1 Introduction

Vegetation restoration is an effective measure for the ecological environment construction. Its implementation has changed the irrational land use, and facilitated the functional use of the soil–plant system in order to recover soil fertility and improve soil structure (Li and Shao 2006; El Kateb et al. 2013; Bienes et al. 2016). Soil aggregates are the basic units of soil structures, and their stability are highly associated with environmental and agricultural problems, such as degradation of organic matter, water storage and movement, carbon turnover, biological activity, crop growth, and soil erosion (Tisdall and Oades 1982; Lenka et al. 2012; Liu et al. 2014; Zhu et al. 2017). Hence, exploring soil aggregate stability is essential for maintaining soil function in agricultural production and terrestrial material cycling.

Previous studies on the mechanisms of soil aggregate stability mainly focused on the slaking effect, differential swelling, raindrop impact, and physicochemical dispersion (Le Bissonnais 1996; Kinnell 2005). Recently, soil internal forces are revealed to be key factors that affect the stability of aggregates in an aqueous system (Li et al. 2013; Hu et al. 2015; Calero et al. 2017; Yu et al. 2017, 2020). Soil internal forces come from the interaction between charged soil particles and adjacent water molecules in solution, which can reach as high as 100–1000 atm, much higher than other forces, such as raindrop impact force (1–3 atm), slaking effect (< 1 atm), and osmotic stress (< 2.5 atm) (Nearing et al. 1987; Zaher et al. 2005; Hu et al. 2015). Soil internal forces include three forces, namely, electrostatic force, van der Waals force, and hydration force. Among these three forces, the electrostatic and hydration forces are repulsive, which mainly induce the breakdown of soil aggregates, while van der Waals force is attractive, which inhibits the dispersion of soil aggregates (Hu et al. 2015; Huang et al. 2016). Aggregate stability is dependent on the balance of electrostatic, hydration, and Van der Waals forces (Hu et al. 2015; Huang et al. 2016).

Under natural condition, soil aggregates are tightly bound together due to the strong van der Waals attractive force between soil particles in dry soils (Li et al. 2013). However, during rainfall or irrigation, soil solution is diluted; strong hydration repulsive force and electrostatic repulsive force build up rapidly among soil particles and break down soil aggregates. Previous studies showed that aggregate stability decreased with increasing electrostatic force and hydration force in simulated rainfall under laboratory conditions (Li et al. 2013, 2018). Hu et al. (2018a, b) quantitatively evaluated the contribution rates of soil internal forces on rainfall splash erosion and found that soil internal forces initiate soil aggregate breakdown and significantly affected splash erosion. Although it has demonstrated that soil internal forces have an important influence on the stability of aggregates, and the soil internal forces have been calculated quantitatively; the results are based on a simplified model soil system (specific cation-saturated aggregates), which can only represent part of the properties of natural soils. Therefore, it is time to study the influence of soil internal forces on the stability of aggregates in the natural soil systems.

Soils represent an extremely complex material with a large number of various anions and cations on their surface. In such circumstances, it remains challenging to quantitatively calculate the internal forces of natural soils. However, some studies employ ethanol to shield the soil internal forces with the aim to study the influence of external forces on some soil processes (Goebel et al. 2012; Hu et al. 2018b; Fu et al. 2019). For example, Hu et al. (2018b) utilized ethanol and electrolyte solutions with different concentrations as rainfall materials to quantitatively separate the effects of soil internal and raindrop impact forces (external) on splash erosion. Ethanol was used to represent the only effect of soil external forces on splash erosion. Meanwhile, an electrolyte solution was used to represent the integrated effects of soil internal and external forces on splash erosion. The obtained results showed that the contribution of soil internal forces to rainfall splash erosion is greater than that of raindrop impact force. Therefore, ethanol is proved to be a good choice to identify the soil internal forces and external forces. In addition, we know that when dry aggregates are fast wetted by deionized water, they will break down because of suffering strong repulsive soil internal forces. However, when dry aggregates are fast wetted by ethanol, the effect of soil internal forces on aggregates can be removed, and soil aggregate will maintain their original state. This phenomenon is due to the fact that ethanol can prevent slaking and swelling due to its weak polarizability and its relatively low surface tension (Concaret 1967; Le Bissonnais 1996). Besides, ethanol has a much smaller dielectric constant with respect to that of water (ethanol vs water 28.4 vs 78.4), which can compress the diffuse layer of particles (Lagaly and Ziesmer 2003) and lower the repulsive soil internal forces (Permien and Lagaly 1994). Therefore, the different properties between deionized water and ethanol can be used to evaluate the effect of soil internal forces on aggregate stability.

