Objective Under global climate change, the role of dryland ecosystems in the carbon-water cycle has become increasingly prominent. As a key indicator for measuring carbon-water coupling, changes in ecosystem water use efficiency (WUE) directly affect the stability of ecosystem functions. Therefore, clarifying the spatiotemporal pattern evolution of WUE in drylands and its climatic drivers is an important prerequisite for accurately predicting the response of the future carbon-water cycle to climate change.

Method This study took ecosystem water use efficiency as the research object. Based on the Google Earth Engine (GEE) platform, long-term MODIS remote sensing data from 2001 to 2020 were retrieved, and a WUE dataset for the Tarim River Basin was calculated and constructed using the ratio method of gross primary productivity (GPP) to evapotranspiration (ET). On this basis, methods such as trend analysis and partial correlation analysis were used to systematically investigate the spatiotemporal changes in WUE and its responses to climatic factors for typical ecosystem types (cropland, forest, and grassland) in the Tarim River Basin from 2001 to 2020.

Result (1) Annual WUE showed a significant decreasing trend at both the basin and ecosystem scales. Relative contribution analysis indicated that the WUE decline was dominated by the negative impact of ET increase (59.12%), which offset the positive contribution of GPP growth (40.88%). (2) Both temporal variation and interannual variability of WUE were positively correlated with temperature and negatively correlated with precipitation. Specifically, long-term trends were mainly regulated by precipitation, while interannual variability was primarily driven by temperature. (3) Cropland and forest WUE also exhibited positive correlations with temperature and negative correlations with precipitation, with the interannual variability of cropland WUE being significantly regulated by temperature. (4) Monthly WUE showed a significant time-lag response to climatic factors, with lag times of 2.37 ± 1.05 months for temperature and 1.08 ± 1.25 months for precipitation. Notably, the lag time of cropland response to precipitation was significantly shorter than that of forests, whereas its response lag to temperature was significantly longer.

Conclusion WUE in the Tarim River Basin showed a significant declining trend. Future warming and increasing precipitation may further suppress WUE in this arid region by accelerating evapotranspiration, implying greater water resource pressure for maintaining carbon sink functions. Therefore, water allocation and ecological management in the basin should account for the time-scale differences and lagged responses of WUE to climatic factors. These findings provide a scientific basis for regulating ecosystem carbon-water functions and promoting sustainable water resource management under climate change.