Climate extremes are expected to increase in frequency and magnitude as a consequence of global warming. Managed permanent grasslands cover a large surface in Europe and contribute substantially to agricultural production. These managed plant communities are dominated by perennial clonal species. Their capacity to adapt to rapidly changing environmental conditions may be limited. We hypothesize that those plant populations that have already been exposed to conditions that are expected to occur due to future climate change, particularly conditions that would be 'extreme' in the target area, are able to cope better with these conditions. For a common-garden experiment we selected ecotypes (provenances as supported by accessions in seed banks) of important European grass species: Arrhenatherum elatius, Festuca pratensis, Holcus lanatus and Alopecurus pratensis. Southern target locations of ecotypes (populations) were identified based on climate model projections for the local site in Northern Bavaria, Germany. In a controlled experiment, the plants were exposed to warming and extreme drought. Drought conditions(16-19 days, depending on the species) were imposed starting from the end of May in combination with and without an increase in the average temperature from May to September 2009 (+1.5 K compared with control; +2.5 K compared with ambient conditions outside of the experimental units). Ecotypes and drought manipulation had significant impacts on biomass production and tissue die-back. Significant interactions between ecotype and drought indicated a different drought tolerance of the ecotypes in some cases. The warming treatment yielded a less significant response. The local ecotype generally did not perform significantly worse than the presumably better-adapted southern ecotypes. Synthesis. The selection of ecotypes that are adapted to more extreme climatic conditions could be an option for maintaining future ecosystem functioning in temperate managed grasslands, as was indicated by the clear differences between ecotypes in our experiment. Based on our data, however, performance cannot be predicted from climatic origin. Therefore, we recommend enhancing the genetic variability within populations of species in general. [ABSTRACT FROM AUTHOR]