Nitrogen losses from the soil/plant system: a review

Soil pH

Calcareous soils with a naturally high pH can lose significant amounts of ammonia gas; however, neutral or acid pH soils may also lose significant amounts of ammonia when urea or animal urine is applied, as shown in Eqn 3 (Black et al., 1985a,b).

When urea is applied to the soil surface the hydrolysis reaction produces a sharp increase in pH resulting in an increase in ammonia volatilisation. The pH of the microsite below the urea granule can increase above a value of 8 (Sommer et al., 2004). This process occurs over the first few days following application; however, once nitrification processes start to become significant then the pH of the soil is decreased, which in turn reduces the volatilisation rate.

Temperature

Soil and atmospheric temperature affect the rate of urea hydrolysis, and the rate of transfer of ammonia from soil solution into the atmosphere (Black et al., 1985a,b; McGarry et al., 1987). Under field conditions, the hydrolysis rate and the ammonia emission rate follow a diurnal sequence, with the highest emissions occurring at the time of highest daily temperature. There is also a seasonal effect with the highest losses occurring during the warmest months. For example, Ball & Keeney (1983) reported ammonia losses from animal urine patches equivalent to 5% and 16% of the N applied under cool and warm conditions, respectively. However, some studies have shown that the total amount of N lost by ammonia volatilisation may not necessarily be different at low versus high temperatures because the volatilisation may continue for longer at low temperatures (Fenn & Kissel, 1974; Harper et al., 1983).

Ammonium concentration

The amount of ammonia emitted from the soil is related to the ammonium concentration in the soil solution. In general, the higher the solution concentration of ammonium the higher the potential emission rate of ammonia gas. Thus, application of N fertiliser (especially urea) and animal urine can significantly enhance the ammonia emission rate. Nevertheless, there are many other factors that affect the concentration of ammonium in the soil solution, including nitrification rate, plant uptake rate, denitrification rate and immobilisation rate, all of which reduce the solution concentration and thus ammonia volatilisation loss potential.

Soil cation exchange capacity

Soil cation exchange reactions retain ammonium ions on the surface of clays and organic matter through electrostatic attraction. This mechanism helps to store ammonium in soil and also reduces the soil solution concentration of ammonium. The cation exchange capacity (CEC) also helps to buffer the soil against pH change; thus, when urea hydrolysis occurs in a clay soil (which has a high CEC) there is a smaller increase in pH than would occur in a sandy soil (which has a low CEC) (Daftardar & Shinde, 1980; Whitehead & Raistrick, 1993). Thus, the ammonia volatilisation potential of a clay soil (high CEC) is generally lower than that of a sandy soil (low CEC).

Soil moisture, rainfall and irrigation

Soil moisture content affects the concentration of ammonia/ammonium in solution, and low soil moisture contents therefore promote high solution concentrations and thus high ammonia emission loss. However, at very low soil moisture levels, the rate of dissolution of urea fertiliser and of urea hydrolysis will be slow and the ammonia loss will also be slow (Al-Kanani et al., 1991). Increasing the soil moisture content of a dry soil can increase the rate of urea hydrolysis and thus increase the rate of ammonia production from urea (Vlek & Carter, 1983; Reynolds & Wolf, 1987). However, significant amounts of rainfall or irrigation water applied soon after urea or urine deposition can also reduce the ammonia volatilisation loss because the rainfall or irrigation water washes the urea below the soil surface and keeps the surface solution concentration low (Black et al., 1987).

Variations in ammonia volatilisation losses are often due to a combination of factors including soil texture, soil moisture content as well as rainfall after fertiliser application. For example, Turner et al. (2012) found that small differences in weather conditions and initial soil moisture content had a significant effect on the size of ammonia volatilisation losses from fertiliser applied 1 week apart at the same location. The ammonia first lost from the application of 46 kg N ha⁻¹ ammonium sulphate represented 12% of the applied N, whereas the loss 1 week later from a similar rate of ammonium sulphate was only 2.8% of the applied N (Turner et al., 2012).

Fertiliser use

The greatest risks of ammonia volatilisation losses occur from ammonium bicarbonate, urea and ammonium hydroxide fertilisers, with lower losses reported from ammonium sulphate or diammonium phosphate fertilisers (Black et al., 1985b; Whitehead & Raistrick, 1990; Sommer & Jensen, 1994; He et al., 1999; Sommer et al., 2004; Turner et al., 2012). Ammonia volatilisation losses from urea applied at <50 kg N ha⁻¹ are generally in the range of 5–15% of the N applied, whereas ammonia losses from ammonium sulphate and calcium ammonium nitrate are typically less than 2–3% of the N applied at this rate (Black et al., 1984, 1985a,b). Examples of ammonia volatilisation losses from urea fertiliser are given in Table 1 (Bishop & Manning, 2010).

