Temporal variations of fluxes of NO, NO2, N2O, CO2, and CH4 in a tropical rain forest ecosystem

Authors

  • Klaus Butterbach-Bahl,

    1. Department of Biogeochemical Cycles and Global Change, Forschungszentrum Karlsruhe, Institute for Meteorology and Climate Research, Atmospheric Environmental Research (IMK-IFU), Garmisch-Partenkirchen, Germany
    Search for more papers by this author
  • Martin Kock,

    1. Department of Biogeochemical Cycles and Global Change, Forschungszentrum Karlsruhe, Institute for Meteorology and Climate Research, Atmospheric Environmental Research (IMK-IFU), Garmisch-Partenkirchen, Germany
    Search for more papers by this author
  • Georg Willibald,

    1. Department of Biogeochemical Cycles and Global Change, Forschungszentrum Karlsruhe, Institute for Meteorology and Climate Research, Atmospheric Environmental Research (IMK-IFU), Garmisch-Partenkirchen, Germany
    Search for more papers by this author
  • Bob Hewett,

    1. Commonwealth Scientific and Industrial Research Organisation, Tropical Forest Research Centre, Atherton, Queensland, Australia
    Search for more papers by this author
  • Spiro Buhagiar,

    1. Bellenden Ker CableCar Inc. (BTS), Bellenden Ker, Queensland, Australia
    Search for more papers by this author
  • Hans Papen,

    1. Department of Biogeochemical Cycles and Global Change, Forschungszentrum Karlsruhe, Institute for Meteorology and Climate Research, Atmospheric Environmental Research (IMK-IFU), Garmisch-Partenkirchen, Germany
    Search for more papers by this author
  • Ralf Kiese

    1. Department of Biogeochemical Cycles and Global Change, Forschungszentrum Karlsruhe, Institute for Meteorology and Climate Research, Atmospheric Environmental Research (IMK-IFU), Garmisch-Partenkirchen, Germany
    Search for more papers by this author

Abstract

[1] Fluxes of N2O, NO, NO2, CO2, and CH4 were measured with high temporal resolution for 3 months at a tropical rain forest site in Queensland, Australia, using automated measuring systems. During this period, representing the transition between dry and wet season, huge pulses of NO emissions from the soil exceeding 500 μg N m−2 h−1 were observed with the onset of the first rainfalls. The magnitude of fluxes was explained by intensive mineralization of accumulated litter from a previous long-lasting dry period. The mean NO emission rate was 207.1 μg N m−2 h−1 (range: 0.1–773.8 μg N m−2 h−1) and thus ∼8 times higher as compared to N2O emissions (25.6 μg N m−2 h−1, range: 0–101.3 μg N m−2 h−1). NO and N2O emissions showed pronounced temporal variations, which were almost exclusively triggered by changes in soil moisture. Total NO and N2O losses summed to ∼3.5 kg N ha−1. Though a significant amount of the NO emitted from the soil was redeposited to the soil surface after its oxidation to NO2 (0.7 kg N ha−1), the observed magnitude of net NOx release from the soil indicates that NOx emissions from tropical rain forest ecosystems are seriously underestimated. The mean CO2 emission was 159.0 mg C m−2 h−1 (range: 36.3–284.8 mg C m−2 h−1) and thus >2 magnitudes higher than NO emissions. Among the C- and N-trace gases measured, the temporal variability of CH4 uptake was the lowest. The mean uptake rate for CH4 was 23.8 μg C m−2 h−1 (range: −50.0–0 μg C m−2 h−1). The emission ratios of NO:N2O, CO2:N2O and CO2:NO varied substantially with time. During dry periods the NO-N:N2O-N ratio was as high as 60:1, whereas for wetter periods it decreased to <7:1. A comparable trend was also observed for the ratio between CO2-C:N2O-N emissions. The largest ratio of CO2-C:NO-N (>1500:1) was observed at intermediate soil moisture values.

