1. Introduction

The relationship between the concentration of greenhouse gases in the atmosphere and the global climate is a current topic of scientific research and cause for alarm (Cairns 2010; Moriarty and Honnery 2010). Since the start of the industrial era around 1750, the concentrations of these gases have increased dramatically and the role of the anthropogenic carbon dioxide emissions has been widely discussed. However, the human impact on the carbon cycle is still relatively small compared to our contribution to nitrogen cycling (Figure 1). To feed the growing human population we have become completely dependent on the use of synthetic fertilizers; nowadays approximately one out of three nitrogen atoms that enter the biosphere originate from the fertilizer industry. Combined with ammonia deposition caused by fossil fuel burning, the anthropogenic input may even be higher than the natural background (Figure 1). The consequences for the climate are poorly understood but we argue that fertilization most likely reinforces global warming by increasing the atmospheric concentrations of methane and nitrous oxide, two powerful greenhouse gases. To understand this possible reinforcement, and make predictions for the future, it is essential to experimentally unravel the complex interactions of the biogeochemical nitrogen network. What follows is a step by step review of these interactions, and finally, the open questions and consequences for the climate are discussed.

2. The Nitrogen Cycle

The nitrogen cycle consists of transport processes and chemical reactions; the latter are mainly catalyzed by bacteria. Except for nitrogen fixation, bacteria perform these reactions to gain energy for chemotrophic growth. The phrase "nitrogen cycle" is generally used, but together the reactions actually form a more complicated "nitrogen network" (Figure 2). To gain insight into this network it is essential to combine two complementary experimental approaches: The first approach, ¹⁵N labeling, can provide information about the rates of the individual processes.

Although other approaches have been used for this purpose in the past (for example use of inhibitors such as acetylene) ¹⁵N labeling is the only one that is still useful in the context of the full complexity of Figure 2. The second complementary and independent approach, molecular ecology, provides information about the presence and activity of the associated bacteria and genes. Because many processes are performed by unrelated bacteria, we are generally dependent on the detection and quantification of functional genes (Figure 3). Technical progress along these lines, for example the application of novel tracer technology and continued mining for functional gene markers, is extremely important.

Nitrogen Fixation and Primary Production

All life depends on atomic nitrogen because it is an essential component of amino acids, nucleic acids, porphyrins, amino sugars, etc. Dinitrogen gas in the atmosphere is the largest reservoir of nitrogen available to life on Earth, and this may even be a consequence of life itself (Capone et al. 2006). Dinitrogen is accessed by microorganisms in a reaction known as nitrogen fixation. In this reaction dinitrogen is reduced to ammonia (NH₃) by the enzyme nitrogenase. Although the reaction is exergonic, the activation energy is very high and it requires 16 molecules of ATP per N₂ fixed. It is especially difficult to fix nitrogen in the presence of oxygen (O₂) because the nitrogenase is destroyed by oxygen. Therefore, during most of the geological history of the Earth the difficulty of accessing atmospheric nitrogen has been one of the factors that constrained primary production.

Only a small proportion of known bacterial species is able to fix nitrogen and possesses the structural genes for nitrogenase (nifDHK). Environmentally important nitrogen fixing organisms are plant symbionts such as Rhizobium and free living organisms such as the cyanobacterium Trichodesmium. Nitrogenase is phylogenetically widespread, e.g. many evolutionary unrelated species have acquired the genes for nitrogenase, although there is little evidence for recent lateral gene transfer (Zehr et al 2003). For this reason, the gene nifH is used as a functional marker to identify nitrogen fixing bacteria in nature, independent of organismal phylogeny as defined by the 16S ribosomal gene. For nitrogen fixation, tracer studies with ¹⁵N labelled N₂ and molecular ecology (targeting NifH) are established methods. Despite the availability of these methods, oceanic nitrogen budgets indicate that we may so far have overlooked key nitrogen fixers in the oceans (Codispoti et al. 2001). Technically, nitrogen fixation is carried out by the Haber Bosch process - essentially the same chemical reaction as nitrogen fixation. The widespread use of fertilizers in agriculture leads to high (hundreds of micromolars) concentrations of nitrate (NO^3-) in many freshwater and coastal surface waters (Mulholland et al., 2008). Here, nitrate has replaced dinitrogen as the main source of nitrogen sustaining growth of bacteria and plants. This is leading to eutrophication, loss of biodiversity and higher rates of primary production (fixation of atmospheric carbon dioxide).

