**Chunfeng Jia 1,2,3, Bao-ping Sun 1,\*, Xinxiao Yu <sup>1</sup> and Xiaohui Yang <sup>4</sup>**


Received: 28 March 2019; Accepted: 13 April 2019; Published: 17 April 2019

**Abstract:** Vegetation restoration plays a significant role in the restoration of expressways in the arid zone of China, but we still do not know which soil and vegetation types are most effective. We investigated soil particle size (SPZ), volume weight of the soil (VWS), soil water content (SWC), total porosity of soil (TP), soil organic matter (SOM), water erosion (WrE), and wind erosion (WdE) of eight sites (S1–S8) and evaluated them using the gray correlation method (GCM). Based on our results, the average SWC of the treatments ranged from 9.6% to 18.8%, following the order S4 > S5 > S8 > S6 > S3 > S7 > S1 > S2. The average SPZ of soils in S1, S2, S4, S5, S6, and S8 was larger, ranging from 0.23 to 0.68 mm, while that of soils in S3 and S7 was smaller, ranging from 0.01 to 0.09 mm. The TP in different treatment areas ranged from 50% to 60%, which is not conducive to soil and water conservation. The SOM levels varied widely among the different soils and were always below the threshold levels established by the second National Soil Census, rendering the soils not suitable for plant growth. The WrE (36–80 t/ha) was greater than the WdE (7–24 t/ha). In general, to achieve high soil and water conservation outcomes in this area, S1 and S7 offered the best protection benefits in terms of soil and water conservation.

**Keywords:** benefit evaluation; gray correlation method; soil erosion; soil physical properties

### **1. Introduction**

The construction of expressways inevitably destroys the surface vegetation, leading to significant ecological problems such as soil erosion and landscape degradation [1–3]. In this sense, vegetation restoration along expressways is crucial to stabilize the roadbed, improve the road environment, and enhance the aesthetic properties of such areas [4–6]. During construction, the soil structure is largely destroyed; the surface soil in the cutting slope was mixed with gravel, parent material, and, in some cases, even rocks [7]. As a result, the surface temperature difference between day and night is relatively high, physical weathering is extreme, and evaporation is high; consequently, precipitation easily initiates runoffs [8,9].

In this sense, vegetation restoration along expressways is challenging. Previous studies have taken a series of measures to restore expressway slopes. The traditional forms of slope vegetation protection include grass planting [10], three-dimensional vegetation networks [11], masonry stone wall establishment [12], skeleton grass planting [13], vine plant protection [14], and geogrid grass planting [15]. Slope grass planting is characterized by a low survival rate, high maintenance costs, and the frequent degradation of grass seeds, while shrub planting is restricted by the water and

nutrient conditions of the slope [14]. Inevitably, such approaches led to a low protection effect of the slope. Some expressway greening projects were only possible with the use of engineering work [15], with large investments. Also, over time, with rock weathering, concrete aging, and steel bar erosion, the protection effects become weaker [16,17]. In the northern regions of China, cutting slopes are also highly vulnerable to wind erosion [18,19]. Especially in the greening of expressways in arid, cold, and desertified areas, this problem is more prominent and mainly manifests in the following aspects: Firstly, the project implementers emphasize short-term effects while neglecting long-term effects [20], which directly leads to the phenomenon of "1 year green, 2 years yellow, 3 years withered, and 4 years dead". Secondly, project implementers overemphasize the quantity of greening. For example, in some arid areas, the emphasis was also on vegetation coverage, and especially in the south, evergreen trees have been planted extensively, often with unsatisfactory results. Thirdly, plant species selection and allocation were unreasonable [21], without any specific norms and standards, mainly copying the model of landscape greening and ignoring the natural laws of ecosystem and plant growth, which often resulted in unsatisfactory outcomes.

To reduce the environmental impacts of expressway construction and to improve the restoration of the damaged areas, it is crucial to employ new technologies, especially in cold, dry, and desertified areas. In this context, we included seven indices, namely volume soil weight, total porosity, soil moisture content, soil particle size, soil organic matter, water erosion, and wind erosion, from eight different treatment areas to evaluate the soil and water conservation benefits, using the gray correlation method (GCM) [22]. In this study, we summarize a set of technical methods for the ecological restoration of expressway areas in cold, dry, and desertified loess areas. The results of our study are of great significance for the restoration of vegetation along expressways, providing a reference for greening technology in other construction projects in cold, arid, and desertified areas and for technological progress of the vegetation restoration industry in general.

