Converting Cropland to Forest Improves Soil Water Retention Capacity by Changing Soil Aggregate Stability and Pore-Size Distribution

Abstract

The semi-arid region of North China has undergone extensive afforestation to prevent land degradation. Although afforestation was considered an effective way to improve soil water retention, the mechanism by which it affects soil hydraulic properties remained uncertain. In this study, soil water retention curve (SWRC), soil water-stable aggregates, and other soil physicochemical properties were determined in short-term abandoned cropland (AC), shrubland (SL), and woodland (WL) that had been converted from cropland for 1, 8, and 24 years, respectively. Pearson correlation analysis and partial least-squares structural equation modeling methods were used to identify the main factors affecting soil hydraulic properties. Results showed that the SWRCs of all three land uses were well-fitted by a double-exponential model. The WL and SL land uses exhibited higher soil field capacity (0.33–0.37 cm³ cm⁻³), wilting point (0.20–0.23 cm³ cm⁻³), and available water content (0.13–0.15 cm³ cm⁻³). Surface soil exhibits a more pronounced trend in water retention capacity changes compared to subsoil under vegetation restoration. The WL and SL land uses showed more soil macroaggregates and intra-aggregate pores at surface layers, which mainly explained the variations in hydraulic properties. The main factors influencing soil hydraulic properties were soil aggregates, matrix and structural porosity, soil organic carbon (SOC), and soil bulk density (BD). Overall, afforestation can improve soil hydraulic properties and could be an effective practice for soil and water conservation in the semi-arid region of North China.

1. Introduction

Land degradation has emerged as a global environmental crisis characterized by the deterioration of land quality and productive capacity, primarily driven by anthropogenic activities such as unsustainable land management practices [1]. This phenomenon adversely impacts 3.2 billion people worldwide (40% of the global population), particularly rural communities and vulnerable groups, while threatening water security, ecosystem stability, and human well-being [2]. Unsound land management practices significantly alter soil hydraulic properties through organic matter depletion, structural deterioration, and erosion processes, ultimately leading to diminished soil fertility and compromised hydrological functions [3,4].

Soil hydraulic properties, particularly water retention capacity, constitute fundamental parameters for characterizing water and solute transport dynamics in vadose zone hydrology. These properties critically determine rainfall partitioning processes by governing hydrological pathways through three primary mechanisms: soil infiltration (enhancing soil water storage), groundwater recharge, and surface runoff (associated with erosional processes) [5,6]. This partitioning efficiency is fundamentally governed by the physical configuration of the soil matrix, with particular emphasis on its structural organization. Two distinct conceptual frameworks have emerged for soil structure characterization [7]: (1) the aggregate perspective focuses on the morphology, dimensions, and spatial arrangement of primary soil particles and aggregates; (2) the pore system perspective prioritizes the integrated architecture of pore networks, specifically analyzing pore-size distribution patterns and their connectivity. These complementary approaches provide critical insights into the relationship between soil architecture and hydraulic functionality.

Land-use changes modify soil pore-size distribution and aggregate stability by altering soil organic matter content and aggregate-associated binding agents, thereby impacting critical soil functions, including water retention, nutrient cycling, plant-available resources, infiltration capacity, and microbial community dynamics [8,9,10]. Previous studies reveal differences in pore-size distributions of contrasting long-term conversion of land use [11,12]. Land-use changes exert direct impacts on soil disturbance intensity and the quantity and quality of organic matter inputs, which in turn influence soil structural stability and hydraulic characteristics [13,14]. Various studies have shown that land use can affect soil hydraulic properties [15,16]. It is, in general, anticipated that conversion from cropland to non-cultivated land (like grassland, shrubland, or woodland) has a positive effect on soil structure [17]. However, the evolution mechanisms of soil hydraulic properties under land-use changes remain a critical knowledge gap in soil hydrology despite extensive research. This complex process involves soil structural reorganization and the impact of restoration time.

To combat soil erosion and desertification, China’s Grain for Green Program has significantly altered land-use patterns in semi-arid northern regions over recent decades [18,19]. This study investigates how these land-use changes affect soil hydraulic properties, specifically testing two hypotheses: (1) vegetation restoration enhances soil water retention capacity depending on restoration time, and (2) this enhancement correlates with aggregate size distribution and pore-size distribution. Through comparative analysis of various land-use types, we characterized soil hydraulic responses to ecological restoration and identified key influencing factors.

2. Materials and Methods

2.1. Study Region

The study area is a typical hilly region in North China, situated in Chongli District, Zhangjiakou City, Hebei Province of China (40°47′–41°17′ N, 114°17′–115°34′ E) (Figure 1). Characterized by a continental monsoon climate, the region exhibits a mean annual temperature of 3.70 °C and receives 483 mm of precipitation annually. As a key venue for the 2022 Winter Olympics, extensive revegetation efforts were implemented to mitigate soil erosion, with shrublands and woodlands prioritized as primary restoration types.