1. Introduction

[2] Our knowledge about the magnitude of C- and N-trace gas exchange between tropical rain forest soils and the atmosphere is still scarce. However, results from recent experiments show that tropical rain forest soils can emit high amounts of N2O as well as NO and can function as a significant sink for atmospheric CH4 [e.g., Verchot et al., 1999; Breuer et al., 2000; Gut et al., 2002; Kiese et al., 2003]. Emissions of N2O from tropical rain forest soils are thought to contribute ∼20% or 2.2–3.7 Tg N2O-N yr−1 to the global atmospheric budget of this primary climate-relevant active trace gas [Intergovernmental Panel on Climate Change, 1997]. The few existing data on NO emissions from these soils suggest that the emission of this secondary climate relevant trace gas is in approximately the same magnitude as N2O emissions and may amount to 1.1 Tg NO-N yr−1 [Davidson and Kingerlee, 1997]. However, part of the NO emitted from the soils is oxidized to NO2 within the canopy by its reaction with O3, and the produced NO2 can be redeposited to plant and soil surfaces or even taken up and metabolized by plant leaves [e.g., Gessler et al., 2001; Rummel et al., 2002; Sparks et al., 2001]. The canopy reactions of NO, NO2, and O3 are assumed to significantly reduce the net NOx release (here NO plus NO2) into the atmosphere by up to 70–80% [Yienger and Levy, 1995]. Nevertheless, soils still remain a key source for atmospheric NOx in tropical regions [Gut et al., 2002]. The emission of both N-gases from tropical rain forest soils, i.e., NO and N2O, strongly depend on the activity of the microbiological N turnover processes of nitrification and denitrification during which NO and N2O are produced as facultative or obligate intermediates. Owing to its biogenic background, emissions of N-trace gases from soils can easily vary by a magnitude within a few days [e.g., Davidson et al., 1991; Gasche and Papen, 1999; Kiese et al., 2003]. These short-term changes, which are mainly driven by changes in environmental parameters such as temperature and soil moisture, are the major cause for the still existing huge uncertainty in the estimation of the source strength of tropical rain forest soils for these N-trace gases. Up to now, only a few measurements have been carried out in which these short-term changes have been studied in detail [e.g., Davidson et al., 1991; Kiese and Butterbach-Bahl, 2002; Gut et al., 2002; Kiese et al., 2003]. However, these studies in tropical rain forests did concentrate either on N2O or NO emissions; that is, simultaneous measurements on short-term changes of NO and N2O emissions are still lacking. Such measurements are urgently needed in order to improve our ability to estimate the source strength of tropical rain forest soils for N-trace gases and to improve our understanding of the underlying mechanisms involved in the emission of both gases, for example, with respect to the development and validation of process oriented models (R. Kiese et al., Regional application of PnET-N-DNDC for estimating the N2O source strength of tropical rainforests in the Wet Tropics of Australia, submitted to Global Change Biology, 2004).

[3] In this study we aimed to generate a detailed database with a high temporal resolution on the emissions of NO as well as N2O from a tropical rain forest soil. Specific objectives addressed in this study were (1) to determine short-term variations of NO and N2O emissions during the transition period from dry to wet season, (2) to evaluate variations of the NO:N2O ratio with time, (3) to compare variations in N-trace gas fluxes to variations in CH4 and CO2 fluxes, and (4) to relate temporal changes in C- and N-trace gas fluxes to changes in temperature and soil moisture.

2. Materials and Methods

2.1. Study Site

[4] The exchange of N2O, NO, CO2, and CH4 between a tropical rain forest soil and the atmosphere was investigated from the beginning of November 2002 to the end of January 2003 at a lowland rain forest site in the “wet tropics” of Australia, close to the village of Bellenden Ker, Queensland (145°54′E, 17°16′S, 80 m above sea level (asl)). The mean annual precipitation at the site is 4360 mm, with 70% of annual precipitation falling during the wet season, which normally lasts from November to April. Mean annual air temperature is 24.3°C, with average daily temperatures exceeding 26°C during the wet season. The soil type is an Ustochrept derived from granite. The soil texture is characterized by a high sand fraction (60%) and medium silt (20%) and clay (20%) contents. In the uppermost soil layers the slightly acidic soil has a soil pH of 4.1, an organic carbon content of 3.1% and a C/N ratio of 12:1 [Kiese and Butterbach-Bahl, 2002]. The vegetation at the site is classified as a complex mesophyll vine forest [Tracey, 1982] with an average canopy height of 20 m. Additional information about site properties is given by Kiese and Butterbach-Bahl [2002] and Kiese et al. [2003].

2.2. Measurements of N2O, CH4, and CO2 Fluxes

[5] N2O, CH4, and CO2 fluxes were measured with a fully automated measuring system operating on the basis of the closed chamber approach. The system consists of five chambers, which are automatically opened and closed in a 1-hour cycle, an automated gas sampling system, which takes air samples every 3 min from the different chambers and a gas chromatograph (Texas Instruments, SRI 8610C) equipped with a 63Ni electron capture detector (ECD, Vichy Valco, Switzerland) for N2O analysis and a flame ionization detector (FID) for CH4 analysis. CO2 concentrations were monitored in 1-min intervals using an infrared gas analyzer (BUSE Guardian II, Buse Anlagenbau, BRD) attached to the end of the gas sampling line. The entire system was calibrated routinely every 2 hours by use of standard gas (0.4 ppmv N2O, 1.5 ppmv CH4 and 500 ppmv CO2 in synthetic air, Messer Griesheim, Germany). Fluxes were calculated from the increase/decrease in the concentration of the respective gases with time. Soil CO2 fluxes were calculated exclusively from the changes of CO2 concentrations during the first 10 min after closure of the chambers. However, for calculation of N2O and CH4 fluxes, concentration changes occurring during the entire period in which chambers were closed (1 hour) were used. The detection limit for N2O, CH4, and CO2 fluxes was 1.5 μg N2O-N m−2 h−1, 2 μg CH4–C m−2 h−1, and 2 mg CO2-C m−2 h−1. Further details on the measuring system and on modes of calculation of fluxes are given by Breuer et al. [2000] and Kiese and Butterbach-Bahl [2002].

