Soil microstructure alterations induced by land use change for sugarcane expansion in Brazil

Land use change (LUC) alters soil structure and, consequently, the functions and services provided by these soils. Conversion from extensive pasture to sugarcane is one of the largest land transitions in Brazil as a result of the growth of the domestic and global demands of bioenergy. However, the impacts of sugarcane expansion on the soil structure under extensive pasture remains unclear, especially when considering changes at the microscale. We investigated whether LUC for sugarcane cultivation impacted soil microstructure quality. Undisturbed soil samples were taken from two soil layers (0–10 and 10–20 cm) under three contrasting land uses (native vegetation—NV, pasture—PA and sugarcane—SC) in three different locations in the central‐southern Brazil. Oriented thin sections (30 μm) were used for micromorphological analysis. The total area of pores decreased following the LUC in the following order: NV > PA > SC in both soil layers. The area of large complex packing pores (>0.01 mm²) also decreased with the LUC sequence: NV>PA>SC. Qualitative and semi‐quantitative micromorphological analysis confirmed porosity reduction was driven by the decrease in complex packing pores and that biological features decreased in the same LUC sequence as the quantitative parameters. Therefore, LUC for sugarcane expansion reduced microscale soil porosity, irrespectively of soil type and site‐specific conditions, indicating that the adoption of more sustainable management practices is imperative to preserve soil structure and sustain soil functions in Brazilian sugarcane fields.

2019). Conversion from extensive and degraded pastureland to sugarcane production is the main scenario of LUC used to support sugarcane expansion in Brazil (Adami et al., 2012;Strassburg et al., 2014).
However, the intensive mechanization used in sugarcane fields, including soil tillage by ploughing and disking and heavy machinery traffic during mechanical harvesting degrades soil structure, affecting multiples of processes and functions in these soils Rabot, Wiesmeier, Schlüter, & Vogel, 2018). Soil structure is typically defined by the arrangement of soil particles and aggregates and the pores among the structural units, which regulates multiple processes and services such as water retention and conductivity, soil aeration, soil organic matter turnover, nutrient cycling (Six, Bossuyt, Degryze, & Denef, 2004), soil erodibility (Barthès & Roose, 2002) and plant growth. Therefore, parameters related to soil structure are considered key indicators of soil quality (Bünemann et al., 2018). Soil microstructure relates to the compositional arrangement of soil at a smaller scale (i.e. at the micron scale) and can be assessed by the use of thin sections, also known as micromorphology (Bullock et al., 1985). Although microstructure assessment by thin section can be time-consuming and generally does not provide 3D structural information, it provides more detail than other approaches where visualization of the soil microfabric is concerned.
Although traditional soil physical properties (e.g. bulk density, soil porosity, soil penetration resistance and soil aggregation) along with visual assessment methods can efficiently infer the stability, and even resilience of soil structure (Castioni et al., 2018;Cherubin et al., 2017), these methods cannot reveal the precise spatial arrangement of soil structure and the geometrical form of pores and aggregates. Imaging methods, such as micromorphology, can be used to further study the dynamics of soil structural development across the time and/or space and help improve understanding concerning the impact of soil structure on soil functioning (Guimarães, Ball, Tormena, Giarola, & Silva, 2013;Pires et al., 2017;Silva, Marinho, Matsura, Cooper, & Ralisch, 2015;Souza, Souza, Cooper, & Tormena, 2015). Whilst other imaging methods such as Xray computed tomography (CT) have become more popular for the analysis of soil pore space in recent years, particularly as they facilitate faster acquisition of images and 3D visualisation, micromorphology is still an important technique for the analysis of soil structure as it permits the microscopic visualization of some soil properties, such as those derived from organic matter, for example faecal deposits, that are currently not straightforward to image by X-ray CT (Helliwell et al., 2013).
Considering the intense mechanization applied to sugarcane soils, we conducted a field study to evaluate the impact of LUC for sugarcane expansion on soil microstructure characteristics using soil thin sections. The hypothesis was that the intensity of sugarcane cultivation had a significant impact of the alteration on soil microstructure and subsequent soil quality.

