Soil Properties under Artificial Mixed Forests in the Desert-Yellow River Coastal Transition Zone, China

Abstract

Mixed forests play a key role in the environmental restoration of desert ecosystems and in order to address the improvement of soil properties by different mixed vegetation types. We selected four typical mixed vegetation types (including: Populus alba var. pyramidalis × Caragana korshinskii, P. pyramidalis × Hedysarum mongdicum, P. pyramidalis × Hedysarum scoparium and Hedysarum scoparium × Salix cheilophila) that have been restored for 22 years and the moving sandy land in the transition zone between the desert and the Yellow River in northern China. We compared the differences in soil properties using a total of 45 soil samples from the 0–30 cm soil layer (10 cm units). We found that revegetation had a significant positive effect on fine particles, soil nutrients, soil bulk density (SBD), and soil fractal dimension (D) values. Soil D values under different types of vegetation range from 2.16 to 2.37. Soil nutrients and fractal dimension showed highly significant or stronger negative correlations with SBD and sand and highly significant or stronger positive correlations with clay and silt. The construction of P. pyramidalis × C. korshinskii improved the soil texture better than other vegetation restoration types. Compared to the mobile sandy land, organic carbon (SOC), available phosphorus (AP), available potassium (AK), alkaline hydrolysis nitrogen (AN), total nitrogen (TN), total potassium (TK), clay, and silt increased by 161%, 238%, 139%, 30%, 125%, 69%, 208%, and 441% respectively. As mentioned above, P. pyramidalis × C. korshinskii is a suitable type of mixed vegetation restoration for the area. In addition, establishing vegetation with high nitrogen fixation rates in desert ecosystems tolerant to drought and aeolian conditions is beneficial in reversing the trend of desertification. This research will suggest vegetation building strategies for controlling desertification.

1. Introduction

Desertification is a major global environmental problem [1]. Generally speaking, desertification is driven by a combination of natural and human factors. The process of desertification involves biological and economic losses and is therefore inextricably linked to ecological security and regional development [2]. In past reports, it has been mentioned that the economic losses due to desertification in China are estimated at US$ 6.8 billion per year. The ability of the land to produce food and fodder, security of food, quantity and quality of fresh water, reduction in the resilience of the land, as well as maintenance costs of production and construction equipment, social costs, poverty, educational problems, and increased political instability, and even damage to human life and health, are some of the obvious effects of desertification [3]. These phenomena threaten the sustainable development goals of human social development in drylands. In addition, strong winds have blown sand particles from arid and semi-arid regions of China to Japan, Korea and even to the United States [4]. As a result, the United Nations Conference on Desertification was held in the 1980s. Since then, China has formally begun to adopt a multidisciplinary and comprehensive approach to combating desertification. Over the past three decades, researchers have made some positive progress in controlling desertification, with much of the work occurring in areas of fragile ecological conditions and high human activity [5]. Some scholars working on wind erosion have successively proposed ecological engineering interventions, such as artificial vegetation, to control these desertification hazards in northern China [6,7]. Moreover, there is a broad consensus worldwide that appropriate vegetation is the most sustainable and effective solution for restoring degraded land and combating desertification [8,9,10]. Vegetation can attenuate aeolian activity by increasing the aerodynamic roughness length of the surface and reducing the near-surface wind speed, capturing soil particles [11,12,13]. The deposition of soil particles supports soil development and vegetation growth [14,15]. Vegetation promotes the cycling of C and N and P in the soil system and improves soil quality through the secretions produced by the root system and the improvement of the soil microenvironment [16,17,18].

The Hobq Desert, the seventh largest desert in China, has seen population growth, lifestyle changes, and economic and social development over the past three decades that have accelerated desertification. A comprehensive assessment of the risk of wind erosion in the area using the IWEMS (Integrated Wind-Erosion Modeling System) and RWEQ (Revised Wind Erosion Equation) models has been carried out by several scholars and compared with observations from local government and hydrological stations. 31 t/m of saltation emission was discharged from the Hobq Desert between 2001 and 2010. Of this, up to 1.39 × 10⁷ t of aoeolian sediment was blown from the Hobq Desert into Ten Tributaries in 2001 [19]. In 2002, China’s Law on the Prevention and Control of Sand and Desertification was implemented with the aim of promoting sustainable development and preventing and combating desertification. The law is aimed at the whole of China and its implementation has boosted the enthusiasm of the whole society to combat drought and promote forestry in the desert [20]. Since then, the annual saltation deposition of sand into the Yellow River in China’s Hobq Desert has been reduced to 2.04 × 10⁶ t and the area of artificial vegetation has increased by 2506.86 km² by 2017 [19,21]. Many studies have demonstrated a reduction in soil erosion and a significant improvement in soil physical and chemical properties following revegetation in the Hobq Desert [22,23]. However, different vegetation types have different effects on soil properties. Populus alba var. pyramidalis, Caragana korshinskii, Hedysarum mongdicum, Hedysarum scoparium and Salix cheilophila are widely planted in the sandy areas of northern China [24,25] because of their fast growth rate, resistance to drought, adaptation to aeolian conditions [26], and improved soil properties [27,28]. The mix of plant species facilitates land restoration, soil conservation, and carbon sequestration, and is superior to single plant species in terms of human production and biodiversity restoration and conservation. Forrester (2019) has also confirmed this in a related report [29].

