An intensive field study on CO2, CH4, and N2O emissions from soils at four land-use types in Sumatra, Indonesia



[1] We measured gas fluxes of carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O) from the soil surface to the atmosphere under various land uses in Sumatra, Indonesia, from September 1997. Four land-use types, i.e., old-growth forest, logged-over forest, burned site after logging, and rubber plantation site, were selected. One logged-over forest was clear-cut and burned in the middle of the experiment. An incubation experiment was also performed to measure the potential of these three gases' emissions by using intact soil cores. The ranges of flux for 1 year for CO2, CH4, and N2O were 51.3–93.7 mg C m−2 h−1, −21.2–4.2 μg C m−2 h−1, and 0.74–26.34 μg N m−2 h−1, respectively. The N2O and CO2 fluxes were among the smallest values in all tropical regions. Clear-cutting and burning of residual trees after logging caused an increase in N2O emissions. N2O emissions correlated highly with the nitrification rate at 0–10 cm soil layer (R2 = 0.7834, p < 0.01). CH4 fluxes correlated with the clay content of 0–10 cm soil layer (R2 = 0.6071, p < 0.15). The results of flux measurements and core incubation strongly suggest that the regeneration of vegetation reduces the impact of land-use/cover changes on trace gas emissions.

1. Introduction

[2] Studies on the effect of land-use/cover change on greenhouse gas emissions in humid tropical ecosystems are limited. An Intergovernmental Panel on Climate Change (IPCC) report pointed out the possible impact of deforestation in tropical regions on global warming [IPCC, 1995]. Land-use/cover change has been rapidly occurring since the 1970s in tropical regions such as the Amazon, and large areas of tropical forest have been converted to agricultural fields and plantations [Myers, 1991]. Land-use/cover change affects these greenhouse gas emissions; for example, N2O emission is stimulated by deforestation and burning [Weitz et al., 1998; Keller et al., 1993; Luizão et al., 1989], and CH4 uptake to the soil decreases after deforestation [Weitz et al., 1998]. The effect of a land-use/cover change on CO2 emission is more obscure than that of the other gases: some reports showed an increase, while others showed a decrease or no change [Raich and Schlesinger, 1992].

[3] The field site locations of these reports are distributed mainly in the Amazon [Weitz et al., 1998; Verchot et al., 2000, 1999] and Costa Rica [Keller and Reiners, 1994]. Southeast Asia, one of the largest tropical regions in the world, is an important area where deforestation has continued during the last 3 decades. For example, Indonesia's original 1,220,000 km2 of forest shrank to 860,000 km2 by 1991 [Myers, 1991]. The rate of decrease was estimated at 12,000 km2 yr−1, which constitutes approximately 9% of the deforestation in the world, the second largest proportion following that of the Amazon [Myers, 1991]. It is important to start a field study in this area to improve the estimate of the greenhouse gas emissions from land-use/cover changes in humid tropical regions in Southeast Asia. In this paper, we will report on trace gas fluxes (CO2, CH4, and N2O) from soils of four land-use types within the sequence of deforestation commonly observed in Sumatra, Indonesia, and compare our results with the results obtained from other regions.

2. Material and Methods

2.1. Site Description

[4] The field site was set up in Jambi province, in Sumatra, Indonesia, a region of humid tropical forest. The intensive field and laboratory studies were performed at the Pasir Mayang Research Site for 1.5 years starting in January 1997. The detailed ecosystems and climate in this area are already described by Murdiyarso and Wasrin [1995]. The temporary field study was performed at Kuamang Kuning, 56 km southwest from the Pasir Mayang Research Site, in October 1998. The total amount of rainfall was 2060 mm yr−1 between September 1997 and August 1998. In 1998 the rainy season ended as usual in April and the dry season started in May (Figure 1). The average monthly rainfall of the dry season (125 mm) was 50% smaller than that of the rainy season (250 mm) during the experimental period.

Figure 1.

(a) Monthly rainfall at the experimental site from September 1997 to August 1998.(b) Water content fluctuation at P1, P2, and L2. (c) Water content fluctuation at L1, R, and Op.

