Plant water resource partitioning and xylem-to-leaf deuterium enrichment in Lanzhou, northwest China

Plants are an important component of the regional water cycles, controlling 50–90% of the evapotranspiration of an ecosystem (Yaseef et al. 2010; Coenders-Gerrits et al. 2014). In arid and semi-arid regions, water is the primary limiting factor affecting the establishment and survival of plants, and interspecific (below ground) competition for water defines plant communities (Schenk & Jackson 2002). Plants must cope with water shortages during their life cycles (Palacio et al. 2017). There is evidence that different species can coexist by using different ranges and proportions of resources (Filella & Peñuelas 2003; Wang et al. 2019), such as the use of water-source segregation between growth forms in several environments (Barbeta & Peñuelas 2017). It has been proposed that trees strongly compete with shrubs for the same water resources (Schnitzer & Bongers 2011). Plant water-use patterns reflect the complex interactions between different functional types and environment conditions. Therefore, knowledge of the processes underlying plant water resource partitioning is critical for understanding soil–plant interactions and evaluating plant adaptability in water-limited ecosystems.

The analysis of stable hydrogen and oxygen isotope ratios (δD and δ¹⁸O) is an effective and powerful approach for investigating plant water resource partitioning (Dawson & Ehleringer 1991; Evaristo & McDonnell 2017). Water uptake from the roots and its transportation through the plant stem is generally considered to occur without fractionation (Stratton et al. 2000), except for some mangrove and halophytic species (Lin & Sternber 1993). In so doing, the isotopic composition of xylem water of plants represents the composition of water sources (Asbjornsen et al. 2007). More field-based plant water studies have shown that plants rarely access rainfall directly and use mostly resident, stored soil pore water (Brooks et al. 2010; Goldsmith et al. 2011; Hervé-Fernández et al. 2016). Soil water isotopic composition is spatially heterogeneous and highly variable (Goldsmith et al. 2018). Thus, explicitly accessing temporal and spatial variations in the stable isotopes of water and plant water uptake is important to improve understanding of the mechanisms behind plant water resource partitioning. Recent investigations have also shown that enhanced temporal and spatial sampling resolution of stable isotopes of δD and δ¹⁸O is key to improved understanding of plant water-use patterns (Goldsmith et al. 2018; Allen et al. 2019; Sprenger et al. 2019).

Due to transpiration, the isotopic composition of leaf water differs markedly from that of xylem water because the lighter isotopologues of water diffuse more readily than the heavier ones (Barbour et al. 2017). Unraveling the complete pathway of water entering a vegetated catchment as precipitation, traveling through soil, and eventually leaving the system either as stream water or as water vapor from plants and bare surfaces is crucial to understanding the regional water cycles (Sohel et al. 2019). Therefore, the transpiration processes from xylem-to-leaf water with isotopic enrichment is one of the important components. The drier and warmer the ambient atmosphere, and the windier local weather conditions, the larger the rate of transpiration will be and thus the required rate of water uptake (Jackson et al. 1995). As regulators of the ecophysiological performance of individual plants, the relative humidity and air temperature are critical environmental parameters for water and carbon cycling at the ecosystem scale (Kahmen et al. 2011). Some studies also show that air temperature, relative humidity- and isotopic composition of water vapor surrounding the plant mainly determine the isotopic enrichment from the xylem to the leaf water (Kahmen et al. 2013). Therefore, it is necessary to understand how the physiological, morphological- and environmental factors affect leaf water enrichment in heavy isotopes during plant transpiration, which is important for many physiological, ecological, and paleoclimate researches (Holloway-Phillips et al. 2016). However, to our knowledge, no studies have been conducted to explore the xylem-to-leaf water deuterium enrichment in urban ecosystems in arid and semi-arid areas of China.

