Impacts of Converting Native Grassland into Arable Land and an Avocado Orchard on Soil Hydraulic Properties at an Experimental Farm in South Africa

Abstract

The main objective of this study was to evaluate the changes in soil water retention curve (WRC) and pore size distribution in the 0–10, 10–20, and 20–30 cm layers following grassland conversion into arable land and an avocado orchard. Undisturbed soil cores were sampled using cylindrical metal cores to determine WRCs. The RETC program was used to fit the van Genuchten equation to the measured water retention data. The maximum equivalent radius (r) of soil pores retaining water at various matric potentials was calculated using the capillary rise equation. Significant differences between treatments were observed mainly in the 10–20 cm layers. Greater θ_s, n, and α in grassland were attributed to low bulk density and high soil organic carbon. Soil compaction in arable land and an avocado orchard was attributed to soil disturbance. The grassland had a greater share of macro- and mesopore volumes and large air capacity than the arable and avocado orchard. Overall, the results indicated that the conversion of native grassland causes substantial changes in soil hydraulic properties that could impact crop growth and the environment.

1. Introduction

Land use (LU) change is driven by human socio-economic needs. In agriculture, LU is usually influenced by the need to cultivate crops for food supply, fibre production, and other economic activities. However, land use changes may lead to the degradation of soil health and ecosystems [1,2]. It is well known that land use can influence soil chemical, physical, and biological properties significantly [3], but the mechanisms involved remain largely unknown. Hence, investigations related to land conversion and soil properties have become the focus of considerable research over the past two decades [4]. Nonetheless, the subject remains not fully understood due to the complexity of soil systems [5]. More so, studies on soil hydraulic properties following land use change have received little attention.

Soil hydraulic properties determine soil quality and its ecosystem functions the most [6]. Soil hydraulic properties are soil characteristics that control water movement and its storage, as well as nutrient and plant water uptake. Hence, these properties are vital for predicting plant–water relationships, irrigation, drainage, nutrient leaching, runoff, catchment management, and overall soil health [7]. The soil–water retention curve (SWRC) is one of the most important soil hydraulic properties. It is defined as the relationship between volumetric water content (θ) and soil matric potential (Ψ). It depends on soil structure, and hence, it is influenced by soil properties like soil texture and pore size distribution [8,9]. Changes in soil structure affect the SWRC and other soil ecosystem functions such as water storage and oxygen supply to the roots [10]. Accordingly, the properties of the SWRC can be used to assess the effect of different soil and crop management practices on the soil’s physical quality [11]. Few studies have examined the impacts of LU change on soil hydraulic properties in semi-arid climates [3,12], and, in particular, the conversion of native grassland to an avocado orchard, which is the focus of this study, is hardly reported. Nevertheless, there exist few studies in the literature whose foci were not on grasslands and avocado orchards per se but on different agroecosystems and forest ecosystems that were found to be relevant for comparison with the current study. For example, Gomez et al. [13] reported that no-till (NT) management in an olive orchard increased soil bulk density and reduced cumulative infiltration compared to conventional tillage (CT) after 25 years on clay loam soil in Southern Spain. They attributed the low infiltration to soil compaction. In contrast, Yimer et al. [12] reported a decreased infiltration rate and an elevated bulk density in cultivated land compared to a forest. Recently, Liao et al. [3] reported that, compared with farmlands, the initial infiltration rates and stable infiltration rates of forestlands, shrublands, and grasslands were higher, while those of orchards were lower in the surface layers in the Loess Plateau of China. They attributed the lower infiltration rates in the orchards to the existence of subsoil plough layers that increased soil bulk density and reduced infiltration rates. However, the impacts of land use change cannot be generalized due to their dependence on soil and vegetation types, as well as the period before measurement is taken after conversion [4]. For instance, Reynolds et al. [14] observed that irrespective of soil type, organic carbon, porosity, and air capacity were lower under NT and CT relative to woodlot (WL), while bulk density was greater. The water retention properties showed no consistent differences among WL, CT, and NT. Meanwhile, Hebb et al. [15] reported no significant effect on the hydraulic conductivity of sandy clay loam soils in their experiment after the conversion of natural meadows into farmland, contrary to the findings of Gajic et al. [5], who observed a significant increase in bulk density (p < 0.05) in the 0–15 cm and 15–30 cm layers in arable land after more than 100 years of continuous use after establishment from deciduous forestland in the meadows of Western Serbia. Similar results have been reported in the literature [16], albeit with inconsistencies over time [2]. Similarly, Owuor et al. [17] observed severe degradation of soil hydraulic properties after the conversion of natural forest to agricultural land in the Highlands of Kenya. They reported lower bulk density (p < 0.001) for native forests compared to tea plantations and pastures, while hydraulic conductivity was highest in native forest soils, echoing results from the literature [5].

