CHANGES IN SOIL ORGANIC MATTER UNDER DIFFERENT LAND MANAGEMENT IN MISIONES PROVINCE ( ARGENTINA )

Highly weathered tropical soils rapidly loose soil organic matter (SOM) and may be affected by water erosion and soil compaction after deforestation and intensive cultivation. With the main objective to estimate the SOM balances in a subtropical soil we determined the dynamics of SOM in a degraded yerba mate (Ilex paraguaiensis Saint Hil.) plantation introduced after deforestation and with elephant grass (Pennisetum purpureum L.) as a cover crop. The study site was in Misiones, Argentina, and we use the natural C abundance methodology and a descriptive model. The study was conducted on three contiguous 50 × 100 m plots of a typic Kandihumult soil with: (i) native forest, (ii) 50 years of continuous yerba mate monoculture with intensive tillage, and (iii) yerba mate associated with elephant grass as a cover crop and no tillage. We determined bulk density, carbon (C), nitrogen (N) and C content of the soil (0 0.05, 0.05 0.15 m layers) and the grass biomass. Yerba mate monoculture reduced soil C and N content as well as porosity at 0 0.15 m depth by 43 and 23%, respectively, as compared to the native forest. After ten years of yerba mate elephant grass association soil C and N contents at the same depth increased by 19 and 12%, respectively, compared to the yerba mate monoculture, while soil porosity remained similar. Total C input, C, and soil organic C were incorporated into a three compartment model to evaluate elephant grass C dynamics. Through the natural C abundance methodology we tracked the elephant grass C incorporation and the “old” soil C loss, and determined the model parameters humification (k1) and mineralization (k) coefficients and stable C (Cs)unambiguously. The high k1 and k predicted by the model are probably explained by elephant grass root system incorporation under no tillage and humid subtropical climate, respectively. In soil under yerba mate monoculture, Cs was counted as 91% of the total soil organic C.


Site description
The study was carried out in Oberá, Misiones province, Argentina (54º80' W 26º40' S, 350 m above sea level).The soil is a Typic Kandihumult; it is highly weathered, red in color and rich in low activity clays (≈ 57%).The climate is sub-tropical humid with no dry season (mean annual precipitation is 1,970 mm) and the regional mean temperature of the coldest and warmest months is 15.6 and 25.5ºC, respectively.Landscape relief is considered as rolling.
The study was conducted in three adjacent 50 × 100 m plots, selected for their uniformity in soil profile development.The three plots are at the same topographical position (mid-hillslope with very long ca.1% slopes) in order to avoid differences in the initial soil profiles that could confound treatment effects.Fertilizer and lime were not applied in the studied sites.
The systems compared were: (i) Virgin soil with the typical subtropical forest (Trichilia elegans, Ocotea acutifolia, Nectandra saligna, Balfourodendrum riedelianum, Chrisophilum marginatum, Matayba eleagnoides, Cabralea oblongifolia, Apuleia leocarpa, Inga afnis, Lonchocarpus leucanthus, Loncocarpus muehlbergianus, Cedrela forsilis, trichilia catigua (Cabrera, 1976).Extraction of the more valuable species enabled the establishment of new ones, maintaining constant the number of species.(ii) A fifty-year yerba mate monoculture.Weeds between rows were continuously controlled with disk-harrow to 0.15 m depth and herbicide.The excessive weed control favored the pruning of yerba mate roots.Leaves and branches of yerba mate were harvested every year leaving scarce surface residue.(iii) A ten-year yerba mate -elephant grass association with no tillage.The elephant grass was inter-cropped on a fifty-year plantation ten years before this study.During summer-time the grass had been cut three times a year and left on the soil surface as a mulch cover.

