Effects of Contour Antislope Terracing on Preferential Soil Flow in Sloping Cropland in the Alpine Valley Area of Southwest China

Abstract

This study was conducted to reveal the response relationship between soil preferential flow characteristics and soil pore structure of sloping cropland under contour antislope step measures in the alpine canyon area of Southwest China. In the sub-watershed of Nantangjing, Yunlong County, the upper and lower slopes of primary sloping cultivated land (PSC) and contour reverse-slope terraced rectified land (CR) were used for the study, and a field staining tracer test was used to compare the differences in preferential flow morphology between different slopes with and without measures. The maximum infiltration depth of preferential flow under the contour reverse-slope terrace land preparation reached 21 cm. The stained area ratio tended to decrease with increasing soil depth. Compared to the original slope farmland, the stable infiltration rate under land preparation increased from 0.017 to 0.244 cm³·s⁻¹, and the maximum macroporosity increased by up to 17.00%. Furthermore, land preparation measures significantly enhanced the correlation between macropore quantity and preferential flow characteristics, with the highest correlation coefficient reaching 0.98. And the soil factors in total porosity, total nitrogen and organic matter were particularly influential on preferential flow. Contour antislope terracing promotes the formation and development of preferential soil flow by remodeling soil structure and optimizing pore network distribution.

1. Introduction

Preferential flow refers to the rapid downward movement of water through macropores or fractures in soil [1]. As the dominant mechanism of water movement in unsaturated soils, it significantly influences water redistribution, nutrient transport, and the migration of subsurface pollutants [2,3]. When precipitation intensity surpasses the soil’s suction threshold (typically a water column pressure of 10–30 cm), water no longer infiltrates following Darcy’s law but instead travels along preferential pathways formed by wormholes, root channels, fractures, and similar structures. The development of preferential flow is strongly affected by soil structure and texture, land use patterns [4,5,6], and the volume of infiltrating water. Soil macropores, key components of soil structure, are shaped by biological activity, human intervention, and physicochemical processes [7,8]. The abundance of macropores plays a critical role in water transport, as a higher density of macropores increases the likelihood of preferential pathways and enhances soil permeability [9]. The high mountain and canyon region of Southwest China lies in the transitional zone between the eastern margin of the Qinghai–Tibet Plateau and the Sichuan Basin. Ecological fragility in this area results from combined geological, climatic, and hydrological pressures. Complex terrain, frequent elevation changes (with slopes often exceeding 10°), and high gravel content restrict the widespread application of slope-to-terrace conversion. Furthermore, the alternating wet and dry climate, along with intense hydraulic erosion, contributes to severe soil degradation. During heavy rainfall events, the risk of landslides and debris flows increases, intensifying soil erosion [10,11]. Previous research has shown that contour reverse-slope steps can effectively conserve soil and water by intercepting runoff, enhancing soil moisture retention, and promoting vertical infiltration [12]. These measures also encourage the development of soil macropores and preferential flow channels. The present study aims to clarify the coupled regulatory effects of contour reverse-slope terrace practices on soil structure, macropore formation, and preferential flow behavior. Findings are expected to provide theoretical support for ecological restoration efforts in the high mountain and canyon areas of Southwest China and offer practical strategies for mitigating soil and water loss in sloping farmland ecosystems while promoting sustainable agricultural development.

In recent years, research on the effects of contour reverse-slope terraces on soil physicochemical properties and preferential flow has gained increasing attention within the academic community. Existing studies have shown that this land preparation technique improves the soil environment through multiple mechanisms. Specifically, regarding soil physicochemical properties, contour reverse-slope terraces significantly increase soil aggregate content [13], optimize pore structure to enhance water retention, and adjust organic matter levels to improve nutrient-holding capacity [14,15]. In a case study by Liu et al. [16], the implementation of contour reverse-slope terraces significantly enhanced soil fertility and improved soil physical properties. In the field of preferential flow research, numerous studies have examined different soil types and ecological zones. For instance, Zhang et al. [17] used dye tracer experiments to characterize preferential flow in loess hilly soils of the Loess Plateau, finding that mixed forest land exhibited the most developed preferential flow, with root distribution being the dominant contributing factor. Hou et al. [18] investigated stony soils in the karst region of Yunnan and demonstrated that rock fragments exert both facilitating and inhibiting effects on preferential flow. Yan et al. [19] investigated the impact of land use types and found that reclamation reduces the depth of preferential flow, thereby contributing to groundwater contamination in karst regions. Wan et al. [20] observed that alternating dry and wet conditions in dry-hot valleys increase the number of soil macropores, thereby enhancing preferential flow. Kan et al. [21] found that artificial forest lands in the karst region of Southwest China contain well-developed root systems and dense channel networks, resulting in more active preferential flow processes. This reflects that the contour antislope step can significantly increase the content of soil aggregates and optimize the pore structure, which is the core carrier of the development of preferential flow. The formation of preferential flow is closely related to the well-connected pore space in the soil, and the contour antislope step can reduce the pore blockage caused by soil compaction by improving the stability of aggregates and optimize the pore size distribution and connectivity to provide a smoother path for preferential flow. Although previous research has provided a broad understanding of preferential flow across soil types and restoration contexts, systematic studies focused on the high mountain and canyon regions of Southwest China remain limited. The quantitative effects of contour reverse-slope terraces on macropore development and preferential flow formation, mediated by changes in soil structure, have not been clearly defined. These unresolved scientific questions require further investigation through integrated field studies, laboratory experiments, and numerical modeling.

