Well-connected trials show low genotype-by-environment interaction in Pinus radiata

• Published in: Forest ecology and management 2024-06, Vol.561, p.121887, Article 121887
• Main Authors: McLean, Duncan, Apiolaza, Luis A., Paget, Mark F., Klápště, Jaroslav
• Format: Article
• Language: English
• Subjects: Breeding strategy; Deployment; Genomic selection; Genotype-by-environment interaction; SNP
• ISSN: 0378-1127
• DOI: 10.1016/j.foreco.2024.121887

Selection in the New Zealand radiata pine (Pinus radiata D. Don) breeding program relies on wide-scale testing to adequately sample environmental variation. The program uses genomic selection for the early selection of parents for the next breeding cycle, but genomic selection may not perform as expected in the presence of crossover-type genotype-by-environment interaction (GxE) if such environments are poorly represented in the training population. This study uses empirical data to assess the magnitude of GxE to guide the selection and deployment strategy for radiata pine in New Zealand. Data was collected from eight well-connected and replicated cloned full-sib progeny trials across major radiata pine growing regions in New Zealand. We applied a second-order factor analytic model (FA2) with additive and non-additive variance components to characterise GxE. Three model types were used: uncorrected pedigree, marker-corrected pedigree and marker-based relatedness. This study found that the average additive genetic correlations among sites were 0.76 for DBH and 0.94 for DEN when estimated with marker-based relatedness. Models that use marker-based relatedness, without considering non-additive effects, provide a marginally superior fit compared to models that use pedigree or incorporate non-additive effects. Our study suggests that while GxE is present, its magnitude does not warrant regionalising (subdividing) radiata pine breeding zones for the North Island of New Zealand. •This study assesses the magnitude of genotype-by-environment interaction in New Zealand’s radiata pine breeding program.•Three model types were used: uncorrected pedigree, marker-corrected pedigree, and marker-based relatedness.•Average additive genetic correlations in genomic models were 0.76 for DBH and 0.94 for DEN.•Models using marker-based relatedness provided a marginally superior fit.•The study found insufficient support to regionalise the radiata pine breeding program.