The experimental area is located at the southeastern edge of the Hobq Desert (106°55′16″–109°16′02″E, 39°22′33″–40°52′14″N), within the jurisdiction of Hangjin Banner in Ordos City (Figure 1). This region experiences a temperate semi-arid continental monsoon climate, characterized by an average annual precipitation of 186 mm and a mean potential evapotranspiration rate of 2,498.7 mm per year. The recorded extreme maximum temperature reaches 38.1°C, while the extreme minimum temperature drops to −30.5°C. The estimated average annual temperature is approximately 6.3°C. The total annual duration of sunshine amounts to 32,115 h. The average annual wind speed is measured at 3.2 m/s, with predominant wind directions originating from the south and southeast. The soil in this area is classified as wind-sand soil, featuring a loose surface layer with limited development and deficient nutrient content. The study area is predominantly characterized by mobile sand dunes with slopes ranging from 10° to 12°. (Data are based on historicalreanalysis datasets from the European Centre for Medium-Range Weather Forecasts (ECMWF)/National Aeronautics and space Administration (NASA), provided by www.xihe-energy.com). The primary morphological types observed in this region include lattice dunes and composite dunes. Notable dune vegetation comprises Salix Psammophila, Artemisia ordosica, Agriophyllum squarrosum, Corispermum hyssopifolium, and Achnatherum splendens. In order to safeguard the sand-penetrating highway against the detrimental impacts of quicksand and guarantee smooth traffic flow, semi-concealed square lattice barriers constructed from Salix Psammophila have been installed on both sides of the highway. These barriers serve to stabilize wind and sand flow dynamics.

Considering the impact of slope, experimental dunes featuring Salix Psammophila sand barriers of varying ages were hereby selected. These dunes exhibited relatively consistent properties, including dune height and orientation within the study area, and flatter sites were specifically chosen for observation. The Salix Psammophila sand barriers consisted entirely of dead branches arranged in a semi-concealed manner, with each barrier measuring 1 m × 1 m. Barriers on dunes aged 1, 3, and 5 years respectively were selected as research factors, while flowing dunes without sand barriers served as controls. Based on previous findings indicating that the windward slope of the dunes was primarily responsible for soil loss—and that different parts of this slope experienced varying degrees of loss along with distinct sorting effects (Zhang, 2022), the present experiments were conducted at different locations along the windward slope. In May 2021, soil samples were collected from the near-surface layer (0–30 cm) at three distinct positions (the bottom, middle, and top) of the sand barriers representing three different ages. A five-point sampling method was employed for this collection. Specifically, soil samples were taken from the four corners and the central point of each block unit. These individual samples were then combined to form a single representative sample, which accurately mirrored the comprehensive soil characteristics of that particular unit. Three barriers constituted a group of replicates for each type based on barrier age, and samples were repeated three times per group. Subsequently, all collected soil samples were packed into ziplock bags and transported back to the laboratory. Ultimately, a total of 108 soil samples were collected for analysis. The samples were air-dried at room temperature prior to pre-treatment, which involved the use of a 2 mm soil sieve to eliminate surface debris.

The soil was processed to remove roots, apoptotic material, and impurities. It was then passed using a 2 mm sieve, followed by the addition of hydrogen peroxide (H₂O₂) at a volume fraction of 30% to remove organic matter present in the soil. Subsequently, a hydrochloric acid (HCl) solution with a volume fraction of 10% was introduced and boiled to extract carbonate constituents from the soil. Following that, a diluted ionized water solution was added and allowed to stand undisturbed for a duration of 12 h. Upon these procedures, the clear upper layer of the solution was carefully removed. This dilution process was repeated until a pH level between 6.5 and 7.0 was achieved (Zhang et al., 2017). The particle size distribution of the soil was evaluated using a laser particle size analyzer (Analysette 22 Nano Tec). In this study, soil particles were classified into the following seven size categories based on American grading standards (USDA Soil Classification System): clay (<0.002 mm), silt (0.002–0.005 mm), very fine sand (0.005–0.01 mm), fine sand (0.01–0.25 mm), medium sand (0.25–0.5 mm), coarse sand (0.5–1.0 mm), and very coarse sand (1.0–2.0 mm). Furthermore, to determine the organic matter content in soil particles larger than 2 mm, the soil samples were first sieved through a 2 mm mesh sieve. Subsequently, the standard potassium dichromate external heating method was employed for quantification. Similarly, when assessing the organic matter content of soil particles with a size exceeding 2 mm, after passing them through a 2 mm sieve, the conventional potassium dichromate external heating method was used for its quantification (Wu and Chen, 2016).

