Estimation of nitrous oxide, nitric oxide and ... - Wiley Online Library

Global Change Biology (2003) 9, 1080±1096

Estimation of nitrous oxide, nitric oxide and ammonia emissions from croplands in East, Southeast and South Asia X I A O Y U A N Y A N , H A J I M E A K I M O T O and T O S H I M A S A O H A R A Frontier Research System for Global Change, Yokohama 236-0001, Japan

Abstract Agricultural activities have greatly altered the global nitrogen (N) cycle and produced nitrogenous gases of environmental significance. More than half of all chemical N fertilizer produced globally is used in crop production in East, Southeast and South Asia, where rice is central to nutrition. Emissions of nitrous oxide (N2O), nitric oxide (NO) and ammonia (NH3) from croplands in this region were estimated by considering background emission and emissions resulting from N added to croplands, including chemical N, animal manure, biologically fixed N and N in crop residues returned to fields. Background emission fluxes of N2O and NO from croplands were estimated to be 1.22 and 0.57 kg N ha21 yr21, respectively. Separate fertilizer-induced emission factors were estimated for upland fields and rice fields. Total N2O emission from croplands in the study region was estimated to be 1.19 Tg N yr21, with 43% contributed by background emissions. The average fertilizer-induced N2O emission, however, accounts for only 0.93% of the applied N, which is less than the default IPCC value of 1.25%, because of the low emission factor from paddy fields. Total NO emission was 591 Gg N yr21 in the study region, with 40% from background emissions. The average fertilizer-induced NO emission factor was 0.48%. Total NH3 emission was estimated to be 11.8 Tg N yr21. The use of urea and ammonium bicarbonate and the cultivation of rice led to a high average NH3 loss rate from chemical N fertilizer in the study region. Emissions were displayed at a 0.5Ê 3 0.5Ê resolution with the use of a global landuse database. Keywords: agricultural soil, ammonia, background emission, nitric oxide, nitrogen fertilizer, nitrous oxide Received 21 November 2002; revised version received 3 March 2003 and accepted 5 March 2003

Introduction Human activity has greatly altered the global nitrogen (N) cycle by fixing as much, or more, N, via energy production, fertilizer production and crop cultivation, than is fixed by natural processes (Galloway et al., 1995). A major consequence of this human-driven change is the increased emission of N-based trace gases, such as nitrous oxide (N2O), nitrogen oxides (NO and NO2, together denoted as NOx) and ammonia (NH3), which impact regional and global atmospheric chemistry. Nitrous oxide is very effective at trapping heat in the atmosphere. In the stratosphere, it triggers reactions that lead Correspondence: Xiaoyuan Yan, fax 81-45-778-5496, e-mail: [email protected]

1080

to the destruction of the ozone layer, which shields the Earth from damaging ultraviolet (UV) radiation (Crutzen, 1970). Nitrogen oxides play an important role in the chemistry of the lower atmosphere by catalyzing the photochemical formation of ground-level ozone (Crutzen, 1979), causing detrimental effects with regard to human health and crop productivity (Bolle et al., 1986; Benton et al., 2000). Nitrogen oxides are also central to the photochemical formation of NO, a component of acid rain (Crutzen, 1979). Ammonia is the only natural alkaline gas in the atmosphere and thus is an important neutralizer of anthropogenic acidity (Brasseur et al., 1999). Many anthropogenic emissions of nitrogenous gases are associated with crop production. Most chemical N fertilizer, the biggest anthropogenic N input, is applied ß 2003 Blackwell Publishing Ltd

