In hybrid maize breeding programs, efficiency of procedures to identify inbreds used to develop outstanding single crosses strongly affects the success of the program (Hallauer and Lopez-Perez, 1979). The best hybrid combinations can be identified using information from diallel or topcrosses to testers (Gonzalez et al., 1997; Terron et al., 1997), pedigree information, morphological traits (which may be greatly influenced by the environment), and molecular markers that detect variation at the DNA sequence level (Smith and Smith, 1992).
With large number of inbreds, diallel crosses are not possible in practice due to a prohibitive number of crosses. Hallauer et al. (1988) suggested relative performance of inbred lines in testcrosses with divergent testers to determine the heterotic patterns among lines, as has been carried out at the International Maize and Wheat Improvement Center (CIMMYT) maize breeding program (Vasal et al., 1992a, 1992b). Grouping of Iranian inbred maize lines into heterotic groups is not yet as well defined as in other temperate maize programs. Information on pedigree would aid in grouping, but is often lacking for Iranian breeding materials. DNA markers could help to establish initial heterotic groups based on GD, allowing breeders to avoid crosses among related inbreds. Choukan et al. (2006) and Choukan and Warburton, (2005) used this approach in grouping 56 late to medium and 36 early maturing Iranian maize inbred lines using 46 and 36 SSR markers, respectively.
Several previous studies have demonstrated the possibility of using molecular markers to assign maize germplasm to heterotic groups or form new groups (Lee et al., 1989; Livini et al., 1992; Dubreuil et al., 1996). Nevertheless, the correlation between GD based on molecular markers and hybrid performance has not been consistent. Low to high correlations between molecular divergence and hybrid performance have been reported (Dhillon et al., 1993; Ajmon-Marsan et al., 1998; Smith et al., 1990), indicating that GD may or may not be useful to predict hybrid performance. The objective of this study was to evaluate the correlation between the SSR markers based on GD and hybrid performance, MPH and SCA in crosses between Iranian maize inbred lines, in order to determine the possible use of marker based GD in Iranian maize improvement programs.
MATERIALS AND METHODS
Plant materials and field experiments: A total of 28 maize inbred lines currently used in Iranian maize breeding programs and five tester lines were used for this study (Table 1). Each of the 28 inbred lines was crossed with five tester lines (Mo17, B73, K74/1, K1264/1 and KL17/2-5) to produce a total of 140 hybrid combinations. These testers have been used in hybrid production in Iran, and are the parents of commercial single cross hybrids, except KL17/2-5. The 140 line x tester crosses and four promising hybrids were evaluated at two locations for grain yield: Karaj in 2003, 2004 and 2005 (35º49´ N, 51º00´ E), and Gorgan in 2004 (35º50´ N, 54º22´ E). Plot data were used to adjust the grain moisture to 14%, calculating the grain yield based on t/ha.
SSR analysis: DNA extraction and SSR assays were previously described in detail by Choukan et al. (2006). Briefly, inbred lines were characterized with forty-six SSR primers, which were chosen based on repeat unit and bin location to provide uniform coverage of the entire maize genome from the MaizeGDB database (http://www.maizegdb.org/ssr.pho). The SSRs were multiplexed for maximum efficiency. DNA fragments were separated using acrylamide gels run on an ABI377 automatic DNA sequencer. Fragment sizes were calculated with GeneScan3.1 (Perkin Elmer/Applied Biosystems) using the Local Southern sizing method. The fragments were assigned to alleles using the category function of Genotype 2.1 (Perkin Elmer/Applied Biosystems), and exported as an excel file recording allele identity for each individual.
Statistical analysis: Line x tester analysis was carried out using the adjusted means of grain yield (t/ha) based on the method described by Kempthorne (1957). Specific combining ability SCA effects for grain yield were calculated according to the line x tester model. The MPH of each cross was calculated as the difference between F1 means and the respective midparent mean across all environments. GDs between all possible pairs of testers and lines, based on the SSR data, were estimated from the Rogers’ distance (Rogers 1972):
where RDij= Rogers’ distance between individuals, pij and qij are the frequency of ith allele at the jth locus in parents x and y, respectively, while aj is the number of alleles at the jth locus, and m is the number of the loci examined. Correlation coefficients (r) were calculated between GD’s and F1 performance and SCA of grain yield for the 140 line x tester testcrosses, as well as within each of five tester lines.