To address the identified knowledge gaps, the present study investigates typical sloping farmland in Yunlong County, located in the high mountain and canyon region of Southwest China. Characterized by complex topography and geology, the area is highly susceptible to soil erosion and water loss. Two land conditions were examined: contour reverse-slope terrace preparation and adjacent undisturbed sloping farmland. Field dye-tracing experiments were conducted to obtain vertical soil profiles, which were subsequently analyzed using image processing techniques and statistical methods. Soil samples were collected to assess physicochemical properties, and water infiltration experiments were performed to quantify macropore number and macropore ratio. The study focuses on the coupled relationships among changes in soil structure, preferential flow path characteristics, dye coverage ratio, and macropore distribution in steeply sloped farmland. The objective is to elucidate the regulatory mechanisms through which contour reverse-slope terraces influence preferential flow dynamics. This research seeks to provide theoretical support for understanding soil hydrological processes in the region and offer a scientific basis for soil and water conservation as well as ecological restoration strategies for sloping farmland.

2. Study Area and Research Methodology

2.1. Study Area

The study was conducted in the Nantangjing small watershed near the Baoluo Reservoir in Yunlong County, Dali Prefecture, Yunnan Province, China (99°29′56″ E, 26°05′34″ N). The region is part of the southwestern high mountain and canyon zone. An overview of the study area is presented in Figure 1. Yunlong County lies in the western part of Yunnan Province, where a significant proportion of arable land is distributed on mountainous slopes. The experimental site is located in the northern section of the county, characterized by red soils with a neutral pH. The local climate is classified as a subtropical plateau monsoon, with pronounced seasonal variation. The rainy season extends from June to October. Vegetation in the area primarily consists of arboreal species such as Juglans, Pinus yunnanensis, along with herbaceous plants including Bidens pilosa, Urtica laetevirens.

2.2. Materials and Methods

2.2.1. Plot Layout and Soil Sampling

To investigate preferential flow dynamics under two land use conditions, four representative plots were selected based on a comprehensive consideration of elevation, rainfall, slope aspect, and gradient. The plots included both upper and lower slope positions of contour reverse-slope terraces and adjacent undisturbed sloping farmland. In July 2024, a 20 m × 20 m experimental plot was established at each of the four sites for baseline habitat assessment. Within each plot, six standard 5 m × 5 m quadrats were delineated to conduct field dye-tracing experiments.

2.2.2. Dye Tracer Experiment

The basic parameters of the plot are shown in

Table 1. Basic status of the sampling site.

Treatment	Hillslope Position	Longitude and Latitude	Altitude/m	Slope (°)	Aspect of Slope	Grow Crops
Contour Reverse-Step (CR)	Upslope	26.090° N	2361.4	18	East by north	Corn
	Downhill	99.5116° W	2351.6	11
Primary Sloping Cropland (PSC)	Upslope	26.091° N	2372.6	18
	Downhill	99.518° W	2361.2	11

. Before conducting the dye-tracing experiment, surface litter and gravel within each quadrat were carefully removed. A 50 cm × 50 cm iron frame with a thickness of 0.5 cm was inserted 30 cm deep into the soil to delineate the test area. The experiment was conducted under dry conditions, with no precipitation occurring on the day before or during the experiment. Referring to the report of Yunlong County Meteorological Station, the heavy rainfall process occurred in the county in July 2024, with rainfall amount reaching 50–60 mm. Considering that, this study needs to select the rainfall amount that can reflect the actual rainfall intensity in the region and avoid the disturbance of surface runoff that may be triggered by extreme heavy rainfall (e.g., 50–60 mm). Therefore, by combining the regional rainfall characteristics and the observation needs of the preferential flow, a 24 h rainfall of 40 mm is finally selected as the rainfall standard for the preferential flow simulation—this value is within the range of the common rainfall intensity in the region, and it can also be used to effectively observe the preferential paths and transport patterns. A 4 g·L⁻¹ solution of Brilliant Blue FCF was prepared as the tracer dye. The dye solution was applied evenly at a constant rate using a custom-built, stable, constant-flow rainfall simulator. Furthermore, the volume of dye solution used and the application rate are consistent. After dye application, the area was covered with waterproof plastic sheeting for 24 h to maintain shade and prevent evaporation or external moisture interference. After 24 h of staining, the plastic sheeting and iron frame were gently removed. Vertical soil profiles were excavated at 10 cm intervals within each quadrat, with the maximum excavation depth determined by the extent of visible dye penetration, following established protocols [22]. Each profile was photographed using a 5-megapixel digital camera, capturing three to five images per profile. Two representative images from each quadrat were selected for further analysis and comparison. Soil samples were collected from each profile layer to assess physicochemical properties. The mean weight diameter (MWD) of soil aggregates was determined using both dry and wet sieving methods. Soil texture was analyzed using a laser particle size analyzer (Malvern MS 2000, Malvern Instruments UK Ltd., Malvern City, UK). Bulk density and porosity were measured using the ring knife method. The basic physical and chemical properties of the 0–20 cm soil layer are presented in

Table 2. Basic physical and chemical properties of soil.

