PHYSICAL QUALITY OF AN OXISOL UNDER AN INTEGRATED CROP-LIVESTOCK-FOREST SYSTEM IN THE BRAZILIAN CERRADO

A qualidade fisica do solo e um importante fator para a sustentabilidade dos sistemas agricolas. O objetivo desta pesquisa foi avaliar as propriedades fisicas do solo e o carbono orgânico em um Latossolo Vermelho distrofico, sob sistema de integracao lavoura-pecuaria-floresta. O experimento foi conduzido no Mato Grosso do Sul, Brasil. Os tratamentos consistiram de sete sistemas: integracao lavoura-pecuaria-floresta, com 357 arvores ha-1 e altura de pastejo de 30 cm (CLF357-30); integracao lavoura-pecuaria-floresta, com 357 arvores ha-1 e altura do pastejo de 45 cm (CLF357-45); integracao lavoura-pecuaria-floresta, com 227 arvores ha-1 e altura do pastejo de 30 cm (CLF227-30); integracao lavoura-pecuaria-floresta, com 227 arvores ha-1 e altura do pastejo de 45 cm (CLF227-45); integracao lavoura-pecuaria, com a altura do pastejo de 30 cm (CL30); integracao lavoura-pecuaria, com a altura do pastejo de 45 cm (CL45); e vegetacao nativa (NV). As propriedades do solo foram avaliadas nas profundidades de 0-10 e 10-20 cm. Todos os tratamentos com pastagem aumentaram a densidade do solo (rb) e resistencia a penetracao (RP) e diminuiram a porosidade total (¦t) e macroporosidade (¦ma), em comparacao com a NV. Os valores de rb (1,18-1,47 Mg m-3), ¦ma (0,14-0,17 m3 m-3) e PR (0,62-0,81 MPa) na profundidade de 0-10 cm nao foram restritivos ao crescimento de plantas. A mudanca de uso do solo a partir NV para CL ou CLF diminuiu o carbono orgânico (SOC) e o estoque de carbono orgânico (SOCpool). Todos os tratamentos com pastagem apresentaram SOCpool similar na profundidade de 0-10 cm e foram menores em comparacao a NV (17,58 Mg ha-1).


