Table I. Comparing different experimental designs.
Comparison of different experimental designs to study G and G:E effects on classic phenotypes in four different conditions. Each E condition is indicated with different gradient of gray in the blocks. In total there are n (164) genetically different RILs, and the data were analyzed in four different ways. The last two rows compare the number of QTLs for main G effect and G:E interacting effect detected using different design strategies. The numbers in parentheses indicate the QTLs that share confidence intervals (1.5 drop-off) with the full-block design.
| Design | QTL | Interacting QTL |
|---|---|---|
| Full-Block | 96 | 30 |
| Best power for G effect | ||
| Most expensive | ||
| Best power for G:E effect | ||
| Single-Block | 93 | 0 |
| Same power for G effect in the selected condition | ||
| Less expensive | ||
| Missing G:E effect | ||
| Random | 78(75) | 17(5) |
| Limited power for G effect | ||
| Less expensive | ||
| Limited power for G:E effect | ||
| GGG | 81(67) | 27(12) |
| Optimal power for G effect | ||
| Less expensive | ||
| Optimal power for G:E effect |