Acknowledgements
4.7 Handling of Hybrid Populations and Selection
Criteria for initial selection
Primary selection from the hybrid progeny is based on: (i) precocity; (ii) fruit size and shape; (iii) skin colour; (iv) fruit characteristics (high pulp to stone ratio and freedom from fi bre and physiological disorders); and (v) fruit qual- ity. Following this preliminary evaluation, selected hybrids are retained for
Breeding and Genetics 81
further screening. It is important to graft the hybrids onto proper rootstocks as early as possible, as grafted plants are precocious. At least ten grafted plants of each selected hybrid are used in the fi nal selection, which is based on yield, regularity in bearing and response to diseases and pests, in addition to other desirable fruit characters. At least 3 consecutive years’ performance data should be collected before deciding on their suitability for release as new cultivars.
Pre-selection
Trees have a long juvenile phase, and the development of pre-selection meth- ods is important for discarding inferior seedlings at a very early stage, obvi- ating the need for maintaining a large number of seedlings for long periods.
This can save time, land and labour. Leaf fl avour has been reported to be directly correlated with fruit fl avour (Majumder et al., 1972; Whiley et al., 1993). Emergence of new growth fl ushes, simultaneously with fruiting or immediately after harvest, is indicative of regular bearing (Sharma et al., 1972). A higher phloem to xylem ratio, associated with dwarfi ng, has been used effectively as a pre-selection criterion. Genotypes in which the ratio exceeds 1.0 are least vigorous, those with a ratio between 0.6 and 1.0 are of medium vigour and those with a ratio of less than 0.6 are most vigorous (Kurian and Iyer, 1992). In addition, higher levels of phenolics in the apical bud is associated with reduced vigour and dwarfi ng (Iyer, 1991). Although Majumder et al. (1981) indicated that low stomatal density is an indicator of dwarfness this has not been confi rmed by other workers (Iyer, 1991). Regular bearing mango cultivars have low polyphenol oxidase (PPO) activity (cate- cholase and cresolase) compared to alternate bearers (Sharma, 2003). Sharma et al. (2000) observed that a strong positive correlation existed between the incidence of fl oral malformation and both enzyme activity (catecholase and cresolase) and phenolic content and speculated that PPO activity can be used as a biochemical index for screening mango germplasm against malforma- tion disease.
Potential for marker assisted selection (MAS)
More than 65 microsatellite markers have been developed for mango and these are easily used to verify parentage using a software package such as cervus (Marshall et al., 1998). When caging trees or using the polycross mat- ing design it is possible to identify the male parent from a set of potential male parents. This has been useful in cacao breeding where mistakes in pol- lination have lead to the estimation of unreliable breeding values for parental clones. The development of linkage maps and identifi cation of quantitative trait loci (QTL) for productivity and quality traits has led to a very successful MAS in cacao (Schnell et al., 2007). This could serve as a model for future mango breeding and selection efforts.
C.P.A. Iyer and R.J. Schnell 82
Molecular markers
Molecular markers can be used for estimating genetic relationships among clones, for parentage analysis and for the development of a saturated linkage map. Isozymes were the fi rst markers to be used for fi ngerprinting mango cultivars, to determine self- versus cross-pollination and to estimate genetic relationships (Degani et al., 1990; Knight and Schnell, 1994). RAPD markers were also used to fi ngerprint cultivars and estimate genetic relationships in mango (Schnell et al., 1995). A group of ‘Haden’ seedlings and a random group of seedlings were evaluated using 11 RAPD primers. This study sup- ported the ‘Haden’ parentage of ‘Eldon’, ‘Lippens’, ‘Tommy Atkins’ and
‘Zill’; however, the parentage of ‘Glenn’ and ‘Osteen’ was questioned. Adato et al. (1995) used DNA fi ngerprinting (DFP) to evaluate genetic relationships between 26 mango cultivars and 14 rootstocks. They provided a pedigree that further confi rmed the relationship between many of the ‘Haden’ seed- lings. Lopez-Valenzuela et al. (1997) used RAPD markers to estimate genetic diversity among 15 rootstock cultivars using 13 markers, and identifi ed a specifi c RAPD band associated only with the polyembryonic types. Eiad- thong et al. (1999) utilized anchored simple sequence repeat markers to anal- yse 22 mango cultivars; they were able to distinguish genotypes, but were unable to fi nd markers unique to either monoembryonic or polyembryonic types, or for the Thai cultivars selected for green harvest (crispy mango) from the cultivars selected for ripe fruit production. Kashkush et al. (2001) utilized amplifi ed fragment length polymorphisms (AFLP) to estimate genetic rela- tionships between 16 cultivars and seven rootstock cultivars. They also anal- ysed 29 progeny from a cross of ‘Tommy-Atkins’ and ‘Keitt’ and produced a crude linkage map that identifi ed 13 of the 20 linkage groups.
