Differential response of plant water consumption to rainwater uptake for dominant tree species in the semiarid Loess Plateau

Whether uptake of rainwater can increase plant water consumption in response to rainfall pulses requires investigation to evaluate the plant adaptability, especially in water limited regions where 15 rainwater is the only replenishable soil water source. In this study, the water sources from rainwater and three soil layers, predawn (Ψpd), midday (Ψm) and gradient (Ψpd−Ψm) of leaf water potential, and water consumption in response to rainfall pulses were analyzed for two dominant tree species, Hippophae rhamnoides and Populus davidiana, in pure and mixed plantations during the growing period (June– September). In pure plantations, the relative response of daily normalized sap flow (SFR) was 20 significantly affected by rainwater uptake proportion (RUP) and Ψpd−Ψm for H. rhamnoides, and was only significantly influenced by Ψpd−Ψm for P. davidiana (P < 0.05). Meanwhile, the large Ψpd−Ψm was consistent with high SFR for H. rhamnoides, and the small Ψpd−Ψm was consistent with the low SFR for P. davidiana, in response to rainfall pulses. Therefore, H. rhamnoides and P. davidiana exhibited sensitive and insensitive responses to rainfall pulses, respectively. Furthermore, mixed afforestation 25 significantly enhanced RUP, SFR, and reduced the water source proportion from the deep soil layer https://doi.org/10.5194/hess-2021-351 Preprint. Discussion started: 30 August 2021 c © Author(s) 2021. CC BY 4.0 License.

Uptaking contrasting water sources between coexisting species usually shows water source separation and can minimize water source competition (Munoz-Villers et al., 2020;Silvertown et al., 2015); however, overlapping water sources among plant species may lead to competition in arid and semiarid regions (Tang et al., 2019;Yang et al., 2020). Rainfall pulses have been observed to relieve or eliminate water competition and thus maintain or increase plant water consumption in some water limited regions 70 (Du et al., 2011;Tfwala et al., 2019). Meanwhile, plant species with strong rainwater uptake ability generally exhibit more competitiveness than coexisting weak rainwater uptake ability species (Stahl et al., 2013;West et al., 2012). However, Liu et al. (2019) attribute opposite rainwater uptake ability to the stable coexistence of species in mixed plantations in semiarid regions, where the rainfall events are variable and less rainwater is uptake by one of the coexisting plant species. In addition, coexisting 75 species may also cope with or minimize water resource competition through plant leaf water potential or root distribution adjustment (Chen et al., 2015;Silvertown et al., 2015). It is still unclear whether these adjustments could influence the rainwater uptake and water consumption for coexisting species in water limited regions.
The "Grain for Green project" has increased vegetation coverage by 25% in the Loess Plateau through afforestation activities since the 1990s, to deal with vegetation degradation and water and soil loss (Tang et al., 2019;Wu et al., 2021). Hippophae rhamnoides and Populus davidiana are typical dominant tree species, with high survival rate and drought tolerance, and occupy nearly 30% of the plantation area in this region (Liu et al., 2017;Tang et al., 2019). In addition to H. rhamnoides and P. davidiana pure plantations, mixed plantations of these two species were also widely promoted due to 85 the higher soil and water conservation capacity than pure plantations in the original afforestation stage (Tang et al., 2019;Wang et al., 2020). Rainwater has obvious seasonal variability and is the only replenished soil water source in this region because of the soil is approximately 100 m deep (Li et al., 2016;Zhang et al., 2017). The imbalance between rainwater input and plant water demand may weaken the sustainability of plantations with further plant growth (Jia et al., 2020;Wu et al., 2021). Previous 90 investigations in the region quantified the water sources from different soil depths Wu et al., 2021) and characterized the water consumption  during drought stress periods for plantation species in pure plantations. To understand the adaptation of plantation species in this study, the water consumption, water sources from rainwater and different soil layers, and plant leaf water potential for H. rhamnoides and P. davidiana in pure and mixed plantations were analyzed. The 95 specific objectives were as follow: (1) to investigate the influence of rainwater uptake and leaf water potential on water consumption after rainfall events, and (2) to assess the mixed afforestation effect on these influences.