Area	Soil Depth (cm)	Total Porosity (%)	Capillary Porosity (%)	Bulk Density (g·cm−3)	Clay (%)	Silt (%)	Sand (%)	MDW (mm)	Organic Matter (%)	Total Phosphorus (g·kg−1)	Total Nitrogen (g·kg−1)	Total Potassium (g·kg−1)
CR	0–10	63.42 ± 2.64 a	50.27 ± 3.91 a	1.34 ± 0.05 a	25.73 ± 5.46 b	55.00 ± 5.40 a	19.27 ± 0.64 a	2.85 ± 0.06 a	14.17 ± 0.53 a	0.30 ± 0.01 a	0.49 ± 0.01 a	9.91 ± 0.27 a
	10–20	60.50 ± 1.64 a	46.44 ± 2.58 a	1.37 ± 0.09 a	33.87 ± 4.47 a	52.45 ± 3.31 a	13.69 ± 3.37 a	2.76 ± 0.03 a	11.49 ± 0.74 a	0.25 ± 0.07 a	0.43 ± 0.01 a	10.00 ± 0.16 a
PSC	0–10	59.08 ± 3.22 a	47.57 ± 4.99 a	1.33 ± 0.09 a	42.27 ± 3.64 a	41.46 ± 4.26 b	16.28 ± 0.71 b	2.73 ± 0.11 a	6.89 ± 0.37 b	0.20 ± 0.02 b	0.26 ± 0.07 b	2.05 ± 0.81 b
	10–20	53.77 ± 6.72 a	36.97 ± 5.97 a	1.30 ± 0.07 a	41.05 ± 9.69 a	49.01 ± 6.95 a	9.93 ± 4.09 a	2.69 ± 0.05 a	8.66 ± 0.32 b	0.17 ± 0.01 a	0.23 ± 0.03 b	3.96 ± 3.52 b

Note: The presence of different lowercase letters in the table indicates that there is a significant difference between the presence or absence of the measure under the same soil layer (p < 0.05). MDW: Average weight diameter of soil aggregates.

.

2.2.3. Determination of Water Infiltration Curve

During the excavation of each soil profile, undisturbed soil samples were collected using cylindrical soil cores with a height of 5.2 cm and a diameter of 7 cm. The constant head method was employed to measure soil permeability. For each plot, three replicate samples were taken from both the 0–10 cm and 10–20 cm soil layers for infiltration experiments. The soil cores were first saturated by soaking in clean water for 12 h. Following saturation, the samples were placed on a coarse sand layer for another 12 h to reach field capacity. Once prepared, the soil cores were installed in a soil water infiltration apparatus. A Mariotte bottle was used to maintain a constant water head of 4.5 cm. Outflow was collected in a beaker positioned on an electronic balance with an accuracy of 0.01 g. The outflow volume was recorded immediately after water began to exit the base of the soil column and subsequently at 10 s intervals until the flow rate reached a stable value. After the experiment, infiltration rates were calculated, and water infiltration curves were generated for each sample plot.

2.2.4. Calculation of Soil Macropore Quantity

In this study, soil macropores are defined as those pores corresponding to the soil moisture content between field capacity and saturation. At field capacity, the matric potential approaches zero, and the infiltration rate is primarily governed by the intensity of external water input. Experimental observations indicated that water movement under these conditions followed the characteristics of laminar flow. By combining water breakthrough curve data with fluid dynamics theory, a mathematical relationship between pore radius and flow rate was established based on Poiseuille’s law.

Flow rate equation under laminar flow conditions are as follows:

$$Q={\displaystyle \frac{\left(\pi r^4∆P\right)}{\left(8\eta \tau L\right)}}$$

(1)

Under steady-state flow conditions, the flow equation transforms into the following:

$$Q={\displaystyle \frac{\pi r^{2}L}{t}}$$

(2)

By simultaneously solving Equations (1) and (2), the formula for calculating the equivalent radius of macropores is derived:

$$r=\tau L{\left({\displaystyle \frac{8\eta }{\left(t∆P\right)}}\right)}^{{\displaystyle \frac{1}{2}}}$$

(3)

where, in the formula, $Q$ represents the unit discharge (cm³·s⁻¹); $r$ denotes the pore radius (cm); $\tau$ is the tortuosity coefficient of the actual flow path, typically set as 1.2; $L$ is the height of the soil column (cm); $\eta$ is the viscosity coefficient of water (g·cm⁻¹·s⁻¹); $∆P$ is the pressure head (cm); and $t$ is the time (s) from the initial start of water addition.

During experimental measurements, the initial seepage time and the stable seepage time are recorded, corresponding to the measurement nodes of the maximum and minimum critical radii, respectively. Based on seepage flow observations across different time intervals, pore classification is conducted at 0.05 mm radius intervals, with the median of each interval serving as the representative radius.

For a specific pore size fraction, the drainage volume is calculated as follows:

$$Q_e=nAv={\displaystyle \frac{n\pi r^2\tau L}{t}}$$

(4)

where, in the formula, v represents the flow velocity (cm·s⁻¹). $A$ is the pore area (cm²).

The formula for calculating the corresponding number of macropores is as follows (n):

$$n={\displaystyle \frac{Q_{e}t}{\left(\pi r^2\tau L\right)}}$$

(5)

Based on the number of macropores, the macroporosity is obtained using the weighted average method:

$$P_r=\sum _{i=1}^n{\displaystyle \frac{n_i\times r_i^2}{R^2}}$$

(6)

where, in the formula, $P_r$ is the macroporosity (%); $r_i$ is the average value of the pore radius range (cm); $n_i$ is the number of macropores within the radius range; $R$ is the radius of the specialized soil core sampler (cm).

2.2.5. Statistical Analysis

Photoshop was used to correct and crop the soil profile images obtained from each sampling site. Image contrast between stained and unstained regions was enhanced by adjusting the grayscale distribution. Stained areas were converted to pure black (grayscale value = 0), while unstained areas were rendered pure white (grayscale value = 255), producing binary images that clearly defined the boundaries of dye infiltration in the vertical profiles. The binarized images were imported into Image-Pro Plus (Version 6.0.0.260) for quantitative analysis. Using the region selection tool and area calculation matrix, the pixel area ratio of stained versus unstained regions was determined. Independent samples t-tests were used to analyze the physicochemical properties of the different soil depths under the different treatment conditions using SPSS software (Version 27), and a one-way analysis of variance (ANOVA) was performed on the stained area ratios of the different slopes under the different treatment conditions as a means of determining which differences between the treatment groups were statistically significant. Results were visualized using Origin 2024 (Version 2024 SR1) and the ChiPlot online platform (https://www.chiplot.online (accessed on 7 July 2025)).

