X-ray Microtomography Analysis of Integrated Crop–Livestock Production’s Impact on Soil Pore Architecture

Abstract

Integrated crop–livestock production (ILP) is an interesting alternative for more sustainable soil use. However, more studies are needed to analyze the soil pore properties under ILP at the micrometer scale. Thus, this study proposes a detailed analysis of the soil pore architecture at the micrometer scale in three dimensions. For this purpose, samples of an Oxisol under ILP subjected to minimum tillage (MT) and no tillage (NT) with ryegrass as the cover crop (C) and silage (S) were studied. The micromorphological properties of the soil were analyzed via X-ray microtomography. The MT(C) system showed the highest values of porosity (c. 20.4%), connectivity (c. 32.8 × 10³), volume (c. 26%), and the number of pores (c. 32%) in a rod-like shape. However, the MT(S), NT(C), and NT(S) systems showed greater tortuosity (c. 2.2, c. 2.0, and c. 2.1) and lower pore connectivity (c. 8.3 × 10³, c. 6.9 × 10³, and c. 6.2 × 10³), especially in S use. Ellipsoidal and rod-shaped pores predominated over spheroidal and disc-shaped pores in all treatments. The results of this study show that the use of ryegrass as a cover crop improves the soil physical properties, especially in MT. For S use, the type of soil management (MT or NT) did not show any differences.

1. Introduction

In recent decades, rapid population growth has been accompanied by several extreme events, including droughts, forest fires, and floods, significantly influencing food production and land use. These pressures have directly impacted agriculture [1,2]. In light of this situation, it is of the utmost importance to direct investments toward the more rational use of natural resources, focusing on soil and water. In this context, one of the strategies by which to achieve an equilibrium between food production and environmental protection is to implement management systems that integrate conservation practices with diverse agricultural crop uses.

Intensive soil cultivation and a lack of diversity in crop production can harm the soil structure [3,4]. The soil pore system’s changes mainly affect water infiltration and water availability for plants [5,6]. In this context, it is essential to understand the pore architecture to seek more environmentally sustainable soil management practices. McBratney et al. [7] demonstrated that the intensification of agriculture aims to meet the global demand for food; however, this intensification must be carried out sustainably, respecting the environment’s limitations. Therefore, soil management practices play a crucial role in ensuring the productivity and sustainability of agricultural activities [8], since they involve a set of techniques and strategies aimed at preserving the soil fertility and quality to reduce harmful processes such as erosion and the contamination of water resources by agrochemicals [9].

In Brazil, it is common practice to adopt management techniques that are considered to be environmentally friendly, particularly the minimum tillage system (MT) and the no tillage system (NT). The former is a system that permits the planting of crops during the rainy season, the more intensive use of agricultural areas, a reduction in soil erosion, and the use of agricultural machinery, as well as greater control of weeds [10]. No-till farming maintains crop residues on the soil surface, thereby preventing erosion and enhancing the organic matter content of the surface layers [11]. Reducing tillage under no-till farming techniques preserves the soil structure, improving its capacity to retain water and nutrients [12].

Several studies have examined the impact of soil management practices on the pore structure and soil quality. Tagar et al. [13] analyzed the effects of different management methods on aggregates’ fractal dimensions and stability, demonstrating the diverse effects on the soil structure. Rusu [14] reviewed the impact of minimum tillage and no-till farming, suggesting that these methods are effective in increasing the soil fertility and water retention capacity and reducing erosion compared to conventional tillage. Specific studies, such as that by Gong et al. [15], have employed X-ray microtomography to examine the soil microstructure, demonstrating how this technique can provide precise details on the shape, tortuosity, and connectivity of soil pores, offering valuable insights into hydrological processes.

The combination of soil management practices with different agricultural crop utilization strategies has the potential to generate noteworthy results in terms of soil conservation. In this context, ryegrass (Lolium multiflorum L.), primarily utilized to produce pasture for livestock, has emerged as a highly versatile crop for the implementation of sustainable agricultural practices [16]. Ryegrass is cultivated in temperate climate regions and employed in ground cover and silage production [17]. The utilization of ryegrass as a cover crop has the combined objective of protecting the soil from erosion, suppressing weed growth, and improving water and nutrient retention, thereby contributing to the sustainability of the production system [18]. During silage production, ryegrass is harvested at different stages of development and stored for use as cattle feed [19].

In this context, integrated crop–livestock systems have emerged as promising agricultural strategies to increase agricultural productivity and sustainability [20]. Integrated livestock production (ILP) is a system that combines the production of grains and other materials that can be used to feed livestock in the form of crop rotation [21]. This technique employs various soil management practices and agricultural crop utilization strategies. The advantages of ILP include soil conservation and improved pasture quality, which can help to reduce the pressure for new production areas [22]. Consequently, implementing ILP may serve as a viable alternative to transform these areas into productive agricultural sites.

Several studies have highlighted the benefits of ILP in improving the soil quality and agricultural sustainability. Tirloni et al. [23] examined how crop–livestock integration can increase the soil organic matter and aggregate stability, contributing to soil conservation. In addition, Ambus et al. [24] concluded that moderate grazing is a more beneficial option than intense grazing. Still, they showed no significant improvements in the soil compared to the ungrazed system when evaluating physical properties such as microporosity and macroporosity. Despite these advances, there is still a significant gap in the literature regarding applying advanced techniques, such as X-ray microtomography, for a detailed analysis of the soil pore structure in ILP systems.

