Introduction

One of the great highlights of the Amazon biome is its great potential in biodiversity, where it has an extensive forest, which houses a fauna and flora, and becomes a great attraction for researchers in the search for the characterization of the dynamics of this ecosystem and can thus gauge its various characteristics. as well as extraction of alternative raw materials for the various industrial activities (Melo et al. 2017).

The Amazon region is located in the northern part of South America, with about 6 million km2 (Vale Júnior et al. 2011). It has the largest tropical forest in the world, being the largest reservoir of biodiversity on the planet and where natural environments still persist. A member of the Amazon region, the southern Amazon region occupies approximately 474,021 km2 of its total area, equivalent to 30% of the state of Amazonas. Among the municipalities that make up the southern region of Amazonas, Humaita is highlighted as the municipality where the junction between the BR 230 and BR 319 occurs, serving as support for travelers from various regions of the country and abroad.

Until the mid-1960s, the exploitation of Amazonian resources was extractive. However, afterward, exploration intensified to integrate the region into the productive and economic process of the country. Livestock farming, one of the most developed economic activities in land use, contributed to the large deforestation and to the cultivated pastures for cattle rearing, and agriculture managed with cutting and burning, causing negative environmental impacts, sometimes even irreversible (Dutra et al. 2000; Rivero et al. 2009).

With the population increase there is a great concern about the world production of food to meet this need (Arraes et al 2012), so it is opted for the advance of agriculture over natural forests (Silva et al. 2015), ie conversion of a natural environment into In this area of cultivation, with this activity comes the doubt of the adverse environmental impacts caused to these environments, mainly related to biodiversity, considering that changes in the environment cause loss of vegetation, change in animal habits, as well as changes in soil attributes, however, At the level of degradation there is still little information (Domingues et al. 2017).

The inadequate use of the soil for agricultural practices directly affects the physical quality of the soil, causing significant modifications in the organic matter content and degradation of soil attributes (Silva et al. 2005; Sá et al. 2010). Thus, knowledge of changes in soil attributes caused by the various anthropogenic uses provides aid for the adoption of management practices that allow to increase the yield of the productive process, with the conservation of the environments, ensuring that future generations can use these environments way we use today, that is, to provide "sustainable development". In this sense, the evaluation of soil attributes in natural and anthropized environments becomes necessary to find subsidies and propose management techniques that aim to minimize and even inhibit possible environmental changes.

Despite the great importance of multivariate statistical methods for interpretations of variations in soil attributes, few studies have used this tool since most use univariate statistical methods (Silva et al. 2010b). However, some studies have applied multivariate techniques to evaluate soil variables and found satisfactory results (Campos et al. 2012b, Pragana et al. 2012; Campos et al. 2013; Oliveira et al. 2013; Mantovanelli et al. 2015, Soares et al. 2016, Cunha et al. 2017).

The proposal of the use of multivariate statistics is justified by having a greater capacity to describe the intra and interdependence relations in agricultural systems ( Marques, 2009). With the multivariate analysis, it is possible to explain the maximum correlation between the variables and find out which ones contribute most to the evaluation of soil quality. In the simultaneous analysis of any information, this technique becomes the best tool, allowing to obtain data and interpretations that could not be perceived with the use of univariate statistical analysis (Cruz and Regazzi 2001). Therefore, the objective of this research was to evaluate the physical attributes and organic carbon of soils in natural environments and anthropic uses in the southern region of Amazonas.

Material and methods

The study was carried out in five rural properties located in the south of Amazonas, more precisely in the municipality of Humaitá (Fig. 1). In these properties, four environments with natural characteristics (native forests—NF) were selected, serving as natural environments, numbered from 1 to 4 for differentiation (NF1, NF2, NF3 and NF4) and also five environments with different anthropic uses, such as pasture—P (natural environment: NF1), agroforestry—AF, cassava—C, açaí—A and reforestation—RF (chart 1).

Fig 1.
figure 1

Location of the environments used in the study, in Humaitá, Amazonas.

Full size image

The environment with pasture (P) is formed with Brachiaria brizantha (cv. Marandu) and several tucumã plants (Astrocaryumaculeatum) scattered over the pasture, with about 20 years of implantation, maintained with a low stocking of cattle. The environment with agroforestry (AF) where species such as Andiroba (Carapaguianensis), Cupuaçu (Theobromagrandiflorum), açaí (Euterpe oleracea), Brazil nut (Bertholletia excelsa), cocoa (Theobromacacao), pupunha (Bactrisgasipaes) and tucumã (Astrocaryumaculeatum). The AF environment has access to small animals (pigs and birds), created in an extensive way. The environment cultivated with cassava (C) (Manihotesculenta) is with approximately two years of cultivation, where the practice of harrowing occurred before planting. The environment with açaí (A) (Euterpe oleracea) cultivation with the beginning of cultivation in the year 2010, the area has anirrigation system and frequently receives cover fertilization. The environment with reforestation (RF) being implanted in 2004 for the cultivation of Teak (Tectonagrandis L.), Mahogany (Swieteniamacrophylla King.), Andiroba (Carapaguianensis Aubl.), Jenipapo (Genipa Americana L.) and Brachiariabrizantha (Marandu cv.) pasture between the lines of these species, characterized at the time as a silvopastoral system. However, currently, the environment does not have pasture, for this reason, was identified as an environment with reforestation.