Building upon the Udden-Wentworth scale grain size criterion, Φ the value was derived utilizing the logarithmic transformation method established by Krumbein (1934), as shown in Equation 1 ϕ = − L og 2 D (1) where, D denotes the fractal dimension.

Ward and Folk (1957) established the cumulative percentage of frequency distribution curves based on the logarithmic values of granularity corresponding to 5%, 10%, 16%, 25%, 50%, 75%, 84%, and 95%, respectively: Φ ₅, Φ ₁₆, Φ ₂₅, Φ ₅₀, Φ ₇₅, Φ ₈₄, and Φ ₉₅. Accordingly, the particle size characteristic parameters were calculated as shown in Equations 2–5:

MEAN ( d 0 ) refers to a parameter that indicates the average condition of characteristics of soil particle sizes, determined using the following formula: d 0 = 1 3 Φ 16 + Φ 50 + Φ 84 (2)

Using characteristics of soil particle sizes volume data for multiple fractal dimensions, multiple fractal parameters calculated before the construction of the partition function, the laser particle size measurement interval in accordance with the form of logarithmic isotropic incremental division of 64 intervals, the interval size of ε = 5 × 2 − k K   is an integer in the range  1 − 6 , u ( ε ) is the probability density of the distribution of characteristics of soil particle sizes in each sub-interval. The group of partition functions was obtained as follows (Hentschel and Procaccia, 1983; Posadas et al., 2001): u i q , ε = u i ε q ∑ i = 1 N u i ε q (3)

The multifractal singularity function was calculated by the following equation (Yang et al., 2021; Dong and Zheng, 2010): α q = lim ε → 0 ∑ i = 1 N ε u i q , ε l g ⁡ p i ε l g ⁡ ε (4)

The corresponding multifractal spectral function was calculated as follows: f α q = lim ε → 0 ∑ i = 1 N ε u i q , ε l g ⁡ u i q , ε l g ε (5) where, q is a real number and ui (q, ε ) is the qth order probability of the interval of the i-th.

The obtained soil data were calculated and organized using Excel 2021. Graphical processing was conducted using MATLAB R2018a software. A one-way ANOVA was performed on the characteristics of soil particle size parameters and soil organic matter (SOM) content using SPSS version 27.0. Additionally, Pearson correlation analysis was used to examine the relationships among multiple fractal parameters of particle size and indicators of SOM. For graphing purposes, Origin 2021 software was utilized.

As shown in Table 1, there existed a significant difference in the composition of soil particle sizes between sand dunes and mobile dunes, particularly when considering the varying years of sand barriers installation within the study area. The soil particle composition on mobile dunes was predominantly characterized by sand and silt particles, with fine sand content ranging from 60.38% to 60.93%. The medium sand content varied between 36.62% and 37.91%, while coarse sand content remains relatively low. Both clay and silt contents were extremely minimal, approaching zero. The grain size distribution of soils from dunes with varying durations of sand barrier installation primarily consisted of fine and medium sands. Specifically, fine sand constituted between 30.87% and 67.26%; very fine sand ranged from 20.69% to 37.83%; medium sand fluctuated between 0.16% and 19.09%; coarse sand spanned from 0.16% to 1.98%; and very coarse sand accounted for approximately 1.10%–1.85%. Notably, an increasing trend in clay and silt content was observed with increasing number of years since the installation of the sandy barriers.

Analysis of grain size composition across various dunes revealed that, in comparison to mobile dunes, the content of clay, silt, and very fine sand in dunes formed over different years of sand barriers deposition increased significantly. Conversely, a notable decrease was observed in the content of medium sand and coarse sand. Furthermore, an examination of different positions on the dune revealed variations in soil particle content across different slopes during the same formation period of the sand barriers. Specifically, for particles such as clay, silt, and very fine sand during this uniform laying period, a pattern emerged: bottom slope > middle slope > top slope. Conversely, for medium and coarse sands within the same timeframe, the distribution followed a different trend: top slope > middle slope > bottom slope.