Thus, for a comprehensive examination of the impact of ILP on the soil structure, three-dimensional (3D) image analysis techniques, such as X-ray microtomography (Micro-CT), can be employed. This technique enables the noninvasive investigation of the soil pore architecture, facilitating the acquisition of high-resolution images. The use of 3D micro-CT images allows for the measurement of numerous morphological and geometric properties of the soil structure, thus enabling the detailed characterization of the distribution of the pores and their complexity in porous media, even under different management practices [25,26,27,28,29]. Applying research approaches such as those offered by micro-CT provides invaluable insights into soil pore systems at various scales, thereby facilitating the implementation of more efficient and sustainable management practices.

This study proposes a 3D image-based analysis of the pore system of an Oxisol under ILP at the transmission and redistribution pore scale. Two management systems (MT and NT) and two forms of ryegrass crop use (cover crop and silage) are investigated. Studies still need to address the detailed analysis of the soil pore architecture under ILP at the micrometer scale based on 3D micro-CT images. Our findings can provide details on how the different management types affect the continuity and sinuosity of the pores. Traditional soil physical analysis methods do not obtain this information in three dimensions by internal structure analysis. In addition, micro-CT allows us to analyze how the management type affects the different pore shapes. This result is fundamental to understanding how water is retained and redistributed in the soil. In this regard, the objective of this study is to provide findings that have not been previously reported, which will assist farmers in selecting agricultural systems that can enhance productivity while promoting the rational use of soil resources.

Thus, this study is based on two hypotheses: the first is that ILP management practices have different effects on the soil micromorphological properties, and the second is that using vegetation cover improves the morphological and geometric properties of the soil pore system, regardless of the type of management adopted. High-resolution three-dimensional images are obtained using X-ray microtomography to evaluate these hypotheses.

2. Materials and Methods

2.1. Study Sites

The study was conducted in an experimental demonstration field characterized by slightly undulated relief, located near the city of Castro, in the state of Paraná, Brazil, at latitude 24°47′53″ S and longitude 49°57′41″ W, with an altitude of 997 m (Figure 1). The local soil type is a Dystrophic Haplohumox, according to the Brazilian soil classification system [30] (USDA Soil Taxonomy: Oxisol), characterized by a loamy texture (

Table 1. Soil physical properties of the samples studied under minimum tillage (MT) and no tillage (NT) management practices with ryegrass as a cover crop (C) and silage (S).

Physical Properties	Soil Use Practices
	MT(C)	MT(S)	NT(C)	NT(S)
Sand (g kg−1)	354	349	342	348
Clay (g kg−1)	420	430	415	445
Silt (g kg−1)	226	221	243	207
BD (g cm−3)	1.13 (0.13)	1.22 (0.05)	1.25 (0.03)	1.22 (0.05)
TP (%)	0.54 (0.05)	0.52 (0.02)	0.51 (0.01)	0.52 (0.02)
MAC (%)	0.14 (0.07)	0.07 (0.05)	0.05 (0.01)	0.09 (0.03)
MIC (%)	0.40 (0.03)	0.45 (0.03)	0.46 (0.01)	0.43 (0.01)
OC (g kg−1)	26 (2)	25 (2)	24 (2)	27 (3)

Sand, clay, and silt content were measured using the densimeter method; BD, bulk density, was measured using the volumetric ring method; TP, total porosity, was measured as the difference between the soil bulk density and the particle density; MAC, macroporosity, was measured as the difference between TP and microporosity; MIC, microporosity, was measured considering the volume of water retained at a pressure of 6 kPa; OC, organic carbon, was measured using the method of oxidizing organic matter in an acidic medium with potassium dichromate (K₂Cr₂O₇). Numbers between parentheses represent the standard deviation of the mean (n = 5).

). The region’s climate is mesothermal with no defined dry season, classified as Cfb, according to the Köppen climate classification [31]. The average annual temperature is 16.8 °C, with yearly rainfall between 1600 and 1800 mm.

Until the 1960s, the experimental area was under native forest. After the forest was cleared, the area was converted into an agricultural area with two types of crop grown yearly: wheat (Triticum aestivum) in the winter and soybean (Glycine max) in the summer. The experiment began in March 2004, when 5.36 Mg ha⁻¹ of lime was applied. In 2009, a new application of 8.36 Mg ha⁻¹ of lime was used to correct the soil’s acidity. The method of application varied according to the type of soil management adopted. Since the experiment began, an annual succession of ryegrass (Lolium multiflorum L.) and corn (Zea mays L.) has been used.

The experiment was carried out in a randomized complete block design divided into bands, with the treatments subdivided into plots and four replications. The bands had an area of 1200 m² (120 × 10 m), while each sub-plot had an area of 100 m² (10 × 10 m). All plots received pH correction and chemical fertilization procedures [32]. The soil preparation systems in the bands studied were (i) minimum tillage (MT), where only one harrowing at 0.10 m with a leveling harrow was carried out, and (ii) no tillage (NT), with sowing without turning the soil and plant remains kept on the surface. The uses of the ryegrass crop in the sub-plots involved (i) a cover crop (C) and (ii) pre-dried silage (S). In all treatments, ryegrass and corn were sown in the last fortnight of May and the first fortnight of October. Annually, 20 kg ha⁻¹ corn seeds were sown at seeding depths up to 0.03 m, with 0.80 m spacing between rows.