Soil internal forces are affected by the electrochemical properties of soil particles, such as soil cation exchange capacity (CEC) and specific surface area (SSA) (Li et al. 2013; Yu et al 2017). During vegetation restoration, the electrochemical properties of soil particles change with the change of soil basic properties, such as soil organic matter (SOM) content, pH, soil particle composition, and clay mineralogy (Hepper et al. 2006; Gruba and Mulder 2015; Liu et al. 2020). Hepper et al. (2006) reported that SSA was positively related with silt contents; organic matter losses due to excessive cultivation would decrease CEC. Liu et al. (2020) found that during the long-term natural grassland restoration, CEC and SSA increased with the increase of SOM and silt contents. Thus, in the process of vegetation restoration, soil internal forces, usually as high as 100–1000 atm, will change with the change of the electrochemical properties based on the classic double layer theory, which will further influence the stability of aggregates. However, the effect of soil internal forces on aggregate stability during vegetation restoration has not been reported, hindering our understanding of the formation and stabilization process of soil aggregates.

Therefore, in this study we investigated the stability of soil aggregates under different succession stages (farmland, grassland, shrubland, early forest, and climax forest) by dry sieving, and wet sieving with ethanol and deionized water prewetting. The purpose of dry sieving is to determine the size distribution and stability of dry aggregates and then compare them with the wet sieving. The wet sieving with ethanol and deionized water prewetting determines the size distribution and stability of wet aggregates in the presence or absence of soil internal forces. This work aims to evaluate the effect of soil internal forces on the stability of natural soil aggregates during vegetation restoration.

2 Materials and methods

2.1 Study area

This study was conducted on the Lianjiabian Forest Farm, Heshui County, Gansu province, China (35°03ʹ–36°37ʹN, 108°10ʹ–109°18ʹE, 1211–1453 m a.s.l.). The location of the study area is present in the region of Ziwuling Forest in the hinterland of the Loess Plateau. The mean annual precipitation and temperature of the area were 587 mm and 7.4 °C, respectively, with a semi-arid monsoon climate. The main soil type is loessial soil (Calcaric Cambisols, WRB classification, 2014) (Jia et al. 2005). In this area, previous research revealed that the Ziwuling forest region is the sole intact natural secondary forest that remains on the Loess Plateau. A series of succession stages have been developed over the past 150 years with different restoration ages. The restoration successive series began from abandoned farmland and advanced as the order of grassland (Bothriochloa ischaemum, Carex lanceolata, Glycyrrhiza, and Stipa bungeana), shrubland (Sophora davidii, Hippophae rhamnoides, and Spiraea pubescens), early forest (Populus davidiana and Betula platyphylla), and climax forest (Quercus liaotungensis) stages (Deng et al. 2013).

