| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Soil Science, Institute of Ecology, Berlin Univ. of Technology, Salzufer 11-12, D-10587 Berlin, Germany
b 3185 Achilles Dr., Reno, NV 89512
* Corresponding author (wolfgang.wilcke{at}tu-berlin.de)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: n.d., not detected • Ntot, total nitrogen
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Pastures currently covering 0.35 to 0.40 million km2, represent the largest land-use practice of agriculture in this region (80% of the total agricultural area, Resck et al., 2000). The intensity of pasture use ranges from productive regularly fertilized to degraded pastures not receiving fertilizers. The second most important land use is cropping (12 million ha), frequently corn (Zea mays L.)–soybean [Glycine max (L.) Merr.] rotations managed with different tillage practices. The agricultural land use also includes forestations, for example, with Pinus. Thirty-five years ago, more than 2000 km2 were transformed into Pinus plantations (IBDF, 1984; Espirito Santo, 1995).
To determine leaching losses of nutrients in the deeply weathered Oxisols of the Cerrado, computer models frequently based on the convection–dispersion equation can be used. However, parameterization of such models requires intensive measurements of soil hydrological and micrometeorological data at a high temporal and spatial resolution. Such measurements are expensive and labor-intensive and frequently cannot easily be conducted in tropical countries. Furthermore, it is difficult to handle fast transport along preferential flow paths in these models. Alternatively, lysimeters can be used. However, lysimeters either do not collect the percolating soil water quantitatively (Jemison and Fox, 1992) or imply a considerable disturbance of the study soil and impose limitations to their management.
A simple and cheap method that can be easily applied in tropical countries is the use of exchange resin cores to cumulatively collect nutrients transported through the soil profile (Schnabel, 1983; Sakadevan et al., 1994; Bischoff et al., 1999; Lehmann et al., 2001). Anion and cation exchangers have frequently been used in laboratory and field experiments to determine nutrient availability and nutrient fluxes in soils (Skogley and Dobermann, 1996). Up to a saturation of 33 to 50% of the exchange sites with the target anion, high retention efficiencies are reached and once nitrate or ammonium are sorbed negligible microbial transformation occurs (Schnabel, 1983; Binkley, 1984). In field experiments, ion exchange resins have, for example, been used to determine cumulative inorganic N fluxes in vegetable soils of the USA and Japan (Jackson, 2000; Pampolino et al., 2000; Allaire-Leung et al., 2001), inorganic N and phosphate fluxes in North American organic forest soil layers (Johnson et al., 2002), sulfate, nitrate, and base metal fluxes in Australian pastures soils exposed to different inputs of excreta (Sakadevan et al., 1993a, 1993b), and the N, P, and base metal budget in an agroforestry system in Kenya (Lehmann et al., 1998; Peter and Lehmann, 2000). The most important limitation of resin cores is the fact that it is difficult to create similar conditions for water flow in the boxes as in the surrounding soil (Schnabel, 1983; Torbert and Elkins, 1992). Bischoff et al. (1999) have optimized the resin core method and found that it works particularly well in sandy soils. It seems therefore to be suitable for measuring nutrient fluxes in Oxisols with their sand-sized aggregates ("pseudo-sand") and high water conductivity.
The objectives of our study were to estimate the fluxes of base metals (Ca, Mg, K) and inorganic N forms (NH4, NO3) in Oxisols under native Cerrado, Pinus caribaea Morelet plantation, productive and degraded Brachiaria decumbens Stapf pasture and conventional and no-till corn–soybean rotation in the Brazilian savanna with the help of resin core collectors. Furthermore, we used recently published data on nutrient inputs from the atmosphere and by fertilization (Lilienfein and Wilcke, 2004) and on the change in nutrient storage during the past 12 to 20 yr (Lilienfein and Wilcke, 2003; Wilcke and Lilienfein, 2004) to set up nutrient budgets of the studied native and land-use systems.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
All studied soils were very-fine isohyperthermic Anionic Acrustoxes (Soil Survey Staff, 1997) or Latossolos vermelhos escuros and Latossolos vermelhos-amarelos according to the Brazilian soil classification (EMBRAPA, 1999). These soils have high clay concentrations (615–885 g kg–1) to a large extent forming stable sand-sized aggregates ("pseudo-sand"). The pH in KCl ranged between 4.0 (Cerrado) and 5.4 (conventional tillage). All study soils developed from fine limnic sediments of the lower Tertiary. The soils were homogeneously weathered to a depth of several meters. Lilienfein et al. (1999) have shown that the soils were sufficiently homogeneous before the beginning of land use by comparing the particle-size distribution and the mineralogical composition (concentrations of dithionite-citrate soluble Fe and oxalate-soluble Al) to attribute most observed changes to the effect of different land use. The soils did not contain stones.