2.2. Field Experimental Design

Due to the gradual implementation of the Grain for Green Program in recent years and the reduction of human activities in mountainous areas, cropland was gradually abandoned and transformed into natural land (grassland, shrubland, or forest land). Experiments were conducted in a typical village experiencing land abandonment, which features a chronological sequence of abandoned croplands. Three land-use types, i.e., short-term abandoned cropland (AC), shrubland (SL), and woodland (WL), were selected in the study area (Figure 2). The selected experimental sites were all located on the same hillside within the village, with consistent soil types and pre-abandonment cultivation practices. The slope positions of AC, SL, and WL on the micro terrain are the lower slope, middle slope, and upper slope, respectively.

According to the villagers, AC, SL, and WL have been converted from cropland for 1, 8, and 24 years, respectively. Broccoli (Brassica oleracea L. var. italica Plenck), cauliflower (Brassica oleracea L. var. botrytis L.), and Chinese cabbage (Brassica bara L.) were mainly planted before conversion. After stopping cultivation, the main vegetation naturally evolved into Potentilla supina L., Tribulus terrestris L., and Medicago Sativa L. (AC), Caragana and Hippophae rhamnoides L. (SL), and Betula platyphylla Suk. (WL) (

Table 1. Characteristics of the four land-use types.

Land Use	Geographical Coordinates	Elevation m	Slope	Plant Species
Abandoned cropland	115°15′11″ E, 40°53′19″ N	1320	5	Potentilla supina L., Tribulus terrestris L., Medicago Sativa Linn
Shrubland	115°15′46″ E, 40°53′7″ N	1337	10	Caragana, Hippophae rhamnoides Linn
Woodland	115°15′40″ E, 40°52′50″ N	1382	8	Betula platyphylla Suk

).

The study area featured three distinct land-use types: short-term abandoned cropland (AC), shrubland (SL), and woodland (WL) (Figure 2). All soil samples were collected throughout 2019. According to residents, these areas were converted from cropland 1, 8, and 24 years before the study, i.e., in 2018, 2011, and 1995, respectively. Prior to abandonment, the croplands primarily cultivated broccoli (Brassica oleracea L. var. italica Plenck), cauliflower (Brassica oleracea L. var. botrytis L.), and Chinese cabbage (Brassica rapa L.). Post-abandonment, vegetation naturally succeeded Potentilla supina L., Tribulus terrestris L., and Medicago sativa L. in AC; Caragana spp., as well as Hippophae rhamnoides L. in SL and Betula platyphylla Suk. in WL (Table 1). The soils were classified as Kastanozem in the World Resource Base [20].

2.3. Soil Sampling

Three soil profiles were randomly selected as replications for each type of land use at each site. Undisturbed soil core samples were collected from soil layers (0–10 cm, 10–20 cm, 20–30 cm, and 30–40 cm) using 100 cm³ stainless-steel cutting rings. Bulk soil samples were collected using a shovel beside the cutting rings and carefully put in the tin box to avoid compaction before the measurement of soil aggregates. Finally, the disturbed soil samples were gathered as well. All the methods applied in this study were listed in the flowchart in Figure 3.

2.4. Saturated Hydraulic Conductivity Measurement

The hydraulic conductivities of land uses were measured by a Guelph Permeameter (Soilmoisture Equipment Corp., Goleta, CA, USA) by determining the cumulative water flux under surface ponding conditions. The depth and radius of the drilled well were 5 cm and 3 cm, respectively, and the depth of the well ponding was 2 cm. Every permeameter test lasted at least 30 min, and a total of four rounds of measurement were taken at each location. The soil saturated hydraulic conductivity (Ks) was calculated following the method in reference [21], which aligns with the steady-state water infiltration rate of soils.

2.5. Measurement of Soil BD, SWRC, Water-Stable Aggregate, and Other Properties

Soil core samples were first subjected to soil water retention curve (SWRC) determination using a CR2G high-speed refrigerated centrifuge (Hitachi Ltd., Tokyo, Japan) following the centrifugal method [22,23]. Subsequently, undisturbed soil bulk density (BD) was measured through the oven-drying method at 105 °C until constant weight. The size separation of soil aggregates was performed in the Yoder apparatus using the method proposed by reference [24]. Briefly, a 100 g air-dried bulk soil sample was kept on the top of a 2 mm sieve before moving the sieve up and down at a depth of 3 cm in deionized water with a constant frequency (25 times per minute). Three sieves were used to get four size fractions consisting of large macroaggregates (>2 mm), small macroaggregates (0.25–2 mm), microaggregates (0.053–0.25 mm), and silt + clay-associated particles (<0.053 mm) [24]. All aggregate fractions were oven-dried at 105 °C to constant weight.

Air-dried and homogenized soil samples (<2 mm sieve fraction) underwent sequential analytical procedures: Soil pH was quantified via PHS-3C pH meter (INESA, Jiading, Shanghai, China) in 1:2.5 (w/v) soil-to-water suspensions with 30 min equilibration. Particle-size distribution analysis employed the sedimentation-pipette method [25]. Soil organic carbon (SOC) and total nitrogen (TN) were measured using a carbon-nitrogen analyzer (vario TOC cube, Elementar, Langenselbold, Hessen, Germany) on samples pretreated with 1 M HCl until effervescence cessation to remove carbonates.

