Selection of trait-specific markers and multi-environment models improve genomic predictive ability in rice

Abstract

Developing high yielding rice varieties that are tolerant to drought stress is crucial for the sustainable livelihood of rice farmers in rainfed rice cropping ecosystems. Genomic selection (GS) promises to be an effective breeding option for these complex traits. We evaluated the effectiveness of two rather new options in the implementation of GS: trait and environment-specific marker selection and the use of multi-environment prediction models. A reference population of 280 rainfed lowland accessions endowed with 215k SNP markers data was phenotyped under a favorable and two managed drought environments. Trait-specific SNP subsets (28k) were selected for each trait under each environment, using results of GWAS performed with the complete genotype dataset. Performances of single-environment and multi-environment genomic prediction models were compared using kernel regression based methods (GBLUP and RKHS) under two cross validation scenarios: availability (CV2) or not (CV1) of phenotypic data for the validation set, in one of the environments. Trait-specific marker selection strategy achieved predictive ability (PA) of genomic prediction up to 22% higher than markers selected on the bases of neutral linkage disequilibrium (LD). Tolerance to drought stress was up to 32% better predicted by multi-environment models (especially RKHS based models) under CV2 strategy. Under the less favorable CV1 strategy, the multi-environment models achieved similar PA than the single-environment predictions. We also showed that reasonable PA could be obtained with as few as 3,000 SNP markers, even in a population of low LD extent, provided marker selection is based on pairwise LD. The implications of these findings for breeding for drought tolerance are discussed. The most resource sparing option would be accurate phenotyping of the reference population in a favorable environment and under a managed drought, while the candidate population would be phenotyped only under one of those environments.

Genomic prediction of arsenic tolerance and grain yield in rice. Contribution of trait-specific markers and multi environment models

“…to 32%. Similar gains of PA were reported in dairy cattle for milk yield and in rice for traits related to drought tolerance (Bhandari et al 2019). Using GRM built with markers identified through GWAS in Holstein-Friesian bulls, Veerkamp et al (2016) reported no increase in the accuracy of genomic predictions compared to GRM built with a large set of markers including or not trait-specific SNPs.…”

“…Hassen et al (2017), using DTF and GY data from two managed environments (AWD and CF), reported PA gains of up to 30% by using multi-environment models compared to their single-environment counterparts. Likewise, Bhandari et al (2019) using rice DTF and GY data from three managed drought experiments, reported PA gains of up to 32% with the RKHS multi-environment models. In the present study, the two multi-environment models (GBLUP and RKHS) provided similar substantial gains of PA, despite the fact that under the M×E model implemented with GBLUP, the environmental covariance is forced to be positive and constant across pairs of environments.…”

“…Analysis of the effect of four levels of marker density, on the PA of three single-environment genomic prediction models, confirmed the superfluity of very dense genotyping when the LD extent is large. Indeed, simulation studies (Habier et al 2009;Zhong et al 2009) and empirical studies in various plant species including rice (Ben Hassen et al 2017;Bhandari et al 2019) have shown very little, if any, effect of marker densities, as long as the distance between adjacent markers remains below the extent of LD. In the case of our Aus population the lowest and highest per-chromosome LD decay were 157 Kb and 499 Kb, respectively (Norton et al 2018).…”

“…They concluded that, when applied to data from managed G × E experiments (E being different levels of a given abiotic stress), multi-environment genomic prediction models allow to breed rice simultaneously for productivity and for adaptation to the target abiotic stress. Lastly, it was shown that accounting for trait-specific markers improves the predictive ability of genomic selection in rice (Bhandari et al 2019).…”

Genome-wide association of rice response to blast fungus identifies loci for robust resistance under high nitrogen.

“…Robustness is measured by comparing the mean phenotype between two environments or by measuring the variance of the phenotype between two environments [28]. A rst approach involves building a model based on large-scale multi-environment phenotyping in order to quantify Genome x Environment effects [29,30]. However, the disadvantage of this approach arises from di culties in nding reliable models that correlate traits and their environment drivers [29].…”