Experimental Design
We replicated a factorial combination of nutrient addition and herbivore exclusion by fencing at 34 sites, which were part of the Nutrient Network distributed experiment (NutNet; Borer et al. 2014a). These sites were the subset of sites that met the following criteria: (1) 5 years of post-treatment; (2) 3 blocks; (3) each block containing a factorial design of nutrient addition and herbivore exclusion by fencing. A block typically spreads over 320 m2, and all three blocks typically spread over >1000 m2. These sites spanned a broad envelope of climate and included a wide range of grassland types such as montane, alpine, semiarid grasslands, prairies, old fields, pastures, savanna, tundra, and shrub-steppe (Table S1).
In each block at each site, two plots were fertilized with nitrogen (N), phosphorus (P), potassium (K), and micronutrients (Fe, S, Mg, Mn, Cu, Zn, B, and Mo in combination). Nutrients were added annually before the start of the growing season at rates of 10 gm-2y-1. N was supplied as time-release urea ((NH2)2CO) or ammonium nitrate (NH4NO3), there were no detectable differences between these N sources on plant biomass or diversity (Seabloom et al. 2015). P was supplied as triple superphosphate (Ca(H2PO4)2), and K as potassium sulfate (K2SO4). The micronutrient mix was applied once at the start of the experiment at a rate of 100 g m-2. Two plots (one in ambient conditions and the other receiving nutrient addition) were enclosed with fences to test the effects of herbivore exclusion (excluding mammalian herbivores > 50 g) on plant communities. Fences were around 230 cm tall, and the lower 90 cm were covered by 1cm woven wire mesh. To further exclude digging animals such as voles, an additional 30 cm outward-facing flange was stapled to the soil. Four strands of barbless wire were strung at similar vertical distances above the wire mesh. Six of the 34 sites deviated from this fence design (Table S2). Wild herbivores such as rodents, lagomorphs, ungulates, marsupials are present at all sites, while domestic herbivores such as sheep, yak, goats, and cattle are also present at a few sites (Table S2). Further details on the design are available in Borer et al. (2014b).