The Badain Jaran Desert (39°04′15″–42°12′23″ N, 99°23′18″–104°34′02″ E) is positioned in the western part of the Inner Mongolia Autonomous Region, China, occupying an area of 5.2 × 10⁴km²(Zhu et al., 2010). The Badain Jaran Desert is situated in the center of the Alashan desert, with a typical continental climate, average annual rainfall of 76.9 mm (Ma et al., 2014), and mean annual temperature of 7–8°C. Northwest and west winds prevail perennially, with an annual mean wind speed of 3.0–4.5 m·s^-1, gradually intensifying from east to west. The desert has a wide range of mega-dunes, generally 150–300 m in height and up to 430 m in height. Numerous small inland lakes are present in the lowlands between the mega-dunes, The lakeside zone consists mainly of saline lakes, and the surrounding zone is dominated by xerophytic and saline-tolerant plants, such as Phragmites australis, Glycyrrhiza uralensis, Achnatherum splendens, and N. tangutorum. Among these, N. tangutorum is the most widespread shrub in the lakeshore zone, and is an important building block species in desert and saline regions with high ecological, medicinal, and nutritional value (Liu et al., 2021).

This experiment was performed in the Badan Lake (39°43′19.31″ N, 102°37′1.74″ E, 1218 m) area in the southeastern part of the Badain Jaran Desert. Badan Lake includes the Badan East Lake and Badan West Lake. The former is a freshwater lake with less surrounding natural vegetation owing to tourism development in the area. The latter is a highly saline lake surrounded by shrubs and wetland grasses (Luo et al., 2017). The shrubs are primarily Nitraria nebkha species. The groundwater table in the lake basin region is 1.5–2 m, with significant daily and seasonal fluctuations (Huang, 2018). Nitraria nebkhas at distances of 20, 50, and 100 m from the Badan West Lake shore were selected as the experimental sample sites.

Samples were collected from May to August 2020 during the growth period, once in the middle of each month (5.15-5.20; 6.11-6.16; 7.15-7.20; 8.13-8.18) for a total of four collections ( Figure 1 ). Suberized and non-green twigs of N. tangutorum were clipped into 3–4 cm branch sections, and the phloem tissues were gently removed using tweezers. Then, these twigs were immediately put into airtight glass vials with screw-tops, sealed with Parafilm to avoid evaporation, and stored in a portable freezer at -20°C. Leaves were collected, stored in envelopes, ventilated, and brought back to the laboratory for oven-drying. In parallel with plant sampling, soil samples were acquired around the marked plant individuals by excavating soil profiles, removing the surface dry sand layer (approximately 30–100 cm thick) and collecting at 30 cm intervals from the appearance of a moist sand layer until groundwater emerged (approximately 1.2–1.5 m in depth). If groundwater was not unearthed, eight layers were collected (0–30, 30–60, 60–90, 90–120, 120–150, 150–180, 180–210, and 210–240 cm), and three replicates were collected for each soil layer at each site. Each soil sample was separated into two pieces: one was immediately put into screw-top glass vials, wrapped with Parafilm, and kept at −20°C; the other was put into an aluminum box to measure the gravimetric soil water content (SWC), which was obtained by weighing in situ and then brought back to the laboratory to be dried in an oven at 105°C to constant weight. After the samples were collected, the holes were backfilled with sand. Groundwater and lake water were sampled from the wells of pastoralists residing near (approximately 1 km) the experimental site and Badan Lake, respectively, using capped vials sealed with parafilm to prevent evaporation. A total of 819 soil isotope, 108 plant stem, 24 lake water, 12 groundwater, and 108 plant leaf samples were collected.

