Host and Environmental Drivers of Foliar Fungal Abundance and Composition in Coastal Douglas-fir

Fungal communities can modulate foliar disease severity in trees, motivating further investigation into the extent that genotypic and environmental factors influence their assembly. Although compositional sequencing data obtained via standard microbiome profiling methods have generated important insights about host-associated fungal communities, these data may lead to misestimates of magnitude and direction of these effects in host systems with limited compositional turnover. Furthermore, the prospect of year-to-year variation in the garden environment and the dynamics of local fungal populations permits only limited interpretation of single-year studies. To address these challenges and test how host genotype and growing environment affect the foliar fungal community, we sampled second-year needles from more than 2,500 trees belonging to one of 15 coastal or 15 montane populations of Pseudotsuga menziesii var. menziesii, each grown at one of three coastal or three montane common gardens. We also sampled the same trees at a pair of coastal and montane common gardens three years in a row to account for annual variation in the environment and/or local fungal populations. Bypassing the issues of compositional data, we adapted the host-associated microbe PCR (hamPCR) metabarcoding technique to quantify a load index for each fungal taxon detected, incorporating fungal abundances into our analyses. Results from multivariate generalized linear modeling showed that needles from coastal gardens achieved higher loads of the biotrophic pathogen Nothophaeocryptopus gaeumannii than those from montane gardens, where loads of the necrotrophic pathogen Rhabdocline pseudotsugae peaked. Needles of trees sourced from coastal populations (grouped at the ecotype level) achieved lower incidences and/or abundances of pathogenic fungi relative to montane populations, but this pattern was reversed for commensalistic and mutualistic fungi. Garden and sampling year effects did not interact with the ecotype effect, but they did interact with each other. Overall, these results support a model where different tree ecotypes apply distinct (but equally plastic) filters that operate separately from fluctuating environmental filters.