GBS and EUCHIP60K SNP data comparison for GWAS on a Eucalyptus dunnii breeding population in Argentina.

NATALIA C. AGUIRRE; PAMELA V. VILLALBA; CARLA FILIPPI; JUAN RIVAS; MARTIN N GARCIA; MARIA CAROLINA MARTINEZ; CINTIA ACUÑA; JAVIER LOPEZ; JUAN LOPEZ; PABLO S. PATHAUER; PALAZZINI, DINO A.; LEONEL HARRAND; JAVIER OBERSCHELP; MARTÍN MARCÓ; ESTEBAN F. CISNERO; ROCIO CARRERAS; HORACIO ESTEBAN HOPP; JOSÉ RODRIGUEZ; DARIO GRATTAPAGLIA; CAPPA, EDUARDO P; NORMA PANIEGO; MARCUCCI POLTRI, SUSANA N.

Raleigh

Conferencia; IUFRO Tree Biotechnology Conference; 2019

North Carolina State University

In Argentina, the Eucalyptus breeding program is mainly carried out by INTA to improve material for the forestry sector. Eucalyptus dunnii is one of the most demanded species by the cellulose industry. Its high growth stress (log end splitting ratio) limits its use as solid wood. Nevertheless, it grows in regions where other species are less adapted, being a good option for breeding proposes. To apply molecular breeding strategies into the E. dunnii genetic program, we compared the ability of an optimized ddRAD-seq protocol (GBS) and of the commercial EUChip60K (NEOGEN, USA) to estimate population parameters and implement a GWAS analysis. A population, coming from six different geographic Australian origins and a local selection of E. dunnii progeny trial, was genotyped and phenotyped for nine traits (DBH, stem straightness, growth stress and wood quality traits through NIR analysis), accounting for 308 individuals. The GBS sequences were mapped against the E. grandis v2.0 reference genome, and SNPs were called using Stacks 1.34 pipeline. Both SNPs matrices were filtered by minor allele frequency (MAF) lower than 0.01 and a percentage of missing data up to 30%, keeping a total of 14129 (GBS) and 19598 (EUChip60K) SNPs. The average number of SNPs per chromosome was 1248 and 1781, respectively. For GBS data, a more disperse distribution pattern was shown as expected due to the characteristics of the methodology. After imputation by the more frequent allele, DAPC population structure (adegenet R package) showed two genetic groups for both genotyping platforms (Fst values of 0.016 for GBS and 0.019 for EUChip60K). The linkage disequilibrium (r2, Synbreed R-package) was by mean 0.0051 (GBS) and 0.0067 (EUChip60K). A GWAS analysis was performed considering both genetic structure and kinship matrices (rrBLUP R-package, threshold = 1e-4), finding 26 and 40 SNPs (GBS and EUChip60K) associated with nine phenotypic traits studied. Although both genotyping strategies have shown similar results in terms of population parameters, they have presented different SNP distribution pattern, genome coverage and consequently associated genomic regions to phenotypic traits. To our knowledge, this is the first report which compares GBS and EUChip60K technologies for GWAS on a breeding E. dunnii population.