Towards association analysis of drought adaptation in a rice diversity panel

Genetic manipulation to enhance drought adaptation of rice cannot be achieved by conventional breeding alone because the key players are quantitative trait loci (QTL) with cumulative effects. This study aims to determine association between marker genotypes and drought tolerance traits. A Diversity and Stress Adaptation (DSA) panel comprised of 201 INGER entries, 54 IRGC traditional accessions, and 45 global diversity inbreds and released cultivars was assembled. These entries were screened in Batac, Ilocos Norte (BIN), Bayombong, Nueva Vizcaya (BNV), and Munoz, Nueva Ecija (CES) in 2010 DS. Drought stress was most severe in BNV (no supplemental irrigation), less severe in BIN (once a week surface irrigation), and least severe in CES (1-2 times a week surface irrigation). For CES where data were more complete, grain yield was analyzed under a mixed model framework, with block and entry treated as random effects terms, and check considered as a fixed effects term. Blocks were allowed to have a diagonal covariance structure to account for spatial heterogeneity in the field. The highest-yielding entry was IR77724-8-2-3-2-2 (0.9 t/ha). Selections from the Umingan, Pangasinan experiment during the previous year (2009 WS) were hybridized with elite varieties in 2010 DS. Considering overall performance in terms of grain yield, vigor, and phenotypic acceptability across locations, a number of entries (28 INGER entries, 2 IRGC accessions, and 4 diversity and released cultivars) were selected for hybridization with elite varieties in 2010 WS. Molecular marker data were generated through genome-wide fingerprinting of 124 SSRs at a targeted density of 1SSR per 3 MB for association mapping. Multi-environmental data were analyzed using mixed models with correction for population structure and genetic covariance among genotypes and environments. Marker-trait association will be carried out by single marker analysis and QTL environment effects will be estimated by fitting a multi-QT1 model.