Effects of Vegetation Type on Soil Shear Strength in Fengyang Mountain Nature Reserve, China

: Shear strength is an important mechanical property of soil, as its mechanical function plays critical roles in reducing land degradation and preventing soil erosion. However, shear strength may be affected by vegetation type through changes in the soil and root patterns. To understand the inﬂuences of different types of vegetation on shear strength, the soil shear indices of three typical vegetation types (broad-leaved forest, coniferous broad-leaved mixed forest, and grassland) were studied and evaluated at the Fengyang Mountain Nature Reserve, China. We employed a direct shear apparatus to measure the soil shear resistance index. We quantiﬁed the soil porosity, moisture content, and composition of particle size to determine the properties of the soil, and a root scanner was used to quantify the root index. The results revealed that there were signiﬁcant differences in shear resistance indices at the stand level. Between the three vegetation types, the internal friction angle of the broad-leaved forest was the largest and the cohesion was the smallest. The soil moisture content and porosity of the coniferous broad-leaved mixed forest were higher than those of the broad-leaved forest, and the root volume density (RVD/cm 3 ) of the broad-leaved forest was higher than that of the coniferous and broad-leaved mixed forest and grassland. Structural equation modeling results show that the soil particle size and root characteristics indirectly impacted the soil water content by affecting porosity, which ﬁnally affected shear strength. In general, there were signiﬁcant differences in soil properties and plant root indices between the different stands, which had an impact on soil shear strength.


Introduction
The impacts of global climate change can lead to extreme precipitation cycles that adversely affect the natural environment. Climate and its modification control or affect some of these phenomena; mainly precipitation and temperature [1]. It is expected that changes in precipitation and its indirect effects (such as changes in soil moisture content) will affect slope stability on geographical scales [2], leading to increased erosion and other degradative impacts on slope stability [3]. Geological hazards associated with mountain slopes (e.g., landslides and soil degradation) are frequently reported [4,5]. Landslides involve the movement of soil down a slope under the action of external forces and gravity. broad-leaved forests) has been seriously degraded or destroyed [35]. Over the last few decades, the Chinese government has conducted forestry ecological projects to promote the vegetation restoration of subtropical primary forests [36]. Consequently, there are many secondary vegetation communities composed of different tree species in subtropical China. It is predicted that the frequency and intensity of extreme climate events (e.g., typhoons, heavy rainfall, etc.) in East and Southeast Asia will increase in the future, leading to an increased risk of landslide in subtropical areas in the coming years [37]. Landslide risks are a function of the probability of occurrence and their consequences. Previous studies have indicated that preventive measures are far better than passive measures to reduce risk; however, such measures are rarely adopted in developing countries [38]. Therefore, it is necessary to study the soil shear resistance of typical vegetation in subtropical China, which can provide a theoretical basis for disaster risk reduction from the perspective of shear strength. The experimental site for this study was the Fengyang Mountain Nature Reserve, China. In 1975, the entire area was declared a nature reserve [39].
Specifically, we tested the following hypotheses: (i) there are differences in the soil moisture characteristics and pH of soil, including the composition of soil particles and root characteristics between different vegetation types; (ii) there are differences in soil shear strength between different vegetation types; and (iii) differences in soil shear strength may be reasonably explained by variances in soil and plant root characteristics.

Study Area
This study was conducted at the Fengyang Mountain Nature Reserve (longitude: 119 • 06 -119 • 15 E, latitude: 27 • 46 -27 • 58 N), Longquan City, Zhejiang Province, China. The reserve is situated between 600 m and 1929 m above sea level. It belongs to a subtropical humid climate with an annual average temperature of 12.3 • C and an annual rainfall of 2400 mm. Historically, the reserve was an evergreen broad-leaved forest, dominated by Schima superba Gardn. et Champ., Cyclobalanopsis glauca (Thunberg) Oersted, and Betula luminifera H. Winkl. In 1975, Fengyang Mountain was designated as a nature reserve, and subsequently, the entire study area was protected from human disturbance. The general vegetation composition is shown in Table 1. The soil type in Fengyang Mountain Nature Reserve is mainly yellow loam, yellow brown, which is widely distributed in the nature reserve, and the soil layer develops early. The general soil configuration is AO-A-B-C. AO stands for litter layer. A layer consists of three layers: litter layer, semi-decomposed organic matter layer, and humus accumulation layer. The litter layer is thicker and the soil is mainly loamy clay or clay loam. The central soil layer of the B layer is 20-40 cm thick, and there is clear delineation between the B layer and A layer. The soil texture is loamy clay, containing partly clay loam and accompanied by gravel. The C layer is a semi-weathered parent rock, the color of which is yellowish-orange, and the soil texture is stony or gravelly sandy loam. The three stands are located in the same geological conditions.

