More than 1000 genotypes are required to derive robust relationships between yield, yield stability and physiological parameters: a computational study on wheat crop.

KEY MESSAGE: Using in silico experiment in crop model, we identified different physiological regulations of yield and yield stability, as well as quantify the genotype and environment numbers required for analysing yield stability convincingly. Identifying target traits for breeding stable and high-yielded cultivars simultaneously is difficult due to limited knowledge of physiological mechanisms behind yield stability. Besides, there is no consensus about the adequacy of a stability index (SI) and the minimal number of environments and genotypes required for evaluating yield stability. We studied this question using the crop model APSIM-Wheat to simulate 9100 virtual genotypes grown under 9000 environments. By analysing the simulated data, we showed that the shape of phenotype distributions affected the correlation between SI and mean yield and the genotypic superiority measure (Pi) was least affected among 11 SI. Pi was used as index to demonstrate that more than 150 environments were required to estimate yield stability of a genotype convincingly and more than 1000 genotypes were necessary to evaluate the contribution of a physiological parameter to yield stability. Network analyses suggested that a physiological parameter contributed preferentially to yield or Pi. For example, soil water absorption efficiency and potential grain filling rate explained better the variations in yield than in Pi; while light extinction coefficient and radiation use efficiency were more correlated with Pi than with yield. The high number of genotypes and environments required for studying Pi highlight the necessity and potential of in silico experiments to better understand the mechanisms behind yield stability.

Optimized cultivar deployment improves the efficiency and stability of sunflower crop production at national scale.

KEY MESSAGE: Crop simulation helps to analyze environmental impacts on crops and provides year-independent context information. This information is of major importance when deciding which cultivar to choose at sowing time. Plant breeding programs design new crop cultivars which, while developed for distinct populations of environments, are nevertheless grown over large areas during their time in the market. Over its cultivation area, the crop is exposed to highly diverse stress patterns caused by climatic uncertainty and multiple management options, which often leads to decreased expected crop performance. In this study, we aim to assess how finer spatial management of genetic resources could reduce the yield variance explained by genotype × environment interactions in a set of cropping environments and ultimately improve the efficiency and stability of crop production. We used modeling and simulation to predict the crop performance resulting from the interaction between cultivar growth and development, climate and soil conditions, and management practices. We designed a computational experiment that evaluated the performance of a collection of commercial sunflower cultivars in a realistic population of cropping conditions in France, built from extensive agricultural surveys. Distinct farming locations sharing similar simulated abiotic stress patterns were clustered together to specify environment types. We then used optimization methods to search for cultivars × environments combinations leading to increased yield expectations. Results showed that a single cultivar choice adapted to the most frequent environment-type in the population is a robust strategy. However, the relevance of cultivar recommendations to specific locations was gradually increasing with the knowledge of pedo-climatic conditions. We argue that this approach while being operational on current genetic material could act synergistically with plant breeding as more diverse material could enable access to cultivars with distinctive traits, more adapted to specific conditions.

Data describing the eco-physiological responses of twenty-four sunflower genotypes to water deficit.

This article presents experimental data describing the physiology and morphology of sunflower plants subjected to water deficit. Twenty-four sunflower genotypes were selected to represent genetic diversity within cultivated sunflower and included both inbred lines and their hybrids. Drought stress was applied to plants in pots at the vegetative stage using the high-throughput phenotyping platform Heliaphen at INRA Toulouse (France). Here, we provide data including specific leaf area, osmotic potential and adjustment, carbon isotope discrimination, leaf transpiration, plant architecture: plant height, leaf number, stem diameter. We also provide leaf areas of individual organs through time and growth rate during the stress period, environmental data such as temperatures, wind and radiation during the experiment. These data differentiate both treatment and the different genotypes and constitute a valuable resource to the community to study adaptation of crops to drought and the physiological basis of heterosis. It is available on the following repository: https://doi.org/10.25794/phenotype/er6lPW7V.

Heliaphen, an Outdoor High-Throughput Phenotyping Platform for Genetic Studies and Crop Modeling.

Heliaphen is an outdoor platform designed for high-throughput phenotyping. It allows the automated management of drought scenarios and monitoring of plants throughout their lifecycles. A robot moving between plants growing in 15-L pots monitors the plant water status and phenotypes the leaf or whole-plant morphology. From these measurements, we can compute more complex traits, such as leaf expansion (LE) or transpiration rate (TR) in response to water deficit. Here, we illustrate the capabilities of the platform with two practical cases in sunflower (Helianthus annuus): a genetic and genomic study of the response of yield-related traits to drought, and a modeling study using measured parameters as inputs for a crop simulation. For the genetic study, classical measurements of thousand-kernel weight (TKW) were performed on a biparental population under automatically managed drought stress and control conditions. These data were used for an association study, which identified five genetic markers of the TKW drought response. A complementary transcriptomic analysis identified candidate genes associated with these markers that were differentially expressed in the parental backgrounds in drought conditions. For the simulation study, we used a crop simulation model to predict the impact on crop yield of two traits measured on the platform (LE and TR) for a large number of environments. We conducted simulations in 42 contrasting locations across Europe using 21 years of climate data. We defined the pattern of abiotic stresses occurring at the continental scale and identified ideotypes (i.e., genotypes with specific trait values) that are more adapted to specific environment types. This study exemplifies how phenotyping platforms can assist the identification of the genetic architecture controlling complex response traits and facilitate the estimation of ecophysiological model parameters to define ideotypes adapted to different environmental conditions.