2.6. Data Analysis

2.6.1. Mean Weight Diameter (MWD), Geometric Mean Diameter (GMD), and Water-Stable Aggregates (WSA)

To assess the impact of different land-use types on soil structure, we calculated three indices expressing soil aggregate stability: MWD, GMD, and WSA. WSA is the weight proportion of aggregates larger than 0.25 mm. MWD and GMD were calculated as [26]:

$$\mathrm{MWD}={\displaystyle {\sum }_{i=1}^j{w}_i\times x_i}$$

(1)

$$\mathrm{GWD}=\exp \left[{\displaystyle \frac{{\displaystyle {\sum }_{i=1}^j{w}_i\times \ln x_i}}{{\displaystyle {\sum }_{i=1}^n{w}_i}}}\right]$$

(2)

where i = 1 to j and j is the number of aggregate ranges, x_i is the mean diameter of each size fraction (mm), and w_i is the proportion (%) of each size class to the total sample.

2.6.2. Soil Hydraulic Properties

The water content and suction data were analyzed using two prominent hydraulic models in soil science: the van Genuchten model [27] (Equation (3)) and the double-exponential model [10] (Equation (4)), both of which have been extensively utilized in soil hydraulic characterization research. The parameter optimization for these models was performed through nonlinear regression analyses employing the built-in “Solver” optimization tool in Microsoft Excel 365, which utilizes a generalized reduced gradient algorithm to minimize residual sum of squares between measured and predicted values.

$$\theta ={\theta }_r+\left({\theta }_s-{\theta }_r\right){\left[{\displaystyle \frac{1}{1+{\left(\alpha \left|h\right|\right)}^n}}\right]}^{1-1/n}$$

(3)

$$\theta =C+A_1\times {\mathrm{e}}^{(-h/h_1)}+A_2\times {\mathrm{e}}^{(-h/h_2)}$$

(4)

where θ is the water content (cm³ cm⁻³); θ_s and θ_r represent the saturated and residual water contents (cm³ cm⁻³), respectively; h is the matric suction (cm); α and n are empirical shape parameters where α corresponds to the inverse of the air entry value (cm⁻¹) and n controls the steepness of the transition zone in the water retention curve; and in the double-exponential model, C denotes the residual water content (cm³ cm⁻³), while A₁ (cm³ cm⁻³) and A₂ (cm³ cm⁻³) quantify the volumetric proportions of structural and matrix porosity, respectively. Correspondingly, h₁ (cm) and h₂ (cm) define the critical matric suction thresholds at which drainage of structural and matrix porosity predominantly occurs. The pore-size distribution predicted by the SWRC model was visualized by differentiating Equation (3) or Equation (4) concerning the logarithm of matric suction.

The water content at suction of 330 cm and 15,000 cm in the fitted SWRC are defined as FC and WP, respectively [28]; that is, FC = θ_330cm and WP = θ_15000cm. AWC is calculated as the difference between FC and WP.

2.6.3. Pore-Size Distribution

The pore-size distribution can be expressed as

$$\mathrm{f}(h)={\displaystyle \frac{\mathrm{d}\theta }{\mathrm{d}\left({\log }_{10}h\right)}}$$

(5)

where θ and h are the variates of the SWRC model (Equation (3) or Equation (4)). Pore-water suction was assumed to relate to an average pore size by the approximate relation:

$$d={\displaystyle \frac{3000}{h}}$$

(6)

where d is the tube-equivalent pore diameter (μm). The equation derives from the physics-based capillary rise equation of Young–Laplace.

2.6.4. Statistical Analysis

One-way analysis of variance (ANOVA) and least significant difference multiple comparisons (p < 0.05) were used to assess the significant effects of land-use types on the soil properties. The means and standard deviations were determined by three separate samples. The ANOVA analyses were conducted with Excel 365, SPSS 25.0. The Pearson correlation analysis was conducted in Origin 9.9.

3. Results

3.1. Soil Basic Properties

Notably, there were no significant effects of land-use type on soil BD (p > 0.05,

Table 2. Soil basic physicochemical characteristics of three land-use types.

