Using soil properties to predict wheat yields on Illinois soils

Updated crop yield estimates for different soil types are required to meet the needs of farmers, land appraisers, and governmental agencies of most st ates within the United States. The objective of this study was to update wh eat [Triticum aestivum L.] yield estimates for all soil types in the state of Illinois under the average management level used by all Illinois farmers in the 1990s. A crop yield soil properties (CYSP) model was developed from established 1970s wheat estimates and soil properties. Correlation and mul tiple regression analysis was used to establish the relationships between 1 6 physical and chemical soil properties and wheat yields. Using step-wise m ultiple regression, a final CYSP model with six variables (% silt in A and E horizons, % organic matter in A and E horizons, cation exchange capacity, rooting depth, bulk density of the B horizon, and Na content) explained 78 % of the variation in 1970s wheat yields of (R-2 = 0.78 [34 soils]). A 22-y ear yield trend was applied to the 1970s predicted and established wheat yi elds to estimate the 1990s yields under an average level of management used by all Illinois farmers, The 1990s wheat yields (model predicted plus tren d) were compared with both established (Circular 1156) plus trend yields an d with farmer-reported yields by Illinois Agricultural Statistics (IAS) Sta ff for 161 soils in nine test counties. Predicted 1990s test county wheat y ields showed no statistical differences between the three data sources. The final 1990s wheat yield estimates for all Illinois soil types were the mea n of the estimates from: (i) the established 1970s yields plus 22-year yiel d trend, (ii) the model predicted plus 22-year yield trend, and (iii) the 1 990s farmer-reported yields. The proposed approach to updating wheat yields worked well in Illinois and should be useful in surrounding states and/or countries.