The construction of artificial vegetation has largely improved the aeolian environment and soil quality conditions in the Hobq Desert, but there is less information on the evaluation of vegetation restoration in the transition area between the desert and the Yellow River. The monsoon climate and human activities have driven the spread and shrinkage of the desert, making this transitional region more sensitive to environmental change [30,31]. The area transformed from barren areas such as hills to desert in the last 30 years is approximately 910.84 km² [21]. Desertification has been effectively controlled following the implementation of the drought afforestation project. Nevertheless, changes in soil properties after revegetation, especially under different mixed vegetation types, remain unclear. In addition, the soil fractal dimension serves as a complementary information representing the physical and chemical properties of the soil and changes in desertification. Previous studies have also demonstrated that soil fractal dimensions are indicative of desertification [32]. However, changes in soil fractal dimension under different types of mixed vegetation have been less studied.

In this study, we investigated the vertical distribution of soil properties characteristics under four typical mixed vegetation types in the desert and the Yellow River coastal transition zone. The objectives of this study were (1) to study the differences and vertical distribution of soil properties under different mixed vegetation types, and to evaluate the most suitable types of vegetation restoration. (2) To investigate the relationship between the role of soil physical and chemical properties on fractal dimension.

2. Materials and Methods

2.1. Description of the Study Area

The study area is located in the northern part of the Hobq Desert in northern China (39°22′22′′~40°52′47′′ N, 106°55′16′′~109°16′08′′ E) (Figure 1). This is a typical temperate continental dry monsoon climate. The average annual temperature is 7.8 °C showing a gradual upward trend in recent years. The average annual precipitation is 188.4 mm, of which about 70% is between July and September. The average annual evaporation is 3155.95 mm. The study area has a high average wind speed and dry climate in spring throughout the year. Mean monthly precipitation, mean monthly temperature, and mean monthly relative humidity also reach their maximum between July and August (Figure 2). The winter and spring seasons in the region are characterized by dry weather and high wind and aeolian activity. Wind erosion causes loss of soil particles, which is a major cause of soil impoverishment. The original landforms in the study area are mobile sand dunes and moving sandy land, and the dunes are mainly barchan dunes and barchan dunes chains, which are more mobile and move faster. Sand-fixing artificial vegetation have been established over the years to mitigate wind and sand damage to towns and industrial parks as well as to the Yellow River basin.

2.2. Study Area Field Investigation

According to the records of the local forestry department. Artificial sand-fixing vegetation in the study area was established in 2000. The number of plant species is planted in a 1:1 ratio in mixed forests. We surveyed the field in August 2021. In this study, we selected fixed sandy areas that had been established with mixed vegetations of Populus alba var. pyramidalis × Caragana korshinskii (YN), Populus alba var. pyramidalis × Hedysarum mongdicum (YY), Populus alba var. pyramidalis × Hedysarum scoparium (YH), and Hedysarum scoparium × Salix cheilophila (HS) as the experimental group, and bare sand areas as the control group (LS). Classical survey methods were used to investigate vegetation features. At each site, three 50 × 50 m areas with five 1 m × 1 m subplots evenly distributed were selected for the survey. All sandy areas were under the same environmental conditions before the establishment of artificial sand-fixing vegetation. These vegetation types represented all habitat types in the area, and the area of each vegetation configuration pattern was large enough to have similar external environmental factors such as slope and elevation.

Table 1. Plant properties of five typical artificially mixed vegetation.

Vegetation Type	Major Plant SpeciesPlant Height (m)		Plant Density (ind m−2)	Understory Herbaceous Species	Vegetation Coverage (%)	Sandy Land Types
YN	Populus alba var. pyramidalis	3.7 ± 1.27	0.35 ± 0.44	Artemisia desertorum, Corispermum hyssopifolium, Glycyrrhiza uralensis, Setaria viridis, Parthenocissus tricuspidata, Bassia dasyphylla	50–60%	fixed sand land
	Caragana korshinskii	1 ± 0.24
YY	Populus alba var. pyramidalis	3.36 + 0.82	0.62 ± 0.47	Corispermum hyssopifolium, Setaria viridis, Parthenocissus tricuspidata, Bassia dasyphylla	45–55%
	Hedysarum mongdicum	1.00 ± 0.32
YH	Populus alba var. pyramidalis	3.4 ± 0.5	0.52 ± 0.32	Sophora alopecuroides, Corispermum hyssopifolium, Phragmites australis, Setaria viridis, Parthenocissus tricuspidata, Bassia dasyphylla	45–55%
	Hedysarum scoparium	1.07 ± 0.35
HS	Hedysarum scoparium	1.62 ± 0.43	0.15 ± 0.13	Corispermum hyssopifolium, Artemisia desertorum, Agriophyllum squarrosum, Bassia dasyphylla	40–50%
	Salix cheilophila	2.06 ± 0.36
LS				Agriophyllum squarrosum, Psammochloa villosa	<5%	moving sandy land

shows the base case of the five vegetation habitats.