Soil Depth	Land Use	SOC	pH	TN	Clay	Silt	Sand	BD
cm		g kg−1	/	g kg−1	%	%	%	g cm−3
0–10	AC	29.11 ± 8.67 A	8.15 ± 0.01 A	2.43 ± 0.03 A	22.43 ± 1.43 A	51.45 ± 10.02 A	26.12 ± 8.77 A	1.26 ± 0.14 A
	SL	32.85 ± 7.67 A	7.89 ± 0.21 A	1.48 ± 0.39 B	19.43 ± 1.14 A	49.62 ± 6.85 A	30.95 ± 6.58 A	1.26 ± 0.13 A
	WL	42.05 ± 0.24 A	6.75 ± 0.18 B	1.56 ± 0.35 B	27.34 ± 1.87 A	49.33 ± 3.41 A	23.34 ± 2.46 A	1.30 ± 0.16 A
10–20	AC	21.04 ± 17.51 A	8.17 ± 0.06 A	2.12 ± 0.67 A	21.79 ± 3.75 A	49.97 ± 7.69 A	28.23 ± 10.00 A	1.34 ± 0.13 A
	SL	34.93 ± 7.11 A	7.93 ± 0.28 A	1.1 ± 0.82 B	23.61 ± 1.90 A	52.08 ± 3.07 A	24.31 ± 1.17 A	1.17 ± 0.10 B
	WL	38 ± 12.98 A	7.30 ± 0.59 B	1.83 ± 0.43 AB	26.16 ± 4.85 A	50.45 ± 6.39 A	23.39 ± 5.70 A	1.23 ± 0.07 AB
20–30	AC	35.02 ± 3.37 A	8.25 ± 0.04 A	1.99 ± 0.14 A	23.53 ± 5.03 A	51.08 ± 6.16 A	25.39 ± 6.79 A	1.44 ± 0.08 A
	SL	14.93 ± 9.19 B	8.04 ± 0.08 A	0.79 ± 0.4 C	24.65 ± 2.37 A	48.40 ± 7.74 A	26.94 ± 10.1 A	1.29 ± 0.07 AB
	WL	30.9 ± 1.51 AB	7.44 ± 0.80 B	1.58 ± 0.07 B	20.46 ± 2.89 A	49.67 ± 7.47 A	29.87 ± 7.17 A	1.21 ± 0.09 B
30–40	AC	20.38 ± 14.35 B	8.21 ± 0.07 A	2.14 ± 0.15 A	25.82 ± 2.31 A	48.35 ± 8.73 A	25.83 ± 8.37 A	1.29 ± 0.02 A
	SL	37.78 ± 17.43 A	7.98 ± 0.04 A	1.12 ± 0.66 B	22.82 ± 5.09 A	50.78 ± 8.85 A	26.41 ± 13.82 A	1.27 ± 0.07 A
	WL	37.37 ± 1.1 AB	7.84 ± 0.40 A	1.87 ± 0.79 AB	19.37 ± 6.29 A	53.62 ± 6.21 A	27.01 ± 9.04 A	1.29 ± 0.04 A

Note: all the data were the means of triplicate with standard deviation; different letters reveal significant differences among different land uses (p < 0.05).

), but shrubland and woodland showed lower soil BD at 20–30 cm depth, compared with that in abandoned cropland (p < 0.05, Table 2). The result could be explained by soil degradation and decreased macropores in the soil of AC due to tillage. The soil texture-class consistently exhibited sandy clay loam dominance across all land-use types, with particle-size fractions (sand: 23.34–30.95%, silt: 49.33–51.45%, clay: 19.43–27.34%) showing no significant inter-group variability (p > 0.05), which indicates that the land-use types had intrinsically similar soil texture. References [29,30] obtained similar findings.

WL demonstrated the maximum SOC (30.90–42.05 g kg⁻¹), higher than the minimum value observed in AC (20.38–29.11 g kg⁻¹). The carbon sequestration in the soil depends on the balance between the accumulation of biomass and soil litter and on the loss of carbon through respiration and decomposition of the litter and carbon in the soil [31,32]. The higher content of SOC in WL may be related to higher biomass and litter amount in woodland. The soil of AC exhibited the highest TN content (2.43 g/kg), a phenomenon primarily attributed to sustained inorganic fertilizer application. Conversely, prolonged fallow duration led to marked declines in total nitrogen levels across both shrubland and woodland ecosystems due to the gradual depletion of fertilization effects.

Soil pH showed significant differences between land-use types (p < 0.05, Table 2) and was lower in WL compared with AC and SL at four soil depths. This reduction in soil pH was explained by the organic matter input into the soil. Similar findings have been obtained in prior studies [29,30].

3.2. Water-Stable Aggregates

In this study, the silt + clay-associated particles (<0.053 mm) and small macroaggregates (2–0.25 mm) dominated the soil mass, with proportions ranging from 37.43–47.87% and 26.26–39.03%. The large macroaggregates (>2 mm) and microaggregates (0.25–0.053 mm) accounted for a very small fraction of the soil mass, with a range of 10.71–14.01% and 9.70–14.90%, respectively (

Table 3. Size distribution, mean weight diameter (MWD), geometric mean diameter (GMD), and water-stable aggregates (WSA) of soil aggregates in three land-use types.

