The C₂H₂ reduction assay (based on the reduction C₂H₂ to C₂H₄ by nitrogenase) is a rapid, sensitive, simple and low cost method which if used with appropriate controls and calibrations can be useful for evaluating nitrogenase activity in time and space [49]. Under controlled conditions it can be extremely useful for comparative purposes where absolute values of N₂ fixation are not critical. Hardy et al. [50] found a direct correlation between N₂ fixation (N₂→2NH₃) and C₂H₂→C₂H₄ in pure cultures of diazotrophs and in legumes, and calculated that the theoretical relationship of C₂H₂ reduced to N₂ fixed was 3. However, measured values of conversion factors in different environments can vary widely; these are summarized in Table 1.

Such variations in ratios of C₂H₄ produced to N₂ fixed have been the basis of most criticism of the acetylene reduction assay, although research has shown that for all systems including pure cultures, legumes, non-legumes and soils that this ratio averaged between 2.6 and 6.9 [51]. The only exception to this was anaerobic soil which had conversion factors of up to 25. Under other conditions a reasonable estimate is possible, although any experimental procedure should always include a calibration of the assay using ¹⁵N₂ gas exposure [52-54].

This method can be sensitive, accurate and provide absolute proof of N₂ fixation and has been used to demonstrate N₂ fixation associated with cereals and grasses [41, 55, 56] and in soils [19, 56, 57]. It is a most useful method for calibrating other measures of N₂ fixation. Although difficulties in controlling environmental conditions can be encountered when using the method to measure N₂ fixation associated with plants, it has been successfully used to estimate N₂ fixation with cereals and sorghum [55, 56].

Demonstrating the incorporation of ¹⁵N₂ into free-living microbial populations in the soil is much more difficult. While N₂ fixation in soils was confirmed both in the laboratory and in the field by incorporation of ¹⁵N₂ [19, 41], absolute measures of N₂ fixation in the field using ¹⁵N₂ gas are costly and are influenced by seasonal factors (M. M. Roper, G. L. Turner, F. J. Bergersen, unpublished data; [58]). Many free-living, diazotrophic bacteria require reduced oxygen concentrations to fix N₂ and are located within microsites of low oxygen tension. Sites that restrict oxygen availability may also limit access by ¹⁵N₂ and this could result in an underestimation of N₂ fixation.

Both ¹⁵N isotope dilution and natural abundance methods depend upon differences in isotopic composition of the sources of N used for plant growth, i.e. atmospheric N, soil N and fertiliser N. Both methods require a non-N₂-fixing reference plant and therefore it is essential that the reference plant and the test plant with associative N₂ fixation have a similar root architecture and can extract N from the soil at the same rate in space and time [59]. Neither ¹⁵N method is suitable for quantification of the total amount of N₂ fixation by free-living bacteria because of the difficulty of separating N₂-fixing microorganisms from the soil for ¹⁵N analysis.

The ¹⁵N isotope dilution technique involves supplying a ¹⁵N enriched (or depleted) source of N to the soil so that it is significantly different from the natural abundance of the atmospheric N₂ [59, 60]. For accurate measurement, the spatial and temporal availability of the isotope should be uniform [60]. The ¹⁵N isotope dilution method has been used to estimate N₂ fixation associated with sugar cane, forage grasses, cereals and actinorhizal plants grown in soil, mostly in tropical systems.

The natural abundance (δ¹⁵N) method is exactly analogous to the isotope dilution method except that endogenous ¹⁵N in the soil is used [61, 62]. The natural abundance method has an advantage over isotope enrichment methods in natural ecosystems because disturbance of the system is unnecessary [62], but other factors can affect δ¹⁵N in plants, such as N from precipitation (NO_x, NH₃), the depths in the soil from which N is taken up and the form of soil N that is used (organic N, NH₄⁺ or NO₃^-) [63]. The ability of the natural abundance method to measure associative N₂ fixation depends on N₂ fixed by associative microorganisms being predominantly taken up by the plant rather than going into the soil N pool [62]. Application of the ¹⁵N natural abundance technique in oil palms in the field in Brazil identified diazotrophs with a high potential for N₂ fixation, but estimates of N₂ fixation could not be calculated because of the absence of a suitable reference plant [64] highlighting a significant limitation of this method.