2.3. Soil Sampling

Soil collection took place in August, 2021. We selected three similar 10 m × 10 m test plots as replicates in each selected mixed vegetation type, each more than 5 m apart, to avoid experimental chance (Figure 3). We randomly selected three soil sampling profiles within three selected similar 10 × 10 m areas. the intact soil samples were collected at three different depths: 0–10, 10–20, 20–30 cm. We also selected areas of bare sand for soil collection. A total of 45 soil samples were collected. Plant roots and plant residues were removed from the soil at each sampling site. Soil samples were placed in sample bags and taken back to the laboratory for air-drying and property measurements.

2.4. Physical and Chemical Analysis

We measured the particle size distribution of soil samples using an Analysette 22 NanoTec laser particle size meter (FRITSCH, Idar-Oberstein, Germany). Measurement range 0.01 to 2000 μm, accuracy better than 0.6%, accuracy/reproducibility better than 0.5% variable. We used the USDA soil size classification criteria, it is divided into clay (<2 μm), silt (2–50 μm), and sand (50–2000 μm). The soil samples were treated with organic matter removal and desalination prior to measurement. Soil water content (SWC) and soil bulk density (SBD) were collected from undisturbed soil with ring knife (diameter 5 cm, high 5 cm, total volume 100 m³), dried at 105 °C for 10 h, and determined by gravimetric method. All collected soils were air dried, and the alkaline hydrolysis nitrogen (AN), available phosphorus (AP), and available potassium (AK) contents were determined through a 1 mm soil sieve, and the soil organic carbon (SOC), total phosphorus (TP), total potassium (TK), and total nitrogen (TN) contents were determined through a 0.25 mm soil sieve. SOC was measured using the volumetric heating method with potassium dichromate. Total N was measured by the semi-micro-Kjeldahl method using a titration of 0.01 mol·L⁻¹ 1/2 H₂SO₄. AN was measured by the Conway diffusion dish method, using 1.0 mol·L⁻¹ NaOH to hydrolyze the soil and 20 g·L⁻¹ H₃BO₃ to absorb the NH₃ released by diffusion and titrated with 0.005 mol·L⁻¹ H₂SO₄ [34]. For total P analysis, soil subsamples were fused with NaOH and P concentrations were measured with the ammonium molybtate method in a U/V spectrophotometer [35]. Available P was extracted with 0.5 mol·L⁻¹ NaHCO₃, measured by Mo-Sb colorimetry, and quantified by spectrophotometry. Total K was determined in the fused soil samples with a flame photometer. Available K was extracted with 1 mol·L⁻¹ NH₄OAc (leaching) and its concentration was measured with a flame photometer.

2.5. Calculation of the Fractal Dimension (D) Values

We used the USDA soil size classification criteria, soil particles are divided into three grades: clay (<2 μm), silt (2–50 μm), and sand (50–2000 μm). Soil fractal dimension (D) values were calculated according to Tyler and Wheatcraft (1992) method using the following equation [36]:

$$\frac{V\left(r<R_i\right)}{V_T}={\left(\frac{R_i}{R_{max}}\right)}^{3-D}$$

(1)

$$\lg \left[\frac{V\left(r<R_i\right)}{V_T}\right]=\left(3-D\right)\lg \left[\frac{R_i}{R_{max}}\right]$$

(2)

where R_i is the average diameter of soil particles of a certain grade and R_max is the average diameter of soil particles. V(r < R_i) is the total volume of soil particles with a soil particle size smaller than R_i, V_T is the total volume of soil particles, and D is the fractal dimension of soil particle volume. In this study, we used a linear regression method to calculate soil D values.

2.6. Statistical Analysis

One-way ANOVA was conducted using SPSS (version 22.0) to determine if statistical significance existed and Duncan and Games-Howell multiple comparison probability levels of 0.05 were used to compare and contrast to determine significant differences in soil properties of different vegetation types and different soil depths. The Shapiro–Wilk test was used to check the normality of the variables before performing the data analysis. The variables that did not conform to the normal distribution were transformed by obtaining the natural logarithm, square root, and derivative of the data to make them conform to the normal distribution. Pearson correlation analysis (two-tailed test) was used to analyze the correlation between soil properties at the depths of the soils we studied. We used path analysis to further assess the direct and indirect effects of soil physical and chemical characteristics on D values. The graphs were created using the mapping software Origin 2021.

3. Results and Analysis

3.1. Soil Physical Property and D Values

The physical properties of the soil demonstrate the distribution of soil particles, the soil bulk density (SBD), and the soil water content (SWC) of our selected sample sites. The characteristics of soil physical properties of various artificially restored mixed vegetations differed to different degrees (

Table 2. Distribution of D values and physical properties of soil.

