Relationship Between Salt Accumulation and Soil Structure Fractals in Cotton Fields in an Arid Inland Basin

Abstract

The relationship between soil structure and salt accumulation is unclear; thus, experiments on salt accumulation under different soil structures were conducted in cotton fields in arid areas of northwest China. Thirty-nine sets of soil samples were collected from the 0 to 180 cm profile of three experimental areas. The total salt content of the soil extracts and the particle size distribution of the soil samples were determined using a JENCO TDS and a laser particle size analyzer, respectively, and the fractal dimension of the soil structure was obtained using fractal theory. Pearson’s correlation analysis and Tukey’s test (p < 0.01) were used to analyze the correlation between soil salinity, soil particle size distribution, and fractal dimensions in the three profiles. The results showed soil salinity accumulation was affected mutually by soil texture and soil structure, and soil salinity tended to accumulate in fine-grained soil. The soil fractal dimension (D) could indicate soil texture and quantify soil salinity content. When the sand content was more than 50%, there was a significant positive correlation between the soil fractal dimension and soil salinity (correlation coefficient R = 0.943). The results provide valuable insights into cotton production in arid areas.

1. Introduction

Cotton is an important fiber and oil crop [1]. In 2023, the cotton output in Xinjiang accounted for 91% of the national total output, and Xinjiang is the main producing area of high-quality cotton in China [2]. Salt stress damages the growth and development of cotton. Although cotton is a moderately salt-tolerant plant, the germination rate decreases when the soil salinity exceeds 10 dSm⁻¹, and when the soil salinity is 15 dSm⁻¹–20 dSm⁻¹, the seedling stage is delayed, which causes a decrease in cotton planting quantity. With the increase in soil salt content, the soil layer with a high content of fine particles clearly inhibits root growth [3]. The high concentration of salt solution in the soil delays the flowering of cotton, inhibits photosynthesis and enzyme activity, reduces boll number and boll weight, decreases cellulose content, and reduces cotton yield [3,4]. Compared with single abiotic stress, the combination of arid climate and salinity in Xinjiang will cause greater yield loss of cotton [1,3,4].

To make up for the low rainfall, drip irrigation under mulch, a relatively new micro-irrigation technology combining surface drip irrigation and drip irrigation under mulch, has been widely used in cotton fields in the arid regions of Xinjiang Province, China, since 1996 [5,6]. Drip irrigation accurately delivers water and nutrients to plant roots. Plastic film mulching can increase soil temperature, reduce contact between soil and turbulent air, and eliminate soil evaporation and transpiration [7,8,9]. Drip irrigation under a plastic film can increase the soil water content and adjust the soil salt distribution [7,8,10]. The distribution of soluble salts in the soil profile follows the water flux model, and the soil near the drip irrigation forms a desalting zone during leaching. The depth of the wetting front of drip irrigation under plastic film is low, and salt accumulates in approximately 50–60 cm of moist soil [6]. With increasing irrigation, salt migrates to the deep soil. With an increase in drip irrigation duration under plastic film, soil salinity undergoes three stages: rapid desalination, slow desalination, and stable desalination [6,11]. The position of accumulated soluble salts differs according to agricultural management methods. An increase in the horizontal planting length of cotton fields reduces the uniformity of drip irrigation, leading to strong spatial variation in the soil salinity content [12]. Under the planting regime of “one tube, one film, and four rows”, the soil salinity content is highest in the inter-film area, while under the planting regime of “two tubes, one film, and four rows”, the soil salinity contents in the inter-film area and wide rows are higher than that in the narrow rows [5]. The initial salt content of the soil is a key factor affecting the temporal and spatial distribution characteristics of soil salinity [12]. The salt content in the 0–100 cm soil layer before cotton sowing is related to the irrigation quota of the cotton growth period in the previous year. The risk of salinization increases if irrigation is not performed in winter or spring [13].

