11 Resource use efficiency, sustainability and environment

Download this chapter in PDF | Download citation

11.3 Nutrient use efficiency

Modern agriculture removes large amounts of nutrients from the land (as product yield), such that nutrient loss must sooner or later be counterbalanced by the addition of chemical and/or organic (largely manure) fertiliser, given the very slow breakdown of soil minerals to give available plant nutrients. Application of fertiliser is also necessary to overcome any plant nutrient deficiency. A comparatively smaller (and variable) amount of available nitrogen (N) and sulfur (S) can also be obtained from atmospheric deposition, while legumes can fix substantial amounts of atmospheric nitrogen (see later).

Fertiliser science has been reviewed by Angus (2012), who gives an agronomist’s up-to-date view across all the essential elements for plant growth and other contemporary issues. Discussion here focuses on fertiliser nutrient use efficiency for the macro-elements nitrogen and phosphorus (P) (see this section, below, for a discussion of P).52 Other major elements—potassium (K) and sulfur (S)—can also be deficient in intensively cropped situations and certain soils (especially with high-demand crop species), but because world supplies are generally adequate, these elements are not discussed here further. Calcium (Ca) and chlorine as chloride (Cl) are additional elements found in larger amounts in plants, but are rarely deficient.

Also important for high crop yield are the minor elements for plants—for example, zinc (Zn), iron (Fe), boron (B) and copper (Cu). However, minor elements are not discussed here because quantities involved are small (usually <1 kg of element is removed per hectare per year) and are unlikely to face limited supply. Suffice to note that constant vigilance is required to avoid micronutrient deficiency; alleviation is inexpensive and automatically boosts efficiency of use of all other inputs. Moreover, there have been notable advances in application efficiency for microelements through sprays or compound fertilisers, and in exploiting genetic tolerances to microelement deficiencies in some crops.


Fertiliser nitrogen is a major cropping input. In 2007–09, FAOSTAT (2013) reported that global annual application of nitrogen was ~100 Mt nitrogen (elemental weight only, i.e. irrespective of the compound applied), of which 55% was used on cereals (Heffer 2009). Furthermore, Smil (2001) estimated that >40% of the protein nitrogen consumed by humans was derived from fertiliser nitrogen.

The global application of fertiliser nitrogen is supplemented by the 31 Mt N annually that is applied as manure (Peoples et al. 2009). In addition, the estimated contribution of atmospheric nitrogen fixed by bacteria in legume roots in agricultural ecosystems—termed ‘biological nitrogen fixation’ (BNF)—is ~46 Mt N, of which 20–22 Mt may derive from grain legumes (Peoples et al. 2009). Non-legume BNF (microbes in other plants or free-living microbes) is considered negligible in crop systems, with few exceptions.

Finally, a significant component of nitrogen input (albeit one that is difficult to estimate) is atmospheric deposition (wet plus dry, meaning in rain and in dust) of reactive nitrogen—i.e. ammonia (NH3) and nitrogen oxides (NOx)—on crops and croplands. US values for atmospheric deposition in 2001 ranged from <5 kg N/ha/yr in sparsely populated areas, to as high as 20 kg N/ha/yr in heavily polluted areas; an average of ~10 kg N/ha/yr was reported for the Mid West (EPA 2011). In China, Liu et al. (2013) reported wet deposition alone contributed 23 kg N/ha/yr in eastern China, with up to 80 kg N/ha/yr total deposition in the North China Plain. Sources of deposited atmospheric nitrogen are identified as industry, transport and agriculture itself (largely NH3 from the animal industry). Emissions from these sources have abated, notably in the USA and Europe through regulation.

Fertiliser nitrogen is often the major input cost for achieving attainable yield with non-leguminous crops. It is largely manufactured from non-renewable energy sources (nowadays mostly natural gas). Priced at US$1–2/kg N, the cost of nitrogen is better appreciated as a ratio to grain price—that is, kilograms of grain to buy one kilogram of nitrogen (at farm gate prices). This number typically ranges from 2 to10 for cereals; low numbers occur in countries where fertiliser is subsidised.

Significant losses of reactive nitrogen to the environment occur in cropping when NH3 and nitrogen oxide gases—especially nitrous oxide (N2O)—are released from the soil, and where nitrate (NO3) is leached from the soil. Non-reactive nitrogen gas (N2) arising from denitrification is a further source of losses, in this case, however, with no direct environmental consequences. Losses can arise both from fertiliser nitrogen and other nitrogen sources in the soil, but NH3 losses arise only from NH3-based fertilisers (and manure), while all losses clearly increase with heavier fertiliser use. Crop leaves can also lose NH3, especially during grain-filling and if the root zone contains high levels of ammonium (NH4+). However, the rate of NH3 loss from leaves is generally low, given (in elemental nitrogen equivalent) as <3 kg N/ha/crop (Hayashi et al. 2011). Losses of N2O, a long-lasting greenhouse gas, are discussed in more detail in Section 11.6.

Given the above considerations, it is not surprising that nitrogen use efficiency (NUE) is a major issue in modern agriculture. NUE refers to kilograms of grain yield per kilogram of nitrogen supply. NUE is complicated by the presence of two nitrogen sources: fertiliser nitrogen applied to the crop in consideration, and nitrogen from the decomposition of soil organic material, usefully referred to as ‘indigenous soil nitrogen’. Because the latter is difficult to measure, this book uses NUE expressed as yield per kilogram of fertiliser-applied nitrogen (kg/kgN) only, abbreviated to NUEf—in other words, the partial factor productivity of nitrogen fertiliser (e.g. Wortmann et al. 2011), without considering the contribution of indigenous soil nitrogen. However, there is no escaping the complications that excluding indigenous soil nitrogen from NUEf causes. The following equations provide a useful separation of NUEf components:

equations (13), (14) and (15) Understanding the component effects on nitrogen fertiliser use efficiency (NUEf)

Source: Moll et al. (1982)

NUEf  =  NUpEf × NUtE (13)

NUtE  =  (DM/NUp) × HI (14)

NUtE  =  NHI/GNC (15)


NUEf is nitrogen fertiliser use efficiency, or kilogram of harvested grain per kilogram of nitrogen fertiliser applied (kg/kg N)

NUpEf is the uptake efficiency of fertiliser nitrogen (kg/kg N), or nitrogen uptake by the crop (NUp) measured in kg N/ha, divided by fertiliser nitrogen supply (kg N/ha)

NUtE is the utilisation efficiency of grain production from nitrogen taken up by the crop (kg/kg N)

DM is the total dry matter of the crop at maturity (kg/ha)

HI is the harvest index (grain dry weight divided by total DM, a dimensionless ratio)

NHI is the nitrogen harvest index (the proportion of crop nitrogen uptake in the grain at maturity, a dimensionless ratio)

GNC is the grain nitrogen concentration (w/w, or often given as a percentage).

