Abstract
Manipulative approaches under natural conditions are fundamental for understanding impacts of climate warming on grassland (agro-) ecosystems. In this paper we present 3 years of data from two simultaneously conducted transplantation experiments, where meadow monoliths were transplanted downwards along an elevation gradient from the subalpine to the montane belt (2000 m to 1500 m a.s.l.), and in parallel from the montane belt to the foothill zone (1500 m to 1000 m a.s.l.) respectively. Each downward transplantation simulated a temperature increase of 2.8 K. Control and downward transplanted mesocosms were compared regarding aboveground phytomass, phytodiversity, and species composition. Downward transplanted mesocosms from the upper transplantation reacted significantly to warming in terms of aboveground phytomass (legumes +213.6%, herbs +128.2%, graminoids +51.7%, total aboveground phytomass +66.2%), but not with regard to species composition. The lower transplantation, however, induced the complete opposite effect, while average species number and species evenness remained unaffected on all treatments. Further analysis based on five plant traits indicated that the observed shifts were both a consequence of warming and methodological artifacts. Interestingly, the relative importance of warming, artifacts and unaffected species changed with elevation: At the higher transplantation 81.2% of the species remained stable in their abundance, 17.5% were affected by the transplantation, and almost no warming effect could be detected. At the lower transplantation percentage of artifact- and warming-affected species increased consistently (37.5% respectively 44.3%). The results showed that transplantation experiments along elevation transects are an appropriate approach to detect warming impact on agriculturally used grassland at different elevations. Nevertheless, the increasing influence of method-caused side effects became more and more evident over time and with decreasing elevation, underlining the importance of quantifying artifacts in in vivo experiments.