Long-term drip irrigation under plastic film not only reduces soil salinity but also changes the salt composition [13,14]. However, long-term mulch drip irrigation and mechanical tillage change the soil structure, making soil particles finer [6]. With the same lithology, the composition of soil particles differs, and the distribution of fine particles in the soil profile is an important factor affecting the vertical distribution of salt [15]. Soil moisture migration changes with fine particle content [16], and the higher the fine particle content, the lower the soil moisture infiltration rate [17]. Soil texture analysis is a key component in evaluating the sustainable development of agriculture. The rising height of capillary water is determined by the soil texture, and the migration and distribution of salt can significantly differ in soils with different textures [18,19,20]. The layered distribution characteristics of soil particle size dominate the vertical distribution of soil salinity [6]. Because of its low porosity and strong water-holding capacity, the distribution of clay interlayers in the profile strongly affects the migration and redistribution of salts [21,22]. Structure is a basic physical and morphological characteristic of soil. In layered soil, water and salt infiltration are not only influenced by stratum lithology and thickness but also by spatial structure [23,24,25]. Soil with a uniform texture has a high cumulative infiltration [17,26]. When water passes through soil layers with different textures, the hydraulic properties are discontinuous owing to changes in the soil properties, which affects the convection, diffusion, and migration of salt [23].

The overall characteristics of the particle size distribution of soil can be used to effectively assess the soil structure, and changes in the profile can be used to accurately estimate the hydraulic characteristics of the soil and describe the migration of soil solutes and the degree of soil chemical weathering [27]. Fractal theory can be used to describe the relationships between the soil particle size distribution and soil physical and chemical properties [28], and is an effective tool for describing soil particle size distribution and soil structure. The fractal dimension (D) is a parameter of soil particle geometry that can quantitatively distinguish subtle differences between different soils, describe the overall characteristics of soil structure, and reflect the uniformity of soil particles [29,30]. The fractal dimension of soil particle size is closely related to the soil salinity content, and the location of accumulated salt in the soil also differs because of topography [30,31] and soil parent material [32].

Gui et al. [33] investigated the soil conditions under different vegetation types on the northern slope of the Kunlun Mountains in Xinjiang at an altitude of 1960–4070 m and found that the fractal dimension of soil particle size had a significant negative correlation with the total salt content of the soil. Zhao et al. [19] studied the variation in the soil particle size distribution and soil salinity in the lower reaches of the Heihe River and found a linear relationship between the fractal dimension and soil salinity. Bai et al. [28] measured the physical and chemical properties of six alkaline soils in the western Songnen Plain, northeast China, and concluded that the fractal dimension had significant positive correlations with the total soil salinity and ${\mathrm{H}\mathrm{C}\mathrm{O}}_3^{-}$, which could be used as a potential indicator of soil alkalization and alkaline soil structure variation.

The distribution characteristics of soil salinity at different depths are important indicators of the normal growth of cotton [34]. Previous studies on the relationship between soil salinity accumulation and soil particle size were focused on revealing the correlation between these across a wide range of research areas. However, comparative analyses of soil particle sizes in different sections of the same area were lacking; thus, the influence of soil structure and soil texture on salt accumulation position was ignored. Consequently, this study set out to explore the relationship between soil D and soil salt content. Three different salt profile types were identified, the vertical distribution characteristics of soil salinity in the Xinjiang irrigation area were thoroughly studied, and the condition of linear correlation between soil D and soil salt content was put forward quantitatively. Describing accumulated soil salinity in different soil texture combinations can provide information regarding differentiated sources of salt accumulation in different soil structures, help prevent secondary salinization caused by inappropriate irrigation practices in Xinjiang, reduce the phenomenon of agricultural land abandonment, and provide guidance for improving cotton yield.

