Mycelial biomass and concentration of loline alkaloids driven by complex population structure in Epichloë uncinata and meadow fescue (Schedonorus pratensis)

Many efforts have been made to select and isolate naturally occurring animal-friendly Epichloë strains for later reinfection into elite cultivars. Often this process involves large-scale screening of Epichloë-infected wild grass populations where strains are characterized and alkaloids measured. Here, we describe for the first time the use of genotyping-by-sequencing (GBS) on a collection of 217 Epichloë-infected grasses (7 S. arundinaceum, 4 L. perenne, and 206 S. pratensis). This genotyping strategy is cheaper than complete genome sequencing, is suitable for a large number of individuals, and, when applied to endophyte-infected grasses, conveniently genotypes both organisms. In total, 6273 single nucleotide polymorphisms (SNPs) in the endophyte data set and 38 323 SNPs in the host data set were obtained. Our findings reveal a composite structure with three distinct endophyte clusters unrelated to the three main S. pratensis gene pools that have most likely spread from different glacial refugia in Eurasia. All three gene pools can establish symbiosis with E. uncinata. A comparison of the endophyte clusters with microsatellite-based fingerprinting of the same samples allows a quick test to discriminate between these clusters using two simple sequence repeats (SSRs). Concentrations of loline alkaloids and mycelial biomass are correlated and differ significantly among the plant and endophyte subpopulations; one endophyte strain has higher levels of lolines than others, and one specific host genotype is particularly suitable to host E. uncinata. These findings pave the way for targeted artificial inoculations of specific host-endophyte combinations to boost loline production in the symbiota and for genome association studies with the aim of isolating genes involved in the compatibility between meadow fescue and E. uncinata.