As equation (13) shows, nitrogen fertiliser use efficiency (NUEf) is a function of the efficiency of both nitrogen uptake and utilisation (NUpEf and NUtE, respectively). NUpEf relates to capture of nitrogen by the crop root system, while NUtE is determined by how efficiently the nitrogen absorbed by the crop is used in growth and ultimately in producing grain. Equations (14) and (15) are two ways of expressing the same thing (NUtE). They assist in understanding how NUtE has been increased by breeding; greater detail is given in Box 11.1 where a specific example is discussed. The key point is that these measures of efficiency are most useful when fertiliser rates are close to the economic optimum (see Box 11.1).

Box 11.1 Nitrogen use efficiency—an example and complications

Assessing how effectively crops take up and use nitrogen needs is complex. A typical grain yield response to nitrogen fertiliser—along with the key component, nitrogen fertiliser use efficiency (NUEf)—for irrigated wheat in south-eastern Australia is used as an example here to show some of the factors that affect such assessments. Components of NUEf (as defined in equations (13), (14) and (15)) appear in part (b) of the figure.

In the example, a low nitrogen uptake (NUp) of only 12 kg N/ha with zero application of fertiliser indicates that the indigenous soil nitrogen supply was very low. As the rate of applied fertiliser nitrogen increased, NUp steadily increased from the fertiliser source. The uptake efficiency of fertiliser nitrogen (NUpEf)—not shown, but calculated by dividing NUp by rate of fertiliser nitrogen—would have declined as nitrogen rate increased. Also, nitrogen utilisation efficiency (NUtE) in part (b) decreased as NUp increased in association with a continuous fall in nitrogen harvest index (NHI) and, after an initial fall, the steady rise in grain nitrogen concentration (GNC).

Obviously neither NUEf nor NUpEf has relevance at zero or low fertiliser rates. At this point indigenous nitrogen dominates NUp, inflates calculated fertiliser efficiencies, and explains much of the early decline in NUEf as nitrogen rates rise. Thus, to be relevant to best practice, improvements in NUEf (and its components) through better agronomy and breeding must be assessed at the economic optimum nitrogen fertiliser rate—and the possible influence from other sources of soil nitrogen always noted. The economic optimum is given when the slope of the grain yield curve in box figure (a) equals the nitrogen-to-grain price ratio; at a typical 5 kg/kg (i.e. 5 kg of grain to buy 1 kg of fertiliser nitrogen), the optimal rate was about 250 kg N/ha in figure (a).

Note that a decrease in the nitrogen-to-grain price ratio will lift optimal nitrogen rates, inevitably lowering NUEf and risking greater nitrogen losses. If the indigenous soil nitrogen supply is greater than the low level apparent in the box figure—for example, through including legumes in the system—the optimal nitrogen fertiliser rate will be lower. Also, because the definition and method of calculation of NUEf and NUpEf ignore the (greater) supply of indigenous soil nitrogen, both these efficiency measures will appear to improve markedly. The reality may be different, and can be seen only if the extra yield per unit fertiliser nitrogen is measured. A proper understanding of NUE must consider all these issues.

Another more direct way of measuring NUpEf is the apparent fertiliser recovery efficiency—nitrogen uptake with fertiliser less uptake without fertiliser, divided by fertiliser rate. For single applications of 120 kg N/ha, this was 70%, while measuring fertiliser nitrogen recovery using fertiliser labelled with the 15N isotope gave a value of 50%. Thus, fertiliser stimulated uptake of additional indigenous soil nitrogen, a not unusual observation. Finally, it can be useful to calculate ‘agronomic nitrogen use efficiency’ which refers to the increase in yield per unit fertiliser nitrogen applied. This latter measure is the true equivalent to WUEi in irrigated cropping, and requires knowledge of yield in the absence of fertiliser nitrogen.

Part a shows grain yield increasing steeply until about 120 kilograms of applied nitrogen per hectare, before flattening out at a yield about 7.7 tonnes per hectare beyond 200 kilograms of applied nitrogen per hectare. NUEf decreases as nitrogen rate increases. Part b shows NHI and NUtE both decreasing with increase in rate of applied nitrogen. GNC decreases until a nitrogen rate of about 80 kilograms per hectare is reached and then GNC increases. Nitrogen uptake, also called NUp, increases linearly from 25 kg/ha NUp at zero nitrogen application to about 230 kg/ha NUp at 300 kg per hectare of nitrogen application.

All variables shown on the Box Figure are defined in the text

Response of irrigated spring wheat to nitrogen fertiliser, showing (a) grain yield and fertiliser nitrogen use efficiency (NUEf), and (b) nitrogen uptake (NUp), nitrogen utilisation efficiency (NUtE), nitrogen harvest index (NHI) and grain nitrogen concentration (GNC). Source: All nitrogen fertiliser treatments in Fischer et al. (1993) and Fischer (1993), except late (poststem elongation) single nitrogen applications

Nitrogen use efficiency—past progress

Figure 11.2 plots US maize grain yield (i.e. FY), nitrogen fertiliser use and NUEf against time and is illustrative of the evolution of NUEf in modern agriculture. Application rates of nitrogen fertiliser to maize rose sharply between 1940 and 1980, after which rates steadied with less attractive grain prices, to be followed by small increases recently with the advent of bioethanol from maize. Initial NUEf values were inflated by reliance on indigenous soil nitrogen (and perhaps manure), and fell sharply to a minimum of 42 kg/kg N as this source of soil organic carbon and nitrogen declined and was replaced by increasing amounts of fertiliser nitrogen. However, as other factors (see below) lifted FY and the efficiency of nitrogen management, in US maize NUEf then began to improve steadily to reach 62 kg/kg N in 2005–10. In comparison to this temporal pattern of NUEf evolution shown for the USA, developing countries generally operate at earlier stages (also see below).