Grazing animals

It is estimated that 85–90% of the N ingested by grazing animals is excreted in urine and dung. This creates discrete patches in grazed pastures, which contain high concentrations of nitrogen. Urine is mostly a highly concentrated solution of urea (e.g. up to 10 g N L⁻¹) that when deposited onto soil during urination can apply large amounts of N (e.g. equivalent to 1000 kg N ha⁻¹ from a dairy cow) (Haynes and Williams, 1993; Di & Cameron, 2002a,b). The urea is quickly hydrolysed creating a temporary high pH, leading to ammonia emissions from the urine patch (Haynes et al., 1986). Ammonia volatilisation can also occur when animal manures are applied to the soil surface (Vanderholm, 1975; Bussink & Oenema, 1998).

Plants

Plant leaves can absorb ammonia directly through the stomata and can therefore reduce the amount of ammonia gas lost from the soil/plant system into the atmosphere (Schjoerring & Mattson, 2001; Denmead et al., 2008). However, plant leaves can also emit ammonia gas. For example, Wang & Schjoerring (2012) reported that green leaves and stems of ryegrass plants generally represent a sink for NH₃, whereas senescent leaves have a large potential for NH₃ emission. Nevertheless, research shows that in general the greater the height and density of the plant cover, the lower the ammonia loss (Denmead et al., 1982).

Nitrification

Nitrate is produced in soil through the process of nitrification, which involves the oxidation of ammonium as described by the following chemical reactions:

(4)

(5)

The two parts of the reaction are dependent on the activity of two different genera of autotrophic bacteria. The first reaction, the oxidation of NH₄⁺ to NO₂⁻, is mainly conducted by the activity of soil ammonia-oxidising bacteria (AOB), such as Nitrosospira and Nitrosomonas. This occurs through oxidation reactions performed by the ammonia monooxygenase (AMO) enzyme associated with the bacteria (Ferguson et al., 2007; Prosser, 2007; Di et al., 2009a,b, 2010a,b). Ammonia-oxidising archaea (AOA) are also present in large numbers in soils, and although they may make an important contribution to nitrification in low-fertility marine and estuarine environments (Leininger et al., 2006), they do not appear to be as important as AOB in the relatively N-rich agricultural soil environment (Di et al., 2009b, 2010b) (Fig. 2).

The second reaction, the oxidation of NO₂⁻ to NO₃⁻, is conducted by Nitrobacter. The conversion of NO₂⁻ to NO₃⁻ takes place very rapidly and therefore nitrite rarely accumulates in soil.

The rate at which ammonium is converted into nitrate is sensitive to soil conditions, such as moisture content, pH and nutrient status. The maximum rate of nitrification takes place at a soil moisture content around ‘field capacity’ (i.e. at a soil matrix potential of −10 kPa) (Haynes et al., 1986). In soils that are wetter than field capacity the nitrification rate is slower. The nitrification rate is also significantly slower when the soil is dry, although it can still occur when soils are at permanent wilting point (−1500 kPa) (Malhi & McGill, 1982; Monaghan & Barraclough, 1992).

The optimum pH for nitrification to occur is between pH 4.5 and 7.5 (Haynes et al., 1986). The optimum soil temperature for nitrifying bacteria is between 25°C and 30°C (Haynes et al., 1986). Nitrification will still occur at soil temperatures below 5°C, but the rate is slower than at higher temperatures.

The amount of ammonium present in the soil also affects the nitrification rate because high concentrations of ammonium/ammonia can restrict the activity of Nitrobacter (Monaghan & Barraclough, 1992). Soil phosphate deficiency can also slow the nitrification rate as can high concentrations of heavy metals in soil (e.g. from sewage sludge applications to land) (Purchase, 1974; Ruyters et al., 2010; Griffiths & Philippot, 2012; Thavamani et al., 2012). Nitrifying bacteria are also sensitive to pesticides (Muñoz-Leoz et al., 2011).

Solute transport mechanisms

If it is assumed, for simplicity, that there are steady-state water conditions in the soil and that there is no interaction between the nitrate ion and the soil, then nitrate leaching can be described by a combination of three main mechanisms: convection, diffusion and dispersion (Hillel, 1980, 1998; Cameron & Haynes, 1986; McLaren & Cameron, 1996).

