Improved prediction of optimal N fertilizer rates for corn (Zea mays L.) can reduce N losses and increase profits. We tested the ability of the Agricultural Production Systems sIMulator (APSIM) to simulate corn and soybean (Glycine max L.) yields, the economic optimum N rate (EONR) using a 16-year field-experiment dataset from central Iowa, USA that included two crop sequences (continuous corn and soybean-corn) and five N fertilizer rates (0, 67, 134, 201, and 268 kg N ha-1) applied to corn. Our objectives were to: a) quantify model prediction accuracy before and after calibration, and report calibration steps; b) compare crop model-based techniques in estimating optimal N rate for corn; and c) utilize the calibrated model to explain factors causing year to year variability in yield and optimal N. Results indicated that the model simultaneously simulated well long-term crop yields response to N (relative root mean square error, RRMSE of 19.6% before and 12.3% after calibration), which provided strong evidence that important soil and crop processes were accounted for in the model. The prediction of EONR was more complex and had greater uncertainty than the prediction of crop yield (RRMSE of 44.5% before and 36.6% after calibration). For long-term site mean EONR predictions, both calibrated and uncalibrated versions can be used as the 16-yr mean differences in EONR’s were within the historical N rate error range (40 to 50 kg N ha-1). However, for accurate year-by-year simulation of EONR the calibrated version should be used. Model analysis revealed that higher EONR values in years with above normal spring precipitation were caused by an exponential increase in N loss (denitrification and leaching) with precipitation. We concluded that long term experimental data were valuable in testing and refining APSIM predictions. The model can be used as a tool to assist N management guidelines in the US Midwest and we identified five avenues on how the model can add value towards agronomic, economic, and environmental sustainability.

Historically crop models have been used to evaluate crop yield responses to nitrogen (N) rates after harvest when it is too late for the farmers to make in-season adjustments. We hypothesize that the use of a crop model as an in-season forecast tool will improve current N decision-making. To explore this, we used the Agricultural Production Systems sIMulator (APSIM) calibrated with long-term experimental data for central Iowa, USA (16-years in continuous corn and 15-years in soybean-corn rotation) combined with actual weather data up to a specific crop stage and historical weather data thereafter. The objectives were to: (1) evaluate the accuracy and uncertainty of corn yield and economic optimum N rate (EONR) predictions at four forecast times (planting time, 6th and 12th leaf, and silking phenological stages); (2) determine whether the use of analogous historical weather years based on precipitation and temperature patterns as opposed to using a 35-year dataset could improve the accuracy of the forecast; and (3) quantify the value added by the crop model in predicting annual EONR and yields using the site-mean EONR and the yield at the EONR to benchmark predicted values. Results indicated that the mean corn yield predictions at planting time (R² = 0.77) using 35-years of historical weather was close to the observed and predicted yield at maturity (R² = 0.81). Across all forecasting times, the EONR predictions were more accurate in corn-corn than soybean-corn rotation (relative root mean square error, RRMSE, of 25 vs. 45%, respectively). At planting time, the APSIM model predicted the direction of optimum N rates (above, below or at average site-mean EONR) in 62% of the cases examined (n = 31) with an average error range of ±38 kg N ha⁻¹ (22% of the average N rate). Across all forecast times, prediction error of EONR was about three times higher than yield predictions. The use of the 35-year weather record was better than using selected historical weather years to forecast (RRMSE was on average 3% lower). Overall, the proposed approach of using the crop model as a forecasting tool could improve year-to-year predictability of corn yields and optimum N rates. Further improvements in modeling and set-up protocols are needed toward more accurate forecast, especially for extreme weather years with the most significant economic and environmental cost.

Determining the optimal nitrogen (N) rate in corn remains a critical issue, mainly due to unaccounted spatial (e.g., soil properties) and temporal (e.g., weather) variability. Unmanned aerial vehicles (UAVs) equipped with multispectral sensors may provide opportunities to improve N management by the timely informing of spatially variable, in-season N applications. Here, we developed a practical decision support system (DSS) to translate spatial field characteristics and normalized difference red edge (NDRE) values into an in-season N application recommendation. On-farm strip-trials were established at three sites over two years to compare farmer’s traditional N management to a split-application N management guided by our UAV sensor-based DSS. The proposed systems increased nitrogen use efficiency 18.3 ± 6.1 kg grain kg N⁻¹ by reducing N rates by 31 ± 6.3 kg N ha⁻¹ with no yield differences compared to the farmers’ traditional management. We identify five avenues for further improvement of the proposed DSS: definition of the initial base N rate, estimation of inputs for sensor algorithms, management zone delineation, high-resolution image normalization approach, and the threshold for triggering N application. Two virtual reference (VR) methods were compared with the high N (HN) reference strip method for normalizing high-resolution sensor data. The VR methods resulted in significantly lower sufficiency index values than those generated by the HN reference, resulting in N fertilization recommendations that were 31.4 ± 10.3 kg ha⁻¹ higher than the HN reference N fertilization recommendation. The use of small HN reference blocks in contrasting management zones may be more appropriate to translate field-scale, high-resolution imagery into in-season N recommendations. In view of a growing interest in using UAVs in commercial fields and the need to improve crop NUE, further work is needed to refine approaches for translating imagery into in-season N recommendations.