3. Results and Analysis

3.1. Characteristics of Preferential Flow in Different Soil Profiles

Figure 2 displays vertical staining profile images from four sample plots, with two representative images selected for each plot. Under steep slope conditions, topographic variation on sloping farmland led to pronounced spatial differentiation in dye distribution. Surface runoff promoted the lateral migration of the staining solution along the slope, resulting in the highest staining intensity in the outer profiles. In contrast, lateral water movement reduced infiltration efficiency in the inner zones, leading to a decline in surface staining coverage. The soil in the study area has outstanding degradation problems and is characterized by large slopes and significantly higher gravel content than other agricultural soils, which makes it difficult for the staining solution to penetrate, and its staining depth is limited to 20 cm. Variations in soil heterogeneity and pore connectivity produced differences in vertical staining patterns within the same treatment group. Preferential flow pathways exhibited irregular morphologies, often appearing as branch-like or funnel-shaped structures [22]. Although maximum staining depth varied across plots, a general decline in dye penetration was observed with increasing depth beyond the 0–10 cm layer. The stained area in the surface 0–10 cm soil layer accounted for 62.3–78.5% of the profile, accompanied by a reduction in macropore frequency. The deepest stained zones were predominantly concentrated in the 10–20 cm layer. Extensive surface staining and frequent occurrences of finger flow indicated a non-uniform infiltration process. Localized increases in staining near fine roots and gravel confirmed the role of biological channels in enhancing preferential flow. Observations during the experiment showed that, beginning at a soil depth of 20 cm, a marked increase in gravel content reduced soil permeability. The corresponding decrease in vertical flow paths further restricted downward water movement.

The CR-S group exhibited a clear downward trend in preferential flow, with staining intensity peaking at a depth of 10–15 cm, indicating strong potential for deep infiltration. The presence of stained zones at this depth suggested the influence of factors such as root systems or embedded stones that promote vertical water movement. The CR-D group exhibited pronounced finger flow patterns, indicating the formation of well-defined preferential channels and potential soil fissures, with flow path widths ranging from 5 to 15 cm. In contrast, the PSC-U and PSC-D groups displayed sparse and discontinuous preferential flow features. Flow paths were narrower, generally within the 5–10 cm range, and appeared as isolated patches, indicating weaker connectivity within the soil matrix and a more compacted structure less conducive to deep infiltration. This result suggests that the contour antislope step measure promotes water infiltration by improving the soil structure and pore network and reflects the phenomenon that the preferential flow area is reduced by increasing soil density with increasing soil depth.

3.2. Soil Vertical Profile Staining Area Ratio

3.2.1. The Variation Pattern of Soil Staining Area

Figure 3 in conjunction with

Table 3. ANOVA analysis between slopes under different measures.

Slope Position	Degrees of Freedom	Mean Square	F	p
Uphill	3	1747.18	3.657	0.029
Downhill	3	949.51	2.088	0.119

demonstrates the distributional characteristics and variability of the stained area ratios of each soil layer with and without land preparation patterns. Specifically, under the CR preparation condition, variations in stained area ratios were observed among plots at the same soil depth. As depth increased, the stained area ratios consistently declined across all profiles, with notable fluctuations in magnitude, indicating a higher degree of preferential flow differentiation and distinct distribution. Improved connectivity of macropores in the CR plots facilitated enhanced vertical water movement. The maximum stained area ratio ranged from 73.52% to 77.84% across profiles. The most pronounced variation occurred at a depth of 6–8 cm, where the difference in stained area among plots was the most pronounced. In both upslope and downslope CR plots, a slight rebound in stained area ratio was observed, likely due to localized disturbances caused by fine roots, gravel, or soil animal activity. The preferential flow staining pattern exhibited an “S”-shaped fluctuation within narrow vertical zones, consistent with the known behavior of lateral seepage in soil [23]. At depths between 7 and 15 cm, stained area ratios dropped below 7.8% in most profiles, with some approaching zero. This sharp decline is attributed to reduced biological and anthropogenic disturbance in the deeper layers of steeply sloping farmland, resulting in fewer macropores and limited formation of preferential flow channels.

In the PSC plots, the average stained area ratio of each profile also decreased with increasing soil depth. The maximum stained area ratio ranged from 59.3% to 94.36%, reflecting considerable spatial heterogeneity in preferential flow and reduced water permeability in the downhill sections. Unlike the CR plots, the PSC-stained area ratio curves displayed limited fluctuation, with smaller variations within the 0–10 cm soil layer. A sharp decline in stained area ratio was observed between 9 and 13 cm, with values approaching zero, suggesting higher soil compaction, reduced macropore abundance, and the presence of hydrophobic soil properties [24].

Substantial differences in the stained area ratio were evident among the sample sites. As shown in Figure 3, the comparison of stained area ratio curves under the two land preparation modes revealed the following trends: the maximum stained area ratio was higher in PSC than in CR, whereas the average stained area ratio and the maximum depth of preferential flow were both greater in CR than in PSC. Across all plots, stained area ratios consistently declined with depth. Preferential flow channels in CR plots showed more uniform development and stronger vertical water transmission capacity compared to PSC plots. In contrast, PSC plots exhibited stronger spatial heterogeneity. Despite differences in infiltration behavior, both treatments showed a tendency for the stained areas to approach zero in the deeper soil layers.

3.2.2. The Variation Pattern of Average Dyeing Area Ratio

The average stained area ratio of soil preferential flow is a key indicator for characterizing the degree of development of preferential flow channels and the concentration of water transport [22]. Figure 4 illustrates the average (av) and standard deviation (std) of the stained area ratio for each soil profile. The black shaded region represents the range defined by the mean ± standard deviation, with its width indicating the extent of spatial variability in preferential flow. Due to soil spatial heterogeneity and water repellency, the degree of variability differed across sampling sites.