| Study sites
Undisturbed soil samples for 0-10 and 10-20 cm soil depth were taken in the central region of southern Brazil at three different locations within the main sugarcane-producing region of the country, as follows: Lat_17S near the city of Jataí-Goiás State (17º56′16″S 51º38′31″W), Lat_21S near the city of Valparaíso-São Paulo State (21º14′48″S 50º47′04″W) and Lat_23S near the city of Ipaussu-São Paulo State (23º05′08″S 49º37′52″W) with soil orders was classified as Oxisol, Alfisol/Ultisol and Oxisol by the USDA Soil Taxonomy (Soil Survey Staff, 2014), respectively. The climate was classified according to Köppen-Geiger's system as mesothermal tropical (Awa), humid tropical (Aw) and tropical (Cwa), respectively. The mean annual temperature and precipitation are 24.0°C and 1600 mm (Awa) at Lat_17S, 23.4°C and 1240 mm (Aw) at Lat_21S and 21.7°C and 1479 mm (Cwa) at Lat_23S, with the rainy season in the spring-summer (October to April) and the dry season during the autumn-winter (May to September). More detailed climate information (mean monthly temperature and precipitation) are available in Cherubin et al. (2015).
In each site, we sampled a LUC sequence, including native vegetation (NV, baseline), pasture (PA) and sugarcane (SC) areas. Selected physical and chemical soil properties are found in Table 1. The land use and management history of each site, as well as chemical and physical characterization of the soils are further described in . For all sugarcane areas, the soil was prepared by ploughing and disking previously to cropping. The SC fields at Lat_17S, Lat_21S and Lat_23S were in the third, third and fourth ratoon, respectively. In SC fields, fertilizer was applied annually and harvesting was performed using a 20 Mg harvester and transported by a tractor and wagon (10 + 30 Mg). A controlled traffic system was not used in these areas.

| Soil sampling and preparation
One undisturbed soil sample (7 × 12 × 6 cm) was collected in the Lat_17S, Lat_21S and Lat_23S for NV, PA and SC in two soil layers (0-10 and 10-20 cm), totalling 18 samples (3 sites × 3 land uses × 2 soil depths). For sugarcane, the soil was sampled in the inter-rows. The soils were air-dried for 35 days and then placed in an oven at 40°C for 48 hr. The dry samples were impregnated with a polyester resin, styrene monomer and fluorescent dye (Tinopal BASF ® ) by capillarity in a vacuum chamber. After impregnation, vertically oriented soil thin sections (c. 30 μm thick) were obtained for qualitative and semi-quantitative description (Bullock et al., 1985;Cooper, Castro, & Coelho, 2017;Stoops, 2003) and quantitative image analysis (Cooper, Boschi, Silva, & Silva, 2016). Figure 1 illustrates the sampling procedure adopted in the field.

| Micromorphological analysis
The thin sections were analysed using a Zeiss petrographic microscope. The qualitative description of thin section was made following the classifications described in Bullock et al. (1985) (2003) only for thin sections from Lat_23S. This method provides reference images for a semi-quantitative assessment of porosity and the description of pore morphology. The pores were classified as packing pores, that is those that result from the loose packing of soil components; channel pores, that is tubular smooth pores with a cylindrical or arched cross section which are uniform over much of the length; vughs, that is more or less equidimensional, irregularly shaped, smooth or rough, usually not interconnected; and planar pores, that is flat, accommodating or not, smooth or rough, resulting from shrinkage or compaction (Stoops, 2003). The soil coarse/fine (c/f) fabric was classified as either porphyric (i.e. coarse grains embedded in fine material), enaulic (i.e. fine material appears as microaggregates between coarser components) or combinations of these as described in Stoops (2003).

