Machine Learning for High-Throughput Stress Phenotyping in Plants

Advances in automated and high-throughput imaging technologies have resulted in a deluge of high-resolution images and sensor data of plants. However, extracting patterns and features from this large corpus of data requires the use of machine learning (ML) tools to enable data assimilation and feature identification for stress phenotyping. Four stages of the decision cycle in plant stress phenotyping and plant breeding activities where different ML approaches can be deployed are (i) identification, (ii) classification, (iii) quantification, and (iv) prediction (ICQP). We provide here a comprehensive overview and user-friendly taxonomy of ML tools to enable the plant community to correctly and easily apply the appropriate ML tools and best-practice guidelines for various biotic and abiotic stress traits.

Conventional and Molecular Techniques from Simple Breeding to Speed Breeding in Crop Plants: Recent Advances and Future Outlook

In most crop breeding programs, the rate of yield increment is insufficient to cope with the increased food demand caused by a rapidly expanding global population. In plant breeding, the development of improved crop varieties is limited by the very long crop duration. Given the many phases of crossing, selection, and testing involved in the production of new plant varieties, it can take one or two decades to create a new cultivar. One possible way of alleviating food scarcity problems and increasing food security is to develop improved plant varieties rapidly. Traditional farming methods practiced since quite some time have decreased the genetic variability of crops. To improve agronomic traits associated with yield, quality, and resistance to biotic and abiotic stresses in crop plants, several conventional and molecular approaches have been used, including genetic selection, mutagenic breeding, somaclonal variations, whole-genome sequence-based approaches, physical maps, and functional genomic tools. However, recent advances in genome editing technology using programmable nucleases, clustered regularly interspaced short palindromic repeats (CRISPR), and CRISPR-associated (Cas) proteins have opened the door to a new plant breeding era. Therefore, to increase the efficiency of crop breeding, plant breeders and researchers around the world are using novel strategies such as speed breeding, genome editing tools, and high-throughput phenotyping. In this review, we summarize recent findings on several aspects of crop breeding to describe the evolution of plant breeding practices, from traditional to modern speed breeding combined with genome editing tools, which aim to produce crop generations with desired traits annually.

Advanced phenotyping and phenotype data analysis for the study of plant growth and development

Due to an increase in the consumption of food, feed, fuel and to meet global food security needs for the rapidly growing human population, there is a necessity to breed high yielding crops that can adapt to the future climate changes, particularly in developing countries. To solve these global challenges, novel approaches are required to identify quantitative phenotypes and to explain the genetic basis of agriculturally important traits. These advances will facilitate the screening of germplasm with high performance characteristics in resource-limited environments. Recently, plant phenomics has offered and integrated a suite of new technologies, and we are on a path to improve the description of complex plant phenotypes. High-throughput phenotyping platforms have also been developed that capture phenotype data from plants in a non-destructive manner. In this review, we discuss recent developments of high-throughput plant phenotyping infrastructure including imaging techniques and corresponding principles for phenotype data analysis.

A Direct Comparison of Remote Sensing Approaches for High-Throughput Phenotyping in Plant Breeding

Remote sensing (RS) of plant canopies permits non-intrusive, high-throughput monitoring of plant physiological characteristics. This study compared three RS approaches using a low flying UAV (unmanned aerial vehicle), with that of proximal sensing, and satellite-based imagery. Two physiological traits were considered, canopy temperature (CT) and a vegetation index (NDVI), to determine the most viable approaches for large scale crop genetic improvement. The UAV-based platform achieves plot-level resolution while measuring several hundred plots in one mission via high-resolution thermal and multispectral imagery measured at altitudes of 30-100 m. The satellite measures multispectral imagery from an altitude of 770 km. Information was compared with proximal measurements using IR thermometers and an NDVI sensor at a distance of 0.5-1 m above plots. For robust comparisons, CT and NDVI were assessed on panels of elite cultivars under irrigated and drought conditions, in different thermal regimes, and on un-adapted genetic resources under water deficit. Correlations between airborne data and yield/biomass at maturity were generally higher than equivalent proximal correlations. NDVI was derived from high-resolution satellite imagery for only larger sized plots (8.5 × 2.4 m) due to restricted pixel density. Results support use of UAV-based RS techniques for high-throughput phenotyping for both precision and efficiency.

