The role of statistics in conducting on-farm experimentation with modern monitoring technology in agriculture

PACCIORETTI, P.; CÓRDOBA, M.; BALZARINI, M.

Barcelona

Conferencia; XXIXth International Biometric Conference; 2018

On-farm experiments are conducted in producers? fields using precision technologies that facilitate trial set up without additional tasks other than the necessary ones for crop development. Current availability of precise machinery equipped with proximal and remote sensors enable automation of changes in the assignment of agricultural inputs (seeds, fertilizers, pesticides) in a plot. The possibility of changing input rates automatically and monitoring associated yields enable local comparisons of the effect of different treatments on production. Advances in design techniques and statistical analyses of this type of trials are essential for rapid development, fine-tuning and evaluation/adoption of precision agriculture. One goal of experimental design in producers? fields is to generate the sufficient amount of data to estimate productivity functions. On-farm estimated productivity function (yield as a response of treatment) for a given number of crop seasons, allows us to make environmentally and economically optimum prescriptions for crop management in the agricultural plot. Design-based and model-based approaches for treatment comparisons can be used in on-farm experiments. Experiment design requires considering the basic principles of randomization, replication and local control within a context of georeferenced data with continuous spatial variation at a fine scale. The underlying spatial variability among sites in a plot (variability in soil properties,topography, water availability, historical yield levels) can be used for stratification or identification of zones, in which treatments are compared under uniform conditions or, rather, as covariables of yield prediction models as a function of treatments. Relative efficiency of completely randomized, randomized complete blocks (RCBD), and strip designswere assessed by simulation. RCBD was the most efficient design to compare treatments, and the differences between RCBD and the other two designs were really high when site covariables were not accounted for. The benefits of adopting a site-specific predictive model approach, including site characteristics as covariables as well as interactions between them with input rates, are discussed through an illustrative case.