An equally important hydraulic parameter is soil pore size distribution [18]. Yet, it has received little attention. Pore size distribution influences many soil properties, including hydraulic conductivity, water retention, and gas transport [9,19], as well as acting as a habitat for roots and soil biota [20]. Furthermore, pore size distribution and the continuity of pores control soil infiltration rates and are hence important for crop water supply [21]. Given its importance and sensitivity to management practice, it is vital to understand how land use change influences pore size distribution.

The foregoing literature review, which is by no means exhaustive, indicated a general trend of mixed and inconclusive results regarding the state of soil hydraulic properties following land use change. Nonetheless, the literature reveals that the effects of LU change, particularly from rangelands to agricultural lands, on soil hydraulic properties are widely reported, but none of these studies evaluated the changes in pore size distribution that might explain the observed hydraulic properties, including hydraulic conductivity, which is highly controlled by soil macropores and their connectivity. Additionally, a higher proportion of mesopores could explain why certain LU systems lead to increased plant-available water than others. Therefore, more research is needed to understand how land use conversion from natural grassland into arable land and an avocado orchard influences pore size distribution. The ability to quantify changes in soil hydraulic properties following land conversion provides insights into how new land use may impact soil health and hence the soil’s ecosystem functions [10]. Studies of this nature are presently not on record and yet are needed in countries such as South Africa, where land use change is a subject of growing importance [22]. Therefore, the objectives of this study were threefold: (1) to characterize the estimated hydraulic properties of the SWRC, (2) to evaluate pore size distribution, and (3) to assess soil bulk density following the conversion of a natural grassland into arable cropland and an avocado orchard in a clayey soil at 0–10, 10–20, and 20–30 cm depths. We hypothesized that land use conversions would alter soil structure and hence soil hydraulic properties due to soil disturbance by tillage implements/human activities and the removal of plant biomass, significantly degrading the hydraulic properties that control water and air movement. To test our hypothesis, we compared the SWRC and its hydraulic properties, pore size distribution, and bulk density after converting native grassland into arable cropland and an avocado orchard on a Ferralsol [23]. Ferralsols are vital soil types for cereal crop production in South Africa [24]. The results of this study could form the foundation for future studies and provide informed land use decisions and policies for similar environmental conditions.

2. Materials and Methods

2.1. Study Site

This study was carried out at the University of Venda Experimental Farm (22°58′ S, 30°26′ E), Thohoyandou, South Africa. The site has a semi-arid climate characterized by low and variable annual rainfall (±500 mm) and high evaporative demand. The mean maximum air temperature in summer (October to March) is ±30 °C, while the winter temperature (May to August) is ±18 °C [25]. Nearly thirty years ago, the site was occupied by savanna grassland, which was cleared to create blocks of land for crop cultivation and an avocado orchard. The avocado (Persea americana) orchard with 5 m × 5 m tree spacing was ploughed once at the time of its establishment, and thereafter, tractors were used to mow the grass between tree rows. The blocks of land each lie adjacent to one another in the west-to-east direction. The form of tillage practiced on the arable land involved ploughing with a disc plough to a depth of 20 to 30 cm, followed by a harrow. Disc ploughing is sometimes alternated with manual digging with hand hoes to shallow soil depths of between 10 cm and 20 cm when the field is used for field trials. The vegetables farmed are mainly butternut squash (Cucurbita moschata), cabbage (Brassica oleracea), and spinach (Spinacia oleracea L.). The grassland is utilized mainly for cattle grazing but may be vacant at times. The grassland is dominated by Rachypogon spicatus and Diheteropogon amplectans grass species [26] and was used as the control in the current study. The Ferralsols at the study site are deeply weathered (>120 cm) and reddish brown (2.5YR 3/4 dry; 2.5YR 3/3 moist) with the texture of clay. They have good internal drainage and a moderate, fine subangular blocky structure. The soil pH is slightly acidic (pH_KCl = 5.4). They have low base saturation (18%) and relatively high cation exchange capacity (15–19 c mol_c (+) kg⁻¹). The clay fraction is dominated by kaolinite mineralogy (99%) and moderate soil organic carbon (<2%) [25]. The soils at the site were classified as Umbric Rhodic Ferralsols according to the World Reference Base for Soil Resources (WRB) [23]. Selected soil characteristics are shown in

Table 1. Particle size distribution and soil organic carbon (SOC) and bulk density (D_b) (mean ± standard deviation) of Umbric Rhodic Ferrallisols in 0–10 cm, 10–20 cm and 20–30 cm depth layers of native grassland, arable land, and avocado orchard soils at the University of Venda.