A cryogenic vacuum distillation system was used to extract water from plant xylem and soil samples. The extraction glass tubes were covered by a heating unit at 95-100°C, and the collection glass tubes were submerged in liquid nitrogen to capture the extracted water vapor. The vacuum pressure was 0.02 Pa. The procedure required 2–3 h depending on the moisture content of the sample. The proportion of water derived from the xylem and soil reached 98%. The hydrogen and oxygen isotope compositions (δD and δ¹⁸O) of all water samples were determined using an isotope ratio infrared spectroscopy (IRIS) system (LWIA, 912-0008-1001, Los Gatos Research Inc., Mountain View, CA, USA) with measured accuracies of 0.3 ‰ for δD, and 0.1 ‰ for δ¹⁸O (Zhou et al., 2019). Given that organic contaminants in water derived from plant twigs interfere with the δD and δ¹⁸O determinations by the IRIS system, the identification and quantification of organic contaminants were performed using spectral contamination post-processing software, and isotopic values of samples were corrected (Schultz et al., 2011; Zhou et al., 2019). The δ¹³C in leaf samples was measured using an isotope ratio mass spectrometer (IRMS) (DELTA V Advantage, Thermo Fisher Scientific, Bremen, Germany) with an accuracy of 0.15 ‰ for δ¹³C (Wang J. et al., 2019). The sample isotopic compositions are expressed as follows:

where X represents D, ¹⁸O or ¹³C, and R_sample and R_standard are the isotopic compositions (D/¹H, ¹⁸O/¹⁶O, and ¹³C/¹²C ratios) of the sample and standard, respectively. The Vienna Standard Mean Ocean Water (V-SMOW) was used as the standard for D and ¹⁸O and the Vienna Pee Dee Belemnite (V-PDB) was used as the standard for ¹³C.

The known δD and δ¹⁸O values of water sources and plant xylem water allow the analysis of water sources absorbed by plants and to quantify the contributions of varying water sources. Based on the similarity in SWC, isotopic composition, and vertical and temporal variations of individual soil layers, the soil layers were aggregated into three main categories: shallow soil layer (0–90 cm), middle soil layer (90–180 cm), and deep soil layer (180–240 cm) to facilitate follow-up analysis and comparison. The IsoSource model (Phillips and Gregg, 2003) was used to assess the proportions of various water sources absorbed by N. tangutorum. Considering the hydrogen isotopic fractionation of xerophytic plants (Ellsworth and Williams, 2007; Zhao et al., 2016) and hydrogen isotope biases produced by the cryogenic vacuum distillation system (Chen et al., 2020), we select only δ¹⁸O values for the analysis and calculation of water sources.

The Ψ _pd, Ψ _md, and evening leaf water potential (Ψ _en) were measured monthly with a pressure chamber water potential meter (Plant Moisture Stress; Corvallis, Oregon, USA) to assess plant water status. Three healthy, uniformly grown individuals from each site were selected and labeled. From each individual, three well-developed leaves without mechanical damage were cut and immediately placed into the pressure chamber to obtain Ψ. Ψ _pd, Ψ _md, and Ψ _enwere measured before sunrise, at solar noon, and during the evening, respectively. Each sample site was measured 27 times per day for 3 consecutive days on sunny days in the middle of each month for a total of 4 months, yielding a total of 972 Ψ data and 324 each of Ψ _pd, Ψ _md, and Ψ _en data.

Variations in the SWC and hydrogen and oxygen isotopic composition of soil water (δD_s, and δ¹⁸O_s) with the soil depth and month were analyzed using two-way analysis of variance (ANOVA). Variances in the hydrogen and oxygen isotopic composition of plant xylem water (δD_x, δ¹⁸O_x), δ¹³C, Ψ values, and water use were also detected using two-way ANOVA with the month and distance from the lake as the fixed effects. One-way ANOVA was conducted to identify monthly variations in the hydrogen and oxygen isotopic composition of groundwater and lake water (δD_g, δ¹⁸O_g, δD_l, and δ¹⁸O_l). Variations in the δ¹³C, Ψ _pd, Ψ _md, and Ψ _envalues among months and distances from the lake were examined using one-way ANOVA with the post hocTukey’s honestly significant difference (HSD) test. Pearson correlation analysis was performed to identify possible correlations between SWC, δ¹³C, Ψ _pd, Ψ _md, and Ψ _envalues and the distance from the lake. Data analyses were conducted using the SPSS software (version 21.0; SPSS Inc., Chicago, IL, USA) and figures were constructed using the Origin 2017 software (OriginLab Corp., Northampton, MA, USA).

