Sap flow characteristics and responses to summer rainfall for Pinus tabulaeformis and Hippophae rhamnoides in the Loess hilly region of China

Abstract As a major driving element of the structure and function of arid and semiarid ecosystems, rainfall is the essential factor limiting plant biological processes. To clarify the characteristics of transpiration and responses to summer rainfall, sap flow density (F d) of Pinus tabulaeformis and Hippophae rhamnoides was monitored using thermal dissipation probes. In addition, midday leaf water potential (ψm) and leaf stomatal conductance (G s) were also analyzed to determine water use strategies. The results indicated that the diurnal variation in the normalized F d values exhibited a single‐peak curve for P. tabulaeformis, while H. rhamnoides showed multiple peaks. The normalized F d for P. tabulaeformis remained relatively stable regardless of rainfall events. However, there was also a significant increase in the normalized F d for H. rhamnoides in response to rainfall in June and August (p < .05), although no significant differences were observed in July. The normalized F d values for P. tabulaeformis and H. rhamnoides fitted well with the derived variable of transpiration, an integrated index calculated from the vapor pressure deficit and solar radiation (R s), using an exponential saturation function. The differences in fitting coefficients suggested that H. rhamnoides showed more sensitivity to summer rainfall (p < .01) than P. tabulaeformis. Furthermore, during the study period, P. tabulaeformis reduced G s as soil water decreased, maintaining a relatively constant ψm; while H. rhamnoides allowed large fluctuations in ψm to maintain G s. Therefore, P. tabulaeformis and H. rhamnoides should be considered isohydric and anisohydric species, respectively. And more consideration should be taken for H. rhamnoides in the afforestation activities and the local plantation management under the context of the frequently seasonal drought in the loess hilly region.


| INTRODUCTION
As a major driving element of the structure and function of arid and semiarid ecosystems, rainfall is the essential factor limiting plant biological processes (Cao, Jiang, Zhang, Zhang, & Han, 2011;Moran et al., 2009). Furthermore, global climate change appears to increase the variability of rainfall patterns in these regions. As a result, plants may endure recurring cycles of water scarcity followed by rainfall events (Jackson et al., 2001;Smith & Nowak, 1990). Understanding the mechanisms that underlie plant responses to rainfall is a key to understand how global climate change will affect arid and semiarid ecosystems (Jiang, 2001;Yuan & Deng, 2004).
Sap flow is a direct indicator of tree transpiration, and it can reflect the physiological characteristics and water use response of individual trees to environmental factors . Some studies have shown that sap flow significantly accelerates after trees absorb water provided by rainfall events (Schwinning & Sala, 2004), but this influence may be related to multiple factors, such as tree species (Cheng et al., 2006) and rainfall amount (Ivans, Hipps, Leffler, & Ivans, 2006). For example, shallow-rooted Isopogon gardneri rapidly increased transpiration up to fivefold after a 34 mm rain event in southern Australia, whereas deep-rooted Eucalyptus species were sufficiently reliant on antecedent soil water and did not respond to summer rainfall (Burgess, 2006). Zeppel, Macinnis-Ng, Ford, and Eamus (2008) found that rainfall pulses of <20 mm did not significantly increase water use of Eucalyptus callitris. In contrast, a rainfall threshold of 10 mm induced a significant response in mesquite shrubs (Fravolini et al., 2005). Pataki, Oren, and Smith (2000) indicated that broad-leaf species, such as Populus tremuloides, showed the greatest increases in sap flow density (F d ) with increasing atmospheric water demand, while the coniferous species Pinus contorta showed the lowest F d . The stomatal regulation of transpiration and the hydraulic structure of plants may account for their different behavior in response to changes in water conditions (Franks, Drake, & Froend, 2007;Hölscher, Koch, Korn, & Leuschner, 2005).
Plants fall into two categories across the continuum of stomatal regulation of water use: isohydric and anisohydric (Tardieu & Simonneau, 1998). Isohydric species, such as Pinus edulis, reduce stomatal conductance (G s ) as soil water decreases, maintaining relatively constant leaf water potential (ψ m ) (West, Hultine, Sperry, Bush, & Ehleringer, 2008;Williams & Ehleringer, 2000). Anisohydric species, such as Juniperus monosperma, allow large fluctuations in ψ m , sustaining higher G s than isohydric species (Tardieu & Simonneau, 1998;West et al., 2008). Compared with isohydric species, anisohydric species tend to occupy more drought-prone habitats to some extent, but under particularly intense droughts or prolonged drought duration, anisohydric species may experience xylem embolism or even mortality (McDowell et al., 2008). Therefore, understanding the characteristics of water use and the responses of tree species to rainfall in arid and semiarid regions will help to build reasonable vegetation restoration models as rainfall patterns change.
The Loess Plateau, located in upper-middle reaches of the Yellow River in northern China, has a number of serious soil erosion challenges, which are largely caused by intensive and unsustainable human activities (Lu & van Ittersum, 2004). Previous studies have indicated that plantations play important roles in reducing soil loss and water loss and maintaining ecological functions (Xu, Xu, Huang, Shan, & Li, 2011).
Pinus tabulaeformis and Hippophae rhamnoides are the dominant woody species on the Loess Plateau, and both species have been widely used for ecological restoration (Chen, Liu, & Hou, 2002;Wu & Yang, 1998). In recent years, however, tree species have experienced water shortages during the course of their life cycle because of low rainfall and high atmospheric evaporative demands in the region (Wang, Fu, Gao, Liu, & Zhou, 2013). Some recent studies have focused on sap flow characteristics and its influencing factors of P. tabulaeformis and H. rhamnoides (Wu, 2016;Zhang, Wei, & Chen, 2015), but little is currently known about the water use strategies and responses to rainfall for these two species. Analyzing the different water use responses of plantation species in the Loess hilly region to summer rainfall can provide a scientific basis for local plantation management in terms of tree water use during ecological restoration (Blackman & Brodribb, 2011).
In this study, we investigated the response to rainfall and the water use types of P. tabulaeformis and H. rhamnoides, by measuring F d , ψ m , and G s under field conditions. Our aims were to (1) understand the different sap flow responses of these two tree species to summer rainfall and (2) analyze their patterns of stomatal regulation to identify their water use types.

