OUP user menu

Nitrous oxide production in a forest soil at low temperatures – processes and environmental controls

Mats G. Öquist , Mats Nilsson , Fred Sörensson , Åsa Kasimir-Klemedtsson , Tryggve Persson , Per Weslien , Leif Klemedtsson
DOI: http://dx.doi.org/10.1016/j.femsec.2004.04.006 371-378 First published online: 1 September 2004


Recent investigations have highlighted the relative importance of the winter season for emissions of N2O from boreal soils. However, our understanding of the processes and environmental controls regulating these emissions is fragmentary. Therefore, we investigated the potential for, and relative importance of, N2O formation at temperatures below 0 °C in laboratory experiments involving incubations of a Swedish boreal forest soil. Our results show that frozen soils have a high potential for N2O formation and subsequent emission. Net N2O production rates at −4 °C equaled those observed at +10 to +15 °C at moisture contents >60% of the soil's water-holding capacity. The source of this N2O was found to be denitrification occurring in anoxic microsites in the frozen soil and temperature per se did not control the denitrification rates at temperatures around 0 °C. Furthermore, both net nitrogenmineralisation and nitrification were observed in the frozen soil samples. Based on these findings we propose a conceptual model for the temperature response of N2O formation in soils at low temperatures.

  • Denitrification
  • Low-temperature
  • Nitrous oxide emission
  • Soil nitrogen transformation

1 Introduction

Prompted to a large extent by the risk and expectation that climate change will follow atmospheric increases in trace gas concentrations (e.g. CO2, CH4, N2O and halocarbon), there is intense interest in the exchange of these gases between terrestrial systems and the atmosphere and in the strength of their respective emissions and consumption [1]. N2O, which has a long atmospheric lifetime (120 y) and a 296 times stronger global warming potential than CO2[2], is of key concern. The annual global emission from soils is estimated to amount up to 9.5 ± 4.5 Tg N, and to contribute ca. 65% of all N2O emissions to the atmosphere [1]. These estimates are mainly based on investigations conducted during the growing season, but there is increasing evidence that wintertime emissions of N2O from northern mid to high latitude soils also have the potential to influence the magnitude of atmospheric loading on an annual basis [310]. N2O fluxes from soils to the atmosphere are regulated by microbial activity through the processes of nitrification and denitrification e.g. [11]. However, the understanding of the environmental controls of N2O production at low temperatures and the processes involved is still, in many respects, speculative.

The cold season in boreal systems can be broadly divided into three different periods, all of which have been shown to be of importance with respect to soil nitrogen dynamics. The first is late autumn, when temperatures gradually decrease towards 0 °C, after which soil freezing is initiated. The second is mid-winter, during which the surface soil remains frozen for a more prolonged period of time, and the third is late winter, with periods of freeze–thaw cycles associated with the degradation of soil ice. The biogeochemical parameters affecting nitrogen transformation and N2O formation processes differ widely in these three periods. The extent and duration of each period in any given boreal ecosystem soil may vary dramatically, depending on latitudinal location, local climate conditions and year-to-year variations in climate, but overall, low temperature conditions prevail during ca. 25–50% of the annual cycle (i.e. 3–6 months).

So far, the main scientific emphasis on low temperature N2O formation and emission has been directed towards the period associated with soil frost thaw during late winter and early spring, but a few recent investigations have emphasized the potential of N2O formation in frozen soils during mid-winter as an important source of atmospheric emissions [6,12]. However, our conceptual understanding of the environmental controls regulating these mid-winter processes is still speculative.

This study has two main aims. The first is to investigate the effects of temperature (ranging from −4 to 25 °C) on the net formation of N2O in order to elucidate the potential of the winter season on forest soil N2O emissions. The second is to determine whether nitrification or denitrification is the main source of N2O originating from frozen soil.

