Abstract: Vegetation gross primary production (GPP), the rate of carbon assimilation via photosynthesis, is a fundamental metric for assessing terrestrial carbon uptake. Functional traits such as the carbon uptake period and maximum photosynthetic capacity (GPP_max) largely determine interannual GPP variability, yet existing frameworks are mainly effective in ecosystems with idealized bell-shaped GPP trajectories. This gap highlights the need for a globally consistent framework that accounts for biome-specific differences. We developed the trait-based ecosystem productivity explanatory model (TEPEM) to disentangle and explain the empirical relationship between total annual GPP (GPP_ann) and vegetation functional traits, particularly GPP_max and the annual average growing season index (GSI_ann). This relationship proved to be remarkably robust across biomes (r = 0.93, p < 0.01) based on flux tower observations. Building on it, we further integrated the light response curve model to dynamically simulate GPP_max, enabling TEPEM to estimate GPP_ann across historical and future climate scenarios. Site-scale evaluations demonstrated the strong performance of TEPEM, with a Pearson’s r of 0.86 and a low root mean square error of 333.8 g m⁻² a⁻¹. At the global scale, TEPEM effectively reproduces the spatiotemporal patterns of GPP_ann across diverse biomes, achieving Pearson’s r values ranging from 0.94 to 0.98 compared to process-based, light use efficiency, and upscaling models. TEPEM projects increasing GPP_ann trends across most terrestrial regions by 2100 under various shared socioeconomic pathways. This study underscores the critical role of vegetation functional traits, particularly phenology and photosynthetic capacity, in explaining GPP variability.

Key words: Interannual variability in gross primary productivity, Growing season index, Seasonal maximum of gross primary production, Terrestrial ecosystem model, Climate change