WATER EROSION IN SOILS UNDER EUCALYPTUS FOREST AS AFFECTED BY DEVELOPMENT STAGES AND MANAGEMENT SYSTEMS

The constant increasing of eucalyptus forest areas in Brazil requires an accurate monitoring of water erosion. The present study aimed to evaluate soil, nutrients and organic carbon losses occasioned by water erosion in eucalyptus planted forests (EPF) at different development stages (2, 3 and 7 years old). Soil erosion sediments were measured and sampled from standard erosion plots installed on Red Argisol-RA (Ultisol) and Haplic Cambisol-HC (Inceptisol). Soil loss decreased as the age of plants increased; at the beginning of plant development, the canopy barely covered the soil surface, exposing the soil to higher erosion at young EPF plantations. Furrow planting system was used in the Red Argisol area and caused higher soil losses (1.1 to 6.2 Mg ha-1 year-1) as compared to pit planting system that was used in the Cambisol area (1.1 Mg ha-1 year-1). It is known that Cambisol is less resistant to erosion than Argisol. However, using pit system in this EPF, resulted in lower erosion and, therefore, nutrients and carbon losses than the traditional furrow system used in Argisol. Concerning the soil loss, this work points to the need of improving soil conservation practices to prevent soil erosion at the earlier stages of eucalyptus plantation. The amount of calcium and potassium were higher than magnesium in the soil sediment.The relatively high amount of carbon found in the erosion sediments raises additional concerns about the environmental sustainability and deserves future research.


INTRODUCTION
Eucalyptus wood plays an important role as a source of renewable energy and other industrial purposes. One of the greatest advantages of eucalyptus crops the possibility of subsequent harvests from the same plantation. Besides, it can grow and produce on marginal or low natural fertility soils (KENYA FOREST SERVICE, 2009). Therefore, eucalyptus crops are sustainable fuelwood sources, thus allowing to prevent deforestation of natural forests, to use higher fertility soils for food production, and to create new job opportunities (CAMPINHOS JUNIOR, 1999). In Brazil, eucalyptus is cultivated on over 4.8 million hectares; 11.2% are in the state of Rio Grande do Sul (ASSOCIAÇÃO BRASILEIRA DE PRODUTORES DE FLORESTAS  PLANTADAS-ABRAF, 2012). In spite of the major advantages mentioned above, the constant increasing of plantation areas also raises concerns about the environmental impact and sustainability of this forestry activity. Site preparation practices prior to planting modify soil properties, affecting most physical, chemical and biological processes. Without an effective soil conservation system, the earlier stages of eucalyptus afforestation can expose the soil to the erosive process, leading to soil degradation and, consequently, low productivity.
Several research have been reported the influence of soil management systems and preparation methods concerning erosion rates.In general terms, the higher the aggregate soil breakdown caused by the soil preparation method or less conservation management systems, greater soil losses occur. Studies also show that better soil protection provides lower soil losses due to less soil exposure to erosive agents. This way, the more the crops develop, reduces soil areas unprotected.
By increasing erosion, some soil preparation methods might reduce organic carbon (OC) in the surface layers of soils (SHUKLA; LAL, 2005). Along with organic carbon, nutrients are also lost (BERTOL et al., 2005), especially in highly weathered soils of the humid tropics where most of the cation exchange capacity (CEC) is due to the soil organic matter. Organic matter oxidation as related to soil erosion is also a real concern. Worldwide, accelerated water erosion is responsible for releasing about 1 gigaton (Gt) of carbon per year to the atmosphere (LAL et al., 2004).
In Brazil, some studies about water erosion in forest plantations indicated soil losses below the tolerance levels. For example, in an eucalyptus reforested watershed, soil losses were below 0.08 Mg ha -1 year -1 in an Argisol (RANZINI;LIMA, 2002). In another Argisol from the Coastal Plain region, the observed soil losses ranged from 0.68 to 0.98 Mg ha -1 year -1 , where eucalyptus was planted under minimum cultivation system (MARTINS et al., 2010). These authors registered the following nutrient concentration in the erosion sediment: Ca (0.32-0.56 g kg -1 ), Mg (0.13-0.22 g kg -1 ), P (0.006-0.008 g kg -1 ) and K (0.019-0.029 g kg -1 ). In a bare condition, an Cambisol lost 205 Mg ha -1 year -1 , and the nutrient contents in the erosion sediment were: Ca (0.22 g kg -1 ), Mg (0.06 g kg -1 ), P (0.003 g kg -1 ) and K (0.002 g kg -1 ) (SILVA et al., 2005). In southern Brazil, soil losses in crops are strongly influenced by tillage systems and time of cultivation (BERTOL et al., 2007). Thus, proper measurement of soil, organic carbon and nutrient losses is very important to support selecting and designing management systems for each condition.
Some land uses and management systems might increase soil erosion rates up to intolerable levels. The term soil loss tolerance (T) denotes the maximum rate of soil erosion that can occur and still allows crop productivity to be economically sustainable (RENARD et al., 1997). In this way, agricultural soils can tolerate a certain amount of erosion without adversely impacting on long-term productivity because new soil material is being constantly formed and compensates the losses (BAZZOFFI, 2009). Thus, soil loss tolerance data are useful to evaluate how sustainable would a given eucalyptus management system be.
Considering the current expansion of eucalyptus afforestation in southern Brazil, which occurs mostly on degraded pastures and other marginal areas, knowledge on water erosion under these forests is very important. Therefore, this study aimed to evaluate soil, nutrients and organic carbon losses caused by water erosion in eucalyptus plantations at different development stage, soils and management systems, as compared to native forest (NF) and bare soil (BS) in southern Brazil. According to the Brazilian common knowledge, it is expected that Cambisol (Inceptisol) would present greater soil loss than Argisol (Ultisol), under similar conditions; however, it depends on how the forest would be planted, since a furrow planting system, for example, can cause soil losses by erosion to be greater even in a more resistant soil.