2.2. Pasir Mayang Research Site

[5] The Pasir Mayang Research Site is a protected forest area, located near a settlement area of a government-sponsored transmigration program combined with forest plantation activities called HTI-Trans. Four land-use types (primary forest, logged-over forest, deforested area, and rubber plantation site) were selected for monitoring the emissions of three gases: CO2, CH4, and N2O. The soil types of the sampling sites belong to gibbsitic Ferralsols [Laumonier et al., 1986] (Ultisols, in USDA Soil Taxonomy) and are strongly acidic. The general soil properties are listed in Table 1. The primary forest (P1 and P2, 1°05.164′S, 102°05.702′E) is an old-growth lowland forest (200 ha), unchanged for more than 50 years by human activity such as logging. The aboveground biomass of this forest is 170 Mg ha−1 [Murdiyarso and Wasrin, 1995]. P1 is located on a slope of about 15° angle, and P2 is on a flat top of the hill. Two plots of logged-over forests were prepared. Because these plots had been logged only once for high-profit trees, there were trees as tall as those in primary forest. One (L1, 1°3.810′S, 102°9.754′E) was a large forest (approximately 6000 ha) in 1996. Logging commenced in September 1997. The remaining slashes, less than approximately 50 cm diameter, were dried on site and burned in early March 1998, followed by the planting of rubber trees (Hevea brasiliensis). The other logged-over forest site (L2, 1°5.235′S, 102°6.586′E), located near the primary forest, had been slightly disturbed by logging, but it was not disturbed by human activity during this experimental period. The deforested site (Op, 1°3.660′S, 102°9.681′E) was clear-cut and burned in August 1996, followed by the planting of a fast-growing tree species Gmelina arborea. The planted trees were about 4 m tall in October 1999. The rubber plantation (R, 1°5.648′S, 102°7.207′E), planted with H. brasiliensis, was managed by a small landholder, an arrangement common in Sumatra. These farmers neither fertilize nor apply herbicide to control the common weed Alang-alang (Imperata cylindrica).

Table 1. Soil Characteristics for Soils Under Each Type of Land-Use at Pasir Mayang and Kuamang Kuninga
 Depth, cmpH (H2O)Total C, mg/gTotal N, mg/gC/NBulk Density, mg/m3Available Phosphate, μgP2O5/gNH4, μg/gNO3, μg/gPhospho-Monoesterase, μmol/g/hMicrobial Biomass C, μgC/g
  • a

    Data are shown with the samples in January 1997 with P1, L1, and Op, in September 1997 with L2 and R, and in September 1998 with L3 and Y, respectively.

Pasir MayangP10–104.2301.916.
Kuamang KuningL30–104.0292.113.80.889.

2.2.1. Flux Measurement

[6] The gas fluxes were measured using a static chamber method. Triplicate chambers with 0.30-m-diameter PVC tubes 0.12 m in length were inserted in the soil at each site to a depth of at least 1 cm. The chambers were fixed during the whole experimental period, except at L1, where we removed the chambers before the burning of the slashes, and then reinserted them after burning. The CO2, CH4, and N2O fluxes were measured at monthly intervals throughout the year from September 1997 to August 1998, except January, April, and May 1998.

[7] The chambers were sealed with PVC tops, which had two ports for gas sampling, one with a sampling bag attached to equilibrate the chamber pressure with the atmospheric pressure. We took four gas samples from the chamber at a time of 0, 10, 20, and 40 min with a disposable syringe and pushed the collected sample into glass vials with butyl rubber stoppers, which had been evacuated beforehand in the laboratory. We brought the glass vials back to our laboratory and analyzed the gas concentrations with a gas chromatograph (Shimazu GC-9A) equipped with a thermal conductivity detector for CO2, a flame ionization detector for CH4, and a second gas chromatograph (Shimazu GC-14A equipped with a precut system) with an electron capture detector for N2O. Because almost the entire accumulation curves appeared linear, we calculated the flux by the slope of gas concentrations with the sampling times. The air temperatures in the chamber and in the ambient air (1 m height above the ground) were measured with a temperature sensor (Sato Co., SK-1250MC). Our analytical system was able to detect a 3 ppb difference in the N2O concentration between 0 and 40 min; consequently, the minimum detectable flux was 0.4 μg N m−2 h−1.

2.2.2. Gas Sampling in the Soils

[8] Soil gas samples were taken with a stainless steel tube (3 mm outer diameter and 1 mm inner diameter) at 5- or 10-cm-depth intervals in January 1997 (at P1, L1, and Op) and October 1998 (at L1, postburn). Because the length of the tube was 50 cm, we could collect the gas samples at a depth of 50 cm at almost all the sites, but at some sites the soil was too hard for insertion to a depth of 50 cm. Twenty mL of gas samples were collected, and the gas concentration was analyzed using the same method as that for the flux measurement.

