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

paired collars within a plot. If slug exclosures had more plant biomass measured than the adja- cent control ring, then slug herbivory had occurred at the measured level of consumption. If the two rings were equal in plant biomass or biomass of the slug exclosure exceeded the con- trol, then no discernible slug herbivory was detected. Slug abundance Live traps were set at regular intervals during the final year of study to quantify differences in slug abundance between the climate treatments. From several preliminary trials (unpublished data), it was determined that the most reliable method of live trapping was to place a 100 mm diameter petri dish with ~10 g of wet cat food (Friskies Classic Pate´, Nestle Purina Petcare Company, St. Louis, MO, U.S.A.) on the ground surface in a plot for 15 minutes just after dawn. This method attracted the greatest number of slugs consistently and is similar to ‘saucer trapping’ in that the area from where the slugs originate is unknown. Therefore, slug density could not be calculated [38]. Trapping was conducted every two weeks from January 3, 2013, until October 10, 2013. Slugs were released in the plot from which they were trapped. Preliminary methods testing over six months (unpublished data) revealed the importance of concluding the trapping no later than 45 minutes after sunrise, as catch efficacy of the traps was significantly reduced after this time. Starting at sunrise, baited dishes were placed ran- domly in each of the four plots within the first of five blocks and left for 15 minutes. Baited

dishes were placed in the second experimental block 5 minutes after dishes were placed in the first block. At any time, two blocks were being trapped so that their 15-minute periods were on a staggered schedule, allowing for all 5 blocks to be trapped within 45 minutes. The order in which blocks were trapped was determined by a random sequence generated by PROC PLAN in SAS (SAS Institute Inc. 9.3, Cary, NC, U.S.A) the day of each trapping event. After 15 minutes, each dish was examined, and the length, number, and species of each slug on the dish was recorded. Additional slug trapping occurred outside the experimental plots from August 15 to Octo- ber 15, 2013. These slugs were taken to the laboratory and used to quantify the relationship between slug length and dry biomass. Slug length was measured in the same way as occurred during plot trapping and then slugs were dried in a 105˚C forced air drying oven for 24 hours and weighed. The length versus biomass curve was used to estimate the biomass of slugs caught during trapping so that it was only necessary to measure the length of slugs when live- trapping as this was a time sensitive process. The curve was determined using the model ln (mass) = A + B�ln(length) in SAS using PROC REG as discussed in Hawkins, Lankester, and Lautenshlager [39]. The model yielded the equation: mass(mg) = 0.007�length(mm)2.92 with an R2 value of 0.5855 (p = 0.0099, n = 10, S2 Fig). The majority of slugs captured (84%) had lengths that fell within the range of the curve used to estimate mass (S3 Fig).

Forage quality & plant community composition Carbon and nitrogen (%C and %N) analysis (FlashEA 1112 NC Soil Analyzer, ThermoFisher Scientific, Waltham, MA, U.S.A.) was performed on a random grab sample of forage taken from each plot after it had been mowed across the study period. Visual cover estimates were recorded by species (to 1%) for each slug exclosure and control ring (n = 40) immediately pre- ceding the final harvest in October 2013. Statistical analysis Slug herbivory, slug count and biomass, and plant biomass quality across climate treatments were analyzed as a 2×2 climate treatment factorial (heat×precipitation) with the lme4 package PLOS ONE Slug herbivory & climate change PLOS ONE | https://doi.org/10.1371/journal.pone.0283128 March 14, 2023 5 / 17