Climate change alters slug abundance but not herbivory in a temperate grassland

in R Studio (Version 4.2.0; [40]). For all models, heat and precipitation climate treatments were fixed effects. The slug herbivory models included additional fixed effects of exclosure and time, and the biomass quality models also had a fixed effect of time. Block and its interaction with the highest order interaction (i.e., heat×precipitation in the slug count model or heat×- precipitation×treatment in the slug herbivory model) were specified as the random terms to properly calculate of degrees of freedom. If a random term returned an estimated variance of 0 (‘boundary is singular’), the random term with an estimated variance of 0 was removed from the model. Type III Analysis of Variance with Satterthwaite’s method was used. Pairwise com- parisons were conducted on significant effects (p < 0.05) with emmeans. Differences in aboveground plant biomass data for modeling slug herbivory were normally distributed as determined by the Anderson-Darling test on four of seven harvest dates. The three harvests which were not normally distributed were determined to be within tolerances of an ANOVA model [41], so no transformation was performed. Slug number was square root transformed. The square root of count and the estimated mass of slugs trapped were not nor- mally distributed but were determined to be appropriate for analysis by ANOVA by visual assessment of histograms [41]. On dates in which no slugs were trapped, no mass, rather than a mass of 0, was used for statistical analysis. Zero values were retained in count data.

Plant species cover data taken from the slug exclosure and control rings prior to the Octo- ber 2013 harvest were analyzed using non-metric multidimensional scaling ordination in PCORD (NMS; PCORD version 6.0, MjM Software, Gleneden Beach, OR). Pairwise compari- sons of the climate treatment plant species cover estimates were performed with MRPP (multi-response permutation procedures) analysis, and p-values were corrected using the Bon- ferroni-adjustment for the number of comparisons (α = 0.05/6: αadj = 0.008). Results Effect of climate on slug herbivory Over the entire two-and-a-half-year duration of the study, slug exclusion had no overall effect on harvested aboveground plant biomass (Table 1; Fig 2). On the May 2013 harvest date, there was a significant effect of slug exclusion in the heated plots with greater plant biomass Table 1. ANOVA results illustrating the effect of slug exclusion on aboveground plant biomass within the climate treatment plots over time. Degrees of freedom are shown as numerator, denominator. Bolding indicates significant main effects and interactions. Effect DF F-Value P-Value heat 1,30 4.88 0.0349 precip 1,30 2.54 0.1210 heat×precip 1,30 2.41 0.1313 exclosure 1,30 0.64 0.4290 heat×exclosure 1,30 0.45 0.5091 precip×exclosure 1,30 1.36 0.2530 heat×precip×exclosure 1,30 0.00 0.9965 time 6,180 60.46 <0.0001 heat×time 6,180 11.68 <0.0001 precip×time 6,180 6.33 <0.0001 heat×precip×time 6,180 3.53 0.0025 exclosure×time 6,180 0.58 0.7447 heat×exclosure×time 6,180 0.21 0.9720

precip×exclosure×time 6,180 0.32 0.9252 heat×precip×exclosure×time 6,180 0.52 0.7945 https://doi.org/10.1371/journal.pone.0283128.t001 PLOS ONE Slug herbivory & climate change PLOS ONE | https://doi.org/10.1371/journal.pone.0283128 March 14, 2023 6 / 17