2. Materials and Methods

2.1. Study Area

The experiment was conducted at the Institute of Water Conservancy and Scientific Research, Tarim River Basin Administration, Bayinguoleng Mongol Autonomous Prefecture, Xinjiang, China. This is located in central Xinjiang, at the southern foot of the Tianshan Mountains and the northeastern margin of the Tarim Basin. The altitude is 988–991 m a.s.l., and the general trend of the terrain is high in the North and low in the South. It has a warm temperate extreme continental arid desert climate with drought, little rain, high evaporation, long sunshine exposure, and sufficient light. The annual average temperature is 11.5 °C, the frost-free period is 191 d, the sunshine hours are 3036.2 h, the annual average rainfall is 53.3–62.7 mm, the potential evaporation is 2273–2788 mm, and the groundwater level of the test station is approximately 16 m. Cotton is planted at Plots 1, 6, and 8 at the station, and irrigation comes from local canal water and groundwater, with canal water sourced from the Peacock River.

2.2. Sampling and Analysis

2.2.1. Soil Sample Collection

Samples were taken from three sampling points at the test station: Plot. 1 (86.180216° E, 41.596967° N, 842.1 m a.s.l.); Plot. 6 (86.166460° E, 41.580605° N, 901.9 m a.s.l.); Plot. 8 (86.174886° E, 41.586451° N, 904 m a.s.l.). The location of the test station and the sampling points is shown in Figure 1.

Soil samples were collected with a stainless-steel auger at depths of 0–10 cm, 10–20 cm, 20–30 cm, 30–40 cm, 40–50 cm, 50–60 cm, 60–70 cm, 70–80 cm, 80–100 cm, 100–120 cm, 120–140 cm, 140–160 cm, and 160–180 cm [5,10,15]. Each soil layer was sampled three times, and thirty-nine sets of soil samples were obtained. Samples from each layer were uniformly collected and placed in a sample bag, with the sampling time, sampling location, soil depth, and other information recorded [28]. After impurities, such as plant roots and gravel, were removed from the collected samples, each layer of the sample was fully mixed, rolled, crushed, and air-dried for later use.

2.2.2. Soil Salinity and pH Measurements

Soil (30 g) was placed in a 250 mL triangular flask, to which 150 mL of ultrapure water was added, and a soil leaching solution was prepared with a water–soil ratio of 5:1 [6,10,13]. The sample was shaken in a water-bath shaker for 60 ± 10 min. After standing for 1 h, 15 g of the extract was transferred into a centrifuge tube and centrifuged for 15 min at 4000 r.min⁻¹. To avoid the interference of tiny impurities in the leachate with the electrode, the centrifuged liquid was filtered using a diaphragm vacuum pump [11]. The filter membrane was an aqueous microporous filter membrane with a diameter of 50 mm and a pore size of 0.22 μm.

The electrical conductivity (EC_1:5) and pH of the soil were measured using a pH-EC water quality tester (H1991301, HANNA instruments, Villafranca, Padovana, Italy), and the total dissolved solids (TDS) of the soil were measured using a JENCO TDS (3010M, Ren Shi Electronics Co., Ltd., Shanghai, China) [10,17]. The measurement was repeated three times for each soil sample and the average value was obtained as the final result. Linear fitting was performed between the total salt content of the soil and conductivity of the leachate of the different plots in the study area, as shown in Figure 2. There was a significant positive correlation between the TDS and EC_1:5 conductivity at Plots 1, 6, and 8.

2.2.3. Soil Particle Size Distribution Determination

After air-drying, the soil samples were screened through a fine screen of 2 mm diameter, and soil particles larger than 2 mm were weighed and recorded [12]. A 30% H₂O₂ solution was added to the screened soil samples, and the organic matter in the soil samples was digested using the heating method. After the reaction was completed, 30% HCL solution was added to remove organic carbon. When the pH of the soil solution was less than 5.5, 0.5 mol/L sodium hydroxide solution was added; when the pH was 6.5–7.5, 0.5 mol/L sodium oxalate solution was added; when the pH was greater than 7.5, 0.5 mol/L sodium hexametaphosphate was added. Then, the aggregates were fully dispersed after standing. After that, the laser particle size analyzer (QT-2002, Channel Scientific Instrument Co., Ltd., Beijing, China) was used to process the dispersed samples for 30 s of shading and ultrasonic treatment to obtain the particle size distribution characteristics of the soil samples in the range of 0.05–800 um [19,27].