High-throughput estimation of incident light, light interception and radiation-use efficiency of thousands of plants in a phenotyping platform

Light interception and radiation-use efficiency (RUE) are essential components of plant performance. Their genetic dissections require novel high-throughput phenotyping methods. We have developed a suite of methods to evaluate the spatial distribution of incident light, as experienced by hundreds of plants in a glasshouse, by simulating sunbeam trajectories through glasshouse structures every day of the year; the amount of light intercepted by maize (Zea mays) plants via a functional-structural model using three-dimensional (3D) reconstructions of each plant placed in a virtual scene reproducing the canopy in the glasshouse; and RUE, as the ratio of plant biomass to intercepted light. The spatial variation of direct and diffuse incident light in the glasshouse (up to 24%) was correctly predicted at the single-plant scale. Light interception largely varied between maize lines that differed in leaf angles (nearly stable between experiments) and area (highly variable between experiments). Estimated RUEs varied between maize lines, but were similar in two experiments with contrasting incident light. They closely correlated with measured gas exchanges. The methods proposed here identified reproducible traits that might be used in further field studies, thereby opening up the way for large-scale genetic analyses of the components of plant performance.

High-Throughput Phenotyping of Sorghum Plant Height Using an Unmanned Aerial Vehicle and Its Application to Genomic Prediction Modeling

Genomics-assisted breeding methods have been rapidly developed with novel technologies such as next-generation sequencing, genomic selection and genome-wide association study. However, phenotyping is still time consuming and is a serious bottleneck in genomics-assisted breeding. In this study, we established a high-throughput phenotyping system for sorghum plant height and its response to nitrogen availability; this system relies on the use of unmanned aerial vehicle (UAV) remote sensing with either an RGB or near-infrared, green and blue (NIR-GB) camera. We evaluated the potential of remote sensing to provide phenotype training data in a genomic prediction model. UAV remote sensing with the NIR-GB camera and the 50th percentile of digital surface model, which is an indicator of height, performed well. The correlation coefficient between plant height measured by UAV remote sensing (PH_UAV) and plant height measured with a ruler (PH_R) was 0.523. Because PH_UAV was overestimated (probably because of the presence of taller plants on adjacent plots), the correlation coefficient between PH_UAV and PH_R was increased to 0.678 by using one of the two replications (that with the lower PH_UAV value). Genomic prediction modeling performed well under the low-fertilization condition, probably because PH_UAV overestimation was smaller under this condition due to a lower plant height. The predicted values of PH_UAV and PH_R were highly correlated with each other (r = 0.842). This result suggests that the genomic prediction models generated with PH_UAV were almost identical and that the performance of UAV remote sensing was similar to that of traditional measurements in genomic prediction modeling. UAV remote sensing has a high potential to increase the throughput of phenotyping and decrease its cost. UAV remote sensing will be an important and indispensable tool for high-throughput genomics-assisted plant breeding.

High-Throughput Non-destructive Phenotyping of Traits that Contribute to Salinity Tolerance in Arabidopsis thaliana

Reproducible and efficient high-throughput phenotyping approaches, combined with advances in genome sequencing, are facilitating the discovery of genes affecting plant performance. Salinity tolerance is a desirable trait that can be achieved through breeding, where most have aimed at selecting for plants that perform effective ion exclusion from the shoots. To determine overall plant performance under salt stress, it is helpful to investigate several plant traits collectively in one experimental setup. Hence, we developed a quantitative phenotyping protocol using a high-throughput phenotyping system, with RGB and chlorophyll fluorescence (ChlF) imaging, which captures the growth, morphology, color and photosynthetic performance of Arabidopsis thaliana plants in response to salt stress. We optimized our salt treatment by controlling the soil-water content prior to introducing salt stress. We investigated these traits over time in two accessions in soil at 150, 100, or 50 mM NaCl to find that the plants subjected to 100 mM NaCl showed the most prominent responses in the absence of symptoms of severe stress. In these plants, salt stress induced significant changes in rosette area and morphology, but less prominent changes in rosette coloring and photosystem II efficiency. Clustering of ChlF traits with plant growth of nine accessions maintained at 100 mM NaCl revealed that in the early stage of salt stress, salinity tolerance correlated with non-photochemical quenching processes and during the later stage, plant performance correlated with quantum yield. This integrative approach allows the simultaneous analysis of several phenotypic traits. In combination with various genetic resources, the phenotyping protocol described here is expected to increase our understanding of plant performance and stress responses, ultimately identifying genes that improve plant performance in salt stress conditions.

