Effects of Xerophytic Vegetation-Salix on Soil Water Redistribution in Semiarid Region

Abstract

Xerophytic vegetation re-regulates and allocates water resources through canopy interception, root water uptake and transpiration, and changes the water budget among precipitation, runoff, interception and infiltration, thus having a significant impact on the processes of the hydrological cycle. In this study, we investigated the effect of xerophytic shrub-Salix on soil water redistribution and water budget through an in situ monitoring experiment combined with two-dimensional vegetation water consumption modeling. The results showed that, due to the interception effect of root water uptake, it was difficult for precipitation infiltration to recharge deep soil water and groundwater. The measured data of soil moisture content, hydraulic head and precipitation were used to verify and calibrate the performance of the soil water flow model in the vadose zone by HYDRUS-2D. The effect of roots system on soil water was simulated, and the appropriate spacing of Salix replanting was estimated. Combined with the relationship between the transverse roots system and the crown width obtained by the investigation, it was determined that the spacing between the Salix should be greater than five times the crown width, so that the balance between the water consumption of Salix and the water supply of deep soil by precipitation could be considered. The results of this study are important for estimating groundwater recharge in arid areas and provide practical vegetation replanting options for similar regions.

1. Introduction

Philip proposed the concept of a soil-plant-atmosphere continuum (SPAC) system in 1966 [1]. The energy flow and material transformation in SPAC systems are extremely intense. In arid and semiarid areas, when the groundwater table depth is shallow, the soil water in the SPAC system is bound to be associated with groundwater under the drive of water potential gradient [2]. However, many studies have not explicitly paid enough attention to groundwater. Yang Jianfeng (1999) incorporated saturated aquifers into SPAC to form a groundwater-soil-plant-atmosphere continuum (GSPAC) system, highlighting the important role of groundwater in the system [3]. There are several interfaces for water exchange in GSPAC, one of which is the soil–root interface that controls the evapotranspiration process. Water exchange at the soil–root interface is the amount of water absorbed by the roots system, which usually accounts for more than two thirds of soil water consumption. More importantly, 65% of precipitation returns to the atmosphere through evapotranspiration, and more than 50% of transpiration flows through the soil–root interface [4]. In the processes of water transformation in GSPAC, root water uptake is affected by atmospheric, soil water conditions and groundwater level, which is one of the most difficult problems in soil water transformation.

Vegetation directly or indirectly changes soil water flow through canopy interception and root water uptake [5,6,7,8]. Especially in dry years, vegetation tends to increase the spatial variability of soil water [9]. Compared with other xerophytic vegetation, shrubs have an extensive roots system and high water consumption characteristics, which result in severe water deficit in deep soil and the formation of soil dry layers [10,11]. Some soil dry layers cannot be recharged through precipitation, which leads to changes in the structure and function of a vegetation ecosystem [12]. At the same time, xerophytic vegetation also affects groundwater recharge. For example, Huang et al. (2010) and Gates et al. (2011) found that the change in land use type from grassland to forest land/forest will lead to the decrease of groundwater recharge [13,14]. Allen (2001) found that although vegetation enhanced the water holding capacity of soil, the groundwater recharge rate was reduced by one tenth [15]. The root water uptake is similar to a water pump, intercepting and extracting a large amount of soil water and groundwater supplemented by precipitation [16,17]. Especially when the root depth exceeds the groundwater table depth, the roots system even directly consumes groundwater, causing the decline in groundwater level [18]. Although previous studies have shown that vegetation affects groundwater recharge, groundwater recharge in arid and semiarid areas is relatively small, and is difficult to quantify accurately [19,20].

To sum up, vegetation, as an important regulator of terrestrial water cycle, plays an important role in water cycle and water budget in the GSPAC system. Therefore, in this study, a typical xerophytic shrub in the Mu Us sandy land, Salix, was selected to study the impact of vegetation water consumption on soil water redistribution, and quantify soil water budget under the influence of vegetation at the individual plant scale. It is expected to provide support for vegetation restoration and water resource management in the study area.

