Abstract: Central Asian grasslands are crucial for regional ecological security and the carbon cycle, yet intensifying climate extremes threaten the stability of their productivity. Seasonal-scale studies that simultaneously incorporate temporal effects and nonlinear relationships remain limited. Using monthly gross primary productivity (GPP) and climate data from 2001 to 2022, we quantified the temporal effects of indices of climate extremes on grassland GPP, including cumulative effects (CE) and lagged effects (LE), and combined correlation analysis with random forest to identify seasonally dominant climatic drivers and their spatial heterogeneity. Results show a slight overall decline in grassland GPP with pronounced spatial variability, alongside an overall warming and drying tendency in climate extremes. Temporal effects were widespread, and for most indices CE exceeded LE; CE for precipitation and temperature extremes was mainly 2.23-2.46 and 1.81-2.80 months, respectively. Temporal effects differed among drought indices, with CE dominating for the Standardized Precipitation Evapotranspiration Index (SPEI) and consecutive dry days (CDD) (2.39 and 2.14 months), whereas LE dominated for vapor pressure deficit (VPD) (1.26 months). At the seasonal scale, precipitation extremes generally promoted GPP, whereas drought generally suppressed GPP. Precipitation extremes and drought played dominant roles in seasonal GPP variability, while the impacts of temperature extremes exhibited pronounced seasonality and spatial heterogeneity. These findings provide a scientific basis for adaptive management of arid grassland ecosystems and offer insights for addressing climate change and promoting long-term regional ecosystem sustainability.

Key words: Gross Primary Productivity (GPP), Climate Extreme Indices, Temporal Effects, Central Asia, Grasslands