Estimativa do número de repetições e medidas de precisão em experimentos com melão

Abstract

The determination of the number of replications constitutes an important stage of the experimental planning because it is directly related to the precision. As the coefficient of variation, a statistic commonly used as a tool for evaluating experimental precision, does not depend on the number of replications, new statistics have been proposed for this purpose. This study was carried with the objective of determining the number of evaluations (replications) necessary to evaluate the characters of productivity and soluble solids in melon hybrids, and were tested the adequacy of the statistics as measures of experimental precision. The study was composed of twelve trials with ten hybrids of Honey Dew melon, eight trials with thirteen hybrids of Yellow melon, nine trials with eight hybrids of Piel del Sapo melon, nine trials evaluating eight hybrids of the Cantaloupe type, and twelve trials evaluating nine hybrids of the Galia melon, totaling fifty experiments. All trials were conducted in the Rio Grande do Norte State, using a randomized block design with three replications. Trials with three repetitions made it possible to identify superior genotypes with more than 80% accuracy in predicting their real value for productivity in the Yellow type, and for productivity and soluble solids in the types Cantaloupe and Galia. The lowest values in the prediction of the real value of the genotypes, using three replicates, were observed in the melon Piel del Sapo type, allowing to identify superior genotypes in relation to productivity and soluble solids, respectively, with precision of 46.6% and 42.0% in the prognosis of real value, requiring a greater number of repetitions for greater prediction. Regarding the evaluation of experimental precision statistics, it was observed that the use of the variation coefficient alone is inadequate, as this statistic does not consider the level of genotypic variation expressed in the character, both for productivity and for soluble solids in melon. Through the use of selective accuracy and the value of the F test for genotypes, it was observed that more precise trials, for productivity, are associated with greater genetic variances and smaller residual variances, and are independent of the mean. For soluble solids, the selective accuracy and F test value for genotype proved to be more adequate for evaluating the quality of the trials because they contemplate the genetic variability