2.2 Sampling and analysis

Field survey and sampling were conducted in mid-June 2018. Based on our previous investigation (Deng et al. 2013), a series of secondary vegetation succession was constructed by choosing five succession stages containing farmland, grassland, shrubland, early forest, and climax forest. Three sampling sites were selected as replicates for each vegetation community, and three plots (20 m × 20 m for forest community, 10 m × 10 m for shrubland community, and 1 m × 1 m for grassland community and farmland) were randomly established within each sampling site. In total, 45 samples (five succession stages × three sampling sites × three plots) were collected. Table 1 lists the basic information of these sites. After removing the surface leaf litter, soil samples were collected from the top layer of 0–20 cm in each plot. Meanwhile, the collection of undisturbed soil was made by cutting the ring (diameter and height of 5 cm each) to measure the bulk density (BD) of soil. After that, we air-dried the soil samples and removed the impurities such as roots and gravels in the soils. Then, we took a small part of soil samples, sieved through 2-mm and 0.25-mm screens, to determine the physicochemical properties of soil. Soil pH (solution/soil ratio: 5/1) was determined via pH meter (Leici, Shanghai, China), while detection of soil organic carbon (SOC) was conducted through the K2Cr2O7 oxidation method (Kalembasa and Jenkinson 1973); CEC and SSA were studied with the combined method for surface property determination suggested by Li et al. (2011). Soil particle size distribution was determined using a Malvern Mastersizer 2000 laser diffraction equipment (Malvern Instruments Ltd, UK).

Table 1 Characteristics and location of the investigation sites (Ma et al. 2020)
Full size table

2.3 Experimental methods

Dry sieving was employed for the determination of the size distribution and stability of dry soil aggregates (Kemper and Rosenau 1986). The protocol involves even placing of air-dried samples (100 g) which have been sieved through a 5-mm sieve present on the top of a set of sieves with an opening of 2, 1, 0.5, 0.25, 0.15, and 0.053 mm from top to bottom, shaking by using a dry-sieving machine with a 75-stroke frequency for 5 min followed by the determination of the sample weight present on each sieve.

To determine the impact of soil internal forces on the aggregate stability, ethanol and deionized water were employed to fast wet the soil aggregates, and then the disintegrated aggregates were sieved in ethanol to obtain the size distribution and stability (Le Bissonnais 1996; Liu et al. 2014). The specific experimental steps are presented as follows: 50 g of mixed soil sample was matched according to the proportion of dry aggregate size, the soil sample gently immersed in ethanol or deionized water for 10 min, the excess ethanol or water piped out, and the remaining soil–water mixture transferred on topmost of a set of six sieves (2, 1, 0.5, 0.25, 0.15, and 0.053 mm) for the separation of aggregate fragments via the wet sieving method. To carry out the wet sieving, 99% ethanol was used to avoid the damage of the aggregate structure during the process of sieving and to inhibit the re-aggregation upon drying. The sieves were shaken in ethanol (amplitude 2 cm, moved up and down 10 times in 1 min) to obtain aggregate fractions of > 2, 2–1, 1–0.5, 0.5–0.15, 0.15–0.053, and < 0.053 mm. The aggregate stability obtained by wet sieving with deionized water prewetting was defined as the water stability of aggregates (Le Bissonnais 1996).

2.4 Index calculation

The mean weight diameter (MWD) and geometric mean diameter (GMD) are common indexes reflecting the soil aggregate stability. The larger the value, the stronger the stability of aggregates and the better agglomeration (Gardner 1956; Youker and McGuinness 1957). Those indices can be calculated according to formulas 1 and 2.

$$MWD=\sum _{i}^{n}{x}_{i}{w}_{i}/\sum _{i}^{n}{w}_{i}$$
(1)
$$GMD=exp\left(\sum _{i}^{n}{w}_{i}ln{x}_{i}/\sum _{i}^{n}{w}_{i}\right)$$
(2)

where wi and xi denote the proportion (%) and mean diameter (mm) of each aggregate size fraction, respectively.

The fractal dimension (D) is considered as a more credible and sensitive parameter. The more stable the soil granule structures, the smaller the fractal dimension (Tyler and Wheatcraft 1992). It can be calculated by Formulas 3 and 4.

$$W/{W}_{T}={\left({\bar{R}}_{i}/{R}_{max}\right)}^{3-D}$$
(3)
$$lgW/{W}_{T}=\left(3-D\right)lg\left({\bar{R}}_{i}/{R}_{max}\right)$$
(4)

where \({\bar{R}}_{i}\) denotes the average diameter of each aggregate size class (mm), W is the weight of the aggregate < \({\bar{R}}_{i}\) (g), WT is the total weight of the aggregates (g), and Rmax represents the maximum diameter of the aggregates (mm).

The transfer matrix method is used to evaluate the preservation rate of individual aggregate size fractions (Shi 2006). The percentage of the aggregates that are placed in sieve i before wet-sieving constructs a matrix Mi, and the percentage of the aggregates that are placed in sieve i after wet-sieving constructs a matrix Ni. The probability of the original aggregates on each sieve preserved in the original sieve during sieving is X1, X2…, Xi and then MX = N. The sum of the preservation rate of each size is used as aggregate stability index (ASI). The ASI is an advanced tool used to analyze the soil aggregate stability. A larger ASI value represents higher aggregate stability (Niewczas and Witkowska-Walczak 2003).

The relative internal force index (RII) was adopted to estimate the impact of the soil internal forces on aggregate stability. The larger the RII, the greater the repulsive soil internal forces generated between soil particles and the greater the degree of disintegration of aggregates. The RII was calculated by following Formula 5.

$$RII=\frac{MWDe-MWDw}{MWDe}\times 100\%$$
(5)

where MWDe is mean weight diameter determined via wet sieving with ethanol prewetting and MWDw is mean weight diameter determined via wet sieving with deionized water prewetting.

2.5 Statistical analysis

In this study, all the statistical analyses were carried out with SPSS statistical software version 19.0. Treatment differences were determined by two-way analysis of variance (ANOVA) in combination with the LSD test. The significance level of the data was set to 0.05. Origin 9.0 was employed to visualize the data.

3 Results

3.1 Basic soil properties under different succession stages

Table 2 shows the basic soil physicochemical properties. The soils in different succession stages were alkaline with pH values ranging from 8.34 to 8.49. Soil BD decreased gradually from 1.27 to 1.04 g cm−3, decreasing by 18.1%, with vegetation succession. The SOC content increased from 7.20 to 16.93 g kg−1, increasing by 135.1%, with vegetation succession. The CEC and SSA increased from 10.90 to 19.52 cmol kg−1 and from 43.92 to 61.39 m2 g−1, respectively. Soil particle composition changed slightly along with the restoration stages, i.e., sand content decreases and silt and clay content increases.

Table 2 Basic properties of testing soil (Ma et al. 2020)
Full size table

3.2 Size distribution and stability of soil aggregates under different succession stages

3.2.1 Size distribution and stability of soil aggregates determined by dry sieving

Figure 1 shows the size distribution of dry aggregates. As shown in Fig. 1, the percentage of different particle sizes aggregates in different succession stages was higher in the 5–2-, 2–1-, 0.15–0.053-, and < 0.053-mm fraction, followed by 1–0.5-mm and 0.5–0.25-mm fraction, the 0.25–0.15-mm fraction being the lowest. Decreasing particle size represented that the percentage of different particle sizes aggregates presents a “V”-type change trend that decreases first and then increases. Besides, with vegetation restoration, the percentage of 5–2-mm aggregates was significantly increased, while the percentage of 0.15–0.053-mm and < 0.053-mm aggregates was significantly decreased, and no significant change was observed for the percentage of 0.25–0.15-mm aggregates.

Fig. 1
figure 1

Soil aggregate size distribution under different succession stages determined by dry sieving. The error bars represent the standard deviations of the means (n = 3). The different lowercase letters indicate significant differences between different succession stages at P < 0.05

Full size image

Table 3 shows the variation of the dry aggregate stability under different succession stages. In Table 3, the highest MWD values were observed in farmland and climax forest followed by early forest. The lowest values were found in shrubland and grassland. The highest values of GMD were observed in farmland followed by climax forest, and the lowest values occurred in grassland. D showed the reverse order of MWD and GMD values, and decreased in the following order: grassland > shrubland > early forest > climax forest > farmland.

Table 3 The stability of dry aggregates under different succession stages determined by dry sieving
Full size table

3.2.2 Size distribution and stability of soil aggregates determined by wet sieving with ethanol prewetting

Figure 2 shows the soil aggregate size distribution measured by wet sieving with ethanol prewetting. As shown in Fig. 2, the percentage of different particle size aggregates in different succession stages was higher in the < 0.053-, 0.15–0.053-, and 5–2-mm fraction, followed by the 2–1-, 1–0.5-, and 0.5–0.25-mm fraction, while it was the lowest in the 0.25–0.15-mm fraction. With decreasing particle size, the percentage of different particle sizes aggregates presents a “V”-type change trend that decreases first and then increases. Besides, with vegetation restoration, the percentage of 5–2-mm aggregates was significantly enhanced, and the percentage of 0.25–0.15-, 0.15–0.053-, and < 0.053-mm aggregates was significantly decreased.

Fig. 2
figure 2

Soil aggregate size distribution under different succession stages determined by wet sieving with ethanol prewetting. The error bars represent the standard deviations of the means (n = 3). The different lowercase letters indicate significant differences between different succession stages at P < 0.05

Full size image

Table 4 shows the variation of soil aggregate stability under different succession stages determined by the wet sieving with ethanol prewetting. In Table 4, the highest values of MWDe occurred in farmland and climax forest followed by early forest. The lowest values were found in shrubland and grassland. The highest values of GMD appeared in farmland and climax forest followed by early forest and shrubland, and the lowest values occurred in grassland. D showed the reverse order of MWDe and GMD values, and decreased in the following order: grassland > shrubland > early forest > climax forest > farmland.

Table 4 Soil aggregate stability under different succession stages determined by wet sieving with ethanol prewetting
Full size table

3.2.3 Size distribution and stability of soil aggregates determined by wet sieving with deionized water prewetting

Figure 3 shows the soil aggregate size distribution measured by wet sieving with deionized water prewetting. As shown in Fig. 3, the percentage of different particle size water-stable aggregates under different succession stages was higher in the < 0.15-mm fraction, followed by 5–0.25-mm fraction, while lowest in the 0.25–0.15-mm fraction. Compared to size distribution investigated by the ethanol prewetting, the process of fast wetting in deionized water increased the percentage of < 0.25-mm aggregates, and decreased the percentage of > 0.25-mm aggregates. Also, vegetation restoration significantly increased the percentage of 5–2- and 2–1-mm aggregates, and decreased the percentage of 0.15–0.053- and < 0.053-mm aggregates.

Fig. 3
figure 3

Soil aggregate size distribution under different succession stages determined by wet sieving with deionized water prewetting. The error bars represent the standard deviations of the means (n = 3). The different lowercase letters indicate significant differences between different succession stages at P < 0.05

Full size image

Table 5 shows the variation of water stability of soil aggregates during vegetation restoration. In Table 5, the MWDw and GMD values showed the same varying tendencies, all of which were increased with vegetation restoration. The highest values of MWDw and GMD were observed in climax forest, followed by early forest, shrubland, and grassland. The lowest values appeared in farmland. However, the D displayed the opposite trend of MWDw and GMD, which was decreased with vegetation restoration.

Table 5 Soil aggregate water stability under different succession stages determined by wet sieving with deionized water prewetting
Full size table

3.3 Preservation rate of soil aggregates and ASI under different succession stages

As shown in Table 6, the higher preservation rate of individual aggregate size fractions was observed in 0.5–0.25-, 0.25–0.15-, and 0.15–0.053-mm aggregates and the lower preservation rate was found in 5–2-, 2–1-, and 1–0.5-mm aggregates, suggesting that soil internal forces exert less effect on the particle size 0.5–0.053-mm fractions, which mainly break down the particle size > 0.5-mm fraction. The preservation rate in each fraction and ASI was increased with the process of vegetation restoration. The ASI values of grassland, shrubland, early forest, and climax forest were 1.93, 2.22, 2.45, and 2.48 times of that of farmland, respectively.

