AB-QTL was proposed by Tanksley and Nelson (1996) as a novel plant breeding scheme designed to integrate the processes of QTL discovery and variety development. Useful QTL alleles from unadapted germplasm (e.g., landraces, wild species) are identified while simultaneously being transferred to elite lines, thus streamlining the QTL identification and utilization pipeline while concurrently broadening the genetic diversity of the cultivated germplasm.
In tomato, the first AB-QTL mapping project started in 1995 as a molecular marker-assisted breeding experiment applied to processing tomatoes, in collaboration with several international processing companies. The main objectives of the project were to:
(i) examine whether wild germplasm could be used as a source of new, agronomically beneficial QTLs,
(ii) test whether the proposed marker-assisted breeding scheme would efficiently maximize QTL discovery, and (iii) develop new lines that would outperform elite commercial varieties for soluble solids content while continuing to maintain or improve other important characteristics for the processing industry, including yield, viscosity, firmness, color and fruit size. For the recurrent parent a commercially acceptable publicly available open-pollinated processing variety (cv. E6203) was agreed upon by all partners, and only wild tomato species were used as donor parents.
Five AB-QTL studies were conducted in tomato involving crosses with five wild Solanum tomato species: S. pimpinellifolium (LA1589) (Grandillo and Tanksley 1996b; Tanksley et al. 1996), S. peruvianum (LA1706) (Fulton et al. 1997), S. habrochaites (LA1777) (Bernac-chi et al. 1998a, b), S. neorickii (LA2133) (Fulton et al. 2000), and S. pennellii (LA1657) (Frary et al. 2004).
These five species were chosen on the basis of genetic diversity and uniqueness, representing the broadest possible spectrum of wild species main tained in gene banks, with the hope that this would increase the frequency of new, previously undiscovered alleles. The use of a common recurrent parent allowed more direct cross-species comparisons. Of the five wild species tested, S. pimpinellifolium was the only red-fruited one and the most closely related to the cultivated tomato. At the other extreme was S. pe-ruvianum, one of the most distantly related and also containing the highest amount of novel DNA and variation when compared to S. lycopersicum (Rick 1986; Miller and Tanksley 1990). All of these wild species have been the source of many major disease resistance genes, but no effort had previously been made to exploit the variation they contain for the improvement of quantitative traits.
For four of the studies marker analysis was conducted on BC2 populations, and BC3 or BC2F1 families or both were used for phenotypic analysis. In the S. peruvianum study genotypic analysis was postponed until the BC3 generation and phenotypes were evaluated on the derived BC4 families. The number of markers used for the molecular analyses ranged from a minimum of 121 in the S. pimpinellifolium study to a maximum of 174 markers in the S. peruvianum study. However, in the S. peruvianum and S. pennellii populations approximately 30% of the scored markers were fixed for the S. lycopersicum allele (SL) and could not be used for QTL mapping. Many of the markers fixed for the SL alleles corresponded to the chromosomal regions for which marker-assisted selection was applied to remove the wild parent allele in the BC1 population (e.g., for the top of chromosome 1 as a result of selection at the self-incompatibility locus, S, to increase the fertility of the plant and, for the bottom of chromosome 6, as a result of selection at the self-pruning locus, Sp, to ensure that the plants would have a determinate growth habit which is essential for mechanical harvesting of processing tomatoes). For other regions of the genome the fixation for the SL allele maybe the result either of sterility in early crosses or the result of genetic drift caused by the small size of the BC1 populations used to develop the correspondent BC2 generations (in the case of the S. peruvianum and S. pennellii studies). For most of the AB-QTL populations analyzed, however, the percentage of markers fixed for the SL allele was lower than 10%.
To reiterate, the main objective of this project was to focus on improving soluble solids content while maintaining or improving other traits important to the processing industry, including yield, viscosity, firmness, color, and fruit size. Therefore, each
AB-QTL population was evaluated for a wide array of traits, ranging from a minimum of 19 in the S. habrochaites study to 35 in the S. peruvianum ABpopulation. Inallcasestotal yield, redyieldandmajor fruit quality characteristics, including soluble solids content or Brix°, fruit color, viscosity, firmness and fruit pH, were measured. Due to frequent negative correlations between Brix° and yield, the derived parameter Brix° x yield is considered to be a more comprehensive biological and agricultural estimate for the productivity of processing tomatoes than yield alone (Eshed and Zamir 1995; Tanksley et al. 1996).
Agronomically favorable QTL alleles were identified in all five of these interspecific AB-QTL populations, for nearly half of the evaluated traits. For certain traits, such as pH and acidity, changes were not categorized as positive or negative, but rather were required to be kept within an acceptable range for processing purposes. But of the QTLs identified for which allelic effects could be deemed as favorable (+) or unfavorable (-), 20 of the 78 identified QTLs (26%), corresponding to 11 out of 23 traits (48%) had trait-improving alleles contributed from S. pennellii (Frary et al. 2004). Approximately the same percentage of traits with favorable wild-alleles were obtained with S. habrochaites (47%, Bernacchi et al. 1998b), while even higher percentages were observed for L. pimpinellifolium (88%, Tanksley et al. 1996), S. peruvianum (73%, Fulton et al. 1997) and S. neorickii (69%, Fulton et al. 2000).
It was possible for ten of the traits for which the effects could be defined either favorable or unfavorable (Brix°, viscosity, fruit color, stem retention, puffiness, firmness, total yield, fruit weight, red yield and Brix° x redyield) to be measured in all five AB-QTL studies; two other traits, cover, and maturity were measured in four of the five studies. Desirable alleles contributed from the wild species were identified not only for traits for which the wild parent showed a superior phenotype (e.g., Brix°, puffiness, cover) but also for those traits for which the wild phenotype was agro-nomically inferior (e.g., total yield, fruit weight, fruit color). The average percentage of favorable wild QTL alleles estimated across the five wild species ranged between a minimum of 3% for red yield to a maximum of 88% for Brix° (Grandillo and Tanksley 2005). Over the ten traits common to all five studies, the highest percentage of positive QTLs was identified in the S. pimpinellifolium study (44%), followed by the S. peruvianum (41%), S. neorickii (28%), S. pennellii (27%) and S. habrochaites (15%) studies (Grandillo and Tanksley 2005). Interestingly, those QTLs which were the most useful in terms of (i) having the most positive agronomic benefits, (ii) confirmation in NILs, and (iii) successful transfer into commercial varieties, are those which came from the most distantly related wild species (SD Tanksley, pers. comm.).
The AB-QTL strategy was also used in the first four AB populations to identify QTLs for biochemical properties that may contribute to the flavor of processed tomatoes, such as sugars and organic acids (Fulton et al. 2002a) (see Sect. 1.15.1). Flavor, a difficult characteristic to define and measure but very important to the industry, was assessed by a taste panel, as well as a derived trait, sugars/glutamic acid, which has been shown to be highly correlated with improved flavor (Bucheli et al. 1999). A total of 222 significant QTLs were identified for 15 evaluated traits. Studies, such as this, are important not only for the obvious purpose of developing new varieties with improved flavor but also for further analyses aimed at improving our knowledge of the biochemical pathways of fruit development.
Overall, these results show that in tomato, on average, for approximately 30% of QTLs for any given trait, an allele superior (from an agricultural viewpoint) to the cultivated parental allele can be identified in the wild species. Furthermore, after having sampled several wild species genomes the rate of discovery of "new" QTL alleles is still approximately 50% (Fulton et al. 2000; Frary et al. 2004), suggesting that continued sampling of exotic germplasm has not yet reached a steady state and there are still new and useful QTL alleles yet to be discovered.
Was this article helpful?