The Relationships between Root Traits and the Soil Erodibility of Farmland Shelterbelts in the Bashang Region of China

Abstract

Soil erodibility by wind is not only affected by the basic physical and chemical properties of the soil but also the functional traits of plant roots. However, the roles played by the morphological and architectural traits of plant roots on wind-based soil erodibility in the Bashang region of China are still unclear. Therefore, two typical tree shelterbelts and two shrub shelterbelts in the Bashang region were selected to assess and determine how the root traits affected soil erodibility, especially characteristics such as dry aggregate, soil organic matter, and shearing resistance. The results showed that the soil dry aggregates of the two shrubs (Lycium barbarum and Caragana korshinskii) had higher geometric mean diameters (0.40 ± 0.03 mm) and mean weight diameters (0.82 ± 0.08 mm) but a lower erodible fraction (81.81% ± 1.62%) compared to the two trees (Populus simonii and Ulmus pumila). The mean weight diameter (MWD_d) and geometric mean diameter (GMD_d) of dry soil aggregates were negatively correlated with the soil erodible fraction (EF_d), but these parameters were positively correlated with shearing resistances. The specific root length (SRL) and surface area (SSA) of plant roots were positively correlated with the GMD_d of the soils, though these two parameters negatively correlated with the soil erodible fraction. The root branching intensity (BI) was negatively correlated with the MWD_d and GMD_d of dry soil aggregates. The total carbon or nitrogen of the soil displayed significantly positive and negative correlations to the geometric mean diameters and erodible fractions of the soils, respectively. The findings showed that plant roots with higher SRLs, as well as lower root diameters and BIs, played positive key roles in soil stability. The same applied to soils with higher nitrogen, carbon, and water content. The results from this study suggest that L. barbarum is superior to the other three species based on root traits and wind erosion resistance. These findings provide critical information for selecting plants for the sustainable management of windbreak and sand fixation.

1. Introduction

Wind erosion is a soil degradation process that frequently occurs in arid and semi-arid regions [1,2]. Soil erodibility refers to the extent to which soils can be readily washed away by erosion factors such as wind. Wind-related soil erodibility mainly depends on wind speed [3], soil moisture [4], soil intrinsic properties (such as soil porosity, organic matter, and aggregates) [5,6], and vegetation [7,8].

Researchers have conducted a series of wind tunnel and field tests and put forward some indicators to effectively characterize soil erodibility by wind. These indicators include the particle size composition and stability of the soil dry aggregates [9,10,11], soil shear strength [12], soil crust [13,14]. Due to the high cost of large-scale field measurements of the index data of soil erodibility by wind erosion, researchers began to use the basic physical and chemical properties of the soil (texture, organic matter, and calcium carbonate content) to predict the wind erosion erodibility index [11,15,16]. This led to the establishment of the intrinsic properties of the soil and wind erosion index model equation [17,18].

Some studies have indicated that stems and leaves of vegetation could effectively reduce wind strength against the soil in addition to holding sand [19,20,21]. It has also been reported that the rate of soil erosion decreases with increased grass coverage [22]. Despite the difficulty in excavating plant roots, increasing attention has been paid to these plant parts due to their contributions to soil erosion phenomena. Roots mainly reduce soil erodibility through the physical binding effect and the biochemical exudate-bonding effect [23,24], both of which are closely related to the mechanical and biological traits of roots. Some studies indicated that higher root densities had a more significant influence on reducing soil detachment [25,26]. This may reduce the net repulsive forces in soil particles, thereby enhancing the integrity of the aggregates [27,28]. Root diameters showed a negative power relationship with tensile strength, while showing a positive power relationship with the tensile force of herbs on the Loess Plateau [29,30]. However, the relationship between root traits and soil aggregate stability has shown inconsistent conclusions. For example, soil stability was positively [24,31,32] or negatively [33,34] correlated with root biomass. On the other hand, positive [34,35,36] or negative [32,37] correlations were reported between specific root length (SRL) or root length density and soil aggregate stability (mean weight diameter). But a study reported that morphological traits of roots, such as the SRL and diameter, were poorly associated with soil stability [38]. These controversial results may be caused by the different types of vegetation in the various studies that were conducted.

