Abstract
Abiotic global change drivers affect ecosystem structure and function, but how they interact with biotic factors such as invasive plants is understudied. Such interactions may be additive, synergistic, or offsetting, and difficult to predict. We present methods to test the individual and interactive effects of drought and plant invasion on native ecosystems. We coupled a factorial common garden experiment containing resident communities exposed to drought (imposed with rainout shelters) and invasion with a field experiment where the invader was removed from sites spanning a natural soil moisture gradient. We detail treatments and their effects on abiotic conditions, including soil moisture, light, temperature, and humidity, which shape community and ecosystem responses. Ambient precipitation during the garden experiment exceeded historic norms despite severe drought in prior years. Soil moisture was 48% lower in drought than ambient plots, but the invader largely offset drought effects. Additionally, temperature and light were lower and humidity higher in invaded plots. Field sites spanned up to a 10-fold range in soil moisture and up to a 2.5-fold range in light availability. Invaded and resident vegetation did not differentially mediate soil moisture, unlike in the garden experiment. Herbicide effectively removed invaded and resident vegetation, with removal having site-specific effects on soil moisture and light availability. However, light was generally higher in invader-removal than control plots, whereas resident removal had less effect on light, similar to the garden experiment. Invasion mitigated a constellation of abiotic conditions associated with drought stress in the garden experiment. In the field, where other factors co-varied, these patterns did not emerge. Still, neither experiment suggested that drought and invasion will have synergistic negative effects on ecosystems, although invasion can limit light availability. Coupling factorial garden experiments with field experiments across environmental gradients will be effective for predicting how multiple stressors interact in natural systems.