2.2.3. Soil Sampling and Analysis

[9] Three soil samples were taken at each site at depths of 0–0.10, 0.10–0.20, and 0.20–0.30 m. All soils were sieved with a 2-mm mesh sieve and stored in a refrigerator at 4°C before analysis. The soil pH (H2O) was measured with a glass electrode (Horiba, F-22). The water content was measured by drying the soil at 105°C for 24 hours. Total carbon and nitrogen were measured with an NC analyzer (Sumitomo Chemical Co., NC-800). The organic and inorganic phosphate contents were measured by ignition method [Olsen and Sommers, 1982] with minor modification of ignition temperature (350°C) [Takami, 1986]. The available phosphate content was measured using Bray II method [Olsen and Sommers, 1982]. The inorganic NH4 and NO3 were measured by a flow-injection analyzer (Tecator, Aquatec 5400 analyzer) with the extractant of 100 mL of 2 M KCl solution to 10 g fresh soil. The phosphomonoesterase (Pase) was measured by a colorimetric method [Tabatabai, 1994] with minor modification [Ishizuka and Ishizuka, 1996]. The microbial biomass carbon (CBio) was determined by a chloroform fumigation extraction method [Vance et al., 1987] using a TOC analyzer (Shimazu TOC-5000). The nitrogen mineralization rates were measured using an aerobic incubation method [Hart et al., 1994] at 25°C for 8 weeks.

2.2.4. Intact Soil Core Analysis

[10] Intact soil cores each of volume 0.1 × 10−3 m3 (0.05 m diameter and 0.051 m height) were collected from depths of 0–0.05, 0.10–0.15, and 0.20–0.25 m at each plot in January 1997 (P1, L1, Op) and September 1997 (L2, R). Triplicate samples were collected from each depth. The gas phase proportion was measured with a pycnometer (Daiki, DIK-1121). Water-filled pore space (WFPS) was calculated by as

equation image

where Ws is the average weight of solid phase in a 0.1 × 10−3 m3 soil core, ω (kg kg−1) is the water content, and Vgas (m3) and Vliq (m3) are the volume of gas and liquid phase in 0.1 × 10−3 m3 soil core, respectively. The gas diffusion coefficients were measured with a gas diffusion analyzer (Fujiwara, KK-320), measuring the diffusion of atmospheric oxygen through the soil [Osozawa and Kubota, 1987]. We incubated soil cores to evaluate the potential of greenhouse gas emission from the soil [Ishizuka et al., 2000]. We set an intact soil core into a 0.5 × 10−3 m3 incubation jar with butyl rubber stopper and monitored the gas concentration in the headspace of the jar for up to 24 hours. The gas concentrations of CO2 and N2O increased linearly, and the emission rates were calculated with linear regression. The CH4 concentration decreased according to first-order kinetics, and the consumption rate was calculated as

equation image

where Ct (m3 m−3) is the CH4 concentration at a time t (hours), C0 (m3 m−3) was the CH4 concentration of the headspace at the beginning, and k (hours−1) is the reaction rate coefficient, respectively. We defined the uptake potential rate by kC0 (namely, C0 was approximately 1.8 μm3 m−3).

2.3. Kuamang Kuning Site

[11] Kuamang Kuning Site is located within an intensively used oil palm plantation. Two land use types were prepared in this area. One is a logged-over forest (L3), much more heavily disturbed than L1 and L2. The other is an oil palm plantation site (Y), managed by a private company. The total area of newly planted oil palm exceeds 10,000 ha.

[12] We set up triplicate chambers temporarily in October 1998 and measured gas fluxes as mentioned above. We took the first gas sample more than 10 min after installing the chambers to minimize the effect of disturbing the soil surfaces. In addition, we took soil gas samples, and soil samples, and analyzed them following the same procedure as in the Pasir Mayang Research Site.

3. Results

3.1. Pasir Mayang Research Site

3.1.1. General Soil Properties

[13] The soil pH was low (4.2–4.9 at P1, 4.3–4.8 at L1, 4.0–4.4 at L2, 4.0–4.3 at Op, and 4.5–4.7 at R, respectively) (Table 1). The soil water content was low, ranging from 0.10 to 0.30 kg kg−1 (Figure 1). The soil water content was higher in a rainy season (from December 1997 to April 1998) than in dry seasons (from September 1997 to early November 1997 and after May 1998). The rainfall between September and October 1997 was much smaller than usual because of a severe drought caused by the El Niño phenomenon, which caused forest fires to produce heavy haze [Yoneda et al., 2000].

