Genetic variation reveals large-scale population expansion and migration during the expansion of Bantu-speaking peoples

Abstract

The majority of sub-Saharan Africans today speak a number of closely related languages collectively referred to as 'Bantu' languages. The current distribution of Bantu-speaking populations has been found to largely be a consequence of the movement of people rather than a diffusion of language alone. Linguistic and single marker genetic studies have generated various hypotheses regarding the timing and the routes of the Bantu expansion, but these hypotheses have not been thoroughly investigated. In this study, we re-analysed microsatellite markers typed for large number of African populations that-owing to their fast mutation rates-capture signatures of recent population history. We confirm the spread of west African people across most of sub-Saharan Africa and estimated the expansion of Bantu-speaking groups, using a Bayesian approach, to around 5600 years ago. We tested four different divergence models for Bantu-speaking populations with a distribution comprising three geographical regions in Africa. We found that the most likely model for the movement of the eastern branch of Bantu-speakers involves migration of Bantu-speaking groups to the east followed by migration to the south. This model, however, is only marginally more likely than other models, which might indicate direct movement from the west and/or significant gene flow with the western Branch of Bantu-speakers. Our study use multi-loci genetic data to explicitly investigate the timing and mode of the Bantu expansion and it demonstrates that west African groups rapidly expanded both in numbers and over a large geographical area, affirming the fact that the Bantu expansion was one of the most dramatic demographic events in human history.

Keywords: Africa; Bantu-speakers; approximate Bayesian computation; migration; population expansion.

Figure 1.

Map of sub-Saharan Africa illustrating (a) the different Bantu-language sub-groups according to the Guthrie classification [15], (b) the route of the Bantu expansions according to the ‘early-split’ linguistic model (redrawn from Pakendorf et al. [4]), and (c) according to the ‘late-split’ linguistic model (redrawn from Pakendorf et al. [4]). (d–g) The different models of the Bantu expansion tested in this study using an ABC approach; (d) the ESW model which posits a primary expansion towards the east (1) and a later expansion to the south (2), (e) the SEW model which posits a primary expansion to the south (1) and a later expansion to the east, (f) the WES model which posits a primary expansion to the east (1) and the southern expansion (2) originated from the populations that migrated to the east, and (g) the STAR model which posits a simulations expansion to the east and the south from the west.

Figure 2.

Population topology of four investigated models: (a) the ESW model where the population topology is (east, (south, west)), (b) the SEW model where the population topology is (south, (east, west)), (c) the WES model where the population topology is (west, (east, south)), and (d) the STAR model where all three groups have a common split time.

Figure 3.

Distribution of the west African genetic component across the African continent: (a) supervised Structure analysis to show the distribution of the west African component (fixed green cluster), in the rest of Africa. Two other fixed clusters are European (yellow) and Middle Eastern/South Asian (brown) to account for non-African admixture into African groups. In total, 10 clusters were assumed (seven free assignments allowed). Increasing the number of clusters, K, from 4 (one free assignment allowed) to 10 (seven free assignments allowed) are shown in the electronic supplementary material, figure S2. Populations in coloured text were used when testing the expansion model using ABC approaches; populations in blue text are Bantu-speakers that were included in the ‘BS’ group during ABC analysis; while populations in green text are Niger–Kordofanian speakers that were included in the ‘NK’ group together with the ‘BS’ populations. Stars indicate populations from east and southern Africa that were used in the ABC analysis which tested different divergence models. (b) Heat map of the west African genetic component on the African continent at K = 10 (electronic supplementary material, figure S3 contains additional heat maps of the west African component with increasing number of clusters allowed in the supervised Structure analysis).

Figure 4.

The posterior distribution of (a) the past population size N_p and (b) expansion time T_EXP and for the Bantu-speaking group (red) and the Niger–Kordofanian-speaking group (blue) group.

Figure 5.

The number of accepted simulated replicates as a function of a fixed tolerance value for all four models.