Intra-Aggregate Pore Network Stability Following Wetting-Drying Cycles in a Subtropical Oxisol Under Contrasting Managements

Abstract

One type of pore fundamental to water dynamics is the intra-aggregate pore, which holds water vital for plant and root system development, mainly in finer-textured soils such as clays. The distribution of intra-aggregate pores also influences the redistribution of water. Thus, it is important to study the dynamics of the intra-aggregate pore network under processes such as wetting and drying cycles (WDC). Changes in these pore types can play essential roles in organic matter protection, water movement, microbial activity, and aggregate stability. To date, there are few studies analyzing the impact of WDC on intra-aggregate pore dynamics. This study aims to provide results in this regard, analyzing changes in the pore architecture of a subtropical Oxisol under no-tillage (NT), conventional tillage (CT), and forest (F) after WDC application. Three-dimensional X-Ray microtomography images of soil aggregate samples (2–4 mm) subjected to 0 and 12 WDC were analyzed. The results showed that WDC did not affect (p > 0.05) the imaged porosity, number of pores, fractal dimension, tortuosity, and pore connectivity for the different soil management types. To analyze the permeability and hydraulic conductivity of the soil pore system, the most voluminous pore (MVP) was examined. No differences were observed in the imaged porosity, fraction of aggregate occupied by the MVP, connectivity, tortuosity, hydraulic radius, permeability, and hydraulic conductivity between 0 and 12 WDC for the MVP. Comparing soil management types after 12 WDCs, for example, F samples became more porous than CT and NT samples. In contrast, the pore system of NT had a lower fractal dimension and was more tortuous than that of CT and F samples. Our results show that for highly weathered soils such as the Brazilian Oxisol studied, the intra-aggregate pore network proved resistant to changes with WDC, regardless of the type of management adopted.

1. Introduction

Different management strategies can lead to significant changes in soil hydraulic and structural properties [1]. Particularly, the choice of an appropriate land-use system plays a crucial role in the productivity and sustainability of agricultural systems. Among the available options, three land-use systems widely used worldwide are sustainable forest (F), conventional tillage (CT), and no-tillage (NT) [2,3]. Sustainable forest is a management practice based on maintaining a dense and diverse forest cover whose vegetation protects the soil from erosion, retains water and nutrients, and promotes soil biodiversity [4]. Studies such as [5] highlight the benefits of forests in soil conservation, restoration of degraded areas, and climate change mitigation. However, for economic purposes, its results are long-term. CT is characterized by the use of chemicals, such as fertilizers, and soil disturbance caused by plowing and harrowing [6]. Although widely used in conventional agriculture, this practice can sometimes lead to problems such as soil compaction, erosion, loss of biodiversity, and degradation of natural resources [7]. However, CT can often be employed for mechanical weed control, reducing the use of herbicides. Another important management is NT, characterized by producing minimum disturbance in the soil structure and by maintaining the plant residues in the topsoil layer. The NT is considered a sustainable agriculture practice, showing important impact on soil conservation due to better water infiltration rate, plant available water, and a decrease in surface runoff [8]. However, inadequacies such as a lack of crop rotation, a lack of intercropping, among others, can cause an imbalance in the population of arthropods and pathogenic microorganisms, a high incidence of pests and diseases, and lead to the intensive use of insecticides and fungicides. Therefore, the selection of an appropriate land-use system depends on the field characteristics, the objectives of the farmer, and the environmental conditions.

Several processes can affect soil structure, such as the influence of wetting and drying cycles (WDC). These cycles are common in soils and play a fundamental role in soil dynamics and the functioning of terrestrial ecosystems [9]. WDC refer to the alternation between periods of increased soil moisture due to water input from rainfall, irrigation, or capillary rise, and periods of drying due to water loss from evaporation and plant transpiration [10]. Water enters the soil and fills the pores during wetting, often altering its structure. This process can lead to changes in plant water and nutrient availability, soil biological activity, organic matter distribution, and soil physical properties [11,12]. On the other hand, during drying, water loss can lead to soil shrinkage depending on its mineralogical composition, which affects its porosity, permeability, and resistance to root penetration [13]. Understanding the effects of WDC on soil structure is crucial for developing conservation strategies that optimize irrigation and select crops better suited to these conditions [14]. This knowledge is essential for effective soil management across various agricultural systems. In addition to influencing soil structural stability, WDC can also affect soil organic carbon fractions and the processes involved in soil aggregate renewal [15]. Therefore, research focused on the impact of WDC on soil structure can provide valuable insights into soil-plant-water interactions and their role in promoting the sustainability of agricultural systems [16].

In addition, analyzing physical properties is fundamental to understanding the processes in the soil pore system. The study of properties such as porosity, fractal dimension, pore number, connectivity, tortuosity, and permeability can provide important information about the structural quality of the soil [17,18,19]. Porosity refers to the ratio between the total volume of a given soil sample and the volume of its pores, and it is directly affected by changes in soil structure [20]. The fractal dimension is used to characterize the complexity or irregularity of an object. Measuring the fractal dimension of any system or porous medium is crucial as it indicates its level of heterogeneity [19]. Connectivity is related to the continuity of pores in the soil, which influences the ability of the soil to allow water infiltration, the distribution of nutrients, and the movement of gases [21]. Tortuosity measures the irregularity and sinuosity of pores, affecting water infiltration and redistribution across the soil profile [22]. Permeability refers to the ability of soil to transmit water and gases, which is primarily influenced by soil porosity and tortuosity of pores [23]. Thus, analyzing how soil management practices affect soil morphological and topological properties can provide information for selecting the best soil management strategy, conserving natural resources, and optimizing agricultural productivity.