Hao et al. (2021) found that root architecture (root fractal dimension, topological index, and radius frequency distribution function) caused insignificant changes in soil texture in a greenhouse experiment with Cynodon dactylon [35]. However, the correlation between the branching ratio (or intensity) of root architectural traits and soil aggregate stability present a gap in the research. Moreover, just a few studies have investigated the impacts of woody plants with different root traits on the resistance of soils to wind erosion in the Hebei Bashang region. In fact, most results focus on the Loess Plateau of China. Therefore, understanding the role of root morphology and architecture on soil erodibility by the wind in the Hebei Bashang region is of paramount significance.

Hebei Bashang is an important agro-pastoral ecotone in northern China, and it is also one of the main dust sources in the Beijing–Tianjin area, which suffers from severe wind erosion [16,39]. The farmland shelterbelts in the Bashang area were built as part of the “three north shelter forest” project in the 1960s and during the Beijing and Tianjin sandstorm source control project of the early 21st century in a bid to reduce wind-based soil erosion and improve the ecological environment [40]. With the degradation and depression of poplar windbreaks and sand fixation, considering the complementarity of species functional traits to match reasonable collocation of forest stands is crucial. Different species with varying root morphological and architectural traits may display different effects on soil erodibility. Therefore, two typical tree shelterbelts and two shrub shelterbelts in the Bashang region in Hebei Province were selected to determine the relationship between root traits and soil erodibility. Based on the controversial correlations between root traits and soil stability [32,34,35], the aims of this study were to evaluate (1) the soil aggregate stability of different shelterbelts in the Bashang region, China, and (2) the correlations between the morphological and architectural traits of plant roots and soil erodibility in the Bashang region of China. We hypothesized that (1) shrub soils would show lower erodibility than tree soils; and (2) plants with smaller root diameters, higher root specific lengths, and lower branches would show positive effects on soil aggregate stability.

2. Materials and Methods

2.1. Study Site and Experimental Design

The study was carried out in the wind erosion observation field of Kangbao County (41°25′–42°08′ N, 114°11′–114°56′ E) in Bashang District, Hebei Province. The area is located in the middle of the agro-pastoral ecotone of northern China, which is the key area of the Beijing–Tianjin–Hebei sandstorm source control project. The region has a temperate continental climate. The annual average precipitation and wind speed are 338.50 mm and 2.99 m/s, respectively. The wind is at its strongest intensity from March to May, with the monthly average wind speed reaching 3.48 m/s. The soil type is a chestnut soil (a mollisol in the United States Department of Agriculture classification) with low organic matter content and a high erosion risk. The mainland can be categorized into cultivated, grassland, forest, and shrubland. Due to long-term overgrazing and irrational land use, wind erosion of soil in this area has progressively aggravated to serious levels. Two typical trees (Populus simonii and Ulmus pumila) and two typical shrubs (Lycium barbarum and Caragana korshinskii) in farmland shelterbelts were selected as study sites. The investigated shelterbelts had the characteristics of flat and open terrain, with large farmlands both before and after the shelterbelt. The farmlands were planted with Hulless oat and Oilseed rape. The sizes of the Lycium, Caragana, Populus, and Ulmus shelterbelts were 500 m (length) × 3 m (width), 5000 m (length) × 8 m (width), 1000 m (length) × 10 m (width), and 1800 m (length) × 25 m (width), respectively. The direction of all shelterbelts was north–south. More information about the shelterbelts is shown in Figure 1 and

Table 1. The basic characteristics of farmland shelterbelts. Diameter means diameter at breast height for trees, and basal diameter for shrubs.