Within an area of about 100 km2, one plot of each of (i) native Cerrado, (ii) pine plantation, (iii) productive and (iv) degraded pasture, and (v) conventional and (vi) no-till cropping was selected. Details on the composition of the Cerrado vegetation are reported in Lilienfein et al. (2001b). The native vegetation was a typical Cerrado (Goodland, 1971). It was characterized by an open grassland with a 15 to 40% cover of 3- to 5-m high trees. Dominant tree species in the layer >2 m were Pouteria torta (Mart.) Radlk., Ouratea spectabilis (Mart.) Engl., Roupala montana Aubl., Byrsonima coccolobifolia H.B. et K., Dalbergia miscolobium Benth., Kielmeyera coriacea Mart., and Caryocar brasiliense Cambess., which together represented 70% of the biomass of the >2-m layer. The dominant grass species were Andropogon minarum Kunth, Axonopus barbigerus Hitchc., Tristachya chrysothrix Nees, and Echinolaena inflexa (Poir.) Chase.
To establish the Pinus caribaea Morelet plantation, natural savanna vegetation was cleared by harvesting the trees, including their large roots. Pinus caribaea trees were planted in 1977 into plant holes and fertilized with about 33 kg Ca ha–1, 13 kg P ha–1, and 20 kg S ha–1 (80 g of superphosphate per tree was applied to 1670 planted trees ha–1) at planting. The soils were not plowed, weeds were not controlled, and there were no further fertilizer applications. At the time of our study (1997–1999), there were about 950 trees ha–1 with an average height of 21 m. The average diameter at approximately 1.4 m ("breast height") was 243 ± 38 mm.
The most important criteria for the selection of the pasture plots were the current visual impression and the high likeliness that the pasture use followed directly after the clearing of the native vegetation. The degraded pasture showed the characteristics Lopes et al. (1999) described as typical for degraded pastures of the Cerrado region: decreased grass cover compared with the productive pastures, followed by the invasion of Cerrado plants. The productive pasture, in contrast, was a pure grass pasture of Brachiaria decumbens, an imported grass species from Africa, with a closed vegetation cover. The pastures under study were established by harvesting the native Cerrado vegetation including the large roots. They were established around 1985. The most common procedure was to plant upland rice (Oryza sativa L.), which was fertilized with about 40 kg P ha–1, 65 kg K ha–1, 32 kg N ha–1, and 1 Mg of dolomite ha–1. The rice fields were undersown with Brachiaria decumbens. In 1996–1997, we fertilized the productive pasture with 17 kg P ha–1 and 33 kg K ha–1. Previously, the plot received 17 kg P ha–1 and 33 kg K ha–1 at 4-yr intervals (i.e., in 1988, 1992, and 1996).
The conventional tillage soil has been plowed with a disk harrow two to three times per year for 12 yr and used for corn–soybean rotation. The no-tillage system was established 3 yr before the beginning of our experiment in 1997 after the plot had been managed with conventional tillage in the way described above for 9 to 11 yr. One major difference between the no-tillage and conventional tillage systems in the study region is that in the no-tillage system, two crops may be grown in one rainy season, while the conventional tillage system only includes one crop. Both cropping systems (no tillage and conventional tillage) were fertilized with an annual average of about 70 kg N ha–1, 100 kg P ha–1, and 160 kg K ha–1. In the rainy season 1998–1999, we planted soybean in both cropping systems on 9 and 10 Nov. 1998. Both cropping systems were fertilized on 29 October with 42 kg P ha–1 and 63 kg K ha–1 applied as Ca(H2PO4)2 and KCl. In the no-tillage system, 1.4 kg ha–1 glyphosate was applied to control weeds. Before planting, conventional tillage soils were manually hoed on 23 October and 5 November to simulate disk harrowing.
