Structural and Chemical Attributes of Soil Inorganic Fractions for Environmental Characterization in the Brazilian Semiarid Region

Soil structure affects plant growth factors in agroecosystems. The relationships of these factors with soil texture and chemical attributes are important to differentiate environments. The objective of this work was to evaluate the relationship of soil structural attributes with physical and chemical attributes of different soils to differentiate environments, using multivariate statistics. The research was carried out in the Terra da Esperança Settlement Project, in Governador Dix-Sept Rosado, RN, semiarid region of Brazil. Ten soil profiles were evaluated: Haplustepts in a conventional area (1CA); Haplustepts in a collective area with (2CA); Haplustepts in an agroecological area with native forest (3AGRO); Eutrustox in a native forest (4NF); Calciustolls in a conventional area of pasture (5CAP); Haplustepts in a native forest (6NF); Haplustepts in an native forest (7NF); Usticfluvents in a permanent conservation area with Licania rigida (8PPA); Haplustepts in a native forest (9MN); and Haplustepts in a collective area with Spondias spp. (10SP). Ten soil samples were collected and analyzed by horizons in volumetric rings and in soil blocks (aggregate analysis), with three replications. Structural, physical, and chemical attributes were evaluated. The data were subjected to multivariate statistical analysis, using correlation matrix, cluster analysis, and factorial analysis considering the factors as principal components. According to the factorial analysis, Factor 1 (F1) and Factor 2 (F2)—F1 consisting of clay, Ca, Na, sum of bases, and base saturation (V), and F2 consisting of cation exchange capacity (CEC), total organic carbon, macroporosity, particle density, K, and microporosity—were essential to differentiate the environments. The cluster analysis formed four groups. The structural groups showed greater similarity, denoting the relationship between source material and land use, followed by the chemical groups P, K, CEC, Ca, V, and pH for the Calciustolls; the structural weighted mean diameter, geometric mean diameter, soil aggregates larger than 2 mm, plasticity index, and plasticity limit for the Eutrustox; and the physical and chemical attributes EC, Na, Silt, macroporosity, total calculated porosity, aluminum saturation, total sand, volumetric moisture, Mg and particle density for the Haplustepts and the Usticfluvents. The clay fraction presented correlation with chemical attributes (F1) for the distinction of soil classes and land uses. The soil classes presented eutrophic character (V ≥ 50%) as a function of the Na and Ca contents, including the Eutrustox (4NF), despite its lower contents when compared to the other soil classes. This is a local peculiarity due to the climatic pattern of the Brazilian semiarid region.

However, despite the soil attributes in this region, concise information of the natural resources and their potentialities and limitations, considering intrinsic characteristics, the environment, and local people is needed for a conscious use of these resources; this can subsidize the implementation of activities, considering the potential of the local physical environment, and land uses to meet production purposes, and ensure soil conservation (Lepesh, 2015).
The structural arrangement of the soil particles can be modified by human actions.It is connected to plant growth, which affects porous space, flows of air, water, and heat to plant roots, nutrient supply, development of the root system, and soil mechanical resistance to penetration, which are important factors for to the development of agricultural crops (Machado & Mielniczuk, 2009).
Thus, characterizing the structural and chemical attributes of the soil and their correlations is necessary to subsidize decision making to improve and maintain the soil productive capacity, considering its potentialities and limitations.
According to Castilho (2017), multivariate analysis has been used successfully in studies involving soil attributes to identify and distinguish environments and know the most sensitive attributes for their distinguishing, considering simultaneously several characteristics, and correlations between them.
In this context, the objective of this work was to evaluate the relationship of soil structural attributes with physical and chemical attributes of different soils to differentiate environments in the Terra da Esperança Settlement Project, in the western mesoregion of the state of Rio Grande do Norte, Brazil, using multivariate analysis.