[14] The bulk densities of logged-over forests at the surface layer (0–5 cm) (0.81 at L1 and 0.88 at L2, respectively) were smaller than those of the other plots (1.12 at P1, 1.20 at Op, and 0.98 at R, respectively). The total soil carbon and nitrogen contents were low (9–45 mg C g−1 and 0.8–6.5 mg N g−1, respectively). The range of CBio was 292–659 μg C g−1, a level comparable with that of a previous report in another tropical forest [Srivastava and Singh, 1988]; it was relatively lower than those of the temperate forest soils ranging from 61 to 1620 μg C g−1 [Sato and Seto, 1995; Ross and Tate, 1993; Srivastava and Singh, 1988]. “Pase,” an index of microbial activities, ranged from 18.2 to 157.2 μmol g−1 h−1 and was higher than that of the soils in temperate forests, ranging from 0.2 to 65.2 μmol g−1 h−1 [Ishizuka and Ishizuka, 1996; S. Ishizuka et al., unpublished data, 1999], suggesting that the activity of microorganisms per the amount of microorganisms was higher in these soils.

[15] At P1, L2, and R, N mineralization products were mostly ammonium-N, and only a part of the NH4 was nitrified into nitrate. In contrast, nitrification activity was observed at L1, Op, L3, and Y (Table 2). At L3, most of all products by N mineralization were detected as nitrate. In September 1998, nitrate accumulated in the soil at the depth of 10–20 cm at L1, but net decrease in the amount of nitrate was observed in the N mineralization analysis (Table 2), indicating that there were some unknown nitrate-consuming processes.

Table 2. Inorganic Nitrogen Content and Nitrogen Mineralization Ratea
Sampling DatePlotDepthNH4, μg/gNO3, μg/gAmmonification (a), μg/g/dNitrification (b), μg/g/dMineralization (a)+(b), μg/g/d
  • a

    N.D.: not determined.

January 1997 (wet season)P10–
September 1997 (dry season)P10–1017.94.90.770.030.80
September 1998 (dry season)L10–1041.510.90.630.831.46

[16] The deforestation caused an increase in organic phosphate, NH4, Pase, and CBio (Table 3). The inorganic NH4 and NO3 in the 0–0.25 m layer were 19 and 4.7 times larger than those prior to clear-cutting, respectively. The slash-and-burn at L1 changed the soil carbon content, nitrogen content, inorganic nitrogen content, and microbial populations (Table 3). The inorganic nitrogen content after slash-and-burn fell to the same level as that before burning (September 1997), but was still greater than that prior to deforestation (January 1997). The nitrate concentration at the 10–20 cm depth in September 1998 showed an increase after slash-and-burn (compared with that in September 1997), while the ammonium concentration at 0–10 cm depth decreased. The level of CBio decreased after burning (Table 3).

Table 3. Changes in Soil Properties Caused by Clear-Cutting and Burning at L1
 0–10 cm Depth10–20 cm Depth20–30 cm Depth
Jan. 1997aSept. 1997bSept. 1998cJan. 1997aSept. 1997bSept. 1998Jan. 1997aSept. 1997bSept. 1998c
  • a

    Before clearcutting.

  • b

    Almost clear-cut but not yet burned.

  • c

    After burning.

W. Cont., kg kg−10.300.
C, mg g−13521162591023289
N, mg g−
Phosphates available, μgP2O5 g−
Org-P, μgP2O5 g−1101.6160.779.377.756.249.067.329.339.6
Inorg-P, μgP2O5 g−122.322.315.818.38.56.617.44.66.5
NH4, μg g−15.5195.941.
NO3, μg g−19.428.810.92.025.6124.82.49.822.7
Phosphomonoesterase, μmol g−1 h−138.6157.238.720.181.420.316.933.013.0
Microbial biomass C, μgC g−1471167090316577144274198178

3.1.2. Gas Fluxes

[17] At all sites, the soil emitted N2O and CO2, but mostly absorbed atmospheric CH4, except in the case of P2, where CH4 was frequently emitted into the atmosphere (Figure 2). There was no apparent seasonal variation in the fluxes of CH4. The N2O flux was slightly greater in the wet season than in the dry season. The CO2 flux and N2O flux showed the same seasonal variation.