The graph shows grain yield increases from just over 2,000 kilograms per hectare in 1956 to almost 10,000 in 2010. Over this same period, nitrogen fertiliser rate increases to about 160 kilograms per hectare around 1980 before it decreases then re-gains a level of 160 by the year 2010. NUEf starts near 150 kilograms per kilogram nitrogen at 1946, decreases sharply to about 40 near the year 1965, then increases to about 60 kilograms per kilogram nitrogen in 2010.

Figure 11.2 US maize grain yield, rate of nitrogen fertiliser applied on maize, and nitrogen use efficiency (NUEf) summarised as 5-year means from 1946 to 2010. Source: USDA (2012)

The NUEf increase seen in maize during the past 30 years has been driven by many synergistic improvements in technology. Importantly, nitrogen loss to leaching and denitrification has been reduced by agronomic techniques that better match timing and amount of nitrogen fertiliser to crop need—examples include tactical topdressing based on measured crop need and variable-rate capability incorporated into fertiliser applicators to account for field variation (often considered part of precision agriculture). However, the most important contribution to increased NUEf has arisen through PY progress driven by breeding gains (Ciampetti and Vyn 2013) and other agronomic improvements such as earlier planting, better soil water conservation, better seed fungicides and higher plant density. Any indirect increase in PY automatically increases NUE—a principle based on Liebscher’s law (de Wit 1992) that is broadly relevant to multiple resource use efficiency. This increase means that the apparent diminishing returns with increased nitrogen initially seen in Figure 11.2 were overridden when other yield increasing factors came together.

Just as with maize in the USA, FY for winter wheat in the United Kingdom (UK) has been rising steadily without any increase in application rates for nitrogen fertiliser since 1981; the current rate of application is ~190 kg N/ha, and NUEf has now reached ~42 kg/kg N. Some manure is also applied to wheat in the UK. Consideration should also be given to atmospheric nitrogen deposition (rain and dust), as the contribution was quite high in the late 1980s in the south-east of the UK—estimated at 35–40 kg N/ha (Goulding 1990)—but this may have since decreased.

Sylvester-Bradley and Kindred (2009) looked closely at NUEf with older (1977–87, n  =  10) and newer (1991–2007, n  =  15) winter wheat varieties in nitrogen fertiliser experiments across the UK, as described in Section 3.8. At a given nitrogen level (e.g. 200 kg N/ha) new winter varieties out-yielded old varieties (9.47 t/ha vs. 8.34 t/ha, respectively), and delivered higher NUEf (47.4 kg/kg NUp vs. 41.7 kg/kg NUp, respectively). The improvement in NUEf was probably due to small gains in NUpEf and in NUtE (i.e. NHI increased slightly and GNC was slightly lower; see equation (15)). An important result was that optimal nitrogen rates (using a nitrogen to grain price ratio of 5) differed as expected (174 kg N/ha for new varieties, vs. 146 kg N/ha for old varieties), as did grain yield (9.54 t/ha for new vs. 8.35 t/ha for old), but NUEf no longer differed noticeably (54.8 kg/kg N vs. 57.2 kg/kg N); neither did GNC (average 2.04%).

Newer, higher yielding and generally shorter varieties of irrigated spring wheat in north-western Mexico were found to have higher NUEf relative to older ones; this was derived through higher NUpEf at lower nitrogen supply, and through higher NUtE at higher nitrogen supply (Ortiz-Monasterio et al. 1997). The NUtE increase with breeding in spring wheat was the result of increases in NHI (associated with increases in HI), and by decreases in GNC (Calderini et al. 1995; Ortiz-Monasterio et al. 1997), which possibly carry a grain quality penalty.

Spring barley released in western Europe between 1931 and 2005 showed a 0.6% p.a. (relative to 2005 values) increase in both PY and in NUEf (Bingham et al. 2012). Forty per cent of NUEf improvement was attributed to increased NUpEf (largely through greater postanthesis nitrogen uptake). The remainder was attributed to NUtE, which was clearly increased by breeding in association with increased HI. These authors saw scope for further modest genetic improvements in NUE in barley.

Nitrogen use efficiency—current status and agronomic practices

Table 11.1 summarises fertiliser nitrogen application rates across the major cereals in some important world regions. Average country NUEf has been calculated by dividing yield by nitrogen application rate. Sub-Saharan Africa is not shown, because the current rate of nitrogen application, at ~7 kg N/ha across all crops, it is too low for meaningful NUEf calculations. Of the crops shown in Table 11.1, NUEf tends to be greatest in maize and lowest in wheat. Very high (i.e. excessive) nitrogen rates are evident in China, as are correspondingly low NUEf values. Rates are more moderate in India, where NUEf is higher than for China. The relatively high NUEf values in the USA, Argentina and Brazil may reflect the extra soil nitrogen from prior soybean crops, commonly grown in rotation with maize in these three countries. Crop and country differences, and the effect that better agronomic practices could have on NUEf, are examined below.

Compared with genetic improvement, agronomic management can more strongly influence NUEf (especially by NUpEf). This approach is encapsulated in the ‘4 Rs’ of best nitrogen management practice promulgated by the IPNI (2009): the right source, the right time, the right place and the right rate. Good and Beatty (2011) reviewed published field trials and found many examples of treatments that increased NUEf in maize and wheat in North and South America, wheat in Europe and rice in Asia, when compared with common farmer practice. Increases in NUEf arose especially through reduced nitrogen rates combined with improved nitrogen timing, and NUEf gains of 20–50% (or even more) were achieved without loss of yield.