The CR-S plot displayed a relatively uniform slope in the stained area ratio curve within the 0–5 cm soil layer, corresponding to the maximum average stained area ratio across all treatments. In the 5–13 cm depth range, the curve exhibited larger fluctuations, indicating the presence of finger flow and uneven lateral water movement, which disrupted the continuity of preferential pathways [25]. In the CR-D plot, significant fluctuations were observed between 3 and 10 cm, suggesting increased development of preferential flow at this depth. However, variations in pore connectivity led to an uneven spatial distribution of macropores, resulting in a higher degree of variability in the stained area ratio. In contrast, the PSC plots showed smaller overall fluctuations in average stained area ratio. The PSC-U treatment demonstrated relatively consistent variation, while PSC-D exhibited the least variability within the 5–13 cm depth range. Notably, the soil depth at which the stained area ratio approached zero was shallower in the PSC plots compared to CR, indicating reduced preferential flow penetration under untreated conditions. Analysis of the standard deviation curves revealed a general decline in the stained area ratio with increasing soil depth across all treatments. However, in the CR plots, artificially induced changes in pore structure occasionally caused localized rebounds in the coefficient of variation. Despite these fluctuations, CR treatments enhanced the overall uniformity and continuity of preferential flow development. This result indicates that land preparation measures significantly promoted the development of preferential flow, and the promotion effect was better on the upper slopes than on the lower slopes. This difference may be related to the slope: the higher slope of the upper slopes makes the soil relatively more permeable, thus providing more favorable conditions for the development of preferential flow.

3.3. The Relationship Between Soil Structure, Macropores, and Preferential Flow with and Without Measures

3.3.1. Soil Moisture Infiltration Curve

As shown by the data curve in Figure 5, soil profiles at different depths exhibited distinct infiltration characteristics during the water penetration experiments. In the 0–10 cm surface soil layer, the stable outflow rates for the sample plots were 0.244, 0.069, 0.017, and 0.021 cm³·s⁻¹, respectively, indicating a strong influence of soil structure on infiltration behavior. In the 10–20 cm layer, the corresponding rates were 0.063, 0.195, 0.058, and 0.065 cm³·s⁻¹, suggesting that changes in texture and macropore distribution with depth affected water movement. In the 0–10 cm layer, CR plots generally exhibited higher infiltration rates than PSC plots, with the maximum rate reaching 0.367 cm³·s⁻¹. In the 10–20 cm layer, CR-D recorded the highest outflow rate among all treatments, peaking at 0.285 cm³·s⁻¹. The outflow rates in PSC-U and PSC-D varied minimally. The stable flow rate of CR increased from 0.017 to 0.244 cm³·s⁻¹ compared to PSC, especially in the 0–10 cm layer, where it increased by an average of approximately 0.23 cm³·s⁻¹ compared to PSC. The average outflow rate in PSC ranged from 0.105 to 0.135 cm³·s⁻¹, consistently lower than that in CR. The results in the figure show that the outflow rate of CR is higher than that of PSC at different slopes, indicating that CR has a higher content of macropores compared to PSC. This difference in results further confirms that the land preparation measures improved the soil structure, thus expanding the distribution range of macropores and effectively enhancing the water infiltration capacity and transport efficiency.

3.3.2. Characteristics of Soil Macropore Quantity

As shown in Figure 6, the radius of soil macropores under both land preparation conditions ranged from 0.15 to 1.3 mm and was categorized into four classes: 0.15–0.45 mm, 0.45–0.75 mm, 0.75–0.95 mm, and 0.95–1.3 mm. The average macropore radius ranged from 0.29 to 0.4 mm. Across all four sample plots, the number of macropores declined with increasing soil depth. Variance analysis revealed that the number of macropores in CR plots exceeded that in PSC plots. Among all treatments, the CR-S plot recorded the highest macropore porosity at 18.45%, while the PSC-U plot showed the lowest value, with a significant difference of 1.46%. Within individual soil profiles, the number of macropores also varied, reflecting spatial heterogeneity in pore distribution. Overall, macropore abundance increased as pore radius decreased. The majority of macropores were concentrated within the 0.15–0.45 mm range, whereas macropores exceeding 0.95 mm were scarce. These results suggest that a higher number of smaller macropores contributes significantly to the formation of preferential flow paths and enhances soil water permeability [26].

The average macropore radius in the CR plots ranged from 0.32 to 0.4 mm, with a mean pore size of 0.34 mm and a macropore ratio between 4.96% and 18.45%. The PSC plots exhibited a narrower radius range of 0.29 to 0.32 mm, with a mean pore size of 0.31 mm and a macropore ratio ranging from 1.46% to 6.32%. The maximum average radius range of CR increased from 0.29 mm to 0.4 mm compared to PSC, while the macroporosity increased from 1.46% to 18.45%. At varying soil depths, both the number of macropores and the average pore size in CR consistently exceeded those in PSC, indicating significant soil spatial heterogeneity. CR significantly enhanced macroporosity characteristics, improved pore structure, enhanced soil spatial heterogeneity compared to PSC, and generalized the phenomenon of decreasing macroporosity with increasing soil depth, suggesting that changes in soil structure affect the number of soil pores.

In this paper, the interactions between the number of macropores and the ratio of stained area of each sample site in different pore radius ranges are extensively explored by means of pass-through analysis. And the indirect pathway coefficient is used as a numerical indicator to quantify the strength of indirect effects, measuring the strength and direction of indirect effects between variables. As shown in

Table 4. Relationship between the number of macropores and the ratio of stained area in different pore size ranges in each sample site. Y: Dyeing area ratio. R₁–R₄: The number of macropores within the radius ranges of 0.95–1.3 mm, 0.75–0.95 mm, 0.45–0.75 mm, and 0.15–0.45 mm.

