Grazing and light modify Silene latifolia responses to nutrients and future climate

looking and well-developed leaves on each Silene transplant. SLA was measured directly after harvesting following the standard protocol [77,78] and averaged per transplant for the statisti- cal analysis. We only sampled leaves that did not show any grazing or other damage. Since most of the leaves on transplants outside the fences were completely eaten, we could not mea- sure foliar traits in grazed plots. Foliar traits were therefore only sampled from transplants inside the fences (i.e., from 40 individuals) and we could not test for a main nor interactive effect of grazing on the foliar traits. For the foliar C:N analysis we pooled the leaf material of each transplant and the chemical analysis was run with a VarioEL CNHS analyzer (Elemental Analyse Systeme GmbH, Hanau, Germany). After measurements of height measurements and collection of leaves, we harvested the aboveground biomass of Silene. Overall, 78 out of 80 indi- viduals were still present at the point of harvest (i.e., after one year). The harvested plants were oven dried at 60˚C for 48 hours and then weighed to obtain data on aboveground biomass allocation as a surrogate for plant performance [7,11]. As with height measurements, we recorded biomass in grazed and ungrazed plots to assess how the net effects of grazing on Silene performance depended on nutrient and light addition and future climate. Contrary to Silene leaves, which we could not sample outside the fences, we could still obtain biomass data from Silene outside the fences. Both the trait sampling and the harvesting took place from the

beginning to the middle of August 2020. Statistical analysis To assess the net effect of grazing under nutrient and light addition, future climate, and their interactions on Silene growth and performance, we applied linear mixed effects models with height and biomass as response variables, both in their own models, and fencing, fertilization, light addition, climate treatments and their interactions as explanatory variables (fixed fac- tors). Plot was nested within block as a random factor to account for the split-plot experimen- tal design. To examine how nutrients, light addition, future climate and their interactions affected foliar traits of Silene, we applied similar linear mixed effect models, where SLA and foliar C:N were response variables, both in their own models, and fertilization, light addition, climate treatments and their interactions as explanatory variables (fixed factors), using similar random variable as in models above. Overall, we ran four different models. All models were restricted to three-way interactions to prevent the models from overfitting. The significance of the single factors and their interactions was tested with F-test. We used model diagnostic plots to check the homogeneity of variances and the normality of errors [79]. Height and biomass were log transformed to meet these assumptions. We used the ‘lme4’ package [80] to run linear mixed effect models, the ‘car’ package for assessing the significance of the treatments effects [81], the ‘tidyverse’ package for data manip-

ulation [82], the ‘ggplot2’ package [83] and the ‘patchwork’ package [84] for plotting in R sta- tistical software [85, version 4.0.3]. Results The responses of Silene performance to grazing, nutrient and light additions, and future climate Silene grew tallest under future climate and when fertilized but not lighted (significant F x C x L interaction on height, Table 1, Fig 1A). This effect, as many other treatment effects, was espe- cially pronounced inside fences, although there was no significant interaction with fencing (Table 1). Fertilization increased Silene height by 52% (* 12 ± 4.4 cm) and aboveground bio- mass by 89% compared to unfertilized conditions (Table 1). Silene also attained greater PLOS ONE Transplant performance and trait responses to abiotic and biotic changes PLOS ONE | https://doi.org/10.1371/journal.pone.0276789 November 8, 2022 6 / 16