Most of the traits of primary importance in sugarcane are quantitative in nature and not highly heritable. Furthermore, when selecting for a primary trait such as cane yield, the selection is being done for several secondary traits that influence the primary trait rather than for the primary trait itself. The genetic and phenotypic associations between those secondary traits are of practical interest since selection for one trait will have a simultaneous effect on the related traits. Understanding that improvement of one trait may cause improvement or deterioration in associated trait(s) serves to highlight the need for simultaneous consideration of all traits that are important in a crop species (Baker, 1986). Because most breeding programs are concerned with simultaneous improvement of several traits, the selection index has become the best alternative, provided that reliable estimates of genetic and phenotypic variances and covariances are available and appropriate economic weights of each trait can be determined (Hallauer and Miranda, 1981; Milligan et al., 2003).
Since the early nineties the Louisiana Sugarcane Variety Development Program (LSVDP) has been using progeny appraisal data to identify sugarcane families with most potential to produce superior individuals. The LSVDP uses an initial 50% family selection and subsequent 20% individual selection within those families.
De Sousa-Vieira and Milligan (1999) examined intra-row plant spacing as a source of variation affecting the efficacy of progeny testing and family selection at the first selection stage of the LSVDP. They showed that intra-row plant spacing affects the variance and reliability of a trait. The work reported in this paper aimed to examine the interrelationships of cane yield components and the effect of intra-row plant spacing on the estimation of a selection index for selecting sugarcane families within the LSVDP.
Materials and Methods
]]> The population and experimental methodology used herein have been described by De Sousa-Vieira and Milligan (1999, 2005), and was planned to simulate the initial family selection stage in a sugarcane breeding program. In brief, 25 bi-parental families were randomly selected. The progeny of those families were then transplanted to the field at two locations: the United States Department of Agriculture (USDA) Ardoyne Farm near Chacahoula, and the St. Gabriel Research Station, both in Louisiana State, USA. Individual plants from each cross were planted in a randomized complete block design using two blocks with a split plot treatment arrangement where the main plots were intra-row plant spacings of 41cm (standard at LSVDP) and 82cm on rows 1.8m apart. Subplots were families. Each subplot consisted of two rows with up to 16 seedlings in each row. Planting and data collection were done during two consecutive years at the same locations. Data were collected in first ratoon cane.Millable stalk number per plant, stalk length, and mid-stalk diameter were recorded. Stalk length was measured from the stalk base to the first visible dewlap (leaf collar) of two random stalks for each plant. The same two stalks were measured for mid-stalk internode diameter using a caliper. Stalk weight was estimated (Miller and James, 1974; Gravois et al., 1991; Chang and Milligan, 1992) as the volume of the stalk assuming a perfect cylinder with specific gravity of one:
Stalk weight = dpr2L
where d: density (1.0gm·cm-3), r: stalk radius (cm), and L: stalk length (cm). Plant weight was estimated as stalk weight times stalk number per plant.
Selection indices
To construct an optimum selection index, one needs the genotypic and phenotypic variances and covariances, and the relative economic values or weights of all traits (Kang, 1994). Symbolically, a selection index (I) takes the form
I= b1X1 + b2X2 +…..+ bnXn
where Xi: observed phenotypic value of the ith trait, and b: weight assigned to that trait in the selection index.
Selection indices were estimated as
b= P-1 G a
]]> where b: vector of index coefficients, P-1: inverse of the phenotypic variance-covariance matrix, G: genotypic variance-covariance matrix, and a: vector of relative economic values or weights (Smith, 1936; Brim et al., 1959; Baker, 1986; Kang, 1994).Using family mean data, covariance components between all possible pairs of traits were estimated. Mean product expectations are analogous to the mean squares expectations for the analysis of variance. Thus, estimates of phenotypic and genetic covariance components were derived in the same fashion as for variance components by using product moment method.
Genetic and phenotypic correlations on a family mean basis, between the traits, were computed as
rij = sij / sisj
where sij : genetic or phenotypic covariance between traits i and j, si: genetic or phenotypic standard deviation for trait i, and sj : genetic or phenotypic standard deviation for trait j.
Twenty-six indices were constructed for each intra-row plant spacing using different trait combinations. Selection indices were constructed according to Smith (1936), as illustrated by Brim et al. (1959). Plant weight was taken as the final product; therefore, a relative economic weight of one was assigned to plant weight and zero to all other traits.
Expected genetic advance (EGA) from selection was estimated (Brim et al., 1959; Miller et al., 1978; Kang, 1994) as
where 0.795: value corresponding to a selection intensity (k) of 50% assuming 200 families (standard procedure), bi: phenotypic weights and Gi: genotypic variance-covariance. An index of relative efficiency (IRE) was estimated based on the assumption that the efficiency of EGA, when selection is based on plant weight alone, is 100. The index of relative efficiency only compares the predicted gain of an index to selection for plant weight alone within the same plant spacing.