SWC in the N. tangutorumhabitats varied significantly with distance from the lake ( Figure 2 ), and a significant negative correlation was identified between SWC and distance from the lake (r= −0.427, P< 0.001, data not shown). Concurrently, SWC in the N. tangutorum habitats exhibited significant variation depending on soil depth (P< 0.001) and month (P< 0.001) at each study site ( Table 1 ). With increasing soil depth, SWC increased gradually with minor undulations ( Figure 2 ), and thereafter rapidly increased as the depth increased closer to the groundwater table, especially at 20 m and 50 m from the lake ( Figures 2A–E ). SWC in the shallow, middle, and deep soil layers varied considerably across months (P< 0.001) and distances from the lake (P< 0.001) ( Table 2 ). The average SWC at each studied site was higher in May (20 m: 12.94%; 50 m: 3.64%; 100 m: 1.87%) and June (20 m: 8.40%; 50 m: 1.71%; 100 m: 1.45%) and lower in July (20 m: 7.00%; 50 m: 1.52%; 100 m: 1.27%) and August (20 m: 6.83%; 50 m: 0.90%; 100 m: 0.85%). The overall SWC value in each month was significantly higher at 20 m from the lake than at 50 m and over 100 m from the lake (P< 0.001).

The δD_svalues in soil water showed considerable differences with distance from the lake (P< 0.001), which were most enriched at distances greater than 100 m from the lake and most depleted at a distance of 20 m from the lake. In addition, the δD_svalues in soil water differed considerably with month and depth at each experimental site (P< 0.001) ( Table 1 ). As indicated, the δD_svalues in the shallow soil layers exhibited strongly pronounced monthly variations (P< 0.001), with smaller values occurring in May and June and larger values occurring in July and August. Conversely, no statistically significant monthly differences were detected in the δD_svalues of the deep soil layers (P> 0.05). Moreover, the δD_svalues declined in fluctuation from the surface of the wet sand layer downward, rather than decreasing monotonically, although a remarkable decrease in the δD_svalues existed from the shallow to middle to deep soil layers (P< 0.001). The variation trend in the δ¹⁸O_svalues of soil water tended to be uniform with the δD_svalues. The slope of the soil water evaporation line (SWL) progressively declined with increasing distance from the lake ( Figure 3 ).

The δD_g and δ¹⁸O_g values of the groundwater varied significantly with time (P< 0.05). However, considering the prominent monthly differences in isotopic signatures in other potential water sources, the δD_g and δ¹⁸O_g in groundwater were relatively stable ( Figures 4 ,  5 ).

Significant monthly variations were detected in the δD_land δ¹⁸O_lvalues in west lake water (P< 0.001). Nevertheless, the east lake water δ¹⁸O_lvalues exhibited notable monthly variations (P< 0.001), although no substantial temporal variation was identified in the δD_lvalues (P> 0.05) ( Figures 4 ,  5 ).

The δD_x and δ¹⁸O_xvalues of N. tangutorum both exhibited pronounced differences for two effects: month (P< 0.001) and distance from the lake (P< 0.01) ( Table 2 ). The patterns of monthly variation in the δD_x and δ¹⁸O_xvalues were similar, with greater depleted values recorded in May and June and greater enriched values noted in July and August ( Figures 4 ,  5 ). In addition, the δD_xand δ¹⁸O_xvalues were the closest to those in the soil water and groundwater ( Figures 4 ,  5 ), and their scatter points were located near the SWL and groundwater regions ( Figure 3 ). Accordingly, N. tangutorum may use soil and groundwater as its principal water sources at different distances from the lake.

Significant differences were identified in the proportion of shallow, middle, and deep soil water, and groundwater that contributed to N. tangutorum (P< 0.05; Figure 6 ). The IsoSource model revealed that N. tangutorum could absorb water from four potential water sources synchronously; however, the relative quantities differed depending on the month and distance from the lake. Groundwater was the primary contributor to N. tangutorum xylem water in May (63.8%) and August (53.5%) but not in June and July. In June, N. tangutorum depended on deep soil water (75.1%), whereas in July, soil water was mostly collected from different layers ( Figure 6 ). From May to June, the percentage of groundwater contributing to N. tangutorum xylem water decreased significantly, with a considerable increase in the contribution of deep soil water. The water uptake pattern of N. tangutorum at 20 m from the lake differed from those at the other two study sites, especially with respect to the proportion of middle soil water, although N. tangutorum at the other two sites exhibited comparable uptake patterns. The proportion of middle soil water contributing to N. tangutorum at 20 m from the lake was higher than that at the other two study sites ( Figure 6 ).