Shelterbelt Types	Abbreviation	Latitude and Longitude	Altitude (m)	Age (Year)	Height (m)	Diameter (cm)	Length of Shelterbelt (m)	Width of Shelterbelt (m)	Line Number	Spacing of Line (m)	Spacing of Trees/Shrubs (m)	Farmland
Lycium barbarum	LB	114°44′13″ E, 41° 55′51″ N	1448	15	2.24 ± 0.16	4.17 ± 0.93	500	3	3	0.1	0.1	Hulless oat and Oilseed rape
Caragana korshinskii	CK	114°48′11″ E, 42°8′5″ N	1283	17	0.89 ± 0.13	1.30 ± 0.15	5000	8	3	0.1	0.1	Hulless oat
Populus simonii	PS	114°44′12″ E, 42°1′21″ N	1445	29	8.10 ± 0.41	25.29 ± 2.16	1000	10	4	2.5	1.0	Hulless oat and Oilseed rape
Ulmus pumila	UP	114°47′22″ E, 42°7′1″ N	1312	36	6.94 ± 0.53	21.16 ± 1.96	1800	25	10	2.5	2.0	Hulless oat

. Understory herbs in the Ulmus shelterbelts included Digitaria sanguinalis, Kochia scoparia, Thalictrum aquilegiifolium, Artemisia scoparia, Setaria viridis, Allium ramosum. Understory herbs in the Populus shelterbelts were Digitaria sanguinalis, Thalictrum aquilegiifolium, Potentilla chinensis, Astragalus adsurgens, Artemisia argyi, Saposhnikovia divaricate. The coverages of herbs under Ulmus and Populus shelterbelts were about 10%. There was no other co-occurring species in the Caragana and Lycium shelterbelts.

Intact fine root segments were sampled following the method described by Kou et al. (2015) and Yan et al. (2019) [41,42]. Four plots of 50 cm (length) × 50 cm (width) × 10 cm (depth) were cut in each shelterbelt using a machete, after removing the floor litter. The harvested soil plots were immediately placed in a plastic bag with ice for temperature control and the samples were transported to the laboratory. In the laboratory, the intact fine root samples were gently separated from the soil using tweezers and placed in deionized water to remove the attached soil and organic matter particles. The roots were stored at 4 °C and their morphology was determined immediately. The separated soils were stored at −20 °C before various physical and chemical properties were measured. The shear strength of the surface soil was measured in situ with four different mass weights (0.5 kg, 1.0 kg, 1.5 kg, and 2.0 kg) using a portable soil direct shear instrument [43]. The four different mass weight were loaded on the shear box to meet the requirements of different levels of normal stress, corresponding to four levels of normal stress (5 N, 10 N, 15 N, and 20 N). The shear rate was set at 2.5 mm/s.

2.2. Root Trait Measurements

The roots were dissected by branch order [44] in a dish with distilled water, spread without overlap in a thin layer of water in a glass container, and scanned using an Epson Expression 10,000 XL desktop scanner with a resolution of 300 dpi. Roots were dried at 65 °C for more than 48 h. The root diameter, total length, surface area, and volume were obtained through image analysis using the W_INR_HIZO Arabido v.2012b (Regents Instruments Inc., Quebec City, QC, Canada). The number of root segments of each root order was obtained by the person through counting scanned pictures. The specific root length (SRL) and specific surface area (SSA) were calculated as the total root length and the surface area divided by the dry mass for each root order. The root tissue density (RTD) was calculated as the ratio of the root dry mass to its volume. The root branching ratio (BR) was determined by calculating the number of first-order roots divided by the number of second-order roots. The root branching intensity (BI) was calculated as the number of first-order roots divided by the total root length of second-order roots. The absorptive roots were used as first- and second-order roots in this study because they were below-ground resource-acquiring units [45].

2.3. Soil Trait Measurements

The soil pH was measured in a soil–water suspension (1:2.5 v/v) using a digital pH meter. The water content of the soil was measured using the aluminum box drying method. Inorganic carbon was removed by adding hydrochloric acid prior to determining the content of organic carbon and total nitrogen in the dried soils using an elemental analyzer (EA 3000, Eurovector, Pavia, Italy). Soil organic matter (SOM) was determined by multiplying the soil organic carbon content by a coefficient of 1.724. The air-dried soil (500 g) was screened for dry aggregates of different particle sizes for five minutes using a vibrating screen machine (SZS, Soil Instrument Factory, Nanjing, China). The sizes of the aggregates were separated as follows: <0.106 mm, 0.106 to 0.25 mm, 0.25 to 0.35 mm, 0.35 to 0.425 mm, 0.425 to 0.85 mm, 0.85 to 2 mm, 2 to 5 mm, and >5 mm. The weights of soil dry aggregate samples with different particle sizes after sieving were measured. The aggregate sizes that were coarser than 0.85 mm were independently sieved for a second time.