Convective transport – This occurs because of the mass flow of water through the soil during drainage events. The convective flux of nitrate can therefore be described by a modified form of Darcy's law:

(6)

where J_c is the convective nitrate flux, c is the nitrate concentration, q is the water flux, K is the hydraulic conductivity and dH/dx is the hydraulic gradient.

The distance transported per unit time by convection depends on the average pore water velocity, U, where:

(7)

and θ is the volumetric water content.

Convective transport implies uniform displacement of the band of nitrate but, in reality, the band of nitrate tends to spread throughout the profile because of the processes of diffusion and hydrodynamic dispersion.

Diffusive transport – Because there is a concentration gradient between the band of nitrate and the surrounding soil, there is a diffusive flux of nitrate from the band of nitrate into the surrounding soil. Movement by diffusion can be described by Fick's law:

(8)

where J_d is the rate of diffusion, D_s is the diffusion coefficient of the nitrate in soil and depends on the soil moisture content and dc/dx is the nitrate concentration gradient.

Hydrodynamic dispersion – This occurs because: (a) the flow velocity within a single pore is not uniform; (b) there are large variations in pore size in soil, which cause a range of pore water velocities and (c) the ‘tortuosity’ of soil pores produces a range of flow path lengths.

Combined convective-diffusive-dispersive transport – The overall equation describing convective-diffusive-dispersive transport of nitrate is:

(9)

where D_a is the apparent diffusion coefficient and represents the sum of molecular diffusion plus hydrodynamic dispersion.

However, there are other soil factors that also affect nitrate leaching and are more difficult to describe mathematically because of their transient nature and complexity. Plant roots, earthworms, freezing and thawing cycles and wetting and drying of soil can produce surface-connected macropores (i.e. large diameter pores) in the soil profile. Water flow through these macropores can have two effects on leaching: (a) when nitrate is present in the infiltrating water then macropore flow will cause leaching to occur at a faster rate than normal, and (b) when nitrate is present within soil aggregates then the bulk of the drainage water may ‘bypass’ them leading to a slower than normal rate of leaching.

Some biological transformation processes, such as immobilisation, can remove N from the soil solution and can therefore reduce the rate of leaching. Other transformation processes, such as mineralisation, may release N into the soil solution and thus increase the rate of leaching loss. Plant uptake of nitrate decreases the concentration in soil solution and therefore reduces the rate of leaching loss (and will be discussed in more detail below).

Season and climate

Nitrate leaching losses are usually greatest during late-autumn, winter and early-spring months when plant uptake of nitrate is low (because of cooler conditions) and drainage is occurring from the soil (because of rainfall inputs exceeding the evapotranspirative demand) (Wild & Cameron, 1980). For example, Di et al. (1999) compared N leaching losses from N fertiliser applied in the autumn with those from N applied in the spring and found that between 15% and 19% of the autumn-applied N (NH₄Cl) was leached compared with an equivalent of 8–11% of N applied in the spring. Autumn rainfall can also leach residual fertiliser nitrate left after the crop has been harvested and nitrate released by mineralisation of soil organic nitrogen. Cameron & Wild (1984) reported that large amounts of nitrate (>100 kg N ha⁻¹ year⁻¹) were produced after ploughing of 3-year-old pasture soil and that this nitrate was leached down through the soil profile over the subsequent winter.

A dry summer can result in an accumulation of nitrate in the surface soil (because of low plant growth and low N uptake) and this nitrate can be leached over the subsequent winter (Scholefield et al., 1993; Goulding, 2000; Stout et al., 2000). In addition, the rewetting of soil after a dry summer can also cause a flush of mineralisation and cause more nitrate to be leached.

Rainfall soon after N fertiliser application can leach N below the active root zone, and this will be discussed in more detail below.

In tropical areas where monsoon rains occur, the nitrate leaching loss is often greatest during the summer monsoon period rather than during the winter period. Arora et al. (1980) reported greater leaching losses because 400–500 mm of rain fell during the summer monsoon compared with only 100 mm of rain during the winter.

Soil properties

Nitrate leaching losses are generally greater from poorly structured sandy soils than from clay soils because of the slower water movement and the greater potential for denitrification to occur in the clay soils. Pratt et al. (1980) presented data that gave an approximate ratio of leaching loss of 5:1 for a silt loam soil compared to a clay loam soil.

