Improving genomic prediction of growth and wood traits in Eucalyptus using phenotypes from non-genotyped trees by single-step GBLUP
Genomic Best Linear Unbiased Prediction (GBLUP) in tree breeding typically only uses information from genotyped trees. However, information from phenotyped but non-genotyped trees can also be highly valuable. The single-step GBLUP approach(ssGBLUP) allows genomic prediction to takeinto account both genotyped and nongenotyped trees simultaneously in a single evaluation. In this study, we investigated the advantage, in terms of breeding value accuracy and bias, of including phenotypic observation from non-genotyped trees in a standard tree GBLUP evaluation. We compared the efficiency of the conventional pedigree-based (ABLUP), GBLUP and ssGBLUP approaches to evaluate eight growth and wood quality traits in a Eucalyptus hybrid population, genotyped with 33,398 single nucleotide polymorphisms (SNPs) using the EucHIP60k. Theoretical accuracies, predictive ability and bias were calculated by ten-fold cross validation on all traits. The use of additional phenotypic information from non-genotyped trees by means of ssGBLUP provided higher predictive ability (from 37% to 75%) and lower prediction bias (from 21% to 73%) for the genetic component of non-phenotyped but genotyped trees when compared to GBLUP. The increase (decrease) in the prediction accuracy (bias) became stronger as trait heritability decreased. We concluded that ssGBLUP is a promising breeding tool to improve accuracies and bias over classical GBLUP for genomic evaluation in Eucalyptus breeding practice.
Main Authors: | , , , , , |
---|---|
Format: | info:ar-repo/semantics/artículo biblioteca |
Language: | eng |
Published: |
Elsevier
2019-03-28
|
Subjects: | Eucalyptus, Evaluación, Información Fenotípico, Genomic Features, Evaluation, Phenotypic Information, Genetics, Genética, Accuracy Bias, Sesgo de Precisión, Características Genómicas, |
Online Access: | https://www.sciencedirect.com/science/article/pii/S0168945218314134 http://hdl.handle.net/20.500.12123/6227 https://doi.org/10.1016/j.plantsci.2019.03.017 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Genomic Best Linear Unbiased Prediction (GBLUP) in tree breeding typically only uses information from genotyped trees. However, information from phenotyped but non-genotyped trees can also be highly valuable. The single-step GBLUP approach(ssGBLUP) allows genomic prediction to takeinto account both genotyped and nongenotyped trees simultaneously in a single evaluation. In this study, we investigated the advantage, in terms of breeding value accuracy and bias, of including phenotypic observation from non-genotyped trees in a standard tree GBLUP evaluation. We compared the efficiency of the conventional pedigree-based (ABLUP), GBLUP and ssGBLUP approaches to evaluate eight growth and wood quality traits in a Eucalyptus hybrid population, genotyped with 33,398 single nucleotide polymorphisms (SNPs) using the EucHIP60k. Theoretical accuracies, predictive ability and bias were calculated by ten-fold cross validation on all traits. The use of additional phenotypic information from non-genotyped trees by means of ssGBLUP provided higher predictive ability (from 37% to 75%) and lower prediction bias (from 21% to 73%) for the genetic component of non-phenotyped but genotyped trees when compared to GBLUP. The increase (decrease) in the prediction accuracy (bias) became stronger as trait heritability decreased. We concluded that ssGBLUP is a promising breeding tool to improve accuracies and bias over classical GBLUP for genomic evaluation in Eucalyptus breeding practice. |
---|