#### **2. Study Sites**

The study area is located along the Dahu Expressway (113.275◦ E and 39.936◦ N) in Datong City, Shanxi Province, a typical loess plateau area in China (Figure 1), at an elevation range from 1320 to 1170 m. The climate is north-temperate semi-arid continental monsoon climate. Due to the monsoon and high pressure on the Siberian and Mongolian plateau, there is little snow; the area experiences mild temperatures in winter, drought and windiness in spring, hot and rainy summers, and cool autumns. Average annual precipitation is 399 mm, of which more than 60% fall from July to September. Average annual temperature is 6.1 ◦C, with minimum and maximum temperatures of −29.5 and 34.5 ◦C, respectively. The daily temperature difference is 13.0 ◦C, with a frost-free period of 125 days. The study area can be divided into five geomorphic units, including loess hilly area, intermountain valley plain area, erosion and denudation high hilly area, valley plain area, and alluvial and flood plain area. The soil is mainly weathered rock soil, coarse silty soil, collapsible loess, and liquefaction soil of saline soil with sandy soil (Reference to China Soil Classification System). The main tree species are *Pinus tabulaeformis*, *Larix gmelinii*, and *Pinus sylvestris var. mongolica*. The shrub species mainly include *Hippophae rhamnoides Linn*. and *Caragana korshinskii*, while the most abundant herb genera are *Carex* and *Artemisia*. The above information was obtained by consulting local historical meteorological data and the respective yearbook.

*Sustainability* **2019**, *11*, x FOR PEER REVIEW 3 of 14

**Figure 1.** Geographic map of the study area. The numbers 1–8 represent the treatments S1–S8. **Figure 1.** Geographic map of the study area. The numbers 1–8 represent the treatments S1–S8.

#### **3. Materials and Methods 3. Materials and Methods**

### *3.1. Experimental Design*

*3.1. Experimental Design*  Because the soil and vegetation were completely destroyed during the construction of roads and slopes, the developers filled the slopes with soil and planted shrubs to restore the ecological environment. During the implementation of the project, four kinds of soils, including weathered rock soil, coarse silty soil, collapsible loess, and liquefaction soil of saline soil with sandy soil were filled (Figure 2), and two plant species (*H. rhamnoides Linn.* and *C. korshinskii*) were planted. To determine which soil and vegetation allocation provided better soil and water conservation benefits, eight representative treatment areas were selected according to the actual situation of the project Because the soil and vegetation were completely destroyed during the construction of roads and slopes, the developers filled the slopes with soil and planted shrubs to restore the ecological environment. During the implementation of the project, four kinds of soils, including weathered rock soil, coarse silty soil, collapsible loess, and liquefaction soil of saline soil with sandy soil were filled (Figure 2), and two plant species (*H. rhamnoides Linn.* and *C. korshinskii*) were planted. To determine which soil and vegetation allocation provided better soil and water conservation benefits, eight representative treatment areas were selected according to the actual situation of the project area. Specific information on the soil and vegetation allocation in each treatment area is shown in Table 1.

area. Specific information on the soil and vegetation allocation in each treatment area is shown in Table 1. **Table 1.** Basic information of the treatments along the expressway in Datong City, Shanxi Province, China.


6 S6 10 × 20 Coarse silty soil 40 7 S7 10 × 20 Collapsible loess 45

8 S8 10 × 20 Liquefaction soil of saline

soil with sandy soil <sup>39</sup>

*H. rhamnoides Linn.* 

*Sustainability* **2019**, *11*, x FOR PEER REVIEW 4 of 14

**Figure 2.** Filled soil in the different treatments: S1 and S5 were the weathered rock soil; S2 and S6 were the coarse silty soil; S3 and S7 were the collapsible loess; S4 and S8 were the liquefaction soil of saline soil with sandy soil. **Figure 2.** Filled soil in the different treatments: S1 and S5 were the weathered rock soil; S2 and S6 were the coarse silty soil; S3 and S7 were the collapsible loess; S4 and S8 were the liquefaction soil of salinesoil with sandy soil.

