The pasture ecosystem is characterized by interactions between plants, animals, soil, climate, and management practices implemented by the farmer. Meat and milk productions are important economic activities in Brazil, and the sustainable practices of these production systems contribute to efficient carbon cycling, thereby improving soil quality (Carvalho et al., 2014; Soussana & Lemaire, 2014). Soils are a carbon pool for terrestrial ecosystems; thus, they are relevant with regard to environmental problems associated with global warming and deforestation (Bao et al., 2015).

Conversion from native forest to livestock grazing has decreased soil C storage (Carvalho et al., 2014) and has been the most common land use change in Brazil (Sousanna & Lemaire, 2014). It is estimated that Brazil releases more than 1.090 Mt CO₂/year into the atmosphere through deforestation, burning grasslands, and enteric fermentation by cattle (Bustamante et al., 2012).

Loss of organic carbon can be minimized by managing soil to reduce disturbances and maximize forage productivity using fertilizers and by integrating pastures and trees, which promotes benefits for the animals, increases productivity, litter inputs, nutrients cycling, and water infiltration (Murgeitio et al., 2011). Therefore, the evaluation of the impact of agricultural systems on pasture, through the quantification of total organic carbon (TOC) stocks and humic fractions, as well as assessing carbon lability, is important to maintain soil quality (Yang et al., 2012).

Specifically in pastures, organic input from vegetation and animal activities can contribute to increase the organic C content and consequently cause an impact on C pools (Lopes et al., 2010). As it may vary according to different pasture system, we hypothetized that the management method applied for different pasture systems could influence the soil carbon pool. In this context, the objective of this study was to assess the C pools of a tropical soil where the native forest was replaced with different pasture systems.

This study was conducted as a long-term experiment on pasturelands belonging to the Animal Science Department, Agriculture Science Center, Federal University of Piauí, Brazil (05°05′21′′ S, 42°48′07′′ W; 74 m asl). The climate is tropical with two seasons: rainy (January to May) and dry (June to December). The mean of precipitation is 1,300 mm/yr. The soil is a Haplic Acrisol. The experimental area presents plots with the following pasture system: a) Andropogon gayanus Kunth (AND) [plots without liming and chemical fertilization; production of 2.1 tons/ha (dry weight); 2.21% N and a C/N ratio of 21]; b) Brachiaria brizantha (BRA) [plots annually fertilized with 120, 180, and 100 kg/ha urea, triple superphosphate, and potassium chloride, respectively; production of 4.35 tons/ha (dry weight); 0.91% N and a C/N ratio of 37]; c) Panicum maximum (PAN) [plots annually fertilized with 70, 80, and 50 kg/ha urea, super triple phosphate, and potassium chloride, respectively; production of 3.0 tons/ha (dry weight); 1.22% N and a C/N ratio of 31]; d) Cynodon dactilon (CYN) [plots annually fertilized with 75, 30, and 30 kg/ha urea, super triple phosphate, and potassium chloride, respectively; production of 1.3 tons/ha (dry weight); 1.37% N and a C/N ratio of 36.9]; e) agroforestry system (AFS) [plots composed of grass (A. gayanus Kunth) and trees (Mimosa sp. and Thiloa glaucocarpa Benth); production of 7.4 tons of plant litter (dry weight)/ha]; and f) native vegetation (NV) (plots composed of native plant species, including Cenostigma macrophyllum, Tabebuia serratifolia, Hymenaea courbaril, Orbignya phalerata, Combretum leprosum, Guarea kunthiana, and Lecythis pisonis; production of 9.5 tons of plant litter (dry weight)/ha).

Soil sampling was carried out in March (rainy season) and September (dry season) 2014. Soil samples were obtained from three transects from each plot (three points per transect) at a depth of 0–20 cm. The soil samples were ground and passed through a 0.21-mm sieve to determine TOC by wet combustion using a mixture of potassium dichromate and sulfuric acid under heating (Yeomans & Bremmer, 1988). Labile organic carbon (LC) was quantified by wet oxidation with 0.33 M KMnO₄, as described by Blair et al. (1995). Non-labile carbon (NLC), that is equivalent to non-oxidized carbon by KMnO₄, was calculated as a difference (NLC = TOC – LC). Based on the difference between TOC-forest (reference) and TOC systems, a carbon pool index was created (CPI) and calculated as CPI = TOC-system/TOC-forest.

According to changes in the proportion of LC (i.e., L = LC/NLC) in the soil, a lability index (LI) was calculated as LI = L system/L reference. These two indices were used to calculate the carbon management index (CMI) using the following expression: CMI = CPI × LI × 100 (Blair et al., 1995). C-CO₂ emission or sequestration rate was estimated for 0–20 cm depth, using native vegetation as a reference (TOC stocks native vegetation – TOC stocks management systems/number of years). A conversion factor of C to CO₂ of 3.67 (molar mass of CO₂/molar mass of C) was used.

