Stochastic simulation of a selection experiment in maize
A selection experiment in maize for increase in yield as described by Genter (1976. Recurrent selection for yield in the F₂ of a maize single cross. Crop Sci. 16:350-352) is simulated and the long term effects of selection intensity, recombination value, number of segregating loci, environmental var...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en |
| Published: |
Virginia Tech
2014
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10919/39330 http://scholar.lib.vt.edu/theses/available/etd-09122012-040023/ |
| Summary: | A selection experiment in maize for increase in yield as described by Genter (1976. Recurrent selection for yield in the F₂ of a maize single cross. Crop Sci. 16:350-352) is simulated and the long term effects of selection intensity, recombination value, number of segregating loci, environmental variance, and gene action model on yield and approach to complete homozygosity are investigated. Selection intensities of 11% and 22%, recombination values of .2 and .05, 100 and 50 segregating loci, and environmental variances of 0 and 60 were chosen and the combinations used represent a 1/2—fraction of a 2⁴ factorial design. Each of these 8 possible situations is combined with 3 gene action models to give 24 different situations for the simulation study. The first gene action model is linear with partial dominance for all loci. For the second and third gene action models, the genotype is partitioned into subgenotypes. The second gene action model has four subgenotypes, each of which is linear with partial dominance and different size effects. Six subgenotypes are combined in the third gene action model. Four of these are linear with partial dominance and different size effects, and the remaining two subgenotypes are epistatic.
In the simulation program, the genotype consists of 100 loci on 10 pairs of chromosomes, which contribute to the genotypic value according to the gene action models used. Each run begins with the F₁ hybrid of two unrelated inbred populations and continues for 30 cycles of selection or until complete homozygosity of the genotype is reached. Genotypic and phenotypic means, genotypic and phenotypic variances, and percent homozygosity are obtained for each cycle of selection.
Comparison of experimental observations and simulated values averaged over all runs for the first four cycles of selection indicate the simulated results are in good agreement with the experimental values. Regression analyses of percent homozygosity, genotypic value, phenotypic value (yield), and heritability in the broad sense using cycle of selection as the independent variable are performed. In addition, the effects and 2-factor interactions of selection intensity, recombination value, number of segregating loci, environmental variance, and gene action model on approach to complete homozygosity and increase in yield are investigated using analysis of variance for cycles 5, 10, 15, 20, and the last cycle of selection.
A recurrence formula for average coefficient of inbreeding is developed for this selection scheme. Also, for each gene action model explicit expressions for partitioning the genotypic variance into additive, dominance, and all epistatic components are given and evaluated for the F₂ generation. Consequences of selection on general combining abilities are being investigated. === Ph. D. |
|---|