The aggregates that were larger than 0.85 mm in diameter were non-erodible [9,46]. The percentage of aggregates whose diameter was less than 0.85 mm, which is the erodible fraction (EF_d) of the dry soil, was calculated using the following equation [10,47]:

$${\mathrm{E}\mathrm{F}}_{\mathrm{d}}\left(\mathrm{\%}\right)=\frac{W_{<0.85}}{T}\times 100\mathrm{\%}$$

(1)

where EF_d is the erodible fraction, W_<0.85 is the weight of aggregates of <0.85 mm from the first sieving (g), and T is the initial weight (g) of the total sample. The dry aggregate stability of the soil (DAS) was estimated using the following equation [10]:

$$\mathrm{D}\mathrm{A}\mathrm{S}=\frac{{\left(W_{>0.85}\right)}_1-{\left(W_{<0.85}\right)}_2}{{\left(W_{>0.85}\right)}_1}\times 100$$

(2)

where (W_>0.85)₁ is the total weight of the aggregates retained by the 0.85 mm sieve after the first sieving, and (W_<0.85)₂ is the weight of the aggregates of >0.85 mm that passed the 0.85 mm sieve after the second round of sieving. The mean weight diameter (MWD_d, mm) and the geometric mean diameter (GMD_d, mm) of soil dry aggregates were calculated using the following equations [9]:

$${\mathrm{M}\mathrm{W}\mathrm{D}}_{\mathrm{d}}=\sum _{i=1}^n\overline{r_i}\times m_i$$

(3)

$${\mathrm{G}\mathrm{M}\mathrm{D}}_{\mathrm{d}}=\exp (\sum _{i=1}^n{m}_i\times \ln \overline{r_i})$$

(4)

where $\overline{r_i}$ (mm) is the mean diameter of dry aggregates (average of r_i and r_i+1), m_i is the proportion of the dry aggregate weight in the corresponding size fraction, and n is the number of size fractions. The aggregate characteristics of EF_d, DAS, MWD_d, and GMD_d related to wind erosion evaluation were determined using the data that were obtained from the dry sieving method.

Another method for obtaining the EF is by using the multiple regression equation that is derived from intrinsic soil properties [15]. This method is widely used in many wind erosion modeling studies [48,49]. Fresh soil was sieved at 2 mm, and its particle size composition was determined using a laser particle size diffractometer (Mastersizer 3000, Malvern, Worcestershire, UK). Gas volumetric analysis was employed for measuring CaCO₃ content. The erodible fraction (EF_w) and the geometric mean diameter (GMD_w) were determined using the regression equations [15,16]:

$${\mathrm{E}\mathrm{F}}_{\mathrm{w}}(\mathrm{\%})=29.09+0.31\times \mathrm{s}\mathrm{a}\mathrm{n}\mathrm{d}+0.17\times \mathrm{s}\mathrm{i}\mathrm{l}\mathrm{t}+0.33\times \frac{\mathrm{s}\mathrm{a}\mathrm{n}\mathrm{d}}{\mathrm{c}\mathrm{l}\mathrm{a}\mathrm{y}}-2.59\times \mathrm{S}\mathrm{O}\mathrm{M}-0.95\times {\mathrm{C}\mathrm{a}\mathrm{C}\mathrm{O}}_3$$

(5)

$${\mathrm{G}\mathrm{M}\mathrm{D}}_{\mathrm{w}}=\exp (1.343-2.235\times \mathrm{s}\mathrm{a}\mathrm{n}\mathrm{d}-1.226\times \mathrm{s}\mathrm{i}\mathrm{l}\mathrm{t}-0.0238\times \frac{\mathrm{s}\mathrm{i}\mathrm{l}\mathrm{t}}{\mathrm{c}\mathrm{l}\mathrm{a}\mathrm{y}}+33.6\times \mathrm{S}\mathrm{O}\mathrm{M}+6.85\times {\mathrm{C}\mathrm{a}\mathrm{C}\mathrm{O}}_3)\times \left(1.0+0.006\times \mathrm{l}\mathrm{a}\mathrm{y}\mathrm{e}\mathrm{r}\mathrm{d}\mathrm{e}\mathrm{p}\mathrm{t}\mathrm{h}\right)$$

