Soil organic matter dynamics in a Cerrado Oxisol

The Brazilian Cerrado is the second largest biome in the country, spreading over 23 % of the national territory. In the last three decades, it has increasingly contributed to the national production, being responsible in 1995 for 25% of the national agricultural production and sheltering 40% of the cattle flock. Development strategies have stimulated strongly mechanised and intensive agricultural practices, which has raised concerns about soil organic matter (SOM) losses and soil degradation.Cerrado soils are dominated by low-activity clays and are characterised by high acidity and low pH, cation exchange capacity (CEC), and available nutrients (especially P and N). SOM exerts important functions in these soils, which are closely related to soil quality and sustainability. It is responsible for most of the CEC, and is involved in process of soil aggregation and water dynamics. Losses of SOM due to cultivation may seriously affect these functions, and also represent an important source of CO 2 to the atmosphere. Studying the effects of different management systems on C dynamics in Cerrado soils may help to develop better ways of using these ecosystems.The initial objective of this thesis was to evaluate the impact of different management systems on SOM dynamics in one of the most representative soil classes in Cerrado, the Dark Red Latosol (18% of the total area; Oxisol - Soil Taxonomy; Ferrasol - FAO Legend). The research strategy adopted was to compare, in the same soil unit, paired plots with different management systems (native vegetation, pasture, no-tillage and conventional tillage). The Dark Red Latosol unit (Typic Haplustox) was located in the experimental research institution of EMBRAPA-CNPMS (Brazilian Institute of Agricultural Research - Maize and Sorghum National Research Centre), in Sete Lagoas - MG, Brazil. To assess SOM dynamics, a combination of two methodologies was proposed, (i) the physical fractionation of soil in particle-size and density separates, and (ii) replacement calculations using the natural abundance of the 13C isotope. The physical fractionation aimed at separating SOM pools with different chemistry, location, and turnover, which hypothetically would be more sensitive to soil alterations than total SOM. The natural abundance of 13C was proposed to assess the turnover time of SOM and its fractions. However, as very little data was available about the use of these methodologies in Cerrado ecosystems, the initial focus of the thesis was changed. The initial general objective was extended to three specific objectives:(i) to adapt the methodologies of physical fractionation of soil and 13C natural abundance to SOM studies in a Cerrado ecosystem;(ii) to contribute to a better understanding of the distribution and dynamics of different SOM pools in this specific environment; and(iii) to evaluate the long-term effect on SOM of converting a cerrado sensu stricto into cultivated pasture or to annual crops (maize and beans) under conventional tillage and no-tillage.In Chapter 2, an exploratory study was conducted to establish ad13C-reference profile under the native cerrado sensu stricto . In a native reserve located at EMBRAPA-CNPMS, three soil profiles were analysed in plots with different fire history. Results showed that the studied cerrado was a C 3 -dominated vegetation, presentingd13C values typical for soils under such a vegetation throughout the soil profile. Fire favoured the grass population and increased the amount of C 4 -derived carbon in the system. No alteration in carbon stocks in soils was observed with increasing fire incidence. It was concluded that different fire regimes may interfere with the establishment of a 13C standard profile for dynamic studies. The profiles under low fire intensity were considered a good reference for the further studies.In Chapter 3, the focus was on establishing a physical fractionation procedure for SOM in the Cerrado Oxisol under investigation. Special attention was given to the dispersion by ultrasonic energy, a crucial point in the methodology. A procedure to calibrate the ultrasonic equipment and to determinate the minimum of energy required for an efficient dispersion is proposed. Increasing ultrasonic energy significantly changed the amounts of C and N, the C:N ratio, and thed13C values of the particle size fractions analysed (0-2mm; 2-50mm; 50-100mm; 100-250mm; 250-2000mm). The results suggest that the soil (< 2 mm) can be divided into unstable (100-2000mm) and stable (50-100mm) aggregates. A threshold energy of 260-275 J ml -1is proposed for the dispersion of unstable