2.2.4. Fractal Analysis Method for Soil PSD

The volume fractal dimension of the soil particle-size distribution was calculated according to the method proposed by Tyler and Wheatcraft [35]:

$${\displaystyle \frac{\mathrm{V}\left(\mathrm{r}<\mathrm{R}\right)}{{\mathrm{V}}_{\mathrm{T}}}}={\left({\displaystyle \frac{\overline{\mathrm{R}}}{{\mathsf{\lambda }}_{\mathrm{V}}}}\right)}^{3-\mathrm{D}}$$

(1)

where r is the soil particle size measured by a laser particle size analyzer, R is a specific soil particle size, $\mathrm{V}\left(\mathrm{r}<\mathrm{R}\right)$ denotes the volume of particles smaller than R, ${\mathrm{V}}_{\mathrm{T}}$ is the total volume of soil particles, $\overline{\mathrm{R}}$ is the particle between two screening fractions ${\mathrm{R}}_{\mathrm{i}}$ and ${\mathrm{R}}_{\mathrm{i}+1}$, ${\mathsf{\lambda }}_{\mathrm{V}}$ is the maximum diameter of soil particles, and soil D is the volume fractal dimension of the soil.

Logarithmic transformation was performed on both sides of the above formula, and the obtained values were plotted in a rectangular coordinate system. Taking $\mathrm{l}\mathrm{g}\left(\overline{\mathrm{R}}/{\mathsf{\lambda }}_{\mathrm{V}}\right)$ as the abscissa and $\mathrm{l}\mathrm{g}\left[\left(\mathrm{r}<\mathrm{R}\right)/{\mathrm{V}}_{\mathrm{T}}\right]$ as the ordinate, the slope k of the straight line was obtained, and then the volume fractal dimension of the soil particles was D = 3 − k.

2.2.5. Soil Grading Evaluation

The soil particle size distribution reflects the range of soil particle sizes, and the Cu and Cc can determine the quality of soil classification. The continuity of the soil particle size grade was evaluated according to the Unified Soil Classification System in ASTM D 2487-11 [36]:

$$\mathrm{C}\mathrm{u}={\displaystyle \frac{{\mathrm{d}}_{60}}{{\mathrm{d}}_{10}}}$$

(2)

$$\mathrm{C}\mathrm{c}={\displaystyle \frac{{\mathrm{d}}_{30}^2}{{\mathrm{d}}_{60}\times {\mathrm{d}}_{10}}}$$

(3)

where ${\mathrm{d}}_{10}$, ${\mathrm{d}}_{30}$, and ${\mathrm{d}}_{60}$ are the effective particle size, median particle size, and constrained particle size (μm), respectively. The soil with Cu ≤ 5 is homogeneous with a single particle size composition and poor grading. When Cu > 5 and 1 < Cc < 3, the soil classification is good; otherwise, it is poor.

2.2.6. Statistical Analysis

Microsoft Excel 2021 and IBM SPSS Statistics 26.0 software (IMB, Chicago, IL, USA) were used for data descriptive statistics. Pearson correlation analysis and Tukey’s test (p < 0.01) were used to analyze the correlations among the soil particle size distribution, fractal dimension, total salt content, and electrical conductivity in the 0–180 cm soil layer. Soil particle size grading was performed using a particle diameter measurement and analysis system. Line charts, histograms, and soil triangle structure charts were drawn using Origin 10.1 (OriginLab Corporation, Northampton, MA, USA) and Corel draw 23.1 (Corel Corporation, Ottawa, ON, Canada).

3. Results

3.1. Distribution Characteristics of Soil Particle Size

The study area is located inland in northwest China and experiences drought, little rain, and obvious wind erosion. Soil particle size composition is a natural property of soil stability and is divided by the soil particle size content [37]. According to the soil texture classification system of the USDA, the soil particle size distribution data in the study area were divided into clay (0–2 μm), silt (2–50 μm), sand (50–200 μm), and gravel (200–300 μm) [38], and the specific distribution characteristics are shown in

Table 1. Soil particle size volume fraction (Means ± SD, n = 13).