The soil source material from the region comes from the alluvial sediments, which are chronologically derived from the Holocene (BRASIL 1978). The climate in Amazonas is equatorial (hot and humid), with relative air humidity ranging from 76 to 89% and average temperatures of 22.0–31.7 ºC, having two well defined seasons: winter, considered the period of the rainfall and summer, dry season or less rainy season (Maia and Marmos 2010).

Soil samples were collected in nine environments, four natural environments (forest fragments) and five environments in different anthropic uses (pasture, agroforestry, cassava, acai, and reforestation). The surveys were carried out between November 2016 and March 2017. Thus, trenches were opened at depths of 0.00–0.05, 0.05–0.10 and 0.10–0.20 m, collecting soil blocks with a minimally altered structure, packed in identified plastic bags, and collected samples with preserved structure, in duplicate, with the aid of volumetric cylinders (69 cm3), and packed in a thermal box so as not to alter the structure of the soils sampled. For each study environment, four replications were performed at randomly selected points. At the end of the sampling, 108 lightly altered soil samples (soil clods) and 108 soil samples were obtained, in duplicate in each layer (to analyze the best of the two), with a preserved structure (volumetric cylinders), being sent to the laboratory where the physical and organic carbon analyzes were carried out. The collect points had their coordinates registered with the aid of a Garmin Global Positioning Satellite (GPS) equipment (GPSmap 64S).

The textural (sand, silt and clay) components were determined by the pipette method. Basically, the time for vertical displacement in the soil suspension with water was fixed after the addition of chemical dispersant (NaOH 0.1N) and slow stirring for 16 h in Wagner-type mechanical stirrer apparatus, with rotation adjusted to about 50 rpm. A 50 ml volume of the suspension was pipetted to determine the clay which was weighed after oven drying. The coarse fractions (fine and coarse sand) were separated by sieving, oven dried and weighed to obtain their bulk fractions. The silt was obtained by difference of the other fractions in relation to the original sample of 20 g of soil (Embrapa 2011).

For the determination of soil density (SD), macroporosity (MAP) and microporosity (MIP) and total porosity (TP), the samples collected in volumetric rings were saturated by gradual elevation until two-thirds of the ring's height from a water slide on a plastic tray. After saturation, the seeds will be weighed and taken to the stress table to determine the soil MIP, being subjected to a tension of − 0.006 MPa. After reaching equilibrium at a matrix potential of − 0.006 MPa, the samples will be re-weighed and then the sample will go to the penetrograph where the penetration resistance of the soil (SRP) was determined, finally the samples were taken to the greenhouse at 105 °C for 16 h, and after removal and cooling in desiccator, these are weighed again, and then obtaining the data will be determined to SD and TP, by the volumetric ring method, and the MAP was determined by the difference between VTP and MiP (Embrapa 2011).

To determine the stability of aggregates, soil blocks with the preserved structure were used, air dried, fragmented into smaller clusters manually and passed through a 9.51 mm mesh screen, using the aggregates retained in the sieve of 4.76 mm. Separation and stability of the aggregates were determined according to Kemper and Chepil (1965), with modifications in the following diameter classes: 4.76–2.0 mm; 2.0–1.0 mm; 1.0–0.50 mm; 0.50–0.25 mm; 0.25–0.125; 0.125–0.063 mm. The aggregates were placed in contact with water on the 2.0 mm sieve and subjected to vertical shaking in Yoder apparatus (Solotest, Bela Vista, São Paulo, Brazil) for 15 min and with 32 oscillations per minute. The material retained in each class of the sieves was placed in an oven at 105 ºC, and then measured the respective masses on a precision digital scale (Embrapa 2011).

