Widespread Time-Lagged and Cumulative Effects Modulate Climatic Controls on Ecosystem Water Use Efficiency

Ecosystem water use efficiency (WUE), an indicator of the trade-off between carbon uptake and water loss, is widely used to assess ecosystem responses to climate change. However, large-scale studies of WUE typically assume a single, fixed lag or accumulation period of climatic drivers across regions. This static assumption neglects spatially heterogeneous temporal responses of WUE to climate, potentially biasing attribution analyses and reducing predictive skill. Here, we developed a pixel-level model to quantify the temporal effects of climatic drivers on WUE by explicitly accounting for no-effect, lagged, cumulative, and combined effects and allowing effect timescales to vary spatially. We found that more than 80% of pixels across China exhibited lagged and/or cumulative effects for each driver, with distinct temporal effect patterns among vegetation types and drivers. In herbaceous cover croplands, precipitation exhibited the shortest lag (0.31 ± 0.56 months) and the longest accumulation time (1.71 ± 0.96 months). Accounting for these spatially heterogeneous temporal effects increased the explanatory power of climatic drivers for WUE variation by 17.7% compared with models without temporal effects. We further showed that for most vegetation types, precipitation and air temperature were more strongly associated with temporal variation in WUE, whereas solar radiation contributed more to spatial variability. These findings indicate that location-specific temporal effects can modulate the climatic controls on WUE. Our framework is readily applicable beyond China and can support a shift toward dynamic climate responses in climate–ecosystem interaction modeling, thereby improving forecasts of ecosystem dynamics and informing climate-adaptive vegetation management.