Plant Soil (2008) 311:1–18 DOI 10.1007/s11104-008-9668-3

MARSCHNER REVIEW

Global inputs of biological nitrogen fixation in agricultural systems David F. Herridge & Mark B. Peoples & Robert M. Boddey

Received: 7 March 2008 / Accepted: 22 May 2008 / Published online: 11 July 2008 # Springer Science + Business Media B.V. 2008

Abstract Biological dinitrogen (N2) fixation is a natural process of significant importance in world agriculture. The demand for accurate determinations of global inputs of biologically-fixed nitrogen (N) is strong and will continue to be fuelled by the need to understand and effectively manage the global N cycle. In this paper we review and update long-standing and more recent estimates of biological N2 fixation for the different agricultural systems, including the extensive, uncultivated tropical savannas used for grazing. Our methodology was to combine data on the areas and yields of legumes and cereals from the Food and Agriculture Organization (FAO) database on world agricultural production (FAOSTAT) with published and unpublished data on N2 fixation. As the FAO lists grain legumes only, and not forage, fodder and Responsible Editor: Yongguan Zhu. D. F. Herridge New South Wales Department of Primary Industries, 4 Marsden Park Rd, Calala, NSW 2340, Australia M. B. Peoples CSIRO Plant Industry, P.O. Box 1600, Canberra, ACT 2601, Australia R. M. Boddey (*) Embrapa-Agrobiologia, Km 7—Rodovia BR 465, Seropédica 23890-000 Rio de Janeiro, Brazil e-mail: [email protected]

green manure legumes, other literature was accessed to obtain approximate estimates in these cases. Below-ground plant N was factored into the estimations. The most important N2-fixing agents in agricultural systems are the symbiotic associations between crop and forage/fodder legumes and rhizobia. Annual inputs of fixed N are calculated to be 2.95 Tg for the pulses and 18.5 Tg for the oilseed legumes. Soybean (Glycine max) is the dominant crop legume, representing 50% of the global crop legume area and 68% of global production. We calculate soybean to fix 16.4 Tg N annually, representing 77% of the N fixed by the crop legumes. Annual N2 fixation by soybean in the U.S., Brazil and Argentina is calculated at 5.7, 4.6 and 3.4 Tg, respectively. Accurately estimating global N2 fixation for the symbioses of the forage and fodder legumes is challenging because statistics on the areas and productivity of these legumes are almost impossible to obtain. The uncertainty increases as we move to the other agricultural-production systems—rice (Oryza sativa), sugar cane (Saccharum spp.), cereal and oilseed (non-legume) crop lands and extensive, grazed savannas. Nonetheless, the estimates of annual N2 fixation inputs are 12–25 Tg (pasture and fodder legumes), 5 Tg (rice), 0.5 Tg (sugar cane), 40% of the world’s soybean with relatively high %Ndfa values (Alves et al. 2003; Hungria et al. 2005) (see also Table 3). To differentiate %Ndfa for the different legumes at smaller scales, i.e. field, catchment, region, according to local soil and plant-growth conditions and then aggregate those estimates to generate country and global values would be extremely difficult and may not improve accuracy. Having said that, %Ndfa of soybean needs to be differentiated for the principal soybean-producing countries as this crop is responsible for most of the N fixed by legumes, and there are considerable differences in soil type, climate and plant-cultural practices amongst those countries (Table 3). In the U.S., soils used for soybean production tend to be fertile, with moderate-high concentrations of clay, organic matter and plant-available N (e.g. Russelle and Birr 2004). As a result, reported Ndfa values mostly range between 40% and 80% (van Kessel and Hartley 2000; Peoples et al. 2008; Salvagiotti et al. 2008), with an overall average value of 60%. The average Ndfa value for soybean in Brazil is calculated at 80%, reflecting the widespread use of rhizobial inoculants, the high N demand of the crops (about 300 kg N/ha) coupled with low inputs of fertiliser N, and the high proportion of the crops that are no-tilled (Hungria and Vargas 2000; Hungria et al. 2005, 2006; Alves et al. 2003; FAOSTAT). Alves et al. (2003) and

8

Plant Soil (2008) 311:1–18

Table 3 Estimates of amounts of N fixed annually by soybean in the major soybean-producing countries, using FAO statistical data for 2005 (FAOSTAT), estimates of country-specific %Ndfa, and estimates of harvest index, %N shoots and below-ground N as % of total crop N Grain yield (Tg)

Shoot DM (Tg)a

Shoot N (Tg)b

Crop N (Tg)c

%Ndfa

Crop N fixed (Tg)