All study sites had slopes below 1°; they have been continuously used for the same purposes for 12 yr (degraded pasture, productive pasture, no tillage, conventional tillage) or 20 yr (Pinus) and passed directly from natural vegetation to the current land-use system except for the no tillage soil.
Installation of the Exchange Resin Cores
We used PVC plastic tubes with a diameter of 0.10 m and a length of 0.12 m as cores. At the lower end of the core, a 1-cm slice was cut and a 0.5-mm polythene net was introduced between the lower (0.01 m) and upper (0.11 m) PVC rings and fixed with a two-component adhesive to hold the soil and exchange resin mixture. The resin cores were filled with a 1:10 (v/v) mixture of exchange resin and field-fresh soil taken from the location where the core was later installed. We used the strongly acidic cation and strongly alkaline anion exchange resin Amberlite MB-20 (Merck, Darmstadt, Germany, cation and anion exchange capacity: each 1 molc L–1). By this choice we followed the recommendation of Lehmann et al. (2001) who tested eight different exchanger resins for field application in resin cores to determine fluxes of NH4, NO3, and Mg in Brazilian Oxisols and found that Amberlite MB-20 showed the best nutrient retention. The nutrient retention by Amberlite MB-20 was not influenced by several drying–rewetting cycles. The exchange sites were completely covered with HBr by batching the exchange resin for 2 h with concentrated HBr (in pro analysi quality). After saturation with HBr, the resins were washed with deionized water to remove remaining HBr until neutral pH was reached in the washing water. The cores were installed into the walls of soil pits in April 1997. The cores at 0.3-m depth were distributed over the whole plot and those at 0.8-, and 2-m depths were installed into the walls of one 2.5-m deep soil pit. For installation, an approximately 0.3 m deep cave was manually dug into the walls of the pits and the core was pressed from bottom to top into the undisturbed soil above the core and fixed by filling the gap between the lower end of the core and the underlying soil with soil material from the same depth. We installed five (no-till, productive pasture) to six (Cerrado, Pinus, degraded pasture, and conventional tillage) replicate resin cores at the 0.3-m depth, three at the 0.8-m depth (all systems) and three (Cerrado, Pinus, pastures, no-till) to six (conventional tillage) at the 2-m depth. The cores at the three different depths had a horizontal distance of at least 0.5 m from each other so that the overlying cores did not affect the deeper ones. All cores were removed in April 1999.
For nutrient analysis, cores were cut into four slices: 0 to 30, 30 to 60, 60 to 90, and 90 to 110 mm and analyzed separately. Furthermore, aliquots of the soil used to fill the cores were collected as control samples.
Physical and Chemical Characterization
Bulk density of the soil was determined gravimetrically using 0.1-m steel rings at the time of installation in April 1997. As an approximate of the surrounding soil, bulk densities of the 0.15- to 0.30-, 0.30- to 0.80-, and 1.2- to 2.0-m layers, determined with five representatively collected replicate steel rings, were taken for the resin cores at the 0.30-, 0.80-, and 2.0-m depths, respectively. Bulk density of the resin core slices was determined by weighing the dried mass of each slice after removal of the cores from the soil.
Resin core slices were air-dried. Accumulated ions were exchanged with 2 M HBr (Ca, Mg, K) or 2 M KCl (NH4, NO3) by sequentially shaking 25-g aliquots of the soil–exchange resin mixtures two times (each 30 min.) on a rotational agitator. In the extracts, Ca, Mg, and K concentrations were determined by atomic absorption spectrometry (Varian SpectrAA 400, Varian Inc., Mulgrave, Australia) and NH4–N and NO3–N concentrations photometrically with the continuous-flow analyzer (SANplus, Skalar, Breda, The Netherlands).
