CARBON CONTENT IN AMAZONIAN OXISOLS AFTER FOREST CONVERSION TO PASTURE ( 1 )

Soil plays an important role in the C cycle, and substitution of tropical forest by cultivated land affects C dynamic and stock. This study was developed in an area of expansion of human settlement in the Eastern Amazon, in Itupiranga, State of Pará, to evaluate the effects of native forest conversion to Brachiaria brizantha pasture on C contents of a dystrophic Oxisol. Soil samples were collected in areas of native forest (NF), of 8 to 10 year old secondary forest (SF), 1 to 2 year old SF (P1–2), 5 to 7 year old SF (P5–7), and of 10 to 12 year old SF (P10–12), and from under pastures, in the layers 0–2, 2–5 and 5–10 cm, to evaluate C levels and stocks and carry out separation of OM based on particle size. After deforestation, soil density increased to a depth of 5 cm, with greater increase in older pastures. Variation in C levels was greatest in the top soil layer; C contents increased with increasing pasture age. In the layers 2–5 and 5–10 cm, C content proved to be stable for the types of plant cover evaluated. Highest C concentrations were found in the silt fraction; however, C contents were highest in the clay fraction, independent of the plant cover. An increase in C associated with the sand fraction in the form of little decomposed organic residues was observed in pastures, confirming greater sensitivity of this fraction to change in soil use.

When native vegetation is substituted by agricultural systems in tropical regions, SOM stocks can be greatly reduced as a consequence of physical disturbance of the soil, exposing the organic matter (OM) to intense microbial activity, which results in rapid decomposition of OM in the soil (Silva & Machado, 2000;Mielniczuk et al., 2003).In the Amazon, although in some cases forest substitution by pasture reduces SOM contents, (Desjardins et al., 1994;Trumbore et al., 1995), in other cases higher C levels were observed several years after pasture implantation, which exceeded even the contents observed in forest areas (Choné et al., 1991;Koutika et al., 1997;Desjardins et al., 2004).According to Desjardins et al. (2004), these opposite tendencies of soil C content development are caused by factors such geographical location and soil type, grass species, and cultural practices and management techniques.
The variation of soil C content has mostly been studied using physical fractionation techniques for OM, to separate fractions that differ from each other in nature, dynamics and function (Desjardins et al., 1994(Desjardins et al., , 2004;;Bird et al., 2000;Christensen, 2001;Roscoe et al., 2001).The less destructive nature of physical than chemical fractionation techniques, for being more related to in situ functions and structure of SOM makes them important tools for SOM dynamics analysis (Christensen, 2001;Rangel & Silva, 2007).Physical fractionation shows that OM dynamics are intimately related to soil texture (Martins et al., 1989;Feller & Beare, 1997;Silver et al., 2000;Desjardins et al., 2004).The proportion of organic C in sandy soil in form of slightly decomposed plant residues is therefore greater, making such soils more fragile from the point of view of land use, since this fraction, devoid of physical protection, is easily mineralized by microorganisms (Feller & Beare, 1997).On the other hand, greater C contents in the silt and clay fractions were verified in clayey soils (Desjardins et al., 2004) due to the greater stability of organic mineral compounds formed (Sánchez, 1981) making these soil fractions more persistent.Different organic compartments respond distinctly to soil use and management practices (Balesdent et al., 1998;Bernoux et al., 1999).In summary, the implantation of pastures by slash and burn agriculture in the Amazon region modifies soil properties, leading in particular to alterations in C stock distribution in the different soil particle-size fractions (Desjardins et al., 1994(Desjardins et al., , 2004;;Koutika et al., 1997;Bernoux et al., 1999).
Research on the cycle and dynamics of C in soil after substitution of forest by pasture in the Amazon region showed varied results depending on region, soil type and land use.The objective of this paper was to study the impact of forest substitution by pasture on C levels and distribution associated with different OM fractions in a dystrophic Oxisol in the Eastern Amazon region.