2 Materials and methods

The forest soil used in this study was sampled from limed plots (8.75 t dolomite ha−1) in a 45-year-old Picea abies (L.) Karst stand (56°24′N, 13°00′E, 185 m above msl, mean air temperature 6.5 °C, annual precipitation 1100 mm). The plots were limed in 1984 and sampled in 1994. The humus layer sampled had 35.5% C, 1.6% N and a pH (H2O) of 6.0. In order to meet the stated objectives of this investigation, two different experimental approaches were adopted. In Experiment 1, the net effect on N2O formation was evaluated by incubating soil samples at temperatures ranging from −4 to 25 °C at different water contents. Experiment 2 was used to distinguish between the importance of nitrification and denitrification as sources of N2O. For this purpose, slurried soil samples were incubated under different O2 regimes with or without dimethyl ether (which inhibits nitrification without disturbing denitrification) and acetylene (which inhibits the reduction of N2O to N2 in denitrification). Furthermore, KNO3 amendments were made to check for potential N limitation associated with denitrification.

2.1 Experiment 1

The humus layer was freshly sifted, and portions of 16 g (dry wt) were incubated (in 0.466 l containers) under constant conditions at 25, 15, 5, 0.5 and −4 °C and at moisture levels of 15%, 30%, 60% and 100% of the soil's water-holding capacity (WHC; n= 3). 100% WHC was defined as the water content obtained when an inundated soil was allowed to drain for 12 h in a 3-cm high cylinder. The N2O emission was measured four times over an 89-day period by replacing open lids (with a 4-mm hole) with gas-tight lids equipped with rubber septa. Gas samples were then taken from the headspace 4–24 h after closure (depending on temperature) and the gas was analysed according to Klemedtsson et al. [13]. The water content of the soil samples was checked once a week by weighing the incubation bottles, and water loss was compensated for by adding water. In addition, net N mineralisation (accumulation of NH4+–N plus NO2/NO3–N) and net nitrification (accumulation of NO2/NO3–N) were also measured, as described by Persson et al. [14].

Net N mineralisation estimates at different temperatures were fitted to the following function for sub-optimal temperatures [15,16]: Embedded Image where r denotes the activity at temperature θ.

2.2 Experiment 2

The slurry experiments involved the incubation of soil slurries in 250-ml flasks (10 g fresh forest soil, pre-incubated aerobically at 0 °C before the start of the experiments; 100 ml solution), stopped with rubber septa. Two different slurry solutions were used in two parallel series of incubations: a mild phosphate buffer solution [17], which was frozen at incubation temperatures below 0 °C, and a synthetic saltwater solution (SSW, with a salinity of 35% and a freezing point of −1.8 °C). Experiment 2 was carried out in two parts, one with a low oxygen atmosphere, to determine whether the N2O formed was of biological origin and also to see whether it came from nitrification or denitrification. The other part was carried out anaerobically in order to determine the temperature response on N2O formation from denitrification.

For the oxic incubations, the soil was first incubated with a headspace O2 concentration of 0.5 kPa to ensure that the cultures were oxygen limited, but not anaerobic (n= 2). Dimethyl ether was used as a nitrification inhibitor, since it does not affect denitrification [20] and is highly soluble in water. By inhibiting nitrification, this treatment allows the potential N2O production from the denitrification process to be separately measured. The dimethyl ether was added just before the start of the incubation. Replicate samples were amended with acetylene (10 kPa) in order to inhibit the final step in denitrification. A control treatment was also used to measure the N2O production from both nitrification and denitrification. The inhibitors were added just before the samples were incubated at −0.8 °C. From the three treatments it was possible to differentiate between the main sources of N2O production and to measure the ratio of N2O:N2 originating from denitrification. In order to determine whether the N2O produced was of biological or non-biological origin, replicate samples were autoclaved (120 °C for 20 min) prior to incubation (n= 2).

For the anoxic incubations, replicate samples (n= 4) were amended with 10 kPa of acetylene (just before incubation) in order to inhibit the final step in denitrification [18]. Control samples (n= 4) without acetylene were used to estimate the ratios between the end products of denitrification. To ensure that electron acceptor availability did not restrict N2O production, 1 mM KNO3 was added to the samples. In addition, replicate samples without KNO3 amendments were incubated at −0.8 °C to evaluate the possibility of N limitation (n= 2). The incubations were carried out under anoxic conditions at −1.6, −0.8 and +1.6 °C in a rotary shaking bath. Gas samples for N2O analyses were taken six times over a 120-h incubation period.

