Are high-throughput root phenotyping platforms suitable for informing root system architecture models with genotype-specific parameters? An evaluation based on the root model ArchiSimple and a small panel of wheat cultivars.

Given the difficulties in accessing plant roots in situ, high-throughput root phenotyping (HTRP) platforms under controlled conditions have been developed to meet the growing demand for characterizing root system architecture (RSA) for genetic analyses. However, a proper evaluation of their capacity to provide the same estimates for strictly identical root traits across platforms has never been achieved. In this study, we performed such an evaluation based on six major parameters of the RSA model ArchiSimple, using a diversity panel of 14 bread wheat cultivars in two HTRP platforms that had different growth media and non-destructive imaging systems together with a conventional set-up that had a solid growth medium and destructive sampling. Significant effects of the experimental set-up were found for all the parameters and no significant correlations across the diversity panel among the three set-ups could be detected. Differences in temperature, irradiance, and/or the medium in which the plants were growing might partly explain both the differences in the parameter values across the experiments as well as the genotype × set-up interactions. Furthermore, the values and the rankings across genotypes of only a subset of parameters were conserved between contrasting growth stages. As the parameters chosen for our analysis are root traits that have strong impacts on RSA and are close to parameters used in a majority of RSA models, our results highlight the need to carefully consider both developmental and environmental drivers in root phenomics studies.

Different wheat loci are associated to heritable free asparagine content in grain grown under different water and nitrogen availability.

KEY MESSAGE: Different wheat QTLs were associated to the free asparagine content of grain grown in four different conditions. Environmental effects are a key factor when selecting for low acrylamide-forming potential. The amount of free asparagine in grain of a wheat genotype determines its potential to form harmful acrylamide in derivative food products. Here, we explored the variation in the free asparagine, aspartate, glutamine and glutamate contents of 485 accessions reflecting wheat worldwide diversity to define the genetic architecture governing the accumulation of these amino acids in grain. Accessions were grown under high and low nitrogen availability and in water-deficient and well-watered conditions, and plant and grain phenotypes were measured. Free amino acid contents of grain varied from 0.01 to 1.02 mg g-1 among genotypes in a highly heritable way that did not correlate strongly with grain yield, protein content, specific weight, thousand-kernel weight or heading date. Mean free asparagine content was 4% higher under high nitrogen and 3% higher in water-deficient conditions. After genotyping the accessions, single-locus and multi-locus genome-wide association study models were used to identify several QTLs for free asparagine content located on nine chromosomes. Each QTL was associated with a single amino acid and growing environment, and none of the QTLs colocalised with genes known to be involved in the corresponding amino acid metabolism. This suggests that free asparagine content is controlled by several loci with minor effects interacting with the environment. We conclude that breeding for reduced asparagine content is feasible, but should be firmly based on multi-environment field trials. KEY MESSAGE: Different wheat QTLs were associated to the free asparagine content of grain grown in four different conditions. Environmental effects are a key factor when selecting for low acrylamide-forming potential.

High-dimensional multi-omics measured in controlled conditions are useful for maize platform and field trait predictions.

KEY MESSAGE: Transcriptomics and proteomics information collected on a platform can predict additive and non-additive effects for platform traits and additive effects for field traits. The effects of climate change in the form of drought, heat stress, and irregular seasonal changes threaten global crop production. The ability of multi-omics data, such as transcripts and proteins, to reflect a plant's response to such climatic factors can be capitalized in prediction models to maximize crop improvement. Implementing multi-omics characterization in field evaluations is challenging due to high costs. It is, however, possible to do it on reference genotypes in controlled conditions. Using omics measured on a platform, we tested different multi-omics-based prediction approaches, using a high dimensional linear mixed model (MegaLMM) to predict genotypes for platform traits and agronomic field traits in a panel of 244 maize hybrids. We considered two prediction scenarios: in the first one, new hybrids are predicted (CV-NH), and in the second one, partially observed hybrids are predicted (CV-POH). For both scenarios, all hybrids were characterized for omics on the platform. We observed that omics can predict both additive and non-additive genetic effects for the platform traits, resulting in much higher predictive abilities than GBLUP. It highlights their efficiency in capturing regulatory processes in relation to growth conditions. For the field traits, we observed that the additive components of omics only slightly improved predictive abilities for predicting new hybrids (CV-NH, model MegaGAO) and for predicting partially observed hybrids (CV-POH, model GAOxW-BLUP) in comparison to GBLUP. We conclude that measuring the omics in the fields would be of considerable interest in predicting productivity if the costs of omics drop significantly.