In this sense, a technique often used to study the above-mentioned physical, morphological, and topological properties is X-Ray computed tomography (XCT) [24,25,26]. XCT allows non-destructive analysis of the internal structure of materials, which is ideal for studying porous systems such as soil. The technique allows the acquisition of high-quality images with different resolutions, permitting the study of inter-aggregate and intra-aggregate porosity. Intra-aggregate pores mainly consist of micropores, which are fundamental for water retention in soil [27]. These pores act by retaining water that, depending on the scale, is readily available to plants. Intra-aggregate pores regulate multiple micro-scale soil processes essential to the biochemical interactions that indirectly influence soil macrostructure. Such pores are indispensable to soil structure’s physical stability due to the micro-scale chemical, physical, and biological interactions. External processes, such as WDC, can also affect the biochemical and microbial activities occurring at the micro-scale that indirectly form soil aggregates and, consequently, the pore space within the aggregates [28]. Differences in soil management can affect intra-aggregate pores, which may influence processes such as water retention, microorganism movement, and soil solution conduction [29,30]. However, only a few studies have examined tomography at the intra-aggregate pore scale, particularly concerning how these pore types change during WDC. Therefore, understanding the role of these pore types in soil processes is essential for improving our knowledge of water retention and movement within small soil aggregates [11,12,31].

A series of studies have examined the effects of different tillage systems on soil pore properties, focusing on the same Oxisol areas in Brazil but at various scales of analysis [29,32,33]. The study of Ferreira et al. [29] specifically investigated the effects of management systems on intra-aggregate pores, utilizing synchrotron-based X-Ray computed tomography to achieve the high spatial resolution required. In contrast, the present study advances this research by investigating how the same soil under contrasting management is affected by WDC at the scale of intra-aggregate pores. Thus, this study will test the following hypotheses: (1) the morphological and hydraulic properties of intra-aggregate pores in a subtropical Oxisol remain structurally stable when subjected to WDC, and (2) different soil management systems can influence the impact of WDC on the soil pore network.

2. Materials and Methods

2.1. Soil Sampling

Undisturbed samples were collected in 2018 from the topsoil layer (0–10 cm) of a Rhodic Hapludox [34] located in experimental sites (25°16′ S, 50°15′ W, 875 m a.s.l) close to Ponta Grossa, Brazil. The soil has a clay texture composed of 58% clay, 29% silt, and 13% sand [27]. For this study, the soil was submitted to three long-term contrasting soil managements: NT, CT, and F. Samples were collected in the crop row to avoid possible compacted regions due to agricultural traffic [35]. More information on experimental sites and soil preparation can be found in [33].

The soil sampling was carried out before soil preparation for new crop production and near its field capacity to avoid damage to soil structure due to sampling [36]. Part of the collected samples were subjected to 12 WDC, and others were not (0 WDC). Samples that did not go through WDC were analyzed in [29] and are used in the present study as a reference to evaluate the effects of WDCs on the architecture of soil intra-aggregate pores. Samples not subjected to WDC were collected in monoliths [29], while those subjected to 12 WDC were collected in volumetric rings, necessary to perform WDC by capillary rise as described in Section 2.2. Following sampling, the samples destined for the 12 WDC had the soil excess outside the volumetric rings trimmed off. This procedure was made to ensure a soil volume similar to the internal volume of the cylinder. It is worth noting that although the collection methods were different, the aggregates selected for tomographic analysis were much smaller (2–4 mm) than the volume of the collected samples, even for the volumetric rings. Furthermore, the aggregates selected for tomography were obtained after sieving the innermost portions of the samples, especially those selected in the volumetric rings. This procedure was performed primarily to mitigate the effects of the walls during soil collection in the cylinders. Therefore, the fact that the samples under 0 W-D were collected in monoliths (volumes between 50 and 100 cm³) and 12 W-D (5 cm height × 5 cm diameter) did not influence the results obtained.

2.2. Wetting-Drying Cycles and Sample Preparation

Following sample preparation, five samples of each soil management were subjected to WDC. A permeable cloth was fixed with rubber rings to the base of the samples to allow water permeability but retain soil particles. In the wetting (W) procedure, samples were saturated by capillary rise. This procedure was carried out slowly by adding around 0.5 cm of water, every two hours, to a tray containing the samples until reaching a water level at half of the sample’s height. After 24 h, another 0.5 cm of water was added to around 80 to 90% cylinder height. The saturation procedure lasted approximately 48 h. The drying (D) procedure was made with the samples placed in a suction table (EijKelkamp—model 08.01 SandBox) under a pressure head (h) of 6 kPa. This pressure head means a diameter up to c. 50 µm (Young-Laplace equation) from the samples are drained [27]. This pore size was chosen because it represents the size that separates the storage pores from the transmission ones [37]. Soil samples were kept in the suction table for approximately 3 to 5 days, when the water ceased to flow out of them (thermodynamic hydraulic equilibrium). Following the equilibrium, the samples were subjected again to a new wetting.