The N. tangutorum leaf δ¹³C values showed remarkable differences between two variables: month (P< 0.001) and distance from the lake (P< 0.001) ( Table 2 ). The mean leaf δ¹³C value of N. tangutorum was −25.70 ± 0.15 ‰, ranging from −27.36 ‰ to −24.42 ‰. The leaf δ¹³C values at each experimental site exhibited monthly variations (P< 0.001), with highly depleted values in May and June and highly enriched values in July and August. Furthermore, the N. tangutorum leaf δ¹³C values showed significant differences among sites at different distances from the lake (P< 0.001). The leaf δ¹³C values at 20 m from the lake were generally lower than those of N. tangutorum at the other two study sites, and a remarkable positive correlation was identified between leaf δ¹³C values and distance from the lake (r= −0.824, P< 0.001) ( Figure 7 ).

The Ψof N. tangutorum exhibited significant differences with the month (P< 0.001) and distance from the lake (P< 0.001) ( Table 2 ). Significant monthly variations in the Ψ _pd, Ψ _md, and Ψ _envalues were present for each experimental site (P< 0.01), and notable differences in the Ψ _pd, Ψ _md, and Ψ _envalues across studied sites were also identified in most months (P< 0.05) apart from the Ψ _mdin July and the Ψ _enin June (P> 0.05). The Ψ _pd, Ψ _md, and Ψ _envalues showed monotonically decreasing temporal patterns, being relatively high in May and June and low in July and August. Moreover, negative correlations existed between the Ψ _pd, Ψ _md, and Ψ _envalues of N. tangutorumand distance from the lake (r= −0.708, P< 0.001; r= −0.425, P< 0.05; r= −0.289, P= 0.087, respectively) ( Figure 8 ).

The δD values of rainfall in the experimental region varied from −227 ‰ to −20 ‰, whereas the δ¹⁸O values ranged from −28.5 ‰ to −1.3 ‰, and the equation of the local meteoric water line (LMWL) was δD=7.90δ¹⁸O+5.02 (Cao et al., 2021). The δD and δ¹⁸O values of the soil water, lake water, and groundwater were all situated at the bottom right of this LMWL ( Figure 3 ), demonstrating that all these water sources were impacted by evaporation enrichment (Chen et al., 2021). As scarce local rainfall inputs cannot offset for the water loss (Ma et al., 2014; Dong et al., 2016), another recharge source is essential. There are conflicting opinions regarding the regional water cycle (Chen et al., 2014; Zhan et al., 2018; Wang et al., 2021). Some researchers have concluded that local rainfall does not contribute substantially to the regional water systems (Chen et al., 2005; Dong et al., 2016; Jin et al., 2018). Wang et al. (2021) demonstrated that infiltration of local precipitation significantly recharges groundwater. Considering the precipitation isotopic values from previous studies (Jin et al., 2018; Cao et al., 2021) and our results, we inferred that direct infiltration of regional rainfall to recharge the water system exists, but it is not a major recharge source. Lake water is primarily replenished by groundwater in desert hinterlands (Dong et al., 2013; Jiao et al., 2015; Luo et al., 2017). However, the groundwater recharge source is inconclusive; in addition to the modern local precipitation recharge hypothesis mentioned above, various other hypotheses can offer plausible explanations, including the neighboring area recharge hypothesis (Wu et al., 2017; Jin et al., 2018), paleowater recharge hypothesis (Zhao et al., 2012), and remote source recharge hypothesis (Chen et al., 2005; Zhan et al., 2018). Furthermore, year-round variations in lake and groundwater levels in this region are primarily constrained by changes in evapotranspiration, with the highest water levels occurring in winter and spring, followed by a decline, a minimum from July to August, and a continuous increase after September (Lu, 2013). Lake basin vegetation exploits the groundwater, leading to diurnal fluctuations in the water table (Zhang et al., 2020), with the highest water level at approximately 08:00 and the lowest level at approximately 16:00 (Huang, 2018).