Number	Soil Particle Size Distribution
	Clay 0–2 μm%	Silt 2–50 μm%	Sand 50–200 μm%	Gravel 200–300 μm%
Plot. 1	0.922 ± 2.198	36.765 ± 29.970	62.289 ± 31.450	0.024 ± 0.087
Plot. 6	12.990 ± 19.414	64.909 ± 25.658	22.062 ± 28.648	0.199 ± 0.345
Plot. 8	5.855 ± 19.021	56.944 ± 24.730	37.195 ± 26.040	0.026 ± 0.072
Universe	6.589 ± 16.128	52.873 ± 28.786	40.515 ± 32.681	0.083 ± 0.221

. The soil was mainly composed of silt and sandy soil, with volume fractions of silt and sandy soil of 52.873% and 40.515%, respectively. The volume fractions of clay and gravel were relatively low at 6.589% and 0.083%, respectively. The soil textures included at the test station were sandy soil, loam sand, sandy loam, clay loam, clay, silty clay, and silt. The soil texture classifications of Plots 1, 6, and 8 are shown in Figure 3.

The proportion of silt in the 0–40 cm layer was highest because of the refinement of soil particles by farming practices. Changes in the clay, silt, sand, and gravel volume fractions in Plots 6 and 8 were similar. Silt had the highest volume fraction, followed by sand, clay, and gravel. Plot. 1 differed from the other two plots in that the volume fraction of sand was the highest, followed by silt, clay, and gravel. In addition, there were clear differences in the soil particle size distribution among the three plots. The volume fractions of clay, silt, and gravel all followed the order of Plot. 6 > Plot. 8 > Plot. 1. In contrast, the volume fraction of silt content followed the order of Plot. 1 > Plot. 8 > Plot. 6.

There was an obvious negative correlation between sand and silt in the three study areas (Plot. 1, p < 0.001, R = −0.999; Plot. 6, p = 0.003, R = 0.749; Plot. 8, p = 0.005, R = −0.072). There was a negative correlation between silt and clay content in Plot. 1 (p = 0.01, R = −0.681), but no such correlation in Plots 6 and 8 (p > 0.05).

To further analyze the distribution of soil particle size, the uniformity coefficient (Cu) and curvature coefficient (Cc) were used to determine whether the distribution of soil particle size was continuous. The average values of Cu and Cc, respectively, obtained by formulas (2) and (3) were as follows: Plot. 1: 4.001, 1.038; Plot. 6: 14.977, 1.004; Plot. 8: 6.359, 1.148. Among these, Plots 6 and 8 met the conditions of Cu > 5 and 1 < Cc < 3. According to the Cc value, they were well-graded, whereas Plot. 1 was poor. The Cu value of Plot. 6 was almost three times greater than five, indicating that the soil lacked some soil particle size distribution [28].

3.2. Fractal Characteristics of Soil Structure

Using fractal theory, the fractal dimension of the soil particle sizes was calculated according to formula 1 to quantify the soil structure, that is, the soil structure fractal. The fractal structures of soil samples (0–180 cm depth) at Plots 1, 6, and 8 were calculated, and the fractal dimension of the soil particle size distribution was statistically described. The variation coefficient (CV) reflects the dispersion degree of sample points, where CV < 0.1 is weak variation, 0.1 ≤ CV ≤ 1 is moderate variation, and CV > 1 is strong variation [39]. The results showed that the fractal dimensions of the soil particles in the three study areas varied moderately, ranging from 0.301 to 2.705, with an average value of 1.566. There were many soil texture types in the test station, which caused the fractal dimension of the soil particles to change over a large range. The average fractal dimension of the soil particles was as follows: Plot. 6 > Plot. 8 > Plot. 1, as shown in

Table 2. Descriptive statistics of fractal dimension of soil structure.

Number	Minimum	Maximum	Average	SD	CV
Plot. 1	0.301	1.900	1.259	0.531	0.422
Plot. 6	0.534	2.655	1.795	0.509	0.284
Plot. 8	0.447	2.705	1.643	0.465	0.283
Universe	0.301	2.705	1.566	0.540	0.345

.

