Uncorrected soil water isotopes through cryogenic vacuum distillation may lead to a false estimation on plant water sources

Successful use of stable isotopes (δ2H and δ18O) in ecohydrological studies relies on the accurate extraction of unfractionated water from different types of soil samples. Cryogenic vacuum distillation (CVD) is a common laboratory‐based technique used for soil water extraction; however, the reliability of this technique in reflecting soil water δ2H and δ18O is still of concern. This study examines the reliability of a newly developed automatic cryogenic vacuum distillation (ACVD) system. We further assessed the impacts of extraction parameters (i.e. extraction time, temperature and vacuum) and soil properties on the recovery of soil water δ2H and δ18O for the ACVD and traditional cryogenic vacuum distillation (TCVD) systems. Finally, we investigated the potential influence of CVD (ACVD and TCVD) technique on the prediction of plant water uptake through a sensitivity analysis. Both ACVD and TCVD similarly extracted water from the rewetted soils, but none of the CVD systems successfully recovered the isotopic signatures of doped water from soil materials. Mean δ2H offsets of extracted soil water were −2.6 ± 1.3‰ and −2.4 ± 1.7‰ for ACVD and TCVD, respectively, while mean δ18O offsets were −0.16 ± 0.14‰ and −0.39 ± 0.37‰. The isotopic offsets of CVD systems were positively correlated with soil clay content, and negatively correlated with soil water content. Using corrected soil data (with CVD offsets) could improve the prediction of plant water uptake based on its high correlation with the environmental factors. This study identifies the isotopic offsets of CVD systems (i.e. ACVD and TCVD) and provides possible solutions for better predicting plant water sources. Even though, the wide use of CVD techniques probably induce noticeable uncertainties in the prediction of plants water uptake depths. The dataset of soil water extraction in this study will have implications for the technological development of CVD techniques.


| INTRODUC TI ON
Soil water hydrogen and oxygen stable isotopes (δ 2 H and δ 18 O, respectively) have broad applications in ecological studies (Coplen, 2011;Ding et al., 2018;Gat, 1996). Such studies have significantly improved our understanding of the eco-hydrological processes, particularly the prediction of pant water sources (Ehleringer & Dawson, 1992;Grossiord et al., 2017), the interpretation of intrainterspecific water competition for water resources (Nie et al., 2019;Schwendenmann et al., 2010) and the identification of hydrological niche segregation among plant communities (Moreno-Gutiérrez et al., 2012;Rodríguez-Robles et al., 2020). Soil water δ 2 H and δ 18 O have also been applied to assume the existence of two water worlds (TWW), whereby plant-available water in the soil is distinct from stream water (Brooks et al., 2010;Chen et al., 2020;Good et al., 2015). Given the technical advancements in isotope ratio mass and infrared spectrometers (IRMS/IRIS), a bottleneck in soil stable isotope studies remains the ability to obtain measurable amounts of liquid water from samples rather than the analysis process itself (Gehre et al., 2004;Griffis, 2013;Wassenaar et al., 2012). Hence, the outcomes of the above-mentioned ecological studies highly rely on the accurate extraction of unfractionated water from different types of soils (Barbeta et al., 2019(Barbeta et al., , 2020Jiang et al., 2022;Orlowski, Breuer, et al., 2016Orlowski, Pratt, et al., 2016;Orlowski, Winkler, et al., 2018;von Freyberg et al., 2020).
The use of stable isotopes as tracers of plant water uptake is typically based on three prerequisites. First (P1), there are vertical gradients in water δ 2 H and δ 18 O within the soil profiles. The isotopic gradients in soil water are mainly caused by two independent processes, which are soil water evaporation and mixing between new and old water (e.g. rain, dew and fog) inputs (Dawson & Simonin, 2011;Gat, 1996). Second (P2), the dual-water isotopes of δ 2 H and δ 18 O would not fractionate during root water uptake and/or xylem redistribution of the source water (Ehleringer & Dawson, 1992;White et al., 1985). This assumption has been widely proved by previous studies apart from the halobiotic and xerophytic species (Ellsworth & Williams, 2007;Lin et al., 1993). Third (P3), the CVD technique is a robust method toward the 'truthful' isotope recovery of plant xylem and soil water (Newberry, Nelson, et al., 2017;West et al., 2006).
Several studies have challenged the last prerequisite and showed isotopic biases for plant and soil materials (Barbeta et al., 2019;Chen et al., 2020;Meißner et al., 2014). The isotopic effects of CVD techniques will further complicate the quantification of source water contributions to plants (Orlowski, Breuer, et al., 2016;Yang et al., 2015).
For instance, Chen et al. (2020) suspected that the TWW hypothesis (i.e. mature riparian trees do not utilize stream water) might result from the artefacts associated with xylem water cryogenic extraction. Therefore, the concept of soil water-excess (SW-excess) has been suggested to correct the isotopic biases of plant xylem water (Barbeta et al., 2019;Landwehr & Coplen, 2006;Li et al., 2021). However, the potential impacts of CVD on plant water source prediction have not been elucidated yet for soil water extraction. In addition, an accurate reflection of soil water δ 2 H and δ 18 O is also the basis of the SWexcess correction (Barbeta et al., 2020).
To fill the above-mentioned knowledge gaps, an inter-laboratory comparison of different CVD systems is required to test their performances on soil water extraction. We examined an automatic 4. This study identifies the isotopic offsets of CVD systems (i.e. ACVD and TCVD) and provides possible solutions for better predicting plant water sources. Even though, the wide use of CVD techniques probably induce noticeable uncertainties in the prediction of plants water uptake depths. The dataset of soil water extraction in this study will have implications for the technological development of CVD techniques.