| Micromorphometrical analysis
Ultraviolet light was used to enhance the contrast between the pore space and soil matrix, and images were obtained using a charged couple device photographic camera (DFW-× 700, Sony ® ). For each soil sample, fifteen images of 180 mm² were randomly obtained ( Figure 1). The images were digitalized with a resolution of 1,024 × 768 pixels in 256 shades of grey in a 10 × amplification giving a pixel size of 12.5 μm. Pore segmentation was undertaken in Noesis Visilog version 5.4 by means of a user-defined threshold (maintained throughout the study), opening and closing filtering, and labelling, which correspond to the individualization of each object followed by its identification. The smallest segmented pore had a diameter of 37.5 μm, which is classified in the meso/macro-pore size range, the size class most sensitive to soil compaction (Richard, Cousin, Sillon, Bruand, & Guérif, 2001).
The total area of pores (TAP) for each image was calculated as the percentage of the sum of the areas of the individual pores divided by the total area of the assessed image (Hallaire & Cointepas, 1993). Pore shape was classified into three groups as in Cooper et al. (2016): rounded, elongated and complex. Two indexes were used to determine the pore shape, as described in Equations 1 and 2: where P is the perimeter of the pore and A is the area.
NI is the number of intercepts of the object in direction i (i = 0°, 45º, 90° and 135º), DF is the Feret diameter of the object in the direction j (j = 0° and 90°), m correspond to the number of i directions and n to the number of j directions. The I2 index was used complementary to I1 for a better pore segregation according to shape.
When morphometric shapes are compared with the micromorphological classification, rounded pores correspond to vughs, elongated pores to channel and planar pores, and complex pores to packing pores.

| Data analysis
The mean soil porosity of each site was derived from 15 subsamples (every image from a single thin section), which were used as pseudo-replicates (Hurlbert, 1984) to compare the difference in LUC porosity for each site; to compare the LUC effect on soil porosity for the central-southern region, each site was considered as a replicate (n = 3). Data normality was tested by Shapiro-Wilk's test (p > 0.05), followed by an analysis of variance (ANOVA) and post hoc via Duncan's test (p < 0.05).

| Micromorphological analysis
Regardless of land use, the soils presented a dominant porphyric relative distribution with secondary areas presenting as porphyric-enaulic, enaulic-porphyric and enaulic-related distributions. The porphyric-enaulic-related distribution areas only occurred in agricultural land uses (PA and SC) whilst the enaulic-porphyric areas were only observed in NV soils (Table 3).
The soil micromorphological descriptions also showed a reduction in soil porosity in both layers as a result of the LUC from native vegetation to pasture (Table 3). Also, the pore morphology observed for native vegetation showed more complex packing pores than in the pasture. In the pasture soils, there was a reduction in complex packing pores and an increase in policoncave vughs in both layers whereas planar pores were generally identified in the subsurface layer (Table 3).
The porosity of soil under sugarcane was lower than pasture only for the 0-10 cm layer. The pore morphology analysis showed a further reduction in complex packing pores from pasture to sugarcane and an increase in spherical and policoncave vughs and channels (Table 3). When pedofeatures were analysed, a reduction in biological features from native vegetation soil to pasture was observed. However, the bio-pores, characterized by the infilling of pores, and aggregates had no clear differences in diameter. The LUC from pasture to sugarcane also led to a reduction in biological features (pores, aggregates and coprolites) and the size of biological-derived aggregates in the 0.1-0.2 m layer (Table 3).

| Micromorphometrical analysis
Considering all sites, the total area of pores (TAP) was 1.2-2.1 times higher in the surface layer (0-10 cm) of NV soils than pasture soils, whereas sugarcane soil had a TAP 1.5-2.2 times lower than pasture soils ( Table 2). The same pattern of change induced by LUC (Table 2) was observed at site scale, except for Lat_21S where PA did not differ from NV. For the subsurface layer (10-20 cm), LUC did not induce changes in TAP (Table 2) when considered at the regional scale. However, for Lat_23S, the NV had a higher porosity than PA and SC, and the TAP of NV was higher than PA, which was higher than SC at Lat_17S (Table 2). For the top soil layer (0-10 cm), the soil pores at NV were rounded, elongated and predominantly, complex pores. A reduction in complex and larger pores was observed in accordance with a reduction in TAP with the LUC sequence; NV > PA > SC. This indicates the reduction in the TAP was Lat_21S 0-10 27.3 aA ± 8.5 23.7 aA ± 6.5 14.5 bA ± 6.0 10-20 15.8 aB ± 6.0 16.9 aB ± 6.8 14.7 aA ± 8.6 Lat_17S 0-10 45.1 aB ± 8.3 21.1 bB ± 3.1 13.8 cB ± 3.0 10-20 59.6 aA ± 10.6 29.6 bA ± 6.0 16.5 cA ± 6.0 Note: Different lowercase letter indicates statistical difference between the land use, and uppercase letter indicates the statistical difference between layers by Duncan test with 5% probability. driven by large and complex pores representing a loss of in the portion of complex packing pores, which is observed in Figure 2, where the 10-20 cm soil layer was less sensitive to this alterations at Lat_21S and Lat_23S (Figures 3 and 4).

