Strong maternal Khoisan contribution to the South African coloured population: a case of gender-biased admixture

Abstract

The study of recently admixed populations provides unique tools for understanding recent population dynamics, socio-cultural factors associated with the founding of emerging populations, and the genetic basis of disease by means of admixture mapping. Historical records and recent autosomal data indicate that the South African Coloured population forms a unique highly admixed population, resulting from the encounter of different peoples from Africa, Europe, and Asia. However, little is known about the mode by which this admixed population was recently founded. Here we show, through detailed phylogeographic analyses of mitochondrial DNA and Y-chromosome variation in a large sample of South African Coloured individuals, that this population derives from at least five different parental populations (Khoisan, Bantus, Europeans, Indians, and Southeast Asians), who have differently contributed to the foundation of the South African Coloured. In addition, our analyses reveal extraordinarily unbalanced gender-specific contributions of the various population genetic components, the most striking being the massive maternal contribution of Khoisan peoples (more than 60%) and the almost negligible maternal contribution of Europeans with respect to their paternal counterparts. The overall picture of gender-biased admixture depicted in this study indicates that the modern South African Coloured population results mainly from the early encounter of European and African males with autochthonous Khoisan females of the Cape of Good Hope around 350 years ago.

Figure 1

Diverse Continental Origins of Maternal and Paternal Contributions to the South African Coloured Bar sizes reflect the combined frequencies of all haplogroups, indicated separately for mtDNA and NRY, associated with the sub-Saharan African, west Eurasian/European, and south/southeast Asian genetic components (Table S1, Tables 1 and 2).

Figure 2

MDS Plots Illustrating the Relationships between the South African Coloured Population and the Five Putative Parental Populations (A) MDS plot based on mtDNA Hg frequencies. (B) MDS plot based on NRY Hg frequencies. The five metapopulations used as putative parental populations include Europeans, Indians, Khoisan, Bantu speakers, and southeast Asians. The European dataset included British, French, Dutch, German, and Austrian populations; the Indian dataset included populations from the Bengal Bay coast (Sri Lanka, Bangladesh, and eastern states of India); the Khoisan dataset included all available Khoisan data; the Bantu dataset included Bantu-speaking populations from South Africa and central Africa; and the southeast Asian dataset included populations from the Malay Peninsula, Sumatra, Java, and neighboring smaller isles. Because different individuals and populations have been analyzed for mtDNA and NRY, sample sizes from these metapopulations differ between the two analyses. Sample sizes of European, Indian, Khoisan, Bantu, and southeast Asian populations were, respectively: n = 6092, n = 1246, n = 298, n = 459, and n = 616 for mtDNA and n = 353, n = 766, n = 122, n = 499, and n = 66 for NRY. The lower population sample sizes for the NRY data result from the nonoverlapping nature of the markers used in the different studies. These comparative data were obtained from the MURKA database, compiled from published sources by O. Balanovsky, V. Zaporozhchenko, R. Sychev, A. Pshenichnov, and E. Balanovska.

Figure 3

Diverse Continental and within-Continental Admixture Proportions in the South African Coloured Population Mean Khoisan, Bantu-speaking, European, Indian, and southeast Asian admixture proportions among the SAC, on the basis of the mY estimator, are represented as bars for the mtDNA and NRY data. Error bars indicate standard deviations. The five parental metapopulations used for this analysis, as well as their sample sizes, were the same as those used for the MDS analysis (Figure 2). Time of the initial admixture event was set to 400 years, and standard deviations were calculated on the basis of 10,000 bootstraps.