INTRODUCTION
Increasing population and its demands have required drastic changes in agricultural systems.Traditional production systems have been replaced by environmentally, socially, and economically sustainable systems.However, projected climate changes demand even more drastic adaptive measures in agriculture production systems (IPCC, 2007).Soils are integral to the sustainability of agricultural systems, and inappropriate management can jeopardize soil quality and sustainability.Land use with traditional agriculture involving excessive use of tillage and degradation of pasture are causing major ecological problems in the Brazilian savanna (Cerrado) and lead to nonsustainability of agricultural systems.However, use of complex systems involving a combination of annual crops and perennial grasses, or annual crops with perennial grasses and trees can restore degraded soils, increase production and improve the environment (Macedo, 2009).
Typically, the process of recovery of degraded soils by integrated crop-livestock or crop-livestock-forestry systems in the Brazilian Cerrado begins in autumn when the soil is tilled and prepared for growing annual crops (corn, soybean) in summer.The sequence continues with the cultivation of grass, or grass and trees, after the grain harvest.In some cases, there is intercropping of annual crops and grass, especially when the annual crop is corn (Borgui & Crusciol, 2007).Livestock grazing occurs in winter and part of autumn; in summer, a no-till system (NT) is used for the production of annual crops.
Assessment of soil physical properties has indicated that integrating crop-livestock systems increases bulk density (ρ b ) and decreases total porosity (ƒ t ) in the first grazing period (Araújo et al., 2010) but does not affect crop yield (Flores et al., 2007).Tracy & Zhang (2008) reported an increase in penetration resistance (PR) and higher corn yield in an integrated croplivestock system after 5 years of use compared to continuous corn.These beneficial effects are attributed to a marked increase in soil organic carbon (SOC) concentration.Moreira et al. (2012) observed improvements in the physical quality of an Oxisol after 8 years of an integrated crop-livestock system and attributed it to the physical quality of resilience.In general, management practices induced changes in soil physical properties, which occur in surface soil (Conte et al., 2011a;Santos et al., 2011).The reversibility and intensity of soil physical degradation by animal trampling depend on grazing intensity (number of animals) and the height of the grass (Conte et al., 2011a).The intensity of soil compaction is related to pasture availability and manifestation of stress by the pasture biomass.Manifestation of stress is directly proportional to the amount of biomass on the soil (Braida et al., 2006).
In addition to pasture height, the factor of spatial arrangement of trees must be considered when they are introduced in integrated systems.The amount of sunlight reaching the soil surface in an agroforestry system depends on the growth and management of trees (Soares et al., 2009).Paciullo et al. (2010) observed a reduction of 22.3 and 41.4 % in green forage and root mass, respectively, in the vicinity of the tree row compared to crop growth without trees.
Experiments carried out in some soils of the Brazilian Cerrado show that the use of an integrated crop-livestock (Carvalho et al., 2010) or eucalypt (Eucalyptus urograndis) forest (Zinn et al., 2011) system may enhance soil organic carbon (SOC) concentration.Lal (2004) reported that integration of crop, livestock and forest (agroforestry) systems can accentuate SOC sequestration.Increases in the SOC concentration and pool through the use of agroforestry systems, as reported by Paudels et al. (2012) in subtropical regions, as well as in some soils of tropical regions (Lima et al., 2011), have been attributed to higher root densities (Kumar et al., 2010),.However, most agroforestry studies involve fruit trees rather than timber wood species.
This study is based on the hypothesis that degradation of soil physical quality and change in the SOC concentration and pool in pastures are influenced by the presence and arrangement of trees and by the height of grass at the time of grazing.Thus, the aim of this study was to assess the soil physical properties and the SOC concentration and pool in a Typic Acrudox managed under crop-livestock and croplivestock-forest systems.

Site and treatments
The study was carried out at the National Cattle Research Center of the Empresa Brasileira de Pesquisa Agropecuária (Brazilian Crop and Livestock Research Company), Mato Grosso do Sul, Brazil (Embrapa/ CNPGC).The climate is Aw according to the Köppen classification, with annual average rainfall of 1,500 + -227 mm and annual average temperature of 22.8 o C. Soil at the study site is classified as a Typic Acrudox, with 389 + -0.22 g kg -1 clay, 71 + -0.16 g kg -1 silt and 540 + -0.15 g kg -1 sand.Soil analyses were performed for seven croplivestock systems as follows: Integrated crop-livestock-forest (CLF 357-30 ): This sequence commenced with the cultivation of soybean (Glycine max L.) from November 2008 to March 2009.Beginning in January 2009, grain crops were grown in the summer in association with eucalypt at a density of 357 trees ha -1 (14 m between rows and 2 m between trees).In addition, palisade grass (Urochloa brizantha Piatã) was sown (4 kg ha -1 ) after soybean harvest.Animals were introduced into the system in May 2010 when the trees reached 7 cm in diameter and the palisade grass was 30 cm high; Integrated crop-livestock-forest (CLF 357-45 ): The sequence used in this treatment was the same as in the CLF 357-30 treatment, except that the pasture height was 45 cm; Integrated crop-livestock-forest (CLF 227-30 ): The sequence used in this treatment was the same as in the CLF 357-30 treatment, except that there were 227 trees ha -1 (21 m between rows and 2 m between trees), and pasture height was 30 cm; Integrated crop-livestock-forest (CLF 227-45 ): The sequence used in this treatment was the same as in the CLF 227-30 treatment, except that the pasture height was 45 cm; Integrated crop-livestock (CL 30 ): The sequence was cultivation of soybean (November 2008 to March 2009) followed by palisade grass.In April 2010, grazing animals were introduced into the grass with a height of 30 cm; Integrated crop-livestock (CL 45 ): This sequence is the same as in CL 30 , except that the pasture height was 45 cm; and Native vegetation (NV): This was an undisturbed site under native vegetation typical to the savanna (Cerrado).
The area occupied by treatments CLF 357-30 , CLF 357-45 , CLF 227-30 , CLF 227-45 , CL 30 and CL 45 had been cleared in 1980 and was grazed continuously for 27 years.In 2008, prior to establishment of present experiment, the site was plowed, and recommended practices were implemented for all crops (Bungenstab, 2012).
All treatments including grass were grazed by cattle of the Nelore breed with individual weight of about 160 kg.The number of animals varied depending on the desired grass height.The area of each experimental plot was about 1.5 ha.

