Bayesian coalescent inference of major human mitochondrial DNA haplogroup expansions in Africa

Abstract

Past population size can be estimated from modern genetic diversity using coalescent theory. Estimates of ancestral human population dynamics in sub-Saharan Africa can tell us about the timing and nature of our first steps towards colonizing the globe. Here, we combine Bayesian coalescent inference with a dataset of 224 complete human mitochondrial DNA (mtDNA) sequences to estimate effective population size through time for each of the four major African mtDNA haplogroups (L0-L3). We find evidence of three distinct demographic histories underlying the four haplogroups. Haplogroups L0 and L1 both show slow, steady exponential growth from 156 to 213kyr ago. By contrast, haplogroups L2 and L3 show evidence of substantial growth beginning 12-20 and 61-86kyr ago, respectively. These later expansions may be associated with contemporaneous environmental and/or cultural changes. The timing of the L3 expansion--8-12kyr prior to the emergence of the first non-African mtDNA lineages--together with high L3 diversity in eastern Africa, strongly supports the proposal that the human exodus from Africa and subsequent colonization of the globe was prefaced by a major expansion within Africa, perhaps driven by some form of cultural innovation.

Figure 1

Genealogy of major African mtDNA haplogroups. This phylogeny shows the genealogical relationships between the L0, L1, L2 and L3 mtDNA lineages of Africa and the position of the two major non-African lineages, M and N.

Figure 2

BSPs of effective population size through time for (a) sub-Saharan Africa (n=224), (b) haplogroup L0 (n=60), (c) haplogroup L1 (n=41), (d) haplogroup L2 (n=43) and (e) haplogroup L3 (n=80). The bold black line represents the median posterior effective population size through time. The grey lines delimit the 95% highest posterior density for effective population size, accounting for uncertainty in the reconstructed phylogeny and substitution model parameters. The dashed line in (a) plots the sum of the separate lineage population estimates through time. Effective population size is plotted on a log scale and assumes a generation time of 20 years. These estimates of effective population size have an inverse relationship with the evolutionary rate of mtDNA used for the calibration, such that they will be lower for faster rates and higher for slower rates. For comparison, all the x-axes have a scale extending to 150 kyr. The plots are truncated to the median estimate of each region's TMRCA.