Soil Depth (cm)	Vegetation Types	SBD (g∙cm−3)	SWC (%)	D Value	D value Determination Coefficient (R2)
0–10	YN	1.35 ± 0.03 Bb	4.91 ± 0.05 Aa	2.39 ± 0.01 Aa	0.94
	YY	1.50 ± 0.05 Aa	3.54 ± 0.08 Cb	2.32 ± 0.01 Ba	0.94
	YH	1.53 ± 0.04 Aa	4.22 ± 0.19 Ba	2.24 ± 0.02 Ca	0.92
	HS	1.50 ± 0.04 Aa	4.29 ± 0.20 Ba	2.16 ± 0.02 Db	0.88
	LS	1.60 ± 0.07 Aa	3.43 ± 0.11 Ca	2.13 ± 0.01 Ea	0.88
10–20	YN	1.51 ± 0.03 Aa	4.94 ± 0.17 Aa	2.27 ± 0.01 Ab	0.90
	YY	1.57 ± 0.04 Aa	4.23 ± 0.10 Ba	2.23 ± 0.00 Bb	0.86
	YH	1.56 ± 0.06 Aa	4.89 ± 0.43 Aa	2.22 ± 0.02 Ba	0.87
	HS	1.51 ± 0.05 Aa	4.57 ± 0.26 ABa	2.19 ± 0.01B ab	0.85
	LS	1.56 ± 0.04 Aa	3.36 ± 0.17 Ca	2.07 ± 0.05 Ba	0.84
20–30	YN	1.51 ± 0.03 Aa	4.93 ± 0.15 Aa	2.27 ± 0.01 Ab	0.84
	YY	1.55 ± 0.04 Aa	0.92 ± 0.05 Dc	2.23 ± 0.01 Bb	0.85
	YH	1.58 ± 0.08 Aa	4.50 ± 0.31 Ba	2.22 ± 0.02B Ca	0.86
	HS	1.57 ± 0.05 Aa	4.19 ± 0.18 Ba	2.20 ± 0.01 Ca	0.85
	LS	1.55 ± 0.05 Aa	3.47 ± 0.07 Ca	2.10 ± 0.01 Da	0.86

Note: The values in the table are means ± SD (n = 3). Different uppercase letters indicate significant differences (p < 0.05, Duncan’s test) between different mixed vegetation types on each soil layer, and different lowercase letters indicate significant differences (p < 0.05) between different soil layers under the same mixed vegetation type. Abbreviations: soil water content (SWC), soil bulk density (SBD), fractal dimension values (D values). YN, YY, YH, and HS represent P. alba var. pyramidalis × C. korshinskii, P. alba var. pyramidalis × H. mongdicum, P. alba var. pyramidalis × H. scoparium, and H. scoparium × S. cheilophila mixed vegetation types, respectively. LS represents mobile sandy.

and

Table A1. Distribution of soil particles in different mixed vegetation types.

Soil Depth(cm)	VegetationTypes	Soil Particle Size Content (%)
		Clay	Silt	Very Fine Sand	Fine Sand	Medium Sand	Coarse Sand	Very Coarse Sand
0–10	YN	1.40 ± 0.14 Aa	15.73 ± 0.43 Aa	11.82 ± 0.21 Aa	59.65 ± 1.08 Bb	9.86 ± 0.24 Ca	0.91 ± 0.62Aa	0.64 ± 0.63 Aa
	YY	1.01 ± 0.05 Ba	7.54 ± 0.20 Ba	2.79 ± 0.03 Ba	60.72 ± 2.50 Bb	22.84 ± 1.53 Ba	2.43 ± 2.41Aa	2.65 ± 1.82 Aa
	YH	0.69 ± 0.08 Ca	4.64 ± 0.13 Ca	2.45 ± 0.07 Ca	70.38 ± 0.31 Ba	21.65 ± 0.27 Ba	0.01 ± 0.01Ab	0.17 ± 0.24 Ac
	HS	0.48 ± 0.05 Db	2.01 ± 0.08 Da	0.94 ± 0.04 Ea	71.88 ± 0.29 Ab	24.07 ± 0.25 Ba	0.08 ± 0.11Aa	0.49 ± 0.31 Aa
	LS	0.41 ± 0.03 Da	1.44 ± 0.04 Ea	1.20 ± 0.04 Da	58.72 ± 0.20 Bb	35.23 ± 0.02 Ab	0.84 ± 0.18Aab	2.14 ± 0.14 Aa
10–20	YN	0.93 ± 0.04 Ab	2.72 ± 0.04 Ab	3.28 ± 0.11 Bb	66.43 ± 2.45 Ac	17.99 ± 1.25 Ca	6.32 ± 2.43Aa	2.31 ± 1.03 Aa
	YY	0.78 ± 0.02 ABb	1.83 ± 0.04 Bb	1.51 ± 0.29 Ab	72.07 ± 0.19 Aa	21.75 ± 0.44 Ba	0.68 ± 0.14Aa	1.36 ± 0.10 Ba
	YH	0.75 ± 0.08 Ba	1.79 ± 0.05 Bb	1.67 ± 0.22 Bb	65.41 ± 2.06 Aa	21.69 ± 1.41 Ba	5.77 ± 1.08Aa	2.91 ± 0.11 Aa
	HS	0.63 ± 0.04 Ca	1.50 ± 0.06 Cb	0.89 ± 0.21 Aa	72.79 ± 2.34 Aab	22.78 ± 2.00 Ba	0.23 ± 0.25Aa	1.12 ± 0.41 Ba
	LS	0.33 ± 0.11 Da	0.99 ± 0.04 Db	0.44 ± 0.01 Cc	59.07 ± 0.31 Ab	36.67 ± 0.17 Aa	0.52 ± 0.10Ab	1.95 ± 0.10 Aa
20–30	YN	1.00 ± 0.04 Ab	2.00 ± 0.05 Ac	3.62 ± 0.37 Ab	76.97 ± 3.10 Aa	14.33 ± 2.40 Ba	0.97 ± 0.77Aa	1.08 ± 0.24 Aa
	YY	0.83 ± 0.05 Bb	1.78 ± 0.01 Bb	1.41 ± 0.23 Bb	72.32 ± 2.05 Aa	22.22 ± 2.30 Ba	0.35 ± 0.13Aa	1.06 ± 0.13 Aa
	YH	0.74 ± 0.09 BCa	1.71 ± 0.06 Bb	1.59 ± 0.24 Bb	67.18 ± 3.16 Ba	24.28 ± 1.40 Ba	2.34 ± 1.36Ab	2.12 ± 0.42 Ab
	HS	0.70 ± 0.04 Ca	1.47 ± 0.05 Cb	1.14 ± 0.17 Ba	76.00 ± 0.68 Aa	19.09 ± 0.90 Ba	0.37 ± 0.22Aa	1.20 ± 0.24 Aa
	LS	0.36 ± 0.02 Da	1.36 ± 0.09 Ca	0.55 ± 0.04 Cb	60.94 ± 0.13 Ca	33.76 ± 0.00 Ac	0.94 ± 0.15Aa	2.04 ± 0.03 Aa

