Rice Genetics IV - IRRI books - International Rice Research Institute

ferently to the environments. Similarly, five E-QTL pairs for heading date showedopposite AAE effects in different environments (Li et al 2001b). Thus, the power(LOD score) by which a QTL is detected and the magnitude of its effect obtained in asingle environment provide little information regarding its performance in another.All of these results strongly justify the advantage of marker-aided selection for QTLsof different behavior over phenotypic selection, even for highly heritable traits.Finally, epistasis plays an important role in GE interactions. In addition to theabove two pieces of evidence, 28.3% of the QTL main effects (27.8% for plant heightand 28.9% for heading date) were detectable only in epistatic models (Tables 6 and7). In other words, nearly 30% of the undetected M-QTLs in a single environmentwere due to epistasis.Differential gene expression to biotic and abiotic stresses is largely responsiblefor GE interactions of quantitative traits. The greater level of QE interactions observedfor QTLs affecting rice heading date is not surprising since the flowering timeof rice plants is known to be affected by many environmental factors, such as daylength,temperature, soil fertility, drought, etc. Strong evidence for the presence of epistaticinteractions between and among different M-QTLs for rice heading date and theirdifferential responses to daylength has been clearly demonstrated using near-isogeniclines (Lin et al 2000). Table 9 shows another more extreme case of QE interactionsaffecting plant height under different submergence conditions. Under nonstress andsubmergence stresses, different sets of M-QTLs affecting plant height (elongation)were detected and the expression of most M-QTLs was much stronger under the morestressful condition (under the submergence of muddy water of the field) than underthe clean water submergence in the greenhouse. It is also interesting to note that themajor gene Sub1 for submergence tolerance was detected as a small QTL when theplants were submerged under the clear water condition (test 1). Similarly, it was clearlyshown that the bacterial blight resistance gene Xa4 acts as a major resistance geneagainst the avirulent races of Xanthomonas oryzae pv. oryzae (Xoo), but as a resistanceM-QTL against the virulent races of Xoo (Li et al 1999b, 2001c).Molecular dissection of trait correlationDetermination of the genetic basis of trait correlation has been a major challenge inquantitative genetics and it has important implications for plant and animal improvement.It is well known that trait correlation may arise from linkage, pleiotropy, andepistasis. In addition to its genetic determinants, trait correlation may have physiologicaland environmental bases (Falconer 1983). In this respect, QTL mapping canhelp in gaining insights into this problem by identifying clustered QTLs mapped forcorrelated traits and comparing their genetic parameters (locations and effects).Table 10 shows some examples of clustered M-QTLs affecting correlated traits inrice. Cases 1 and 2 represent examples of pleiotropy, which is often responsible forpositive correlation between a complex phenotype (spikelets per panicle) and its componenttraits (number of primary and secondary branches per panicle). Cases 3 and 4present a situation of QTLs affecting developmentally related traits. For QLusi12QTL mapping in rice: . . . 163