Crop Nitrogen Status Research Articles

Predicting leaf nitrogen content of coffee trees using the canopy hyperspectral reflectance feature bands, vegetation index and machine learning

ABSTRACT Leaf nitrogen content (LNC) is an essential indicator of crop nitrogen status. To rapidly and correctly estimate the LNC using hyperspectral remote sensing, the canopy hyperspectral reflectance of coffee trees treated with five levels of nitrogen fertilization in a greenhouse was obtained in this study. Five methods were used for hyperspectral data preprocessing, namely, Savitzky–Golay (SG) smoothing, a combination of SG and standard normal variate transformation (SG-SNV), a combination of SG and first-order derivative (SG-FD), a combination of SG and second-order derivative (SG-SD), and a combination of SG and multiplicative scatter correction (SG-MSC). Feature wavelengths were extracted using variables combination population analysis (VCPA), competitive adaptive reweighted sampling (CARS), and the combination (CARS-SPA) of CARS and successive projections algorithm (SPA). Vegetation indexes (VIs) were constructed and subjected to correlation analysis and variance inflation factor (VIF) analysis. Linear and nonlinear models including partial least squares regression (PLSR), back propagation neural network (BPNN), extreme learning machine (ELM), random forest regression (RFR), and support vector regression (SVR), were adopted to construct LNC retrieval models for coffee trees. The results indicated that SG-MSC could increase the signal-to-noise ratio of hyperspectral data well. The wavelengths selected by CARS-SPA were more relevant to LNC, and combined with ELM resulted in the best performance of LNC prediction (R2 P = 0.901, RMSEP = 0.825 g·kg−1, RPD = 3.229). Ten VIs were obtained through correlation analysis and VIF, and the VIs-based ELM prediction model also performed moderately well (R2 P = 0.814, RMSEP = 1.131 g·kg−1, RPD = 2.354). By comparing the coffee LNC prediction models established by different methods, two coffee LNC inversion models with better prediction accuracy were obtained, which provide a scientific basis for accurate diagnosis of coffee trees LNC, and are of great significance for optimizing the field management.

In-season dynamic diagnosis of maize nitrogen status across the growing season by integrating proximal sensing and crop growth modeling

Efficient and accurate in-season diagnosis of crop nitrogen (N) status is crucially important for precision N management. The main objective of this study was to develop a strategy for in-season dynamic diagnosis of maize (Zea mays L.) N status across the growing season by integrating proximal sensing and crop growth modeling. In this study, we integrated plant N concentration (PNC) derived from leaf fluorescence sensor data and aboveground biomass (AGB) based on the best-performing spectral index calculated from active canopy reflectance sensor data with simulated PNC and AGB using a crop growth model, DSSAT-CERES-Maize, for dynamic in-season maize N status diagnosis across the growing season. The results confirmed the applicability of leaf fluorescence sensing for PNC estimation and active canopy reflectance sensing for AGB estimation, respectively. The calibrated DSSAT CERES-Maize model performed well for simulating AGB (R2 = 0.96), which could be used for calculating the N status indicator, N nutrition index (NNI). However, the model did not perform satisfactorily for PNC simulation, with significant discrepancies between the simulated and measured PNC values. The data integration method using both proximal sensing and crop growth modeling produced accurate predictions of NNI (R2 = 0.95) and N status diagnostic outcomes (Kappa statistics = 0.64) for key growth stages in this study and could be used to simulate maize N status across the growing season, showing the potential for in-season dynamic N status diagnosis and management decision support. More studies are needed to further improve this approach by multi-sensor and multi-source data fusion using machine learning models.

Cultivar effect on proximal optical sensor measurements and estimation of leaf N content in muskmelon and sweet pepper