2. Methods and Materials

2.1. Study Area

The study area is located in the southern edge of the Mu Us sandy land, which is 108°17′36″–109°40′22″ E and 37°38′54″–39°23′50″ N (Figure 1). Its administrative division belongs to Galutu Town, Wushen Banner, Ordos City, Inner Mongolia. Ordos city is an important energy and chemical base, and a typical agricultural and pastoral interlacing area. With the development of the economy and society, the demand for water resources is increasing, and the shortage of groundwater resources has become the main bottleneck restricting economic and social development and ecological construction. The study area belongs to a temperate continental monsoon climate, with large diurnal temperature difference, low humidity, sparse and uneven distribution of precipitation and strong evaporation. The annual potential evapotranspiration ranged from 1900 mm to 2500 mm. The average annual precipitation is 350–400 mm, mainly concentrated in July to September, with large inter-annual variation. The annual average temperature is 7.8 °C, and the annual sunshine is 2800~3000 h. In the past 20 years (2000–2020), the average annual precipitation is 378 mm, the average annual temperature is 8.7 °C and the climate is developing towards a warm and humid trend.

The study area is dominated by aeolian sand landform, with fixed and semi-fixed dunes distributed. The dune topography is highly undulating, and most of them appear alone or in the form of dune–dune depression, with shrubs or no vegetation cover. The soil of aeolian sand landform in Mu Us sandy land is mainly aeolian sand, and its constituent particles have coarse particle size, loose structure, uniform texture, little organic matter content and soil bulk density is 1.45 g·cm³, which is measured by SEDIMAT4-12 (UGT, Müncheberg, Germany) (

Table 1. Physical parameters of aeolian sand.

Grain Composition			Dry Density (g·cm−3)	Porosity (cm3·cm−3)	Name
0.5–0.25	0.25–0.075	0.075–0.05
5.7	88.7	5.6	1.45	0.37	silty-fine sand

). The water retention capacity of aeolian sand is weak, and the field water retention capacity is 6.85%, and the maximum water retention capacity is 14.82%, but the hydraulic permeability is strong, and the saturated hydraulic permeability coefficient is 580 cm·day⁻¹. The growth period of most vegetation in Mu Us sandy land is from May to November every year, among which May to mid-June is the germination period, mid-June to late July is the growing period, August to September is the maturing period and October to November is the wilting period. During the growth period, the roots system of vegetation has a strong water consumption, and the internal transformation of precipitation, soil water and groundwater is intense. The period from November to April of the following year is the dormant period of vegetation and the soil freezing period, during which precipitation and vegetation water consumption are almost zero.

2.2. Field Monitoring Experiment

The field monitoring experiment site was set on a dune with a relatively flat surface of about 25 m² (Figure 2), and the vegetation type was dominated by Salix. There was a mature perennial Salix with a height of about 3.5 m and a crown width of about 5.6 m in the experiment site, surrounded by bare aeolian sand. The soil moisture content and temperature sensors (ECH2O-5 TM (±1–2%), Decagon, Pullman, WA, USA) were installed in the soil inside the experiment site, and the location was as follows: the horizontal direction was 60 cm, 200 cm, 500 cm away from the center of Salix; the vertical depth was 3 cm, 10 cm, 20 cm, 30 cm, 50 cm, 100 cm, 200 cm, 300 cm. The monitoring frequency was 1 h/time. The specific field monitoring experiment site and the sensor installation details could be seen in Figure 2. The experiment was implemented from 2 June 2019 to 6 November 2019.

There was a weather station (Metpro^TM, Campbell, CA, USA) approximately 200 m away from the experiment site, which could provide atmospheric pressure, net solar radiation, precipitation, temperature, humidity, wind speed and direction, sunshine hours, etc., and the recording frequency was 1 h/time.

2.3. Determination of Soil Hydraulic Parameters

Laboratory experiments were conducted according to the soil samples obtained in the field to determine the soil hydraulic parameters. The saturated soil moisture content and residual soil moisture content of aeolian sand were obtained by a ring knife and immersed in water until saturated and then measured by drying method. The saturation permeability coefficient of aeolian sand was measured in situ by double ring experiment. The soil–water characteristic curve of aeolian sand was observed by a Ku-pF unsaturated conductivity measurement system (ku-pF MP10, UGT, Müncheberg, Germany), and the soil–water curve was fitted by Matlab (9.2.0.538062(R2017a)). The results of soil hydraulic parameters measurement are detailed in

Table 2. Hydraulic parameters of the aeolian sand.

