Deciphering early human history using Approximate Bayesian Computation and 74 whole genomes from Central and Southern Africa

Human evolutionary history in Africa before and after the out-of-Africa event remains largely unexplored, due to lack of genome sequence data, limited representation of populations and limitations of presently available inference methods. We generated high-coverage genomes from 49 Central African individuals, from five rainforest hunter-gatherer populations and four neighboring populations, and from 25 Khoe-San individuals, from five populations. We analyzed these genomes jointly with 104 comparative genomes from worldwide populations. We showed that rainforest hunter-gatherers and Khoe-San populations define two distinct major axes of genetic variation both at the worldwide and Sub-Saharan scales. This new data provides unprecedented resolution to unravel complex genetic differentiation among rainforest hunter-gatherer populations in particular. Using both deterministic and Approximate Bayesian Computation inferences, we found strong support for gene flow throughout the entire history of Central and Southern Africa, and an early divergence, some 250-370 kya ago, of Khoe-San ancestors from the lineage ancestral to all Central African populations. This event was followed, still in the presence of gene-flow, some 80-240 kya, by the divergence of lineages ancestral to rainforest hunter-gatherers and their neighbors. Finally, divergence between the different Khoe-San populations likely predated that of eastern and western rainforest hunter-gatherers which occurred 16-44 kya. Altogether, our results indicate that a tree-like history of Central Africa incorporating gene-flow among ancient lineages as well as among recent lineages can explain genomic variation observed among populations today.