Effects of fine grinding on mid-infrared spectroscopic analysis of plant leaf nutrient content.

Fourier transform mid infrared (FT-MIR) spectroscopy combined with modeling techniques has been studied as a useful tool for multivariate chemical analysis in agricultural research. A drawback of this method is the sample preparation requirement, in which samples must be dried and fine ground for accurate model calibrations. For research involving large sample sets, this may dramatically increase the time and cost of analysis. This study investigates the effect of fine grinding on model performance using leaf tissue from a variety of crop species. Dried leaf samples (N = 300) from various environmental conditions were obtained with data on 11 nutrients measured using chemical methods. The samples were scanned with attenuated total reflectance (ATR) and diffuse reflectance (DRIFT) FT-MIR techniques. Scanning was repeated after fine grinding for 2, 5, and 10 min. The spectra were analyzed for the 11 nutrients using partial least squares regression with a 75%/25% split for calibration and validation and repeated for 50 iterations. All analytes except for boron, iron, and zinc were well-modeled (average R2 > 0.7), with higher R2 values on ATR spectra. The 5 min level of fine grinding was found to be most optimal considering overall model performance and sample preparation time.

A leaf-level spectral library to support high-throughput plant phenotyping: predictive accuracy and model transfer.

Leaf-level hyperspectral reflectance has become an effective tool for high-throughput phenotyping of plant leaf traits due to its rapid, low-cost, multi-sensing, and non-destructive nature. However, collecting samples for model calibration can still be expensive, and models show poor transferability among different datasets. This study had three specific objectives: first, to assemble a large library of leaf hyperspectral data (n=2460) from maize and sorghum; second, to evaluate two machine-learning approaches to estimate nine leaf properties (chlorophyll, thickness, water content, nitrogen, phosphorus, potassium, calcium, magnesium, and sulfur); and third, to investigate the usefulness of this spectral library for predicting external datasets (n=445) including soybean and camelina using extra-weighted spiking. Internal cross-validation showed satisfactory performance of the spectral library to estimate all nine traits (mean R2=0.688), with partial least-squares regression outperforming deep neural network models. Models calibrated solely using the spectral library showed degraded performance on external datasets (mean R2=0.159 for camelina, 0.337 for soybean). Models improved significantly when a small portion of external samples (n=20) was added to the library via extra-weighted spiking (mean R2=0.574 for camelina, 0.536 for soybean). The leaf-level spectral library greatly benefits plant physiological and biochemical phenotyping, whilst extra-weight spiking improves model transferability and extends its utility.

Resilience of soybean cultivars to drought stress during flowering and early-seed setting stages.

Drought stress during the reproductive stage and declining soybean yield potential raise concerns about yield loss and economic return. In this study, ten cultivars were characterized for 20 traits to identify reproductive stage (R1-R6) drought-tolerant soybean. Drought stress resulted in a marked reduction (17%) in pollen germination. The reduced stomatal conductance coupled with high canopy temperature resulted in reduced seed number (45%) and seed weight (35%). Drought stress followed by rehydration increased the hundred seed weight at the compensation of seed number. Further, soybean oil decreased, protein increased, and cultivars responded differently under drought compared to control. In general, cultivars with high tolerance scores for yield displayed lower tolerance scores for quality content and vice versa. Among ten cultivars, LS5009XS and G4620RX showed maximum stress tolerance scores for seed number and seed weight. The observed variability in leaf reflectance properties and their relationship with physiological or yield components suggested that leaf-level sensing information can be used for differentiating drought-sensitive soybean cultivars from tolerant ones. The study led to the identification of drought-resilient cultivars/promising traits which can be exploited in breeding to develop multi-stress tolerant cultivars.

Variation in morpho-physiological and metabolic responses to low nitrogen stress across the sorghum association panel.

BACKGROUND: Access to biologically available nitrogen is a key constraint on plant growth in both natural and agricultural settings. Variation in tolerance to nitrogen deficit stress and productivity in nitrogen limited conditions exists both within and between plant species. However, our understanding of changes in different phenotypes under long term low nitrogen stress and their impact on important agronomic traits, such as yield, is still limited. RESULTS: Here we quantified variation in the metabolic, physiological, and morphological responses of a sorghum association panel assembled to represent global genetic diversity to long term, nitrogen deficit stress and the relationship of these responses to grain yield under both conditions. Grain yield exhibits substantial genotype by environment interaction while many other morphological and physiological traits exhibited consistent responses to nitrogen stress across the population. Large scale nontargeted metabolic profiling for a subset of lines in both conditions identified a range of metabolic responses to long term nitrogen deficit stress. Several metabolites were associated with yield under high and low nitrogen conditions. CONCLUSION: Our results highlight that grain yield in sorghum, unlike many morpho-physiological traits, exhibits substantial variability of genotype specific responses to long term low severity nitrogen deficit stress. Metabolic response to long term nitrogen stress shown higher proportion of variability explained by genotype specific responses than did morpho-pysiological traits and several metabolites were correlated with yield. This suggest, that it might be possible to build predictive models using metabolite abundance to estimate which sorghum genotypes will exhibit greater or lesser decreases in yield in response to nitrogen deficit, however further research needs to be done to evaluate such model.