Lanzhou is the capital of Gansu Province and is located in the semi-arid northwest China, which has a typical semi-arid temperate continental climate. The city also lies at the transitional zone between monsoon climate zone and non-monsoon climate zone, and the Yellow River, which passes through the inner city from west to east, has formed a special ecological environment. The city is representative of arid and semi-arid urban ecosystems, with sparse vegetation and limited precipitation. The vegetation of this area has dramatically changed in recent years (Chen et al. 2019). Despite this, little is known about the temporal and spatial variations in water resource partitioning of the main plant species that inhabit this region, and plant adaptability and hydrogen isotopic fractionation from xylem to leaf water in this region remain poorly understood. In this study, we therefore measured the δD and δ¹⁸O of precipitation, river water, soil water, and plant xylem and leaf water in two plant members of each of the eight plant species at four sampling sites with different elevations around Lanzhou. Our primary research aims were to determine the temporal and spatial variations of plant water resource partitioning in Lanzhou and to explore the influence that these environmental drivers have on the leaf-water isotopic fractionation associated with evaporation at each of Lanzhou's sampling sites. This study can provide a basis for the study in water resource partitioning of plants in arid and semi-arid urban ecosystems.

This study was conducted in Lanzhou, which is located in the upper reaches of the Yellow River and the geometric center of China's land territory. The city is encircled by mountains, with a mean elevation of 1,500 m. According to the long-term climatology from 1981 to 2010, the mean annual air temperature is 10.4 °C, ranging from −4.5 °C in January to 23.1 °C in July. The annual precipitation amount is less than 400 mm because the area is located deep in the mainland, far from the surrounding oceans, and precipitation occurs mostly during summer (Chen et al. 2015).

Lanzhou is a typical valley city with obvious elevation gradient. We assume that elevation has a significant influence on the plant survival and growth; therefore, according to the principle of elevation gradient, the four sites, Jiuzhoutai (N1: 103.78°E, 36.09°N, 2,054 m), Beishan (N2: 103.73°E, 36.11°N, 1,667 m), the Northwest Normal University (N3: 103.73°E, 36.10°N, 1,553 m), and Wetland park (N4: 103.72°E, 36.08°N, 1,509 m) were located north of the Yellow River and selected as the sampling sites (Figure 1). The changes in elevation show that: N1 > N2 > N3 > N4. The site N4 is very close to the Yellow River. Eight typical plant species with varying growth form (tree, shrub, or herb) and leaf phenology (evergreen or deciduous) were collected in the four sampling sites (Table 1). Shrubs were defined as woody plants with multiple stems, while trees had one erect perennial stem.

From April to October 2016 (growing season), river water, soil, and plant xylem and leaf samples were collected once a month, precipitation samples were collected on a per event basis. Sampling was performed as follows. (1) plant xylem and leaf samples: plant samples were collected between 08:00 and 11:00 a.m. to minimize the influence of external factors, such as illumination, on the results of isotopic analysis. Two plants of the same species with good growth and high consistency were selected. From each plant, five to seven bolted branches were cut, with diameters of about 0.3 cm and lengths of about 5 cm. After peeling off the epidermis quickly, the bolted branches were put into 10-mL glass vials with threaded opening and sealed with Parafilm sealing film in order to be transported to the laboratory where the samples continued to be kept in the cooler until the water was quantitatively extracted (Wu et al. 2016). If twigs were sampled, the leaves on those twigs were also sampled. (2) Soil samples were collected simultaneously with plant tissues, in the proximity of the sampled plants, using a hand auger. Soil samples were taken from the surface and up to a depth of 100 cm, at 10-cm intervals, which were divided into two: one subsample was immediately packed into 10-mL glass vials with threads sealed with a screw-lid and Parafilm wrap and placed in a cooler where it was refrigerated until water extraction for isotopic analysis; the other subsample was sealed in an aluminum box and placed into a cooler for measuring soil water content (SWC, %). (3) Event-based precipitation samples were collected using a 300-mL rain gauge near the site N3 where neither trees nor buildings were present. The samples were immediately bottled in plastic bottles, wrapped in Parafilm and refrigerated at 2 °C until their analysis. A total of 35 precipitation samples were collected during the growth season. In addition, the meteorological parameters (precipitation amount and air temperature) were recorded by a weather station near site N3. (4) Monthly river water samples were collected from the Yellow River, near site N4.

In this study, seasonal characterization was performed as follows: samples collected from June to September were attributed to the wet season and those collected from October to May to the dry season. Since the SWC and isotopic composition of soil water varied significantly along the entire soil profile, three potential soil water source layers (0–30 cm, 30–60 cm, and 60–100 cm) were defined for the needs of this analysis (Wu et al. 2016) as follows:

• (1)

  Shallow soil layer (0–30 cm): SWC and soil water isotopic composition varied significantly and were sensitive to precipitation pulse input and evaporation depending on the season and depth.