The results were expressed as a percentage of the aggregates retained in each of the sieve classes and the stability of the aggregates evaluated by the weighted average diameter (WAD) obtained by the formula proposed by Castro Filho et al (1998), and the geometric mean diameter (GMD), according to Maia and Marmos (2010), cited by Alvarenga et al (1986), according to the equations:

$$\mathrm{W}\mathrm{A}\mathrm{D}=\frac{{\sum }_{i=1}^{N}{n}_{i}{D}_{i}}{\sum {n}_{i}}$$
(1)
$$\mathrm{G}\mathrm{M}\mathrm{D}={10}^{\frac{{\sum }_{i=1}^{N}{n}_{i}log{D}_{i}}{\sum {n}_{i}}} (2)$$
(2)

where ni is the percentage of aggregates retained in a given sieve, Di is the average diameter of a given sieve and N is the number of sieve classes.

For the determination of soil density (SD), macroporosity (MAP) and microporosity (MIP), total porosity (TP) and soil gravimetric moisture (SM), samples collected in volumetric cylinders were prepared and saturated with a gradual elevation of a water slide, up to two-thirds the height of the ring, in a plastic tray. After saturation, the samples were weighed on a digital scale and taken to the tension table to determine the soil MIP, being subjected to a matrix potential of − 0.006 MPa (Embrapa 2011).

After reaching equilibrium at a matrix potential of − 0.006 MPa, the samples were again weighed and then the measurements of soil resistance to penetration (SRP) were made using an electronic bench pedometer (MA-933, Marconi, SP, BR), with a constant velocity of 0.0667 mm s−1, 4 mm diameter base cone and 30° semiangle, receiver and interface coupled to a microcomputer to record the readings.

Afterward, the samples were taken to the oven at 105 ºC for determination of the soil moisture (SM), SD and MAP by the volumetric cylinder method, and the TP was determined by the sum of MAP with MIP (Embrapa 2011).

Total organic carbon (OC) was determined by the Walkley–Black method, modified by Yeomans and Bremner (1988). The organic matter is determined by the OC product with 1724 (Embrapa 2011). The carbon stock (CS) is defined by the equation below:

$$\mathrm{C}\mathrm{S}= Sd\times h\times \mathrm{C}\mathrm{O}\mathrm{T}$$
(3)

where: CS = carbon stock (Mg ha−1); Sd = soil density (g cm−3); h is the thickness of the soil layer sampled (cm); COT = C content (%).

After the determination of the physical attributes and soil organic carbon, univariate and multivariate statistical analyzes were performed. The univariate analysis of variance (ANOVA) was used to compare attribute means individually using the Scott-Knott test at 5% probability, comparing all the environments and then comparing the environments in anthropic uses with their respective natural environments. These analyzes were conducted using a spreadsheet and ASSISTAT 7.7.

Then, the factorial analysis of the main components was performed to find statistical significance of the sets of soil attributes that more discriminate the study environments, obtaining as answer which are the environments whose attributes are most influenced by the anthropic action. These statistical analyzes were performed using the statistical software STATISTICA 7.0 (StatSolft 2004).

The adequacy of the factorial analysis was made by the KMO (Kaiser–Meyer–Olkin) measure, which evaluates the simple and partial correlations of the variables, and the Bartlett sphericity test, which is intended to reject the equality between the correlation matrix and identity. The extraction of the factors was done by the analysis of the main components (PCA), incorporating the variables that presented commonalities equal or superior to five (5). The choice of the number of factors to be used was made by the Kaiser criteria (factors that have eigenvalues greater than 1). To simplify the factorial analysis, orthogonal rotation (varimax) was performed and represented in a factorial plane of the variables and the scores for the main components. In the scatter plot of the PCA after varimax rotation, the scores were constructed with standardized values such that the mean is zero and the distance between the scores are measured in terms of the standard deviation. In this way, the variables in the same quadrant (1°, 2°, 3° and 4°) and closer to the dispersion graph of the PCA are better correlated. Likewise, scores attributed to the samples that are close and in the same quadrant are related to the variables of that quadrant (Burak et al. 2010).

Results and discussion

The environments with NF1, P and AF presented a silt franc texture, with sand contents varying between 11 and 21%, silt between 60 and 70% and clay between 16 and 24%, in the three depths (Table 1). The other environments presented a clay texture, with sand contents between 10 and 27%, silt between 20 and 47% and clay between 29 and 65%. In the depth of 0.00–0.05 m, there were no significant differences in the sand contents between NF4 and C, as they presented significantly similar averages, as did also between NF1, NF2 and, NF3. At the depth of 0.05–0.10 m, the environments with P and NF4 had similar values, as did C and AF and also as NF1 and NF3. For the depth of 0.10–0.20 m, the environments with NF4 and C were statically equal, as well as P and NF2, A and AF and also NF1 and NF3. The proximity to the texture values of these three environments (NF1, P and AF), besides being close environments, can be explained by having a direct relation of relief (CAMPOS et al. 2012a). Similar results were found by Martins et al. (2006) and Campos et al. (2012a), in studies carried out near this region.