hctsa: A Computational Framework for Automated Time-Series Phenotyping Using Massive Feature Extraction

Phenotype measurements frequently take the form of time series, but we currently lack a systematic method for relating these complex data streams to scientifically meaningful outcomes, such as relating the movement dynamics of organisms to their genotype or measurements of brain dynamics of a patient to their disease diagnosis. Previous work addressed this problem by comparing implementations of thousands of diverse scientific time-series analysis methods in an approach termed highly comparative time-series analysis. Here, we introduce hctsa, a software tool for applying this methodological approach to data. hctsa includes an architecture for computing over 7,700 time-series features and a suite of analysis and visualization algorithms to automatically select useful and interpretable time-series features for a given application. Using exemplar applications to high-throughput phenotyping experiments, we show how hctsa allows researchers to leverage decades of time-series research to quantify and understand informative structure in time-series data.

LeasyScan: a novel concept combining 3D imaging and lysimetry for high-throughput phenotyping of traits controlling plant water budget

In this paper, we describe the thought process and initial data behind the development of an imaging platform (LeasyScan) combined with lysimetric capacity, to assess canopy traits affecting water use (leaf area, leaf area index, transpiration). LeasyScan is based on a novel 3D scanning technique to capture leaf area development continuously, a scanner-to-plant concept to increase imaging throughput and analytical scales to combine gravimetric transpiration measurements. The paper presents how the technology functions, how data are visualised via a web-based interface and how data extraction and analysis is interfaced through 'R' libraries. Close agreement between scanned and observed leaf area data of individual plants in different crops was found (R(2) between 0.86 and 0.94). Similar agreement was found when comparing scanned and observed area of plants cultivated at densities reflecting field conditions (R(2) between 0.80 and 0.96). An example in monitoring plant transpiration by the analytical scales is presented. The last section illustrates some of the early ongoing applications of the platform to target key phenotypes: (i) the comparison of the leaf area development pattern of fine mapping recombinants of pearl millet; (ii) the leaf area development pattern of pearl millet breeding material targeted to different agro-ecological zones; (iii) the assessment of the transpiration response to high VPD in sorghum and pearl millet. This new platform has the potential to phenotype for traits controlling plant water use at a high rate and precision, of critical importance for drought adaptation, and creates an opportunity to harness their genetics for the breeding of improved varieties.

Enabling phenotypic big data with PheNorm

Objective

Electronic health record (EHR)-based phenotyping infers whether a patient has a disease based on the information in his or her EHR. A human-annotated training set with gold-standard disease status labels is usually required to build an algorithm for phenotyping based on a set of predictive features. The time intensiveness of annotation and feature curation severely limits the ability to achieve high-throughput phenotyping. While previous studies have successfully automated feature curation, annotation remains a major bottleneck. In this paper, we present PheNorm, a phenotyping algorithm that does not require expert-labeled samples for training.

Methods

The most predictive features, such as the number of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes or mentions of the target phenotype, are normalized to resemble a normal mixture distribution with high area under the receiver operating curve (AUC) for prediction. The transformed features are then denoised and combined into a score for accurate disease classification.

Results

We validated the accuracy of PheNorm with 4 phenotypes: coronary artery disease, rheumatoid arthritis, Crohn's disease, and ulcerative colitis. The AUCs of the PheNorm score reached 0.90, 0.94, 0.95, and 0.94 for the 4 phenotypes, respectively, which were comparable to the accuracy of supervised algorithms trained with sample sizes of 100-300, with no statistically significant difference.

Conclusion

The accuracy of the PheNorm algorithms is on par with algorithms trained with annotated samples. PheNorm fully automates the generation of accurate phenotyping algorithms and demonstrates the capacity for EHR-driven annotations to scale to the next level - phenotypic big data.