N I T R O G E N O U S G A S E M I S S I O N S F R O M C R O P L A N D S 1081 to croplands in order to increase productivity. Emissions are further increased by the cultivation of N-fixing crops, and by tillage and other agricultural activities that speed up the release of N from long-term storage in soils and organic matter. Agricultural soil has been estimated as giving rise to half of the total global anthropogenic emission of N2O (Mosier et al., 1998; Olivier et al., 1998; Kroeze et al., 1999), but its contribution to NOx emission is much less certain. Using an empirical model, Yienger & Levy II (1995) estimated a global emission of 2.25 Tg NOx N yr 1. Davidson & Kingerlee (1997) estimated an emission of 3.9 Tg NOx N yr 1 by scaling up field flux measurements. Both estimates considered canopy effects. Because the source strength of NOx from fossil fuels varies greatly according to region, agricultural soils could be the dominant source of NOx in certain regions. Of the total global anthropogenic NH3 emission of 43 Tg N yr 1, 12.6 Tg N is from croplands, not including emission from animal manure spread on croplands (Olivier et al., 1998). East, Southeast and South Asia make up one of the most densely populated areas in the world. Collectively, these regions contain 36% of the world's total crop harvest area and are responsible for more than half of the total global chemical N fertilizer consumption, as of 1999 (FAOSTAT, 2002). Agriculture in these regions is unique; rice, central to nutrition in Asia, is a very important crop. Production and harvest area of rice in these regions account for almost 90% of the total world rice production and harvest (FAOSTAT, 2002). Urea and ammonium bicarbonate account for 85% of the total chemical N fertilizers, which is unlike the situation in the rest of the world. Although current field data are not enough to quantify N2O, NOx and NH3 emissions per crop, they suggest that the emission fluxes of these gases from upland fields could differ significantly from those of paddy fields (Xing, 1998). Emissions, especially of NH3, are also influenced by fertilizer type (Asman, 1992; Xing & Zhu, 2000). This study estimated cropland N2O, NOx and NH3 emissions per country in East, Southeast and South Asia, and at provincial or state levels in China and in India. Emissions from uplands and paddy fields were distinguished. Ammonia emissions from different fertilizer types were also distinguished.

Estimation method Emission sources considered N2O emission Nitrous oxide emission from croplands was estimated primarily following IPCC Guidelines (IPCC, 1997); that is, an emission factor was multiplied by various N inputs. Nitrogen sources considered were ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

synthetic N fertilizer (FSN), animal excreta used as fertilizer (FAW), biologically fixed N (FBN) and N from crop residue returned to croplands (FCR). Owing to the limited availability of data on the area of Histosols under cultivation, and the relatively small proportion of N2O emission from Histosols (less than 5% of the direct global emissions from agricultural soils (Mosier et al., 1998)), these N2O emissions were not considered in the current estimate. Background emission ± that is, emission when no external nitrogen is added ± was considered. In order to calculate the amount of N input, some modifications were made to the IPCC method. In the IPCC Guidelines (IPCC, 1997), FSN is calculated as the total N fertilizer consumption minus the fraction volatilized as NH3 and NOx, which has a default value of 10%. As the fertilizer-induced N2O emission factor is based on the actual amount of fertilizer applied, rather than on the amount of fertilizer remaining on soil after NH3 volatilization, this N was not subtracted from total fertilizer consumption in the calculations for this study. The calculation for FAW was slightly modified from the IPCC methods, as shown in the following equation: X FAW Nt Next 1 FracGRAZt FracFUEL t 1

FracGASM

where Nt is the population size of animal t in a prescribed area, Next is the annual N excretion rate of animal t, FracGRAZt is the fraction of manure directly deposited on soil during grazing for animal t, FracFUEL is the fraction burned for fuel and FracGASM is the fraction volatilized as NH3 and NOx before being applied to fields. Default values for FracFUEL and FracGASM are 0 and 0.2, respectively. Values of Nex and FracGRAZ for Asia were taken from IPCC (1997). The IPCC Guidelines for the calculation of FBN consider the amount of N contained in the aboveground plant material of N-fixing crops as a reasonable proxy for the total amount of N fixed by the crop. However, the N contents of seed and residue differ (see Table 1). The following equation therefore ± which distinguishes between N contained in crop product and crop

Table 1 Parameters for estimating nitrogen (N) fixed by crops Crop

Ratio* FracDM

Soybean Beans Peas Groundnut

2.1 2.1 1.5 1

0.93 0.89 0.87 0.96

Si

{

FracN 0.0693 0.0397 0.0398 0.0470

{ Si

FracDM 0.86 0.86 0.87 0.86

Ri*

FracN

Ri*

0.0230 0.0150{ 0.0142 0.0106

*Data from IPCC (2000). { Values cited or calculated from Summerfield & Roberts (1985).

