Silicon exportation by crops alters soil mineralogy and clay fraction

Silicon (Si) dynamic in system controls mineral evolution. We expected that the Si exported from soil due to soybean cultivation would affect Si forms and clay minerals. The objective of this study was to evaluate Si forms in the soil-plant system in areas with different soybean cultivation times in order to respond how Si exportation affects soil mineralogy. Oxisols under soybean cultivation for 2, 8 and 40 years were evaluated and an adjacent area with native vegetation was used as the control treatment. The total and available Si in the soil and in the roots, aerial part of the plants and in the soybeans were evaluated, as well as the physical, chemical and mineralogical attributes of the soil.


Introduction
Silicon oxide (SiO 2 ) is the most abundant cation in the earth's crust. The presence of Si in terrestrial ecosystem compartments depends on environmental factors such as: climate, topography, soil source material, weathering and biotic factors (i.e. vegetation, microorganisms) as well as soil use and management (Alexandre et al. 1997;Sommer et al. 2006;Street-Perrott and Barker, 2008;Alexandre et al. 2011;White et al. 2012).
In the soil/vegetation system, minerals are the reserves of silicon (Si), represented by tectosilicates, which are resistant to weathering, called lithogenic silicates (Henriet et al. 2008) while the phyllosilicates start to control the primary source of Si for the plants because they are susceptible to rapid dissolution (Henriet et al. 2008). The total Si content in the soil depends on the mineralogy of the soil and it ranges from 5 to 470 g kg -1 (Mckeague and Cline, 1963). Part of the Si in the ecosystem is in the vegetable form of biopals (phytoliths). These structures are originated from plants that present the ability to store Si in their tissues. Phytoliths are formed from monosilisic acid (H 4 SiO 4 ) present in the soil solution that is absorbed by the root system of vegetables and, subsequently, deposited in the inter and intracellular spaces along the leaves and stems, forming silici ed structures representing phytolites (Parr and Sullivan 2005). Plants store Si in the form of phytolites in many plant tissues, especially at the end points of transpiration (Hodson et al. 2005;Schaller et al. 2013).
The Si biogeochemical cycle in the rock/soil/vegetation system determines the mineralogical evolution of these compartments (Cornelis and Delvaux 2016). The amount of Si in the compartments is also determined by human activity, for example, irrigation, and the change of native vegetation by agricultural crops (Cornelis and Delvaux 2016). In arable land, it is known that the Si cycle, as well as its magnitude in two different ways, is in uenced by the export of Si from the soil by harvesting grains, which reduces the storage of relatively soluble phytolites in the topsoil (Vandevenne et al. 2012).
Phytoliths represent an important reservoir of biogenic silica in acidic soils from tropical and subtropical regions. Phytoliths are released during the degradation of organic matter in the litter horizon and later transferred to deeper horizons or evacuated by air or hydrographic route (Bartoli 1981). Studies show that the solubility of phytolites is close to that of amorphous silica, while their dissolution rates are between those of quartz and vitreous silica, with similar pH dependence. The production of phytoliths more susceptible to dissolution, in the words of Alexandre et al. (1997), may be an evolutionary response to environmental pressure, perhaps a shortage of Si, as for example in Oxisols.
In the literature, several authors have studied mineral changes due to plant growth and development. Korchagin et al. (2019) evidenced the dissolution of the clay fraction, releasing Al and forming HIMs in response to the action of eucalyptus root. Bortoluzzi et al. (2012) evidenced the loss of the clay fraction and mineral alteration in response to the extraction of K in grapevine. Low composition of 2:1 clay mineral was found in experiments involving no fertilization of potassium (K), as well as with the addition of K, when testing the absorption by vegetables (Moterle et al. 2016;. Oort et al. (2016) in a 42-plot experiment in Versailles observed the partial dissolution of smectite by microdivision through the alteration of ne phyllosilicate minerals or a process of neoformation of secondary phases of Si, Al and/or Fe released by dissolution of clays in acidic medium. Studies on Si cycling had already been performed, and some of them have elucidated the plant absorption and the bioavailability of Si in soils (Blecker et al. 2006;Henriet et al. 2008), as well as the Si balance in terrestrial environments (Chadwick and Chorover 2001), however, none of them were associated to the Si exportation from commercial crops and related to soil mineralogical alteration. In this sense we expected that the Si exportation by plants play an important role in the mineral evolution of the soil, which the mechanism is still poorly understood.
This information may aid the fertilization managements considering Si in new formulations, although this element is not an essential nutrient for plants.
We know that plants absorb Si from the soil solution in the form of monosilicic acid Si(OH) 4 (Tisdale et al. 1993).
The return of Si to the soil can occur through the plant debris after death and the release of Si in the form of monosilicic acid and in the form of phytolites, structures that are rich in Si and more stable (Piperno 2006). In the Si biogeochemical cycle, if there is a change in the amounts of Si in these compartments, there may be degradation of biogenic (biopals) and inorganic (clay) silicate minerals releasing Si from the reserve compartments for more available forms of Si. There may also be a recombination of Si in coprecipitation with other elements and neoforming other clay minerals (Bortoluzzi et al. 2018). The small clay minerals, in turn, can dissolve, as evidenced by Korchagin et al. (2019) and Bortoluzzi et al. (2012), which contributes to replenish the available Si in the soil.
The objective of this study was to evaluate if the mineral elements of the soil are affected by agricultural cultivation, discussing the Si exportation as a mechanism of mineralogical alteration. For this, we evaluated Si forms in the soil and plant system in areas with different times of soybean cultivation.