Figure 2.

Fluctuation of (left) CH4, (middle) CO2, and (right) N2O flux from September 1997 to August 1998 at (a) primary forest, (b) logged-over forest, and (c) clear-cut and rubber plantation. Data are the average of three chambers; error bars are standard deviations. Vertical dotted line shows the period of burning.

[18] Annual mean N2O fluxes (n = 9) are 1.47 ± 0.59 (P1), 4.43 ± 4.07 (P2), 16.09 ± 12.17 (L1, preburning 7.90 ± 5.72, n = 5; postburning 26.34 ± 9.95, n = 4), 6.37 ± 4.92 (L2), 2.78 ± 2.19 (Op), and 0.74 ± 0.41 (R) μg N m−2 h−1. The N2O flux at L1, however, drastically increased after slash-and-burn between February and March 1998, as shown in Figure 2.

[19] Annual mean CO2 fluxes (n = 9) are 63.3 ± 15.4 (P1), 93.7 ± 26.9 (P2), 65.3 ± 17.6 (L1, preburning 74.5 ± 15.2, n = 5; postburning 53.9 ± 14.3, n= 4), 81.1 ± 23.4 (L2), 51.3 ± 27.4 (Op), and 74.6 ± 26.5 (R) mg C m−2 h−1. Annual mean CH4 fluxes (n = 9) are −21.2 ± 8.7 (P1), 4.2 ± 24.9 (P2), −4.5 ± 5.3 (L1, preburning −5.1 ± 7.2, n = 5; postburning −3.7 ± 2.0, n = 4), −17.6 ± 9.7 (L2), −6.2 ± 2.0 (Op), and −12.8 ± 5.3 (R) μg C m−2 h−1.

[20] The relationship between gas fluxes and WFPS was not clear (Figure 3), although the WFPS correlated with the CH4 uptake rate at R (R = 0.5891, p < 0.20) and Op (R = 0.7255, p < 0.15), with the N2O emission rate at Op (R = 0.7413, p< 0.10), and with the CO2 emission rate at P1 (R = 0.6358, p < 0.10) and Op (R = 0.6610, p < 0.20).

Figure 3.

Relationship between WFPS and the flux of three trace gases.

3.1.3. Soil Gases

[21] At increasing depths in the soil, CH4 concentration decreased, while CO2 and N2O concentrations increased (Figure 4). The N2O at P1 and L1 in 1997 showed nearly the same concentration as that of the ambient air. Slash-and-burn affected soil gas concentration remarkably, a change which was indicated in the difference between L1 in 1997 and L1 in 1998; that is, CH4 concentration decreased while CO2 and N2O concentration increased.

Figure 4.

CH4, CO2, and N2O concentration profiles of soil gas in January 1997 (L1, Op, P1) and in October 1998 (L1, L3, Y). Profile at L1 in 1997 was before slash-and-burn, and profile at L1 in 1998 was 7 months after slash-and-burn.

3.1.4. Soil Core Incubation

[22] At all sites, the emission rate of CO2 at a depth of 0–5 cm was the largest rate of all layers, except at Op (Table 4). The emission of N2O from the soil was observed at L1, L2, and Op, especially at 0–5 cm depth, although it was much smaller than from the soil at P1 and R. At Op there was no CH4 oxidation by any layers (Table 4); instead, CH4 was emitted. At all other sites, CH4 was oxidized by the soil. The soil at a depth 0–5 cm showed less capacity to oxidize CH4 compared to deeper layers.

Table 4. Trace Gas Emission Rates Calculated from Core Incubationa
 PlotDepth, cmCO2, mg C m−2 h−1CH4, μg C m−2 h−1N2O, μg N m−2 h−1
  • a

    N.D.: not determined.

Pasir MayangP10–532.113.9−1.592.890.390.28
Kuamang KuningL30–599.429.91.090.18N.D.N.D.

3.2. Kuamang Kuning Site

[23] The bulk density of L3 was as low as the other two logged-over forest soils at the Pasir Mayang Research Site (Table 1). Other soil properties were similar to those of the soils at Pasir Mayang Research Site. The CO2 flux determined by the chamber method (Table 5) was 1.5 times larger than the average of the logged-over forest in Pasir Mayang, where the average values of L1 (preburn) and L2 were 77.8 mg C m−2 h−1, although CH4 and N2O fluxes (Table 5) were nearly equal to the average of the logged-over forest in Pasir Mayang, where the average value of L1 (preburn) and L2 were −11.4 μg C m−2 h−1 and 7.1 μg N m−2 h−1, respectively. Compared to the average values of R (Table 4), the CH4 uptake rate of Y was smaller and the N2O emission rate larger.

