Aboveground net primary productivity not CO2 exchange remain stable under three timing of extreme drought in a semi-arid steppe

Precipitation patterns are expected to change in the semi-arid region within the next decades, with projected increasing in extreme drought events. Meanwhile, the timing of extreme drought also shows great uncertainty, suggesting that the timing of drought, especially during growing season, may subsequently impose stronger stress on ecosystem functions than drought itself. However, how the timing of extreme drought will impact on community productivity and carbon cycle is still not clear. In this study, three timing of extreme drought (a consecutive 30-day period without precipitation event) experiments were set up separately in early-, mid- and late-growing season in a temperate steppe in Inner Mongolia since 2013. The data, including soil water content (SWC), soil temperature (ST) chlorophyll fluorescence parameter (F-t/F-m), ecosystem respiration (Re), gross primary productivity (GPP), net ecosystem carbon absorption (NEE) and aboveground net primary productivity (ANPP) were collected in growing season (from May to September) of 2016. In this study, extreme drought significantly decreased SWC during the drought treatment but not for the whole growing season. Extreme drought decreased maximum quantum efficiency of plant photosystem II (F-t/F-m) under optimum value (0.75-0.85) of two dominant species (Leymus chinensis and Stipa grandis). While ANPP kept stable under extreme drought treatments due to the different responses of two dominant species, which brought a compensating effect in relative abundance and biomass. In addition, only early-growing season drought significantly decreased the average Re (P < 0.01) and GPP (P < 0.01) and depressed net CO2 uptake (P < 0.01) than mid- and late-growing season drought. ST and SWC influenced the changes of GPP directly and indirectly through photosynthetic ability of the dominant species by path analysis. Our results indicated that the timing of drought should be considered in carbon cycle models to accurately estimate carbon exchange and productivity of semi-arid grasslands in the context of changing climate.