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

harvested from the slug exclosure rings than the control rings. May 2013 was the only harvest during the study in which the heat treatment had significantly more total plant biomass than the other climate treatments (Fig 3). Seasonal variation in harvested plant biomass was not consistent across years, with the least amount of plant biomass occurring in July 2012 (11.3 ± 1.1g average plant biomass per ring) and the greatest amount in August 2013 (45.5 ± 3.3g plant biomass; Fig 3). While additional precipitation tended to increase plant bio- mass at most harvests, the effect was only significant in Fall 2013 (Fig 3). In May 2012 & 2013, additional precipitation significantly reduced plant biomass in the combined climatic treat- ment (+heat+precipitation) compared to the +heat alone. Effect of climate on slug abundance and biomass Slug count, but not biomass, was significantly reduced in heated vs. non-heated treatments, but this climate treatment effect was modified by time (Tables 2 and 3). At one sampling date in winter, heated plots had greater slug populations than non-heated plots (average of 0.6 v. 0 slugs), but the heat effect was reversed in the late spring, mid-summer, and early fall in which control plots contained 62–95% more slugs (Fig 4). A significant time×precipitation interaction was also apparent for slug counts. Plots receiving additional precipitation had more slugs than plots that received no additional precipitation but only in mid-summer and early fall, but this trend was reversed on a single day in May (Fig 4). Overall time effects were

apparent, as number of slugs trapped was greater in late spring and early fall than throughout the rest of the experiment (Fig 4). Slug species was not considered in the analysis as only 4 of 325 slugs trapped in 2013 (1.2%) were not D. reticulatum. Arion hortensis (Fe´russac) was the only other slug species present and occurred in both control and precipitation plots, albeit very rarely. Forage quality Ordination analysis of the slug exclosure and control ring plant cover estimates showed two main groupings, which were defined by the heat treatment. Heated plots had distinctly differ- ent vegetative communities compared to the non-heated plots at the October 2013 harvest (Fig 5). Heated plots had greater abundance of the warm-season (C4 physiology) Bermuda grass (Cynodon dactylon (L.) Pers), while non-heated plots were dominated by cool-season (C3) spe- cies—tall fescue (Schedonorus arundinaceus (Schreb.) Durmort), buckthorn plantain (Plantago lanceolata L.), and Kentucky bluegrass (Poa pratensis L.). Pairwise comparison analysis using MRPP confirmed this result, yielding significant differences in all comparisons between heated and non-heated treatments (S1 Table). There was no grouping that suggested differences between control and slug exclosure rings, illustrating that slug exclusion for >2 years had no detectable effect on the plant community composition. Significant effects of heat, heat×precipitation, time, and heat×time were observed for C:N ratios of grab-sampled forage for the year in which slugs were trapped (Table 4). Heated plots

had higher average C:N values (± S.E.) than non-heated plots (30.3 ± 1.0 v. 27.3 ± 0.7, respec- tively). However, the effect of heat was only significant in summer and autumn harvests (Fig 6) when the plots were dominated by C4 grasses (unpublished data). The addition of precipita- tion to +heat and ambient (control) plots affected C:N differently. In +heat+precipitation treatments, C:N was subtly augmented compared to +heat alone; while in +precipitation treat- ments, C:N was slightly lower than in ambient control plots (Fig 6). In the spring, when plant community composition was more similar across climate treatments, C:N ratios did not differ (Fig 6). PLOS ONE Slug herbivory & climate change PLOS ONE | https://doi.org/10.1371/journal.pone.0283128 March 14, 2023 7 / 17