marker-assisted selection in wheat

290Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishof the trait and the larger the number ofgenes affecting the trait, the larger thesample size required (see Strauss, Landeand Namkoong, 1992). As shown byBrown et al. (2003), the use of small mappingpopulations of 100–200 segregatingindividuals, typical of most QTL studiesin trees, is likely to cause an upward biasin the estimated phenotypic effect of QTL.Simulation and practical studies have shownthat, in addition to sample size, QTL detectionis affected by genetic background,environment and interactions among QTL.The location of QTL is imprecise as theycan only be mapped to 5–10 cM. This maytranslate into a physical distance of severalmegabases, which may contain severalhundred genes. The effect of a QTL is alsolikely to vary over time in perennial plantswith changing biotic and abiotic factors(Brown et al., 2003). This highlights thenecessity of verifying QTL in differentseasons, environments and genetic backgrounds(Sewell and Neale, 2000). Thechallenges of developing and genotypingthe large progeny arrays required to locateQTL accurately in outbred pedigrees, andof verifying these QTL in different environmentsand ages, are such that MAS hasnot yet been applied in any commercial treebreeding programme.In one of the most intensive studies onapplying MAS to date, and based on datafrom over 1 300 individuals for wood density,4 400 individuals for wood diameterfrom a single pedigree and using selectivegenotyping of the 50 highest and lowestscoring individuals for density and 100of each for diameter, Devey et al. (2004a)were able to validate (in the same pedigree)two out of 13 QTL for diameter and eightout of 27 QTL for wood density in Pinusradiata. The effect of each QTL rangedfrom 0.8 to 3.6 percent of phenotypicvariation, implying that these traits werecontrolled by a large number of genes, eachof small effect. Using a different approach,Brown et al. (2003) used a verificationpopulation of 447 progeny (derived fromre-mating the parents of the QTL pedigree)and an “unrelated population” of 445progeny from the base pedigree to verifyQTL in Pinus taeda. They found about halfthe QTL were detected in multiple seasonsand fewer QTL were common to differentpopulations.An area where QTL mapping may assistbreeders is in breaking linkages betweennegatively correlated traits. For example inE. grandis and E. urophylla, Verhaegen et al.(1997) reported co-location of QTL for thenegatively correlated traits of wood densityand growth. If these traits are controlled bytightly linked genes, markers could be usedto select favourable recombinants.Most markers used in QTL mappinghave been anonymous markers that areunlikely to occur in a gene influencing aquantitative trait. In an attempt to increasethe power of QTL mapping, candidategenes that may control the trait in questionare being used as molecular markers.Candidate genes are typically sourced fromthe tissue of interest (e.g. xylem or leaves)and have either a known function intuitivelyrelated to the trait, or are of interestfrom studies of their expression usingDNA microarrays. Comparative mappingand candidate gene approaches can utilizesuch information to search for homologousgenes in different genomes. Candidategenes have been mapped to QTL for woodquality in E. urophylla and E. grandis (Gionet al., 2000), Pinus taeda (Neale, Sewell andBrown, 2002), and E. globulus (Thamaruset al., 2004). They have also been mappedto QTL for bud set and bud flush inPopulus deltoides (Frewen et al., 2000).