Solar-induced chlorophyll fluorescence (SIF) has been used to estimate leaf-level net CO 2 assimilation by a mechanistic light reaction (MLR-SIF) equation. However, the application of this model would be limited by the challenging measurement and estimation of input parameters (e.g. fraction of open PSII reaction centres, q L). We modified the MLR-SIF model by replacing q L by the easily obtained parameters (non-photochemical quenching [NPQ]) to facilitate its application. We employed synchronous measurements of gas exchanges, ChlF parameters and SIF for Leymus chinensis, Populus tomentosa Carrières and Ulmus pumila var. sabulosa under the soil–water deficit and rehydration process to test the robustness of the modified MLR-SIF model. Our results demonstrated that for L. chinensis the net photosynthesis rate dynamics under severe soil–water stress and saturated water condition were effectively captured by the modified MLR-SIF model ( R 2 = 0.75–0.92, RMSE = 1.11–3.56) . For P. tomentosa Carrières and U. pumila var. sabulosa, the net photosynthesis rates were predicted by the modified MLR-SIF model with good accuracy ( R 2 = 0.86, RMSE = 9.44; R 2 = 0.88, RMSE = 4.16) across the water deficit and rehydration condition . However, the electron transport rate estimated by the modified MLR-SIF model uncoupled with the photosynthetic capacity ( r 2 = -0.13) and lowered the net photosynthesis rate simulation precision ( R 2 = 0.35, RMSE = 3.41) for L. chinensis under mild drought stress and saturated light intensities. The electron transport rate estimated by the modified MLR-SIF model downregulated the photosynthetic capacity for P. tomentosa Carrières ( r 2 = 0.32) and U. pumila var. sabulosa ( r 2 = 0.22) under mild drought stress. The shift of the Rubisco and RUBP limited state cross-points, the dynamic photosynthesis parameters across the plant species and the alternative electron sinks under soil–water deficit and rehydration process influenced the simulation precision of the modified MLR-SIF model. Our modified MLR-SIF model provided a basis for understanding and inferring the photosynthetic rate by SIF and NPQ under water stress.

The interannual variation (IAV) of net ecosystem carbon production (NEP) plays an important role in understanding the mechanisms of the carbon cycle in the agriculture ecosystem. NEP is usually partitioned into the diffecence between gross ecosystem productivity (GEP) and ecosystem respiration (RE), or the integration of the carbon uptake or release peak and the corresponding duration. In this study, the climatic and biotic controls of the IAV of NEP, which were expressed as annual values and anomalies, were investigated based on an eddy covariance dataset of rain-fed spring maize during 2005–2018 in the northeast of China. The annual NEP was 270±115 g C m−2yr −1. Annual values and anomalies of NEP were positively correlated with that of precipitation (PPT), GEP and daily maximum NEP (NEPmax). 78.9% of annual anomalies of NEP were explained by the interaction of climate, soil and plant variables, and the atmospheric water vapor deficit (VPD) played a dominated role. Annual anomalies of NEP were dominantly and positively controlled by the soil water content (SWC) through GEP and the soil temperature (Ts) through RE. In comparison, annual anomalies of NEP were dominantly and negatively controlled by summer VPD through the NEPmax, positively adjusted by spring precipitation and the effective accumulative temperature through the beginning date (BDOY) of the affecting carbon uptake period (CUP), and by autumn precipitation and leaf area index through the end date (EDOY) of the affecting CUP. Residues restrained the carbon release at the beginning of the year, and accelerated the carbon release at the end of the year. Our results hightlight that NEP might be more sensitive to the change of water condition (such as PPT, SWC and VPD) induced by the climate changes.