Data Evaluation
To determine ion fluxes, the ion concentrations of the control samples (aliquots of the soil used to produce the resin cores collected on the date of installation of the resin cores) were subtracted from those of the material in the cores. Ions stored in the 0- to 30- and 30- to 60-mm slices were considered as resulting from downward fluxes, those stored in the 90- to 110-mm slices were attributed to upward fluxes and the ions in the 60- to 90-mm slice were 50:50 partitioned between upward and downward fluxes. Mean values of ion fluxes were tested for differences between plots by using Tukey's honest significant difference means separation test (Spjotvoll/Stoline modification for uneven sample sizes). Statistical analyses were performed with STATISTICA for Windows 5.1 (StatSoft, Loll and Nielsen, Hamburg, Germany). Significance was set at p 
0.05.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
To ensure that the hydrological conditions in the exchange resin cores are similar to surrounding undisturbed soil, it is important to match as closely as possible the bulk density of the core with the bulk density of the undisturbed surrounding soil.
Nutrient Concentrations
With few exceptions, there were relative concentration minima in the third (60–90 mm) layer. However, in none of the resin cores, ion concentrations in this layer were negligible. We assume that this was the consequence of the long accumulation time of 2 yr allowing for multiple desorption/sorption processes associated with vertical ion movement in the resin core. Furthermore, soil organic matter was mineralized inside the resin cores and retained on the adsorber resin. There were six cases (of a total of 6 systems x 3 depths x 5 ions = 90 cases), in which the highest concentrations occurred in the third layer. In all of these cases but one (NH4 at 0.8-m depth under no-till), the second highest concentration occurred in the lowermost slice. It is likely that upward movement of ions played a major role in these six resin cores. If this was true, we might have underestimated the upward flux and overestimated the downward flux in these six cases.
In the native Cerrado, most maximum mean concentrations of the replicate resin cores occurred in one of the uppermost two slices. Figure 1a shows the mean depth distribution of the ion concentrations in the resin cores from the 0.3-m depth under native Cerrado as an example for the non-fertilized soils. In these non-fertilized soils, there were several cases in which we observed highest mean ion concentrations of a sampling depth in the lowermost slice, particularly at greater soil depth. Under native Cerrado, highest mean NH4–N concentrations occurred in the lowermost slice at the 0.3- and 0.8-m depths and highest mean Mg concentrations at the 2.0-m depth. At the 0.3-m soil depth, highest mean NO3–N concentrations and at the 2.0-m soil depth, highest mean Ca concentrations occurred in the 60- to 90-mm layers, followed by the second highest concentrations in the 90- to 110-mm layers. Under Pinus, most mean concentration maxima of the replicate resin cores occurred in the lowermost slice, except K at the 0.3- and 0.8-m depth, Mg at the 0.3-m depth, Ca at the 0.8-m depth, and NO3–N at the 0.3- and 2.0-m soil depths. This indicated that upward movement of nutrients played an important role under the forests.
|
In the only initially or regularly low level fertilized pastures, in about half the cases maximum mean concentrations were found in the uppermost two slices, and in the other cases the lowermost two slices, indicating that upward and downward ion fluxes were similar.
Overall, depth distribution of ion concentrations in resin cores met the prerequisites that allow for quantitatively interpreting the results although differences in ion concentrations between various slices of the resin cores were partly small. The frequent occurrence of concentration maxima in the lower part of the resin cores points at an important role of capillary rise of water moving plant nutrients upward in this soil. The fact that there was no slice with negligible ion concentrations and the assumed important role of upward ion fluxes did not allow for unambiguously separating downward and upward fluxes. It cannot be completely ruled out that a complete breakthrough of downward ion fluxes through the resin cores occurred or that capillary rise of ions reached the uppermost two slices. Our decision to partition the third slice equally among downward and upward fluxes is arbitrary and introduces an unknown bias into our flux estimate. We assessed this error by calculating a minimum and a maximum downward flux. The concentration changes in the uppermost two slices relative to the controls were considered as a minimum estimate of ion fluxes in the study soils and those in the whole resin core as a maximum estimate (neglecting upward fluxes). Furthermore, our estimate of N fluxes was biased by N mineralization inside the boxes. We therefore estimated the contribution of N mineralization to the total inorganic N retained in the resin cores.
