Dimensionamento amostral na cultura do feijão de porco

Abstract

Jack bean is an agricultural crop used in crop rotation systems, with an important environmental contribution in biological nitrogen fixation and soil improvement. The objective of this work was to determine the sample size necessary for estimating the mean of twelve jack bean variables at twenty-five levels of precision. Three uniformity tests were carried out in an experimental area of the Federal University of Santa Maria, campus of Frederico WestphalenRS. Trials 1, 2 and 3 were sown, respectively, on November 8, 2020, December 9, 2020 and January 8, 2021. Each trial had five seeding rows of 20 meters in length with 100 plants per row and spacing between rows of 0.45 m and between plants of 0.20 m and sowing performed manually and without the use of fertilizers. At harvest, carried out between May and August 2021, depending on the sowing date and maturity of the plants, twelve morphological and productive variables were evaluated: plant height; stem diameter; number of nodes; number of productive branches; number of pods; number of grains; number of grains per pod; total mass; plant mass without pods; mass of grain less pods; grain mass; and, mass of one hundred grains. The final number of evaluated plants was 351, 330 and 256, respectively, for trials 1, 2 and 3. Afterwards, for each time period and variable, measures of central tendency (mean and median), dispersion (minimum, maximum, standard error, standard deviation and coefficient of variation), distribution (kurtosis and skewness) and the 95% confidence interval for the population mean were calculated. Then, the sample size necessary for estimating the mean of each variable was determined, assuming twenty-five levels of precision (estimation errors equal to 1, 2, ..., 25% of the mean) and defining the estimation error with based on the original sample. In jack bean, an average of 50 plants, are required for the estimation of the mean of twelve variables with an error of 10% of the mean. At this level of precision, the sample size varied between seven and 140 plants, depending on the sowing date and the evaluated variable.