2.3. Measurements of NO and NO2 Fluxes

[6] NO and NO2 fluxes were determined using the dynamic chamber method. For this we developed a fully automated, transportable measuring system with five measuring chambers and one reference chamber, which was closed to the soil surface by Plexiglas (dimension of the chambers: 0.5 m × 0.5 m × 0.3 m). The chambers, which were sunk ∼1–2 cm into the soil, were made up of an aluminum frame in which Plexiglas panes were fitted. A scheme of this system is given Figure 1. The entire system consisted of a chemoluminescence detector for NO detection (CLD 770 AL ppt, Ecophysics AG, Dürnten, Switzerland), a photolysis converter for NO2 detection (PLC 760, Ecophysics AG, Dürnten, Switzerland), a gas phase titration unit for calibration of the chemoluminescence detector (Environics 100, Environics Inc., West-Wellington, New Zealand), an UV absorption analyzer for O3 detection (TE 49 C, Thermo-Environmental Instruments Inc., Franklin, Massachusetts) for correction of NO/NO2 concentrations due to fast reactions with O3 [Butterbach-Bahl et al., 1997], a laptop for data acquisition and control, automated chambers, flow controllers, and pumps for generating a continuous flow of ambient air through the chambers of ∼130 L min−1. Owing to sensitivity of the measuring instruments to temperature, all instruments were packed into one aluminum box, which was cooled down to ∼18°C by an air-conditioning system (Figure 1). Calibration of the NO/NO2-analyzer was performed at least every other day using 10 ppb NO in synthetic air produced by dilution of recalibrated standard gas (0.997 ppm NO in N2, Messer Griesheim, Olching, Germany, recalibrated according to TOR standard by the IFU calibration laboratory). Efficiency of photolytic cleavage of NO2 into NO was determined as described by Butterbach-Bahl et al. [1997]. Fluxes were calculated on the basis of concentration differences between measuring chambers and the reference chamber [Butterbach-Bahl et al., 1997, 2002]. The detection limit of NO, NO2, and O3 was 50 pptv, 100 pptv, and 100 pptv, respectively, resulting in a minimum detection limit for NO and NO2 flux rates of 1.0 μg NO-N m−2 h−1 and 1.5 μg NO2-N m−2 h−1. Further details of the measuring system and methods for calculation of fluxes are given by Butterbach-Bahl et al. [1997] and Gasche and Papen [1999].

Figure 1.

Scheme of the fully automatic, mobile measuring system for online determination of NO/NO2 fluxes. CLD: Chemoluminescence detector, PLC: photolysis converter; GPTU: gas phase titration unit (calibration); MC1-MC5: measuring chambers 1–5; RC: reference chamber.

2.4. Auxiliary Measurements

[7] Measurements of temperature were carried out 2 m above ground level (air) and at various soil depths (litter layer, mineral soil: 1 cm, 5 cm, 10 cm) using PT100 thermocouples (IMKO, Germany). Changes in soil moisture were monitored by three TDR probes (UMS, Munich, Germany) placed in the direct vicinity of the measuring chambers. Volumetric soil moisture values were transferred into water-filled pore space values (WFPS) as described by Breuer et al. [2000]. These data were stored at hourly intervals on a separate data logger (Keithley Instruments, Germany). Furthermore, an automatic climate station (Campbell 21x) was installed at the top of a 25-m-high tower ∼100 m apart from the measuring area. This meteorological station acquired hourly data on the following parameters: global radiation, air pressure, air temperature, relative air humidity, precipitation, wind direction, and speed. During the measuring period the amount of litter fall was determined three times (mid-November, mid-December, and mid-January). Details for these measurements are given by Kiese et al. [2003].

2.5. Statistical Analysis

[8] Linear and nonlinear correlation analyses were used to examine relationships between the fluxes of N2O, NO, NO2, CO2, and CH4 and environmental parameters such as temperature and soil moisture. All statistical analyses were performed with SPSS 8.0 (SPSS Inc., United States) and Microcal Origin 6.1. Tests of significant differences (p < 0.05) between individual chambers were performed using the paired nonparametric Wilcoxon-test.