| Study site
The experiments were conducted at the Ansai Station of the Chinese Academy of Sciences, located in Shaanxi Province, China (36°51′N, 109°19′E, elevation ranges from 1068 to 1309 m a.s.l.). The climate is a typical temperate conditional monsoon; mean annual rainfall amounts are about 500 mm, with great seasonal variations (Shan & Chen, 1993). Average annual temperature is 8.8°C, with a low of −6.9°C in January and a high of 22.6°C in July. The range of average annual potential evaporation is 600-800 mm. The soil type is classified as Calcic Cambisols that developed on wind-deposited loess parent material (Wang, Fu, Qiu, & Chen, 2003 (James, Clearwater, Meinzer, & Goldstein, 2002). The temperature difference between two probes was measured every 30 s, and 30 min averages were recorded on a data logger (CR1000; Campbell Scientific Inc., Logan, UT, USA) with a multiplexer (AM16/32A; Campbell Scientific). The sensor was mounted with waterproof silicone and covered with an aluminum box cover to avoid physical damage and thermal influences from radiation (Wu, Chen, & Tang, 2015).
According to the empirical relationship between sap flow and the temperature difference between the probes established by Grainer and revalidated by other studies (Clearwater, Meinzer, Andrade, Goldstein, & Holbrook, 1999;Granier, 1987), the uncorrected sap F d was calculated as: where F d is sap flow density (ml m −2 s −1 ), ∆T is the temperature difference between the two probes, and ∆T max is the maximum value of ∆T recorded at the no-transpiration state when F d is zero.