Nitrogen fertilization was conducted on corn crops with an application rate of 165 kg ha⁻¹ on installment. The initial application was 40 kg ha⁻¹ of 15–30–0 N–P₂O₅–K₂O at sowing, followed by another application of 25–00–25 N–P₂O₅–K₂O at the V4 stage of the crop. Phosphorus and potassium were applied to the planting furrow at 80 kg ha⁻¹ P₂O₅ and 125 kg ha⁻¹ of K₂O. Moreover, 600 g ha⁻¹ of the herbicide nicosulfuron was applied to plants in the post-emergence phase at the crop stadium V3 to control weeds.

Corn was harvested mechanically for silage when the grains were milky or solid with starch in February. The soil was maintained in a fallow state until ryegrass sowing in the first half of May. The ryegrass sowing was conducted annually, utilizing a no-till fertilizer with a 60 kg ha⁻¹ seed dosage at a soil depth of 0.03 m and row spacing of 0.17 m. In all harvests, 200 kg ha⁻¹ of the formulated N–P₂O₅–K₂O 10–20–20 was applied as basic fertilization, and 120 kg ha⁻¹ N and 36 kg ha⁻¹ K₂O were used during the tillering phase. After the first cutting, the ryegrass silage treatment received 55 kg ha⁻¹ N to enable two cuts per crop.

When harvesting ryegrass for silage, the plants were cut at a height of approximately 0.30 m, 0.10 m above the soil surface, using a silage machine. When treating ryegrass for soil cover, plants were desiccated with glyphosate (1200 g ha⁻¹) when they reached a height of 0.30 m. The shoots and roots were then left on the soil until corn planting.

Table 1 shows the results of some of the physical properties of the soil studied under the different treatments for the 0–0.20 m layer.

2.2. Soil Sampling and Preparation

In October 2012, following the annual ryegrass phytomass management, undisturbed soil samples were collected in the crop row from the 0.05 to 0.15 m depth layers. Six pseudo-replications were sampled in the sub-plots of one of the experiment replications. This sampling aimed to reduce the spatial variability in the soil properties due to the size of the experimental area. The sampling process involved using trowels, hoes, and shovels designed for this purpose. The samples were collected three days after heavy rainfall, when the soil was at field capacity moisture. Six irregularly shaped clods of soil were obtained per treatment, yielding 24 samples (6 samples × 2 soil managements × 2 uses of ryegrass = 24 clods). The samples were randomly collected from the experimental area, one meter from the border on each side of the sub-plot. After sampling, the irregularly shaped clods with a size of more than a thousand cubic centimeters were wrapped in plastic wrap and transported in plastic boxes to the laboratory to avoid disintegration and loss of moisture during transport.

The soil samples were permitted to air-dry for several weeks until they reached a constant mass. After this, the clods were carefully trimmed to produce more regular-shaped blocks (parallelepipeds) using steel saws and spatulas. The blocks were wrapped in plastic to prevent moisture absorption following this procedure. The volumes of the blocks were adjusted according to the dimensions of the sample holders (diameters of 3 and 5 cm) that were used in the microtomography analysis. The samples exhibited a length of approximately 3 cm and a width and height of between 3 and 5 cm. Foam was employed to secure the samples within the holder to prevent movement, which could introduce image artifacts.

2.3. X-ray Computed Tomography

The soil samples were scanned via an X-ray tomography analysis facility at C-LABMU (UEPG, Ponta Grossa) using a Nikon X-ray microtomograph, model XT V 130C. The equipment settings were adjusted to a voltage of 125 kV, a current of 140 µA, an acquisition time of 250 ms per image, and eight frames. A 0.25-mm-thick copper filter was used to minimize the hardening of the X-ray beam. Scanning the samples generated 1583 projections with a pixel size of 50 μm each. After acquiring the images, 1008 two-dimensional (2D) images per sample were used to construct the 3D images (1008 × 1008 × 1008 voxels). The 2D grayscale images were saved in 16-bit format (TIFF format) for the remaining processing steps.

The following steps were carried out after image reconstruction: (i) the selection and cropping of cubic subvolumes, (ii) segmentation, and (iii) the binarization of the 2D images. All of these steps were performed using the ImageJ software (v. 1.54i) [33], considering only the selected region of interest within the scanned soil aggregate (Figure 2a). The subvolume of interest (VOI) cropping was carried out in ImageJ using the crop tool (Figure 2b). Subsequently, the cropped images were converted to 8-bit format (256 shades of gray) to facilitate the selection of the peaks referring to the contributions of air and solids in the aggregates. Filters were applied to reduce noise and improve the separation between the phases (air and solids). The first procedure involved the application of a 3D median filter (Figure 2c), with the following parameters: x = 2.0, y = 2.0, and z = 2.0. This was followed by the Unsharp Mask tool (Figure 2d), which was configured with the following parameters: radius = 1.0 pixels and mask weight = 0.9.