Measurement and analyses
Soil samples were obtained in January 2011.Seven bulk and seven core samples were obtained from the 0-10 and 10-20 cm depths at each site.
Aggregate stability was determined by the wet sieving method (Nimmo & Perkins, 2002).Fifty grams of air-dried aggregates at a size of 4.75-8.00mm were placed on top of a nest of sieves of 4.75, 2.00, 1.00, 0.50 and 0.25 mm size and oscillated vertically at 30 oscillations per minute for 30 min.The fractions retained on each sieve were separated and oven dried at 105 o C for 24 h.The percent of water stable aggregates (WSA) in each size were calculated according to equation 1: where wi is the weight of the aggregates in each size classification (g).
The cores of 100 cm 3 volume (5.04 cm diameter and 5 cm height) were saturated for 48 h and subjected to pressures of 0 and 0.006 MPa using a tension table (Romano et al., 2002), and 0.01 MPa using a pressure chamber (Dane & Hopmans, 2002).After desorption at 0.01 MPa, the soil cores were removed from the pressure chamber and were oven dried at 105 o C to determine soil bulk density (ρ b) (Grossman & Reinsch, 2002).Grab samples, collected from an area adjacent to where the soil-core samples were obtained, were air dried and crushed to pass through a 2 mm sieve.Soil particle density (ρ s ) was measured by the pycnometer method (Flint & Flint, 2002).The average ρ s of 98 soil samples (seven treatments × two depths × seven replications) was 2.58+0.09Mg m -3 .Soil total porosity (ƒ t ) was determined by equation 2. Microporosity (ƒ mi ) was estimated corresponding to volumetric moisture content at 0.006 MPa suction in the tension table according to the method of Embrapa (2011), and macroporosity (ƒ ma ) (pores with equivalent diameter >50 µm) was determined by the difference between ƒ t and ƒ mi .
Soil cores at an equilibrium of water potential of 0.01 MPa (corresponding to field moisture capacity) were wrapped in a cellophane membrane and stored in a refrigerator (5 o C) for 5 days to homogenize soil moisture content.Penetration resistance (PR) was then measured in two replications for each core using a static electronic cone penetrometer with a cone angle of 30 o and basal area of 0.1167 cm 2 at a constant penetration speed of 1 cm min -1 .This device was equipped with a 20-kg load cell coupled to a microcomputer for data input (Tormena et al., 1998).
The SOC concentration was determined by the potassium dichromate (K 2 Cr 2 O 7 ) oxidation method at 170-180 o C, followed by titration with 0.1 mol L -1 ferrous sulfate (Walkley & Black, 1934).The SOC pool (SOC pool ) was computed on an equal mass basis by using NV as a reference soil (Ellert & Bettany, 1996) and was calculated by using equation 3: where SOC pool is expressed in Mg ha -1 , SOC is in kg Mg -1 , ρ b is bulk density in Mg m -3 , and depth of the soil layer is in m.

