This paper reviews options for the design and analysis of on-farm experiments. It covers both older approaches that have been popular since the Green Revolution, and more recent developments made possible by the availability of online monitoring systems as used in precision farming. The roles of randomisation as well as of geostatistical methods of analysis for these kinds of experiments are critically discussed. Two case studies are provided for illustration.
In recent years, real-time technology has been introduced into the practice of spraying variable fungicide rates in cereal fields. Plant parameters for characterising heterogeneous plant growth such as biomass or plant surface area can be indirectly detected by the sensor CROP-Meter. The sensor signal is correlated with the Leaf Area Index, which can be used to adapt the application rate. However, this relatively simple method of controlling variable-rate fungicide application does not take into account the differences in disease distribution. In practice, decision support systems such as proPlant expert.classic can provide information about disease infection probabilities, application time, fungicide products and application rates for uniform spraying. A prototype of the system proPlant expert.precise was developed to estimate infection risks from fungal diseases using weather and field-specific data for up to three management areas with different yield expectations. The system also considers economic factors such as expected yield and costs of the fungicide products in generating a spraying map with different fungicide dosages. The information from the CROP-Meter (sensor) and from the decision support system proPlant expert.precise (map) was combined to provide a real-time spraying system with map overlay. The system was tested in 2007 in three winter wheat fields. Compared with conventional uniform spraying the CROP-Meter with map overlay treatment resulted in up to 32.6% fungicide savings (CROP-Meter versus uniform: up to 20.3%). There was no yield reduction on average when the sensor-controlled spraying technologies were used.
A joint assessment of two separate approaches investigated the occurrence of volunteer oilseed rape (Brassica napus L.; OSR) as affected by cultivar, field history and environment. Approach I comprised surveys for volunteers on >100 farmers' fields in Germany in the years 2009 and 2010. Volunteers were assigned to the cultivars grown in previous years by inter-simple sequence repeat-PCR and cluster analysis. High-dormancy cultivars resulted in 0-7 and low-dormancy cultivars in 0-1.3 volunteers m À2 . Highest numbers originated from the most recent harvests 2007 and 2006. Approach II was a meta-analysis based on 116 data sets from field trials and farmers' fields in Germany to evaluate and to rank the impact of management factors on the soil seedbank and volunteers in following crops. Varietal disposition to seed dormancy turned out to be the significantly most relevant factor. The contribution of varietal dormancy to variation in the soil seedbank and of volunteers in the 1st and 2nd following crop was 1.2-2.3 times as great as the contribution of post-harvest tillage. Up to 45% of the variation in the observations originated from factors that can be controlled by human actions on a given location in a given year. The overall analysis confirmed the results from independent short-term trials and showed that both agronomists and breeders need to contribute to reducing OSR volunteers.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.