• (2)

  Middle soil layer (30–60 cm): lower soil water isotopic composition and milder monthly changes than in the shallow soil layer.

• (3)

  Deep soil layer (60–100 cm): relatively stable variations in isotopic composition of soil water and SWC through the entire soil profile.

The hydrogen isotopic signatures of xylem samples follow an evaporation line (LEL), and slope_LEL represents the slope of LEL. The slope_LMWL and intercept_LMWL represent the slope and intercept of LMWL, respectively.

The higher this ED value, the further away from the LMWL and the more evaporated the water will be, i.e. the concentration of heavy isotopes will be higher.

Slopes and intercepts of the LMWL and LELs were calculated using linear regression. To explore the different sources of plant water use, a discussion about the different slopes and intercepts is provided in this paper.

During this study (2016), the monthly average air temperature ranged from −12.9 °C to 29.8 °C (annual mean value: 11.4 ± 10 °C), where the air temperature maximum (24.2 °C) occurred in July and the minimum (−5.4 °C) in January (Figure 2(a)). The total amount of local precipitation was 372 mm, of which 66.5% (231.3 mm) occurred in the wet season and 33.5% (116.4 mm) in the dry season. The event-based isotopic composition of precipitation (δD_prec and δ¹⁸O_prec) varied significantly within the plant growing season (Figure 2(b)). The δD_prec and δ¹⁸O_prec ranged from −90.52‰ to 34.82‰ and −12.48‰ to 4.44‰, respectively, with monthly amount-weighted values of −39.3‰ and −6.31‰, respectively. The isotopic values of precipitation were most enriched during the wet season of June, with values of 10.09‰ for δD_prec and 0.37‰ for δ¹⁸O_prec, and most depleted during the dry season of April, with values of −55.46‰ for δD_prec and −7.43‰ for δ¹⁸O_prec, respectively. The LMWL (δD = 7.08 δ¹⁸O +5.12, n= 35) in Figure 3 had a slightly lower slope and about half the intercept value compared with the global meteoric water line (GMWL: δD = 8.1 δ¹⁸O + 10.3 (Rozanski et al. 1993)), indicating that an apparent evaporation enrichment occurred in our study area.

The isotopic compositions of river water revealed very little seasonal variations in δD (δD_river) and δ¹⁸O (δ¹⁸O_river), mainly concentrated near the LMWL (Figure 3). The mean δD_river (ranging from −71.90‰ to −60.31‰) and δ¹⁸O_river (ranging from −10.52‰ to −9.20‰) value were −66.83 ± 4.18‰ and −9.81 ± 0.71‰ during seven consecutive months, respectively (Table 2). The soil water isotopic values were generally distributed along and below the right of the LMWL, which suggested that soil water had undergone evaporation enrichment, especially in shallow soil layer.

At the four sampling sites, the dual isotopic signature of xylem water, leaf water, soil water, and river water were plotted along the LMWL (Figure 3). The δD of xylem water (δD_xylem) in a total of 117 analyzed samples (no grasses) had an overall mean value of −65.67 ± 11.12‰ (ranging from −88.67‰ to −35.29‰) across all plants during sampling period, and δ¹⁸O of xylem water (δ¹⁸O_xylem) ranged from −10.66‰ to 5.68‰ with an average of −6.90 ± 2.35‰. The variations of δD_xylem showed that N2 (−55.23 ± 8.14‰) > N1 (−65.56 ± 14.22‰) > N4 (−68.31 ± 8.28‰) > N3 (−72.90 ± 4.96‰). At sites N1 and N4, the δD_xylem and δ¹⁸O_xylem values of trees were higher than the shrubs, but no obvious differences of isotopic values were observed between the trees and shrubs at sites N2 and N3. Across all sampled plants, leaf phenology (deciduous or evergreen) had significant influence on the δD_xylem value. In terms of different seasons (Table 2), the δ¹⁸O_xylem and δD_xylem of trees or shrubs showed no significant seasonality across the four sampling sites.

