Trait-mediated community assembly during experimental grassland restoration is altered by planting year rainfall

1. Ecological restoration outcomes are highly variable, undermining efforts to recover biodiversity and ecosystem functions. One poorly understood source of variability is `year effects'-interannual variation in environmental conditions during the first year of restoration that alter successional trajectories of plant communities. 2. There have been few experimental tests disentangling planting years from other differences among restoration projects (e.g. edaphic conditions, restoration approach), particularly those resolving mechanisms for year effects such as planting-year rainfall. Moreover, past year effect studies focused almost exclusively on species-level consequences. Therefore, the extent to which year effects influence the traits of communities is unknown. 3. To address these gaps and provide a mechanistic test of how precipitation contributes to year effects, we conducted an experiment where we manipulated rainfall (drought, average and high levels) during the first growing season, replicated across three establishment year treatments to disentangle the effects of precipitation from other drivers of year effects. In each establishment year, we seeded the same species mix to initiate grassland restoration. We then surveyed plant community compositions annually for 5 years to quantify trait responses of restored communities to planting year rainfall. 4. We found that variation in planting-year precipitation altered community assembly trajectories by influencing community-weighted mean (CWM) trait composition, and these effects persisted for at least 5 years. Over time, CWM specific leaf area and CWM seed mass decreased and CWM plant height increased. The effect of age on CWM plant height was stronger in plots that received mean and high watering treatments compared to drought treatments. This effect was also observed for CWM seed mass, albeit weaker. 5. We also found some evidence for planting year effects unrelated to planting-year rainfall for the three CWM traits, illustrating how interannually varying environmental conditions besides rainfall can generate persistent year effect on plant communities through their traits. 6. Synthesis and applications. Our results provide evidence for planting year rainfall interacting with community assembly to alter the functional trait composition of restored grasslands. This suggests that interannual variation in rainfall during establishment is an important source of divergent biodiversity and functional outcomes in restored grasslands.