In intensive vegetable crops, there is a requirement for monitoring tools that assess crop nitrogen (N) status to reduce N use and the potential for nitrate (NO3-) leaching loss. Monitoring with proximal optical sensors can regularly provide information to vegetable growers to assist with optimal crop N management. Reference values for proximal optical sensor measurements that indicate crop N sufficiency or deficiency are increasingly available in the literature. Available values are species specific; however, their use may be problematic if proximal optical sensor measurements are affected by cultivar. Little is known on whether differences between cultivars, within a species, affect optical sensor measurements and their relationships with crop N status, particularly in vegetable crops. This study evaluated the effect of cultivar on a) measurements with the SPAD-502 chlorophyll meter, Normalized Difference Vegetation Index (NDVI) measured with the Crop Circle ACS 470 canopy reflectance sensor, and Nitrogen Balance Index (NBI-R) measured with the Multiplex fluorimeter, and b) the relationships of these measurements with leaf N content. The study was conducted with four cultivars of muskmelon (Tezac, Bosito, Magiar and Jacobo) and three of sweet pepper (Melchor, Machado and CLX PLRJ 731). In both muskmelon and sweet pepper, there were significant differences between cultivars in SPAD and in NBI-R. The range of differences between the cultivars was 2.2−9.7 and 0.6−2.8 SPAD units in muskmelon and sweet pepper, respectively, and 0−0.16 and 0.020.08 NBI-R units in muskmelon and sweet pepper, respectively. Significant differences between cultivars in the NDVI were only obtained in muskmelon; the range of differences between the four cultivars was 0.01–0.04 NDVI units. The relatively large cultivar effect with SPAD and NBI-R values with muskmelon may limit the use of reference values in this species. For avoiding cultivar effects in the two species evaluated, NDVI is preferrable to SPAD and NBI-R. In general, the slopes and intercepts of linear relationships between optical sensor measurements and leaf N content, for each cultivar of each species, were not significantly different to the slope and intercept of the relationship calculated with pooled data of all cultivars of each species. This indicated that the sensitivity of optical sensors to assess leaf N content was similar between cultivars of each species.

Improving Wheat Leaf Nitrogen Concentration (LNC) Estimation across Multiple Growth Stages Using Feature Combination Indices (FCIs) from UAV Multispectral Imagery

Leaf nitrogen concentration (LNC) is a primary indicator of crop nitrogen status, closely related to the growth and development dynamics of crops. Accurate and efficient monitoring of LNC is significant for precision field crop management and enhancing crop productivity. However, the biochemical properties and canopy structure of wheat change across different growth stages, leading to variations in spectral responses that significantly impact the estimation of wheat LNC. This study aims to investigate the construction of feature combination indices (FCIs) sensitive to LNC across multiple wheat growth stages, using remote sensing data to develop an LNC estimation model that is suitable for multiple growth stages. The research employs UAV multispectral remote sensing technology to acquire canopy imagery of wheat during the early (Jointing stage and Booting stage) and late (Early filling and Late filling stages) in 2021 and 2022, extracting spectral band reflectance and texture metrics. Initially, twelve sensitive spectral feature combination indices (SFCIs) were constructed using spectral band information. Subsequently, sensitive texture feature combination indices (TFCIs) were created using texture metrics as an alternative to spectral bands. Machine learning algorithms, including partial least squares regression (PLSR), random forest regression (RFR), support vector regression (SVR), and Gaussian process regression (GPR), were used to integrate spectral and texture information, enhancing the estimation performance of wheat LNC across growth stages. Results show that the combination of Red, Red edge, and Near-infrared bands, along with texture metrics such as Mean, Correlation, Contrast, and Dissimilarity, has significant potential for LNC estimation. The constructed SFCIs and TFCIs both enhanced the responsiveness to LNC across multiple growth stages. Additionally, a sensitive index, the Modified Vegetation Index (MVI), demonstrated significant improvement over NDVI, correcting the over-saturation concerns of NDVI in time-series analysis and displaying outstanding potential for LNC estimation. Spectral information outperforms texture information in estimation capability, and their integration, particularly with SVR, achieves the highest precision (coefficient of determination (R2) = 0.786, root mean square error (RMSE) = 0.589%, and relative prediction deviation (RPD) = 2.162). In conclusion, the sensitive FCIs developed in this study improve LNC estimation performance across multiple growth stages, enabling precise monitoring of wheat LNC. This research provides insights and technical support for the construction of sensitive indices and the precise management of nitrogen nutrition status in field crops.