Note: The values in the table are means ± SD (n = 3). Different uppercase letters indicate significant differences (p < 0.05) between different mixed vegetation types on each soil layer, and different lowercase letters indicate significant differences (p < 0.05, Duncan’s test) between different soil layers under the same mixed vegetation type. YN, YY, YH and HS represent P. alba var. pyramidalis × C. korshinskii, P. alba var. pyramidalis × H. mongdicum, P. alba var. pyramidalis × H. scoparium, and H. scoparium × S. cheilophila mixed vegetation types, respectively. LS represents mobile sandy.

). The clay, silt, very fine sand, and medium sand were significantly different among vegetation types (p < 0.05). The soil particles are finer than bare sandy land after vegetation construction, and the content of clay, silt, and very fine sand increases to 0.48–1.40%, 1.47–15.73%, and 0.94–11.82%, respectively, and the content of medium sand decreases to 9.86–24.28%. The distribution of soil particles had obvious hierarchical characteristics, and the content of clay, silt, and very fine sand was significantly higher in the 0–10 cm soil layer than in the 10–20 cm and 20–30 cm soil layers (p < 0.05). In the 0–10 cm soil layer, the order of clay, silt, and very fine sand contents of different vegetation types were YN > YY > YH > HS > LS, and the medium sand content showed LS > HS > YH > YY > YN, and similar trends were observed in the 10–20 cm and 20–30 cm soil layers. As the distribution of soil particles changed, the soil bulk density (SBD) also changed accordingly (Table 2). In contrast, the vegetation construction promoted a decrease in soil bulk density (SBD), and lower than that of bare sandy land, and the soil bulk density (SBD) roughly showed an increase with the increase of soil layer. Among them, the soil bulk density of YN under the 0–10 cm soil layer was significantly lower than the other sample plots (p < 0.05). The soil water content also differed significantly (p < 0.05) between the different artificial mixed vegetation types. D values were significantly different (p < 0.05) at different soil depths and with different mixed vegetation types. There was a significant increasing trend in soil D value after vegetation construction, and D value decreased with increasing soil depth. Overall, the variation in D values under varying types of vegetation ranged from 2.18–2.31 (

Table 3. Soil characteristics of different mixed vegetation types.