Statistical analyses
The experimental site including all seven treatments and seven replications was located on land with homogeneous topographic and edaphic-climatic conditions.The data were analyzed using a completely randomized factorial design and seven replications.
The F test was used to determine the significance of the main effects by computing ANOVA.Significant differences among treatments were established using the least significant difference (LSD) (p<0.05).
The multivariate structure was examined by hierarchical cluster analysis and principal component analysis (PCA).For cluster analyses, data standardization was performed in order to provide null means and variance equal to one for each sample.The Euclidean distance was used as a dissimilarity distance (Sneath & Sokal, 1973).Ward's method was used as the grouping technique (Ward, 1963).
In this study, analysis was made of principal components with eigenvalues >1 (Kaiser, 1958).Bidimensional representation was created with the principal components for better visualization of the structure of specific soil properties and treatments.

Soil physical properties
All grazing treatments increased soil ρ b , and decreased ƒ t and ƒ ma compared to NV (p<0.05)(Table 1).At the 0-10 cm depth, the treatments with animal grazing had similar ρ b , with values ranging from 1.18 to 1.32 Mg m -3 , i.e., from around 25 to 40 % higher than that of NV (0.94 Mg m -3 ).The use of trees did not cause differences in ρ b at the 0-10 cm depth.At the 10-20 cm depth, the ρ b of treatments with trees (CLF 357-30 , CLF 357-45 , CLF 227-30 and CLF 227-45 ) ranged from 1.30 to 1.45 Mg m -3 , which was from 15 to 65 % higher than those for CL (CL 30 and CL 45 ) and NV, respectively.The increase in ρ b at the 10-20 cm depth was accompanied by a reduction in ƒ t .Although ƒ ma also decreased, lower value of ƒ ma (0.07 m 3 m -3 ) was observed in CLF 357-30 .The less ƒ mi was found in NV for two depths -0.30 and 0.32 m 3 m -3 .At the 0-10 cm depth, ƒ mi was similar for the CL 30 (0.39 m 3 m -3 ) and CL 45 (0.38 m 3 m -3 ) treatments; these values were higher than those for the other treatments (Table 1).
Soil PR was greater in all the grazing treatments, and it was lower at the 0-10 cm depth than at the 10-20 cm depth.For the 0-10 cm depth, treatments with trees and the lower pasture height of CLF 357-30 (0.81 MPa) and CLF 227-30 (0.88 MPa) had higher PR values than those for other treatments, and they were 279 and 303 % higher than the value for NV.For the 10-20 cm depth, the CLF 357-30 and CLF 227-30 treatments had the highest PR; the values observed of 0.92 and 0.96 MPa were 438 and 457 % higher, respectively, than the value for NV (Table 2).

Soil organic carbon concentration and pool
The change in land use from NV to CL or CLF decreased the SOC and SOC pool (p<0.05)(Table 4).For the 0-10 cm depth, treatments with trees (CLF 357-30 , CLF 357-45 , CLF 227-30 and CLF 227-45 ) tended to have lower SOC than those without trees (though statistically not different).All grazing treatments had a similar SOC pool in the 0-10 cm depth, and values were lower than the value for NV (17.58 Mg ha -1 ).In the 10-20 cm depth, however, CLF 357-30 with 11.64 Mg ha -1 and CLF 227-30 with 11.49 Mg ha -1 had lower values for SOC pool ; the values were about 16 % lower than the value for NV.The sum of SOC pool at both depths for CLF 357-30 (24.67 Mg ha -1 ), CLF 357-45 (25.32 Mg ha -1 ) and CLF 227-45 (25.61 Mg ha -1 ) were lower than SOC pool for NV (31.28 Mg ha -1 ) and other treatments (Table 4).
Two dimensional representations of PC1 and PC2 showed similar trends of cluster analysis.The formation of three clusters made it possible to identify the treatments together with soil properties (Figure 3).With negative values and towards the left of CP1 was the NV treatment, for soil properties such as ƒ ma at the 0-10 and 10-20 cm depths and ƒ t , WSA 1.00-2.00 ,WSA 0.50-1.00and SOC pool at the 0-10 cm depth.On the right side of the graph with positive values for PC1 included treatments involving crop-livestock-forest for PR and ρ b at the 0-10 and 10-20 cm depths and WSA 4.76-8.00and WSA 0.21-0.50 at the 0-10 cm depth.On the lower side of the graph, negative values of PC2 were established for the CL 30 and CL 45 treatments and for ƒ mi at the 0-10 and 10-20 cm depths, SOC pool at the 10-20 cm depth, and WSA 2.00-4.76 and WSA 2.00-0.50 at the 0-10 cm depth (Figure 3).