Table 1 Soil texture (sand, silt and, clay) in natural environments and anthropic uses in the southern region of Amazonas, Brazil
Full size table

The lowest clay fractions were in AF at all three depths. There were no significant differences between the environments with A and NF4 and between P and AF in the depth of 0.00–0.05 m, NF2 with C and NF1 with P in the depth of 0.05–0.10 m, and NF2 with C at the depth of 0.10–0.20 m. The high silt content is justified by the alluvial nature of the sediments that make up the original material (BRASIL 1978). According to Rosolen and Herpin (2008), which have already studied soils in the region, the occurrence of small depressions in soil topography favors the movement and deposition of thinner sediments into lower relief parts. According to Campos et al. (2012a) and Santos et al. (2012), in studies with soils in toposequences under alluvial terraces in the region of the middle Madeira river, found high levels of silt, with values close to 600 g kg−1, corroborating that high levels of silt are common in the soils of the region.

In the comparison of the environments in anthropic uses with their respective natural environments (Table 1), it was verified that there were significant differences in the texture of most environments, with tendencies of increases in sand content and reduction of the finer particles in the most superficial layer. although the texture is considered a constant attribute in the medium and long term, variations such as those of the environments in anthropogenic uses can happen due to the natural action and, to a lesser extent, by human action, and there may be changes in the contents of clay, suggesting its vertical migration since this parameter is influenced by the texture and the organic carbon in the soil (Araújo et al., 2011; Reinert and Reichert 2006).

The majority of the environments presented SD greater than 1.50 g cm−3 (Table 2). The exception was for the environments with NF1, P, NF2, C, and, NF4 in the depth of 0.00–0.05 m and NF1 in the depths of 0.05–0.10 m and 0.10–0.20 m, which presented values lower than 1.50 g cm−3. It can be seen from the data in Table 2 that most environments increased SD in relation to their natural environments, except for the environment with C that did not present significant differences in relation to NF2 in all depths, and the environment with A, which had no differences at depths of 0.05–0.10 and 0.10–0.20 relative to NF3. The increase in SD generally indicates an environment with greater resistance to root growth, reduction of aeration and reduction in soil hydraulic capacity. The decrease in SD values may be related to the increase of organic matter in the soil. In this way, the low soil density may be related to the high levels of organic carbon and of intense biological activity (fauna and roots), that construct canals, cavities and galleries in the subsoil (STEINBEISS et al. 2009; Soares et al. 2016).

Table 2 Humidity, density and soil resistance to penetration in natural and anthropogenic environments in Humaita, Amazonas
Full size table

In this study, the natural environments presented an increase of SD as the depth progressed (Table 2), with this understanding that this effect can occur naturally for all the environments studied. This is probably due to the pressures exerted by the upper layers on the lower ones, which cause their compression and reduction of porosity (Cunha et al. 2011). However, studies have already shown that SD has an increase due to its intense use coupled with mechanization and inadequate soil management, promoting the degradation of soil physical quality, commonly identified by the increase in compaction level ( Soares, 1992; Marchão et al. 2009; Collares et al. 2006; Bergamin et al. 2010; Soares et al. 2015), mainly in soils with high clay contents (SECCO et al. 2004), being a mechanical impediment for the growth of roots, affecting the development of the plants (Bergamin et al. 2010).

The highest SRP was recorded for the RF environment, followed by P and A, with significant differences in the three depths in relation to their natural environments (Table 2). There was a correlation with the increase in SD and MAP decrease. Increases in SRP and SD are associated with soil compaction, which is usually caused by the intensification of their uses and management. In the case of RF, there may be an association with the previous use of the environment as pasture, where animal traffic caused soil compaction in the most superficial layer. Despite this, these values are below the limit value of 2.0 MPa for compacted soils, defined by Tavares Filho and Tessier (2009). However, Soares et al. (2015), in TPA soils under pasture, verified values above 2.0 MPa in the depth of 0.00–0.05 mm, which, according to the authors, increased due to soil compaction by trampling animals.

A different effect of the other environments in AF and C, in the first layer, was observed in relation to their natural environments (NF1 and NF2, respectively), where there was a decrease in SPR, probably due to the increase in sand content and the development of roots of the plants in the first case, and in the second case, because of the use of harrowing before planting the cassava, which breaks the surface layer of the soil making it less compact. The increase of the SPR for the environment with C was associated to the increase of the MAP in the depth of 0.00–0.05 m, a fact that is not observed in the 0.10–0.20 m layer, in which the increase occurs of the SPR. According to Silva et al. (2005), soil tillage usually promotes a temporary increase in macroporosity. This fact may indicate the onset of the phenomenon called "foot-of-grid", where the most superficial layer of the soil is revolved and, at the same time, the counterbalance below forces and compacts the deeper layers, very common where this type is practiced of management.

