Abstract: Aims To clarify how water use in rubber plantations (Hevea brasiliensis) varies across leaf phenophases and to identify the key environmental drivers of stem sap flow dynamics, thereby improving assessment of impacts on regional water cycling and informing water-resource management. Methods We continuously monitored stem sap-flux density (Fd) with the thermal dissipation probe (TDP) method across four phenophases—defoliation, early leaf-expansion, mid leaf-expansion, and late leaf-expansion. Environmental variables (soil and atmospheric temperature and humidity, etc.) were recorded synchronously. We then applied XGBoost coupled with the SHAP interpretability framework to quantify the relative contributions of candidate drivers. Important findings (1) Phenology strongly modulates water use: Fd and tree-level water use differed significantly among phenophases (p < 0.05), in the order of late leaf-expansion > mid leaf-expansion > defoliation > early leaf-expansion; the daily mean Fd in late leaf-expansion was 4.4–6.1 times that of the leafless periods (defoliation and early leaf-expansion), indicating a significant increase in water consumption (p < 0.05). (2) The dominant environmental drivers shift with phenology. During defoliation, air temperature (Ta; 32.39%) and volumetric soil water content at 5 cm depth (VWC₅; 24.38%) were the primary drivers. In the early leaf-expansion, soil temperature (Ts; 36.61%) and photosynthetically active radiation (PAR; 19.55%) prevailed. During mid leaf-expansion, vapor pressure deficit (VPD; 52.16%) dominated. In late leaf-expansion, Ta again was the key driver (62.75%). Mechanistically, leaf phenology modifies canopy leaf area (and thus canopy conductance), reshaping the soil–plant–atmosphere water-potential gradient, so that limitation of water use shifts from soil water availability and temperature toward atmospheric evaporative demand; this shift amplifies phenophase-dependent differences in sap flux and water consumption. These results delineate the phenology‐dependent dynamics and driver transitions of sap flow in H. brasiliensis, and provide an evidence base for water resource management and strategy development in tropical plantations.

Key words: Key words rubber trees, phenophase, machine learning model, environmental factors, water use