Genome-wide association reveals novel insights into the molecular mechanisms regulating stem volume inPinus taeda

Abstract

Pinus taeda(loblolly pine [LP]), a long-lived tree species, is one of the world's most economically important forest trees. Genetic improvement programs for loblolly and otherPinusspecies have focused on survival, biomass growth, resistance to diseases and pests, and stem shape. Among the growth traits, volume is the most widely considered in tree improvement programs. Despite the great interest in increasing volume growth, there are significant challenges to unraveling the molecular mechanisms behind this quantitative trait since it is presumably influenced by the action of large numbers of genes that may interact epistatically through unknown molecular mechanisms. The challenge of uncovering the genetic variants involved in variation for growth traits and their interaction is even greater in conifers such as LP because of the extremely large size and high complexity ofPinusgenomes. Here we present a comprehensive genetic analysis of LP involving genetic markers in association with volume (via genome-wide association studies [GWAS] and machine learning [ML]) to uncover pathways involved with good phenotypes for volume. The objective of this data integration was to provide a functional characterization of the genetic marker regions selected by GWAS and ML. We used a population of LP in the 2nd cycle of breeding and testing composed of full-sib progenies established at seven sites by the Cooperative Forest Genetics Research Program at the University of Florida. A total of 1,692 individuals were phenotyped and genotyped using sequence capture probes targeting putative genes. Probes were developed based on an elite germplasm transcriptome from LP. A total of 31,589 SNPs were identified and used to perform a GWAS through a multilocus mixed model. A precision core marker set with 7,864 SNPs selected through ML, i.e., feature selection (FS), was used for genomic selection (GS), providing a predictive accuracy (R Pearson coefficient) of approximately 0.79. For genome annotation and the construction of gene coexpression networks, three transcriptomes were assembled based on data from different pine species (Pinus taeda,Pinus elliottiiandPinus radiata). We selected genes associated with the target trait and assessed the cascade of related molecular mechanisms within coexpression networks. These results advance our understanding of the genetics influencing wood traits and reveal candidate genes for future functional studies as well as increase our understanding of quantitative genetics and the genomics of complex phenotypic variations in LP.