K E Y W O R D S
eco-hydrology, extraction technique, inter-laboratory comparison, isotopic offsets, stable isotopes, water uptake cryogenic vacuum distillation (ACVD) and two traditional extraction (TCVD) systems. This study aims to identify the reliability of soil water CVD techniques and evaluate the isotopic effects of soil water extraction during the determination of plant water sources. We also introduced the SW-excess correction of Barbeta et al. (2019) and re-analysed the dual-water isotopes in plant xylem and soils across seven sites, including one grassland, two croplands, three plantations and one evergreen broadleaf forest in China. We hypothesized that (i) the ACVD and TCVD have different performances on recovering the soil water δ 2 H and δ 18 O, and (ii) the partitioning results of plant water sources will be affected by local environmental factors at the research sites.

| Overview
This study assessed the CVD systems for stable isotope analysis and plant water source prediction through the following experiments. Experiment (I): 225 soil samples containing various rates of clay particles were oven-dried, doped with deionized water, and cryogenically extracted using ACVD and TCVD systems. Experiment

| Description of two types of CVD systems
The automatic cryogenic vacuum distillation (ACVD) system (LI- The traditional mode of cryogenic vacuum distillation (TCVD) system consists of five independent extraction lines (West et al., 2006). The collection tubes are cooled by manually adding liquid nitrogen. The extraction time was visually determined for the traditional extraction system when no additional water vapour appeared in the collection tubes. The TCVD system generally performs better in terms of vacuum threshold (0.3-2 Pa) than ACVD (Orlowski et al., 2013;West et al., 2006). However, the performance of TCVD system can highly depend on the operator's experience and skills.
Two TCVD systems are installed in the Technical Laboratory of LICA United Technology Limited (LICA) and the Key Laboratory of Northwest A & F University (NAFU).
Before formal water extraction, the extraction system should be evaluated to check its manifold tightness. The ACVD system can automatically check leakage according to a predetermined program.
Within 10 min, the vacuum degrees of ACVD and TCVD systems must be lower than 600 and 8 Pa (i.e. the threshold for the leakcheck). Once the xylem/soil samples are frozen in the extraction tubes, the extraction lines will be pumped into vacuum again. A certain amount of glass wool or fleece can be packed above the extracted samples to prevent the spread of soil particles (Orlowski et al., 2013;West et al., 2006). If the vacuum degree is better than the leak-check at this step, the extraction tubes will be heated until water vapour emanating from the samples is completely trapped in the collection tubes.

