QTL-by-QTL, QTL-by-environment, and QTL-by-QTL-by-environment interactions of loci controlling grain length in rice

In order to obtain novel information about QTL-by-QTL, QTL-by-environment, and QTL-by-QTL-by-environment interactions (QQ, QE, and QQE, respectively), four loci controlling grain length in rice (Oryza sativa L.), that is, GW2, OsLG3b, and GS3, derived from a large-grain cultivar, BG 1, and LKF, from another cultivar, Fusayoshi, were targeted as a model system of quantitative trait. First, four single-QTL near-isogenic lines (NILs), each of which was a homozygote of a long-grain allele at one of the four loci in a common genetic background of cultivar Gimbozu, were developed. Then, pyramiding of these long-grain alleles was conducted through intercrossing between single-QTL NILs and selection of segregants of the longest grains, leading to another series of double-QTL NILs. All of these NILs, which were confirmed genotypes using molecular markers, were grown with their parents over eight years (environments). QQ, QE and QQE, as well as the additive effect and environmental effect, were estimated based on a linear model and analyzed using interaction principal component analysis (IPCA). All genetic terms and environmental effects showed highly significant variations. QQs showed all negative values and their magnitude tended to be smaller than those of the additive effects on average. The IPCA clearly showed that the additive effects and QQs responded to environments differentially and not randomly. In addition, a similar differential response was also observed between QQs including the long-grain allele at LKF and those excluding this allele. Such a demonstrated nature of these interactions may provide valuable suggestions on practical breeding.

Automated summary

This study investigated the QTL-by-QTL, QTL-by-environment, and QTL-by-QTL-by-environment interactions by targeting four loci controlling grain length in rice. Through the development of near-isogenic lines with different genetic combinations and the analysis using linear models and interaction principal component analysis, the variations of additive effects, QQ effects, and environmental effects were estimated. The results showed differential responses of QQ effects to different environments, as well as differences between QQ effects including the long-grain allele at LKF and those excluding this allele.