Area	Independent Variable	Simple Correlation Coefficient with Y	Path Coefficient (Direct Action)	Indirect Path Coefficient (Indirect Action)
				R1	R2	R3	R4
CR-S	$R_1$	0.613	1.851	-	1.334	−1.553	−1.666
	$R_2$	0.108	−2.444	−1.762	-	−0.809	−1.249
	$R_3$	0.871	9.669	−8.112	−3.200	-	9.456
	$R_4$	0.802	−8.237	−7.413	−4.209	8.055	-
CR-D	$R_1$	-	-	-	-	-	-
	$R_2$	−0.572	−3.001	-	-	2.608	2.707
	$R_3$	−0.221	−7.487	-	6.506	-	7.465
	$R_4$	−0.269	9.904	-	8.933	9.874	-
PSC-U	$R_1$	-	-	-	-	-	-
	$R_2$	−0.818	−0.725	-	-	0.624	0.239
	$R_3$	−0.752	−0.096	-	0.082	-	0.010
	$R_4$	−0.356	−0.322	-	0.011	0.033	-
PSC-D	$R_1$	-	-	-	-	-	-
	$R_2$	−0.702	2.546	-	-	2.538	2.380
	$R_3$	−0.829	−4.209	-	4.197	-	4.083
	$R_4$	−0.779	0.923	-	0.863	0.895	-

, the number of macropores in the range of R₁–R₄ pore radii showed a positive correlation with the stained area ratio, and the simple correlation coefficient of R₃ in the CR-S sample site was higher than that in the other sample sites. The maximum value of the simple correlation coefficient in PSC was −0.829, the flux coefficient ranged from 0.096 to 4.209, and the indirect effects were all positive; CR flux coefficients were the largest for R₃ in the CR-S sample site and R₄ in CR-S sample site, which ranged from 1.851 to 9.904, the maximum simple correlation coefficient was up to 0.871, and the indirect correlation coefficients were all 1.249 and above. It indicates that the correlation between the number of macropores and the ratio of stained area was improved by the contour antislope step preparation measure, and the number of macropores under the diameter grades of 0.45–0.75 mm and 0.15–0.45 mm had the greatest effect on the occurrence of the preferential flow phenomenon. By distinguishing the functional differences of R₁–R₄ pore radii, we can precisely reveal the direct/indirect influence of macropores on preferential flow and provide a scientific basis for understanding the soil water transport law and optimizing soil management measures.

3.3.3. The Relationship Between Preferential Flow and Macropores

As shown in Figure 7, all four sample plots demonstrated a positive correlation between the number of macropores and the stained area ratio of preferential flow, highlighting the regulatory role of soil structure in controlling water movement. The correlation coefficients under contour reverse-slope terrace conditions were consistently higher than those observed in the original sloping farmland. The maximum correlation coefficient reached 0.98 (p < 0.0001). The value close to the coefficient of complete positive correlation confirms a very strong relationship between the two. Both the CR-S and CR-D plots exhibited strong linear correlations, with coefficients ranging from 0.91 to 0.98, suggesting a high goodness of fit. These results suggest that land preparation significantly improves macropore connectivity, forming a more continuous and distributed pore network conducive to preferential flow development. In contrast, the PSC-U and PSC-D plots showed lower correlation coefficients, ranging from 0.86 to 0.97. Although still positive, the relationship was weaker compared to CR plots, indicating less efficient water transport and a limited capacity to form extensive preferential pathways [27]. These findings confirm that contour reverse-slope terrace preparation promotes a marked increase in macropore number and connectivity, thereby enhancing the likelihood and efficiency of preferential flow. As a result, water infiltration and overall soil permeability are significantly improved under the CR treatment.

3.3.4. Comparative Analysis of the Relationship Between Soil Structure, Macropores, and Preferential Flow

As shown in Figure 8, notable differences were observed in the interaction intensity of soil factors between treated (CR) and untreated (PSC) conditions. Both the number of macropores (DKX) and the stained area ratio of preferential flow (RS) exhibited strong positive correlations with organic matter (OM) (p < 0.05), suggesting that OM plays a regulatory role in macropore formation and influences preferential flow development. In the PSC plots, CP, TPS, and clay were identified as key factors contributing to soil heterogeneity (p < 0.05). CP showed positive correlations with clay, TN with sand, and TN with clay and CP (p < 0.05). Clay content demonstrated a negative correlation with macropore number, while sand displayed the opposite trend. These results indicate that increased clay content may elevate soil bulk density, restricting water movement and reducing macropore development. The relatively limited number of significant correlation pathways between DKX, RS, and other variables in the PSC plots reflected weak interrelationships among soil structural factors under natural slope conditions. In contrast, the CR plots exhibited more complex and significant correlations. Both DKX and RS showed strong positive associations with OM and TN (p < 0.05), while MWD and TSP were highly correlated with multiple physicochemical properties, underscoring their central roles in soil structure-function interactions. These findings indicate that land preparation measures modified the internal relationships among soil structural components, promoting an increase in macropore formation and enhancing the development of preferential flow pathways. CP, TN, silt, and sand showed significant positive correlations with TSP (p < 0.05), and sand, TN, TK, TP, and OM also exhibited positive correlations with MWD (p < 0.05). These results suggest that land preparation enhanced the connectivity and interaction intensity among soil variables, strengthening internal system coordination compared to the PSC condition. In the PSC site, RS and sand demonstrated a highly significant correlation (p < 0.01), likely driven by the regulatory influence of gravel content on soil pore architecture. Reduced gravel content decreases soil particle radius and limits the number of pore channels, thereby restricting preferential flow infiltration and reducing stained area extent. In comparison, CR measures improved soil structure, increased pore network complexity, and enhanced the hydraulic performance of the soil system relative to the PSC condition.