Item	θr [cm3·cm−3]	θs [cm3·cm−3]	α [cm−1]	n [-]	Ks [cm·day−1]	l [-]
Aeolian sand	0.023	0.32	0.036	2.56	560	0.5

.

2.4. Modeling

The Hydrus-2D model can simulate water flow, heat, and solute transport in a two-dimensional variable-saturation watershed. In addition, the model can be used to estimate root water uptake and thus assess the spatial distribution of soil moisture content. Under the assumption of ignoring the air phase and the soil thermal gradient, the modified form of Richards equation is used to describe the water movement in the soil, and the Richards equation is solved numerically by Galerkin finite element method.

$${\displaystyle \frac{\partial \theta \left(h\right)}{\partial t}}={\displaystyle \frac{\partial }{\partial x}}\left[K\left(h\right){\displaystyle \frac{\partial h}{\partial x}}\right]+{\displaystyle \frac{\partial }{\partial z}}\left[K\left(h\right){\displaystyle \frac{\partial h}{\partial z}}+K\left(h\right)\right]-S\left(h\right)$$

(1)

where θ(h) is the volumetric soil moisture content (cm³·cm⁻³); h is the pressure head (cm); K(h) is the unsaturated hydraulic conductivity (cm·day⁻¹); t is time (d); x and z are the spatial locations (cm).

The root water uptake is calculated according to the Feddes model:

$$S\left(h\right)=\tau \left(h\right)\beta \left(x,z\right)T_P{L}_t$$

(2)

where T_p is the potential transpiration rate (cm·day⁻¹); L_t is the surface length associated with transpiration (cm); β(x, z) is the root water uptake distribution function (cm⁻²); and τ(h) is the root water uptake stress reduction function.

The soil water retention curve θ(h) and hydraulic conductivity function K(h) need to be known for the model. The van Genuchten equation is used to describe the hydraulic characteristics between soil moisture content and pressure head, as well as the relationship between unsaturated hydraulic conductivity coefficient and soil pressure head [7]:

$$\mathsf{\theta }\left(h\right)=\left\{\begin{array}{c}{\theta }_r+{\displaystyle \frac{{\theta }_s-{\theta }_r}{{\left[1+{\left|\alpha h\right|}^n\right]}^{1-\frac{1}{n}}}} h<0\\ \\ {\theta }_s h\ge 0\end{array}\right.$$

(3)

$$\mathrm{K}\left(h\right)={\displaystyle \frac{K_s{\left[1-\left({\left|\alpha h\right|)}^{n-1}{\left[\left(1+{\left|\alpha h\right|}^n\right)\right]}^{-m}\right.\right]}^2}{{\left[\left(1+{\left|\alpha h\right|}^n\right)\right]}^{\raisebox{1ex}{$m$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}}}$$

(4)

where θ_s is the saturated soil moisture content (cm³·cm⁻³); θ_r is the residual soil moisture content (cm³·cm⁻³); K_s is the saturated hydraulic conductivity (cm·day⁻¹); α is the air inlet parameter (cm⁻¹); n is the pore size distribution parameter (-). α and n are the empirical coefficients that determine the shape of the soil water retention curve.

HYDRUS-2D also includes a Marquardt–Levenberg-type parameter optimization algorithm for retrieving soil hydraulic parameters from measured data. In this study, HYDRUS-2D was used to construct a simplified conceptual model to analyze the transport process of soil water in vegetated and non-vegetated soils, and to calibrate the model parameters. The purpose of calibration is not to obtain a model that can accurately reproduce the in situ monitoring experiment process, but to obtain a reasonable and credible model. The model domain is 10 m × 10 m vertical two-dimensional volume, which is divided by triangular mesh. The model distinguishes the upper soil, which is dominated by rapid evaporation and infiltration, from the deep soil, where mainly root water uptake occurs, and has four soil zones: the shallow and deep layer of non-vegetated soil, and the shallow and deep layer of vegetated soil, as shown in Figure 3. In this study, zone I was defined as the root water uptake zone, zone II was the infiltration zone, zone III was the subroot zone and zone IV was the deep recharge zone. This partition can easily represent evaporation from shallow soil layers and distinguish soil water from vegetated land and bare land.