Granier sensors
There has been concern about the validity of the empirical formula when the sensor design deviates from the original-type Granier sensors (Lu, Urban, & Zhao, 2004). In practice, there are three situations where differences in the sapwood (D sw ) and the length of the probe may lead to a deviation from the true F d value: The ideal case is when the D sw depth is equal to the length of the probe, which means that F d can be calculated using Formula 1.
If the length of the probe is less than D sw , the ideal case calculation can be used if F d is assumed to be the same in the sapwood beyond the length of the probe. This assumption may not always be true as there is often a substantial variation in F d across the whole cross section.
If the length of the probe is greater than the D sw , the probe does not provide a true F d value, and the measured F d always underestimates the true F d value. Based on a similar analysis, Clearwater et al. (1999) proposed an alternative method to calculate F d in the active sapwood. If parts of the probe are inserted into nonconducting xylem while the remainder is in sapwood with a relatively uniform F d , then it can be assumed that the measured ∆T is a weighted mean for ∆T in the sapwood (∆T sw ) and ∆T in the inactive xylem (∆T max ): where a and b are the parts of the probe in the sapwood and inactive xylem (b = 1-a), respectively. This approach assumes that the thermal properties of inactive xylem are the same as sapwood when F d = 0.
If the depth of the sapwood is known, then the corrected F d for that portion of the sapwood can be calculated by replacing ∆T in Formula 2 with ∆T sw .
There has also been a recommendation that species-specific calibration should be applied to the Granier-type probes data Smith & Allen, 1996). Furthermore, a recent report suggests that F d is significantly underestimated by heat techniques, particularly by the thermal dissipation method (Du et al., 2011;Steppe, De Pauw, Doody, & Teskey, 2010

| Environmental factors measurements
Meteorological data were obtained from the meteorological station

| Measurement of leaf water potential and stomatal conductance
The transpiration characteristics of the species were investigated by measuring the daytime temporal change patterns for canopy conductance and leaf stomatal conductance. Canopy conductance (G t ) changes were estimated using the sap flow data as presented below. Leaf G s was measured in situ for canopy sunlit leaves using a Li-6400XT portable photosynthesis system (LI-COR) under approximately natural conditions. The ψ m was measured at 11:00-13:00 using a Pressure Chamber (1515D; PMS Instruments, and Corvallis, OR, USA) for both species. In each measurement, three fully extended healthy leaves were selected from three replicate trees.

| Investigation of the root systems
Excavation and drilling methods were used to measure the root systems of the two species in order to investigate their functional types (2) ΔT = aΔT sw + bΔT max and water use strategies. The roots were sampled from 0 to 2 m in depth at 0.1 m intervals.

| Data analyses
To investigate the response of sap flow to rainfall in P. tabulaeformis and H. rhamnoides from June to August, three rainfall events were selected, and the sap flow characteristics were compared before and after rainfall to analyze changes in water use for the two species. We chose a derived variable of transpiration (VT) to represent R s and VPD, which was calculated as a simplified combination of R s and VPD (Iida, Nakatani, & Tanaka, 2006).
where VPD is the vapor pressure deficit (kPa), and R s is solar radiation Previous researchers have shown that the relationship between F d and VT can reflect the response patterns of sap flow to atmospheric factors for each species under different soil water conditions after rainfall events (Du et al., 2011). They analyzed the relationship between F d and VT using the following exponential saturation function (Ewers, Mackay, & Samanta, 2007;Kumagai, Tateishi, Shimizu, & Otsuki, 2008): where VT is the variable of transpiration, F d is the normalized sap flow density, and a and b are the fitting parameters. The differences in fitting coefficients were used to roughly divide the species into two types based on their rainfall sensitivity (Du et al., 2011). According to Oren, Zimmermann, and Terbough (1996), the hydraulic conductance from soil to atmosphere can be estimated from the slope of the F d -VT relationship. A steep slope in the relationship indicates a high conductance. This is consistent with the model coefficient (parameter b) (Du et al., 2011;Ewers et al., 2007).
The normalized F d was also used to estimate the relative canopy conductance. The total canopy conductance of individual trees is usually estimated from the ratio between the whole-tree F d and VPD. The calculation also includes a unit conservation coefficient (Kostner et al., 1992): where G t is total canopy conductance (μm/s), E is the whole-tree F d (μm/s), k is the unit conversion coefficient (kPa) calculated from the water density, gas constant for water vapor, and air temperature. As we used a normalized F d in this study, we calculated a simplified relative canopy conductance as follows: In addition, this is because VPD usually contributes more than two-thirds of the total transpiration, making it the dominant environmental variable, with the remainder from the radiation component (Green, 1993; Zhang, Simmonds, Morison, & Payne, 1997). Therefore, combined formulas of (3) and (6), the law of stomatal regulation can be roughly determined by the following formula: One-way analyses of variance (ANOVA) was used to compare significant differences in the responses of sap flow to rainfall events for these two tree species and reanalyze the significant differences in leaf water potential (ψ m ) and stomatal conductance (G s ) within species. In addition, the general linear model was used to test the significance of the fitting curve between F d and VT before and after the rainfall events. All statistical analyses were performed using SPSS 16.0 (SPSS Inc., Chicago, IL, USA); p < .05 was considered statistically significant.

