Germplasm conserved in gene banks is underutilized, owing mainly to the cost of characterization. Genomic prediction can be applied to predict the genetic merit of germplasm. Germplasm utilization could be greatly accelerated if prediction accuracy were sufficiently high with a training population of practical size. Large-scale resequencing projects in rice have generated high quality genome-wide variation information for many diverse accessions, making it possible to investigate the potential of genomic prediction in rice germplasm management and exploitation. We phenotyped six traits in nearly 2000 indica (XI) and japonica (GJ) accessions from the Rice 3K project and investigated different scenarios for forming training populations. A composite core training set was considered in two levels which targets used for prediction of subpopulations within subspecies or prediction across subspecies. Composite training sets incorporating 400 or 200 accessions from either subpopulation of XI or GJ showed satisfactory prediction accuracy. A composite training set of 600 XI and GJ accessions showed sufficiently high prediction accuracy for both XI and GJ subspecies. Comparable or even higher prediction accuracy was observed for the composite training set than for the corresponding homogeneous training sets comprising accessions only of specific subpopulations of XI or GJ (within-subspecies level) or pure XI or GJ accessions (across-subspecies level) that were included in the composite training set. Validation using an independent population of 281 rice cultivars supported the predictive ability of the composite training set. Reliability, which reflects the robustness of a training set, was markedly higher for the composite training set than for the corresponding homogeneous training sets. A core training set formed from diverse accessions could accurately predict the genetic merit of rice germplasm.

Despite the great success achieved by the exploitation of heterosis in rice, the genetic basis of heterosis is still not well understood. We adopted an advanced-backcross breeding strategy to dissect the genetic basis of heterosis for yield and eight related traits. Four testcross (TC) populations with 228 testcross F₁ combinations were developed by crossing 57 introgression lines with four types of widely used male sterile lines using a North Carolina II mating design. Analysis of variance indicated that the effects of testcross F₁ combinations and their parents were significant or highly significant for most of the traits in both years, and all interaction effects with year were significant for most of the traits. Positive midparent heterosis (H_MP) was observed for most traits in the four TC populations in the two years. The relative H_MP levels for most traits varied from highly negative to highly positive. Sixty-two dominant-effect QTL were identified for H_MP of the nine traits in the four TC populations in the two years. Of these, 22 QTL were also identified for the performance of testcross F₁. Most dominant-effect QTL could individually explain more than 10% of the phenotypic variation. Four QTL clusters were observed including the region surrounding the RM9–RM297 region on chromosome 1, the RM110–RM279–RM8–RM5699–RM452 region on chromosome 2, the RM5463 locus on chromosome 6 and the RM1146–RM147 region on chromosome 10. The identified QTL for heterosis provide valuable information for dissecting the genetic basis of heterosis.