QTL Analysis

QTL analysis refers to a genetic analysis of a trait in which genes that infl uence the trait are located on a genetic map. The quantitative value of a trait is normally con-

sidered to be infl uenced by one to many genes and the environment. Analytical approaches are now quite sophisticated, but all depend upon particular marker alle- les being associated with particular trait alleles (linkage disequilibrium). If we con- sider a genetic linkage map, each progeny individual has a particular genotype for each locus. If a marker locus is near a gene that infl uences the trait, then there will have been little recombination between the marker alleles and the trait alleles. Various statistical tests are applied to identify those marker loci or positions for which values of the trait differ signifi cantly. These are puta- tive QTLs. Many different computer pack- ages are available to conduct QTL analysis, but not all are easy to use for citrus crosses because some require populations (such as F2, backcross or recombinant inbred lines) derived from homozygous parents. QTL

CARTOGRAPHER (Basten et al., 1998) and MAPQTL (Van Ooijen and Maliepaard, 1996) have been used in previous QTL mapping studies. See Liu (1998) for a through dis- cussion of QTL analysis.

The conventional approach to QTL analysis is to study a measurable trait on progeny (replicated if possible) in a segre- gating population for which a linkage map has been developed. Marker and trait data are then jointly analysed to identify QTLs. However, citrus crosses are more complex than generally assumed by programs designed to analyse populations derived from inbred lines. Consider for example a ‘backcross’ between pummelo, as the recur- rent parent, and trifoliate orange. Both pummelo and trifoliate orange are some- what heterozygous, so their potential geno- types are A1A2 and A3A4, although for any

locus A1 may be identical to A2 and/or A3 to

A4. This means that the ‘backcross’ parents could be A1A2 ´ A1A3, A1A2 ´ A1A4, A1A2 ´ A2A3, A1A2 ´ A2A4, etc. instead of the simple A1A1 ´ A1A3 confi guration assumed by models developed for a true backcross.

Thus this design does not detect effects of all alleles in trifoliate orange, only the par- ticular alleles that are passed to the F1 parent. Furthermore, the effects of trifoliate

alleles are not cleanly separated from effects of pummelo alleles if the pummelo parent is heterozygous for a QTL.

An alternative approach, linkage dise- quilibrium mapping, is now being devel- oped; it depends only on natural linkage disequilibrium between traits and markers. These methods do not require a mapping population derived from specifi c parents. Instead, existing individuals (such as a variety collection) are measured for the trait and genotyped for previously mapped markers. In this approach, it is essential that marker genotypes refl ect the specifi c DNA sequence (haplotype) present at the locus because the marker information is used to infer that individuals of unknown ancestry but having the marker allele also share alleles for genes in the surrounding region. Statistically signifi cant associations between marker alleles and trait values are then identifi ed. The extent of linkage dise- quilibrium varies among plants depending on their history, mating system and other factors. If disequilibrium extends for large distances (hundreds of kilobases), then it is fairly easy to locate regions containing QTLs, but fi ne-scale mapping of the QTL is not possible. Conversely, if disequilibrium covers only short distances, fi ne mapping is easier if an initial QTL–marker association can be found. We know little about the extent of disequilibrium in citrus, but it is likely to vary greatly with the population. With a species group having sexual repro- duction, such as pummelo or mandarin, disequilibrium is likely to cover smaller regions than if interspecifi c hybrids such as orange are included. In general, current citrus maps are not suffi ciently dense to apply these methods, but this is likely to improve in the future. See Flint Garcia et al. (2003) for a review of linkage disequilib- rium mapping in plants.

Several QTL studies have been reported in citrus. Tozlu et al. (1999a, b) studied QTLs for many traits related to growth and ion accumulation under normal and saline conditions in 47 progeny from a backcross of pummelo and trifoliate orange. Despite the rather small population size,

they identifi ed 70 putative QTLs that infl u- enced various traits. Some QTLs infl uence growth under both normal and saline con- ditions, while others were specifi c to only one environment. In some cases, QTLs for ion accumulation localized to the same region as QTLs for growth under saline con- ditions.

A second example is provided by Garcia et al., (1999) who studied QTLs asso- ciated with nucellar embryony (apomixis) in 50 progeny of a cross of C. volkameriana

´ P. trifoliata. They found QTLs affecting polyembryony in both parents, indicating that genes infl uencing the trait segregate in both parents. No QTLs of major effect were found. However, the single-parent maps on which this analysis was based included only 45 and 38 markers, and it is not clear what portion of the genome is covered by these maps. Therefore, it is likely that only some of the heterozygous genes infl uencing nucellar embryony in these parents were discovered. Any QTL analysis is limited to discovery of QTLs in the portion of the genome covered by segregating markers, so map coverage is an important issue.