To understand the types of gene action controlling seven quantitative traits in rice, we carried out quantitative trait locus (QTL) mapping in order to distinguish between the main-effect QTLs (M-QTLs) and digenic epistatic QTLs (E-QTLs) responsible for the trait performance of 254 recombinant inbred lines (RILs) from rice varieties Lemont/Teqing and two backcross hybrid (BCF1) populations derived from these RILs. We identified 44 M-QTL and 95 E-QTL pairs in the RI and BCF1 populations as having significant effects on the mean values and mid-parental heterosis of heading date, plant height, flag leaf length, flag leaf width, panicle length, spikelet number and spikelet fertility. The E-QTLs detected collectively explained a larger portion of the total phenotypic variation than the M-QTLs in both the RI and BCF1 populations. In both BCF1 populations, over-dominant (or under-dominant) loci were more important than additive and complete or partially dominant loci for M-QTLs and E-QTL pairs, thereby supporting prior findings that overdominance resulting from epistatic loci are the primary genetic basis of inbreeding depression and heterosis in rice.

One hundred twenty six doubled-haploid (DH) rice lines were evaluated in nine diverse Asian environments to reveal the genetic basis of genotype x environment interactions (GEI) for plant height (PH) and heading date (HD). A subset of lines was also evaluated in four water-limited environments, where the environmental basis of G x E could be more precisely defined. Responses to the environments were resolved into individual QTL x environment interactions using replicated phenotyping and the mixed linear-model approach. A total of 37 main-effect QTLs and 29 epistatic QTLs were identified. On average, these QTLs were detectable in 56% of the environments. When detected in multiple environments, the main effects of most QTLs were consistent in direction but varied considerably in magnitude across environments. Some QTLs had opposite effects in different environments, particularly in water-limited environments, indicating that they responded to the environments differently. Inconsistent QTL detection across environments was due primarily to non- or weak-expression of the QTL, and in part to significant QTL x environment interaction effects in the opposite direction to QTL main effects, and to pronounced epistasis. QTL x environment interactions were trait- and gene-specific. The greater GEI for HD than for PH in rice were reflected by more environment-specific QTLs, greater frequency and magnitude of QTL x environment interaction effects, and more pronounced epistasis for HD than for PH. Our results demonstrated that QTL x environment interaction is an important property of many QTLs, even for highly heritable traits such as height and maturity. Information about QTL x environment interaction is essential if marker-assisted selection is to be applied to the manipulation of quantitative traits.

One hundred and ninety three parental lines obtained from 26 countries for an international rice molecular breeding program were evaluated using 101 well-distributed simple sequence repeat (SSR) markers. An overall genetic diversity of 0.68 and an average of 6.3 alleles per locus were revealed, indicating a high level of genetic variation in these lines. Cluster analysis of the 193 accessions showed three major groups and nine subgroups. Group I corresponded to the classical indica subspecies, whereas groups II and III belong to the japonica subspecies. Indica and japonica differentiation accounted for only 6.5% of the total variation in the entire sample and 93.5% was due to within-subspecies diversity. Differentiation among eco-geographic regions accounted for 24% of the diversity within the subspecies. Larger amounts of the eco-geographical differentiation were resolved within japonica than within indica. The largest indica-japonica differentiation based on the single locus level was detected by markers on chromosomes 9 and 12, while the smallest differentiation was detected by markers on chromosomes 4 and 8. Furthermore, genetic differences at the single-locus and two-locus levels, as well as components due to allelic and gametic differentiation, were revealed between indica and japonica and among the main geographic regions. The multilocus analysis in genetic diversity showed a higher proportion of variation caused by predominant non-random associations of different loci within and among the classified subspecies and geographic subdivisions. The results suggest that selection for eco-geographical adaptation on multilocus associations was largely responsible for the maintenance of the extensive variation in the primary gene pool of rice.

To understand the types of gene action controlling seven quantitative traits in rice, QTL mapping was performed to dissect the main effect (M-QTLs) and digenic epistatic (E-QTLs) QTLs responsible for the trait performance of 254 recombinant inbred lines (RILs) of "Lemont/Teqing", and two testcross (TC) F(1) populations derived from these RILs. The correlation analyses reveal a general pattern, i.e. trait heritability in the RILs was negatively correlated to trait heterosis in the TC hybrids. A large number of M-QTLs and E-QTLs affecting seven traits, including heading date (HD), plant height (PH), flag leaf length (FLL), flag leaf width (FLW), panicle length (PL), spikelet number per panicle (SN) and spikelet fertility (SF), were identified and could be classified into two predominant groups, additive QTLs detected primarily in the RILs, and overdominant QTLs identified exclusively in the TC populations. There is little overlap between QTLs identified in the RILs and in the TC populations. This result implied that additive gene action is largely independent from non-additive gene action in the genetic control of quantitative traits of rice. The detected E-QTLs collectively explained a much greater portion of the total phenotypic variation than the M-QTLs, supporting prior findings that epistasis has played an important role in the genetic control of quantitative traits in rice. The implications of these results to the development of inbred and hybrid cultivars were discussed.

