Inferring complex demographies from PSMC coalescent rate estimates: African substructure and the Out-of-Africa event. S. Gopalakrishnan1, P. Grabowski2, M. C. Turchin1, B. Henn7, J. Kidd5, G. Perry3, A. Gebremedhin6, C. Beall4, C. D. Bustamante7, A. Di Rienzo1, Y. Gilad1, A. A. Palmer1, J. K. Pritchard1 1) Human Genetics, University of Chicago, Chicago, IL; 2) Ecology & Evolution, University of Chicago, Chicago, IL; 3) Anthropology, Penn State University, University Park, PA; 4) Anthropology, Case Western Reserve University, Cleveland, OH; 5) Human Genetics, University of Michigan, Ann Arbor, MI; 6) Medicine, Addis Ababa University, Addis Ababa, Ethiopia; 7) Genetics, Stanford University, Stanford, CA.

   Human population history is an intriguing and complex story with many events like population growth, bottlenecks, time-dependent/non-homogeneous migration, population splits and mixtures. Estimating complete demographies with population sizes, rates of gene flow and population split times has proven to be a challenging endeavor. We propose a framework for jointly estimating the demography parameters, especially gene-flow rates and split times, for a large number of populations. We use coalescent rate estimates obtained from Pairwise Sequentially Markovian Coalescent (PSMC) as the starting point for our analysis. Since PSMC works on only two chromosomes at a time, we apply PSMC to all pairs of individuals to obtain the pairwise coalescent rates for lineages from every pair of sampled populations. Using a mathematical model for calculating coalescent probabilites given population parameters, we estimate demography using the parameters that best fit the observed coalesecent rates.
   In this study, we focus on two aspects of African population genetics, 1. the nature of population structure in Africa going back in time and 2. the timing of the Out-of-Africa event. To address these questions, we assembled a dataset with whole genome sequences from 162 individuals using both in-house sequencing and publicly available sources. These samples span 22 populations worldwide. These include eleven African populations which we use to dissect the population substructure in Africa. In addition, we also have 2 Middle Eastern, 5 European and 4 East/Central Asian populations which inform the population split time estimates for the Out-of-Africa event and the European-Asian split.
   We find extensive population structure in Africa extending back to before the Out-of-Africa event. The Ethiopian populations, Amhara and Oromo, show evidence of mixing beyond 15 kya. The Maasai and Luhye merge with the Ethiopian populations to form a panmictic East African population ~40kya. We find evidence for extensive mixing between east and west African populations before 50kya. Among the pygmy populations, we see recent gene flow between the Batwa and Mbuti. All African populations except the San merge into a single population around 110 kya. The San exchange migrants with the other African populations beginning ~120 kya. We estimate the Out-of-Africa event to have occurred ~75kya and the European-Asian split to ~25kya.

You may contact the first author (during and after the meeting) at