The relationship between α∼f (α) for different slopes of mobile dunes, as well as sand barriers laid for 1 year, 3 years, and 5 years, is illustrated in Figure 2. As shown in Figure 2, the multifractal spectral functions for both the mobile dune and the dunes with different installation durations of sand barriers displayed asymmetric arch-shaped curves. Additionally, there was a local superposition of soil particle content, indicating that as the time since the sand barriers were installed increased, external disturbing forces had caused non-uniform alterations in the distribution characteristics of soil particle sizes. The shape features Δf were consistently exceeding zero. Thus, the graphs displayed a leftward bias, and all demonstrated a left-hooked appearance. This suggested that a large probability subset predominated within these distributions. For the mobile dune and sand barriers laid for 5 years, Δα was wider at the slope base yet narrower at its apex. In contrast, for those laid for only 1 year, it was wider at the bottom of the dune slope yet narrowed towards its top. Meanwhile, for barriers installed over a period of 3 years, Δα exhibits a broader range at mid-slope while being narrower at both ends. Δα represents an indicator of non-uniformity within soil particle-size distribution. Specifically, a wider Δα signifies more pronounced changes in dune particle size and indicates overall enhanced heterogeneity within particle size distribution.

The distribution of organic matter at the top, middle, and bottom of the slopes of sand dunes and mobile dunes with varying years of sand barriers are illustrated in Figure 3. Obviously, the average soil organic matter (SOM) content for mobile dunes, as well as for sand barriers established for 1 year, 3 years, and 5 years, were 0.29 g/kg, 0.55 g/kg, 0.75 g/kg, and 0.78 g/kg, respectively. Notably, with an increase in the duration of barrier establishment, there was a significant upward trend in SOM content. The differences in SOM content between mobile dunes and sand barriers established for different durations (1 year, 3 years, and 5 years) were statistically significant (P < 0.05). However, no significant difference was observed between the SOM contents of sand barriers laid down for 3 years versus those laid down for 5 years. Furthermore, analysis revealed that within different slope positions on dunes with varying durations of barrier establishment, the SOM content tended to accumulate at the bottom slope position, specifically: bottom slope > middle slope > top slope.

As shown in Figure 4, the correlation between soil grain size composition and the SOM content varied significantly across different years after the establishment of sand sand willow barriers. The correlation coefficient between SOM and clay was determined to be 0.74, while the coefficients for fine sand and very fine sand were 0.86 and 0.84, respectively. These values indicated a strong positive correlation in all cases. Conversely, the correlation coefficient between SOM and coarse sand was −0.87, whereas that between SOM and medium sand was −0.75, both reflecting a significant negative correlation. These findings suggested that the enhanced accumulation of fine-grained materials within sand dunes was correlated with elevated levels of SOM content. Moreover, the clay and silt fractions in sand dunes could be concluded as key grain sizes impacting organic matter accumulation.

The study area is situated on the southeastern edge of the Kubuqi Desert, flanking both sides of a sand-penetrating highway in an ecologically fragile and sensitive region characterized by delicate soil structure and overall poor soil nutrients (Gao et al., 2006). As indicated by the present findings, following 3–5 years of establishing sand barriers, the contents of fine sand, clay, and silt particles within these barriers increased significantly compared to those in mobile dunes. Conversely, the contents of medium sand and coarse sand decreased slightly. This change could be primarily attributed to the implementation of Salix Psammophila sand barriers, which altered subsurface characteristics and increased ground surface roughness. As a result, wind-sand flow dynamics as well as surface erosion and accumulation processes were correspondingly affected. The result was an effective reduction in wind speed and intensity of sand transport while simultaneously mitigating nutrient depletion in near-surface layers. Overall, these interventions significantly decreased near-surface wind speeds while reducing the intensity of sand transport. They also slowed down the movement frequency of sand particles within the barriers, stabilized quicksand surfaces, and preserved fine-grained materials contained within the constructed sand barriers (Wen and Zhao, 2004).

