Drought occurrence is increasing due to anthropogenic climate change. Drought can negatively affect plants via reduced water below‐ground and increased evaporative demand or vapour pressure deficit (VPD) above‐ground. Past work has shown that plant diversity can ameliorate the negative effects of drought in plant communities, but these results are inconsistent between experimental and natural drought studies. Furthermore, while studies on the negative effects of reduced soil moisture on plant growth in drought experiments are abundant, the effects of predicted increases in atmospheric VPD have been neglected. We directly manipulated atmospheric relative humidity in a biodiversity and drought experiment at the California State University, Los Angeles (CA, USA) under three atmospheric conditions (ambient, dehumidified and humidified), two treatments of native perennial grass diversity (monoculture and eight species polyculture) and two soil drought treatments (control and drought). We assessed both polyculture plant community and individual species (Poa secunda) responses to atmospheric drought and soil drought. We found that soil drought only limits above‐ground biomass production when atmospheric conditions are also dry. We also found that P. secunda was limited by increased competition in polyculture when ambient atmospheric conditions were humid but was facilitated by diversity when atmospheric conditions were dry. Synthesis. Higher diversity ecosystems may be capable of protecting individual species from the negative effects of drought (facilitation). Without careful experimental manipulation of atmospheric drought, this important mechanism will be missed.

Climate change alters mean global surface temperatures, precipitation regimes, and atmospheric moisture. Resultant drought affects the composition and diversity of terrestrial ecosystems worldwide. To date, there have been no assessments of the combined impacts of reduced precipitation and atmospheric drying on functional trait distributions of any species in an outdoor experiment. Here, we examined whether soil and atmospheric drought affected the functional traits of a focal grass species (Poa secunda) growing in monoculture and eight-species grass communities in outdoor mesocosms. We focused on specific leaf area (SLA), leaf area, stomatal density, root:shoot ratio, and fine root:coarse root ratio responses. Leaf area and overall growth were reduced with soil drying. Root:shoot ratio only increased for P. secunda growing in monoculture under combined atmospheric and soil drought. Plant energy allocation strategy (measured using principal components) differed when P. secunda was grown in combined soil and atmospheric drought conditions compared with soil drought alone. Given a lack of outdoor manipulations of this kind, our results emphasize the importance of atmospheric drying on functional trait responses more broadly. We suggest that drought methods focused purely on soil water inputs may be imprecisely predicting drought effects on other terrestrial organisms as well (other plants, arthropods, and higher trophic levels).

Climate change alters global mean surface temperatures and precipitation regimes. This changes the composition and diversity of grasslands worldwide. To date, there are no assessments of plant functional trait responses to experimentally controlled atmospheric drought (separate from soil drought). We examined whether soil and atmospheric drought result in changes to Poa secunda functional traits in monoculture and 8-species grass communities. We focused on SLA, leaf area, stomatal density, root:shoot ratio, and fine root:coarse root ratios. Leaf area and overall growth were higher with increased watering and higher atmospheric moisture in most cases. However, root:shoot ratio significantly increased in atmospheric drought combined with drought watering. Plant energy allocation strategy (measured using PCA) differed when P. secunda was grown in combination soil and atmospheric drought conditions. Our results indicate that methods focused purely on soil water inputs, such as rainout shelters, may be imprecisely predicting drought effects.