Enviromic Assembly Increases Accuracy and Reduces Costs of the Genomic Prediction for Yield Plasticity in Maize

Abstract

Quantitative genetics states that phenotypic variation is a consequence of the interaction between genetic and environmental factors. Predictive breeding is based on this statement, and because of this, ways of modeling genetic effects are still evolving. At the same time, the same refinement must be used for processing environmental information. Here, we present an “enviromic assembly approach,” which includes using ecophysiology knowledge in shaping environmental relatedness into whole-genome predictions (GP) for plant breeding (referred to as enviromic-aided genomic prediction, E-GP). We propose that the quality of an environment is defined by the core of environmental typologies and their frequencies, which describe different zones of plant adaptation. From this, we derived markers of environmental similarity cost-effectively. Combined with the traditional additive and non-additive effects, this approach may better represent the putative phenotypic variation observed across diverse growing conditions (i.e., phenotypic plasticity). Then, we designed optimized multi-environment trials coupling genetic algorithms, enviromic assembly, and genomic kinships capable of providing in-silico realization of the genotype-environment combinations that must be phenotyped in the field. As proof of concept, we highlighted two E-GP applications: (1) managing the lack of phenotypic information in training accurate GP models across diverse environments and (2) guiding an early screening for yield plasticity exerting optimized phenotyping efforts. Our approach was tested using two tropical maize sets, two types of enviromics assembly, six experimental network sizes, and two types of optimized training set across environments. We observed that E-GP outperforms benchmark GP in all scenarios, especially when considering smaller training sets. The representativeness of genotype-environment combinations is more critical than the size of multi-environment trials (METs). The conventional genomic best-unbiased prediction (GBLUP) is inefficient in predicting the quality of a yet-to-be-seen environment, while enviromic assembly enabled it by increasing the accuracy of yield plasticity predictions. Furthermore, we discussed theoretical backgrounds underlying how intrinsic envirotype-phenotype covariances within the phenotypic records can impact the accuracy of GP. The E-GP is an efficient approach to better use environmental databases to deliver climate-smart solutions, reduce field costs, and anticipate future scenarios.

Genomic prediction for canopy height and dry matter yield in alfalfa using family bulks

“…Genome-wide family prediction can be used in crops bred as family bulks because this approach uses family pools as the basis for measuring genotypes and phenotypes (Ashraf et al, 2016;Fè et al, 2015Fè et al, , 2016Rios et al, 2021). Several GS pipelines were developed to predict the phenotypic performance of genotypes under specific environmental conditions (Costa-Neto et al, 2021;Lorenzana & Bernardo, 2009). However, alfalfa stands can last up to 5 yr, and predicting the performance of genotypes using a single harvest might not represent the performance over time.…”

Envirome-Wide Associations Enhance Multi-Year Genome-Based Prediction of Historical Wheat Breeding Data

“…The so-called linear kernel-based reaction norm model of Jarquin et al (2014) was utilized by Pérez-Rodríguez et al (2015), employing pedigree and environmental covariates in cotton trials. Similar substantial gains in prediction accuracy of genotype performance were obtained by Cuevas et al (2016;, He et al (2019) and Costa-Neto et al (2021a) using non-linear genomic Gaussian kernel models for modeling G×E interaction over the conventional linear GBLUP kernel. However, to date, these studies have focused on genomicenabled predictions but have neglected two important components: first, the uncertainty of predictions in new environments, which can be substantial, as demonstrated by de los Campos et al (2020); and second, the detailed enviromics assessment needed for the quantification of soil and climatic variables.…”

“…All models were fitted using the EnvRtype ::kernel_model() functions, which is based on the kernel-optimized Hierarchical Bayesian approach implemented by the BGGE package (Granato et al, 2018). More details about this approach can be found at Costa-Neto et al (2021a,b). We considered 15,000 iterations, in which the first 5,000 were used as burn-in considering a thinning of 5.…”

“…Three statistics were used to measure the quality of the prediction: (1) average predictive ability ( pa ) at “model level”, as is conventionally done for testing models in genomic prediction; (2) genotype-specific predictability, based on the Spearman rank correlation between the predicted and observed values for each genotype in a new year, as suggested by each genotype (Costa-Neto et al, 2021a,b); (3) the relation between the predictive ability of the model and the proportion of the wheat population that was considered “predictable”, that is, with pa ≥ 0.2. Here we consider the combination of predictive ability and resolution as a practical measure of the accuracy and usefulness for plant breeders when making decisions for the following years.…”

“…To address this situation, we made some efforts in the past to develop public envirotyping pipelines, such as EnvRtype (Costa-Neto et al, 2021b), also tested the merit of nonlinear kernels for modeling the environmental similarities in genomic prediction (Costa-Neto et al, 2021a), and for designing new "environmental markers" to add value in the prediction platforms (Costa-Neto et al, 2021c) by increasing the resolution of the predictive models in reproducing the phenotypic plasticity of the plants, while supporting the optimization of training sets with less phenotyping effort. Other authors have also addressed this situation by including crop growth models as a supervised machine learning tool, deep learning approaches using long-term field trial data and using robust experimental trialing networks for training reaction-norm models using in-field sensors.…”