The studied area and soil description
The studied area is located at the Porto Alegre Metropolitan Region (Figure 1), Rio Grande do Sul state, Brazil (30º06'50" S, 51º19'30" W), at 23 m altitude. According to the Köppen classification, the regional climate is Cfa, subtropical humid, characterized by rain in all the months of the year, with temperatures of the warmest and coldest months above 22° C and 3° C, respectively (ARACRUZ CELULOSE, 2006).
The rainfall erosivity was calculated by Faculdade de Agronomia Eliseu Maciel-Faem (2008). The author used the pluviograph data from 1986 to 2005, located at Porto Alegre weather station, and calculated the EI 30 according to Wischmeier (1959). For the 18 boundary counties, within the same physiographic zone, pluviometric data were used to calculate the modified Fournier coefficient, considering rainfall higher than 10 mm as being erosive. Thus, the modified Fournier coefficient can be described as: (1) The areas chosen for this study represent the main soils used for Eucalyptus saligna Sm. cultivation in the Rio Grande do Sul state. These soils were classified according to Brazilian Soil Classification (EMPRESA BRASILEIRA DE PESQUISA AGROPECUÁRIA-EMBRAPA, 2006) and USA Soil Taxonomy (SOIL SURVEY STAFF, 1999). The studied soils were three Red Argisols (RAs) (Ultisols), represented by RA1, RA2, and RA3, which are the main soils in the studied area, and an Haplic Cambisol (HC) (Inceptisol), which is considered as a marginal soil and more likely to be affected by the erosion processes. The slope for each soil is: RA1 (8%); RA2 and RA3 (20%); and HC (45%).

Erosion measurements
The soil erosion experiments were carried out in each of the previously mentioned soil classes under E. saligna cultivation (EUC) at three development stages (2, 3 and 7 years old), bare soil (BS) and subtropical perennial forest -native forest (NF) (reference). This forest has medium and large size trees, forming a dense and closed according to Brazilian Soil Classification (EMPRESA BRASILEIRA DE PESQUISA AGROPECUÁRIA-EMBRAPA, 2006) and USA Soil Taxonomy (SOIL SURVEY STAFF, 1999). The studied soils were three Red Argisols (RAs) (Ultisols), represented by RA1, RA2, and RA3, which are the main soils in the studied area, and an Haplic Cambisol (HC) (Inceptisol), which is considered as a marginal soil and more likely to be affected by the erosion processes. The slope for each soil is: RA1 (8%); RA2 and RA3 (20%); and HC (45%).