3. Results

3.1. Environmental Conditions

[9] In 2002, total annual rainfall amounted to 2248 mm, which is only half of the sum normally observed at the Bellenden Ker site. From June to October, which is the dry period, total rainfall was 238 mm, ∼3 times lower than in normal years. Rainfall in October was less than 5 mm. As a consequence, values of WFPS of the light textured soil were as low as 15% at the beginning of the measurements. At the start of the measurements a significant, 3- to 5-cm-thick litter layer was present on top of the mineral soil, the mass of which amounted to 5300 ± 408 kg dry matter ha−1 or ∼2650 kg C ha−1 or 43 kg N ha−1, respectively (N content of leaf litter is 1.63% [Kiese et al., 2003]). The litter layer mass declined until the end of the measurements to 4400 ± 321 kg dry matter ha−1. If the litter fall during the measuring period (3700 ± 110 kg ha−1) is additionally taken into account, total mineralization in the period November 2002 to the end of January 2003 must have amounted to ∼4600 kg dry matter (equivalent to ∼2300 kg C ha−1 and 37.4 kg N ha−1; excluding root litter production). The strong mineralization of organic matter was initialized by increasing frequencies and intensities of rainfall, which increased values of WFPS to >25% from mid-December onward (Figure 2). With regard to soil temperature, there was no major change during the entire observation period (mean soil temperatures in the litter layer: 23.6 ± 1.4°C).

Figure 2.

Course of daily totals for precipitation and of hourly measured values of soil temperatures (litter, mineral soil) and of soil moisture at the rain forest site near Bellenden Ker, Australia (11 January 2002 to 22 January 2003).

3.2. Temporal and Spatial Variability of N2O, NO, and NO2 Fluxes

[10] The temporal variability of N-trace gas fluxes at the Bellenden Ker site was pronounced and, obviously, mainly driven by changes in soil moisture due to rainfall and subsequent drying out of the soil (Figure 3). The mean N2O emission during the observation period was 25.6 ± 0.7 μg N m−2 h−1 (range: 0–101.3 μg N m−2 h−1, Table 1). Lowest N2O emissions (<5 μg N m−2 h−1) were observed during the first few days of measurements, i.e., at the end of a dry period lasting more than 6 months. Following the first rainfall event on 6 November, N2O emissions increased up to 36 μg N m−2 h−1, but decreased quickly after a few hours to values <7 μg N m−2 h−1. Comparable patterns could also be observed during the next few weeks when singular rainfall events increased N2O emissions for a few hours or even a few days (Figure 3). However, following a period with no rainfall (25 November to 16 December), N2O emissions dropped down again to values <7 μg N m−2 h−1. After this period, an increased frequency of rainfall led to a marked increase in N2O emissions to values mostly well above 25 μg N m−2 h−1. Still, each rainfall event that resulted in an increase in soil moisture also led to a significant increase in N2O emissions (Figure 3). During this period, which can still be called a transition period from dry to wet season and which lasted until the end of the measuring period, maximum values of N2O emissions were observed (e.g., 101.3 μg N m−2 h−1, chamber 3, Table 1).

Figure 3.

Temporal variability of fluxes of N2O, NO, NO2, CH4, and CO2 and of soil moisture and soil temperature at the rain forest site near Bellenden Ker, Australia, during the period 1 November to 22 January. Given are hourly (NO, NO2) or 2-hourly (N2O, CO2, CH4) mean values (±SE) calculated from five measuring chambers.

Table 1. Means (±SE), Minimum and Maximum of Fluxes of N2O, NO, NO2, and CH4 for the Different Chambers at the Rain Forest Site Near Bellenden Ker, Australiaa
 Chamber 1Chamber 2Chamber 3Chamber 4Chamber 5Chambers 1–5
  • a

    Different lowercase letters in the Mean±SE entries indicate significant differences between individual chambers (P < 0.05); cV, coefficient of variation; N, number of valid N2O, NO, NO2, CO2, and CH4 fluxes.

N2O, μg N m2h1
Minimum0.00.08.54.91.10.0
Maximum82.242.3101.372.270.5101.3
Mean ± SE20.3 ± 1.1a18.0 ± 0.6a30.7 ± 1.0b28.8 ± 1.0c30.3 ± 1.0b25.6 ± 0.7
N2132122132122131063
cv, %76.452.047.246.749.5 
 
NO, μg N m2h1
Minimum7.729.60.10.148.20.1
Maximum736.7761.8620.6773.8703.2773.8
Mean ± SE202.1 ± 3.6a307.3 ± 4.2b128.5 ± 3.1c109.5 ± 3.1d248.4 ± 4.1e207.1 ± 3.1
N10691097894106810305258
cv, %58.545.472.693.152.7 
 
