Nitrogenase gene diversity and microbial community structure: a cross-system comparison

Major nifH clusters

Nitrogenase genes cluster into four basic groups, previously designated as Clusters I-IV (Chien and Zinder, 1996)(Fig. 1A). The composition of these major clusters consist of the ‘conventional’ Mo-containing nifH and some vnfH (Cluster I), the ‘second alternative’ non-Mo, non-V containing anfH as well as nitrogenases from some Archaea (Cluster II), nifH sequences from a diverse group of distantly related microorganisms many of which are strict anaerobes (e.g. clostridia, sulphate reducers; Cluster III), and a divergent loosely coherent group of nif-like sequences from Archaea and distantly related chlorophyllide reductase genes (Cluster IV). It is not clear whether the vnfH genes within Cluster I (Table 1) can be distinguished from conventional nifH. However, the vnfH genes of Azotobacter species cluster together (Cluster 1 N, Fig. 1), and as more vnfH genes are identified and sequenced it may be possible to resolve nifH from vnfH sequences.

Many microorganisms have multiple copies of nitrogenase genes or homologues of nitrogenase genes (Cluster IV). Clostridium pasteurianum was shown to have a nifH gene family (Wang et al., 1988), and has sequences in Clusters II and III (Table 1). Azotobacter was the first genus to be shown to have both alternative nitrogenases, vnfH and anfH (Bishop et al., 1985). A number of other examples of microorganisms with multiple copies of nitrogenase genes have been uncovered. Some cyanobacteria have multiple copies of nitrogenase, including a vnfH as well as a second distinct copy of nifH (Thiel, 1993; Thiel et al., 1995). Sulphate reducers contain multiple homologues of nifH (Braun et al., 1999) in which group in Clusters II and III. Archaea often have two nifH homologues. In some cases, the two copies are in Cluster IV and Cluster III, and in other cases, the homologues are in Cluster II and Cluster III (Table 1). The placement of the multiple copies in the different phylogenetic clusters might be a phylogenetic diagnostic that could be exploited for the development of probes and primers.

NifH phylogeny of cultivated diazotrophs

nifH gene sequences from cultivated microorganisms, and their cluster affiliation are shown in Table 1. The proteobacteria nifH sequences form a number of distinct clusters (1, 2) that correspond approximately, but not perfectly, with ribosomal RNA phylogeny (Table 1, Fig. 1B). The gamma and alpha proteobacterial clusters are generally well-defined (Fig. 1B, Table 1), although there are a few cultivated microorganisms that do not consistently cluster within the appropriate group.

Cultivation techniques, although biasing for viable, cultivable cells, have provided crucial landmarks to ground-truth sequences obtained by amplification (Steppe et al., 1996; Bagwell et al., 1998; Egener et al., 1999; Olson et al., 1999; Lilburn et al., 2001; Steppe and Paerl, 2002). For example, the nif sequences of spirochaetes isolated from termites were very similar to sequences obtained by direct amplification (Ohkuma et al., 1996; Lilburn et al., 2001). Isolates have also been useful for developing in situ hybridization assays for enumerating and visualizing N₂-fixing cells in situ (Hurek et al., 1997; Egener et al., 1998).

Archaeal nitrogenases have thus far only been found in methanogens, and though the presence of nitrogenase in other Archaeal lineages is likely, it has yet to be documented. Archaeal nitrogenases include homologues of both functional nitrogenases and nif-like genes of unknown function (Chien et al., 2000).

Methylotrophs of both types I and II have now been shown to have nifH and to be capable of N₂ fixation (Auman et al., 2001) (Table 1, Groups 1K, 1M, 1P). Interestingly, nifH sequences from strains isolated from the environment are very similar (>94% identical at the amino acid level) to sequences amplified from a number of habitats, implying that methylotrophs may be found to be important in N as well as carbon cycling (Auman et al., 2001).