| Environmental factors, root distribution, and sap flow density characteristics
The diurnal courses of the normalized F d and VT values for these two tree species from June to August are shown in Figure 1a Figure 1d-f. During the study period, there were 33 rainfall events, which produced a total rainfall of 78 mm and accounted for 35% of the annual rainfall. Rainfall had a greater effect on the soil moisture content at 10 cm depth more frequently than on the other four soil depths in the two study plots. Soil water content below 20 cm greatly decreased between June and August, which might not be significantly recharged by the rainfall events. Soil water content at 10 cm depth responded to rainfall events if the cumulative rainfall total over a 3-to 5day period exceeded 10 mm. A single rainfall event of <10 mm had little effect on soil water conditions (Figure 1d-f). Data collected along vertical profiles in the two study plots both before and after rainfall showed that rainfall had a significant effect on surface soil moisture, and the short-term changes in soil moisture below a depth of 50 cm were subject to little influence from rainfall during the study period (Figure 2).
The vertical distribution of fine roots in the root system is shown for the two tree species and is based on the surface area of the lateral roots in each 10 cm soil section (Figure 3). The root distribution with soil depth showed that P. tabulaeformis roots tended to be concentrated at around 10 and 40 cm depth, H. rhamnoides roots were concentrated around 50 cm depth. Furthermore, the surface area of the P. tabulaeformis fine roots was greater than H. rhamnoides at the same soil depth. (

| The response of sap flow density to rainfall
The diurnal courses of the normalized F d values exhibited a single-peak curve for P. tabulaeformis, while H. rhamnoides exhibited a multipeak curve before and after rainfall (Figures 4a,b). The normalized F d values were approximately zero between 0:00 and 6:00 hr, and it in-  8 mm), 15 to 20 July (11.6 mm), and 10 to 14 August (10.9 mm). Arrows represent the observed time of the leaf water potential and stomatal conductance. (e) and (f), SWC10, SWC20, SWC50, SWC100, and SWC150 are soil volumetric moisture content (m 3 /m 3 ) at 10, 20, 50, 100, and 150 cm below the soil surface for P. tabulaeformis and H. rhamnoides plot, respectively rainfall was significantly higher than that before rainfall (except in July) ( Figure 4b).
In order to further explore the differences in the pattern of daytime normalized F d values between these two tree species, we counted the frequency of F d peak times during the study periods ( Figure 5). The results showed that the F d peak time of P. tabulaeformis was mainly concentrated at around 10:00 (9:30-10:30) ( Figure 5a). H. rhamnoides had a relatively wide band of F d peak time distribution, a little more than 1/5 of peaks appeared around 9:00 and 13:00, and about 1/5 of peaks appeared around 15:30 (15:00-16:00) (Figure 5b).
There was no obvious increase in the normalized F d for P. tabulaeformis in response to the three rainfall events. The normalized F d for P. tabulaeformis reached its highest level in June before and after rainfall periods, and there was a significant difference between July and August, but no significant difference between July and August ( Figure 6a). There was a significant increase in the normalized F d for H. rhamnoides after the rainfall in June and August (except in July).
The normalized F d of H. rhamnoides also reached its highest level in June, and there was a significant difference between June and August, but there was no significant difference between June and July before rainfall events. However, after rainfall periods, there was a significant difference between June and other 2 months (July and August), but there was no significant difference between July and August ( Figure 6b).

F I G U R E 2 Soil volumetric water content along vertical profiles (a-c) in Pinus tabulaeformis and (d-f)
Hippophae rhamnoides plot, respectively, collected three days during before and after rainfall events (mean ± SE) over the experimental periods, respectively F I G U R E 3 Vertical distribution of fine roots of Pinus tabulaeformis and Hippophae rhamnoides