Layer (cm)	Land Use	Particle Size Distribution (%)			SOC (%)
		Sand	Silt	Clay
0–10	Arable	23.6 a ±3.0	20.1 ±2.5	56.3 a ±3.6	1.48 ±0.58
	Avocado	26.2 a ±2.3	25.3 ±2.3	48.5 b ±1.8	1.34 ±0.19
	Grassland	27.3 b ±3.9	25.4 ±2.0	47.3 b ±3.3	1.30 ±0.63
10–20	Arable	25.0 ±2.5	23.0 a ±3.6	52.0 a ±4.7	1.38 ab ±0.42
	Avocado	25.1 ±2.4	19.1 b ±2.8	55.8 a ±2.8	1.19 a ±0.15
	Grassland	25.6 ±3.0	24.7 a ±1.4	49.7 b ±0.7	1.55 b ±0.25
20–30	Arable	24.2 a ±1.7	20.3 a ±3.2	55.5 a ±5.8	1.24 ±0.18
	Avocado	24.0 a ±1.6	20.0 a ±2.2	56.0 a ±2.2	0.79 ±0.38
	Grassland	24.1 a ±4.4	28.2 b ±4.5	48.7 b ±1.3	0.94 ±0.63

Sand = 2.0–0.05 mm, silt = 0.05–0.002 mm, and clay = <0.002 mm. Different letters within the soil depth indicate significant differences between land use at p ≤ 0.05.

.

2.2. Soil Sampling and Laboratory Procedures

For each land use type, 9 representative soil profiles were sited along a transect in the middle of each land unit on flat topography and dug to a depth of 100 cm in July 2023. In the case of the avocado orchard, the profiles were sited between tree rows. Undisturbed soil samples were collected in triplicates from 0–10, 10–20, and 20–30 cm layers using cylindrical stainless steel core rings (100.1 cm³) for the determination of bulk density, water retention, and pore size distribution. Disturbed samples of about 1 kg per soil layer were collected from the same layers for the determination of particle size distribution using the hydrometer method [27], and soil organic carbon was determined using the Wakely and Black method after passing the soil through a 0.5 mm sieve [28].

2.3. Determination of the Soil Water Retention Curve and Pore Size Distribution

The bottoms of the soil core samples were covered at the lower end with a piece of nylon cloth fixed with an elastic O-ring, as described in the manual. The marked samples were placed in the sandbox and saturated by raising the water level to 1 cm below the top of the sample rings and allowed to stand for 1 week. Water retention was measured at 0, −1, −2.5, −10, −31.6, −63.1, and −100 cm matric potentials using sandbox equipment (Eijkelkamp, Giiesebeek, The Netherlands). Disturbed samples were subjected to −330 and −15,000 cm matric potentials using 5-bar and 15-bar pressure plates, respectively. Both pressure chambers are products of Eijkelkamp, Giesebeek, The Netherlands. Wet masses of soil samples were measured after equilibrating at each matric potential. At the end of desaturation at predetermined matric potentials, the samples were dried at 105 °C for 24 h, and the gravimetric water content at each matric potential was calculated as the difference between wet mass and dry mass. Similarly, dry bulk density was calculated as the ratio of dry mass and sample volume (100.1 cm³). Then, the volumetric water content (cm³ cm⁻³) was calculated by multiplying the gravimetric water content and the dry bulk density.

The van Genuchten (vG) model [29] was fitted to the measured water retention data to estimate its parameters. The vG model is a smooth function, which is differentiable and provides a few parameters that are easy to interpret. Furthermore, the vG model is applicable over a wide range of soil types. Fitting was achieved using the RETC program. This is a computer program developed to analyse the hydraulic conductivity and water retention of unsaturated soils [29]. The program is widely used in soil science and hydrology. The vG is expressed in Equation (1) as follows:

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

(1)

where θ is the volumetric water content (cm³ cm⁻³), θ_s is the saturated water content (cm³ cm⁻³), θ_r is the residual water content (cm³ cm⁻³), h is the matric potential (cm), α is the scaling parameter as an inverse matric potential at air entry (cm⁻¹), n is a fitting dimensionless parameter related to curve shape, porosity, and pore size distribution, and m is a constant defined in Equation (2) as follows:

$$m=1-{\displaystyle \frac{1}{n}}$$

(2)

Field capacity (FC) and permanent wilting point (PWP) were determined at −330 and −15,000 cm matric potential, respectively, from the water retention curve. Plant-available water (PAW) was calculated as the difference between FC and PWP. Air capacity (AC), which indicates soil aeration, was calculated as the difference between θ_s and FC [6,10].

The maximum equivalent radius (r) of soil pores retaining water at the potential (h) of water in soils was calculated using the Young–Laplace capillary rise equation [5,11]:

$$r={\displaystyle \frac{2ɣcos\theta }{\rho g\left|h\right|}}$$

(3)

where r is the equivalent pore radius (m), h is the matric potential (m), ɣ is the surface tension of water (72.75 mJ m⁻²), θ is the pore water contact angle (taken to be zero), ρ is the density of water (0.998 Mg m⁻³), and g is the gravitational acceleration (9.8 m s⁻²). After substituting the constants into Equation (3), the equivalent pore diameter (d, μm) at corresponding matric potentials is approximated as in Equation (4):

$$d\approx {\displaystyle \frac{3000}{\left|h\right|}}$$

(4)

where h is the matric potential (cm). Equation (4) was used to calculate the diameter d of the largest equivalent pore sizes at any given soil water potential h [11].