Using numerical plant models and phenotypic correlation space to design achievable ideotypes.

Numerical plant models can predict the outcome of plant traits modifications resulting from genetic variations, on plant performance, by simulating physiological processes and their interaction with the environment. Optimization methods complement those models to design ideotypes, that is, ideal values of a set of plant traits, resulting in optimal adaptation for given combinations of environment and management, mainly through the maximization of performance criteria (e.g. yield and light interception). As use of simulation models gains momentum in plant breeding, numerical experiments must be carefully engineered to provide accurate and attainable results, rooting them in biological reality. Here, we propose a multi-objective optimization formulation that includes a metric of performance, returned by the numerical model, and a metric of feasibility, accounting for correlations between traits based on field observations. We applied this approach to two contrasting models: a process-based crop model of sunflower and a functional-structural plant model of apple trees. In both cases, the method successfully characterized key plant traits and identified a continuum of optimal solutions, ranging from the most feasible to the most efficient. The present study thus provides successful proof of concept for this enhanced modelling approach, which identified paths for desirable trait modification, including direction and intensity.

Optimization of black-box models with uncertain climatic inputs-Application to sunflower ideotype design.

Accounting for the interannual climatic variations is a well-known issue for simulation-based studies of environmental systems. It often requires intensive sampling (e.g., averaging the simulation outputs over many climatic series), which hinders many sequential processes, in particular optimization algorithms. We propose here an approach based on a subset selection in a large basis of climatic series, using an ad-hoc similarity function and clustering. A non-parametric reconstruction technique is introduced to estimate accurately the distribution of the output of interest using only the subset sampling. The proposed strategy is non-intrusive and generic (i.e. transposable to most models with climatic data inputs), and can be combined to most "off-the-shelf" optimization solvers. We apply our approach to sunflower ideotype design using the crop model SUNFLO. The underlying optimization problem is formulated as a multi-objective one to account for risk-aversion. Our approach achieves good performances even for limited computational budgets, outperforming significantly standard strategies.

Genetic control of plasticity of oil yield for combined abiotic stresses using a joint approach of crop modelling and genome-wide association.

Understanding the genetic basis of phenotypic plasticity is crucial for predicting and managing climate change effects on wild plants and crops. Here, we combined crop modelling and quantitative genetics to study the genetic control of oil yield plasticity for multiple abiotic stresses in sunflower. First, we developed stress indicators to characterize 14 environments for three abiotic stresses (cold, drought and nitrogen) using the SUNFLO crop model and phenotypic variations of three commercial varieties. The computed plant stress indicators better explain yield variation than descriptors at the climatic or crop levels. In those environments, we observed oil yield of 317 sunflower hybrids and regressed it with three selected stress indicators. The slopes of cold stress norm reaction were used as plasticity phenotypes in the following genome-wide association study. Among the 65 534 tested Single Nucleotide Polymorphisms (SNPs), we identified nine quantitative trait loci controlling oil yield plasticity to cold stress. Associated single nucleotide polymorphisms are localized in genes previously shown to be involved in cold stress responses: oligopeptide transporters, lipid transfer protein, cystatin, alternative oxidase or root development. This novel approach opens new perspectives to identify genomic regions involved in genotype-by-environment interaction of a complex traits to multiple stresses in realistic natural or agronomical conditions.

Assessment of the Potential Impacts of Wheat Plant Traits across Environments by Combining Crop Modeling and Global Sensitivity Analysis.

A crop can be viewed as a complex system with outputs (e.g. yield) that are affected by inputs of genetic, physiology, pedo-climatic and management information. Application of numerical methods for model exploration assist in evaluating the major most influential inputs, providing the simulation model is a credible description of the biological system. A sensitivity analysis was used to assess the simulated impact on yield of a suite of traits involved in major processes of crop growth and development, and to evaluate how the simulated value of such traits varies across environments and in relation to other traits (which can be interpreted as a virtual change in genetic background). The study focused on wheat in Australia, with an emphasis on adaptation to low rainfall conditions. A large set of traits (90) was evaluated in a wide target population of environments (4 sites × 125 years), management practices (3 sowing dates × 3 nitrogen fertilization levels) and CO2 (2 levels). The Morris sensitivity analysis method was used to sample the parameter space and reduce computational requirements, while maintaining a realistic representation of the targeted trait × environment × management landscape (∼ 82 million individual simulations in total). The patterns of parameter × environment × management interactions were investigated for the most influential parameters, considering a potential genetic range of +/- 20% compared to a reference cultivar. Main (i.e. linear) and interaction (i.e. non-linear and interaction) sensitivity indices calculated for most of APSIM-Wheat parameters allowed the identification of 42 parameters substantially impacting yield in most target environments. Among these, a subset of parameters related to phenology, resource acquisition, resource use efficiency and biomass allocation were identified as potential candidates for crop (and model) improvement.