Next, the images were segmented using the nonparameterized Otsu method (Figure 2e), where the threshold values were set based on the gray tones of the images. The Otsu algorithm method divides pixels into classes based on a threshold chosen from the histogram: class 1, gray tones [0, t]; class 2, gray tones [t, 255]. When there is clear separation between the peaks due to the characteristics of the samples, this segmentation method yields good results. As the last stage of image processing, the remaining noise was minimized using the Remove Outliers tool (Figure 2f), configured for particles and pores with values of bright = 4.0 (for particles) and dark = 2.0 (for pores).

2.4. Soil Morphological and Geometric Properties

The porosity ($\phi$) is an indicator that describes the amount of empty space present in the soil [34]. This parameter was calculated by dividing the volume of pores ($V_{pores}$) by the total volume of the sample ($V_{sample}$):

$$\phi (\mathrm{\%})=\sum {\displaystyle \frac{V_{pores}}{V_{sample}}}\times 100$$

(1)

The imaged porosity was determined using the voxel counter function in the ImageJ software (v. 1.54i). The plugin counts the voxels in black with a threshold value within a region of interest (pores with a value of 255) in a stack of 8-bit binarized images. The image porosity is then calculated as the ratio between the voxels with a threshold value (255) and all of the voxels in the sample (0 and 255).

The fractal dimension ($FD$) is a property that characterizes the complexity and irregularity of patterns at different scales [35]. This parameter was determined using the fractal dimension plugin using the box counting technique. The $FD$ is determined by the slope of the line on the $\log n$ versus $\log 1/R$ graph, where $n$ represents the number of boxes and $R$ is the lateral length of the box:

$$FD=\underset{R\to 0}{\lim }{\displaystyle \frac{\log n}{\log \frac{1}{R}}}$$

(2)

The degree of anisotropy ($DA$) is an indicator used to measure the directionality or preferential orientation of a given property in a system. The BoneJ plugin [36] was used to calculate the $DA$ of the soil pore system. The anisotropy function was chosen. The $DA$ is based on a series of vectors with $\eta$ directions originating from random positions in 3D images that intersect the pores (black = 255):

$$DA=1-{\displaystyle \frac{IC}{IL}}$$

(3)

where $IC$ represents the average length of the shortest interception vectors and $IL$ represents the average length of the longest interception vectors. Values close to 0 indicate isotropic pore systems, while values closer to 1 indicate greater anisotropy in the orientation of the pores. When the $DA$ is equal to 0, the ratio between $IC$ and $IL$ is equal to 1 ($IC$ = $IL$), whereas, when the $DA$ is equal to 1, the ratio between $IC$ and $IL$ is equal to 0.

Pore connectivity ($C$) is a geometric measure that quantifies the number of connected structures and interconnected paths in a porous system [37]. The procedure for the determination of $C$ involved using the purify filter to remove isolated cavities and the connectivity plugin in BoneJ, which analyzes neighboring voxels to calculate the Euler number ($EN$) and determine the contribution to $C$. The pore connectivity was calculated by

$$EN=N_{ob}-C+H$$

(4)

$$C=1-EN$$

(5)

where $N_{ob}$ denotes the number of isolated objects, $C$ represents the connectivity, and $H$ corresponds to the number of completely closed cavities. The $EN$ quantity will be positive if the number of isolated pores is greater than the number of connections between the pores ($N_{ob}>C$). In a fully connected pore network, the $EN$ value will be negative ($C>>N_{ob}=1$). In this context, the $EN$ parameter counts the number of multiple connections and is associated with the number of ramifications in the pore network. It is important to note that in porous systems such as soil, the contribution of the $H$ parameter is usually negligible.

Tortuosity ($\mathrm{Ʈ}$) is a measure that assesses the sinuosity of a pore [38]. The calculation of this parameter involves the ratio between the geodesic length ($L_g$) and the Euclidean length ($L_e$), along a path of connected pores, as expressed by

$$\mathrm{Ʈ}={\displaystyle \frac{L_g}{L_e}}$$

(6)

Tortuosity was calculated using the tortuosity plugin in ImageJ, based on the geodesic reconstruction algorithm (RG) developed by Gommes et al. [39]. The analyses were carried out along three different axes ($\pm x, \pm y, \pm z$).

The pore volume ($VP$) and pore number ($NP$) were calculated based on the distribution of pore shapes determined from the ellipsoid axes plotted inside them. The particle analyser plugin in ImageJ software (v. 1.54i) was used for this analysis, following the pore classification system template proposed by Bullock et al. [40]. Four main pore shapes were considered: (i) equant—pores with an approximately spherical shape; (ii) prolate—elongated pores with a longer main axis; (iii) oblate—flattened pores with a smaller main axis and two larger perpendicular axes; and (iv) triaxial—elongated pores with an ellipsoid shape, presenting three axes of different sizes.

Table 2. Definition of the shapes of soil pores based on the ratio between the main axes of the ellipsoids. $IA$: intermediate axis; $LA$: large axis; $SA$: short axis.

Ellipsoid Axis	Shape
	Equant	Prolate	Oblate	Triaxial
$IA/LA$	≥0.65	<0.65	≥0.65	<0.65
$SA/IA$	≥0.65	≥0.65	<0.65	<0.65

shows the relationship between the ellipsoid axes for the calculation of the pore shape [41]. During the analysis, some pores were considered unclassified due to pore complexities. It was impossible to compute all or one of the axes for these pores.