The vertical profile of the soil water isotope composition integrates the processes involved in rainfall replenishment, groundwater recharge, blending with pre-existing water, and evaporation (Tang and Feng, 2001; Brooks et al., 2015; Wang J. et al., 2019). Isotopic enrichment of the surface soil ( Figures 4 ,  5 ) was influenced by evaporation (Zhou et al., 2019), as well as evaporative water vapor condensation (Chen et al., 2014), of which strong evaporation resulted in a lower SWC in this layer ( Figure 2 ). In comparison, the deeper soil water contains less heavy isotopes than the shallower layers, primarily attributable to the fact that deep soil layers were less influenced by evaporation and more affected by capillary upward recharge of groundwater containing light isotopes (Wu et al., 2018; Zhou et al., 2019). Additionally, plant water uptake from deep soil layers and fluctuating variations in the groundwater table accounted for the fluctuating changes in SWC in this layer. The SWC and isotopic composition of the deep soils exhibited higher temporal stability than those of the shallow soils, which is in accordance with previous studies (Penna et al., 2013; Yang et al., 2014; Li et al., 2019). The δD_xand δ¹⁸O_xvalues in N. tangutorumxylem water were more similar to those in the soil water and groundwater ( Figures 4 ,  5 ), exhibiting significant monthly variations according to different uptake ratios. A non-negligible point is that δD_xvalues were more depleted than δ¹⁸O_xvalues compared with those of potential water sources, which numerous scholars have suggested may be due to the isotopic fractionation that occurs during root water uptake (Ellsworth and Williams, 2007; Zhao et al., 2016). However, a recent study found that this phenomenon is related to the cryogenic vacuum distillation system used to extract plant xylem water (Chen et al., 2020). This is something that should be explicitly considered when conducting research on plant water sources.

In this desert lake basin habitat, fluctuating variations in groundwater levels are responsible for significant differences in N. tangutorum water-use patterns and directly affected its growth (Li et al., 2013; Zolfaghar et al., 2015; Zhou et al., 2019). The IsoSource model revealed that N. tangutorum could access different water sources simultaneously; however, the percentages of the four potential water sources absorbed by N. tangutorum showed remarkable differences over time ( Figure 6 ). Overall, N. tangutorum mainly assimilated deep soil water and groundwater, which is in accordance with the inference of Dong et al. (2019)regarding the water source of N. tangutorum. Due to the shallow groundwater level in May, N. tangutorum absorbed groundwater extensively at different distances from the lake during this time ( Figures 6B, C ). The surface soil layer of the N. tangutorum nebkha, 20 m from the lake, was the closest to the groundwater table, and groundwater was unearthed in the 120 cm soil layer in May. Soil layers close to the groundwater table remained moist owing to the capillary rise of groundwater (Naumburg et al., 2005); therefore, a sufficient water supply was observed in this layer in May, but the source of its uptake remained groundwater-dependent ( Figure 6A ). As the groundwater table decreased, N. tangutorum alternated to drawing on deeper soil water in June to alleviate the water deficit triggered by a falling water table. Upon the depletion of deep soil water ( Figure 2 ), this layer no longer met the growth requirements of N. tangutorum. Following this, N. tangutorum increased the uptake percentage of the middle soil water in July ( Figure 6 ), which might be attributed to N. tangutorum activating fine roots in this soil layer. Subsequently, as N. tangutorum absorbed water from the soil profile, soil water availability was significantly diminished, and water uptake by N. tangutorum was hindered. This phenomenon might be primarily due to incomplete root-soil contact and increased hydraulic resistance, which constrainswater movement between the roots and soil (Wu et al., 2019). Temperature and drought stress further result in the dormancy of the surface root system or dehydration and death of fine roots, stimulating root growth in deep soils to absorb deep water sources (Dai et al., 2015). Thus, N. tangutorum may have increased the availability of deep soil water and groundwater in August ( Figure 6 ) by developing its deep tap roots (Granda et al., 2022). Continuous staged root growth is essential for improving soil WUE and overcoming water uptake restrictions caused by decreased soil water availability (Fernandez and Caldwell, 1975). The maximum growth rate of shrub roots in the arid zones is 3–15 mm·d^-1(Fernandez and Caldwell, 1975). Variations in water availability may result in variations in root hydraulic architecture to integrate water signals from heterogeneous soil environments and continuously adjust water acquisition strategies (Maurel and Nacry, 2020). The hydraulic lifting effect of N. tangutorum’s deep root system should not be ignored when adjusting water-use patterns (Zhao, 2007). Furthermore, in ecosystems where deep-rooted plants coexist with shallow-rooted plants that rely on the hydraulic lifting capacity of deep-rooted plants, shallow-rooted plants can indirectly use deep-water sources. In conclusion, N. tangutorum may respond to changes in groundwater depth by adjusting its root distribution to adapt to seasonal shifts in available water sources through rapid growth, activation, or dormancy of the root system at diverse depths ( Figure 9 ) to maximize water acquisition (Brooks et al., 2015; Brunner et al., 2015; Granda et al., 2022). This phenomenon of exploiting different water sources in different seasons is prevalent in desert plants ((Dai et al., 2015; Hao and Li, 2021; Granda et al., 2022), which enhances water stress tolerance under natural conditions and serves as an invaluable strategy for allowing desert plants to survive in arid habitats.

