DIVISÃO 1-SOLO NO ESPAÇO E NO TEMPO Comissão 1 . 1 – Gênese e morfologia do solo BOuNDAry BETwEEN SOIL AND SAPrOLITE IN ALISOLS IN ThE SOuTh Of BrAzIL

Despite numerous studies conducted on the lower limit of soil and its contact with saprolite layers, a great deal of work is left to standardize identification and annotation of these variables in the field. In shallow soils, the appropriately noting these limits or contacts is essential for determining their behavior and potential use. The aims of this study were to identify and define the field contact and/or transition zone between soil and saprolite in profiles of an Alisol derived from fine sandstone and siltstone/claystone in subtropical southern Brazil and to subsequently validate the field observations through a multivariate analysis of laboratory analytical data. In the six Alisol profiles evaluated, the sequence of horizons found was A, Bt, C, and Cr, where C was considered part of the soil due to its pedogenetic structure, and Cr was considered saprolite due to its rock structure. The morphological properties that were determined in the field and that were different between the B and C horizons and the Cr layer were color, structure, texture, and fragments of saprolite. According to the test of means, the properties that support the inclusion of the C horizon as part of the soil are sand, clay, water-dispersible clay, silt/clay ratio, macroporosity, total porosity, resistance to penetration, cation exchange capacity, fe extracted by DCB, Al, h+Al, and cation exchange capacity of clay. The properties that support the C horizon as a transition zone are silt, Ca, total organic C, and fe extracted by ammonium oxalate. Discriminant analysis indicated differences among the three horizons evaluated.


INTrODuCTION
The Alisol class in the international soil classification of the "World Reference Base for Soil Resources" (FAO), is characterized by the presence of an argillic horizon that has a cation exchange capacity (CEC) value of 24 cmol c kg -1 clay or greater, associated with base saturation of less than 50 % in most of the argillic horizon (IUSS Working Group WRB, 2014).The WRB Alisols are mainly included in the "Argissolo" classes of the Brazilian Soil Classification System (SiBCS), according to Embrapa (2013), and are considered a broadly distributed soil with intermediate agricultural potential.
The effective soil depth in Alisols is not generally a limiting factor for agricultural use since the depth usually exceeds 150 cm.However, in imperfectly drained soils developed from fine sediments, such as fine sandstone and siltstone/claystone in southern Brazil, soil vertical development is less, reaching depths of approximately 60 to 100 cm (Pedron et al., 2012).Thus, the contact or transition zone between soil and saprolite is more important.In soils developed from sandstone parent materials, thinness of the soil requires proper management to prevent environmental damage to the aquifer.Stolt and Baker (1994) considered identification of the boundary between soil and saprolite as a complex task due to the diversity of processes that modify the transition from rock to soil.Studies published by Lietzke and Weber (1981), O'Brien and Buol (1984), Rice et al. (1985), Stolt et al. (1991Stolt et al. ( , 1992)), Schoeneberger et al. (1995), and Stolt and Baker (1994) define "C" as a saprolite horizon.When saprolite maintains a rock structure, it is denominated "Cr".The most widely used soil classification systems and morphological guides in Brazil and around the world, such as Schoeneberguer et al. (2012), Embrapa (2013), Santos et al. (2013), USDA (2014), and the IUSS Working Group WRB (2014), follow the same idea, but they separate the C horizon and layer, designating saprolite as the "C layer".
According to Stolt et al. (1991) and O' Brien and Buol (1984), in some situations, the advanced stage of modification results in a C horizon structure with greater similarity to soil than to rock.In this case, due to evidence of the pedogenesis structure, authors such as Pedron et al. (2009Pedron et al. ( , 2010Pedron et al. ( , 2011) ) classified the C horizon as part of the soil, differing from a Cr layer that represents saprolite at different stage of alteration, but always with the structure of the parent rock.
The same situation has been observed in Alisols of southern Brazil, where it is very common to find a horizon that has a pedogenetic structure, differing from the saprolite layer (Cr) and also from the B horizon.This has been named the "C horizon", and it is not contemplated in field morphological guides, which only mention the "C horizon"that is little affected or unaffected by pedogenic processes or with strong gleying (Santos et al., 2013;USDA, 2014).