It is unlikely that fertilization actually drives a net removal of carbon dioxide from the atmosphere, because surface waters contain no long term sink for carbon dioxide. The extra biomass that is produced is rapidly consumed and recycled - to carbon dioxide. Long term studies addressing this issue have even reported additional release of carbon dioxide by enhanced mineralization of refractile organic matter in the presence of nitrate (Mack et al. 2004). The only long term biological sink for atmospheric carbon dioxide is the biological pump – which requires sinking of biomass into the deep ocean and storage of the carbon in deep marine sediments.

Therefore, ammonia deposition to the open ocean is more important as a possible negative feedback on global warming than fertilization. However, even in the open ocean the beneficial climate effects caused by increased primary production is likely to be neutralized by increases in nitrous oxide production (Duce et al 2008).

In surface waters it is more likely that fertilization reinforces global warming by stimulating biological methane production (and nitrous oxide production, see below). Increased primary production leads to more buildup of biomass in shallow sediments where most of the biomass is degraded anaerobically, leading to the production of methane. Because shallow sediments are already a major source of methane emissions to the atmosphere (e.g. wetlands, rice fields) it is likely that a part of the extra carbon dioxide removed from the atmosphere by fertilization is returned as methane. As a greenhouse gas, methane is much stronger than carbon dioxide.

The nitrogen incorporated into biomass by primary producers enters the biological food chain. At each trophic level of the food chain most of the biomass is used as an energy source; sugars, proteins and lipids are mainly oxidized to carbon dioxide and only a small part is used for growth. Therefore, most of the nitrogen in the biomass is set free as ammonia. This ammonia release is known as ammonification. Presumably, many organisms are involved in this process. It is rarely investigated experimentally. Rates of ammonification are generally inferred from Redfield stoichiometry.

Nitrification

Ammonia can be oxidized to nitrate with oxygen. This aerobic process is known as nitrification and consists of two steps performed by two different groups of chemolithoautotrophs: the ammonium-oxidizers oxidize ammonia to nitrite and the nitrite-oxidizers oxidize nitrite to nitrate. Environmentally important ammoniaoxidizers are affiliated with beta- (e.g. Nitrosomonas europaea) and Gammaproteobacteria (e.g. Nitrosococcus oceani) and Crenarchaea (e.g. Nitrosopumilus maritimus, Könnecke et al. 2005). Presently, five different unrelated groups of nitrite oxidizers (Figure 3) are known, all affiliated to different bacterial phyla.

Biochemically, ammonia is activated by the enzyme complex ammonia monooxygenase. The genes encoding this complex are amoABC and they serve as functional genetic markers to assess the diversity and abundance of ammonia oxidizers. Hydroxylamine is oxidized to nitrite by the octaheme enzyme hydroxylamine oxidoreductase, encoded by the gene hao (Klotz et al. 2008). However, this gene is not present in the genome of the crenarchaeal ammonia oxidizers (Walker et al. 2010). Nitrite oxidation is catalyzed by the enzyme complex nitrite:nitrate oxidoreductase, a member of the molybdopterin oxidoreductase superfamily. The genes are known as nxrAB but they are homologous to the genes used by denitrifiers to perform the reverse reaction (e.g. narGH see below).