Study site
The study was conducted in the Ansai Ecological Station in the semiarid Loess Plateau (36.55°N, 109.16°E), Northern China. The study area has a semiarid continental climate. The annual average (mean ± SD) rainfall amount and air temperature are 493.1 ± 127.9 mm and 10.7 ± 0.5 °C (1985-2017), respectively. The soil is characterized as a silt loam soil according to United States soil taxonomy (

Sap flow observation
Three standard individuals, with approximately mean height and trunk diameter, for specific species were chosen in each of the nine plots (Table S1). In each plot in the mixed plantation, three individuals of H. rhamnoides were chosen firstly, then a neighboring P. davidiana individual was selected at approximately 2 m distance from each chosen H. rhamnoides individual. The sap flow was monitored 140 by a pair of Granier-type thermal dissipation probes 10 mm in length and 2 mm in diameter in 36 selected individuals. During the plant growing season and ranging from 1 June (DOY 152) to 30 September (DOY 273) in 2018, the 30 s original and 30 min average sap flow values were monitored using a CR3000 data logger (Campbell Scientific Inc.). Waterproof silicone and aluminum foil were used to avoid the impact of the external environment on and physical damage to TDPs (Du et al., 2011).

145
The standard sap flow density (F d , ml m −2 s −1 ) was calculated as follows (Granier, 1987): where Δt and Δt max are the temperature difference of heated and unheated probes at 30 min intervals and the maximum Δt in each day, respectively. Steppe et al. (2010) suggested that F d should have a species specific calibration to validate Eq. (2).

150
Meanwhile, the possibility of underestimating the F d value with the Granier-type thermal dissipation method (Du et al., 2011) should be considered when the whole tree water consumption is calculated. From April to October 2018, at the end of each rainfall event, 19 rainwater samples were collected immediately using a rain gauge cylinder placed in the middle of the plantation plots, and stored at 4 °C .
To avoid the influence of sample collection on sap flow observation, one standard individual for the specific species nearby each sap flow monitored individual was selected for plant stem and soil water collection. In the mixed plantation, the distance was approximately 2 m between the selected H. an interpulse period longer than 7 days to eliminate the potential influence of the previous rainfall event.
At each of successive three days after every selected rainfall event, one suberized stem after removing A vacuum line (LI-2100, LICA Inc., China) was used to extract water from soil samples and plant stems. The water isotopic values of rainwater, soil samples, and plant stems were determined using a DLT-100 water isotope analyzer (LGR Inc., USA), with accuracy of ± 0.1 (δ 18 O) and ± 0.3 ‰ (δD).
The potential influence of organic matter on water isotopic values produced during water extraction 180 from stems was eliminated using the method of Yang et al. (2015). proportion of rainwater in plant stem as follows (Cheng et al., 2006): Equations (4) and (5) In addition, on the first day after rainfall, the relative water uptake proportions from different soil depths were calculated using the MixSIR program ( Moore and Semmens, 2008). The model input 205 parameters were the average δ 18 O and δD values in plant stem water, soil water at seven depths in each plot, and rainfall water. The SD for δ 18 O and δD at each soil depth was used to accommodate the uncertainties of these values, and no fractionation was considered during water source uptake by plant roots. In addition, the calculated water uptake proportions from seven soil depths were combined into three soil layers (shallow, middle, and deep) to facilitate water source comparisons, for soil depths of 0-210 30, 30-100, and 100-200 cm, respectively.
In this study, on the first day after rainfall, the water uptake proportions from rainwater and soil layers were calculated separately. The sum of RUP and relative water uptake proportions from three soil layers were larger than 100%. Thus, no significant difference was determined between RUP and water sources from different soil layers in the following analysis.

