Population sizes for QTL studies
Choosing a population size is a compromise between what is theoretically desirable and what is feasible in practice. Theory and computer simulations argue for large population sizes (at least several hundred) in order to adequately sample the population, to identify QTL of both large and small effect, and to accurately estimate the size of QTL effects (Beavis, 1994, 1998). In practice, it is often difficult to evaluate more than 200 or 300 progeny, especially when multiple replications and environments are needed. For corn, 250 progeny is considered a reasonable compromise by many researchers.
The effect of population size on QTL detection was shown in studies by Bradshaw et al. (1995, 1998). They evaluated floral morphology traits in interspecific crosses of monkeyflower (Mimulusspp.), using populations of 93 and 465 F2 individuals. In the smaller population, 12 QTLs of relatively large effect were detected, while in the larger population, 11 of the same QTLs, plus an additional 16 QTLs, were revealed. The larger population allowed the detection of QTLs of smaller effect. For QTLs common to the two populations, the estimate of effect size was reduced in the larger population, supporting the notion that the magnitude of QTL effects is overestimated in small populations.
One strategy to reduce the work involved with large populations is to obtain marker genotypes only for progeny at the tails of the phenotypic trait distribution, e.g., the 20% highest and 20% lowest families. However, this will work only if a single trait is being analyzed, as each trait is likely to have a different distribution. (Paterson, 1998).
Uses of QTL Information in Genetics and Breeding After all the work involved in detecting QTLs, how are the results put to use? The major strategies for exploiting QTL information are described below.
- Selecting plants or families on the basis of their marker genotypes, a procedure known as marker-assisted selection. For example, if a QTL study indicates that favorable QTL alleles are associated with pattern A at markers 1 and 2, and with pattern B at marker 3, then plants having those marker patterns are selected. In theory, the technique should be useful for traits that are expensive or logistically difficult to measure directly or that need to be measured on mature plants. Marker data can be obtained on very young seedlings, resulting in a significant time savings in some cases. The cost effectiveness of marker-assisted selection is a key consideration that needs to be considered individually for every trait, population, and laboratory. In practice, there are only a few examples of successful use of marker-assisted selection based on QTL information (e.g., Ribaut et al., 2002; Young, 1999). For more information, see the lesson on Marker-Assisted Selection in the Library of Crop Technology.
- Understanding trait “architecture” , the number of genes, size of their effects, and type of gene action governing a trait. This information is potentially valuable to breeders in helping them decide upon appropriate breeding methods and populations sizes.
- Providing insights into genetic relationships among traits, the physiological mechanisms or biochemical pathways that contribute to a trait, and environmental effects on QTL expression. For example, if QTLs for different traits overlap at one or more genome locations, this suggests that the traits may be related genetically, either through pleiotropy, physiological trade-offs, or some other interaction. See Remington and Purugganan (2003) for more discussion.
- Identifying chromosome regions for isolating and cloning genes, sometimes known as map-based cloning. As mentioned previously, a QTL is initially detected in a rather broad section of a chromosome, far too large a region from which to isolate a gene. However, strategies have been developed to map the initial QTL at finer and finer resolution, until a relatively small DNA segment is identified. After sequencing the segment, it has been possible to determine which gene in the segment is responsible for the QTL effect. Two of the first examples of QTL cloning are described in Frary et al. (2000) and Yano et al. (2000).
Limitations of QTL Analysis
Although QTL studies provide interesting and useful information about the genetic factors that control trait inheritance, there are limitations to the technique that must be acknowledged. Among the most important limitations are the following (See also Kearsey, 2002; Remington and Purugganan, 2003; Ribaut et al., 2002).
- Information on QTL locations and effects is specific to a particular population and cannot be readily transferred to another population. This is because QTLs can be detected only when the loci influencing a trait are polymorphic, and each population is likely to be polymorphic at different sets of loci.
- QTL analysis (including population development, marker genotyping, trait evaluation, and statistical analysis) is expensive in time and materials. Therefore, it can only be used in a very limited number of populations.
- QTL analysis detects chromosome regions, not genes, that influence traits. Moreover, QTL locations have large confidence intervals, often greater than 30 cM. Such large regions encompass many candidate genes, so it is difficult to deduce which specific gene might be influencing the trait. Therefore, in most cases little information is provided on the mechanisms or pathways involved in trait expression.
- It is difficult to distinguish two closely linked QTLs, those that are less than 20 cM apart.
- When two QTLs are linked “in repulsion”, i.e., alleles at loci on the same parental chromosome have opposite effects on the trait, it may not be possible to detect the QTL, because the effects of the associated alleles cancel each other out.