By the end of the 12 WDC, the samples were oven dried for around two weeks at 40 °C. Next, they were carefully extracted from the volumetric rings, disaggregated, and passed through sieves with mesh sizes between 2 and 4 mm, as performed for samples under 0 WDC [29]. One small aggregate (size of c. 3 mm) was selected from each soil sample under the contrasting soil managements, taking into account both 0 and 12 WDC. Thus, thirty samples were scanned by computed tomography (5 aggregates × 2 levels of WDC × 3 soil managements). It’s important to recognize that the high cost of analyses and the significant computational requirements for accurately assessing soil morphological properties through microtomography make it impractical to handle a large number of samples in our study. Factors such as the limited access to tomography equipment, the expenses associated with analysis, the need for robust computational infrastructure, and the large volume of data generated typically result in a low number of samples being scanned in many studies within this field [8,11,12,13,25,26].

2.3. Synchrotron-Based X-Ray Computed Tomography: Image Acquisition and Segmentation

Each soil aggregate was scanned using synchrotron X-Ray computed tomography at the Brazilian Synchrotron Light Source Facility (LNLS/CNPEM) using the IMX Beamline (Campinas, Brazil). A set of 16-bit images (raw type) with a maximum array of 2048 × 2048 × 2048 voxels was acquired. Thus, each image had a volume of 3.36 × 3.36 × 3.36 mm³ with a voxel side length of 1.64 µm. The image segmentation was based on machine learning combined with traditional local segmentation. The steps utilizing machine learning were carried out through the software Annotat3D 2.0, which was developed by the LNLS/CNPEM team [38]. Details of image acquisition and segmentation can be found in [29,39].

2.4. Morphological Analysis

The image-based porosity (P) was calculated as the ratio between the volume of pores (V_p) and total sample volume (V_t) (Equation (1)). The Volume Fraction module (Avizo software 9.7) was employed to determine V_p and solids, which together give V_t in the segmented images.

$$\mathrm{P}\left(\%\right)=\left({\displaystyle \frac{{\mathrm{V}}_{\mathrm{p}}}{{\mathrm{V}}_{\mathrm{t}}}}\right)\times 100$$

(1)

The number of pores divided by volume (NP/V), defined as the total number of unconnected pores, and the volume of isolated pores, was obtained using the Label Analysis module (Avizo software). The module Fractal Dimension (Avizo software) was employed to characterize the porous system of the aggregates based on their complexity. This module provides numbers between 2 and 3, with numbers close to 2 representing two-dimensional surfaces [19]. For a 3D image, the fractal dimension is an effective indicator to measure and compare the irregularity of a surface. In the case of porous system segmented images, this surface can be understood as the pore walls. The less regular the surface (which characterizes a more complex porous system), the higher the fractal dimension [19].

The pore connectivity (Euler Number, EN) and global tortuosity (τ) were determined by applying the modules Euler 3D and Centroid Path Tortuosity (Avizo software) on the entire pore system of each aggregate. Concerning tortuosity, a rectilinear pore has a value of 1, while tortuous pores are characterized by values larger than 1 [29]. The Euler number indicates how connected a pore is to other pores: the smaller (more negative) the Euler number, the greater the connectivity [40].

The permeability (k) was evaluated following the image-based parametrization of the Kozeny-Carman equation (IBP-KC) (Equation (2)), described in detail in [29,39]. For that, each sample’s most voluminous pore (MVP) was selected using the Avizo software function Sieve Analysis (Figure 1).

$$\mathrm{k}={\displaystyle \frac{{{\mathrm{P}}_{\mathrm{M}\mathrm{V}\mathrm{P}}\mathrm{R}}^2}{\mathrm{\beta }{{\mathrm{\tau }}_{\mathrm{M}\mathrm{V}\mathrm{P}}}^2}}={\displaystyle \frac{{\mathrm{P}}_{\mathrm{M}\mathrm{V}\mathrm{P}}{\left(\frac{{\mathrm{V}}_{\mathrm{M}\mathrm{V}\mathrm{P}}}{{\mathrm{A}}_{\mathrm{M}\mathrm{V}\mathrm{P}}}\right)}^2}{\mathrm{\beta }{{\mathrm{\tau }}_{\mathrm{M}\mathrm{V}\mathrm{P}}}^2}}$$

(2)

where P_MVP is the porosity of a sub volume of the aggregate (Figure 1), which includes the MVP, that is, the ratio between the volume of the MVP (V_MVP) and the sub volume of the aggregate that contains the MVP (V_sb). Note that for continuous percolation to occur, the orthogonal boundaries of V_sb must intersect the MVP. This is inherent to the method. Thus, the MVP is used because permeability can only be measured using this method when the pore percolates the V_sb limits. R is the hydraulic radius (µm), β is a dimensionless shape factor that depends on the geometry of the porous space, and τ_MVP is the tortuosity of the MVP.