Soil Properties	Vegetation Types					Mean	Coefficient of Variation
	YN	YY	YH	HS	LS
SOC (g∙kg−1)	1.59 ± 0.42 a	1.20 ± 0.27 a	0.72 ± 0.13 b	0.79 ± 0.04 b	0.61 ± 0.07 b	0.96 ± 0.43	44.79
AN (mg∙kg−1)	19.83 ± 3.45 a	16.22 ± 3.33 ab	12.21 ± 5.37 b	17.03 ± 2.14 a	5.87 ± 2.00 c	13.85 ± 6.03	43.54
AP (mg∙kg−1)	2.01 ± 0.44 a	1.15 ± 0.27 b	1.04 ± 0.16 b	0.96 ± 0.06 b	0.84 ± 0.11 b	1.20 ± 0.49	40.83
AK (mg∙kg−1)	101.78 ± 10.58 a	84.22 ± 4.18 b	78.00 ± 7.45 b	81.67 ± 9.89 b	78.56 ± 1.64 b	84.84 ± 11.70	13.79
TN (g∙kg−1)	0.18 ± 0.08 a	0.17 ± 0.01 a	0.13 ± 0.02 b	0.14 ± 0.01 b	0.08 ± 0.00 c	0.14 ± 0.05	35.71
TP (g∙kg−1)	0.36 ± 0.07 a	0.37 ± 0.06 a	0.35 ± 0.03 a	0.36 ± 0.04 a	0.35 ± 0.02 a	0.36 ± 0.05	13.89
TK (g∙kg−1)	35.71 ± 1.11 a	25.22 ± 6.20 b	24.73 ± 2.39 b	27.47 ± 2.07 b	21.07 ± 2.70 b	26.84 ± 5.94	22.13
C/N	9.35 ± 1.34 a	6.82 ± 1.18 b	5.66 ± 1.21 c	5.84 ± 0.53 bc	8.05 ± 0.91 a	7.04 ± 1.62	23.01
C/P	4.28 ± 0.46 a	3.17 ± 0.33 b	2.06 ± 0.36 c	2.23 ± 0.25 c	1.75 ± 0.12 c	2.65 ± 0.94	35.47
N/P	0.47 ± 0.12 a	0.47 ± 0.05 a	0.37 ± 0.06 b	0.38 ± 0.05 b	0.22 ± 0.02 c	0.38 ± 0.12	31.58
D	2.31 ± 0.06 a	2.26 ± 0.04 a	2.23 ± 0.02 a	2.18 ± 0.02 b	2.10 ± 0.04 c	2.22 ± 0.08	3.60
SWC (%)	4.93 ± 0.13 a	2.90 ± 1.43 b	4.54 ± 0.43 a	4.35 ± 0.27 a	3.42 ± 0.13 b	4.03 ± 1.01	25.19
SBD (g∙cm−3)	1.46 ± 0.08 b	1.54 ± 0.05 a	1.56 ± 0.06 a	1.53 ± 0.05 a	1.57 ± 0.06 a	1.53 ± 0.07	4.82
Clay (%)	1.11 ± 0.24 a	0.87 ± 0.11 a	0.73 ± 0.09 a	0.61 ± 0.10 a	0.36 ± 0.085 b	0.74 ± 0.28	37.89
Silt (%)	6.82 ± 6.70 a	3.72 ± 2.87 a	2.72 ± 1.45 a	1.66 ± 0.27 a	1.26 ± 0.21 b	3.27 ± 3.71	113.38
Sand (%)	92.06 ± 6.91 b	95.39 ± 2.97 b	96.54 ± 1.42 b	97.69 ± 0.22 b	98.34 ± 0.27 a	95.96 ± 3.92	4.08

Note: Values in the table for soil properties in the 0–30 cm soil layer, are mean ± SD (n = 9). Different letters indicate significant differences (p < 0.05, Games–Howell’s test) among different mixed vegetation types. Abbreviations: soil organic carbon (SOC), total phosphorus (TP), total potassium (TK), total nitrogen (TN), alkaline hydrolysis nitrogen (AN), available phosphorus (AP), available potassium (AK), carbon/nitrogen ratio (C/N), carbon/phosphorus ratio (C/P), nitrogen/phosphorus ratio (N/P), soil water content (SWC), soil bulk density (SBD). YN, YY, YH and HS represent P. alba var. pyramidalis × C. korshinskii, P. alba var. pyramidalis × H. mongdicum, P. alba var. pyramidalis × H. scoparium, and H. scoparium × S. cheilophila mixed vegetation types, respectively. LS represents mobile sandy.

) and were all significantly higher than LS (2.10 ± 0.04) (p < 0.05). In addition, these variances in soil water content, sand content, and D value were weak, with coefficients of variation of 4.82%, 4.08%, and 3.65%, respectively. Clay and soil bulk density (SBD) were of moderate variability with coefficients of variation of 37.89% and 25.19%, respectively. However, silt showed a strong degree of variability with a coefficient of variation of 113.38%. It is clear that the establishment of mixed vegetation types is important for the improvement of soil texture.

3.2. Soil Nutrient Distribution

As the physical characteristics of the soil improved, the soil nutrient content was enriched substantially. Soil nutrient contents also differed to varying degrees under different mixed vegetation types, with significant differences in soil nutrient contents at different types of sites, except for soil TP and AP (Table 3 and Figure 4). The soil nutrient contents of the YN sample plots reached the maximum. Except for soil TN, TP, N/P, and SOC, other soil nutrient contents were significantly higher in YN sample plots than in other mixed vegetation types (p < 0.05). After revegetation, soil AN and TN contents were significantly higher than LS in each mixed vegetation type (p < 0.05), and soil AN and TN increased from 5.87 mg∙kg⁻¹ and 0.08 g∙kg⁻¹ to 12.21–19.83 mg∙kg⁻¹ and 0.13–0.18 g∙kg⁻¹, respectively. Soil AP, AK, TP, and TK contents showed no significant difference from LS in YY, YH, and HS sample plots (p > 0.05).

The distribution of soil nutrient content under different mixed vegetation types has obvious hierarchical characteristics. The distribution of soil nutrient content was 0–10 cm > 10–20 cm > 20–30 cm in order, which reflected the surface enrichment effect of soil nutrient content. The soil nutrient content of each mixed vegetation type showed different variability in different soil layers (Figure 4). The order of individual soil nutrient contents in the 0–10 cm soil layer was roughly YN > YY > YH > HS > SL, and similar trends were observed in the 10–20 cm and 20–30 cm soil layers. There were significant differences (p < 0.05) in the SOC and AN contents of soil of different vegetation types in each soil layer. After the establishment of mixed vegetation, the increase in SOC content of each soil layer was 1.18–3.22, 1.48–2.39, and 1.02–1.65 times, while the increase in AN content was 3.51–4.72, 2–3.70, and 1.17–2.16 times, respectively. The improvement effect of soil TN, C/N, and C/p contents after revegetation was small in the 20–30 cm soil layer, and there was no significant difference between different vegetation types (p > 0.05). All soil nutrient contents were of moderate variability, with higher coefficients of variation of 44.79%, 43.54%, 40.83%, and 35.71% for SOC, AN, AP, and TN contents, respectively.