The upper boundary condition of the model is the realistic atmospheric boundary condition, and daily precipitation, as well as potential evaporation and transpiration rates, needs to be specified. The lower boundary condition of the model is defined as the free drainage boundary. The lateral boundary condition is set to the zero flux boundary. The initial conditions of the model are linear differences based on the pressure head converted from the measured soil moisture content through the soil water retention curve.

Since the Salix selected for the experiment was an adult vegetation, it was assumed that the roots system of the Salix was fully developed and did not change with time. The observed root density distribution profile of Salix: the root density decreased linearly in both vertical and horizontal directions, and the maximum root density was located at a certain deep position in the vertical direction within the root zone. Therefore, the maximum root depth of Salix is set as 5.0 m, the location of maximum root density is 0.4 m, the radius of horizontal root distribution is 3.0 m and the location of the maximum radius is 0.4 m.

The reliability of vegetation root water uptake model was analyzed by R², RMSE and MAE. The higher correlation coefficient R² indicates that the simulated value is more reliable. The accuracy of the model can be reflected in RMSE because it is very sensitive to any error (large or small) in the measurement set. MAE is the simulated value that deviates from the measured value and reflects the reliability of the model. The calculation formula of RMSE and MAE is as follows:

$$RMSE=\sqrt{{\displaystyle \frac{1}{n}}{\sum }_{i=1}^n{\left(S_i-C_i\right)}^2}$$

(5)

$$MAE={\displaystyle \frac{1}{n}}{\sum }_{i=1}^n\left|{\displaystyle \frac{S_i-C_i}{C_i}}\right|$$

(6)

where n is the number of observation points, S is the simulated value and C is the calculated value.

2.5. Water Budget

According to the soil water balance method in vertical two dimensions, the relationship among precipitation inputs, changes in soil water storage, evapotranspiration (ETa) and deep recharge are determined as follows:

$$∆W=Q_{in}-Q_{out}$$

(7)

where $∆W$ is changes in soil water storage (cm²), $Q_{in}$ is the flux of water into the soil medium (cm²) and $Q_{out}$ is the flux of water flowing out of the soil medium (cm²). When the upper boundary of the soil medium is the surface, $Q_{in}$ represents precipitation, and $Q_{out}$ represents evaporation and water flux flowing out of the bottom boundary.

In order to analyze the drive processes of rainfall on soil water better, the variability of all rainfall events in this year was analyzed. The result shows that the rainfall events; variation coefficient over 15-days on average is large (see Figure 4), which will bring uncertainty; although the rainfall events variation coefficient less than 10 days on average is small, it will increase the computing efforts. So, rainfall events at 15 days on average is the optimal and 15 days was set as the time step in this study.

3. Results and Discussion

3.1. Identification of the Model

During the simulation, the soil hydraulic parameters, such as the residual soil moisture content (θ_r) and saturated soil moisture content (θ_s), were not sensitive to the model results. Therefore, in the process of parameter inversion, θ_r and θ_s were set as fixed values. The other parameters were optimized by inversion. Moreover, the medium was homogeneous silty sand and its hydraulic parameters were uniform in theory. But the soil characteristics varied slightly with depth after many simulation analyses. Therefore, the domain was stratified properly in order to achieve the best fitting effect. The aeolian soil land was divided into five layers: 0–60 cm, 60–120 cm, 120–200 cm, 200–250 cm, 250–300 cm. The hydraulic parameters of each soil layer were slightly different, and the optimized hydraulic parameters are shown in

Table 3. Optimized hydraulic parameters of aeolian sand.

Soil Layers	Depth (cm)	α [cm−1]	n [-]	Ks [cm·day−1]
1	0–60	0.0110	1.440	450.0
2	60–120	0.0366	1.611	660.0
3	120–200	0.0199	3.230	760.0
4	200–280	0.1551	4.617	410.0
5	280–400	0.0170	2.800	760.0

.