Materials and Methods
The research was developed in the Terra da Esperança Settlement Project, in the municipality of Governador Dix Sept Rosado (05°27′32″ S and 37°31′15.6″W), which is in the Chapada do Apodi microregion, western mesoregion of the state of Rio Grande do Norte, semiarid region of Brazil.The climate of the region is hot semiarid (BSw'h), according to the Köpper classification, with annual average precipitation of 712 mm from February to May (Alvares et al., 2013).The native vegetation of the region is hyper-xerophilic Caatinga.
The environments were characterized by sampling deformed and undisturbed soil samples in different horizons.They were placed in plastic bags and analyzed in the Laboratory of Soil, Water, and Plant Analyses of the Center for Agrarian Sciences of the Federal Rural University of the Semi-arid Region (LASAP-CCA-UFERSA), following the methodology proposed by Teixeira et al. (2017).
The granulometry was obtained by the pipette method using chemical dispersant (sodium hexametaphosphate) and distilled water in 20 g of the air-dried fine earth, with slow mechanical agitation (Wagner 50 rpm) for 16 hours.The sand fraction (2 to 0.05 mm) was quantified by sieving, clay (< 0.002 mm) by sedimentation, and silt (0.05 to 0.002 mm) by the difference between the total sand and clay fractions.Particle density analysis was carried out using the volumetric flask method described by Kiehl (1979), oven-dried fine earth at 105 °C, and ethyl alcohol.The soil consistency tests were evaluated based on LL, with the use of the Casagrande device.
PL was evaluated with four replications, using a representative sample of the central part of the shear of the soil in the metallic sphere of the equipment from the determination of the liquidity limit, which formed a compressed sphere on a glass plate, resulting in a cylindrical rod of 3.0 to 4.0 mm in diameter without breaking or flowing.PI was evaluated by the difference between LL and PL.Gravimetric moisture was evaluated in the plasticity condition for the soil rods.
Undisturbed samples were collected using Uhland type device, and rings with height of 0.05 m and diameter of 0.05 m.Ten samples were collected per horizon in the soil profiles.Soil density, macroporosity, microporosity, and total porosity were evaluated.Soil density was evaluated using a volumetric ring with known volume, as described by Forsythe (1975), with averages represented by the quotient between the soil solid particle weight and the total volume of the soil, and expressed in kg dm -3 .Macroporosity, microporosity, and total porosity were evaluated using undeformed samples in volumetric rings; they were saturated for 48 hours and weighed (total porosity).The tension table method was used, according to Kiehl (1979), considering 6 kPa stress (microporosity), with the total soil porosity (Pt) evaluated by measuring the soil moisture saturation according to equation: Pt = [(Msat -Ms)/VT] × 100, wherein MSat is the saturated soil weight; Ms is the dried soil weight; and VT is the sample volume.Microporosity was evaluated by the water volume retained in the soil for the h value equal to 60 cm of water (pore radius of approximately 25 μm).Soil macroporosity was evaluated by the difference between total porosity and microporosity.
Soil blocks from the horizons of each soil class were sampled and sieved in 4.0 mm and 2.0 mm mesh sieves, preserving the aggregate structure for the study of the aggregates, using units of weakness.The wet sieve method, with mesh sieves of 4.76, 2.00, 1.00, 0.50, and 0.25 mm and a vertical oscillation device with 42 oscillations per minute, was used according to Kemper and Rosenau (1986).The soil aggregates were separated by shaking in water, and the samples were dried in an oven at 105 °C; the sand fraction was discarded, and the aggregate size distribution, percentage of aggregation, and stable aggregates in water, weighted mean diameter (WMD), and geometric mean diameter (GMD) were evaluated.
Deformed samples were used for the chemical analyses.The attributes evaluated were pH in water, electrical conductivity (EC) in water, TOC by digestion of organic matter, and macronutrients-exchangeable calcium (Ca 2+ ) and exchangeable magnesium (Mg 2+ ) using potassium chloride as extractant, and phosphorus (P), sodium (Na + ) and potassium (K + ) using Mehlich 1. Subsequently, the cation exchange capacity (CEC) and base saturation (V) were evaluated and interpreted according to the Manual of Recommendations for Use of Correctives and Fertilizers of Minas Gerais (Ribeiro & Guimarães, 1999).
The structure attributes of the deformed samples were evaluated with four replications; they were subjected to statistical analysis, using multivariate analysis to identify the more sensitive attributes for differentiating the environments, considering the different land uses, using the Statistica 7.0 program (Stansoft, 2004).
The analytical results from the correlation matrix and multivariate analysis were based on cluster analysis (CA), factorial analysis (FA) and principal component analysis (PCA).The CA was represented by a dendrogram of the distance matrix, using the Euclidean distance to measure similarity, and the Ward's method to connect the cases.The importance of each variable in the CA for differentiating the environments was measured as a function of its smaller distance from the reference axis, x or y, and the axis that contains the largest value of accumulated variance that explains most of the causes of variance (Sá Paye et al., 2012).
The established connecting distance in the dendrogram from the cluster analysis was 12, which formed four large groups of soils (Figure 1).The FA showed the contribution of each variable to each factor; the factors were defined by differentiating the attributes of the environments.The environments were distinguished using six diagrams of the principal components (Factors 1 and 2, Factors 3 and 4, and Factors 5 and 6) considering physical and chemical attributes.These data were used to create two-dimensional diagrams to distinguish areas and vector projection diagrams to identify sensitive soil attributes for differentiating the environments (Hair et al., 2009).
The pH of the samples presented significant correlations with electrical conductivity (EC), Ca 2+ , V, and Als.These correlations are explained by the soil classes, geological formation of the study region, and pedogenetic processes.Thus, the high correlations between pH and EC, Ca 2+ , and V denoted the effect of these variables on the high alkalinity levels found, especially in the Calciustolls.EC represents the amount of soluble salts in the exchange complex and is direct proportional to the number of ions; high amount of exchangeable bases increases soil alkalinity, which hinders the development of crops (Ribeiro et al., 2007).
