In theory, the plot size is in reverse proportion to the experimental error, and it depends on the soil heterogenity. It is essential to estimate its optimum size as the minimum area that represents a genotype within the experiment on condition that the experimental error is not significantly increased. With that aim 7 experiments with approx. 60 maize hybrids of different maturity groups (4 of earlier and 3 of later groups) were carried out by using randomised complete-block design in 6 replications throughout 3 independent years at 2 locations. In the statistical analysis the oversized plot (8 rows x 6 m) facilitated the artificial decrease of the plot size by 1 row each time. Thus, it was possible to carry out data analysis for 78 different plot sizes (from 8 to 2 rows) for all experiments. The analyses confirmed the theoretical knowledge of the relation between the plot size and the experimental error. In the majority of cases the two-row plot (at times even three-row plot) experimental error was significantly greater, which confirms their unreliability in the evaluation of hybrids. The simulation of the split-plot design proved the significant plot size and interaction hybrid x plot size influence on the later maturity groups yield whereas this did not occur with the earlier ones, which could be of considerable interest in the future investigations. The researched material did not establish the significance of competition effects. Four to five-row plot size showed to be the optimum one for hybrids and environments involved in these investigations.

Agronomija