Abstract: Boreal forests play a pivotal role in the global carbon balance and climate change mitigation. Understanding their carbon exchange dynamics is essential for advancing knowledge of forest carbon cycling. However, studies addressing multi-scale carbon flux variability and its drivers in China’s boreal forests remain limited. This study analyzed five years (2019–2023) of growing-season eddy covariance data from a Larix gmelinii forest in the Greater Khingan Mountains to elucidate seasonal and interannual variations in carbon fluxes and their driving mechanisms. On average, the growing season net ecosystem carbon exchange (NEE), gross primary productivity (GPP), and ecosystem respiration (Re) were − 376.99 ± 44.24, 856.15 ± 37.99 and 479.16 ± 51.46 g m⁻², respectively, indicating that the forest consistently functioned as a moderate carbon sink. At seasonal variations, NEE exhibited a “U”-shaped pattern, predominantly driven by GPP. Path analysis showed that photosynthetically active radiation (PAR) mainly influenced seasonal variations of NEE, leaf area index (LAI) was the key factor for GPP seasonal variations, while vapor pressure deficit (VPD) limited both NEE and GPP variations. Re was primarily influenced by soil temperature (T_s). In contrast, interannual NEE variations were relatively stable and predominantly controlled by Re, with higher PAR and fewer precipitation days enhancing carbon sequestration. These findings demonstrate the divergent controls of environmental and biological factors across timescales. Long-term monitoring of boreal forest carbon dynamics is critical for predicting their carbon sink potential under future climate warming and improving global carbon cycle models.

Key words: Boreal forests, Eddy covariance, Carbon exchange processes, Seasonal and interannual scales, Environmental and biological controls