The variance associated to the production of plants with multiple harvests is not always homogeneous, what compromises the experimental precision. To address this issue, in this work we identified (1) the behavior of the means and variance for fruit mass in zucchini (Curcubita pepo) along harvests, between planting rows within and betweens harvests, and (2) the interference of different crop managements over these parameters. The experiment was carried out in two seasons, winter-spring (I/P) and summer-autumn (V/O), 2004/2005, using cultivar Caserta and two plastic tunnels (T1 and T2), with three 25-plant rows in each tunnel. In T1, drip irrigation was used, combined with mulching over the ridges, and less biocide spraying than in T2. In T2, sprinkler irrigation was used, placed 1,8 m height, without mulching. The Bartlett test was applied to the six planting row variances within each harvest and to the average variance of each row to check the homogeneity between harvests in each crop season. The t test was used to compare the production means in each planting row, in each harvest, in each tunnel, and also between planting rows, for plants with the same position in the tunnel. In the V/O season, variances were homogeneous in two out of the eleven harvests. In season I/P, variances were homogeneous in ten of the thirteen harvests. In both seasons, variance homogeneity was higher in the first half of the cycle. In V/O, variances in T1 were in general smaller than in T2 in the whole cycle. In both seasons there was variance heterogeneity between harvests. In V/O, significant differences were observed in six and three harvests, respectively in T1 and T2, with production means reaching 481.0 and 454.0 g plant-1 T1 and T2, respectively. In I/P, significant differences were observed in the first (T1) and fourth and fifth harvests (T2), with the means for fruit mass reaching 598.2 (T1) and 597.4 g plant-1 (T2).
Curcubita pepo; greenhouse; experiment planning; size plot; homogeneity of variances
