Predicting ecosystem primary productivity on plant community traits: Theoretical basis and research progress

Plants contribute significantly to ecosystem primary productivity, serving as the basis of material cycling and energy flow. How to improve the accuracy of ecosystem productivity predictions is a classic topic in ecology. For decades, researchers have employed radiation-based remote sensing models or big-leaf-based process models to predict the spatiotemporal variations in ecosystem productivity. However, large discrepancies among model outputs constrain our understanding of the carbon sequestration capacity of ecosystems under global change. Recently, plant functional traits, as key parameters in next-generation process models, have received extensive attention. However, the scale mismatch between traditionally measured individual-level traits and community-level productivity constitutes an important source of model uncertainties. To address these challenges, we introduced the classical engine power model from physics and developed a novel trait-based productivity (TBP) framework centered on the two-dimensionality of plant community traits (quantity traits and efficiency traits). Contrary to the traditional models, all parameters in the TBP framework were defined at the community scale, with environmental factors influencing ecosystem productivity both directly and indirectly by regulating plant community traits. On this basis, using an in situ multi-trait database of Chinese ecosystems (including leaf chlorophyll concentration, leaf area, specific leaf area, leaf dry mass, and leaf nitrogen and phosphorus concentrations), we used three empirical studies to demonstrate the application scenarios of TBP theory. The TBP framework effectively bridges the scale gap between traits at individual level and ecosystem primary productivity. This framework is compatible with massive spatial data generated by flux observations, hyperspectral sensing, remote sensing, machine learning technologies, thus holding considerable application potential. Currently, the TBP framework is at an early stage. Besides requiring further theoretical innovation and methodological improvement, it also necessitates extensive support and validation from ground-based and remote sensing data. This will lay the foundation for the development of new-generation mechanistic process models and effectively improve the prediction accuracy of ecosystem productivity.