| Relationship between normalized F d and VT in response to rainfall
The diurnal courses of the estimated relative canopy conductance for P. tabulaeformis and H. rhamnoides had a single-peak curve before and after rainfall (Figure 7). These two tree species showed rapid increases and decreases in canopy conductance during the morning, but remained low in the afternoon. However, P. tabulaeformis reached its maximum earlier than H. rhamnoides, and this was followed by a steep decline before and after rainfall. By midday, the canopy conductance for this species had fallen to levels that were lower than those recorded in the early morning (e.g., 6:00-7:00).
H. rhamnoides showed gradual declines after they had peaked and maintained their conductance at or above the levels recorded during the early morning.
To further elucidate the response patterns of F d to atmospheric factors for these two tree species under different soil water conditions, data sets of normalized F d and VT values were analyzed before and after rainfall over 3 months. The normalized F d values for P. tabulaeformis and H. rhamnoides correlated with transpiration (VT) using an exponential saturation function (Formula 4).
Generally, F d increased in response to rising VT, while these values tended to stabilize when VT reached about 50 kPa (W/m 2 ) 1/2 ( Figure 8).  (a and b), and related significance tests. There was a relatively large difference in the variation coefficients (a and b) before and after rainfall periods for H. rhamnoides, which suggests the transpiration process is sensitive to soil water conditions (Table 2). In contrast, there were lower variation coefficient differences before and after rainfall, for P. tabulaeformis, suggesting less amplitude in the response of transpiration to the soil water changes (Figure 8). Parameter b was greater in H. rhamnoides than in P. tabulaeformis for each of the three rainfall events, suggesting a high hydraulic conductance for H. rhamnoides (Table 2).
F I G U R E 4 Changes in normalized F d of (a) Pinus tabulaeformis and (b) Hippophae rhamnoides during each before and after rainfall period over the study period. The different periods are separated by dotted lines

| Stomatal regulation patterns
During the study period, there were no significant differences in ψ m between the different measurement periods for P. tabulaeformis and H. rhamnoides. The ψ m of P. tabulaeformis was approximately −1.7 MPa, and the coefficient of variation was 8.34%. The ψ m of H. rhamnoides had a wide range of −2.06 to −2.62 MPa, and the coefficient of variation was 13.19%. The ψ m in P. tabulaeformis was significantly higher than that in H. rhamnoides (p < .05) (Figure 9a). In contrast, there was a significant difference in G s between June and the other 2 months, but there was no significant difference between the other 2 months. The average G s of H. rhamnoides (0.12 mol/m 2 / s 1 ) was 1.5 times greater than that in P. tabulaeformis (0.08 mol/m 2 / s 1 ); however, the coefficient of variation of H. rhamnoides (16.25%) was significantly lower than P. tabulaeformis (45.51%) (Figure 9b). This indicated that as soil water changes (Figure 1e

| Sap flow characteristics
Typical daytime F d patterns for tree species may indicate water use strategies (Du et al., 2011). Yin, Cheng, and Zhang (2011) showed that the daily changes of F d for P. tabulaeformis were in the form of a single-peak curve that reached a peak at 12:00-14:00. Our results also showed that F d exhibited a single-peak curve for P. tabulaeformis (Figure 4a), and the peak was mainly concentrated at 10:00 ( Figure 5a). The diurnal variation of F d for H. rhamnoides exhibited a multipeak curve (Figure 4b), with peaks concentrated at 9:00, 13:00, and 16:00, which implies active sap flow in this species until late afternoon ( Figure 5b). Yu, Yang, Zang, and Xu (2008) found similar results for H. rhamnoides in Huangfuchuan Basin, but Shen et al. (2014) showed a typical single-peak pattern for H. rhamnoides. This may depend on meteorological factors and water conditions. Furthermore, environmental factors affect broad-leaf species more than coniferous species (Cheng et al., 2015;Yang et al., 2012). The diurnal courses of the normalized F d for P. tabulaeformis remained relatively stable before and after rainfall events over the 3 months (Figure 4a). The normalized F d for H. rhamnoides (Figure 4b) after rainfall was significantly higher than before rainfall (except in July), which indicated significant differences in the responses of these two tree species to the same rainfall event. This might be related to the anatomical structure of the xylem in these two tree species. Zhai, Li, and Nie (2003) indicate that the P. tabulaeformis belongs to nonpore wood and the xylem conducting tissues that are composed of tracheids, whereas the H. rhamnoides belongs to porous wood and has an advanced form of xylem F I G U R E 5 Frequency of normalized F d peak times of (a) Pinus tabulaeformis and (b) Hippophae rhamnoides during the research periods F I G U R E 6 Normalized F d of (a) Pinus tabulaeformis and (b) Hippophae rhamnoides during each before and after rainfall period over the 3 months (mean ± SE). Capital letters indicate the significant difference between before and after rainfall at 0.05 levels, and small letters indicate the significant difference among months for before and after rainfall events at 0.05 levels conducting tissue that is composed of vessel elements (Ai, Li, Chen, & Chen, 2015). According to the Hagen-Poiseuille law, the value for sap flow is proportional to the fourth power of the hydraulic radius (Tyree & Zimmermann, 2002). Therefore, in the water transport process, the resistance in P. tabulaeformis was relatively greater than in H. rhamnoides. Furthermore, the canopy structure and the hydraulic architecture of roots may also be the reason for the sensitive response differences to rainfall events among tree species (Cheng et al., 2015;Zhu et al., 2010). Extensive investigations of the hydraulic architecture characteristics of P. tabulaeformis and H. rhamnoides are needed to clarify the difference in water use by these two tree species.