Multivariate analysis using Partial Least Squares Structural Equation Modeling (PLS-SEM) further clarified the intensity of influence exerted by various factors on soil macropore quantity and preferential flow stained area before and after the implementation of land preparation measures. Path coefficient analysis was used to quantify the direct effects among variables. As shown in Figure 9, physicochemical indicators exerted highly significant direct effects on macropore number (p < 0.001), with path coefficients β of 0.928 (CR) and −0.972 (PSC), respectively. In the CR site, total porosity, organic matter, and total nitrogen emerged as the primary drivers of soil spatial heterogeneity. Among these, total nitrogen showed the strongest direct effect on preferential flow, with a coefficient of 0.173. In the PSC site, capillary porosity, total nitrogen, and gravel content constituted the core influencing factors, with gravel having the highest direct effect (β = 0.169). As an intermediate variable, the role of macropore quantity varies depending on the presence of land preparation measures. The correlation coefficient for land preparation reached a highly significant level (p < 0.001), exceeding that of the untreated slope farmland. In the CR plots, physicochemical factors primarily influenced macropores with radii in the range of 0.45–0.75 mm (weight = 0.426), thereby indirectly promoting preferential flow development. In contrast, the PSC plots showed the highest weight (0.501) in the 0.95–1.3 mm radius range, indicating its influence on preferential flow. Weight comparisons between subfigures (b,c) and (e,f) indicates that the PSC plots were influenced by a greater number of soil variables with negative associations. In subfigure (b), the preferential flow stained area ratio shows negative correlations with clay content and bulk density (weights of −0.155 and −0.018), indicating that increases in these parameters may suppress preferential flow to some extent. In subfigure (e), organic matter, total potassium, and silt content are negatively associated with preferential flow infiltration (weights of −0.206, −0.116, and −0.184), suggesting that the combined effects of these factors constrained the development of preferential flow under natural slope conditions.

4. Discussion

4.1. The Impact of Land Leveling on Preferential Flow

This study investigated the morphological characteristics of soil preferential flow through field dye-tracing experiments, supported by systematic image analysis of vertical soil profiles across different treatment plots. Comprehensive evaluation of dye coverage ratio, macropore number, average pore radius, macropore ratio, and soil physicochemical properties demonstrated that all parameters under the contour reverse-slope terrace treatment significantly exceeded those of untreated sloping farmland. Notably, the maximum dye infiltration depth under contour reverse-slope terraces reached 21 cm, compared to 14 cm for untreated slopes, indicating a 50% improvement with an absolute difference of 7 cm. The results of this study are consistent with the findings of Huang et al. that contour antislope grading measures promote downward runoff transfer and can effectively reduce soil erosion [28]. In a study by Li et al., the maximum infiltration depth of preferential flow in farmland under varying rainfall conditions remained within 20 cm [29]. However, other research has reported deeper preferential flow pathways in oasis farmland, with infiltration depths exceeding 40 cm under irrigation conditions [30], differing from the findings of the present study. This discrepancy can be attributed to the steep-slope terrain of the current study area, where strong surface runoff erosion, alternating wet and dry conditions, and high subsurface gravel content (especially below 20 cm) collectively restrict vertical water infiltration. As a result, preferential flow tends to develop within shallower soil layers. Experimental data revealed a consistent decline in stained area with increasing depth across all plots, eventually approaching zero, aligning with previous findings [21,22,23,24]. This pattern is primarily attributed to increased soil bulk density at greater depths, which results in a more compact structure and reduced porosity. Concurrently, declining organic matter content leads to a reduction in root density. Biological activity, including that of earthworms and other soil fauna, along with anthropogenic disturbances, also decreases with depth [31]. These combined factors significantly reduce soil permeability, preventing the staining solution from infiltrating deeper layers. In addition, lateral movement of soil moisture was frequently observed in the vertical profiles, producing dendritic staining patterns, which are consistent with the findings of Guan et al. [23] in the karst regions of Southwest China. This phenomenon is attributed to the presence of vertical pore structures and lateral pore structures in the soil, which can improve the heterogeneity of soil moisture in sloping farmland.

4.2. The Impact of Land Leveling on Soil Macropores

This study conducted a quantitative assessment by integrating dye-tracing characteristics from vertical soil profiles with macropore counts derived from water infiltration experiments. The stained area ratio, number of macropores, and macropore rate under contour reverse-slope terrace treatments were consistently higher than those observed in untreated slope farmland. Under contour reverse-slope terraces, the stable seepage rate increased significantly, rising from 0.017 to 0.244 cm³·s⁻¹ compared to the original slope farmland, which is consistent with the research conclusions of many scholars [20,31,32]. However, the infiltration patterns observed across different soil layers between the upper and lower slopes of the terraces differed from those reported by Wan et al. [20]. On the lower slope, soil permeability increased with depth, likely due to the presence of larger soil pore structures [33] and a gentler slope gradient relative to the upper slope, which together promoted more effective vertical water movement. A comparison between treated and untreated plots confirmed that contour reverse-slope terraces measure enhanced soil permeability, improved soil physicochemical properties, and effectively reduced surface runoff, consistent with the findings of Li et al. [13]. In contrast, the outflow rate observed in the original slope farmland was low and exhibited minimal variation, likely due to enhanced surface runoff on steep slopes and higher soil bulk density, both of which inhibited effective water infiltration. As a result, fewer macropores formed, and the stained area of preferential flow was smaller compared to the plots treated with contour reverse-slope terrace measures. These findings are consistent with those of Zhang et al. [34], who reported improved soil moisture response under similar land preparation interventions. The number of macropores within the pore radius ranges of 0.45~0.75 mm and 0.15~0.45 mm under the contour reverse-slope terrace plays a primary role in the occurrence of preferential flow. However, both the macropore radius and quantity observed in this study were lower than those reported in other research, such as the investigation on saturated permeability in Chinese fir forest soils conducted by Qi et al. [35]. This discrepancy can be attributed to substantial differences in site characteristics. The present study focused on sloping farmland, leading to variations in the number of pores and the macropore ratio between the two studies. The pore size distribution of macropores is a central determinant of the process of soil water infiltration, retention and loss. It also indirectly affects crop growth and yield by improving water and air permeability in the root zone. The regulation of macroporous distribution by land preparation measures and optimizes the water transport process, significantly improving the soil water retention capacity to improve crop yield.