Results and Discussion
]]> Twenty-six different indices were constructed for each of the two intra-row spacings, and the estimated index relative efficiency (IRE) for each individual index was calculated (Table I). In each case, it was assumed that the top 50% of 200 sugarcane families would be selected, as is currently practiced in the LSVDP.
Expected genetic advance for plant weight was 0.502kg/plant for narrow spaced plants and 1.138kg/plant for wide spaced plants when selection was based on plant weight alone. This value is used as a basis for comparison of the relative efficiency of the indices (IRE). The IRE compared only within row spacing.
Expected genetic advance values for plant weight were greater in the wide-spaced indices than in the narrow-spaced ones. Ranking of the indices did not appear to be greatly affected by spacing; the best indices in wide-spaced plants were also the best indices in narrow-spaced plants. Even though spacing may not affect indices ranks, it might affect both family performance and individual selection effectiveness within a family.
The highest gain in expected genetic advance, both in narrow (121.91%), and wide (106.15%) intra-row plant spacings, was obtained when selection was based on Index 1, which incorporated information of all five traits. Indices 4 and 6 had the same expected genetic advance as Index 1 did for widely spaced families and was comparable to Index 1 for narrow spaced families. Index 4 did not include stalk diameter, and Index 6 did not include plant weight. The relatively high expected genetic advance for these two indices was probably due to the presence of intermediate variables that contain some information about the traits that were excluded. In Index 4, stalk weight was an intermediate variable which contained information about stalk diameter, and in Index 6 the information about plant weight is contained in the genotypic covariances of plant weight and all the other traits.
The range of index relative efficiency for the twenty-six indices was between 81.32 and 121.76% for narrow intra-row plant spacing and between 94.75 and 106.15% for wide intra-row plant spacing. As a general rule, expected genetic advance increased over selection for plant weight alone, when selection indices were used. Just two indices for wide planted and one index for narrow planted families had a lower expected genetic advance than selecting for plant weight alone. Expected genetic advance for several selection indices was comparable to that of plant weight alone. Indices that included number of stalks per plant, stalk diameter, and/or stalk length resulted in the highest expected genetic advance.
Irrespective of plant spacing, Indices 21 and 26 gave the highest expected gain from selection when just two traits were included in an index. The use of any of these indices would probably select for taller, larger diameter, and heavier stalks, which is good, but it may not take into consideration the number of stalks per plant. In the same fashion, Index 15 was comparable with Index 1 without the inclusion of stalk diameter and plant weight. This index will probably give favorable results since it took into account, directly or indirectly all five traits.
The results suggest that a combination of traits, such as the ones indicated in this study will lead to a higher efficiency in family selection, which is definitely superior to selection for plant weight alone. The right combination would be the one from which sugarcane breeders at LSVDP could get the maximum expected gain from selection with minimum cost, resources and efforts. The study also suggests that selection using indices obtained from wide-spaced sugarcane families would be more effective than selection using indices obtained from narrow-spaced families where competition between a plant and surrounding plants might influence a plant character.
REFERENCES
1. Baker RJ (1986) Selection Indices in Plant Breeding. CRC Press. Boca Raton, FL, USA. 218 pp. [ Links ]
2. Brim CA, Johnson HW, Cockerham CC (1959) Multiple selection criteria in soybeans. Agron. J. 51: 42-46. [ Links ]
3. Chang YS, Milligan SB (1992) Estimating the potential of sugarcane families to produce elite genotypes using univariate cross prediction methods. Theor. Appl. Genet. 84: 662-671. [ Links ]
4. De Sousa-Vieira O, Milligan SB (1999) Intra-row plant spacing and family x environment interaction effects on sugarcane family evaluation. Crop Sci. 39:358-364. [ Links ]
5. De Sousa-Vieira O, Milligan SB (2005) Interrelationships of cane yield components and their utility in sugarcane family selection: path coefficient analysis. Interciencia 30: 93-96. [ Links ]
6. Gravois KA, Milligan SB, Martin FA (1991) Indirect selection for increased sucrose yield in early sugarcane testing stages. Field Crops Res. 26: 67-73. [ Links ]
7. Hallauer AR, Miranda JB (1981) Quantitative Genetics in Maize Breeding. 1st ed. The Iowa State University Press. Ames, IO, USA. 468 pp. [ Links ]
8. Kang MS (1994) Applied Quantitative Genetics. Louisiana State University, Baton Rouge, LA, USA. 157 pp. [ Links ]
9. Miller JD, James NI (1974) The influence of stalk density on cane yield. Proc. Int. Soc. Sugar Cane Technol. 15: 177-184. [ Links ]
10. Miller JD, James NI, Lyrene PM (1978) Selection indices in sugarcane. Crop Sci. 18: 369-372. [ Links ]
11. Milligan SB, Balzarini M, White WH (2003) Broad-Sense heritabilities, genetic correlations, and selection indices for sugarcane borer resistance and their relation to yield loss. Crop Sci. 43: 1729-1735. [ Links ]
12. Smith HF (1936) A discriminant function for plant selection. Ann. Eugenics 7: 240-250. [ Links ] ]]>