aggregates. The use of this threshold energy, combined with particle-size fractionation, was not satisfactory for all purposes, since litter-like material and relatively recalcitrant organic carbon present in stable aggregates > 100mm were recovered in the same pool. An ultrasonic energy of 825 J ml -1was not sufficient to stabilize the redistribution of soil mass and organic matter among particle-size fractions, but at energies above 260-275 J ml -1relatively stable aggregates would fall apart and cause a mix of carbon with varied nature in the clay fraction.In order to better understand the dynamics of soil organic matter (SOM) in Oxisols, and the impact of converting the cerrado sensu stricto into pasture, the dynamics of physically separated SOM pools at different depths in a cerrado Oxisol (Typic Haplustox), under natural conditions and after 23 years of cultivated pasture ( Brachiariaspp . ) was studied via the replacement of the native C (C 3 -derived) by pasture C (C 4 -derived) (Chapter 4). Organic C stocks of the original cerrado (15±3 kg m -2) and pasture (17±3 kg m -2) were not significantly different, which was attributed to the high biomass production of the tropical grasses and the protective effect of the high clay content (> 800g kg -1). The clay + silt fraction accumulated 89-91% of the total organic C. The replacement of cerrado-derived C by pasture-derived C was in average 36%, 34%, and 19% for A p , AB 1 , and B w2 horizons respectively, suggesting a fast turnover rate of organic C, regardless of the high clay content. The replacement decreased in the order: free low-density organic matter (LDOM) > heavy fractions (sand, silt, clay) > occluded-LDOM. The lower replacement of the occluded-LDOM compared to the heavy fractions was attributed to protection inside aggregates and to a possible accumulation of C 3 -derived charcoal (black carbon) in that fraction. After 23 years of pasture, about 50% of the total organic C in the free-LDOM in the topsoil was still from cerrado, indicating that a significant part of this fraction was relatively recalcitrant. Charcoal fragments observed in the fraction suggested that the recalcitrance was probably due to charred material.In Chapter 5, the spatial continuity ofd13C and other soil organic matter (SOM) related variables (organic C, total N, and C:N ratio) were analysed in the native cerrado sensu stricto and in a nearby cultivated area with neighbouring plots under conventional and no-tillage systems. The aims were to describe the spatial variability of these properties in the cerrado, with especial attention to fire effects, and to analyse how management systems affect their spatial structure. Global, within strata, and stratified kriging were used to model the spatial variability in the areas. In cerrado, the total variability of all variables was relatively small, which was attributed to the high textural and mineralogical homogeneity of the clayey soil. Nevertheless, part of the variability was spatially structured. In cerrado locations with more open vegetation, long-term cumulative effect of repeated fires seemed to determine the spatial structure ofd13C and SOM-related variables. Cultivation reduced the variability of most of the variables and changed their spatial structure. The variables tended to be less spatially structured in no-tillage than in conventional tillage, due to small-scale variability. The spatial structure observed ford13C in the cultivated area was probably inherited from the former cerrado vegetation. This implies that, in studies of SOM dynamics, the variability of the replacements would be overestimated if the trends in both areas were not taken into account.In Chapter 6, as the different particle-size fraction in Chapter 4 presented similar dynamics, only density fractionation was used to assess changes in SOM upon 30 years of cultivation. The objectives of the study were (i) to evaluate the long-term impact of conventional and no-tillage systems on SOM stocks in the soil, and (ii) to better understand the dynamics of SOM in different density fractions of this soil. It was observed that cultivation led to compaction, significantly increasing soil bulk density. This resulted in a systematic overestimate of C and N stocks in cultivated areas when compared to the natural cerrado. Conversion of cerrado into conventional tillage (CT) or no-tillage (NT) system did not alter the total C (~100 Mg ha -1) and N (~7 Mg ha -1) stocks of the first 45cm depth in 30 years of cultivation. However, about 22% of the total carbon was replaced by maize material in this period. In accordance with results from Chapter 4, the relative