An evaporation line (LEL: δD = 5.31 × δ¹⁸O − 28.11; n= 117) was obtained from all xylem samples, with a slope and intercept lower than those of LMWL. LELs across four sampling sites were used to calculate δD_LMWL-int. It was found that the δD_LMWL-int (mean value: −98.23 ± 17‰) ranged from −146.22‰ to −57.91‰ for all plant samples. Comparing the results between the four sites (Figure 4), the δD_LMWL-int decreased as follows: N2 (mean value: −88.26 ± 19‰) > N1 (mean value: −96.15 ± 20‰) > N4 (mean value: −102.00 ± 16‰) > N3 (mean value: −105.47 ± 11‰), indicating that the higher the elevation, the greater the δD_LMWL-int value. The plants at the four sites showed negative δD_LMWL-int values in the dry season, which was consistent with the lower isotope values of precipitation during the dry season. The lowest δD_LMWL-int value at all sites was observed in April and the highest in September.

In this study, the ED values across all plant samples ranged from 11‰ to 136‰ (mean value: 38.04 ± 16.15‰) (Figure 4). The plants from site N4 showed systematically higher ED values (mean value: 41.6 ± 23.84‰) than those from site N3 (mean value: 38.33 ± 9.95‰), N2 (mean value: 38.04 ± 16.15‰), and N1 (mean value: 36 ± 11.38‰). The ED values decreased in the sense N4 > N3 > N2 > N1, i.e. the higher the elevation, the lower the ED value. The ED value at site N4 was highest because plants from this site used a substantial fraction of the Yellow River water, which had relatively enriched δD_river. The ED value was maximum at all sampling sites during the dry season, especially in April, and minimum during the wet season.

A total of 201 plant samples were collected to analyze leaf water δD (δD_leaf) and δ¹⁸O (δ¹⁸O_leaf) variations during the growing season, including three species of grasses (Agropyron cristatum, Medicago, Phragmites australis). The isotopic compositions of leaf water varied obviously among plant species, sampling sites, and seasons. The δD_leaf ranged from −64.52‰ to 59.35‰ (mean value: −6.05 ± 23.26‰) across all samples (Figure 4), and δ¹⁸O_leaf ranged from −7.67‰ to 51.42‰ (mean value: 11.06 ± 10.52‰). In relation to the growth form of plants, the δD_leaf of trees (mean value: −11.61 ± 14.31‰) were less enriched than those of shrubs (mean value: −1.31 ± 29.90‰), while grasses were intermediate (mean value: −2.88 ± 25.26‰).

The deuterium enrichment factor (ɛ_l/x) of xylem-to-leaf water was calculated based on 117 pairs of δD_xylem and δD_leaf values (Figure 5). The ɛ_l/x for δ¹⁸O was not discussed here because it follows the same trend as the δD. Our results showed that the average ɛ_l/x was −0.91 ± 0.36‰ at all sampling sites and for all plant species (trees and shrubs only). The ɛ_l/x exhibited significant differences between four sites plants, with mean values: −0.89 ± 0.16‰ at N1, −1.07 ± 0.46‰ at N2, −0.76 ± 0.19‰ at N3, and −0.62 ± 0.12‰ at N4. The ɛ_l/x values were classified as follows: N4 > N3 > N1 > N2, indicating that the evaporation of leaf water decreased with elevation. Kahmen et al. (2013) noted that the enrichment of δD was stronger in arid biomes (40–100%), intermediate in temperate biomes (10–30%), and weaker in humid tropical biomes (0–20%). Some studies have also shown that air temperature, relative humidity, wind speed, and δD of atmospheric water vapor influences the leaf water deuterium enrichment (Sachse et al. 2004). In our study, the broad range of measured δD_leaf values (ranging from −44.25‰ to 59.35‰) was not surprising because several environmental variables, the monthly average air temperature, and the precipitation amount varied significantly (Figure 2) and affected greatly the variability of δD_leaf and ɛ_l/x between different months.

In arid and semi-arid regions, precipitation is the primary source of water for plants. Water use competition between plants is more significant during the dry season, and in order to avoid fierce competition for water, plants have developed different water-use strategies to cope with water shortages or water stress. In our study area, over 76% of precipitation occurred in the 2016 summer, and the δD_prec and δ¹⁸O_prec values exhibited the rainfall-amount effect (Dansgaard 1964). The mean δ¹⁸O_prec value showed significant seasonal fluctuations, with enriched values occurring in summer and autumn, and depleted values in spring. However, xylem water (both trees and shrubs) isotopic compositions across all sites were significantly depleted compared to the precipitation values in spring (Table 2). Soil water also experienced significant enrichment during this season.

