Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Comparison of un-replicated check plot designs
Early generation breeding experiments often are un-replicated due to insufficiency of seed. Due to the large numbers of lines involved, row-column designs are used with a replicated well-known check variety included in an attempt to overcome difficulties associated with the trials. Both arrangement and frequency of check variety in the grid are important in improving the line estimates. In these simulations square plots of unit size were used and, the effectiveness of three designs of check plots (good, A, medium, B, and poor, C), four check plot densities (5%, 10%, 20% and 50%); two ranges of influence which are a measure of variability (3.5 and 20) were studied. Also the effects of considering lines as a fixed or random effect in the analysis were studied in three grid sizes. Spherically correlated errors were assumed over the entire grid. The designs were compared based on the correlation of the treatment (genotype) effects to best linear unbiased predictors (BLUPs), least squares means (LSMEANs) and observed values, Ys. Comparisons were also made based on the proportion of true top ranking lines in the top ranked 20%. ^ Based on A-optimality criterion results, the A-designs were more optimal than B or C with optimality increasing with both range and density. This result was confirmed also by trends in correlations, proportions and smaller average variances.^ BLUPs correlated better to the genotype effects than LSMEANs and Ys at all check plot densities, designs, grid sizes and ranges. Ranking lines on BLUPs also captured the highest proportion of true top ranking lines compared to LSMEANs and Y in top ranked 20%. These responses increased with an increase in range while trends were constant. For LSMEANs check densities between 5 and 10% could be effectively used by plant breeders in selection experiments where genotypes are fixed effect. Since BLUPs performed the same at all densities, zero check simulations were included. Their results showed that checks may not be needed if genotypes are random and the spatial correlation is incorporated in the analysis. ^
Agriculture, Agronomy|Biology, Biostatistics|Statistics|Agriculture, Plant Culture
Sebolai, Boingotlo, "Comparison of un-replicated check plot designs" (2003). ETD collection for University of Nebraska - Lincoln. AAI3116606.