replacement of carbon decreased the order: free light fraction (F-LF) > heavy fraction (HF) > occluded light fraction (O-LF). The low substitution in the O-LF was attributed to the possible presence of charcoal. The F-LF showed the highest sensitivity to changes in management system, and converting cerrado into cropland significantly decreased its quantity. The proportion of C replacement in this fraction was higher in CT than NT, suggesting a faster turnover in the first. Nevertheless, because most carbon (~95%) was held in the HF, carbon dynamics in the whole soil was controlled by the behaviour of this fraction.In Chapter 7, results form Chapter 4 and Chapter 6 were used to calculate proportions and stocks of carbon derived from a newly introduced C 4 vegetation in a C 3 ecosystem, by two different linear mixing models. One model assumed no 13C discrimination upon humification for the new introduced C 4 material, and the other assumed equal 13C discrimination for both C 3 and C 4 materials. The aims were to evaluate (i) how the assumption of equal 13C discrimination for C 3 and C 4 material may affect estimates of proportions and stocks of C 4 -derived carbon for different soil depths and soil organic matter fractions; (ii) the significance of a possible difference between outputs using the different models; and (iii) the sensitivity of this difference to variations in input parameters. The assumption of equal discrimination was discussed in the light of the current theoretical understanding of processes leading to 13C discrimination in soils. Taylor series approximation was applied to estimate the output uncertainties of the models. Sensitivity analysis was used to test the influence of these outputs in the magnitude of 13C discrimination with decomposition for the standard C 3 vegetation, the population standard deviation of soil and litterd13C, and the sample size for soil and litter. The two models may generate significantly discrepant outputs. This difference was most sensitive to the standard deviation of the soild13C population and the number of soil samples. A critical analysis of the current understanding of processes leading to 13C discrimination in soils suggests that the assumption of absence of 13C discrimination for the newly C 4 introduced material would provide the most realistic results.Carbon stocks in the studied soil did not change with land-use (pasture, conventional tillage, and no-tillage). The low capacity of occlusion in soil aggregates and their high stability, as well as the massive concentration of SOM in the heavy fraction (largely dominated by clay-sized organo-mineral complexes) explained the relative resilience of this soil upon disturbance. These characteristics were probably related to the large contents of oxi-hydroxides of Fe and Al in the studied soil. Nevertheless, a significant amount of SOM (about 1/4 to 1/3) was replaced in 30 years, suggesting that large part of the carbon is rather active. Conversely, the remaining 2/3 to 3/4 seemed to be fairly stable, since it was not recycled even after 30 years of intensive cultivation. This fraction probably represents the passive compartment used in dynamic models such as CENTURY and RothC.In the studied Oxisol, density fractionation separated SOM fractions with contrasting dynamics. This could potentially be used for separating measurable SOM pools, which would represent an important advance in the verification of current carbon dynamic models. However, these fractions were a mixture of labile and recalcitrant components, and procedures for distinguishing these components are needed. Quantification of charcoal in the light fraction and non-hydrolysable materials in the heavy fractions represents a promising alternative.The 13C natural abundance approach proved efficient for tracing SOM dynamics in the Cerrado Oxisol. The spatial variability ofd13C in different plots (e.g. cerrado vs. pasture or cultivated plots) is responsible for most of the final uncertainty in replacement calculations. To reduce such an uncertainty, samples should be taken at the same locations before and after conversion for a new vegetation. This implies that, ideally, experiments should be planed for long-term evaluation in the same area, instead of comparing neighbouring plots.The assumptions in the calculations can significantly affect the outputs (e.g. source proportions). When a C 4 plant is introduced into a C 3 ecosystem, the assumption of no fractionation for the newly introduced material seems the most reliable option. Nevertheless, further research is needed to better understand the process of isotope fractionation during decomposition.