MATERIALS AND METHODS
The study was carried out in the Southeast of the State of Pará, microregion of Marabá, in the settlement Benfica (5 ° 16 ' 00 " S; 49 ° 50 ' 00 " W), at a distance of 80 km from the center of Itupiranga, PA, in the Eastern Amazon.The landscape of Benfica consists of elongated hills developed from a biotite monzogranite mother-rock of the Brazilian Shield (Paleoproterozoic).The soil in the study area is predominantly dystrophic Oxisol with clayey texture, which covers approximately 2/3 of the surface and occurs in areas with higher elevation.Cambissols are found on lower parts of slopes and Gleissols are found in low-lying areas (Simões, 2004).The climate of the microrregion Marabá is Awi, according to Köppen, corresponding to humid tropical climates with a definite dry season between May and September, average annual temperature of 26 °C and average annual precipitation around 2,000 mm (De Reynal, 1995).All studied sites were located in the upper part of the landscape, where dystrophic Oxisol with clayey texture occurs.
Soil sampling was carried out in May 2004, at depths of 0-2, 2-5 and 5-10 cm, in 15 sample areas selected at the tops of slopes (three areas for each soil management and use system), in a 100 m long transect perpendicular to the slope declivity at 20 sampling points to form composite samples that were air-dried, ground and filtered through 2 mm mesh sieve (airdried fine earth -ADFE).Soil management and use types were: (a) native forest (NF): a selectively logged, humid tropical forest area; this area was chosen as a reference for comparison of the study properties; (b) capoeira or secondary forest (SF): secondary forest of 8-10 years after slash and burn of original forest and cultivation of one year of annual crops (rice and cassava); (c) grass pastures (Brachiaria brizantha) for 1 to 2 years (P1-2), 5 to 7 years (P5-7) and 10 to 12 years (P10-12): areas used for extensive cattle-raising without control of the stocking rate, with a density of approximately 1 au ha -1 , using the conventional livestock system in which the original vegetation had been removed, and after burning, grass was planted by broadcasting seed.In this system no phytosanitary control, no liming or fertilization is applied (Biri Kassoum & Maitre d'Hotel, 2002).ADFE samples were utilized in textural analysis and particle-size fractionation of OM.The textural analysis was performed after OM destruction with H 2 O 2 , using the pipette method (Embrapa, 1997).For soil density analysis, six undisturbed samples were collected from each system and depth, using 2, 3 and 5 cm high volumetric rings.The mean soil chemical properties per depth of the acidic, dystrophic Oxisol poor in exchangeable bases under study are shown in table 1 (Embrapa, 1997).Particle-size fractionation of soil OM was performed by the method of Gavinelli et al. (1995), using 20 g of ADFE, initially dispersed by mechanical agitation in 400 mL water with 100 cm 3 soda resin (VETEC Amberlite Ira 120) for 16 h.The fraction > 53 μm, containing plant residues along with mineral particles, was wet-separated in a 53 μm sieve.The 0-2 μm and 0-53 μm fractions were separated by sedimentation in water.The silt + clay fraction (0-53 μm) was obtained by total collection of the suspension.The clay fraction (0-2 μm) was obtained by collecting aliquots from the suspension based on the sedimentation time of this fraction.All fractions were dried in a forced-air oven at 60 °C.
Total organic C was determined by dry combustion of a sample aliquot of each soil particle-size fraction, sieved through 0.149 mm mesh, in a CHNS elemental analyzer (LECO), and expressed in g kg -1 .The C stock in the different systems and at each soil depth was calculated based on the following formula: C stock (Mg ha -1 ) = C level (g kg -1 ) x Ds x T/10, where Ds = soil density at depth (g cm -3 ) and T = thickness of soil layer (cm).The silt fraction C (2-53 μm) was calculated as the difference between silt + clay fraction C (0-53 μm) and clay fraction C (0-2 μm).
A completely randomized design was used and the results for the variables studied were subjected to variance analysis of the effects of land use type for each depth.Averages were compared by Duncan's test (5 % error probability) using the statistical program SAEG 8.1.