1082 X . Y A N et al. residue ± was used in this study in order to calculate N fixed by crops. X Cropi FracDM Si FracN Si Cropi FBN i Ratioi FracDM

Ri

FracN

Ri

where Cropi and Ratioi are the production and residue/ crop ratio of N-fixing crop i, FracDM±Si and FracN±Si are dry matter content and N content in crop i and FracDM±Ri and FracN±Ri are dry matter content and N content in residue of crop i. The method, in IPCC Good Practice Guidance, for calculating N returned to soils in crop residue is rather complex; it specifies, per crop, the fractions of residue burned in the field, burned as fuel, used for construction, and used as fodder (IPCC, 2000). In this study, the fraction of residue returned to soils was directly specified per crop. Thus, the equation was modified to be: X FBN Cropj Ratioj FracDM Rj FracN Rj j FracRETURNj where Cropj is the production of crop j (including both Nfixing and non-fixing crops); Ratioj is the mass ratio of residue/crop for crop j; FracDM±Rj and FracN±Rj are the dry matter fraction and N fraction in residue of crop j, respectively, and FracRETURNj is the fraction of residue returned to the field for crop j. Parameters for each crop are listed in Table 2. For crops not included, it was assumed that no residue was returned to soils as fertilizer.

Table 2

Parameters for estimating FCR*

Crop

Residue/seed

FracDM±R

FracN±R

FracRETURN

Rice Wheat Maize Barley Oat Beans Peas Peanut Rapeseed Sugarcane Sugarbeet Potato

1.4 1.3 1.0 1.2 1.3 2.1 1.5 1.0 2.5{ 0.8{ 0.2** 0.4

0.85 0.85 0.78 0.85 0.92 0.86 0.87 0.86 0.80§ 0.62{ 0.2** 0.6**

0.0067 0.0028 0.0081 0.0043 0.007 0.015{ 0.0142 0.0106 0.0067 0.004 0.018{ 0.011

0.3{ 0.45{ 0.2{ 0.45{ 0.45{ 0.8{ 0.8{ 0.42{{ 0.4{ 0.42{{ 0.42{{ 0.42{{

*Data from IPCC (2000) as otherwise indicated. { Data from Xing & Yan (1999). { Calculated from Summerfield & Roberts (1985). § From Lardy & Anderson (1999). { From EPA (2002). **From IPCC (1997). {{ From Li et al. (1998).

NOx emission Nitrogen oxides are mainly released from soils in the form of NO (Conrad, 1990), thus only NO will be discussed. Nitric oxide and N2O are primarily produced biologically, as intermediate products of nitrification and denitrification. A simple conceptual model relates the sum of NO and N2O production to indices of N availability; that is, NH4 and NO3± in soils, and relates the ratio of NO to N2O production to environmental factors, such as soil water content (Firestone & Davidson, 1989; Davidson & Verchot, 2000). Therefore, the same method was used in order to estimate NO emission from croplands as was used for N2O. The background emission rate of NO and the fertilizer-induced NO emission factor, however, were different from those of N2O, as will be discussed later. NH3 volatilization The following N sources were considered for NH3 volatilization from croplands: synthetic N fertilizer, animal manure used as fertilizer and N fixed by N-fixing crops. The sizes of the N sources were the same as for N2O emission estimates. Emission factors of NH3, unlike N2O, varied with N source, however. Background emission levels of NH3 from croplands are not known for certain, because plants can act both as a source and as a sink of NH3 (Farquhar et al., 1983; Schjùrring, 1991). After reviewing a large number of reports, however, Holtan-Hartwig & Bùckman (1994) concluded that crops are net emitters of NH3 to the atmosphere and suggested that net losses of NH3 from arable crops in temperate regions to the atmosphere are approximately 1±2 kg NH3±N ha 1 yr 1. This value is, therefore, considered as background NH3 emission from croplands. Ammonia emission rates from decomposing crop residue are also not certain and there are not enough data to make a reliable estimate. Whitehead and Lockyer (1989) have shown that NH3 was emitted only from the decomposition of herbage with a high N content, whereas others have shown significant NH3 emission from decomposing Brassica leaves (Sutton et al., 1996). In the current estimate, the emission from decomposing crop residue, if any, was included in the background emission.