Two sets of reciprocal introgression line (IL) populations, i.e., ILs with Lemont as recurrent parent (IL_LT) and ILs with Teqing as recurrent parent (IL_TQ), were developed and evaluated for traits representing panicle size, including primary branch number (PBN), secondary branch number (SBN), and spikelet number per panicle (SNP). Together with the regression to recurrent parent by advanced backcross, transgressive segregations were observed for all traits. Correlation and regression analysis showed that SBN had much higher contribution to SNP than PBN. It was confirmed by the QTL analysis that many common loci were detected between SBN and SNP, in comparison with single common locus between PBN and SNP. One and three main effect QTLs (M-QTLs) were detected for PBN in IL_LT and IL_TQ, respectively. Six M-QTLs per trait per populations were associated with SBN and SNP. Less number and lower contribution of epistasis were detected in IL populations in comparison with mapping result from F2 or RI population. There were only four QTLs in fourteen loci (near 30%) commonly detected in both reciprocal IL populations implying the large impact of genetic background on QTLs expression.

Drought is the most important factor limiting rice productivity in the rainfed areas of Asia. In this study, 48 pyramiding lines (PLs) and their recurrent parent, IR64, were evaluated over two years for their yield performances and related traits under severe drought stress at the reproductive stage (RS), the vegetative stage (VS) and irrigated control in order to understand the relationship between drought tolerance (DT) and yield potential (YP) in rice and their underlying mechanisms. When compared with IR64, all PLs had significantly improved DT to RS and 36 PLs also had significantly improved DT to VS. In addition, 17 PLs had higher YP than IR64 and the remaining 31 PLs had a similar YP IR64 under irrigated conditions. Detailed characterization of the PLs revealed three possible mechanisms that functioned together to contribute to their improved DT. The most important mechanism was dehydration avoidance (DA), characterized by significantly higher growth rate and biomass of all PLs than IR64 under stress and no reduction in biomass under control conditions. The second mechanism was efficient partitioning, characterized by improved harvest index in all PLs compared with IR64, resulting primarily from heavier grain weight and/or higher spikelet fertility under control conditions, which was the major constituent of the improved YP in the 17 best performing PLs. Drought escape (DE) by accelerated heading under drought was the third mechanism that contributed to DT of the PLs to RS. The considerable variation in the measured traits among the PLs with similar levels of DT and YP implies the complex genetic control of the mechanisms for DT/YP and offers opportunities to improve DT and YP further by fine-tuning of a small number of QTLs segregating among the PLs using MAS. Finally, our results indicate that selection for yield plus some secondary traits under appropriate type(s) of stress and non-stress conditions similar to the target environments are critically important for improving both DT and YP in rice.

A large set of 254 introgression lines in an elite indica genetic background were evaluated for grain yield (GY) and related traits under the irrigated (control) and drought (stress) conditions in two consecutive years for genetic dissection of adaptive strategies of rice to water stress. A total of 36 quantitative trait loci (QTLs) affecting heading date (HD), plant height (PH), GY and yield components were identified and most QTLs showed pronounced differential expression either qualitatively or quantitatively in response to drought. These QTLs could be grouped into three major types based on their behaviors under control and stress conditions. Type I included 12 QTLs that expressed under both the stress and non-stress conditions. Type II comprised 17 QTLs that expressed under irrigation but not under stress. Type III included seven QTLs that were apparently induced by stress. The observation that the Lemont (japonica) alleles at all HD QTLs except QHd5 resulted in early heading under stress appeared to be responsible for the putative adaptation of Lemont to drought by escaping, whereas the Teqing (indica) alleles at most PH/GY QTLs were consistently associated with increased yield potential and trait stability and thus contributed to DT. Our result that most DT QTLs were non-allelic with QTLs for drought escaping suggests that the two adaptive strategies in the parental lines are under possible negative regulation of two largely non-overlapping genetic systems.

Complete resistance (CR) and partial resistance (PR) of rice (Oryza sativa L.) to its bacterial pathogen, Xanthomonas oryzae pv. oryzae (Xoo), was genetically dissected by using 2 mapping populations and 10 Xoo races. Two CR genes, 50 quantitative resistance loci, and 60 digenic interactions were identified, which showed various degrees of race specificity to the Xoo races. The complex epistasis between these loci led us to the discovery of complex genetic networks underlying the rice defensive system to Xoo. The networks consisted of two major components: one representing interactions between alleles at the R loci of rice and alleles at the corresponding avirulence loci of Xoo for CR and the other comprising interactions between quantitative resistance loci in rice and their corresponding aggressiveness loci in Xoo for PR. The race specificity of PR and its strong genetic overlap with CR indicate that PR is essentially ''weaker'' CR. The genetic networks discovered are expected to maintain a high level of the allelic diversity at avirulent loci in the pathogen by stabilizing selection, which may maintain a high allelic diversity at R loci in the host by the frequencydependent selection.complete resistance ͉ epistasis ͉ partial resistance ͉ plant-pathogen interaction ͉ rice bacterial blight