Sampling
For this study, in August 2017, areas at similar altitudes with three different vegetation types (evergreen broad-leaved forest, mixed forest, and shrub-grass forest) were randomly selected. In each of the three forest types, we selected three stands, resulting in a total of 9 sampling stands. In each sampled stand, a 20 × 20 m 2 plot was randomly established to represent the stand and was at least 500 m away from the forest edge. During sampling, two points were taken along a diagonal line in each sample plot and soil profiles were excavated. The soil at a distance of 5 m from the trunk on the diagonal of the quadrate was selected. Since the soil layer below 30 cm contained a lot of gravel, the samples were extracted from the 0-10 cm, 10-20 cm, and 20-30 cm soil layers of slope soil, where each layer was (20 mm high × Ø 61.8 mm). Four replicate samples were extracted from each layer, with a total of 216 soil samples taken from each layer. The samples were sealed with plastic film to determine the shear strength of undisturbed soil and were placed in hermetic boxes and immediately taken to the laboratory. In order to prevent the sampling of the ring knife samples from being disturbed during transportation, we used plastic foam to fix the ring knife to prevent the soil from spreading. Each soil sample was removed from a self-sealed bag and returned to the laboratory for air drying and screening, after which the soil pH and soil particle composition were determined. Next, the undisturbed soil of each layer was immediately transported to the laboratory to quantify the soil physical properties (bulk density, water content, and porosity).

Soil Moisture Characteristics and pH Analysis
The soil water content (MC, %) was measured by taking the proportion of the loss of mass, after oven-drying at 105 • C, to the constant mass of dry soil. The soil bulk density (BD, g cm −3 ) was measured using a stainless-steel cylindrical ring of 100 cm 3 volume to collect the samples. According to Coulomb's formula: where W (g) is the weight of the ring cutter + the weight of dry soil after drying, Wr (g) is the weight of the empty ring cutter, and V (cm 3 ) is the volume of the ring cutter. The upper and lower covers of the ring knife were taken off indoors, one end was replaced with a bottom cover with mesh and filter paper, and the ring knife was put into the porcelain plate with a water depth of 2-3 mm. After immersing for 12 h, the water outside the ring knife was wiped off and it was weighed (W 1 ). After weighing, it was immersed in water again. The water surface was up to the upper edge of the ring knife. The soaking time was until the filter paper on the ring knife was fully wet. At that time, the water outside the ring knife was wiped off and it was weighed (W 2 ), and then transferred to the oven at 105 • C [40] for constant weight (W 3 ).
The pH value was determined by the potential method (water/soil ratio 2.5:1, pH meter).

Determination and Calculation of Shear Strength
The soil shear strength was measured using a ZJ-2 Isostrain Direct Shear Apparatus (Nanjing Soil instrument factory, Nanjing, China), where the unconsolidated and undrained fast shear test was performed in strict accordance with the "geotechnical experiment specification SL237-1999". During the determination, it was necessary to keep the natural dry density and moisture content of soil samples within the ring cutter. During shearing, vertical pressures of 100, 200, 300, and 400 kPa were applied, respectively, and the handwheel was rotated at a speed of 10 s/R (six revolutions per minute) until the specimen was sheared.
The shear strength parameters, the cohesion (c) and internal friction angle (ϕ), were calculated automatically by the software installed on the PC. The calculations were in strict accordance with the Chinese Standards for Soil Test Methods (GB/T50123-1999), with the following equations being used to calculate the results [42]: where τ is the shear stress in kPa and C is the calibration coefficient of the dynamometer in kPa/0.01 mm. The calibration coefficient of the dynamometer was 1.825 kPa/0.01 mm; R is the degree of the dynamometer to 0.01 mm. The shear stresses under different pressures were obtained, according to Coulomb's formula: where σ is the initial normal stress (kPa). The cohesion (c; kPa) and the internal friction angle (ϕ; • ) of each group of samples were calculated.