Sample collection and analysis
Soil sampling was done at the time of the elephant grass implantation and at six and ten years later

INTRODUCTION
Soil organic matter (SOM) is an important determinant of tropical soils fertility (Tiessen et al., 1994) and the main carbon (C) storage in terrestrial ecosystems (Post et al., 1982;Eswaran et al., 1993;TSBF-CIAT, 2004;Zingore et al., 2005).Highly weathered soils of the tropics are characterized by a rapid decline in SOM when land is cleared for cultivation on a slash-burn crop sequence (Ogle et al., 2005;Jimenez & Lal, 2006).Deforestation is frequently accompanied by loss of essential nutrients through soil erosion and by soil compaction (Eneji et al., 2003;Cerri et al., 2007), particulary in Misiones (NE Argentina), where about 200,000 ha of yerba mate (Ilex paraguaiensis Saint Hil.) have been planted.SOM and fertility loss following clear-cutting have been reported for the soils of Misiones (Piccolo et al., 1998) and fifty percent of the soils under yerba mate are affected by moderate or severe water erosion (Grúner, 1955;Musto et al., 1983;Casas et al., 1988).It is necessary to develop local sustainable agriculture systems that minimize SOM losses and optimize the inputs efficiency.Grasses add large amounts of SOM in the soil.Elephant grass (Pennisetum purpureum L.), a frequently used grass for cover crops (Opara-Nadi, 1993;Scopel et al., 2004), was introduced in NE Argentina in the last few years, Soil C balance under grasses depends on the stability of the SOM derived from the previous vegetation, the rates of SOM input from the new established grass and its stability.When trees and shrubs (dominantly C 3 species) are replaced by grasses (predominantly C 4 ) it is possible to determine the origin of the SOM from its isotopic composition (Balesdent et al., 1987;Chonè et al., 1991;Bonde et al., 1992;McDonagh et al., 2001;Roscoe et al., 2004;Desjardins et al., 2004;Schwendenmann & Pendall, 2006).Furtherover, the use of 13 C natural abundance technique together with simple mathematical approaches have further improved SOM turnover studies (Noordwijk et al., 1997;Bernoux et al., 1998;Wynn et al., 2006).
The main objective of this study was to compare the dynamics of the stable and labile C pools of SOM, using the 13 C abundance technique in: (i) native forest, (ii) degraded yerba mate plantation and (iii) a similar plantation associated with elephant grass.We also propose a descriptive model to estimate the stable in the three situations.Four samples of forest litter, elephant grass mulch and yerba mate residues were collected from each plot using a 0.25 m 2 ring sampler.Four composite soil samples (of seven sub samples each) were extracted from each system at 0 -0.05, 0.05 -0.15 m, and also at 0.15 -0.30 and 0.30 -0.60 m under the forest.The 0 -0.15 m depth corresponds to the A horizon and the two others to the B1 and part of B21 horizons, respectively.Sampling depths were chosen considering that SOM does not differ between these systems below the 0.05 m (Morrás & Piccolo, 1998) and that soil tillage was always superficial (≤ 0.15 m).Samples were dried at 70ºC, passed through a 0.002 m sieve to remove plant residues and mortar ground to pass through a 0.0001 m sieve.C, N and 13 C contents were determined with a mass spectrometer (Fisons/Isochrom).A 2.50 m long ditch was dug in each system and four soil cores ( ≅ 0.07 dm 3 ) were extracted at the same sampling depths to determine soil bulk density.