N2O analysis of the gas phase was performed by GC-ECD according to [23], and dissolved N2O gas was corrected for according to [19]. The N2O production rates were calculated from the slope of increase in N2O concentration, obtained from linear regression of the data in the anoxic incubations. Because there was a short lag phase, the initial data points were omitted. In the frozen buffer samples, we also calculated the total production rates, measured after thawing and including the N2O trapped in the ice during the incubations. The frozen buffer samples were rapidly thawed in a microwave oven and allowed to equilibrate at 0 °C before sampling the headspace. In this case, we only had data from the end-points of the experiment. Thus, the rates were calculated by assuming zero-order kinetics of N2O formation from no N2O at the start to the measured total amount of N2O at the end of the experiment.

3 Results

3.1 Incubations at different temperatures and WHC permutations

In soil incubations in which temperature and moisture conditions were varied, the amount of N2O formed was strongly dependent on temperature at temperatures above zero (Fig. 1). Total amounts were highest at 25 °C (ca. 3–4 μg N2O–N g−1 89 d−1) and decreased with each drop in incubation temperature (1.5, 1 and 0.25 μg N2O–N g−1 89 d−1 for 15, 5, and 0.5 °C, respectively). For samples incubated at 15% and 30% WHC, this decline continued below 0 °C, and at −4 °C the amounts of N2O–N formed ranged between 0.10 and 0.15 μg N2O–N g−1 89 d−1. However, at high water contents a significant increase in potential N2O formation was observed at temperatures below zero. At 60% WHC, N2O–N formed at −4 °C ranged around 1 μg N2O–N g−1 (comparable to the amounts formed at 5 °C) and at 100% WHC the samples produced ca. 2 μg N2O–N g−1 (comparable to the amounts formed at 15 °C; Fig. 1), clearly demonstrating the potential for N2O production in frozen soils. The incubations of autoclaved soil samples confirmed that the N2O formed at sub-zero temperatures was of biological origin (data not shown).


Cumulative evolution of N2O after 89 days of incubation of humus-layer samples from a spruce forest (Hasslöv, Sweden) at different temperatures and moisture levels: 15% WHC, corresponding to the wilting point and 60% WHC, corresponding to the optimum moisture content for aerobic soil micro-organisms. Error bars are SE (except for 15% WHC, the SE were less than 20%).

At temperatures above 0 °C, mean net N mineralisation rates increased with increasing water content up to 60% WHC (Table 1). However, at 100% WHC net N mineralization was not observed in any of the samples. Fitting the net N mineralization measurements to the function for sub-optimal temperatures yielded an estimate of −9.3 °C for the minimum temperature at which mineralization could occur. At temperatures above zero, the proportion of nitrate N as a percentage of total inorganic N (NH4+–N + NO2/NO3–N) at the end of the incubation ranged between ca. 86% and 100% (Table 2). At −4 °C, however, this ratio dropped to ca. 40%. Furthermore, no net accumulation of inorganic N was found at 100% WHC (Table 2), showing that the gaseous N-losses from the samples always exceeded the net mineralisation, regardless of incubation temperature.

View this table:

Mean net nitrogen mineralisation rates (μg N g soil−1 d−1) at different temperatures and moisture levels. No net N mineralisation was found at 100% WHC, probably because of extensive gaseous N losses

% WHC−4 °C0.5 °C5 °C15 °C25 °C
150.6 (0.2)0.8 (0.2)2.2 (0.7)4.8 (0.7)10.5 (1.2)
300.8 (0.1)1.2 (0.3)3.9 (0.1)9.2 (0.2)16.3 (0.3)
600.2 (0.1)2.2 (0.5)4.7 (0.3)13.4 (0.1)26.4 (1.5)
  • SE are shown in parenthesis, n= 3.