Leaf water potential measurement
On the same day as plant stem and soil sample collections, the Ψ pd and Ψ m were measured by a PMS1515D analyzer (PMS Instrument, Corvallis Inc., OR, USA) at 4:30-5:30 (predawn) and 11:20-12:40 (midday), respectively. One leaf was selected for each sap flow monitored individual, and the 220 average value for each species in each plot was used for further analysis. The diurnal variation in leaf water potential (Ψ pd -Ψ m ) was used to illustrate the leaf water potential gradient.

Plant fine root investigation
In August 2018, six soil cores were dug around each selected standard individual for plant stem and soil

Statistical analysis
In the present study, the first day after rainfall was the maximum normalized F d within 3 days for H. rhamnoides and P. davidiana in both plantation types, except after 24 and 35.2 mm for P. davidiana in 235 pure plantation. The maximum normalized F d for P. davidiana in pure plantation was observed on the second day after these two rainfall events. However, for P. davidiana in pure plantation, there was no significant difference (P > 0.05) in diurnal sap flow between the first and second day after each of these two rainfall events based on independent-sample t-test ( Fig S1). Therefore, the normalized F d on the first day after each selected rainfall amount was used in Eq. (7) to calculate the relative response of 240 daily normalized F d (SF R , %) to rainfall pulses: where X after and X before are the normalized F d on the first day after and on the day before the rainfall event, respectively.
Meanwhile, none of Ψ pd , Ψ m nor Ψ pd -Ψ m showed significant differences between the first and second 245 day after each rainfall events (P > 0.05) for these two species in both plantation types (Table S2). The Ψ pd , Ψ m , and Ψ pd -Ψ m on the first day after each rainfall event were used in the following analysis to illustrate the influence of leaf water potential on SF R in response to rainfall pulses.
A repeated ANOVA (ANOVAR) was used to analyze the differences in water consumption, water sources, and plant physiological parameters between these species in pure and mixed plantations, 250 respectively. This analysis was conducted with SF R , RUP, relative water uptake proportions from three soil depths, and Ψ pd -Ψ m as response variables, and "species" and "rainfall" as between-subject and within-subject factors. The same analysis was used to detect mixed afforestation effect on response variables for each plant species, with "plantation type" and "rainfall" as the between-subject factor. Furthermore, significant differences in fine root proportion for each soil layer (shallow, middle, and 255 deep) for each species between pure and mixed plantations were detected through independent-sample t-test. All of these analyses were calculated with SPSS 18 (IBM Inc., New York, US), after data normal distribution and homogeneity of variance analysis were tested.

Variation in environmental parameters and plant fine root vertical distribution
The rainfall amount during the study period (265.7 mm, DOY 152-273) was 11.8% lower than the average value during 2008-2017. Rainfall varied seasonally with 36 consecutive days had no rainfall event  and 5 days had successive rainfall events (DOY 237-241) (Fig 1). The ET 0 (554.7 mm) was approximately twice the rainfall amount during the study period, with the higher and 265 lower values during the low and high rainfall event periods, respectively (Fig 1). The SW increased and https://doi.org/10.5194/hess-2021-351 Preprint. Discussion started: 30 August 2021 c Author(s) 2021. CC BY 4.0 License.
subsequently decreased by different degrees following rainfall events, with shallow soil layer (0-30 cm) exhibited higher variation than the corresponding value below 30 cm in the three plantations (Fig 1).
The coefficients of variation (CVs) in the shallow soil layer were 19.22%, 18.56%, and 16.61% in H. rhamnoides and P. davidiana pure plantations and the mixed plantation, respectively. The SW for

