The dataset is from an incubation experiment which was carried out in the dark at the temperature of 15°C and the relative humidity of 40% for approximately three months. The predominant soil sampled from a temperate grassland of northern China, it is chernozem and belongs to sandy loam, with sand, silt and clay percentages of 54%, 41% and 5%, respectively. The SOC content of the 0-10 cm layer is 4.2% and the pH is 6.9. This study manipulated soil water levels and the intensity of water variability with a total of six treatments. There were four replicates for each treatment for gas sampling. The average cumulative CO₂ release and the soil aggregate of the four replicates and its standard error (SE) are provided in the dataset. One additional group of destructive samplings and the soils for the gas sampling were collected at the wet end and dry end in the last drying-rewetting cycle, respectively. 14 phospholipid fatty acids (FLFA) markers were selected from the combination results of the wet and dry ends in the last drying-rewetting cycle.

The soil samples were collected from the 0-10 cm layer at four locations several meters apart in September 2020. Plants and surface litter were removed before sampling. The collected soil samples were transported to the laboratory in containers with ice. In the laboratory, the soil samples were passed through a 2 mm sieve, homogenized, and stored at -20℃ until the incubation experiment started. During the incubation, 66.3 g fresh soil (equivalent to 50 g air-dried soil) was put in a 200 ml open-top mason jar. All the jars were pre-incubated at 60% of field water holding capacity for seven days. The soil water content was monitored daily by weighing the jars and adjusted by adding the corresponding amount of water if needed. During gas measurement, the jar was covered by a rubber lid with two gas paths and connected to an LI-8100A infrared gas analysis system (LI-COR, USA). The CO₂ concentration changes within 90 seconds in the headspace of each jar were recorded to calculate the CO₂ emission rate. The gas measurement was conducted daily in the first eight days, every two days for the next 19 days, and every 2-4 days thereafter.

The soils for the gas measurements were destructively sampled at the end of the experiment. The experiment ended at their respective driest ends for intensified water variability treatments. One additional group of destructive samplings was collected at the wet end in the last drying-rewetting cycle to reveal whether there was a quick shift of microbial community between the dry and wet ends. Based on the microbial community and the total PLFA did not change within the last cycle, 14 FLFA markers were selected from the combination results of the wet and dry ends in the last drying-rewetting cycle. We chose PLFAs 15:0, 17:0, i15:0, a15:0, i16:0, i17:0, a17:0, cy17:0, 16:1ω7, 18:1ω7 to represent bacterial biomass, PLFAs 18:2ω6 and 18:1ω9 to indicate fungi biomass and PLFAs 10Me16:0, 10Me18:0 to represent actinomycetes biomass. The water-stable aggregates composition was measured by Passing the soil samples through 250 μm and 53μm sieves to get macro-aggregates (>250 μm), micro-aggregates (53-250 μm) and mineral-associated SOC fraction (<53 μm).