Effects of transient anaerobic conditions in the

0 downloads 0 Views 265KB Size Report
whether N2O came from nitrification or denitrification. ... variations in respiration, N2O production and N2O reduction ... [email protected] (P. Renault). ... by the same volume of Kr, in order to check for gas leakage. ..... of experiment 2; actual rate is calculated using equation 8 for three values of k3: k3Z2, 3 and 4%,.
Soil Biology & Biochemistry 37 (2005) 1333–1342 www.elsevier.com/locate/soilbio

Effects of transient anaerobic conditions in the presence of acetylene on subsequent aerobic respiration and N2O emission by soil aggregates K. Khalila,b,1, P. Renaulta,*, B. Maryb a

INRA, Unite´ Climat, Sol et Environnement, Site Agroparc, 84914 Avignon Cedex 9, France b INRA, Unite´ d’Agronomie, rue Fernand Christ, 02007 Laon Cedex, France

Received 14 January 2003; received in revised form 28 July 2004; accepted 23 November 2004

Abstract Our objective was to assess the effect of anaerobic conditioning in the presence of acetylene on subsequent aerobic respiration and N2O emission at the scale of soil aggregates. Nitrous oxide production was measured in intact soil aggregates D (compacted aggregates without visible porosity) and G (aggregates with visible porosity) incubated under oxic conditions, with or without anaerobic conditioning for 6 d. N2O emissions were much higher in aggregates that had been submitted to anaerobic conditioning than in aggregates that did not experience 15 this conditioning, although very little NOK N isotope tracing technique was used to check 3 remained in soil after the anaerobic period. whether N2O came from nitrification or denitrification. The results showed that denitrification was the major process responsible for N2O emissions. The aerobic CO2 production rate was also measured in intact soil aggregates. It was greater in aggregates submitted to anaerobic conditioning than in those that were not, suggesting that the anaerobic conditioning lead to an accumulation of small compounds including fatty acids that are readily available for microbial decomposition in aerobic conditions. This process increases the aerobic CO2 production and favours the N2O emissions through denitrification. q 2005 Elsevier Ltd. All rights reserved. Keywords: Denitrification; Anaerobiosis; Nitrous oxide; Aerobic respiration; Soil aggregates

1. Introduction Nitrous oxide (N2O) is a trace gas involved in atmospheric pollution; it contributes to the greenhouse effect (Smith, 1990; IPCC, 1996), and affects the chemistry of O3 in the upper troposphere and lower stratosphere (Graedel and Crutzen, 1992). N2O is mainly produced in soils during biological denitrification and nitrification (Tortoso and Hutchinson, 1990; Groffman, 1991; Conrad, 1996). Various models, more or less complex, have been proposed to estimate N2O emissions through nitrification * Corresponding author. Address: INRA, Unite´ Climat, Sol et Environnement, Site Agroparc, 84914 Avignon Cedex 9, France. Tel.: C33 43272 2223; fax: C33 43272 2212. E-mail address: [email protected] (P. Renault). 1 Present address: Laboratoire des Sciences du Climat et de l’Environnement, Unite´ mixte de Recherche CNRS-CEA, Domaine du CNRS—Baˆt. 12—Avenue de la Terrasse, 91198 Gif-sur-Yvette, France. 0038-0717/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.soilbio.2004.11.029

and denitrification. Most of them account for the variations with time in environmental variables such as soil water, NOK 3 content and temperature. Simplified models (Parton et al., 1988; He´nault and Germon, 1995; Parton et al., 1996) do not account for the microbial dynamics, while more complex ones (Grant, 1995) explicitly consider these dynamics. However, the latter models do not consider variations in potential microbial activities, particularly the variations in respiration, N2O production and N2O reduction activities, whereas they are likely to vary with time in arable soils. Potential denitrification has been shown to be correlated with soluble organic matter and easily mineralisable C (Burford and Bremner, 1975). Anaerobic conditions may lead to the accumulation of small organic compounds, including acetate (Tsusuki and Ponnamperuma, 1987; Dassonville et al., 2004) that may be consumed later in aerobic conditions and decrease temporarily the pH of the soil solution (Dassonville et al., 2004). Such changes in

1334

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

easily mineralisable C compounds and pH might affect potential respiration and denitrification activities. Our aim was to assess the consequence of a prolonged anaerobic period (6 d) on subsequent aerobic respiration and net N2O emission through denitrification. Experiments were performed at the soil aggregate level. The micro-scale approach to study denitrification was motivated by the fact that, in many cases, the conditions experienced by soil organisms at the microscale are not reflected by measurements on bulk soil samples (Parry et al., 2000). For example, O2 concentrations may decrease from values nearly equal to the atmospheric concentration to zero within a few millimetres in soil aggregates (Greenwood, 1961; Greenwood and Berry, 1962; Sextone et al., 1985; Sierra et al., 1995).

2. Materials and methods 2.1. Soil aggregate sampling and conservation Experiments were performed on an Orthic Luvisol (FAO classification) sampled at Mons-en-Chausse´e in Northern France (49880’ N, 3860’ E). The soil was cropped with maize in 2000. The properties of the soil were as follows: clay, 194 g kgK1; silt, 706 g kgK1; sand, 68 g kgK1; total CaCO3, 32 g kgK1; pH (water), 8.2; organic C, 8.52 g kgK1; total N, 1.00 g kgK1. At sampling time, the soil contained K1 4.7 mg NOK . Large aggregates were sampled in the 3 -N kg ploughed layer (10–30 cm depth) on 12 September 2000. Two sets of aggregates were distinguished according to Richard et al. (1999): aggregates D, with a massive structure and no visible porosity resulting from compaction due to traffic (Fig. 1a), and aggregates G, with a fragmentary structure and visible porosity (Fig. 1b). The larger aggregates were gently broken down immediately after sampling and then calibrated: we kept aggregates between 25 and 30 mm diameter. In order to reduce microbial activity during storage, the aggregates were air-dried over 3 d to obtain a residual moisture close to 0.11 g gK1 soil, and then stored at 2 8C until the beginning of the experiments, i.e. until November 2000, January 2001 and May 2001, for experiments 2, 3 and 1, respectively. Because of water evaporation during storage, the soil moisture at the beginning of experiments was 0.11, 0.07 and 0.06 g gK1, for experiments 2, 3 and 1, respectively. 2.2. Batch incubations and measurements Three experiments were carried out. Experiment 1 was performed to check the effect of anaerobic conditioning on the subsequent aerobic respiration, by measuring CO2 production. Aggregates were rewetted with water; half of them were then submitted to 6.6 d of anaerobic conditioning. Experiment 2 was performed to check the effect of anaerobic conditioning on the subsequent N2O emissions.