Determination of fertilizer-induced emission factors and background emission rates by surveying published data N2O from uplands The default IPCC fertilizer-induced N2O emission factor of 1.25% was primarily based on the regression analysis of Bouwman (1996). In order to determine the effect of fertilization on N2O emission in an experiment, unfertilized plot is usually required. We have assembled most of the recent reports that have included unfertilized treatments. The subsequent database was combined with Bouwman's; after removing ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

N I T R O G E N O U S G A S E M I S S I O N S F R O M C R O P L A N D S 1083 experiments performed on unplanted soil, organic soil, N-fixing crop fields, paddy rice fields and with organic N input, a data set remained of 134 data points from 34 reports of fertilizer-induced N2O emission (data not shown). Fertilization rates were from 35 to 880 kg N ha 1. Measurement periods were from 7 to 840 days, with more than 90% of measurement periods longer than 1 month. Fertilizer-induced N2O emission varied from 1.17 to 12%, with an average of 1.19%. The average fertilizerinduced N2O emission, for experiments that lasted at least 1 year, was 1.31% (n 23), slightly higher than the average from all measurements. As both values are close to the IPCC default emission factor of 1.25% (IPCC, 1997), the IPCC value was adopted as the N2O emission factor for fertilizers applied to uplands. Bouwman (1996) derived a background emission rate of 1.0 kg N2O±N ha 1 yr 1, based on five measurements from unfertilized grasslands, which was highly uncertain, as the author acknowledged. For the current study, available data on N2O emissions from unfertilized croplands was compiled (Table 3). On an unfertilized winter wheat field, RoÈver et al. (1998) measured 0.41 kg N2O± N ha 1 in the growing season and 1.43 kg N2O±N ha 1 in the following fallow season. Pulses of N2O emission are often driven by rainfall and thaws (Wagner-Riddle et al., 1997; Dobbie et al., 1999). These results suggest that it is necessary, when estimating background N2O emission, to consider emissions over the entire year, rather than only during a cropping season. There were two ways, therefore, to estimate background N2O emission. One was to average those annual emission measurements that spanned at least a full year. The average of the seven such measurements available (Table 3) was 1.22 kg N2O±N ha 1 yr 1, with a coefficient variance of 53%. The other method was to view emission from unfertilized soil as a function of time and to calculate emission on the basis of 365 days. As shown in Fig. 1, background emission did relate linearly to measurement time, with a slope of 0.0041 kg N ha 1 day 1, which resulted in an annual emission of 1.5 kg N ha 1. The latter measurement is understandably higher than the former, because it is an extrapolation of seasonal emission, generally the cropping season, to the entire year; background emission in the cropping season could be higher than in the fallow season because agricultural activities accelerate the mineralization of organic N. The former measurement, 1.22 kg N ha 1 yr 1, was, therefore, adopted in this study as the estimate of background N2O emission from uplands. N2O from paddy fields Paddy fields differ from upland fields in that the soil is generally flooded, a condition unfavourable to nitrification that leads to low N2O production ratios in denitrification products. Early ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

studies reported that N2O emission from paddy fields was negligible (Denmead et al., 1979; Freney et al., 1981). However, rather than being flooded for the entire ricegrowing season, many paddy fields are aerated at least once during the cropping season, resulting in significant N2O emission. Many studies have examined N2O emission from fertilized paddy fields, but only a part of them has included unfertilized plots. As unfertilized data are needed to derive fertilizer-induced emissions, we compiled only those experiments that included unfertilized plots (Table 4). The average fertilizer-induced emission factor was 0.25% (+ 0.24%, n 21). All these studies covered a full rice-growing season; therefore, this value was used for fertilizer-induced N2O emission from paddy fields in the growing season. The average N2O emission from unfertilized paddy fields in the cropping season was 0.26 kg N ha 1 (calculated from data in Table 4), with an average season lasting 117 days. If emission flux in the fallow season were equal to the flux in the cropping season, then the annual N2O emission from unfertilized plots would be 0.81 kg N ha 1. However, many researchers have found that N2O flux from fertilized paddy fields in the fallow season is much higher than N2O flux in the cropping season (Bronson et al., 1997a, b; Tsuruta et al., 1997; Abao et al., 2000). If this pattern were also valid for unfertilized fields, then background N2O emission from paddy fields would be comparable to that from uplands. Moreover, paddy fields cannot be completely differentiated from uplands, because some fields are rotated with rice and upland crops in the same year. For these two reasons, the same background N2O emission rate was assigned to both paddy fields and uplands. NO emission from uplands A recently derived mean fertilizer-induced NO emission factor of 0.66% for upland soils, based on statistical analysis of individual field reports of fertilizer-induced NO emissions (Yan et al., submitted), was used in the current estimate. Only five field measurements of NO emission from unfertilized non-leguminous croplands were available (Table 3) and none of these measurements spanned a full year; therefore, the average flux of these five measurements, 6.55 mg NO±N m 2 h 1, was extrapolated to an entire year; the result was an annual emission of 0.57 kg N ha 1 yr 1. This linear extrapolation could lead to an overestimation, because N transformation in the cropping season is more intense than in the fallow season. However, the extrapolation could also lead to an underestimation of emissions, because crops compete with microorganisms for available N. However, this value was used for background NO emission from uplands, as not enough field measurements were available to derive a better estimate.