The horizons/layers designated as C and/or Cr, when located near the surface, play a critical role in the environment (Schafer et al., 1979;Lietzke and Weber, 1981).However, in many soil surveys, these layers are not adequately described (Tandarich et al., 1994), and the field soil description guides (Schoeneberger et al., 2012;Santos et al., 2013) do not clearly state the procedures for identification of the soil-saprolite boundary.This may lead to an incomplete description that compromises evaluation of land use potential.The aims of this study were to identify and define the field contact and/or transition zone between soil and saprolite in the Alisols derived from fine sandstone and clayey siltstone in subtropical southern Brazil for the purpose of contributing to knowledge of the horizons that make up this contact/transition area.Subsequently, field observations and laboratory data were validated through multivariate analysis, highlighting the soil properties that might help pedologists in identifying those horizons and layers in the field.

MATErIAL AND METhODS
The area under study is in the central region of the State of Rio Grande do Sul (RS), Brazil, in the municipalities of São Pedro do Sul and Santa Maria (Figure 1).The altitude ranges from 132 to 160 m a.s.l, with a predominantly softly undulating to undulating relief.The geology of the area is dominated by sedimentary rocks of the Sanga do Cabral Formation, which consists of fine sandstones, subarcosians to arcosians, and the Santa Maria Formation (Alemoa member) composed of clayey siltstones, both in the presence of carbonate concretions (Sartori, 2009).The climate is Cfa -humid subtropical without a dry season with an average annual rainfall of 1,708 mm and an average annual temperature of 19.2 °C (Maluf, 2000).
Table 1 provides landscape data and soil classification according to the Brazilian Soil Classification System (Embrapa, 2013) and the WRB -IUSS/FAO (IUSS Working Group WRB, 2014).The six profiles, identified as P1 to P6, were sampled in natural pasture areas, and all of the soils had shallow depth and somewhat poor drainage.
The landscape characteristics and morphology of the profiles were evaluated according to the field description guides of Schoeneberguer et al. (2012) and Santos et al. (2013).The contacts/transitions between soil and saprolite were initially identified in the field through observation of the structure, color, and volume of fragments of the altered parent material.
The particle size distribution of the fine earth fraction (<2 mm) was determined by the pipette method as described in Donagema et al. (2011), using 1mol L -1 NaOH as the chemical dispersant.Water-dispersible clay (WD clay) was determined by the same method without the use of a chemical dispersant.The particle size distribution of the coarse fraction (>2 mm), performed by dry sieving, the silt-to-clay ratio, the degree of flocculation (DG), and hydraulic conductivity (Ks) were all calculated according to Donagema et al. (2011).
After the values of Ks were determined, macroporosity, microporosity, and bulk density were measured in the same samples.The water contents corresponding to the saturation point and water tension of 6kPa were assumed to be equal to total porosity (TP) and microporosity, respectively.Macroporosity was determined by the difference between total porosity and microporosity, and bulk density (BD) was obtained by the ratio of the mass of dry matter (kg) and its volume (dm 3 ) (Donagema et al., 2011).
Resistance to penetration (RP) was determined using an impact penetrometer with a conical tip at a 30° angle.The final values of RP were obtained using Holland's equations, suggested by Stolf (1991).
The water-stable aggregates were analyzed in undisturbed soil samples collected in each horizon and layer of the six profiles.The procedure followed the proposal of Kemper and Chepil (1965).The fraction that passed through the 8-mm sieve and was retained in the 4.76-mm sieve was used to calculate the percentages of water-stable aggregates >2 and <2 mm.