Fig. 1. Photographs of thin sections of D (a) and G (b) aggregates, The thin sections were obtained after the inclusion of dry clods in resin. Care was then taken to center the section on the center of the clods.

Net N2O emissions were measured in air for all aggregates, and N2O gross emissions for aggregates that did not experience an anaerobic conditioning. The objective of Experiment 3 was to verify that most N2O emissions came from denitrification, by using a 15N isotope tracing technique. In all experiments, D and G aggregates were first rewetted with either water or KNO3 solution (4 g LK1) at 20 8C on tension tables successively at K10 kPa suction for 1 d, K5 kPa for 1 d, K1 kPa for 1 d, and K0.5 kPa for 4 d. This procedure ensured a slow rewetting process, which prevented crack formation. The soil moistures obtained after 7 d was 0.21G0.01 and 0.24G0.01 g gK1 for D and G aggregates, respectively. Half of the aggregates rewetted with water then experienced a 6.6 d period of anaerobic conditioning: they were incubated in anaerobic conditions in flasks that received 3 successive cycles of 3 min vacuum and 3 min of pure N2 gas addition. Approximately 5% of N2 (2% in experiment 1) was removed and replaced by the same volume of C2H2 in order to record NOK 3 consumption through denitrification in experiment 2 and create the same conditions in experiments 1 and 3. In experiment 2, approximately 1% additional N2 was removed and replaced by the same volume of Kr, in order to check for gas leakage. During anaerobic conditioning in experiment 2, gas samples

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

were withdrawn at d 1, 2, 3, 4, 5 and 6 and analysed for N2O, N2, CO2, O2, Kr and C2H2 concentrations. All the flasks were incubated at 20 8C in the dark for 6.6 d. 2.2.1. Experiment 1 The anaerobic conditioning was conducted on 32 aggregates of each type (D and G), in 500 ml flasks each containing four aggregates. After this period, the flasks were flushed with air; the aggregates were transferred 3 h later in a 1 l airtight jars with a beaker containing 10 ml NaOH 0.1 N in order to trap CO2. Eight replicate jars (each containing four aggregates that had followed the anaerobic conditioning) were thus incubated at 20 8C in aerobic condition over 4 d. The same procedure was applied to the same number of aggregates that did not experience the anaerobic conditioning, i.e. that were transferred directly from the tension table to the 1 l jars. The jars were opened every 24 h during 4 d, aerated for 5 min and the NaOH beaker was replaced. Trapped CO2 was precipitated as barium carbonate by adding excess of BaCl2 solution. The remaining NaOH was then titrated with HCl 0.1 N at pH 8.62. Each soil aggregate was cut into three at the end of the aerobic incubation period. Soluble organic C was extracted on one third of them. It was also measured at the end of the rewetting phase and at the end of the anaerobic conditioning in additional aggregates, always with eight replicates. Organic C was extracted with a 30 mm K2SO4 solution and measured with an organic carbon analyser (O.I. Analytical, College Station, TX, USA) using the persulfate oxidation method at 100 8C (Barcelona, 1984). The second third of the aggregates was air-dried and finely ground, and total C and N contents were measured with an automatic CN analyser (Carlo Erba, NA1500, Milan, Italy). The last third of the aggregates was used to measure the pH of soil solution. pH was also measured after the rewetting phase for aggregates that did not experience anaerobic conditioning, and after the anaerobic conditioning for aggregates that experienced anaerobic conditioning. We added a mass of ultra pure water equal to twice that of the soil mass. The flask was closed and shaken for 10 min, then transferred to a beaker and left for 2 min. The electrode (Calomel electrode K401, Glass electrode G202B, Copenhagen, Denmark) was used to read pH every minute for 5 min. 2.2.2. Experiment 2 16 aggregates of each type (D and G) rewetted with water experienced anaerobic conditioning in 150 ml flasks (1 aggregate per flask), and 10 aggregates rewetted with KNO3 were not conditioned K K The mineral N (NHC 4 , NO2 and NO3 ) content of the aggregates that experienced anaerobic conditioning was measured at d 0 and 6, with six replicates. At d 6.6, the atmosphere of the flasks of the 10 remaining aggregates of each type was replaced with air as indicated before;