Plasticity QTLs specifically contribute to the genotype × water availability interaction in maize.

KEY MESSAGE: Multi-trial genome wide association study of plasticity indices allow to detect QTLs specifically involved in the genotype x water availability interaction. Concerns regarding high maize yield losses due to increasing occurrences of drought events are growing, and breeders are still looking for molecular markers for drought tolerance. However, the genetic determinism of traits in response to drought is highly complex and identification of causal regions is a tremendous task. Here, we exploit the phenotypic data obtained from four trials carried out on a phenotyping platform, where a diversity panel of 254 maize hybrids was grown under well-watered and water deficit conditions, to investigate the genetic bases of the drought response in maize. To dissociate drought effect from other environmental factors, we performed multi-trial genome-wide association study on well-watered and water deficit phenotypic means, and on phenotypic plasticity indices computed from measurements made for six ecophysiological traits. We identify 102 QTLs and 40 plasticity QTLs. Most of them were new compared to those obtained from a previous study on the same dataset. Our results show that plasticity QTLs cover genetic regions not identified by QTLs. Furthermore, for all ecophysiological traits, except one, plasticity QTLs are specifically involved in the genotype by water availability interaction, for which they explain between 60 and 100% of the variance. Altogether, QTLs and plasticity QTLs captured more than 75% of the genotype by water availability interaction variance, and allowed to find new genetic regions. Overall, our results demonstrate the importance of considering phenotypic plasticity to decipher the genetic architecture of trait response to stress.

Genetic regions determine tolerance to nitrogen deficiency in European elite bread wheats grown under contrasting nitrogen stress scenarios.

KEY MESSAGE: Clustering 24 environments in four contrasting nitrogen stress scenarios enabled the detection of genetic regions determining tolerance to nitrogen deficiency in European elite bread wheats. Increasing the nitrogen use efficiency of wheat varieties is an important goal for breeding. However, most genetic studies of wheat grown at different nitrogen levels in the field report significant interactions with the genotype. The chromosomal regions possibly involved in these interactions are largely unknown. The objective of this study was to quantify the response of elite bread wheat cultivars to different nitrogen field stress scenarios and identify genomic regions involved in this response. For this purpose, 212 elite bread wheat varieties were grown in a multi-environment trial at different nitrogen levels. Genomic regions associated with grain yield, protein concentration and grain protein deviation responses to nitrogen deficiency were identified. Environments were clustered according to adjusted means for grain yield, yield components and grain protein concentration. Four nitrogen availability scenarios were identified: optimal condition, moderate early deficiency, severe late deficiency, and severe continuous deficiency. A large range of tolerance to nitrogen deficiency was observed among varieties, which were ranked differently in different nitrogen deficiency scenarios. The well-known negative correlation between grain yield and grain protein concentration also existed between their respective tolerance indices. Interestingly, the tolerance indices for grain yield and grain protein deviation were either null or weakly positive meaning that breeding for the two traits should be less difficult than expected. Twenty-two QTL regions were identified for the tolerance indices. By selecting associated markers, these regions may be selected separately or combined to improve the tolerance to N deficiency within a breeding programme.

Phenomic selection in wheat breeding: identification and optimisation of factors influencing prediction accuracy and comparison to genomic selection.