Combining spectrum, thermal, and texture features using machine learning algorithms for wheat nitrogen nutrient index estimation and model transferability analysis

The nitrogen nutrition index (NNI) has been extensively applied for the diagnosis of crop nitrogen status, providing insights into efficient nitrogen utilization and plant growth. In this study, we utilized a low-altitude unmanned aerial vehicle (UAV) platform, equipped with multispectral (MS), red–green–blue (RGB), and thermal infrared (TIR) cameras, to comprehensively capture wheat spectral information. The analysis of the relationship between NNI and relative yield revealed an initially linear relationship, which saturated for high NNI values. To enhance accuracy and minimize complexity, we employed a random forest (RF) – recursive feature elimination (RFE) method to select features as inputs for four machine learning (ML) models: back propagation neural network (BPNN), extreme learning machine (ELM), support vector regression (SVR), and Gaussian process regression (GPR). After feature selection, the prediction accuracies of single-sensor models were ranked as: MS > RGB > TIR. The R2 values for the four ML models were in the range of 0.54–0.75. Among multi-sensor combinations, the GPR with MS + RGB + TIR input features achieved the best results with R2 = 0.89 and RPD = 2.52. Further, the dataset was partitioned into six subsets based on location and cultivar variety to evaluate model transferability. The results showed that the transferability largely suffered during the bivariate conditions of different varieties at different locations; the transferability of the model was average improved by 11 % when GPR was combined with transfer component analysis (TCA). The accuracy and transferability of the NNI estimation models significantly improved, offering valuable guidance and methodological support for diagnosing the nitrogen nutrient status of wheat.

UAV-based multispectral image analytics and machine learning for predicting crop nitrogen in rice

Assessment of crop nitrogen status is essential for efficient crop growth management. Existing nitrogen measurements are accurate but destructive, laborious, and time-consuming. Therefore, the soil plant analysis development (SPAD) meter approach is commonly used to address these challenges along with location-specific measurements. The study aims to develop a robust machine learning-based model for predicting rice crop SPAD values using spectral data and to generate spatial maps of SPAD values and nitrogen content. The SPAD meter data, UAV-based multispectral images, and spectroradiometer-based data were collected during Rabi 2021/22 and 2022/23 seasons. The red and red-edge bands, Normalized Difference Vegetation Index, and Normalized Pigment Chlorophyll Index correlated well with SPAD values. The random forest regressor model performed well with UAV-based data compared to support vector regression and partial least square regression and achieved good accuracy with the ground truth spectroradiometer data. This generalized model demonstrates adaptability in precisely assessing crop nitrogen status.

Optimizing Nitrogen Sources in Top Dressing for Wheat: Field Study on Growth, Yield, and Ammonia Volatilization.

In alkaline calcareous soils, ammonia volatilization is the primary nitrogen (N) loss process, resulting in the reduced N use efficiency of crops. This study aimed at assessing the impact of different N sources for top dressing on ammonia volatilization, as well as their effects on wheat growth and yield over two years. In each year, half of the recommended N was applied as a basal dose using diammonium phosphate (DAP) and urea. The remaining half was top-dressed 35 days after sowing with various sources: prilled urea (PU), granular urea (GU), ammonium sulfate (AS), and calcium ammonium nitrate (CAN) in the first year; PU, urea coated with a urease inhibitor from 20 g (VnU-20) and 40 g (VnU-40) leaves of Vachellia nilotica, biochar-coated urea (BU), and urease inhibitor paraphenylenediamine-coated urea (PPDU) in the second year. Ammonia volatilization losses were tracked for up to 12 weeks from sowing. Ammonia losses from basal-applied N remained consistent in both years, comprising around 4% of the applied N. In the first year, top-dressed AS resulted in the highest losses, followed by GU, while losses from urea and CAN were statistically similar. In the second year, coated fertilizers showed lower ammonia losses compared to PU, with VnU-40 displaying the least losses, 48% less than PU. Nitrogen concentration in wheat grain and straw exhibited a negative correlation with ammonia losses. The choice of top-dressed N source influenced tillering, biological, straw, and grain yields of wheat. In the first year, CAN provided maximum yield benefits, and in the second year, VnU-20 exhibited 27% more grain yield than PU. These findings suggest that top dressing with coated urea, especially VnU-20, has the potential to reduce ammonia losses, improve crop nitrogen status, and enhance economic yield compared to other nitrogen sources.

Synthetic data generation methods for training neural networks in the task of segmenting the level of crop nitrogen status on UAV images of agricultural fields

This study is devoted to the automatization of the image masks’ construction of large sized agricultural objects in precision farming tasks for training neural network methods for crop’s nitrogen status analysis using georeferenced images. The scientific direction is extremely relevant because it allows to automate and replace the manual process of data labeling, significantly reducing the cost of preparing training samples. In the paper, four new synthetic data generation methods are proposed for training neural networks aimed at UAV image segmentation by the level of crop nitrogen supply on an agricultural field. In particular, the paper gives a description of synthetic data generation algorithms based on nitrogen covering with lines, parabolas, and areas. Experiments were carried out to test and evaluate the quality of these algorithms using eight modern image segmentation methods: two classical machine learning methods (Random Forest and XGBoost), four convolutional neural network methods based on U-Net architecture, and two transformers (TransUnet and UnetR). The results showed that two algorithms based on areas gave the best accuracy for convolutional neural networks and transformers — 98–100 %. Classical machine learning methods showed very low values for all quality metrics — 27–44 %.