Soil Depth	Land Use	Large Macroaggregates (2 mm)	Macroaggregates (2–0.25 mm)	Microaggregates (0.25–0.053 mm)	Associated Particles (0.053 mm)	MWD	GMD	WSA
cm		----------------------------------------%---------------------------------------				mm	mm	%
0–10	AC	10.97 ± 6.44 A	26.26 ± 2.09 B	14.90 ± 5.41 A	47.87 ± 2.08 A	0.99 ± 0.38 A	0.17 ± 0.05 B	37.23 ± 7.22 B
	SL	14.01 ± 6.00 A	31.2 ± 5.74 AB	9.70 ± 1.93 B	45.09 ± 2.36 AB	1.22 ± 0.30 A	0.22 ± 0.03 AB	45.21 ± 1.00 AB
	WL	10.71 ± 3.74 A	39.03 ± 1.55 A	12.83 ± 1.07 AB	37.43 ± 4.44 B	1.11 ± 0.24 A	0.26 ± 0.07 A	49.74 ± 5.29 A
10–20	AC	10.48 ± 5.54 A	28.78 ± 2.85 A	14.42 ± 2.91 A	46.32 ± 6.33 A	0.99 ± 0.32 A	0.18 ± 0.05 B	39.26 ± 4.73 B
	SL	15.88 ± 5.49 A	31.61 ± 3.25 A	11.96 ± 0.96 A	40.55 ± 7.47 A	1.34 ± 0.34 A	0.26 ± 0.09 A	47.50 ± 7.33 AB
	WL	13.71 ± 3.44 A	34.27 ± 8.36 A	12.08 ± 1.44 A	39.94 ± 3.73 A	1.24 ± 0.12 A	0.25 ± 0.03 A	47.99 ± 5.01 A
20–30	AC	8.78 ± 3.63 AB	30.54 ± 0.59 A	14.36 ± 2.19 A	46.32 ± 3.13 A	0.90 ± 0.21 A	0.17 ± 0.03 AB	39.33 ± 3.39 A
	SL	5.70 ± 2.23 B	29.76 ± 9.15 A	13.71 ± 0.51 A	50.83 ± 7.01 A	0.71 ± 0.06 B	0.14 ± 0.03 B	35.45 ± 7.18 A
	WL	12.12 ± 2.79 A	31.17 ± 6.66 A	10.32 ± 3.78 A	46.38 ± 6.05 A	1.11 ± 0.09 A	0.20 ± 0.03 A	43.29 ± 4.00 A
30–40	AC	9.25 ± 3.85 A	36.57 ± 0.49 A	13.27 ± 4.31 A	40.90 ± 1.96 B	1.00 ± 0.22 A	0.22 ± 0.03 A	45.82 ± 3.40 A
	SL	3.61 ± 3.44 B	21.21 ± 3.09 B	13.23 ± 0.75 A	61.95 ± 5.42 A	0.49 ± 0.22 B	0.09 ± 0.02 B	24.82 ± 5.38 B
	WL	9.72 ± 3.75 A	36.97 ± 6.82 A	12.06 ± 1.88 A	41.26 ± 1.29 B	1.03 ± 0.15 A	0.22 ± 0.01 A	46.69 ± 3.13 A

Note: all the data were the means of triplicates with standard deviation; different letters reveal significant differences among different land uses (p < 0.05).

).

Soil aggregate stability was quantified through three established indices: mean weight diameter (MWD), geometric mean diameter (GWD), and water-stable aggregates (WSA) [33]. Comparative analysis demonstrated significantly elevated values of all three indices in woodland and shrubland ecosystems relative to anthropogenically disturbed sites. This observation aligns with established pedological principles where undisturbed natural soils typically exhibit greater macroaggregate abundance and enhanced structural integrity compared to artificially modified systems [34]. The effect of soil depth on large macroaggregates and silt + clay-associated particles was remarkable (Table 3, p < 0.05). For example, the contents of large macroaggregates seemed to be higher at the surface layer and decreased with soil depth. The superior aggregate stability in woodland soils arises from synergistic mechanisms: (1) minimal anthropogenic disturbance preserving soil architecture, (2) continuous input of root exudates and organic detritus fostering cementing agents, and (3) robust microbial communities facilitating organic matter stabilization [35]. These processes collectively enhance organo-mineral complexation, thereby promoting macroaggregate formation and mechanical resilience. In addition, the litter floor in woodland and shrubland protected the soil from rainfall disturbance and soil erosion [36].

3.3. Soil Water Retention Curve (SWRC)

All the data of soil water contents and suctions were used to fit either the van Genuchten model or the double-exponential model. The parameters of the fitted soil water retention curves of three land uses are presented in

Table 4. Determination coefficient (R²) and root mean square error (RMSE) of van Genuchten model and double-exponential model fitting to the soil water retention curves (SWRCs) of three land-uses.

Soil Depth	Land Use	van Genuchten Model		Double-Exponential Model
		R2	RMSE	R2	RMSE
cm		/	m3 m−3	/	m3 m−3
0–10	Abandoned Cropland	0.982	0.012	0.995	0.008
	Shrubland	0.985	0.011	0.996	0.007
	Woodland	0.994	0.007	0.998	0.005
10–20	Abandoned Cropland	0.978	0.013	0.995	0.007
	Shrubland	0.989	0.009	0.996	0.007
	Woodland	0.989	0.008	0.995	0.007
20–30	Abandoned Cropland	0.985	0.010	0.996	0.006
	Shrubland	0.981	0.012	0.995	0.008
	Woodland	0.982	0.013	0.996	0.007
30–40	Abandoned Cropland	0.989	0.010	0.996	0.007
	Shrubland	0.989	0.010	0.997	0.006
	Woodland	0.992	0.010	0.993	0.009

. The double-exponential model was better than the fit of the van Genuchten model of three land uses when comparing their determination coefficient (R²) (Table 4). The values of R² ranged from 0.993 to 0.998, indicating a good reliability of the double-exponential model in modelling the retention curves.