The R factor was obtained from the ratio between the volume of the MVP (V_MVP) and its internal surface area (A_MVP) [39], which was determined using the Label Analysis function in Avizo. As the cross-sections obtained were irregular, the β-factor was taken as 2.5, considering a value of two for flow through a small capillary tube and three for flow over an infinite flat plate [41]. The τ_MVP was determined via the Centerline Tree function in the Avizo software by means of the skeletonization technique. The tube parameters (slope and zeroVal) were set to 5 and 4, respectively, and the number of parts (segments) was set to 70, following the same methodology used in [29,39]. This method delivers a skeleton representing the 3D pore structure and computes the tortuosity and radius for each segment of the skeleton. Therefore, the tortuosity of each segment was obtained by the ratio between the sinuous length between the ends of the segment (Le) and the length of a straight line connecting the ends of that segment (L). The mean tortuosity of all segments (τ_MVP) was applied in Equation (2) to estimate k for each aggregate. The criterion of convergence between the radius R and an alternative mean hydraulic radius R’, proposed by [29], was adopted to identify outliers, i.e., samples for which the IBP-KC method was not suitable; where R’ is computed as the ratio between each segment’s volume and superficial area (2π × r × Le, r representing the mean segment radius).

In turn, the saturated hydraulic conductivity (K) was obtained based on the permeability data [42] according to Equation (3):

$$\mathrm{K}={\displaystyle \frac{\mathrm{k}\mathrm{\rho }\mathrm{g}}{\mathrm{\mu }}}$$

(3)

where ρ, g, and μ are the water density (0.998 g cm⁻³), the acceleration of gravity (978.7 cm s⁻²), and the water dynamic viscosity (1.005 × 10⁻² g cm⁻¹ s⁻¹), respectively.

The fraction of the aggregate occupied by the MVP (FA_MVP) was determined as the ratio between V_MVP and V_t. The Euler Number was also evaluated for the MVP. The variability among samples from different soil managements and WDC was analyzed by distributions of segment radius and tortuosity.

The statistical analysis was conducted using the PAST3 (PAleontological Statistics) software. We evaluated the differences among soil management systems using a one-way ANOVA (p ≤ 0.05), followed by Tukey’s HSD post hoc tests. To compare the differences between the WDCs, we utilized a t-test (p ≤ 0.05). We also checked for normality and homogeneity of the data after diagnosing the residuals. When necessary, data transformation was performed. The means and standard deviations were also calculated to illustrate the differences between management systems.

In Figure 2, we present a schematic diagram of the procedures performed in our study.

3. Results and Discussion

The results presented in this study aim to compare the effect of WDC on the morphological and hydraulic properties of aggregates in a subtropical soil. Although we applied statistical analysis between soil managements, the focus of the study was not to compare pore architecture between soil managements. A comparison between these different soil managements has already been studied in detail by [29]. However, we chose to apply statistics between uses before and after WDC to verify the effect of WDC on soil structure between uses, especially after 12 WDC. In this sense, we only observed specific differences for some of the properties analyzed, as shown next.

3.1. Morphological Analysis Considering All Pores in Each Aggregate

The imaged porosity did not present significant differences among soil managements before the WDC application (Figure 3a), which is mainly associated with the high P variability for the samples under F. However, 12 WDC significantly decreased P by c. 31% (CT) and c. 28% (NT) compared to F (p ≤ 0.05), while no differences were found between CT and NT. The average values of imaged porosity showed the following trend after 12 WDC among soil managements (F > CT = NT). Interestingly, the variability in P among samples was reduced after continuous wetting and drying in all soil managements. When the effect of WDC is analyzed for each soil management, only minor variations were observed (p > 0.05). Although the samples under F decreased P by c. 46%, the high variability in porosity by imaging, especially for the 0 WDC samples, influenced the results. In the case of the samples under CT, the decrease in P was c. 22%, whereas, for the samples under NT, it was only c. 4%.

Other authors have observed that WDCs often favor soil aggregation by altering P, especially in clayey soils [43,44]. The action of capillary forces inside the pores in the aggregates has also been a factor in altering soil structure after WDC [45]. In our study, although there were no significant differences from each factor alone (soil management type and number of WDC), interesting associated effects were identified, i.e., F samples under continuous WDC become more porous than CT and NT samples under the same conditions. The higher P for the area under F is mainly associated with decomposing organic matter and the action of tree roots, which also explains the greater P variability for 0 WDC [29,46]. The soil under NT showed greater stability in terms of P variability with WDC, which would be expected due to the lack of soil disturbance and the maintenance of vegetation cover and crop residues on the soil surface, which increases aggregate stability [8,47]. Regarding CT [48], lower stability was found in soil aggregates of varying sizes under this soil management when compared to soils with higher organic matter content, such as NT and F, which is responsible for increasing cohesion between particles. Therefore, even slow hydration during WDC could significantly reduce soil porosity. Our results show, however, that the CT samples were resistant to P changes due to WDC, which was somewhat unexpected.

The 3D porous system images can be qualitatively assessed in Figure 4, where white regions are attributed to the absence of pores; that is, a higher appearance of white regions through the blue pore structure represents less porous samples. Although samples F0-4 and F0-5 were responsible for the high deviation found in the P results, qualitatively, only F0-4 presented a noticeable difference when compared to other samples from F and CT (for 0 and 12 WDC). Samples under NT visually looked less porous compared to CT and F, both for 0 and 12 WDC, which agrees with the quantitative results. Regarding the effects of WDC, qualitative analysis proves difficult, since the samples are quite similar between the same soil managements.