1335

the procedure was repeated 1 h later in order to ensure that C2H2 had been totally removed. The aggregates were then incubated aerobically for 7 h in air at 20 8C, 1 ml of air being replaced with Kr at the beginning of the incubation. Gas samples were taken with a syringe either 1, 3, 5 and 7 h after the addition of Kr for aggregates rewetted with water, or 14, 16, 18 and 20 h after the addition of Kr for aggregates rewetted with KNO3. The samples were analysed for N2O, N2, CO2, O2 and Kr concentrations. The final mineral N K K (NHC 4 , NO2 and NO3 ) content of the soil was determined on each of the 10 replicates. The same procedure was applied to other 30 aggregates of each type (D and G) rewetted with KNO3 solution that did not experience the anaerobic conditioning, i.e. that were transferred directly from the tension table to the 150 ml flask jars; gas samples were taken with a syringe 14, 16, 18 and 20 h after the addition of Kr. The flasks were then opened for a few minutes in order to release the trapped gases. After closing the flasks, 7 and 1 ml of gas were replaced by C2H2 and Kr, respectively. The 30 D and 30 G replicates were incubated at 20 8C once more. Gas samples were withdrawn with a syringe 14, 16, 18 and 20 h after the addition of C2H2 and analysed for N2O, N2, CO2, O2, Kr and C2H2 concentrations. The final mineral contents were measured on all the D and G replicates. N2O concentration was determined using a gas chromatograph equipped with an electron capture detector (HP 5890 Series II, USA) fitted with a Porapak Q column (80–100 mesh, 2 m). The carrier gas was Ar–CH4 (95/5); the oven and detector temperatures were set at 50 and 300 8C, respectively. N2, CO2, O2, Kr and C2H2 concentrations were measured on a TCD gas chromatograph (HP 5890 Series II, USA) fitted with Porapak Q ˚ , 1.8 m) (80–100mesh, 1.8 m) and molecular sieve (1–5 A columns. The carrier gas was He; the oven and detector temperatures were fixed at 50 and 120 8C, respectively. The relative precision of each chromatograph was 0.5–1%. The mineral N content of aggregates was extracted with a 1 M KCl solution (soil:solution ratioZ1:5). Measurements were performed with a continuous flow colorimeter (Skalar Analytical, Breda, The Netherlands) using the method proposed by Kamphake et al. (1967); Krom (1980). 2.2.3. Experiment 3 15 N isotope was applied either as KNO3 or urea, both having a 50% atom enrichment. Urea was chosen instead of NHC 4 because of its higher diffusion rate in soil and its rapid hydrolysis into NHC 4 already observed in this soil (Recous et al., 1988). A small amount of tracer was applied in order to minimise changes in mineral N in soil, i. e. 0.5 mg N kgK1 soil as urea-N and 2.0 mg N kgK1 soil as NOK 3 -N. We added the labelled urea or KNO3 solutions (0.3 ml per aggregate) to D aggregates first rewetted with water that had experienced anaerobic conditioning for 6 d. At the end of this conditioning period, three aggregates were

1336

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

K extracted for their initial concentrations of NOK 3 , NO2 , 15 C C K K NH4 , and the N atom% excess of NH4 and (NO2 CNO3 ). The N2 atmosphere of the flasks containing the three other aggregates was replaced with air after 14 additional h by alternating three successive cycles of 3 min vacuum and 3 min air replacement. This procedure was repeated 1 h later; the aggregates were then incubated in aerobic conditions at 20 8C. The atmosphere in the flasks was sampled after 8 h with 10 ml Venojectw vacuum tubes and dual-ended sampling flasks of 250 ml. An aliquot of 0.2 ml was taken from the Venojectw tubes and analysed for N(N2CN2O) measurement and 15N composition. The 250 ml sampling was used for N2O and 15N2O measurement. Three aggregates were used to extract and measure the final soil mineral N content and 15N enrichment. The 15N composition of the mixture (N2CN2O) was determined by automatic CN analyser coupled to a mass spectrometer (Fisons, Isochrom, Manchester, England). N2O and 15N2O were analysed on the mass spectrometer after cryogenic concentration with a trace gas system (Micromass, Manchester, England). Mineral N in the aggregates was extracted with a 1 M KCl solution K (soil:solution ratioZ1:5). Measurements of NHC 4 , NO2 K and NO3 concentrations were performed with a TRAACS 2000 analyser (Bran & Luebbe, Norderstedt, Germany) using the methods proposed by Kamphake et al. (1967) for K C NOK 3 and NO2 analysis and Krom (1980) for NH4 analysis. 15 C K K The N analysis of NH4 and (NO2 CNO3 ) were carried out after separation by micro-diffusion and evaporation (Brooks et al., 1989).

2.3. Data analysis 2.3.1. Model of aerobic CO2 production with time We assume that C decomposition rate by microbes during the aerobic condition was proportional to the amount of substrates, i.e. decomposable organic pools. For aggregates without anaerobic conditioning, only one substrate pool S1 (mol C kgK1 soil) was taken into account, which disappeared at the following rate: dS1 Z Kk1 S1 dt

(1)

where k1 is a rate constant (hK1), and t the time (h). Variation of this pool with time is then: S1 Z S10 expðKk1 tÞ

(2)

The rate of CO2 production can be written: v1 Z Kð1 K Y1 Þ

dS1 Z k1 ð1 K Y1 ÞS10 expðKk1 tÞ dt

(3)

where Y1 is the assimilation yield of S1 by microbes. In the case of aggregates previously submitted to anaerobic conditions, our results suggest that this conditioning led to the creation of an additional decomposable

pool S2 (mol C kgK1 soil) decomposing with a rate constant k2 (hK1): the CO2 production rate in this treatment is then: v2 Z Kð1 K Y1 Þ

dS1 dS K ð1 K Y2 Þ 2 dt dt

Z k1 ð1 K Y1 Þ expðKk1 tÞ C k2 ð1 K Y2 Þ expðKk2 tÞ

(4)

where Y2 is the assimilation yield of S2 by microbes. We first estimated the parameters k1 and the product (1KY1) S10, using the Marquardt–Levenberg algorithm (Marquardt, 1963), by fitting simulated CO2 production rates (Eq. (3)) to values measured in D and G aggregates, which had not been submitted to anaerobic conditioning. The optimisation was run simultaneously in the two types of aggregates, assuming that the decomposition rate k1 was the same in both soils. The procedure was applied to the mean of the eight replicates. Using these values, we then estimated the other two parameters (k2 and (1KY2) S20) by fitting simulated CO2 production rates (Eq. (4)) to values measured in D and G aggregates, which had been submitted to anaerobic conditioning. Again we assumed that the decomposition rate k2 was the same in both types of aggregates. 2.3.2. Relationship between aerobic CO2 production and soluble organic C We have tested whether there was a relationship between the aerobic CO2 production and the soluble organic C. Soluble organic C was measured in aggregates submitted to anaerobic conditioning (i) after rewetting on the tension table, (ii) after anaerobic pre-incubation, and (iii) at the end of the measurement periods. Soluble organic C was measured only at the last date in aggregates, which had not been submitted to anaerobiosis.

