Application of ABC to Infer the Genetic History of Pygmy Hunter-Gatherer Populations from Western Central Africa

International audience

英语. In evolutionary biology, approximate Bayesian computation (ABC) methods are well adapted to the complex models of species and population history in which serial or independent divergence events, change of population sizes, and genetic admixture, or migration events are often suspected. Here, we present the results of using a set of recent ABC-based methods to analyse a microsatellite genetic dataset composed of Pygmy populations from Western Central Africa and non-Pygmy human populations, partly stemming from the dataset originally investigated in Verdu et al. (2009). Although archaeological remains attest to Homo sapiens’ presence in the Congo Basin for at least 40,000 years, the demographic history of these groups, including divergence and admixture, remains little known. In this chapter, we will successively describe (1) the observed microsatellite dataset for which one wants to conduct a Bayesian analysis, the set(s) of models we compared, with their parameters and associated priors, and the way datasets were simulated for ABC analyses; (2) the model choice analyses we carried out using a recently developed method, ABC random forest (Pudlo et al., 2015); (3) the estimates we obtained for Pygmy historical and demographic parameters under the most likely model; and (4) model-posterior checking to evaluate the goodness of fit between the final inferred genetic history and the observed dataset. We found a probable recent (about 4,900 YBP) common origin of all Western Central African Pygmy populations, despite the vast cultural, morphological, and genetic diversity observed today among these populations. We also confirmed recent asymmetrical and heterogeneous genetic introgressions from non-Pygmies into each Pygmy population. These results are consistent with previous population genetics studies on Pygmies from Western Central Africa by Verdu et al. (2009), Batini et al. (2011), and Patin et al. (2009).