Determination of Root Characteristics
At the conclusion of the shear strength test, the soil in the ring cutter was removed and washed using a 0.05 mm mesh screen to remove all of the roots in the ring knife. Parameters including the root length, root surface area, and root volume were also determined. Finally, the root system was placed into a paper envelope and dried at 90 • C for 72 h to obtain the root biomass.
The roots were separated from soil samples using a 5 mm mesh sieve, after which the roots were placed in basins with water to be gently washed to remove soil particles and other debris. The roots were scanned and analyzed using the WinRHIZO root analysis system for morphological parameter measurements. The root length density (RLD, cm.cm −3 ), defined as the total root length per soil volume (mm (100 cm 3 ) −1 ), root surface area density (RSAD; cm 2 .cm −3 ), root volume density (RVD; cm 3 .cm −3 ), and root weight density (RWD, mg. cm −3 ) were estimated [40].

Statistical Analyses
The Duncan test was used when one-way ANOVA (SPSS Inc., Chicago, IL, USA) revealed that the effects of land use type on shear strength were significant. We examined model residuals and confirmed that the assumptions of normality and homogeneity were met. Two-way ANOVA was employed to test the main effects and interactions of land use type, soil layers on root soil, physical and chemical soil properties, and root index by SPSS 19.0. All histograms were fit using Origin8.5 software. Redundancy discriminant analysis (RDA) was performed to reveal the relationships between the shear strength parameters of root soil, the physical and chemical properties, and root index, using a Canoco 5.0 (Microcomputer Power, Ithaca, NY, USA). Each arrow (usually representing a numeric variable) points in the direction of the steepest increase in environmental variable values. The angle between arrows (alpha) indicates the correlation between individual environmental variables. The approximated correlation is positive when the angle is sharp and negative when the angle is larger than 90 degrees. The length of the arrow is a measure of fit for the indicators.
Structural equation modelling (SEM) was used to investigate the effects of soil characteristics and plant root characteristics on the soil shear strength. The model was used to test whether different vegetation types directly or indirectly affected the soil shear strength by influencing soil and root characteristics. SEM analyses were performed using AMOS 24.0 (SPSS Inc., Chicago, IL, USA).

Soil Moisture Characteristics and pH Value
There were significant differences in the capillary porosity (CAP) (p < 0.01), total porosity (SAT), and pH between different land use types and soil layers (Figure 1b,c,f). Within the deeper soil layers, the non-capillary porosity (NCAP), CAP, and SAT exhibited a downward trend (Figure 1a-c), contrary to the trend in pH (Figure 1f). Compared with the broad-leaved forest, mixed forest and grassland soil samples had larger CAP and SAT values, while the pH and NCAP values were the opposite; however, the differences were not significant. Compared with the native broad-leaved forest and grassland, the moisture content (MC) of mixed forest was slightly larger and the difference was not significant ( Figure 1d). able) points in the direction of the steepest increase in environmental variable values. The angle between arrows (alpha) indicates the correlation between individual environmental variables. The approximated correlation is positive when the angle is sharp and negative when the angle is larger than 90 degrees. The length of the arrow is a measure of fit for the indicators.
Structural equation modelling (SEM) was used to investigate the effects of soil characteristics and plant root characteristics on the soil shear strength. The model was used to test whether different vegetation types directly or indirectly affected the soil shear strength by influencing soil and root characteristics. SEM analyses were performed using AMOS 24.0 (SPSS Inc., Chicago, IL, USA).