Erosion measurements
The soil erosion experiments were carried out in each of the previously mentioned soil classes under E. saligna cultivation (EUC) at three development stages (2, 3 and 7 years old), bare soil (BS) and subtropical perennial forest -native forest (NF) (reference). This forest has medium and large size trees, forming a dense and closed canopy, with about 20 thousand species (for example, Andira sp., Aspidosperma spruceanum Benth.  where Rc m (mm) is the Fournier modified coefficient; p >10mm 2 represents the monthly erosive precipitation (mm); and P >10mm is the annual erosive precipitation (mm).
Correlation between Rc m and EI 30 was calculated in order to estimate the rainfall erosivity factor for each county. Then, the isoerodent map for the studied region was drawn ( Figure 1). All the erosion plots were situated in the same range of rainfall erosivity (4,500 to 5,000 MJ mm ha -1 h -1 year -1 ). During the year, the lower rainfall erosivity values range from July to September and the higher ones range from November to March (FAEM, 2008).
The Argisols were subsoiled at 60 cm depth, and planted in a furrow system in a 3 x 3 m spacing among plants. Planting pit system was used in the Cambisol, in order to have a minimum tillage in this area.
One soil erosion plot (4 x 12m) was built in each place, in order to quantify erosion in E. saligna, BS and NF conditions. The plots were monitored from November 2006 to October 2007. The plots were located at the steepest slope in each condition, so that the worst erosion scenario could be measured in each case. The boundaries of the erosion plots were 0.40 m wide steel sheets, being 0.20 m buried into the soil in order to isolate surface and subsurface water flows. A runoff collector gutter was placed at the lower end of each plot; water and sediments were driven into a collector system consisting of two sedimentation tanks. The first tank (250 L capacity) had a volumetric container inside it were the sediments were collected for low-volume runoff events, and had a Geib type splitter connected with the second tank; the second tank (500 L capacity) was used to collect the excess runoff from the first tank (GEIB, 1933). The amount of runoff in the second tank represents 1/15 of the excess runoff from the first tank.
Before crop plantation, 3 Mg ha -1 of dolomitic limestone was applied in the whole area. This operation is considered essential for eucalyptus cultivation in southern Brazil.
After each rainfall event, water and sediments in the tanks were vigorously and homogeneously shaken; three portions of 250 mL were sampled and transferred to the laboratory for gravimetric processing. These samples were flocculated with an aluminum sulfate saturated solution, the excess water was decanted and the solids were oven-dried at 105°C until constant weight was achieved; erosion rate (Mg ha -1 ) was then obtained for each event.

Soil analyses
Soils under the native forest were sampled. Three replicates of disturbed and undisturbed samples from 0-10 cm depth were collected in September 2006 and individually analyzed. Undisturbed soil samples were collected using an Uhland sampler, for determining the bulk density (BD), according to Grossman and Reinsch (2002), and the total porosity (TP), according to Flint and Flint (2002a). Disturbed samples were used for chemical characterization, which was accomplished according to Embrapa (1997), including: pH in water 1:2.5), extractable cations (1M KCl extractant for Ca and Mg; 0.05M HCl + 0.05M H 2 SO 4 extractant for K and P), and the organic C determinated by potassium dichromate oxidation and ferrous sulfate titration method, according to van Raij and Quaggio (1983). Particle size distribution was determined after dispersion with NaOH, measuring the clay content in suspension by the hydrometer, the sand content by sieving, and the silt content was calculated by the difference (EMBRAPA, 1997). Particle density (PD) was determined according to Flint and Flint (2002b). Soil permeability was measured in the field through a constant rate of water infiltration at 15cm depth, with two loads at 3-and 6-cm water column, using the Guelph permeameter (REYNOLDS; VIEIRA; TOPP, 1992).