MAP values lower than 0.10 m3 m−3, such as those occurring in the three depths in NF 3 and A, mainly in AF, NF−4 and RF (0.00–0.05 m), in NF 2 (0.05–0.10 m) and in NF2, C and NF4 (0.10–0.20 m) (Table 3) indicate critical values with probable limitations to soil aeration in wetter times (Baver et al. 1972; Pagliai et al. 2003; Bergamin et al. 2010). According to Feng et al. (2002), values equal to or very close to that in clayey soils may already cause inhibition to the adequate supply of oxygen to the plants, being ideal values higher than 0.10 m3 m−3 of MAP. According to Lima et al. (2007), soils considered ideal have values equal to or greater than 0.50 m3 m-3 of total porosity, in which the microporosity would oscillate between 0.25 and 0.33 m3 m−3 and the macroporosity between 0.17 and 0.25 m3 m−3. Adverse conditions of TP are found in environments with AF, A and RF, in the depth of 0.00–0.05 m, and the environments with P, A and RF in the depth of 0.05–0.10 m. In the depth of 0.10–0.20 m only the environments with NF1 and AF had TP equal to or greater than 0.50 m3 m−3 (Table 4).

Table 3 Macroporosity, microporosity and total porosity in natural environments and anthropic uses, in the southern region of Amazonas, Brazil
Full size table
Table 4 Geometric mean diameter and weighted mean diameter of soil aggregates in natural environments and anthropic uses in the southern region of Amazonas, Brazil
Full size table

In Table 3, it was observed that the environment with RF was the one that presented the greatest difference of TP in relation to its natural environment (NF4), before the other environments, probably because of the influence of trampling of animals in the period when it was used in the silvopastoral system. The reduction of the total pore volume in pasture areas may be a reflection of the reduction of macroporosity, promoting a possible increase in soil density and microporosity, as well as a possible decrease in water infiltration rate, especially in the more superficial layer (Salton et al. 2002; Giarola et al. 2007; Goulart et al. 2010).

For the smaller aggregates (< 1 mm), the highest percentage was found in the environment with NF 3 (33.94%) and the lowest with AF (2.91%) in the upper layer. At the depth of 0.05–0.10 m, the highest and the lowest percentage continued with the same environments, with 58.90% for NF 3 and 7.49% for AF. In the depth of 0.10–0.20 m, the highest value was in the environment with NF 4 (70.02%) and the lowest value was in the environment with C (16.66%). The surface horizons are usually characterized by the rounded granular structure that presents a hierarchy in which aggregates > 2 mm are composed of smaller aggregates (Bronick and Lal 2005). In Table 5, it was clear that the percentage of aggregates > 2 mm has a downward trend in depth, while aggregates < 1 mm tend to increase with in-depth feed. The explanation for this is related to the OM content in soils, since there is a positive correlation between OM and aggregates > 2 mm and negative between OM and aggregates < 1 mm.

Table 5 Stability of aggregates in natural environments and anthropic uses in the southern region of Amazonas, Brazil
Full size table

The conversion of natural environments to anthropogenic uses favored the alteration of the stability of aggregates in all study environments and in most of the studied depths, observing that there was a significant reduction in the larger aggregates (> 2 mm) only in the environments with P and C, in the depth of 0.00–0.05 m. This decrease in aggregate size may have been due to decreased OM (Table 6) and a possible increase in SD (Table 2).

Table 6 Organic carbon, organic matter and soil carbon stock in natural environments and anthropic uses in the southern region of Amazonas, Brazil
Full size table

In Table 6, in the depths of 0.00–0.05 and 0.05–0.10 m, the AF environment presented the highest OC content, with 28.84 g kg−1 and 21.75 g kg−1, respectively. At the depth of 0.00–0.05 m, the RF environment presented the lowest OC content (10.63 g kg−1). In the 0.05–0.10 m depth, the environments with the lowest OC contents were A (9.77 g kg−1) and RF (9.87 g kg−1), without significant differences between them. At the depth of 0.10–0.20 m, the highest OC content was found in P (18.74 g kg−1) and the lowest NF3 content (8.20 g kg−1), followed by RF (8.42 g kg−1) and A (8.81 g kg−1), their means being considered equal by the test. Soils with less than 2% of organic carbon may be considered erodible and soil erodibility decreases linearly with OC content (LIU et al. 2010). Other literature considers the content of 40 g kg−1 as a critical limit for OM ( EMBRAPA, 2013).