Table 11.1 Average nitrogen fertiliser application rate and nitrogen use efficiency (NUEf)a across world regions and countries in 2007–09







Fertiliser rate (kg N/ha)




NUEf (kg/kg N)





Fertiliser rate (kg N/ha)




NUEf (kg/kg N)





Fertiliser rate (kg N/ha)




NUEf (kg/kg N)




European Union

Fertiliser rate (kg N/ha)




NUEf (kg/kg N)





Fertiliser rate (kg N/ha)




NUEf (kg/kg N)





Fertiliser rate (kg N/ha)




NUEf (kg/kg N)




  1. Given as kilogram grain yield per kilogram nitrogen applied as fertiliser (kg/kg N)
  2. na  =  not available
  3. Source: FAOSTAT (2013) for crop yield to calculate NUEf and overall fertiliser use; Heffer (2009) for fertiliser use distribution among crops at country level

However, best practice agronomic management of nitrogen fertiliser is not easy. Dobermann et al. (2011) conducted detailed nitrogen rate trials in 32 irrigated maize fields (2002–04) in the central part of southern Nebraska, USA. They illustrated this point by calculating the economically optimal nitrogen rate (N to grain price ratio of 7:1), and the yield and NUEf value at that nitrogen rate. For maize-following-maize rotations, the average values were 171 kg N/ha optimal application, 14.9 t/ha grain yield and 83 kg/kg N for NUEf at the optimal nitrogen rate. For maize-following-soybean rotations, the corresponding figures were 122 kg N/ha, 14.5 t/ha and 117 kg/kg N, respectively.

Unfortunately, the measured economically optimal nitrogen rates varied considerably among fields in each of the Dobermann et al. (2011) groups (standard deviation ~40 kg N/ha), and were overestimated by existing predictive functions used by crop advisers. It was also evident that there was greater potential for economic return to be reduced by using insufficient nitrogen fertiliser, compared to the potential for reduced returns by using excess fertiliser—a common worldwide observation that favours excess nitrogen application ‘as an insurance’, especially at low nitrogen to grain price ratios (with detrimental environmental consequences; see Section 11.5 on sustainability). The southern Nebraskan soils are currently maintained at a high level of fertility, because in the absence of nitrogen fertiliser, average yield was 10 t/ha (range 6–14 t/ha) and indigenous nitrogen uptake was ~160 kg N/ha. At the economically optimal nitrogen rate for these crops (given above), apparent recovery of fertiliser nitrogen was 64%, and NUtE averaged 57 kg/kg NUp, associated with a NHI value of 65% and GNC of 1.3% (Wortmann et al. 2011).

In the same Nebraskan region, Grassini et al. (2011a) examined 777 irrigated maize crops in 2005–07, and reported an average FY of 13.0 t/ha for an average nitrogen fertiliser rate of 183 kg N/ha;53 the average NUEf was therefore 71 kg/kg N, which is well above the Nebraskan state average of 64 kg/kg N. The nitrogen rate used by farmers ranged widely from 100 to 250 kg N/ha across the 777 crops; this range occurred without apparent explanation and without any observable relationship to yield, except for the higher rate (+21 kg N/ha) used for maize-following-maize (representing 38% of maize crops), compared with maize-following-soybean. All of these Nebraskan studies point to scope for better temporal matching of fertiliser nitrogen applications to crop demand in order to further lift NUE.

There are no reports on NUE progress in rice on a broad scale, but rice NUtE should be higher than other cereals because of the intrinsically low GNC of ~1.2% (see equation (15)). Rice is, however, also well known for its low NUpEf values that are commonly <50% and sometimes as low as 30% (e.g. Spiertz 2010; Angus 2012) because of high losses from NH3 volatilisation and denitrification associated with flooding. The NUE situation in China is extreme with rice—and other crops (see Table 11.1)—because nitrogen fertiliser prices are subsidised, nitrogen application rates are very high, and the forms of nitrogen fertiliser primarily used (i.e. urea and ammonium bicarbonate) favour NH3 losses (L. Ma et al. 2010).

Rice agronomists have developed several techniques to improve NUpEf (and hence NUE) that once again involve better matching of supply and timing to crop need and which can be measured at panicle initiation by tissue analysis or leaf colour. Horie et al. (2005a) describe the steady improvement in rice NUE in Japan, such that overall NUpEf improved from 25% to 80%, by moving successively through the following steps:

  1. single application of nitrogen fertiliser at sowing
  2. split application of nitrogen fertiliser (partial application at sowing)
  3. localised application of nitrogen fertiliser
  4. nitrogen fertiliser rates adjusted according to leaf colour
  5. introduction of controlled-release urea.

Among farmers of smallholdings growing rice in Asia, the Japanese experience has evolved into site-specific nutrient management based on crop nitrogen fertilisation according to leaf colour (Dobermann et al. 2002). With site-specific nutrient management, yield can be improved even as nitrogen rates are reduced—such that NUEf can be improved by 20–30%, particularly because in-season nitrogen management lifts NUpEf. Another technique is the use of deep placement of urea super-granules, which lessen nitrogen losses by concentrating the urea several centimetres below the soil surface (IFDC 2012).

The low NUEf at farm level in Asia partly reflects the majority production by smallholder farmers who are yet to adopt the more complex techniques for efficient nitrogen management. Recent developments using mobile phones to relay leaf colour pictures and then deliver recommendations for in-crop nitrogen addition may help the techniques to be adopted, because there remains large scope for improved NUEf in rice (particularly in China) by improving management of nitrogen fertiliser.

While experiments show opportunities to increase NUEf, surveys of observed practice tend to reveal another picture. For example, the USDA’s Agricultural Resource Management Survey (ARMS; Ribaudo et al. 2011) of nitrogen use on all field crops (excluding rice) across the USA in 2006 found that relative to best practices for application rate, timing and method:

  • 21% of the area exceeded best practice rate for nitrogen application by >10%
  • 25% missed best timing for nitrogen application
  • 38% missed best method of nitrogen application.

For maize, the dominant crop in the sample, the figures were not greatly different from the average of all crops. But 14% of maize crops received manure as a source of applied nitrogen, while 2% received manure as the only source, and (for maize) those crops receiving manure scored lower on best management practices than those receiving only nitrogen fertiliser. A later (2010) ARMS report found only slight improvements in the adoption of best practice (Ribaudo et al. 2012).

Overall the ARMS data indicate considerable scope for NUE improvement through better agronomy, even in the USA. It is notable that regulations regarding best nitrogen practices in various European Union countries are beginning to affect nitrogen fertiliser rates, which now tend to be static or decreasing (Good and Beatty 2011; Mikkelsen et al. 2011). At the other extreme, the scope for improvement in China is clearly large, and marked progress in NUEf has been demonstrated in on-farm experiments and demonstrations with smarter management—often achieving higher wheat, rice and maize yields with less nitrogen fertiliser use—but the challenge is to reach the millions of tiny farms there (Zhang et al. 2013). In most other developing countries NUEf is deceptively high because nitrogen fertiliser rates are usually suboptimal and indigenous soil nitrogen is a larger component (see Box 11.1).