Figure 3 shows a flowchart summarizing the main steps followed in this study, to facilitate an understanding of the results that will be presented next.

2.5. Statistical Analysis

The data on the morphological and geometric parameters of the soil pores were subjected to an analysis of variance (ANOVA), followed by the Tukey test at the 5% significance level. These tests were carried out to compare the means of the different treatments. Pearson’s correlation coefficients were calculated for the variables of the different management practices, the ryegrass crop uses, and the morphological and geometric properties measured. In addition, the variables studied were also analyzed using multivariate analysis. The method was applied to the mean values of six samples from each property measured. The raw data were auto-scaled before calculation. The sample scores were represented by the coordinates of the first principal component (PC1) and the second principal component (PC2), which were linearly dependent on their respective variables, represented by the loading axes. All statistical analyses were performed using the PAST (PAleontological STatistics) software, version 4.03 [42].

3. Results and Discussion

Representative 3D grayscale images of the pore systems of the samples are shown in Figure 4. We chose to select only one sample for each of the treatments studied. The MT(C) system is characterized by higher porosity and the presence of regions with connected pores in its porous system (Figure 4). In the case of MT(S), it is possible to verify a smaller frequency of pores compared to MT(C), which indicates lower porosity and the presence of connected pores, which may be associated with dead roots (Figure 4). The NT(C) (Figure 4) and NT(S) (Figure 4) systems also have a smaller volume of pores compared to MT(S). In the former, the pores are more concentrated in some samples, indicating the greater anisotropy of the pore system.

In general, the images show the existence of pores with branches and interconnections in all of the treatments studied. This finding suggests these systems’ predominant presence of elongated and interconnected pores. When the results of the qualitative analysis are compared with the data in Table 1, we observe that the sample with the lowest bulk density (MT(C)) has the largest pore volume in the 3D images. This result is consistent with the macroporosity values found for this system (Table 1). The other treatments have similar bulk density values, explaining the similarities in the pore volumes.

The $\phi$ results (Figure 5a) show that the MT(C) system differed significantly (p < 0.05) from the other treatments. The greater porosity of MT(C) compared to MT(S) suggests that the cover crops positively affected the pore soil system. This fact highlights the importance of conservation practices that maintain cover crops at the surface of the soil structure [43]. There were no significant differences (p > 0.05) between NT(C) and NT(S), although the adoption of no tillage reduced $\phi$. When soils are managed for long periods under NT, the compaction of the surface layer can occur due to the transport of agricultural equipment and animals, as in the case of crop–livestock integration [44,45].

Using ryegrass as a cover positively affects $\phi$, especially when combined with minimum tillage [46,47]. In addition, maintaining vegetative cover benefits the soil system by providing organic matter (Table 1) to the surface layers of the soil [48]. The presence of vegetative cover facilitates water and air movement, contributes to water retention, and reduces the impact of raindrops, which can seal the soil surface and increase the risk of erosion [49].

Holthusen et al. [50] demonstrated the importance of plant residues in maintaining the soil structure based on porosity measurements. The presence of cover crops, such as ryegrass, can improve the soil porosity by increasing the amount of organic matter and stimulating biological activity [46]. Earthworms and microorganisms use organic matter as a source of energy, creating biopores that are essential for water and air movement in the soil [51]. These biopores help to maintain a good, permeable soil structure, contributing to water retention and soil aeration. In addition, the roots of cover crops can also penetrate and decompress the soil, creating pathways for water infiltration and root growth [48]. Conversely, silage removes biomass from the surface, which can lead to the lower addition of organic matter [52] and, consequently, lower soil porosity. Auler et al. [32], who conducted measurements in the same experimental area as the present study, found that the intensive use of ryegrass for grazing or silage production, regardless of the planting system, negatively affects the soil structure compared to the use of cover crops. These results are consistent with those obtained in our study.

The $FD$ results were similar between the treatments analyzed (Figure 5b). The most significant variability (error bars) observed between samples was related to the sensitivity of the $FD$ to small changes in pore geometry [53,54]. It is important to note that this physical parameter provides an assessment of the complexity and degree of irregularity of the soil pore structure at different scales [55,56]. Thus, changes in the complexity of the pore geometry, especially at the micrometric scale, can be analyzed using the FD [57].

In this study, the MT treatment had greater $FD$ values for cover and silage than the NT treatment but had little effect on NT (p > 0.05). This result may be related to the greater resilience of NT to structural changes due to the lack of soil disturbance, as pointed out by Fiorini et al. [58] and Zhang et al. [59]. The physical stability of NT may contribute to the maintenance of a more stable porous structure that is less subject to significant changes in pore geometry [60]. However, in the silage process, the procedures used to manage the area can reduce $\phi$, which affects the pore complexity [32,61]. It has been observed that the growth of ryegrass for silage can increase the soil bulk density due to the passage of the silage maker [62]. This increase in bulk density can potentially contribute to flattening the pores, thereby reducing their complexity. Papadopoulos et al. [63] reported that the $FD$ is sensitive to several factors, including soil compaction and biological activity. Biological activity in the soil, influenced by management and vegetation cover such as ryegrass, can contribute to the formation of complex and irregular pores [51], increasing the $FD$.