Soil humic substances (humic acids, fulvic acids, and humin) were extracted and fractionated using the method recommended by the International Humic Substances Society, as described by Swift (1996). The carbon content of the fulvic acid (C-FAF), humic acid (C-HAF), and humin (C-HF) fractions was measured using the dichromate oxidation method (Yeomans & Bremmer, 1988). The ratios of C-HAF by C-FAF and the alkali soluble fractions (C-FAF + C-HAF = AE) by H (AE/C-HF) were calculated to characterize the humified fraction of soil organic matter (SOM). Additionally, the humification index (HI) was calculated using the following formula to estimate the proportion of humified organic matter in relation to TOC content: HI = (C-FAF + C-HAF + C-HF) / TOC × 100.

Data were analyzed using one-way analysis of variance, and means were compared using the least significant difference (LSD) values calculated at the 5% level of significance. All analyses were performed using STATISTICA 7.1 (StatSoft).

The C-HAF was lower than the C-FAF during the wet and dry seasons in all systems (Table 1). The lower C-HAF found in all systems may have occurred because this humic fraction can easily migrate in soil with high porosity (Martins et al., 2009). Therefore, in the evaluated areas the soil porosity may facilitate the movement of C-HAF through the soil horizons. The C-HAF values in plots under Brachiaria, Panicum and Cynodon were similar to those in native vegetation during the dry season. Greater C-FAF availability was detected in the agroforestry system during the rainy and dry seasons because this plot showed higher TOC and Ca⁺² content, which is a favorable condition for the complexation of this C pool in the soil (Barros et al., 2012). The high carbon content from FAF in this system may have stimulated microbial diversity because that fraction is more easily used as an energy source by soil microorganisms, generating more negative charges and improving nutrient cycling and ecosystem productivity (Moraes et al., 2011).

The C-HF was higher than the C-HAF and C-FAF in all evaluated plots during both seasons, suggesting a greater stability of humin to mineralization than of humic and fulvic acids (Barros et al., 2012). The C-HF was higher in the native vegetation (dry season) and agroforestry system (rainy season), indicating the presence of more stable humus, low degradation, and strong stimulus to soil microbial activity (Barros et al., 2012). The presence of lignin derived from plant residues increased also the humin in soil (Carvalho et al., 2014). This increase in humin in soil is resulting from the loss of oxidative C and, at the same time, an increase in the C stable (Moraes et al., 2011). Therefore, the permanent inputs of organic C from herbaceous plants and trees in the agroforestry system may indicate a high potential for nutrient cycling and increased fertility.

No significant difference in the HA:FA ratio was observed between areas. The ratio (HAF + FAF)/HF in the native vegetation (rainy season) and agroforestry system (dry season) was higher than in the pasture systems, indicating that these systems contain more chemically stable organic C, for which the turnover time is approximately 2,000 years (Chan et al., 2001). The humification index did not differ between systems, which was likely because of the low proportion of labile C found in all plots. These results are indicative of soils with low organic matter input; however, with potential to stimulate the microbial growth in these ecosystems because of the complexity of their organic molecules (Bausenwein et al., 2008).

Higher lability of C was detected in the soil with Andropogon during the dry and rainy seasons, whereas soil with Cynodon showed higher lability of C during the dry season (Table 2). All plots showed a CMI < 100 (Table 2), indicating high plant input and minimal soil disturbance (Leite et al., 2014). According to Blair et al. (1995), a CMI < 100 indicates a strong negative impact of management practices on a soil ecosystem. Specifically, the agroforestry system showed a higher CMI than in the Brachiaria, Panicum, and Cynodon. The absence of trees and different pasture management practices reduce the C management index over time, reflecting a decrease in the potential to restore pre-existing carbon stocks (Leite et al., 2014).

The agroforestry system showed highest carbon stock values and high C sequestration rate for the rainy and dry seasons, whereas Brachiaria, Panicum, and Cynodon showed the lowest C stock and a high C loss (Fig. 1). It means that in multicropping systems, such as agroforestry system, the presence of pioneer tree species and the lesser removal of plant residues contribute significantly to mitigate the loss of C and decrease greenhouse gas emissions. Also, agroforestry systems have higher potential to build up and sequester C in soils because of the increased rates of organic matter addition and retention (Lenka et al., 2012).

In conclusion, conversion of native vegetation to pasture system causes changes in C pools, increasing CO₂ emissions into the atmosphere. The agroforestry system has the potential to sequester more carbon in the soil than the pasture system, and it may be an alternative to produce forage for animal production.