View this table:

NO3–N as % of total inorganic N (NH4+–N + NO2–N/NO3–N) at the end of the incubation (307 days). No net accumulation of NH4+–N or NO3–N was found at 100% WHC

% WHC−4 °C0.5 °C5 °C15 °C25 °C

3.2 Factors controlling nitrification and denitrification

In the soil sample slurries incubated at low oxygen concentrations (0.5 kPa), the nitrification inhibitor dimethyl ether did not significantly reduce N2O production (formation rates: ca. 5 ng N2O–N (g soil)−1 h−1), showing that denitrification was the main source of N2O. Furthermore, the autoclaved samples showed no detectable N2O formation.

Anoxic conditions only allow denitrification, and in the anoxic soil sample buffer slurries formation rates of N2O released to the gas phase declined with decreasing temperature: ca. 25, 10 and 5 ng N2O–N (g soil)−1 was produced at 1.6, −0.8 and −1.6 °C, respectively (Fig. 2). However, this was not the case when soil samples were incubated with SSW (where the slurry did not freeze during incubation). Although SSW decreased overall N2O formation rates, the rates measured in SSW incubations did not differ significantly with respect to temperature. Furthermore, after rapidly thawing the samples incubated in buffer solution and measuring the total amount of N2O released (including that trapped in the ice), it was found that they followed the same pattern as the soils incubated in SSW. The total amounts of N2O formed at different temperatures did not differ significantly (Fig. 2). Thus, the temperature per se had no effect on the N2O-forming potential of the soil microorganisms in this temperature range. Moreover, the temperature did not affect the ratio of N2/N2O resulting from denitrification in the anoxic buffer solutions, and ca. 70% of the gaseous N-loss could be attributed to N2O across the investigated temperature range (Fig. 2). KNO3 addition did not significantly affect the N2/N2O ratio or the N2O formation rates at −0.8 °C (which averaged around 8 ng N2O–N (g soil)−1 h−1), demonstrating that denitrification was not subject to N-limitation.


Production rates of N2O from anaerobic incubations of limed forest soil suspensions at different temperatures. The soil was suspended in Synthetic Salt Water (triangles) or a phosphate buffer (squares). The solid and broken lines indicate the presence and absence of acetylene (which inhibits further reduction of N2O to N2), respectively. The filled squares represent data points from the incubations with buffer after thawing, and show the total amounts of gas that were produced during the incubation at −0.8 and −1.6 °C, but to a large extent were trapped in the ice. Error bars show 95% confidence intervals.

4 Discussion

The observed potential for N2O formation and its response to temperature changes above 0 °C are consistent with the results of earlier investigations on soils e.g. [11]. However, the significant increase in formation rates we found in soil samples with high water contents incubated at −4 °C deviates from the expected pattern, because they were comparable to rates observed at higher temperatures. This detection of N2O formation in frozen soil supports earlier reports of similar observations in both field [6,12,2123] and laboratory investigations [24]. However, several different mechanisms have been proposed to explain the soil N2O emissions from frozen soils that are, to some extent, contradictory. For example, Kaiser and Heinemeyer [22] proposed that constant emissions from frozen soils occurred because N2O formed in the unfrozen subsoil was transported through frost-induced cracks in the surface soil, while Goodroad and Keeney [25] suggested that the emissions may be caused by microbial activity at the soil surface in association with diurnal temperature fluctuations. In contrast, Teepe et al. [24] suggested that the production of N2O in the frozen soil emanates from microorganisms harboured in thin water films, where the soil water remains unfrozen. Our results support an explanation based on microbial activity in unfrozen water, because we detected considerable N2O formation from frozen bulk soil samples at −4 °C. The results from the slurry incubations at low oxygen concentrations, where the presence of acetylene increased N2O formation while dimethyl ether had no effect, also supports the postulated theory [24] that denitrification rather than nitrification is the dominating process for N2O production at low temperatures. Thus, it appears that the theory of anoxic microsite formation in frozen bulk soils is valid for our samples.