Different pore size categories can be defined depending on one’s objectives, but there is no standardization in the separation of pores [11]. For this study, 3 pore size categories were defined based on their equivalent diameters according to Luxmore [30] as follows: >1000 μm (macropores or transmission pores), pores draining at |h| < 3 cm, 10–1000 μm (mesopores or storage pores), pores that drain at |h| = 3–300 cm and <10 μm (micropores or residual pores), and pores that drain at |h| > 300 cm. Draining matric potentials for each category of pore size diameters were calculated using Equation (4). Each pore size category has specific functions relative to water transport. Macropores hold water loosely, such that it freely drains under gravity and is unavailable to plants; mesopores hold water strongly enough that it does not drain away under gravity and is readily available to plants; and residual pores hold water so tightly that it is unavailable to plants [31].

2.4. Statistical Analyses

The difference between the hydraulic properties of the three land uses at three soil depths (0–10, 10–20, and 20–30 cm) was tested separately using one-way analysis of variance (ANOVA version: Precision Pro). The means between the land uses were compared using Fischer’s Least Significant Difference (LSD) at p ≤ 0.05 using the SPSS 29 statistical package. The derived parameters of the vG model from each set of measured moisture data were tested for normality using the Shapiro–Wilk test and subjected to ANOVA, and the means between the land uses were separated using the LSD (p ≤ 0.05).

3. Results and Discussion

3.1. Soil Particle Size Distribution and Soil Organic Carbon

The results of the one-way ANOVA indicated significant (p ≤ 0.05) variations in soil particle size distributions between land uses in the 0–30 cm layer (Table 1). Overall, the soils in all treatments had similar textures, which were classified as clayey according to the USDA Soil Classification System. Apparent differences in particle size distribution in adjacent fields were attributed to inherent soil variability and not a result of different land uses. However, since the soil textural group was similar across the land use types, the soils’ physical characteristics, like water-holding capacity and drainage, tend to be similar. Nonetheless, variations in soil organic matter content can cause the soils in the same textural group to differ slightly. In the present study, though, SOC, which is a proxy for soil organic matter, showed little variation between land use types. Therefore, changes in soil properties were largely attributed to land use conversion. The range of means of sand, silt, and clay concentration in the 0–30 cm layer for all treatments is 23.6–27.3%, 19.1–28.2%, and 47.3–56.3%, respectively. Soil organic carbon in the 0–30 cm layer was less than 2%, which is typical for South African soils [24]. Similar results have been reported [25]. The average SOC content was 1.37%, 1.37%, and 0.99% in the 0–10, 10–20, and 20–30 cm layers, respectively. The SOC concentration decreased with depth, as expected. Significant differences in SOC between land uses were observed in the 10–20 cm layer, but the difference was not statistically significant in the 0–10 and 20–30 cm layers. Compared to the grassland, SOC decreased by 11% and 23% in the arable land and avocado orchard, respectively. Grasslands are associated with the accumulation of organic matter compared to croplands due to their extensive rooting system and continuous input of plant biomass. Comparable to our results, Ghimire et al. [32] reported 37% greater soil organic carbon in the grasslands than the croplands in the 0–20 cm layer of clay loam soil in a semi-arid climate of New Mexico. Furthermore, they reported no significant difference between the treatments in the lower subsoil layers, in agreement with our results. Conversely, tilling the land has been shown to destroy soil organic matter by exposing the organic matter to microbial activity [33]. Hence, losses of SOC in the cultivated land are consistent with findings presented in the literature. The avocado orchard exhibited an SOC concentration like the arable land, probably due to losses of tree leaves by blowing wind, which left the soil surface bare and consequently without organic matter inputs. The tree-shedding effect that prevented grasses from proliferating could be another plausible reason for the low organic matter content, like the arable land.