1084 X . Y A N et al. Table 3

Measurements of N2O, NO emissions from unfertilized uplands

Location

Crop/treatment

Sample period, day

NC, USA Rothamsted, UK Oxon, UK Oxon, UK Oxon, UK Oxon, UK Edinburgh, UK Ontario, Canada New York, USA Colorado, USA Colorado, USA Colorado, USA Edinburgh, UK Tsukuba, Japan Colorado, USA Tsukuba, Japan Tennessee, USA Modena, Italy Chalons, France Longchamp, France Messigny, France Colorado, USA Colorado, USA Ottawa, Canada Tennessee, USA Costa Rica Costa Rica Costa Rica Costa Rica Lethbridge, Canada Yucheng, China Hebei, China Braunschweig, Germany Nebraska, USA Nebraska, USA Nebraska, USA

Soybean Wheat Wheat Wheat Wheat Wheat Oilseed rape Maize Maize Barley Maize Maize Wheat Carrot Maize Maize No-till corn Maize Oilseed rape Oilseed rape Oilseed rape Barley Barley Maize No-till corn Papaya Taro Balsa Maize/taro Barley Maize-wheat Wheat-maize Wheat

N2O 24 25 28 28 30 31 54 80 85 86 97 97 104 116 120 123 129 150 150 150 150 153 153 185 210 298 298 348 348 365 365 365 365

Colorado, USA Tsukuba, Japan Tennessee, USA Sweden Tennessee, USA

Wheat Maize Maize Barley Maize

Wheat Wheat Wheat

840 840 840 NO 63 123 129 150 210

Flux, mg Nm

4.22 46.44

37.44 12.2 14.3 23.4 10.4

7.6 0.89 11.16 2.28 12.6

2

h

1

Seasonal emission, kg N ha 1

Reference*

0.23 0.02 0.07 0.07 0.07 1.49 0.32 0.1 0.3 0.45 0.11 0.12 0.05 0.08 2.23 0.13 1.43 0.65 0.01 0.01 0.07 0.52 0.82 0.3 1.98 1.07 1.25 2.05 0.91 0.7 0.1 1.03 1.84

1 2 3 3 3 3 4 5 6 7 8 8 4 9 7 10 11 12 13 13 13 14 14 15 16 17 17 17 17 18 19 20 21

3.16 3.94 4.05

22 22 22

0.03 0.34 0.63

23 10 11 24 16

*1, Whalen et al. (2000); 2, Harrison et al. (1995); 3, Colbourm & Harper (1987); 4, Baggs et al. (2000); 5, McKenney et al. (1980); 6, Duxbury & McConnughey (1986); 7, Mosier et al. (1986); 8, Bronson et al. (1992); 9, Minami (1990); 10, Yan et al. (2001); 11, Thornton et al. (1996); 12, Arcara et al. (1999); 13, Henault et al. (1998); 14, Mosier et al. (1982); 15, Lessard et al. (1996); 16, Thornton & Valente (1996); 17, Weitz et al. (2001); 18, Chang et al. (1998); 19, Dong et al. (2001); 20, Song et al. (1997); 21, RoÈver et al. (1998); 22, Kessavalou et al. (1998); 23, Andserson & Levine (1987); 24, Johansson & Granat (1984).

ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

N I T R O G E N O U S G A S E M I S S I O N S F R O M C R O P L A N D S 1085

Seasonal emission (kg N ha−1)

NO emission from paddy soils The only available field measurement of NO emission from paddy fields (Galbally et al., 1987) had a negligible flux. This measurement, 5 y = 0.0041x