NO2, μg N m2h1
Minimum−382.4−308.0−372.9−270.8−305.4−382.4
Maximum−0.1−0.1−0.1−0.1−0.1−0.1
Mean ± SE−48.3 ± 1.6a−51.8 ± 1.7a,b−48.7 ± 1.7c−31.1 ± 1.1d−42.9 ± 1.5b,c−46.1 ± 1.0e
N9549647899219094537
cv, %94.999.997.8113.2105.9 
 
CO2, mg C m2h1
Minimum41.647.536.376.776.236.3
Maximum206.7239.3270.9284.8276.2284.8
Mean ± SE123.7 ± 0.1a152.5 ± 2.5b170.9 ± 3.4c185.9 ± 3.5d182.0 ± 3.8d159.0 ± 2.6
N148150150149152749
cv, %20.620.124.323.225.4 
 
CH4, μg C m2h1
Minimum−50.0−28.2−36.0−39.2−31.1−50.0
Maximum−21.80.0−18.50.0−9.60.0
Mean ± SE−33.6 ± 0.4−19.4 ± 0.3−25.4 ± 0.3−17.5 ± 0.3−22.4 ± 0.4−23.8 ± 0.3
N2132122122082081053
cv, %15.119.213.526.124.0 

[11] The temporal variability of NO emissions was comparable to those of the N2O emissions (Figure 3). However, NO emissions, with a mean value of 207.1 ± 3.1 μg N m−2 h−1 (range: 0.1–773.8 μg N m−2 h−1, Table 1), were ∼8 times higher as compared to N2O emissions. Moreover, during periods when values of WFPS exceeded 25%, additional rainfall events did not lead to further increases in NO emissions as was observed for N2O. This is evident in the period starting in December and lasting until the end of January, when the first rainfall events increased NO emissions to maximum values of up to 773.8 μg N m−2 h−1 (Table 1). However, all subsequent rainfall events did not further increase NO emissions. Rather, a steady decline of emission rates until the end of the measuring period was observed. Nevertheless, even then, NO emissions were still around 40–50 μg N m−2 h−1 (Figure 4).

Figure 4.

Temporal variability of N2O, NO, and NO2 fluxes during the period 1 November to 22 January. Given are the hourly (NO, NO2) or 2-hourly (N2O) fluxes for the individual measuring chambers.

[12] The observed pattern of NO2 deposition mirrored the pattern of NO emission. High values of NO2 deposition were observed during time periods when NO emissions were also high (Figure 3). The mean value for NO2 deposition for the entire measuring period was −46.1 ± 1.0 μg N m−2 h−1 (range: −382.4–−0.1 μg N m−2 h−1, Table 1), ∼20% of the magnitude of NO emissions.

[13] The spatial variability of N2O emissions, i.e., the differences in N2O emissions between single measuring chambers, was pronounced (Figure 4), but did not exceed 55% on average over the entire measuring period(span of mean values of N2O emissions for chambers 1–5: 18.0–30.7 μg N m−2 h−1, Table 1). Compared to N2O, the spatial variability of NO emissions was much higher and differences in mean emission rates between individual chambers were up to ∼300% (span of mean values of NO emissions for chambers 1–5: 109.5–307.3 μg N m−2 h−1) (Table 1). The spatial variability of NO2 deposition was within ∼40%, the lowest for all three N-trace gases investigated (span of mean values of NO2 deposition for chambers 1–5: 31.1–51.8 μg N m−2 h−1).

3.3. Temporal and Spatial Variability of CO2 and CH4 Fluxes

[14] The temporal variability of soil CO2 emissions was generally comparable to that observed for the NO and N2O emissions, though the relative increase in CO2 emissions due to the wetting of the soil was less pronounced (approximately a factor of 2) and delayed when compared to the changes observed for N-trace gas fluxes. For N-trace gas fluxes, the increases in emissions could be observed immediately after rainfall events, i.e., within 2–4 hours, where these rainfall-induced increases reached a factor of 6–10 (Figure 3). At the beginning of the measurements soil CO2 emissions were <50 mg C m−2 h−1, but increased with increasing soil moisture to values of up to 200 mg C m−2 h−1. Mean CO2 emission during the entire measuring period was 159.0 ± 2.6 mg C m−2 h−1 (range: 36.3–284.8 mg C m−2 h−1, Table 1) and thus was ∼3 magnitudes higher than NO emissions.

[15] Among all C- and N-trace gases studied, the temporal variability of CH4 uptake rates was the least pronounced. This is also indicated by low values for the coefficient of variation (cv) of individual chambers, which were all <26% (e.g., span of cv for N2O measurements: 46.7–76.4%, Table 1). Only during the period from 16 December to 4 January when values of WFPS were constantly above 25%, a decrease in CH4 uptake rates from values of ∼−27 μg C m−2 h−1 to values of ∼−20 μg C m−2 h−1 was observed (Figure 3). The mean value of CH4 uptake rates for the entire observation period was −23.8 ± 0.3 μg C m−2 h−1 (range: −50.0-0 μg C m−2 h−1, Table 1).

