(1) Design on-farm trials. In the proposed project, we will follow a novel approach that will allow us to (i) efficiently locate field experiments, (ii) determine which practice(s) to evaluate, (ii) scale out results from these experiments to other farmer fields, and (iii) quantify regional production and ROI impact from targeted production changes. The project will leverage from the outputs of our previous 3-year benchmarking project. First, our approach will determine trial locations using the Technology Extrapolation Domain (TED) spatial framework that delineates geographic regions with similar climate-soil conditions (Rattalino Edreira et al., 2018). Second, we will determine which practice(s) to test in each TED based on a list of candidate explanatory factors for yield gaps derived from our previous analysis of farmer survey data (Mourtzinis et al, 2018). We will first use legacy data (existing data from on-farm research groups and previous NCSRP-funded benchmarking project) to evaluate the robustness of the proposed approach. Such evaluation will compare the ROI in on-farm trials following the current “business-as-usual” model versus the prescient selection model proposed here.
Our prescient approach will strategically locate on-farm trials to represent TEDs with largest soybean planted area in each state. In other words, we will prioritize environments where the potential impact of on-farm research is the largest. Fields will be chosen to be representative of the “average” farmer in each TED, that is, with yields and practices that do not deviate substantially from the average in the region. We will use data from our previous project as a benchmark to determine average yield and dominant set of practices for each TED (Mourtzinis et al, 2018). We will ensure that fields are located near an automatic weather station—understanding on-site weather conditions will allow proper interpretation of the results. Once we have determined the location of the field trials, we will determine what practice to test for (or to omit) based on (i) what the farmer is currently doing, versus (ii) a list of candidate management practices explaining yield gaps in each TED derived from our previous NCSRP benchmarking project (Rattalino Edreira et al 2016; Mourtzinis et al 2017). When designing the specific treatment for each TED, the aim will be to have a ‘system comparison’ in which we modify a management practice, but we also fine-tune other practices so that we fully capture the yield benefit associated with that change. The treatment will aim to increase farmer profit by increasing yield, or by reducing costs, or both, and doing so in a way that maximize profit and minimize environmental footprint.
The preliminary selection of TEDs where we would like to conduct on-farm trials account for the majority of soybean acreage in the US NC region, and we have already identified a set of candidate explanatory factors for yield gaps in each of them through our previous NCSRP-funded project. The TED framework will help disseminate results from the on-farm trials to other producer analog soybean fields with similar climate-soil conditions. As we mentioned previously, because on-farm trials will be run across a wide range of climates and soils, results from the project will also be useful to producers in states other than those included in the proposal. Using the TED framework as basis for site selection will help on-farm research groups to better complement and coordinate their field trials to make sure that they will not have an excessive number of field trials located in one single region with similar climate and soil, while other important regions (in relation with soybean area) are not covered.
(2) Conduct on-farm trials. Based on our previous experience conducting field experiments, a large (but not excessive) number of field trials (that are not concentrated too much in a local area) is needed to detect statistically significant effect of changes in management practices and make robust recommendations. We would like to conduct a minimum of 20 field trials (one trial per farmer field) in each year and each state. Each field trial will consist of 3-4 replicated strips (size: ~40 by 500 ft) where the treatment determined by the UW-UNL core team, in consultation with the state collaborator, will be implemented. The goal is to compare the yield and profit measured for that treatment against the one attained by the producer for the rest of the field using his/her average management. We will conduct the on-farm trials over 2 years to account for year-to-year weather variation, particularly, in-season precipitation, which can be locally variant. The collaborator in each state and his/her technician will be responsible for conducting field trials following UW-UNL guidelines, input results into a digital file, and send it to the core team. Our collaborators will also request farmers to report information about their yield, field location, and detailed information on crop/field/input management, such as planting date, soybean variety, tillage method, etc. as well as to submit grain samples to UW for seed composition. Individual field data and producer contact information will remain confidential. Indeed, the TED framework will ensure confidentiality of producer data because, once a field has been contextualized relative to its climate and soil, the exact field location has no value. Participating farmers will also complete a grower production survey to get input costs for economic analysis. A fully funded third year of on-farm research would be preferred; however, this will not be possible given the granting cycle of NCSRP. We propose initiating a third year for the project where we can establish research plots and identify farms to serve as on-farm learning laboratories where collaborators can sponsor field days and events to communicate results locally. Collaborators will seek local QSSB or industry funding to help finalize harvest and data synthesis for year 3.
(3) Data analysis. Once the data are provided, they will need to be standardized into a single, consistent format, error-checked, and then inputted into a digital database. We will use a range of state-of-art methods to analyze the data from field experiments, including remote sensing, crop modeling, spatial analysis, and advanced statistical techniques (e.g., machine learning). We will make use of the expertise on analysis of farmer data acquired by the UW-UNL team during the previous benchmarking project to retrieve detailed data on weather, soil, and topography for each field-year trial. UW and UNL will be responsible for data analysis and will collaborate with faculty at the Statistical Departments at UNL, UW and ISA to validate our statistical analyses. Our analysis will be based on the aggregated database and results of the analysis will not specifically pinpoint individual producer fields. The database will be saved in a secured server, which will be accessible only to those involved in the project. After the end of the project, the state-specific databases (yield, management, soil, weather) will be (with NCSRP permission) turned over to the on-farm groups for use by them, particularly if they want to continue the annual on-farm trials to build longer term databases for their use in knowing more about their producer constituents.
(4) Communication and dissemination of results. Results of the proposed project will be disseminated to producers and public via peer-reviewed scientific and Extension publications, presentations at scientific conferences and Extension events sponsored by universities, natural resources districts, growers’ associations, and proprietary organizations that market their products to soybean producers. Individual state and combined regional reports will be posted and distributed through various webpage portals such as www.coolbean.info, SRRI, and the NE CropWatch website (http://cropwatch.unl.edu/). UW will organize team meetings at the ASA/CSSA/SSSA meetings and/or Commodity Classic. Moreover, participating farmers will be actively involved with this research project. Farmers from each state will be identified to participate in farmer video profiles which will be shared with individual state boards to highlight local research and translate results. The participating farmers will also be linked in a farmer-to-farmer network where they can share ideas in a non-competitive environment. Lastly, in year 3 of this project participating farmers and on-farm groups will plant year 3 of the research plots. These plots will also serve as field based “learning laboratories” where collaborators will hold field days to highlight, communicate and promote the results of this work locally. Given the project timeline, additional funding will be required to see this third year of research through harvest.