290
Daily normalized F d for H. rhamnoides and P. davidiana fluctuated with rainfall events in pure and mixed plantations (Fig 2). The variation of normalized F d for H. rhamnoides and P. davidiana in mixed plantation was higher than the specific species in pure plantations, with corresponding CVs of 30.99% and 34.88% in the mixed plantation, and 24. 64% and 27.44% in pure plantations (Fig 2). The relative response of water consumption to rainfall pulses was significantly influenced by both rainfall amount 295 and plant species (P < 0.001) (Fig 2, Table 1). Following large rainfall amounts (≥15.4 mm), the diurnal variation of sap flow was significantly higher than the value before rainfall (P < 0.05) for H. rhamnoides in pure plantation and for P. davidiana in both plantation types (Figs S3 and S4). The lowest rainfall amount (7.9 mm) that significantly increased the diurnal variation of sap flow was observed for H. rhamnoides in the mixed plantation ( Fig S3). Furthermore, in response to rainfall pulses, the SF R for H. 300 rhamnoides in pure (range 6.69 ± 1.22% to 106.34 ± 4.7%) and mixed (range 2.23 ± 0.54% to 190.89 ± 15.49%) plantations was significantly higher (P < 0.001) than corresponding values for P. davidiana: ranges 4.24 ± 0.52% to 60.28 ± 5.72% and 3.14 ± 0.53% to 83.04 ± 14.23% (Table 1). Mixed afforestation significantly enhanced SF R for both species (P < 0.001) (

Variations in plant water sources
The soil water δ 18 O and δD for pure H. rhamnoides, pure P. davidiana, and mixed plantations showed large vertical variation following small rainfall events (≤ 7.9 mm), and exhibited relatively small vertical variations following large rainfall events (≥ 15.4 mm) (Fig S5). Generally, the isotopic values of soil water depleted from shallow to deep soil layers, and water isotopic values in shallow and middle soil layer were close to rainfall water in the three plantations following large rainfall events.
Although no significant difference in RUP was observed between H. rhamnoides (14.2 ± 7.81%) and P. davidiana (12.43 ± 7.33%) in pure plantations (Fig 3, Table 2), the RUP was significantly higher for 325 H. rhamnoides (19.17 ± 8.6%) than P. davidiana (14.59 ± 5.86%) in the mixed plantation (P < 0.05) ( Table 2). In addition, H. rhamnoides mainly uptake water from the middle soil layer in pure and mixed plantations based on the MixSIR result, with corresponding average values of 36.27 ± 2.43% and 44.14 ± 3.06% (Fig 4). The water source for P. davidiana in pure and mixed plantations was mainly from the deep and middle soil layers, respectively, with corresponding average values of 41.4 ± 15. 18% and 330 40.17 ± 5.9%. In pure plantation, the water source from shallow and middle soil layers for H. rhamnoides was significantly higher than P. davidiana; however, the water source from the deep soil layer was significantly lower for the former species (P < 0.05) ( Table 3). No significant differences in water sources from each soil layer were observed between these species in the mixed plantation (Table   3). In addition, mixed afforestation significantly enhanced RUP and decreased the deep soil water 335 uptake proportion for H. rhamnoides and P. davidiana (P < 0.05) (Tables 2 and 3

Variations in plant leaf water potential
In response to rainfall pulses, H. rhamnoides exhibited higher CV for Ψ pd , Ψ m , and Ψ pd −Ψ m than corresponding value for P. davidiana in both plantation types, except that H. rhamnoides exhibited lower CVs for Ψ pd than P. davidiana (12. 99% and 18.33%, respectively) in the mixed plantation (Fig 5).
Compared with P. davidiana, H. rhamnoides exhibited significantly positive Ψ pd in pure plantation, negative Ψ m in the mixed plantation, and larger Ψ pd −Ψ m in both plantation types (P < 0.05) (Table 4).
Meanwhile, mixed afforestation significantly reduced the Ψ m and increased the Ψ pd for H. rhamnoides 365 and P. davidiana (P < 0.05), respectively, and significantly increased Ψ pd −Ψ m for both species (Table 4).  Table 4. Repeated ANOVA (ANOVAR) parameters for predawn (Ψ pd ), midday leaf water potential (Ψ m ), and leaf water potential gradient (Ψ pd −Ψ m ) for H. rhamnoides and P. davidiana (n = 30 indicate the mixed afforestation effect on leaf water potential for these species.