| Impacts of conversion from native vegetation to pasture on soil microstructure
Land transition from native vegetation to pasture promoted reduction in porosity in surface and subsurface soil layers at Lat_23S and Lat_17S. However, considering the data at the regional scale, this conversion induced a reduction in the soil porosity only for the superficial layers (Table 2). These results are in agreement with a higher soil bulk density (BD), reduced macroporosity (MaP) and hydraulic conductivity (K fs ) of these same pasture soils found by Cherubin, Karlen, Franco et al. (2016).
In addition, despite the contrasting scales of evaluation, our micromorphometric analysis confirmed the results obtained by on-farm visual evaluation by Cherubin et al. (2017), using the Visual Evaluation of Soil Structure (VESS) method (Guimarães et al., 2011). Based on VESS assessment, pasture soils presented larger, harder and less porous aggregates than native vegetation soils, resulting in lower overall soil physical quality in the 0-to 25-cm layer (Cherubin et al., 2017). Cattle trampling may be the main driver of soil porosity reduction in pastures. Mulholland and Fullen (1991) observed higher BD and penetration resistance in pastureland soil after trampling using a thin section evaluation. Also, soils under native vegetation can have higher organic matter inputs than the anthropic land uses, increasing organic matter content , which is responsible for aggregate formation and stabilization (Six et al., 2004), providing better soil physical conditions . The quantitative pore shape results showed a reduction in larger complex pores (Figures 3 and 4). This reduction did not alter the soil microstructure between these LUC's, but changes were identified in the qualitative pore morphology analysis showing a decrease in complex packing pores and an increase in spherical and policoncave vughs and fissures from NV to PA (Table 3). These changes in the quantitative and qualitative pore morphology assessments are also reflected in the changes in the related c/f distribution with a transformation of enaulic and enaulic-porphyric-related distribution in NV to a porphyric-enaulic-related distribution in PA. This morphological evidence suggests an incipient compaction process in PA that caused by animal trampling and poor pasture management that may reduce the benefits of soil macrofauna bioturbation, which is partly responsible for the formation of these morphological features. Compaction causes a reduction in the total volume of pores, and this reduction not only alters pore morphology but changes the pore size distribution (Boivin, Schäffer, Temgoua, Gratier, & Steinman, 2006). Therefore, the pore size and shape results obtained in this study can be useful indicators or proxies for pore connectivity and tortuosity properties, which are important for the evaluation of changes in key soil functions and services (Rabot, Wiesmeier, Schluter, & Vogel, 2018;Silva et al., 2015), such as regulation of water fluxes and soil aeration, induced by land use change and soil management practices. Although, the observation in 2D is a limitation in this instance as an assessment of pore connectivity in 3D is more appropriate for prediction of some soil functions, for example soil hydraulic behaviour. Further investigations combining both the data from thin sections and X-ray imaging would improve our understanding concerning the soil structure changes induced by agricultural land uses, as well as to better establish the linkage between soil structure dynamics and the provision of soil functions and ecosystem services.