It is important to note that complex porous structures due to the presence of vegetation cover have been verified by several authors, confirming the results of our study [64,65]. The process of soil disturbance, even if minimal, as in the case of MT, can also favor the appearance of more complex and irregular pores, as observed by Zhang et al. [66]. Moreover, as described by Balesdent et al. [67], processes involving minimal soil disturbance and the incorporation of organic matter into the soil matrix can facilitate the formation of more complex and irregular pores, as observed in MT, in contrast to NT, where structural stability can mitigate such changes.

Notably, the $FD$ values found (between 2.50 and 2.90) are consistent with other studies for 3D structures [25,53]. However, Dhaliwal and Kumar [41] and Singh et al. [68] reported slightly lower values than those presented here. Nevertheless, the authors found that management practices that include cover crops have the potential to improve the complexity of the soil pore system. In this sense, the higher $FD$ values reported in our study are evidence of complex pores, regardless of the treatment analyzed.

The degree of anisotropy (Figure 5c) reflects the arrangement of soil pores in different directions [69,70]. Our study revealed the highest $DA$ values for NT, although minor differences (p > 0.05) were observed between treatments. Similar results were reported by Polich et al. [71] in a study that combined management practices with winter cover crops. The higher $DA$ observed for NT under cover crops is associated with less soil disturbance, improved root development, and the accumulation of plant residues on the soil surface, which favors the appearance of pore clusters [58]. Garbout et al. [72] highlight that no tillage practices facilitate the formation of more oriented and connected pore systems than practices that till the soil. These practices, typically associated with the presence of soil fauna and plant remains on the soil surface, directly influence the anisotropy of the soil pore system [73].

The low $DA$ values observed in our study (≤0.25) indicate the presence of more isotropic pore structures. The stability of the microaggregates, which results in low anisotropy between the different treatments, can be explained by the formation of bridges between organic colloids and the increase in multivalent cations. This phenomenon is often observed in clay soils, as described by Oades [74]. These findings are consistent with those of previous studies in soil science [25,72,75]. As Tseng et al. [75] stated, lower $DA$ values can indicate that the soil pore network extends relatively homogeneously in all directions. This fact suggests that the capacity for water transport, aeration, and nutrient movement in the soil can occur more uniformly, which is vital from the point of view of water percolation and redistribution [76].

Compared with the other treatments, MT(C) exhibited the highest pore connectivity (Figure 5d). This parameter is related to the continuity of the pore network through connections between pores of different sizes, which are fundamental to water dynamics and the transport of nutrients and gases in the soil profile [77,78]. The $C$ values observed for the different treatments used are consistent with the findings of Ferreira et al. [79]. Castro Filho et al. [80] and Dexter [81] emphasized that management practices considered conservationist, such as minimum tillage associated with vegetation cover, tend to result in soils with a stronger structure and more stable aggregates. However, when the soil structure remains stable over the long term, better connectivity between pores is observed [82], especially considering the beneficial effects of organic matter from vegetation cover, which promotes the formation of interconnected pores [83].

The organic matter added by cover crops serves as food for soil microorganisms and earthworms, creating biopores during their activities [51]. These biopores enhance the pore connectivity, facilitating the infiltration and movement of water and air in the soil [46]. Another relevant aspect of MT is the positive effect of cover crops compared to silage. Some authors have proposed that silage may damage the pore structure due to the movement of the forage harvester [61,84,85], resulting in the densification of the soil and a reduction in its connectivity (Table 1). Concerning NT, the lower $C$ values may also be associated with the transport of agricultural equipment and animals during grazing, which reduces the pore connectivity [86,87].

Figure 6 shows the tortuosity results for the different directions ($x,y,z$) and their mean values (considering all directions).

It is fundamental to acknowledge that the $\mathrm{Ʈ}$ results are not related to the orientation of the soil pores, as the aggregates were extracted without indicating the direction. Moreover, the $\mathrm{Ʈ}$ values in this study provide an understanding of the variability of this physical parameter across different directions. Notably, properties such as tortuosity serve to describe the degree of sinuosity of the pores [88], which directly influences the dynamics of fluids and gases in the soil [89,90].

The $\mathrm{Ʈ}$ values showed minor differences (p > 0.05) between the treatments in the three directions (Figure 6a–c). However, the $\mathrm{Ʈ}$ considering all directions showed differences between MT(C) and the other treatments (Figure 6d), except for NT(C). When the different directions were analyzed, NT(S) and MT(S) showed the highest average $\mathrm{Ʈ}x$ and $\mathrm{Ʈ}z$ values (Figure 6a,c). On the other hand, for the y-axis (Figure 6b), the MT(S) and NT(C) systems were found to have the highest $\mathrm{Ʈ}$ values. When $\mathrm{Ʈ}$ was analyzed (Figure 6d), the MT(S) and NT(S) treatments had the highest average values, with the lowest being observed for MT(C).

