Ecosystem water use efficiency (WUE ), the core metric of terrestrial carbon-water coupling, is increasingly studied, yet three gaps persist: nonlinear thresholds have been quantified mainly for single drivers, while thresholds of multi-driver interactions remain largely unquantified; basin-scale assessments rarely disaggregate broad climatic gradients, masking region-specific driver regimes; and identified thresholds are seldom translated into spatially explicit management zones. These gaps motivate a climate-stratified, interaction-aware assessment of WUE thresholds that can inform precision ecological zoning in water-limited basins. Taking the Yellow River Basin ( YRB ), a water-limited basin spanning arid to semi-humid climates, we quantified WUE dynamics from 2001 to 2020 and deciphered its drivers by integrating trend persistence analysis (Theil-Sen, Mann-Kendall, Hurst exponent, coefficient of variation) with an interpretable machine learning framework ( CatBoost / XGBoost-SHAP ). WUE exhibited three climate-mediated regimes: high-efficiency growth with moderate stability in the humid southeast, high fluctuation with low stability in the semi-arid center, and low-level, high-stability in the arid northwest. Dominant drivers reorganized systematically along this gradient, from vegetation-atmosphere interactions, through energy-vegetation co-limitation, to physical environment-moisture control under spatially heterogeneous nonlinear thresholds (e.g., PRES = 59.8 kPa in the Plateau Climate Zone; SWVL = 0.08 m 3 ·m −3 in the Middle Temperate Zone). By nesting four management zones within three climatic strata, the three-tier framework converts climate-regime-specific interaction thresholds into actionable, risk-based zoning; over half of the basin (55.6%) falls into early-warning zones. Coupling multi-driver interaction thresholds with climatic stratification, this framework supports a shift from uniform basin-wide policies to proactive, threshold-driven ecological management and is transferable to other water-limited regions.
No takes yet. Share an insight, caveat, or question.
Huang et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: