Introduction

Desertification is a type of land degradation caused by climate and unsustainable land use. Desertification causes a series of environmental, economic, and social problems worldwide, especially in arid areas, and seriously threatens the sustainable development of human society1,2,3. The damage caused by the combination of wind and wind-blown sand not only affects the productivity and life of local people, but also affects the surrounding countries and regions worldwide4. Reducing land degradation and improving soil texture are important objectives when combating desertification5. More and more countries in the world choose to develop plantation forests in arid and semi-arid areas to improve the environment, and people have gradually come to realize the important role of planted vegetation and habitat restoration in regional ecological restoration6,7,8,9. Practice has proven that the restoration of vegetation on sandy land can significantly amend the properties of sandy soil10, increase soil nutrient contents11, promote soil organic carbon accumulation12, and accelerate soil development and recovery13. At present, appropriate revegetation measures have been universally recognized as one of the ways to effectively improve degraded land in desertified areas14,15,16.

China has the largest area of desertification worldwide and experiences the most serious damage from this desertification17. Annual losses of about US $6.8 billion are caused by land desertification in China18. Starting in 1978, China began implementing large-scale afforestation and ecological restoration projects in areas experiencing desertification19. Our study area is located in the Hunshandak Sandy Land, one of the four major sandy areas in China. This study was conducted in the ecological transitional zone between the eastern agricultural and western pastoral areas of China. The diversity of human activities and the special geographic location of the study area have led to frequent disturbances and extensive degradation of the ecosystem there, and land degradation has become an environmental, social, and economic problem in this region20,21. To effectively fix the quicksand and reduce wind-sand disasters, this region has become a key area for the construction of the Three-North Protective Forest, and the goverrnment has planted large areas of different types of plantation forests22. However, after a long period of natural succession, the ecological benefits of the different types of planted forests have been found to exhibit obvious differences23,24. For any degraded land affected by desertification around the world, the development and restoration of soil is a high-priority issue that cannot be ignored. Soil is the basic environment for the growth and living of soil organisms and other species that depend on them. Changes in soil physical and chemical properties are signs of soil formation, development, and evolution, which directly affect the growth, reproduction, and metabolic functions of soil organisms25,26. Since soil is a porous medium composed of soil particles with different shapes and sizes, its structural properties are very complex. Therefore, since the 1980s, the fractal theory has been widely used as a tool to reveal the local structure and morphology of substances in the study of soil structure27,28. The soil fractal dimension is an important index for describing the complexity of the soil structure, which reflects the distribution and morphological characteristics of the internal spaces in soil29. The larger the soil fractal dimension is, the more complex the soil structure is, and the more uniform the pore distribution is30,31. In a large number of previous studies, the fractal dimension of soil particles has been used as a quantitative index to evaluate the soil structure and fertility characteristics32,33,34,35, and some studies have confirmed that the fractal dimension can be used as a representative indicator of the degree of desertification36.

Although several studies have reported the effects of planting different vegetation in sandy areas with different soil properties37,38,39, relevant studies of sandy lands in agro-pastoral ecotones with frequent human behavioral disturbances still need to be supplemented. After the afforestation of desertified land, people tend to pay more attention to changes in soil indicators, and studies on the soil fractal dimension under different vegetation types are not common. In addition, long-term manual management of planted forests has been conducted in many desertified areas, and the research sample sites in our study contain planted forests under natural restoration and rain-fed conditions. The results of our study provide a scientific reference for the restoration of vegetation in agricultural and pastoral transition zones, ecologically fragile areas, and desertification areas. The purpose of this study was to test the following hypotheses: (1) the effects of different vegetation types on soil properties are significantly different under natural restoration conditions and there is a certain regularity in the vertical distribution of the soil properties; and (2) the soil fractal dimension is affected by the vegetation type, and there is a strong correlation between the size of the soil fractals and the soil physicochemical properties. The fractal dimension plays a certain role in characterizing regional soil properties.

Materials and methods

Description of the study area

The study area is located 180 km north of Beijing in Duolun County, Xilin Gol League (City), Eastern Inner Mongolia Autonomous Region (Province), China. The physical geographical unit is located at the intersection of the residual vein of the Greater Hinggan Mountains, the northern foot of Yinshan Mountain, and the residual vein of Yanshan Mountain, at the southeastern edge of the Hunshandake Sandy Land. The study area is a typical area used for a combination of agriculture and animal husbandry in the ecotone of northern China. The landforms mainly include low mountains, hills, valley depressions, sloping plains in front of mountains, high platforms, and stacked dunes. The typical continental climate experiences a temperate semi-arid to semi-humid transition. The average elevation is 1350 m, while the average annual precipitation, windspeed, and temperature are 385 mm, 3.6 m/s, and 1.6 °C, respectively. The long and cold winters are part of a climate with no obvious summer, strong winds, and sufficient light, water, and heat in the warm season23. According to the relevant data of the local government, in the 1970s and 1980s, the joint impact of natural and human factors caused the environment of Duolun County to continuously deteriorate, to the point where the area of desertification exceeded 70% of the total area of the county. By October 2023, 274,533 hm2 of desertified land in Duolun County had been effectively managed for sand stabilization, and only 1,267 hm2 of mobile and semi-fixed sandy areas remained to be restored. The government mainly manages desertified land by banning grazing, returning farmland to forest, and planting a large amount of sand-fixing forest22.

Sample plot setting

In August 2023, a field survey of the study area revealed that the main tree species in the surviving relatively complete plantation in this area were Ulmus pumila, Pinus sylvestris var. mongolica, and Populus simonii. The researchers selected U. pumila, P. sylvestris var. mongolica, and P. simonii forests with the same site conditions, similar age in years since planting, and similar planting density as the study samples, while a nearby bare sandy land without afforestation was selected as the control plot. There were three replicate plots of each type of plot, a total of 12 plots. Basic information on sample plots are shown in Table 1. The selected forest lands had not been artificially tended or managed since planting.

Table 1 Basic information on sample plots (data are the mean ± standard deviation).
Full size table

Soil sample collection and analysis

Soil samples were collected in August 2023. For each plot, one large 30 m × 30 m plot was set up, a total of 12 large sample plots, and three small 10 m × 10 m plots were set up along the diagonal direction of each large plot, a total of 36 small sample plots. Three soil sampling points were evenly selected on the diagonal of each small sample plot. A soil profile with a depth of 60 cm was excavated at each sampling point, and then undisturbed soil was collected in three layers from soil depths of 0–20 cm, 20–40 cm, and 40–60 cm. Soil samples were collected from the same soil layer in three sampling points in each small sample plot and were mixed together prior to analysis of the soil properties. Finally, nine soil samples from each large sample plot were analyzed, 27 soil samples were from each sample plot type were analyzed, and a total of 108 soil samples were analyzed.

The soil samples used to determine water content (SWC) and bulk density (SBD) were sampled with a ring knife of 100 m3 and determined by dry weight. Samples used to determine other soil properties are brought back to the laboratory in sealed bags, plant residues were removed, and soil was air-dried in dark conditions. The volume fraction of the soil particle size of each different plantation sample was determined by an Analysette 22 MicroTecPlus laser particle size analyzer manufactured by Fritsch GmbH in Idar-Oberstein, Germany. The soil organic matter (SOM) was measured using the volumetric heating method with potassium dichromate. The available nitrogen (AN) was measured using the alkali-diffusion method. The available phosphorus (AP) was measured using the 0.5 mol/L NaHCO3-H2SO4 leaching z-molybdenum antimony anti-colorimetric assay method. The available potassium (AK) was measured using the 1 mol/L CH3COONH4 leaching-flame photometric method. The total nitrogen (TN) was measured using the concentrated sulfuric acid with perchloric acid mixed cooking by the Kjeldahl method. The total phosphorus (TP) was measured using the concentrated sulfuric acid-perchloric acid mixed cooking-molybdenum antimony anti-colorimetric method. The total potassium (TK) was measured using the acid dissolution-flame photometric method. This experiment consisted of three forest lands with different tree species and a bare sand plot.

Soil fractal dimension

The soil was divided into: clay (r < 0.02 mm), silt (0.02 mm < r < 0.05 mm), very fine sand (0.05 mm < r < 0.1 mm), fine sand (0.1 mm < r < 0.25 mm), medium sand (0.25 mm < r < 0.5 mm), coarse sand (0.5 mm < r < 1 mm), and very coarse sand (1 mm < r < 2 mm). This was based on the United States Department of Agriculture soil texture grading standards. Based on the particle size volume data obtained from the laser particle size analyzer, the following volume fractal dimension formula was obtained based on the concept of volume fractal dimension of soil particles using calculation formula deduced by Tyler et al.40:

$$\frac{{V_{{(r < R)}} }}{{VT}} = \left( {\frac{R}{{R_{{\max }} }}} \right)^{{3 - D}}$$
(1)

Taking logarithms on both sides of Eq. (1) at the same time gives the formula for the dimensionality of the soil fractal:

$$\lg \left( {\frac{{V_{{(r < R)}} }}{{VT}}} \right) = (3 - D)\lg \left( {\frac{R}{{R_{{\max }} }}} \right)$$
(2)

where V is the total volume of soil smaller than the particle size R (%); VT is the total volume of soil measured (%); R is the average value of the particle size between the two sieve particle size classes Ri and Ri +1 (mm); Rmax is the largest particle size in the soil particle size grading, and the largest particle size of the soil in this study is 2 mm; D is the volumetric fractal dimension of the soil particles, where the left and right sides of the Eq. (2) has the longitudinal and transverse coordinates, respectively, of the linear regression fitting equation, to find the slope value, the difference between 3 and the slope value of the straight line is the value of the soil fractal dimension D.

Data processing and analysis

The normality of the data variables was tested using the Kolmogorov-Smirnova test at a significance level of α = 0.05. For the variables that did not conform to a normal distribution, we transformed the natural logarithm, square root, and derivative of the data to make the data variables obey a normal distribution (P > 0.05). The variance homogeneity test was also performed on the data variables, which yielded P > 0.05, indicating that the fluctuations of the data in the different groups were consistent and met the requirements of variance homogeneity. One-way analysis of variance (ANOVA) and least significant difference multiple comparisons were used to determine if the differences in the soil physicochemical properties across the sites were statistically significant (at the 0.05 probability level for the Duncan and Games-Howell multiple comparisons). The relationship between the fractal dimension of soil particles and the percentage of particles in each soil grain size was analyzed using linear regression analysis, while the effect of soil properties on the fractal dimension of soil was analyzed using path analysis. The correlations between the soil physical and chemical properties were analyzed via Pearson correlation analysis (two-tailed). Statistical significance was defined as significant at P < 0.05 and highly significant at P < 0.01. All of the above data analysis processes were conducted using SPSS 26.0 software. All of the data presented in this paper are the mean ± standard deviation of three replicate determinations for all of the types of sample plots, and on the graphs, the standard deviation is presented in the form of error bars. The plots were created using Origin Pro 2021.

Results

Soil particle size characterization

Afforestation in sandy land significantly reduced soil particle size (Table 2). Compared with the bare sandy land, the volume percentages of the clay, silt, and very fine sand in the plantations were 0.24–0.88%, 3.16-11.91%, and 1.61–6.59% higher, respectively. Moreover, the percentages of fine, medium, coarse, and very coarse sand decreased by 0.96–5.26%, 0.98–9.38%, 0.89–2.5%, and 1.41–1.79%, respectively. At the same soil depth, the volume percentage content of soil clay, silt, and very fine sand was: U. pumila > P. sylvestris var. mongolica > P. simonii > bare sandy land, while the volume percentage content of soil fine sand and medium sand was: U. pumila < P. sylvestris var. mongolica < P. simonii < bare sandy land. The volume contents of the clay, silt, and very fine sand decreased with increasing soil depth, ad the contents of the clay, silt, medium sand, and coarse sand in the different plantation plots were significantly different in each soil depth layer (p < 0.05). The soil texture of U. pumila forest was mainly sandy and loamy sandy soils, while that of P. sylvestris var. mongolica, P. simonii and bare sandy land were sandy soil (Fig. 1).

Table 2 Distribution of soil particles in different plantations.
Full size table
Fig. 1
figure 1

Ternary diagrams of the soil texture.

Full size image

Soil particle fractal dimensions

The average fractal dimension (D) of the soil particles in the three plantations and that of the bare sandy land are shown in Table 3. The D values of all of the soil depth layers exhibited the following change pattern: U. pumila > P. sylvestris var. mongolica > P. simonii > bare sandy land. The fractal dimension of the same sample plot decreased with increasing soil depth, and the fractal dimension of all soil depths had significant differences (p < 0.05).

Table 3 Soil fractal dimensions of different plantations.
Full size table

Characterization of soil water content and bulk density

Soil water content and bulk density in different plots had significant differences in each soil depth (p < 0.05) (Fig. 2). Compared with the bare sandy land, the soil water contents in the plantation sample plots were 1.57–4% higher. Soil bulk density was reduced by 0.1–0.35 g·cm− 3. The variations in the soil water content and bulk density were 0–20 cm < 20–40 cm < 40–60 cm. The soil water contents at the same soil depth were as follows: U. pumila > P. sylvestris var. mongolica > P. simonii > bare sandy land. In addition, the order of the soil bulk density was as follows: U. pumila < P. sylvestris var. mongolica < P. simonii < bare sandy land.

Fig. 2
figure 2

(a) Soil water content and (b) bulk density in different sample plots.

Full size image

Soil nutrient characterization

Afforestation on sandy land significantly increased soil nutrient content (Fig. 3). At the same soil depth, the contents of the soil organic matter, total nutrients, available nitrogen, and available phosphorus exhibited the following change regularity: U. pumila > P. sylvestris var. mongolica > P. simonii > bare sandy land, and the difference was significant among all plots (p < 0.05). The soil nutrients in all of the types of land decreased with increasing depth. The soil available potassium contents of the three plantations were higher than that of the bare sandy land, but there was no significant difference in the soil available potassium among the three types of forest land (p > 0.05). In the same plantation sample, the soil nutrients, except for the available potassium, were significantly different at different soil depths (p < 0.05), but the available potassium was not significantly different at soil depths of 20–40 cm and 40–60 cm in the plantations (p > 0.05).

Fig. 3
figure 3

Soil nutrient content in different sample plots: soil (a) total nitrogen; (b) total phosphorus; (c) total potassium; (d) available nitrogen; (e) available phosphorus; (f) available potassium; (g) organic matter.

Full size image

Relationship between soil properties

Regression analysis was performed on the soil particle percentage content and fractal dimension (D) (Fig. 4). The D values were highly significantly positively correlated with the contents of the clay, silt, and very fine sand (p < 0.01). It was highly significant negative correlated with the percentage of fine and medium sand (p < 0.01). There was a significant negative correlation with the percentage of coarse and very coarse sand (p < 0.05). The D values were mainly affected by the soil particles smaller than 0.5 mm. The particle size of 0.1 mm was the boundary of the change in the D value. The greater the content of the fine particulate matter less than 0.1 mm in the soil was, the higher the D value was.

Fig. 4
figure 4

The relationship between the soil fraction grain percentage and fractal dimension: (a) clay (r < 0.02 mm); (b) silt (0.02 mm < r < 0.05 mm); (c) very fine sand (0.05 mm < r < 0.1 mm); (d) fine sand (0.1 mm < r < 0.25 mm); (e) medium sand (0.25 mm < r < 0.5 mm); (f) coarse sand (0.5 mm < r < 1 mm); and (g) very coarse sand (1 mm < r < 2 mm).

Full size image

The soil fractal dimension was highly significantly positive correlated with soil clay, silt, organic matter, total available nutrients, and available nutrients, and highly significantly negative correlated with soil sand and bulk density (Fig. 5). The index that had the smallest correlation with soil fractal dimension and grain size distribution was water content, in which the soil fractal dimension and clay were significantly positive correlated with water content, while silt and sand were not significantly correlated with water content. Moreover, soil water content was not significantly correlated with bulk density, total phosphorus, and available potassium, but was significantly positive correlated with organic matter, total nitrogen, total potassium, available nitrogen, and available phosphorus. In addition, the bulk density and sand content were highly significantly negative correlated with organic matter, total available nutrients, and available nutrients, and the organic matter, total available nutrients, and available nutrients were highly significantly positive correlated with each other.

Fig. 5
figure 5

Pearson correlations between the soil properties. Note: The numbers displayed represent correlation coefficients. Blue and red indicate positive and negative correlations between parameters, respectively. The significance levels are * p < 0.05; ** p < 0.01.

Full size image

The influence of soil characteristics on fractal dimension (D) was further explained by the path analysis method. The independent variables of path analysis were the soil physical and chemical properties; and the dependent variable was soil fractal dimension. The normal distribution test of the dependent variables showed that p was greater than 0.05 (Table 4); the dependent variables followed a normal distribution. The analysis results showed that soil fractal dimension was mainly affected by soil bulk density and soil clay content (Table 5). The bulk density had a significant negative effect on the D value, while the clay content had a significant positive effect on the fractal dimension (Table 5).

Table 4 Regression model test results.
Full size table
Table 5 Path analysis results.
Full size table

Discussion

The results of this study reveal that the construction of the types of sand fixation forests in mobile sandy areas had a positive effect on the soil properties, but there was a significant difference in the effects of the different vegetation types on the soil. The soil nutrient accumulation was mainly concentrated in the top soil layer (0–20 cm), and. With increasing soil depth, the soil nutrient content and fine particle content gradually decreased, and the soil capacity and water content gradually increased (Fig. 6). Many recent related studies on sand fixation vegetation in sandy areas have yielded similar conclusions. Haonian Li et al.38. found that the construction of mixed vegetation types in the desert-Yellow River transition zone had a good positive effect on the soil physical properties and nutrient contents, but the effects of different mixing patterns on the soil were different. Other scholars39 have studied the effects of different types of plantation forests on alpine sandy soils on the Tibetan Plateau and have found that soil improvement by different plantation forests occurred mainly in the shallow layer. This is mainly due to the fact that the sand-fixing forests not only prevented the movement of wind-blown sand but also protected the soil and created a special microclimate that allowed for soil development. In the sheltered area of the sand-fixation forests, part of the air-borne dust naturally settles on the surface due to gravity, while the other parts adhere to leaves and branches. This imported soil then resupplies to the nutrients and fine soil particles to the soil surface along with “stem flow” along tree trunks or “canopy rain”, adding to the aeolian sand soil41,42.

The results of this study revealed that the soil fractal dimension was strongly correlated with the soil physicochemical properties (Fig. 6), especially the soil nutrients and fine particle content. which both exhibited highly significant correlations (p < 0.01), and the correlation coefficients were all greater than 0.8. The correlations between the soil fine particle content and soil nutrients were also highly significant (p < 0.01). Therefore, the higher the soil fine particle content was, the higher the soil fractal dimension was, and the better the soil nutrient status in the plantation forest was. Changes in the soil properties often represent the degree of land degradation in desertified areas, and the soil fractal dimension can quantitatively describe this phenomenon. In summary, we believe that the soil fractal dimension can be used as an important reference index to characterize soil properties in sandy areas. In forested ecosystems, strong correlations usually exist among the various soil physicochemical properties13,43. Related studies have also analyzed soil properties under different vegetation types or different densities of the same vegetation in sandy areas in arid and semi-arid regions, and they have concluded that a reduction in soil coarse particles increases the soil nutrient contents and the fractal dimension38,44. The soil structure plays an important role in regulating the soil quality and its ecological function, which directly affect the soil water movement (infiltration and evaporation), soil air quantity and quality, and soil nutrient conversion and supply45,46. Soil particles are the basis of the soil structure, and the soil particle size and the combination and ratio of various particles directly affect the basic properties of soil. Strong adsorption and particle combinations exist between soil particles and organic matter. The loose porous structure of soil organic matter not only plays a key role in the formation of soil aggregates but also improves the exchange and absorption capacities of water and nutrients47,48. Scholars in industry have conducted relevant research on the influences of land use types on the soil particle size distribution and fractal dimension. They have concluded that the fractal dimension of soil particles can better reflect the characteristics of the soil structure and fertility32. Jifeng Deng et al.36 studied sand-fixing forests with different stand densities in the Mu Us Desert and confirmed that the fractal dimension is a sensitive and useful indicator for evaluating the degree of desertification. This is consistent with the results of our study.

This research shows that U. pumila is the most suitable sand fixing species in the Hunshandake Sandy Land when compared with the sand fixing of P. sylvestris var. mongolica and P. simonii. It has long been reported that Populus and Pinus plantations are not sustainable for vegetation restoration in sandy lands because these species require more water than is naturally available in many landscapes49. In fact, in some sandy areas of China, U. pumila plant communities have formed a unique natural landscape of sparse U. pumila forest through long-term natural succession. In some cases, even a large number of ancient U. pumila trees that have survived for more than 100 years. These precious ancient U. pumila resources are the result of long-term natural selection. Previous studies have shown that U. pumila is the climax community of psammophytic succession because it is adapted to arid and semi-arid climates in typical steppe landscapes, and is also the main retrogressive succession community in forest steppe during desertification50,51. The economy of the present research area is driven by a combination of agriculture and animal husbandry in an area with a typical continental climate in a temperate semi-arid to semi-humid transition. Ulmus pumila trees can well adapt to the natural environment of the research area. In recent years, sand-fixing forests have become increasingly important to the local government, and these forests have been well protected from human interference. A positive interaction exists between good forest growth and soil quality.

During the restoration of vegetation in sandy land habitats, the development and restoration of soil is a long and complicated process. The forest density, plantation age, and different management measures in a forested landscape will change the growth of individual trees and the forest community structure. The change of surface vegetation will affect the soil microenvironment, which in turn affects the ability of the soil to retain water and fertilizer. The present research is based on the conclusion that different types of sand-fixing forests have similar densities and ages but do not require active forest management. Additional studies and analysis of the changes of sand-fixing forests after long-term natural succession or active forest management are still needed and the direction of future research.

Fig. 6
figure 6

Graphical Abstract.

Full size image

Conclusions

Ulmus pumila, P. sylvestris var. mongolica and P. simonii all had positive effects on the soil properties of sandy land in eastern Inner Mongolia, China. Afforestation on sandy land significantly reduced soil particle size and soil bulk density, increased soil fine particle, water, and nutrient content. The overall soil improvement effect was U. pumila > Pinus sylvestris var. mongolica > P. simonii. Compared with the bare sandy land, the soil bulk density under the U. pumila plantation was 19% lower, and the soil water content, soil organic matter, total nitrogen, total phosphorus, total potassium, available nitrogen, available phosphorus, available potassium, clay, and silt content were 74%, 336%, 207%, 106%, 31%, 41%, 125%, 21%, 498%, and 387% higher, respectively. The plantation effectively improved the surface soil in all of the forest treatments. The volume contents of the clay and silt and the soil nutrient contents of the same sample plot decreased with increasing depth. The soil water content and bulk density increased with increasing depth. These findings prove that our first hypothesis is valid. The soil fractal dimension was obviously different in each type of forest and exhibited the following order in all of the soil layers: U. pumila > Pinus sylvestris var. mongolica > P. simonii > bare sandy land. The clay and silt contents and bulk density of the soil were closely related to the accumulation of soil nutrients. The soil fractal dimension was strongly correlated with the soil physicochemical properties, especially with soil nutrients and fine particle content, which exhibited highly significant correlations (p < 0.01), and the correlation coefficients were all greater than 0.8. The particle size of 0.1 mm was the boundary particle size of the change in the fractal dimension. The greater the content of fine particulate matter less than 0.1 mm in the soil, the higher the fractal dimension of the soil. These findings suggest that our second hypothesis is also valid. Therefore, we believe that U. pumila is a suitable sand-fixing plant species in this area. In addition, the soil fractal dimension can be used as an important reference index to characterize soil properties in sandy areas. The results of this study provide a scientific reference for the restoration of vegetation in desertification areas.