Cyanobacteria are a morphologically and physiologically diverse group within the Bacteria. The cyanobacteria nifH genes cluster together (Zehr et al., 1997) (Fig. 1B, Cluster 1B) although sequences from some unicellular and filamentous non-heterocystous cyanobacteria form deep branches. Heterocyst-forming cyanobacterial nifH form a tight cluster within the cyanobacterial group. Some cyanobacteria have a nitrogenase that is expressed in vegetative cells of filamentous heterocyst-forming species (Thiel et al., 1995). Vanadium nitrogenases have also been reported in cyanobacteria and the Anabaena variabilis vnfH is found within the heterocyst-forming cyanobacterial nifH clade (Thiel, 1993). As most nitrogenases from cyanobacteria have not yet been characterized by mutant or deletion analysis, it is not yet possible to determine whether the vegetative, heterocyst or vnfH nitrogenases in cyanobacteria (Cluster 1B) can be resolved by nifH phylogeny. Thus, the cyanobacterial nifH cluster (Fig. 1B, cluster 1B) contains nifH genes expressed in vegetative cells and heterocysts, as well as the cyanobacterial vnfH genes.

The cyanobacteria gene cluster can intermingle with Frankia nifH sequences of the firmicutes (Hirsch et al., 1995), although as more sequences have been added to the database, the resolution of these groups has improved.

Paenibacillus species are Gram-positive bacteria which are typically found in rhizosphere zones. Analysis of nifH genes in Paenibacillus species with a nested PCR approach showed that these organisms also contain multiple copies, with one homologue in Cluster II (Table 1)(Rosado et al., 1998).

The sequences from a number of cultivated diazotrophs are in Cluster III. These sequences include those from Gram-positive microorganisms, delta proteobacteria, green sulphur bacteria and Archaea. The organisms represented by nifH within this cluster are mostly, if not all, strict anaerobes. The known organisms in this cluster are Clostridium, Desulfovibrio (and other sulphate reducing genera), Chlorobium, and the Archaea Methanosarcina barkeri. Although they cluster together, Cluster III is characterized by deep bifurcations and long branch lengths, and the distances between sequences are large relative to distances within Cluster I (Fig. 2). The clustering of these sequences is not as aberrant with ribosomal phylogeny as it might first appear, as phylogenetic trees based on partial 16S ribosomal sequences from these microorganisms groups them together as well (Fig. 3).

Sulphate reducers are known to be N₂ fixers and have nitrogenase genes (Kent et al., 1989) that are in Cluster III. Experimental data implicated them in N₂ fixation in estuarine sediments (Dicker and Smith, 1980). However, it is curious that sulphate reducers fix N₂ in many cases, because they are active in highly reduced sediments where decomposition processes release ammonium. Steppe and Paerl (2002) investigated the involvement of sulphate reducers in nitrogen fixation in marine sediments and provided evidence for nifH gene expression.

Cluster II and III nifH sequences from spirochaetes have been recently reported (Lilburn et al., 2001) (Table 1). nifH genes from termites guts that are closely related to spirochaete nifH genes (Lilburn et al., 2001) were shown to be transcribed using RT-PCR (Noda et al., 1999; Ohkuma et al., 1999).

Representative sequences for several major lineages are still not represented on the phylogenetic tree. For example, there are no nitrogenases from bacterial thermophiles (except from the green sulphur bacteria), yet it seems likely, given the dispersion of nifH through the prokaryotes, that they will be present in this group. Although it is possible that some of these lineages do not have N₂-fixing representatives, it is probably still true that we simply have not yet provided cultures with the right conditions to observe N₂-fixation.

Subclusters of nifH sequences from cultivated and uncultivated microorganisms

Within the four major clusters, sequences group within subclusters that have been tentatively defined on the basis of amino acid sequences for the purposes of this cross-system comparison (Table 1, 1, 2). The sequences from cultivated microorganisms help to classify sequence types obtained from the environment and evaluate the topology of the tree with respect to the evolution of nitrogenase. The conclusion of many surveys of diazotroph diversity is that assemblages are diverse, and that the phylotypes obtained directly from the environment are not closely related to sequences from previously characterized strains and environments.

Lateral gene transfer

Recently, particularly in light of genomic information, there has been discussion of the potential for lateral gene transfer among microorganisms (Doolittle, 1999; Nesbo et al., 2001). Dissimilatory sulphite reductase (DSR), the gene encoding the protein catalysing the last step in respiratory sulphate reduction, is one candidate for lateral gene transfer based on comparison of DSR and 16S rRNA phylogenies, and between DSR subunits (Klein et al., 2001). Nitrogenase is another functional gene that could easily be subject to lateral gene transfer, and in fact, exists on plasmids in some rhizobia (Barbour et al., 1992). However, most of the nitrogenase genes that have been characterized by genomic analysis are not plasmid-borne, despite the fact that diazotroph assemblages do contain plasmids (Beeson et al., 2002). It is clear that true evolutionary relationships cannot necessarily be inferred from either single gene trees or by integrating the information from multiple gene trees (Cavalier-Smith, 2002), but here we compare the phylogeny of nifH to that of 16S rRNA genes for cultivated microorganisms for which both sequences are available, in order to examine the evidence for lateral gene transfer of nifH.

