A δ2H offset correction method for quantifying root water uptake of riparian trees

Root water uptake plays an important role in water cycle in Groundwater-Soil-Plant-Atmosphere-Continuum. Stable isotopes (δ2H and δ18O) are effective tools to quantify the use of different water sources by plant roots. However, the widespread δ2H offsets of stem water from its potential sources due to δ2H fractionation during root water uptake result in conflicting interpretations of water utilization using stable isotopes. In this study, a potential water source line (PWL), i.e., a linear regression line between δ18O and δ2H data of both soil water at different depths and groundwater, was proposed to correct δ2H offsets of stem water. The PWL-corrected δ2H was determined by subtracting the deviation between δ2H in stem water and PWL from the original value. The MixSIAR model coupled with seven types of input data (i.e. various combinations of single or dual isotopes with uncorrected or corrected δ2H data by PWL or soil water line (SWL)) were used to determine seasonal variations in water uptake patterns of riparian tree of Salix babylonica (L.) along the Jian and Chaobai River in Beijing, China. These methods were evaluated via three criteria including Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and root mean square error (RMSE). Results showed that different types of input data led to considerable differences in the contributions of soil water at upper 30 cm (9.9–57.6%) and below 80 cm depths (29.0–76.4%). Seasonal water uptake patterns were significantly different especially when δ2H offset was pronounced (p < 0.05). The dual-isotopes method with uncorrected δ2H underestimated the contributions of soil water in the 0–30 cm layer (by 30.4%) and groundwater (by 56.3%) compared to that with PWL-corrected δ2H. The PWL correction method obtained a higher groundwater contribution (mean of 29.5%) than that estimated by the SWL correction method (mean of 24.5%). The MixSIAR model using dual-isotopes with PWL-corrected δ2H produced the smallest AIC (94.1), BIC (91.9) and RMSE values (4.8%) than other methods (94.9–101.7, 92.6–99.5 and 5.3–12.4%, respectively), which underlined the best performance of PWL correction method. The present study provides crucial insights into quantifying accurate root water uptake sources even if δ2H offset exists.