KEY MESSAGE: Phenomic selection is a promising alternative or complement to genomic selection in wheat breeding. Models combining spectra from different environments maximise the predictive ability of grain yield and heading date of wheat breeding lines. Phenomic selection (PS) is a recent breeding approach similar to genomic selection (GS) except that genotyping is replaced by near-infrared (NIR) spectroscopy. PS can potentially account for non-additive effects and has the major advantage of being low cost and high throughput. Factors influencing GS predictive abilities have been intensively studied, but little is known about PS. We tested and compared the abilities of PS and GS to predict grain yield and heading date from several datasets of bread wheat lines corresponding to the first or second years of trial evaluation from two breeding companies and one research institute in France. We evaluated several factors affecting PS predictive abilities including the possibility of combining spectra collected in different environments. A simple H-BLUP model predicted both traits with prediction ability from 0.26 to 0.62 and with an efficient computation time. Our results showed that the environments in which lines are grown had a crucial impact on predictive ability based on the spectra acquired and was specific to the trait considered. Models combining NIR spectra from different environments were the best PS models and were at least as accurate as GS in most of the datasets. Furthermore, a GH-BLUP model combining genotyping and NIR spectra was the best model of all (prediction ability from 0.31 to 0.73). We demonstrated also that as for GS, the size and the composition of the training set have a crucial impact on predictive ability. PS could therefore replace or complement GS for efficient wheat breeding programs.

Phenomic selection in wheat breeding: prediction of the genotype-by-environment interaction in multi-environment breeding trials.

KEY MESSAGE: Phenomic prediction of wheat grain yield and heading date in different multi-environmental trial scenarios is accurate. Modelling the genotype-by-environment interaction effect using phenomic data is a potentially low-cost complement to genomic prediction. The performance of wheat cultivars in multi-environmental trials (MET) is difficult to predict because of the genotype-by-environment interactions (G × E). Phenomic selection is supposed to be efficient for modelling the G × E effect because it accounts for non-additive effects. Here, phenomic data are near-infrared (NIR) spectra obtained from plant material. While phenomic selection has recently been shown to accurately predict wheat grain yield in single environments, its accuracy needs to be investigated for MET. We used four datasets from two winter wheat breeding programs to test and compare the predictive abilities of phenomic and genomic models for grain yield and heading date in different MET scenarios. We also compared different methods to model the G × E using different covariance matrices based on spectra. On average, phenomic and genomic prediction abilities are similar in all different MET scenarios. Better predictive abilities were obtained when G × E effects were modelled with NIR spectra than without them, and it was better to use all the spectra of all genotypes in all environments for modelling the G × E. To facilitate the implementation of phenomic prediction, we tested MET designs where the NIR spectra were measured only on the genotype-environment combinations phenotyped for the target trait. Missing spectra were predicted with a weighted multivariate ridge regression. Intermediate predictive abilities for grain yield were obtained in a sparse testing scenario and for new genotypes, which shows that phenomic selection is an efficient and practicable prediction method for dealing with G × E.

Using environmental clustering to identify specific drought tolerance QTLs in bread wheat (T. aestivum L.).

KEY MESSAGE: Environmental clustering helps to identify QTLs associated with grain yield in different water stress scenarios. These QTLs could be useful for breeders to improve grain yields and increase genetic resilience in marginal environments. Drought is one of the main abiotic stresses limiting winter bread wheat growth and productivity around the world. The acquisition of new high-yielding and stress-tolerant varieties is therefore necessary and requires improved understanding of the physiological and genetic bases of drought resistance. A panel of 210 elite European varieties was evaluated in 35 field trials. Grain yield and its components were scored in each trial. A crop model was then run with detailed climatic data and soil water status to assess the dynamics of water stress in each environment. Varieties were registered from 1992 to 2011, allowing us to test timewise genetic progress. Finally, a genome-wide association study (GWAS) was carried out using genotyping data from a 280 K SNP chip. The crop model simulation allowed us to group the environments into four water stress scenarios: an optimal condition with no water stress, a post-anthesis water stress, a moderate-anthesis water stress and a high pre-anthesis water stress. Compared to the optimal water condition, grain yield losses in the stressed conditions were 3.3%, 12.4% and 31.2%, respectively. This environmental clustering improved understanding of the effect of drought on grain yields and explained 20% of the G × E interaction. The greatest genetic progress was obtained in the optimal condition, mostly represented in France. The GWAS identified several QTLs, some of which were specific of the different water stress patterns. Our results make breeding for improved drought resistance to specific environmental scenarios easier and will facilitate genetic progress in future environments, i.e., water stress environments.

Association mapping for cold tolerance in two large maize inbred panels.