Soil Moisture Characteristics and pH Value
There were significant differences in the capillary porosity (CAP) (p < 0.01), total porosity (SAT), and pH between different land use types and soil layers (Figure 1b,c,f). Within the deeper soil layers, the non-capillary porosity (NCAP), CAP, and SAT exhibited a downward trend (Figure 1a-c), contrary to the trend in pH (Figure 1f). Compared with the broad-leaved forest, mixed forest and grassland soil samples had larger CAP and SAT values, while the pH and NCAP values were the opposite; however, the differences were not significant. Compared with the native broad-leaved forest and grassland, the moisture content (MC) of mixed forest was slightly larger and the difference was not significant (Figure 1d). . Soil water-physical properties of three vegetation types and three soil layers (mean ± SE, n = 4). SE is the variance of the index. The error line indicates standard error; upper case letters indicate a significant difference among different soil layers of the same land use types (p < 0.05); the lower case letter indicates a significant difference between different land use types of the same soil layer (p < 0.05). EBF stands for evergreen broad-leaved forest; CBF stands for mixed forest; Figure 1. Soil water-physical properties of three vegetation types and three soil layers (mean ± SE, n = 4). SE is the variance of the index. The error line indicates standard error; upper case letters indicate a significant difference among different soil layers of the same land use types (p < 0.05); the lower case letter indicates a significant difference between different land use types of the same soil layer (p < 0.05). EBF stands for evergreen broad-leaved forest; CBF stands for mixed forest; SGF stands for shrub-grass forest; NCAP stands for non-capillary porosity; CAP stands for capillary porosity; SAT stands for total porosity; BD stands for bulk density; MC stands for moisture content; VT stands for vegetation type; SL stands for soil layer. Two-way ANOVA was applied to indicate significant differences between variances.

Soil Particle Composition
The clay, silt, and sand contents of different stands were significantly different (p < 0.05). The contents of clay and silt in the broad-leaved forest were lower than those in the coniferous broad-leaved forest and grassland (Figure 2a,b); however, the content of sand was the highest (Figure 2c). In the deeper soil layers, the content of clay and silt in the three stands exhibited the same change trends. The change trend of sand content in the stands was the opposite, however; the difference was not significant. The clay, silt, and sand contents of different stands were significantly different (p < 0.05). The contents of clay and silt in the broad-leaved forest were lower than those in the coniferous broad-leaved forest and grassland (Figure 2a,b); however, the content of sand was the highest (Figure 2c). In the deeper soil layers, the content of clay and silt in the three stands exhibited the same change trends. The change trend of sand content in the stands was the opposite, however; the difference was not significant.

Root Distribution Characteristics
With deeper soil layers, changes in the root surface area density (RSAD), root volume density (RVD), and root weight density (RWD) in different stands exhibited a downward trend with significant differences (p < 0.01) (Figure 3b-d). In addition to the native broadleaved forest, the RLD of the coniferous and broad-leaved mixed forest and grassland initially decreased and then increased to attain a significant level in the coniferous and broad-leaved mixed forest (p < 0.05) (Figure 3a).

Root Distribution Characteristics
With deeper soil layers, changes in the root surface area density (RSAD), root volume density (RVD), and root weight density (RWD) in different stands exhibited a downward trend with significant differences (p < 0.01) (Figure 3b-d). In addition to the native broadleaved forest, the RLD of the coniferous and broad-leaved mixed forest and grassland initially decreased and then increased to attain a significant level in the coniferous and broad-leaved mixed forest (p < 0.05) (Figure 3a).  Figure 3. Parameters of root characteristics of three vegetation types and three soil layers (mean ± SE, n = 4), SE is the variance of the index. The error line indicates standard error; upper case letters indicate a significant difference among different soil layers of the same land use types (p < 0.05); the lower case letter indicates a significant difference between different land use types of the same soil layer (p < 0.05). EBF stands for evergreen broad-leaved forest; CBF stands for mixed forest; SGF stands for shrub-grass forest; RLD stands for root length density; RSAD stands for root surface area density; RVD stands for root volume density; RWD stands for root weight density; VT stands for vegetation type; SL stands for soil layer. Two-way ANOVA was applied to indicate significant differences between variances. (a) shows the RLD in different soil layers in different stands; (b) shows the RSAD in different soil layers in different stands; (c) shows  different soil layers of the same land use types (p < 0.05); the lower case letter indicates a significant difference between different land use types of the same soil layer (p < 0.05). EBF stands for evergreen broad-leaved forest; CBF stands for mixed forest; SGF stands for shrub-grass forest; RLD stands for root length density; RSAD stands for root surface area density; RVD stands for root volume density; RWD stands for root weight density; VT stands for vegetation type; SL stands for soil layer. Two-way ANOVA was applied to indicate significant differences between variances. (a) shows the RLD in different soil layers in different stands; (b) shows the RSAD in different soil layers in different stands; (c) shows the RVD in different soil layers in different stands; (d) shows the RWD in different soil layers in different stands.
Compared with broad-leaved forest and grassland, the coniferous and broad-leaved mixed forest had the lowest root length density (RLD), RSAD, and RVD in the 10-20 cm and 20-30 cm soil layers, while the RWD was just the opposite, with no significant difference. The RLD, RSAD, and RWD in the 0-10 cm soil layer of the mixed coniferous and broadleaved forest were the highest; however, the difference was not significant.