4

R 2= 0 . 6 8 7 6 3 2 1 0

0

200

400 600 Measurement days

800

1000

Fig. 1 N2O emission from unfertilized agricultural soils. Data are from Table 3. Experiments performed on organic soil, unplanted soil, N-fixing crop field, or with crop residue incorporated are excluded. Table 4

however, only spanned several days in the flooding period; many paddy fields go through dry-wet cycles, during which intense nitrification±denitrification occurs. Although field measurements are lacking, NO emission from paddy fields during the rice-growing season was assumed to be non-negligible for the following reasons. Firstly, both N2O and NO are intermediate products of nitrification±denitrification and significant N2O emissions have been observed during the dry-wet cycles of paddy soils (Yan et al., 2000). Secondly, emission peaks of NO from upland fields have been observed after wetting dry soil (Anderson & Levine, 1987). Finally, in a laboratory incubation of paddy soil, Kodo et al. (2002) found significant NO emission. The fertilizer-induced emission factor for N2O from paddy fields is one-fifths that of upland fields. Assuming this also applies to NO emission, this results in a fertilizer-induced emission factor of 0.13% for NO from paddy fields. As with background N2O emission, background NO emission from

Field measurements of N2O emission from paddy fields as affected by nitrogen (N) fertilization*

Location

Season days

New Delhi, India New Delhi, India New Delhi, India New Delhi, India New Delhi, India New Delhi, India Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Nanjing, China Nanjing, China Nanjing, China Nanjing, China Nanjing, China Louisiana, USA Louisiana, USA Louisiana, USA Louisiana, USA Louisiana, USA

98 98 98 98 98 98 126 126 126 126 126 126 126 126 126 126 126 126 112 112 112 112 112 110 110 110 110 110

N type{

U AS U AS U U U U U U U U U U U AS AS U U U U U

N rate, kg N ha 0 140 140 140 140 140 0 86 86 0 86 86 0 86 86 0 86 86 0 100 300 100 300 0 90 180 90 180

1

N2O flux, mg Nm 2 h

11.42 23.52 32.8 9.45 23.13 26.51 15.41 22.41 25.81 14.94 33.7 31.24 5.1 6.3 23 6.5 36.5

1

Seasonal emission, kg N ha 1 0.05 0.16 0.24 0.14 0.17 0.15 0.35 0.7 0.99 0.29 0.70 0.80 0.47 0.68 0.78 0.45 1.02 0.95 0.14 0.17 0.62 0.18 0.98 0.07 0.10 0.17 0.11 0.09

*Only those that included unfertilized plots are listed. { U, urea; AS, ammonium sulfate; AB, ammonium bicarbonate; SN, sodium nitrate. { 1, Kumar et al. (2000); 2, Suratno et al. (1998); 3, Cai et al. (1997); 4, Smith et al. (1982). ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

Fertilizer-induced N2O±N, %

0.086 0.135 0.068 0.091 0.072 0.425 0.752 0.481 0.599 0.246 0.366 0.659 0.573 0.033 0.16 0.038 0.28 0.04 0.05 0.04 0.01

Reference{ 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4

1086 X . Y A N et al. topdressed to floodwaters. Table 5 shows that the average NH3 loss rate from urea incorporated into topsoil is 20% and that from topdressed urea is much higher, up to 36%. However, all data in Table 5 were obtained during the rice-transplanting period; about 40% of N fertilizer is applied at the time of panicle initiation. Studies have consistently demonstrated that NH3 loss from fertilizer topdressed at panicle initiation is much less than the loss from fertilizer topdressed at the time of rice transplanting (Fillery et al., 1984, 1986; Humphreys et al., 1988; Dhyani & Mishra, 1992). Field experiments performed in the Philippines and Australia have shown that the emission rate at panicle initiation is approximately one-third of that during the transplanting period (Table 6). As the NH3 loss rate from urea topdressed at transplanting is 36% of the total (Table 5), the emission factor of urea topdressed at panicle initiation was assumed to be 12%. By assuming that 30% of fertilizer is incorporated into topsoil as basal fertilizer, 30% topdressed at time of transplanting and 40% topdressed at panicle initiation (Xing & Zhu, 2000), the overall average NH3 emission from urea applied to rice fields was estimated to be 22% of the total. For upland fields at the transplanting stage, the average NH3 emission from incorporated urea was 11.5% of the total and the average from surface applied urea was 23.5% (Table 5). As with rice fields, because of the rapid N uptake by plant roots and the reduced wind speed at the soil surface in later growth stages, it was supposed that NH3 loss from fertilizer topdressed at later stages was also less than loss from fertilizer applied at the time

