Early split between African and European populations of Drosophila melanogaster

Natural populations of the fruit fly Drosophila melanogaster have been used extensively as a model system to investigate the effect of neutral and selective processes on genetic variation. The species expanded outside its Afrotropical ancestral range during the last glacial period and numerous studies have focused on identifying molecular adaptations associated with the colonization of northern habitats. The sequencing of many genomes from African and non-African natural populations has facilitated the analysis of the interplay between adaptive and demographic processes. However, most of the non-African sequenced material has been sampled from American and Australian populations that have been introduced within the last hundred years following recent human dispersal and are also affected by recent genetic admixture with African populations. Northern European populations, at the contrary, are expected to be older and less affected by complex admixture patterns and are therefore more appropriate to investigate neutral and adaptive processes. Here we present a new dataset consisting of 14 fully sequenced haploid genomes sampled from a natural population in Umeå, Sweden. We co-analyzed this new data with an African population to compare the likelihood of several competing demographic scenarios for European and African populations. We show that allowing for gene flow between populations in neutral demographic models leads to a significantly better fit to the data and strongly affects estimates of the divergence time and of the size of the bottleneck in the European population. Our results indicate that the time of divergence between cosmopolitan and ancestral populations is 30,000 years older than reported by previous studies.