Hybrid performance: Grain yields for 140 line x tester combinations are presented in Table 2. Mean grain yield of testcrosses averaged over environments varied from 3.940 t/ha to 11.200 t/ha. The highest grain yield was seen in the cross K3615/2 x Mo17. Six of the ten highest yielding hybrids correspond to crosses with Mo17, two with tester B73, and two with K74/1 (including one commercial hybrid). Among the ten lowest yielding hybrids, three correspond to KL17/2-5, five to K74/1 and two to K1264/1 crosses. Most of the inbred lines which presented the highest grain yield had Mo17 or B73 tester lines as a parent. This was probably due to previous selection for combining ability of inbred lines in the earlier stages of line development using Mo17 as a tester line, and possibly the higher yield per se of B73. Estimates of positive SCA effects (Table 3) were greatest for Mo17 sister lines [lancuster suree crop (LSC) related lines], K18, K19 and K19/1 with the K74/1 tester line. Crosses among Late Synthetic related inbred lines had in several instances a relatively high yield and significantly positive SCA estimates.
Genetic distances among inbred lines and testers: The average GD coefficient based on the polymorphic SSRs among inbred lines with testers was 0.576, ranging from 0.231 to 0.740 with significant (p<0.01) differences between GD estimates (Table 2). The highest GD (0.740) was detected between K2818 and K74/1, which are of unknown origin/pedigree. The lowest GD (0.231) was observed between the sister lines KL17/2-5 and KL17/2-3. In general, lower distances were observed between sisters lines (KL17/2-5 with Kl17/2-3 and KL17/2-2, Mo17 with K19, K19/1 and K18). However, two notable exceptions to this observation were seen in crosses between Mo17, an LSC derived inbred, and the two LSC sister lines K18 and K19 (which showed GD estimates of 0.447 and 0.721). The latter, in particular, was unexpectedly high, and may indicate a mis-reported pedigree. K2818 was the most distant line from all the testers, except B73. The GDs between inbred lines and testers spanned a great range, indicating the presence of a broad genetic base when comparing inbred lines to testers.
Relationship of genetic distance with hybrid performance: The correlation coefficients between SSR based GD and grain yield, SCA and MPH for the entire set of 140 line x tester combinations were moderate but significant (r= 0.27**, 0.39** and 0.28, respectively), showing a positive association between GD and grain yield and MPH (Table 4). Correlation between GD and grain yield, SCA and MPH across testers varied from 0.03 to 0.54, 0.10 to 0.61 and 0.09 to 0.61, respectively. Grain yield, SCA and MPH correlated significantly with GD in Mo17 and KL17/2-5 testcrosses to 28 inbred lines, as well as K74/1 testcrosses only for SCA. In general, correlations between GD and yield or heterosis were low in B73 crosses and highest in Mo17 crosses. Although GD between parents was significantly correlated with hybrid performance, it did not consistently identify the best crosses. This is similar to results already published (Godshalk et al., 1990; Melchinger et al., 1990; Ajmon-Marsan et al., 1998).
The difference in predictive ability of markers in different genetic backgrounds has been reported earlier (Melchinger, 1993). Bernardo (1992) stated that molecular marker heterozygosity would be most valuable for predicting hybrid performance in crop species under conditions such as strong dominance effects, and high trait heritability. Charcosset et al. (1991) anticipated the need for linkage disequilibrium between marker loci and quantitative trait loci (QTL) for grain yield in order to associate marker heterozygosity with heterosis. In other studies, correlations of GD based on marker heterozygosity with performance and heterosis differed from one tester to another and depended on the genetic background of the germplasm (Lee et al., 1989; Boppemmaier et al., 1993; Zhang et al., 1996).
In the current study, although the magnitude of the correlation is very low to be predictive, a significant conclusion can be made favoring the use of SSR based GD estimates in some Iranian testers. Because many productive hybrids have been made in Iran with Mo17, it will continue to be a major tester in hybrid maize programs. It would be beneficial to make all hybrids with Mo17 (and possibly KL17/2-5) only after fingerprinting to determine highest the GD, allowing a significant reduction in number of crosses tested, (and thus a reduction in costs and labor). The GDs based on SSR data for crosses to B73 and other testers with low correlations may also predict poor hybrid performance, which can then these can be avoided, thus reducing the number of crosses required under field evaluation.