SWRCs obtained by the double-exponential model are shown in Figure 4. The observation of this study showed that SWRCs of different land uses performed differently in terms of soil depths. Soil water contents were higher in WL and SL, rather than AC, at 0–10, 10–20, and 20–30 cm (Figure 4). However, there seemed to be no significant difference among the three land uses at 30–40 cm. This was mainly because plant roots are mainly distributed in the soil surface layer. FC showed significantly higher values in WL compared to AC in 0–30 cm soil depth (

Table 5. Parameters of the double-exponential model fitted to the soil water retention curves (SWRCs), field capacity (FC), permanent wilting point (WP), and available water content (AWC) of three land uses.

Soil Depth	Land Use	A1	A2	h1	h2	C	FC	WP	AWC
cm		cm3 cm−3	cm3 cm−3	cm	cm	cm3 cm−3	cm3 cm−3	cm3 cm−3	cm3 cm−3
0–10	AC	0.21	0.14	60.51	4823.81	0.17	0.30	0.18	0.12
	SL	0.18	0.15	71.88	4554.15	0.19	0.33	0.20	0.14
	WL	0.13	0.17	80.50	4061.99	0.21	0.37	0.22	0.15
10–20	AC	0.19	0.14	81.90	6085.32	0.17	0.30	0.18	0.12
	SL	0.21	0.15	43.60	4392.98	0.20	0.34	0.21	0.14
	WL	0.16	0.16	45.58	4465.47	0.22	0.36	0.22	0.14
20–30	AC	0.13	0.15	68.48	5007.89	0.18	0.32	0.18	0.13
	SL	0.17	0.15	78.03	5386.34	0.20	0.34	0.21	0.13
	WL	0.18	0.14	87.47	5669.00	0.21	0.36	0.23	0.13
30–40	AC	0.16	0.17	82.57	5657.56	0.18	0.34	0.20	0.15
	SL	0.16	0.17	62.69	4694.08	0.19	0.36	0.20	0.15
	WL	0.15	0.17	61.02	4335.70	0.20	0.35	0.20	0.15

). However, SL showed no significant difference in FC compared to WL and AC in 0–10 cm, 10–20 cm, and 30–40 cm. FC of SL showed even lower than AC in 20–30 cm. In terms of AWC, WL and SL showed higher AWC than AC in 0–10 cm and 10–20 cm soil depth. No significant difference was observed between AWCs of three land uses in 20–30 cm and 30–40 cm (Table 5). These results suggested that woodland was the more effective land-use type for water conservation in this region, which was similar to results in Slovakia [37].

Pore-size distributions (Figure 5) were determined by the double-exponential model in water retention curves. Pore-size distributions of soil in AC, SL and WL showed a dual porosity curve, with two peaks in the structural and matrix pores areas. The two peaks referred to structural and matrix porosity, respectively. Compared to AC, SL and WL showed lower structural porosity, but more matrix porosity for 0–10 cm soil layers. However, in 10–40 cm soil layers, pore size was larger in WL and SL than in AC. The reason might be that soil macroaggregates increased and soil BD decreased in WL and SL. In previous studies, several researchers found that soil pore-size distributions depend on the soil aggregate hierarchy [14,38,39].

3.4. Soil Saturated Hydraulic Conductivity

Compared to AC, SL and WL showed higher soil-saturated hydraulic conductivity (Figure 6). WL showed the highest saturated hydraulic connectivity, ranging from 0.008 cm s⁻¹ to 0.019 cm s⁻¹, with an average of 0.013 cm s⁻¹. A high variety of plant species in the forest improved hydraulic conductivity by increasing root biomass, soil organic carbon, soil structure, and soil porosity [40]. Intensive tillage practices in agricultural systems induce chronic soil compaction and structural degradation, thereby significantly reducing saturated hydraulic conductivity through pore-connectivity disruption [39].

3.5. Relationships Between Soil Properties and Hydraulic Properties

Relationships between soil hydraulic properties and associated physicochemical parameters were statistically evaluated through Pearson correlation analysis, with results visualized in the correlation matrix presented in Figure 7. For the surface soils, structural porosity (A₁) showed significantly positive correlations with soil-saturated water content, and negative correlations with FC, WP, and AWC, while soil matrix porosity showed positive correlations with FC and AWC. There were strong and significant positive correlations between GWD, WSA and FC, WP, and AWC. This suggests that soil aggregate and porosity might be the main factors influencing soil hydraulic properties (e.g., soil field capacity, available water content, and hydraulic connectivity), which is consistent with previous studies from references [1,41]. However, the relationships between soil hydraulic properties and other parameters in 30–40 cm layers were weak (except for saturated water content vs. bulk density, and wilting point vs. matrix porosity).

Taking the surface 0–10 cm soil layer, which is most significantly influenced by land use types, as an example, this study analyzes the driving factors behind the evolution of soil hydraulic properties. A partial least-squares structural equation modeling (PLS-SEM) (Figure 8) was established to clarify the direct and indirect effects of soil properties on hydraulic properties in surface soil (0–10 cm). The PLS-SEM explained 93.3%, 95.8, and 69.4% of the variance in FC, AWC, and Ks. The main factors influencing soil hydraulic properties were soil aggregates, matrix and structural porosity, SOC, soil BD, etc.