The NP/V followed the same pattern of P for samples not submitted to WDC (Figure 3b). This finding indicates a close relation between P and NP/V for the different soil managements investigated for 0 WDC. Only minor variations (p > 0.05) were observed in NP/V for both 0 and 12 WDC when the contrasting soil managements were compared (CT and NT decreased c. 6% and c. 25% compared to F after 0 WDC and decreased by c. 17% (CT) and c. 29% (NT) compared to F after 12 WDC). The analysis of the WDC effect for each of the contrasting soil management shows that none presented significant differences (p ≤ 0.05). Samples under F and NT increased by c. 7% and 1% respectively, whereas samples under CT decreased by c. 6%. For F, the increase in NP/V is an indication of the presence of smaller and more isolated pores after the action of the WDC [11]. In the case of the CT, the decrease in P was accompanied by a decrease in NP/V. For NT, the aggregates showed practically the same NP/V values after the WDC, showing that NP/V, like P, is not very sensitive to changes after 12 WDC in this pore size range (R between 0.5 and 25 µm). Other authors have demonstrated that WDC can cause changes in the number of isolated pores, but generally, when larger pores are analyzed [49,50]. Thus, our results show that at the intra-aggregate pore scale, NP/V is not influenced by WDC in each soil management, nor between soil managements associated with WDC.

The FD has proven to be an essential property for verifying the complexity and heterogeneity of porous systems [19,51]. The FD values identified in our study align with findings from the scientific literature regarding the 3D analysis of soil pores [52,53,54]. When comparing different management systems, samples exhibiting the highest FD values also showed the p values, indicating a relationship between these two micromorphological properties, as noted in other studies [24,53,55]. The higher FD values in areas under F are linked to the intricate network of roots, wormholes, and reduced human interference, which contributes to a more complex and heterogeneous pore structure [19,46].

The FD did not present significant differences (p > 0.05) among soil managements when samples were not submitted to WDC (Figure 3c). Samples under CT and NT presented FD values c. 2% and c. 6% smaller than F, considering 0 WDC (p > 0.05 in both cases). For 12 WDC, the differences were c. 1% (CT) and c. 5% (NT) (p ≤ 0.05 in the latter). When the effect of WDC for each of the soil managements is analyzed, only minor variations were observed for the samples submitted to continuous wetting and drying cycles (p > 0.05). This finding means that the application of WDC alone does not influence the complexity of the soil pore system from each soil management, at the scale of intra-aggregate pores. However, associated effects were identified, i.e., the pore system of NT samples under continuous WDC becomes less complex than that of CT and F samples under the same conditions. Compared to F, the rationale may be that NT has less natural heterogeneity than F, possibly having fewer modifications/ramifications in the fluid path at the intra-aggregate scale. Compared to CT, NT undergoes less disturbance, i.e., the aggregate breakdown during soil tillage in CT may be associated with a higher complexity of the pore system, particularly confirmed by the significant difference between these two soil managements after 12 WDC [51].

Samples from 0 WDC did not exhibit differences (p > 0.05) in pore τ for the different soil managements (Figure 3d). This finding is mainly related to the high variability of τ under F, similar to the result of P. On the other hand, the 12 WDC application affected τ when the contrasting soil managements were analyzed. For instance, significant differences were noticed between F and NT and between CT and NT (p ≤ 0.05). For samples under 0 WDC, CT and NT presented c. 2% and c. 35% larger values of τ compared to F (p > 0.05 in both cases). For 12 WDC, these differences were even more considerable, being c. 17% (CT) and c. 51% (NT) compared to F (p ≤ 0.05 in the latter). The reduced variability caused by WDC, i.e., lower deviations among samples under 12 WDC, certainly contributed to revealing significant differences. When the effect of WDC on pore τ is analyzed for each of the soil managements, only minor variations were observed (p > 0.05) (Figure 3d). With the WDC application, the pore tortuosity decreased by c. 2% for samples under F, while for samples under CT and NT, it increased by c. 13% and c. 10%, respectively. Therefore, although there were no significant differences from each factor alone (soil management and number of WDC), associated effects were observed, i.e., the pore system of samples from NT under continuous WDC became more tortuous than CT and F samples under the same conditions.

Interestingly, there seems to be an opposite correlation between FD (F > CT > NT) and τ (F < CT < NT), for both 0 and 12 WDC, i.e., more tortuous pores seem to be associated with a lower complexity. Few studies compare FD and τ, which do not present a linear relationship. When compared, P and FD follow the same behavior; that is, FD increases with the increase in P [56]. On the other hand, P and τ show an inverse relationship; that is, when P increases, τ decreases [57,58]. Although the inverse relationship between FD and τ found in our study is counterintuitive, it is worth mentioning that a more complex structure (larger FD) means greater filling of a given space by this structure [59]. In addition, highly complex tortuous branches can form this structure.

Our tortuosity calculation takes into account the center of mass of the image. As a result, even for a pore structure characterized as complex, tortuous pores might show lower tortuosity when assessed as a whole. It is important to note that the FD calculated using the box-counting method also evaluates the structure globally. The FD can be understood as a measure of how much space a structure occupies as it is examined at progressively smaller scales [60]. Therefore, the FD values obtained through the box-counting method might not be the most effective for identifying complexities in specific parts of the analyzed structure. Tools such as multifractal analysis could yield more insightful results to compare the relation between properties such as FD and τ [51].

