method to estimate soybean seed protein and oil concentration before harvest

Temperature and precipitation variables during linear seed fill are known to be environmental determinants of protein and oil composition of the soybean [Glycine max (L.) Merr.] seed. However, the contribution of other precipitation and temperature events during the growing season and a method that would determine the precipitation and temperature variables most related to protein and oil concentration values of the seed has not been fully explored. The former was evaluated by comparing monthly temperature and precipitation variables of the growing seasons to protein and oil data for the years 1959 to 1996 from three locations listed in the Uniform Soybean Tests, Northern Region. The data set comprised locations from Maturity Groups II and III and consisted of 186 location-years. Classification and regression “tree-based” analysis were conducted to determine the month, environmental variable, and “splitting” points that correctly classified most of the 186 location-years for below-vs.-above-median protein or oil composition. The protein concentrations from the location-years were separated into these two median-boundary categories most readily by temperature variables from the months of April and August. The oil concentrations from the location-years were classified best by August and September temperature variables and precipitation in May and September. The sum of protein and oil concentrations from the location-years were best separated by August and July temperature variables and precipitation in May and July. The protein-to-oil ratios from the location-years were best separated by September precipitation and July and June temperature variables. These data demonstrate that tree-based models can use monthly temperature and precipitation variables during linear seed fill and other specific months of the crop year and relate them to the final protein and oil concentration in the seed. These results could be used by the processing industry to estimate seed composition before harvest.