Shear Strength Parameters of Root Soil
The internal friction angle and cohesion of the different classes of soil were significantly different (p < 0.05) (Figure 4a,b). In different soil layers, the internal friction angle of mixed coniferous and broad-leaved forest was significantly lower than that of the grassland and native broad-leaved forest, while the internal friction angle of broad-leaved forest was the largest (Figure 4a). Moreover, the cohesiveness of the broad-leaved forest soil was significantly lower than that of the mixed broad-leaved forest and grassland (Figure 4b).
Except for the 10-20 cm soil layer, the soil cohesiveness of the mixed broad-leaved forest and grassland were the largest, while that of the other soil layers was grassland.  Figure 4. Change law of shear strength parameters of three vegetation types and three soil layers (mean ± SE, n = 4), SE is the variance of the index. The error line indicates standard error; the upper case letter indicates a significant difference among different soil layers of the same land use types (p < 0.05); lower case letters indicate a significant difference between different land use types of the same soil layer (p < 0.05). EBF stands for evergreen broad-leaved forest; CBF stands for mixed forest; SGF stands for shrub-grass forest; φ stands for internal frictional angle; c stands for cohesive force; VT stands for vegetation type; SL stands for soil layer. Two-way ANOVA was applied to indicate significant differences among variances. (a) shows the φ in different soil layers in different stands; (b) shows the c in different soil layers in different stands.

Linking Physical Properties of Soil Moisture and pH, Soil Particle Composition, Root Distribution Characteristics, and Shear Strength Parameters of Root Soil
In the RDA of the shear strength parameters of the root soil, with soil and root properties as the explanatory variables, Axis 1 accounted for 40.61% of the variation in the dataset, with 2.72% of the variation accounted for by Axis 2 (Figure 5). High φ with high sand, pH, RVD, RVD, and RLD, was found at the right-hand end of the ordination plots and associated with lower clay, silt, CAP, and SAT. The c, BD, clay, and silt increased along the y-axis, whereas the RWD, RLD, and sand decreased. case letter indicates a significant difference among different soil layers of the same land use types (p < 0.05); lower case letters indicate a significant difference between different land use types of the same soil layer (p < 0.05). EBF stands for evergreen broad-leaved forest; CBF stands for mixed forest; SGF stands for shrub-grass forest; ϕ stands for internal frictional angle; c stands for cohesive force; VT stands for vegetation type; SL stands for soil layer. Two-way ANOVA was applied to indicate significant differences among variances. (a) shows the ϕ in different soil layers in different stands; (b) shows the c in different soil layers in different stands.

Linking Physical Properties of Soil Moisture and pH, Soil Particle Composition, Root Distribution Characteristics, and Shear Strength Parameters of Root Soil
In the RDA of the shear strength parameters of the root soil, with soil and root properties as the explanatory variables, Axis 1 accounted for 40.61% of the variation in the dataset, with 2.72% of the variation accounted for by Axis 2 (Figure 5). High ϕ with high sand, pH, RVD, RVD, and RLD, was found at the right-hand end of the ordination plots and associated with lower clay, silt, CAP, and SAT. The c, BD, clay, and silt increased along the y-axis, whereas the RWD, RLD, and sand decreased.
In the RDA of the shear strength parameters of the root soil, with soil and root properties as the explanatory variables, Axis 1 accounted for 40.61% of the variation in the dataset, with 2.72% of the variation accounted for by Axis 2 (Figure 5). High φ with high sand, pH, RVD, RVD, and RLD, was found at the right-hand end of the ordination plots and associated with lower clay, silt, CAP, and SAT. The c, BD, clay, and silt increased along the y-axis, whereas the RWD, RLD, and sand decreased. properties and root index. RLD stands for root length density; RSAD stands for roots surface area density; RVD stands for root volume density; RWD stands for root weight density; NCAP stands for non-capillary porosity; CAP stands for capillary porosity; SAT stands for total porosity; BD stands for bulk density; MC stands for moisture content; ϕ stands for internal frictional angle; c stands for cohesive force. Figure 6 shows the structural equation modeling (SEM) as estimated by AMOS. Each of the observed variables is displayed in a rectangle. The χ 2 test revealed that the model generated χ 2 = 6.160, df = 6, and p = 0.406 (>0.05). The goodness-of-fit index (GFI) was 0.963 (>0.900), and the root mean square error of approximation (RMSEA) was 0.023 (<0.080). Table 2 displays the standardized direct and indirect influence parameters of the RVD, MC, sand, and SAT on the ϕ and c. of the observed variables is displayed in a rectangle. The χ 2 test revealed that the model generated χ 2 = 6.160, df = 6, and p = 0.406 (>0.05). The goodness-of-fit index (GFI) was 0.963 (>0.900), and the root mean square error of approximation (RMSEA) was 0.023 (<0.080). Table 2 displays the standardized direct and indirect influence parameters of the RVD, MC, sand, and SAT on the φ and c.