Moreover, the more similar τ results for the aggregates under F with 0 and 12 WDC indicate a higher structural stability to changes in tortuosity of pores in the soil management with a more complex porous system, given its higher FD (Figure 3c). Regarding CT, the τ results differed from those found in the literature, in which the tortuosity was greater for CT than in soils with a greater amount of organic matter [61]. For NT, the typically higher organic content may favor changes in the volume of aggregates due to coalescence affecting the smallest pores [62], which may favor an increase in tortuosity. It is also worth emphasizing that the τ values found in our study (Figure 3d), for intra-aggregate pores, are higher than those found in previous studies of the same soil areas when samples contain inter-aggregate pores [32,33]. It should be noted, though, that this is not a fair comparison as the referenced studies analyzed tortuosity with different methods [29]. On the other hand, applying the same method used here for the intra-aggregate pores of another studied soil. These authors found values varying in a similar range to that from our results in Figure 3d.

The EN was not significantly affected by contrasting soil managements for both 0 and 12 WDC, nor by WDC for each of the contrasting soil management (p > 0.05) (Figure 3e). Again, the large variability in EN values, especially for samples from F with 0 WDC, may explain the lack of significant difference between soil managements. Among samples not submitted to WDC, although no significant differences were observed, the lowest average EN (higher connectivity) was found for F, followed by CT (c. 26%) and NT (c. 31%). On the other hand, the 12 WDC application made samples under NT present the lowest EN (highest connectivity), followed by F (c. 6%) and CT (c. 17%). The 12 WDC increased EN (decreased connectivity) for samples under F (c. 54%) and CT (c. 33%), while keeping EN stable for NT (c. 1% increase). During wetting and drying, water can carry loose particles or sub-unit aggregates from the process of soil breakdown or homogenization and lead to pore clogging, which can explain the increase in EN (decrease in connectivity) for CT between 0 and 12 WDC. This shows that the absence of tillage, the maintenance of vegetation cover, the organic matter content at the soil surface, and the plant root system under continuous WDC in NT tend to promote a similar intra-aggregate connectivity compared to F, and slightly more connected intra-aggregate pore structure compared to CT, thus favoring the water infiltration and nutrient transport [25,63]. Ultimately, contrary to what would be expected according to [11,53], the 12 WDC in our study did not favor new connections between pores that would reflect in an increase in EN for all the soil managements.

3.2. Permeability and Hydraulic Conductivity Considering the Most Voluminous Pore (MVP) in Each Aggregate

The WDC did not significantly influence the P_MVP between soil managements for both 0 WDC and 12 WDC (p > 0.05). As discussed in [29], the P_MVP (Figure 5a) followed the same behavior as the aggregate porosity, considering the different soil managements under 0 WDC, F > CT > NT (Figure 3a). On the other hand, when the samples were subjected to 12 WDC, there was a higher P_MVP value for NT compared to CT (p > 0.05). When comparing the cycles between the same soil managements, no significant differences were found (p > 0.05), although there was a reduction in P_MVP of c. 57% in F and c. 50% in CT and an increase of c. 10% in NT. Interestingly, according to [29,64], the clay content among the contrasting soil managements in the present study varied in a narrow interval (53–61%), following the sequence: CT > F > NT. Therefore, we hypothesize that CT and F were more prone to the transport of finer material (clay particles) during soil wetting and drying, leading to a reduction in P_MVP due to pore clogging at the intra-aggregate pore scale [18].

Regarding FA_MVP (Figure 5b), for the 0 WDC samples, NT significantly differed from F and CT (p ≤ 0.05) [29]. For 12 WDC samples, though, no significant differences were found concerning FA_MVP between soil managements, that is, this parameter became more similar among soil managements after 12 WDC. The cycles did not affect FA_MVP within the same soil management (p > 0.05). This result supports the hypothesis of the porous system’s structural stability at the intra-aggregate scale, particularly the MVP. A significant variation in FA_MVP would indicate a change in the MVP’s size relative to the aggregate. Notably, the MVP is the dominant pore for water conduction within the aggregate [29]. Furthermore, the fact that there is no difference between FA_MVP means that there are no differences in the contributions of the MVP for each management after the cycles, thus mitigating possible discrepancies that could be found when using the MVP to estimate hydraulic permeability. However, similar to P_MVP, there was a downward trend in FA_MVP when cycles were applied for F and CT of c. 33% and c. 40%, respectively, and an increase for NT of c. 33%.

The Euler number of the MVP (EN_MVP) showed no significant differences between cycles and soil managements (p > 0.05), but followed the sequence NT > CT > F for both 0 WDC and 12 WDC, where the higher the EN_MVP, the lower the MVP connectivity (Figure 5c). Although not significant, the cycles provided an increase of EN_MVP of c. 84% and c. 35% (decrease of connectivity) in F and CT, respectively, and a decrease of c. 68% (increase of connectivity) in NT, reproducing the same pattern observed for EN considering all pores (Figure 3e). A direct relation between porosity (both P and P_MVP) and pore connectivity was observed, i.e., the decrease in P and P_MVP when comparing 0 and 12 WDC was followed by a reduction of pore connectivity (increase of EN and EN_MVP) for F and CT; and vice versa for NT (Figure 3a,e and Figure 5a,c). This finding indicates that, at this level of analysis, the pore connectivity is tied to changes in porosity, in accordance to [26,29,65] It is note-worthy that MVP may stem from the joining of several other smaller pores [66], which can be reduced when the aggregate is subjected to WDC, resulting in more isolated and less connected pores [50]. As noticed in our study [67], this corroborates that WDC can curtail the fraction of conduction pores affecting the pore connectivity and tortuosity.