A Powerful New Quantitative Genetics Platform, Combining Caenorhabditis elegans High-Throughput Fitness Assays with a Large Collection of Recombinant Strains

The genetic variants underlying complex traits are often elusive even in powerful model organisms such as Caenorhabditis elegans with controlled genetic backgrounds and environmental conditions. Two major contributing factors are: (1) the lack of statistical power from measuring the phenotypes of small numbers of individuals, and (2) the use of phenotyping platforms that do not scale to hundreds of individuals and are prone to noisy measurements. Here, we generated a new resource of 359 recombinant inbred strains that augments the existing C. elegans N2xCB4856 recombinant inbred advanced intercross line population. This new strain collection removes variation in the neuropeptide receptor gene npr-1, known to have large physiological and behavioral effects on C. elegans and mitigates the hybrid strain incompatibility caused by zeel-1 and peel-1, allowing for identification of quantitative trait loci that otherwise would have been masked by those effects. Additionally, we optimized highly scalable and accurate high-throughput assays of fecundity and body size using the COPAS BIOSORT large particle nematode sorter. Using these assays, we identified quantitative trait loci involved in fecundity and growth under normal growth conditions and after exposure to the herbicide paraquat, including independent genetic loci that regulate different stages of larval growth. Our results offer a powerful platform for the discovery of the genetic variants that control differences in responses to drugs, other aqueous compounds, bacterial foods, and pathogenic stresses.

Comparative Aerial and Ground Based High Throughput Phenotyping for the Genetic Dissection of NDVI as a Proxy for Drought Adaptive Traits in Durum Wheat

High-throughput phenotyping platforms (HTPPs) provide novel opportunities to more effectively dissect the genetic basis of drought-adaptive traits. This genome-wide association study (GWAS) compares the results obtained with two Unmanned Aerial Vehicles (UAVs) and a ground-based platform used to measure Normalized Difference Vegetation Index (NDVI) in a panel of 248 elite durum wheat (Triticum turgidum L. ssp. durum Desf.) accessions at different growth stages and water regimes. Our results suggest increased ability of aerial over ground-based platforms to detect quantitative trait loci (QTL) for NDVI, particularly under terminal drought stress, with 22 and 16 single QTLs detected, respectively, and accounting for 89.6 vs. 64.7% phenotypic variance based on multiple QTL models. Additionally, the durum panel was investigated for leaf chlorophyll content (SPAD), leaf rolling and dry biomass under terminal drought stress. In total, 46 significant QTLs affected NDVI across platforms, 22 of which showed concomitant effects on leaf greenness, 2 on leaf rolling and 10 on biomass. Among 9 QTL hotspots on chromosomes 1A, 1B, 2B, 4B, 5B, 6B, and 7B that influenced NDVI and other drought-adaptive traits, 8 showed per se effects unrelated to phenology.

High-Throughput Phenotyping Enabled Genetic Dissection of Crop Lodging in Wheat

Novel high-throughput phenotyping (HTP) approaches are needed to advance the understanding of genotype-to-phenotype and accelerate plant breeding. The first generation of HTP has examined simple spectral reflectance traits from images and sensors but is limited in advancing our understanding of crop development and architecture. Lodging is a complex trait that significantly impacts yield and quality in many crops including wheat. Conventional visual assessment methods for lodging are time-consuming, relatively low-throughput, and subjective, limiting phenotyping accuracy and population sizes in breeding and genetics studies. Here, we demonstrate the considerable power of unmanned aerial systems (UAS) or drone-based phenotyping as a high-throughput alternative to visual assessments for the complex phenological trait of lodging, which significantly impacts yield and quality in many crops including wheat. We tested and validated quantitative assessment of lodging on 2,640 wheat breeding plots over the course of 2 years using differential digital elevation models from UAS. High correlations of digital measures of lodging to visual estimates and equivalent broad-sense heritability demonstrate this approach is amenable for reproducible assessment of lodging in large breeding nurseries. Using these high-throughput measures to assess the underlying genetic architecture of lodging in wheat, we applied genome-wide association analysis and identified a key genomic region on chromosome 2A, consistent across digital and visual scores of lodging. However, these associations accounted for a very minor portion of the total phenotypic variance. We therefore investigated whole genome prediction models and found high prediction accuracies across populations and environments. This adequately accounted for the highly polygenic genetic architecture of numerous small effect loci, consistent with the previously described complex genetic architecture of lodging in wheat. Our study provides a proof-of-concept application of UAS-based phenomics that is scalable to tens-of-thousands of plots in breeding and genetic studies as will be needed to uncover the genetic factors and increase the rate of gain for complex traits in crop breeding.