The list of cultivated organisms for which there are nifH gene sequences in GenBank was used to extract aligned 16S rRNA sequences for the same or closely related representatives from the Ribosomal Database Project (Maidak et al., 2001). The aligned sequences were used to generate a phylogenetic tree using ARB, so that the ribosomal RNA phylogeny could be directly compared to nifH phylogeny.

Trees constructed for nifH and 16S genes based on the same lineages represented by both genes generate remarkably consistent trees (Fig. 3). It is difficult to assess transfers in Archaea, because Clusters II and IV include many of the Archaeal nifH genes, but are paralogous to the nifH genes in Clusters I and III. In fact, Cluster III could itself be a paralogue arising from a duplication of nifH early in evolution (Hirsch et al., 1995). Nonetheless, the topology between the two trees is largely consistent (Fig. 3). The gamma group including Vibrio and Klebsiella is congruent between trees, as is the cyanobacterial group. Frankia and Paenibacillus form distinct groups in both trees (Fig. 3), although Paenibacillus clusters with a larger group including Clostridium in the 16S rRNA tree, but does not on the basis of nifH.

Rhizobium, Mesorhizobium, Azospirillum Rhodobacter and Methylosinus cluster together on both trees as a group of alpha proteobacteria. Herbaspirillum is perhaps the best example of aberrant phylogeny, grouping with Alcaligenes and Azoarcus in the ribosomal RNA tree. The Cluster III sequences in the nifH tree include sequences from Clostridium, Chlorobium, some spirochaetes, and a number of sulphate respirers (e.g. Desulfonema and Desulfobacter). These sequences also form distinct clusters in the 16S tree when only sequences from N₂ fixing microorganisms are analysed (Fig. 3). Thus, although this cluster of sequences initially seemed to contradict rRNA phylogeny (Zehr et al., 1995), it is perhaps largely a result of the sequences included in phylogenetic analysis.

Genetic distances among sequences of major clusters are greater on the basis of nifH than 16S rRNA sequences (Table 2). The regions used for nifH PCR amplification (included in this analysis) include the more divergent regions of the nifH gene. The nitrogenase gene is likely not to be as conserved as rRNA, and this data shows the difference in rates of evolution between genes.

This analysis is based on a relatively short (∼350 nt), but phylogenetically informative, region of the nifH gene, and the resolution of the issue of lateral gene transfer will depend on more extensive genomic analyses, or possibly analyses based on the molecular sequences of the other nitrogenase structural genes. Nonetheless, there is not strong support or evidence for lateral gene transfer, by plasmid transfer or other mechanisms, and if such transfers occurred, they must have occurred early in evolution. Finally, we are aware of at least two probable gene duplications, giving rise to the alternative nitrogenases. It is possible that other duplications exist within the nifH phylogenetic tree, that could lead to false conclusions about phylogeny based on paralogous comparisons. NifD and nifK analyses may help in this regard, as well as genomic context information (Young, 1992).

Cross-system comparison of nifH diversity

The question of how diazotrophs are distributed across ecosystems and habitats can be investigated now that there is an extensive dataset of nifH genes. The diversity of nifH genes as a function of habitat is at the heart of biocomplexity, and encompasses the issue of how genetic redundancy relates to ecosystem function. This issue is also central to understanding how organisms and genes are selected in habitats, and how selection versus other factors determine the distribution and diversity of genotypes. Table 3 depicts the distribution of nifH genes obtained from different habitats within subclusters of nifH genes that contain cultivated representatives (Table 1). Table 4 shows phylogenetic clusters that are currently composed of only environmental sequences. The vertical distribution of shaded areas shows how subcluster phylotypes are distributed in relation to the phylogeny of nifH (1, 2).

