Hill vs. row plots in predicting full-season and late-planted soybean yield

Hill plots have been shown to be useful as initial yield test plots for soybean [Glycine max (L.) Merr.] under full-season production conditions. The objectives of this research were to determine the value of hill plots for yield selection at double-crop planting dates and to characterize the number of plants per hill needed to maximize yield at these delayed planting dates. Twenty-eight genotypes in each of three maturity classifications were grown in full-season (average planting date 21 May and late-planted (average planting date 4 July) hill plots and row plots at one location in each of 3 yr as prediction environments. These same genotypes were grown in row plots in full-season and late-planted evaluation tests to establish genotype yield rankings. In both the full-season and late-planted tests, correlation coefficients were calculated for genotype yields in row plots and hill plots in each prediction environment with genotype mean yields in row plots in the evaluation tests. These correlation coefficients were used to assess the value of the different plot types as predictors of genotype yield rankings. Hill plots were as good as row plots as a predictor both for full-season yields and for late-planted yields based on mean productivity [(full-season yield + late-planted yield)/2]. Hill plots, however, were inadequate for predicting late-planted yields when the selection criterion was late-planted yield itself. Plant populations of three cultivars were varied from 1 to 10 plants per hill plot in full-season and late-planted tests at one location for 2 yr. Seven plants per hill plot maximized yield in the late-planted test.