MODELOS DE PREDIÇÃO GENÔMICA ADITIVO E ADITIVO-DOMINANTE, PARA VOLUME E QUALIDADE DA MADEIRA EM Eucalyptus benthamii Maiden et Cambage

• Autor(a) principal: Paludeto, João Gabriel Zanon
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações do UNICENTRO
• Texto Completo: http://tede.unicentro.br:8080/jspui/handle/jspui/1310
• topic: Capacidade preditiva; efeito de dominância; viés de predição; GBLUP singlestep; Predictive ability; dominance effect; prediction bias; single-step GBLUP; CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL; RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL::MANEJO FLORESTAL

Eucalyptus benthamii is often indicated as a potential species for use in areas with severe frost. Although several species of the Eucalyptus genus, including E. benthamii, are considered fastgrowing, one of the major challenges in traditional forest breeding of the genus is the time required to complete a selective cycle. Genomic selection has shown great potential to significantly accelerate breeding programs, enabling a trustworthy early selection of genetically superior individuals. Thus, this study aimed to construct additive (GA), additive-dominant (GAD), single-step hybrid (H) and pedigree-based (At and Ag) predictive models for lignin (LIG), extractives (EXT), carbohydrates (CBO) and basic wood density (DBM) at four years of age and individual volume at six years of age (VOL6). The genomic association study (GWAS) presented a low number of associations, which explained from 0.9% to 6.5% of the genetic variation of traits. Genomic narrow-sense heritability ( ˆ2 a h ) ranged from 0.15 (EXT - GAD) to 0.70 (DBM – At). Genomic models performed better than pedigree models. The predictive abilities ( gy r ) ranged from 0.12 (VOL6 - Ag) to 0.44 (DBM – H/GA/GAD). The inclusion of the dominance effect (GAD model) positively influenced predictive ability and prediction bias for VOL6. The single-step hybrid model (H) has been shown to be applicable for genomic selection on a larger scale, including non-genotyped individuals, with a predictive ability equivalent or superior to the other evaluated models.