The inner Niger delta has been suggested as the center of origin but molecular data to support this hypothesis is lacking. Here, we present a comprehensive analysis of the evolutionary and domestication history of African rice. By analyzing whole genome re-sequencing data from 282 individuals of domesticated African rice Oryza glaberrima and its progenitor O. barthii, we hypothesize a non-centric (i.e. multiregional) domestication origin for African rice. Our analyses showed genetic structure within O. glaberrima that has a geographical association. Furthermore, we have evidence that the previously hypothesized O. barthii progenitor populations in West Africa have evolutionary signatures similar to domesticated rice and carried causal domestication mutations, suggesting those progenitors were either mislabeled or may actually represent feral wilddomesticated hybrids. Phylogeographic analysis of genes involved in the core domestication process suggests that the origins of causal domestication mutations could be traced to wild progenitors in multiple different locations in West and Central Africa. In addition, measurements of panicle threshability, a key early domestication trait for seed shattering, were consistent with the gene phylogeographic results. We suggest seed non-shattering was selected from multiple genotypes, possibly arising from different geographical regions. Based on our evidence, O. glaberrima was not domesticated from a single centric location but was a result of diffuse process where multiple regions contributed key alleles for different domestication traits.
Figure 3: Phylogenomic and principal component analysis of African rice. (A) Neighbor-joining tree built using a distance matrix estimated from NGSdist. Color strips represent group of O. glaberrima individuals sharing a common ancestor. Internal branches with red circle denote bootstrap support of greater then 80%. Dark arrows indicates the O. glaberrima grouping with divergent O. barthii groups, and white arrows indicates the O. glaberrima grouping with OB-G group O. barthii. (B) Geographical distribution of each individual and colored by its grouping status as outlined in (A). (C) Principal component analysis conducted using NGScovar. Individuals are color coded according to (A)