F2 population size for resistance to root and stalk lodging in maize
--Stojkov, S, Rosulj, M, Stankovic G
Root and stalk lodging is one of the most important traits in commercial maize breeding. One of the main causes of increased root and stalk lodging is the presence of pathogens from the species Fusarium, so the maize breeder must pay special attention to resistance to these pathogens during the process of selection. An F2 population is a generation of maximum gene recombination. The influence of F2 population size has been studied in many papers for grain yield and other traits, but there is no literature on the optimal size of the F2 population necessary to develop hybrid combinations resistant to pathogens of maize root and stalk lodging. The objectives of this study were to estimate changes in genetic parameters with changes in population sizes and to obtain an F2 population size adequate for traits such as resistance to root and stalk lodging
The genetic material evaluated in the present study was F2 population S-5892 derived from a cross of two inbred lines, L588 and B92. Inbred line L588 is a dent type, derived from crossing B84 × Yugoslavian germplasm. FAO maturity group is 550. L588 has good general combining ability and is tolerant to root and stalk lodging. Public inbred line B92 is a semi-dent type. FAO maturity group is 700. B92 has excellent general combining ability, but is sensitive to root lodging in Yugoslavian conditions. The F1 generation L588 × B92 was self-pollinated in 1992 to obtain an F2 population. In 1993, 500 S0 plants from F2 population S-5892 were self-pollinated (plants were randomly selected) and crossed to six plants of inbred line L1325 as a tester. Inbred line L1325 is a flint type, derived from Argentinean germplasm. FAO maturity group is 450. L1325 shows high heterotic effects with both L588 and B92 inbred lines, and is tolerant to root lodging.
A total of 500 entries (half-sib progenies) were evaluated within 25 sets (Cochran and Cox, 1957). Each set consisted of 20 half-sib progenies completely randomised within each of three replications. The entries were grown at Zemun Polje, Velika Plana, Indjija and Becej in 1994, and Zemun Polje, Velika Plana and Becej in 1995. A plot consisted of 9.20m long hand-planted rows with 0.70m between rows. Over-planted plots were thinned to a uniform plant density of approximately 62.112 plants ha-1. All experiments were machine-cultivated and manually weeded as necessary for proper weed control. Data were collected at harvest for root and stalk lodging according to the following scale: 1-stalk broken bellow the tassel, 2-stalk broken above the ear, 3-stalk broken at the level of the ear, 4-stalk broken bellow the ear, 5-totally lodged stalk.
The analysis of data was based on plot means. Data were analysed by pooling over sets and combining across environments. From a basic population size of 500 half-sib progenies (25 sets with 20 half-sib progenies), 53,130 (25/5) populations with a size of 100 half-sib progenies, 3.268,760 (25/10) populations with a size of 200 half-sib progenies and 3.268,760 (25/15) populations with a size of 300 half-sib progenies were obtained by computer simulation. From the total number of combinations, 30 samples for each population sizes were randomly selected (except the 500 where only one sample is possible).
Comparisons of mean values between different population sizes were done by t or t′ test in relation to whether variances were homogenous or not (Steel and Torrie, Principles and procedures of statistics, McGraw-Hill Book Co., New York, 1960). Half-sib family means from each sample were used to construct the distribution histogram for each population. The Komogorov-Smirnov one-sample test was applied to test distribution. Values of D that are significant indicate non-normality of the distribution (Snedecor and Cochran, Statistical Methods, 8th ed., Iowa State University Press, Ames, 1989).
The analyses of individual populations pooled over sets and combined across environments were calculated to partition the within population variation for each population size into environments, sets, environments × sets interaction, replications, genotypes, genotypes × environments and error sources of variation. Genotypes × environments interaction mean squares were used to test significance of the genotype source of variation. Error mean squares were used to test significance of the genotypes × environments interaction source of variation.
Estimates of genetic variance components were calculated by equating observed mean squares with expected mean squares and solving the resulting system of equations. Heritability was estimated on a half-sib progeny mean basis within each population size. Genetic variance component and heritability estimates were declared significant if their values were 2 times greater than their standard errors (Falconer, Introduction to quantitative genetics, Longman, London and New York, 1989). Additive, dominance, and epistatic variance components are confounded in the genetic variance estimates for half-sib families; hence, heritability estimates should be considered an upper limit of the narrow-sense heritability (Lamkey and Hallauer, Maydica, 32:64–78, 1999).
Average estimates ranged from 1.712 for 200 HS progenies population size to 1.730 for 100 HS progenies population size. The maximum (1.826) and minimum (1.631) sample average estimate was found in the population size of 100 HS progenies (Table 1). There is no significant difference for average estimates between estimated population sizes (Table 2).
According to the Komogorov-Smirnov test, a lower value of parameter D indicates a greater normality of distribution. The values of parameter D became greater with the decrease of population size (500 HS, D = 0.0469; 300 HS, D = 0.0529; 200 HS, D = 0.0543; 100 HS D = 0.0469), but there is no evidence of statistically significant deviation from normality in any sample (Table 3).
Genetic variability of estimated half-sib progenies was at a satisfactory level for all population sizes investigated. Estimates of genetic variances were statistically significant for all samples and population sizes and ranged from 0.192 for 100 HS progenies population size to 0.232 for 300 HS progenies population size (Tables 4, 5, 6 and 7).
Values for genetic × environment variance interaction were also statistically significant for all samples and population sizes and ranged from 0.257 for 200 HS progenies population size to 0.283 for 100 HS population size (Tables 4, 5, 6 and 7). Statistically significant estimates of heritability were found for all samples in all population sizes. Their values ranged from 0.585 (100 HS progenies population size) to 0.647 (500 HS progenies population size) (Tables 4, 5, 6 and 7).
These results point to the possibility of working with a lower number of plants per F2 population for traits such as tolerance to root and stalk lodging.
Table 1. Mean values and standard errors for different population sizes.
Table 2. Differences between mean values from estimated population sizes.
ns = statistically non-significant difference
Table 3. Values of parameter D from the Komogorov-Smirnov one sample test for different population sizes.
Table 4. Estimate of genetic variance, genetic × environment interaction variance, heritability, and their standard error for a population size of 100 half-sib progenies.
|Population size 100 half-sib progeny|
Table 5. Estimate of genetic variance, genetic × environment interaction variance, heritability, and their standard error for a population size of 200 half-sib progenies.
|Population size 200 half-sib progeny|
Table 6. Estimate of genetic variance, genetic × environment interaction variance, heritability, and their standard error for a population size of 300 half-sib progenies.
|Population size 300 half-sib progeny|
Table 7. Estimate of genetic variance, genetic × environment interaction variance, heritability, and their standard error for a population size of 500 half-sib progenies
|Population size 500 half-sib progeny|