Identification of genomic regions for grain yield and yield stability and their epistatic interactions.

The task of identifying genomic regions conferring yield stability is challenging in any crop and requires large experimental data sets in conjunction with complex analytical approaches. We report findings of a first attempt to identify genomic regions with stable expression and their individual epistatic interactions for grain yield and yield stability in a large elite panel of wheat under multiple environments via a genome wide association mapping (GWAM) approach. Seven hundred and twenty lines were genotyped using genotyping-by-sequencing technology and phenotyped for grain yield and phenological traits. High gene diversity (0.250) and a moderate genetic structure (five groups) in the panel provided an excellent base for GWAM. The mixed linear model and multi-locus mixed model analyses identified key genomic regions on chromosomes 2B, 3A, 4A, 5B, 7A and 7B. Further, significant epistatic interactions were observed among loci with and without main effects that contributed to additional variation of up to 10%. Simple stepwise regression provided the most significant main effect and epistatic markers resulting in up to 20% variation for yield stability and up to 17% gain in yield with the best allelic combination.

Genomic Bayesian functional regression models with interactions for predicting wheat grain yield using hyper-spectral image data.

BACKGROUND: Modern agriculture uses hyperspectral cameras that provide hundreds of reflectance data at discrete narrow bands in many environments. These bands often cover the whole visible light spectrum and part of the infrared and ultraviolet light spectra. With the bands, vegetation indices are constructed for predicting agronomically important traits such as grain yield and biomass. However, since vegetation indices only use some wavelengths (referred to as bands), we propose using all bands simultaneously as predictor variables for the primary trait grain yield; results of several multi-environment maize (Aguate et al. in Crop Sci 57(5):1-8, 2017) and wheat (Montesinos-López et al. in Plant Methods 13(4):1-23, 2017) breeding trials indicated that using all bands produced better prediction accuracy than vegetation indices. However, until now, these prediction models have not accounted for the effects of genotype × environment (G × E) and band × environment (B × E) interactions incorporating genomic or pedigree information. RESULTS: In this study, we propose Bayesian functional regression models that take into account all available bands, genomic or pedigree information, the main effects of lines and environments, as well as G × E and B × E interaction effects. The data set used is comprised of 976 wheat lines evaluated for grain yield in three environments (Drought, Irrigated and Reduced Irrigation). The reflectance data were measured in 250 discrete narrow bands ranging from 392 to 851 nm (nm). The proposed Bayesian functional regression models were implemented using two types of basis: B-splines and Fourier. Results of the proposed Bayesian functional regression models, including all the wavelengths for predicting grain yield, were compared with results from conventional models with and without bands. CONCLUSIONS: We observed that the models with B × E interaction terms were the most accurate models, whereas the functional regression models (with B-splines and Fourier basis) and the conventional models performed similarly in terms of prediction accuracy. However, the functional regression models are more parsimonious and computationally more efficient because the number of beta coefficients to be estimated is 21 (number of basis), rather than estimating the 250 regression coefficients for all bands. In this study adding pedigree or genomic information did not increase prediction accuracy.

Genetic Yield Gains In CIMMYT's International Elite Spring Wheat Yield Trials By Modeling The Genotype × Environment Interaction.

We calculated the annual genetic gains for grain yield (GY) of wheat (Triticum aestivum L.) achieved over 8 yr of international Elite Spring Wheat Yield Trials (ESWYT), from 2006-2007 (27th ESWYT) to 2014-2015 (34th ESWYT). In total, 426 locations were classified within three main megaenvironments (MEs): ME1 (optimally irrigated environments), ME4 (drought-stressed environments), and ME5 (heat-stressed environments). By fitting a factor analytical structure for modeling the genotype × environment (G × E) interaction, we measured GY gains relative to the widely grown cultivar Attila (GYA) and to the local checks (GYLC). Genetic gains for GYA and GYLC across locations were 1.67 and 0.53% (90.1 and 28.7 kg ha-1 yr-1), respectively. In ME1, genetic gains were 1.63 and 0.72% (102.7 and 46.65 kg ha-1 yr-1) for GYA and GYLC, respectively. In ME4, genetic gains were 2.7 and 0.41% (88 and 15.45 kg ha-1 yr-1) for GYA and GYLC, respectively. In ME5, genetic gains were 0.31 and 1.0% (11.28 and 36.6 kg ha-1 yr-1) for GYA and GYLC, respectively. The high GYA in ME1 and ME4 can be partially attributed to yellow rust races that affect Attila. When G × E interactions were not modeled, genetic gains were lower. Analyses showed that CIMMYT's location at Ciudad Obregon, Mexico, is highly correlated with locations in other countries in ME1. Lines that were top performers in more than one ME and more than one country were identified. CIMMYT's breeding program continues to deliver improved and widely adapted germplasm for target environments.

Single-Step Genomic and Pedigree Genotype × Environment Interaction Models for Predicting Wheat Lines in International Environments.