3.2. Soil Bulk Density

The mean bulk densities for all land use types were 1.29 g cm⁻³, 1.28 g cm⁻³, and 1.34 g cm⁻³ in the 0–10 cm, 10–20 cm, and 20–30 cm layers, respectively, and significantly different in the 10–20 cm and 20–30 cm layers (Figure 1). Compared to the grassland, the bulk density was significantly higher (p = 0.032) in the 10–20 cm layer by nearly 8% and 11% in the arable land and the avocado orchard, respectively. Relative to the grassland, the bulk density increased even more in the 20–30 cm layer by 13% and 16% in the arable land and the avocado orchard, respectively. When averaged over three depths, the bulk density in the grassland (1.23 g cm⁻³) was significantly lower than in the arable land (1.33 g cm⁻³) and avocado orchard (1.34 g cm⁻³) (Figure 1b). Like in the current study, Buruso et al. [34] reported high bulk density under cultivation compared to grazing and natural forests, further confirming that tillage operations elevate bulk density. In contrast, Yimer et al. [12] reported a similar bulk density between grazing land and cultivated land on soils with mixed textures developed from volcanic ash, indicating the importance of soil types when interpreting results of this kind. The increased bulk density in the arable land and avocado orchard compared to the grassland, particularly in the 10–20 and 20–30 cm layers, was attributed to soil compaction due to the formation of plough layers by farm machinery, as previously reported [5,13]. However, others obtained the opposite results (Ghimire et al. [32]).

Meanwhile, Jarvis et al. [35] reported that bulk density showed no apparent trend with land use, contrary to Jabro and Stevens [11], who reported lower bulk densities in conventionally tilled soils compared to no-till soils. They attributed this to loosening by tillage operations, thereby lowering bulk density. However, it is difficult to compare the effects of tillage, such as CT practices, because they are defined differently between studies.

3.3. Soil Water Retention Curve and Fitted vG Model Parameters

The saturated water content (θ_s) represents the highest amount of water the soil can hold at saturation [29].

Table 2. Fitted parameters (mean ± standard deviation) of the vG model of the SWRC as influenced by land use in the 0–10, 10–20, and 20–30 cm layers at the University of Venda, South Africa. θ_s is the saturated water content (cm³ cm⁻³), θ_r is the residual water content (cm³ cm⁻³), α is the scaling parameter as an inverse matric potential at air entry (cm⁻¹), and n is a fitting dimensionless parameter related to curve shape, porosity, and pore size distribution. The parameters were fitted using the RETC program.

Depth (cm)	Land Use	θs (cm3 cm−3)	θr (cm3 cm−3)	n	α (cm−1)
0–10	Arable	0.528 ±0.02286	0.0006 ±0.00044	1.3538 a ±0.05198	0.0133 ±0.00956
	Avocado	0.487 ±0.01465	0.0005 ±0.00038	1.3824 a ±0.05361	0.0130 ±0.00481
	Grassland	0.5355 ±0.03223	0.0452 ±0.03498	1.6549 b ±0.0391	0.0081 ±0.00115
10–20	Arable	0.4892 a ±0.01616	0.0005 ±0.00020	1.3507 ±0.04708	0.0099 a ±0.00162
	Avocado	0.4548 a ±0.00550	0.0006 ±0.00020	1.3993 ±0.08577	0.0107 a ±0.00601
	Grassland	0.5726 b ±0.02506	0.0001 ±0.00015	1.3428 ±0.03364	0.0465 b ±0.01153
20–30	Arable	0.4791 ±0.08140	0.0002 ±0.00017	1.3856 ±0.17952	0.0567 ±0.09140
	Avocado	0.4737 ±0.00576	0.0007 ±0.00025	1.3696 ±0.03211	0.0142 ±0.00269
	Grassland	0.5448 ±0.00366	0.0025 ±0.00416	1.4270 ±0.04545	0.0160 ±0.00162

Different letters in a column within a soil depth indicate significant differences between land uses at p ≤ 0.05.

shows the fitted vG parameters obtained using the RETC program. The highest (0.5726 cm³ cm⁻³) and lowest (0.4548 cm³ cm⁻³) mean θ_s for all layers were observed in the grassland and avocado orchard, respectively. Significant variations in θ_s were observed in the 10–20 cm layer, where the grassland stored 8% and 12% more water at saturation than the arable land and avocado orchard, respectively (Table 2). Averaged over the three layers (0–30 cmthe grassland contained significantly (p < 0.001) more volumetric water content (0.5510 cm³ cm⁻³) at saturation than the arable land (0.4988 cm³ cm⁻³) and avocado orchard (0.4691 cm³ cm⁻³), which were statistically similar. This result is consistent with the low bulk density and high SOC of the grassland observed in the current study (Table 1). Our results concur with those of Smith et al. [36], who observed that volumetric water contents at saturation always decreased with increasing bulk density, regardless of soil texture in South African soils. They ascribed this to the creation of meso- and micropores at the expense of macropores, leading to reduced porosity. Contrary to our results, tillage showed no significant effect on θ_s in the 0–15 and 15–30 cm depths in an experiment spanning over 4 years. However, when the results were averaged across 4 years and 2 depths (0–30 cm), they reported that CT had more saturated volumetric water than the NT on sandy loam [11]. They associated θ_s with soil loosening and disturbance induced by CT practice. Meanwhile, Gajic et al. [5] observed similar θ_s values among forest, meadow, and arable land.

