Responses of plant leaf economic and hydraulic traits mediate the effects of early- and late-season drought on grassland productivity

Drought can occur at different times during the grassland growing season, likely having contrasting effects on forage production when happening early or later in the season. However, knowledge about the interacting effects of the timing of drought and the development stage of the vegetation during the growing season is still scarce, thus limiting our ability to accurately predict forage quantity losses. To investigate plant community responses to drought seasonality (early-vs. late-season), we established a drought experiment in two permanent grasslands of the Swiss Jura Mountains that are used for forage production. We measured three plant functional traits, including two leaf traits related to plant economics (specific leaf area, SLA; leaf dry matter content, LDMC) and one hydraulic trait related to physiological function (predicted percentage loss of hydraulic conductance, PLCp), of the most abundant species, and plant above-ground biomass production. Plant species composition was also determined to calculate community-weighted mean (CWM) traits. First, we observed that CWM trait values strongly varied during the growing season. Second, we found that late-season drought had stronger effects on CWM trait values than early-season drought and that the plant hydraulic trait was the most variable functional trait. Using a structural equation model, we also showed that reduction in soil moisture had no direct impacts on above-ground biomass production. Instead, we observed that the drought-induced decrease in above-ground biomass production was mediated by a higher CWM PLCp (i.e. higher risk of hydraulic failure) and lower CWM SLA under drought. Change in CWM SLA in response to drought was the best predictor of community above-ground biomass production. Our findings reveal the importance of drought timing together with the plant trait responses to assess drought impacts on grassland biomass production and suggest that incorporating these factors into mechanistic models could considerably improve predictions of climate change impacts.