Table 5. Trace Gas Emission Fluxes at L3 and Ya
PlotCO2, mg C m−2 h−1CH4, μg C m−2 h−1N2O, μg N m−2 h−1
  • a

    Data are shown by the average and standard deviation of three chambers.


4. Discussion

4.1. Comparison With Other Studies on Flux Measurement and Soil Properties

4.1.1. N2O Flux

[24] The N2O flux of primary forest (0.13 kg N ha−1 yr−1 at P1 and 0.39 kg N ha−1 yr−1 at P2, respectively) was in a smaller group of the tropical rain forests (0.01–7.68 kg N ha−1 yr−1, summarized by Breuer et al. [2000]). The N2O fluxes measured at the sampling sites correlated highly with the nitrification rate of 0–10 cm soils collected during flux measurement (Figure 5), a trend suggesting that N2O emission is regulated by the nitrification process. This result is comparable to those of previous reports [Verchot et al., 1999, Keller and Reiners, 1994, Matson and Vitousek, 1987], and the relationship between nitrification rate and N2O emission rate had a strong correlation at different ecosystems throughout the world (Figure 6). The N2O emission rate in this study was on the same order as that of a previous report from Hawaii [Matson and Vitousek, 1987], one of the lowest rates in the world. These results suggest that the low N2O emission rates in this study result from the small rate of nitrate production, due to the infertile and acidic soil properties. Because the spatial and seasonal variation of nitrification rates in this province is not known, information about them will help us obtain a better understanding of N2O emission properties.

Figure 5.

Relationship between N2O fluxes and nitrification rates (soil at 0–10 cm depth); open circle: January 1997; solid circle: September 1997; open square: October 1998. N2O fluxes were measured in the sites at the same time in soil sampling for the aerobic incubation experiment. Nitrification rates mean the rates of nitrate formation calculated from the aerobic incubation experiment. R2 = 0.7834 (p < 0.01), except Op (January 1997) and L1 (October 1998, postburn), which were strongly affected by land-use/cover change.

Figure 6.

Comparison of the ratio of nitrification rate to N2O flux with other ecosystems; open circles, Costa Rica [Matson and Vitousek, 1987]; triangles, Amazon [Livingston et al., 1988; Keller et al., 1988]; open squares, Hawaii [Matson and Vitousek, 1987].

[25] Our study suggests that the previous N2O emission rate of humid tropical forest is probably an overestimate. The global annual N2O emission from natural soils is estimated at 6–7 Tg N2O-N yr−1 [Potter et al., 1996a; Bouwman et al., 1993], a figure which suggests that tropical forests contribute the main portion of N2O emissions. The estimate of annual N2O flux of tropical rain forests was 136.4 mg N m−2 yr−1 [Potter et al., 1996a], while our result showed the value ranging from 25.8 to 62.5 mg N m−2 yr−1 (the average of annual fluxes of P1 and P2, and of L1 (preburn) and L2, respectively). Another estimate of N2O emissions from tropical rain forest was examined from a result in Australia [Breuer et al., 2000]. They obtained 10 times larger N2O emissions (13.1–74.5 μg N m−2 h−1) in Australia than that of our study (1.5–7.9 μg N m−2 h−1). In their report, a possibility of extrapolating their result to the tropical ecosystems outside neotropics, for example, in Southeast Asia was addressed, but because the soil (Ferralsols or Ultisols) and vegetation type in this study is typical of the islands Sumatra and Kalimantan and the N2O fluxes in these area were estimated to be lower than the result in Australia, the possibility should be low. Extrapolating the N2O emission rate in this study to the whole Indonesian forest (estimated 530,000 km2 [Myers, 1991]), the N2O emission rate from humid tropical forest soil in the world decreases by 0.039–0.058 Tg N yr−1 from the estimation [Potter et al., 1996a], which is about 3–4% of the total global emissions from tropical rain forest. However, this is a rough estimate, and a larger data set is needed in order to improve it.