Influence of water sources and Ψ pd −Ψ m on plant water consumption
The SF R significantly increased with increasing RUP and decreasing Ψ pd −Ψ m for H. rhamnoides (P < 0.01) in both plantation types (Fig 6). Meanwhile, SF R significantly increased with decreasing Ψ pd −Ψ m 380 for P. davidiana in both plantation types (P < 0.05). However, a significant relationship between SF R and RUP was observed for P. davidiana in the mixed (P < 0.05) but not in pure plantations (Fig 6).
Furthermore, SF R significantly increased with decreasing water uptake proportion from the deep soil layer for H. rhamnoides in both plantation types and P. davidiana in mixed plantation (P < 0.05) (Table   S3). No significant relationship was observed between SF R and water uptake proportion from shallow or 385 middle soil layers for both species in both plantation types.

Rainwater uptake enhances water consumption for H. rhamnoides but not P. davidiana in pure plantations
Rainwater is the only replenished soil water source in the studied region, because plants cannot uptake 395 ground water of approximately 100 m depth below the surface (Wu et al., 2021). Small rainfall events generally only wet the soil surface and may evaporate before plant root uptake ( Zhao and Liu, 2010).
However, large rainfall events are most likely recharge soil moisture and enhance the metabolic activity of plant fine roots (Hudson et al., 2018), thus enhancing plant water uptake. Similar to Salix psammophila and Caragana korshinskii in the studied region (Zhao et al., 2021), both H. rhamnoides 400 and P. davidiana exhibited plasticity in water sources in pure plantations (Fig 4), with H. rhamnoides exhibiting the greater plasticity. In pure plantations, the obviously lower SWC at all soil depths (Fig 1) and large water uptake proportion from the deep soil layer (Fig 4) after 3.4 mm of rainfall for these two species, suggested that this rainfall amount did not relieve the drought caused by 36 days (DOY 157-https://doi.org/10.5194/hess-2021-351 Preprint. Discussion started: 30 August 2021 c Author(s) 2021. CC BY 4.0 License. 192) of no rainfall. The RUP for H. rhamnoides but not P. davidiana significantly increased following 405 an increase in rainfall amount (P < 0.05) (Fig S6), indicating that water uptake was more sensitive to rainfall for H. rhamnoides. This may be mainly due to the greater proportions of fine root surface area distributed in the shallow soil layer for H. rhamnoides (40.85 ± 3.14%) compared to P. davidiana (21.94 ± 2.3%) (Fig S2).
Rainwater uptake does not permit water consumption increase after rainfall pulses especially in 410 semiarid and arid environments (Dai et al., 2020;Grossiord et al., 2017;West et al., 2007), and the influence of water potential gradient (Ψ pd −Ψ m ) on plant water consumption should also be considered (Hudson et al., 2018;Kumagai and Porporato, 2012). For example, although Juniperus osteosperma, a deep rooted plant species, could uptake rainwater after large events in the west of the United States, the sap flux did not increase with increasing rainfall amount (West et al., 2007). The synchronization 415 between rainwater uptake and water consumption for J. osteosperma was mainly attributed to the uptake of rainwater by plant being unable to reverse the cavitation in its roots and stems (Grossiord et al., 2017;West et al., 2007). Our previous investigations in the studied region indicated that P.
davidiana is relatively more vulnerable to cavitation than H. rhamnoides, with water potential at 50% loss of conductivity of −1.15 MPa (Zhang et al., 2013) and−1.49 MPa (Dang et al., 2017), respectively, 420 based on stem vulnerability curves. Being less vulnerable to stem cavitation allowed H. rhamnoides to experience a significantly lower Ψ m and larger Ψ pd −Ψ m compared with P. davidiana in response to soil water conditions after rainfall pulses. The large Ψ pd −Ψ m for H. rhamnoides was consistent with the high SF R and CVs of normalized sap flow, indicating that this species exhibited a rainfall sensitive mechanism. The relative constant Ψ pd −Ψ m for P. davidiana was consistent with the relatively small SF R 425 and CVs of normalized sap flow, indicating that this species exhibited a rainfall insensitive mechanism.
Furthermore, after rainfall events, the SF R for H. rhamnoides but not for P. davidiana significantly increased following rainfall amount increases (P < 0.05) (Fig S6), also indicating that water consumption was more sensitive to rainfall for H. rhamnoides.
The SF R was significantly influenced by RUP and Ψ pd  and 7). However, the SF R was only significantly influenced by Ψ pd −Ψ m for P. davidiana (Fig 7), suggesting that its water use was mainly constrained by plant physiological characteristics. The ET 0 represents the atmospheric evaporative demand, and has been observed to influence plant water consumption in water limited (Li et al., 2021) and non-water limited regions (Iida et al., 2016). 435 However, in the present study, neither ET 0 after rainfall nor relative response of ET 0 significantly influenced SF R for either species in pure plantations ( Table S4). The influence of plant physiological characteristics (i.e. Ψ pd −Ψ m ) on SF R for both species, may partially contribute to the lack of atmosphere evaporative demand effect on plant water consumption in the studied region, although these species exhibited different rainfall pulse sensitivity. the bottom half of the schematic, with "increase", "decrease" or "enlarge" indicating a significant difference (P < 0.05) for a species between pure and mixed plantations. Mixed afforestation significantly enhanced RUP and plant water consumption, decreased Ψ m , and enlarged Ψ pd −Ψ m for H. 450 rhamnoides, and also significantly enhanced the RUP and water consumption, increased Ψ pd , and enlarged Ψ pd −Ψ m for P. davidiana.