The fractal dimension of soil particles in Plot. 1 decreased with increasing soil depth (Figure 4B), whereas the fractal dimension of the soil in Plot. 8 did not change significantly with depth (Figure 4A). Although the range of soil D values between Plots 6 and 8 at 0–180 cm depth was small, their changes with depth differed. In Plot. 6, the fractal dimension of soil particles at 60–70 cm and 80–100 cm was higher, while it was lower at 70–80 cm. In contrast, the fractal dimension of soil particles in Plot. 8 was highest at 60–70 cm and lowest at 70–80 cm.

The fractal dimension of the soil in Plots 1, 6, and 8 was the same as the change in the fine particle content in the soil. The higher the fine particle content, the greater the fractal dimension. The fractal dimension decreased with an increase in the sand content. However, there were some differences between soil D and soil particle size distributions in the three study areas. There was a significant positive correlation between soil D and the clay content at Plots 6 and 8 (

Table 3. Correlation analysis between soil D and soil particle size in different research areas.

Number	Clay		Silt		Sand
	p	R	p	R	p	R
Plot. 1	0.081	0.501	<0.001	0.910	<0.001	−0.903
Plot. 6	0.001	0.790	0.411	0.249	0.003	−0.759
Plot. 8	0.009	0.694	0.274	0.328	<0.001	−0.818

). Plot. 1 was different, and soil D was positively correlated with the silt content. This indicates that soil particle composition is an important factor affecting soil D. However, the fractal dimension of the soil was not related to the total soil particle size, primarily because the volume percentages of clay, silt, and sand were different.

3.3. Soil Structure Fractal and Salt Distribution

The salt distribution characteristics corresponding to soil profiles of Plots 1, 6, and 8 are shown in Figure 5. Salt storage in the 0–10 cm soil layer in Plot. 1 was the lowest (57.51 mg/L), accounting for 5.57% of the salt storage of the whole profile. The change in the soil salinity content at 20–180 cm depth was complex and did not simply increase with depth. The highest salt content was 99.47 mg/L in the 30–40 cm soil layer. The salt distribution characteristics belonged to the bottom aggregation type, and a soil desalination effect was evident. The distribution characteristics of the soil salinity content at Plots 6 and 8 showed a decrease along the profile. Evaporation was high in the study area. After spring irrigation, with the evaporation of soil moisture, salt accumulated above the soil layer, and the salt storage capacity of the surface soil was the largest. The surface soil salinity of Plots 6 and 8 was 118.5 mg/L and 150.43 mg/L, accounting for 10.53% and 15.33% of the total salt content, respectively.

Plots 1, 6, and 8 are located in the same area and experience the same climatic conditions and groundwater depth, which fluctuates between 16 and 20 m throughout the year. Because of the different distributions of soil particle size, the distribution of soil salinity in the profile also showed significant differences among plots. The difference in the initial salt content resulted in the total soil salinity in Plot. 6 being the highest, followed by Plots 1 and 8 (

Table 4. Descriptive statistics of total soil salinity content.

Number	Average (mg/L)	Maximum (mg/L)	Minimum (mg/L)	SD	Aggregate (mg/L)	CV
Plot. 1	79.377	99.470	57.510	14.067	1031.9	0.177
Plot. 6	86.593	118.500	52.140	20.559	1125.71	0.237
Plot. 8	75.496	150.430	48.320	25.906	981.45	0.343

). The distribution of soil salinity at 0–180 cm in the three plots showed moderate variation, and the coefficient of variation of the soil salinity content was in the order Plot. 8 > Plot. 6 > Plot. 1. As shown in the fractal dimension results, the variability in the soil structure in Plot. 1 was the largest, followed by Plot. 6 and Plot. 8, indicating that soil structure affected the distribution of soil salinity at this experimental station.

