Premium
Boosting predictive ability of tropical maize hybrids via genotype‐by‐environment interaction under multivariate GBLUP models
Author(s) -
Dalsente Krause Matheus,
Dias Kaio Olímpio das Graças,
Pedroso Rigal dos Santos Jhonathan,
Oliveira Amanda Avelar,
Guimarães Lauro José Moreira,
Pastina Maria Marta,
Margarido Gabriel Rodrigues Alves,
Garcia Antonio Augusto Franco
Publication year - 2020
Publication title -
crop science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.76
H-Index - 147
eISSN - 1435-0653
pISSN - 0011-183X
DOI - 10.1002/csc2.20253
Subject(s) - hybrid , biology , best linear unbiased prediction , zea mays , multivariate statistics , boosting (machine learning) , missing data , predictive modelling , genotype , inbred strain , single nucleotide polymorphism , gene–environment interaction , grain yield , statistics , microbiology and biotechnology , agronomy , genetics , selection (genetic algorithm) , mathematics , computer science , machine learning , gene
Genomic selection has been implemented in several plant and animal breeding programs and it has proven to improve efficiency and maximize genetic gains. Phenotypic data of grain yield was measured in 147 maize ( Zea mays L.) single‐cross hybrids at 12 environments. Single‐cross hybrids genotypes were inferred based on their parents (inbred lines) via single nucleotide polymorphism (SNP) markers obtained from genotyping‐by‐sequencing (GBS). Factor analytic multiplicative genomic best linear unbiased prediction (GBLUP) models, in the framework of multienvironment trials, were used to predict grain yield performance of unobserved tropical maize single‐cross hybrids. Predictions were performed for two situations: untested hybrids (CV1), and hybrids evaluated in some environments but missing in others (CV2). Models that borrowed information across individuals through genomic relationships and within individuals across environments presented higher predictive accuracy than those models that ignored it. For these models, predictive accuracies were up to 0.4 until eight environments were considered as missing for the validation set, which represents 67% of missing data for a given hybrid. These results highlight the importance of including genotype‐by‐environment interactions and genomic relationship information for boosting predictions of tropical maize single‐cross hybrids for grain yield.