Genomic prediction models have been commonly used in plant breeding but only in reduced datasets comprising a few hundred genotyped individuals. However, pedigree information for an entire breeding population is frequently available, as are historical data on the performance of a large number of selection candidates. The single-step method extends the genomic relationship information from genotyped individuals to pedigree information from a larger number of phenotyped individuals in order to combine relationship information on all members of the breeding population. Furthermore, genomic prediction models that incorporate genotype × environment interactions (G × E) have produced substantial increases in prediction accuracy compared with single-environment genomic prediction models. Our main objective was to show how to use single-step genomic and pedigree models to assess the prediction accuracy of 58,798 CIMMYT wheat ( L.) lines evaluated in several simulated environments in Ciudad Obregon, Mexico, and to predict the grain yield performance of some of them in several sites in South Asia (India, Pakistan, and Bangladesh) using a reaction norm model that incorporated G × E. Another objective was to describe the statistical and computational challenges encountered when developing the pedigree and single-step models in such large datasets. Results indicate that the genomic prediction accuracy achieved by models using pedigree only, markers only, or both pedigree and markers to predict various environments in India, Pakistan, and Bangladesh is higher (0.25-0.38) than prediction accuracy of models that use only phenotypic prediction (0.20) or do not include the G × E term.

Canopy Temperature and Vegetation Indices from High-Throughput Phenotyping Improve Accuracy of Pedigree and Genomic Selection for Grain Yield in Wheat.

Genomic selection can be applied prior to phenotyping, enabling shorter breeding cycles and greater rates of genetic gain relative to phenotypic selection. Traits measured using high-throughput phenotyping based on proximal or remote sensing could be useful for improving pedigree and genomic prediction model accuracies for traits not yet possible to phenotype directly. We tested if using aerial measurements of canopy temperature, and green and red normalized difference vegetation index as secondary traits in pedigree and genomic best linear unbiased prediction models could increase accuracy for grain yield in wheat, Triticum aestivum L., using 557 lines in five environments. Secondary traits on training and test sets, and grain yield on the training set were modeled as multivariate, and compared to univariate models with grain yield on the training set only. Cross validation accuracies were estimated within and across-environment, with and without replication, and with and without correcting for days to heading. We observed that, within environment, with unreplicated secondary trait data, and without correcting for days to heading, secondary traits increased accuracies for grain yield by 56% in pedigree, and 70% in genomic prediction models, on average. Secondary traits increased accuracy slightly more when replicated, and considerably less when models corrected for days to heading. In across-environment prediction, trends were similar but less consistent. These results show that secondary traits measured in high-throughput could be used in pedigree and genomic prediction to improve accuracy. This approach could improve selection in wheat during early stages if validated in early-generation breeding plots.

Wheat quality improvement at CIMMYT and the use of genomic selection on it.

The International Center for Maize and Wheat Improvement (CIMMYT) leads the Global Wheat Program, whose main objective is to increase the productivity of wheat cropping systems to reduce poverty in developing countries. The priorities of the program are high grain yield, disease resistance, tolerance to abiotic stresses (drought and heat), and desirable quality. The Wheat Chemistry and Quality Laboratory has been continuously evolving to be able to analyze the largest number of samples possible, in the shortest time, at lowest cost, in order to deliver data on diverse quality traits on time to the breeders for making selections for advancement in the breeding pipeline. The participation of wheat quality analysis/selection is carried out in two stages of the breeding process: evaluation of the parental lines for new crosses and advanced lines in preliminary and elite yield trials. Thousands of lines are analyzed which requires a big investment in resources. Genomic selection has been proposed to assist in selecting for quality and other traits in breeding programs. Genomic selection can predict quantitative traits and is applicable to multiple quantitative traits in a breeding pipeline by attaining historical phenotypes and adding high-density genotypic information. Due to advances in sequencing technology, genome-wide single nucleotide polymorphism markers are available through genotyping-by-sequencing at a cost conducive to application for genomic selection. At CIMMYT, genomic selection has been applied to predict all of the processing and end-use quality traits regularly tested in the spring wheat breeding program. These traits have variable levels of prediction accuracy, however, they demonstrated that most expensive traits, dough rheology and baking final product, can be predicted with a high degree of confidence. Currently it is being explored how to combine both phenotypic and genomic selection to make more efficient the genetic improvement for quality traits at CIMMYT spring wheat breeding program.

Increased prediction accuracy in wheat breeding trials using a marker × environment interaction genomic selection model.

Genomic selection (GS) models use genome-wide genetic information to predict genetic values of candidates of selection. Originally, these models were developed without considering genotype × environment interaction(G×E). Several authors have proposed extensions of the single-environment GS model that accommodate G×E using either covariance functions or environmental covariates. In this study, we model G×E using a marker × environment interaction (M×E) GS model; the approach is conceptually simple and can be implemented with existing GS software. We discuss how the model can be implemented by using an explicit regression of phenotypes on markers or using co-variance structures (a genomic best linear unbiased prediction-type model). We used the M×E model to analyze three CIMMYT wheat data sets (W1, W2, and W3), where more than 1000 lines were genotyped using genotyping-by-sequencing and evaluated at CIMMYT's research station in Ciudad Obregon, Mexico, under simulated environmental conditions that covered different irrigation levels, sowing dates and planting systems. We compared the M×E model with a stratified (i.e., within-environment) analysis and with a standard (across-environment) GS model that assumes that effects are constant across environments (i.e., ignoring G×E). The prediction accuracy of the M×E model was substantially greater of that of an across-environment analysis that ignores G×E. Depending on the prediction problem, the M×E model had either similar or greater levels of prediction accuracy than the stratified analyses. The M×E model decomposes marker effects and genomic values into components that are stable across environments (main effects) and others that are environment-specific (interactions). Therefore, in principle, the interaction model could shed light over which variants have effects that are stable across environments and which ones are responsible for G×E. The data set and the scripts required to reproduce the analysis are publicly available as Supporting Information.