Physical properties
No significant variation in soil texture within each layer of the studied areas was observed (Table 2).On these areas, located in flat places or with slight slopes, surface erosion which can occur very rapidly after forest substitution when rainfall intensity is high had no significant effect and did not affect the soil textural composition, providing good experimental conditions for a comparison of the texture-dependent variables studied, e.g., soil density and C content (Martins et al., 1990;Desjardins et al., 2004).The lowest and highest clay levels, respectively, were observed in the secondary forest (SF) and the 10 to 12 year old (P10-12) pasture area.The clay content varying from 320 to 500 g kg -1 soil confirmed the soil classification as clayey, except in the 0-2 cm layer of the SF and in the 1-to-2-year-old pasture (P1-2) areas (Simões, 2004).
Soil density (Ds) values varied from 1.05 to 1.45 g cm -3 (Table 2).In the surface layer, the lowest value occurred in native forest (NF), followed by SF and P1-2.Values were highest in the pastures P5-7 and P10-12, differing significantly from NF and SF areas.In the 2-5 cm layer, soil density of the native forest area only differed from pasture areas P5-7 and P10-12.In the 5-10 cm layer, there was no Ds variation in function of plant cover.Note also that Ds increased with greater depth under all soil use types, with greatest increases between the layers  Araújo et al. (2004) observed that in the surface layer of a Yellow dystrophic Ultisol in the State of Acre (Western Amazon region), soil density increased after substitution of native plants (Ds = 1.40 g cm -3 ) by pasture (Ds = 1.73 g cm -3 ).The authors attributed this increase to cattle trampling and the greater susceptibility of the soil to wetting and drying cycles than under native forest.Soil density increase in function of increasing pasture age planted in former forest areas was also observed by other authors in the Amazon region (Bernoux et al., 1999;Desjardins et al., 2004).

Particle size fractionation of organic material
There was no significant difference in the weight distribution resulting from soil particle-size fractionation for all of soil use types and layers (Table 3).Greatest variation was observed in the 2-53 μm or silt fraction, which is probably due to an incomplete dispersion of micro-aggregate particles with the soda resin used in the particle-size fractionation procedure.The method of Gavinelli et al. (1995) was only able to recover an average of 86 % of the clay fraction for the textural analysis, showing that a small part of the clay fraction remained in the silt fraction.This effect could have occurred on account of the linkage of clay particles or Fe and Al oxides and hydroxides to the humified OM, forming stable micro-aggregates with diameters > 2 μm, in which SOM would be protected from the attack by chemical and microbial agents (Feller & Beare, 1997;Balesdent et al., 1998;Christensen, 2001), making these micro-aggregates very stable and difficult to rupture.
Several studies on particle-sized fractionation of soil OM using similar or different fractionation methods drew similar conclusions (Desjardins et al., 1994;2004;Roscoe et al., 2001).Similarly, Balesdent et al. (1998) observed lower levels for finer soil fractions (< 20 μm) when OM was not destroyed in a study of OM content variability in areas of a forest-maize sequence, indicating that such fractions were not completely dispersed.

Carbon concentration in soil particle-size fractions
Carbon concentrations differ significantly in the particle-size fractions; highest values were found in the silt fraction (2-53 μm) at all depths studied (Table 4).In this fraction, the C concentration varied from 48.9 to 112.9 g kg -1 .In the sand fraction (53-2,000 μm), the C concentration was lowest and the reduction with increasing depth greatest, evidencing the effect of deposition of plant residues (leaves, branches, flowers) on the soil surface.On the other hand, the percentage of reduction in C concentration in this fraction with increasing depth was lowest in NF, probably as a result of a more uniform distribution of OM by organic particle transportation processes in the soil profile.
The reduction of C concentration with increasing depth in the clay fraction (0-2 μm) was the lowest in all fractions, with an average of 27 % for the 2-5 cm layer and 38 % for the 5-10 cm layer.This more gradual reduction was partly due to the greater stability of OM associated with clays, consisting of humified compounds which are chemically or physically protected and more resistant to mineralization (Tisdall & Oades, 1982;Desjardins et al., 1994;Feller & Beare, 1997).