To further analyze the accuracy of the model simulation results, the statistical values, ME, RMSE and MAE, of the measured soil moisture content and its simulated values were calculated (

Table 4. Statistical analysis of error of model simulation results.

	Depth (cm)
	3	10	20	30	50	100	200	300
ME	0.017	0.013	0.011	0.012	0.009	0.018	0.014	0.007
RMSE	0.022	0.016	0.014	0.014	0.011	0.020	0.019	0.009
MAE	0.235	0.111	0.109	0.103	0.075	0.173	0.100	0.026

). The results indicate that the simulated results are basically consistent with the measured values at each depth except for the slight deviation for shallow soil moisture content. Therefore, this model can be used to analyze the soil water redistribution and water budget under the influence of Salix.

3.2. Distribution of Soil Water in Bare Land and Vegetated Land

Figure 5 shows the evolution of soil moisture content under the influence of root water uptake. It can be seen that the soil moisture content in the root zone changed mostly, and the root water uptake resulted in serious soil water loss in the root zone. Soil moisture content in the non-root zone was generally higher than that in the root zone, and soil water was driven to transfer to the root zone under the action of hydraulic head gradient, resulting in a decrease in soil moisture content in the non-root zone indirectly. In the vertical direction, the maximum depth of root water uptake was more than 7 m. In the horizontal direction, the influence range of root water uptake on soil water reached 8 m, which was 1.7 times the horizontal distribution of the roots system. Under continuous drought conditions, the range of soil water deficit induced by root water uptake was similar to the distribution of roots density. When precipitation occurred, the soil moisture content near the edge of horizontal roots system was higher than that in the core root zone of the same depth. This is because the root density in the edge of horizontal roots system is small, and the interception of roots system on infiltration precipitation is limited. Therefore, the infiltrated precipitation can penetrate down and replenish the deep soil layers and even groundwater.

Figure 6 shows the hydraulic head distribution of the soil profile under the influence of root water uptake. It can be seen that the hydraulic head in the root zone was the lowest, and when precipitation occurred, the hydraulic head in the shallow root zone increased significantly. The hydraulic head of soil in the non-root zone was higher than that in the root zone, and soil water transferred to the root zone under the effect of hydraulic head gradient, which supports the results shown in Figure 4.

The movement of soil water can be generalized to the pattern shown in Figure 7. That is, the infiltrated precipitation was mainly intercepted and absorbed by the roots system in the root water uptake zone, and soil water was consumed, resulting in a water deficit zone, without excess soil water transferring to deeper layers. While in the infiltration zone, infiltrated precipitation was used for evaporation and infiltration mainly. Under the indirect effect of root water uptake, soil water in this zone gradually transferred to the root zone under the drive of horizontal hydraulic head gradient. In the subroot zone, soil water transferred to the water deficit zone driven by vertical hydraulic head gradient, which was caused by root water uptake. In the deep recharge zone, the infiltrated precipitation can further transport to the deep soil until it recharges to the groundwater.

3.3. Effects of Root Water Uptake on Water Budget

Water budget refers to the change in the amount of water inflow, outflow and storage within the GSPAC system. Soil water budget refers specifically to the relationship between the water obtained in a certain soil volume and the water consumed by vegetation in a certain period of time, which is an important factor affecting the ecological environment [21,22]. Figure 8 shows the water budget of the four sub-regions of the land. The water budget in both the root water uptake zone and infiltration zone showed a trend of first increasing and then decreasing, which was attributed to the fact that soil water was supplemented with the increase in precipitation, and then was consumed due to the continuous effect of vegetation transpiration and soil evaporation after a precipitation event. In addition, the increment of soil water in the infiltration zone is higher than that in the root water uptake zone, which means that root water uptake is not conducive to soil water storage. The soil water increment in the subroot zone is much smaller than that in the deep recharge area under the bare soil, indicating that the infiltrated precipitation can effectively recharge the deep soil in the bare land, while in the vegetated land, the infiltrated precipitation is intercepted and absorbed in the root zone, which weakens the recharge amount to the deep soil layers.

