Abstract: Ecosystems in dryland mountains occur along pronounced elevational gradients and exhibit high environmental heterogeneity, playing key roles in the global carbon cycle. However, their CO₂ flux dynamics and environmental controls remain insufficiently understood. In this study, we established three eddy covariance flux sites along an elevational gradient in a dryland mountain region of Northwest China, representing a Larix principis-rupprechtii plantation, a Rhododendron lapponicum shrubland, and an alpine meadow. Continuous observations from 2020 to 2022 quantified seasonal and cross-ecosystem variations in net ecosystem productivity (NEP), gross primary productivity (GPP), and ecosystem respiration (Re), and examined their responses to environmental and physiological variables. Annual GPP and Re decreased significantly with increasing elevation. The plantation showed the strongest net CO₂ uptake, the shrubland maintained variable net CO₂ uptake, and the alpine meadow remained close to CO₂ neutrality, shifting between weak net CO₂ uptake and weak net CO₂ release. Monthly anomaly analysis showed that NEP anomalies were most pronounced during the peak growing season and were mainly driven by GPP anomalies, while Re anomalies either amplified or offset GPP effects on NEP. Random forest and structural equation modeling analyses revealed that dominant controls shifted along the elevational gradient, with lower-elevation ecosystems being primarily regulated by vegetation activity, photosynthetic capacity, and atmospheric water demand, whereas higher-elevation ecosystems showed stronger temperature and water limitations. These results highlight that magnitude of net CO₂ uptake was jointly influenced by vegetation activity, thermal conditions, and water availability through their contrasting effects on photosynthetic CO₂ uptake and ecosystem respiration.

Key words: elevational gradient, eddy covariance, net ecosystem productivity, environmental drivers, alpine meadow, forest ecosystems