QTL-based analysis of genotype-by-environment interaction for grain yield of rice in stress and non-stress environments

Use of DNA-based markers can accelerate cultivar development in variable cultivation environments since, in contrast to phenotype, DNA markers are environment-independent. In an effort to elucidate the genetic basis of genotype-by-environment interaction (G x E) for yield of rice (Oryza sativa L.), the associations between 139 AFLP markers and grain yield were determined for rice grown in fresh water (EC of 0.65 dS m(-1) and saline conditions (EC of 4-8 dS m(-1)) with 0 kg ha(-1) or 100 kg ha(-1) nitrogen fertilizer in the years 2000 and 2001. A population of recombinant inbred lines of rice, developed from an IR29 x Pokkali cross, was used in the study. Both genotype x salinity and genotype x nitrogen level interactions were significant, with the genotype x salinity interaction being stronger. Through multiple regression analysis using a stepwise procedure for selecting markers, 36 markers were detected for grain yield in the four test conditions and of these 28 were detected in only one test condition implying strong environmental specificity for yield QTL expression. However, the fact that eight QTLs were detected in more than one test condition points to the existence of wide-adaptability genes in this cross. Markers with significant associations with yield explained between 37% and 48% of the yield variation in each test condition. Superior genotypes of rice were identified in all four test conditions based on their marker signatures. Furthermore, across N fertilizer regimes, yield predicted from summed additive effects of QTLs were significantly correlated with observed yield in the same year and across years. Thus marker-assisted selection can help breeders overcome the problem of low selection efficiency encountered during phenotypic selection for yield in stress environments.