Rainwater uptake enhances water consumption for coexisting species in mixed plantation
Spatial water resource partitioning is considered one of the essential plant strategies to maintain 455 coexistence in mixed plantations, especially in semiarid and arid regions (Munoz-Villers et al., 2020;Silvertown et al., 2015;Yang et al., 2020). However, water source competition has widely been observed among coexisting plant species according to the literature surveys by Silvertown et al. (2015) and Tang et al. (2018), regardless of annual average rainfall amount. In the present study, the non-significant differences in xylem δ 18 O and δD (P > 0.05) and plant water sources for the three soil 460 layers (Table 3, Fig 4) indicated water competition between these species in the mixed plantation, although the RUP was significantly higher for H. rhamnoides (Table 2).
Generally, two types of adaptation can be adopted by plants to cope with resource competition: increased competition ability or minimized competition interactions (West et al., 2007). Consistent with the first adaptation type, mixed afforestation enhanced the RUP for H. rhamnoides and P. davidiana 465 (Figs 3 and 7, Table 2). Although mixed afforestation did not significantly alter the Ψ pd and Ψ m for H. rhamnoides and P. davidiana, respectively, significantly negative Ψ m and positive Ψ pd were observed for corresponding species (P < 0.01) ( Table 4). Mixed afforestation significant increased Ψ pd for P. davidiana, possibly due to the advantage of access to soil moisture recharged by rainwater through an increased root surface area in the shallow soil layer for this species in the mixed plantation ( Fig S2). 470 Thus, plant physiological (Ψ m ) and root morphological adjustments were adopted by H. rhamnoides and P. davidiana in the mixed plantation, respectively, to significantly enlarge Ψ pd −Ψ m and increase RUP (Fig 7). Similar to the result in pure plantations, no significant relationship between SF R and ET 0 after rainfall and relative response of ET 0 was observed for these species in the mixed plantation (Table S4). This result also confirmed the influence of physiological or morphological factors on water 475 consumption for these species in the mixed plantation in response to rainfall pulses.
Furthermore, consistent with the second adaptation type, mixed afforestation significantly decreased the water uptake proportion from the deep soil layer for these species (Table 3). The increasing rainfall amount significantly decreased water source proportion from deep soil layer (P<0.05) for H. rhamnoides and P. davidiana in the mixed plantation (Table S3), with the corresponding values 480 decreasing from 43.13 ± 13.74% and 47.07 ± 5.39% (both after 3.4 mm), respectively, to 21.54 ± 8.9% (after 35.2 mm) and 28.66 ± 12.26% (after 24 mm) (Fig 4). Thus, both increased rainwater uptake and decreased water source competition from the deep soil layer were adopted by these species in the mixed plantation to minimize water sources competition under water limited conditions.