paddy fields was assumed to be the same as that from uplands. NH3 emission factor A significant part of the synthetic N fertilizer applied to agricultural soils is lost through NH3 volatilization. The loss rate strongly depends on fertilizer type, timing and method of fertilization, soil properties, and meteorological conditions. While data are insufficient to quantify the effects of all influencing factors, there are relatively better data for distinguishing the effects of fertilizer type. In the study area, urea and ammonium bicarbonate are the dominant N fertilizers. According to FAOSTAT (2002), in 1995, urea accounted for approximately 62% of the total chemical N fertilizer consumption in the study area and ammonium bicarbonate (NH4HCO3) accounted for approximately 23%. We derived emission factors for these two fertilizers based on experimental data from the area. For other N fertilizers, we adopted emission factors from the European emission inventory guidebook (EEA, 1999), or used the emission factor for ammonium bicarbonate as a default. While many field experiments have been performed in order to measure NH3 emission from urea applied to croplands in this area, only a limited number of them used micrometeorological methods, thought to be the only acceptable method for quantifying NH3 loss from fields. Emission factors were derived from those measurements (Table 5). In rice fields, fertilizer is either incorporated into topsoil and subsequently flooded or

Table 5

Ammonia (NH3) emission from urea applied to rice fields and uplands (% of total N applied) Application method

Site

Incorporation

Munoz, Phillipines Los Banos, Phillipnes Munoz, Philippines Laguna, Philippines Laguna, Philippines Laguna, Philippines Zambles, Philippines Fenqiu, China Danyang, China Fuyang, China Yingtan, China Avereage

15 13

Shanghai, China Fengqiu, China Average

11 12 11.5

18 28 30 9 7 40 20

Topdressing Rice field 47 27 36 54 46 27 36

11 36 Upland 17 30 23.5

Reference

Fillery et al. (1984, 1986) Fillery et al. (1984, 1986) Fillery & De Datta (1986) De Datta et al. (1991) De Datta et al. (1991) Obcemea et al. (1988) Obcemea et al. (1988) Zhu et al. (1989) Cai et al. (1986) Cai et al. (1992a) Cai et al. (1992b)

Xi et al. (1987) Zhang et al. (1992)


N I T R O G E N O U S G A S E M I S S I O N S F R O M C R O P L A N D S 1087 Table 6

Comparison of NH3 volatilization from nitrogen fertilizer applied at transplanting and panicle initiation periods (%)

Site

Transplanting

Panicle initiation

Reference

Munoz, Philippines Los Banos, Philippines Average

47 27 37

10 15 12.5

Fillery et al. (1984, 1986) Fillery et al. (1984, 1986)

Australia Australia Australia Average

11 21 18 16.6

3 8

Humphreys et al. (1988) Humphreys et al. (1988) Humphreys et al. (1988)

of transplanting. Bian et al. (1997) observed a very low NH3 emission rate following the topdressing of N fertilizer to wheat in later stage; therefore, it is also assumed that NH3 emission from urea topdressed in later stages (surface applied) is one-third of the emission when topdressed at the early stages. By assuming a similar allocation of fertilizer to early and later stages for upland fields and for rice fields, we estimated a NH3 emission factor of 13.7% for urea applied to upland fields. Ammonium bicarbonate accounts for about 40% of the total N fertilizer consumption in China; two field studies have compared its effect on NH3 emissions from rice fields to that of urea (Cai et al., 1986; Zhu et al., 1989). The NH3 emission factor for NH4HCO3 was approximately 1.5 times greater than that from urea; therefore, an emission factor of 33% was assigned to NH4HCO3 applied to rice fields and a factor of 20.5% was assigned to NH4HCO3 applied to upland fields. Fillery & De Datta (1986) found that while the patterns of NH3 loss, in a paddy field, from urea and ammonium sulfate differed, the accumulated losses were similar. Therefore, the same emission factors were used in this study for both urea and ammonium sulfate. For other N fertilizers, the emission factors of the simple methodology in EEA (1999) were adopted, without distinguishing between upland and paddy fields, because of the relatively low consumption of these fertilizers. For animal excreta applied to croplands as fertilizer, distinctions were not made between paddy fields and uplands. Emission factors were adopted from EEA (1999), in which, because of differences in mineral N content of manure from different animal types, manure from cattle, pigs and poultry was assigned a NH3 volatilization rate of 20%, and manure from other animals was assigned a rate of 10%. Leguminous crops fix atmospheric N; this N could also be volatilized. Furthermore, the residue of leguminous crops contains higher N levels than that of other crops. Whitehead and Lockyer (1989) demonstrated that no NH3 is emitted during the decomposition of herbage with low N content, but NH3 is emitted during the ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

5.5

decomposition of herbage with higher N concentrations. Thus, the EEA (1999) method for estimating NH3 emission from leguminous crops was adopted; emissions are 1% of the fixed N. As mentioned above, background emission of NH3 from croplands is approximately 1±2 kg N ha 1 yr 1. The value 1.5 kg N ha 1 yr 1 is used here. All emission factors for NH3 are summarized in Table 7.