Elephant grass C and N inputs were calculated as the product between biomass (aerial + root) and total C and N concentrations.Elephant grass biomass was measured during its sixth year after implantation, when it had reached constant biomass production, by harvesting 0.5 m 2 before each cut.Roots from the 0 -0.15 m soil layer were collected (from the same surface area) and separated from the bulk soil using a fine jet of water, on a stacked sieve (0.2 mm mesh screen) set above a sedimentation vat.Then they were dried at 70ºC for two days, weighed and ground to a fine powder.C and N concentration were determined by dry combustion using the same mass spectrometer.
Measurements of natural δ 13 C abundance were used to establish the origin of the SOM in each system using the technique developed by Cerri et al. (1985) and Balesdent et al. (1987).The proportion of soil C derived from elephant grass residues (α P ) was calculated as: where δ m and δ F are the δ 13 C values of the yerba mate -elephant grass association and the yerba mate monoculture soils, respectively, and δ P is the δ 13 C value of the elephant grass material.δ F was measured in soil samples from the yerba mate monoculture before elephant grass implantation because it is in this soil that elephant grass is introduced.The C contents (g kg -1 soil) derived from elephant grass residues (C DP ) and native forest (C DF ) was calculated as: (2) where C m represents the total C content of bulk soil (g kg -1 soil) The C stocks (total, originated from elephant grass and from native forest, in Mg ha -1 ) were calculated using the equation ( 4): where C x represents C m , C DP or C DF , d is the depth (m) and ρ is the bulk density (Mg m -3 ) of the corresponding soil depth.The same equation was used for soil N stock.
To study the dynamics of C under the association of yerba mate and elephant grass, the model proposed by Andriulo et al. (1999) was used.In this three-compartment model (Figure 1), SOM is separated into stable and active fractions and the evolution of the C stock is described by the following equation: For time (t) = 0 where C s is the stable C content (Mg ha -1 ), C a0 is the initial active C content (Mg ha -1 ), C m0 is the initial total C content (Mg ha -1 ), m is the annual mass of C input (Mg ha -1 year -1 ), k 1 is the humification or isohumic coefficient (unit-less), k is the annual coefficient of mineralization of the active fraction (year -1 ), q is the mass of C humifed annually (Mg ha -1 year -1 ), C max is the maximum quantity of soil C originated from elephant grass (Mg ha -1 ) and C meq is the total quantity of soil C at equilibrium (Mg ha -1 ).Natural 13 C isotopic abundance was used to estimate the size of the model compartments and their evolution.The model considers the changes in "old" C stock (C 1 ), originated from forest, and "young" C stock (C 2 ), originated from elephant grass, separately.Contribution of yerba mate was considered null.Then, equation ( 5) can be separated in two terms: with and The term C 1 describes the decrease of the C content present at time zero; we call it "old C", since it includes the stable carbon pool C s .The term C 2 represents the net amount of active carbon produced since time zero.We have fitted the two models simultaneously, by minimizing the sums of squares of model deviations for the variables C 1 and C 2 .Whilst optimizing, we have taken into account the variance of the data by weighting the sums of squares of deviations.The following quantity (unitless) was minimized: where SSQ 1 and SSQ 2 are the sums of squares of deviations (calculated -simulated) of C 1 and C 2 , respectively and 2 1 s and 2 2 s are the mean calculated variances obtained for C 1 and C 2 , respectively.Fitting was performed by fixing C m0 and m and optimizing three parameters: C s , k 1 and k.
Carbon and N contents, bulk density and natural δ 13 C abundance at 0 -0.05 and 0.05 -0.15 m soil depths and C and N stocks at 0 -0.15 m were compared among systems using the Student test.Statistical significance were considered at P < 0.05 (*) and P < 0.01 (**).