Determining nitrogen topdressing rates in accordance with actual seedling density and crop nitrogen status in direct-seeded rice.

Adjusting nitrogen (N) input based on actual seedling density (ASD) and plant N status is a practical approach for improving the yield stability of direct-seeded rice. However, the adjustment of topdressing N rates has been empirical in the past. This study aimed to establish a quantitative approach for determining N topdressing rates during tillering (Ntil ) and panicle development (NPI ) based on ASD and crop N status in direct-seeded rice. Field experiments were conducted involving 12 treatments, consisting of four Ntil and three seeding rates in 2017, and eight treatments combining seeding rate, Ntil , and NPI in 2020. Linear regression analysis revealed that the tiller number at panicle initiation (TILPI ) was predominantly influenced by ASD and Ntil . The determination coefficients (R2 ) of the regression models ranged from 0.887 to 0.936 across the four-season experiments. The results indicated that Ntil could be determined accurately using ASD and the target maximum tiller number. Similarly, grain yield was influenced significantly by the N uptake at panicle initiation (NUPPI ) and NPI , with R2 of 0.814 and 0.783 in the early and late seasons of 2020, respectively. This suggested that NPI could be calculated based on NUPPI and the target grain yield. The findings offer a quantitative method for establishing N topdressing rates for tillering and panicle development, relying on the monitoring of actual seedling density and plant N status in direct-seeded rice production. © 2023 Society of Chemical Industry.

Incremental learning for crop growth parameters estimation and nitrogen diagnosis from hyperspectral data

Nitrogen is an essential nutrient in crop growth cycle and directly affects the photosynthesis of crops. The leaf chlorophyll content (LCC) and leaf area index (LAI) are important for characterizing photosynthetic capacity of crops and are critical indicators for diagnosing nitrogen status of crops. Hyperspectral remote sensing technology provided a means to achieve crop LCC and LAI estimation. However, the redundancy of spectral data and canopy structure effect can cause poor robustness of the estimating models, further hindering the development and application of estimating models across different crop species. In this study, a method based on incremental learning was proposed for the simultaneous estimation of LCC and LAI, and for nitrogen diagnosis across crops. First, for the spectral dataset generated by the PROSAIL model, a deep neural network was used to construct LCC and LAI estimation models (called DNNCA model). Secondly, for the hyperspectral data collected from field crops (soybean, canola and wheat), incremental learning using regularization (LwF algorithm) was used to update the DNNCA model parameters. Finally, the dilution curve model based on the LCC-LAI anisotropic growth relationship was developed to assess crop nitrogen status. The results showed that: (1) The constrained bi-objective optimizated DNNCA model can consider the interactive effect of LAI and LCC on spectral reflectance, and achieved reliable estimation on PROSAIL simulation data set (LAI:R2 = 0.82, RMSE = 0.77 m2/m2; LCC:R2 = 0.91, RMSE = 6.4 ug/cm2). (2) By incremental learning, DNNCA model has continuous learning capability and stable estimation on cross-crop (canola, soybean and wheat) field-measured data (LAI:R2 = 0.64–0.82, RMSE = 0.58–1.02 m2/m2; LCC:R2 = 0.56–0.82, RMSE = 3.9–10.5 ug/cm2). (3) The relationship between the NNILCC and the NNILNC was significant. The NNILCC derived from the anisotropy relationship between crop LAI and LCC was an effective tool for crop nitrogen status diagnosis. (4) The process of crop LAI, LCC and NNILCC reflected the effect of water and nitrogen supply on crop growth. The appropriate water-nitrogen treatments contributed to LAI increase and LCC accumulation. The study demonstrated that hyperspectral remote sensing technology combined with incremental learning is an effective method for cross-crop growth monitoring and nitrogen diagnosis. These results provide a reference and basis for filed water-nitrogen supply and management.