(6)

where sand is the soil sand (0.05–2.0 mm) content, silt is the soil silt (0.002–0.05 mm) content, clay is the soil clay (<0.002 mm) content, SOM is the organic matter content, and CaCO₃ is the calcium carbonate content. All variables were expressed as percentages. The layer depth was 10 cm in this case. The meanings of all the abbreviations are shown in

Table 2. Abbreviations and descriptions of soil characteristics and root morphological and architectural traits.

Parameters		Abbreviation	Units	Description
Soil characteristics	Soil water content	SWC	%	Soil water content
	Soil total nitrogen	STN	g kg−1	Soil total nitrogen
	Soil total organic carbon	STC	g kg−1	Soil total organic carbon
	Mean weight diameter of dry aggregates	MWDd	mm	Mean weight diameter of soil dry aggregates using the dry sieving method
	Geometric mean diameter of dry aggregates	GMDd	mm	Geometric mean diameter of soil dry aggregates using the dry sieving method
	Erodible fraction of dry aggregates	EFd	%	Erodible fraction of the dry aggregates using the dry sieving method
	The dry aggregate stability	DAS	%	The dry aggregate stability using the dry sieving method
	The erodible fraction	EFw	%	The erodible fraction of soils using regression equations
	Geometric mean diameter	GMDw	mm	The geometric mean diameter of soils using regression equations
	Shear strength with 0.5 kg	SR0.5	J	Shear strength of the surface soil when mass weight was 0.5 kg
	Shear strength with 2.0 kg	SR2	J	Shear strength of the surface soil when mass weight was 2.0 kg
Root morphological traits	Average root diameter	RD	mm	Average root diameter
	Specific root length	SRL	m g−1	The ratio of root cluster length to root dry mass
	Specific surface area	SSA	cm2 g−1	The ratio of root surface area to root dry mass
	Root tissue density	RTD	g cm−3	The ratio of root dry mass to root volume
Root architectural traits	Branching ratio	BR	none	The ratio of 1st order root number to 2nd order root number
	Branching intensity	BI	cm−1	The ratio of 1st order root number to 2nd order root length

.

2.4. Statistical Analyses

Differences in root and soil traits among treatments were analyzed using one-way analysis of variance (ANOVA) with Fisher’s least significant difference (LSD) at a significance level of p < 0.05 using SPSS 20 (SPSS Inc., Chicago, IL, USA). The data were log-transformed to meet normality and homoscedasticity requirements if necessary. Redundancy analysis (RDA) was used to describe the importance of root functional traits and soil characteristics on soil erodibility using the R 4.2.0 statistic platform with the R package vegan (R Development Core Team, 2013). All figures were plotted using Origin 2021 (OriginLab Software Inc., Northampton, MA, USA) and SigmaPlot 12.0 (Systat Software Inc., San Jose, CA, USA).

3. Results

3.1. Differences of Root Functional Traits across Species

Root diameter (RD) and tissue density (RTD) increased, but specific root length (SRL), specific surface area (SSA), and branching intensity (BI) significantly decreased with increasing root order (Figure 2, p < 0.05). The root diameter (0.40 ± 0.00 mm) and RTD (0.17 ± 0.01 g cm⁻³) in the first-order roots of C. korshinskii were significantly higher than those of the other three shelterbelts (Figure 2a,b, p < 0.05). The L. barbarum shrub shelterbelt had the highest SRL (236.48 ± 37.05 m g⁻¹) and SSA (2.57 ± 0.40 cm² mg⁻¹) of first-order roots, followed by the P. simonii and U. pumila tree shelterbelts, with the shrub C. korshinskii being the lowest (47.80 ± 3.82 m g⁻¹ for SRL and 0.59 ± 0.04 cm² mg⁻¹ for SSA, Figure 2c,d). The root branching intensities of U. pumila (10.50 ± 1.01 cm⁻¹ for 1st/2nd, 5.49 ± 1.07 cm⁻¹ for 2nd/3rd, 1.41 ± 0.29 cm⁻¹ for 3rd/4th) were obviously higher than those of P. simonii, L. barbarum, and C. korshinskii (Figure 2e, p < 0.05).