BACKGROUND: Breeding for cold tolerance in maize promises to allow increasing growth area and production in temperate zones. The objective of this research was to conduct genome-wide association analyses (GWAS) in temperate maize inbred lines and to find strategies for pyramiding genes for cold tolerance. Two panels of 306 dent and 292 European flint maize inbred lines were evaluated per se and in testcrosses under cold and control conditions in a growth chamber. We recorded indirect measures for cold tolerance as the traits number of days from sowing to emergence, relative leaf chlorophyll content or quantum efficiency of photosystem II. Association mapping for identifying genes associated to cold tolerance in both panels was based on genotyping with 49,585 genome-wide single nucleotide polymorphism (SNP) markers. RESULTS: We found 275 significant associations, most of them in the inbreds evaluated per se, in the flint panel, and under control conditions. A few candidate genes coincided between the current research and previous reports. A total of 47 flint inbreds harbored the favorable alleles for six significant quantitative trait loci (QTL) detected for inbreds per se evaluated under cold conditions, four of them had also the favorable alleles for the main QTL detected from the testcrosses. Only four dent inbreds (EZ47, F924, NK807 and PHJ40) harbored the favorable alleles for three main QTL detected from the evaluation of the dent inbreds per se under cold conditions. There were more QTL in the flint panel and most of the QTL were associated with days to emergence and ΦPSII. CONCLUSIONS: These results open new possibilities to genetically improve cold tolerance either with genome-wide selection or with marker assisted selection.

Phenomic Selection: A New and Efficient Alternative to Genomic Selection.

Recently, it has been proposed to switch molecular markers to near-infrared (NIR) spectra for inferring relationships between individuals and further performing phenomic selection (PS), analogous to genomic selection (GS). The PS concept is similar to genomic-like omics-based (GLOB) selection, in which molecular markers are replaced by endophenotypes, such as metabolites or transcript levels, except that the phenomic information obtained for instance by near-infrared spectroscopy (NIRS ) has usually a much lower cost than other omics. Though NIRS has been routinely used in breeding for several decades, especially to deal with end-product quality traits, its use to predict other traits of interest and further make selections is new. Since the seminal paper on PS , several publications have advocated the use of spectral acquisition (including NIRS and hyperspectral imaging) in plant breeding towards PS , potentially providing a scope of what is possible. In the present chapter, we first come back to the concept of PS as originally proposed and provide a classification of selected papers related to the use of phenomics in breeding. We further provide a review of the selected literature concerning the type of technology used, the preprocessing of the spectra, and the statistical modeling to make predictions. We discuss the factors that likely affect the efficiency of PS and compare it to GS in terms of predictive ability. Finally, we propose several prospects for future work and application of PS in the context of plant breeding.

Building a Calibration Set for Genomic Prediction, Characteristics to Be Considered, and Optimization Approaches.

The efficiency of genomic selection strongly depends on the prediction accuracy of the genetic merit of candidates. Numerous papers have shown that the composition of the calibration set is a key contributor to prediction accuracy. A poorly defined calibration set can result in low accuracies, whereas an optimized one can considerably increase accuracy compared to random sampling, for a same size. Alternatively, optimizing the calibration set can be a way of decreasing the costs of phenotyping by enabling similar levels of accuracy compared to random sampling but with fewer phenotypic units. We present here the different factors that have to be considered when designing a calibration set, and review the different criteria proposed in the literature. We classified these criteria into two groups: model-free criteria based on relatedness, and criteria derived from the linear mixed model. We introduce criteria targeting specific prediction objectives including the prediction of highly diverse panels, biparental families, or hybrids. We also review different ways of updating the calibration set, and different procedures for optimizing phenotyping experimental designs.

Genomic prediction for grain yield and micro-environmental sensitivity in winter wheat.