Correction to: In-Season Potato Crop Nitrogen Status Assessment from Satellite and Meteorological Data

Correction to: In-Season Potato Crop Nitrogen Status Assessment from Satellite and Meteorological Data

Toward potential applications of the nitrogen nutrition index in Sweet basil (Ocimum basilicum L.) production under greenhouse conditions

The nitrogen nutrition index (NNI) as an effective indicator for monitoring crop nitrogen (N) status has multiple applications in agriculture, such as estimating in-season crop N requirement (NR), predicting crop biomass/yield, and improving N use efficiency. The present study was aimed to develop basil (Ocimum basilicum L.) NR and relative biomass (RB) prediction models based on the concept of critical N concentration. Two experiments were carried out with different N application rates in the research greenhouse of University of Tehran, Iran. Basil NNI values were calculated using the previously developed critical N dilution curve. Subsequently, the in-season basil NR estimation model was developed as a function of NNI, N recovery efficiency (NRE) and plant age. Additionally, basil RB was simulated as a function of the averaged and integrated NNI based on a linear-plateau model. Moreover, a theoretical relationship between N utilization efficiency (NUtE) and the NNI was established. The validation results of the developed NR-NNI model demonstrated a good performance with a root mean square error (RMSE) and a normalized root mean square error (NRMSE) less than 5.28 kg ha−1 and 4.72%, respectively. The results showed that crop biomass and NNI were the most significant explanatory factors for the variations in basil NUtE. Therefore, the crop NNI can serve as a practical in-season diagnostic tool for basil NR and RB prediction, while also contributing to the design of decision support systems and sustainable agro-ecosystems.

Potato Crop Nitrogen Status Monitoring for Sustainable N Fertilisation Management: Last 15 Years and Future-Expected Developments with Reference Method and Use of Optical Sensors

Potato Crop Nitrogen Status Monitoring for Sustainable N Fertilisation Management: Last 15 Years and Future-Expected Developments with Reference Method and Use of Optical Sensors

Tracing the nitrogen nutrient status of crop based on solar-induced chlorophyll fluorescence

Accurate and non-destructive monitoring of wheat nitrogen nutrition is of great significance for field fertilizer management to ensure crop yield and quality, reduce environmental pollution, and improve economic benefits. Compared with spectral vegetation indices (which are sensitive to greenness and structural parameters), or active fluorescence (which is limited to small-scale studies), solar-induced chlorophyll fluorescence (SIF) provides a direct measure of crop response to environmental stress and photosynthetic characteristics. However, there has been few studies comparing agronomic parameters, photosynthetic parameters, vegetation indices and SIF as an indicator of nitrogen status. In this paper, we therefore explore these measures as tools for monitoring nitrogen nutrition. During the 2016–2017 growing season, we conducted a field experiment in Rugao, Jiangsu Province, China, using winter wheat (Triticum aestivum L.) and different nitrogen treatments. The sensitivity of SIF indices, vegetation indices, photosynthetic parameters and agronomic parameters to crop nitrogen status were compared. Our results demonstrated that, compared with vegetation indices and agronomic parameters, the ratio of SIF emission peaks (FY687/FY761) responded to nitrogen status most rapidly at both the leaf and canopy scales, as soon as the fourth day after treatment (DAT4). A wheat nitrogen nutrition index (NNI), based on FY687/FY761, was used to construct a leaf dry matter (LDM-based NNI) diagnostic model, which will be beneficial for monitoring and diagnosing the nitrogen nutrition status of wheat leaves. Our results also illuminate the physiological mechanism that enables SIF to be used as a tool to monitor nitrogen nutrient status, primarily through changes in the proportion of light energy distribution. These findings provide theoretical and technical support for monitoring and diagnosing wheat nitrogen nutrition status based on SIF technology.

Improving Nitrogen Status Diagnosis and Recommendation of Maize Using UAV Remote Sensing Data

Effective in-season crop nitrogen (N) status diagnosis is important for precision crop N management, and remote sensing using an unmanned aerial vehicle (UAV) is one efficient means of conducting crop N nutrient diagnosis. Here, field experiments were conducted with six N levels and six maize hybrids to determine the nitrogen nutrition index (NNI) and yield, and to diagnose the N status of the hybrids combined with multi-spectral data. The NNI threshold values varied with hybrids and years, ranging from 0.99 to 1.17 in 2018 and 0.60 to 0.71 in 2019. A proper agronomic optimal N rate (AONR) was constructed and confirmed based on the measured NNI and yield. The NNI (R2 = 0.64–0.79) and grain yield (R2 = 0.70–0.73) were predicted well across hybrids using a random forest model with spectral, structural, and textural data (UAV). The AONRs calculated using the predicted NNI and yield were significantly correlated with the measured NNI (R2 = 0.70 and 0.71 in 2018 and 2019, respectively) and yield (R2 = 0.68 and 0.54 in 2018 and 2019, respectively). It is concluded that data fusion can improve in-season N status diagnosis for different maize hybrids compared to using only spectral data.