Materials And Methods
Experimental place and soil classi cation The soils were collected in January 2018 in the town of Água Santa -RS (Figure 1 a), at 650 m of altitude, 28°10'37" S, 52°02'02" O. The climatic classi cation of the place is according to Köppen-Geiger, with an average temperature of 20.6 °C and an average annual rainfall of 1409 mm, belonging to the Atlantic Forest biome.
The soil was classi ed as typical Latossolo Bruno Distró co (Oxysoil) after morphological description and classi cation of a soil pro le located in the native eld, at an altitude of 740 m, 28º16'0.408" S, 52º1'57.396" W ( Figure 1), according to SBCS (2018).
The history of the studied places, reported by the owner, shows that the areas cultivated for 2 and 40 years were occupied by native forest and the area cultivated for 8 years was covered by native grassland prior to cultivation. However, before native forest, it is believed that the predominant vegetation would be the grassland. Soybean crop is grown annually in the summer season and grasses such as ryegrass or oats for grazing in the winter. There are no records of the exact rates of limestone and fertilizers used in each area annually.

Soil collection
Soil sampling was carried out on January 17, 2018, in all areas with different times of land use. Ten replicates were performed, each one comprising 10 subsamples collected with an auger at a depth of 0-20 cm. The experimental design was completely randomized.

Collection of plants
Plant sampling was performed on the same day of soil collection and in all areas with different times of land use. Whole plants were collected, that is, aerial part together with the root. In the three cultivation areas, soybean plants were collected in the R1 stage. In the native eld, the natural cover plants present in that period were collected, where grass species prevailed.
The plants were collected in order to have three replicates, being each one composed of ve soybean plants and plants from the native eld present in the area of 1 m 2 . After collection, the samples were sent to the laboratory, where the plant material was washed in a detergent solution and rinsed in distilled water to remove the detergent.
The plant material was dried in a forced circulation chamber at 65 o C until constant weight, after which the root and aerial part were separated and grounded in a Willey mill. After grinding, the plant material was packed in paper bags and sent to predetermined analyzes.

Grain collection and sample preparation
At the end of the soybean crop cycle, grain samples were collected from three replicates of 5 plants from each treatment. The grains were manually traced and the samples were sent to the XRF analysis.

Soil size
The granulometric analysis was performed on the soil samples, using the pipette method (Gee and Bauder 1986).
The samples were subjected to the burning of organic matter and subsequently to chemical (NaOH 0.1 mol L -1 + NaPO 3 0.07 mol L -1 ) and mechanical (stirring for 16 h on a horizontal shaker) dispersion. Afterwards, the sand (> 0.053 mm) fractions were obtained by sieving and silt (0.002 -0.053 mm) and clay (<0.002 mm) by pipetting.

Soil density analysis
Soil density was determined by the volumetric cylinder method, with undisturbed samples, using the mass of the dry soil divided by the volume of soil collected in the the cylinder. The density was obtained by averaging soil samples collected every 5 cm in the 0 -20 cm layer.

Chemical characterization of the soil
The organic matter content of the soil was determined according to Embrapa (2011), where the pH and SST indices and the levels of P, K, Ca, Mg, Mn and Al were determined according to Tedesco et al. (1995).

Elementary analysis by X-ray uorescence
The analysis of the total contents of the chemical elements was performed in the source rock, in the total fraction of the soil and in the clay fraction. The elements expressed in the form of oxides were determined using the X-ray Fluorescence (XRF) technique in the soil samples, being all of them analyzed in the form of powder pressed to 25 tons, forming tablets.
In addition, samples of undisturbed soybean plants were analyzed in the XRF with the main purpose of quantifying the Si content exported by the grains. Besides that, the contents of other chemical elements present in the grains of the crop were also veri ed.
The determination was made using a Bruker S2 Ranger XRF equipment, which is basically a compact Energy Dispersible X-ray spectrometer (EDX), with a 50W high-voltage maximum power generator, 50 KV voltage and power intensity, current of 2 mA, with X-ray tube of anode material (Pd -Lead). It has a XFLASH detector with 129 eV resolution and Berilio window, operation in air, He and vacuum, analyzes from Na (Sodium) to U (Uranium).

