Identifying and testing marker-trait associations for growth and phenology in three pine species: Implications for genomic prediction.

In tree species, genomic prediction offers the potential to forecast mature trait values in early growth stages, if robust marker–trait associations can be identified. Here we apply a novel multispecies approach using genotypes from a new genotyping array, based on 20,795 single nucleotide polymorphisms (SNPs) from three closely related pine species (Pinus sylvestris, Pinus uncinata and Pinus mugo), to test for associations with growth and phenology data from a common garden study. Predictive models constructed using significantly associated SNPs were then tested and applied to an independent multisite field trial of P. sylvestris and the capability to predict trait values was evaluated. One hundred and eighteen SNPs showed significant associations with the traits in the pine species. Common SNPs (MAF > 0.05) associated with bud set were only found in genes putatively involved in growth and development, whereas those associated with growth and budburst were also located in genes putatively involved in response to environment and, to a lesser extent, reproduction. At one of the two independent sites, the model we developed produced highly significant correlations between predicted values and observed height data (YA, height 2020: r = 0.376, p < 0.001). Predicted values estimated with our budburst model were weakly but positively correlated with duration of budburst at one of the sites (GS, 2015: r = 0.204, p = 0.034; 2018: r = 0.205, p = 0.034–0.037) and negatively associated with budburst timing at the other (YA: r = −0.202, p = 0.046). Genomic prediction resulted in the selection of sets of trees whose mean height was taller than the average for each site. Our results provide tentative support for the capability of prediction models to forecast trait values in trees, while highlighting the need for caution in applying them to trees grown in different environments.

在树种中，如果能够确定可靠的标记 - 性状关联，基因组预测就有可能预测早期生长阶段的成熟性状值。在此，我们应用一种新的多物种方法，使用来自一种新的基因分型阵列的基因型，该阵列基于来自三个密切相关的松树物种（欧洲赤松、偃松和中欧山松）的20,795个单核苷酸多态性（SNP），来测试与一个公共园林研究中的生长和物候数据的关联。然后对使用显著相关的SNP构建的预测模型进行测试，并将其应用于欧洲赤松的一个独立多地点田间试验，同时评估预测性状值的能力。118个SNP与松树物种的性状显示出显著关联。与芽形成相关的常见SNP（最小等位基因频率>0.05）仅在假定参与生长和发育的基因中被发现，而与生长和芽萌发相关的SNP也位于假定参与对环境响应的基因中，在较小程度上还位于与繁殖相关的基因中。在两个独立地点中的一个，我们开发的模型在预测值和观测到的树高数据之间产生了极显著的相关性（YA，2020年树高：r = 0.376，p < 0.001）。用我们的芽萌发模型估计的预测值在其中一个地点与芽萌发持续时间呈弱正相关（GS，2015年：r = 0.204，p = 0.034；2018年：r = 0.205，p = 0.034 - 0.037），在另一个地点与芽萌发时间呈负相关（YA：r = -0.202，p = 0.046）。基因组预测导致所选树木组的平均树高高于每个地点的平均值。我们的结果为预测模型预测树木性状值的能力提供了初步支持，同时强调在将其应用于不同环境中生长的树木时需要谨慎。