As discussed above, macropores created by earthworms, plant roots or wetting and drying cycles can allow nitrate to be leached quickly through the soil profile and can therefore have a large effect on the rate of nitrate leaching (e.g. Silva et al., 2000). Agricultural drainage systems can also have an effect on the NO₃⁻ leaching loss. For example, Scholefield et al. (1993) showed that NO₃⁻ leaching increased in the presence of an efficient drainage system, probably because of increased N mineralisation. Drainage systems also shorten the distance that NO₃⁻ travels through the soil resulting in faster transport into rivers and lakes. Drainage systems also increase the aeration status of the soil, which decreases the denitrification potential meaning that more nitrogen stays in the form of nitrate.

Soil organic matter content and mineralisation of that organic matter can affect the nitrate leaching loss. Substantial leaching losses can occur from unfertilised bare fallow soil, for example, at Rothamsted Experimental Station in the UK unfertilised bare fallow lysimeters were found to have lost 42 kg N ha⁻¹ year⁻¹ (Miller, 1906; Russell & Richards, 1920; Addiscott, 1988). Low & Armitage (1970) also measured large leaching losses of 102 kg N ha⁻¹ year⁻¹ from bare fallow soil lysimeters, emphasising the increased risk of nitrate leaching when the opportunity for plant uptake is removed.

Forests

Nitrate leaching losses are generally lowest from forest systems because: (a) there is usually zero or only low rates of N fertiliser applied; (b) the N is cycled efficiently through the forest ecosystem and (c) because the soils are generally acidic and thus mostly contain ammonium-N rather than nitrate-N. Leaching losses from forest soils may range from about 1 to 15 kg N ha⁻¹ year⁻¹ (Gosz, 1981; Juergens-Gschwind & Follett, 1989; Kiese et al., 2011).

However, harvesting the trees can release large amounts of N that can be leached, or washed off slopes through soil erosion (Smith et al., 1994; Hazlett et al., 2007). Burning of forests can also release large amounts of N that can be leached (Stark, 1977). Hornbeck et al. (1975) reported a total N loss of over 340 kg N ha⁻¹ over a 3-year period following the clear-cutting of a North America hardwood forest. Nevertheless, the C : N ratio of the forest litter and soil dictates the amount of N mineralisation (and thus potential risk of N leaching) that occurs from the forest residues. When the C : N ratio is high the potential leaching loss is low because net immobilisation rather than mineralisation occurs.

Grassland

In general, extensively grazed grassland systems have low nitrate leaching losses (Di & Cameron, 2002a). This is due to rapid plant uptake and immobilisation of N, which keeps the nitrate concentration in soil solution low (Woodmansee et al., 1981) and the low animal stocking density of extensive systems means that there are fewer urine patches per unit area (Moir et al., 2006). Grasslands that are mown or cut for hay or silage also have very low leaching losses (e.g. Di et al., 1998b) because grass and pasture plants are usually very efficient at capturing the nitrogen applied in fertiliser or N fixed by legumes such as clovers that are grown in the pasture sward. A large amount of the N that is applied is removed by pasture plants (300–400 kg N ha⁻¹ year⁻¹) (Di et al., 1998b). Our studies have shown that when 200 or 400 kg N ha⁻¹ is applied in four split applications (either as urea fertiliser or as animal manure) the nitrate leaching losses ranged from 6 to 17 kg N ha⁻¹ year⁻¹ (Silva et al., 1999). However, if the N applied was only split into two rather than four applications, the nitrate leaching loss increased to 13–49 kg N ha⁻¹ year⁻¹ (Di et al., 1998a,b). The actual amounts leached will of course depend on the amount of fertiliser applied as well as the soil and climatic conditions (Steenvoorden et al., 1986; Addiscott, 1996; Wachendorf et al., 2004; Monaghan et al., 2005). Some examples of the amounts of nitrate leaching losses under a range of grassland systems are given in Table 2.