For the determination of permeability, we considered the tortuosity of the MVP (τ_MVP) computed based on a local analysis using the skeletonization procedure (Equation (2)). This decision followed the same rationale adopted in [29] to enable a fair comparison considering the effects of WDC on samples from different soil managements. The local analysis-based τ_MVP, depicted in Figure 5d, followed the behavior CT > F > NT for 0 and 12 WDC. Samples under NT showed a significant difference relative to CT (p ≤ 0.05) only for 0 WDC, previously discussed in Ferreira et al. [29]. For 12 WDC, no significant differences were found concerning the τ_MVP between soil managements; that is, similarly to FA_MVP, this parameter became more similar among soil managements after 12 WDC. With the application of the cycles, the τ_MVP was reduced (p > 0.05) in F and CT by only c. 3% and c. 4%, respectively, and slightly increased by c. 2% for NT.

However, the behavior of τ_MVP (Figure 5d) differs from that observed for τ (Figure 3d). That is, while NT presented the highest τ values, CT presented the highest τ_MVP values. Therefore, when analyzing the largest pore in each aggregate, which is the pore that dominates fluid flow at this scale, we found that CT presents a more tortuous pore system, which is crucial information for assessing permeability and hydraulic conductivity. Concerning the effects of WDC, one could say from the tortuosity analyses that F is more structurally stable to changes of both τ and τ_MVP, since the overlap in deviations is more clearly defined for F than for CT and NT (Figure 5d). However, we consider that the τ_MVP is a more detailed approach for considering the local variations of the tortuosity.

The hydraulic radius of the MVP (R) showed no significant differences (p > 0.05) between soil managements and WDC (Figure 5e). The range of R assessed for all soil managements falls within the classification of storage pores (i.e., R between 0.5 and 25 µm) [37], as corroborated by the frequency distribution of the mean segment radius of the MVP. Storage pores hold water against gravity, so as not to drain readily, and contribute to supplying water to plants [68,69]. At the intra-aggregate scale, although no significant differences were observed, CT presented smaller R, and therefore, pores that hold water more strongly at the considered scale than the other soil managements. When analyzing the effect of the cycles within the same soil management, only minor differences (p > 0.05) were observed between the R values. There were increases of c. 7% and c. 11% for the samples under F and NT, respectively, from 0 to 12 WDC, while it remained stable for CT. It is important to note that the criterion for identifying outlier samples involved calculating the ratio between R and R’ [29]. If this ratio exceeded 35%, the sample was excluded from the analysis. As a result, only the samples F12-R4 and NT12-R5 were disregarded in the evaluation of k and K due to the significant divergence between R and R’ (Figure 6). This cutoff value was chosen based on statistical criteria that took into account the variability of the pore network in the analyzed samples.

The k of the MVP (Figure 5f) showed a pattern similar to R and P_MVP and opposite to τ_MVP, i.e., with F > NT > CT, as we expected based on Equation (2). Here, it should be emphasized that if the global analysis of tortuosity (τ) considering all pores of each aggregate from Figure 3d was used, the permeability analysis would be biased. Currently, we have observed that there is no trustworthy image-based procedure available to evaluate permeability without focusing on the dominant pore. As a result, caution is necessary to avoid misinterpreting the data. While computational fluid dynamics simulations can be employed for this evaluation, they tend to select a continuous path through the imaged volume, which typically corresponds to the dominant pore [70]. It is important to note that if the flow and mass volume proportions (FA_MVPs) are significantly different from the MVP, this can result in inaccurate comparisons. However, this was not an issue in our study, as illustrated in Figure 5b. Additionally, it should be noted that this method can sometimes overestimate permeability compared to other techniques [39,71]; however, such a comparison was not the goal of our research.

The K of the MVP (Figure 5g) followed the same pattern as k, as these two properties are directly related, as shown in Equation (3). A significant difference in K (p ≤ 0.05) only occurred between CT relative to F, considering 0 WDC [29]. Interestingly, K was linked to the connectivity of MVP, i.e., an increase in EN_MVP (decrease of connectivity) is associated with a decrease in K between 0 and 12 WDC, for F and CT, and vice versa for NT, corroborating the findings of An et al. [65].

Although no significant differences were found when the effects of the WDC within the same soil management were analyzed (p > 0.05), we noticed that K decreased for F (c. 14%) and CT (c. 50%) and increased for NT (c. 42%). Similar behaviors for conventional and conservation tillage systems were observed by Horn [72] over about eight years. In the same study, for the CT, the decrease in K was accompanied by a reduction in the internal soil strength and cohesion, which are associated with the homogenization of the pore system in CT during soil breakdown. On the other hand, for the conservation tillage, the increase in K was accompanied by an increase in soil strength and cohesion [72]. While the pore system in NT has experienced less disturbance than in CT over the years, it still undergoes particle rearrangement. Consequently, our results do not allow us to clearly determine the difference in permeability efficiency between the managements after WDCs. This highlights the necessity for further analysis to draw reliable conclusions.