Discussion
In accordance with previous research results, soil properties changed significantly in the subtropics due to the presence of different vegetation cover types, which directly reflected soil structural changes [43]. The results show that the soil pH and bulk density had minor differences along with soil depths for each vegetation type. This result may have been due to a common parent material and type [44]. Compared with the broad-leaved forest soil, the pH values of the grassland, coniferous forest, and coniferous broad-leaved forest soil decreased. This might have been due to the release of organic acids from tree roots, the conversion of exchangeable cations during plant uptake and/or soil microbial processes, the transformation of organic matter, and the release of organic acids [45,46]. The water content of coniferous broad-leaved mixed forest was higher than that of broadleaved forest and grassland, and was positively correlated with soil porosity and root characteristics. This result was consistent with that observed in Li et al. [47]. These changes may be attributed to the stimulating effect of the increase in water content on plant growth. At the same time, good root permeability helps to store soil water [48,49].
In contrast to the broad-leaved forest with only a single tree species, the root biomass, root specific surface area, and root length density of the coniferous and broad-leaved mixed forest with diverse species were higher, which may have been due to more abundant fine roots in the mixed forest [50]. With deeper soil layer, changes in the root surface area, root volume, and root biomass showed similar trends (from high to low), and were similar between different vegetation types. It was verified by research that the fine root density in the upper 0-10 cm soil layer was higher than that at 10-20 cm across all forest types [40,51]. Simultaneously, root characteristics and soil porosity were significantly positively correlated. Therefore, the root system of vegetation was related to the improvement of soil structure. The results show that the capillary porosity and total porosity of coniferous and broad-leaved mixed forest were higher than those of the grassland and broad-leaved forest. This corresponded to previous results showing that root systems in coniferous and broad-leaved mixed forests were more developed This difference may also be attributed to the difference in the amount of litter, which is easily decomposed for different vegetation types [52,53].
The results of the two factors test showed that the soil particle size and soil shear index were also affected by the vegetation type, and there were significant differences (p < 0.05). SEM analysis indicated that the soil particle size, soil water physical properties, and plant roots could directly affect the soil internal soil friction angle, and the direct effect of soil particle size was the largest. In addition to the soil water content, soil cohesion was directly affected by plant root system characteristics and particle size, indirectly through the soil water content, and the final comprehensive effect.
The results show that the soil particle size had the greatest effect on soil shear resistance, followed by the soil water physical properties and plant root index. The internal friction angle of broad-leaved forest was significantly larger than that of the grassland and coniferous broad-leaved forest; however, the cohesion was the opposite. Previous studies have shown that fine roots, such as those of most herb roots, and the root hairs and fibrous roots of the lateral roots and taproots of woody plants, can enhance the cohesion of the soil but not the friction angle of the soil. In contrast, the coarse roots of woody plants can enhance both the soil's cohesion and its internal friction angle. This is because the lignification degree and the stiffness of the coarse roots of woody plants are greater than those of fine roots [54]. Compared with coniferous broad-leaved forest and grassland, the clay content of broad-leaved forest was the lowest, whereas the sand content was the highest, which was consistent with reported results [55].
The RDA results show that the internal friction angle was positively correlated with the characteristics of soil sand and vegetation roots, and negatively correlated with the fine grain content and porosity, while the cohesion was the opposite. However, its internal friction angle was the highest, which have been because the higher the content of fine particles, the more unfavorable the sliding and occluding friction between particles, and the smaller the internal friction angle. The results reveal that there was a positive correlation between clay content and cohesion, but it was not significant. This result is reasonable, as the fineness of the colloid, the high surface area, and the large number of chemical bonds enhanced the population of bonds between the particulates and promoted aggregation, which resulted in an increase in cohesion [56]. Consequently, coarse particles (i.e., gravel) reduced soil cohesion, and fine particles (i.e., fine clay) greatly reduced the friction strength of the soil in this area [57]. Simultaneously, the soil particle size was significantly related to soil water characteristics.