Analysis of available Si in the soil
The determination of Si in the soil was performed by colorimetry. After extraction with 0.01 mol L -1 of calcium chloride, 10 g of soil were used, and 100 ml of the extracting solution were added. The reading was performed on a UV-Visible Spectrophotometer at a wavelength of 660 ηm (Korndörfer et al. 2004).

Available Si in the plant samples
The Si extraction process in the plant was performed through the oxidation of the organic matter, that is, the elimination of carbon from plant tissue with hydrogen peroxide (digestion). 0.1000 g of the ground vegetable material was weighed and 3 ml of NaOH (500 g L -1 ) were added. The reading was performed on a Visible UV Spectrophotometer at a wavelength of 410 ηm (Korndörfer et al. 2004).

Analysis of phytoliths in the soil
The separation of phytoliths in soil samples from the native eld and from the area cultivated for 40 years was carried out using the method proposed by Madella (1998) for simpli ed extraction of phytolites, using sodium polythungstate with a density of 2.35 g/cm 3 , and it was performed directly on air-dried ne textured soil. Three slides per horizon of each soil pro le were examined in order to identify phytoliths in a petrographic microscope with 400 x magni cation. Phytoliths were identi ed based on the International Code for Phytolith Nomenclature -ICPN 1.0 (Madella et al. 2005) and International Code for Phytolith Nomenclature -ICPN 2.0 (Neumann et al. 2019). The extracted phytoliths were wheighed and the percentage of phytoliths present in the analyzed portion of the soil was estimated.

Mineralogical characterization of soil samples
The clay and silt fractions of the soil samples were characterized mineralogically using the X-ray diffraction equipment (XRD), available at the Centro Tecnológico de Pedras, Gemas e Jóias of Rio Grande do Sul (CT Pedras), in the city of Soledade/RS. The detailed methodology for mineralogical analysis can be found in Poleto (2013) and the identi cation keys are found in Brindley and Brown (1980). The samples must be ground beforehand.
Afterwards, they were subjected to pre-treatments that consisted of: dispersion by mechanical agitation or ultrasound; granulometric separation by sedimentation or centrifugation; the aliquots containing the representative fractions of the particle sizes of the ground material were subjected to powder XRD analysis (disoriented sample, scanned in 3-55º 2θ XRD) and in samples oriented on air-dried glass slides (XRD scan 3-55º 2θ) and after solvation in ethylene glycol (DRX scan of 3-55º 2θ) and after heating under temperatures of 150 ºC , 350 ºC and 550 ºC in a mu e furnace (DRX scan of 3-55º 2θ).

Statistical analysis
The results of the analysis of granulometry, soil density, organic matter, XRF, Si available in the soil and Si in the plant were subjected to analysis of variance using the F test. For the treatments that showed a signi cant difference, the comparison of means was performed by the Tukey test (p <0.05).

Effect of cultivation time on soil physical and chemical properties
The area cultivated for 40 years showed 14% less clay and 25% more silt compared to the control (native eld), demonstrating that the longer the soil cultivation time, the lower the clay content and the greater the silt content. There was no signi cant difference between the sand content from the 40-year crop area and the native eld ( Figure 2). The lowest organic matter content and the highest density of the soil were observed in the soil cultivated for 40 years (Table 1). The highest pH value and potential acidity were found in the place cultivated for 40 years. The pH of the 2 and 8-year cultivated areas and the native eld showed no statistical difference. The lowest values of potential acidity were found in the native eld and in the 8-year crop, consequently the highest levels of exchangeable aluminum were found in these same treatments. The levels of manganese were higher in the native eld and in areas cultivated for 2 and 8 years (Figure 3). The XRF analysis of the total soil fraction showed that the highest SiO 2 content was found in the 40-year crop, followed by the 2 and 8-year crops. The native eld and the rock were the treatments with the lowest SiO 2 content. The highest content of Al 2 O 3 , Fe 2 O 3 and K 2 O were also found in the rock and in the native eld (Table 2). The XRF analysis of the total soil fraction showed the highest content of CaO and P 2 O 5 in the area cultivated for 40 years, followed by the 2 and 8-year crop, while the lowest levels were found in the native eld and in the rock.
The contents of MgO, ZrO 2 and BaO showed no difference between the treatments studied ( Table 2).
The XRF analysis of the clay fraction of the soil showed the highest SiO 2 content in the 2-year crop, followed by the native eld. The areas of 8 and 40-year crop presented the lowest SiO 2 content. The highest levels of Al 2 O 3 and Fe 2 O 3 were found in the native eld (Table 3).  (Table 4).