One conceivable explanation for the observed increase in N2O formation rates at sub-zero temperatures is that the N2O/N2 ratio of total denitrification may increase as temperatures decrease below 0 °C, resulting in a relative increase in N2O formation and emission, although the overall denitrifying activity decreases. This is consistent with some previous findings, since it has been shown that low temperatures suppress N2O reductase [26]. However, the results from the anoxic slurry incubations with and without inhibition of the step from N2O to N2 contradict this explanation, because the N2O/N2 formation ratio was not significantly affected over the temperature range investigated. Similarly, Holtan-Hartwig et al. [27] found that the differences in activation energies of N2O production and reduction (as estimated from Arrhenius functions) could not explain higher N2O emission rates from soils at low temperatures. Another reported source of N2O in frozen soils is chemodenitrification from NO2[28]. However, autoclaving our soil samples inhibited N2O formation, showing that chemodenitrification was negligible and this is also consistent with observations from γ-ray treated soil [6] and CHCl3-treated soil [25]. Nevertheless, it should be mentioned that the knowledge of chemical reactions in soils at frozen–unfrozen interfaces, where transiently high solute concentrations may arise, is still fragmentary e.g. [29]. One might argue that the temperature range investigated in this study (i.e. −1.6 to +1.6 °C) is too narrow to draw far-reaching conclusions concerning winter N2O formation. However, it is a range typically observed in the surface soil layers of boreal forests during the cold season, and it also facilitates observation of the microbial response as the soil is subjected to the environmental changes associated with soil water freezing and ice crystallisation.

The soil pore water content prior to freezing appears to have a major influence on N2O formation in frozen soils. This contradicts the findings of Teepe et al. [24] who found no correlation between water-filled pore spaces and N2O formation rates from frozen soil columns, but it does support indications gained from field investigations of N2O emissions from frozen ground reported by Röver et al. [6]. We conclude that the degree of water saturation of the soil prior to freezing is an important regulating factor for N2O formation in frozen soils, probably because it favours the formation of anoxic microsites and water films associated with soil aggregates by reducing the diffusion of oxygen into the soil matrix. Furthermore, the slurry incubations revealed that temperature per se had no effect on N2O formation rates over the investigated temperature range (i.e. 1.6 to −1.6 °C), strengthening the hypothesis that anoxic microsite formation is the main factor governing N2O emissions from frozen soils. That frozen soils can contain water in the liquid state has been shown by previous investigations e.g. [40]. Indeed, as much as 8–20% of the soil water can remain unfrozen, even if the soil temperature is kept at −5 °C for several days [30]. Dorland and Beauchamp [31] concluded that denitrification in a silt loam soil could occur down to −2 °C, providing that the soil was in a super-cooled, unfrozen state. Our results conflict with this conclusion, because the potential for denitrification did not differ at temperatures below 0 °C, regardless of whether the soil samples were frozen or not.

The regulation of N2O formation in a soil system is complex, since the regulating factors are not independent of each other. In addition to the development of anoxic microsites, resulting from the reduced rates of gas diffusion through the ice, a number of other properties also change. For example, during ice crystallization ions and nutrients are excluded from the growing ice grid, with consequent increases in concentration in the remaining unfrozen water [29,30]. Moreover, the availability of carbon substrates and organic N fractions may rise, due to the lysis of ruptured cell structures associated with freezing [32,33]. An important finding revealed by our investigation is the fact that both net nitrification and net N mineralisation occur at temperatures down to −4 °C, since N2O production requires inorganic N to be available. This implies that N2O formation in frozen soils can be sustained over several months and should be viewed in a long-term perspective. This potential for dynamic transformation of soil N compounds at low temperatures is consistent with earlier studies that have reported both significant accumulation and reductions in the concentrations of N species during the course of a winter [34]. Ryan et al. [34] conclude that the metabolic substrates following soil freezing do not originate from soil organic matter but from the fraction of the soil microbial biomass that is killed and lysed during this process. Numerous investigations have highlighted the importance of this phenomenon e.g. [3538], and killed micro-organisms (and other, ruptured cell structures) may also have provided sources of carbon or nitrogen substrates in our investigation. However, the fact that our frozen samples maintained net N mineralisation, nitrification and denitrification over several months indicates that soil organic matter is an important N source during periods with frozen conditions. The fact that no N accumulation was detected at 100% WHC further emphasises the importance of water content as a major factor controlling gaseous N losses at sub-zero temperatures. The high fraction of inorganic N present in the form of NO3 at temperatures above zero shows that the investigated soil has a high potential for nitrification. Although this potential decreased in frozen soil samples, it remained high enough to sustain the denitrifying community with NO3 over a period of several months. The extrapolation of our data using the function for suboptimal temperatures [15] suggested that −9.3 °C was the temperature limit at which net nitrogen mineralization could occur. However, it is conceivable that microbial activity in most soils would cease as temperature falls before this limit is reached, due to diminishing amounts of free water. Nevertheless, our results show that N mineralisation, nitrification and denitrification can occur simultaneously in frozen soils down to −4 °C. Thus, N2O formation in frozen soil has the potential to affect annual dynamics and budgets of N.