Predicting how cells spread and migrate: focal adhesion size does matter

Efficient cell migration is central to the normal development of tissues and organs and is involved in a wide range of human diseases, including cancer metastasis, immune responses, and cardiovascular disorders. Mesenchymal migration is modulated by focal-adhesion proteins, which organize into large integrin-rich protein complexes at the basal surface of adherent cells. Whether the extent of clustering of focal-adhesion proteins is actually required for effective migration is unclear. We recently demonstrated that the depletion of major focal-adhesion proteins, as well as modulation of matrix compliance, actin assembly, mitochondrial activity, and DNA recombination, all converged into highly predictable, inter-related, biphasic changes in focal adhesion size and cell migration. Herein, we further discuss the role of focal adhesions in controlling cell spreading and test their potential role in cell migration.

Dynamic quantification of canopy structure to characterize early plant vigour in wheat genotypes

Early vigour is an important physiological trait to improve establishment, water-use efficiency, and grain yield for wheat. Phenotyping large numbers of lines is challenging due to the fast growth and development of wheat seedlings. Here we developed a new photo-based workflow to monitor dynamically the growth and development of the wheat canopy of two wheat lines with a contrasting early vigour trait. Multiview images were taken using a 'vegetation stress' camera at 2 d intervals from emergence to the sixth leaf stage. Point clouds were extracted using the Multi-View Stereo and Structure From Motion (MVS-SFM) algorithm, and segmented into individual organs using the Octree method, with leaf midribs fitted using local polynomial function. Finally, phenotypic parameters were calculated from the reconstructed point cloud including: tiller and leaf number, plant height, Haun index, phyllochron, leaf length, angle, and leaf elongation rate. There was good agreement between the observed and estimated leaf length (RMSE=8.6mm, R (2)=0.98, n=322) across both lines. Significant contrasts of phenotyping parameters were observed between the two lines and were consistent with manual observations. The early vigour line had fewer tillers (2.4±0.6) and larger leaves (308.0±38.4mm and 17.1±2.7mm for leaf length and width, respectively). While the phyllochron of both lines was quite similar, the non-vigorous line had a greater Haun index (more leaves on the main stem) on any date, as the vigorous line had slower development of its first two leaves. The workflow presented in this study provides an efficient method to phenotype individual plants using a low-cost camera (an RGB camera is also suitable) and could be applied in phenotyping for applications in both simulation modelling and breeding. The rapidity and accuracy of this novel method can characterize the results of specific selection criteria (e.g. width of leaf three, number of tillers, rate of leaf appearance) that have been or can now be utilized to breed for early leaf growth and tillering in wheat.

A Combined Phenotypic and Metabolomic Approach for Elucidating the Biostimulant Action of a Plant-Derived Protein Hydrolysate on Tomato Grown Under Limited Water Availability

Plant-derived protein hydrolysates (PHs) are an important category of biostimulants able to increase plant growth and crop yield especially under environmental stress conditions. PHs can be applied as foliar spray or soil drench. Foliar spray is generally applied to achieve a relatively short-term response, whereas soil drench is used when a long-term effect is desired. The aim of the study was to elucidate the biostimulant action of PH application method (foliar spray or substrate drench) on morpho-physiological traits and metabolic profile of tomato grown under limited water availability. An untreated control was also included. A high-throughput image-based phenotyping (HTP) approach was used to non-destructively monitor the crop response under limited water availability (40% of container capacity) in a controlled environment. Moreover, metabolic profile of leaves was determined at the end of the trial. Dry biomass of shoots at the end of the trial was significantly correlated with number of green pixels (R ² = 0.90) and projected shoot area, respectively. Both drench and foliar treatments had a positive impact on the digital biomass compared to control while the photosynthetic performance of the plants was slightly influenced by treatments. Overall drench application under limited water availability more positively influenced biomass accumulation and metabolic profile than foliar application. Significantly higher transpiration use efficiency was observed with PH-drench applications indicating better stomatal conductance. The mass-spectrometry based metabolomic analysis allowed the identification of distinct biochemical signatures in PH-treated plants. Metabolomic changes involved a wide and organized range of biochemical processes that included, among others, phytohormones (notably a decrease in cytokinins and an accumulation of salicylates) and lipids (including membrane lipids, sterols, and terpenes). From a general perspective, treated tomato plants exhibited an improved tolerance to reactive oxygen species (ROS)-mediated oxidative imbalance. Such capability to cope with oxidative stress might have resulted from a coordinated action of signaling compounds (salicylic acid and hydroxycinnamic amides), radical scavengers such as carotenoids and prenyl quinones, as well as a reduced biosynthesis of tetrapyrrole coproporphyrins.