Giller and Merckx [23] suggested that the ultimate test of the contribution of N from fixation is to measure net inputs of N over long periods (>10 years) in the field, i.e. an N budget. However, this may be difficult as it requires measuring all inputs and outputs of N over this period, including inputs from fertilisers, wet N deposition, dry N deposition, run-on and uptake from lateral flow, outputs from crop/animal removal, gaseous losses, N leaching and soil erosion. A number of studies using long-term field experimental data have shown considerable N gains which were attributed to inputs from NS N₂ fixation (for example [18, 22, 65-67]).

In the Rothamsted Broadbalk experiment, during the period from 1852-1967, Jenkinson [68] calculated that inputs from N₂ fixation were between 18-28 kg N ha^-1 year^-1 in a plot that received no fertiliser, and 23-35 kg N ha^-1 year^-1 in a plot that received inorganic fertiliser without N. These estimates were obtained after adjusting for wet deposition from rainfall (5 kg N ha^-1 year^-1), dry N deposition (10 kg N ha^-1 year^-1) and for inputs in the seed (3 kg N ha^-1 year^-1) all of which were measured at some time during the course of the experiment [68]. Estimates indicate N deposition in rainfall and dust may range between 3-5 kg N ha^-1 year^-1 over much of Africa and Australia, and 10-50 kg N ha^-1 year^-1 in more densely populated areas such as in Europe [23, 69].

To achieve a reliable N balance it is necessary to have a very high repeatability and accuracy of N measurements through strict sampling protocols and extremely high sample numbers to enable the mean soil N to be precise enough to determine statistically significant changes in soil N [70, 71]. From a scientific viewpoint, N balance studies only give an indirect measure of N gains due to NS N₂ fixation which can be useful for supporting other more direct measures. However, from a grower’s perspective, statistically significant measures of N accumulation provide valuable information for planning N fertiliser inputs for a crop.

Used in conjunction with other measures such as the C₂H₂ reduction assay, N budgets may increase the certainty of estimates. For example, Shearman et al. [72] found that in grass pastures inoculated with a range of known N₂-fixing bacteria, rates of C₂H₂ reduction were strongly correlated (r=0.92) with N accumulation measured by the Kjeldahl method. ¹⁵N aided N balance studies have been used to strengthen evidence for associative biological N₂ fixation in sugar cane [65, 73, 74], where there was good agreement between estimates of biological N₂ fixation from N balance and isotope dilution. However, the authors were careful not to assume the same rates of fixation occurred in the field because the conditions of the experiment differed from those in the field [73].

Much of the information on estimates of N₂ fixation using techniques that are currently available apply to one instant in space and time or over a short period of assay [58]. However, knowledge of the conditions that favour N₂ fixation and the rates at which fixation responds to changes in environmental conditions can be used to obtain estimates for a wider region if environmental conditions in those regions are known (e.g. meteorological records; cropping statistics and soil maps). Gupta et al. [22] used this principle to derive estimates for parts of the southern agroecological zones of Australia. Using information from other studies on the effects of different soil moistures, temperatures and carbon sources, potential N₂ fixation in different zones was determined using a spatial analytical tool (ArcviewGIS Spatial Analyst, v3.1). Use of this principle with a range of measurement strategies may provide useful information about regions that are most likely to benefit from NS N₂ fixation and where new advances can be made.

