Use of Hill and Short‐Row Plots for Selection of Soybean Genotypes

Abstract

Insufficient seed production rules out the use of large, replicated plots for the evaluation of the progenies of single soybean [Glycine max (L.) Merr.] plants. This experiment was conducted to compare two types of small, unbordered plots, (i) hill plots and (ii) single, short (1.5 m) row plots, for selection of soybean lines, and to determine the optimum allocation of testing resources for each type of plot. The genetic material consisted of 33 F2‐derived lines, selected only for similar maturity. Small‐plot tests were conducted in eight environments, using a split‐plot design with plot type as the wholeplot factor. The lines were also tested in four‐row, bordered plots at three locations for 3 yr. Maturity, height, lodging, and seed yield were determined in each test. Genotypic correlations were obtained between each trait in each plot type and each trait in four‐row plots. Variance components were used to calculate expected gains in four row plots following selection in each type of plot. Some of the genotypic correlations involving lodging and height in small plots differed from the corresponding correlations in four‐row plots. Direct selection for maturity, height, and lodging was more effective in fourrow plots than in small plots. Response to direct selection for yield was approximately 30% greater in short‐row plots than in hills. Increased height and lodging was an indirect response associated with selection for yield in hill plots, but to a lesser extent in short rows and still less in four‐row plots. We conclude that short rows are an attractive alternative to hill plots for testing single‐plant progenies. Based on expected gains for different resource allocations and on practical considerations, we recommend that short‐row plot tests be conducted at two or three locations with a single replication per location.