RESULTS AND DISCUSSION
Mean soil C (N) stocks (0 -0.15 m soil depth) of the yerba mate monoculture and the yerba mateelephant grass association were 50 -70% of those from the natural forest (Table 1).These results demonstrate the rapid decrease in the SOM reserves of this tropical highly weathered soil under management.The C and N content of the forest soil was three and two times larger than the one under continuous yerba mate cultivation in the 0 -0.05 and 0.05 -0.15 m layers, respectively.Forest soil bulk density was 65 and 78% lower than in the yerba mate monoculture in the 0 -0.05 and 0.05 -0.15 m layers, respectively.The sharp decrease in C and N content and in soil porosity following the slash and burn crop sequence is typical of tropical low activity-clay soils (Casas et al., 1983;Gupta et al., 1989;Cerri et al., 2007)   Although soil erosion was not measured in this study, it has been reported in Misiones under similar conditions: Casas et al. (1988) and SAGPyA (1995) found soil losses of 25 Mg ha -1 year -1 .Applying this rate of soil loss to our condition, we would have lost 1,250 Mg ha -1 after fifty years.Calculated soil mass from 0 -0.30m depth under forest (3,455 Mg ha -1 ) is similar to soil mass of the Ap horizon plus the soil mass loss under yerba mate cultivation (2065 + 1250 = 3375 Mg ha -1 ).Assuming this to be true, the original C and N stocks under forest were 83.5 and 9.0 Mg ha -1 , respectively.Then, soil organic C and N would have been lost not only by biological processes but also by water erosion.Soil erosion together with continuous machinery traffic could explain the higher bulk density measured in the soil under yerba mate monoculture as compared to the forest soil.
Soil C and N contents increased under elephant grass -yerba mate association in the 0 -0.05 m soil layer after six years of grass implantation (P < 0.01) and in 0.05 -0.15 m soil after ten years of grass implantation (P < 0.05).However, this SOM increase did not improve soil porosity.The increase in soil C and N with grass is mainly due to a great residue returned to the soil, including the belowground ones (18.0 ± 2 and 5.3 ± 0.3 Mg ha -1 year -1 for aerial and root biomasses, respectively) and this has also been reported in other tropical soils (Scopel et al., 2004).Additionally, cover crop mulch associated with no tillage increases SOM in surface soil by reducing soil disturbance which favors the protection of soil aggregates from rapid oxidation and by modifying the local edaphic environment: bulk density, pore distribution, temperature, water and air regime that might also restrict SOM biodegradation (Puget & Lal, 2004).Working in similar soils of Misiones, Morrás & Piccolo (1998) found more aggregation and more SOM in the top 0.05 m of the soil and more aggregation degree but no differences in SOM content at 0.05 -0.15 m soil depth after six years of elephant grass -yerba mate association as compared to a long term conventionally tilled yerba mate plantation.In contrast to temperate soils, the better aggregation in tropical soils does not lead to an increase in C sequestration: aggregates formed by mineral-mineral bindings form faster but also protect less C upon conversion to no tillage than aggregates in temperate soils (Six et al., 2002).
After ten years of elephant grass -yerba mate association, soil C storage was 6.4 Mg ha -1 higher than under yerba mate in the 0 -0.15 m layer, i.e. an increase of 0.6 Mg C ha -1 year -1 .Our results are in accordance with other authors that found that more sus-tainable agricultural systems can lead to an annual C accumulation of 0.3 -0.6 Mg ha -1 under humid subtropics and tropics (Pretty & Ball, 2001).Soil N stock increased 0.4 Mg ha -1 after ten years of elephant grass yerba -mate association.Elephant grass is a luxury feeder of soil N due to its high protein content (1.67% N) and biomass production (23.3 Mg ha -1 year -1 ) (Goorahoo et al., 2005).Since we did not graze on the elephant grass and its leaves were cut and left on soil surface as mulch, it became a significant addition of N to the soil (389 kg ha -1 year -1 ).Therefore, elephant grass -yerba mate association should be considered a sustainable system because it increased the soil C and N stock in the upper 0.15 m on a short period of time due to the very important C and N annual contributions, and also to marked decrease on soil disturbance.
The δ 13 C values obtained for litter were similar for the forest and yerba mate plots because forest and yerba mate trees have the same C3 photosynthetic pathway while the δ 13 C value of elephant grass was in accordance with a C4 grass.δ 13 C values showed a consistent increase (7.0 ‰) with depth in the forest soil profile (Table 2).This has been noted previously (Vitorello et al., 1989;Martin et al., 1990;Desjardins et al., 1994) and may be related to the isotopic fractionation occurring during humification (Balesdent et al., 1987;Martin et al., 1990).
In the soil under continuous yerba mate cultivation, the 0 -0.05 m layer presented a high enrichment in δ 13 C (-30.05 ‰ in litter and -22.47 ‰ in the 0 -0.05 m soil depth).However, soil δ 13 C contents usually resemble closely those of the original plant tissues.A reasonable explanation for this result is the wash out of, at least, part of the original A1 horizon.