4.1.2. CH4 Uptake

[26] The CH4 uptake rates were lower than those of previous reports about tropical rain forests (32.8 ± 5.2 μg C m−2 h−1, n = 22) [Potter et al., 1996b]. The CH4 flux correlated with the percentage of clay content of surface soils (A. Iswandi et al., unpublished data, 2000) (Figure 7). Many researchers have suggested that the gas transportation mainly regulates methane oxidation in various soil types [Dobbie and Smith, 1996; Striegl, 1993; Dörr et al., 1993], and methane flux shows a good correlation with the gas phase ratio [Keller and Reiners, 1994]. Our study shows, however, that the correlation coefficient between CH4 flux and gas phase ratio of 0–10 cm depth is low (R2 = 0.1321). This low value suggests that the limitation of gas diffusion in the soil at a depth of 10–20 cm (which had the highest potential of CH4 oxidation (Table 3)) could not be the dominant factor affecting the CH4 uptake rate. The positive methane fluxes that we also observed at some sampling sites mean that the CH4 flux is formed by the balance of the consumption and production of methane. These results suggest that the soil texture characteristics are more important because they are the key factors of not only gas transportation but also the microenvironment formation to affect the microbial activity on CH4 production and oxidation. Some researchers have reported that CH4 is emitted from soils of tropical forests even in nonflooded conditions [Verchot et al., 2000]. In our study, 23 fluxes were positive out of the 162 total measurements (9 times × 3 chambers × 6 sites), in which 15 positive fluxes were observed in the rainy season. At P2, methane emissions were frequently observed from one chamber. We have no data to explain the reason why CH4 was emitted, but it is plausible that termites or methanogenesis affected the CH4 fluxes [Verchot et al., 2000].

Figure 7.

Relationship between CH4 fluxes and the percentage of clay content of soil at a depth of 0–10 cm (R2 = 0.6071, p < 0.15). Value of CH4 flux is the mean of nine measurements in 1997 and 1998.

4.1.3. CO2 Emission

[27] In the primary and logged-over forest soils, annual mean fluxes of CO2 (63.3–93.7 mg C m−2 h−1) were lower than the average values in tropical and subtropical lowland moist forests (143 ± 21 mg C m−2 h−1, n = 10) [Raich and Schlesinger, 1992] and were comparable with relatively low rates observed in a tropical mountain areas (74–102 mg C m−2 h−1 [Raich, 1998]). It is suggested that this relatively low CO2 flux depended on the relatively low vegetation biomass (170 Mg ha−1, compared to a reasonable average of 300 Mg ha−1 for tropical rain forests [Laurance et al., 1998]), which made the turnover rate of the carbon cycles of the ecosystem relatively small. As a result, the CO2 flux was smaller than those of other areas [Raich, 1998].

4.2. Impact of Land-Use/Cover Change on Greenhouse Gas Emissions

[28] The long-term trend of the greenhouse gas fluxes due to land-use changes can be estimated by using the mean fluxes at these sites, as shown in Figure 8. In Figure 8 the short-term impact of land-use change appears in the difference between L1 (preburn) and L1 (postburn).

Figure 8.

Average of three gas fluxes at six sampling sites arranged in time series of land-use/cover change. L1 (preburn) and L1 (postburn) are the average of five and four observations, respectively.

[29] The order of the average fluxes for all the sites (Figure 8) was N2O emission

equation image

CH4 uptake

equation image

CO2 emission

equation image

This order is nearly identical to the results of the core incubation potential, a similarity which suggests that the emission/uptake rate of these trace gases in the field reflects the potential greenhouse gas emissions of soils. This result indicated that the core incubation method is useful tool for estimating the annual emission rate of these trace gases in the field condition. Slash-and-burn strongly stimulated N2O emissions. The N2O emission rate, which was very low in logged-over forests, drastically increased after slash-and-burn and then decreased to the same level in the rubber plantation as that of the primary forest. The stimulation effect on N2O emission by slash-and-burn was supported by the results of the short-term flux changes before and after slash-and-burn at L1 (Figure 8).