Individuals within a common environment experience variations due to unique and non-identifiable micro-environmental factors. Genetic sensitivity to micro-environmental variation (i.e. micro-environmental sensitivity) can be identified in residuals, and genotypes with lower micro-environmental sensitivity can show greater resilience towards environmental perturbations. Micro-environmental sensitivity has been studied in animals; however, research on this topic is limited in plants and lacking in wheat. In this article, we aimed to (i) quantify the influence of genetic variation on residual dispersion and the genetic correlation between genetic effects on (expressed) phenotypes and residual dispersion for wheat grain yield using a double hierarchical generalized linear model (DHGLM); and (ii) evaluate the predictive performance of the proposed DHGLM for prediction of additive genetic effects on (expressed) phenotypes and its residual dispersion. Analyses were based on 2,456 advanced breeding lines tested in replicated trials within and across different environments in Denmark and genotyped with a 15K SNP-Illumina-BeadChip. We found that micro-environmental sensitivity for grain yield is heritable, and there is potential for its reduction. The genetic correlation between additive effects on (expressed) phenotypes and dispersion was investigated, and we observed an intermediate correlation. From these results, we concluded that breeding for reduced micro-environmental sensitivity is possible and can be included within breeding objectives without compromising selection for increased yield. The predictive ability and variance inflation for predictions of the DHGLM and a linear mixed model allowing heteroscedasticity of residual variance in different environments (LMM-HET) were evaluated using leave-one-line-out cross-validation. The LMM-HET and DHGLM showed good and similar performance for predicting additive effects on (expressed) phenotypes. In addition, the accuracy of predicting genetic effects on residual dispersion was sufficient to allow genetic selection for resilience. Such findings suggests that DHGLM may be a good choice to increase grain yield and reduce its micro-environmental sensitivity.

Genome and Environment Based Prediction Models and Methods of Complex Traits Incorporating Genotype × Environment Interaction.

Genomic-enabled prediction models are of paramount importance for the successful implementation of genomic selection (GS) based on breeding values. As opposed to animal breeding, plant breeding includes extensive multienvironment and multiyear field trial data. Hence, genomic-enabled prediction models should include genotype × environment (G × E) interaction, which most of the time increases the prediction performance when the response of lines are different from environment to environment. In this chapter, we describe a historical timeline since 2012 related to advances of the GS models that take into account G × E interaction. We describe theoretical and practical aspects of those GS models, including the gains in prediction performance when including G × E structures for both complex continuous and categorical scale traits. Then, we detailed and explained the main G × E genomic prediction models for complex traits measured in continuous and noncontinuous (categorical) scale. Related to G × E interaction models this review also examine the analyses of the information generated with high-throughput phenotype data (phenomic) and the joint analyses of multitrait and multienvironment field trial data that is also employed in the general assessment of multitrait G × E interaction. The inclusion of nongenomic data in increasing the accuracy and biological reliability of the G × E approach is also outlined. We show the recent advances in large-scale envirotyping (enviromics), and how the use of mechanistic computational modeling can derive the crop growth and development aspects useful for predicting phenotypes and explaining G × E.

Genetic Analysis of Platform-Phenotyped Root System Architecture of Bread and Durum Wheat in Relation to Agronomic Traits.

Roots are essential for water and nutrient uptake but are rarely the direct target of breeding efforts. To characterize the genetic variability of wheat root architecture, the root and shoot traits of 200 durum and 715 bread wheat varieties were measured at a young stage on a high-throughput phenotyping platform. Heritability of platform traits ranged from 0.40 for root biomass in durum wheat to 0.82 for the number of tillers. Field phenotyping data for yield components and SNP genotyping were already available for all the genotypes. Taking differences in earliness into account, several significant correlations between root traits and field agronomic performances were found, suggesting that plants investing more resources in roots in some stressed environments favored water and nutrient uptake, with improved wheat yield. We identified 100 quantitative trait locus (QTLs) of root traits in the bread wheat panels and 34 in the durum wheat panel. Most colocalized with QTLs of traits measured in field conditions, including yield components and earliness for bread wheat, but only in a few environments. Stress and climatic indicators explained the differential effect of some platform QTLs on yield, which was positive, null, or negative depending on the environmental conditions. Modern breeding has led to deeper rooting but fewer seminal roots in bread wheat. The number of tillers has been increased in bread wheat, but decreased in durum wheat, and while the root-shoot ratio for bread wheat has remained stable, for durum wheat it has been increased. Breeding for root traits or designing ideotypes might help to maintain current yield while adapting to specific drought scenarios.

Scalable Sparse Testing Genomic Selection Strategy for Early Yield Testing Stage.