The nitrate leaching potential increases when grassland is grazed rather than harvested (Ball & Ryden, 1984; Ryden et al., 1984; Stout et al., 1997; Monaghan et al., 2007). This is because a large proportion (60–90%) of the nitrogen ingested by the grazing animal is excreted back onto the soil in the small concentrated areas of urine and dung patches (Haynes & Williams, 1993; Jarvis et al., 1995). For example, a grazing cow may deposit the equivalent of 700–1200 kg N ha⁻¹ (Haynes & Williams, 1993; Di & Cameron, 2002a) in each urine patch area of around 0.4 m³ (Moir et al., 2012). A cow may urinate 10–12 times per day and thus, depending on stocking rate, these concentrated urine patch areas may cover 20–30% of the grazed field each year (Moir et al., 2011). Some of the urine-N may be volatilised (Sherlock & Goh, 1983, 1984; Di & Cameron, 2000) but most of it will be nitrified into nitrate (Moir et al., 2012) and is thus at risk of leaching if it is not utilised by plants or denitrified (Fraser et al., 1994). Unfortunately, the urine-N loading rate is usually well above the ability of the pasture plants to retain the N (usually less than 600 kg N ha⁻¹ year⁻¹; Moir et al., 2007, 2012) and the excess nitrate remains in the soil until drainage and leaching occurs. The amount of nitrate-N leached from animal urine patches can therefore be very high and losses equivalent to over 120 kg N ha⁻¹ year⁻¹ have been recorded from below urine patches (Silva et al., 1999; Di & Cameron, 2004a,b).

However, because the whole of the grazed field is not covered by urine patches, the overall leaching loss is ‘diluted’ by the lower leaching loss from the interurine patch areas (often <8 kg N ha⁻¹ year⁻¹) (Di & Cameron, 2002a). The whole field leaching loss can be calculated on the basis of the relative area of urine patch coverage (Di & Cameron, 2000, 2002a; Lilburne et al., 2012):

(10)

where N_L is the annual average NO₃⁻–N leaching losses from a grazed field, N_L1 and N_L2 are the leaching losses from the urine and non-urine patch areas, respectively, and P₁ and P₂ are the proportion of areas covered by urine and non-urine patch areas, respectively. The values of P₁ and P₂ will vary depending on the stocking rate and the urine patch area coverage.

In the study by Silva et al. (1999), although the leaching loss was 124 kg N ha⁻¹ year⁻¹ directly under the urine patches, the leaching loss from the whole field when calculated according to Eqn 10 was equivalent to 33 kg N ha⁻¹ year⁻¹ (when no N fertiliser was applied). However, the addition of N fertiliser to the field increased the nitrate leaching loss to between 36 and 60 kg N ha⁻¹ year⁻¹, depending on the rate of fertiliser-N applied. Monaghan et al. (2005) showed that in cattle grazed pasture systems the leaching loss increased when the rate of fertiliser increased, with an average of 30, 34, 46 and 56 kg N ha⁻¹ year⁻¹ being leached when 0, 100, 200 and 400 kg N ha⁻¹ year⁻¹ of fertiliser were applied, respectively (Table 2). The effect of rate of N input (fertiliser or clover fixation) on nitrate leaching loss is illustrated in Fig. 3.

Nitrate leaching losses under sheep grazing are generally lower than under cattle grazing because sheep urine patches contain less nitrogen (equivalent to about 300–500 kg N ha⁻¹) compared to cow urine patches (Di & Cameron, 2002a).

The age and thus ability of plants to capture N from urine and fertiliser has a significant influence on the leaching loss. For example, Jarvis (2000) and Scholefield et al. (1993) reported average annual leaching losses over an 8-year period to be 129 kg N ha⁻¹ year⁻¹ from old grazed pasture receiving 400 kg N ha⁻¹ year⁻¹, whereas the leaching loss was reduced to about 50 kg N ha⁻¹ year⁻¹ when the same rate of N fertiliser was applied to new pasture (this was owing to the greater uptake and growth of the new pasture plants) (Fig. 4).

Arable and horticultural crops

There is a large amount of literature that reports leaching losses from cropping soils and examples of the losses are shown in Table 3. Nitrate leaching losses are generally greater from horticultural crops than from arable crops because of the higher rates of N fertiliser that are used in horticultural crops and the shallow root systems of horticultural plants compared to arable plants. However, the loss is also affected by the climate, soil type as well as the fertiliser rate. As shown in Table 3, leaching losses between 4 and 155 kg N ha⁻¹ year⁻¹ have been reported, depending on the rate of N fertiliser applied, soil type and crop rotation system.

Goulding (2000) reviewed nitrate leaching losses from arable and horticultural land in the UK and reported that the greatest losses occurred from peas and potatoes as well as from rotational set-aside land (i.e. land that is left in grass and only cultivated occasionally). The lowest leaching losses were reported to occur from sugar beet, grass pasture (with low inputs of N fertiliser) and from winter cereals (Fig. 5).