No significant variations in residual water content (θ_r) were observed across all land uses and soil layers (Table 2). Similar results have been reported in the literature [5,10,11]. However, it is noteworthy that the θ_r values were close to zero for all land uses and soil layers. The residual water content, θ_r, defines the maximum amount of water in a soil that will not contribute to liquid flow either due to blockage from the flow paths or strong adsorption by the soil matrix and is an extrapolated parameter and hence may not necessarily represent the lowest possible water content in the soil [29].

The average fitted parameter, n, related to curve shape, porosity, and pore size distribution was 1.4637, 1.3497, and 1.3941 in the 0–10 cm, 10–20 cm, and 20–30 cm layers, respectively (Table 2). Figure 2, Figure 3 and Figure 4 show the vG-fitted and measured soil water retention curves for different land uses. The average α fitting parameter, representing the inverse of the air-entry potential, in the 0–10 cm, 10–20 cm, and 20–30 cm layers was 0.0344 cm⁻¹, 0.0221 cm⁻¹, and 0.0869 cm⁻¹, respectively. Land use affected the shape of the SWRC significantly only in the 0–10 cm layer where the n fitting parameter for the grassland (1.6549) was 1.2 times higher than that of the arable land and the avocado orchard (Table 2 and Figure 2a, Figure 3a and Figure 4a), contrary to the results of Gajic et al. [5], who reported higher n values in arable land compared to the forest and meadow land use types. The α parameter was affected significantly in the 10–20 cm layer, with the value in the grassland approximately more than four times that of the arable land and the avocado orchard (Figure 2b, Figure 3b and Figure 4b). The α values in this layer were equivalent to the air-entry potential values (AEVs) (i.e., 1/α) of 21.5 cm, 93.5 cm, and 101 cm in the arable land, the avocado orchard, and the grassland, respectively. The results indicate that more suction must be applied to empty large pores from the grassland soil [8]. However, the AEVs obtained in this study increased with decreasing bulk density, contrary to the report by Jabro and Stevens [34], who showed that AEVs increased with an increase in soil bulk density, suggesting that more studies are needed to establish undisputable relationships between bulk density and α values. There were no significant differences between the vG fitting parameters in the arable land and the avocado orchard in this study, probably showing that the grassland promoted better distribution of soil pores.

Table 3. Mean ± standard deviation of field capacity (FC), permanent wilting point (PWP), plant-available water (PAW), and air capacity (AC) as influenced by land use in the 0–10, 10–20, and 20–30 cm depths. Field capacity was determined as the volumetric water content at |h| = 330 cm, where h is the matric potential (cm); PWP is the water content at |h| = 15,000 cm; PAW is the water stored between |h| = 330 and 15,000 cm; and AC is the water stored between |h| = 0 (saturation) and 330 cm.

Depth (cm)	Land Use	Total Porosity	FC	PWP	PAW	AC
		cm3 cm−3
0–10	Arable	0.528 ±0.02286	0.3119 a ±0.02795	0.0867 ±0.01198	0.2252 ±0.02837	0.1931 ±0.08093
	Avocado	0.487 ±0.01465	0.2719 b ±0.01084	0.0662 ±0.00925	0.2056 ±0.01949	0.2069 ±0.04360
	Grassland	0.5355 ±0.03223	0.2684 b ±0.01176	0.0359 ±0.00549	0.2325 ±0.00929	0.2671 ±0.05833
10–20	Arable	0.4892 ±0.01616	0.3095 a ±0.01308	0.0864 ±0.01881	0.2231 a ±0.00670	0.1797 a ±0.00744
	Avocado	0.4548 ±0.00550	0.2775 b ±0.01606	0.0654 ±0.01622	0.2121 a ±0.03071	0.1773 a ±0.01806
	Grassland	0.5726 b ±0.02506	0.2257 c ±0.01450	0.0559 ±0.01839	0.1698 b ±0.01242	0.3469 b ±0.01557
20–30	Arable	0.4791 ±0.08140	0.3079 ±0.03651	0.0856 ±0.04355	0.2223 ±0.07705	0.1712 ±0.10898
	Avocado	0.4737 ±0.00576	0.2613 ±0.00459	0.0473 ±0.0230	0.2141 ±0.02737	0.2124 ±0.00450
	Grassland	0.5448 ±0.00366	0.2636 ±0.00755	0.0554 ±0.00858	0.2082 ±0.00284	0.2812 0.01031
		Average across 3 depths
0–30	Arable	0.4988 ±0.04852	0.3098 a ±0.02397	0.0862 a ±0.02447	0.2235 ±0.04121	0.1813 a ±0.06864
	Avocado	0.4691 ±0.01376	0.2702 b ±0.01223	0.0596 b ±0.01748	0.2106 ±0.02848	0.1988 a ±0.02245
	Grassland	0.5510 ±0.0264	0.2526 b ±0.02261	0.0491 b ±0.01442	0.2035 ±0.04848	0.2984 b ±0.04384