Application of Machine Learning Algorithms in Plant Breeding: Predicting Yield From Hyperspectral Reflectance in Soybean

Recent substantial advances in high-throughput field phenotyping have provided plant breeders with affordable and efficient tools for evaluating a large number of genotypes for important agronomic traits at early growth stages. Nevertheless, the implementation of large datasets generated by high-throughput phenotyping tools such as hyperspectral reflectance in cultivar development programs is still challenging due to the essential need for intensive knowledge in computational and statistical analyses. In this study, the robustness of three common machine learning (ML) algorithms, multilayer perceptron (MLP), support vector machine (SVM), and random forest (RF), were evaluated for predicting soybean (Glycine max) seed yield using hyperspectral reflectance. For this aim, the hyperspectral reflectance data for the whole spectra ranged from 395 to 1005 nm, which were collected at the R4 and R5 growth stages on 250 soybean genotypes grown in four environments. The recursive feature elimination (RFE) approach was performed to reduce the dimensionality of the hyperspectral reflectance data and select variables with the largest importance values. The results indicated that R5 is more informative stage for measuring hyperspectral reflectance to predict seed yields. The 395 nm reflectance band was also identified as the high ranked band in predicting the soybean seed yield. By considering either full or selected variables as the input variables, the ML algorithms were evaluated individually and combined-version using the ensemble-stacking (E-S) method to predict the soybean yield. The RF algorithm had the highest performance with a value of 84% yield classification accuracy among all the individual tested algorithms. Therefore, by selecting RF as the metaClassifier for E-S method, the prediction accuracy increased to 0.93, using all variables, and 0.87, using selected variables showing the success of using E-S as one of the ensemble techniques. This study demonstrated that soybean breeders could implement E-S algorithm using either the full or selected spectra reflectance to select the high-yielding soybean genotypes, among a large number of genotypes, at early growth stages.

Hyperspectral Leaf Reflectance as Proxy for Photosynthetic Capacities: An Ensemble Approach Based on Multiple Machine Learning Algorithms

Global agriculture production is challenged by increasing demands from rising population and a changing climate, which may be alleviated through development of genetically improved crop cultivars. Research into increasing photosynthetic energy conversion efficiency has proposed many strategies to improve production but have yet to yield real-world solutions, largely because of a phenotyping bottleneck. Partial least squares regression (PLSR) is a statistical technique that is increasingly used to relate hyperspectral reflectance to key photosynthetic capacities associated with carbon uptake (maximum carboxylation rate of Rubisco, Vc,max ) and conversion of light energy (maximum electron transport rate supporting RuBP regeneration, Jmax ) to alleviate this bottleneck. However, its performance varies significantly across different plant species, regions, and growth environments. Thus, to cope with the heterogeneous performances of PLSR, this study aims to develop a new approach to estimate photosynthetic capacities. A framework was developed that combines six machine learning algorithms, including artificial neural network (ANN), support vector machine (SVM), least absolute shrinkage and selection operator (LASSO), random forest (RF), Gaussian process (GP), and PLSR to optimize high-throughput analysis of the two photosynthetic variables. Six tobacco genotypes, including both transgenic and wild-type lines, with a range of photosynthetic capacities were used to test the framework. Leaf reflectance spectra were measured from 400 to 2500 nm using a high-spectral-resolution spectroradiometer. Corresponding photosynthesis vs. intercellular CO2 concentration response curves were measured for each leaf using a leaf gas-exchange system. Results suggested that the mean R 2 value of the six regression techniques for predicting Vc,max (Jmax ) ranged from 0.60 (0.45) to 0.65 (0.56) with the mean RMSE value varying from 47.1 (40.1) to 54.0 (44.7) μmol m-2 s-1. Regression stacking for Vc,max (Jmax ) performed better than the individual regression techniques with increases in R 2 of 0.1 (0.08) and decreases in RMSE by 4.1 (6.6) μmol m-2 s-1, equal to 8% (15%) reduction in RMSE. Better predictive performance of the regression stacking is likely attributed to the varying coefficients (or weights) in the level-2 model (the LASSO model) and the diverse ability of each individual regression technique to utilize spectral information for the best modeling performance. Further refinements can be made to apply this stacked regression technique to other plant phenotypic traits.