Nutrients and carbon organic losses
For measuring organic C and nutrients losses, samples of the erosion sediments collected in the sedimentation tanks, after each rainfall event, were mixed to get an annual composed sample. Available P and K, exchangeable Ca 2+ and Mg 2+ (EMBRAPA, 1997), and organic C (van RAIJ; QUAGGIO, 1983) were determined. It was assumed that the sediment removed by erosion and measured from the erosion plots corresponded to the 0.0-0.10m depth of the soils; therefore we compared the chemical composition of the erosion sediment with that of the original soil depth. Nutrients and OC losses by runoff was not counted in the present study.
The enrichment ratio (ER) was calculated as a measure of the selectiveness of the erosion processes. The ER represents the concentration of the components in the eroded material divided by their concentrations in the soil (MASSEY; JACKSON, 1952). Such a ratio might represent an estimative of components carried by the erosion sediment.

Soil loss tolerance (T)
Based on the morphological, physical and chemical properties of the typical soil profiles, the T value was estimated for both soil classes (Cambisol and Argisol), as proposed by Lombardi Neto and Bertoni (1975), Galindo and Margolis (1989), and Bertol and Almeida (2000), which is more appropriated to the Brazilian conditions; soil effective depth for root growing, clay content ratio between the subsurface and the surface horizons, permeability and organic matter content are the parameters considered for estimating the T value. Average T values were compared to the soil losses in order to know if land use and the management systems were appropriate or if they overexposed the soil to degradation.

Soil losses in reference systems
The total soil losses ranged from 0.030 to 63.795 Mg ha -1 year -1 (Table 1). These values occurred in RA1 under NF and in HC under BS treatments, respectively. Comparing NF and BS systems, the greatest soil loss was found in HC and the lowest in RA1, as a consequence of the higher resistance to erosion of RA1 due to more advanced pedological development stage, under equilibrium conditions (native forest), as opposed to HC which is more susceptible to erosion under these conditions (SILVA et al., 2005;BESKOW et al., 2009), due to the steepness of the relief, the higher silt/clay ratio, and lower soil effective depth of the latter soil (BONO et al., 1996).
For the bare soil condition, where soil exposure imposes higher erosion, soil losses in the HC soil were greater than its corresponding T value (7.3 Mg ha -1 year -1 ,table 1). The steepness of slope (45%), the high silt/clay ratio, and predominance of fine and very fine sand fractions in this soil (Table 2), favored the great soil loss. Soil crusting (impermeable surface layer) helped to increase soil erosion, which is peculiar for this soil. Silva et al. (2005) reported soil losses of 205 Mg ha -1 year -1 for Table 1 -Soil losses by water erosion for the soils and reference systems studied. such a soil class. Even under this bare soil condition, only one of the three studied Argisols showed soil loss value greater than its T value (Table 1). Analyzing BS conditionin the Argisols (Table 1), it can be verified that the soil loss for RA1 was more than four times lesser that the soils RA2 and RA3, most likely due to more softened slope in the RA1.