Different letters in a column within the soil depth indicate significant differences between land uses at p ≤ 0.05.

shows water retention properties estimated from SWRCs. The field capacity (FC) ranged from 0.226 cm³ cm⁻³ to 0.3119 cm³ cm⁻³, falling within the range of South African soils [37] with averages of 0.2841 cm³ cm⁻³, 0.2709 cm³ cm⁻³, and 0.2776 cm³ cm⁻³ in the 0–10 cm, 10–20 cm, and 20–30 cm layers, respectively. Compared to the grassland, the FC increased by 16% in the arable land. However, the difference was not significant between the grassland and the avocado orchard. When averaged between the three depths, the results followed a similar trend whereby the FC in arable land was significantly greater than that of the avocado orchard and the grassland, which were not statistically different (Table 3). The permanent wilting point (PWP) did not differ between land uses and was between 0.0359 and 0.0867 cm³ cm⁻³. However, when averaged over the three depths, that of the arable land was significantly higher than that of the avocado orchard and the grassland, which were statistically similar (Table 3). The PWP range in the 0–30 cm layer was within the expected range of 0.0 to 0.30 cm³ cm⁻³ of South African soils [37]. High FC in the 0–30 cm layer could partly be explained by a high proportion of macropores in the grassland, specifically in the 10–20 cm layer, compared to the arable land (Figure 5). Macropores promote water drainage at field capacity [11]. Conversely, a higher proportion of micropores in the arable land than in the grassland could result in more water storage in the arable land. Similar results were reported by Gajic et al. [5], who attributed the destruction of macropores, resulting in the formation of micropores in tilled soils, to the destruction of structural aggregates by tillage implements. Meanwhile, in other studies, land use showed no significant effect on FC, whereas PWP was statistically significant in the 0–15, 15–30, and 30–45 cm depths when forest, meadow, and arable land were compared [5]. Furthermore, Jabro and Stevens [11] reported that NT showed significantly higher volumetric water contents at FC than CT after 4 years of establishment.

Plant-available water (PAW) ranged from 0.1698 to 0.2252 cm³ cm⁻³ with significant differences observed in the 10–20 cm layer (Table 3), where PAW in the grassland (0.1698 cm³ cm⁻³) was significantly lower than in the arable land (0.2231 cm³ cm⁻³) and the avocado orchard (0.2121 cm³ cm⁻³), which were statistically similar. Compared to the grassland, the PAW in arable land was 31% higher. However, there was no significant difference in PAW when the volumetric water contents of the three layers were averaged (Table 3). The air capacity (AC), which is also called air-filled porosity, ranged from 0.1712 to 0.3469 cm³ cm⁻³ with averages of 0.2224 cm³ cm⁻³, 0.2346 cm³ cm⁻³, and 0.2216 cm³ cm⁻³ in the 0–10 cm, 10–20 cm, and 20–30 cm layers, respectively. Significant differences in AC between land uses were observed in the 10–20 cm layer, where the grassland stored close to 50% more air than the arable land and the avocado orchard (Table 3). Additionally, when averaging AC over the three soil layers, compared to the grassland, the arable land and the avocado orchard contained 39% and 33% significantly less air in the 0–30 cm layer. The higher PAW in the arable land was associated with high FC. However, it must be noted that discussing PAW is difficult because it depends on field capacity, which is an arbitrarily defined parameter varying with soil and crop characteristics [36,37]. Moreover, field capacity is expressed either on a mass basis or volumetric basis, making it difficult to compare [36]. AC was greater in the grassland due to a decrease in FC and an increase in total porosity. Similarly, Schwarzel et al. [10] attributed the decrease in air capacity under the pastures relative to the croplands to a decrease in porosity, although their FC was at a matric potential of −100 cm instead of −330 cm as in our study. Lal and Shukla [38] reported that the lower critical limit of air-filled porosity is 10%. Therefore, the studied land use types do not limit crop growth as far as oxygen supply is concerned. Our results agree with those of Gajic et al. [5], who reported lower air-filled porosity in arable land compared to a forest and natural meadow.