Country

Area (Mha)

U.S. Brazil Argentina China

30.0 22.9 14.0 9.6

85.0 51.2 38.3 16.8

212.6 128.0 95.8 42.0

6.38 3.84 2.87 1.26

9.56 5.76 4.31 1.88

60 80 80 50

5.74 4.61 3.44 0.95

Soybean

93.4

214.8

537.1

16.12

24.17

68

16.44

a Using harvest index (grain dry matter as a proportion of total above-ground dry matter) value of 0.4 (Jefing et al. 1992; Herridge and Holland 1992; Guafa et al. 1993; Herridge and Peoples 2002; Shutsrirung et al. 2002; Gan et al. 2002, 2003; Salvagiotti et al. 2008) b

Using %N shoots of 3.0% (Herridge et al. 1990; Herridge and Holland 1992; Herridge and Peoples 2002; Shutsrirung et al. 2002; Gan et al. 2002, 2003; Salvagiotti et al. 2008) c

Multiplying shoot N by 1.5 (Rochester et al. 1998)

others (see review by van Kessel and Hartley 2000) reported consistent increases in nodulation and N2 fixation of no-tilled soybean compared with crops grown under cultivation. The increases under no till were thought to be due principally to reduced levels of nitrate coupled with improved moisture conditions in the soil. Thus, Alves et al. (2003) reported that Brazilian soybean derived 70–85% of crop N from N2 fixation, equivalent to 70–250 kg N/ha. In the case of high-yielding crops, i.e. >4.0 t/ha, as much as 350–400 kg N/ha may be fixed. Similarly, Hungria et al. (2005) reported Ndfa values of 69–94% for inoculated soybean in Brazil. There are very few reports quantifying N2 fixation of soybean in Argentina. Published Ndfa values are 30– 70% (Garcia 2004) and 40–50% (Gutiérrez-Boem et al. 2004; Di Ciocco et al. 2004), but these estimates were from experimental sites and not farmer’s fields. However, Argentinian soybean farmers, like the Brazilian farmers, commonly use inoculants and no-tillage production systems with negligible fertiliser N (Garcia 2004; Hungria et al. 2005; Peloni 2006; FAOSTAT). Garcia (2004) also noted that most of the soils used for soybean production in Argentina have nutrient deficiencies, including N. Taken together, these reports suggest that the high N demand crops would need to fix a large proportion of their N requirements. We therefore assume the same average Ndfa value for soybean in Argentina as for soybean in Brazil, i.e. 80%. Chinese farmers reportedly apply fertiliser N to soybean and rely on the naturalised soil rhizobia to nodulate the crops rather than use inoculants (Gan et al. 2002; Ruiz Sainz et al. 2005). P.W. Singleton (personal communication) estimated that about

0.54 Tg fertiliser N was applied to 10.5 Mha soybean and groundnut in 1994. The fertiliser N inputs plus residual mineral N in the soil from previous crops would depress N2 fixation activity substantially. Thus, we estimate the average Ndfa value for China at 50% (Ruiz Sainz et al. 2005). The total amount of N2 fixed by soybean for each of the four major soybean-producing countries can now be estimated by combining the %Ndfa values with production statistics from FAOSTAT. First, the total amount of soybean N is calculated by dividing the FAOSTAT crop production data (Column 3, Table 3) by an average harvest index value (0.4) to determine shoot dry matter (DM) (Column 4). Shoot N (Column 5) and crop N (Column 6) are then calculated using 3% for the N concentration of shoots and a multiplication factor of 1.5 to account for below-ground N (Rochester et al. 1998). Crop N fixed (final column) is then determined as crop N×%Ndfa. Thus, estimates of total crop N fixed by soybean range between 0.95 Tg annually for China to 3.4 Tg for Argentina, 4.6 Tg for Brazil and 5.7 Tg for the U.S. We used the same series of calculations to estimate global N2 fixation of the major pulse and oilseed legumes (Table 4). The final column contains the calculated values for annual crop N fixed for each species plus total values for the pulse legumes (2.95 Tg), oilseed legumes (18.5 Tg) and all crop legumes (21.5 Tg). In a previous publication we calculated global N2 fixation by the pulse and oilseed legumes by using estimates of average amounts of N fixed per unit shoot biomass (Peoples et al. 2008). This approach was based on the observation that amounts of N2

