Quantifying the sensitivity of trees to environmental variables such as vapor pressure deficit (D) is important for understanding the hydrological responses to land use changes and the role of underlying physiological mechanisms. Sap flux (Js), water transport, and sensitivity of canopy stomatal conductance (Gs) to D (m) were determined during the early dry season in 2012 and 2013 for young timber trees (5 years), secondary forest trees (∼25 years), and mature forest trees (∼50 years) at three sites of lower subtropical China. The Js was found to be negatively correlated with tree biometric parameters, and whole tree water transport for a given tree size decreased with age or size, which was partially attributed to the lower sapwood quantity for a given diameter. The m and reference Gs at 1 kPa (Gsref) of the seven tree species decreased from young timber to mature trees, although not significantly. The D, which is the extrapolated value when stomata closure (eb/m), significantly varied among species, which decreased from isohydric Eucalyptus grandis × urophylla to more anisohydric Acacia auriculiformis and Eucalyptus citriodora. Js and maximum sap flux (Js,max) decreased with wood density, suggesting that water flux was regulated by biophysical constraints. Wood density was a good predictor of Gsref (r2 = 0.664, P = 0.026), indicating that water flux to D is an intrinsic adaptive characteristic. To the best of our knowledge, this is the first study linking wood density and Gsref. The responses of water flux to D varied among species, size or age, and hydraulic architecture, which highlights that modeling water budget to land cover transition based on the same hydraulic parameter from one land use scenario can potentially cause extensive errors in model prediction. Our empirical research could be used for the evaluation of the consequences of land use changes on the hydrological process of forest ecosystems.