Only the environments with NF1, P and AF presented values higher than the critical limits for both OC and OM. The structural stability of the aggregates decreases when inadequate management practices result in the reduction of OM content for most soils (Paul et al. 2013). CS values were influenced by both OC and SD values. However, the environments with higher or lower CSs were the same with OC and OM. Thus, the same environments that presented higher or lower OC contents also presented higher or lower CS levels, that is, they were proportional.

In the comparison of the environments in anthropic uses with their natural environments, it was observed that there were significant differences in OC in AF at depths of 0.00–0.05 and 0.05–0.10 m in relation to NF1in A only in the depth of 0.05–0.10 m in relation to NF3, in P and C only in the depth of 0.10–0.20 m in relation to NF1 and NF2, respectively, and in RF in the three depths in relation to NF4. The maintenance of natural environments and environments with agroforestry depends on the recycling of the nutrients contained in litter and soil OM (Moreira and Malavolta 2004).

High OM contents maintain a well-structured soil with a balanced distribution of the particles (sand, silt, clay), with the appearance of pores where water and air are stored, constituting an ideal place for the development of the root system and plants. Probably, the increase of OM in the AF environment was due to the accumulation of plant residues (leaves, branches and roots) from the local plants, associated with ideal humidity and temperature, in the presence of decomposing microorganisms, resulting in the increase of OM content in this environment. environment. In the case of RF, the effect was contrary, since there was the suppression of the vegetation cover, with a higher percentage of vegetation, and the absence of anthropic uses (Siqueira Neto et al. 2009). insertion of a silvopastoral system, where there were soil erosion and pasture elimination due to inadequate management, promoting the low input of vegetal residues, which may justify the low levels of carbon in the soil. Silva et al. (2004) verified that pastures of low productivity favored the reduction of OM in the soil.

Soils with native forest present higher OM content, which gives them higher fertility due to the higher amount of organic residues (Morais et al. 2012). The greatest relation with OM is also due to the fact that it is directly associated with anthropic interference, without the use of agricultural implements and cultural practices, and does not degrade the stability of soil aggregates. According to Portugal et al. (2010) and Freitas et al. (2011), there is a decline in OM stock after the conversion of native forests into agricultural systems, as a consequence of increased soil erosion, mineralization of organic matter from the soil, and lower inputs of organic waste.

The highest values of carbon stocks (CS) were found in AF and P, in relation to the natural environment (NF1¬) and the other environments. According to Salimon et al. (2007), high values of carbon stock (CS) in pasture environments compared to native forest may occur due to the higher density and organic carbon content in these areas, and in some cases decreases in the first years of implantation, increasing in the following years until reaching values close to or greater than those existing before conversion, a process that probably occurred with the AF environment. Several studies have already verified a greater amount of CS in pasture environments in relation to the native forest.

Desjardins et al. (2004) and Araújo et al. (2011), verified that among these environments it is possible to observe that the evolution of the organic carbon of the soil obeys two simultaneous processes, one being the continuous mineralization of the carbon derived from the native vegetation due to the cycles of wetting and drying of the soil and the other to progressive incorporation of the carbon derived from the remnants of the crop introduced by the pasture, mainly by the grassroots. Souza et al. (2012), add that the high values of CS of pasture areas can occur in the presence of grasses that have an abundant root system and intense rhizosphere effect, where after their decomposition they release nutrients and contribute to the formation of soil OM, favouring their aggregation. Given these results, it is clear that the anthropic actions cause diverse alterations in the natural environments and that these alterations vary according to the uses and management affected by these environments.

The correlations between the studied soil attributes and the fractal dimension of the soil texture are presented in Table 7. According to Jakob (1999), the correlation analysis between variables shows the attributes that can be represented by others with a certain degree of correlation. tolerable information.

Table 7 Spearman's correlation coefficient (n = 108) between physical attributes, organic carbon and fractal dimension of texture in soils under natural environments and in anthropic uses, at a depth of 0.00–0.20 m in Humaitá, Amazonas
Full size table

In Table 7, among the attributes that showed highest correlation coefficient values, the OC had a higher correlation (p < 0.01) positive with silt (0.73) GMD (0.71), WAD (0.71) and with aggregates > 2 mm (0.69), evidencing that the OC has an important contribution in the aggregation of the soil particles, also presenting a negative correlation (p < 0.01) with SD (− 0.74), clay (− 0.73) and aggregates < 1 mm (− 0.71), confirming the importance of OC in soil structuring. These results are in agreement with those obtained by Cunha et al. (2017), working with different uses of Archaeological Black Earth (ABE) in the southern region of Amazonas.