To enable a scalable sparse testing genomic selection (GS) strategy at preliminary yield trials in the CIMMYT maize breeding program, optimal approaches to incorporate genotype by environment interaction (GEI) in genomic prediction models are explored. Two cross-validation schemes were evaluated: CV1, predicting the genetic merit of new bi-parental populations that have been evaluated in some environments and not others, and CV2, predicting the genetic merit of half of a bi-parental population that has been phenotyped in some environments and not others using the coefficient of determination (CDmean) to determine optimized subsets of a full-sib family to be evaluated in each environment. We report similar prediction accuracies in CV1 and CV2, however, CV2 has an intuitive appeal in that all bi-parental populations have representation across environments, allowing efficient use of information across environments. It is also ideal for building robust historical data because all individuals of a full-sib family have phenotypic data, albeit in different environments. Results show that grouping of environments according to similar growing/management conditions improved prediction accuracy and reduced computational requirements, providing a scalable, parsimonious approach to multi-environmental trials and GS in early testing stages. We further demonstrate that complementing the full-sib calibration set with optimized historical data results in improved prediction accuracy for the cross-validation schemes.

Combining Crop Growth Modeling With Trait-Assisted Prediction Improved the Prediction of Genotype by Environment Interactions.

Plant breeders evaluate their selection candidates in multi-environment trials to estimate their performance in contrasted environments. The number of genotype/environment combinations that can be evaluated is strongly constrained by phenotyping costs and by the necessity to limit the evaluation to a few years. Genomic prediction models taking the genotype by environment interactions (GEI) into account can help breeders identify combination of (possibly unphenotyped) genotypes and target environments optimizing the traits under selection. We propose a new prediction approach in which a secondary trait available on both the calibration and the test sets is introduced as an environment specific covariate in the prediction model (trait-assisted prediction, TAP). The originality of this approach is that the phenotyping of the test set for the secondary trait is replaced by crop-growth model (CGM) predictions. So there is no need to sow and phenotype the test set in each environment which is a clear advantage over the classical trait-assisted prediction models. The interest of this approach, called CGM-TAP, is highest if the secondary trait is easy to predict with CGM and strongly related to the target trait in each environment (and thus capturing GEI). We tested CGM-TAP on bread wheat with heading date as secondary trait and grain yield as target trait. Simple CGM-TAP model with a linear effect of heading date resulted in high predictive abilities in three prediction scenarios (sparse testing, or prediction of new genotypes or of new environments). It increased predictive abilities of all reference GEI models, even those involving sophisticated environmental covariates.

Wheat individual grain-size variance originates from crop development and from specific genetic determinism.

Wheat grain yield is usually decomposed in the yield components: number of spikes / m2, number of grains / spike, number of grains / m2 and thousand kernel weight (TKW). These are correlated one with another due to yield component compensation. Under optimal conditions, the number of grains per m2 has been identified as the main determinant of yield. However, with increasing occurrences of post-flowering abiotic stress associated with climate change, TKW may become severely limiting and hence a target for breeding. TKW is usually studied at the plot scale as it represents the average mass of a grain. However, this view disregards the large intra-genotypic variance of individual grain mass and its effect on TKW. The aim of this study is to investigate the determinism of the variance of individual grain size. We measured yield components and individual grain size variances of two large genetic wheat panels grown in two environments. We also carried out a genome-wide association study using a dense SNPs array. We show that the variance of individual grain size partly originates from the pre-flowering components of grain yield; in particular it is driven by canopy structure via its negative correlation with the number of spikes per m2. But the variance of final grain size also has a specific genetic basis. The genome-wide analysis revealed the existence of QTL with strong effects on the variance of individual grain size, independently from the other yield components. Finally, our results reveal some interesting drivers for manipulating individual grain size variance either through canopy structure or through specific chromosomal regions.

BWGS: A R package for genomic selection and its application to a wheat breeding programme.

