A Method to Estimate the Probability that Subsoiling will Increase Crop Yields1
Crop yield responses to subsoiling are determined by the amount and distribution of rainfall and soil properties. Because growing season rainfalls are not known in advance, the likelihood of subsoiling increasing yields at a given site can only be stated in terms of probabilities based on current soil properties and historic rainfall patterns. In this study, the probability that subsoiling would increase tobacco (Nicotiana tabacum L.) yield as compared to a nonsubsoiled treatment was estimated for 20 coarse-textured soils in the southeastern USA where significant yield increases from subsoiling had been observed previously or were expected. Using data collected from 1981 through 1984 for yields of subsoiled and nonsubsoiled treatments, soil physical properties, and rainfall from the 20 sites, a regression model was first developed to describe yield differences between treatments on the basis of specific weekly rainfalls and the sand percentage and water retention of selected horizons. The regression model was then used to estimate yield differences between treatments using physical properties of each of the 20 sites and rainfall histories of two Coastal Plain locations for the years 1930 through 1986. For each rainfall history, the number of years that the estimated yield differences between treatments were ⩾300 kg ha 1(significant differences in yields at the 0.05 level) was tabulated for each site and divided by 57 to compute the proportion of years that subsoiling would have significantly increased yields in the soils considered. These proportions were assumed to be the probabilities that subsoiling would increase yields for a given rainfall history and set of soil properties. The estimated probability mean and standard deviation were 0.69 ± 0.12 across all 20 sites and both rainfall histories. This indicated that on the average, subsoiling would be expected to increase tobacco yield in approximately 7 out of every 10 yr. Probabilities were < 0.50 in only two sites that had relatively high water retention values. The probability values decreased as E-B horizon water retention increased and Ap horizon sand percentage decreased. These results pertain to Costal Plain soils that had bulk densities ⩾ 1.63 Mg m−3 in either the Ap or E-B horizon, and sand percentages ⩾ 73% in the Ap horizon.
Copyright © .