Tracing pastoralist migrations to southern Africa with lactase persistence alleles

Abstract

Although southern African Khoisan populations are often assumed to have remained largely isolated during prehistory, there is growing evidence for a migration of pastoralists from eastern Africa some 2,000 years ago, prior to the arrival of Bantu-speaking populations in southern Africa. Eastern Africa harbors distinctive lactase persistence (LP) alleles, and therefore LP alleles in southern African populations may be derived from this eastern African pastoralist migration. We sequenced the lactase enhancer region in 457 individuals from 18 Khoisan and seven Bantu-speaking groups from Botswana, Namibia, and Zambia and additionally genotyped four short tandem repeat (STR) loci that flank the lactase enhancer region. We found nine single-nucleotide polymorphisms, of which the most frequent is -14010(∗)C, which was previously found to be associated with LP in Kenya and Tanzania and to exhibit a strong signal of positive selection. This allele occurs in significantly higher frequency in pastoralist groups and in Khoe-speaking groups in our study, supporting the hypothesis of a migration of eastern African pastoralists that was primarily associated with Khoe speakers. Moreover, we find a signal of ongoing positive selection in all three pastoralist groups in our study, as well as (surprisingly) in two foraging groups.

Figure 1

Map of the Approximate Locations of the Groups Included in This Study

Figure 2

Frequency Distribution of the −14010*C Allele and Networks of Associated Haplotypes (A) Surfer map of the −14010*C allele frequency. Circles indicate sample locations. (B) Network of STR haplotypes associated with the −14010*C allele, colored according to language family. (C) Network of STR haplotypes associated with the −14010*C allele, colored according to subsistence. Circles denote haplotypes. The size of the circle indicates the number of times that haplotype was observed, and each branch connecting two haplotypes consists of a single step-mutation at one STR locus.

Figure 3

Estimated Selection Coefficient and Age of The−14010*C Allele in Southern African Populations Results are shown for a population growth rate of r = 0.01; see also Figure S2 for comparable results assuming no population growth. Top: a graph of the composite log likelihood [log(L)] for the four STR loci on the y axis versus the selection coefficient (s) on the x axis. Bottom: a graph of the age of the −14010*C allele (in generations) on the y axis versus the selection coefficient (s) on the x axis. The solid line shows the expected age given s, and the dashed lines show the expectation ± 2 SDs of the posterior distribution of age, given s.