Si balance
In order to assess the Si contents of the plant, rock, clay fraction and soybean grains, the Si balance was performed. After 8 years of cultivation, the Si contents of the clay fraction decreased by 2 g/kg and after 40 years, these contents were reduced by 6 g/kg ( Figure 6). The Si content in the dry matter seems to increase in the area cultivated for 8 years, however, after 40 years the Si content was lower, as a consequence, the Si exported by grains was around 12 to 20 kg/ha per year of cultivation, totaling a negative balance of 500 kg/ha of Si extracted from the grains in 40 years of cultivation (Figure 7). Note: The estimate of the dry mass production of 1200 kg/ha was used for native elds and 7945 kg/ha for soybean according to literature data. The annual soybean grain production was estimated in 3600 kg per hectare.

Phytolytes analysis
Phytolytic analysis identi ed the presence of the morphotypes: elongate cavate, elongate sinuate, cylindric sinuate, rectangular aerolate, rectangular nodulate and globular nodulate in the native eld soil ( Figure 8A) and the morphotypes: acicular, saddle, elongate smooth, bilobate, rondel, rectangular and saddle in the soil cultivated for 40 years ( Figure 8B). The quanti cation of phytoliths determined by weighing showed that 0.63% of the native eld soil and 0.36% of the cultivated soil were composed of phytolites.   (Figure 9 a).
The mineralogical analysis of the soil of the area cultivated for 2 years showed that there was no alteration of the mineralogical composition of the clay fraction when compared to the native eld soil. Thus, the minerals found there were: muscovite (d=1.001 nm), kaolinite (d=0.716 nm; d=0.357 nm; d=0.228 nm), quartz (d=0.425 nm; d=0.334 nm), cristobalite (d=0.405 nm; d=0.248 nm; d=0.187 nm) and calcite (d=0.386 nm; d=0.302 nm; d=0.209 nm; d = 0.191 nm) (Figure 9 b).
Mineralogical analysis of the soil cultivated for 8 years showed that the minerals that were part of the clay fraction are muscovite (d=1.001 nm), zeolite (d=0.935 nm), kaolinite (d=0.716 nm; d=0.357 nm; d=0.228 nm), quartz (d=0.425 nm; d=0.334 nm), cristobalite (d=0.405 nm; d=0.248 nm) and calcite (d=0.386 nm; d=0.302 nm; d=0.209 nm) (Figure 9 c). The soil cultivated for 8 years showed absence of cristobalite peaks at a distance of d=0.187 nm and calcite at a distance of d=0.191 nm in relation to the native eld (Figure 9 a)  The mineralogical evaluation of the samples saturated with calcium in the air-dried treatment showed that kaolinite (d=0.716 nm and d=0.357 nm) has decreased its intensity in areas with 8 and 40 years of cultivation. The peaks of quartz (d=0.425 nm and d=0.334 nm), cristobalite (d=0.405 nm and d=0.248 nm) and calcite (d=0.209 nm) increased their intensity in the soil cultivated for 8 and 40 years (Figure 9 a). It was not possible to observe differences in the mineralogical evaluation of the powdered soil (Figure 10 b).
Fig 10 X-ray diffraction of the clay fraction of the soil: a) saturated with calcium and air-dried in oriented coverslips; b) powdered soil, in the native eld and soil submitted to cultivation for 2, 8 and 40 years, at a depth of 0-20 cm Figure 11 shows the X-ray diffraction patterns of the decomposed clay fraction saturated with Ca and solvated with ethylene glycol (Ca -EG). In the native eld and areas of 2 and 8 years of cultivation, the peak of muscovite was present (d = 1.001), and in the soil cultivated for 40 years, the peak of muscovite was not found. The relative proportion of diffracted intensity attributable to muscovite varied between 5.01-8.70%, while the surface area of kaolinite varied between 91.3-100% ( Figure 11). In addition, the decomposition of the soil cultivated for 8 years showed the peak centered at d = 0.935 nm, demonstrating that the zeolite group is present, which contributed to 0.84% of the diffracted intensity ( Figure 11). The mineralogical assessment of the native eld and from the soil cultivated for 2, 8 and 40 years showed that the minerals that were part of the silt fraction are basically quartz (d=0.425 nm; d=0.334 nm; d=0.245 nm; d=0.228 nm; d=0.223 nm; d=0.213 nm, d=0.197 nm and d=0.182 nm) and cristobalite (d=0.405 nm), with no difference among the treatments applied and the studied soils (Figure 12 a, 12 b, 12 c and 12 d). The lower clay levels found in the 8 and 40-year cultivated areas indicated that the absorption of Si by the vegetation and subsequent exportation contribute to the dissolution of minerals from the soil. The increase in silt content, resulting from the transformation of sand to silt and later to clay, demonstrates the degradation of primary minerals from lithogenic silicates, which when released can be recombinant to synthesize new-formed clay. These clay-size minerals can, in turn, dissolve and contribute to replenish the availability of Si in the soil (Mckeague and Cline 1963).