Table 6 Preservation rate and of soil aggregates and ASI
Full size table

3.4 Effect of soil internal forces on the stability of soil aggregates during vegetation restoration

The RII was adopted to estimate the effect of soil internal forces on aggregate stability. As shown in Fig. 4, the RII was decreased with vegetation restoration, and the RII of grassland, shrubland, early forest, and climax forest decreased by 47%, 54%, 55%, and 64% compared with that of farmland, respectively. The results indicated that vegetation restoration process decreased the degree of disintegration of the aggregates caused by soil internal forces.

Fig. 4
figure 4

Relative internal forces index (RII) under different succession stages. The error bars represent the standard deviations of the means (n = 3). The different lowercase letters indicate significant differences between different succession stages at P < 0.05

Full size image

4 Discussion

Vegetation restoration can significantly improve soil physicochemical properties (Li and Shao 2006; Lenka et al. 2012; Deng et al. 2013; Bienes et al. 2016). In this study, the SOC content increased and soil BD decreased with vegetation restoration, which were consistent with the findings reported previously (Burri et al. 2009; Dou et al. 2020). Vegetation restoration increases the amount of vegetation biomass, and large quantities of leaf litter and plant residues are associated with the enhancement of soil fertility, the development of physical characteristics, and the improvement of the soil structure (Li and Shao 2006). Besides, the surface charge properties of soil particles, such as CEC and SSA, increased during vegetation restoration, which was in agreement with the conclusion of Liu et al. (2020). Changes in surface charge properties of soil particles may be associated with the increase of organic matter content during vegetation restoration. SOM is a kind of colloidal particles, which has a lot of surface charges (300–600 cmol kg−1) and a high specific surface area (500–800 m2 g−1) (Leinweber et al. 1993; Pennell et al. 1995). SOM transferred into the soil can bind to the mineral particle surface through chemical bonding and form organo-mineral complexes. The interaction between SOM and clay minerals can greatly alter soil particle surface charge properties (Thompson et al. 1989; Oorts et al. 2003).

The results of dry sieving revealed that revegetation favored the macroaggregate (1–5 mm) formation and promoted the stability of the dry aggregates. This was consistent with the findings obtained by Tang et al. (2016) and Mahesh et al. (2017) who suggested that vegetation restoration promoted the accumulation of aggregates with small particle sizes into large sizes due to the increase of SOM content. The stability of dry aggregates was higher in farmland, which was due to the fact that long-term application of inorganic fertilizer and continuous tillage caused the soil structure to be destroyed and compacted (Haynes and Naidu 1998; Wang et al. 2015). In addition, the results of the aggregate size distribution and stability determined by dry sieving were similar as those determined by wet sieving with ethanol prewetting, both showing that with the decrease of particle size, the percentage of different particle size aggregates decreased first and then increased, and the indexes of aggregate stability, including MWD, GMD, and D, also showed the same change rule with vegetation restoration. These results indicated that ethanol could shield soil internal forces and remove the influence of soil internal forces on soil aggregates, consequently maintaining soil aggregates in their original state (Le Bissonnais 1996; Lagaly and Ziesmer 2003). Besides, the MWD and GMD measured by wet sieving in ethanol treatment were slightly smaller than by dry sieving and the D measured by wet sieving in ethanol treatment was slightly larger than that by dry sieving. This is due to the smaller agitating force used in the process of wet sieving that exerts an effect on aggregate stability.

The results of wet sieving in deionized water prewetting demonstrated that vegetation restoration increased the content of water-stable macroaggregate (1–5 mm) and decreased the content of water-stable microaggregate (< 0.15 mm), and increased the water stability of soil aggregates. These results were consistent with previous researches (Lenka et al. 2012; Liu et al. 2014). These results are commonly attributed to the increase of organic matter content, since SOM performs the function of binding agent, contributing to aggregate stabilization and formation through binding soil mineral particles (Chaplot and Cooper 2015). Besides, plant roots are also involved in the formation and stabilization of soil aggregates. Fine roots interact with external hyphae and generate exudates and temporary binding agents, promoting the soil aggregate formation (Six et al. 2004; Eisenhauer et al. 2011). Additionally, certain studies authenticated that afforestation could stimulate the activity and growth of microorganisms and yield transient binding agents required for the agglomeration, which could thus affect soil aggregate stability (Six et al. 2000; Zhu et al. 2017).