Total C distribution in soil and particle-size fractions
The soils of the five land use systems did not differ largely within each fraction with respect to C content, whether for soil, corresponding to the sum of the three fractions, nor for each particle-size fraction (Figure 1).
In the 0-2 cm layer, the C content was lowest in NF.A value similar to native forest was found in P1-2 and there was a tendency of increasing C content with pasture age, as also observed in SF.In the 2-5 cm layer, this C increase was not repeated in the SF use system, and the tendency to increase with (AM), Choné et al. (1991) observed high C inputs derived from pasture into the total C stock, in the 0-3 cm layer.
On the other hand, Falesi (1976) and Desjardins et al. (1994) verified a reduction in C content after forest substitution by pasture in the Brazilian Amazon, corroborated by Desjardins et al. (2003) in the Peruvian Amazon after maize cultivation.These results indicate that variations in C content of soil under grass is mainly a function of the land use, as shown in the study of Fearnside & Barbosa (1998).
In the 53-2,000 μm fraction, substitution of forest provoked a significant change in C content in the surface layer only.The C content was lowest in NF, which differed significantly from all other land use systems.In SF, the C level was 1.7 times higher than in NF; in pasture use systems it was 1.5-1.9times higher in P10-12 and P5-7, respectively.However, no such effect was observed for finer fractions, whether in the surface nor in deeper layers.
The soil C content increase in pastures and secondary forest could be explained by the C increase in the sand fraction.This could be due to: (1) C increase caused by OM that is little humified, produced pasture age was less prevalent; in the 5-10 cm layer, the values for all land use systems were similar.
The C content reduction with depth was more pronounced in use systems under human influence, mainly pastures.The percent decrease in C content of the particle-size fractions between the soil surface and the 5-10 cm layer was lowest for the clay fraction: 7 % in P1-2 to 21 % in NF.For larger particle sizes reduction was greater: 45 % in NF to 66 % in P10-12 for the 2-53 μm fraction and 69 % in NF to 80 % in P1-2 for the 53-2,000 μm fraction.
There was an increasing tendency for soil C content after substitution of forest by pasture in all fractions studied.The tendency of C content to increase in the surface layer after forest conversion to pasture observed in this study was also verified by Trumbore et al. (1995), Bernoux et al. (1999), andDesjardins et al. (2004) in the Brazilian Amazon.In a study evaluating the effect of deforestation followed by pasture establishment, carried out in chronosequences of Brachiaria brizantha and Panicum maximum pastures on sandy soils in Rondônia, Moraes et al. (1996) verified that after 20 years, pasture soil C content was 17-20 % higher than in forest soil.In clayey soils in the region of Manaus  by the aerial part of the secondary forest and by elevated production, accumulation and cycling of grass roots (Moraes, 1995); (2) and/or by alterations of soil microbial activity, which was reduced in pasture areas compared with forest, decreasing OM decomposition and mineralization processes, as shown in the study of Melo (2007) for the study area.
Generally, substitution of forest-derived C by pasture-derived C occurs more quickly in the soil OM fractions than in finer fractions since the elevated cycling process of grass roots and reduction in decomposition processes favors accumulation of little decomposed fragments in the fraction > 53 μm (Feller & Beare, 1997;Desjardins et al., 2004).
The C distribution pattern in the particle-size fractions observed in this study was similar to results of Desjardins et al. (1994Desjardins et al. ( , 2004)), Borges & Kiehl (1996), Silver et al. (2000) and Roscoe et al. (2001).Generally, the contribution to total soil C was highest in the finest fraction.This contribution increased with soil depth, varying from 34 % in the surface layer to 67 % in the 5-10 cm layer.On the other hand, C content decreased as depth increased in soil fractions with larger particle sizes.The reduction was greatest in the sand fraction, diminishing from 19 % in the 0-2 cm layer to 8 % in the 5-10 cm layer.These results show a strong correlation between soil texture and C content, with a higher C content in finer-textured soils (Desjardins et al., 2004;Luca et al., 2008).