| Impacts of conversion from pasture to sugarcane on soil microstructure
Our results indicated a reduction in total porosity, mainly in the surface soil layer (0-10 cm), when sugarcane was converted from pasture ( Figure 2). The decrease in packing pores observed in the micromorphological analyses (Table 3) confirms the reduction in porosity and complex pores observed in the quantitative image analyses. Overall, land transition from pasture to sugarcane increases the mechanical compressive stresses applied on the soil surface, causing microstructural degradation as a result of the coalescence of aggregates by compaction. The effect of this microstructural degradation in this study is evidenced by the significant reduction in the complex pore areas as a result of LUC, and in some sites, by the increase in less connected and more rounded pores (Figures 3 and 4). This pore morphology change was also observed in the decrease in the percentage of complex packing pores and increased percentage of spherical and policoncave pores from PA to SC (Table 3). Microstructure changes from a microgranular to blocky structure, both with well-developed aggregates, and an increase in porphyric c/f distributions, were also observed. These modifications in microstructure, c/f distribution and pore morphology occur as result of mechanical stress (Silva et al., 2015) and reduce soil aeration, water and nutrient uptake and crop yield (Lipiec,   Ishioka, Szustak, Pietrusiewicz, & Stepniewski, 1996). Soil compaction creates a restrictive environment for plant growth due the physical impediment for roots development (Lipiec & Hatano, 2003) and the reduction in soil aeration and consequentially, the redox potential (Eh) (Czyz, 2004), creating a poor bio-chemical environment (Husson, 2013). Otto, Silva, Franco, Oliveira, & Trivelin (2011) showed the inverse relationship between soil penetration resistance and diverse root parameters (root length, area and density). The background for these limitations for plant and root growth could lie in changes in microstructure and pore morphology as a result of LUC as we have shown in this study.
Our results highlighted the urgent need for more sustainable management practices to improve soil physical quality, especially those related to the improvement of soil microstructure and pore morphology, mitigating the negative impact of biofuel production. As sugarcane planting typically occurs between September and March (in the central region of southern region in Brazil), which is also the rainy season, it is important to avoid, or at least restrict, machinery traffic under high soil moisture conditions and to encourage the introduction of conservation agriculture cropping systems that reduce or eliminate soil tillage  and recommend the use of cover crops as an alternative to prevent soil structure degradation and mitigate other agronomic issues, such as weeds, pests and soil fertility. In this context, cover crops can also be used to improve soil structure at scales as fine as considered here through root modification of the soil porous architecture (Bacq-Labreuil et al. 2019). As there is an increasing interest in sugarcane straw to cogenerate bioelectricity or produce 2G ethanol, maintaining part of the sugarcane straw in the field is an important practice to improve several soil physical quality properties, such as soil structure, pore size and morphology, BD, resistance to penetration, among others (Castioni et al., 2018. Other soil parameters, such as soil organic matter, soil fauna and soil texture (Bonetti, Anghinoni, Moraes, & Fink, 2017;Porre, van Groenigen, De Deyn, de Goede, & Lubbers, 2016;Six et al., 2004;Vreeken-Buijs, Hassink, & Brussaard, 1998), are important for soil structuring and may contribute to the differences in changes in pore morphology and size observed in this study. However, irrespectively of the site-specific conditions (climatic, biological, chemical and physical), the results of the micromorphological and micromorphometrical analysis, together with the physical attributes provided by , Cherubin et al. (2017), show that the soil compaction process occurs following LUC. More sustainable management practices are necessary to maintain the soil physical properties (e.g. soil structure, pore morphology and size, and pore connectivity) that influence soil functions (e.g. hydraulic conductivity, air permeability, C storage and physical stability to resist against degradation) in Brazilian sugarcane fields to achieve the expected productivities.

| CONCLUSIONS
Land use change from native vegetation to pasture to sugarcane degraded the soil microstructure, reducing the porosity of the soil and negatively influencing the pore shape and size distribution, irrespectively of the soil texture and site environmental conditions. As changes in soil microstructure and pore morphology affect important soil hydrological and physical attributes, which in turn can negatively affect crop yield, the adoption of more sustainable management practices in sugarcane fields (e.g. reduced soil tillage, cover crop incorporation, straw retention and machinery traffic control) is imperative to preserve and/or enhances soil structure and, consequently, sustain soil function in a productive capacity.