3.2. Differences in the Soil Erodibility of Four Land Types

Significantly higher geometric mean diameters (GMD_d) and mean weight diameters (MWD_d) of dry aggregates were observed for both shrub soils, compared to the two tree soils (Figure 3a,b). However, the erodible fractions of dry soil aggregates (EF_d) in the tree shelterbelts were higher than those of the shrub soils (Figure 3c). The dry aggregate stability (DAS) of P. simonii soil was the lowest (56.76% ± 6.52%) among the samples (Figure 3d). Based on the regression equations, the U. Pumila and C. korshinskii soils were associated with higher erodible fractions (EF_w) and lower geometric mean diameters (GMD_w) than the other two species, L. barbarum and P. simonii (Figure 3e,f). This pattern is slightly different from the EF_d and GMD_d that were obtained using the dry vibrating screen. The U. pumila soil had a higher percentage of sand content and lower silt content than the other three shelterbelt soils (

Table 3. Characteristics of soil particle size composition, pH, water content, organic carbon, and total nitrogen in four farmland shelterbelts.

Shelterbelt	pH	Water Content (%)	Soil Organic Carbon (g kg−1)	Soil Total Nitrogen (g kg−1)	Soil Particle Volume Content Percentage (%)
					Clay (<0.002 mm)	Silt (0.002–0.05 mm)	Sand (0.05–2 mm)
Lycium barbarum	8.07 ± 0.11 b	21.02 ± 0.93 a	18.06 ± 1.27 a	2.23 ± 0.18 a	3.78 ± 0.38 a	30.33 ± 0.93 a	65.89 ± 0.87 b
Caragana korshinskii	8.49 ± 0.01 a	16.98 ± 0.62 b	11.55 ± 0.50 b	1.61 ± 0.12 b	3.14 ± 0.15 ab	28.63 ± 1.84 a	68.22 ± 1.84 b
Populus simonii	8.31 ± 0.04 a	14.18 ± 0.93 b	13.19 ± 2.14 b	1.81 ± 0.24 ab	2.82 ± 0.27 b	30.00 ± 1.22 a	67.18 ± 1.44 b
Ulmus pumila	8.41 ± 0.09 a	16.72 ± 2.03 b	14.46 ± 0.87 ab	1.50 ± 0.13 b	2.75 ± 0.16 b	21.98 ± 0.93 b	75.27 ± 0.94 a

Data are expressed as means ± standard error (n = 4). Different letters represent statistical significances among species.

).

When the normal stress was 5 N, the U. pumila soil exhibited the highest shearing resistance (22.35 ± 1.47 J), followed by the L. barbarum (16.47 ± 0.86 J) and C. korshinskii (16.87 ± 1.74 J) shrubs, with the lowest values being recorded for P. simonii (10.76 ± 1.34 J, Figure 4). When the normal stress increased, the shearing resistance varied. The two shrub soils displayed higher shearing resistance compared with both tree soils when the normal stress was 20 N (Figure 4).

3.3. The Correlations among Soil Characteristics

There were obvious correlations between the indexes of soil erodibility. The MWD_d and GMD_d of dry soil aggregates were negatively correlated with EF_d to a significant extent (p < 0.001, Figure 5), but these parameters were positively correlated with the shearing resistances with 2 N normal stress (p < 0.05, Figure 5). The GMD_w was negatively correlated with the EF_w (p < 0.001, Figure 5). The erodible fraction (EF_d) of soils measured using the dry vibrating screen displayed a significant positive correlation to the EF_w that was determined using the regression equation (p < 0.05, Figure 5).