3. Results 3.1. Experiment 1: aerobic CO2 production with or without anaerobic conditioning The CO2 production rate during the aerobic incubation decreased slowly with time in aggregates that had not been submitted to anaerobic conditioning: from 7.0 to 5.2 nmol kgK1 sK1 in D aggregates, and from 8.2 to 6.1 nmol kgK1 sK1 in G aggregates, between 12 and 84 h, respectively (Fig. 2). The rate decreased much faster in aggregates that had been submitted to anaerobic preincubation. It varied from 14.5 to 4.9 nmol kgK1 sK1 in D aggregates, and from 17.9 to 6.0 nmol kgK1 sK1 in G aggregates, between 12 and 84 h, respectively. The model previously described (Eq. (1–4)) could be satisfactorily fitted to the observed data, even with the assumption that the constant rates k1 and k2 did not differ between the two treatments (Fig. 2). The estimated parameters (Table 1) show that the anaerobic conditioning has resulted in

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

the formation of a small additional decomposable pool (S2), which decomposed more rapidly than the decomposable C present in the control soil (S1). Its turnover time (1/k2Z14.5 h) is 18 fold smaller than that of pool one (1/k 1Z255 h). Assuming a microbial assimilation YZ0.60 g C gK1 C, the size of the pool two can be assessed at 2.45 and 2.72 mmol C kgK1 in D and G aggregates, corresponding to 29.4 and 32.6 mg C kgK1 soil, respectively. The rate of CO2 production at time 0, i.e. the sum k1(1KY1) S10Ck2(1KY2) S20, can also be calculated. It was equal to 26.2 and 29.4 nmol kgK1 sK1, for D and G aggregates submitted to anaerobic conditioning, respectively (Table 1). These rates were 3.5 times higher than the corresponding rates in aggregates that had not experienced anaerobic conditioning. Table 1 Parameters obtained by fitting the CO2 production rate model (Eqs. 3–4) to the CO2 rates measured during a 4 d aerobic incubation with or without a preliminary anaerobic conditioning

(1KY1) S10 (mol kgK1) 1/k1 (h) (1KY2) S20 (mol kgK1) 1/k2 (h)

Aggregates D

Aggregates G

6.82!10K3 254 0.98!10K3 14.5

7.81!10K3 254 1.09!10K3 14.5

Accounting for the CO2 solubilisation in water and its first dissociation constant (see Appendix), the actual aerobic CO2 production rates were also calculated using the apparent CO2 values measured in experiment 2. The actual aerobic CO2 production rates were higher in experiment 1 than in experiment 2 (Table 4). The soluble organic C contents varied between treatments from 17.9 to 24.7 mg C kgK1 soil. They increased from the end of the rewetting period to the end of the anaerobic pre-incubation by 19 and 11% for D and G aggregates, respectively. We obtained linear relationships between measured soluble organic C contents and the more proximal CO2 production rate measured at the same date (Fig. 3), except for one value that seemed unrealistic, since it was obtained at the end of incubation and corresponded to the highest soluble organic C content for G aggregates. Without this value, the correlation coefficients were rZ0.954 and 0.979 for D and G aggregates, respectively. The linear regressions suggest that there is a threshold for soluble organic C below which there is no more aerobic respiration: the difference between the initial soluble organic C and this estimation represented approximately 23% and 9% of the initial soluble organic C for D and G aggregates, respectively. The pH of aggregates that were not submitted to anaerobic conditioning was, after the rewetting phase, 8.31G0.08 and 8.24G0.03 for D and G aggregates, respectively. After the aerobic incubation, the mean value of pH was 8.38 G 0.12 and 8.29G0.04, respectively. The pH of aggregates that experienced anaerobic conditions was, after the anaerobic conditioning, 8.34G0.02 and 8.37G0.05 for D and G aggregates, respectively. After the subsequent aerobic incubation, the mean value of pH was 8.74G0.01 and 8.47G0.03 for D and G aggregates, respectively. The results indicated no variation with time or aggregate type (D and G). 25 20

CO2 (nmol kg–1s–1)

Fig. 2. Rate of CO2 production (nmol kgK1 sK1) measured or simulated (Eq. (1–4)) in soil aggregates (type D or G) during an aerobic incubation following either no conditioning or a 6 d anaerobic pre-incubation. Each point is the mean of eight replicates made of four soil aggregates.

1337



15

Γ

10 5 0

0

5

10

15

20

25

SOLUBLE ORGANIC C (mg kg–1)

Fig. 3. Relationship between the rate of CO2 production (nmol kgK1 sK1) measured at various 24 h periods and the soluble organic C (mg C kgK1) measured at the same time, in the two types of aggregates. The periods are: after rewetting on tension table (d K6), after anaerobic conditioning (d 0), and at the end of the aerobic incubation (d 4).

1338

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

Table 2 Mean net N2O emission rates (pmol kgK1 sK1) measured in soil aggregates during aerobic incubation, for various rewetting conditions and pre-treatments Symbol D, G D*, G* D, G DC, GC

Rewetting KNO3 solution KNO3 solution KNO3 solution Deionised water

Conditioning None None None Anaerobic with C2H2

na

Incubation No C2H2 With C2H2 No C2H2 No C2H2

N2O emission rate (pmol kgK1 sK1)

40 40 10 10

D aggregates

G aggregates

K1.0 (0.4) 5.2 (3.2) K1.2 (0.8) 72.1 (9.4)

K0.3 (1.2) 3.3 (5.4) K0.9 (0.1) 48.3 (8.3)

Values in brackets represent the confidence intervals (P!0.05). a n, number of aggregates (replicates).