Water is used as a carrier for salt transport. Soil texture and structure jointly determine the height and intensity of soil water movement in the vertical direction and indirectly affect the spatial distribution of soil salinity [40,41]. Soil texture heterogeneity controls the location of salt accumulation in the soil [16,42,43], and soil structure determines the infiltration rate of soil moisture. Coarse-grained soil has high permeability, which is beneficial for salt leaching, whereas fine-grained soil has strong water-holding and adsorption capacities with a slow infiltration rate, and usually has high water and salt contents [17,23,41].

Leaching and evaporation affected the accumulation and distribution of salt in Plot. 1. As sampling occurred five days after spring irrigation, the influence of evaporation was less than that of leaching. The soil particles in Plot. 1 gradually became thicker with increasing depth, and the soil salinity in the 0–20 cm soil layer was significantly reduced by spring irrigation. The lithological boundary that formed at 40 cm hindered the downward migration of water [16]. When water moves from fine-grained to coarse-grained soil, it cannot directly pass through the interface because of weak water absorption by the 0–40 cm capillary barrier layer. After spring irrigation stops and once the capillary breaks, the 30–40 cm stores more soil moisture [23]. The sand content of the 40–180 cm soil layer was higher, which was conducive to the continuous infiltration of soil moisture, and salt gradually accumulated at 180 cm. Under the combined influence of evaporation and leaching, 99.47 mg/L of salt accumulated in the 30–40 cm soil layer, which was as high as 9.64% of the total salt.

Plots 6 and 8 were mainly affected by evaporation. During evaporation, the soil structure of the coarse particles in the upper layer and fine particles in the lower layer produced capillary fractures. For example, the salt content of the 50–60 cm layer of silt in Plot. 8 was 72.6 mg/L, which was higher than that of the 40–50 cm silty loam (56.51 mg/L). However, the salt accumulation between these plots was very different, and the salt content at 0–10 cm depth was higher in Plot. 8 than it was in Plot. 6. This is because the overall clay content in Plot. 6 was greater than that of Plot. 8, while the sand content was opposite this. In the soil structure, the clay layer acts as a semi-permeable barrier, which hinders the movement of water and salt [16,44]. The higher coarse-grained soil content in Plot. 8 enhanced the permeability of the soil [45,46], making the hydraulic properties of the soil profile continuous during evaporation, and thus more salt accumulated on the surface of the soil. After spring irrigation, the soil salinity of Plots 1, 6, and 8 was less than the salt tolerance threshold of cotton [1,3,4]. Therefore, the soil salinity in the study area will not affect the normal germination of cotton.

Plots 1, 6, and 8 were sampled in spring. To avoid the influence of mechanical tillage, correlations between soil salinity content, soil particle size distribution, and soil D were analyzed after removing the 0–40 cm soil layer [47]. In Plot. 1, soil salinity positively correlated with the silt content (R = 0.696, p = 0.037) and negatively correlated with the sand content (R = −0.694, p = 0.038). There was a negative correlation between soil salinity and sand content in Plot. 6 (R = −0.784, p = 0.012). There was no correlation between the soil salinity content and soil particle size distribution in Plot. 8. Soil D had a significant positive correlation with soil salinity content in Plot 1 (R = 0.943, p < 0.001), but this correlation was not significant in Plots 6 and 8. A scatter plot of the soil salinity content and fractal dimensions of the three plots is shown in Figure 6. The fractal-dimension values of Plots 6 and 8 showed little fluctuation, while the soil salinity content changed significantly. From the soil particle size distribution data, it can be seen that Plots 6 and 8 had higher fine-grained soil contents and lower fractal dimension variability. Compared with the soil structure, the distribution of soil salinity in Plots 6 and 8 was more easily affected by soil texture. Plot. 1 was classified as poor and had high sand content. The distribution of salt in Plot. 1 was significantly affected by the soil structure, and the correlation between the fractal dimension and salt content was significant.