4.3. Investigation into the Formation Mechanisms of Preferential Flow

This study revealed the driving mechanisms of soil physicochemical factors and pore structure on preferential flow through Mantel analysis and PLS-SEM model. The study found that soil physicochemical indicators showed extremely significant correlations with the number of pores (β = 0.928, −0.972) regardless of whether land preparation measures were implemented. This is consistent with the results of Dai et al.’s study, which showed that organic matter and total porosity enhance soil water holding capacity and water supply capacity by improving soil structure, but it is not consistent with the results of capacity as a key factor, which may be due to the environmental differences in the study area as well as the implementation of land preparation measures to improve the soil conditions in this study area [36]. In the CR plots, total porosity, organic matter, and total nitrogen emerged as the key factors influencing soil spatial heterogeneity, with total nitrogen exhibiting the strongest direct effect on preferential flow (β = 0.173). This result aligns with previous studies suggesting that chemical properties can improve aggregate stability and promote macropore formation [37]. In the PSC plot, capillary porosity, total nitrogen, and gravel emerged as the key factors in the coupling effect, with gravel showing the greatest effect (β = 0.169). This effect is likely attributed to the physical role of gravel in enhancing soil permeability, providing flow channels, and increasing hydrodynamic potential [38], consistent with findings reported by Zhang et al. [16] and other studies. As an intermediate variable, the influence of macropore quantity varied significantly between treatments. The correlation between macropore quantity and preferential flow under contour reverse-slope terrace conditions was highly significant (p < 0.001) and notably stronger than that observed in the original slope cropland. These results indicate that the implementation of land preparation measures promotes pore network formation by improving soil structure, thereby enhancing the facilitating role of macropores in the development of preferential flow. Further analysis of negatively correlated factors revealed that, in the CR plots, clay content and bulk density exerted inhibitory effects on the stained area of preferential flow. Increased clay content intensified soil compaction, reduced pore connectivity, increased hardness, and compressed pore volume, collectively restricting water movement. In the PSC plots, negative correlations with organic matter, total potassium, and silt suggested that land preparation measures could influence the spatial distribution of these components, thereby mitigating their obstructive impact on water flow pathways. Overall, contour reverse-slope terrace land preparation reshaped key soil structural parameters, diminishing the inhibitory effects of organic matter and silt and promoting the development of preferential flow channels. These findings demonstrate that contour reverse-slope terrace land preparation enhances the soil and water conservation capacity of sloping farmland by optimizing soil structure and improving the conditions necessary for preferential flow. The intervention strengthens the interrelationships among soil physicochemical factors, facilitates the formation of macropore networks, improves vertical infiltration, and reduces surface runoff. The study provides a theoretical basis for implementing effective erosion control strategies in steep-slope agricultural regions.

5. Conclusions

This study systematically quantified the effects of contour reverse-slope terrace land preparation on soil structure, macroporosity, and preferential flow characteristics through brilliant blue dye-tracing tests and water infiltration experiments. The core conclusions are as follows: The average dye coverage ratio in the contour reverse-slope terrace plots reached 73.3%, markedly higher than in the original slope farmland. The infiltration depth of preferential flow extended to 21 cm, showing a significant increase compared to the original slope farmland. Dye coverage ratios decreased with increasing soil depth, exhibiting slight fluctuations before approaching zero. Finger flow and block-like staining patterns were observed under both treatments, while lateral water movement was evident in the terrace-prepared plots. Under soil preparation measures, the outflow rate of the sample plots was significantly higher than that of the original slope farmland, with the maximum outflow concentrated in the 0–10 cm soil layer. In contrast, no distinct stratification was observed in the outflow of untreated plots. The macroporosity and the number of pores under soil preparation measures were significantly higher than those of the original slope farmland. The macroporosity of contour terraced land increased from 1.46% to 18.45% compared to the original slope farmland, and the maximum average radius range increased from 0.29 mm to 0.4 mm. Moreover, the number of macropores showed a significant correlation with the dye-stained area ratio (p < 0.05), with stronger correlations observed in the contour reverse-slope terrace plots than in the original slope farmland. Under land preparation measures, macropores within the radius ranges of 0.45–0.75 mm and 0.15–0.45 mm were identified as the primary contributors to the formation of preferential flow. Significant differences were observed in the correlation patterns of soil structural factors between treated and untreated plots. In the terrace-treated plots, total porosity, organic matter, and total nitrogen served as key variables, demonstrating stronger correlations with other physicochemical properties compared to those in the original sloping farmland. These findings suggest that contour reverse-slope terrace preparation alters the structural characteristics of the soil ecosystem, promotes the development of macropore networks, and significantly enhances the preferential flow capacity. This study provides quantitative evidence for optimizing soil and water conservation measures in sloping farmland, confirming the regulatory mechanism of structural land consolidation on soil moisture movement processes. It also supports the wider adoption of contour reverse-slope terraces in mountainous agriculture prone to erosion.