Soil losses in eucalyptus systems of different development stages
Under the eucalyptus systems, greater soil losses were observed for RA soils rather than HC soil (Table 3), even though, the rainfall erosivity for both sites was the same (Figure 1). This scenario was caused by the furrow planting system in RA soils, which is a much less conservationist system than the pitting planting used in the HC soil, discussed thereafter.
Systems with EUC had intermediate soil losses (Table 3), between the bare soil condition and the native forest (Table 1). The NF system had very low erosion rates (Table 1), as a result of the adequate protection by the canopy of the forest and the plant cover (litter) on the soil surface. The native vegetation cover has been recognized as providing high soil protection against water erosion (ROOSE, 1977;KOULI et al., 2009). In table 3, the influence of the eucalyptus age on soil loss, comparing the same soil and slope classes, can be observed. Comparing RA2 and RA3 (same slope), the younger the plants (EUC 2 years, for both RA2 and RA3) the higher were the soil losses. The 7-year-old plantations had the lowest soil loss among the plant ages, indicating that soil protection increases with increasing the development stage of plant. Like in the NF system, the efficiency of the EUC in protecting the soil against the HC: Haplic Cambisol (Inceptisol); RA: Red Argisol (Ultisol); BS: bare soil (maximum soil loss); NF: native forest (equilibrium system).  direct impact of raindrops is a result of the relatively long crop cycle (7 years), with increasing accumulation of stable plant material on the ground, forming the litter. Likewise, the maintenance of crop residues on the soil surface after harvesting the eucalyptus plants increases the protection against erosion. On the other hand, mechanized harvesting can degrade soil structure; this is considered the main disturbance in planted forests (AJAYI et al., 2009). Harvesting activities are responsible for increasing erosion rates by up to four times (SOUZA; SEIXAS, 2001). In native eucalyptus forests of Australia, the maximum impact on the production of runoff and sediments occurs immediately after road construction and timber harvesting, significantly decreasing after five years (CROKE; HAIRSINE;FOGARTY, 2001). In the present study, the soil loss in the system with EUC (7 years) was about six times lower than that observed for the EUC (2 years) in RA2 soil (Table 3). The largest soil loss observed for EUC (2 years) (Table 3), and also for BS treatment (Table 1) in RA2, compared to RA3 soil is due to the much lower permeability and greater silt/clay ratio of RA2 compared to RA3 soil (Table 2). Thus, plantings implanted right after the harvesting operations will increase erosion as the plants are younger. As the plants grow up, there is a reduction of soil loss and the values tend to approach those observed at the equilibrium condition (NF) ( Table 1).
In regard to the different development stages of eucalyptus, there is a difference between the EUC (2 years) and both EUC (3 years) and EUC (7 years), showing a significant decrease of soil loss after the first two years of the crop development (Table 3).
The soil loss in EUC (3 years) for RA1 was lower than EUC (2 years) for RA2 and RA3, which is due to a   combination of factors, which are the eucalyptus plantation being slightly old (more developed) in RA1, and also situated in a more gentle relief ( Table 2). The lowest soil loss for EUC observed in HC (Table 3) is due to the lower impact of the pit planting system as compared to the furrow system used in RA. In the same region, Baptista and Levien (2010) found similar results, with soil losses up to 1.57 Mg ha -1 year -1 in pit-planted eucalyptus in HC area.
This planting system was so effective in preventing higher erosion that even in the Cambisol, which is known as being naturally more susceptible to erosion, the soil losses were relatively close to that found in RA3 for the 7year-old EUC. These findings are in agreement with practical field observations, where, due to the steeper relief of HC, the harvest operations leave much more residues on its soil surface than in RA, providing additional protection against erosion.
For both soil and eucalyptus ages (Table 3), the soil losses were below the tolerance values (Table 1), indicating that the crop management performed had been adequate regarding the erosion point of view.