Oxygen is the most important controller of N2O emissions from terrestrial and aquatic environments, and any factor affecting the oxygen regime may have a stronger impact than the anticipated direct effect of the factor [11]. This reasoning is consistent both with our findings and the hypothesis mentioned earlier that the anoxic microsite distribution has a major influence on N2O formation in frozen soils. Based on the results of this investigation and observations reported in the literature [6,12,24] we propose the following conceptual model (Fig. 3) for the relationship between temperature and the formation of N2O in frozen soil. During freezing, the diffusion barrier increases with a concomitant increase in the number of anoxic microsites. This leads to higher rates of N2O formation than would be expected if temperature was the only factor involved. We cannot describe how the temperature will affect the production below −4 °C, but it is likely to drop, due to decreasing amounts of free water. This conceptual model for the N2O production response to low temperatures is similar to the established mechanism accounting for anomalous formation rates associated with higher temperatures [39], where the temperature enhances soil oxygen consumption and induces anoxic conditions that increase the production of N2O. While the high-temperature response is included in most mechanistic models designed to simulate N2O emissions from soils [39], the anomaly with respect to low temperature and soil freezing, as demonstrated by our results, is not. In fact, most models and global budget estimates assume that trace gas exchange stops when soil is covered by snow or soil temperatures drop to near, or below, 0 °C [40,41]. It should be mentioned that our data only support this conceptual model at soil moisture contents above 60% WHC (prior to freezing). However, humus soils in the boreal climate zone typically have high water content before soil frost development.


Conceptual illustration of the temperature effect on the production of N2O in soils. The dotted line includes only the direct temperature effects, while the solid line combines both the direct temperature response and increased anoxic conditions from either reduced diffusion due to ice development or from increased oxygen consumption due to increased heterotrophic activity at increasing temperature.

Although some investigations have shown that the N2O emissions from frozen soils may contribute as much as 70% of the annual atmospheric load [6], most studies on emissions from soil at low temperatures have concentrated on periods of soil thaw, when peaks in emission rates are typically observed [4,5,7,25,33,42,43]. Consequently, these events have been given high priority in research aimed at providing a scientific foundation for strategies for mitigating N2O emissions. However, the factors that most need to be identified in order to develop effective measures to reduce emissions are those affecting production, rather than those that influence emissions. Our results show that a large fraction of the N2O emitted from thawing soil may have been formed in frozen soil prior to the thawing. Thus, research aimed at resolving the factors controlling N2O formation in frozen soil is essential for generating satisfactory estimates of annual budgets and for the development of efficient mitigation strategies for reducing N2O emissions from terrestrial ecosystems.


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
  30. [30]
  31. [31]
  32. [32]
  33. [33]
  34. [34]
  35. [35]
  36. [36]
  37. [37]
  38. [38]
  39. [39]
  40. [40]
  41. [41]
  42. [42]
  43. [43]
View Abstract