Chemical analyses were performed on the fine earth fraction (<2 mm).The pH was determined in water using the soil:water ratio of 1:2.5.Calcium and Mg contents were determined by atomic absorption spectrophotometry (AAS) after extraction with a 1.0 mol L -1 KCl solution.Exchangeable sodium was extracted with 0.05 mol L -1 HCl and determined by flame photometry.Potential acidity (H+Al) was determined in extracts of 1.0 mol L -1 (pH 7.0) Ca(OAc) 2 and titrated with 0.0606 mol L -1 NaOH.Exchangeable Al was extracted with 1.0 mol L -1 KCl solution and titrated with 0.025 mol L -1 NaOH.Total organic carbon (TOC) was determined by wet oxidation with 0.4 mol L -1 K 2 Cr 2 O 7 and titrated with 0.1 mol L -1 Fe(NH 4 ) 2 .(SO 4 ) 2 .These analyses followed the procedures of Donagema et al. (2011).
From the analytical data, the potential cation exchange capacity of the soil (CEC) was calculated by the sum of bases (Ca+Mg+K+Na) and H+Al, and clay reactivity (Tclay) was calculated by (CEC × 100)/% clay, as proposed by Donagema et al. (2011).
The soil samples were also subjected to extraction of Fe with H 2 SO 4 concentrate in the ratio 1:1 (Donagema et al., 2011), with tribasic sodium dithionite, sodium citrate dihydrate, and sodium bicarbonate (DCB) (Mehra and Jackson, 1960), and with a 0.2 mol L -1 ammonium oxalate (AO) solution in acid medium (pH 3.0) in the dark (McKeague and Day, 1966).The Fe concentration in the H 2 SO 4 , the DCB, and the AO extracts were determined by AAS.
Chemical dissolutions were performed on the samples of fine earth to determine the K content, according to Jackson et al. (1986).Exchangeable K (K ex ) contents were estimated using the Mehlich-1 extractant (0.05 mol L -1 HCl + 0.025 mol L -1 H 2 SO 4 ) and the levels of non-exchangeable K (K st ) were estimated by the 1.0 mol L -1 HNO 3 method adapted by Melo et al. (2000).
Initially, chemical and physical soil properties were evaluated through the values of average and standard deviation in each horizon/layer, and then the differences of each property were evaluated using a one-way ANOVA test, followed by the Tukey test (p≤0.05)or Dunnett T3 test (p≤0.05)when the data showed a non significant or significant heterogeneity, respectively, by the Levene test (p≤0.01).The Kolmogorov-Smirnov test (p≤0.05)was used to verify the normality of the data.
To conduct the Fisher multiple discriminant analysis (MDA), the database was standardized using the procedure of mean=0 and standard deviation=1 (Manly, 2005).Two MDAs were run using all of the independent variables (chemical and physical), and two MDAs were run using independent variables selected by the Wilks Lambda test (λ) and the F statistic (Brown and Wicker, 2000).All of the calculations for obtaining discriminant analysis phases can be found in Manly (2005).SPSS 13.0 software was used for discriminant analysis.

rESuLTS
The sequence of the horizons and layers was A, E, Bt, C, and Cr for the P3 and P4 profiles and A, Bt, C, and Cr for the P1, P2, P5, and P6 profiles (data from the A and E horizons not shown).The profiles showed depths ranging from 90 to 118 cm from the top of the A horizon to the upper limit of the Cr layer.Color changes occurred mainly with increasing depths, varying from red to yellow.The horizons showed clear and abrupt boundaries, with horizontal boundaries ranging from smooth to irregular to wavy (Table 2).
The moist consistency ranged from firm in the Bt horizons to friable in the C horizon and Cr layerin the profiles derived from Clayey siltstone.In the profiles derived from fine sandstone, the moist consistency in Bt was predominantly firm to friable, and very firm to extremely firm in C and Cr.As for wet consistency, in the profiles derived from fine sandstone, sticky and plastic conditions predominated for the Bt horizons, and non plastic and non sticky for the Cr layers. (1)Relief: SU -softly undulating (3-8 %); U -undulating (8-20 %). ( 2) Parent Material: CS -clayey siltstones (Sanga do Cabral Formation); FS -fine sandstone (Santa Maria Formation).