Implications for plantation species and type selection based on rainwater uptake and consumption
Rainwater uptake by plant and water consumption response to rainfall pulses may influence plant physiological process and the water cycle (Meier et al., 2018;Zhao et al., 2021). In pure plantations, H. rhamnoides rather than P. davidiana showed rainwater uptake advantage due to the large Ψ pd −Ψ m for 490 the former species, although both species exhibited plasticity in water sources. The excessive water uptake from the deep soil may desiccate deep soil (Wu et al., 2021), weakening plant resilience to drought stress and thus plant community sustainability in this Loess Plateau region (Song et al., 2018;Zhao et al., 2021). Whether rainwater uptake can reduce plant water uptake from deep soil layers is essential for plantation adaptation (West et al., 2012;Wu et al., 2021). In the present study, the 495 proportion of water sources from deep soil layers was significantly decreased with increased rainfall amount for these species in both pure and mixed plantations (P < 0.05), except for P. davidiana in pure plantation. Physiological (e.g., Ψ m ) and morphological (fine root distribution) adjustments were observed for H. rhamnoides and P. davidiana in the mixed plantation, respectively, to enlarge Ψ pd −Ψ m and enhance the rainwater uptake and water consumption (Tables 1 and 2; Fig 7). Mixed afforestation 500 also significantly decreased the deep soil water uptake proportion for both species (Table 3).  (Table   S5). Thus, rainfall pulse sensitive species in pure plantation, and plant species in mixed plantation that can adopt physiological or morphological adjustment to enhance rainwater uptake and reduce excessive 505 water uptake from deep soil layers, should be more considered for use in the studied region.

Conclusions
The influence of water sources and Ψ pd −Ψ m on water consumption in response to rainfall pulse was determined for H. rhamnoides and P. davidiana in the semiarid Loess Plateau region. In pure plantations, 510 the SF R was significantly influenced by RUP and Ψ pd −Ψ m for H. rhamnoides, but the SF R was only significantly influenced by Ψ pd −Ψ m for P. davidiana. Meanwhile, the lower value Ψ pd −Ψ m was consistent with the high SF R for H. rhamnoides, and the higher value Ψ pd −Ψ m was consistent with the low SF R for P. davidiana, in response to rainfall pulses. Thus, H. rhamnoides and P. davidiana exhibited sensitive and insensitive response to rainfall pulses, respectively. Furthermore, mixed afforestation 515 enhanced the rainwater uptake and water consumption for both species. Significantly lower plant Ψ m and increased fine root surface area were adopted by H. rhamnoides and P. davidiana in the mixed plantation, respectively, to enlarge Ψ pd −Ψ m and enhance rainwater uptake and decrease water source competition from the deep soil layer. The SF R was significantly influenced by RUP and Ψ pd −Ψ m for both species in the mixed plantation, and rainwater uptake enhanced plant water consumption in the 520 mixed plantation regardless of species sensitivity to rainfall pulses.

Data availability
The data that support the findings of this study are available from the corresponding author upon request.

Author contribution
YKT designed the study, performed the statistical analyses and wrote the original manuscript draft.

Declaration of Competing Interest
The authors declare that they have no conflict of interest.