Ca 2+ was significant due to the local lithology.According to Angelim et al. (2006), the Chapada do Apodi region is in the Jandaíra geological formation, constituted basically of limestone, which is rich in carbonate rocks that provide high amounts of calcium and magnesium that contribute to the soil alkalinity.Mota et al. (2008) characterized soils of the Chapada do Apodi and found effects of its lithology on the pH, denoting that soils from limestones have neutral to alkaline pH because of their high carbonate contents, which affects exchangeable bases, as also observed by Marinho et al. (2016).
The pH and Als presented correlation because aluminum is soluble in a wide pH range; Als represents the proportion of soluble aluminum to exchangeable bases and aluminum in the CEC of the soil.These variables are inversely proportional, with a lowest critical pH value of 5.5 for Al 3+ becoming more soluble, generating loss of exchangeable bases and silica by leaching, predominating the pH generated by the Al 3+ and H+Al released from the rocks to the soil solution (Perin et al., 2010) due to the latolization process, typical in Eutrustox soils.
TOC showed significant correlations with P, K + , Ca 2+ , and CEC, contributing to differentiate Group IV (Table 1, and Figure 1), characterized by the Calciustolls.These correlations are explained by the contribution of TOC to these nutrients.Moreover, K + is not metabolized by plants but incorporated into the plant biomass, forming connections with organic complexes of easy reversibility and, consequently, rapid release rate by plant residues (Perin et al., 2010).
The high correlation between P and SOM is due to the cycling of this nutrient in the soil by microbial activity due to the plant energy support, in the labile and non-labile forms, especially in surface horizons, with accentuated effect in areas of native forest due to constant deposition of plant residues (Turrion et al., 2010).The contribution of TOC to P release is more significant in weathered soils in tropical regions, where 1:1 clay minerals of low colloidal activity present high adsorption force, favoring complexation of organic and inorganic forms of P in the soil matrix, making this element little labile, resulting in low fertility (Machado et al., 2012).Thus, organic P affects the availability of this element in the soil.Pereira et al. (2011) evaluated light SOM and remaining P for agriculture purposes and found increased light SOM content in areas under no-tillage system with cover plants, and reduced adsorption of P because of the adsorbed SOM, or blocked adsorption sites that occur on the surface of clays with low colloidal activity.Leite et al. (2017) reported the contribution of SOM to labile organic P in the soil; the maintenance of plant cover on the area of cultivation (unburned sugarcane plants) generated higher organic P contents and other forms of P (total P, available P, inorganic P, and P in humic substances) when compared to the areas with burned sugarcane plants.Note.WMD = weighted mean diameter; GMD = geometric mean diameter; AGR>2mm = soil aggregates larger than 2 mm; LL = liquidity limit; PL = plasticity limit; PI = plasticity index; TPd = total porosity determined; TPc = total porosity calculated; Microp = microporosity; Macrop = macroporosity; TS = total sand; Dp = particle density; θ = volumetric moisture; EC = electrical conductivity; TOC = total organic carbon; CEC = cation exchange capacity; V = base saturation; Als = aluminum saturation.
In the dendrogram generated by the cluster analysis (Figure 1), the reading is made from right to left, in which the y axis indicates the distances between the formed groups, and the x axis represents the groups joined in descending order of similarity.A high dissimilarity indicates that two individuals are distinct from the set.The cluster analysis formed four groups, with Groups I and IV formed from a single soil class (I from Haplustepts; and IV from Calciustolls), while Group II were formed from Haplustepts and Eutrustox, and Group III from Haplustepts, and Usticfluvents.The physical attributes were determinant for the dissimilarity of the horizons between soil classes for Groups I, II, and III, and chemical attributes were determinant for Group IV (Figure 1).
Group I was represented by areas with native forest, where the consolidation of the surface due to absence of soil turning of Haplustepts affected directly the physical attributes WMD, TPd, microporosity, aeration, clay, AGR>2mm, and the consistence limits (PL, LL, PI), agreeing with the results of the correlation matrix (Table 1) and the dendrogram (Figure 2).These results were similar to those of the Haplustepts due to active clay fraction of the soil; different types of clay confer different characteristics of cohesion and adhesion, affecting structural attributes.These properties were affected by the origin material of the Haplustepts, which belong to the Jandaíra formation, conferring them expressive adhesion and cohesion forces.These forces are dependent on the soil moisture, consistency, and aggregation, and affect the structural stability, depending on the clay content and type by the formation of primary aggregates; and the most stable aggregates (secondary aggregates) are dependent on the energy input (SOM) through microbial activity (Brady & Weil, 2013).
WMD, Dp, Mg 2+ , volumetric moisture (θ), TS, and Als were significant for Group II (Figure 2), with predominance of sand fraction in the surface layer, which was more expressive in the Eutrustox.This result is explained by the intense weathering that occurs in this soil class, compared to the others.These correlations are due to the clay fraction, which affects various physical attributes and influences the soil structure.Ghidin et al., (2006) evaluated the correlation between clay fraction and structural properties and reported that the mineral content in the clay fraction is more important to defining the soil structure and porosity than its crystallographic characteristics.Similarly, Vasconcelos et al. (2010)   Factorial analysis showed that most of the variance (F1 = 26.72%)were due to interrelations between inorganic fractions (especially clay fraction) and chemical attributes (Tables 1 and 2, and Figures 1 and 2).Similarly, Souza et al. (2018) reported clay, total sand, Ca 2+ , Mg 2+ , OM, and SB as the variables that contributed the most to discriminate environments.Macroporosity, microporosity, Dp, K + , and CEC was significant for F2, presenting positive factorial loads (17.30%);WMD, geographic mean diameter (GMD) and AGR>2mm was significant for F3, describing 12.39% of the data of aggregation of soils; and PI, TPc, TS were the most significant (10.13%) for F4.The other factors (F5 and F6) were less expressive, presenting lower factorial loads (7.39% and 4.73%, respectively) for the grouped attributes θ, pH, EC, and PL, and were affected by F1 and F2.
The inorganic fractions (sand, silt, and clay) directly affected the chemical (F1) and structural (F2) attributes; different clay contents were essential to differentiate the environments, explaining 26.72% of the causes of variance.The sum of the accumulated variances (F1 and F2) explained 44.03% of the causes of variance, therefore, they were the factors that contributed the most to distinguish the studied environments.Note.Factorial axes rotated by the Varimax method.

