Abstract Yield and its components were investigated by using a population of241 recombinant inbred lines (F 9 RILs) derived from an elite hybridrice cross of Zhenshan 97 · Minghui 63. Quantitative trait loci(QTLs) for causal analysing of yield traits were detected at differentyield component (YC) influences by conditional and unconditionalQTL mapping methods. The number of QTLs significantly affectingyield was different at component-special influence. Some QTLscontrolling yield identified in one component influence were undetect-able at the others. More QTLs for yield could be detected at differentYC influences. It is possible to reveal that causal gene expression foryield could be different at different YC influences. Mapping QTLs forcomponent effects of yield could help us in understanding the nature ofcause-effect traits for the formation of grain yield. Key words: Oryza sativa — quantitative trait loci — recom-binant inbred line — yield — yield componentGrain yield and its four components in cereal crops are alwaysone of the most important target traits in research andbreeding programmes. Faced with the same yield in differentcultivars, there mainly are three model systems of high yieldfrom the view of yield components (YC), such as heavy-panicle, multipanicle and mixture. As Donald (1968) advoca-ted the ideotype concept to describe the potential of enhancingyield by improving the individual component trait, manycomponent analyses have been studied for complex traitsbased on morphological and physiological characterization(Sparnaaij and Bos 1993, Gravois and Helms 1992, Piepho1995, Guo et al. 2002a). In addition, Sidwell et al. (1976)reported that through YC improvement to increase yield (YD)would be most effective because of lower heritability (0.12) forYD (Xiong 1992) and higher heritability for tillers per plant(TP; 0.45) and 1000-grain weight (GW; 0.50) (Li et al. 1998).However, selection for YCs in early segregating generationswas not a highly effective method because of the negativecorrelation between YCs (Li et al. 1998).To solve this contradiction, Backes et al. (1995) and Zhu(1995) suggested that focusing on component gene/quantita-tive trait locus (QTL) for YD should be a more effectiveapproach than looking for QTL for YD directly, because theformer not only allows complex traits to be divided intocomponent QTL that may be under separated and probablysimpler genetic control, but can also exclude the influence ofother components.Because the first restriction fragment length polymorphism(RFLP) map of rice was constructed, hundreds of genes andQTLs have been mapped (Xiao et al. 1995, Lin et al. 1996,Veldboom and Lee 1996, Zhuang et al. 2002). Some yieldQTLs coincided with those for component traits, explainedfrequently as pleiotropic effects or linkage. In these studies, acommon problem associated with the analyses of the datareported so far is that the analysis of QTLs for YD and itscomponents was conducted based on the phenotypic valuesseparately. Although it can provides statistical estimates forthe magnitude of the effects and the proportions of variationexplained, it is nonetheless impossible to detect the componentQTLs controlling complex yield trait based on the phenotypicvalue of yield conditional on the phenotypic value of thiscomponent (Zhu 1995).According to the theory of developmental genetics, genes areexpressed selectively at different time and position (Atchleyand Zhu 1997). Piepho (1995) proposed that the complex yieldcharacter may be regarded as the end-point of a process ofwhich the successive stages are represented by observedprimary characters TP, grains per panicle (GN), percentageseed set and GW. At different YC influences, the yield geneexpressing might behave differentially. It is necessary, there-fore, to understand the component gene expression for yield atdifferent yield factor influences as a basis for quantitative traitmanipulation (Xu 1997). So far, no report has explored thecomponent QTL expression in yield formation in rice. Thesubject of this study is by combining the newly developedstatistical procedures of analysing conditional genetic effects(Zhu 1995, Yan et al. 1998) and the composite intervalmapping method (Zeng 1993, 1994), to dissect the componentQTLs for yield at different component influences, to improveour understanding for gene expression affecting grain yield,and to provide an outline of genetic basis for the cause-resulttraits and useful information for genetic improvement of yieldpotential in rice.
Read full abstract