Abstract
The application of quantitative genetics in plant and animal breeding has largely focused on additive models to estimate breeding values. The availability of dense panels of SNPs (Single Nucleotide Polymorphisms) have made it possible to estimate the realized genomic relationship which in turn allows partitioning the genetic variance into additive and non-additive components and the prediction of the total genetic value of individuals. We used and compared a systematic series of genomic prediction models accounting for additive, dominance, and first-order epistatic interactions (additive by additive, dominance by dominance, and additive by dominance), considering either pedigree- or genomic-based information for growth traits, wood basic density and pulp yield in a hybrid population of Eucalyptus urophylla × E.grandis. We showed that, compared to pedigree-derived information, the use of a realized genomic-based relationship matrix yields a substantially more precise separation of additive and non-additive components of genetic variance. In addition, phenotypic prediction accuracies were increased by including dominance effects for growth traits due to the large contribution of non-additive effects. This novel result improves our current understanding of the architecture of quantitative traits and recommends accounting for dominance variance when developing genomic selection strategies in hybrid Eucalyptus.
- Abbreviations
- AIC
- Akaike Information Criterion
- CBH
- circumference at breast height
- F1
- first generation population
- FDR
- false discovery rate
- GBLUP
- genomic-based best linear unbiased prediction
- h2
- narrow-sense heritability
- H2
- broad sense heritability
- LD
- linkage disequilibrium
- PCA
- principal component analysis
- REML
- residual maximum likelihood
- RRS-SF
- reciprocal recurrent selection with forward selection
- SNP
- single nucleotide polymorphism