Bulk Densities
Most mean bulk densities of all slices of the resin cores at all depths in the six study sites were lower than those of the undisturbed soil (Table 1). Mean bulk densities of the 0- to 30- and 90- to 110-mm slices in the six land-use systems were frequently significantly lower than those of the surrounding soils at all study depths. These uppermost and lowermost slices at the 0.3-m soil depth only reached, 50 to 91 and 51 to 111%, respectively, of the bulk density of the undisturbed soil. Mean bulk densities of the 30- to 60- and 60- to 90-mm slices of the resin cores (70–110% of the bulk density in undisturbed soils) were close to those of the undisturbed soils with some exceptions, particularly at the 0.3- and 0.8-m soil depths. Generally, lower bulk densities in the resin cores indicate that we were unable to compact the exchange resin/soil mixture to the original density. The low measured bulk density of the upper- and lowermost slices of the resin cores, particularly at the 0.3- and 0.8-m soil depths may additionally be attributed to some loss of soil material during recovery of the resin cores. The lower bulk density could have resulted in a higher water conductivity implying a locally increased water flow through the resin cores and thus an overestimation of the fluxes. However, the overall water conductivity of the resin cores is determined by that of the two central slices (30–90 mm), which in most land-use systems only had a slightly lower bulk density than the surrounding soils. The loss of resin core material during sampling would, in contrast, result in an underestimation of the fluxes.
|
|
At the 0.3-m depth under native Cerrado and degraded pasture and at the 0.8- and 2.0-m depths under productive pasture the entire inorganic N accumulated in the resin core can be attributed to mineralization. Thus, for these resin cores no quantification of N fluxes was possible. At the 0.3-m depth under productive pasture, cropping systems, and Pinus, mineralization contributed approximately 20 to 40% to the downward fluxes and approximately 30 to 60% to the upward fluxes. At the 0.8-m depth under native Cerrado and cropping systems mineralization contributed approximately 20% to the downward flux estimates and approximately 15 to 30% to the upward flux estimates but only <10% to upward and downward fluxes under Pinus and degraded pasture. At the 2.0-m depth, the contribution of mineralization to N flux estimates was <10% for Pinus, degraded pasture, and cropping systems, and approximately 10% for Cerrado. Thus, the error of our flux estimated at the lower border of the ecosystem (2.0-m soil depth) used for the balance because of organic matter mineralization in the resin cores is small except for the productive pasture where, however, almost no N fluxes occurred.
We do not expect a similar effect on the base metals because they are not released during mineralization. As all base metals were exchangeable in these highly weathered soil containing no weatherable minerals, we also do not expect that the addition of the resin resulted in a more pronounced base metal mobilization than would occur without resin.