Leaching losses from cropping systems can be high when compared with pasture systems (Francis et al., 1995). Williams (1975) reported that 51 kg N ha⁻¹ year⁻¹ was leached from cultivated cropland compared to only 18 kg N ha⁻¹ year⁻¹ from grassland. In crop rotation systems where pasture is included for only a few years and then ploughed up to establish a cereal crop then leaching losses can be high (Cameron & Wild, 1984; Addiscott et al., 1991; Francis et al., 1992; Whitmore et al., 1992; Shepherd et al., 2001). Francis et al. (1992) reported losses of 60 kg N ha⁻¹ year⁻¹ when a 3-year grass ley was cultivated. Johnston et al. (1994) measured 126 kg N ha⁻¹ in the top 90 cm of soil after cultivation of a 1-year grass ley and 252 kg N ha⁻¹ after cultivation of a 6-year grass ley. Deep borehole studies in England showed that ploughing of grassland was responsible for the release of peak nitrate concentrations that are moving slowly through the unsaturated zone of chalk aquifers (Young et al., 1976).

Mineralisation of soil organic matter and nitrification of residual fertiliser can lead to large leaching losses when there is no crop to remove the nitrate from the soil solution (Cameron & Wild, 1984; Francis et al., 1992; Shepherd et al., 2001).

In organic farming systems the highest nitrate leaching losses have also been found to occur following the ploughing of grass–clover leys because of the large amounts of N that are released by mineralisation of organic-N (Stopes et al., 2002; Eriksen et al., 2004; Andrews et al., 2007). This lack of vegetation for part of the year is a key factor that affects the amount of nitrate leached from cropping soils.

Legumes such as faba beans (Vicia faba L.) can nevertheless fix large amounts of N (up to 100–200 kg N ha⁻¹) and can thus reduce the need for N fertiliser on succeeding crops (Jensen et al., 2010). This in turn can reduce the risk of nitrate leaching as less N fertiliser may be required.

Plant uptake of N during the winter drainage period can reduce the risk of nitrate leaching loss, and cover crops can therefore reduce the risk of N leaching from ploughed or cultivated land (Shepherd & May, 1992; McLenaghen et al., 1996; Eriksen et al., 2004; Hansen et al., 2007)

Plant species can influence the amount of nitrate leaching that occurs from the soil profile. For example, alfalfa has a deep rooting system that is capable of taking up nitrate from considerable depth (up to 3 m), whereas shallow root crops such as potatoes are unable to capture nitrate that is deep in the soil profile and therefore leave large residues of N in the soil that can be leached over the drainage period (e.g. Webb et al., 1997). Brassica crops have been reported to leave more than 300 kg N ha⁻¹ in some soils, most of which can be leached over the next winter (Rahn et al., 1992).

Studies in North America have reported annual leaching losses of 11–107 kg N ha⁻¹ year⁻¹ from corn production soils (Bjorneberg et al., 1996) with much of this loss being attributed to the period of bare fallow before the crop is grown (November to May). Most of the nitrate-N that is leached often comes from the mineralisation of organic-N and remineralisation of fertiliser-N that was immobilised earlier during the previous cropping phase.

Wheat crops remove a large amount of N from the soil each year. Studies in the UK have shown that when 230 kg N ha⁻¹ year⁻¹ of fertiliser was applied to wheat in the spring, 50–80% of the N applied was recovered in the crop, 10–25% was retained in the soil and only about 5% of the N applied was lost by leaching (Macdonald et al., 1989; Addiscott, 1996).

Biomass crops have recently been promoted as an environmentally friendly alternative to fossil fuels, and crops such as maize (Zea mays L.) have been planted for ethanol production (Dale et al., 2010). In 2008, about 30% of all maize grown in the USA was used for ethanol fuel production and this is projected to increase in the future (McIsaac et al., 2010), and the effect of this on the wider environment is now being assessed. McIsaac et al. (2010) have reported average nitrate leaching losses of 40 kg N ha⁻¹ year⁻¹ from a typical maize/soybean (Glycine max L.) system and compared this to losses from switchgrass of just 1.4 kg N ha⁻¹ year⁻¹ when grown in the same location. However, as Christian & Riche (1998) and Cadoux et al. (2012) have shown, nitrate-N losses from biofuel crops such as Miscanthus × giganteus can be significant, with losses of over 60 kg N ha⁻¹ year⁻¹ occurring at fertiliser rates of 120 kg N ha⁻¹ year⁻¹.