Microtopography represents a significant factor influencing the spatial heterogeneity of soil characteristics (Vivekananthan et al., 2018). As a complement to the broader concept of topography, the characterization of microtopography varies across different studies. One aspect focuses on the unique topographical features of gullies and ravines that have developed due to long-term rainfall and surface runoff erosion. This perspective is frequently applied in research regarding loess hilly regions and soil erosion in China (Zhao, 2014; Puyang et al., 2020; Wang et al., 2012). Another aspect pertains to small-scale topographic variations, where microtopography in hilly areas can be classified into seven categories, including slope tops, upper side slopes, valley head depressions, lower side slopes, foothill slopes, floodplain terraces, and valley beds. The third aspect addresses distinct microtopographical features resulting from changes in surface microelevation and undulation, i.e., variations that occur with minimal alterations in relative elevation at small scales on slopes. These changes significantly influence the redistribution of water, heat, light, and nutrients. This perspective is commonly utilized in studies examining vegetation elemental stoichiometry, community composition, diversity distribution, and ecological functionality (Qi et al., 2016; Chen et al., 2020; Zhang et al., 2016). While each approach emphasizes distinct elements of microtopography, they all have in common their concentration on phenomena occurring at the microscopic scale.

The present study falls into the third category, focusing on the analysis of soil nutrients across three dune slopes: the top of the windward slope, the middle slope, and the bottom slope within the study area. During the years when sand barriers were installed, the contents of clay, silt, and very fine sand followed a consistent trend: bottom slope > middle slope > top slope. Conversely, for medium sand and coarse sand, an opposing pattern emerged, with top slope > middle slope > bottom slope. This finding aligned with Xu Tingting’s conclusions (Xu et al., 2014), which indicated that upon installation of sand barriers, there was an increase in clay and silt particle content in the soil. Furthermore with the prolonged installation of sand barriers, the grain size distribution across different slopes gradually approached stability.

Multifractal parameters effectively capture the interception effects of sand barriers on fine-grained materials and variations in soil particle composition, thereby elucidating the local variability and nonuniformity in the characteristics of soil particle size distribution (Meng et al., 2024). In this study area, a comparison of mobile dunes with sand barriers installed for different durations revealed that the singularity spectral function exhibited a phenomenon of superposition in local distributions. This observation further substantiated that as the duration of barrier installation increased, the distribution characteristics of soil particle sizes became increasingly non-uniform. The study by Posadas et al. (2001) conducted an analysis on multifractal spectra derived from 30 particle-size distributions encompassing a broad spectrum of soil textural classes. The findings demonstrated multifractal parameters as promising indicators for describing particle size distribution. Furthermore, various sand control measures exhibited distinct pore structures, which consequently altered the spatial patterns associated with soil particle size distribution characteristics (Howell et al., 2018). The results indicated a significant probability subset demonstrating a predominance of fine sand intervals.

Characteristics of soil particle size distribution and SOM content are critical parameters for evaluating composite soil indices. The relationship between these two factors directly affects the stability of soil structure as well as the advantages and disadvantages of various soil properties. There exists a varying degree of correlation between the characteristics of soil particle size distribution and organic matter content. In this study, the results demonstrated that as the number of years since the establishment of sand barriers increased, a significant correlation developed between the characteristics of soil particle sizes and organic matter content (Dong et al., 2017). The rise in fine particle content within wind-sand areas indicated an improvement in soil texture, revealing a positive correlation between fine particulate matter and organic matter accumulation. As organic matter content increased, it enhanced the bonding strength within the microstructure of the soil, thereby improving its resistance to external forces and increasing its overall resilience against wind erosion. Furthermore, fine-grained materials could adsorb organic matter, contributing positively to SOM accumulation. The underground portion of sand barriers was particularly affected by prolonged exposure to a sand-buried environment, leading to microbial decomposition processes influenced by microorganisms present in the surrounding soils (Wang et al., 2021). Additionally, fluctuations in moisture levels within lower layers of sandy land contributed to a cyclical process involving moisture absorption and desorption. This cyclic moisture movement led to repeated expansion and contraction cycles for the barrier structures, causing ruptures in internal cell walls along with structural damage. As a consequence, the barriers became more vulnerable to microbial decomposition (Liang et al., 2022a; Liang et al., 2022b). Such phenomena might significantly account for notable increases in organic matter content observed in surrounding soils during later stages following barrier establishment.