Lessa and Anderson (1996) estimated annual nutrient leaching rates in laboratory experiments with packed columns of soil material from a Brazilian Oxisol used for slash and burn agriculture to be 2.8 g Ca m–2 yr–1, 1.3 g Mg m–2 yr–1, 6.0 g K m–2 yr–1, and 7 to 10 g N m–2 yr–1. While the Ca leaching rate is well within the range of downward fluxes in our study, those of the other elements are at the upper end of the range of fluxes in our study (Mg) or by a factor of three to four higher than our highest leaching rate (K, NO3–N, Table 1). Higher leaching rates in the column experiment than in our field experiment may be attributed to the disturbance by packing the columns. In a Togolese Oxisol used for corn cultivation, Poss et al. (1996) found K leaching rates determined with the help of ceramic suction cups and a drainage estimate to be 0.45 g K m–2 yr–1 (no K fertilizer) and 0.75 g K m–2 yr–1 (application of 13.7 g K m–2 yr–1), which fell well within the range of K leaching rates observed at our study sites. The downward NO3–N fluxes under no- and conventional tillage are similar to those reported by Lehmann et al. (2004), for the same study sites. They determined total N (Ntot) fluxes of 5.7 to 7.8 g m–2 yr–1 at the 0.15-m soil depth and 1.9 to 2.6 g m–2 yr–1, based on a simple water balance model. In the study of Lehmann et al. (2004), NO3–N contributed 30 to 80% to Ntot in soil solution. In a column experiment using a Malaysian Oxisol from a rubber plantation, 9.2 g NO3–N m–2 yr–1 was leached, a factor of three to four higher than in our study (Wong and Rowell, 1994). However, higher NO3–N leaching rates in this column experiment can again partly be attributed to the disturbance of the soil during column packing. Sierra et al. (2003), in contrast, did not detect NO3–N leaching to below 0.8 m in an Oxisol of the French West Indies, being similar to our finding of low or no net NO3–N leaching losses from Cerrado and productive pasture (Table 2). From the above compilation of nutrient leaching data collected in field and laboratory experiments with Oxisols from other locations, we conclude that our flux estimates are realistic.
Fluxes of Ca, Mg, and K determined with the resin core method were significantly correlated with their mean concentrations in soil solution (r = 0.82 for Ca, 0.81 for Mg, and 0.77 for K). For NO3–N, there was still a positive but not significant correlation between mean concentrations in soil solution and fluxes estimated with resin cores reflecting the larger error of the N flux estimates (data were log-transformed before correlation analysis because they were not normally distributed, plots for which N fluxes could not be quantified were removed). The soil solution data are unpublished. Soil solution was collected with five replicate ceramic suction cups at each of the 0.3-, 0.8-, and 2.0-m depths at all study plots. Metals were analyzed with atomic adsorption spectrometry and NO3–N with photochemical methods. More details on collection and analytical methods can be found in Lilienfein et al. (2000a). Ammonium concentrations in soil solution were consistently below the detection limit. We conclude that the resin core method reflected well the chemical composition of the soil solution.
To further confirm that the flux estimates with the resin core method are plausible, we roughly estimated leaching rates of Ca, Mg, K, and NO3–N by assuming an annual drainage rate of approximately 400 mm as suggested by Lehmann et al. (2004), based on calculations with a simple water budget model at the same study sites. Rough measures of element fluxes were then derived as product of the mean concentration in soil solution and the drainage rate. Figure 2 illustrates data points are scattered around the 1:1 line. At low mean concentrations of Ca, Mg, K, and NO3–N in the soil solution, the resin core method tends to measure higher fluxes. At high mean concentrations in the soil solution, it tends to measure lower fluxes. We nevertheless consider this result as further confirmation of the reliability of the ion flux estimates with the resin cores.
|
Nutrient Leaching
With few exceptions, there were higher downward than upward fluxes, indicating that ions were leached, on balance, to the deeper subsoil in all studied systems (Table 2). Net downward fluxes of Ca at the 0.3-m depth were significantly higher in the cropping systems and productive pasture than in all other systems, those of K and NO3–N at the 0.3-m depth were significantly higher in cropping than in all other systems, except that net NO3–N fluxes at the 0.3-m depth were not significantly different between cropping systems and Pinus (Fig. 3) . There were no significant differences in Mg and NH4–N fluxes at any depth and in the fluxes of all studied ions at the 0.8- and 2-m soil depths. There was nevertheless a clear trend of highest net downward fluxes under no-till for all ions at the 0.8- and 2.0-m soil depths, indicating this system lost the most nutrients of all systems (Table 2). This finding corroborates earlier observations of a faster ion transport in no-till than in conventional tillage soils (Lilienfein et al., 2000a, 2000b). It is also in line with findings of Wu et al. (1995) and Azooz and Arshad (1996) who reported that increased infiltration rates and higher saturated hydraulic conductivity under no-till than under conventional tillage may lead to faster solute transport. However, contrasting results are reported concerning the effects of tillage practices on nutrient leaching. In a lysimeter study in Kentucky, Tyler and Thomas (1977) found higher NO3 and Cl leaching under no-till than under conventional tillage. Drury et al. (1993) and Angle et al. (1989)(1993) reported higher NO3 concentrations in drainage water, soil solution or ground water under conventional tillage than under no-till in Canada and Maryland. This suggests that nutrient leaching does not only depend on tillage practices but also on environmental conditions and other management practices.
