Arid regions, being climate-sensitive and ecologically fragile, are among the most significantly impacted areas by changes in carbon and water fluxes. Researching the dynamic variations in these fluxes and their driving factors is crucial for global ecological balance, climate change, and sustainable development. Therefore, exploring the dynamic changes in carbon and water fluxes, as well as their driving and process effects in mountainous grasslands and forests in arid regions, is particularly important. This study, based on eddy covariance (EC) system measurements, employs Structural Equation Modeling (SEM) and Generalized Additive Models (GAM) to investigate the differences in carbon and water flux balances, driving mechanisms, and process effects in the central Tianshan grassland and Schrenk spruce ecosystems from 2019 to 2022. The findings reveal: (1) the Schrenk spruce ecosystem sequesters more carbon annually (-517.35 g C m-2 y-1) than the Tianshan grassland (-319.54 g C m-2 y-1); (2) the Schrenk spruce ecosystem has a higher water use efficiency (WUE) of 1.52 g C kg-1 H2O compared to the Tianshan grassland (0.88 g C kg-1 H2O), with lower evapotranspiration (ET) of 439.38 mm compared to 565.67 mm in the grassland; (3) precipitation intensity has different effects on the carbon flux of both ecosystems; (4) factors such as soil temperature (Ts), soil moisture content (SWC), photosynthetically active radiation (PAR), and precipitation (Pre) influence carbon and water fluxes through pathway and process effects, showing distinct carbon sink "turning points" under different environmental conditions. These findings provide essential insights for understanding the dynamic changes in carbon and water cycles, revealing complex driving relationships, and predicting future climate impacts. Additionally, they offer valuable references for formulating sustainable vegetation and water resource management strategies in the central Tianshan arid region.