Seasonal variations of carbon fluxes and the drivers in typical Chinese forest ecosystems.

We integrated monthly eddy-covariance carbon flux data from 27 forest flux towers across China, compiled from ChinaFLUX during 2003 to 2023, literature, and our own observations. The study sites covered four forest biomes, namely boreal, temperate, subtropical, and tropical forests. Based on long-term mean monthly series, we quantified seasonal metrics of gross primary productivity (GPP), ecosystem respiration (Re), and net ecosystem productivity (NEP), including seasonal amplitude, seasonality index, and seasonal standard deviation, as well as flux magnitude metrics, including maximum monthly flux and growing-season mean monthly flux. Generalized additive models were used to identify the main explanatory variables of spatial variations in these metrics to examine the seasonal variation and spatial differentiation of forest carbon fluxes across different climatic zones in China. The results showed that the long-term mean monthly fluxes of GPP, Re, and NEP across the four forest biomes ranged from 0 to 298.3, 11.8 to 227.1, and -27.5 to 106.8 g C·m-2·month-1, respectively. GPP and Re exhibited unimodal seasonal patterns, with peaks mainly occurring from June to August. NEP showed a "carbon source-carbon sink-carbon source" transition pattern in boreal and temperate forests, whereas subtropical and tropical forests were mostly carbon sinks throughout the year. The seasonality of GPP and Re increased significantly with latitude, and the seasonality indices of which were significantly and positively correlated with latitude (R2 values were 0.74 and 0.56, respectively). Generalized additive model analyses showed that the seasonal metrics of forest carbon fluxes were mainly associated with climatic factors. Annual temperature range contributed 52.3%-68.7% to the GPP seasonal metrics. The flux magnitude metrics of forest carbon fluxes were mainly associated with stand structural factors. Canopy height contributed 74.3% and 70.4% to the maximum monthly fluxes of GPP and NEP, respectively, and contributed 28.6%-57.7% to the growing-season mean monthly fluxes of carbon flux components. Across the typical forest sites compiled in this study, the spatial differentiation of carbon fluxes was expressed mainly in the strength of seasonality rather than in flux magnitude, providing a scientific basis for forest carbon balance assessment and seasonal parameterization of ecosystem models.