Intensive horticultural crop production systems have a high risk of nitrate leaching. High inputs of fertiliser-N, frequent cultivation, short periods of plant growth, high-N-containing crop residues, shallow rooted plants and low nutrient use efficiency all contribute to this high risk of nitrate leaching. Fertiliser-N rates are often as high as 600–900 kg N ha⁻¹ year⁻¹ with plant N recovery as low as 20% (Pionke et al., 1990; Di & Cameron, 2002a). Nitrate leaching losses of 300 kg N ha⁻¹ year⁻¹ have been recorded in some studies (e.g. Spiers et al., 1996) and are typically over 100 kg N ha⁻¹ year⁻¹ (Di & Cameron, 2002a). The nitrate concentration of water draining below market garden crops in sandy soil in Western Australia ranged from 71 to 209 mg N L⁻¹ (Pionke et al., 1990).

There are far less data on nitrate leaching losses from orchard crops compared with vegetable crops. Kramer et al. (2006) reported 80% lower nitrate leaching losses from organically fertilised apple orchard soils compared to those that received conventional N fertilisers.

In wetland rice production systems in Asia, the risk of nitrate leaching is generally low because of the low permeability layer at about 15 cm depth, which has been formed by puddling the soil. However, in coarse textured paddy soils, nitrate leaching losses can be significant (e.g. Singh et al., 1991).

Nitrification

The nitrification process described in Eqns 2 and 3 can be expanded to show that the oxidation of ammonium can also produce nitrous oxide gas (adapted from Kool et al., 2009):

The activity of AOB such as Nitrosospira and Nitrosomonas can therefore produce N₂O from NH₄⁺ and from NH₂OH (which is an intermediary compound shown in Eqn 11; Bremner, 1997; Saggar et al., 2009). It is thought that the intermediate compound nitroxyl (HNO) may dismutate chemically under low O₂ concentrations to N₂O, or that the nitrate reductase enzyme may produce N₂O when O₂ concentrations are low and NO₂⁻ replaces O₂ as the terminal electron acceptor during metabolic processes (Schmidt, 1982; Haynes et al., 1986). The first article that showed that nitrification could produce N₂O was published by Corbet (1935).

Denitrification

Biological denitrification mainly occurs in poorly drained soils where there are anaerobic soil conditions, low oxygen availability and low redox conditions (Firestone, 1982; Dobbie & Smith, 2001; Bateman & Baggs, 2005). Under these circumstances, facultative anaerobic bacteria can use NO₃⁻, instead of oxygen, as the terminal electron acceptor during respiration. This causes NO₃⁻ to be reduced producing, in turn, nitrite, nitric oxide (NO), nitrous oxide and finally dinitrogen, as illustrated in the following equation:

The enzymes responsible for the reduction are NO₃⁻ reductase, NO₂⁻ reductase, NO reductase and N₂O reductase. These enzymes are associated with most denitrifying bacteria; however, some bacteria are only able to complete part of the reduction (i.e. they are partial denitrifiers).

The amounts of NO gas that are emitted from soil are generally low, but under field conditions significant amounts of nitrous oxide gas can be emitted before it can be converted into dinitrogen. There is currently considerable research interest in the effect of NO on plant development, metabolism and disease responses (Gupta et al., 2011). In addition to the NO produced within the plant itself, NO emissions from the soil may also affect plant functions.

Although denitrification is generally highest under anaerobic conditions it can also occur in soils that are not completely water logged and can also occur within soil aggregates, where oxygen diffusion rates are slow and the soil redox potential is low. Highest denitrification losses generally occur in situations where there are high concentrations of NO₃⁻ in soil solution, a readily available source of carbon and the soil temperature is sufficiently warm for microbial activity to occur (Firestone, 1982; Saggar et al., 2009).

Fungal denitrification can also occur in soils, and as shown by Shoun et al. (1992) and Thomson et al. (2012), although most fungal-denitrifying activity reduces only NO₂⁻ to N₂O, some fungi can reduce both NO₂⁻ and NO₃⁻ to N₂O (Shoun & Tanimoto, 1991). In wet soils, anaerobic ammonium oxidation, called ‘anammox’, can also occur, which involves ammonium being converted directly into N₂O and N₂ gas (Hayatsu et al., 2008; Thomson et al., 2012). In addition, N₂ and/or N₂O can also be produced through chemical reactions between nitrous acid (HNO₂) in soil solution and soil organic matter, ammonium, NH₂OH and compounds containing free amino groups (such as amino acids, urea and amines).

The factors that affect denitrification have been reviewed previously (e.g. Firestone, 1982; Haynes & Sherlock, 1986; Bouwman et al., 1993; Bouwman, 1996; Bremner, 1997; Freney, 1997; Mosier et al., 1998; de Klein et al., 2001; Bolan et al., 2004; Clough et al., 2005; Saggar et al., 2009; Thomson et al., 2012) and examples of N₂O emissions are given in Table 4.