#### *3.2. Soil Sample Collection 3.2. Soil Sample Collection*

All samplings and erosion observations were performed in 2013. For the eight treatments along the expressway, nine sampling points were selected for each treatment. At each sampling point, soil samples were collected from the undisturbed original slope surface using a ring knife at depths of 0–20 cm, 20–40 cm, 40–60 cm, and 60–80 cm. The samples were brought to the laboratory to determine the volume weight of soil (VWS) and total porosity (TP). Additional soil samples were collected using an aluminum box to measure the soil moisture content (SWC) at the four soil depths, using the dry weighing method [23], and to determine soil organic matter levels via the potassium dichromate method [24]. Three soil samples were collected from each soil layer for each All samplings and erosion observations were performed in 2013. For the eight treatments along the expressway, nine sampling points were selected for each treatment. At each sampling point, soil samples were collected from the undisturbed original slope surface using a ring knife at depths of 0–20 cm, 20–40 cm, 40–60 cm, and 60–80 cm. The samples were brought to the laboratory to determine the volume weight of soil (VWS) and total porosity (TP). Additional soil samples were collected using an aluminum box to measure the soil moisture content (SWC) at the four soil depths, using the dry weighing method [23], and to determine soil organic matter levels via the potassium dichromate method [24]. Three soil samples were collected from each soil layer for each analysis.

#### analysis. *3.3. Determination of Soil Physical Properties*

#### *3.3. Determination of Soil Physical Properties*  3.3.1. Total Soil Porosity

3.3.1. Total Soil Porosity To measure soil porosity, we opened the lower cover of the ring knife (with the mesh end), put it onto filter paper, and immediately weighed the sample (m1). Subsequently, we removed the upper cover of the ring cutter and placed the mesh end of the ring cutter into the basin. The height of the water layer in the bowl was up to the edge of the knife (not submerged). After 12 hours of water absorption, we removed the ring knife, covered the lid, and weighed the sample again (m2). After weighing, the ring cutter (bottom end of the mesh) was placed in a flat pan with dry sand and weighed (m3) after 12 h. After weighing, the ring knife (mesh side down) was placed in a flat pan again and weighed (m4) immediately after infiltration for 12 h. The weighed ring knife cover was To measure soil porosity, we opened the lower cover of the ring knife (with the mesh end), put it onto filter paper, and immediately weighed the sample (m1). Subsequently, we removed the upper cover of the ring cutter and placed the mesh end of the ring cutter into the basin. The height of the water layer in the bowl was up to the edge of the knife (not submerged). After 12 hours of water absorption, we removed the ring knife, covered the lid, and weighed the sample again (m2). After weighing, the ring cutter (bottom end of the mesh) was placed in a flat pan with dry sand and weighed (m3) after 12 h. After weighing, the ring knife (mesh side down) was placed in a flat pan again and weighed (m4) immediately after infiltration for 12 h. The weighed ring knife cover was opened, placed in an oven at 105 ◦C for 24 h, covered, and weighed (m5). Subsequently, soil porosity was calculated as follows:

$$\mathbf{P} = \left(\frac{\mathbf{m}\_2 - \mathbf{m}\_5}{\mathbf{v}}\right) \times 100\% \tag{1}$$

m2 5 P= 100% *<sup>m</sup>* <sup>−</sup> <sup>×</sup> (1) where P is total porosity (g/cm<sup>3</sup> ) and v is the volume of the ring cutter.

#### 3.3.2. Volume Weight of The Soil

follows:

porosity was calculated as follows:

where P is total porosity (g/cm3) and v is the volume of the ring cutter. 3.3.2. Volume Weight of The Soil The VWS was measured using the ring knife method [25]. We ensured that the soil in the ring The VWS was measured using the ring knife method [25]. We ensured that the soil in the ring cutter was not disturbed during each operational step. The ring cutter with the soil sample was placed in an oven at 105 ◦C and dried to constant weight; afterwards, the ring knife with soil samples (m1j) and the soil samples (m2j) were weighed separately, and the VWS was calculated as follows:

*v*

$$\text{VWS} = \frac{\sum\_{j=1}^{3} (\mathbf{m}\_{\text{lj}} - \mathbf{m}\_{\text{2j}})}{\mathbf{3v}} \tag{2}$$

3 1 2 j 1 m m VWS *j j* = − = ( ) (2) where the v is the volume of the ring cutter, and J = 1, 2, and 3 is the repeat number of the same soil layer.