High throughput analysis of leaf chlorophyll content in sorghum using RGB, hyperspectral, and fluorescence imaging and sensor fusion.

BACKGROUND: Leaf chlorophyll content plays an important role in indicating plant stresses and nutrient status. Traditional approaches for the quantification of chlorophyll content mainly include acetone ethanol extraction, spectrophotometry and high-performance liquid chromatography. Such destructive methods based on laboratory procedures are time consuming, expensive, and not suitable for high-throughput analysis. High throughput imaging techniques are now widely used for non-destructive analysis of plant phenotypic traits. In this study three imaging modules (RGB, hyperspectral, and fluorescence imaging) were, separately and in combination, used to estimate chlorophyll content of sorghum plants in a greenhouse environment. Color features, spectral indices, and chlorophyll fluorescence intensity were extracted from these three types of images, and multiple linear regression models and PLSR (partial least squares regression) models were built to predict leaf chlorophyll content (measured by a handheld leaf chlorophyll meter) from the image features. RESULTS: The models with a single color feature from RGB images predicted chlorophyll content with R2 ranging from 0.67 to 0.88. The models using the three spectral indices extracted from hyperspectral images (Ration Vegetation Index, Normalized Difference Vegetation Index, and Modified Chlorophyll Absorption Ratio Index) predicted chlorophyll content with R2 ranging from 0.77 to 0.78. The model using the fluorescence intensity extracted from fluorescence images predicted chlorophyll content with R2 of 0.79. The PLSR model that involved all the image features extracted from the three different imaging modules exhibited the best performance for predicting chlorophyll content, with R2 of 0.90. It was also found that inclusion of SLW (Specific Leaf Weight) into the image-based models further improved the chlorophyll prediction accuracy. CONCLUSION: All three imaging modules (RGB, hyperspectral, and fluorescence) tested in our study alone could estimate chlorophyll content of sorghum plants reasonably well. Fusing image features from different imaging modules with PLSR modeling significantly improved the predictive performance. Image-based phenotyping could provide a rapid and non-destructive approach for estimating chlorophyll content in sorghum.

Hyperspectral reflectance-based phenotyping for quantitative genetics in crops: Progress and challenges.

Many biochemical and physiological properties of plants that are of interest to breeders and geneticists have extremely low throughput and/or can only be measured destructively. This has limited the use of information on natural variation in nutrient and metabolite abundance, as well as photosynthetic capacity in quantitative genetic contexts where it is necessary to collect data from hundreds or thousands of plants. A number of recent studies have demonstrated the potential to estimate many of these traits from hyperspectral reflectance data, primarily in ecophysiological contexts. Here, we summarize recent advances in the use of hyperspectral reflectance data for plant phenotyping, and discuss both the potential benefits and remaining challenges to its application in plant genetics contexts. The performances of previously published models in estimating six traits from hyperspectral reflectance data in maize were evaluated on new sample datasets, and the resulting predicted trait values shown to be heritable (e.g., explained by genetic factors) were estimated. The adoption of hyperspectral reflectance-based phenotyping beyond its current uses may accelerate the study of genes controlling natural variation in biochemical and physiological traits.

Predicting total petroleum hydrocarbons in field soils with Vis-NIR models developed on laboratory-constructed samples.

Accurate quantification of petroleum hydrocarbons (PHCs) is required for optimizing remedial efforts at oil spill sites. While evaluating total petroleum hydrocarbons (TPH) in soils is often conducted using costly and time-consuming laboratory methods, visible and near-infrared reflectance spectroscopy (Vis-NIR) has been proven to be a rapid and cost-effective field-based method for soil TPH quantification. This study investigated whether Vis-NIR models calibrated from laboratory-constructed PHC soil samples could be used to accurately estimate TPH concentration of field samples. To evaluate this, a laboratory sample set was constructed by mixing crude oil with uncontaminated soil samples, and two field sample sets (F1 and F2) were collected from three PHC-impacted sites. The Vis-NIR TPH models were calibrated with four different techniques (partial least squares regression, random forest, artificial neural network, and support vector regression), and two model improvement methods (spiking and spiking with extra weight) were compared. Results showed that laboratory-based Vis-NIR models could predict TPH in field sample set F1 with moderate accuracy (R2  > .53) but failed to predict TPH in field sample set F2 (R2  < .13). Both spiking and spiking with extra weight improved the prediction of TPH in both field sample sets (R2 ranged from .63 to .88, respectively); the improvement was most pronounced for F2. This study suggests that Vis-NIR models developed from laboratory-constructed PHC soil samples, spiked by a small number of field sample analyses, can be used to estimate TPH concentrations more efficiently and cost effectively compared with generating site-specific calibrations.