Sequences have been obtained that group in Cluster I from many environments, consistent with the widespread distribution of proteobacteria (purple bacteria) and cyanobacteria. Cluster III is particularly interesting as multiple divergent sequences within Cluster III have been retrieved from environments that may be anaerobic at times or may contain anaerobic or microaerophilic microsites. These environments include estuaries, a hypersaline lake, sediments, microbial mats, termite guts and planktonic crustaceans (Table 3).

Interestingly, relatively few Cluster II alternative nitrogenase gene sequences have been obtained from the environment (Table 3). Primers directed at anfG, which is diagnostic for alternative nitrogenases, have demonstrated that alternative nitrogenase containing microorganisms are present in a few environments (Loveless and Bishop, 1999). Few sequences of this group have been obtained from amplification using nifH primers (Table 3), indicating their probable low relative abundance (on a gene level).

Very few sequences have been obtained from Cluster IV (Table 3). These sequences were amplified from termites, and could be homologues from spirochaetes or methanogens, both typical inhabitants of the termite gut.

It is clear that diazotroph diversity is not distributed equally among habitats, and that there are differences in distribution of specific types and the number of types across habitats (Table 3). There are many clusters for which we do not yet have cultivated isolates (Table 4), and some of these clusters are found in multiple environments (e.g. 1 A, 1G, 3H, 3I; Table 4). Some habitats have less nifH sequence diversity, although this may be partially due to discrepancies in the number of sequences obtained in each study. Marshes, mats, termites and the aquatic water column have received the most sequencing efforts. However, a number of distinct patterns emerge independent of the numbers of sequences obtained from a given environment.

Oligotrophic oceans

In oligotrophic aquatic habitats, nifH genes do not appear to be as abundant as in productive lakes or coastal waters or estuaries, or sediments. Interestingly, different groups of nifH genes have been obtained from the open ocean than from lakewaters, such as from Lake Michigan or Lake George (Tables 3 and 4). Relatively few different major groups have been found in the plankton (Tables 3 and 4), but interestingly new species of diazotrophs have been discovered on the basis of the nifH gene and transcripts (Zehr et al., 2001). Cyanobacteria nifH genes observed have included the well-known marine diazotroph Trichodesmium, but new groups of genes from unicellular cyanobacteria, some of which are now in culture (Zehr et al., 2001), have routinely been detected in Pacific Ocean waters at station ALOHA. Other groups include nifH genes from yet undescribed heterocyst-forming cyanobacteria and gamma proteobacteria.

Lakes, rivers and estuaries

N₂ fixing cyanobacteria species are often involved in algal blooms in lakes and rivers (Paerl, 1988). Although the amplification of nifH can be useful in identifying N₂ fixing species (Zehr and McReynolds, 1989; Olson et al., 1999; Steppe et al., 2001; Zehr et al., 2001), or identifying species expressing nitrogenase (Zani et al., 2000; 2001), it is rather short for a useful phylogenetic marker in closely related strains (Zehr et al., 1997; Tamas et al., 2000; Dyble et al., 2002). Nonetheless, Dyble et al. (2002) found that amplified nifH sequences from cyanobacterial isolates reflected their geographic origin. In depth profiles in Lake Michigan, cyanobacteria sequences were prominent, although sequences from bacterial groups were also found. Similarly both cyanobacterial and bacterial nifH genes were found in meso-oligotrophic Lake George, New York (Zani et al., 2000).

An interesting, contrasting aquatic environment is Mono Lake, a hypersaline, currently meromictic lake in the eastern Sierra Nevada mountains of Northern California (Jellison and Melack, 2001). This lake is believed to be N limited, although attempts to detect N₂ fixation in the water column have been unsuccessful. Diversity of diazotrophs is high, and there is a substantial difference between the communities of the mixolimnion and monimolimnion (G. Steward, unpub. data). Composition of these communities is substantially different than that found in other aquatic environments. For example, about two-thirds of the sequences obtained from Mono lake fall in Cluster III, a cluster that contains only two sequences from freshwater lakes and none from the open ocean. Several sequences group closely to Desulfovibrio sequences and one appears to be closely related to Spirochaetes, which is of interest because there were 16S sequences recovered from Mono Lake that group with Spirochaetes (Humayoun et al., 2003). Of the Cluster III sequences recovered, some cluster within groups that contain only sequences recovered from the environment (3M,N,O,Q), specifically estuaries, salt marshes, sea grasses and microbial mats.