Nitrogen use efficiency—novel approaches and limits

What is not clear from the above discussion of NUE and FY progress is whether further PY increases through breeding will continue to improve (or at least maintain) NUE under modern farming at optimal nitrogen rates, and whether improved nitrogen management has more to offer. Breeding specifically targeted at NUE (i.e. more yield with low nitrogen supply) has attracted attention in maize from CIMMYT, and there appear to be large gains reflected across all nitrogen supply levels (see Section 5.4 on maize in Sub-Saharan Africa). The International Rice Research Institute (IRRI) has targeted incorporation of nitrogen fixation capacity into rice roots for several years with limited progress, although sugarcane (another grass crop) can fix appreciable nitrogen through bacteria in its stems (see Section 7.6).

Some predict that NUpE can be further increased, for example, by breeding for root systems that will vigorously grow to capture early mineral nitrogen—when it is most vulnerable to leaching and denitrification losses (Palta and Watt 2009)—or that will penetrate deeper in the soil to retrieve NO3 that has leached. Other gains could come from root exudates that suppress nitrification. This process, termed ‘biological nitrification inhibition’, has been found in the tropical perennial grass Brachiaria spp. (Subbarao et al. 2013), with promising effects on NUEf for maize following Brachiaria. Gains could also come from engineering of ion transporter proteins to increase the rate of NO3 uptake at the root surface (Good and Beatty 2010).

When breeding to increase NUtE—the second component of NUE—then the two simple relationships (equations (14) and (15)) must be satisfied. Equation (14) indicates that increased DM production or increased HI is required without extra NUp if NUtE is to be raised. Unless genetic engineering (GE) can deliver a complete switch in C3 crops to more nitrogen-efficient C4 photosynthesis—which seems very unlikely (see Section 9.4 on increasing RUE)—it is difficult to envisage greater photosynthetic activity and DM production without a heavy nitrogen demand and a corresponding increase in NUp. In addition, as has already been noted in Section 9.2, HI may be reaching its limits.

Similarly, conditions in equation (15) must also be satisfied to raise NUtE. NHI, which is related to, but always greater than, HI, could be independently increased if less nitrogen than the typical proportion of ~30% of crop NUp in a modern variety is left behind in non-grain crop parts at maturity. This is a difficult trade-off for the plant, since nitrogen is needed in leaves to continue photosynthesis during grain-filling at the same time that it is needed in the grain to build NHI. If NHI is held constant, NUE can only increase only if GNC continues to fall, but this brings quality penalties in wheat and possibly nutritional penalties in rice. Barraclough et al. (2010) pointed to the influence of grain classification of recent UK winter wheat varieties: both yield and NUtE were inevitably lower in bread-making varieties because selection has sought higher GNC than for biscuit and feed varieties.

On the agronomic front, there is undoubtedly scope to lift NUE through adopting existing best practices. Newer technologies such as controlled-release urea have been mentioned in the context of rice in Japan. In addition, Yang et al. (2011) used controlled-release urea in irrigated winter wheat in the province of Shandong, China, and reached the economically optimal nitrogen rate with at least 33% less urea nitrogen than for straight urea. Adding nitrification inhibitors with fertiliser granules is another strategy. For example, in Iowa, USA, Parkin and Hatfield (2010) significantly lifted NUE of maize by 7% with the nitrification-inhibitor nitrapyrin applied with the autumn anhydrous NH3, but currently such inhibitors are relatively expensive and not used.

Upper limits to NUEf at the economically optimal fertiliser rate can be determined by using equation (15). Assuming that GNC remains (for grain quality reasons) at typical minimal acceptable values of 1.5% for wheat, 1.2% for paddy rice and 1.3% for maize, and that NHI is fixed at 0.7 for all three crops, then NUtE will be limited to 47 kg/kg N for wheat, 58 kg/kg N for rice and 54 kg/kg N for maize. If NUpEf were to reach 100%, then these NUtEf values would equate to NUEf. But this is not the whole picture, because apart from current applications of fertiliser, NUEf is also influenced by changes in the indigenous supply of soil nitrogen. These nitrogen sources are:

  • accessions from atmospheric deposition, irrigation water and applied manure (see also Section 11.5 on sustainability)
  • biological nitrogen fixation
  • net change in soil inorganic nitrogen fraction involving soil organic matter mineralisation, NO3 leaching and denitrification
  • residual nitrogen from previous fertilisation and from crop residue.

The fact that NUEf in Table 11.1 exceeds the above estimated limits for wheat in the European Union (50.6 kg/kg N) and maize in the USA (60.2 kg/kg N) must reflect that such sources of nitrogen have been excluded from the above estimation. This nitrogen supply could be sustainable if the sources are carryover fertiliser nitrogen, manure, accessions or legume nitrogen. Otherwise, if they arise from net loss of soil organic matter through mineralisation, such high NUEf results will be unsustainable.

Furthermore, for any cropping system there will be an upper limit to NUE when nitrogen removal by grain yield equals nitrogen replacement by fertilisation. In this book this is called the ‘balancing limit’ and suggests another way of looking at the NUEf limit. Grain yield divided by GNC gives this limit if it can be assumed that: (1) accessions in water and dust, manure and nitrogen fixation balance any nitrogen losses due to gaseous nitrogen losses and NO3 leaching; (2) all crop residue nitrogen is returned to soil; and (3) soil organic matter (and thus total soil nitrogen content) is stable. Using the GNC values specified above, the balancing limit NUEf is 67 kg/kg N for wheat, 85 kg/kg N for rice and 77 kg/kg N for maize. These values are higher than would be calculated assuming 100% fertiliser uptake above, because the latter approach ignores return of nitrogen to soil in crop residue.