Soil moisture and aeration

Changes in soil moisture content affect the aeration status of the soil and this influences the denitrification rate. In particular, when the moisture content of the soil is greater than field capacity there is a significant increase in the potential denitrification rate (Bremner & Shaw, 1958; Haynes et al., 1986; Mosier et al., 1986; Bouwman, 1996; de Klein & van Logtestijn, 1996; Müller & Sherlock, 2004; Saggar et al., 2009). Denitrification losses are therefore greatest during the late-autumn, winter and early-spring periods when soil moisture contents are highest. However, heavy rainfall or irrigation can also cause denitrification at other times of the year (Di & Cameron, 2003; Phillips et al., 2007). Soil texture influences the drainage rate of the soil, and thus the amount of denitrification loss is often greater in clay soils than in free draining sandy soils.

The reductase enzymes that operate in the early part of the reaction sequence shown in Eqn 12 appear to be less sensitive to the availability of O₂ than the later reductase enzymes, so that when O₂ concentrations are high N₂O is produced, whereas at low O₂ levels N₂ gas is produced in greater quantities.

Nitrate and ammonium

The availability of N (NH₄⁺ and NO₃⁻) has a large influence on the denitrification rate with emissions generally increasing resulting in increases in mineral-N availability (Haynes & Sherlock, 1986; Ball et al., 1997; Ambus, 2005; Saggar et al., 2009). In particular, the application of N fertiliser and animal excreta (urine and dung) increases the mineral-N status of the soil and can induce very large increases in denitrification (de Klein et al., 2001; Di & Cameron, 2003; Di et al., 2007). The proportion of the N fertiliser or excreta that is emitted as N₂O is called the ‘emission factor’ (IPCC, 2007) and this value is used in calculations of greenhouse gas emissions by individual countries. For fertiliser, this value is usually in the range of 0.1–2.0%, although it can be higher under some circumstances (de Klein et al., 2001). For animal manures, the emission factor value generally ranges from 0% to 5% of the amount of manure N applied (de Klein et al., 2001). Animal excreta that is deposited during grazing can apply relatively large amounts of N within discrete urine or dung patches and this can produce large amounts of N₂O and N₂ emissions (generally within the range of 0.1–4% of the N applied) (de Klein et al., 2001). The default IPCC emission factor for N₂O emissions from urine deposited on grazed pasture soil is 2%; however, some countries such as New Zealand have a country-specific emission factor of 1% based on detailed inventory measurements (de Klein et al., 2001).

Carbon

It is well established that there is a strong relationship between readily available organic carbon in soil and the denitrification rate (Bremner & Shaw, 1958; Burford & Bremner, 1975). Thus, land management practices that increase the amount of readily available organic carbon (such as application of organic wastes or animal excreta deposited during grazing) stimulate the denitrification process (Cameron et al., 1997; Chadwick, 1997; de Klein et al., 2001; Di & Cameron, 2003). The addition of organic carbon to soil not only stimulates microbial growth and respiration but also provides the organic carbon needed by the soil denitrifiers. Taghizadeh-Toosi et al. (2011) have recently shown that the incorporation of biochar into soil can suppress in situ N₂O emissions.

Soil pH

Soil pH influences both the nitrification rate and the denitrification rate and therefore has an influence on the amount of N₂O and N₂ gas emitted (Parkin et al., 1985; Gödde & Conrad, 2000). The N₂O/N₂ + N₂O product ratio is influenced by soil pH and increases with acidity (Thomson et al., 2012). Therefore, acid soils with a pH <5.0 have lower denitrification rates than agricultural soils with a typical pH around 6.0 (Haynes & Sherlock, 1986). Increasing soil pH can also influence the ratio of N₂O to N₂ emitted (Saggar et al., 2009).

Temperature

The denitrification rate increases with temperature (Ryden, 1986; Dobbie & Smith, 2001). For example, de Klein & van Logtestijn (1996) found that the denitrification rate increased 10-fold in a grassland soil when temperature increased from 10°C to 20°C. Nommik & Larsson (1989) also found that denitrification increased 10 to 20-fold in a forest soil when soil temperatures increased from 6°C to 21°C. Dobbie & Smith (2001) as well as de Klein & van Logtestijn (1996) showed that the effect of soil temperature was greatest on non-irrigated dry soil compared to irrigated soil.