Population structure of the rice composite set

The population structure of the composite rice collection of 2757 accessions was investigated using 50 SSR markers through collaboration between IRRI, Agropolis-CIRAD, CIAT, EMBRAPA, CAAS, WARDA and Cornell University. To construct the composite rice set, nominations of materials were solicited from the genebank and other holdings among the partners and from within the breeding programs and INGER at IRRI. Accessions were chosen to balance coverage across 7 eco-regional gene pools that reflect centers of diversity and its dispersion throughout the world: South, East, Southeast, West Asia, Oceania, Africa plus Central and South America, and Europe plus North America. This set included special purpose sets of germplasm that have been extensively characterized, both phenotypically and by molecular markers, as well as reference lines for drought mapping populations and the Cornell panel for SSR analysis. The SSRs used are a subset of the Cornell panel developed by Coburn et al. (2002, Crop Sci. 42:2092-2099). Statistics on the SSRs were computed using PowerMaker version 3.25. Clustering was accomplished using DARwin 5 with the simple matching coefficient for distances followed by unweighted Neighbor Joining. Model-based inference of population structure was attempted using Structure 2.1 using a burn-in of 5,000 to 10,000 run lengths from 15,000 to 40,000 and a model allowing for admixture and correlated allele frequencies. Entries with less than 80 percent contribution from a single group are considered admixed. The structures observed in these initial analyses are analogous to those for sativa (Garris et al., 2005, Genetics 169:1631-1638) and for Glaberrima (Semon et al, 2005, Genetics 169:1639-1647). Currently, allele calls are being checked for consistency, and the re-analyzed data will be used to clarify the degree of admixture. Once publicly available, new tools developed specifically for modeling inbred population structure will be applied.