3v

#### 3.3.3. Soil Organic Matter

We used the potassium dichromate oxidation-external heating method to determine soil organic matter (SOM). Briefly, the soil mixture is boiled with potassium dichromate oxidizer and sulfuric acid (95%) in an oil bath at 170–180 ◦C for 5 min. The carbon in the SOM was oxidized to carbon dioxide by potassium dichromate, while the hexavalent chromium in the potassium dichromate was reduced to trivalent chromium. The remaining potassium dichromate was then titrated with standard solution of ferrous oxide. According to the amount of ferrous sulfate consumed by potassium dichromate before and after oxidation of organic carbon, the content of organic carbon was calculated and converted into SOM.

#### 3.3.4. Soil Erosion

During the observation period, runoff barrels were used to collect runoff and sediment samples in the eight treatment plots. Weighing–sedimentation–drying–weighing and sampling–drying–weighing were used to correct the runoff and sediment yields, respectively. Hydraulic erosion per hectare was obtained through the area of the plot [26]. Secondly, wind erosion was measured via the cutting method [27]. The first choice is to record the depth of cutting, measure the change depth of cutting after the observation period, and, finally, calculate the wind erosion.

#### 3.3.5. Evaluation of Benefits of Soil and Water Conservation of Different Treatments by the GCM

Here, X<sup>0</sup> is the reference series and X<sup>i</sup> the comparison series. The value of the reference series represents the maximum value of each test indicator. First, we calculated the absolute value of the difference between each corresponding point of comparison sequence X<sup>i</sup> and reference sequence X0, and the GCM was evaluated as follows:

$$\Delta \mathbf{i}(k) = \left| \mathbb{X}\_0(k) - \mathbb{X}\_i(k) \right| \tag{3}$$

$$\mathbf{X}\_{0} = \{ \mathbf{X}\_{0}(1), \mathbf{X}\_{0}(2), \mathbf{X}\_{0}(3), \dots, \dots, \mathbf{X}\_{0}(n) \} \tag{4}$$

$$\mathbf{X}\_{i} = \{ \mathbf{X}\_{i}(1), \mathbf{X}\_{i}(2), \mathbf{X}\_{i}(3), \dots, \dots, \mathbf{X}\_{i}(n) \}, i = 1, 2, 3, \dots, \dots, m. \tag{5}$$

When using this approach, the actual value of each index must be converted into an evaluation value, and the original data should be processed without a dimension, according to the reference series data. On the one hand, the influence brought by each index dimension should be removed [28]; however, on the other hand, this indicator can also reflect the relative dominance of the community.

We used the formula X*<sup>i</sup>* (k) = original data sequence/reference data sequence to perform non-dimensionalization, reducing all data return to the [0,1] interval.

Subsequently, we determined the second-order maximum difference and the second-level minimum difference and calculated the correlation coefficient:

$$\mathbb{E}\_{i}(k) = r\mathbb{X}\_{0}(k), \mathbb{X}\_{i}(k) = \frac{\text{minmin}\Delta\_{i}(k) + \rho \text{maxmax}\Delta\_{i}(k)}{\Delta\_{i}(i) + \rho \text{maxmax}\Delta\_{i}(k)} = \frac{\Delta\_{\text{min}} + \rho\Delta\_{\text{max}}}{\Delta\_{0i}(k) + \rho\Delta\_{\text{max}}} \tag{6}$$

where ξ*<sup>i</sup>* (*k*) represents the relative value of the k point comparison curve (X*<sup>i</sup>* ) and the reference curve (X0); that is, the X*<sup>i</sup>* 's correlation coefficient of X<sup>o</sup> at point k. When ∆0, *i*(k) represents the absolute value of the X<sup>0</sup> sequence and the X*<sup>i</sup>* sequence at the k point, 1 ≤ i ≤ m, m was positive; ∆min and ∆max, respectively, represent the minimum and maximum values of the absolute difference of all comparison points in each point; ρ is the resolution coefficient ranging from 0–1. Here, ρ was artificially set to 0.5 (the artificial coefficient for qualitative analysis) to weaken the distortion effect caused by the excessively large maximum value to increase the significance of the difference between the correlation coefficients:

$$\mathbf{r}\_{l} = \frac{1}{\mathbf{n}} \sum\_{k=1}^{n} \xi\_{l}(k) \tag{7}$$

where r*<sup>i</sup>* indicates the degree of gray correlation; ξ indicates the gray correlation coefficient.