Digital agriculture (DA) can contribute solutions to meet an increase in healthy, nutritious, and affordable food demands in an efficient and sustainable way. South America (SA) is one of the main grain and protein producers in the world but the status of DA in the region is unknown. A systematic review and case studies from Brazil, Argentina, Uruguay, and Chile were conducted to address the following objectives: (1) quantify adoption of existing DA technologies, (2) identify limitations for DA adoption; and (3) summarise existing metrics to benchmark DA benefits. Level of DA adoption was led by Brazil and Argentina followed by Uruguay and at a slower rate, Chile. GPS guidance systems, mapping tools, mobile apps and remote sensing were the most adopted DA technologies in SA. The most reported limitations to adoption were technology cost, lack of training, limited number of companies providing services, and unclear benefits from DA. Across the case studies, there was no clear definition of DA. To mitigate some of these limitations, our findings suggest the need for a DA educational curriculum that can fulfill the demand for job skills such as data processing, analysis and interpretation. Regional efforts are needed to standardise these metrics. This will allow stakeholders to design targeted initiatives to promote DA towards sustainability of food production in the region.

Applying at the economic optimal nitrogen rate (EONR) has the potential to increase nitrogen (N) fertilization efficiency and profits while reducing negative environmental impacts. On-farm precision experimentation (OFPE) provides the opportunity to collect large amounts of data to estimate the EONR. Machine learning (ML) methods such as generalized additive models (GAM) and random forest (RF) are promising methods for estimating yields and EONR. Twenty OFPE N trials in wheat and barley were conducted and analyzed with soil, terrain and remote-sensed variables to address the following objectives: (1) to quantify the spatial variability of winter crops yield and the yield response to N using OFPE, (2) to evaluate and compare the performance of GAM and RF models to predict yield and yield response to N and, (3) to quantify the impact of soil, crop and field characteristics on the EONR estimation. Machine learning techniques were able to model wheat and barley yield with an average error of 13.7% (624 kg ha⁻¹). However, similar yield prediction accuracy from RF and GAM resulted in widely different economic optimal nitrogen rates. Across sites, soil available phosphorus and soil organic matter were the most influential variables; however, the magnitude and direction of the effect varied between fields. These indicate that training a model using data coming from different fields may lead to unreliable site-specific EONR when it is applied to another field. Further evaluation of ML methods is needed to ensure a robust automation of N recommendation while producers transition into the digital ag era.

Maize breeding programs have indirectly altered many plant traits; however, our knowledge of some important phenological traits remains unexplored. One such trait is leaf appearance rate, which is crucial for predicting maize development. We studied 40 short‐season (103‐day) and 38 long‐season (111‐day) hybrids released from 1980 to 2020 by Bayer Crop Science. Measurements included weekly counting of collared leaves across 13 experiments in the US Corn Belt. The progression of leaf number was expressed as a function of thermal time and described with a trilinear model. Results indicated that new 111‐day hybrids produce leaves faster than old hybrids throughout the vegetative phase (7.4% and 3.1% faster before and after the ninth leaf stage, respectively), whereas new 103‐day hybrids produce leaves faster only after the ninth leaf stage (9.4%). Thermal time to silking and anthesis decreased by about 1 and 0.56°C day year⁻¹, respectively. Our data revealed that silking and anthesis can precede the final collared leaf by 96°C day (3.3 days under optimal conditions), which indicates an overlap between vegetative and reproductive phases. We concluded that maize breeding has indirectly altered the rate of vegetative development of maize hybrids without affecting the final leaf number. Present results expand our knowledge base on the genotypic variability in maize development traits, which can improve empirical and process‐based models used for crop stage and yield prediction.