Coordination of available soil water content and root distribution modifies water source apportionment of the shrub plant Caragana korshinskii

Abstract

Stable isotopes have been widely used to identify root water uptake (RWU) by classifying potential water sources as distinct endmembers and evaluating their contributions to xylem water. However, the estimated contributions of endmembers (mainly soil layers) are usually based on variations in soil water isotopes alone. Available soil water and root distributions are key limiting factors of RWU but are rarely considered in water source apportionment. Thus, we have compared the relative contributions of distinct soil layers based on mean soil water isotope values, and values weighted by both available soil water content (AWC) and root weight density (RWD), to RWU of Caragana korshinskii. We derived these values (hereafter mean and weighted contributions, respectively) using three Bayesian mixing models (SIAR, simmr and MixSIAR) at three sites with different water conditions. We calculated the differences between the mean and weighted contributions (DC) and the accumulation of the absolute value of DC (AADC) to analyse the differences between them and their relationships with AWC and RWD. Both the weighted and mean contributions varied with sites and models. We obtained the following AADC values: 27, 8 and 11 % for Sites 1–3, respectively, using SIAR; 39, 13 and 14 %, respectively, using simmr; 68, 40 and 25 %, respectively, using MixSIAR. We detected a significant correlation between DC and RWD when AWC ≤ 6 %, as well as a significant correlation between DC and AWC when AWC > 6 %, indicating that the influence of RWD on DC depended on soil water conditions. Based on our findings, endmembers weighted by AWC and RWD altered the proportion of water source allocation relative to non-weighted endmembers, while the magnitude of the effect was related to the model used. Thus, we suggest careful consideration of the characterisation of endmember isotopes and model selection when partitioning plant water sources using δ2H and δ18O.