3.4. Soil Pore Size Distribution

The pore size distribution derived from the SWRC is depicted in Figure 5. The average volume of the macropores was 0.007 cm³ cm⁻³ and 0.0201 cm³ cm⁻³ in the 0–10 and 20–30 cm layers, with no significant differences within each depth. However, significant differences in the macropores were observed in the 10–20 cm layer, in which the grassland recorded a volume of 0.0193 cm³ cm⁻³, which was higher than that of the arable land and avocado orchard by 60% and 75%, respectively. A similar trend was observed in the mesopores, where the grassland recorded about 50% more pore volume than both arable land and the avocado orchard. In the 20–30 cm layer, statistically significant differences between land uses were observed in the mesopore fraction, whereby the grassland recorded an average of 0.2846 cm³ cm⁻³, which was 50% and 23% more than that of the arable land and avocado orchard, respectively. A reverse trend was observed in the volume of the micropore fraction in the 10–20 cm layer, whereby that of the arable land (0.3000 cm³ cm⁻³) and the avocado orchard (0.2867 cm³ cm⁻³) significantly exceeded that of the grassland (0.2267 cm³ cm⁻³) although the total porosity in the latter was significantly higher than in the former. When averaged over the three depths, no significant differences due to land use were observed in the macropores. However, the grassland recorded 42% and 33% greater mesopore volumes compared to the arable land and avocado orchard, respectively, which were statistically similar (Figure 5d). Overall, when the entire measuring depth was considered, our results indicated that the volume of macropores (>1000 µm) and mesopores (10–1000 µm) was considerably higher in the grassland than in both the avocado orchard and the arable land. Yet, the opposite was observed for micropores (<10 µm). The results suggest that tillage damages structural aggregates and destroys macropores by shrinking pores within aggregates, as previously reported [5]. Hence, the supply of water may be limited in the arable land and the avocado orchard, thereby impacting crop growth. Furthermore, Lipiece et al. [39] reported that 10% of macropores (>0.5 mm) and mesopores (0.06–0.5 mm) contributed approximately 89% of the water flux. Therefore, our results further suggest that water runoff losses are likely to occur in the arable land and the avocado orchard, worsening crop water shortages. The results of this study agree with those of Gajic et al. [5], who reported lower macro-porosities and meso-porosities in arable land compared to a forest in the 0–30 cm depth, but they observed no significant differences between treatments in the 30–45 cm depth. However, their pore size categories were defined differently from those of the current study. Unlike our results, Lipiece et al. [39] observed differences in pore size distribution to a greater degree in the 0–10 cm layer than 10–20. In addition, they reported a greater proportion of large pores under CT than NT and attributed it to annual mixing and homogenization by the plough. However, our CT was sometimes performed manually using hand hoes; thus, it differed largely from the CT described in their study.

4. Conclusions

In this study, the impacts of converting native grassland to arable land and an avocado orchard at a semi-arid site in South Africa were examined. The soils at the site were classified as Umbric Rhodic Ferralsols according to the World Reference Base for Soil Resources (WRB). The objective of this study was to evaluate the changes in hydraulic soil properties, such as soil bulk density, soil water retention curve and its vG parameters, and pore size distribution in the 0–10, 10–20, and 20–30 cm layers. Generally, the hydraulic properties in the control (grassland) were more favourable compared to the arable land and the avocado orchard, which were invariably similar. The results of this study indicate that the conversion of native grassland resulted in a significant increase in bulk density in the arable land and the avocado orchard relative to the grassland, particularly in the subsoil layers. This was attributed to the formation of plough layers. Therefore, management strategies must include introducing reduced tillage with frequent use of a ripper plough to break the pan in the arable land and the application of organic amendments in the orchard. The fitted parameters of the vG equation showed that the saturated volumetric water content (θ_s) was significantly higher in the grassland than in the arable land and the avocado orchard in the 10–20 and 0–30 cm layers. In addition, converting the native grassland to arable land and an avocado orchard influenced the shape of the SWRC, with n being significantly greater in the grassland than in the arable and avocado orchard, particularly in the 0–10 cm layer. Meanwhile, the α parameter was significantly greater in the grassland than in the arable land and the avocado orchard, particularly in the 10–20 cm layer. The field capacity (FC) was significantly greater in the arable land than in the grassland and the avocado orchard in the 0–10 and 0–30 cm layers. The differences in FC were attributed to the high proportion of macropores that promoted water drainage in the grassland. The increase in FC in the arable land was accompanied by an increase in plant-available water, particularly in the 10–20 cm layers. However, the air capacity (AC) was significantly greater in the grassland compared to the arable land and the avocado orchard in the 10–20 cm layer. This was attributed to a higher volumetric water content at saturation, accompanied by a lower FC in the grassland compared to the avocado orchard and arable land. There was no significant variation in the pore sizes of all categories in the 0–10 cm depth. However, the volume of macropores was significantly greater in the grassland than in the arable land and the avocado orchard in the 10–20 cm layer. This trend was repeated in the mesopore fraction in the 10–20, 20–30, and 0–30 cm layers. Overall, the volume of micropores was greater in the arable land than in the grassland and the avocado orchard, suggesting that the grassland stored more water for plant uptake compared to both the arable land and the avocado orchard. Overall, the grassland had favourable properties that would promote plant growth, confirming our hypothesis that converting a grassland to arable land and an avocado orchard can result in substantial degradation of hydraulic soil properties. Nevertheless, more multi-location studies are needed to validate these results.