[30] The mechanism possibly responsible for the enhancement of N2O emission after slash-and-burn at L1 depended on the enhancement of nitrification. Nitrification was one result of mineralization of organic nitrogen after burning. Is there any possibility that N2O emission could be enhanced by denitrification in addition to nitrification? The possibility of denitrification is unclear. The addition of organic matter to the soil after deforestation, like leaf litter and twigs, increased NH4 concentration through nitrogen mineralization, followed by nitrate formation (Table 3). The nitrate moved down to the subsoil from the topsoil (Table 3) with no vegetation uptake for several months after slash-and-burn. It is possible that this excess nitrate first stimulates the denitrifying activity in the subsoils and then stimulates the emission of N2O. In support of this hypothesis, denitrifying bacteria were detected at Op (Y. Nakajima et al., unpublished data, 2001). At L1 (postburn), however, only a small number of denitrifying bacteria were detected (Nakajima et al., unpublished data) even though the N2O emission rate (20 μg N m−2 h−1, September 1998) was higher than the estimated rate (8.1 μg N m−2 h−1) calculated, as follows, by regression analysis without the data at L1 (postburn) (Figure 6):

equation image

where F is the N2O flux (μg N m−2 h−1) and N is the nitrification rate (μg N g−1 d−1).

[31] The net nitrification rate at L1 in September 1998 was negative (Table 2), which indicates that there was an NO3 consuming process in the soils. After slash-and-burn, the gas concentration below the 10 cm depth level increased with the depth (Figure 4), suggesting that the deeper soil evolved N2O through denitrification. The mechanism is still obscure, but we can suggest that in some cases the excess nitrate after slash-and-burn causes denitrification, followed by the N2O emission.

[32] The impact of land-use/cover change on CH4 flux was obscure. At L1, the CH4 uptake rate after slash-and-burn was no different from that prior to slash-and-burn, while the CH4 uptake rates at L1 both before and after burning were smaller than those at L2. We propose two possible explanations: (1) Because the difference in CH4 uptake rates depends on the difference in clay content (Figure 7), the CH4 uptake rates at L1, where the clay content was higher than at L2, were lower than those at L2; that is, the impact of land-use/cover change affects the CH4 uptake rate little; or (2) because the CH4 uptake rates decreased as a result of logging, the CH4 uptake rates at L1 were lower than those at L2; that is, the impact of deforestation decreases the CH4 uptake rate. Although the clay content at Op and R were equivalent, the CH4 uptake rate at R was higher than that at Op, a difference meaning that the regeneration of vegetation increases the CH4 uptake rate. This result suggests that the latter explanation is convincing.

[33] The long-term trend of CO2 flux after deforestation is expected to be a decrease. The short-term decrease (the decrease before and after slash-and-burn in Figure 8) supports this trend. The value of CBio decreased remarkably after burning (Table 3), a change suggesting that the fire sterilized the surface soil and caused a decrease in the population of microorganisms. Slash-and-burn removes most of the substrate for microorganisms, such as litter and organic exudates from roots, in turn decreasing the turnover rate of the carbon cycle.

[34] The impact of slash-and-burn, however, decreased over several years according to Figure 8. If the difference between the fluxes at Op and R was attributed to that in the vegetation regeneration, the planting of rubber trees would have some effect on the emission of the greenhouse gases, i.e., increasing CO2 emissions, decreasing N2O emissions, and increasing CH4 uptake (Figure 8). These effects show that the recovery of greenhouse gas emissions reduces the negative impact of a land-use/cover change such as deforestation of the primary forest or logged-over forest. This indicates that the growth of rubber plantation can mitigate the impact on these greenhouse gas emissions. The long-term effects of planting are important when considering the effects of a long-term succession of land-use/cover changes, because the restoration of greenhouse gas emissions to the level of the original vegetation takes several decades, during which planting has a large potential to affect greenhouse gas emissions. In this research we were not able to obtain the precise data to estimate the long-term effects; more research is needed.

5. Conclusion

[35] Our study suggests that slash-and-burn causes an increase in N2O emissions and a decrease in CO2 emissions, and the regeneration of vegetation decreases the impact of these land-use/cover changes. The N2O emissions are controlled by the nitrification rate of surface soil, but the emission rate was smaller than those previously known. The CH4 flux is affected by the clay content of the topsoil (0–10 cm). The CO2 flux in this area was 50% smaller than the average of humid tropical areas. We need more spatial data to clarify greenhouse gas emission/absorption and the controlling factors before and after land-use/cover change.


[36] This study was managed under the project “The Impact of Land-Use/Cover Change on Greenhouse Gas Emissions in Asia-Pacific Region” in the Global Environmental Research Program, supported by a grant from the Ministry of the Environment, Japan. We would like to express our appreciation to the timber company (PT IFA Barito Pacific Group) for allowing us to conduct this research at their experimental site, especially to Harmen, the administrator of the research site. We would also like to thank the staff at ICRAF in Muarabungo and BIOTROP in Bogor.