Converging phenomics and genomics to study natural variation in plant photosynthetic efficiency

In recent years developments in plant phenomic approaches and facilities have gradually caught up with genomic approaches. An opportunity lies ahead to dissect complex, quantitative traits when both genotype and phenotype can be assessed at a high level of detail. This is especially true for the study of natural variation in photosynthetic efficiency, for which forward genetics studies have yielded only a little progress in our understanding of the genetic layout of the trait. High-throughput phenotyping, primarily from chlorophyll fluorescence imaging, should help to dissect the genetics of photosynthesis at the different levels of both plant physiology and development. Specific emphasis should be directed towards understanding the acclimation of the photosynthetic machinery in fluctuating environments, which may be crucial for the identification of genetic variation for relevant traits in food crops. Facilities should preferably be designed to accommodate phenotyping of photosynthesis-related traits in such environments. The use of forward genetics to study the genetic architecture of photosynthesis is likely to lead to the discovery of novel traits and/or genes that may be targeted in breeding or bio-engineering approaches to improve crop photosynthetic efficiency. In the near future, big data approaches will play a pivotal role in data processing and streamlining the phenotype-to-gene identification pipeline.

Biomass for thermochemical conversion: targets and challenges

Bioenergy will be one component of a suite of alternatives to fossil fuels. Effective conversion of biomass to energy will require the careful pairing of advanced conversion technologies with biomass feedstocks optimized for the purpose. Lignocellulosic biomass can be converted to useful energy products via two distinct pathways: enzymatic or thermochemical conversion. The thermochemical pathways are reviewed and potential biotechnology or breeding targets to improve feedstocks for pyrolysis, gasification, and combustion are identified. Biomass traits influencing the effectiveness of the thermochemical process (cell wall composition, mineral and moisture content) differ from those important for enzymatic conversion and so properties are discussed in the language of biologists (biochemical analysis) as well as that of engineers (proximate and ultimate analysis). We discuss the genetic control, potential environmental influence, and consequences of modification of these traits. Improving feedstocks for thermochemical conversion can be accomplished by the optimization of lignin levels, and the reduction of ash and moisture content. We suggest that ultimate analysis and associated properties such as H:C, O:C, and heating value might be more amenable than traditional biochemical analysis to the high-throughput necessary for the phenotyping of large plant populations. Expanding our knowledge of these biomass traits will play a critical role in the utilization of biomass for energy production globally, and add to our understanding of how plants tailor their composition with their environment.

Genome-wide association mapping unravels the genetic control of seed germination and vigor in Brassica napus

Rapid and uniform seed germination is a crucial prerequisite for crop establishment and high yield levels in crop production. A disclosure of genetic factors contributing to adequate seed vigor would help to further increase yield potential and stability. Here we carried out a genome-wide association study in order to define genomic regions influencing seed germination and early seedling growth in oilseed rape (Brassica napus L.). A population of 248 genetically diverse winter-type B. napus accessions was genotyped with the Brassica 60k SNP Illumina genotyping array. Automated high-throughput in vitro phenotyping provided extensive data for multiple traits related to germination and early vigor, such as germination speed, absolute germination rate and radicle elongation. The data obtained indicate that seed germination and radicle growth are strongly environmentally dependent, but could nevertheless be substantially improved by genomic-based breeding. Conditions during seed production and storage were shown to have a profound effect on seed vigor, and a variable manifestation of seed dormancy appears to contribute to differences in germination performance in B. napus. Several promising positional and functional candidate genes could be identified within the genomic regions associated with germination speed, absolute germination rate, radicle growth and thousand seed weight. These include B. napus orthologs of the Arabidopsis thaliana genes SNOWY COTYLEDON 1 (SCO1), ARABIDOPSIS TWO-COMPONENT RESPONSE REGULATOR (ARR4), and ARGINYL-t-RNA PROTEIN TRANSFERASE 1 (ATE1), which have been shown previously to play a role in seed germination and seedling growth in A. thaliana.