Monitoring leaf nitrogen content in rice based on information fusion of multi-sensor imagery from UAV

Timely and accurately monitoring leaf nitrogen content (LNC) is essential for evaluating crop nutrition status. Currently, Unmanned Aerial Vehicles (UAV) imagery is becoming a potentially powerful tool of assessing crop nitrogen status in fields, but most of crop nitrogen estimates based on UAV remote sensing usually use single type imagery, the fusion information from different types of imagery has rarely been considered. In this study, the fusion images were firstly made from the simultaneously acquired digital RGB and multi-spectral images from UAV at three growth stages of rice, and then couple the selecting methods of optimal features with machine learning algorithms for the fusion images to estimate LNC in rice. Results showed that the combination with different types of features could improve the models’ accuracy effectively, the combined inputs with bands, vegetation indices (VIs) and Grey Level Co-occurrence Matrices (GLCMs) have the better performance. The LNC estimation of using fusion images was improved more obviously than multispectral those, and there was the best estimation at jointing stage based on Lasso Regression (LR), with R2 of 0.66 and RMSE of 11.96%. Gaussian Process Regression (GPR) algorithm used in combination with one feature-screening method of Minimum Redundancy Maximum Correlation (mRMR) for the fusion images, showed the better improvement to LNC estimation, with R2 of 0.68 and RMSE of 11.45%. It indicates that the information fusion from UAV multi-sensor imagery can significantly improve crop LNC estimates and the combination with multiple types of features also has a great potential for evaluating LNC in crops.

Assessment of Nitrogen Nutrition Index of Winter Wheat Canopy from Visible Images for a Dynamic Monitoring of N Requirements

Hand-held chlorophyll meters or leaf-clip-type sensors indirectly and instantaneously measure leaf N content. They can provide an N nutrition index (NNI) value that is crucial information for adjusting the amount of N fertilizer to the actual N status of the plant. Although these measurements are non-invasive and non-destructive, they require numerous repetitions at the canopy scale. The objective of this work was to explore the potential of visible images to predict nitrogen status in winter wheat crops from estimating NNI and to compare these results with those deduced from classical methods. Based on a dark green colour index (DGCI), which combines hue, saturation and brightness, a normalized DGCI (nDGCI) was proposed as the ratio between the measurements of the study microplot and those of the over-fertilized microplot. The methodology was performed on winter wheat microplots with a nitrogen gradient. Half of the microplots were grown with a single cultivar (LG Absalon) and the other half with a mixture of four wheat cultivars. The impact of optical device (digital camera or smartphone), the white balance (Manual or Automatic), the crop growth stage (two-nodes or heading) and cultivars (single or mixed) on the relationship between (DGCI, nDGCI) and NNI was evaluated. The results showed a close correlation between the nDGCI values and the NNI_NTester values, especially on a single cultivar (LG Absalon; R2 = 0.73 up to 0.91 with smartphone). It suggested that the relationship is highly sensitive to the wheat cultivar. This approach with no specific calibration of images is promising for the estimation of N requirements in wheat field.

Combining a multi-environment trial and a diagnosis method to assess potential yield and main limiting factors of three highly different pea types