The proportion of C derived from elephant grass residues (α P ) increased with time in the 0 -0.05 and 0.05 -0.15 m depths (Table 2).After ten years, α P reached 39% and 11% of bulk C in the upper 0 -0.05 and 0.05 -0.15 m layers, respectively.Studies of pasture chronosequences in the Amazon Basin showed that large amounts of soil C of forest origin were maintained after several years of pasture: 65% of the SOM in the 0 -0.20 m layer was of forest origin in a ten-year-old elephant grass pasture (Desjardins et al., 1994) and about 80% of the total soil C in the upper 0.30 m was originated from forest material after eight years of pasture (Bonde et al., 1992).In our study, the contribution of elephant grass material to the total C stock was smaller (20% at 0 -0.15 m; Table 2), probably due to a lower annual C input and/or to a microbial population of the degraded soil (yerba mate) not adapted to efficiently process the SOM. Figure 2 shows the evolution of C DP and C DF calculated by 13 C data and predicted using Andriulo et al. (1999) model.The same exponential kinetics, which take into account the hypothesis of a stable fraction of SOM, have been used in other SOM turnover studies (Jenkinson & Rayner, 1977;Parton et al., 1987;Balesdent et al., 1988).The proposed model adjusted very well to the measured data (Q = 0.11).As bulk density soil values did not change during ten years under elephant grass -yerba mate association, comparison of SOC stocks at fixed depth was appropriate between cultivated treatments.We assumed that water erosion became negligible with the elephant grass implementation.The model indicates that the C content at the equilibrium point (equation 8) would increase markedly (C meq = 40 Mg ha -1 ) because of the high quantity of humified SOM (q = 2.5 Mg ha -1 year -1 ) (equation 7).In fact, the predicted increase in C DP was very important (from 0 to 9.0 Mg ha -1 ), while the predicted decrease of the C DF was smaller (from 34 to 31 Mg ha -1 ).Humified SOM represented 26% of the annual applied C (k 1 = 0.26).The k 1 values estimated in other studies range between 0.08 and 0.40 (Hénin & Dupuis, 1945;Janssen, 1984;Jenkinson, 1990;Bolinder et al., 1999).In general, k 1 values for aboveground residues average 0.12 (Bolinder et al., 1999) and roots present higher k 1 values because of the higher lignin content and the physical protection imparted by soil aggregates (Bayer et al., 2006).The isohumic coefficient obtained in this study would be mainly due to grass abundant system under no tillage.The active fraction mineralized with a turnover (TRM) of ca.four years.The coefficient of mineralization (k) of our model (0.28 year -1 ) may be compared with the annual rate of SOC loss (k 2 ) using equation 13 (Andriulo et al., 1999): The resulting k 2 is ≈ 0.06 year -1 , which is in agreement with the climatic characteristics and texture of these soils.Bayer et al. (2006) reported a k 2 of 0.10 year -1 in a tropical Ultisol of SE Brazil (São Paulo) where mean temperaturas are higher than in Misiones, and a k 2 of 0.01 year -1 in a clayey ultisol (> 600 g clay kg -1 ) in another region of Brazil (Rio Grande do Sul) with lower temperatures and rainfall.
The model estimated that the organic C stock stabilized at 40 Mg ha -1 .The active C pool represented 22% of the total C at the equilibrium level.Total C was 34 Mg ha -1 at t 0 and C s was 31 Mg ha -1 (Figure 2 and Table 1).The difference between these two pools represents C a0 .This low C a0 (3 Mg ha -1 ) could have origi-  nated either from yerba mate root biomass or from forest after fifty years of continuous yerba mate cultivation.
If the active fraction has a high turnover then, it would have disappeared unless the active soil C from forest soil was progressively exposed due to water erosion.The use of natural abundance 13 C has proved to be a reliable method to follow the young C (from elephant grass) incorporation into the soil and the old C (from forest) evolution.Its combination with a simple model to describe decrease C kinetic from forest and accumulation C kinetic from elephant grass is promising.Although the descriptive model we used is easier to parameterize than other more powerful models which describe SOM bio-transformations taking into account a great number of parameters and compartments (Smith et al., 1997;Nicolardot et al., 2001;Petersen et al., 2005), the validation of this one in other experimental sites of the Ultisols of Misiones province is still pending.Thus it should be used carefully in long-term simulations, due to the need of continuous experimental data, particularly to establish the stable SOM pool size.Furthermore, the probable negative impact of water erosion on soil C content should be taken into account.

Figure 1 -
Figure 1 -Diagram of the three-compartment model of soil C evolution.k 1 = crop residues humification coefficient.k = annual coefficient of mineralization of active SOM.

Figure 2 -
Figure 2 -Evolution of simulated and calculated old and young C fractions of SOM (0-0.15 m layer) under ten years of continuous elephant grass -yerba mate association and after fifty years of yerba mate.
due to the low dry matter input and continuous mechani-

Table 1 -
Studied properties of a Kandihumult soil under three different soil management treatments in Misiones, Argentina.

290-297, May/June 2008 cal
weed control that characterize most of the yerba mate plantations.

Table 2 -
Natural soil13C abundance and proportion of soil C derived from elephant grass under three different soil management treatments in Misiones.Argentina.: standard deviation; !: mean value from leaves and branches of yerba mate exported as yield; *means followed by the same letter within a line do not differ (Student test at P < 0.05). SD