Fertilizer is generally applied in the form of ammonium. Ammonium is positively charged and binds to (negatively charged) clay. After the nitrifiers convert ammonium to nitrate (negatively charged) it desorbs from the clay and is readily transported to surface or groundwater. For this reason, nitrification inhibitors are frequently added to the fertilizer mixture but these are only partially effective (Welte, 1994).

Nitrification is also a major source of nitrous oxide (N₂O) emissions. Although nitrous oxide is not an intermediate of nitrification, it is still produced by ammonia oxidizers, either as a by-product of the hydroxylamine oxidoreductase or by reduction of the produced nitrite ("nitrifier-denitrification") by denitrification enzymes expressed by ammonia oxidizers at low oxygen levels (Meyer et al., 2008). Both the application of ¹⁵N tracers and molecular gene markers are established for nitrification (e.g. Lam et al., 2009).

Denitrification

The first process that recycles nitrate back to dinitrogen gas is known as denitrification or "nitrate respiration". The latter term results from the fact that nitrate instead of oxygen serves as the final electron acceptor. During denitrification, nitrate (NO^3-) is reduced via nitrite (NO^2-), nitric oxide (NO), and nitrous oxide (N₂O) to dinitrogen (N²). Denitrification generally occurs in the absence of oxygen but may sometimes proceed even when oxygen is present (Gao et al 2009).

Denitrification is carried out by many different unrelated species. The "model denitrifiers" are affiliated with Proteobacteria (e.g. Pseudomonas) but environmentally important denitrifiers have yet to be found. The first step, the reduction of nitrate to nitrite, is catalyzed by the enzyme complex nitrate reductase encoded by narGH or napAB. The next step, the reduction of nitrite to nitric oxide, is catalyzed by the enzyme nitrite reductase. Nitrite reductase occurs in two forms, a multicopper oxidase type enzyme encoded by nirK or aniA and a heme cd enyzme encoded by nirS. The reduction of two nitric oxide molecules to nitrous oxide is catalyzed by Nitric oxide reductase encoded by norB or norZ. This enzyme is part of the heme/copper family of oxygen reductases. Nitrous oxide is reduced to dinitrogen by the copper enyzme nitrous oxide reductase, encoded by nosZ. nirK, nirS and nosZ are most often used as functional gene markers (Jones et al., 2008). However whole genome sequencing is providing evidence that with the primers currently in use (e.g. Baker et al., 1997) a substantial portion of these functional genes may be overlooked in the environment.

Recently, it was shown that denitrification may also proceed via a different pathway. In this pathway, performed by the bacterium Candidatus "Methylomirabilis oxyfera", nitrate is first reduced to nitric oxide as described above. Subsequently, two molecules of nitric oxide (NO) are dismutated into dinitrogen (N₂) and oxygen (O₂) (Ettwig et al., 2010). Oxygen can then be respired aerobically or used in a monooxygenation reaction to activate hydrocarbons such as methane. It is unknown which enzymes and genes are responsible for this dismutation reaction. It is also unknown how important this reaction is in nature. With all present tracer methods it cannot be distinguished from "normal" denitrification. The interesting implication of the discovery of this pathway is that methane emissions resulting from fertilization may be restrained this way.

Together with nitrification, denitrification is an important source of nitrous oxide emissions to the atmosphere. Apparently, under some conditions denitrification is incomplete and nitrous oxide is not further reduced to dinitrogen. The chemical or biological conditions that affect nitrous oxide production are actively researched but no clear causal relationships have become apparent so far.

Anammox

The second process that recycles nitrate back to dinitrogen gas is known as anaerobic ammonium oxidation (anammox), a relatively recent discovery (Mulder et al. 1995). In this process, ammonia and nitrite are combined into dinitrogen. As far as we know, anammox is performed by one monophyletic group of bacteria associated with the phylum planctomycetes (Strous et al., 1999).