|
Nutrient Balances
To set up nutrient balances we partly used unpublished data for our study plots. Briefly, regional above-canopy deposition was determined with five Hellmann-type rainfall collectors at three stations in the study area. Dry deposition was estimated with the help of a canopy budget approach based on above- and below-canopy deposition rates. Throughfall was measured at each plot with five collectors of the same type as the rainfall collectors at 0.3 m above the soil surface. Fertilizer input was estimated based on farmer interviews and records. The methods are described in full detail in Lilienfein and Wilcke (2004). Nutrient removal by harvest was determined by analyzing nutrient concentrations in five representatively collected plants and multiplying by the harvested biomass. To estimate N fixation it was assumed that all N in soybean was fixed and that every second year soybeans were grown (Lilienfein and Wilcke, 2003). The flux balance was furthermore compared with a mean annual accumulation–depletion rate based on the determination of total nutrient storages in the studied systems and comparison of the current nutrient storages in the five land-use systems with those in the native Cerrado. The accumulation–depletion rates were derived by dividing the differences in nutrient storage between land-use system and native Cerrado by the age of the land-use system (12–20 yr). More details on this approach are presented in Lilienfein and Wilcke (2003) and Wilcke and Lilienfein (2004). We did not calculate N balances because we were unable to determine organic N leaching with the resin cores and at some soil depths under native Cerrado and the pasture systems N fluxes were so small that even inorganic N fluxes could not be determined.
In the Cerrado soil, base metal (Ca, Mg, K) leaching to below the 2.0-m soil depth was smaller than their input, indicating that native Cerrado gained nutrients (Table 3). This is unexpected because it can be assumed that native ecosystems are in steady state, that is, inputs and outputs are balanced. The result therefore suggests that there are anthropogenic element inputs. This might be the result of emissions from Uberlândia, a nearby city with approximately 400000 inhabitants and some agricultural industry, local charcoal production, and the agricultural activity in the surroundings of the study plots. In the top 2 m of the Pinus soil, all base metals accumulated. The accumulation rates derived from the flux balance match closely those estimated by dividing the difference in total nutrient storages of the systems between Pinus and native Cerrado by the age of the Pinus stands (20 yr, Lilienfein et al., 2001b; Lilienfein and Wilcke, 2003; Wilcke and Lilienfein, 2004). Pasture systems also gained base metals on balance. However, base metal accumulation rates based on flux estimates do not as closely match those derived from comparison of total nutrient storages of pastures with that of native Cerrado (Lilienfein and Wilcke, 2003; Wilcke and Lilienfein, 2004), although they are similar in size. The base metal budgets of the cropping systems were balanced (Ca in conventional tillage) or negative. As for pasture systems, accumulation–depletion rates determined via total base metal storages in systems partly differ from those derived with our balance approach. Nevertheless, net nutrient fluxes determined with the two approaches are again similar in size.
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
We conclude that only in the cropping systems, particularly the no-till system, is there a risk of some nutrient depletion because of the current management. To reduce these losses, timing of fertilizer amendments should be optimized and in the no-till system evaporation and fast water fluxes reduced, for example, by stubble mulching.
|
ACKNOWLEDGMENTS |
|---|
Received for publication November 8, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Minoshima, L.E. Jackson, T.R. Cavagnaro, S. Sanchez-Moreno, H. Ferris, S.R. Temple, S. Goyal, and J.P. Mitchell Soil Food Webs and Carbon Dynamics in Response to Conservation Tillage in California Soil Sci. Soc. Am. J., May 16, 2007; 71(3): 952 - 963. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education</ | |||||
