Combined digital and standard methods to optimize nitrogen (N) management and reduce N surplus in winter wheat (T. aestivum) production

Precision agriculture practices have the potential to support the transition towards more resource efficient and productive agriculture while at the same time reducing the negative impact on the environment. The process is supported by agronomical and environmental data that can be collected through a multitude of sensors, which allow monitoring crop growth in the field and nutrient availability from the soil in real-time and at high spatial and temporal resolution. Wheat is one of the three most cultivated crops globally, with about 750 Mt year−1 production and the most cultivated cereal crop in Europe and Switzerland. At the same time, the nitrogen use efficiency (NUE) in wheat cultivation (and other crops) is relatively low in Europe and globally, with values between 30% and 70%, whereas an optimal NUE should be about 80–90% to balance a low N surplus with a low risk of soil N depletion. Nitrogen (N) is often the most limiting nutrient for plant growth and accounts for most fertilizer applied in agriculture to promote high yield. The inaccurate use of N fertilizers globally and in Switzerland causes some of the major environmental and health problems related to agricultural crop production, namely the emission of greenhouse gas nitrous oxide (N2O) to the atmosphere and nitrate leaching (NO3 – ) into groundwater. Furthermore, the sources of N available to the plant during the season are affected by the natural spatial variability at field scale and the timely synchronization between N inputs from fertilization and crop N demand. The concept of observing, measuring, and responding to inter- and intra-field variability in crops during the season is known as site-specific N management. This thesis's objective was to develop and test a combination of digital and standard methods to provide decision-support for N fertilization in winter wheat production in Switzerland. The basic idea is that monitoring the agro-ecosystem's dynamic processes leads to a more automated quantification of the N balance, including the N supplied by the soil system. By precisely estimating the required, additional N input in relation to the expected N outputs, a better-informed N management would result in increased N use efficiency and reduced risk of N emission to the environment. This thesis's central hypothesis was that the implementation of site-specific N fertilization using variable rate (VR) technology would reduce average N application compared to the standard (ST) uniform application without affecting yield, thus increasing N use efficiency and reducing the risk of N surplus. The methods established were based on sensor data such as multi-spectral images and soil properties combined with standard plant and soil analysis. The experimental setup consisted of a multi-plot design with two treatments (VR and ST) and two controls (no fertilizer and additional fertilizer) in seven fields over three years (2018–2020) for a total of seven site-years (F1–F7) located in northeast Switzerland.ii Wheat was monitored throughout the season to support three split fertilizer applications. The combined imaging and ground truth plant data were used to validate the sensitivity of spectral vegetation indices, suitable for the sensor used, with biomass and N status traits. Grain yield was mainly in the expected range (6–7 t ha−1), and no significant difference between VR and ST was observed. In contrast, N fertilizer application was reduced in the VR treatments between 5 and 40%, depending on the field. In most site-years, N use efficiency was improved (13% on average) by redistributing and reducing the amount of N fertilizer applied. However, a better prediction of N mineralization in the soil and related N uptake by the plants was necessary to further optimize in season N fertilization. To better understand the relation between the components of the N balance and crop growth, part of the study concentrated on the ST treatment of six site-years (F2–F7). Two indicators were tested to link the weekly remote sensing-based monitoring of N uptake and nitrate-N dynamics in soil pore water measured with suction cups and an ion-selective electrode. The N balance was calculated and compared with the fertilization recommended by the adjusted N fertilization norm (ANFN) a good practice to determine N fertilizer requirement in Switzerland. The remotely estimated Nuptake (REN) allowed high temporal resolution N monitoring. The changes in nitrate-N in the soil solution (NSS) could be monitored weekly. The REN and NSS showed a distinct relationship for high- and low yielding fields. The timely integration of both was related to improved NUE and reduced N surplus. For Swiss wheat production, NUE can be increased and surpluses reduced with the ANFN method. A concept was developed to assess the economic and environmental performance of the two treatments VR and ST. The N balance was linked to the economic optima. This relation was used to calculate N fertilizer's price that would need to be charged to maximize revenues and minimize the risk of N surplus. On average, net revenues in VR were 4% (94 CHF ha−1) higher than in the ST treatment. The residual N was on average 32% (21 kg N ha–1 ) lower in VR compared to ST, due to a reduction in N inputs of 13% without significant yield differences. The economic optima ranged from 200 to 240 kg N ha–1 over the three years (2018–2020). The point where increasing N surplus occurs varied from 180 to 205 kg N ha–1 (2018–2020). The results imply that if N fertilizer's price increases by 20–40%, the gap between the economic optimum and the optimal N fertilization could be reduced. In conclusion, by monitoring and managing the variability observed in crop and soil, the established methods showed possible solutions to reduce N surplus in medium- to small-scale wheat production in Switzerland. Ultimately, the progress necessary to further improve wheat cultivation will rely on identifying genotypes and precision management practices better adapted to the changing environment. Sensor-based monitoring of agro-ecosystems can lead the way to these practices, aiming to maintain profitability, high yields, and quality, and reduce ecological impact by reducing emissions to the environment