We developed an integrated R library called BWGS to enable easy computation of Genomic Estimates of Breeding values (GEBV) for genomic selection. BWGS, for BreedWheat Genomic selection, was developed in the framework of a cooperative private-public partnership project called Breedwheat (https://breedwheat.fr) and relies on existing R-libraries, all freely available from CRAN servers. The two main functions enable to run 1) replicated random cross validations within a training set of genotyped and phenotyped lines and 2) GEBV prediction, for a set of genotyped-only lines. Options are available for 1) missing data imputation, 2) markers and training set selection and 3) genomic prediction with 15 different methods, either parametric or semi-parametric. The usefulness and efficiency of BWGS are illustrated using a population of wheat lines from a real breeding programme. Adjusted yield data from historical trials (highly unbalanced design) were used for testing the options of BWGS. On the whole, 760 candidate lines with adjusted phenotypes and genotypes for 47 839 robust SNP were used. With a simple desktop computer, we obtained results which compared with previously published results on wheat genomic selection. As predicted by the theory, factors that are most influencing predictive ability, for a given trait of moderate heritability, are the size of the training population and a minimum number of markers for capturing every QTL information. Missing data up to 40%, if randomly distributed, do not degrade predictive ability once imputed, and up to 80% randomly distributed missing data are still acceptable once imputed with Expectation-Maximization method of package rrBLUP. It is worth noticing that selecting markers that are most associated to the trait do improve predictive ability, compared with the whole set of markers, but only when marker selection is made on the whole population. When marker selection is made only on the sampled training set, this advantage nearly disappeared, since it was clearly due to overfitting. Few differences are observed between the 15 prediction models with this dataset. Although non-parametric methods that are supposed to capture non-additive effects have slightly better predictive accuracy, differences remain small. Finally, the GEBV from the 15 prediction models are all highly correlated to each other. These results are encouraging for an efficient use of genomic selection in applied breeding programmes and BWGS is a simple and powerful toolbox to apply in breeding programmes or training activities.

Training set optimization of genomic prediction by means of EthAcc.

Genomic prediction is a useful tool for plant and animal breeding programs and is starting to be used to predict human diseases as well. A shortcoming that slows down the genomic selection deployment is that the accuracy of the prediction is not known a priori. We propose EthAcc (Estimated THeoretical ACCuracy) as a method for estimating the accuracy given a training set that is genotyped and phenotyped. EthAcc is based on a causal quantitative trait loci model estimated by a genome-wide association study. This estimated causal model is crucial; therefore, we compared different methods to find the one yielding the best EthAcc. The multilocus mixed model was found to perform the best. We compared EthAcc to accuracy estimators that can be derived via a mixed marker model. We showed that EthAcc is the only approach to correctly estimate the accuracy. Moreover, in case of a structured population, in accordance with the achieved accuracy, EthAcc showed that the biggest training set is not always better than a smaller and closer training set. We then performed training set optimization with EthAcc and compared it to CDmean. EthAcc outperformed CDmean on real datasets from sugar beet, maize, and wheat. Nonetheless, its performance was mainly due to the use of an optimal but inaccessible set as a start of the optimization algorithm. EthAcc's precision and algorithm issues prevent it from reaching a good training set with a random start. Despite this drawback, we demonstrated that a substantial gain in accuracy can be obtained by performing training set optimization.

Phenomic Selection Is a Low-Cost and High-Throughput Method Based on Indirect Predictions: Proof of Concept on Wheat and Poplar.

Genomic selection - the prediction of breeding values using DNA polymorphisms - is a disruptive method that has widely been adopted by animal and plant breeders to increase productivity. It was recently shown that other sources of molecular variations such as those resulting from transcripts or metabolites could be used to accurately predict complex traits. These endophenotypes have the advantage of capturing the expressed genotypes and consequently the complex regulatory networks that occur in the different layers between the genome and the phenotype. However, obtaining such omics data at very large scales, such as those typically experienced in breeding, remains challenging. As an alternative, we proposed using near-infrared spectroscopy (NIRS) as a high-throughput, low cost and non-destructive tool to indirectly capture endophenotypic variants and compute relationship matrices for predicting complex traits, and coined this new approach "phenomic selection" (PS). We tested PS on two species of economic interest (Triticum aestivum L. and Populus nigra L.) using NIRS on various tissues (grains, leaves, wood). We showed that one could reach predictions as accurate as with molecular markers, for developmental, tolerance and productivity traits, even in environments radically different from the one in which NIRS were collected. Our work constitutes a proof of concept and provides new perspectives for the breeding community, as PS is theoretically applicable to any organism at low cost and does not require any molecular information.