| Sap flow responses to rainfall
The function of exponential saturation has been used to elucidate the response types of F d to atmospheric water demand for tree species (Ewers et al., 2007;Kumagai et al., 2008). In a comparative study on three forest species in the semiarid Loess Plateau region of China, large differences were found in the fitting parameters during before and after rainfall periods in R. pseudoacacia, which suggests that the transpiration process is sensitive to soil water conditions.
In contrast, the other species showed less amplitude in the response of transpiration to soil drought (Du et al., 2011). Further study on the effects of stomatal conductance on tree water use will improve understanding of the eco-hydrological relationships across large spatial scales because stomatal conductance responds to various environmental factors, such as solar radiation, vapor pressure deficit, and the water status of the root (Buckley, 2005). In general, stomatal conductance increased as the atmospheric evaporative demands rose, but when the atmospheric evaporative demands increased continuously up to a certain value, the stomatal conductance decreased or the stomata closed to avoid xylem embolisms (Franks et al., 2007;Whitehead & Beadle, 2004 (Figure 7). These two tree species showed rapidly rising and declining canopy conductance in the morning, but the levels remained low in the afternoon. The results indicated that these two tree species regulate their conductance in response to high VT values (Köcher, Gebauer, Horna, & Leuschner, 2009). In addition, the normalized F d for these two tree species slowly increased with VT before rainfall, indicating that the ability of the tree to transmit water was low and transpiration was restrained, but this situation improved slightly after rainfall, which may be explained by a release of xylem hydraulic conductivity under the temporal recharge of soil water (Eberbach & Burrows, 2006;Pataki et al., 2003).
The large difference in each coefficient (parameters a or b) before and after rainfall in H. rhamnoides suggests that there was a high water demand and that soil water conditions had a major influence on the transpiration process (Table 2). In contrast, P. tabulaeformis was less sensitive to soil moisture change (Figure 5a). The fine root distribution characteristics may be a reason for the sensitive response to soil water changes (Kume, Takizawa, Yoshifuji, & Suzuki, 2007). A preliminary survey of the fine root distribution in the soil profiles at the study sites showed that P. tabulaeformis roots tended to be concentrated at around 10 and 40 cm depth, respectively; H. rhamnoides roots were concentrated around 50 cm depth. Furthermore, the surface area of P. tabulaeformis fine roots was greater than that of H. rhamnoides at the same soil depth (Figure 3). However, the relationship between the presence of roots in a particular soil layer and the magnitude of their contribution to water absorption remains unclear (Moreira, Sternberg, & Nepstad, 2000). Therefore, in the further study, the plant root F I G U R E 8 Relationship between normalized F d and variable of transpiration (VT) of (a-c) Pinus tabulaeformis and (d-f) Hippophae rhamnoides during each before and after rainfall event for June, July, and August dynamics need to be investigated, and the hydrogen and oxygen stable isotopes should be used to quantify the water sources along soil profile (West et al., 2012). Meanwhile, the differences in the biological characteristics of conifers and broad-leaf trees (e.g., the leaf anatomical structures and xylem anatomical characteristics) may also affect the sensitive responses to soil moisture recovery (Sun, Sun, Wang, & Zhou, 2005). Furthermore, the different sensitivity of sap flow responses to temporary recharge may also come from species-specific effects, such as the stomatal regulation of transpiration (Köcher et al., 2009). Several studies have found a trade-off relationship between tree water transport, leaf anatomical structure, stomatal conductance, and the distribution of root systems in ecosystems with water shortages, which would enable tree species to reach a homeostasis equilibrium and adapt to the environment under climate change (Sun et al., 2005).