The respiration of the roots releases CO 2 and forms carbonic acid, a weak acid that signi cantly changes the pH of the soil (Cotter-Howels and Paterson 2000). When in extreme, acidic or mainly basic pH media, their dissolution potential is signi cantly increased (Feth et al. 1961), these pH conditions can also favor ion coprecipitation and mineral crystallization ).
The pH and TSM values, as well as the levels of Al, Mn, P, K, Ca and Mg are the result of the use of limestone and fertilizers in the crops, considering that there was no application of these products in the native eld. The 8-year crop has the highest P and K content, these elements are made available quickly by soluble fertilizers. The higher levels of Ca and Mg present in areas of 2 and 40 years of cultivation, are due to the use of higher doses of agricultural fertilizers that contain these elements, as informed by the farmer.
The organic matter content of the soil decreases over the years of cultivation and, consequently, the density of the soil increases, which was expected due to the maintenance of the carbon content of native areas compared to cultivated areas, as well as the tra c of agricultural machines in tillage areas, which can increase the density of the soil.
Total and available Si in the soil and Si in the plant at different times of soil cultivation The highest values of Si available in the soil were found in the native eld and in the 2-year crop, which is why Si cycling occurs through the current natural vegetation (native eld) or recent land use (2-year cultivation). The native eld, being basically composed of grasses, becomes more aggressive in the absorption of Si, explaining the higher levels of Si found in the aerial part and roots of the plants from the native eld, however, as there is no export outside the system, this Si returns to the soil in the form of phytoliths, which are dissolved over time and the Si can be reused (Chadwick and Chorover 2001;Blecker et al. 2006;Henriet et al. 2008). When the native eld area is cultivated, the Si contents present in the vegetation return to the soil with the total senescence of the native vegetation, thus explaining the higher Si content available in the 2-year crop.
Areas cultivated for 2 and 8 years had a higher Si content in soybean plants than that cultivated for 40 years, as the levels of Si available in the soil are higher, enabling luxury consumption by the plant, through transfer from soil to the grain, demonstrating that the greater Si export by the grain, the greater the Si available in the soil.
The highest total Si content of the entire fraction of the soil was found in the 40-year crop, present in poorly soluble primary minerals such as quartz and mostly in the sand and silt fractions. The clay fraction had the lowest total Si content, demonstrating that there was dissolution of the most soluble silicate minerals and the release of Si, which was absorbed by the plants and exported by the grains. In addition, over the years, there was a reduction in the storage of phytolites in the topsoil of 0.63 to 0.36%, which was also observed by Vandevenne et al., 2012, evidenced by the wear of phytoliths in the soil cultivated for 40 years (Fig 8B), reducing the Si content from the clay fraction.

Soil mineralogy at different times of soil cultivation
In the mineralogical evaluation of the clay fraction, the soil cultivated for 40 years showed an absence of muscovite, cristobalite, calcite and zeolite peaks in relation to the other treatments ( Figure 11). In addition, quartz peaks were found in greater quantities in this area (Figure 9d). The kaolinite peaks decreased their intensity while the quartz peaks increased their magnitude over the years of cultivation ( Figure 10). The silt fraction showed no difference in mineral composition between the native eld and cultivated soil ( Figure 12).
The increase in the presence of quartz peaks occurs in areas cultivated for 40 years, due to its compact arrangement, it has high resistance to weathering, and is therefore the most abundant mineral in soils and in the sand and silt fractions, and often in the clay fraction (Allen and Hajek 1989;Drees et al. 1989;Inda Jr. et al. 2006). Si in soils is present in several minerals, mainly crystalline silicates, such as quartz, plagioclase, feldspar, minerals rich in clay and amorphous silica (Sauer et al. 2006). The partial dissolution of minerals, such as muscovite and zeolite solubilizes structural elements like K and Al, transforming the minerals into quartz.