The SD had negative correlation (p < 0.01) with MAP (− 0.75), TP (− 0.76), OC (− 0.74), SM (− 0.68) and silt (− 0.63), and positive (p < 0.01) correlation with clay (0.58) and (p < 0.05) with sand (0.23). This can be related to the different types of soils and uses of the environments, since the soils with greater clay content can be more easily altered, with influences of the handling practices (traffic of machines and animals) that can compact the soil, by the reduction of porous spaces and consequent increase in SD. Normally, SD correlates better with sand, as occurred in the work of Cunha et al. (2017), where SD had a better positive correlation with sand (0.48) and then with clay (0.17), an inverse effect to what occurred in the present work. This can be justified by filling the spaces between the sand particles by the clay, making the soil more susceptible to compaction.

Soil moisture had a higher positive correlation (p < 0.01) with TP (0.86), MIP (0.68), OC (0.46) and MAP (0.31), corroborating with the results of Cunha et al. (2017) and Bergamin et al. (2015) with MIP. The SM also had a higher negative correlation (p < 0.01) with SD (− 0.68) and with SPR (− 0.52), which were similar to the results obtained both by Cunha et al. (2017) and by Bergamin et al. (2015). However, SPR tends to decrease in moist soils, favouring the correlation obtained in this research. According to Silveira et al. (2010), soils with low water content, present particles closer and difficult to be separated, with the increase of SPR, being evident the power of lubrication of water in the soil. These results only confirm that, during the compaction, the porous spaces decrease and increase in SD and SPR and decrease in SM. This can be observed in Table 7, where MAP had a positive (P < 0.01) correlation with TP (0.63), OC (0.53) and silt (0.48) and correlation (p < 0.01) negative with SD (− 0.75), SRP (− 0.43) and clay (− 0.48).

Concerning soil texture, a negative correlation (p < 0.01) between clay and silt (− 0.95) and a positive correlation (p < 0.01) with OC (0.73) was observed. It can be attributed to the process of displacement of finer particles, both horizontally and vertically, influencing the distribution of soil particle size (texture), which was also verified in the work of Cunha et al. (2017).

Through the factorial analysis it was possible to verify that the results were significant (KMO = 0.74 and p < 0.05 for the Barlett sphericity test), indicating, therefore, that it was adequate for the evaluated attributes. For the principal component analysis (PCA), the number of factors to be extracted was established in such a way that they explained at least 70% of the total data variance (Table 8 and Fig. 2), which presented covariance matrix eigenvalues greater than one (1) (Manly 2008), with 6.47 at CP1 and 3.62 at CP2. From the percentage of variance explained, it was observed that CP1 is responsible for 46.18% of the total variance of 72.01%, while CP2 accounts for 25.83%, which was sufficient to explain the variability of the data originals, Since the components only work in two axes, in the two axes that present the largest variance, the remainder of the variation value is not significant for the analysis of principal components. It was verified that Oliveira et al. (2013) and Cunha et al. (2017) found values of variance above 70% in soil physical and chemical attributes, and these values can be attributed to the variability of these attributes. Both factors (CP1 and CP2) presented high coefficients of determination for textural characteristics, structural stability of the aggregates and organic carbon of soil study (Table 8).

Table 8 Correlations between each main component and analyzed variables and factorial analysis of soil attributes referring to a depth of 0.00–0.20 m, with rotational factors (varimax) (Factor CP1 and CP2) corresponding to the environments studied in the southern region of Amazonas, Brazil
Full size table
Fig 2.
figure 2

Proportion of variation in the data set explained by the main component (CP) and contribution of each variable to explain the total variance by the screeplot method. SM soil moisture, SD soil density, MAP macroporosity, TP total porosity, SPR soil penetration resistance, GMD geometric mean diameter, WAD weighted average diameter, OC organic carbon, FD fractal dimension

Full size image

Figure 2, with “screeplot”, graph, can also be used to verify the importance and contribution of each variable to explain the total variance. This graph together with the eigenvalues can be used to make the decision of how many components should be retained for later application of the principal component analysis (PCA). The weights of the attributes of each environment in the first and second retained components show that the most significant attributes for 72.01% of the variability explained in the 0.00–0.20 m depths were: silt (3.96%), clay (4.50%), SM (5.39%), SD (6.21%), MAP (4.22%), TP (6.48%), SPR (4.39%), GMD (6.10%), WAD (5.73%), > 2 mm (5.63%), 2–1 mm (4,09%), < 1 mm (5.62%) e OC (5.34%) (Fig. 2). Given that these attributes possibly have a greater impact or change, in relation to the other attributes.

Figure 3 shows the factorial plan of distribution of the scores of the different environments studied and the arrangement of the factorial loads of the soil attributes, collected at a depth of 0.00–0.20 m, formed by CP1 and CP2. For a geometric interpretation, the weights assigned to each variable correspond to the projections to each of the coordinate axes represented by the main components (Manly 2008).