Plant Soil (2008) 311:1–18

9

Table 4 Estimates of amounts of N fixed annually by the major pulse and oilseed (crop) legumes, using FAO statistical data for 2005 (FAOSTAT), values for average %Ndfa from Table 2 and estimates of values for harvest index, %N shoots and below-ground N as % of total crop N Legume

Area (Mha) Grain yield (Tg) Shoot DM (Tg)a Shoot N (Tg)b Crop N (Tg)c %Ndfa Crop N fixed (Tg)

Common bean Cowpea Chickpea Pea Lentil Fababean Other pulses Total pulses

25.1 9.2 10.4 6.6 4.1 2.7 11.4 69.7

18.1 4.6 8.4 11.3 4.1 4.3 9.4 60.2

51.7 13.3 23.9 32.3 11.8 12.4 26.8 171.9

1.03 0.27 0.48 0.65 0.24 0.27 0.54 3.48

1.45 0.37 0.96 0.90 0.33 0.38 0.75 5.14

40 63 63 63 63 75 63 57

0.58 0.23 0.60 0.57 0.21 0.29 0.47 2.95

Groundnut Soybean Total oilseeds

23.4 93.4 116.7

37.6 214.8 252.4

93.9 537.1 707.8

2.16 16.11 18.27

3.03 24.17 27.20

68 68 68

2.06 16.44 18.50

Total crop legumes 186.4

312.6

879.7

21.75

32.34

66

21.45

a

Using harvest index (grain dry matter as a proportion of total above-ground dry matter) values of 0.4 for groundnut and soybean and 0.35 for the remainder (see references in footnote Table 3; also Schwenke et al. 1998; Evans et al. 2001; Hiep et al. 2002; Hoa et al. 2002; MJ Unkovich, personal communication) b Using %N shoots of 3.0% for soybean, 2.3% for groundnut, 2.2% for fababean and 2.0% for the remainder (see references in footnote Table 3; also Schwenke et al. 1998; Evans et al. 2001; Hiep et al. 2002; Hoa et al. 2002) c

Multiplying shoot N by 2.0 (chickpea), 1.5 (soybean) and 1.4 (remainder) to account for below-ground N.

fixed by legumes in any agroecosystem were primarily regulated by plant growth and DM production. The provisos were that effective rhizobia were present in the soil and concentrations of soil mineral N were not excessive. Data collected from both experimental trials and farmers’ crops indicated that crop legumes generally fix 15–25 kg shoot N for every Mg shoot DM accumulated, with averages of 20 kg shoot N/Mg shoot DM (Fig. 2; see also Evans et al. 2001; Maskey et al. 2001; Peoples et al. 2001). Fixed N associated with the nodulated roots increased the value to 30 kg total crop N/Mg shoot DM. Common bean, chickpea and soybean were identified as the exceptions, with values for common bean of 15 kg total crop N fixed/Mg shoot DM, and for chickpea and soybean of 40 kg crop N fixed/Mg shoot DM. We used these values to calculate global N2 fixation of 4 and 18 Tg N (total 22 Tg N) annually by the pulses and oilseed legumes, respectively, using FAOSTAT production statistics for 2000–2004. Smil (1999) used yet another approach to calculate average annual values for global N2 fixation by the crop legumes. Ranges of values (minimum, mean, maximum) for crop N fixed for each species were estimated on an area basis (kg N/ha), then applied to the global areas of the legumes from FAOSTAT.

Comparisons of the Smil (1999) estimates of legume N2 fixation (area basis, kg N/ha) and estimates using the data in Table 3 are shown in Table 5. There is generally good agreement between the Smil (1999) values for crop N2 fixed (kg/ha) and our values calculated from Table 4, except for soybean and pea (Table 5). The difference in the case of soybean can be explained by the recent expansion of production in Argentina and Brazil where the use of fertiliser N is low, inoculation is widespread and the N demands of the predominantly no-tilled crops are large because of relatively high grain yields (2.73 Mg/ha for Argentina and 2.23 Mg/ha for Brazil, FAOSTAT for 2005). The longstanding notion that soybean fix, on average, about 50% of their N needs would appear to be no longer valid. Smil (1999) estimated crop legumes to fix a total of 10 Tg N annually, compared with our estimate of 21.5 Tg annually. As mentioned above, the discrepancy results mainly from the different values of % Ndfa for pea and soybean, our inclusion of estimates of below-ground fixed N associated with, or released from, roots and nodules, and the use of updated FAOSTAT statistics, i.e. 2005 data used for calculations in Tables 3 and 4 compared with mid 1990s data used by Smil (1999).

