausa proportion of the maroons (12,857) was comparable or adequate for the entire colonial era, then we would use the maroon rate of 1.607% (calculated by dividing 12857 by 800,000). The Hausa proportion would then increase to 5165, of which it is very likely high mortality rates significantly reduced. Applying this to our total sample of "Central Sudan" captives, 134, would bring us to an estimate ranging from 5211 to 8338. Again, high mortality rates would have reduced the number of survivors. 

Overall, assuming these captives arrived on French slave ships, they would represent about 3.4% to 5.4% of African captives purchased from the Bight of Benin in Saint-Domingue (based on Geggus's figures in "The French Slave Trade: An Overview." Those figures are actually plausible. Moreover, these lower figures seem more likely based on the data. If, we were to assume 4% of the captives of Saint Domingue were Nupe, Hausa, and Bariba, based on Geggus's survey of slave nations and gender ratios from 1727-1797, then their totals could have ranged from perhaps as many as 19,949 to 31,918. That would significantly boost their proportion in the "Bight of Benin" captives in Saint-Domingue, from about 13% to nearly 21%, a figure that seems far too high.

Of course, high mortality rates and the lack of adequate early data makes all of this very speculative. Our method requires one to assume the proportion of, say, Hausa slaves, in the maroon population was proportional to the total slave population's share of Hausa. Yet slaveholders held certain stereotypes about which "nations" were best for specific occupations, and additional factors such as location, type of plantation and urban versus rural must have shaped actual rates of marronage (at least, those reported in the press). So, the aforementioned estimates are only useful as very broad ranges that might provide a general idea. But it is interesting to note, nonetheless, that the "Hausa" significantly outnumbered other groups in Geggus's survey data from the areas north of Oyo. The Taqua or Tapa (Nupe) came second at 161, while there were only 84 Bariba. The Hausa appear more frequently as runaways, too, representing nearly 62% of the reported maroons from this region. It is possible that their higher rate of marronage was both due to a more distorted sex ratio as well as their higher numbers. If this is the case, then the Hausa may not have significantly outnumbered the Nupe and Bariba as our data seems to suggest. 

Toby Green on the Trans-Atlantic Slave Trade

I first read Toby Green's "The Rise of the Transatlantic Slave Trade in Western Africa: 1300-1589" a while back, but I love how Green emphasizes African agency and history that shaped creolization, trade, and slavery in the broader Atlantic World context.  He emphases internal African diasporas that linked North Africa with West Africa and global markets, which paved the way for international trade with Europeans, the slave trade, and cultural mixing that facilitated the exchange of people, goods, and culture that shaped the Atlantic World. Listen to podcasts with Toby Green here and here.