(a) 100 80

N2O (pmol kg–1s–1)

The net N2O emission rates measured in experiment 2 were small and even slightly negative in aggregates that did not experience anaerobic conditioning: K1.2 and K0.9 pmol N2O kgK1 sK1 in the sets of 10 D and 10 G aggregates, respectively, and K1.0 and K0.3 pmol N2O kgK1 sK1 in the sets of 40 D and 40 G aggregates, respectively (Table 2). In contrast, the net N2O emissions of aggregates that had been previously submitted to anaerobic conditions were high: 72.1 and 48.3 pmol N2O kgK1 sK1 in D and G aggregates, respectively. These rates were also much higher than the gross emission rates through denitrification (i.e. with C2H2) measured on 40 D and 40 G aggregates that did not experience anaerobic conditioning (5.2 and 3.3 pmol N2O kgK1 sK1 in D and G aggregates, respectively), although these rates were measured in favourable conditions (NOK 3 was supplied in large amount to promote denitrification and acetylene was added to prevent N2O reduction). The mean N2O emission rate following anaerobic incubation was higher (P!0.05) in D (72.1 pmol kgK1 sK1) than in G aggregates (48.3 pmol kgK1 sK1). The mean weight of D aggregates (25.8 g) was also significantly higher (P!0.05) than that of G aggregates (20.2 g). Fig. 4b suggests that N2O emission rate does not depend directly on aggregate type (D and G), but would rather depend on mass: this relationship could be due to a larger anoxic volume in the larger aggregates (Renault and Stengel, 1994; Sierra et al., 1995). K The NHC 4 and NO3 contents of aggregates that had been rewetted with water and submitted to anaerobic conditioning were low and did not differ between aggregates or dates: the K1 soil and average was 2.2G0.3 mg NHC 4 -N kg K1 K 1.5G0.9 mg NO3 -N kg soil (Table 3). In aggregates that had been rewetted with KNO3 solution and that did not experience anaerobic conditioning, the NHC 4 concentration was similar (2.2G0.3 mg N kgK1 soil), but the NOK 3 concentration was higher as expected: 96.0G8.9 mg N kgK1 soil, without significant change between the initial and final measurement (P!0.05). N2O emissions were low in these aggregates in spite of their high NOK 3 concentration.

The apparent CO2 production rate was also much higher in aggregates that had been submitted to anaerobic conditioning than in the others: 3.8 and 2.6 times higher in D and G aggregates, respectively (Table 4). The apparent rates were not statistically different between D and G aggregates not submitted to anaerobic conditions, but could be distinguished in aggregates submitted to anaerobic preincubation (significant at P!0.10). The calculated actual CO2 production rates, corrected for carbonates present in soil solution, were 1.8 to 2.8 times greater. The factor depends both on the aggregate type and its pre-treatment and on the constant k3 that characterises the relationship

∆ Γ ∆+ Γ+

60 40 20 0 –20

0

10

20

30

40

50

(b) 100

N2O (pmol kg–1s–1)

3.2. Experiment 2: N2O production with or without anaerobic conditioning

80

∆ Γ ∆+ Γ+

60 40 20 0 –20

0

10

20

30

40

50

Aggregate mass (g) Fig. 4. Net N2O emission rates (pmol kgK1 sK1) as a function of aggregate mass, measured: (a) in soil aggregates rewetted with nitrate solution and incubated aerobically with C2H2 (DC, GC) or without C2H2 (D, G); (b) in soil aggregates rewetted with deionised water and incubated aerobically without C2H2 either after anerobic conditioning and C2H2 (DC, GC) or without (D, G).

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

1339

Table 3 Amounts of mineral N (in mol mK3 solution or mg kgK1 soil) measured at the beginning and the end of 1 d aerobic incubation following either soil water rewetting and anaerobic conditioning (6 d) or nitrate rewetting alone (experiment 2) Aggregates

Rewetting

Conditioning

Mineral N (mol N mK3) tZ0

D G DC

KNO3 solution KNO3 solution Deionised water

GC

Deionised water

None None Yes, anaerobic with C2H2 Yes, anaerobic with C2H2

Mineral N (mg N kgK1) tZ24 h

tZ0

NHC 4

NOK 3

NHC 4

NOK 3

0.44 (.06) 0.43 (.03) 0.57 (.03)

32.9 (3.1) 0.97 (.14) 29.1 (1.8) 0.81 (.12) 0.38 (.18) 0.74 (.05)

0.71 (.09)

0.28 (.03) 0.73 (.05)

tZ24 h

NHC 4

26.5 (4.3) 1.4 (.2) 27.8 (4.2) 1.5 (.1) 0.35 (.03) 1.8 (.1) 0.85 (.4)

2.3 (.3)

NOK 3

NHC 4

NOK 3

106.2 (10.1) 99.2 (6.1) 1.2 (.6)

3.1 (.5) 2.7 (.4) 2.3 (.2)

85.5 (13.9) 92.8 (15.2) 1.1 (.1)

0.9 (.1)

2.4 (.2)

2.8 (1.3)

Values in brackets represent the standard errors.

between soil pH and partial pressure of CO2. These values remain lower than the values obtained for the same aggregates in experiment 1. 3.3. Experiment 3: origin of the N2O emissions following anaerobic conditioning The N2O emission of D aggregates supplied with 15 NHC 4 was about twice that in aggregates supplied with 15 NOK 3, although the difference between these two values was not statistically significant (P!0.05). However, for aggregates submitted to anaerobic conditioning and thereafter enriched 15 15 either with 15 NOK NHC N of 3 or 4 , the isotopic excess in produced N2O was similar to the mean excess of NOK 3 defined as the average between the initial and final excess of NOK 3 (Table 5).

4. Discussion

suggesting a dependence on aeration within the aggregate. Surprisingly, we found that the net N2O emission rate was high in aggregates submitted to anaerobic conditioning and K3 having a small NOK solution) 3 content (0.46 mol m and low in aggregates that did not experience anaerobic conditioning but had a high NOK concentration 3 (29.1 mol mK3 solution). The lower NOK 3 concentration is comparable or smaller than the Michaelis constant values reported in the literature: 11.4–31.4 mol mK3 (Malhi et al., 1990), approximately 0.22 mol mK3 (Klemedtsson et al., 1977), 1.8–16.6 mol mK3 for soil and sediment slurries (Murray et al., 1989) and 0.3–3.8 mol mK3 for purified NOK 3 reductase (Zumft, 1997). Although NOK 3 concentration was likely to be a limiting factor of N2O emissions in aggregates conditioned anaerobically, there was no correlation between individual aggregate emission and its final NOK 3 concentration (results not shown). Under these conditions, the other factors affecting denitrification, particularly the aeration status of the aggregate and its content in available C, were more important under these conditions.