The evaluation of varieties of cultivated species is based on the implementation of multi-environmental trials that show high performances or strong limitations in some environments. To know the cause of these limitations, and to reason the adaptation of varieties to different environments, we propose a combined approach between the classical analysis of experimental results and a diagnostic approach using the DiagVar tool, which allows to assess the impact of limiting factors in each trial, the potential yield and the sensitivity of cultivars to those factors. This procedure is applied in a large and diversified field trial network including eight locations and three years to the comparison of two to four cultivars of the three agronomical pea types: spring peas, hr-winter peas (non-reactive to photoperiod) and Hr-winter peas (reactive to photoperiod).Estimates of potential yields from DiagVar for the three pea types were consistent with the experimental results, with higher values for hr-winter cultivars than for spring and Hr-winter ones. We shed light on specific limiting factors for each type, some of them being rarely quantified previously, which allows us to better target specific regions or environments for each pea type. Thus, spring types were more impacted by early limiting factors (loss of plants, low temperatures and radiation, lack of water, leading to reduced growth) and end-of-cycle stresses (high temperatures, water stress). Diseases had higher impacts on both winter types. Here we show for the first time that the lack of solar radiation affected particularly hr-winter type and the low crop nitrogen status affected more frequently spring peas. These results highlight the interest of carrying out an agronomic diagnosis in the analysis of varietal multi-environment trials, revealing the most important limiting factors to take into account while breeding new pea cultivars, particularly limiting factors that are difficult to directly observe. This method proves to be reliable and informative for breeders, technicians and agricultural advisors to discuss the adaptation of varieties to specific environment.

Estimation of Winter Wheat Canopy Chlorophyll Content Based on Canopy Spectral Transformation and Machine Learning Method

Canopy chlorophyll content (CCC) is closely related to crop nitrogen status, crop growth and productivity, detection of diseases and pests, and final yield. Thus, accurate monitoring of chlorophyll content in crops is of great significance for decision support in precision agriculture. In this study, winter wheat in the Guanzhong Plain area of the Shaanxi Province, China, was selected as the research subject to explore the feasibility of canopy spectral transformation (CST) combined with a machine learning method to estimate CCC. A hyperspectral canopy ground dataset in situ was measured to construct CCC prediction models for winter wheat over three growth seasons from 2014 to 2017. Sensitive-band reflectance (SR) and narrow-band spectral index (NSI) were established based on the original spectrum (OS) and CSTs, including the first derivative spectrum (FDS) and continuum removal spectrum (CRS). Winter wheat CCC estimation models were constructed using univariate regression, partial least squares (PLS) regression, and random forest (RF) regression based on SR and NSI. The results demonstrated the reliability of CST combined with the machine learning method to estimate winter wheat CCC. First, compared with OS-SR (683 nm), FDS-SR (630 nm) and CRS-SR (699 nm) had a larger correlation coefficient between canopy reflectance and CCC; secondly, among the parametric regression methods, the univariate regression method with CRS-NDSI as the independent variable achieved satisfactory results in estimating the CCC of winter wheat; thirdly, as a machine learning regression method, RF regression combined with multiple independent variables had the best winter wheat CCC estimation accuracy (the determination coefficient of the validation set (Rv2) was 0.88, the RMSE of the validation set (RMSEv) was 3.35 and relative prediction deviation (RPD) was 2.88). Thus, this modeling method could be used as a basic method to predict the CCC of winter wheat in the Guanzhong Plain area.

Comparison of statistical methods to fit critical nitrogen dilution curves

The critical nitrogen dilution curve (CNDC) is one of the most important models underpinning diagnostic tools of crop nitrogen (N) status. The CNDC relates plant N content (%N) to plant biomass under optimal nutritional conditions, using a simple nonlinear function with two parameters. More precisely, the CNDC estimates the critical %N (%NC) - the amount of %N needed to achieve maximum biomass at a certain moment. Thus, the comparison of a measured %N to %NC diagnoses if a plant is N deficient and if application of N fertilizer is necessary. Because of the practical importance of the CNDC, its parameters should be estimated as accurately as possible from field data. Two contrasting frameworks have been proposed in the literature for parameter estimation of the CNDC. The most popular approach is a sequential framework that consists of fitting two statistical models sequentially. Recently, an alternative framework has been introduced to estimate the parameters in a single step, using a hierarchical Bayesian model (hierarchical framework). In this study, we compare both methods and evaluate their performances for a large range of experimental designs. We consider datasets with different numbers of observation times (4, 8, 16), N fertilizer rates (3, 4, 5), and levels of measurement accuracy (low, medium, high). Our results show that the hierarchical framework outperforms the sequential framework under most scenarios. The sequential framework often resulted in poor statistical properties of the parameter estimates (e.g., increased bias and poorly calibrated confidence intervals) and to overestimated %NC values. The estimates for the amount of N needed to sustain N sufficiency levels were between 30 and 100 kg N ha−1 greater under the sequential framework, compared to the hierarchical framework. Thus, adopting the hierarchical framework could contribute to decreasing the risk of overfertilization and increasing the sustainability of agricultural systems.