New techniques using microarray technologies have the potential to simultaneously measure the dynamics and/or activities of most microbial populations in the complex soil environment [75, 76]. Zhang et al. [37], used a nifH-based short oligonucleotide microarray and showed that this technique allows quantification and mapping of the abundance, diversity and activities of N₂-fixing populations. This approach is likely to assist in the identification of regions and managements that favour inputs of N from NS N₂ fixation. With the availability of complete genomes for several diazotrophic rhizobacteria and our increased ability to conduct in-depth genomic and functional analysis of candidate genes, we can now interrogate the specific features of diazotrophic endophytes. Such information on the molecular mechanisms of NS N₂ fixation should enable the development of agronomic options to improve N₂ fixation in non-leguminous crops [77].

Edaphic, environmental and management factors have a significant impact on the composition of diazotrophic communities and potentially their function [41,78, 79]. Varietal-based differences in the diversity of diazotrophic communities have been reported for wheat, barley, rice, sorghum [80-82] and perennial grasses [41]. Differences in soil environments at the micro-scale can also influence the composition of diazotrophic communities [83].

Nitrogenase proteins are extremely sensitive to O₂ and on exposure to air they are rapidly and irreversibly inactivated [84]. Therefore, NS N₂-fixing bacteria need mechanisms to exclude O₂ for N₂ fixation (nitrogenase activity) to occur. Some bacteria, e.g. Azotobacter, Azomonas, Beijerinckia and Derxia exclude O₂ through rapid respiration or the formation of extracellular polysaccharide [29, 85]. However, most culturable diazotrophic bacteria will only fix N₂ under microaerophilic or anaerobic conditions [29]. Anaerobic conditions can be created by saturating soil moistures and substantial amounts of N₂ fixation have been measured under these conditions [86]. In aerated soils, aggregate formation in the soil allows microaerophilic and anaerobic conditions to coexist simultaneously under aerobic conditions. Furthermore, substrates such as dissolved organic C can be allocated into both aerobic and anaerobic fractions and processes [87] and so, it is possible for soluble products from organic matter decomposed under aerobic conditions, to supply C energy to microaerophilic and anaerobic N₂-fixing bacteria within aggregates.

The second major condition required for NS N₂ fixation is the availability of C as an energy source. Free-living N₂-fixing bacteria generally rely on decomposing plant material above and below ground from crops and pastures. Associative N₂-fixing bacteria utilise root exudates within a rhizosphere association with plants and other organisms. In both environments, other microbial groups compete for limited energy resources. Endophytic N₂-fixing bacteria, on the other hand, have ready access to C and nutrients from within the plant [46].

Crop residues contain cellulose and hemicellulose which comprise 50-70% of its dry weight [88]. A few species of N₂-fixing bacteria (Azospirillum spp.) are able to use straw directly for fixation [89], but most N₂-fixing bacteria rely on decomposition to smaller components by other organisms [90]. Almost all N₂-fixing heterotrophic bacteria are able to utilise the products of cellulose decomposition including carbohydrates and some organic acids and alcohols [91, 92]. Rates of N₂ fixation are proportional to the amount of crop residue available and to rates of decomposition [19]. The retention of crop residues can alter the composition of diazotrophic community structure, increase nifH gene abundance and N₂ fixation [82, 93, 94]. Root exudates, e.g. carbon containing compounds and quorum-sensing compounds, have been shown to influence the composition and function of nifH communities in the rhizosphere, and N₂ fixation rates of 4-20 kg N ha^-1 year^-1 were predicted by Jones et al. [95]. Gupta et al. [41] observed a plant-based selection of nifH communities in the root environments of different summer-active perennial grass species. They found that diversity of diazotrophic bacteria was significantly higher in the rhizosphere than in the roots and that both the rhizosphere and roots supported higher N₂ fixation than in cropping soils during summer.