The results of this study demonstrate that different soil management practices influence pore tortuosity. Previous studies by Elliot et al. [91] and Eltz and Norton [92] have identified variations in $\mathrm{Ʈ}$ in response to different management practices, indicating that the pore network is susceptible to changes due to the action of agricultural equipment. When examining systems under crop–livestock integration, Dhaliwal and Kumar [41] and Peth et al. [93] reported lower $\mathrm{Ʈ}$ values and well-connected pores, which align with the findings of our study. The lower $\mathrm{Ʈ}$ value for MT(C) can be attributed to the positive impact of this management type on the porosity and pore connectivity (Table 1). The addition of organic matter through cover crops improves the soil aggregation and aggregate stability, resulting in more connected and less tortuous pores [46]. Recent studies have demonstrated that management practices with less connected pores and low porosity tend to exhibit the highest tortuosity [49,88,94]. It can be observed that processes involving the densification of the soil (Table 1) by the transport of agricultural machinery or animals tend to increase the tortuosity [95,96]. This fact may explain why higher tortuosity values were observed mainly for areas under silage.

Notably, the $\mathrm{Ʈ}$ values that we found are consistent with those reported in the scientific literature. Pires et al. [82] investigated no tillage and conventional tillage systems and found $\mathrm{Ʈ}$ values ranging from 1.5 to 1.7. Ferreira et al. [79] examined soil under no tillage and pasture and found $\mathrm{Ʈ}$ values ranging from 2.0 to 2.7. Conversely, Galdos et al. [49] reported lower $\mathrm{Ʈ}$ values, varying from 1.3 to 1.5, for tropical soils under no tillage and conventional tillage. These findings demonstrate that $\mathrm{Ʈ}$ is a valuable indicator for the monitoring of changes in the soil pore network, as it is influenced by soil properties such as the texture and structure (Table 1), as well as by the management practices and vegetation cover crop systems. It is also important to note that the method for the measurement of $\mathrm{Ʈ}$ can also affect the results obtained [97].

The effects of the treatments studied on the contribution of the volume and the number of pores as a function of their shape are shown in Figure 7. It is of fundamental importance to study the pore shape, as the porous system is susceptible to changes due to the different processes that occur in the soil. These changes are influenced by natural and human-induced processes [72,94]. In the $VP$ and $NP$ results (Figure 7a,b), the contributions of unclassified pores, characterized by higher complexity, were not included. Notably, these pores represented approximately 65% of the $VP$ and $NP$. These pores are designated as unclassified when it is impossible to identify at least one of the semi-axes of the ellipsoids used to describe their shape.

The results show that the different treatments affected the more elongated (triaxial-shaped and prolate-shaped) pores (Figure 7). MT(C) favored the formation of rod-shaped pores, which were significantly different from those in the other treatments (p < 0.05), but with a reduction in the contribution of ellipsoidal-shaped pores (Figure 7a). Pores with rod and ellipsoidal shapes are essential for water infiltration, soil aeration, erosion resistance, and organic matter decomposition [41,82]. The elongated transmission pores have an important role in the processes of water conduction in the soil [6]. In the case of MT(C), the appearance of these pores corroborates the greater porosity and connectivity of the pores and the lower tortuosity found for this treatment. As previously discussed, elongated pores can be formed due to biological mechanisms. The activity of soil organisms, such as earthworms and other macro- and microorganisms, plays a crucial role in creating and modifying the pore structure [98]. These organisms move the soil and decompose organic matter through their burrowing activities [99], forming elongated pores. Other authors working with clay soils have also observed the predominance of $VP$ and $NP$ with an ellipsoidal shape in soil structures [41,49].

Although disk and spheroidal pore shapes are essential in the soil structure, they were attributed less to $VP$ than to the ellipsoidal and rod-shaped pores (Figure 7a). No significant differences (p > 0.05) were observed between the treatments for these pore types. The minor differences observed for some pore shapes analyzed between treatments indicate pores more resilient to changes [41]. Spheroidal and disc-shaped pores are generally found isolated from other pores and can be produced by the action of agricultural equipment, air entrapment during soil drying, and the activity of soil fauna [100,101]. The prevalence of these types of pores can indicate a soil with a damaged structure [6,102]. Authors such Pietola et al. [103] and Posadas et al. [104] have indicated that the transport of agricultural equipment and the inadequate use of pasture can result in more isolated and flattened pores. Consequently, the low contribution of these pores in the soil demonstrates that the area had good structural conditions for the treatments analyzed. It is reported that spheroidal and disk-shaped pores are generally more resilient to changes and have lower contributions due to being disconnected from other pores or having a very flat shape [101,104].

Elongated pores provide a greater capacity for the infiltration and storage of water and nutrients in the soil, avoiding air entrapment, and are important in drought conditions, where the storage of available water becomes crucial for plant development [105,106]. However, it is worth mentioning that, in the case of our study, unclassified pores were also vital, especially concerning the pore connectivity and tortuosity. These pores are generally formed by joining several other pores and, therefore, have complex shapes, making it difficult to classify them in terms of shape. For this reason, the relationship between the pore shape and different soil properties should be analyzed carefully.

