Alternative data managements and interpretations for strip trials harvested with yield monitors.
Antonio P. Mallarino, Manuel Bermudez, D. J. Wittry, Paul N. Hinz, P. C. Robert, Richard H. Rust, W. E. Larson · 2000
This research demonstrates possible field methods and alternative data analyses for on-farm comparisons of fertilization practices for grain crops using precision agriculture technologies. Soil samples and grain yields were collected using intensive soil sampling techniques and yield monitors from field-scale, replicated strip trials. The data were managed and analyzed using various geographical information systems (GIS) and statistical procedures. Use of yield monitors allowed for yield measurements for different areas of the field, which is a major advantage compared with the traditional method of weighing grain harvested from strips as long as the field length. This capability lead to various statistically valid estimates of treatment effects on crop yield and different interpretations of the results. These estimates arise from dividing the field into different areas in ways that still conform to sound field experimental design. Consideration of the spatial correlation of yield in analyses of variance by means of nearest neighbor analysis or modeled semivariograms improved the assessment of treatment effects compared with classic analyses. However, accounting for spatial correlation usually resulted in similar treatment means. Subjective scientific and agronomic judgment continues to play a major role in selecting the most appropriate analysis to achieve the objectives of the experiments and in interpreting the results.