In addition to the above factors, soil internal forces also exert an important impact on the stability of aggregates. Our results showed that the fast wetting process in deionized water increased the percentage of < 0.25-mm aggregates, and decreased the percentage of > 0.25-mm aggregates, and the aggregate stability in ethanol treatment was higher than that in deionized water treatment. For instance, the MWDe values of farmland, grassland, shrubland, early forest, and climax forest soils were 4.62, 1.45, 1.31, 1.32, and 1.17 times of their MWDw values. The higher aggregate stability in ethanol treatment than that in deionized water treatment was due to the fact that ethanol can reduce the influence of soil internal forces on soil aggregates and maintain soil aggregates in their original state (Le Bissonnais 1996; Lagaly and Ziesmer 2003). By contrast, soil aggregates in deionized water treatment suffered strong repulsive soil internal forces and disintegrated severely (Li et al. 2013; Hu et al. 2015). Therefore, the MWDw values were significantly lower than MWDe values. These results indicated that the repulsive soil internal forces significantly broke down the aggregates. Many studies have investigated the effect of soil internal forces on soil aggregate stability. They used model soil systems demonstrated that soil internal forces could produce interparticle forces as high as 100–1000 atm, and significantly affect aggregate stability (Li et al. 2013; Hu et al. 2015; Gong et al. 2018). In the current work, we also prove that soil internal forces have an important influence on the stability of aggregates in natural soil systems. Unfortunately, due to the complexity of soil systems, we can only qualitatively evaluate the effect of soil internal forces on aggregate stability. In addition, our results also showed that soil internal forces mainly broke down the particle size > 0.5-mm fraction. This result was similar to the outcomes of Niewczas and Witkowska-Walczak (2003). The reason is that the > 0.5-mm aggregates are more prone to degradation by water erosion, since the erosion can physically disrupt the formation of water-stable aggregates with larger sizes (Ayoubi et al. 2012).

During the process of vegetation restoration, soil internal forces will change with the change of soil properties. Our results showed that RII decreased with vegetation restoration, indicating that the vegetation restoration process decreased the repulsive soil internal forces, thereby decreasing the degree of disintegration of aggregates. The reason for the decrease of soil internal forces may be associated with SOM accumulation during the process of vegetation recovering. SOM accumulation can increase soil CEC and SSA (Table 2), which in turn can increase electrostatic repulsive force and more importantly increase the van der Waals attractive force between soil particles, finally decreasing the net repulsive force of soil internal forces and increasing aggregate stability (Thompson et al. 1989; Oorts et al. 2003; Huang 2004; Hu et al. 2015; Liu et al. 2020). A similar result was also reported by Yu et al. (2017), who found that the increase of soil organic matter content through straw incubation would greatly increase the interparticle attractive forces, resulting in the decrease of the repulsive soil internal forces and ultimately improve soil aggregate stability. Therefore, it may be that the increase of organic matter content in the process of vegetation restoration reduces the repulsive soil internal forces and thus the aggregate stability is improved.

5 Conclusions

In this study, we qualitatively evaluated the effect of soil internal forces on the stability of natural soil aggregates during vegetation restoration with ethanol and deionized water. The results showed that soil aggregates would not break down as they were fast wetted in ethanol because of removal of the effect of soil internal forces, while soil aggregates would severely break down as they were fast wetted in deionized water due to suffering strong repulsive soil internal forces, and the soil internal forces mainly broke down > 0.5-mm aggregates during fast wetting. Moreover, it could also be found that with the vegetation restoration, the degree of disintegration of aggregates caused by soil internal forces decreased; thus, the water stability of aggregates increased. The obtained findings reveal the reasons for the improvement of water stability of aggregates in the process of vegetation restoration from the perspective of soil internal forces, and provide theoretical references for further study of the formation and stability of good soil structure.