When we compared the relationships between the analyzed properties, we found that MT(C) had the highest porosity and fractal dimension (Figure 5a,b). The correlation analysis comparing all treatments revealed a moderate positive linear correlation (r = 0.61, p < 0.05), indicating that lower $\phi$ is related to less complex porous media [56,107]. For $DA$, no relationship was observed with $\phi$ (r = −0.13, p < 0.05). However, there was a weak negative correlation with $FD$ (r = −0.47, p < 0.05). For pore connectivity, MT(C) was found to have the highest values, as was observed for $\phi$ and $FD$ (Figure 5d). Connectivity showed a strong (r = 0.88, p < 0.05) and moderate (r = 0.63, p < 0.05) positive linear correlation with $\phi$ and $FD$, respectively. This result suggests that soil systems with higher porosity and greater complexity are characterized by greater connectivity between pores [108,109]. However, $C$ and $DA$ did not show any relationship (r = −0.05, p < 0.05). The pore tortuosity was inversely related to $C$ (r = −0.58, p < 0.05) and $\phi$ (r = −0.63, p < 0.05). These results suggest a tendency for $\mathrm{Ʈ}$ to increase when the pores are less connected and the soil porosity is lower [109,110].

For the different pore shapes, positive correlations were found for the rod-shaped $VP$ and $\phi$ (r = 0.72, p < 0.05), $FD$ (r = 0.61, p < 0.05), and $C$ (r = 0.54, p < 0.05) and negative relationships with $\mathrm{Ʈ}$ (r = −0.65, p < 0.05). For the rod-shaped $NP$, positive correlations were found with $\phi$ (r = 0.61, p < 0.05), $FD$ (r = 0.86, p < 0.05), and $C$ (r = 0.51, p < 0.05) and negative with $\mathrm{Ʈ}$ (r = −0.72, p < 0.05). On the other hand, for the ellipsoidal-shaped pores, the largest contribution was to $VP$ and $NP$, revealing surprising correlations with these properties. For the ellipsoidal-shaped $VP$, negative correlations were identified with $\phi$ (r = −0.72, p < 0.05), $FD$ (r = −0.42, p < 0.05), and $C$ (r = −0.79, p < 0.05) and a positive correlation with $\mathrm{Ʈ}$ (r = 0.48, p < 0.05). The correlation between $NP$ and the ellipsoidal pores was weak for the different properties analyzed: $\phi$ (r = −0.09, p < 0.05), $FD$ (r = −0.06, p < 0.05), $C$ (r = −0.17, p < 0.05), and $\mathrm{Ʈ}$ (r = 0.05, p < 0.05).

A multivariate analysis was run to compare the influences of the different variables studied on the treatments and to complement the results of the other statistical analyses (Figure 8).

Figure 8 illustrates the principal component analysis (PCA) graph generated from the soil morphological and geometric property results, totaling 13 variables. The PCA reduced the two principal components (PC1 and PC2), with total variance of 65.9%. The axes are related to the loadings, which indicate the trends concerning the variables analyzed in the PCA for each quadrant. The full lines represent these trends.

The biplot graph (Figure 8) shows that MT(C) favored porosity, pore connectivity, and fractal dimension increases relative to the other treatments. The number and volume of rod-shaped and spheroidal pores were also positively influenced. On the other hand, the degree of anisotropy, tortuosity, ellipsoidal pore volume, and number of disk-shaped pores showed a reduction under MT(C). The graph also indicates an inverse relationship, mainly between the pore connectivity and porosity and the tortuosity. However, the fractal dimension, pore connectivity, and porosity showed positive relationships. An increase in the number and volume of rod-shaped pores is also associated with greater pore complexity due to its relationship with the fractal dimension. Other pore shapes tend to reduce the pore complexity, especially when the number of pores is considered.

NT management under cover crops and silage decreased the pore connectivity, porosity, and fractal dimension and reduced the spheroidal ($VP$) and rod-shaped ($VP$ and $NP$) pores compared to MT(C). However, this management type under the two uses of ryegrass caused an increase in the porous system’s anisotropy and the number of spheroidal and disk-shaped pores. Regarding MT(S), it was found that this treatment positively influenced the tortuosity and the volume of ellipsoidal pores when compared to MT(C). However, this treatment had a negative effect on the porosity and pore connectivity. In addition, there was a decrease in the volume of spheroidal and rod-shaped pores and the number of prolate pores.

4. Conclusions

The results of this study revealed that the MT(C) management system showed significant improvements in various soil properties, such as the porosity, fractal dimension, pore connectivity, volume, and number of rod-shaped pores. These results indicate a soil structure that favors water infiltration, air circulation, and nutrient transport processes. On the other hand, the MT(S), NT(C), and NT(S) systems showed results indicating that they may restrict water infiltration in the range of pore sizes assessed by microtomography in our study. These treatments were characterized by higher tortuosity and lower porosity and pore connectivity, especially when using ryegrass as silage.

The fractal dimension and anisotropy results show that all treatments exhibited complex and isotropic pore structures, regardless of ryegrass management or use. These results allow us to infer that solutes can infiltrate the soil without preferential paths, which is essential for the better redistribution of solutes and water retention for plants. The study also analyzed the number and volume of pores of different shapes. It was observed that ellipsoidal and rod-shaped pores were predominant, which are essential for soil water conduction and storage processes. On the other hand, spheroidal and disc-shaped pores were also identified, which contributed little to the soil porosity and showed minor differences between the treatments, proving resilient to changes.

Finally, our results enable a detailed analysis of the soil pore architecture in integrated crop–livestock production systems. These types of findings are fundamental in understanding how different land uses at the micrometer scale affect the soil pore system, considering the function of soil pores. Thus, the information obtained here can support new studies on the relationship between soil micromorphological properties and water infiltration and retention, which are fundamental for the development of crops under integrated production systems.