NUEf values for maize in the USA shown in Table 11.1 (60.2 kg/kg N) and in the Nebraskan maize examples above (71 kg/kg N) are close to the balancing limit NUEf—a surprising outcome given how difficult it is to avoid nitrogen leaching from occasional heavy rainfall events in the US Corn Belt. Of course, much US maize is now planted in a rotation that follows soybean, in part to exploit the small net positive effect that soybean contributes to the soil nitrogen balance. Soybean receives little nitrogen fertiliser under normal practice and 10% of maize crops receive some nitrogen from manure applications. Moreover, atmospheric deposition and nitrogen in irrigation water are not insignificant. Nitrogen input in irrigation water in Nebraska amounts to ~30 kg/ha/yr (P. Grassini, pers. comm. 2012). EPA (2011) maps suggest that total nitrogen deposition in the US Corn Belt is ~10 kg/ha/yr. On the other hand, for economic reasons much fertiliser nitrogen for maize is applied in the autumn ahead of planting and is at greater risk of loss by leaching with spring rains. So the conclusion could be that although the system may be close to the balancing limit for NUEf, there are still moderate non-fertiliser nitrogen inputs balanced by moderate amounts of nitrogen escaping into the environment.

For wheat in Table 11.1, the NUEf result for the European Union (50.6 kg/kg N) is closest to the balancing limit, possibly reflecting the best example of nitrogen management. For rice, ignoring the figure of 73 kg/kg N from Brazil, which may reflect soils with a short history of cropping and high organic matter, the best NUEf occurred in India and the USA (each close to 47 kg/kg N). But these figures are still were well below the balancing limit (85 kg/kg N), probably reflecting high losses of fertiliser nitrogen in irrigated rice.

Stability of soil organic matter is the biggest uncertainty in these calculations. Carbon sequesters nitrogen at a ratio of about 12:1 in soil organic matter—that is, one tonne of extra carbon sequesters 100 kg of nitrogen. If soil organic matter were to increase (say, if the cropping system switches to conservation tillage), more soil nitrogen would be sequestered and NUEf would decline. On the other hand (and other things equal), as climates warm, soil organic matter may decline (Sentilkumar et al. 2009), which will initially boost nitrogen supply to crops and raise NUEf. As difficult as it may be to increase NUpEf and NUtE, none of these important nitrogen balance complications and uncertainties lessen the urgency for seeking new techniques to achieve this goal, in particular, easier and less-expensive ways for farmers to meet the ‘4 R’s’ of best management practice.

Nitrogen—role of legumes

If managed to maximise nitrogen fixation, legume grain crops generally receive little or no fertiliser nitrogen; therefore NUEf is irrelevant for these crops. Further, depending on crop type, grain legume crops can make small net contributions of around 10–50 kg N/ha to soil nitrogen. As a rule of thumb, legumes may fix 20 kg of atmospheric nitrogen per tonne of total DM produced (Peoples et al. 2009). This production is generally more than the amount of nitrogen removed when the grain is harvested.54

Nitrogen in remaining underground DM (roots, nodules, rhizosphere exudates) adds to the legume contribution of nitrogen to the soil. Peoples et al. (2009) estimated that 190 Mha of legume grain crops globally contribute ~5–7 Mt N to soil after allowing for removal of about 6 Mt through grain harvest; this is a contribution of about 30 kg N/ha. For this and other reasons, legume crops are almost always followed by non-legumes, as shown, for example, by the soybean–maize rotation in the USA and Brazil, and pulse–wheat or pulse–canola rotations in Europe, Canada and Australia. In such systems, proper nitrogen fertiliser recommendations acknowledge the nitrogen contribution from the legume, so that less nitrogen fertiliser is applied to post-legume crops, and NUEf correspondingly increases as discussed previously for maize-following-soybean.

Soil organic matter and nitrogen accumulate much faster under legumes sown as green manure crops or pasture (rather than legume grain crops), with rates of 50–300 kg N/ha/yr depending on legume DM production. However, these practices have declined in popularity everywhere because of negative economic return in the green manure year, or reduced returns from grazed pastures (relative to crops), and the widespread availability of cheap nitrogen fertiliser (except generally in Sub-Saharan Africa). For example, the rainfed wheat cropping system in Australia did not use fertiliser nitrogen until the 1980s, but relied on nitrogen from preceding years of leguminous pasture, with the ratio of leguminous pasture to wheat area approximately 1:1. Since then, economics favoured cropping, and rotational pastures and livestock numbers on wheat farms have declined, so that by 2000 wheat was receiving fertiliser nitrogen at an average rate of 30 kg N/ha (Angus 2001). However, the very high national average NUEf of ~63 kg/kg N suggests that legumes still contribute an important amount of nitrogen to wheat production in Australia. When the ratio of livestock to crop prices changes, so does the attractiveness of including legume forage years in a cereal crop rotation, both in Australia’s wheat–sheep system (e.g. Thomas et al. 2009) and in the maize–leguminous forage system in Iowa, USA (e.g. Davis et al. 2012). However, what discourages retaining traditional mixed farming or switching to it are the capital costs and the much greater management complexity of such systems.

Legume research has been neglected relative to cereal research so there is considerable scope to redress this imbalance, with benefits for cropping system NUE. The success of soybean as a new crop confirms the responsiveness of legume yield to investment in research and development. Furthermore, recent work suggests that while nitrogen fixation in legume root nodules consumes photosynthate, it appears that rate of photosynthesis is stimulated more by the nodule sink resulting in no net cost (in terms of DM) from nitrogen fixation (Kaschuk et al. 2009). It is, however, wishful thinking to expect legume-fixed nitrogen to supply all the nitrogen demands of the global non-legume grain crop. Green manures probably account for <1% of current world crop area, and would need to expand to perhaps equal the non-legume crop area (i.e.1,200 Mha) to supply all nitrogen demands for these crops (Connor 2008; see also Box 11.2). Even so, grazed legumes (ley farming) could produce high-protein animal products during the legume phase of rotation with little reduction in the accumulation of soil nitrogen relative to green manure legumes.

Nitrogen—some conclusions

Cereals and other non-legume crops need large amounts of nitrogen for current economically optimal or attainable yields. The amounts of nitrogen are proportional to target yield (~20 kg N/t for wheat and rice, and ~15 kg N/t for maize), but inevitably increase the risk of some losses of nitrogen from the cropping system. If soil nitrogen is not to be depleted, nitrogen demand must be met by nitrogen fertiliser. In a stable system, the amount of applied nitrogen should be more or less equal to nitrogen removed in grain, provided losses from leaching and denitrification can be balanced by the small accessions from dust, rain, irrigation water, legume and non-symbiotic nitrogen fixation.