Indicators with significant effects on soil FC were soil BD, matrix porosity, and structural porosity in surface soils. SOC and WSA showed indirect effects on soil FC through soil BD, matrix porosity, and structural porosity (Figure 8 and Figure 9). SOC and matrix porosity showed positive direct effects on soil AWC, while soil BD, WSA, and structural porosity showed negative direct effects. In addition, soil BD and WSA showed positive indirect effects on AWC by affecting soil matrix porosity. The main factors affecting soil Ks were soil BD, soil macroaggregates, and saturated water content (Figure 8 and Figure 9).

4. Discussion

4.1. Main Factors in the Evolution of Soil Hydraulic Properties During Revegetation

Converting cropland to forests improved soil water retention capacity and infiltration. Woodland showed the highest FC, AWC, and hydraulic connectivity compared with cropland and shrubland. Vegetation restoration primarily enhances soil water retention properties in surface layers, while subsoil (30–40 cm depth) shows no significant differences compared to AC, which can be explained by the evolution of soil structure. In general, soil aggregates and soil pores are two different perspectives on soil structure, which have significant effects on soil hydraulic properties [7]. Soil pore-size distribution can directly affect the storage and transport of soil water, which is also influenced by soil aggregates. Soil pores of different sizes play different roles, affecting soil hydraulic properties. Structural porosity, constituting the inter-aggregate pore network between aggregates or soil peds, encompasses macropores including interpedal voids, biopores, and desiccation cracks [40]. In contrast, matrix porosity refers to the intra-aggregate pore space between primary soil particles, predominantly governed by capillary forces that dictate soil water retention characteristics [10]. The more soil macroaggregates lead to a higher matrix porosity.

Revegetation improves soil structure through mechanisms such as root penetration, litter accumulation, and canopy interception, thereby influencing the movement and retention of soil water. For instance, tree and shrub–herb layers can mitigate rainfall impact and enhance water infiltration [41]. Forest ecosystems, with their abundant litter input, significantly increase soil organic matter content and aggregate stability [42]. The increase in water-stable aggregates can reduce soil erodibility factors and further reduce erosion and protect soil structure through the indirect action of aboveground biomass. The agglomerate structure affects the water permeation path. For example, the increase in the proportion of large aggregates in converted forest land promotes the formation of preferred flows and enhances water infiltration [38]. The concentration of both root systems and surface-litter accumulation in the topsoil layer explains the observed variation in surface soil water retention capacity.

Soil aggregates serve as the fundamental building blocks of soil architecture, whose formation and stability critically govern soil hydrological dynamics and erosion resistance. Functioning as the primary binding nucleus, organic matter orchestrates aggregate formation through multifaceted mechanisms encompassing physical cementation, chemical complexation, and biological mediation. The incorporation of organic substances emerges as a pivotal determinant in regulating the assembly and persistence of these structural units within soil ecosystems. After the conversion of cropland to forests, plant litter and root decomposition stimulated soil organic matter accrual, thereby decreasing bulk density and enhancing matrix porosity through improved aggregate formation and pore network reorganization (Table 2 and Table 5). The decomposition of plant residues (litter and roots) releases substantial plant-derived organic compounds into soil systems. These organic constituents, with densities (0.6–1.3 g/cm³) significantly lower than mineral particles like clay (2.6–2.8 g/cm³), directly reduce soil mass per unit volume, thereby lowering bulk density. Concurrently, biochemical byproducts of decomposition, particularly humic acids and microbial-derived binding agents, act as organic cementing materials that bridge mineral particles, promoting the formation of water-stable aggregates. The resulting aggregate hierarchy increased matrix porosity and improved hydraulic conductivity while maintaining moisture retention. Moreover, when cropland was abandoned, tillage disturbance was reduced, and physical damage to soil aggregates was reduced [39]. Tillage could accelerate organic matter decomposition and agglomerate fragmentation, while after returning to farmland, artificial disturbance was reduced, and the proportion of large aggregates increased significantly (Table 3).

Land-use conversion significantly changed the matrix and structural porosity (Table 5). The dual-pore system in soil creates complex dynamics of water and nutrient transport through the rapid channeling effect of macropores and the slow-release water retention of matrix pores [43]. This system exhibits dual impacts: enhancing infiltration rates (especially during initial rainfall/irrigation through gravity-driven preferential flow in macropores) and maintaining plant-available water via capillary action in micropores. The interlocking of large and small pores in soil jointly affects the water holding and conducting properties, which have a significant impact on the retention and transport of water and nutrient ions. Large pores in soil acted as fast pathways for water and nutrient transport, while small pores (matrix pores) primarily retained water and nutrients and served as internal solute sources of soil. Previous studies proposed water retention and conductivity models under a dual-porosity system [43,44]. A general solution for coexisting internal (matrix pores) and external boundary sources was also proposed [45]. In this study, WL exhibited greater matrix porosity, more water-stable soil aggregates, and superior water retention and infiltration capabilities compared to AC and SL, indicating enhanced water conservation functionality in woodland areas. The reason might be that soil macroaggregates increase while soil BD decreases in WL. In previous studies, several researchers found that soil pore-size distributions depend on the soil aggregate hierarchy [10,46,47]. Soil BD and total porosity might be another reason. Conversion from cropland to shrubland and woodland was beneficial for soil structure recovery from compaction [31], thus promoting the growth of plant roots. Root hole was considered an important source for inter-aggregate porosity [10]. The SOC effects came from providing energy and nutrients to soil organisms and, thus, improved soil aggregate stability [35,48]. Soil BD directly affected soil total porosity, including soil structural and matrix porosity, and, thus, influenced soil hydraulic properties.