For a more detailed assessment of the variability of the soil pore system at the scale of the aggregates studied, pore size distributions were analyzed concerning the mean segment radius (Figure 7) and segment tortuosity (Figure 8) determined by the skeletonization analysis of the MVP. In the case of samples from F with 0 WDC, the most substantial variability for the mean segment radius between samples was found up to c. 40 µm. The application of 12 WDC tended to reduce the variability between samples from F, but it occurred until c. 50 µm (Figure 7a,b). For the samples under CT (Figure 7c,d) and NT (Figure 7e,f), minor variability was observed between the samples, which mostly occurred for the mean segment radius up to c. 40 µm, for both 0 and 12 WDC. Therefore, although intra-aggregate pores are in general structurally stable to changes [29], our results show that samples from F seem to be the least stable to modifications in the MVP radii caused by the application of WDC. This result is interesting because earlier in the text, we discussed how F was the more stable soil management, among those studied here, in response to changes in τ and τ_MVP due to WDC (Figure 3d and Figure 5d). Integrating these observations, intra-aggregate pores of F samples tend to be more stable to changes of tortuosity (from both global and local analyses) and, contrastingly, more prone to changes in radii of the MVP (from a local analysis) than CT and NT due to application of WDC. In addition, we also discussed earlier that our results seem to show that FD and τ are oppositely related. Therefore, one could also say that intra-aggregate pores of F samples tend to be more complex and less tortuous (based on global analyses) than CT and NT. Interestingly, the frequency curve of average segment radius is steeper in CT than F and NT, in both 0 and 12 WDC, showing the prevalence of smaller pore segments in CT (Figure 7), which is in accordance with the smaller hydraulic radius found for CT in both 0 and 12 WDC (Figure 5e).

Regarding the frequency distribution for τ of MVP skeleton segments (Figure 8), overall small variability between samples with a slight tendency for a bimodal distribution was found for F under 0 and 12 WDC (Figure 8a,b). For both 0 and 12 WDC, the most common value of τ was c. 2, and it should be recalled that the mean values from these distributions are those presented in Figure 5d. In the case of the samples under CT (Figure 8c,d), the bimodality of the τ distribution was more evident among the samples, especially for the 0 WDC [29]. The most frequent value for the first peak was close to 2, while for the second peak it was close to 4. For CT, a more substantial variability between samples was observed close to the second peak, although with a lower frequency with respect to the first peak. The fact that the MVP shows two or more peaks for τ (per skeleton segment) indicates that the MVP in aggregates from this soil management has more tortuous segments, as reflected in the τ_MVP (Figure 5d). In the case of the samples under NT (Figure 8e,f), the bimodal character in the distribution of τ was the least pronounced among soil managements. The distributions for these samples were more similar to those under F (Figure 8a,b). As observed for the F samples, the most common value of τ for NT was around 2. The broader distribution of τ values for the samples under CT (Figure 8c,d) indicates that the porous system of the MVP tends to hinder the flow of solutes [73]. This result agrees with the lower K values (Figure 5g) observed for CT [29]. In another study [74], lower conductivities were also found in more tortuous porous media, in accordance with our results. According to these authors, the decrease in connections between pores tends to increase τ, directly influencing K. This influence between connectivity and K for the MVP of intra-aggregate pores was also observed in our studies, as previously reported.

4. Conclusions

The results of our study confirm our first hypothesis that WDC do not affect the intra-aggregate pore network, even when different management systems are analyzed, which negates our second hypothesis. Several properties were investigated at the intra-aggregate scale, and none presented significant differences between 0 and 12 WDC for each soil management type, among them: imaged porosity, number of pores, fractal dimension, tortuosity (both global for all pores and local for MVP), connectivity, fraction of the aggregate occupied by the MVP, hydraulic radius, permeability, and hydraulic conductivity. However, some interesting associated effects were found, i.e., comparing soil managements after 12 WDCs. For instance, under 12 WDC, F samples became more porous than CT and NT samples; the pore system of NT samples became less complex (FD) and more tortuous (from a global analysis considering all pores) than CT and F samples. On the other hand, some parameters that had presented significant differences between soil managements under 0 WDC became not significant under 12 WDC, for instance, the fraction of the aggregate occupied by the MVP, the tortuosity of the MVP (local analysis), permeability, and hydraulic conductivity.

Although not statistically significant when samples following WDC are compared, some observations highlight the need for future studies to consider the effects of soil management systems and WDC on the soil intra-aggregate pore structure. For example, the continuous application of WDC in NT systems promotes similar intra-aggregate pore connectivity compared to F and a slightly more connected pore structure than CT. In contrast, CT exhibited a smaller hydraulic radius, which retains water more effectively at the scale considered, as well as lower permeability and hydraulic conductivity.

Specifically, while not statistically significant, permeability and hydraulic conductivity decreased by approximately 50% in CT and increased by about 42% in NT due to the application of WDC. According to existing literature, these variations may be linked to differences in soil strength. The hydraulic radius range observed in our study fell within the classification of storage pores. However, the distributions of the mean segment radius derived from the MVP skeletonization revealed radii varying up to 50 µm, with some sizes classified as transmission pores.

In this context, the F system showed the least structural stability to changes in MVP radii due to WDC application, while the distribution curves from CT indicated a dominance of smaller pores. Additionally, a broader distribution of segment tortuosity was observed in the MVP skeleton for CT. Although some results suggest that the CT soil management type is associated with less efficient fluid transport and poorer physical quality, we acknowledge the need for further investigation. This should include an increased number of samples and different soil types to better understand the impact of WDC.

Maintaining stability against changes following WDC can help preserve the physical structure of the soil. This, in turn, supports adequate porosity for infiltration and water retention, even under adverse weather conditions, while reducing compaction and erosion. It can also promote greater mechanical resistance and protection against soil loss.