| Types of water use
According to McDowell et al. (2008), the terms "isohydric" and "anisohydric" are used to divide the continuum of stomatal regulation of water status into two categories. Isohydry is generally attributed to strong stomatal control of the transpiration rate, which results in the observed similarity in ψ m . Isohydric behavior has been observed in temperature hardwoods, C 4 grasses, Australasian and neotropical trees, and other species of gymnosperms (Bonal & Guehl, 2001;Fisher, Williams, Do Vale, Da Costa, & Meir, 2006). Anisohydry typically exhibits less stomatal sensitivity to soil moisture, allowing large fluctuations in ψ m . Anisohydric behavior has also been observed in sugar maple (Acer saccharum), sunflower (Helianthus annuus), and eucalyptus (Eucalyptus gomphocephala) (Franks et al., 2007;West et al., 2008). Our results showed that during the study period, as the soil water decreased (Figure 1e,f), P. tabulaeformis reduced G s to maintain a relatively constant ψ m . In contrast, H. rhamnoides allowed large fluctuations in ψ m to maintain G s (Figure 9). Therefore, P. tabulaeformis should be considered an isohydric species, while H. rhamnoides should be considered an anisohydric species.
Anisohydric and isohydric regulation of water status may be a critical factor in the regulation of survival and mortality during drought (McDowell et al., 2008). Compared with isohydric species, anisohydric species tend to occupy more drought-prone habitats to some extent. In this study, soil water content decreased from June to August (Figure 1e,f). Meanwhile, June to August represents a period of vigorous growth for trees, resulting in great demand for soil water.
Therefore, under short-term drought conditions, P. tabulaeformis and H. rhamnoides may use a continuum of stomatal regulation of water status to maintain their growth, but under long-term and intense drought conditions, the advantages of isohydry are readily apparent. The stomatal control system prevents their xylem water potential from T A B L E 2 Regression analysis results for sap flow density and variation in transpiration (VT) for P. tabulaeformis and H. rhamnoides before and after rainfall events during June, July, and August falling below a critical threshold, which is an advantage in an environment with fluctuating evaporative demand or soil moisture (Franks et al., 2007).

| CONCLUSIONS
Due to low rainfall and high atmospheric evaporative demands, tree species in arid and semiarid regions experience water shortages during the course of their life cycle. This study discussed the effects of summer rainfall on the main tree species in the Loess hilly region.
Results showed differences in sap flow characteristics and responses to rainfall for P. tabulaeformis and H. rhamnoides. H. rhamnoides belongs to an anisohydric type, as F d in this species is more sensitive to rainfall and allows large fluctuations in ψ m to maintain G s with soil moisture decreases. In contrast, P. tabulaeformis can be categorized into an isohydric type, which mainly reduces G s to maintain a relatively constant ψ m . The results have implications for evaluating water use by different tree species, creating regional hydrological models, and selecting sustainable reforestation species in the Loess Plateau. The water use characteristics in the context of changes in rainfall pattern changes need further investigation to understand the long-term hydrological regime of these species.

ACKNOWLEDGMENTS
The study presented in this manuscript was a collaborative effort and would not have been possible without the help of Dr. Tang and Professor Chen. The study was conducted at the Ansai Station of the Chinese Academy of Sciences, located in Shaanxi Province, China. We thank Jie Wen, Yuli Xie, and Senbao Lu for their assistance with the study.

CONFLICT OF INTEREST
None declared.

AUTHORS CONTRIBUTION
Xu Wu performed the experiment, analyzed data, and wrote the manuscript. YKT and YMC designed the experiment and edited the manuscript. All authors read and approved the final manuscript.

Xu Wu
http://orcid.org/0000-0001-5543-7742 F I G U R E 9 (a) Midday leaf water potential for Pinus tabulaeformis (filled symbols) and Hippophae rhamnoides (open symbols), and (b) stomatal conductance during the study period. Different lowercase letters indicate the significant differences among species at .05 levels. "*" indicates the significant differences among measurements over 3 months at .05 levels for specific species. The data represent mean ±SE