Estuaries have diverse groups of microorganisms including those that are similar to the distribution in marshes and mats. Two estuarine habitats where approximately the same number of nifH sequences have been analysed are the Chesapeake Bay (B. Jenkins, unpub. data) and the Neuse River (Affourtit et al., 2001). Both of these environments contain large numbers of Cluster 1J and 1K alpha proteobacteria in regions where salinity is low. Both estuaries also contain other Cluster I beta and gamma proteobacteria sequences (1P,1U) and the Neuse River contains a small number of epsilon proteobacteria sequences (cluster 1F). Unlike the Chesapeake Bay, where no cyanobacterial sequences (IB) have been recovered to date, about one-quarter of the Neuse River nifH sequences are from cyanobacteria. Both the Chesapeake Bay and the Neuse River have sequences in Cluster III (E,J,K,L,P). Sequences from Chesapeake Bay and Neuse River sediments are also found in Cluster III, which suggests that some of these Cluster III sequences may be derived from sediment mixing (Affourtit et al., 2001; Burns et al., 2002).

Mats and sediments

Mats and stromatolites are representative of microbial communities on the early Earth (Des Marais and Walter, 1999). These complex assemblages include heterotrophic and phototrophic microorganisms, and diazotrophs are found in both physiological types (Zehr et al., 2001; Steppe and Paerl 2002). We have done extensive sequence analysis of nifH genes in mats in the hypersaline lagoons of Guerrero Negro (E. Omoregie and B. Bebout, unpubl. data). These sequences include Cluster I proteobacteria and cyanobacterial sequences as well as many sequences in Cluster III. Sequences from both Cluster I and III are expressed in the mats (E. Omoregie and B. Bebout, unpubl. data).

Sediment and salt marsh environments have diverse assemblages of nifH sequences, although there appear to be selection for different phylotypes between habitat types. Salt marsh ecosystems are important wetland environments at the land–sea interface, characterized by salt tolerant grass species. N₂ fixing microbial assemblages are diverse in these environments (Piceno and Lovell, 2000a,b), with a variety of phylogenetic and physiological types represented (Bagwell et al., 1998). nifH genes cluster in Cluster I alpha and gamma proteobacteria groups as well as in Cluster III (Lovell et al., 2001). Interestingly, diazotroph assemblages appear to be relatively stable even in the face of N enrichments (Piceno and Lovell, 2000b) and environmental variability (Bagwell and Lovell, 2000), as is N₂ fixation as assayed by acetylene reduction (Bagwell and Lovell, 2000; Piceno and Lovell, 2000a). Despite species stability, there are differences in species assemblages for different grasses (Bergholz et al., 2001), above-ground versus below-ground (Lovell et al., 2001) and for dead salt marsh wrack (Lovell et al., 2000).

Soils

Interestingly, soils have fewer phylogenetic clusters represented than some aquatic environments (Tables 3 and 4). Using DGGE, it was found that the composition of the diazotroph community varied as a function of soil type on a range of scales (Poly et al., 2001b). Phylotypes recognized by RFLP of amplified fragments indicated that diazotroph species in forest litter shift in clearcut communities (Shaffer et al., 2000).

Rhizosphere soils contain bacteria of a number of anaerobic and aerobic bacteria that have been found in other environments including rice root rhizospheres and aquatic environments (Hamelin et al., 2002). Azoarcus is an interesting genus composed of disparate lineages, some of which cluster in different proteobacterial clusters on the basis of nifH (Hurek et al., 1997). Azoarcus sp. BH72 was shown to express nitrogenase in rice root rhizospheres, using a green fluorescent protein reporter (Egener et al., 1998) and also inside rice roots (Egener et al., 1999). Analysis of expression of nifH phylotypes in conjunction with natural abundance of ¹⁵N showed that Azoarcus in Kallar grass (a productive grass which requires little N fertilization) was contributing to N₂ fixation.

Invertebrate-associated

Termite guts represent an extremely N-limited environment, as some termites can feed solely on cellulose and must obtain N elsewhere (Ohkuma et al., 1999). Diverse nifH genes were recovered from termite guts using nifH PCR primers, including sequences that could be derived from methanogens (Ohkuma et al., 1996). Sequences were recovered from Clusters II, III and IV. Undoubtedly some of these sequences represent multiple copies, because in many organisms, there are nifH homologues in groups IV and III, or in II and III such as in Methanosarcina (Chien and Zinder, 1996). Lilburn et al., (2001) isolated spirochaetes of the genera Treponema and Spirochaeta from termites, showed that some of the isolates were capable of N₂ fixation, and found that the nifH genes clustered with previously reported nifH sequences from termite guts, as well as marine copepods.