[16] Compared to the relatively low temporal variability of CH4 uptake rates, the spatial variability of CH4 uptake was pronounced and reached approximately a factor of 2 (see Figure 5 and Table 1). In this respect, the spatial variability of soil CO2 emissions was lowest for all investigated gases; that is, the mean value for individual chambers only varied in a relatively narrow range of 123.7–185.9 mg C m−2 h−1.

Figure 5.

Temporal variability of CO2 and CH4 fluxes during the period 1 November to 22 January. Given are the 2-hourly fluxes for the individual measuring chambers.

3.4. Correlation of C- and N-Trace Gas Fluxes With Changes in Temperature and Moisture

[17] The dominant environmental factor affecting the magnitude of C- and N-trace gas fluxes at the rain forest site at Bellenden Ker during the observation period was soil moisture. Depending on the trace gas analyzed, the relationship could be described best by either a simple linear function (N2O and CH4), a quadratic function (NO), or an exponential decay function (CO2) (Figure 6). With regard to N2O and CO2, an increase in WFPS led to an increase in emissions. For NO a quadratic function provided the best fit for the relationship between NO emissions and soil moisture, which points to the fact that NO emissions were in most cases highest when WFPS reached an optimum value of ∼30–35% (Figure 6).

Figure 6.

Dependency of N2O, NO, CH4, and CO2 fluxes to changes in soil moisture. For these analyses, mean hourly (NO) or 2-hourly values (N2O, CH4, CO2) of five measuring chambers for the entire observation periods were used.

[18] Also, the magnitude of CH4 uptake was significantly related to changes in WFPS; that is, at increased values of WFPS the uptake of atmospheric CH4 by the soil decreased (f(x) = −28.2 + 0.14x; r2 = 0.1624; note that uptake rates are defined here as negative values). The changes in the ratio between NO-N and N2O-N emissions in dependency from WFPS could be described best with an exponential decay function (Figure 7). Here increased values of WFPS led to a decrease of the ratio; that is, N2O emissions increased faster with increasing values of WFPS as compared to NO emissions. At WFPS values <5%, the NO-N:N2O-N ratio was as high as 60:1, whereas at values of WFPS >30% this ratio decreased to <7:1. A comparable trend was also observed for the ratio between CO2-C:N2O-N emissions. Here, at WFPS values <20%, the ratio was >20000:1; that is, the emission of 20000 CO2-C atoms was only accompanied by the emission of one atom of N2O-N. At WFPS values >30% this ratio decreased to <8000:1 (Figure 7). The ratio between CO2-C and CH4−C decreased linear with increasing WFPS. For the ratio between CO2-C and NO-N emissions and WFPS, no significant relationship was found (Figure 7). The ratios between the emissions of NO-N:N2O-N and CO2-C:N2O-N as well as CO2-C and NO-N showed a pronounced temporal variation during the observation period as a result of changes in WFPS (Figure 8). Dryer periods were always characterized by high NO-N:N2O-N ratios (>10) and CO2-C:N2O-N ratios (>10,000), whereas during wetter periods these ratios decreased to <10 and <10,000, respectively (Figure 8). The greatest ratio of CO2-C:NO-N (>1500:1) was observed at intermediate values of WFPS, whereas during dryer periods with WFPS values <18% the CO2-C to NO-N ratio decreased to values of ∼500:1.

Figure 7.

Dependency of the ratios of N2O-N:NO-N, CO2-C:N2O-N, CH4-C:CO2-C, and CO2-C:NO-N on changes in soil moisture. For these analyses, mean hourly (NO) or 2-hourly values (N2O, CH4, CO2) of five measuring chambers for the entire observation periods were used.

Figure 8.

Temporal variability of the ratios of emissions of NO-N:N2O-N, CO2-C:N2O-N, and CO2-C:NO-N.

[19] Except for NO/NO2 fluxes, no significant relationship between N- and C-trace gas fluxes and soil temperature could be demonstrated. However, the relationship between NO/NO2 fluxes and temperature could only be demonstrated to exist for a short time period between 29 November and 16 December, when the soil was drying out and the diurnal soil temperature changes were accompanied by a pronounced diurnal pattern of NO/NO2 fluxes (Figure 9). Therefore it may also be hypothesized that diurnal changes in NO/NO2 fluxes may more frequently be observed during the dry season. The correlation between temperature and NO/NO2 fluxes during the period between 29 November and 16 December was described best by a linear relationship to the temperature in 1 cm soil depth (NO) or to air temperature (NO2), respectively (NO emissions (μg N m−2 h−1) = −1743 + 79.9 × t_soil_1cm; r2 = 0.722; NO2 deposition (μg N m−2 h−1) = 62.1 - 3.3 × t_air; r2 = 0.756).