| Functionality test of CVD systems on soil water extraction
In Experiment (I), we selected three soil textures for rehydration treatments: sandy loam soil from Loess Plateau, loam soil from Northeast Plain and clay soil from Yunnan-Guizhou Plateau ( Figure 1; Table 1). Soil textures were classified based on United States Department of Agriculture (USDA) classification procedures (Baillie, 2001). Each soil texture (~3 kg each) was homogenized, sieved (2 mm) and divided into 75 parallel sets (75 sets/ samples × 3 sites/textures).
Step by step, 15 soil samples were randomly picked from each of the three textures. Thus, we obtained five subsets of soil materials (5 subsets × 3 textures × 15 replications), which were pre-dried (105°C, 72 h) and shipped to the following laboratories: three sets to the Institutional Center for Shared Technologies and Facilities of XTBG, one set to the Technical Laboratory of LICA and one set to the Key Laboratory of NAFU. Simultaneously, 2.5 L of deionized (DI) water (δ 2 H = −144.9 ± 0.5‰; δ 18 O = −19.42 ± 0.12‰) was divided into quadrants, filled into plastic bottles (100 mL) and also delivered to these laboratories. The remaining bottle of DI water was kept as a standby reference. In each laboratory, the sub-samples of soils were packed into 10-mL screw-cap glass vials, oven-dried (105°C, 12 h) and doped with DI water in a decreasing order from 3.0 ± 0.6 to 1.2 ± 0.2 mL (SWC = ~10%-30%). Finally, the remoistened subsamples were extracted using: (i) ACVD system: soil water extraction was first conducted at 180 min, 85°C and ≤400 Pa (ACVD 3H ). Then, the extraction time was extended to 240 min (ACVD 4H ). Finally, the extraction time (240 min) and heating temperature (105°C) were both improved (ACVD 4HT ).
The water recovery rate (WRR) of the rehydration experiment was carefully determined (precision 0.0001 g) to check whether the DI water was completely removed from the soil materials (ME204; Mettler-Toledo Inc.): where W wet is the weight of remoistened soil (g), W extracted is the weight of soil after extraction (g) and W DI water is the initial weight of DI water (g). All samples were weighed after equalizing to room temperatures.

| Sensitivity analysis of CVD on plant water source prediction
In Experiment (II), we re-analysed the dual-water isotopes of

| Measurement of stable isotopes of hydrogen and oxygen
In experiments (I), the δ 2 H and δ 18 O of water samples were meas- where R sample is the isotope ratio of a water sample and R VSMOW is the isotope ratio of the VSMOW standard.
It is critical for the inter-laboratory comparison to have normalized water standards and measurement procedures (Coleman & Meier-Augenstein, 2014). Therefore, we pre-calibrated the laboratory water For the BN and JD sites, measurements of water samples were conducted using a DELTA-V-Advantage isotope ratio mass spectrometer (Thermo Fisher Scientific) (Song et al., 2022;Yang et al., 2020). The isotope ratio mass spectrometer was coupled with a high-temperature conversion elemental analyser (HTC/EA) and an AS1310 auto-sampler (Thermo Fisher Scientific). According to our test results, the precision of the IRMS was ±0.4‰ for δ 2 H and ±0.14‰ for δ 18 O.

| Statistical analysis
The statistical analyses were performed with R software (Version logarithmic transformations. The model estimates were computed using 'nlme' package (Pinheiro et al., 2021;Stoffel et al., 2017).

| Effects of soil properties on soil water isotopes
The isotopic offsets of CVD techniques were influenced by soil textures and soil moisture conditions (Figure 3). Here, only the F I G U R E 2 Hydrogen (a-c) and oxygen (d-f) isotope offsets between deionized (DI) water and the same water cryogenically extracted from rehydrated soils (clay loam, loam, and clay, n = 15). Isotopic offsets are differences between the extracted soil water and the spiked DI water (δ 2 H/ 18 O Soil water − δ 2 H/ 18 O DI water ). ACVD: automatic cryogenic vacuum distillation (3H: 85°C and 180 min; 4H: 85°C and 240 min; 4HT: 105°C and 240 min). TCVD: traditional cryogenic vacuum distillation (1# and 2#: 105°C and 120 min in two laboratories). Box size represents the interquartile range, whiskers indicate variability outside the upper and lower quartiles, black lines are the medians, and individual points are outliers. Box values not sharing the same letters are significantly different (p < 0.05). The presence of an asterisk below a box indicates that the value is significantly different from zero (ns: p > 0.05, *p < 0.05, **p < 0.001).
results of ACVD 4HT were used for further comparisons with TCVD 1# and TCVD 2# . The isotopic offsets of ACVD 3H and ACVD 4H were higher than those of the other treatments under lower extraction temperatures. All the CVD systems tended to produce more isotopic offsets for soils with higher clay contents (Figure 3a,b).

