A Simple Method for the Analysis of On-Farm Strip Trials

The design and analysis of experiments in farmers' fields using commercial yield monitoring equipment has received considerable attention in the last decade. Complex design and analytical techniques have been developed to maximize the utility of the trial to the producer while retaining the potential for robust statistical analysis. However, implementation of such approaches requires access to specialists in spatial analysis, geostatistics, and geographic information systems (GIS) and so is rarely implemented by farmers and their advisors. Here we explore a new and simple approach to the analysis of farmer strip trials and the spatial variability of treatment response. Strip trials evaluating different fertilizer treatments were conducted in farmers' fields in South Australia and Western Australia. Yield data were subjected to analysis with a linear model that accounted for the spatial autocorrelation in the data, a linear model that did not account for the spatial correlation and a moving pairwise comparison of treatments. Results suggest the pairwise comparison adequately identified treatment differences and their significance. Since this method can be readily implemented in a simple spreadsheet, it offers an important advance in facilitating on-farm experimentation using precision agriculture technologies.