Fig 3.
figure 3

Factorial plan of soil attributes collected at a depth of 0.00-0.20 m in soils under natural environments and in anthropic uses in the southern region of Amazonas, Brazil. Values are standardized so that the mean is zero and the distances between the scores are measured by the standard deviation. NF1 native forest 1, P pasture, AF agroforestry, NF2 native forest 2, C cassava, NF3 native forest 3, A-acai, NF4 native forest 4, RF reforestation

Full size image

It is observed a greater densification of the scores in the environments with NF1, P and AF distributed between the first and fourth quadrants, which discriminates these three environments in a significantly homogeneous group. Thus, soil samples collected in the environments under NF1, P and AF resulted in values for OC, silt, GMD, WAD and aggregate classes > 2 mm above the mean, in comparison to the other environments, positively correlated with to CP1, clay, aggregate classes 2–1 mm and < 1 mm below average, compared to the other environments, negatively correlated with CP1. On the other hand, the soil samples collected in the environments under NF3, A and NF4 are more distributed between the second and third quadrants, with attributes clay, classes of aggregates 2–1 mm and < 1 mm above the mean, compared to other environments, negatively correlated with CP1, and OC, silt, GMD, WAD, and aggregate classes > 2 mm below the mean and positively correlated with CP1.

The soil samples collected in the environments under NF2 and C were more distributed in the first quadrant and resulted in values for the attributes SM, TP and MAP above the average, in comparison to the other environments, positively correlated with CP2, and SD and SPR attributes below the mean, compared to other environments, negatively correlated with CP2. On the other hand, the soil samples collected in RF environments are more distributed between the third and fourth quadrants, with SD and SPR above average, compared to the other environments, negatively correlated with CP2, and the SM attributes, TP and MAP below the mean and positively correlated with CP2.

With this perspective, the characterization of the environments with NF2, C and RF were summarized in terms of the structural characteristics (SM, MAP, TP, SD and SRP), and the first two environments presented better values of SM, MAP and TP with low values of SD and SRP, whereas RF showed higher values in SD and SRP with lower values in SM, MAP and TP, which, in this case, may be associated to a higher level of compaction and resistance to soil rupture (Soares et al. 2015). In the environments with NF1, P, AF, NF3, NF4 and A, the characterization of these environments was summarized in terms of the texture, aggregate stability, organic carbon and soil texture, having higher values of OC, silt, GMD, WAD and aggregates > 2 mm the environments with NF1, P and AF, which presented lower values of clay and aggregates of smaller sizes (2–1 and < 1 mm), while NF3, NF4 and A presented the exact opposite.

It was possible to observe that the acai berry was in an area with greater similarity to all areas of the natural forest, configuring that few changes are being made due to the substitution of the natural environment. This fact is clearly associated with the fact that an acai is a palm tree native to the Amazon rainforest, that is, its ecosystem mechanism is the same as the general forest species of the region, and when and when a substituted being is displayed to the ground, it is suffering only since it is a cultivar with characteristics similar to the original vegetation (Yuyama et al. 2011).

It is also possible to verify that the data related to the soil aggregation correlate well with the pasture area, bringing good results, since this use is the most used machinery material for its implementation (Chioderoli et al., 2012), as well as As the attributes related to carbon and porous spaces were positively related to the cassava area, this can be justified by taking into account the rusticity of this cultivar, that is, it requires few cultural treatments related to soil preparation, thus freeing of factors. conditioning factors that affect porous spaces (Oliveira et al., 2013).

Conclusions

The different anthropic uses of the environments cause significant changes, both positive and negative, in the texture and structure of the soils when compared to their natural environments. The agroforestry environment presents the highest gains of organic carbon, organic matter and carbon stock up to 0.10 m depth and the environment with reforestation with the greatest loss up to 0.20 m depth, compared to its environments natural. The structural improvement of the analysed soils is closely related to the increase of organic carbon.

The environments with native forest 1, pasture and agroforestry are characterized by higher values of organic carbon, silt, geometric mean diameter, weighted average diameter and aggregates > 2 mm, while native forest 3, clay and aggregates between 2 and 1 mm and < 1 mm. The organic carbon has a positive correlation with silt, geometric mean diameter, weighted average diameter and aggregates > 2 mm, and negative with soil density and clay.

The environments with native forest 2 and with cassava are characterized by the highest values of soil gravimetric moisture, macroporosity and total porosity, while the environment with reforestation by the highest values of soil density and soil resistance to penetration.

What defines the degree of soil degradation that a particular crop can cause is the way it is planted, that is, when done incorrectly even a simple crop can cause serious damage to the soil, as well as others, crops that are totally different from the natural forest but properly implemented, can be the least harmful, such as pasture at work that most closely resembled its forest environment.