Molecular Plant Breeding 2024, Vol.15, No.5, 317-327 http://genbreedpublisher.com/index.php/mpb 322 primarily involved in cell wall biosynthesis and lignification processes, which are crucial for tree growth and wood quality. Furthermore, the joint GWAS method demonstrated repeated independent detection of the same SNP associations across different populations, providing unprecedented validation results for GWAS in forest trees. This approach has provided unprecedented validation results in forest trees, identifying genes involved in cell wall biosynthesis and lignification, which are critical for wood quality. Figure 3 Manhattan plots of the associations for total height (HT) using gene-based (31 770 genes) and segment-based (4 766 windows) joint genome-wide association study (Joint-GWAS) for the combined dataset using four unrelated Eucalyptus grandis ×E. urophyllahybrid breeding populations (Adopted from Müller et al., 2018) Image caption: (a) Gene-based Joint-GWAS adjusted for kinship matrix, age of measurements and population of origin. (b) Gene-based Joint-GWAS adjusted for all covariates mentioned before with the inclusion of population structure. (c) Segment-based Joint-GWAS adjusted for kinship matrix, age of measurements and population of origin. (d) Segment-based Joint-GWAS adjusted for all other covariates with the inclusion of population structure. Red line indicates Bonferroni-corrected threshold with an experimental type I error rate at α = 0.05, blue line indicates a false discovery rate (FDR) at 5% and green dashed line represents an ad hoc threshold of -log10 (P) = 4.0 (Adopted from Müller et al., 2018) 5.3 Genomic selection models for wood traits Implementing genomic selection models for predicting wood traits in Eucalyptus breeding programs has shown significant potential. For example, a study on Eucalyptus benthamii used various genomic models, including GBLUP and HBLUP, to estimate genetic parameters for lignin, extractives, carbohydrates, and wood density (Paludeto et al., 2021). The study found that genomic models provided more accurate predictions of trait values compared to pedigree-based models, with considerable dominance variance observed for all traits. This highlights the importance of considering non-additive genetic effects in genomic selection to improve overall selection efficiency. Another study on Eucalyptus globulus indicates that genomic prediction models such as Ridge Regression-BLUP (RRBLUP) and supervised PCR have high predictive ability for growth and wood quality traits, further supporting the use of genomic selection in breeding programs (Ballesta et al., 2018). The study assessed various GS methods, including RRBLUP, Bayes-A, Bayes-B, BLASSO, and PCR, using a chip containing 60 000 single nucleotide polymorphism (SNP) markers. The research found that RRBLUP-B and supervised PCR models exhibited the highest predictive ability for most traits studied, particularly for branching quality (PA approximately 0.7) and
RkJQdWJsaXNoZXIy MjQ4ODYzMg==