The presence of red fragments, vestiges of saprolite in the B and C horizons, ranged from 5 to 25 %, while in the Cr, the values ranged from 70 to 90 %.The size of the fragments also increased with depth in the profile.The quantity of roots ranged from many to few in the Bt and C horizons, and were rare in the Cr.The predominant structures were angular and subangular blocks in the Bt horizons; and angular to massive blocks, typical of a geogenic structure, in the Cr layers (Table 2).The C horizons showed an intermediate morphology of predominating angular blocks.
Analysis of variance for the chemical variables (Table 3) showed significant differences among the three horizons/layers (B, C and Cr) for TOC, Ca, and Fe AO .The averages of the Cr layers differed from those of the other horizons (B and C) for Al, H+Al, CEC, Tclay, and Fe DCB .For the Mg variable, the B horizon differed from the Cr layer, while the C horizon was not different from the other horizons.For the variable K ex , the C horizon differed from Cr, while the B horizon was similar to the other horizons.
For the physical variables, ANOVA reported significant differences among the three horizons only for BD.The Cr horizon differed significantly from the B and C horizons for seven variables (sand, WD clay, clay, macroporosity, RP, TP, and Ks).The C horizon differed from the others for aggregates >2 mm, while the C horizon differed from Cr for microporosity.There were no differences in the silt averages.
The multiple discriminant models performed for the chemical properties (MDA1 and MDA2) generated similar results.The main discriminant functions (FD1) had high Eigenvalues, explaining the percentage of variance (greater than 95 % power of discrimination between groups) (Table 4).For the physical property models, MDA4 had higher variation, explained by its first Eigenvalue (81.9 % power of discrimination between the groups) than that of MDA3 (76.4 %).The physical and chemical properties selected may have good discriminatory potential due to the high correlations for the canonical discriminant functions with the properties in all four models.
In MDA2, the stepwise method indicated that six of the 13 chemical properties best classified the soil layers.The properties that contributed most to distinguishing the groups were, in descending order of importance, TOC, Ca, Fe DCB , Mg, Al, and Fe H2SO4 (Table 5).In MDA4, the stepwise method selected six of the 14 physical variables used for discrimination of the soil horizons.The most important properties were BD, FD, RP, microporosity, and wet aggregates (>2 mm and <2 mm).Considering the whole group of properties, the models selected 12 physical and chemical properties with high power for distinguishing the three horizons in the soils evaluated.
In the two-dimensional perceptual maps (Figure 2), it can be seen that the discriminant model for both the chemical and physical properties showed a clear separation of the three groups.It likewise evidenced the similarity between points of the B and C horizons and C horizon and Cr layers by the distribution of samples.Another noteworthy aspect in this figure is the similarity between the B and C horizons, both for chemical and physical properties.

Identification of the boundary between soil and saprolite in the field
Initially, the A, E, Bt, and Cr horizons/layers were identified based on field morphology; and an apparently transitional layer between the Bt horizon and Cr layer was verified in all of the profiles.Initially, this layer was called the "C horizon" due to differences in the color, structure, texture, root penetration, and amount of saprolite fragments.
The Bt horizon showed strong to moderately developed structures and angular and subangular blocks with an increase in clay content, characterizing the lessivage process (Quénard et al., 2011).The predominant moist color was yellow, and the amount of saprolite fragments was smaller than in the underlying layers.In the C horizon, a clear grayish color, with moderate and angular blocks predominated, with up to 25 % of saprolite fragments.In the Cr layer, the predominant color was red, due to the occurrence of red saprolite above 70 % in volume.The modified structurefrom angular to massive blocks, with a larger size  DF1 3.506 81.9 (81.9) 0.882 DF2 0.775 18.1 (100) 0.661 (1) MDA1 and MDA3: all chemical and physical soil properties used in the model; (2) MDA2 and MDA4: using properties selected by the stepwise procedure.DF: discriminant functions.Fabrício de Araújo Pedron et al. than in the overlying horizons -is characteristic of geogenic processes (Lietzke and Weber, 1981;Stolt et al., 1991).
