This experiment was based on data obtained from a factorial experiment with a split-plot design in which three sou fertility levels (FL) were assigned as main plots, and two maize varieties as 20 sub-plots. The study was conducted to evaluate the variability among sub-plots grouped with different sizes after subtraction of the variety effect, as well as the soil heterogeneity index for each replication of the soil fertility level. The soil fertility levels were: FL1 - fertilization recommended for grain yields up to 3.0 t ha-1 ; FL2 - fertilization for up to 6.0 t ha-1 plus Zn treated seeds; and FL3 - equal to FL2 plus incorporation of an organic compost into the soil. Comparing the the variance between different plot sizes was observed interaction between plot size and fertility level. The smallest plot size (5m²) and FL1 showed the smallest variance. However, for fertilitv level no statistic difference was observed when 25m² plot size was used. The estimation of the soil heterogeneity index due to soil fertilization level indicates that the addition of organic mather increases the variability as compared to the addition of mineral fertilizers.
soil fertility; plot size; variability