Effects of plant growth stage and leaf aging on the response of transpiration and photosynthesis to water deficit in sunflower.

Water deficit influences leaf transpiration rate and photosynthetic activity. The genotype-dependent response of the latter has not been assessed in sunflower (Helianthus annuus L.), particularly during the reproductive period when grain filling and lipogenesis depend greatly on photosynthate availability. To evaluate genotypic responses to water deficit before and after flowering, two greenhouse experiments were performed. Four genotypes-two inbred lines (PSC8, XRQ) and two cultivars (Inedi, Melody)-were subjected to progressive water deficit. Non-linear regression was used to calculate the soil water deficit threshold (FTSWt) at which processes (transpiration and photosynthetic activity) were affected by water deficit. In the vegetative growth stage, photosynthetic activity was affected at a lower mean value of FTSWt (0.39) than transpiration (0.55). However, in the reproductive stage, photosynthetic activity was more sensitive to soil water deficit (FTSWt=0.45). We found a significant (P=0.02) effect of plant growth stage on the difference between photosynthesis and transpiration rate thresholds and, a significant (P=0.03) effect of leaf age on transpiration. Such results will improve phenotyping methods and provide paths for integrating genotypic variability into crop models.

A biomarker based on gene expression indicates plant water status in controlled and natural environments.

Plant or soil water status is required in many scientific fields to understand plant responses to drought. Because the transcriptomic response to abiotic conditions, such as water deficit, reflects plant water status, genomic tools could be used to develop a new type of molecular biomarker. Using the sunflower (Helianthus annuus L.) as a model species to study the transcriptomic response to water deficit both in greenhouse and field conditions, we specifically identified three genes that showed an expression pattern highly correlated to plant water status as estimated by the pre-dawn leaf water potential, fraction of transpirable soil water, soil water content or fraction of total soil water in controlled conditions. We developed a generalized linear model to estimate these classical water status indicators from the expression levels of the three selected genes under controlled conditions. This estimation was independent of the four tested genotypes and the stage (pre- or post-flowering) of the plant. We further validated this gene expression biomarker under field conditions for four genotypes in three different trials, over a large range of water status, and we were able to correct their expression values for a large diurnal sampling period.

A generic model to simulate air-borne diseases as a function of crop architecture.

In a context of pesticide use reduction, alternatives to chemical-based crop protection strategies are needed to control diseases. Crop and plant architectures can be viewed as levers to control disease outbreaks by affecting microclimate within the canopy or pathogen transmission between plants. Modeling and simulation is a key approach to help analyze the behaviour of such systems where direct observations are difficult and tedious. Modeling permits the joining of concepts from ecophysiology and epidemiology to define structures and functions generic enough to describe a wide range of epidemiological dynamics. Additionally, this conception should minimize computing time by both limiting the complexity and setting an efficient software implementation. In this paper, our aim was to present a model that suited these constraints so it could first be used as a research and teaching tool to promote discussions about epidemic management in cropping systems. The system was modelled as a combination of individual hosts (population of plants or organs) and infectious agents (pathogens) whose contacts are restricted through a network of connections. The system dynamics were described at an individual scale. Additional attention was given to the identification of generic properties of host-pathogen systems to widen the model's applicability domain. Two specific pathosystems with contrasted crop architectures were considered: ascochyta blight on pea (homogeneously layered canopy) and potato late blight (lattice of individualized plants). The model behavior was assessed by simulation and sensitivity analysis and these results were discussed against the model ability to discriminate between the defined types of epidemics. Crop traits related to disease avoidance resulting in a low exposure, a slow dispersal or a de-synchronization of plant and pathogen cycles were shown to strongly impact the disease severity at the crop scale.

Quantifying physiological determinants of genetic variation for yield potential in sunflower. SUNFLO: a model-based analysis.

Present work focussed on improving the description of organogenesis, morphogenesis and metabolism in a biophysical plant model (SUNFLO) applied to sunflower (Helianthus annuus L.). This first version of the model is designed for potential growth conditions without any abiotic or biotic stresses. A greenhouse experiment was conducted to identify and estimate the phenotypic traits involved in plant productivity variability of 26 sunflower genotypes. The ability of SUNFLO to discriminate the genotypes was tested on previous results of a field survey aimed at evaluating the genetic progress since 1960. Plants were phenotyped in four directions; phenology, architecture, photosynthesis and biomass allocation. Twelve genotypic parameters were chosen to account for the phenotypic variability. SUNFLO was built to evaluate their respective contribution to the variability of yield potential. A large phenotypic variability was found for all genotypic parameters. SUNFLO was able to account for 80% of observed variability in yield potential and to analyse the phenotypic variability of complex plant traits such as light interception efficiency or seed yield. It suggested that several ways are possible to reach high yields in sunflower. Unlike classical statistical analysis, this modelling approach highlights some efficient parameter combinations used by the most productive genotypes. The next steps will be to evaluate the genetic determinisms of the genotypic parameters.