| Impacts of CVD techniques on plant water source prediction
Soil water δ 2 H and δ 18 O of different sites occupied the dual-isotope plots from the LWMLs to their right sides (Figure 4). The slopes of the F I G U R E 3 Soil water δ 2 H offsets as a function of soil water δ 18 O offsets (a) for two types of cryogenic vacuum distillation systems (n = 15). Confidence ellipses and light grey shaded areas indicate the 95% prediction areas and 95% confidence intervals of linear regressions, respectively. The insert panel depicts the relationship between soil water δ 2 H offsets and soil clay content (b) for automatic cryogenic vacuum distillation (ACVD: 105°C and 240 min) and traditional cryogenic vacuum distillation (TCVD 1# and TCVD 2# : 105°C and 120 min in two laboratories). Relationships of soil water δ 2 H (c, d) and δ 18 O (e, f) offsets to soil water content (SWC) are also presented in (c)-(f) (ns: p > 0.05, *p < 0.05, **p < 0.001).
LMWLs increased gradually from 6.5 to 8.6, reflecting the climate characteristics of the study sites from inland oasis (ZY) to coastal hills (TH). Most of xylem water fell within the δ 2 Hδ 18 O spaces of soil water.
However, some part of xylem water δ 2 H and δ 18 O distributed besides the ranges of soil water. Thus, raw soil water data were corrected based on the isotopic correction equations (details in Appendix B): (i) ACVD system: (ii) TCVD system: where δ 2 H offset and δ 18 O offset are the potential isotopic offsets of soil water extraction caused by TCVD and ACVD systems, SCC and SWC are the soil clay contents (%) and soil water content (%), respectively.
However, all the models using corrected data (i.e. SW-excess and/ or CVD offset correction) reasonably predicted plant water uptake depths than the input of raw data (marginal R 2 = 0.078). The potential isotopic biases in plant xylem and soil water could alter the definition of principal environmental factors that related to plant water uptake.

| DISCUSS ION
This study investigates the reliability of two laboratory-based CVD techniques for soil water extraction. To avoid incomplete extraction caused by artificial manipulations, the amounts of the input water into soil materials were strictly tracked before and after extraction. Theoretically, an incomplete extraction leads to more negative δ-values in the collected water (Orlowski et al., 2013). Based on the Rayleigh distillation mechanisms, the δ 2 H and δ 18 O of the removed water will be 'lighter' than the remaining fraction (Gat, 1996;Lee et al., 2005). A small leakage in the vacuum system may also cause the condensation of laboratory ambient air moisture in the extraction lines, which could commonly conduct to more negative δ 2 H and δ 18 O of extracted water. Here, outputs of the spectralcontamination-identifier were 0.0614 ± 0.0089 for methanol (NB) and 1.0001 ± 0.0002 for ethanol (BB) interferences. This spectral contamination probably conducts to more positive soil water δ 2 H (0.14 ± 0.15‰) and δ 18 O (0.07 ± 0.09‰), which was not considered because these isotopic errors were smaller than the measurement F I G U R E 5 Variations of xylem water SW-excess at seven experimental sites. Box size represents the interquartile range, whiskers indicate variability outside the upper and lower quartiles, black lines is the median, and individual points are outliers. The presence of an asterisk below a box indicates that the difference is significant between pairs of data (ns: p > 0.05, *p < 0.05, **p < 0.001).