Morphological features, such as transitions between horizons and the soil consistency, showed greater variation, which was difficult to interpret.However, the tendency was reduction in the degree of wet consistency with depth; and transitions were smooth between the Bt and C horizons and irregular to wavy between the C horizons and Cr layers.Irregular and wavy transitions are common among less developed layers, as registered in the Cr layers, reflecting variability of alteration in the weathering processes (Stolt et al., 1993;Pedron et al., 2009Pedron et al., , 2010)).Thus, although incipient, a major variation in the C horizon can be interpreted as resulting from pedogenetic processes.

Statistical analyses of the chemical and physical data
Physical and chemical analyses are in agreement with field morphology.Discriminant analysis showed that the B and C horizons and the Cr layers can be distinguished based on chemical and physical properties.However, ANOVA analysis (Table 3) indicated similarities between the B and C horizons in regard to some chemical and physical properties.
The MDA specified which properties were most important in distinguishing between the layers and in determining how they differ from each other by a linear combination of p-dependent properties.The use of chemical and physical properties selected in the stepwise system allowed for a better fit of the classification and distinction of layers evaluated in MDA2 and MDA4.However, all of the other properties (not used by stepwise) did not significantly influence the classification, as indicated by the power of MDA1 and MDA3to represent the variability of the data for both groups of chemical and physical properties (Tables 4 and 5).
The average test results for sand, clay, WD clay, silt/clay ratio, microporosity, TP, and RP indicate that the C horizons are closer to the B horizons than to the Cr layers.The reduction in BD, RP, and Ks in the C horizons compared to the Cr layers is the greatest alteration, matching the structural development and its smaller size and the presence of biochannels, which contribute to increased porosity, water flow, and root penetration (Pedron et al., 2011).
The mean Ks show the similarity between the B and C horizons and their difference from the Cr layers.However, this difference was not statistically significant, due to high standard deviation values, especially in the C horizons and Cr layers.These results corroborate those of O' Brien and Buol (1984) and Schoeneberger et al. (1995).The low conductivity of saturated layers in all of the Cr layers is associated with a significant reduction in macroporosity and TP and an increase in BD.
The resistance of aggregates decreased for the >2mm aggregates from the Bt to the C horizon, possibly related to the decrease in organic matter and Fe oxides (Hickman et al., 2011;Vezzani and Mielnickzuk, 2011).The >2 mm aggregates increased from the C horizons to the Cr layers as a result of the increased volume of saprolite, which has a higher natural resistance (Stolt et al., 1991).
The properties CEC, Fe DCB , Al, H+Al, and Tclay statistically differed between the C horizons and Cr layers, and were similar in the C and B horizons.The other chemical properties, such as Ca, TOC, and Fe AO , showed differences from B to Cr, separating C as transitional.
The concentration of illuvial Fe in the Bt horizons match the lessivage process.The K values showed variation within the profile, with a significant reduction from Cr to C, and an increase from C to Bt.However, no significant differences were found between the mean values in C when compared with the mean values in Bt and Cr for either K st or K ex .Considering the high amounts of mica and K-feldspar in the fine sandstone and the rapid alteration of these minerals in tropical and subtropical environments, it was expected that the dissolution data of K would be a good indicator of the soil-saprolite boundary, which was not observed by the test of means or by the discriminant models.

Boundary between soil and saprolite
The data presented in this study suggest two perspectives.In the first, the C horizon provides stronge vidence of pedogenesis, based on morphological, physical, and chemical properties that have more similarities to the B horizon than to the saprolite.The second indicates that the C horizon is a transitional zone between saprolite (Cr) and soil (B), separated from both but with some mixed properties.
The data obtained for Alisols in the South of Brazil revealed that the C horizon should not be considered as saprolite as it does not show morphological or physical and chemical properties compatible with that layer, validating the results of Pedron et al. (2009Pedron et al. ( , 2010)).In this study, the C horizon was identified according to clear evidence of a pedogenic process, which is not in agreement with the description found in the guides used to describe and classify soil profiles (Schoeneberguer et al., 2012;Santos et al., 2013;IUSS Working Group WRB, 2014;USDA, 2014).