Although the need for nitrogen fertiliser can be reduced in a cereal cropping system by increasing the frequency of legumes in rotations, the nitrogen contribution from legumes is limited if the grain is harvested, and legumes do not reduce the risk of nitrogen losses to the environment. Improved agronomic techniques have improved NUEf (largely through greater NUpEf) and genetic improvement has lifted NUE (largely through greater NUtE). However, modern cropping operates at higher optimal soil nitrogen levels, which somewhat counters these gains.

Further progress in NUtE through breeding seems difficult, because it will require further increase in NHI, or reduction in GNC, which could threaten grain quality. However, scope exists for raising NUpE—perhaps through genetics, but especially through improved agronomy. Such improved practice would ensure that nitrogen is supplied only to meet crop needs at each stage of the crop cycle while avoiding excess supply. Besides research, more expensive nitrogen and regulation are probably necessary to achieve more efficient nitrogen fertiliser use.


The current global amount of phosphorus applied annually as fertiliser is ~17 Mt (average for 2007–09; FAOSTAT 2013), of which ~46% is applied to cereals (Heffer 2009).55

The global supply of non-renewable rock phosphate—the source of all fertiliser phosphorus—is adequate for many years to come, although grades are falling and costs are rising (Van Kauwenbergh 2010). The alarming price spike of 2007–08 should not be attributed to long-term supply shortages but rather to a temporary shortage in the face of heightened demand, along with profit-seeking behaviour of suppliers (Cornish 2010). On a weight basis, phosphorus costs about three times as much as nitrogen, and although grain phosphorus concentration is usually only one-fifth of that of GNC, phosphorus can be an expensive input to cropping in deficient or phosphorus-fixing soils. Thus an important productivity objective is to increase phosphorus use efficiency (PUE) or more specifically, PUEf (defined similarly to NUEf).

Table 11.2 shows average elemental phosphorus rates and PUEf for the key cereals in several world countries or regions. Sub-Saharan Africa is again not shown because rates, at ~1.5 kg P/ha across all crops, are too low. Compared with the USA and the European Union, phosphorus rates in China, India, Argentina and Brazil are higher and PUEf is clearly lower for both wheat and rice, suggesting inefficiencies in those countries. For maize, rates are higher in the USA, the European Union and Brazil, but efficiencies are similar across all countries except for the low value in Brazil. The low PUEf values may be caused by excessive application rates, especially in China, but the uniquely low PUEf in Brazil may reflect the highly P-fixing soils in Brazil in general and the new maize area of the Cerrado in particular.

PUE can usefully be dissected in the same way as NUE (Box 11.1)—thus PUE increases as all other factors contributing to PY are optimised—yet there are notable differences between phosphorus and nitrogen. Phosphorus is relatively immobile in soil and, apart from topsoil erosion, or leaching risks in very sandy soils, it is not subject to losses. However, depending on soil chemistry, fertiliser phosphorus is fairly readily fixed into inorganic forms (with iron, aluminium and calcium) and/or organic forms, and thus plant-available phosphorus will gradually decrease over time (McLaughlin et al. 2011)—this represents another type of loss, at least to the plant, if not to the environment. Because of this, soil testing for levels of available phosphorus can provide a useful guide to optimal fertiliser rates; the practice is especially appropriate because phosphorus must be applied before crop planting.

Highly phosphorus-fixing soils require phosphorus to be applied for many years at rates well in excess of crop demand until available soil phosphorus is high enough to permit only replacement (or maintenance) rates of phosphorus fertiliser to balance phosphorus removal as grain. In theory, crops with a lower critical requirement for available phosphorus should reduce the phosphorus investment needed before ‘balancing’ phosphorus applications are reached, but this approach is more complex in practice (Simpson et al. 2011).

Table 11.2 Average phosphorus fertiliser application ratesa and phosphorus use efficiency (PUEf)b across world regions and countries in 2007–09







Fertiliser rate (kg P/ha)




PUEf (kg/kg P)





Fertiliser rate (kg P/ha)




PUEf (kg/kg P)





Fertiliser rate (kg P/ha)




PUEf (kg/kg P)




European Union

Fertiliser rate (kg P/ha)




PUEf (kg/kg P)





Fertiliser rate (kg P/ha)




PUEf (kg/kg P)





Fertiliser rate (kg P/ha)




PUEf (kg/kg P)




  1. Elemental phosphorus weights are used, not phosphorus pentoxide (P2O5) weight
  2. Given as kilogram of grain yield per kilogram of phosphorus applied as fertiliser (kg/kg P)
  3. Source: FAOSTAT (2013) for total fertiliser used and for yields to calculate PUEf; fertiliser use distribution among crops at country level based on Heffer (2009)

PUpEf increases as phosphorus-fixing sinks in soils become saturated with continuing application in excess of removal. Separately, improvements to PUpEf have arisen through manufacturing techniques that have improved availability of fertiliser phosphorus—for example, superphosphate compared with rock phosphate, and liquid instead of granular forms (at least on calcareous soils)—or which place fertiliser phosphorus below (or otherwise close to) the crop seeding row. Soil phosphorus tends to be concentrated in the upper soil levels, especially with modern zero-till systems; thus deeper location of fertiliser phosphorus to where soil water is more consistently available (even into the subsoil) may be advantageous in rainfed situations, and may be made possible without deep tillage if new formulations are used (McLaughlin et al. 2011).

Applying NH3 forms of nitrogen fertiliser along with the phosphorus fertiliser can increase early growth in phosphorus-deficient calcareous soils (Jing et al. 2012), possibly because acidification of the rhizosphere in the fertiliser band aids phosphorus solubilisation. Intercropping cereal and certain legumes in phosphorus-deficient situations can markedly improve the cereal yield, leading to overyielding56 of the intercrop. For example, this effect occurred with an intercrop of maize and faba bean because of the intermingling of the rhizospheres of each crop, and probably because of acidification arising from the faba bean rhizosphere (Li et al. 2007). Note that the overyielding effect in this example was present with high nitrogen fertiliser, but not with high phosphorus fertiliser, when the maize presumably had less need for the solubilising effect of the faba beans. Thus the general but poorly understood phenomenon of legumes complementing or aiding phosphorus supply to intercropped cereals (Hinsinger et al. 2011) may operate only in phosphorus-deficient situations, thus reducing its usefulness.