Viruel et al. (2005) developed the fi rst reported set of 16 microsatellite markers for mango, of which 14 produced the expected one or two amplifi - cation products per genotype. These 14 microsatellites were used to evaluate 28 mango genotypes that included 14 Florida cultivars. Discrimination of all 28 genotypes was possible and the average number of alleles per locus was 5.3. Previously known pedigree information for the ‘Haden’ family of man- goes was confi rmed and was in agreement with previously published RAPD and DFP analyses (Adato et al., 1995; Schnell et al., 1995) with one exception.
Viruel’s clone of ‘Zill’ was not resolved as a seedling of ‘Haden’. Schnell et al.
(2005) developed a second set of 15 microsatellite markers and analysed 59 Florida cultivars and four related species. Two of the microsatellites were monomorphic among the Florida cultivars; the other 13 had an average num- ber of alleles per locus of 4.2 with polymorphism information content (PIC) values varying from 0.21 to 0.63.
Schnell et al. (2006) used 25 microsatellite loci to estimate genetic diversity among 203 unique mangoes (M. indica L.), two M. griffi thii Hook. f.
and three M. odorata Griff. accessions maintained at the National Germplasm Repository (NGR) and by Fairchild Tropical Botanic Garden (FTBG) in Miami, Florida. The 25 microsatellite loci had an average of 6.96 alleles per locus and an average PIC value of 0.552 for the M. indica population. The total
Breeding and Genetics 83
propagation error in the collection (i.e. plants that had been incorrectly labelled or grafted) was estimated to be 6.13%. When compared by origin, the Florida cultivars were more closely related to Indian than to South-east Asian cultivars. Unbiased gene diversity (Hnb) of 0.600 and 0.582 was found for Indian and South-east Asian cultivars, respectively, and both were higher than Hnb among Florida cultivars (0.538). When compared by horticultural type, Hnb was higher among the polyembryonic types (0.596) than in the monoembryonic types (0.571).
To date 63 microsatellite markers have been developed for mango (Duval et al., 2005; Honsho et al., 2005; Schnell et al., 2005; Viruel et al., 2005).
This number is more than adequate for genetic diversity studies and for par- entage analysis as has been demonstrated by Schnell et al. (2006); however, it is not enough to develop a saturated linkage map for the 20 linkage groups of mango. Developing an additional 200 microsatellite or single nucleotide polymorphic markers is a major objective of the USDA Agriculture Research Service (ARS) programme in Miami over the next 2 years. Three experimen- tal populations have been developed and planted in the fi eld as mapping populations. The fi rst population is an F2 population derived from self-polli- nation of ‘Tommy Atkins’ consisting of 168 seedlings that was planted in the fi eld in 1995. The second population is an F2 population derived from self- pollination of ‘Haden’. A total of 224 seedlings from a single isolated ‘Haden’
tree have been in the fi eld for 3 years. Phenotypic data collection is in prog- ress for both of these populations. The development of a saturated linkage map and the identifi cation of QTL for important traits are objectives for the USDA-ARS programme in Miami for the next 5 years.