Soil physical properties
Management induced changes in soil physical properties for crop-livestock systems compared to soil physical properties for the NV were also reported by Santos et al. (2011), especially for the surface soil.The values of ρ b (1.18-1.47Mg dm -3 ), ƒ ma (0.14-0.17 m 3 m -3 ) and PR (0.62-0.81 MPa) observed at the 0-10 cm depth are not necessarily restrictive to plant growth (Xu et al., 1992;Reichert et al., 2009).These data are in accord with the data reported by Marchão et al. (2007), and support the conclusion that the integrated crop-livestock system is sustainable for the Brazilian Cerrado.
The data on soil physical properties for CLF or CL systems indicate favorable soil physical quality.These favorable trends may be due to the effect that soil samples were obtained during the second year after establishing the treatments.In general, drastic adverse effects on soil physical properties by animal trampling in crop-livestock systems normally occur during the first grazings, and decrease over time (Moreira et al., 2012).Better soil physical quality observed in CLF or CL may be attributed to soil resilience and restoration by wetting and drying cycles and root growth (Gregory et al., 2007).The CL (soybean and palisade) and CLF (eucalypt, soybean and palisade) systems with greater species diversity may have a better root system (Kumar et al., 2010), with a beneficial impact on soil physical quality.Silva et al. (2011) observed that two years is an optimum time for management of a crop-livestock system for improving soil physical properties.Similar or higher WSA in the 4.76-8.00mm size class for treatments involving palisade compared to NV indicated the importance of grass in enhancing soil aggregation (Carvalho et al., 2010).Inda Júnior et al. (2007) reported that grasses can have an important function in formation of large aggregates by cementing together of micro aggregates.A relatively low effect of animal grazing on aggregates (1.0-2.0,0.5-1.0 and 0.21-0.50mm) was also reported by Souza et al. (2010) for an Oxisol under an integrated crop-livestock system.It is possible that the forces that hold particles together in small aggregates are higher than those for larger aggregates because of higher protection of organic matter in small size aggregates.
The data of the present study indicated a higher WSA in a crop-livestock-forestry system (CLF 357-30 and CLF 357-45 ) than for other systems.These results are in accord with those of Lenka et al. (2012) from central India, who also reported a higher WSA in soils under forest based pasture.Tree roots can penetrate deeper into the soil compared to those of annual crops (Mekonnen et al., 1997) and they release enzymes, which accentuate the aggregation process (Paudels et al., 2012).

Soil organic carbon concentration and pool
A higher SOC concentration in soils with pasture compared to those under NV and eucalypt forest have been reported by Pulrolnik et al. (2009).Such trends are attributed to a prolific root system and more biomass on the surface soil under grass.An average difference of 2.15 g kg -1 of SOC concentration between CLF and CL treatments may be attributed to the loss of dry matter from forage by grazing and, consequently, to the loss of the root system.Established trees (eucalypt) alter the microclimate near the grass and affect that development.Soares et al. (2009) reported a decline in dry matter yield of some grasses when planted between the rows of pine (Pinus taeda).
Changes in SOC pool upon conversion of NV to pasture in the Brazilian Cerrado have also been reported by Maia et al. (2009).Reduction in SOC pool in the present study by 23 % (3.95 Mg ha -1 ) and 15 % (2.62 Mg ha -1 ) for CLF and the CL, respectively, compared to NV are more than the values reported for an Oxisol in central Brazil by Zinn et al. (2005).However, the SOC pool of 13.63 and 14.96 Mg ha -1 at the 0-10 cm depth for CLF and CL, respectively, are similar to those reported for degraded pasture by Carvalho et al. (2010).Furthermore, the SOC pool under the crop-livestock system increased over time.
Soil samples in the present study were obtained two years after the conversion of degraded pasture to CLF or CL.Thus, the SOC pool may have increased over the 2-year period.For example, Tonucci et al. (2011) reported an increase in the SOC pool in CLF systems after 14 years of establishment, and the SOC pool in CLF was similar to values of soils under pasture for 33 years and for native vegetation.

