At present, using plant traits to predict ecosystem functions is of great significance as trait-based ecology recasts classic ecological propositions. However, functional traits at the individual level do not predict higher-level aggregated ecosystem processes (e.g., gross primary productivity) accurately because vital contextual information on specific communities or ecosystems is not considered. Here, we integrate individual or species level plant traits with background information, such as total leaf mass, to form plant community traits that explain spatial variability in productivity. We use a systematic dataset of leaf nitrogen (N) content in 73 typical ecosystems of China that were classified based on community weighted mean leaf N content (N_content, mg g⁻¹) and the total leaf N reserves per land area (N_quantity, g m⁻²) to verify their variations at a large scale across biomes and their responses to changing environmental factors. The results showed that using plant community traits, even without factoring in climatic factors, explains 81% and 68% of the spatial variability in annual gross primary productivities and monthly gross primary productivities, respectively. In addition, both N_content and N_quantity were significantly affected by environmental factors (climatic and soil factors). Our findings highlight that two-dimensional plant community traits of content vs. quantity on land area can strengthen the predictability of variations of primary productivity and even other ecosystem functions.