Marker-assisted Selection

MAS can increase the effi ciency of citrus breeding, and may speed release of new cultivars. Essentially, the procedure involves identifi cation of a marker that is polymorphic in the target population and closely linked to a gene that is segregating and the target of selection. In such situa- tions, selection for the marker allows the breeder to identify seedlings that carry the trait gene with a low error rate. Only seedlings with desirable genotypes are car- ried forward for additional testing. Conditions in which MAS is more effi cient that conventional breeding have been dis- cussed by a number of authors (Luby and Shaw, 2001). Traits that are costly to score by phenotype, such as traits that require fruiting trees, or those that have low heri- tability and can be accurately scored only with a progeny test (percentage of nucellar

seedlings in a rootstock) are likely to be good candidates for MAS.

Markers must be informative about the trait, i.e. the marker genotype must predict the progeny genotype. A dominant marker

(M) that is in repulsion phase with a domi- nant trait gene (T) is not effi cient in an F2

confi guration: Mt / mT ´ Mt / mT will give

25% Mt / Mt, 50% Mt / mT and 25% mT / mT in the progeny if there is no recombination between the marker and the trait gene. All mm progeny will be TT, but 75% of the population will be discarded, including 50% of the T -carrying progeny. If the marker and trait gene are in coupling phase, MT / mt ´ MT / mt, the progeny will contain 25% MT / MT, 50% MT / mt and 25% mt / mt; all progeny with the M phenotype will carry the T gene and therefore selection will only require discarding 25% of the popula- tion, and no progeny with the desirable allele T will be discarded. Since it is not easy to generate very large populations in citrus, discarding progeny with desirable genes reduces the overall effi ciency of the program. Co-dominant, multiallelic mark- ers are more informative in general, but still can be ineffi cient in certain crosses.

A common situation where MAS is useful is in transferring a gene from one species to another. It is easy to fi nd inform- ative markers in such crosses.

The breeder should always consider the economic effi ciency of MAS before implementing it. Such a program may require identifi cation of markers, and this

may be costly and take at least 1–2 years. The cost of DNA extraction and marker analysis should also be considered. If the same degree of improvement can be achieved by phenotypic or family selection, then MAS is not justifi ed (Luby and Shaw, 2001).

As markers for more traits become available, the costs of DNA extraction and marker testing will effectively decrease because they will be spread over multiple traits. Citrus biology limits what can be achieved with these methods. If we start with a population of 1000 hybrids, which is larger than usually obtained in citrus breed- ing, and impose MAS for three independent traits that segregate 1: 1, we will have 1000 (1–0.53) = 125 selected hybrids to evaluate in the fi eld for other traits. If the target traits segregated 3: 1 and the recessive was selected, then we would expect only 1000 (1–0.753) = 15.6 progeny to have the desired marker genotype. This is too few to have much chance of fi nding hybrids with addi- tional desired traits. Therefore, it is not realistic to expect that we can select for a large number of traits by MAS in a single generation. Similarly, it may often be diffi - cult to select for a complex QTL and still have suffi cient plants to select for other traits. A two-generation strategy involving MAS for different sets of traits in two pop- ulations, then intercrossing selections from these populations, might be more effective, but the long generation time of citrus reduces the value of this approach.