N. tangutorum in the Ebinur Lake Basin mainly relied on surface soil water in spring, deep soil water in summer, and an increased percentage of middle soil water in autumn (Hao and Li, 2021), which is partially dissimilar to the results of this study. This difference may be primarily relevant to the replenishment of shallow soil water by spring snowmelt in the Ebinur Lake Basin, whereas the extremely low winter precipitation in the Badain Jaran Desert did not replenish shallow soil moisture. N. tangutorum on the Loess Plateau predominantly exploits water from the 0–40 cm soil layer, principally because natural precipitation is the only water replenishment source in this habitat (Wang et al., 2013). Consequently, plants adopt various water-use patterns in different habitats to adapt to different water conditions. N. tangutorum mainly depended on groundwater for survival in this study ( Figure 6 ), and its growth required a suitable groundwater level. Numerous scholars have explored suitable water table depths for N. tangutorum between 1.6-4.7 m using different methods (Ma and Wang, 2005; Song, 2012; Zhong et al., 2002). The groundwater in the lake basin region of the Badain Jaran Desert is buried at a depth of approximately 1.5–2 m (Huang, 2018). However, N. tangutorum scrub dunes in the lake-basin region were taller, generally at 3.0–5.0 m or higher, essentially exhibiting higher height with increasing distance from the lake. This finding, coupled with groundwater level fluctuations, may have contributed to the monthly variations in the water utilization patterns of N. tangutorum.

WUE is intimately associated with water-use patterns, revealing the competitive strategies of plants under restricted water availability (Su and Shangguan, 2020). To a certain extent, δ¹³C values can reflect the degree of plant water stress and WUE (Easlon et al., 2014; Mohale et al., 2014). Numerous studies have reported a significant positive correlation between δ¹³C values and WUE (Ehleringer and Cooper, 1988; Ebdon et al., 1998). In this study, the space-time variations in the WUE (δ¹³C) of N. tangutorum were fundamentally demonstrated as a possible response to the groundwater table, showing a highly significant positive correlation with the groundwater burial depth (Dong et al., 2019). As the groundwater table decreases, decreasing soil water availability or increasing temperatures throughout the growing season can cause structural and physiological changes in plants (Tong et al., 2019), which can alter their water-use strategies and improve WUE. Examples of these changes include smaller and thicker leaves, smaller specific leaf area, and higher leaf nitrogen content, thus achieving higher photosynthesis rates while reducing the evaporative area of a single leaf (Wang et al., 2017); development of longer fine roots, increased specific root length, and decreased water consumption (Li et al., 2023); and control of stomatal openness (Graham and Zhang, 2014), among others. As drought progresses, some plants display greater stomatal conductance variation, resulting in an increased plant WUE (Lázaro-Nogal et al., 2013; Jiang et al., 2021). Some species obtain high carbon assimilation rates and maintain high WUE by reducing water loss through reduced stomatal conductance (Yang et al., 2018). Moreover, C₃ plant tissues subjected to water stress have higher WUE or δ¹³C values than non-water stressed plant tissues (Santesteban et al., 2015; Bchir et al., 2016). In addition to water conditions, light and temperature can affect plant WUE (Zhu et al., 2015). The monthly variations in the WUE (δ¹³C) of N. tangutorum in this research confirmed that the differences in light and water due to seasonal changes affected the plant WUE (Shen et al., 2017).