PUEf has also increased as breeding has lifted PY, probably largely because phosphorus-utilisation efficiency (PUtE) by the plant has increased for the same reasons as has NUtE. Certainly, modern short wheat varieties show higher values for phosphorus HI (and lower values for grain phosphorus concentration) than older, taller varieties (Batten 1993; Calderini et al. 1995), without difference in total phosphorus uptake. In contrast to grain nitrogen, much grain phosphorus is in the form of inositol or phytate phosphorus—forms that do not benefit (and can actually reduce) grain food and feed quality because phytate reduces calcium uptake by animals. Thus PUtE gains made through lower grain phosphorus concentration may carry no hidden cost, and such changes have arisen as the indirect consequence of genetic yield progress, at least in wheat. Researchers now directly pursue reduced wheat grain phytate content (and hence lower grain phosphorus concentration) to further increase PUE. However, to date, at least with wheat, low grain phytate lines do not necessarily show lower grain phosphorus concentration, and hence improved PUEf; neither do they show improved nutritional quality.

Most efforts for the future focus on increasing phosphorus uptake efficiency (PUpE) through increasing the availability of soil-fixed phosphorus. Several mechanisms for achieving this exist in plants naturally adapted to soils with low available phosphorus, so there is considerable variation in PUpE among crop species. For example, crops like lupins (Lupinus spp.) or buckwheat (Fragopyrum esculentum) can extract several times more indigenous soil phosphorus than wheat (McLachlan 1976). Mechanisms showing intraspecific variation have been extensively reviewed (Richardson et al. 2011). They include root foraging and root mining characteristics such as:

  • more and finer roots (especially close to the soil surface, where phosphorus is concentrated)
  • longer root hairs
  • propensity to harbour root associations with fungal arbuscular mycorrhiza
  • root exudates such as malic, citric or piscidic acid as well as phosphatase enzymes, often associated with cluster root formation.

Some progress has been made in identifying varieties with roots that are more efficient phosphorus foragers; this has been achieved in beans (Phaseolus vulgaris) by Liao et al. (2001) and in wheat by Manske et al. (2000). Rice is known to contain a natural gene (Pup1) that increases phosphorus uptake and confers tolerance to phosphorus deficiency. Known for some time, Pup1 has been mapped and studied, and appears to operate by increased early root growth and hence greater phosphorus acquisition (Gamuyao et al. 2012). Other research groups are now pursuing GE approaches such as improved phytase exudation to release soil phytate phosphorus, a major component of the unavailable soil organic phosphorus. So far success in the field has been very limited (Richardson et al. 2011). These mechanisms will not be useful if they are facultative—that is, if they operate only when the plant is seriously phosphorus deficient. By comparison, microbial inoculants have been available for some time, and are promoted as increasing phosphorus solubility in the rhizosphere. On occasions these inoculants appear to stimulate root growth rather than increase phosphorus solubility, but either way, their performance in the field is still very unreliable (Richardson et al. 2011).

Note that improved PUpE can only reduce the need for fertiliser phosphorus until soil phosphorus levels build to a point sufficient to supply PY, when phosphorus applications can then fall to levels balancing phosphorus removal by grain. As applies for nitrogen—although free from complicated gains and losses from the system, except for possibly slow decrease in the availability of fixed forms of soil phosphorus—the upper balancing limit to PUE is reached in a stable cropping system, and is given by grain yield divided by grain phosphorus concentration. Based on typical grain phosphorus concentrations of 0.35%, 0.27% and 0.40% for wheat, paddy rice and maize, the upper balancing PUEf limits are determined to be 290 kg/kg P, 370 kg/kg P and 250 kg/kg P, respectively.

Observed farm values from Table 11.2 that are lower than these upper limits (e.g. rice and wheat in China and India, maize in Brazil) suggest either overfertilisation or ongoing fixation of phosphorus, undoubtedly the latter in Brazil. Table 11.2 values that are higher (rice in the USA, wheat in the European Union and maize in all regions except Brazil) indicate that soil phosphorus is being mined—possibly from excess phosphorus applied to the systems at earlier times. However, considerable uncertainty surrounds observed values of PUEf and the lower limits of grain phosphorus content.

In conclusion, just as has occurred for NUE, PUE has benefited from the multiple ways in which PY (and PYw) have been improved, but the existence of soil phosphorus fixation brings unique considerations, including very low soil phosphorus losses if erosion is controlled. In the phosphorus build-up phase, best practice management of phosphorus fertiliser can increase uptake efficiency and reduce application rates, so that less investment in soil phosphorus accumulation is needed before a stage of balance is reached. While large differences appear to exist between plant species in uptake efficiency, it has proved very difficult to exploit such traits in crop plants and much more research is needed. At the same time, direct selection for lower grain phosphorus concentration could bring further small gains in PUtE (and PUEf), both benefits which would be seen in the build-up phase as well as afterwards, when phosphorus application should only need to balance product removal.

52 Elemental phosphorus (P) content is used throughout this book, although the fertiliser industry often refers to phosphorus pentoxide (P2O5) content, of which 43.66% is elemental phosphorus, so that 1 kg P = 2.29 kg P2O5.

53 Additional nitrogen input in irrigation water for maize in Nebraska, USA, amounts to ~30 kg/ha/yr but atmospheric deposition is less than 10 kg/ha (P. Grassini, pers. comm. 2012).

54 Kilograms of grain nitrogen removed can be calculated per tonne of total DM (above-ground only) by multiplying HI by GNC by 1,000. GNC is lower for pulses (3–4%) than soybean (6.0–6.5%). A crop with HI  =  0.4 and GNC  =  4% removes as grain 16 kg N/t DM, leaving a net gain of only 4 kg N/t.

55 Elemental phosphorus (P) content is used throughout this book, although the fertiliser industry often refers to phosphorus pentoxide (P2O5) content, of which 43.7% is elemental phosphorus, so that 1 kg P  =  2.29 kg P2O5.

56 ‘Overyielding’ is best expressed as a ‘land equivalent’ ratio of >1.0—the ratio of the land area of the two crops as sole crops needed to produce the same amount as unit area of intercrop; in the maize–faba bean example used above, the land equivalent ratio was 1.35 under phosphorus deficiency.

Download this chapter in PDF | Download citation | Back to top