3.4. The Correlations between Root Traits and Soil Erodibility

Soil water content displayed significantly positive correlations with the SRL and SSA of absorptive roots (p < 0.05) but negatively correlated with the BI of absorptive roots (p < 0.05, Figure 5). The total carbon of the soil exhibited a significantly positive relationship to SRL and SSA (p < 0.001), though it negatively correlated with the RD and RTD (p < 0.01) of absorptive roots.

There were correlations between soil erodibility and root functional traits and soil characteristics. Negative relationships were noted between the root diameter and the GMD_w (p < 0.05, Figure 5). The SRL and SSA of absorptive roots showed negative correlations to the soil EF_w (p < 0.05), though these two parameters positively correlated with GMD_w (p < 0.001). The total carbon or nitrogen of the soil displayed significantly positive and negative correlations to the soil GMD_w and EF_w, respectively (p < 0.05, Figure 5). There were significantly positive correlations between the soil water content and the soil DAS (p < 0.01).

RDA showed that the first dimension was mainly represented by negative GMD_d (or MWD_d) and EF_d values, and the second dimension was mainly represented by negative GMD_w and EF_w values (Figure 6). The relative eigenvalues of axis 1 and axis 2 were 42.95% and 17.40%, respectively. Based on a permutation test (Figure 6), specific root length (p < 0.001), root diameter (p = 0.002), and root branching intensity (p = 0.016) were the primary biotic factors which had significant effects on soil aggregate stability. As for the abiotic factors, the total nitrogen (p = 0.014), organic carbon (p = 0.016), and water content (p = 0.034) of the soil displayed important roles in the soil aggregate stability (Figure 6).

4. Discussion

Lower-order roots were associated with smaller diameters and tissue densities but higher specific root lengths, specific surface areas, and branching intensity in comparison with higher-order roots. This is consistent with the findings from previous studies [44,50], and it indicates that the structure and function of different root orders vary. Absorptive roots mainly contain the first two non-lignified roots, which mainly perform the function of acquiring soil resources [51,52]. Higher-order roots mainly perform functions associated with transporting, in addition to aiding structural support and protection against adverse environments [51,52]. In this study, the results showed that the L. barbarum shrubs had smaller root diameters and tissue density but higher specific root lengths and surface areas than the other three species. These traits suggested that L. barbarum roots had a stronger capacity for absorbing soil nutrients and water, with higher nitrogen content compared to the other three species (Table 3). Previous studies showed that nutrient addition increased specific root length and decreased root diameter, thereby resulting in higher nutrient uptake capacity of the roots in fertile environments [53,54]. The results from this study showed that there are positive correlations between soil nitrogen and the specific root length (or specific surface area) of absorptive roots. This indicated that in nutrient-rich soils, plants tend to develop more favorable morphological traits for better resource acquisition.

The soil dry aggregates of the two shrubs had higher geometric mean diameters and mean weight diameters than those of the two trees. This suggested that the soil aggregates of the shrubs exhibited more stability than those of the trees when subjected to wind erosion. These findings are relatively different from what was reported from other research that investigated the soil erodibility of trees and shrubs. Ma et al. [55] found that the stability of the soil around Pinaceae and Cupressaceae trees was higher than that of Leguminosae and Rosacea shrubs in the Loess hilly region. Higher aggregate stability was reported for woodlands compared to the grasslands and shrublands in the Karst Valley region of China [56,57]. This might be partly explained by the more complex community structure of the tree forest, which is more effective in improving the microclimate and soil properties, in addition to preserving water and soil. The above-mentioned studies used the composition of soil aggregates as an indicator of anti-erodibility properties against water. However, in this study, we examined the sensitivity of soil to erosion caused by the wind. Zhang et al. (2021) reported that the wind-based soil erosion sand transport rate in the shrublands was notably lower than that of land where there are grasses and trees [20]. This result indirectly supported our conclusion that shrub soil stability was higher than that of tree soil during wind erosion. The differences in results may be due to variations in the study areas, species, and type of erosion.

