Analysis of seedling survival
In this study, seedling survival was monitored over three years (from
March 2016 to early September 2018), which differs from most previous
studies that only recorded seedling death or survival at the end of an
experiment. The survival time analysis offers a more powerful test and
inference on the JC effect. The Cox PH model was used to quantify hazard
ratios (HRs) of seedling mortality under different treatments. In the
Cox PH model, the three selected parent trees for each of the two focal
tree species were treated as a “cluster” (i.e., as a random effect) in
the ‘coxph’ function in the R ‘survival’ package
(http://survival.r-forge.r-project.org/).
Treatment effects on relative
abundances of plant-pathogenic and EcM
fungi
To quantify the treatment effects on the relative abundances
(frequencies) of plant-pathogenic and EcM fungi, which were presumed to
be responsible for seedling mortality, negative binomial GLMMs were used
to estimate the treatment effects of warming (OTC versuscontrol), pesticides (versus distilled water), and distance to
selected adult trees on the relative abundances of the two fungal guilds
(i.e., plant-pathogenic and EcM fungi). The total sequence counts of
fungal OTUs in each soil core were included as “weights” in the
negative binomial GLMMs ‘glmmTMB’ function in R package. Since our
interest was to test the effects of the three treatments on the relative
abundances of the two fungal guilds, variations between the two focal
tree species and among the three selected adult trees of each focal tree
species were treated as random effects.