Genetic and genomic resources of sorghum to connect genotype with phenotype in contrasting environments

With the recent development of genomic resources and high-throughput phenotyping platforms, the 21st century is primed for major breakthroughs in the discovery, understanding and utilization of plant genetic variation. Significant advances in agriculture remain at the forefront to increase crop production and quality to satisfy the global food demand in a changing climate all while reducing the environmental impacts of the world's food production. Sorghum, a resilient C₄ grain and grass important for food and energy production, is being extensively dissected genetically and phenomically to help connect the relationship between genetic and phenotypic variation. Unlike genetically modified crops such as corn or soybean, sorghum improvement has relied heavily on public research; thus, many of the genetic resources serve a dual purpose for both academic and commercial pursuits. Genetic and genomic resources not only provide the foundation to identify and understand the genes underlying variation, but also serve as novel sources of genetic and phenotypic diversity in plant breeding programs. To better disseminate the collective information of this community, we discuss: (i) the genomic resources of sorghum that are at the disposal of the research community; (ii) the suite of sorghum traits as potential targets for increasing productivity in contrasting environments; and (iii) the prospective approaches and technologies that will help to dissect the genotype-phenotype relationship as well as those that will apply foundational knowledge for sorghum improvement.

Predicting dark respiration rates of wheat leaves from hyperspectral reflectance

Greater availability of leaf dark respiration (R_dark ) data could facilitate breeding efforts to raise crop yield and improve global carbon cycle modelling. However, the availability of R_dark data is limited because it is cumbersome, time consuming, or destructive to measure. We report a non-destructive and high-throughput method of estimating R_dark from leaf hyperspectral reflectance data that was derived from leaf R_dark measured by a destructive high-throughput oxygen consumption technique. We generated a large dataset of leaf R_dark for wheat (1380 samples) from 90 genotypes, multiple growth stages, and growth conditions to generate models for R_dark . Leaf R_dark (per unit leaf area, fresh mass, dry mass or nitrogen, N) varied 7- to 15-fold among individual plants, whereas traits known to scale with R_dark , leaf N, and leaf mass per area (LMA) only varied twofold to fivefold. Our models predicted leaf R_dark , N, and LMA with r² values of 0.50-0.63, 0.91, and 0.75, respectively, and relative bias of 17-18% for R_dark and 7-12% for N and LMA. Our results suggest that hyperspectral model prediction of wheat leaf R_dark is largely independent of leaf N and LMA. Potential drivers of hyperspectral signatures of R_dark are discussed.

Assessing Wheat Traits by Spectral Reflectance: Do We Really Need to Focus on Predicted Trait-Values or Directly Identify the Elite Genotypes Group?

Phenotyping, via remote and proximal sensing techniques, of the agronomic and physiological traits associated with yield potential and drought adaptation could contribute to improvements in breeding programs. In the present study, 384 genotypes of wheat (Triticum aestivum L.) were tested under fully irrigated (FI) and water stress (WS) conditions. The following traits were evaluated and assessed via spectral reflectance: Grain yield (GY), spikes per square meter (SM2), kernels per spike (KPS), thousand-kernel weight (TKW), chlorophyll content (SPAD), stem water soluble carbohydrate concentration and content (WSC and WSCC, respectively), carbon isotope discrimination (Δ13C), and leaf area index (LAI). The performances of spectral reflectance indices (SRIs), four regression algorithms (PCR, PLSR, ridge regression RR, and SVR), and three classification methods (PCA-LDA, PLS-DA, and kNN) were evaluated for the prediction of each trait. For the classification approaches, two classes were established for each trait: The lower 80% of the trait variability range (Class 1) and the remaining 20% (Class 2 or elite genotypes). Both the SRIs and regression methods performed better when data from FI and WS were combined. The traits that were best estimated by SRIs and regression methods were GY and Δ13C. For most traits and conditions, the estimations provided by RR and SVR were the same, or better than, those provided by the SRIs. PLS-DA showed the best performance among the categorical methods and, unlike the SRI and regression models, most traits were relatively well-classified within a specific hydric condition (FI or WS), proving that classification approach is an effective tool to be explored in future studies related to genotype selection.