Molecular markers allow the analysis of the genetic control of quantitative traits by mapping QTL. Mapping QTL is based on a systematic search for association between the genotype at a marker locus and the average value of a phenotypic trait. It requires a segregating population, e.g., derived from the cross between two individuals with different values of the trait of interest. For each individual within the population, the genotype of marker loci distributed over the entire genome is determined and a saturated genetic map is constructed. Simultaneously, the value of the trait of interest is measured for each individual. Statistical methods are then used to find marker loci whose genotype is correlated with the trait and to estimate the genetic parameters of the QTL detected. Several biometric techniques have been proposed to find QTL, from the most simple, based on analysis of variance (ANOVA), applied marker by marker, to methods that simultaneously take into account two or more markers (Lynch and Walsh 1998). As a QTL can only be detected if the corresponding gene is polymorphic, the sample of QTL detected is specific to each progeny. The choice of parental lines is thus very important. For distantly related parents QTL detection is easiest, but major effect QTL may hamper the detection of lower effect QTL that would be detected if the major QTL were fixed.
Populations showing the highest efficiency in mapping QTL are those derived from crosses between two homozygous lines, such as F2, RILs and BC. F2s are the only populations allowing the dominance effect to be estimated, while a mixture of additivity and dominance is estimated with BC. Tanksley and Nelson (1996) proposed to search for QTL in advanced back-cross (AB) BC2, BC3, and BC4 populations. Although the power of QTL detection is reduced in comparison to F2 or RIL, this strategy is attractive when screening positive alleles from a wild species, as it allows the identification of additive effects, it reduces linkage with unfavorable alleles around the QTL, and simultaneously advances the production of enhanced lines (see Sect. 1.8). The efficiency of detecting a particular
QTL in a segregating population is low partly because other QTLs are segregating and major QTLs are masking the minor ones. For this reason, Eshed and Zamir (1995) proposed the use of ILs in which each line possesses a unique segment from a wild progenitor introgressed into the same genetic background (see Sect. 1.8).
Wild species produce low quality fruit but can carry alleles at QTL which may improve agronomic traits. Such transgressive QTLs have been discovered frequently. Even when highly contrasted individuals have been chosen as parents of a population, it is not rare to find a QTL showing an effect opposite of that expected from the value of the parents. Results from advance backcross experiments in tomato have shown, for example, unexpected positive transgressions from wild relatives for various fruit traits (Bernacchi et al. 1998b).
Epistasis between QTLs has rarely been detected with classical populations (Tanksley 1993b), but this is mostly due to statistical limitations for the populations studied. One way of increasing the reliability of epistasis analysis is to eliminate the "background noise" due to other QTLs, using crosses between NILs (see Sect. 1.8.2) differing only by a chromosome fragment carrying a QTL (Eshed and Zamir 1996).
Was this article helpful?