Climate change is altering the frequency and severity of drought events. Recent evidence indicates that drought may produce legacy effects on soil microbial communities. However, it is unclear whether precedent drought events lead to ecological memory formation, i.e., the capacity of past events to influence current ecosystem response trajectories. Here, we utilize a long-term field experiment in a mountain grassland in central Austria with an experimental layout comparing 10 years of recurrent drought events to a single drought event and ambient conditions. We show that recurrent droughts increase the dissimilarity of microbial communities compared to control and single drought events, and enhance soil multifunctionality during drought (calculated via measurements of potential enzymatic activities, soil nutrients, microbial biomass stoichiometry and belowground net primary productivity). Our results indicate that soil microbial community composition changes in concert with its functioning, with consequences for soil processes. The formation of ecological memory in soil under recurrent drought may enhance the resilience of ecosystem functioning against future drought events.

Increasing climatic pressures such as drought and flooding challenge agricultural systems and their management globally. How agricultural soils respond to soil water extremes will influence biogeochemical cycles of carbon and nitrogen in these systems. We investigated the response of soils from long-term agricultural field sites under varying crop rotational complexity to either drought or flooding stress. Focusing on these contrasting stressors separately, we investigated soil heterotrophic respiration during single and repeated stress cycles in soils from four different sites along a precipitation gradient (Colorado, MAP 421 mm; South Dakota, MAP 580 mm; Michigan, MAP 893 mm; Maryland, MAP 1192 mm); each site had two crop rotational complexity treatments. At the driest (Colorado) and wettest (Maryland) of these sites, we also analyzed microbial biomass, six potential enzyme activities, and N2O production during and after individual and repeated stress cycles. In general, we found site specific responses to soil water extremes, irrespective of crop rotational complexity and precipitation history. Drought usually caused more severe changes in respiration rates and potential enzyme activities than flooding. All soils returned to control levels for most measured parameters as soon as soils returned to control water levels following drought or flood stress, suggesting that the investigated soils were highly resilient to the applied stresses. The lack of sustained responses following the removal of the stressors may be because they are well in the range of natural in situ soil water fluctuations at the investigated sites. Without the inclusion of plants in our experiment, we found that irrespective of crop rotation complexity, soil and microbial properties in the investigated agricultural soils were more resistant to flooding but highly resilient to drought and flooding during single or repeated stress pulses.