From the changes in water budget in the vegetation land and the bare land shown in Figure 8b, it can be seen that the changes in water budget in the bare land show a trend of first increasing and then decreasing, and the water volume increases after a growing period. However, the water budget in the vegetation land changed slightly, and after one growing period, there was almost no change in water volume. It further confirms the interception effect of vegetation roots on soil water, which is very unfavorable for maintaining available soil water and groundwater recharge. Soil water budget is affected by climate, vegetation cover and soil types [23,24,25]. Vegetation has a strong influence on soil water budget through their water acquisition, transport and transpiration [26,27,28,29,30], and deep-rooted vegetation in particular consumes a large amount of deep soil water [31,32]. In addition, vegetation restoration and vegetation types will also have an impact on soil water budget [33,34]. In arid and semiarid regions, vegetation restoration is conducive to soil and water conservation, but after vegetation restoration, the underlying surface conditions will be changed. Due to the change in infiltration conditions and root water uptake, soil moisture content will be affected, thus affecting water budget. Therefore, it is of great significance to study the impact of vegetation on soil water budget for vegetation replanting, soil and water conservation and water resources protection [22,35,36].

3.4. Reasonable Estimation of Vegetation Replanting Interval

Vegetation reduces soil evaporation through canopy shading, and root water uptake prevents soil water infiltration, which all affect the distribution and transport processes of soil water. Therefore, vegetation spacing is a very important parameter. If the vegetation is close to each other, the shadow under the canopy will reduce the evaporation of shallow soil water [37,38,39,40]. If the vegetation is far apart, the surface soil temperature in the inter-space is higher than that in the canopy cover area, which promotes the increase in soil evaporation and aggravates water loss. However, for xerophytic vegetation, its lateral root system development, which exceeds the crown width, and has a wide range of influence on soil water. If vegetation spacing is close, the attack of soil water by vegetation will be intensified [40], while large vegetation spacing can reduce the water consumption by root water uptake. On the other hand, the soil’s hydraulic permeability under the vegetation canopy is enhanced due to the root channel effect [41,42], and the precipitation infiltration capacity decreases gradually with the increase in canopy distance. However, whether precipitation can effectively recharge groundwater in the vegetation land is largely unknown and has only been studied in a few xerophytic shrubs [43]. It can be seen that vegetation has many effects on soil water, especially in soil water conservation. There are both favorable and unfavorable effects, which should be analyzed synchronously in the study of hydrological functions of xerophytic vegetation.

The height of mature Salix is generally more than 2.5 m, the canopy width is more than 3 m, the extension depth of taproot can reach 5 m, the ratio of root depth to canopy can reach more than 2.34 times and the ratio of lateral root length to canopy can reach more than 1.5 times [44,45]. According to the results of this study, the lateral distance of roots affecting soil water is nearly 1.7 times that of horizontal roots length. Therefore, in order to ensure that precipitation can effectively infiltrate in the deep soil layers without being intercepted by the roots system, the spacing of Salix should be more than five times that of the canopy width. At present, a common measure to prevent wind and sand fixation is to plant Salix in the central position of the artificially laid “grass grid”. The side length of the “grass grid” is usually 1 m, so the spacing between Salix is also 1 m. When Salix survived, it could develop into mature vegetation in 2~3 years. If the vegetation spacing was not controlled reasonably, the recharge of deep soil water or groundwater would be affected, which would not be conducive to the sustainability of vegetation replanting.

4. Conclusions

In this study, a field in situ experiment was conducted on a sand dune in the Mu Us sandy land to investigate soil water redistribution affected by Salix. The measurements of soil moisture content, hydraulic head and precipitation were used to verify and calibrate the performance of a vadose zone soil water flow model. The modeling results highlight the role of the local precipitation as an essential source of vegetation and groundwater in the Mu Us sandy land. Due to the interception effect of root water uptake, it is difficult for precipitation infiltration to recharge deep soil water or even groundwater. Combined with the relationship between the transverse roots system and the canopy width obtained by the investigation, it was determined that the spacing between the Salix should be greater than five times that of the canopy width. The results tradeoff the balance between the water consumption of vegetation and the deep soil water supplied by precipitation.