4.1. N2O production with or without anaerobic conditioning During the aerobic incubation of soil aggregates, net N2O emission rates were higher in aggregates that were submitted to anaerobic conditioning (60.2 pmol N2O kgK1 sK1) than in those that did not (K1.05 pmol N2O kgK1 sK1). The 15N measurements clearly demonstrated that most of these emissions resulted from denitrification. We also observed that these emissions increased with the aggregate size,

4.2. Why N2O emission through nitrification should be neglected? Aggregates conditioned anaerobically were simultaneously submitted to C2H2 (at 5 kPa) to check whether denitrification stopped at the end of this period. It has been recognised that both NHC 4 oxidation and N2O production through nitrification are inhibited by C2H2 when its partial

Table 4 Apparent and actual CO2 production rates (nmol CO2 kgK1 sK1) during 6 h of aerobic incubation of D and G aggregates Aggregate type

Apparent CO2 production

Actual CO2 production k3Z0.83 (nmol CO2 kg 12.5 8.99 2.72 2.86

s

K1 K1

DC GC D G

5.42 (1.14) 4.23 (0.70) 1.41 (0.31) 1.63 (0.99)

k3Z1.25

k3Z1.66

14.3 10.1 3.12 3.30

15.3 10.8 3.36 3.56

)

Apparent rate is the accumulation rate measured in flasks of experiment 2; actual rate is calculated using equation 8 for three values of k3: k3Z2, 3 and 4%, corresponding to 0.83, 1.25 and 1.66 mol mK3. DC and GC aggregates have been submitted to conditioning (anaerobiosis incubation for 6 d with C2H2), whereas D and G aggregates have not. Values in brackets are the confidence intervals.

1340

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

Table 5 N2O emission rate, amounts of mineral N and isotopic composition of N2O and mineral N measured in D aggregates, during a short aerobic incubation (7 h) following a 6.6 d anaerobic incubation with C2H2

15

N-urea NO3 15 N-urea 15 NO3 15

No C2H2 No C2H2 C2H2 C2H2

N2 O (pmol kgK1 sK1)

N2 O (mg kgK1)

NH4 (mg kgK1)

NO3 (mg kgK1)

N2O (at%)

NH4 (at%)

NO3 (at%)

287 (283) 173 (292) 195 (122) 114 (28)

0.10 (.09) 0.06 (.09) 0.07 (.04) 0.04 (.01)

3.35 (1.65) 2.94 (1.06) 3.36 (.22) 2.97 (.87)

1.45 (1.29) 1.32 (.69) 0.12 (.12) 0.65 (.14)

0.33 (.33) 3.15 (4.08) 0.31 (.14) 4.48 (.57)

2.02 (.92) 0.66 (.11) 2.81 (.25) 0.70 (.28)

0.36 (.15) 3.40 (1.14) 0.24 (.09) 4.61 (.58)

15

Pool sizes and isotopic composition of mineral N are the mean of values measured at tZ0 and tZ7 h. N labelling was made using the anaerobiosis period. Values in brackets are the standard deviations (3 replicates).

pressure is higher than 1–10 kPa (Mosier, 1980). Moreover, the inhibitory effect of C2H2 may not cease after its removal C2H2: Berg et al. (1982) noted that C2H2 effects has not disappeared 7 d after removing the C2H2 initially applied at 10 Pa. We carried out additional 15N measurements in experiments 2 and 3 on aggregates submitted to anaerobic conditioning without C2H2. Nitrification rates in aggregates submitted or not submitted to C 2H 2 were 0.54 nmol kgK1 sK1 and 5.6 nmol kgK1 sK1, respectively (results not shown), and confirm the effect of C2H2 on nitrification. 4.3. Required conditions to increase N2O emissions through denitrification The use of C2H2 could overestimate gross denitrification, because the reduction of NOK 3 to N2O involves a lower electron flow than the reduction of NOK 3 to N2. Since this estimate of the gross denitrification rate of aggregates without anaerobic conditioning was negligible compared with the net N2O emission rate (i.e. without C2H2) of aggregates that experienced anaerobic conditioning, it appears that the aeration status of aggregates placed in airfilled flasks is affected by the anaerobic conditioning. Indeed, experimental measurements of aerobic CO2 production showed that aggregates with anaerobic conditioning increased CO2 production by a factor of 2.2 and 1.9 (experiment 1), and 3.8 and 2.6 (experiment 2) in D and G aggregates, respectively. However, these two methods lead to different rates of aerobic CO2 production. As shown in Table 3, the main difference between these two experiments lies in the method (trapping CO2 in NaOH solution or accumulation in the gas phase). When CO2 accumulates in the gas phase, it also partly accumulates in the solution mainly as hydrated H2CO3 and HCOK (Stumm and 3 Morgan, 1996). The actual rate of CO2 production was calculated from its apparent value measured in experiment 2 using unproved hypotheses on (i) the relationship between pH and air CO2 concentration that was proposed from geochemical simulations for hypothetical calcareous and non calcareous soil using the AQUA model (Valle`s and Bourgeat, 1988) and (ii) the dissolution of calcite that was neglected. Aerobic CO2 production was lower in D than in G aggregates in experiment 1, whereas the inverse occurred in

