Genome-wide analysis of essential oil yield variation in Eucalyptus polybractea

Abstract

Essential oil found in the leaves of Myrtaceous species, stored in specialised sub-epidermic secretory cavities, consists mostly of a large variety of terpenoid compounds. One such oil, Eucalyptus oil, is produced from a number of “oil mallee” species with high total foliar oil concentration, high proportion of the monoterpene 1,8-cineole and the ability to re-sprout with multiple stems from lignotubers after coppicing. The yield of foliar oil in such commercially harvested perennial species (e.g. eucalypts, Tea Trees and Hop) is dependent on complex quantitative traits such as foliar oil concentration, leafy biomass accumulation and adaptability. These often show large natural variation and some are highly heritable, which has enabled significant gains in oil yield via traditional phenotypic recurrent selection. However, molecular breeding techniques could increase gains per unit time by improving the accuracy of selection and reducing cycle time.

In this thesis I explore the pathway to implementing genomic selection for essential oil traits in Eucalyptus polybractea (blue mallee). This begins with a general review of the challenges of breeding in perennial essential oil crops. I discuss the potential for applying genomic selection (GS) to improve oil yield, while noting the factors that affect GS accuracy and how they may manifest in openpollinated tree populations. Next, using non-destructive methods I assess traits relating to oil yield (quantitative and qualitative variation of foliar essential oils and biomass-related parameters) for their variability, heritability as well as phenotypic and genetic interactions in an open-pollinated progeny trial with 40 families and 480 individuals of E. polybractea. From raw phenotypes I develop a model that is able to predict future harvest oil yield performance at the family-level with a rank correlation of r = 0.74. This study shows that relying on oil concentration and 1,8-cineole proportion alone is not ideal for selection of top performing families for oil yield. Rather a mixture of biomass related traits, foliar oil concentration, 1,8-cineole proportion and leaf architecture contribute to family-level oil yield in varying ways.

To implement genomic selection it is important to understand the genetic architecture of the trait under selection. To this end I use whole genome re-sequencing of 480 blue mallees to perform a GWAS of eleven oil yield traits. I find that allelic variants in the pathways involved in the biosynthesis of terpenes are not necessarily the major driver of foliar oil concentration when viewed at the genome-wide level rather than at candidate-gene level. I also reveal additional candidate genes that may be involved in precursor availability for terpene biosynthesis, terpene transport and the formation of oil secretory cavities. The GWAS widens our understanding of the genetic basis of essential oil variation to the genomic scale, while also providing an informative set of priors for advanced genomic selection models that make use of such information.

GS models face a problem of over-parameterization when fitting large numbers of SNPs obtained from whole genome sequencing since most SNPs are uninformative. Therefore I implement a modified G-BLUP model that weights specific SNPs according to the trait genetic architecture. I show that by using curated candidate gene information the accuracy of prediction for total oil concentration can be improved by 15-50% over standard G-BLUP. Finally, this philosophy of partitioning genomic data into parts to be modelled differently based on a-priori knowledge is well established in phylogenetics. I explore the effects of different approaches to partitioning in the context of phylogenetics, noting that poor partitioning can result in misleading outcomes.

In general, this thesis broadens our understanding of the genetic basis of quantitative oil traits, and shows how that information can be used to more accurately predict genetic value in breeding populations. Specific terpenes are increasingly sought after for industrial purposes, such as advanced biofuels, so this knowledge may facilitate increased production of key terpenes through either plant-based systems or engineered pathways