Soil water is the immediate water source for plants. The isotopic compositions of soil water at four sampling sites showed a clear isotopic depletion with depth (Table 2). The shallow soil water at sites N1, N2, and N3 isotopically enriched in the wet season and depleted in the dry season, which was synchronized with the seasonal variations of the isotopic composition in precipitation. This pattern was the result of both precipitation pulse inputs and evaporation enrichment. The isotopic compositions of middle soil water presented a depletion in the wet season and an enrichment in the dry season at sites N1, N2, and N3. The deep soil water isotopically depleted in the wet season and enriched in the dry season at sites N1 and N3, while site N2 was the opposite. At site N4, the isotopic value was gradually depleted with soil depth, but the layers were close to the isotopic values of the river water in both the dry and wet seasons.

The average isotopic composition of the source of xylem water was reflected in the intersection points of individual xylem sample LEL with the LMWL. Plants that relied mostly on water from isotopically enriched precipitation during the wet season would exhibit relatively high δD_LMWL-int values (Figure 6), while the opposite was true for plants that relied on isotopically depleted precipitation in the dry season. The ED value was proportional to the relative degree of evaporation before uptake by the plants. The higher this ED, the greater the relative importance of shallow soil water (which was prone to evaporation) compared to deeper soil water. In our study area, Chen et al. (2015) found that the δ¹⁸O and δD values of precipitation enriched from May to September and depleted from October to April of the next year. As shown in Table 2, at the sites N1, N2, and N3, shallow soil water isotopically depleted in the dry season and enriched in the wet season, which was similar to the variations of precipitation isotopic compositions.

Whether in the dry or wet season, the leaf phenology of plant had significant impact on δD_LMWL-int and ED at sites N1 and N2 (Figure 6 and 7), where the ED value of evergreen plants was lower than that of the deciduous, suggesting that evergreen plants would have been adapted to water tapping from deep soil water sources (e.g. isotopic depleted water), which allows them to survive the long dry season, as observed by Jackson et al. (1995) in a tropical moist lowland forest in Panama.

At sites N1 and N2, the mean ED value of trees was lower than that of shrubs during the dry season or wet season (Figure 7), indicating that trees used more deep soil water. In contrast, at site N3, the mean ED value of trees was higher than that of shrubs, indicating that trees used more shallow soil water enriched in heavy isotopes by evaporation. It was generally accepted that the deeper root systems of trees compared to shrubs allowed them to access deeper soil water or groundwater (Evaristo & McDonnell 2017). However, Meinzer et al. (1999) found that smaller trees used deeper sources of water than larger trees, and attributed this to three possible factors. In our study, the ED values of plants (both trees and shrubs) at site N4 were strikingly higher than that of plants growing at sites N1 and N2 (Figure 7). This was because plants in this site used more river water, which had relatively depleted δD_river and δ¹⁸O_river. In general, the ED value of trees and shrubs were closer in the wet season because there was more precipitation for plants to uptake. During the dry season, the differences of ED values between trees and shrubs were relatively large, indicating differences in water source depths. This suggests that water competition among different plant species was mitigated during the dry season due to water resource partitioning.

Plants in all four sites showed similar, very high δD_LMWL-int values from June to September, indicating that plants relied mostly on enriched precipitation during this period. The δD_LMWL-int value of plants sampled in October was close to the δD_LMWL-int in September because the precipitation represented the onset of the short wet season following a dry season and the soil was expected to recharge. This is following the ‘two water world’ hypothesis of Brooks et al. (2010). From observations it was obvious that site N4 was the wettest of all four local sites, and plants remained green in the dry season because there was plenty of river water available. Plants at sites N1, N2, and N3 showed lowest δD_LMWL-int values in April, which then increased during the following months to reach their peak values in September. This indicates that the water pool for the plants was replenished stepwise by the isotopically more enriched precipitation. At site N1 and N2, plants showed strong seasonal trends in δD_LMWL-int, and the seasonal trends were more pronounced than at site N3. This is probably because site N3 was affected by irrigation water coming from the Yellow River. We did not measure the isotopic composition of irrigation water at site N3, but in this study the isotopic value of irrigation water is close to river water.