Multivariate analyses
The formation of three clusters indicated the grouping of treatments by management (NV, CL and CLF) rather than by the height of the grass.These trends are similar to those reported by Conte et al. (2011a), who also did not observe any difference in soil physical properties because of pasture height in the crop-livestock systems in an Oxisol in Brazil.It is probable that the palisade height (30 and 45 cm) is sufficient to minimize the adverse effects of trampling by the cattle.Souza et al. (2010) also did not observe any degradation in soil structure when the height of grazing of black oat (Avena nativa) + Italian ryegrass (Lolium multiflorum Lam.) was about 20 cm.
Both graphic presentation and correlation coefficients indicate that favorable soil physical quality under NV (group 1, in cluster analysis) is associated with higher ƒ ma and ƒ t at the 0-10 and 10-20 cm depths and SOC and SOC pool at the 0-10 cm depth (Table 4).Furthermore, higher ρ b and PR at the 0-10 and 10-20 cm depths are useful indicators for separating CLF treatments (group 3, in cluster analysis).Tormena et al. (1998) also observed that management can easily change PR as a soil physical attribute, and any differences in PR can be measured immediately after pasture establishment (Conte et al., 2011b).
Naturally, pastures are heterogeneous in the amount and quality of forage.Paciullo et al. (2011) observed the formation of a shadow zone by trees in agroforestry systems; the shadow zone near the tress has a smaller amount of forage, but it is of better quality (higher crude protein); and the zone away from the tree has a higher amount of forage, but of poorer quality.Animals seek better forages, which in CLF systems is in a zone parallel to and between the trees.Therefore, higher trampling by animals increased soil ρ b between the tree rows.
The SOC concentration and SOC pool at the 10-20 cm depth are index soil properties for characterizing group 2 (CL).For C input by the root system, some CLF treatments may have a larger root density than others.The growth of tree species eucalypts lower, Kuzyakov & Domansky (2000) reported that grasses could sequester SOC pool at the rate of 2.2 Mg ha -1 year -1 C. Furthermore, SOC and SOC pool at the 0-20 cm depth can also differentiate group 2 (CL) because the yield of palisade with trees is less than the yield without trees (Paciullo et al., 2011).There is a strong relationship between yield and the root system.Both of these scenarios are in agreement with the hypothesis reported by Carvalho et al. (2010) that palisade has a deep root system and adds biomass-C to the surface layers.

CONCLUSIONS
1.The hypothesis that the presence of trees in pastures affects soil physical properties and SOC concentration and SOC pool is indicated by the data presented by multivariate analyses.Changes in soil physical properties by CLF or CL were not significant under the soil and management conditions of this experiment.Thus, crop-livestock systems can be used for long-term sustainability.
2. Additionally, long-term studies are needed to examine a range of combinations of trees, grasses and cropping systems.The potential of integrated croplivestock-forest systems must be observed for diverse soils and ecoregions.

Figure 2 .
Figure 2. Dendogram of warm method showing similarities in soil physical attributes, SOC and SOC pool in all treatments.

Table 3 .
Water stable aggregate size distribution at the depths and in the treatments studied Mean values with the same uppercase letter at the depth, and lowercase letter in the treatment do not differ among themselves in the LSD test (p<0.05).

Table 2 .
Penetration resistance and soil water content at the soil depths and treatments studied Mean values with same uppercase letter at the depth and lowercase letter in the treatment do not differ among themselves in the LSD test (p<0.05).

Table 4 .
Soil organic carbon (SOC) and soil organic carbon pool (SOC pool ) at the soil depths and treatments studied