Conclus
The physic (GMD), so (θ), and pl organic ca for the di macroporo The chemical attributes P, K + , CEC, Ca 2+ , V, and pH were the most sensitives for the differentiation of the Calciustolls, whereas the physical attributes WMD, GMD, AGR>2mm, PI, and PL distinguished the Eutrustox, and the physicochemical attributes EC, Na + , silt, macroporosity, PTc, aluminum saturation, total sand, θ, Mg 2+ , and Dp distinguished the Haplustepts and the Usticfluvents.
The clay fraction presented interrelation with the chemical attributes (Factor 1), distinguishing the soil classes and land uses depending on their origin material, and local peculiarities.
The soils classes evaluated presented a eutrophic character (V ≥ 50%), including the Eutrustox, especially due to the Na + and Ca 2+ , despite its lower contents when compared to the other soil classes, which denotes a local peculiarity because of the climatic pattern of the semi-arid region.

Figure
Figure Figure 4 , and the c K + , cation exch of the soil cl TOC, K + , and plastic butes Na + , Ca 2 y (CEC), pH, a nd uses.How e

Table 1 .
Correlation matrix for soil attributes in the study areas.Terra da Esperança Settlement Project, Governador Dix-Sept Rosado, Rio Grande do Norte, Brazil emphasized the contribution of different types of clay to the description of soil physical attributes.

Table 2 .
Factor axes extracted for soil attributes and their respective factorial loads, eigenvalues, and total and accumulated variances