Study site and experimental design
We initiated this study in 2016, in a glasshouse and in large mesocosm pots set in a fallow field on the Lincoln University campus (Lincoln, New Zealand. 43.6434o S, 172.4678oE, elevation 10 m), and completed it in 2019 (Waller et al.2020). We designed 20 communities, each containing eight unique plant species taken from a pool of 39 species (species details listed in Table S1), and grown in 125 L pots (575 mm dia.). Full methods are described in Waller et al. (2020), but briefly summarized here. Communities varied orthogonally in their proportion of exotic plant species (0, 25, 50, 75, 100%) and woody plant species (0, 25, 38, 63%). Each of the 20 plant communities was replicated four times (80 pots total) and received one of two soil treatments, described below. We included an additional 80 pots with a treatment with addition of herbivores (reported in Walleret al. 2020 and Allen et al. 2021). However, that treatment was not relevant for our hypotheses, so is excluded from the current analysis.