15

NOK 3

15 N2O/ (15N2C15N2O)

0.04 (.06) 0.78 (.39) or 15N-urea before

experiment 2. Therefore, other causes are surely to be invoked, including the storage conditions before measurements: the period of storage varied in experiments 1 and 2 as well as the water content of aggregates (0.06 g gK1 and 0.11 g gK1, respectively). In addition to an increase in aerobic respiration, the N2O/(N2OCN2) ratio may have increased too. However, such an increase could not have been sufficient alone to explain the difference between aggregates with or without anaerobic conditioning, since total denitrification of aggregates without anaerobic conditioning was lower than the N2O emissions of other aggregates. An increase in N2O/(N2OCN2) ratio could a priori result from a decrease in the pH of the soil solution (Knowles, 1982), an increase in [NOK 3 ] (Blackmer and Bremner, 1979), or a decrease in aerobic respiration (Parkin, 1987). In our studies, the last two hypotheses can not be retained, since [NOK 3 ] and aerobic respiration were lower and greater, respectively, for aggregates with anaerobic conditioning. Therefore, only a pH decrease might be possible; it could have resulted from the accumulation of fatty acids during anaerobic conditioning (Dassonville and Renault, 2002) or from an increase in CO2 partial pressure (in experiment 2) (Tsusuki and Ponnamperuma, 1987; Stumm and Morgan, 1996). However, the pH measured after re-equilibration of the soil solution with atmospheric CO2 did not change with the anaerobic conditioning; the CO2 concentration at the end of the measurement period (1.65 mmol mK3 air) could induce a decrease in soil pH greater than 1 unit (Dassonville et al., 2004). In addition, the presence of C2H2 during anaerobic conditioning inhibits both N2O reduction and the synthesis of N2O reductase (Klemedtsson et al., 1977). C2H2 is likely to have maintained N2O/(N2OCN2) ratio at a high value, and may even have increased it since denitrifier activity was stimulated by the long anaerobic conditioning. 4.4. Biogeochemical changes that could appear during the prolonged anaerobic period Anaerobic conditions often lead to the incomplete degradation of organic matter and the accumulation of small compounds, particularly fatty acids (Tsusuki and Ponnamperuma, 1987; Dassonville and Renault, 2002), which can be easily mineralised during subsequent aerobic

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

conditions (Burford and Bremner, 1975). This accumulation of small organic compounds may be reflected indirectly by the increase of soluble organic C during the anaerobic conditioning. It justified our choice to consider two organic pools in Eq. (4) to explain aerobic CO2 production: the first pool being for native organic C, and the second one for small organic compounds. In agreement with this interpretation, the period constant 1/k1 for mineralisation of the first pool was about 200–260 h; in contrast, the corresponding value 1/k2 was only about 4–20 h, indicating that some organic compounds can easily be mineralised in aerobic conditions after the anaerobic pre-incubation.

Acknowledgements This work was supported by INRA and Re´gion Picardie, and the programs PNSE and GESSOL. We thank O. Delfosse and G. Alavoine for their technical assistance. We would like to acknowledge J. Chadœuf for helpful discussions. We are grateful to A.M. Wall for reviewing the English version of the manuscript.

Appendix A. Estimating actual CO2 production from its accumulation in experiment 2 Since CO2 is soluble and dissociated in water, gaseous CO2 is only a part of total content present in the flask when the gas is not trapped as in Experiment 2. The total amount Q of CO2 (mol) in the flask is equal to: Q Z Cf ðVg C Vw Sð1 C 10pHf KpKa ÞÞ where Cf is the CO2 concentration in the flask atmosphere (mol mK3), Vg and Vw are the gas and soil solution volumes (m3), respectively; S is the liquid H2CO3 concentration to gaseous CO2 concentration ratio equal to 0.87 at 20 8C, pKa is the Log of the first acidity constant equal to 6.35 at 20 8C (Stumm and Morgan, 1996). If we neglect the contribution of calcite in the production of HCOK 3 anions, the actual CO2 production rate Pa (mol kgK1 sK1) can be deduced from the apparent production P^ (mol kgK1 sK1):   Vw V S ^ Pa Z P 1 C S C w ðCf 10pHf KpKa K Ci 10pHiKpKa Þ m Dt Vg where m is the aggregate mass (kg), Dt the incubation time (s), and i and f are subscript for initial and final values of pH. In addition, we assume the following dependence of the pH to CO2 concentration in the gas phase:   C pH Z pHmin C ðpHmax K pHmin Þexp K k3 where pHmax is the maximum pH value in absence of CO2, pHmin its minimum value at high CO2 partial pressure, and

1341

k3 a constant (mol mK3), which characterises the dependence on CO2 concentration (C). For the calculations, we put pHmaxZ8.20, pHminZ5.75 and we successively estimated the actual aerobic CO2 production rate for k3 equal to 0.83, 1.25 or 1.66 mol mK3, corresponding to 2, 3 or 4% CO2, respectively. These values were obtained by fitting simulations of more complete geochemical models to this simple relationship for soils with traces of calcite.

