Molecular Breeding For Wood Physical Traits In Canadian Spruces: What Can We Expect From Association Genetics and Genomic Selection?

Spruces extend over most of Canada's boreal biome. Most active spruce breeding programs are now focusing on wood quality traits. Given the time delays and high costs for evaluating these traits, genomics-assisted selection (GAS) could contribute to increase gains per time unit. To develop GAS for wood traits in white spruce, a discovery population of 1700 trees has been assembled from a 30-year-old provenance-progeny trial. Increment cores were collected and wood physical traits assessed using SilviScan. A genotyping assay including SNPs belonging to about 2500 gene loci was developed to genotype the population for a total of 7000 SNPs. Association studies were carried out on single and multiple SNPs. Results showed that between 400 and 600 SNPs were significantly associated (P ≤ 0.05) with single traits before correction for false-discovery rate. The variation explained for each trait varied between 1.1 and 3.3%.

Prediction of genetic effects was also tested with models that included either all SNPs or only those that were significant at P ≤ 0.05 after association testing using cross-validation. Predictive ability varied between 0.2 and 0.3 for models based on the complete set of SNPs, whereas accuracy estimates were about 0.4 for the same traits.  When only significant SNPs were used, both predictive ability and accuracy estimates shifted to 0.6. These results suggest that significant gene SNPs cover a significant part of the QTLs underlying wood traits. Gains per time unit estimated were largely superior to those obtained by traditional approaches, hence resulting in reduced selection delays.