Researchers established model equations for soil texture, soil organic matter, and calcium carbonate content to predict the soil erodibility index [15,18]. In the present study, the erodible fraction of the soils was calculated based on the regression equation, and the results were as follows: L. barbarum ≤ P. simonii ≤ C. korshinskii ≤ U. pumila. The highest geometric mean diameter and lowest erodible fraction of the L. barbarum shrub indicated the highest soil stability around this plant. On the other hand, the U. pumila tree had the lowest soil stability. These results are similar to the findings that were obtained using the dry vibrating screen. The positive correlation between the EF_d and EF_w further supports that the results of the different types of soil erodibility obtained using the two methods are consistent.

The shear strength of surface soil is an important mechanical property that is useful in predicting its resistance levels against the shear force that is created by the wind [12]. In the present study, the shearing resistance of soils varied when normal stress increased. This indicated that the soils had varying shear strength in response, based on the type of wind. The soil shearing resistance was positively related to the mean weight diameter and geometric mean diameter. These results were consistent with what was reported from a previous study [43]. The larger average weight diameter of soil aggregates was associated with stronger stability and better soil structure. When the shear strength was presented as the index of soil erodibility by the wind, the differences in the anti-erodibility of the soils were inconsistent with the conclusion that was obtained from the dry vibrating screen or the regression formula. This may be due to a small sample size. A positive correlation was noted between soil shear strength and water content, although this was not significant. Previous studies suggested that soil shear strength increased [58] or decreased with rising levels of water content [59,60,61]. Another study showed that the correlation was complicated [62]. Such variations may be due to the different experimental materials and measurement methods.

Previous studies showed that plant root systems could affect soil stability and resistance [63,64]. The findings in this study indicated that roots with higher specific lengths and smaller diameters showed positive effects on the soil GMD_w. The results supported our hypothesis and were consistent with the concept that the soil aggregate stability increased with a greater root length density or SRL [34,36,65]. This may be attributed to fine root functions, such as soil exploration and nutrient uptake. The finding that there was a positive correlation between the soil total nitrogen and the GMD_w was consistent with the fact that the soil in severely eroded areas always had low nutrient content [66]. Thus, plants on fertile soils have higher specific root lengths and lower root diameters as an adaptive strategy for acquiring more nutrients [67]. Previous studies suggested that soil detachment capacity was greatly reduced, not only by higher clay content, soil bulk density, aggregate stability, and shear strength [24,68], but also by increases in specific root morphology traits and the percentage of fine roots [69,70]. Fibrous roots contributed more to soil cohesion than thick taproots due to the larger contact area and stronger bonding forces of the former [71].

There was a positive correlation between the erodible fraction of the soil and the root branching intensity, and this supports our hypothesis. This may be because more root branches tend to weaken, physically binding and connecting fine particles. Generally, plant roots contribute to the soil’s resistance against erosion in two ways. First, the exploratory effect of root systems can form a dense mesh structure in the soil matrix and directly reinforce the soil by physically binding and connecting the fine particles [72,73]. Second, root exudates (like binding agents) can adhere to soil particles, thus enhancing the soil stability through biochemical bonding effects [74,75]. The contributions of the physical effects of the roots were greater (70%) than those of the roots’ biochemical effects in resistance to erosion [74,76,77]. Based on the findings from this study, roots with higher SRLs and lower RDs in their morphology and lower branching intensity in their architecture show positive effects on soil aggregate stability.

5. Conclusions

The dry soil aggregates under two shrubs, L. barbarum and C. korshinskii, had higher geometric mean diameters and mean weight diameters but lower erodible fractions compared to those of the two trees, P. simonii and U. pumila. There was a positive correlation between the EF_d and EF_w that were obtained from a dry vibrating screen and the regression equation, respectively. In addition, the geometric mean diameter of the soils was positively correlated with the specific root length and specific surface area, though it negatively correlated with the root tissue density and root diameter. Root branching intensity was positively correlated with the soil’s erodible fraction and negatively correlated with the mean weight diameter and geometric mean diameter of the dry soil aggregates. Both plant root traits (root diameter, specific root length, and branching intensity) and soil characteristics (carbon, nitrogen, and water content) displayed important positive effects on the soil stability. Our results suggested that L. barbarum was superior to the other three species, based on root traits and wind erosion resistance. The complementarity of root and soil traits can provide critical information for the reasonable collocation of forest stands as windbreaks and for sand fixation.