| Site-specific, or variable-rate management of nitrogen (N), an important crop nutrient, provides the opportunity to improve N fertilizer use efficiency and reduce N losses to the environment. Field studies were initiated in 1993 to quantify the effects of variable-rate N management on corn (Zea mays L.) yield and the potential for nitrate loss to groundwater. In 1994 and 1995, two treatments, uniform and variable application of N fertilizer, were imposed on two center-pivot irrigated corn fields in south central Kansas, USA. The soil at Site 1 (50 ha) was a sandy, mixed, slightly acid, thermic Psammentic Haplustalf while the soil at Site 2 (60 ha) was a complex of the soil at Site 1 and a sandy, mixed, slightly acid, thermic Typic Ustipsamment. Corn yields and residual soil nitrate concentrations were similar between treatments at both sites and both years; however, N use efficiency was slightly higher with variable N application. Temporal instability in spatial yield patterns, especially at Site 1, made it difficult to develop reliable site-specific yield goals based on 1-2 years of yield mapping. This result is particularly important for precision farming because the N fertilizer recommendation model was primarily driven by yield goal for these sandy fields. Additional years of yield mapping will be required to develop yield goal maps that can be used to successfully implement variable-rate N management. Variable-rate nitrogen management requires intensive soil sampling. Two grid-based sampling methods were compared. Single-core and composite soil sampling at grid points resulted in similar measurements for soil phosphorus and potassium, but quite dissimilar in the measurements of other soil properties. The effects of grid sampling intensity and location were also investigated. For optimally mapping soil properties on these sites, a grid size of 0.6 ha is recommended. Regression analysis was used to show that soil and topographic properties accounted for only 39, 23, and 20% (Site 1) and 59, 57, and 48% (Site 2) of the variation in yield for 1993, 1994, and 1995, respectively. The soil variables in the regression equations were different for each year. Other unmeasured site-specific properties could have accounted for more of the yield variation. |
