Bootstrap confidence intervals as tool to classify races of soybean's cyst nematode

Abstract

Bootstrap estimates of arithmetic mean of Pickett, Peking, PI88788 and PI90763 soybean genotypes and the confidence intervals obtained through normal theory and bootstrap distribution of this estimator, like the percentile bootstrap and the BCa (bias-corrected and accelerated) of the differentiation parameter of the standard cultivar Lee for susceptibility, are utilized to classify races of soybean cyst nematode. The confidence intervals obtained from bootstrap distribution showed smaller amplitude and were very similar. Therefore, the lower limit of the bootstrap percentile confidence interval was taken as reference level in the bootstrap distribution of the arithmetic mean estimator of the differential genotypes allowing to estimate the empirical probability of a positive or negative reaction and, consequently, to identify the most probable race under specific test.