Long term experimental drought alters community plant trait variation, not trait means, across three semiarid grasslands

Background and aims Grasslands are expected to experience droughts of unprecedented magnitude and duration in this century. Plant traits can be useful for understanding community and ecosystem responses to climate extremes. Few studies, however, have investigated the response of community-scale traits to extreme drought on broad spatial/temporal scales, with even less research on the relative contribution of species turnover vs. intraspecific trait variation to such responses. Methods We experimentally removed similar to 66% of growing season rainfall for three years across three semi-arid grasslands of northern China and tracked changes in community functional composition, defined as the community mean and variation of several leaf economic traits. Results Community trait variations were more sensitive to drought than community trait means, which suggests this component of functional composition may be a better indicator of initial community drought responses than trait values themselves. The greatest change in trait variation was observed at the high aridity site and was driven largely by intraspecific trait variability. Apart from specific leaf area, trait variability increased with increasing aridity across sites, largely due to species turnover. Variations in soil moisture and fertility likely mediated the responses of community trait variations to water stress. Conclusions These results highlight the importance of measuring community trait variability in response to drought and support the well-documented pattern of increased community drought sensitivity of more arid ecosystems.