The pathway presumably proceeds via nitric oxide and hydrazine (N₂H₄) and is inhibited by oxygen. Functional gene markers are currently being established (Strous et al. 2006). For example, hydrazine is presumably oxidized by a homologue of hydroxylamine oxidoreductase encoded by hzo (Schmid et al. 2008). A general property of anammox bacteria is that they can also reduce nitrate to ammonia (dissimilatory nitrate reduction, see below). For this reason it can be difficult in tracer studies to discriminate their overall activity from denitrification (Kartal et al. 2007).

However, because anammox is performed by only a single group of bacteria and these bacteria have a unique lipid biomarker in the form of ladderanes (Sinninghe Damste et al. 2002). Anammox bacteria can be detected in the environment with relative ease targeting both the 16S gene and the ladderanes. Presently, it is estimated that in the marine environment approximately 50% of the dinitrogen is produced by the anammox bacteria of the genus Scalindua (e.g. Lam et al 2009; but see Ward et al 2010).

Dissimilatory Nitrate Reduction to Ammonia

Where denitrification and anammox close the nitrogen cycle by recycling nitrate to dinitrogen, dissimilatory nitrate reduction closes the cycle by recycling nitrate to ammonia. In contrast to denitrification and anammox, this process does not remove the nitrogen from the habitat – it remains available to primary producers. Like denitrification and anammox, it is a form of anaerobic respiration, where nitrate is used as electron acceptor instead of oxygen. Many different unrelated bacteria are capable of this process, but proteobacteria have been most extensively studied; the best known dissimilatory nitrate reducers are Escherichia coli and giant sulfur bacteria such as Thioploca (Otte et al. 1999).

The first step of the pathway is shared with denitrification, the reduction of nitrate to nitrite by a molybdopterin enzyme complex. Next the five-electron reduction of nitrite to ammonia is performed by pentaheme nitrite reductase encoded by nrfAB. Recently, it was found that octaheme enzymes evolutionary related to hydroxylamine oxidoreductase (see nitrification above) are reversible and can also reduce nitrite to ammonia (Atkinson et al., 2007; Klotz et al., 2008). Thus, a functional gene marker for this process is still work in progress. Dissimilatory nitrate reduction rates can be measured in natural ecosystems by tracer studies, but this is rarely performed and the same is true for the use of functional gene markers (e.g. Lam et al. 2009; Dong et al., 2009). Therefore, it is unknown how important this process is compared to denitrification. It is also unknown how much this process contributes to nitrous oxide emissions.

3. Conclusion and Discussion

Humanity mainly impacts the nitrogen cycle by agricultural fertilization and fossil fuel burning (resulting in ammonia deposition). Together these anthropogenic inputs are estimated to be more important than natural nitrogen fixation (Figure 1). From field and budget studies the current effects of these emissions can be estimated: It is possible that oceanic ammonia deposition has minor negative feedback on the atmospheric carbon dioxide concentration. However, the positive feedback in the form of increased emissions of nitrous oxide and methane, as well as increased mobilization of stored terrestrial carbon is more important. It can be calculated that the total anthropogenic nitrogen inputs currently contribute 5-10% of the current enhanced greenhouse effect, with methane and nitrous oxide contributing approximately equally. It is difficult to estimate the contribution of enhanced mobilization of refractile organic carbon in soils.

It is impossible to draft scenarios for future trends because our current understanding of the microbial nitrogen network is far from complete: For many processes environmentally significant microorganisms and genes are simply unknown (Figures 2 and 3). It is also unknown how the environmental conditions affect the interplay between the different processes and the outcome in the form of changes in primary productivity, and emission of methane and nitrous oxide. These unknowns can only be addressed by disentangling the different branches of the nitrogen network and by identifying the missing microbial players. To do just that, the combination and continued development of ¹⁵N labeling approaches and molecular ecology is essential.

Acknowledgements: Anna Hanke and Marc Strous are supported by the European Research Council (ERC) Starting Grant “MASEM”.