Joint GWAS Analysis for Growth Traits Across Four Eucalyptus Breeding Populations

GWAS studies in forest trees have targeted collections of individuals derived from natural populations aiming at gene discovery and the identification of the elusive quantitative trait nucleotide. All studies to date, however, have suffered severely from lack of power, especially for growth traits, the pillars of every tree improvement program. Notwithstanding the fact that less than a handful associations have been found for growth, explaining negligible fractions of heritability, it still remains to be seen how such associations found in natural populations, far removed from advanced material, would be translated into useful breeding information. In this study, we attempted a GWAS for growth by assembling a considerably larger association population. We leveraged the power of the EuCHIP60K, a 60K SNP chip developed for Eucalyptus, to assemble genome-wide SNP and growth traits data for 3,417 trees across four unrelated E. grandis x E. urophylla breeding populations. A mixed linear model association analysis was performed with SNPs fitted as fixed effects and the polygenic effect captured by the genomic relationship matrix as random effect. No significant association was found using each population individually. Despite the increased power of the joint GWAS experiment, few significant associations were detected for growth traits. Alternative analytical approaches accounting for relatedness in family-based data and also meta-analysis of GWAS from GBLUP evaluations are being explored. This study further corroborates the complex architecture of growth traits in Eucalyptus and suggests that considerably larger samples will be necessary to capture any relevant portion of the genetic variance for growth.