The abandoned time from cropland may play a crucial role in the evolution of soil hydraulic properties. Our studies showed soil marcoporosity decreased, while matrix porosity and bulk density increased in surface layers after cropland abandonment. However, some studies [49,50] have pointed out that woodland showed lower soil bulk density, higher macroporosity than that of cropland, and included no matter in the surface soil or subsoil, which is different from the results of this study. The possible reason might be the short time after conversion. Some studies on no-till suggest that it may take substantial time to develop a good structure when changing from a tilled system to a system with less disturbance, and that topsoil may experience a period with increasing density [51]. In the early stage, the soil bulk density of surface layers increased significantly and porosity decreased, which led to the decrease of water-holding capacity and the deterioration of soil structure. Long-term restoration can improve soil structure: volume weight decreased, porosity increased, saturated water content increased, and soil aggregate stability increased [52]. For the 30–40 cm soil layer, 24 years of vegetation restoration did not significantly alter soil water retention capacity, indicating a weak influence of restoration on subsoil hydrology.

4.2. Implications for Future Research and Land Management

Extensive research has demonstrated that revegetation serves as a proven strategy for maintaining ecosystem hydrological integrity. These restored systems provide critical ecosystem services, including agricultural productivity, freshwater provisioning, and aquifer replenishment [1].

Our study indicated that woodland may be more effective in improving soil hydraulic properties than shrubland during revegetation. The synergistic enhancement of soil organic carbon content, aggregate stability, and matrix porosity in woodland ecosystems significantly improved soil water retention capacity and saturated hydraulic conductivity. This pedostructural optimization thereby enhances rainfall infiltration capacity while reducing preferential flow pathways, effectively mitigating surface runoff and increasing plant-available water-residence time. Consequently, these hydrological modifications substantially elevate precipitation utilization efficiency through improved soil moisture redistribution dynamics [51,53]. At the regional scale, it is possible to improve hydrological conditions, regulate regional water circulation, prevent water and drought disasters, and develop and utilize water resources rationally. Therefore, from the perspective of water-resources management in a water-limited region with large seasonal variations in precipitation, woodland might be a more suitable option for soil conservation and better ecosystem function.

Notably, this study controlled baseline conditions through matched soil types and cultivation histories, yet microtopographic parameters (e.g., surface roughness) under different treatments were not quantitatively characterized. These unresolved microtopographic variations constitute a methodological limitation that contributes to the uncertainty of the results. Furthermore, soil hydraulic property evolution analysis with cultivation abandonment timelines implicitly assumes inter-decadal climatic stability. However, observed precipitation fluctuations and long-term decreasing trends over multi-decadal scales (as per regional meteorological records) in North China [54] could exert cumulative impacts on soil characteristic development, a factor we recommend future studies explicitly incorporate through climate–soil interaction modeling.

To further elucidate the impacts of land-use change on soil hydraulics, future research should also focus on elucidating the dynamic interactions between soil aggregates, pore distribution, and vegetation restoration. These processes govern the evolution of soil structure and biodiversity and warrant investigation across multiple spatial scales to understand regional hydrological redistribution. The bidirectional interactions between soil and vegetation warrant deeper exploration: soil nutrients and moisture shape plant growth, while root exudates and plant litter reciprocally remodel soil structure. Plant metabolic pathways, such as γ-aminobutyric acid (GABA)-enhanced tolerance to saline-alkaline stress [55], further drive the co-evolution of soil–plant systems through physiological feedbacks. By conducting long-term temporal analyses and in-depth mechanistic investigations across diverse vegetation cover types, scholars can elucidate mechanisms of soil structure formation, stability, and resilience to environmental perturbations. These investigations have significant implications for promoting soil health and ecosystem service resilience. Theoretical advancements in this field will provide a robust scientific basis for evidence-based ecosystem management strategies, particularly in contexts of accelerated land-use transitions.

5. Conclusions

The study demonstrates that land-use conversion significantly alters soil hydraulic properties in semi-arid mountainous regions of North China, with woodland outperforming abandoned cropland and shrubland in soil–water conservation. Specifically, woodland soils exhibited higher field capacity, available water content, and saturated hydraulic conductivity, attributed to improved soil aggregates (>0.25 mm water-stable aggregates), optimized matrix porosity, and enhanced organic carbon. These findings highlight woodland restoration as an effective strategy for ecological programs like the Grain to Green Project in semi-arid North China, mitigating land degradation while enhancing hydrological resilience. This study also provides a valuable reference for formulating vegetation restoration strategies in other semi-arid regions and mountainous areas.