Figure 9.

Diurnal pattern of air and soil temperatures and fluxes of NO and NO2. Given are mean values from five individual chambers, which were aggregated for different times of the day during the period 29 November to 16 December.

4. Discussion

[20] The range, magnitude, and spatial variability of N2O and CH4 fluxes observed in this study for the transition period from dry to wet season 2002/2003 at the rain forest site at Bellenden Ker is in agreement with previous observations at this site, which covered the period November 2002 to October 2003 [Kiese et al., 2003]. However, during the observation period reported here, rates of N2O emissions are still lower by at least a factor of 2–10 as compared to N2O emissions at the Bellenden Ker site in February/March 2000 and December/January 2001, when frequent and high rainfall had increased WFPS to values >40% and thus increased N2O emissions over several weeks and months [Kiese and Butterbach-Bahl, 2002]. These findings further support the statement of Kiese et al. [2003] that the interannual variability of rainfall is a key driving factor for the pronounced interannual variability of N2O emissions from tropical ecosystems. Besides the temporal variability of N2O emissions, the spatial variability of N2O emissions also contributes significantly to the uncertainty when estimating the source strength of tropical rain forest ecosystems for the global atmospheric N2O budget. In this study, spatial variability of N2O emissions and especially of NO emissions was pronounced, resulting in differences between mean values for individuals chambers of approximately a factor of 2 for N2O and approximately a factor of 3 for NO. Even though reasons for microsite variability have not been studied in detail in this work, we do have to assume that this variability is due to small-scale differences in WFPS, mineralization activity, NO3 availability, C/N ratio, or pH. This has already been demonstrated in previous studies carried out, for example, in mountainous tropical rain forests in Australia [Breuer et al., 2000].

[21] With regard to soil CO2 emissions the observed emissions rates exceeded those observed previously during much wetter periods in 2000/2001 [Kiese and Butterbach-Bahl, 2002]. In this study we calculated a mean value of CO2 emissions of 159.0 ± 2.6 mg C m−2 h−1 (range: 36.3–284.8 mg C m−2 h−1), whereas in the previous study, significantly lower values were reported (February/March 2000: 102.2 ± 3.2 mg C m−2 h−1; December 2000/January 2001: 133.3 ± 1.8 mg C m−2 h−1). This indicates that during the observation period of this study the huge amount of litter that had accumulated during the long-lasting dry period of the year 2002 was intensively decomposed with the onset of rainfall. Total soil CO2 emissions summed up to 3243 kg C for the period 1 November 2002 to 24 January 2003. If this figure is compared to the estimated mineralization rate of 2300 kg C ha−1 for the observation period, which was derived from measurements of the mass of the litter layer and litter fall amount (see sections 2 and 3), it becomes evident that during the observation period, autotrophic respiration may have contributed one third to total soil CO2 emissions, at best.