It is well known that inorganic mineral N in soil can inhibit N₂ fixation by NS microorganisms [96]. However, the dynamics of N₂-fixing microbial populations are linked to available C:N ratios. For example, when C is abundant, excess ammonium N can be assimilated by other microbial populations allowing N₂ fixation to occur, but with low C, excess ammonium N concentrations inhibit N₂-fixing populations [97]. In the presence of large amounts of crop residue with wide C:N ratios, decomposition can be slow. But the addition of N increases the rate of decomposition (making C available for use by N₂-fixing bacteria [30, 90]. Other mineral nutrients may influence NS N₂ fixation. Mo and Fe are components of the nitrogenase enzyme, but they are rarely limiting in natural environments [98]. On the other hand, applications of P can significantly increase NS N₂ fixation in crops [99, 100] and in grasslands [101, 102] particularly in nutrient poor soils. Reed et al. [99] and Smith [100] concluded that the strong inverse relationship between N₂ fixation and mineral N content in the soil is mitigated by the availability of P.

High soil water contents have been used to promote N₂ fixation in soils by reducing O₂ at the sites of fixation [86]. However, in unsaturated soils, it is important to maintain aggregate structure and O₂ gradients. In disturbed soils in the laboratory, a minimum of 50% water holding capacity was required for nitrogenase activity [103], whereas in in situ assays in undisturbed soils in the field, nitrogenase activity occurred at soil water contents below 30% water holding capacity [19]. In some environments, N₂-fixing bacteria have adapted to harsh semi-arid environments e.g. lichens (containing cyanobacteria or other free-living bacteria in association with a fungus) [104, 105] and rhizosheaths around the roots of perennial grass species, and can contribute significant amounts of biologically fixed N [106, 107]. Rhizosheaths support enriched organic materials, greater water contents and a higher density of microorganisms including associative diazotrophs [106] and up to 9 kg N ha^-1 year^-1 has been measured [108].

N₂ fixation has been shown to occur in situ in temperature extremes from near 0^oC in Antarctica [109, 110] and in the Arctic [111] to desert environments where N₂-fixing bacteria utilise morning dew or summer rains [105] but must survive during intervening hot dry conditions up to 60^oC [98]. Laboratory experiments indicated that the most favourable temperatures for N₂ fixation were between 30 and 35^oC, with a range from 4-45^oC [103]. The variation for the best temperature range for activity may depend upon the organisms present and the climatic conditions at each environment [98, 103].

Soil characteristics can significantly alter the potential for NS N₂ fixation. For example, nitrogenase activity by free-living bacteria extracted from soil is best at pH 7-7.5 regardless of the pH of the original soil [112], and liming can increase the abundance of nifH-containing rhizobacteria in acidic soils [94]. Clays are highly reactive colloidal particles that interact strongly with microorganisms [113]. Roper and Smith [112] observed that the presence of montmorillonite clay increased N₂ fixation by free-living bacteria. Clays increase macroaggregate formation [114] which creates sites of low O₂ concentration [115] thus favouring N₂ fixation. Microsites within intra-aggregate pore spaces and interior parts of aggregates not only provide suitable environments for nitrogenase activity but also protect bacteria from environmental extremes [116]. Although halophilic bacteria (e.g. Halomonas maura) isolated from saline soils have been shown to contain nifH genes and fix N₂ [117], there is little other information on the impact of salt on NS N₂-fixing bacteria in terms of growth and N₂ fixation [118] particularly in agricultural soils.

Heavy metal (Zn, Cu, Ni, Cd, Cr, Pb, Hg, As) contamination can reduce abundance of NS N₂-fixing bacteria [119, 120] and N₂-fixing activity [121]. However, some strains of Azospirillum brasilense showed adaptation to heavy metals (Co, Cu and Zn [122]). The response of associative N₂ fixation to heavy metals seems dependent on the tolerance of the plants themselves to the contaminant [123]. For example, roots of aluminium-tolerant plants exude significantly higher amounts of low molecular weight dicarboxylic acids which not only chelate Al³⁺ protecting the plant, but are also C sources for Azospirillum spp. and other N₂-fixing bacteria [36, 91].

