In breeding applications, molecular markers may be either diagnostic (i.e. perfect markers of a specific allele within a specific gene sequence) or putative (e.g. markers associated with or flanking a QTL that has been discovered via mapping in biparental crosses or in association panels of related or unrelated lines). Diagnostic markers are preferable as they can be used to select desired alleles in any parental or progeny line of a species under any crossing strategy. Alternatively, such markers can be used to prepare a gene for transgene (genetic modification, GM) approaches within or across crop species. QTL markers from mapping studies are not perfectly linked to genes (i.e. they are 'nearby') and are frequently difficult to transfer between crosses, unless there is substantial research investment in crossing and mapping to 'fine map' the QTL to locate markers that are within a gene. The identification of QTLs for complex traits is further confounded by substantial genotype x environment interaction effects that occur for traits like heat and drought adaptation. Crossa et al. (Chapter 14, this volume) highlight the new capabilities in statistical methods, such as modelling of QTL x environment effects (Boer et al., 2007) that enable the more robust detection of useful genomic regions for selection.
MAS is routinely applied for traits such as disease resistance and grain quality characters when diagnostic gene-based markers have been identified (William et al., 2007; Whitford et al., Chapter 12, this volume). Such markers, located within a gene sequence, are discovered through either fine mapping around QTLs of large effect, or by looking for gene candidates that are part of known pathways (e.g. functional disease resistance genes). Only few QTLs of large effect have been documented for performance-related traits under heat or drought and no candidate genes in known biochemical pathways of response to heat or drought have been shown to have large effects on performance traits such as yield. Using fine-mapping approaches, genes have been identified and cloned for a number of abiotic stresses, including salinity, flooding, Al tolerance and B tolerance (Collins et al., 2008), but none has been cloned from QTLs associated with drought or heat stress. In part, the low success rate for these stresses relates partly to the genetic and environmental complexity of adaptation. Mapping populations have frequently been made by crossing highly contrasting parents to maximize genetic polymorphisms in the progeny. Therefore, performance QTLs identified in random lines or in deliberately contrasting lines are likely to be associated with traits that have already been optimized by breeding. Furthermore, in crops like wheat and barley, it has been demonstrated that segregation for genes of major agronomic effect (height and maturity) within experimental populations makes it more difficult to identify QTLs of minor effect that may be associated with more direct mechanisms of adaptation (Reynolds et al., 2009). Using mapping populations with more uniform flowering time (Olivares-Villegas et al., 2007) both trait and yield QTLs were readily detected independent of loci associated with phenology (Pinto et al., 2010).
Discovery and utilization of QTLs for drought and heat tolerance requires further investment in development of genetic resources and in more detailed phenotypic 'dissection' of complex performance traits.
To assist with gene discovery, several precision phenotyping protocols based on remote sensing can be applied, including spectral reflectance indices for a range of growth-related parameters (Montes et al., 2007) and infrared thermometry as mentioned earlier. Application of these principles in wheat has led to the identification of a number of QTLs that are associated with both drought and heat adaptation, which suggests some common genetic basis for adaptation to these two stresses (Pinto et al., 2010). The usefulness of such traits in selection requires the development of a comprehensive understanding of the genetic and environmental influences that determine their effect on yield and other performance characteristics (see Crossa et al., Chapter 14, this volume).
Molecular breeding is benefiting from the rapidly decreasing cost of genotyping, and points to a more pragmatic future in which phenotyping is again highly valued. Commercial breeding programmes (especially in maize and soybean) are now beginning to release germplasm that has been developed for yield through the application of marker-assisted recurrent selection (MARS) (Eathington et al., 2007). This approach relies on cheap abundant marker systems being applied to a large number of accurately phenotyped biparental populations, followed by rigorous statistical methods. Breeders either estimate QTLs using the types of methods described by Boer et al. (2007) or apply techniques such as genome-wide selection (GWS) to assign predictive values to every marker used in the analysis (e.g. Heffner et al., 2009). Favourable QTLs and/or markers are then used in several cycles of glasshouse selection to quickly assemble new inbred lines as complexes of useful genomic regions, although without direct knowledge of the genes or their mechanisms (i.e. fine mapping and gene discovery is not utilized at all in the breeding, although these may follow at a later date to locate genes for future use as diagnostic markers or in gene transformation). Heffner et al. (2009) have argued that the typical lack of success in breeding with QTLs means that genome-selection methods, where every marker has a value (positive or negative), will probably take over from QTL approaches. MARS molecular breeding methods accelerate the traditional phenotyping approach of breeding and are being deployed in many breeding programmes as an adjunct to phenotypic methods. Readers are referred to the chapter by Whitford et al. (Chapter 12, this volume) for further examples of applications of biotechnology in breeding.
Was this article helpful?