TA B L E 2
Parameter estimates of fixed effects and R squares of the generalized linear mixed models on source water contributions for each input data. VPD is vapour pressure deficit and SWC is soil water content. Bold numbers are the largest estimates of fixed effects (i.e. VPD and SWC at two soil depths) and marginal R squares (reflecting variances in source water contributions explained by independent fixed variables) for each model.
Under a longer extraction time (240 min), the ACVD system performed equally or even better than TCVD systems ( Figure 3). In fact, the constant δ 2 H and δ 18 O of extracts can still be inconsistent with their 'actual values' (Orlowski et al., 2013;Orlowski, Winkler, et al., 2018), although a minimum time of 30-40 min is sufficient for field sample extraction (West et al., 2006). The results also showed that the isotopic offsets of ACVD decreased with the increasing extraction temperature (85-105°C), highlighting the importance of extraction temperature for a low vacuum extraction system. The dependence of isotopic bias on the extraction temperature has also previously been reported by Walker et al. (1994), Araguás-Araguás et al. (1995) and Palacio et al. (2014). Higher extraction temperature (100-150°C) will speed up the rate at which extraction is completed; however, it might also decompose the soil organic matter and release crystallized water (Walker et al., 1994). Orlowski et al., 1995;Orlowski, Pratt, et al., 2016). The isotopic effects of this heat-labile water were more important when the tightly bound water composed large fraction of the total added water (Ingraham & Shadel, 1992). Our results also supported this conclusion since the 2 H and 18 O of extracted water were more depleted for clay soils (Figure 3a,b), which commonly have a high amount of bound water (Araguás-Araguás et al., 1995;Koeniger et al., 2011;. In our rehydration experiment, the soil materials were collected from the Loess Plateau (AS), Northeast Plain These findings could primary infer that isotopic background values were not the key driving factors behind the depleted water for clay soils, because the soil water (including precipitation) δ 2 H and δ 18 O of BN generally varied within the ranges of AS and SY sites. However, the soil materials should be rewetted with repeated addition of the spiked water to diminish the possible background influences (Newberry, Prechsl, et al., 2017;Orlowski, Pratt, et al., 2016;Thielemann et al., 2019). If more spiking waters with two strongly different isotopic compositions are used in the rehydration experiments, it will also increase the confidence in assessing the soil-bound water effects. Therefore, a better understanding of the isotopic exchange effect between soil-bound water and spiked water is urgently needed in future research.
Nowadays, several studies highlighted the substantial 'unexplained' uncertainties concerning isotope-based plant water source partitioning (Barbeta et al., 2019;Li et al., 2007;Meißner et al., 2014;Orlowski et al., 2013;Yang et al., 2015). Therefore, one should pay attention to the isotopic offsets of CVD techniques. Our results showed that the ACVD and TCVD systems resulted in noticeable biases in soil water δ 2 H and δ 18 O (Figure 3). The isotopic offsets of CVD might vary significantly across study locations and laboratories Orlowski, Pratt, et al., 2016). Here, among the seven experimental sites in China, the isotopic offsets of CVD systems ranged from −3.7‰ to −1.7‰ for soil water δ 2 H, and from −0.27‰ to −0.10‰ for soil water δ 18 O (Figure 4). Using soil data corrected by CVD offsets improved the prediction of plant water sources according to its correction with environmental factors ( Table 2). The SW-excess can also be used to reduce the possible impacts of different isotopic biases in plant water source partitioning (Barbeta et al., 2019;Li et al., 2021). Therefore, the effective use of SW-excess would highly depend on the accurate extraction of unfractionated water from different types of soils ( Figure 5). The sensitivity analysis of this study suggested that the isotopic biases of CVD systems would lead to a high risk of miscalculating the water uptake fractions (e.g. at BN and JD sites). These findings suggest that one should pay more attention for examining and optimizing CVD techniques for ecohydrological studies.

| CON CLUS IONS
The present study assessed the reliability of CVD techniques on soil water extraction, and evaluated the potential influences of cryogenic extraction on plant water source partitioning. Our results showed that the newly designed ACVD technique could have similar or better performance compared with TCVD system. However, the two CVD techniques did not achieve unfractionated soil water

ACK N O WLE D G E M ENTS
We are thankful to Xue-Fa Wen, Ya-Kun Tang, Bin Chen, Jie Zhao and Corlett Richard for providing scientific data and suggestions.
We further acknowledge the supports from the Institutional Center

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare that there are no known competing financial interests or personal relationships that will influence the current work reported in this paper.