To measure the importance of each index more realistically and objectively, the coefficient of variation W*<sup>i</sup>* was calculated using the coefficient of variation method:

$$\mathbf{W}\_{i} = \frac{\mathbf{r}\_{i}}{\sum \mathbf{r}\_{i}}.\tag{8}$$

Subsequently, we calculated the gray comprehensive evaluation value G (k) with the correlation degree value:

$$\mathbf{G}\_k = \sum\_{k=1}^n \xi\_i(k)\mathbf{W}\_i. \tag{9}$$

#### *3.4. Statistical Analysis*

Statistical analyses were performed using the software package SPSS 16.0. Descriptive statistics was used to calculate the mean and standard deviations for each set of replicates. First, a two-way ANOVA was used to analyze the differences in the SOM and SWC (*n* = 27 for each soil stratum), with treatment and soil depth as the independent factors. Two-way ANOVA was also used to analyze the differences in the SPT (*n* = 27 for each soil stratum), with treatment and soil particle size as independent factors. One-way ANOVA was also performed to test the effects of VWS, TP, WrE, and WdE on all treatments. All data were tested for normal distribution and homogeneity of variance analysis, meeting the requirements of variance analysis.

#### **4. Results and Discussion**

#### *4.1. Basic Meteorological Characteristics*

The study area is a temperate continental monsoon climate region, with an annual average precipitation of 384.02 mm. In 2013, average precipitation reached 418.7 mm, which was 9.03% higher than annual average precipitation (Figure 3). Precipitation from June to September was 296.1 mm, accounting for 70.71% of the total precipitation. Daily average precipitation was 1.14 mm, and maximum precipitation was 45.9 mm on September 9. During 2013, average daily temperature was 2.1 ◦C, with a single-peak "convex" pattern with seasonal changes. This gradually increased from January to July, reaching a maximum of 26.7 ◦C on July 30. After this peak, temperatures gradually *Sustainability*  decreased. The average daily wind speed was 1.04 m **2019**, *11*, x FOR PEER REVIEW /s, with distinct differences between the seasons. 7 of 14

**Figure 3.** Precipitation, temperature, and wind speed in the study area along the expressway in Datong City, Shanxi Province, China, in 2013. **Figure 3.** Precipitation, temperature, and wind speed in the study area along the expressway in Datong City, Shanxi Province, China, in 2013.

*4.2. Weight Percentage of Soil Particle Size Under Different Treatments* 

water-holding capacity of the soil will be reduced [36].

distribution of the particle size of the samples greatly varied among the different treatments (Figure 4). Site S2 contained the largest amount of gravel among all treatments, accounting for 25.35%. In the other treatments, the proportion of gravel was smaller, or gravel was completely absent. Particles with a diameter of 0.05–1 mm were most abundant in S8, with a significant difference when compared to the other treatments (*p* < 0.05). The weight percentages of d = 0.05–1 mm were 48.86%, 35.18%, 41.35%, 50.28%, and 61.72%, respectively, for the treatments S1, S2, S3, S7, and S8, which were significantly higher than those for the other particle size compositions (*p* < 0.05). The proportions of particles with a diameter of <0.005 mm were 39.52% and 38.65%, respectively, in the treatments S4 and S5, which were significantly higher than those in the other treatments (*p* < 0.05). Lu et al. also found that the size of the soil particles in the soil matrix differed; this was not only the case for the surface soil layers, but also for soil porosity in general [33]. Sandy soil contains coarse gravel and has a high soil porosity. On the contrary, the permeability of loam or clayey soil is lower than that of sandy soil, facilitating surface runoff [34]. Soil particle size and runoff are closely related, and a favorable soil particle size composition can effectively maintain water, nutrients, and organic matter [35]. According to our results, most of the particles with a diameter of more than 2 mm retain the original mineral composition of the parent rock. There are few available mineral nutrients, and the ability to absorb water was also poor. When the content of gravel in the soil exceeds 20% of the total volume of sample, changes in the temperature of the sample will be aggravated, and the

Previous studies have stated that with decreasing soil particle diameter, the soil cohesiveness