Total soil carbon stock
In native forest soil, 5.0 Mg ha -1 C are stocked in the 0-2 cm layer (Figure 2).Carbon stocks in SF and P1-2 were practically equal (≈ 6.2 Mg ha -1 ), with a slight increase of 1.2 Mg ha -1 in relation to NF.However, there was an increasing tendency of C stock with pasture age; the soil C stock of P10-12 in this same layer increased significantly in relation to NF, reaching 8.0 Mg ha -1 .In the layers 2-5 and 5-10 cm, for pasture use systems, soil C stock tended to increase with age, although there was no significant effect of land use system on C stock.In the surface layer, there was an increase of 60 % in C stock for P10-12 soil compared with NF which decreased to nearly the half (33 %) in the 2-5 cm layer, and to 12 % in the deepest layer (5-10 cm).
Part of the C stock increase could be explained by the greater density of pasture soil.However, calculating the C stock based on soil mass and not volume, the same tendencies were observed, though to a lesser extent (results not shown).The strong entry of organic residue from the aerial part and the well-developed root system in pasture grass (Roscoe et al., 2001) could also explain this increase.Accumulation of organic residues with larger particle size (Figure 1) and decrease in microbial activity observed in the same areas by Melo (2007) suggest a slowing down of the SOM humification processes.
Quantities of C stocked in the surface layer verified in this study exceeded the values observed by Bernoux et al. (1999) in a chronological sequence of forestpasture in Paragominas (PA) in clayey soil.These authors observed a lower C stock in forest soil in the 0-10 cm layer (21.9 Mg ha -1 ) whereas in 10 year old pasture, they obtained lower values than for forests (20.4 Mg ha -1 ).They attributed this reduction to the fact that the area had been used for maize for two years before it was converted to pasture, since according to Cerri (1986), maize returns less plant material than pasture.On the other hand, Bernoux et al. (1999) verified a significant increase in C stock in a 15 year old pasture (26.5 Mg ha -1 ) where maize had been grown before being used as pasture.
The results in figure 2 show that pastures could be great C sinks for soil, in agreement with Bernoux et al. (1999), confirming that pasture implantation could lead to an increase in soil C content over time in the Amazon, as verified by Silver et al. (2000) and Desjardins et al. (2004).

CONCLUSIONS
1.After deforestation, soil C stocks increased with time of land use as pastures in the layer 0-10 cm, particularly in the layer 0-2 cm. 2. In the surface layer of pasture soils, slightly decomposed organic residues are accumulated in the sand fraction, while the quantity of C in the finer fraction did not vary, indicating that the humification processes of SOM had slowed down.
3. No quantitative alterations were observed after deforestation and pasture implementation in the humified organic material associated with clay particles, which represents the majority of organic carbon (OC) in the soil.

Figure 1 .
Figure 1.Carbon content in particle-size fractions at different depths in a dystrophic Oxisol, in function of soil use.

Figure 2 .
Figure 2. Carbon stock in a dystrophic Oxisol with clayey texture at three depths in function of soil use.

Table 2 . Average grades of total clay, silt, and sand and soil density (Ds) in different layers of a dystrophic Oxisol with clayey texture in function of land use system 0
-2 cm and 5-10 cm in NF (+ 23 %) and SF (+ 21 %), and the lowest increase in P5-7 (+ 1.4 %).Increasing pressure on the soil by animal trampling, mainly in the top layer largely explains the density increase over time of use in pastures.

Table 3 . Average mass of particle size fractions for the three soil layers, collected down to 10 cm depth in a dystrophic Oxisol with clayey texture in function of land use system
ns: not significant for the same layer by the Duncan test with an error margin of 5 %.R. Bras.Ci.Solo, 33:1603Solo, 33: -1611Solo, 33: , 2009

Table 4 . Carbon of fractions in different layers of Oxisol, with different land use system
Averages followed by the same lower-case letter in the column and capital letter in the line did not differ by the Duncan test with an error margin of 5 %; ns : not significant.