Tamanho de amostra para a avaliação de doenças em experimentos com arroz e trigo

Abstract

The blast and the yellow leaf spot are the major rice and wheat, respectively, both are commonly found as for its destructive potential. Because of this, proving the efficiency of control methods through experiences is of paramount importance, since these results will be used as reference for technicians and producers. The precision is related to the choice of suitable design (local control), the number of repetitions, the sample size in the field, among others. In the case of control experiments involving fungicides, which is the main method of controlling both diseases, leaf sampling in installments is necessary because measuring the entire population becomes unviable. Sampling generates a new error (sampling) within the plot, and this should be minimized by appropriate sampling strategy. Thus, this study aimed to determine sample sizes needed to assess the severity of rice leaf blast and the yellow spot of wheat. For this reason, results of chemical control experiments performed during harvests and 2009/2010 2010/2011 were used. The procedure for collection and analysis of data was identical in all experiments, regardless of culture. In all, the diseases were measured at seven, 14 and 21 days after application of fungicides by sampling 10 leaves in the plots. The variables were disease severity and area under the disease progress curve (AUDPC). Data were subjected to analysis of variance to obtain the experimental and sampling errors, and so, by hypothesis test to check whether the sample dimension and the number of repetitions were adequate. The departure from randomness of variable severity was tested in order to determine whether the methodology used to calculate the sample size was adequate. The departure from randomness test showed that both diseases behaved were distinct, both among treatments between assessments, sometimes distributing randomly in the field, sometimes not. Thus, the combination of theoretical distributions (indicating random or clustered distribution of the disease in the field) to the formula used to calculate the sample size is inadequate. The sample size necessary to measure the average disease severity in the plots was not the same in all treatments and between assessments. This result is expected since, in most of the experiments, the average of the disease was different between treatments and between evaluations. This result leads to a constant change in the relationship between the variance and the mean, which is indicative of the disease in the field of dispersion, this dispersion that is related to the sample intensity. Finally it was observed that, to measure the average AUDPC variable, it is necessary to evaluate fewer leaves in the plots.