Efficacy of multivariate analysis and zone soil sampling to study relationships between site variables affecting crop yield and yield response to phosphorus and potassium fertilization.

摘要

Crop yield and yield response to fertilization is frequently heterogeneous within-fields. A better understanding of this variability and the development of cost-effective soil sampling techniques to better represent plant nutrient levels are needed to improve nutrient management practices. Two studies were conducted in Iowa to address this general objective.;One study focused on correlations between several site variables and their relationship with soybean [Glycine max (L.) Merr.] yield using two multivariate techniques. Soil and crop variables were measured at five fields using a dense, 0.2-ha grid-point sampling approach. Factor analysis grouped correlated variables into three common factors across fields, which represented conditions for early crop growth, interactions between intrinsic soil properties and landscape position, and P and K availability. These factors explained 10--65% of the yield variability across fields. Principal component analysis explained a similar portion of the yield variability, but relationships between site variables were more difficult to interpret.;In the second study, soil-test P and K results from samples collected using a dense grid sampling approach (0.08--0.24-ha) before applying P or K treatments to long strips in seven fields was used to assess the effectiveness of various zone sampling approaches. Zones were delineated using a systematic, 1.0 grid-cell approach, soil survey maps, remotely sensed soil electrical conductivity, elevation, and slope. The efficacy of the zoning approaches was assessed based on soil-test values and corn (Zea mays L.) and soybean yield response to fertilization. We used yield monitors, global positioning systems, and geographical information systems software. Zones based on elevation and (or) electrical conductivity identified field areas with high and low yield. However, soil-test values and crop response to fertilization seldom differed across zones, except for the grid-sampling approaches. A systematic 1.0-ha grid-cell approach was more effective for identifying field areas with different crop response to fertilization, but the numerous samples required may result in excessive soil testing costs. Because long-term nutrient removal with harvest also is a key element for P and K fertilizer management, less costly zoning approaches used in conjunction with yield maps could be useful for improving long-term nutrient management.