Minimum tillage systems support the stability of the aggregates especially macroaggregates that are critical for the development and maintenance of microsites of reduced O₂ tension and for protection against biocidal exposure [116]. Any increase in soil disturbance reduces aggregation, reduces soil C and disrupts the soil pore network by which soil organisms interact [124-126]. As a result of all these factors, NS N₂ fixation under no-till is characteristically higher than in cultivated soils [126]. However, biological changes in NS N₂ fixation in response to adopting reduced/no- tillage practices can be slow sometimes taking several years to develop [20, 116, 127].

In no-till systems, populations of soil macrofauna such as ants (and termites) and earthworms are generally more abundant [126]. Significant amounts of N₂ fixation can occur in the guts of earthworms [128], termites [129, 130] and arthropods [131], e.g. 4-10 kg N ha^-1 year^-1 [131, 132]. Crop rotations can profoundly modify the soil environment by influencing the removal of nutrients from the soil, return of crop residues (including quality and quantity), development and distribution of bio-pores, and dynamics of microbial communities [133] and therefore, are likely to affect the potential for N₂ fixation. Information on the impact of pesticides on NS N₂ fixation is relatively sparse and the effects are mixed. Among the pesticides, herbicides appear to have least significant effects on soil organisms, whereas insecticides and especially copper fungicides can be quite toxic [134]. Fungicides (methyl N-(1H-benzimidazo-2yl) carbamate and tetramethylthiuram disulfide) [135] and the herbicide glyphosate [136] have been shown to have negative impacts on the abundance of diazotrophs, whereas other studies indicated stimulation of populations and activities of N₂-fixing bacteria, e.g. with the insecticide (hexachlorcyclohexane) [137] and a range of herbicides [138].

Associative N₂ fixation has been suggested to be under the genetic control of the host plant [12, 139]. Differences in associative N₂ fixation have been observed between different lines of rice [140], wheat [82, 141], maize and sorghum [142, 143], millet [144], and among various species of grasses [41, 145] and weeds [146]. Characteristics which contribute to high N₂-fixing genotypes include a reduced transpiration rate, lower numbers of stomata and increased root exudates with a high concentration of dicarboxylic acids [143, 147]. Wood et al. [148] suggested that plants with an increased release of photosynthate to the rhizosphere should be a priority for the future development of broad-acre agricultural systems that are more self-sufficient for N nutrition.

There have been many studies on inoculation with N₂-fixing bacteria of non-legumes (predominantly cereals and grasses), with reported above- and below-ground increases in total plant growth and N content [157]. The most successful inoculation responses have been in pot trials under controlled conditions (e.g. [158-161], but inoculation experiments in the field have been less consistent [162]. Andrews et al. [163] concluded that currently no NS N₂-fixing bacterial inoculant is available that can match the consistency of N fertilisers for reducing soil N deficiencies.

One of the difficulties of inoculating soils with bacteria is that the inoculants generally decline rapidly due to competition with the native microflora [164, 165]. Inoculants compete with other microflora for available nutrients or become food for indigenous micro- and macro-fauna [166]. Hence the ultimate test for even the most effective beneficial organism is the ability to survive and colonise plant roots in the presence of much larger populations of indigenous microorganisms [157]. Inoculum formulation and application technology, e.g. along with organic matter (compost or peat) or micro-granulated inoculum, are likely to be crucial for inoculant survival and success [167].

Development of combinations of management practices should maximize NS N₂ fixation. Cropping systems which combine high C inputs and good soil structure, e.g. conservation farming practices or perennial grass systems, are likely to be ideal. No-tillage practices combined with crop residue retention have increased rapidly world-wide in response to a range of pressures, and by 2003 for example, it was estimated that in Western Australia, 86% of farmers had adopted no-tillage [28]. Many of the field studies on NS N₂ fixation in the past were done with soils where stubble retention had recently been adopted (e.g. [19]). Soil microbial composition and functions respond slowly to changed managements [116] and therefore it is likely that if similar experiments were conducted today on sites with long-term no-tillage and crop residue retention, different rates of N₂ fixation might be found, particularly in areas under continuous cropping.