From all plants at four sampling sites, trees and shrubs exhibited different water-use patterns that might be related to the distribution of their roots and to the physiological characteristics of the same habitat (Ward et al. 2013). For example, the surface roots of Platycladus orientalis are more developed–about 50% of their root biomass is located in the surface layer (Evaristo & McDonnell 2017). The surface roots developed by P. orientalis are sensitive to precipitation responses and can be used to absorb surface water with higher water content through their developed root tip after a rain event (Liu et al. 2016). In addition to the effect of the roots on the water-use patterns of plants, the plants themselves affect the soil. The threshold value of response of SWC also affects the water use of plants (Han et al. 2009).

Leaf phenology significantly influenced the ɛ_l/x. The ɛ_l/x values of evergreen plants were generally higher than those of the deciduous plants at sites N1 and N2 (Figure 5). Evergreen plants that keep their foliage during the dry season must be protected against drought stress by a high degree of succulence or sclerophylly (thickened or hardened leaves) in order to reduce moisture loss (Chabot & Hicks 1982). Thus, the adaptive traits of the evergreens that reduce water loss and transpiration rates causes a lower xylem-to-leaf deuterium enrichment. The differences in ɛ_l/x between the four sampling sites were obvious: the ɛ_l/x values were lower at sites N1 and N2 but higher at sites N3 and N4, indicating that the location (elevation) of the plants significantly affected the ɛ_l/x. For example, the ɛ_l/x values of Caragana korshinskii (sites N1 and N2), Sophora japonica, and Rosa xanthine (sites N3 and N4) varied significantly between the different sites. However, the ɛ_l/x values of Salix babylonica and P. orientalis changed slightly, irrespective of their sampling sites. We can see that the overall difference between ɛ_l/x values among the different sampling sites was due to differences between plant assemblage or features at each site rather than to habitat-specific (environment conditions) factors. Growth forms did not significantly affect the ɛ_l/x in the study area. For example, the mean ɛ_l/x values of trees and shrubs were −0.84 ± 0.22‰ and −1.03 ± 0.50‰, respectively. The variations of season also influenced the ɛ_l/x value of the plant species (Figure 8): during April and June all species were characterized by lower ɛ_l/x values, but were higher during May and July. In August, September, and October, the difference between the ɛ_l/x values were not significant. The ɛ_l/x value reached its peak in July with higher temperatures and more precipitation, but in June with higher temperatures, the ɛ_l/x value reached a minimum. There was no obvious correlation between the ɛ_l/x value and temperature or precipitation.

In summary, our results showed that on the local scale of a single study area, with several distinct plant sampling sites, the plant species assemblage and leaf phenology were the primary factors influencing xylem-to-leaf water deuterium enrichment. Seasonality also influences the ɛ_l/x values, while local elevation and growth form had negligible effects.

In this study, we measured the δD and δ¹⁸O of precipitation, river water, soil water, and plant xylem and leaf water during the 2016 growth season at four sampling sites in Lanzhou. The conclusions are summarized as follows:

• (1)

  Plant water resource partitioning during the 2016 growing season varied both in time and space, as inferred by the relationship between the δD_xylem values and the different water sources and ED values. During the wet season, plants exhibiting relatively high δD_LMWL-int values relied mostly on water from isotopically enriched precipitation. During the dry season, the differences of ED values between trees and shrubs were relatively large, indicating differences in water source depths. The leaf phenology of the plant had a significant impact on δD_LMWL-int and ED values, whereas growth form did not significantly influence the values.

• (2)

  The mean value of xylem-to-leaf water deuterium enrichment (ɛ_l/x) was −0.91 ± 0.36‰ for all plant species, seasons, and sampling sites. Plant species, leaf phenology, and seasons were the primary factors influencing the ɛ_l/x, while growth form and elevation had negligible effects. In terms of plants leaf phenology, the deciduous species gave the highest enrichment. In terms of different seasons, plants showed the highest isotopic enrichment in July.

The authors greatly thank the colleagues in the Northwest Normal University for their help in laboratory analysis. Thanks very much for Athanassios A. Argiriou's editorial suggestions on the spelling and grammar of the manuscript.