[22] The most striking observation made during this observation period was the magnitude of NO emissions from the soil. The observed NO emission rates (range: 0.1–773.8 μg N m−2 h−1, mean: 207.1 ± 3.1 μg N m−2 h−1) are exceeding by far other data sets of observed NO emissions from tropical rain forest ecosystems, which are mostly well below 100 μg N m−2 h−1 [e.g., Verchot et al., 1999; Gut et al., 2002]. Furthermore, NO emissions also exceeded N2O emissions in average by ∼1 magnitude, which is in agreement with previous observations in rain forest ecosystems of Costa Rica [Keller and Reiners, 1994]. However, it must be pointed out that measurement results of NO emissions from tropical rain forest ecosystems are scarce and that except for the measurements of Verchot et al. [1999] and Gut et al. [2002], reported measurements of the sampling frequency and number were much lower than in this study [e.g., Bakwin et al., 1990; Serça et al., 1994; Keller and Reiners, 1994]. The question arises why NO emissions have been so high during the observation period and why N2O emissions were comparably low. If we sum up NO emissions, a figure of ∼3 kg N ha−1 for a period of roughly 3 months is obtained, whereas for N2O a sum of 0.35 kg N was calculated (Figure 10). This number is ∼2 times higher than the estimate of annual NO emissions from tropical evergreen forests in eastern Amazonia (∼1.5 kg N ha−1 yr−1 [Verchot et al., 1999]) and is also ∼3 times higher than the average value for tropical rain forest that has been used by Davidson and Kingerlee [1997] for the calculation of a global inventory of soil NO emissions. To explain the high rates of NO emissions, we do have to interpret the NO emissions in the context of the observed high mineralization rates during the observation period, which, if calculated for N, were equivalent to 37.3 kg N ha−1. This means that ∼10% of the mineralized N was released during litter mineralization as NO. It is a high figure, but nevertheless is reasonable considering that the first rainfall events mainly moistened the litter layer rather than the mineral soil. For this reason, N uptake by plant roots must have been constrained, so that microbes were supplied in excess with nitrogen, which in consequence led to high loss rates of N gases from the site. The dominance of NO emissions versus N2O emissions indicates that nitrification was the main pathway of N-trace gas production, since NO production in soils is mainly resulting from nitrification rather than from denitrification [Conrad, 1996, 2002; Skiba et al., 1997]. This statement is in agreement with the observed ratios of NO-N versus N2O-N emissions, which was >10 during dryer periods. Conversely, the lowering of the ratio of NO-N:N2O-N with increasing values of WFPS may be used as an indicator that at higher WFPS (1) denitrification or coupled nitrification-denitrification processes gain in importance as a source for N-trace gases (especially for N2O) as compared to nitrification, (2) the mineral soil get involved in N2O production, and (3) NO consumption processes are gaining importance. Furthermore, the relatively narrow ratio of CO2-C to NO-N during dryer periods (<700) indicates either that NO losses from nitrification are higher at dry soil conditions or that NO consumption processes, such as oxidative NO uptake via nitrification or reductive NO uptake via denitrification [e.g., Rudolph et al., 1996; Conrad, 2002], are more sensitive against dry stress than the NO production via nitrification. Since we do have to assume that nitrification is closely coupled to mineralization, and mineralization decreased with decreasing soil moisture, nitrification and thus NO production via nitrification should do so as well. Therefore it is most likely that the observed decrease in the CO2-C to NO-N ratio during low soil moisture conditions is mainly due to reduced NO consumption in the mineral soil and forest floor, which even overcompensate the reduced NO production by nitrification.

Figure 10.

Cumulative sum of N-trace gas fluxes for the observation period November 2002 to January 2003.

[23] The calculated loss rate of 3 kg N ha−1 during the observation period is only ∼0.3% of the observed gross nitrification rate at our site (>900 kg N ha−1 for a 3-month period in the transition period from dry to wet season [Kiese et al., 2002]), which compares quite well with previous estimates by Davidson et al. [1993], who observed that NO and N2O losses after remoistening of a seasonally dry tropical forests were ∼0.1%.

[24] It must be pointed out that the observed high rates of NO emissions and the quick temporal responses of NO emissions to the first rainfall events are in excellent agreement with observed pulses of NO emissions with reported values >500 μg N m−2 h−1 from seasonally dry tropical forests in Mexico occurring after a long dry period with the first rainfalls [Davidson et al., 1991, 1993]. Furthermore, the observation by Davidson et al. [1991] that wetting events during the wet season resulted only in small responses of NO emissions is confirmed by our measurements. As was already pointed out by Davidson et al. [1993], it is the duration and severity of antecedent dry periods that finally determines the magnitude of response of NO emissions to wetting events. The results obtained in this study suggest that the emission of NO from tropical rain forest soils may still be underestimated by at least a factor of 3, i.e., that the figure is >3 TG NO-N yr−1 rather than 1.1 Tg NO-N yr−1 as suggested by Davidson and Kingerlee [1997]. However, to answer this environmentally important question, more long-term measurements at different tropical forest sites are urgently needed.

[25] Even though it can be assumed that due to intensive biological N-fixation [e.g., Cleveland et al., 1999], the growth of tropical rain forests is not primarily constrained by nitrogen limitation, the enormous net NOx loss rates from the soil surface during the transition period from dry to wet season brought us to the idea that NOx emissions may be a significant and important mechanism of N redistribution within tropical forest ecosystems. As has been demonstrated by others, up to 90% of the NO emitted from the soil during daytime is converted by the reaction with O3 to NO2 within the canopy [Rummel et al., 2002], and part of this NO2 is subject to redeposition and plant uptake [e.g., Sparks et al., 2001; Thoene et al., 1991]. Here we hypothesize that assimilatory NO2 uptake by plants via the leaves may be a significant source of N for epiphytes but also for other plants, which may have only limited access to soil nitrogen, for example, due to interspecies and intraspecies competition for nitrogen. However, to better understand the ecological importance of soil NOx emissions and to further validate the above hypothesis, additional field and laboratory measurements are required. One major conclusion from the results presented here is that we do urgently need more long-term studies of N2O and NO emissions, and ideally also of N2 emissions, from soils, covering entire years, to better understand N cycling in tropical rain forest ecosystems.

Acknowledgments

[26] This research was supported by the Deutsche Forschungsgemeinschaft (DFG) under contracts BU 1173-3 and BU 1173-4.

Ancillary

Advertisement