References Barcelona, M.J., 1984. TOC determinations in ground water. Groundwater 22, 18–24. Berg, P., Klemedtsson, L., Rosswall, T., 1982. Inhibitory effect of low partial pressures of acetylene on nitrification. Soil Biology & Biochemistry 14, 301–303. Blackmer, A.M., Bremner, J.M., 1979. Stimulatory effect of nitrate on reduction of N2O to N2 by soil microorganisms. Soil Biology & Biochemistry 11, 313–315. Brooks, P.D., Star, J.M., McInteer, B.B., Preston, T., 1989. Diffusion method to prepare soil extracts for automated nitrogen-15 analysis. Soil Science Society of America Journal 53, 1707–1711. Burford, J., Bremner, J.M., 1975. Relationships between the denitrification capacities of soils and total water soluble and readily decomposable soil organic matter. Soil Biology & Biochemistry 11, 261–267. Conrad, R., 1996. Soil micro-organisms as controllers of atmospheric trace gases (H2, CO, CH4, OCS, N2O and NO). Microbiological Reviews 60, 609–640. Dassonville, F., Renault, P., 2002. Interactions between microbial processes and geochemical transformations under anaerobic conditions: a review. Agronomie 22, 51–68. Dassonville, F., Renault, P., Valle`s, V., 2004. A model describing the interactions between anaerobic microbiology and geochemistry in a soil amended with glucose and nitrate. European Journal of Soil Science 55, 29–45. Graedel, T.E., Crutzen, P.J., 1992. Atmospheric change, An earth system perspective. W.H. Freeman, New York p. 446. Grant, R.F., 1995. Mathematical modelling of nitrous oxide evolution during nitrification. Soil Biology & Biochemistry 27, 1117–1125. Greenwood, D.J., 1961. The effect of oxygen concentration on the decomposition of organic materials in soil. Plant and Soil 14, 360–376. Greenwood, D.J., Berry, G., 1962. Aerobic respiration in soil crumbs. Nature (London) 195, 161–163. Groffman, P.M., 1991. Ecology of nitrification and denitrification in soil evaluated at scales relevant to atmospheric chemistry, in: Rogers, J.E., Whitman, W.B. (Eds.), Microbial Production and Consumption of Greenhouse Gases: Methane, Nitrogen Oxides and Halomethanes. American Society for Microbiology, Washington, pp. 201–217. He´nault, C., Germon, J.C., 1995. Quantification de la de´nitrification et des e´missions de protoxyde d’azote (N2O) par les sols. Agronomie 15, 321–355. IPCC, 1996. Radiative forcing of climate change, in: Houghton, J.T., Meira Filho, L.G., Callander, B.A., Harris, N., Kattenberg, A., Maskell, K. (Eds.), Climate Change 1995: The Science of Climate Change. Cambridge University Press, Cambridge, pp. 65–131. Kamphake, L.J., Hannah, S.A., Cohen, J.M., 1967. Automated analysis for nitrate by hydrazine reduction. Water Research 1, 205–216. Klemedtsson, L., Svensson, B.H., Lindberg, T., Rosswall, T., 1977. The use of acetylene inhibition of nitrous oxide reductase in quantifying denitrification in soils. Swedish Journal of Agricultural Research 7, 179–185.

1342

K. Khalil et al. / Soil Biology & Biochemistry 37 (2005) 1333–1342

Knowles, R., 1982. Denitrification. Microbiological Reviews 46, 43–70. Krom, M.D., 1980. Spectrometric determination of ammonia: a study of a modified Berthelot reaction using salicylate and dichloroisocyanurate. The Analyst 105, 305–316. Malhi, S.S., McGill, W.B., Nyborg, M., 1990. Nitrate losses in soils: effect of temperature, moisture and substrate concentration. Soil Biology & Biochemistry 22, 733–737. Marquardt, D., 1963. An algorithm for least-squares estimation of non linear parameters. SIAM Journal of Applied Mathematics 11, 431–441. Mosier, A.R., 1980. Acetylene inhibition of ammonium oxidation in soil. Soil Biology & Biochemistry 12, 443–444. Murray, R.E., Parsons, L.L., Smith, M.S., 1989. Kinetics of nitrate utilization by mixed populations of denitrifying bacteria. Applied and Environmental Microbiology 55, 717–721. Parkin, T.B., 1987. Soil microsites as a source of denitrification variability. Soil Science Society of America Journal 51, 1194–1199. Parry, S., Renault, P., Chadoeuf, J., Chenu, C., Lensi, R., 2000. Particulate organic matter as a source of variation in denitrification in clods of soil. European Journal of Soil Science 51, 271–281. Parton, W.J., Mosier, A.R., Schimel, D.S., 1988. Rates and pathways of nitrous oxide production in a shortgrass steppe. Biogeochemistry 6 (3), 45–58. Parton, W.J., Mosier, A.R., Ojima, D.S., Valentine, D.W., Schimel, D.S., Weier, K., Kulmala, A.E., 1996. Generalized model for N2 and N2O production from nitrification and denitrification. Global Biogeochemical Cycles 10, 401–412. Recous, S., Fresneau, C., Faurie, G., Mary, B., 1988. The fate of labelled 15 N urea and ammonium nitrate applied to a winter wheat crop. Plant and Soil 112, 205–214. Renault, P., Stengel, P., 1994. Modeling oxygen diffusion in aggregated soils: I. Anaerobiosis inside the aggregates. Soil Science Society of America Journal 58, 1017–1023.

Richard, G., Boizard, H., Roger-Estrade, J., Boiffin, J., Gue´rif, J., 1999. Study of soil compaction due to traffic in field conditions: pore space and morphological analysis. Soil and Tillage Research 51, 151–160. Sextone, A.J., Revsbech, N.P., Parkin, T.B., Tiedje, J.M., 1985. Direct measurement of oxygen profiles and denitrification rates in soil aggregates. Soil Science Society of America Journal 46, 68–74. Sierra, J., Renault, P., Valle`s, V., 1995. Anaerobiosis in saturated soil aggregates: modelling and experiment. European Journal of Soil Science 46, 519–531. Smith, K.A., 1990. Greenhouse gas fluxes between land surfaces and the atmosphere. Progress in Physical Geography 14, 349–372. Stumm, W., Morgan, J.J., 1996. Aquatic chemistry, Third ed, Chemical Equilibra and Rates in Natural Waters. John Wiley, New York p. 1022. Tortoso, A.C., Hutchinson, G.L., 1990. Contributions of autotrophic and heterotrophic nitrifiers to soil NO and N2O emissions. Applied and Environmental Microbiology 56, 1799–1805. Tsusuki, K., Ponnamperuma, F.N., 1987. Behavior of anaerobic decomposition products in submerged soils. Effects of organic material amendment, soil properties and temperature. Soil Science and Plant Nutrition 33, 13–33. Valle`s, V., Bourgeat, F., 1988. Geochemical determination of the gypsum requirement of cultivated sodic soils. I. Development of the thermodynamic model Gypsol simulating the irrigation water- soil chemical interactions. Arid Soil Research and Rehabilitation 2, 165–177. Zumft, G., 1997. Cell biology and molecular basis of denitrification. Microbiology and Molecular Biology Reviews 61, 533–616.