The soil cohesion and internal friction angle exhibited a nonlinear attenuation trend with the soil water content. Previous studies revealed that the distribution of soil particle dimensions determined soil hydraulic properties [58], the actual effective soil stress, and soil shear strength. A higher clay content can retain additional soil moisture [59], which in turn increases the distance between soil particles. At the same time, the internal structures of the soil samples with different porosities were different, where the contact and arrangement of particulates were also different. The greater the total porosity and capillary porosity, the smaller the friction force between soil particles [60]. This was consistent with the conclusion of this study. Therefore, soil fine particles have a strong adhesion and adsorption capacity, and have a good expansibility, viscosity, and water holding capacity, which can increase the available soil moisture of plants to a certain extent, thus promoting their growth.
The correlation between the root index and shear index was not significant; however it had an indirect effect on the shear strength, which may have been because the interactions between roots and soil involve a very complex process, not only related to the root index, but also including the influences of root exudates. Several reports have verified that plant root systems are the main driving factor in the physical properties of soil water [61], and indirectly so through root-induced changes in soil properties (e.g., soil bulk density and porosity) [62]. Previous studies have shown that roots can physically change particle contacts in soil by either loosening or infiltration [63]. The cohesive strength of the soil in the grassland and coniferous broad-leaved mixed forest was higher than that of broadleaved forest, which may have been due to the higher fine roots (Ø < 2 mm) turnover rate, compared with coarse roots (Ø > 2 mm) in the 0-30 cm soil layer [64,65]. Furthermore, the roots generate fungal hyphae and globulins that indirectly bind to soil particles, which contribute to soil stability by producing exudates (e.g., gum and arabinogalactan) that bind to these bonds [66]. Therefore, a greater focus on functional root traits should be considered in future research of the influences of root shear strength, to explain the changes in the physical properties of soil and deepen our understanding of the complex relationships of root soil shear resistance [67].
Subtropical forests have long been significantly impacted by anthropogenic land use habits and deforestation due to climatic and economic drivers, deforestation (e.g., access to timber or agricultural land), or changes in forest coverage (e.g., conversion from local vegetation to mixed forest) in response to climatic and economic drivers [68]. Meanwhile, a 2014 IPCC report indicated that extreme precipitation events in most midlatitude and tropical humid regions are expected to become more intense and frequent [69]. Consequently, the frequency of landslides is anticipated to increase. In these cases, it is particularly important to study the shear erosion resistance of plantation soil. At present, people are paying more and more attention to the ecological benefits of plantations [70]. The results of this study show that the shear strength of different vegetation types is different, and this difference can be explained by soil properties and plant root characteristics. The difference in forest root causes the change of soil structure, enhances soil water holding capacity, and finally leads to the difference in forest shear strength. However, the effect of different forest roots on soil shear strength is not deep enough. There is no research on root exudation and the tensile strength of the root system itself, which will be further studied in combination with rainfall in the future.

Conclusions
The results reveal that the soil moisture characteristics and pH of soil and vegetation root indices of different vegetation types were variable. The direct and indirect impacts of soil properties and plant roots modified the soil shear strength. Compared with the broad-leaved forest, the internal friction angle of the coniferous broad-leaved mixed forest and grassland soil was smaller, and the cohesion was greater, which improved the soil shear strength to a certain extent. These results are of great significance for the effective restoration of soils that have been subject to human disturbances, to reduce land water flow and erosion.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.