1. Net ecosystem gas exchange (NEE), a balance between gross ecosystem primary productivity (GPP) and ecosystem respiration (ER), is an important indicator of terrestrial ecosystem CO2 sink or source. Increasing frequency of droughts during different periods of the growing season may affect terrestrial ecosystem carbon balance. However, detecting how drought timing controls ecosystem carbon processes is insufficiently explored because it is a challenge to accurately monitor and forecast drought dynamics.

2. In a five-year (2015-2019) precipitation manipulation experiment in a temperate steppe in northern China, we imposed a 60% decrease in precipitation in the early (April-June, DEP) and late (July-September, DLP) growing seasons in plots under rainout shelters to simulate drought-occurrence timing. Responses of GPP, ER, and NEE to DEP and DLP were examined to determine how the timing of decreased precipitation affects CO2 fluxes.

3. Both DEP and DLP reduced GPP and ER. Decreased precipitation in the late growing season reduced net ecosystem gas exchange (NEE) due to a greater decline of GPP than ER. In contrast, the lack of an effect of DEP on NEE can be attributed to a proportional decline of GPP and ER. Reduced normalized difference vegetation index (NDVI) and GPP induced by decreased precipitation explained the decline of GPP and ER, respectively. Drought in the late growing season weakened the ecosystem carbon sink. The result indicates a low resistance in net carbon uptake to decreased precipitation in the late growing season. However, we did not find evidence that previous precipitation decreases resulted in a lower overall rate of ecosystem carbon exchange, indicating a high resilience in carbon processes in this semi-arid grassland.

4. Synthesis and applications. Our study provides the rare evidence for the important role of drought timing in regulating ecosystem carbon balance. The findings are significant for management practices aimed at assessing source-sink functional conversion in the grassland ecosystems under climate change scenarios.

MEASUREMENTS OF ECOSYSTEM C FLUX

Gross ecosystem primary productivity and ER were measured with an LI-6400 Portable Photosynthesis System (Li-Cor, Lincoln, NE, USA) attached to a transparent chamber (0.5 × 0.5 × 0.5 m³). This was always conducted on a sunny morning (8:00 -12:00 pm) three times a month over the growing seasons. For the measurement of NEE and evapotranspiration (ET), the transparent chamber was placed on a preset iron frame (0.5 × 0.5 m²), which had been inserted into the soil at the depth of 3 cm, and provided a flat surface between the soil surface and the sampling chamber (Niu et al. 2012). There were two fans inside the chambers running continuously to mix the air during measurements. GPP measurement were taken at 10-sec intervals during the 90-sec after steady-state conditions were achieved within the chamber after 2 minutes. After measurement of GPP, the box was lifted and ventilated for 30 seconds, the original iron frame was re-covered, and the box was covered with an opaque cloth to block plant photosynthesis for ER measurement (Niu et al. 2012). Net Ecosystem Exchange was estimated as the difference between ER and GPP.

Each block had seven treatments, including 1) control (C), 2) 60% decrease in precipitation in the early growing season (April-June, DEP), 3) 60% decrease in precipitation in the late growing season (July-September, DLP), 4) 60% decrease in precipitation during the entire growing season (April-September, DP), 5) 60% increase in precipitation in the early growing season (April-June, IEP), 6) 60% increase in precipitation in the late growing season (July-September, ILP), and 7) 60% increase in precipitation during the entire growing season (April-September, IP). Each treatment had five replicates, resulting in a total of 35 plots.