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

Abstract

Identifying the additive genetic variation underlying complex traits is important for species of economic and/or ecological value. In particular, where DNA markers can be associated with trait variation they can be used to develop models to predict phenotypes as the basis of future selection and conservation programmes. Here, SNPs associated with growth (height and annual increment) and phenology (budburst and bud set) were identified in three closely related pine species including Pinus sylvestris (Scots pine). A genotyping array was used to screen 20,795 SNPs from coding regions for their association with trait variation using mixed linear model (MLM) and multilocus mixed model (MLMM) approaches: 113 SNPs located at 111 loci were significantly associated with the traits, with the majority associated with either budburst or growth increment in P. sylvestris. Common SNPs (MAF > 0.05) identified as significantly associated with bud set were found in genes putatively involved in only growth and development, whereas SNPs associated with growth and budburst were located in genes putatively involved in growth and development, response to environment and, to a lesser extent, reproduction. Predicted values estimated using the model for growth had highly significant correlations with phenotypes quantified in a P. sylvestris common environment experiment established at two sites in Scotland (YA and GS), but only at one of the sites (YA, height at 2020: r = 0.376, p < 0.001). Predicted values estimated with the model for budburst were found to be weakly but significantly correlated with duration of budburst at one of the field sites (GS, duration at 2018: r = 0.242, p = 0.012) and negatively associated with timing of budburst at the other (YA, stage six: r = −0.216, p = 0.033). Genomic prediction using the model for growth was more successful than random selection as a method of selecting tall trees at both sites. This study provides tentative support for the development of prediction models for traits that are of interest to both foresters and conservationists, while highlighting the need for caution when applying them to trees growing in different environments.