10

Plant Soil (2008) 311:1–18

Fig. 2 Examples of the relationship between amounts of N2 fixed (as kg N/ha in shoots) and shoot dry matter (Mg/ha) for crop legumes growing in different geographic regions. Data includes both rainfed and irrigated cool-season (open circles) and warm-season legumes (closed triangles). The dashed lines indicate 15 and 25 kg N fixed per Mg dry matter. Relationship modified from Peoples et al. (2008) who used published and unpublished data collated from studies undertaken in the Middle East and Asia (Syria, Nepal, Pakistan, Thailand), Oceania (Australia), South America (Brazil), North America (Canada and USA), and Europe (Austria, Denmark and France)

Forage/fodder legumes–rhizobia Accurately estimating global N2 fixation for the symbioses of the forage and fodder legumes is challenging because statistics on the areas and productivity of these legumes are almost impossible

to obtain. Smil (1999) reported 100–120 Mha of land in fodder and forage legumes and green manure crops. He assumed average annual N2 fixation rates of 200 kg N/ha for alfalfa, 150 kg N/ha for the clovers (Trifolium spp.), 100 kg N/ha for other forages and 50 kg N/ha for legume–grass pastures. Thus, total N2

Table 5 Comparing estimates of N2 fixation/unit area (kg/ha) by Smil (1999) with estimates calculated from legume global areas (Table 4, column 2) and crop N fixed (Table 4, column 8) Legume

Common bean Chickpea Pea Lentil Fababean Other pulses Groundnut Soybean

Smil (1999) ranges of values (kg N/ha/year) Minimum

Mean

Maximum

30 40 30 30 80 40 60 60

40 50 40 40 100 60 80 80

50 60 50 50 120 80 100 100

Calculated from Table 4 (kg N/ha/year) 23 58 86 51 107 41 88 176

Plant Soil (2008) 311:1–18

11

fixation for the forage and fodder legumes was calculated at 12 Tg annually (average of about 110 kg N/ha/year) (Table 6). A substantial body of work in Australia and northern Europe shows that forage/fodder legumes have an average Ndfa value of about 70% and 25 kg N is fixed in the shoots for every Mg shoot biomass produced (Peoples and Baldock 2001; Carlsson and Huss-Danell 2003). It should be noted that Peoples and Baldock (2001) reported wide variations for this value, ranging 8–53 kg shoot N fixed/Mg shoot biomass. Such variation would have been caused by differences in soil nitrate levels and pasture vigour, as well as species differences in foliage-N content, experimental treatment and error. Assuming 50% of forage legume nitrogen is below-ground (McNeill et al. 1997; Peoples and Baldock 2001), the overall average for N2 fixation by forage legumes becomes 50 kg N fixed/Mg shoot biomass. Smil (1999) estimated global shoot productivity of the forages at 500 Tg from the 100–120 Mha, equivalent to 4.2–5.0 Mg/ha. Global N fixed by the

forage and fodder legumes can be calculated by combing the overall annual production of 500 Tg with the rate of N2 fixation per unit of forage (50 kg N fixed/Mg shoot biomass). Thus, a value of 25 Tg N/annually is obtained, a value about double that of Smil (1999). The same value of 25 Tg N can be calculated if the following figures and assumptions are used: globally 110 Mha legumes with an average Ndfa of 70%, average shoot DM production of 4.5 Mg/ha, shoot N concentration of 3.6% and below-ground N of 50%. Thus, average annual N2 fixation is calculated at 227 kg/ha and global N2 fixation at 25 Tg. So, what is a realistic figure for N2 fixation by the forage and fodder legumes in agricultural systems? The Smil (1999) figure of 12 Tg annually may be low because it does not reasonably account for belowground N, but without reliable data on global forage and fodder legume areas and production statistics for those areas, it is impossible to provide an alternative. The real figure may lie somewhere between 12 and 25 Tg annually (Table 6).

Table 6 Summary of estimates of N fixed annually in agricultural systems by rhizobia in symbiosis with crop, pasture and fodder legumes, numerous genera of bacteria associated with non-leguminous species and free-living bacteria Areaa (Mha)

Rate of N2 fixation (kg N/ha/year)

Crop N fixed (Tg/year)

Comments on validity of global N2 fixation estimates

186

115

21

Pasture and fodder legumes

110

110–227

12–25

Azolla– cyanobacteria, cyanobacteria Endophytic, associative & free-living bacteria

Rice

150

33

5

20

25

0.5

Endophytic, associative & free-living bacteria

Crop lands other than used for legumes and rice

800