Variations and drivers of CO2 fluxes at multiple temporal scales of subtropical agricultural systems in the Huaihe river Basin

Understanding of the crop carbon balance across different time scales and corresponding responses to abiotic and biotic factors is crucial for improving carbon cycle models in the context of future climate change and management practices. In this study, we employed the Random Forest (RF) algorithm, Kolmogorov-Zurbenko filtering method and structural equation modeling (SEM) to quantify the effects of abiotic and biotic factors on CO2 fluxes at various time scales based on 7-years measurements. Our results revealed that O3 primarily manifested indirect effects on NEE and GPP via altering LAI on the daily and monthly scale, and that overall regulatory effect on CO2 fluxes developed greater as the time scale increased. Net radiation (Rn) was the most critical abiotic factor altering net ecosystem exchange (NEE) and gross primary productivity (GPP) at the half-hourly, daily, and monthly scales, with the exception of photosynthetically active radiation (PAR) controlling daily NEE and GPP in the rice system. It was innovatively found that LAI had little control on detrended daily CO2 fluxes, which was much lower than the monthly CO2 fluxes. Air temperature (Ta) was the most important abiotic factor for ecosystem respiration (Reco) at half-hourly and daily scale. For NEE, Reco, and GPP, the maximum explanation of SEM models was 70.10%, 79.60% and 76.20%, respectively. The SEM results indicated that at multiple time scales, Rn exerted significant direct and indirect effects on both NEE and GPP. LAI only showed a strong direct leading effect on NEE and GPP on the monthly scale. The findings we reported have the potential to further develop carbon cycle models of cropland ecosystems under climate change by clarifying the influence path of O3 on CO2 fluxes and highlighting the factors that dominate CO2 fluxes on various time scales.