Plant leaf δ¹³C values provided information about photosynthesis and transpiration (Shen et al., 2017), the main processes of water consumption in plants, while δ¹⁸O_xvalues in plants xylem water in this study indicated the sources of water. There was a non-significant positive correlation between the two ( Figure 10A ), indicating that N. tangutorum exhibited some increase in photosynthesis and WUE when using shallow water sources. The fractionation patterns of δ¹⁸O values in plant water were driven by the transpiration process, which required us to obtain the δ¹⁸O values of different plant parts and fit them with leaf δ¹³C values to determine the dominant water use process. Although the δ¹⁸O fractionation patterns of water in different parts of the plant were not obtained in this study, they ware supplemented with Ψ at different times that were closely related to transpiration. The negative correlation between plant leaf δ¹³C values and Ψ ( Figures 10B–D ) suggested that efficient plant water utilization is closely related to stomatal conductance and is dominated by transpiration processes. The highly significant negative correlation between Ψ_pd and δ¹³C values also suggested that water utilization is more efficient when plants are water-deprived. Although plant transpiration water consumption accounts for the majority of plant water consumption, photosynthesis water consumption should not be neglected. In the future, we should clarify the transpiration and photosynthesis water consumption and confirm the existence of stem water storage to establish the plant water balance equation in this habitat by determining the water source, water absorption, and the water consumption of each process, among others, and to more accurately determine the plant WUE and the dominant process of water use.

The Ψof N. tangutorum decreased according to month ( Figure 7 ) in response to an intensifying water deficit due to the declining groundwater table, which is related to N. tangutorum water use patterns. As water stress intensifies and the effective soil water is depleted, plants experience increasing difficulty in obtaining water. Under such circumstances, some plants exhibit sensitive stomatal behavior to keep a certain Ψ by rapidly diminishing stomata opening, which reduces the photosynthetic rate, and it is typically referred to as isohydric behavior. In contrast, other plants leave a certain extent of stomata open to preserve a high photosynthetic rate, which is known as anisohydric behavior (Lanning et al., 2020). Martínez-Vilalta et al. (2014)developed a theoretical model by measuring Ψ _pd and Ψ _md to classify plants into four categories: strictly isohydric, partially isohydric, strictly anisohydric, and extremely anisohydric. However, in reality, plant water regulation always occurs on a continuum from conservative isohydric to adventurous anisohydric regulation. In the present study, based on the Ψ _pd and Ψ _mdvalues, N. tangutorum mainly used anisohydric regulation and gradually shifted to isohydric regulation as soil water availability decreased. This phenomenon explains the space-time changes in the WUE (δ¹³C) of N. tangutorum in this study. Long (2022)also discovered that the water regulation strategy of N. tangutorum gradually switched from anisohydric to isohydric regulation with decreasing SWC, and that its lethal mechanism changed from hydrodynamic failure into carbon starvation.

The Ψ _enof N. tangutorum increased in May compared to that Ψ _md, whereas it remained low in June–August. This finding was predominantly associated with the daily temperature variation in different months in this region, with lower evening temperatures in May and higher evening temperatures in June–August. In addition to water potential, hydraulic conductivity is a crucial indicator for analyzing water regulation strategies. To better understand the water regulation strategies of N. tangutorum under water stress conditions, future analyses should be conducted in conjunction with hydraulic characteristics such as hydraulic conductivity. Root, stem, and leaf hydraulic conductivity all affect the Ψ. If the stem exhibits high water flow resistance, the root hydraulic conductivity characteristics have a restrictive impact on Ψ (Turner, 1982). Even in cases of high stem water potential, the Ψ will probably remain low when the inner hydraulic conductivity of the leaf remains low (Nissanka et al., 1997).