It was shown that second alternative nitrogenases were expressed in termite guts on the basis of RT-PCR. Unexpectedly, the anf genes were not regulated by Mo as anticipated, as Mo usually represses the transcription of alternative nitrogenases (Noda et al., 1999). Using a primer pair that bracketed nifH and nifK, PCR showed that the alternative nitrogenase in the termite guts contained anfH, two ORFs, and anfDGK, an arrangement consistent with Archaeal Cluster II nitrogenases. Nitrogenases were also found that are very similar to sequences found in other environments that cluster within Cluster I gamma proteobacteria sequences.

Factors limiting nitrogenase expression and environmental selection

N₂ is a potential source of N that should ultimately alleviate N limitation in the environment, unless other factors constrain N₂ fixation (Redfield, 1958; Howarth and Marino, 1988; Vitousek and Howarth, 1991; Vitousek, 1999). It is usually assumed that genes are ultimately not retained by microorganisms, unless they are functional and thus, are selected for in the environment. In the case of nitrogenase, this would seem to be particularly true because N₂ fixation can involve around 20 genes, even in free-living microorganisms (Fig. 4). It would seem likely that the presence and diversity of nitrogenase genes might reflect selection due to the extent of N limitation in the environment. One of the interesting patterns seen in Table 3 is that the diversity of nitrogenase genes is not directly linked to the degree of N limitation. The open ocean studies were performed in oligotrophic subtropical gyres where there are almost undetectable levels of fixed inorganic N (Karl and Michaels, 1996), and yet the diversity of nifH subclusters is less than in estuarine habitats such as the Chesapeake Bay and the Neuse River. Although the Neuse River is N limited in the outer reaches, upstream regions receive high N loading from agricultural activities (Steppe et al., 2001). These data indicate that gene distributions in natural assemblages are a function of not only environmental selection but also physical-chemical factors including physical transport of cells between and among habitats. Microbial assemblages in estuaries may reflect an important allochthonous component, which has implications for the structure and function of planktonic microbial communities. Clearly presence of a gene in a habitat, be it nitrogenase or other ‘functional’ genes, must be interpreted cautiously, and may not be directly linked to the process catalysed by the expressed protein. Nonetheless, the presence of functional genes can provide markers for the movement and composition of microbial communities.

Analysis of gene expression has indicated that not all nif genes that are present are expressed in the environment (Noda et al., 1999; Zani et al., 2000), at least not at the same time. Although the regulatory cascade controlling nitrogenase gene expression is relatively well known in some enteric bacteria, little is known of the regulatory mechanisms in most prokaryotes. Interactions between microorganisms, such as the supply of photosynthate from autotrophs to heterotrophic bacteria as well as endogenous rhythms can drive the patterns of nif gene expression in the environment. Although it is difficult to work with mRNA from many natural sample matrices, much more work is needed in the area of gene regulation in the environment.

Genomics

The use of nitrogenase gene sequences to describe diazotroph diversity is somewhat limited by the use of nifH, as it is very highly conserved and the use of one gene for phylogenetic analysis can be misleading. A large number of microbial genomes have now been sequenced, although there are still relatively few different major branches of the prokaryotes represented by genomes that contain the structural nif genes (nifHDK) in the microbial genome database. Nonetheless, it is clear that the genomic context provides information for studying phylogenetic relationships based on gene organization rather than molecular sequence of individual genes, as well as studying lateral gene transfer (Fig. 4). Furthermore, the different nitrogenase homologues can be differentiated on the basis nif gene arrangements (Dominic et al., 2000) and by the presence and absence of specific nif and related genes in the genome (Fig. 4). In methanogenic Archaea, the nifH and nifDK genes are separated by glnB homologues, the gene for the regulatory protein PII (Fig. 4). These differences provide targets for directed molecular assays. It will be interesting to see if the gene arrangements from cultivated microorganisms are representative of the arrangements found in natural assemblages. It is likely that many surprises are in store when the genomic context of nif genes is investigated in the environment.