Rice Genetics IV - IRRI books - International Rice Research Institute

wild and cultivated types (Est10, Wang XK et al 1992) and indica and japonica types(such as Cat1 and Amp2) were found in or near the QTL clusters. Similar clusterphenomena of domestication-related QTLs were reported by Koinange et al (1996)in common bean and by Xiong et al (1999) in rice.To determine whether QTL clusters are loosely linked loci or a single locus witha pleiotropic effect, more precise analysis is necessary. It is interesting to note thatQTLs for mesocotyl length, which is considered to reflect endogenous hormone level,tended to join those clusters. This might suggest a possibility that the clustering phenomenonis partly due to the pleiotropic effect of an unknown key factor controllingvarious traits through hormonal regulation.In this O. sativa × O. rufipogon cross, character correlations were generally weakamong RILs that represent essentially F 2 variation. This implies that nonrandom characterassociations of the domestication syndrome are mainly due to natural selectionfor coadapted traits. The present analysis revealed several QTL linkages reflectingthe domestication syndrome. Since they were located on different chromosomes, theysegregated into RILs, resulting in decreased correlations. This phenomenon could beunderstood by “multifactorial linkages” advocated by Grant (1975). This phenomenonis inevitably caused by random distribution of multiple factors over the limitednumber of chromosomes that determine the differences in two or more quantitativecharacters.QTL clusters thus identified were mapped on the regions in which cultivar-derivedalleles segregated at higher frequencies than expected. This coincidence couldbe explained by unconscious selection, which worked under cultivation pressure whenestablishing RILs favoring the combination of cultivar-derived gene blocks.Inference for domestication dynamicsThe domestication process is undoubtedly a gradual process directed by humans. Inrice, Oka and Morishima (1971) demonstrated early that “cultivation pressure” broughtabout a rapid change in population genotype toward the cultivated type as suggestedby Harlan (1975). Seed shedding is the critical trait disruptively selected in naturaland cultivated fields. The genetic basis of seed shedding is simpler than other domestication-relatedtraits. Eiguchi and Sano (1990) identified two loci conferring highseed shedding on O. rufipogon, one of which was linked to the spreading paniclelocus on chromosome 4. Our QTL analysis detected five loci for seed shedding andone of them located on chromosome 1 was linked with another spreading paniclelocus (Cai and Morishima 2000b). A spreading panicle is advantageous for wild riceto disperse seeds but disadvantageous for cultivated rice. It may be reasonable toinfer that selection for nonshedding genes with relatively large effects played an importantrole as a trigger for domestication.Our experiments with another cross (O. sativa × O. rufipogon) showed that associationamong seed shedding, seed dormancy, and awn length tended to increase inlater hybrid generations without deliberate selection (Table 3). In the cross propagatedin bulk, the outcrossing rate was higher in the lines with wild characteristicsthan in those with cultivated-rice characters (Table 3). This suggests that flower char-Evolution and domestication of rice 73