This situation was indicated by most of the morphological and physical properties used to identify saprolite in the field.These properties have practical applicability in the field related to water and root penetration, which are inferred by properties such as morphology (color, structure, texture, and roots), density, porosity, excavation and penetration strength, and hydraulic conductivity.If the C horizon is part of the soil, its lower limit becomes the lower limit of the soil, also called the soil-saprolite contact.
However, as indicated by some chemical properties, if C is considered a separate horizon as a transition between soil and saprolite, the profile no longer displays the lower limit or soil-saprolite contact, but rather represents a transition zone, as proposed by Lietzke and Weber (1981) and Stolt et al. (1993).In both perspectives presented in this article, the concept of the C horizon is different.So, considering our limited database, further studies should be conducted to verify the representative nature and practical applicability of the perspectives in the field.
In the case of the Alisols studied, where the saprolite is close to the surface, it performs important environmental functions (Lietzke and Weber, 1981).According to Schafer et al. (1979), Sternberg et al. (1996), andPedron et al. (2011), the saprolite layers of shallow soils may have equal or greater water retention and availability to plants than the soil itself.Thus, considering the C horizon as part of the soil and, additionally, as part of the solum would demand greater detail in its morphological description in soil surveys.
Landscape complexity limits clear identification of the soil lower limit, as highlighted by Tandarich et al. (2002) and USDA (2014).However, validating morphological and physical procedures that can be used to separate these layers in the field and better distinguish the soils are fundamental for technical and taxonomic soil interpretations.The results of this study suggest that the C horizonin Alisols can be identified in the field by the presence and development of soil structure and root penetration into the aggregates because, in saprolite, root penetration is limited to pre-existing fractures.Fabrício de Araújo Pedron et al.
The C horizons and the Cr layers included in this study showed distinct natures.The C horizons showed clear evidence of pedogenesis, which was not the case in the Cr layers.In addition, there are all the morphological, physical, and chemical differences perceived in the field and validated by statistical tests.In this case, designation of both with the letter "C" becomes inadequate.The authors of this study suggest the use of the letter "Cr" for saprolite, as cited in Tandarich et al. (2002).However, as already mentioned, more studies are needed to verify the representative nature of these data in terms of occurrence in the field.

CONCLuSIONS
The results indicated two situations: the C horizon developed with strong evidence of pedogenesis, showing morphological, physical, and chemical behaviors similar to those of the B horizon; and the C horizon is a transition zone between saprolite (Cr) and soil (B), independent of both.
The morphological variables analyzed in the field that showed differences between the B and C horizons and Cr layer were color, structure, texture, roots, and fragments of saprolite.The variables that support inclusion of the C horizon as part of the soil are sand, clay, WD clay, silt/clay ratio, macroporosity, TP, RP, FD, CTC, Fe DCB , Al, H+Al, and Tclay.The variables that support the C horizon as a transition zone are silt, Ca, TOC, and Fe AO .
Multiple discriminant analysis was effective in separating the three layers studied, and the discriminant functions of the chemical variables classified the samples with an accuracy greater than 93 %, while the discriminant functions of the physical variables classified the samples with an accuracy greater than 85 %.
The discriminant models generated by the stepwise method suggested the variables TOC, Ca, Fe DCB , Mg, Al, Fe H2SO4 , BD, RP, microporosity, and wet aggregates (>2 and <2 mm) for validating the morphological data in the field in identification of the limit between soil and saprolite.
figure 1.Localization of the study area.

figure 2 .
figure 2. Biplot for the fisher discriminant model for all the chemical (a) and physical (b) properties.within each frame: B and C: horizons, and Cr: saprolite layers.

Table 3 . Average, standard deviation, and ANOVA test of the soil chemical and physical properties
* rejected the hypothesis of normality of the data by the Kolmogorov-Smirnov test with p≤0.05.s significant properties for heterogeneity of variances by Levene's test (p≤0.01), the others showed homogeneity of variances (Levene p>0.01).Lowercase letters in the same row mean differences in means by the Tukey test (properties with homogeneity of variances) or Dunnett T3 test (properties with heterogeneity of variance) (p≤0.05).