Nutrients and organic carbon (OC) losses
Losses of nutrients and OC are proportional to soil losses. The soil under NF (Table 2) was used as a reference for natural fertility in the discussion of the nutrient contents in the erosion sediment. The contents of OC and nutrients in the erosion sediments for each system are shown in table 4. Generally, the sediments are richer in nutrients than the original soils (Table 2), as also noted by some authors (SEGANFREDO; ELTZ; BRUM, 1997;BERTOL et al., 2004;MARTINS et al., 2010). Phosphorus was the nutrient presented in greater contents in sediments as compared to the original soil. Phosphate has strong interaction with clay particles, which are the main size fraction removed by the water erosion. This increased P content in the sediments, as compared to the original soil, can be due to redox reactions (SILVA et al., 2005) and increased pH (Table 2 versus table 4) in the runoff collecting tanks, that can increase P solubility and can also release P from water soluble organic carbon.
For all the situations in this study, there was more Ca than Mg in the sediments, due to the higher Ca concentration in the original soils (Table 2), as also shown by Silva et al. (2005). There was also more K than P in the sediments, because K is effectively more mobile in the soil and less concentrated in the upper layer than P, under the conservationist system, as indicated by Schick et al. (2000) and Bertol et al. (2004).
Organic carbon is one of the first components of the soil to be carried away, and bound to it may be K and P (SEGANFREDO; ELTZ; BRUM, 1997). For eucalyptus systems, the highest contents of Ca, Mg and OC in the erosion sediment were registered for EUC (3 years) in RA1 (Table 4).
The highest concentration of OC in the upper layer of the soils, and the consequent low density of the soil material in this layer, account for the higher contents of OC in the sediment as compared to the topsoil from where it was removed, as noted by Schick et al. (2000). Generally, the lowest concentration of OC in the sediments collected Table 4 -Contents of nutrients and organic carbon (OC) in the erosion sediment.    (Table 5), in agreement with the data from the literature, where such behavior is explained by the erosion selective nature that takes more of the clay fraction where P is adsorbed (SEGANFREDO; ELTZ; BRUM, 1997). Among the soil classes, the highest values of ER were observed in RA1 under EUC (3 years), except for P. In general, systems with 2-3-year-old plantations had ER values greater than 1.0. The ER for EUC (7 years) decreased, when compared to younger eucalyptus systems, again except for P.
The EUC (7 years) system had the lowest nutrient contents in sediment, while the greatest contents were found in the sediments of HC under BS treatment. Nutrient contents tend to be greater in the sediments of eucalyptus systems than in the BS (Table 6), due to the lack of nutrients bio-cycling and the occurrence of crusting observed in the latter condition in the field (RESENDE; CURI; LANI, 1999).  Red Latosols (Oxisols) (between 6 and 10 g kg -1 Water erosion in soils under eucalyptus forest...

167
Ciênc. agrotec., Lavras, v. 37, n. 2, p. 159-169, mar./abr., 2013 Calcium was the nutrient that showed the greatest values in the sediment in all studied systems, ranging from 153 to 3,252 g ha -1 for eucalyptus systems ( Table 6). The highest Ca contents in sediment were due to the higher concentration of this nutrient in the previously limed soil (3 Mg ha -1 of dolomitic limestone) which was applied in the whole area prior to crop plantation; this operation is fundamental for eucalyptus nutrition in subtropical Brazilian conditions.
Organic carbon was highly found in the sediments, especially for HC under BS tratment (957 kg ha -1 ). In the 2-3-year-old plantations, the OC values ranged from 29 to 144kg ha -1 (Table 6). These values are greater than those observed in the literature for eucalyptus plantations in Brazil (OLIVEIRA et al., 2010;MARTINS et al., 2010), which confirms the need for improving conservation practices that reduce erosion, maintaining and/or increasing this organic fraction in the soil, since it is important in enhancing soil structure, moisture retention and CEC, among other important soil properties. In the RA2, the EUC (2 years) was the system with greater amounts of nutrients, especially Ca, which was the nutrient mostly lost in all the eucalyptus systems ( Table 6). The amounts of nutrients observed for EUC (7 years) were the lowest among the eucalyptus systems, which are due to the most protective condition of the crop at this age.

CONCLUSIONS
The Cambisol was more susceptible to erosion, either under the bare soil or the native forest. However, the pit planting system used in this soil area prevented higher losses of soil, nutrients and organic carbon. There is a reduction in soil losses as the development stage of eucalyptus forest increases after the first two years, with soil loss rates close to those observed for native forest, indicating a tendency to achieve sustainability in terms of water erosion. The plantings installed after the harvest operations will favor erosion as the eucalyptus trees are younger.
This paper also showed that soil losses in eucalyptus systems were much lower than the tolerable values, indicating the adequacy of the adopted minimum cultivation regarding water erosion. The results suggest the need for additional care in the management system for the Argisol, mainly related to soil preparation for planting.
There were higher losses of K as compared to P in the erosion sediment; Ca was the nutrient that showed the greatest values in the erosion sediment in all the management systems. The highest concentration of OC in