Water-use efficiency in a humid karstic forest in southwestern China: Interactive responses to the environmental drivers

Water-use efficiency (WUE) is a key parameter that measures the balance between carbon gain and water loss in an ecosystem. Achieving better understanding of how WUE responds to environmental drivers is critical for revealing ecohydrological processes and mechanisms, and for biophysical modelling to investigate the interactions between vegetation and the changing environment. Karst landforms cover similar to 10 % of the Earth's land surface; however, to date few studies have focused on the behaviours of WUE in karstic ecosystems. To fill these knowledge gaps, a two-year experiment was conducted to investigate the WUE and its drivers in a climax karst forest in southwestern China. Results showed that both the GPP (Gross Primary Production) and ET (evapotranspiration) showed exponential relationships to Ta (air temperature, P < 0.05), but linear relationships to Rn (net radiation, R-2 = 0.38 (GPP) and 0.78 (ET), P < 0.05) and VPD (water vapour pressure deficit, R-2 = 0.36 (GPP) and 0.28 (ET), P < 0.05), being quite different from those in other ecosystems. The reasons would be the different habits of the karstic plants, and the facts that the root zone water availability rarely dropped to low levels due to the frequent precipitation. The WUE, with the mean of 8.0 +/- 4.8 mg CO2 g(-1) H2O, exhibited negative relationships to VPD (R-2 = 0.23, P < 0.05), Rn (R-2 = 0.28, P < 0.05), and Ta (air temperature, R-2 = 0.42, P < 0.05), and a positive linear relationship to the Bowen ratio (R-2 = 0.17, P < 0.05). Besides these, results also showed that the effects of a specific driver (e.g., VPD) on WUE was affected obviously by other factors (e.g., Ta, Rn), indicating that whether a driver is important or not to WUE, depends strongly on the environmental conditions. As the WUE-response formulas are often used to investigate and model the interactions between vegetation and environment, these findings suggested that the modelers should caution the versatility of the WUE-response formulas.