A system which supports 100% ground cover 100% of the time is likely to provide continuous inputs of C either by rhizodeposition or by inputs from above-ground residues. In ‘pasture-cropping systems’ where native summer-active perennial grasses are coupled with winter cereals it is proposed that biological N inputs are sufficient to supply the needs of the cereal crop. Using δ¹⁵N techniques, Mordelet et al. [186] and Abbadie et al. [187] were able to identify contributions from NS N₂ fixation of up to 17 % of the annual savannah requirement for N. In pasture-cropping systems, fixed N is likely to be protected from leaching losses due to uptake by the grasses in autumn and later release by mineralisation from stubble and decomposing roots to a cereal crop during winter when the grass is not active [188]. Further research is needed to understand the N dynamics of pasture-cropping systems and to evaluate their potential in agriculture.

Little is known about the potential for N inputs via N₂ fixation with other plants including weeds and other non-legume components of pastures except for a small study by Conklin and Biswas [146] who observed NS N₂-fixing bacteria and nitrogenase activity (C₂H₂ reduction) associated with 20 weed species.

Only in Brazil are there varieties of sugar cane that have been shown to fix over 60% of their nitrogen (>150 kg N ha^-1 year^-1 [65]). Elsewhere in the world, measurements of contributions to N supply in sugar cane via biological N₂ fixation have been small [189] although specific associations between diazotrophic bacteria and sugar cane have been observed [190]. It has been argued that this may be due to sugar crops in Brazil being systematically bred for high yields with low fertiliser inputs [65, 150]. Baldani et al. [150] suggested that such a breeding process with low fertiliser inputs has led to the development of (or preserved) an effective association between N₂-fixing bacteria and the plant. Almost all of our modern crop varieties have been developed in conjunction with the use of nitrogen fertilisers suggesting that the capacity for significant associative N₂ fixation may have been lost during breeding processes. Therefore, examination of the capacity for associative and endophytic N₂ fixation in the wild relatives of wheat and other cereals, and the possibility of transferring this capability into modern varieties may have merit. Support for this notion can be seen with rice, for example Knauth et al. [140] examined the composition of diazotrophic communities associated with related rice cultivars (Oryza sativa) and wild species (Oryza brachyantha) and found that when grown under identical conditions in the same soil without N fertiliser there were remarkable differences in root associated nifH-gene expressing communities between the two cultivars. Furthermore, NifH fragments expressed in the wild species of rice roots indicated that the active diazotrophs were not related to cultured strains. In a separate study, Engelhard et al. [191] found that endophytic populations of diazotrophs differed with rice genotype and that the natural host range of the non-culturable Azoarcus spp. included rice, with wild and old rice varieties being preferred over modern cultivars. On the other hand, culturable species such as Azospirillum spp., Klebsiella sp., Sphingomonas paucimobilis, Burkholderia spp. were associated with more modern cultivars of Oryza sativa.

Evidence that ecosystems with low N promote NS N₂ fixation occurs in a perennial grass (Molinia coerulea) which grows in oligotrophic environments. In another example, Gupta et al. [41] reported diazotrophic N₂ fixation of 0.92 to 2.35 mg N / kg root / day with summer active perennial grasses such as Panicum species and Rhodes grass (Chloris gayana) in low organic matter soils of southern Australia. Hamelin et al. [34] observed that the rhizosphere of the perennial grass Molinia coerulea supported a diversity of N₂-fixing bacteria, 56% of which contained NifH sequences that did not match any cultivated diazotrophs, but were dominant in the roots and surrounding soil. Further examination of such oligotrophic systems may yield diazotrophic communities that could be adapted to agricultural systems where they might increase the contribution from associative N₂ fixation in agricultural crops.