The Atlantic slave trade resulted in the forced migration of an estimated 11 million Africans to the Americas. Only 9 million are thought to have survived the passage, and many more died in the early years of captivity. Historical accounts indicate that virtually all enslaved Africans brought to North America came from either West or West Central Africa. A recent comparison of mtDNA sequences from 1148 African Americans living in the US with a database of African mtDNA sequences showed that more than 55% of the US lineages have a West African ancestor, while fewer than 41% came from West Central or South West Africa 1. In North America, different constellations of African groups were brought to various staging areas 2. Among the important staging areas for the arrival and distribution of enslaved Africans were the ports of Savannah, GA and Charleston, SC. Estimates of the origin of enslaved Africans received at these sites are presented in Figure 1, with the largest African regional contributions coming from West Central Africa (40%; contemporary Angola, the Congos, Equatorial Guinea, and Gabon), and the West African regions of Senegambia (23%; contemporary Senegal, Gambia, and northern Guinea), and Upper Guinea (18%; contemporary Guinea and Sierra Leone and northwestern Liberia). Africans in the Carolina coast region were intentionally mixed to reduce the possibilities for successful revolts and to facilitate their assimilation into plantation-slave society. The contemporary Gullah/Geechee culture emerged from these Africans.

Because mitochondrial DNA (mtDNA) is passed from mother to daughter with few, if any, changes occurring over many generations, it is possible to compare contemporary African-American mtDNA haplotypes with contemporary mtDNA haplotypes in a worldwide database to obtain information about the ancestral origins of these mtDNAs. In such a comparison, continent-specific haplotypes are readily observed, and the assignment of mtDNAs to continent of origin is relatively straightforward. The more difficult task is to tie particular mtDNA haplotypes to specific geographical regions and ethnic groups within a continent. This task is particularly difficult for Africa, as there is more genetic diversity among Africans than among people from any other continent and because humanity has resided in Africa longer than anywhere else.

Comparisons of individual mtDNA haplotypes could be used to identify a geographical region, particular country, or even an ethnic group from which a maternal ancestor emanated. However, the geographic distribution of mtDNA haplotypes is greatly influenced by the migration of individuals or population groups. These movements often result in the assimilation of people from other ethnic groups. Intermarriage also causes mtDNA haplotypes to move from one ethnic group to another. Over time, mtDNA haplotypes that originated in a single ethnic group are distributed among many ethnic groups. Despite these complications, mtDNA analyses for the purposes of ancestry reconstruction are increasing in popularity. Many people have had their mtDNA tested with the hope that the test will match their DNA to an mtDNA haplotype found in a particular ethnic group. For African Americans, who have been disenfranchised from their specific African roots, such a test might provide a clue about the ethnic group or country in Africa where one of their maternal ancestors originated. However, if identical mtDNA haplotypes are shared among many ethnic groups from different parts of Africa, it would be impossible to use DNA sequence information to determine which single ethnic group was the source of a particular maternal ancestor. To date, there are no published assessments that provide quantitative information about how often African-American mtDNAs are exact matches to multiple African ethnic groups. Therefore, we decided to compare samples of Carolina coast and other African-American mtDNAs to a database of sub-Saharan African mtDNAs to generate such an assessment.

Because only a small fraction of the sub-Saharan African ethnic groups have been sampled, and there are parts of sub-Saharan Africa that are poorly represented in our database (Figure 2), our database cannot be considered a representative subset of the sub-Saharan mtDNA gene pool. Nevertheless, it is clear that a much larger database is needed since 40% of the African-American samples analyzed have no exact match in our database. The extensive sharing of mtDNA haplotypes among ethnic groups from different regions of Africa is consistent with the historical evidence of extensive migration and mixing of African ethnic groups. Indeed, the well-documented Bantu migrations appear to have had a major impact 4, as have the formation of the historic empires and kingdoms of the region (such as the historic empires of Ghana, Mali, and the Songhai, Bakongo, and Ashanti Kingdoms). Despite the limitations of our database of sub-Saharan mtDNA sequences, it is likely that we have identified the most common haplotypes found in this region. Some are found throughout the region that includes the Bantu migrations, and others are found primarily in either the western or the eastern parts of the continent. We intend to continue to increase the size of our database, because a significantly larger database would provide more information about haplotypes that are present at lower frequencies than the most common haplotypes. Some of these lower-frequency haplotypes are likely to be shared among widely distributed ethnic groups, while others may have a more localized distribution.

Another way to assess our sub-Saharan mtDNA database would be to see how well African-American mtDNAs match database sequences. Historical accounts of the trans-Atlantic slave trade indicate that most North American slaves came from the western coast of Africa, including the geographical regions from present-day Angola to Senegal. When African-American mitochondrial DNA HVS-I sequences were studied, nearly half were identical to those from two or more African ethnic groups in our expanded database. Furthermore, the average number of perfect matches per matching African-American mtDNA increased from 3.6 different ethnic groups to 6.1 different ethnic groups when the size of the database was increased by 53% to its present size of 3725 sequences. These results reflect the fact that approximately half the mtDNA sequences in our sub-Saharan database are shared by members of three or more ethnic groups.

In both of the African-American samples, approximately 40% of the mtDNA sequences did not match any sequence in any other ethnic group (Table 3). However, more than half of these sequences differed from multiple database sequences at a single position (Table 4). Because it is unlikely that more than a few of these differences result from new mutations that occurred in North America or that more than a few lineages went extinct in Africa after being introduced to the new world, this result suggests that only a small fraction of the mtDNA diversity present in sub-Saharan Africa has been sampled, and that much of the unsampled diversity is due to single mutations that have occurred in the common haplotypes.

Many African Americans are interested in learning more about their African roots and are willing to pay to have their mtDNA analyzed in the hope that it will match DNA from a particular African ethnic group. However, as more than half of the mtDNA sequences in the African database are identical to sequences from other ethnic groups, African-American mtDNAs will be much more likely to match sequences from multiple ethnic groups than sequences from a single ethnic group. When this result is coupled with the fact that 40% of African-American mtDNAs did not match any sequence in the database, it is clear that matches to a single African ethnic group will not be the outcome for most African Americans, and even when a match to a single ethnic group is obtained, multiple matches may occur in a larger database. Furthermore, for the typical African American, the maternal ancestor who was the source of the mtDNA was just one of hundreds of enslaved African ancestors. In fact, it likely that there has been more mixing of African ethnic groups in the Americas than has ever occurred elsewhere. Thus, the ancestors of virtually all contemporary African Americans came from a large number of ethnic groups located throughout the region from Senegal to Angola.

Half of the sub-Saharan mtDNA sequences in our database are common haplotypes that are shared among ethnic groups from multiple regions of sub-Saharan Africa. The finding that fewer than 10% of African-American mtDNAs matched mtDNA sequences from a single African region suggests that as few as one in nine African Americans may be able to trace their mtDNA lineage to a particular region of Africa. However, no firm conclusions should be made until a much larger database is available. It is clear, however, that nearly half of contemporary African-American mtDNAs are identical to African haplotypes that are found in multiple ethnic groups throughout sub-Saharan Africa. For these mtDNAs, it is impossible to use only mtDNA sequence information to determine which single ethnic group was the source of the maternal ancestor.

BE, JLW, BAJ participated in the assembly of the database. The database comparisons were performed by BE. All authors contributed to the interpretation of the data and the writing of the manuscript.