3.3. Relationship between Soil Properties

The Pearson correlation test showed a link between the physical characteristics of the soil and the chemical properties of the soil (Figure 5). There was a highly significant (p < 0.01) or stronger significant positive correlation (p < 0.001) between soil nutrient content and soil clay and silt content, and a highly significant (p < 0.01) or stronger significant negative correlation (p < 0.001) with sand content. The correlation coefficient between soil bulk density and soil nutrients was low, but still showed a highly significant (p < 0.01) or stronger significant negative correlation (p < 0.001). Soil bulk density (SBD) was also strongly and significantly negatively correlated with clay and silt content (p < 0.001), while a strong and significant positive correlation existed with sand content (p < 0.001). Soil physical and chemical characteristics are closely related to D values. Soil D values showed varying degrees of significant positive correlation with soil nutrients. Soil D values were more strongly and significantly positively correlated with clay and silt, and showed a stronger and significantly negative correlation with sand and soil bulk density (SBD).

To further show the influence of soil physical and chemical characteristics on D values, path analysis was used to determine the direct and indirect effects of independent variables (e.g., soil physical and chemical parameters) on the dependent variable (D value). We tested the dependent variable (D value) for normal distribution and the significance was all greater than 0.05, indicating that the distribution of D values obeyed a normal distribution (

Table 4. Regression model test results.

	Kolmogorov-Smirnova			Shapiro-Wilk			R2
	Statistics	df	P	Statistics	df	P
D Value	0.098	45	0.200	0.975	45	0.429	0.953

). The parameters of soil physical and chemical parameters that have a significant influence on the D values through path analysis of soil properties are soil AN content and clay content (

Table 5. Linear regression analysis of parameters with significant contribution to D values.

Soil Properties	Direct Path Coefficient	t	P
SOC	0.03	0.25	0.81
AN	0.16	2.21	0.03
AP	−0.02	−0.25	0.80
AK	−0.04	−0.80	0.43
TN	-0.03	−0.26	0.80
TP	0.09	1.50	0.14
TK	−0.01	−0.14	0.89
SWC	0.03	0.80	0.43
SBD	0.06	1.20	0.24
Clay	0.92	10.52	0.00
Sand	0.03	0.26	0.80

Abbreviations: soil organic carbon (SOC), total phosphorus (TP), total potassium (TK), total nitrogen (TN), alkaline hydrolysis nitrogen (AN), available phosphorus (AP), available potassium (AK), carbon/nitrogen ratio (C/N), carbon/phosphorus ratio (C/P), nitrogen/phosphorus ratio (N/P), soil water content (SWC), soil bulk density (SBD).

). The path analysis shows that the D value is mainly influenced by the AN and clay. Clay content and AN have a positive effect on D values with direct path coefficients of 0.92 and 0.16 respectively (

Table 6. Path coefficient between soil parameters and D values.

Independent Variable	Direct Path Coefficient	Indirect Path Coefficients
		AN	Clay	Total
AN	0.16		0.57	0.57
Clay	0.92	0.10		0.10

). Clay and AN also have an indirect positive effect on D values. It is worth noting that the indirect effect of clay and AN on D values is lower than the direct effect, and that clay have a higher positive effect. As mentioned above, clay and AN had a positive effect on soil D values, with clay having a much greater effect on D values than AN (Table 6).

4. Discussion

Soil and vegetation are two essential components of terrestrial ecosystems [37]. Land use change and agricultural activities have caused degradation of soil quality and the building up of vegetation has contributed to the accumulation of soil organic matter, increased soil fertility, and improved soil structure [38,39]. Wind erosion causes loss of soil particles, which is another major cause of soil impoverishment. Sterk et al. (1996) also reports that the loss of soil C, N, and P elemental content due to wind erosion under strong storms is very significant by means of sample plot estimation [40]. In a similar report, Shao also demonstrated that soil nutrients and structure are influenced by vegetation type and human intervention under the same climatic and biological conditions [41]. Changes in vegetation type have altered the distribution pattern of soil nutrients, which in previous studies have contributed to soil nutrient content through root exudates and vegetation litters [42]. Differences in the quantity and quality of root exudates from different plants and the rate of decomposition of litters from different vegetation are important factors in the differences in soil nutrients [43,44]. However, the factors of drought and frequent aeolian activity in this study area have resulted in low litters, and the soil nutrient content is mainly secreted by the plant roots. The mixed vegetation type of YN has a higher content of all soil nutrients than the other mixed vegetation types and has a higher soil SOC and N content. The high growth rate and nitrogen fixation capacity of C. korshinskii [45,46] and the high vegetation cover of the area (Table 1) are important factors in protecting the soil from erosion and enrichment of soil SOC and N content. The distribution of soil nutrient content under different mixed vegetation types shows 0–10 cm soil layer > 10–20 cm soil layer > 20–30 cm soil layer, reflecting the superficial effect of soil nutrient content distribution. Zhang et al. (2019) mentions that the nutrient content of the surface layer of the soil is higher than that of the deeper layers, a phenomenon that may be related to the higher biological activity and root penetration of the surface layer in his report [47]. In addition, the construction of the vegetation increases the aerodynamic roughness of the surface, which helps to protect the soil fine particle content, and also traps soil particles and provides a stable microclimate. Soil particles are finer under different vegetation types than in bare sandy areas, with clay, silt, and very fine sand content increasing to 0.48–1.40%, 1.47–15.73%, and 0.94–11.82% respectively. The accumulation of clay and silt content contributed to the increase in soil nutrient content [48], which resulted in clay, silt, and soil nutrient content showing a highly significant (p < 0.01) or stronger significant positive correlation (p < 0.001) (Figure 5). In addition, soil bulk density and soil moisture content can also reflect soil texture, as there is an interdependence between the physical and chemical properties of the soil [49,50]. The distribution of the physical properties of the soil of the different mixed vegetation types is therefore broadly similar to the distribution pattern of soil nutrients. Sand and soil bulk density showed either a highly significant (p < 0.01) or a stronger significant negative (p < 0.001) correlation with soil nutrient content.

