Plot size and shape for soybean yield trials

Abstract

A uniformity trial with soybean was used for investigating the effect of size and shape of plots on the precision and cost of field experiments. The measure of soil variability was the regression coefficient b for the regression of the logarithm of the reduced variance of the different sized plots on the logarithm of the number of units per plot. The coefficient determined was b = 0.4942. The estimate of optimum plot size, taking in consideration the costs K₁ = 70 and K₂ = 30 was x = 2.3 times the basic unit (0.60 m x 0.60 m). Considering the plot size in its components, x₁ (length) and x₂ (width), the best plot shape was studied too, and it has been determined that the "length and width of plots are factors acting independently on the reduced variance". In this study we have found that b₁ = 0.2673 and b₂ = 0.5808 are the parameters of a multiple regression equation, where b₁ influences the length and b₂ the width of plots. Hence we determined that both length and width influence the plot variance significantly, and that the influence of the width was greater than that of the length, and hence that long and marrow plots are more efficient. The size and shape of plots that reduced to a minimum the experimental error and the experimental costs were x = x₁.x₂ = 2.3 x 1.0 times the basic unit.