Relationships between molecular marker polymorphism and hybrid performance in rice

This study investigated the relationship of molecular marker heterozygosity with hybrid performance and heterosis in rice in an attempt to make use of molecular markers in predicting hybrid performance. Four sets of diallel crosses, 147 in total, were made using parents encompassing a wide range of germplasm, including 1) a set of 28 crosses using 8 elite parental lines commonly used in hybrid rice production in China, 2) a set of 36 crosses using 9 indica varieties representing a wide range of germplasm from three countries, 3) a set of 55 crosses using 11 japonica varieties also with a wide range of representation from three countries, and 4) a set of 28 crosses using 8 long-grained rice lines from the southern United States. All F1s and their parents were evaluated for agronomic performance in replicated field trials. The parental lines were surveyed for DNA polymorphisms using more than 100 molecular markers including restriction fragment length polymorphisms and simple sequence repeats. The marker genotypes of the F1s were deduced from the genotypes of their parents. The results showed that there are extensive genetic polymorphisms among the parental lines of each diallel set. High levels of heterosis are common in yield and yield component traits in crosses of all four diallel sets. The correlations between marker genotype divergence and hybrid performance varied greatly from one data set to another depending on the materials studied. The highest correlation was observed between hybrid yield and general heterozygosity in the set of southern U.S. long-grained varieties. A high level of correlation was also observed between specific heterozygosity and heterosis in the set of Chinese hybrid rices, whereas the correlations were much lower in the sets of indica and japonica mixtures. Detailed characterization of the rice germplasm and in-depth understanding of the genetic basis of heterosis are needed for molecular markers to have practical utility in predicting hybrid performance.