The fractal dimension of a soil reflects the distribution of soil particles or soil properties, and it is often used to assess changes in the soil environment. Wind-induced loss of soil fine fractions causes damage to soil structure, loss of nutrients in aeolian environment. After vegetation reconstruction, the content of clay and silt increased, while the content of sand decreased (Appendix A, Table A1). D value and clay, silt content showed a stronger significant positive correlation. There is a stronger negative correlation between D value and sand content [51]. Similar findings were reported by Feng et al. (2019): an increase in the clay and silt content and a decrease in the sand content indicated a richer soil nutrient content. The increase in clay and silt content and the decrease in sand content indicated an increase in D values [52]. Soil nutrient content and D value showed varying degrees of significant positive correlation, while soil bulk density and D value showed a stronger significant negative correlation. This shows that the distribution pattern of soil D values is generally consistent with the distribution pattern of soil nutrients and physical structure. Feng also provides an explanation for the increase in soil D values with increasing soil depth. The vertical variability of the soil environment is characterized by complex factors such as soil evolution, plant cover, decomposition of litters, and wind erosion and deposition [52]. In addition, we used path analysis to assess the direct and indirect effects of soil physical and chemical characteristics on D values. AN and clay content have different direct and indirect effects on D value. Soil nutrients and clay interact with each other, with a highly significant or stronger significant correlation. The direct path coefficients for clay and AN on soil D values were 0.92 and 0.16, indicating that clay had a greater effect on D values than AN (Table 6). The indirect path coefficient indicates that a change in one indicator of the independent variable causes a change in another indicator of the independent variable, which ultimately leads to a change in the dependent variable. AN and clay have different indirect effects on soil D values. As mentioned above, the accumulation of fine soil particles and soil nutrients indicates an increase in D values. In addition, the promotion of soil nutrient accumulation by fine soil particles can also increase D values. According to our research, the planting of vegetation adapted to arid, aeolian environment with a rapid rate of nitrogen fixation is conducive to the reversal of desertification. The scientific conservation and management of artificial vegetation in desert ecosystems and the ongoing vegetation construction projects are equally important.

5. Conclusions

Vegetation is vital to desert ecosystems and has an important role to play in the restoration of desert soil. The mixed vegetation patterns of Populus alba var. pyramidalis × Caragana korshinskii, Populus alba var. pyramidalis × Hedysarum mongdicum, Populus alba var. pyramidalis × Hedysarum scoparium and Hedysarum scoparium × Salix cheilophila have been widely used in the desertified areas of the northern arid zone of China. According to our study, vegetation type positively influenced the physical and chemical properties of our selected soil, with vegetation type being more sensitive to improvements in soil organic carbon (SOC), alkaline hydrolysis nitrogen (AN), total nitrogen (TN), clay and silt content, with coefficients of variation of 44.79%, 43.54%, 35.71%, 37.89%, and 113.38% respectively. The improvement in the physical structure of the soil and the accumulation of soil nutrient content are closely related, with soil bulk density, sand, clay, and silt showing highly significant or more strongly significant relationships with each of the soil nutrient indicators. Different mixed vegetation types have good positive effects on soil physical properties and nutrient content, with mixed vegetation type of Populus alba var. pyramidalis × Caragana korshinskii having a better effect than other mixed vegetation types in increasing soil C, N, P content and improving soil structure. Compared to the mobile sandy land, organic carbon (SOC), available phosphorus (AP), available potassium (AK), alkaline hydrolysis nitrogen (AN), total nitrogen (TN), total potassium (TK), clay, and silt increased by 161%, 238%, 139%, 30%, 125%, 69%, 208%, and 441% respectively. Therefore, we consider Populus alba var. pyramidalis × Caragana korshinskii mixed vegetation to be a suitable vegetation restoration type for the area. In addition, changes in the physical and chemical properties of the soil ultimately led to changes in fractal dimension values. We further demonstrated that soil AN and clay content mainly influenced the variation in fractal dimension values. The higher soil AN and clay content contributed to the increase in fractal dimension values. This has important implications for the reversal of desertification levels. Therefore, the scientific evaluation of artificial vegetation construction project is helpful to the ecological restoration of desert ecosystem.