4. Discussion

Soil particle size distribution can indicate the processes of soil compaction and formation [26], in which land-use type is the main factor causing changes in the soil particle size [48,49,50]. Coarse-grained soil gradually becomes finer after human cultivation and agricultural machinery polishing. The sand content in all three plots was negatively correlated with the fine-grained soil content. The high fine particle content in Plots 6 and 8 gave the soil a large specific surface area; therefore, it had a strong adsorption effect on salt. Under the same terrain, the soil salinity content of Plots 1 and 6 increased with an increase in the fine grain content, which was consistent with previous research [5,16]. However, the coefficient of variation of soil salinity in Plot. 8 was very large, and there was no correlation between it and the particle size distribution.

The fractal dimension represents the spatial structural characteristics of the soil formed by long-term evolution [41], which is closely related to the heterogeneity of soil particle sizes [28,31,51]. In this study, the single fractal dimension of the soil in the three cotton fields under the same tillage conditions was calculated to compare the soil’s structural characteristics. The results showed that the spatial variability of the soil’s physical properties was large within a small range, and the fractal dimension decreased with an increase in the sand content. The correlation between the soil D and soil particle size distribution also varies with the proportion of soil particles [52,53]; that is, there was a positive correlation between the soil salinity content and silt content in Plot. 1 soil, while the salt content from Plots 6 and 8 was positively correlated with clay.

Sand was the main component of the soil particle size distribution in Plot 1, and there was an obvious positive correlation between the soil salinity content and fractal dimension here, which was consistent with the results of Zhao et al. [19] and Bai et al. [28]. Zhao et al. [19] found that the linear relationship between soil salinity (Y) and soil fractal dimension (X) was Y = 432.05X − 1080.30, R² = 0.44. Bai et al. [28] reported that soil fractal dimension was significantly positively correlated with salt content (p < 0.05), and a significant positive correlation with the HCO³⁻ (p < 0.01). Bai et al. [28] subdivided the salt types; however, these two studies only described the mathematical features of the positive correlation between soil salt content and fractal dimension. When the proportion of sand was more than 50%, the soil structure dominated the vertical distribution characteristics of salt, and the soil salinity content increased with increasing fractal dimension. With an increase in the fine particle content, the correlation between the soil salinity content and fractal dimension weakened or even disappeared. Gui et al. [33] tested soil samples from the northern slope of the middle section of the Kunlun Mountains in Xinjiang and found that the sand content of Plots 1–7 plots was large, while clay in Plots 8–21 plots dominated the soil particle size distribution. After calculating soil D, no obvious differences were observed in some of the sample plots. Although the soil salinity content was negatively correlated with the fractal dimension, the R-value was lower (0.421). This is the same as the research results of Plots 6 and 8. In this study, the silt contents in Plots 6 and 8 were higher, and the fractal dimension did not change significantly with depth. The distribution of soil salinity was affected by the soil texture, and there was no correlation between the fractal dimension and salt content.

In addition, the research results of Gui et al. [33] are contrary to those of Zhao et al. [19] and Bai et al. [28]. Gui et al. [33] found that soil salinity mainly accumulated in coarse-grained soil at low altitudes, which was negatively correlated with the soil salinity content and fractal dimension. This is because the soil parent material and geographical location also influence the distribution of salt accumulation [16]. Therefore, when the sand content in the soil is greater than 50%, the fractal dimension can quantitatively describe the accumulation of soil salinity.

5. Conclusions

By sampling the 0–180 cm soil profile in three cotton fields in the northwest arid area of China, the distribution of soil salinity and soil particle size was statistically analyzed using fractal theory, Pearson correlation analysis, and Tukey’s tests, and the accumulation characteristics of soil salinity under different structures were revealed. It is clear from the present study that there was a significant negative correlation between soil D and the sand content. Furthermore, when the soil sand content was more than 50%, soil D was positively correlated with the salt content of the soil. Our results show that fractal dimension can not only indicate soil texture but also quantify soil salinity content. In the vertical section, different soil textures and soil structures control the migration and distribution of soil salinity. Therefore, in order to ensure the timely emergence of cotton and improve the germination rate, the amount of spring irrigation should be controlled at 1800–3600 m³·hm⁻². It can reduce the salt content in the fine-grained soil and provide a normal soil environment for cotton growth and development.