Emissions estimates In order to summarize, per area, N2O emission from croplands per year was estimated to be: 1:22kg N ha 1 Area 1:25% FSNnon FAWnon r FCRnon r 0:25%

r

FBN

r

FBN

FSNr FAWr FCRr : NO emission was estimated to be: 0:57kg N ha 1 Area 0:66% FSNnon FAWnon r FCRnon r 0:13% FSNr FAWr FCRr : where Area is the total area of cropland, FSNnon r, FAWnon r and FCRnon r are chemical fertilizer N, animal manure N and crop residue N used on non-rice crops, respectively, and FSNr, FAWr and FCRr are the corresponding nitrogens used on rice fields. Ammonia emission was estimated by summing the products of N sources and emission factors listed in Table 7. Animal manure and crop residue were assumed as evenly applied to all crop areas. In order to allocate synthetic N fertilizer to rice and non-rice fields, it was necessary to know the fertilization rate for rice and non-rice crops. Nitrogen fertilization rates for major crops, for most of the countries in the study area, are available in IFA/IFDC/FAO (1999), but the data for different countries are not all for 1995. For some countries, total consumption of synthetic N fertilizer

1088 X . Y A N et al. Table 7 Parameters for estimating NH3 emission factors from croplands Source

NH3 emission factor (%)

Ammonium nitrate Ammonium phosphate Ammonium sulfate, applied to paddy fields Ammonium sulfate, applied to uplands Calcium ammonium nitrate Other complex nitrogen fertilizers Urea, applied to paddy fields Urea, applied to uplands Ammonium bicarbonate, applied to paddy fields Ammonium bicarbonate, applied to uplands Manure nitrogen used as fertilizer for cattle, pigs and poultry Manure nitrogen used as fertilizer for other animals Nitrogen fixed by leguminous crops Background emission

2* 5* 23.5{ 13.7{ 2* 2* 23.5{ 13.7{ 34.5{ 20.5{ 20* 10* 1* 1.5 kg N ha

1

yr

1

*Adopted from EEA (1999). { The same value as for urea. { Estimated for this study.

varies greatly from year to year, thus, fertilization rate should also change. In this study, the ratio of N fertilization rate for non-rice to rice crops, per country, was assumed to not change with time. The ratios were derived from IFA/IFDC/FAO (1999), for those countries that have significant consumption of synthetic N fertilizer, with the exception of North Korea. Information on fertilizer application rate for North Korea is not available in IFA/IFDC/FAO (1999); a ratio of 0.6 was used, which is the average ratio of N fertilization rate for non-rice to rice for the developing countries in the study area. The N fertilization rate for rice crops was then calculated using the following equation: Frice FSN=Arice R Atotal Arice where Frice is the N fertilization rate for rice; FSN is the total consumption of synthetic N fertilizer in a country, province, or state in 1995; Atotal and Arice are the total crop cultivation area and rice cultivation area of a country, province, or state in 1995; and R is the ratio of N fertilization rate for non-rice to rice, derived from IFA/ IFDC/FAO (1999). Activity data needed in the estimation included information on N consumption, animal populations, crop production, cropland area (arable land plus permanent crops in the FAO database), total crop cultivation area and rice cultivation area of 1995. All the data were downloaded from the FAO website, except the data for China and India. Data for each province of China were obtained from the China Agriculture Yearbook 1995 (Editorial Board of China Agriculture Yearbook, 1996) and the Taiwan Agriculture Yearbook (1997). Data for each state

of India were obtained from Indian Agriculture in Brief 27th Edition (2000) and Fertiliser and Allied Agricultural Statistics, 1995±96 (1996). Data for each animal type per Indian state were not available for 1995. The 1995 values were estimated based on the detailed data of 1992 and the total animal population of 1995, by assuming linear growth.

Results The estimated total N2O emission from croplands in the study area in 1995 was 1.22 Tg N. Of those countries with significant N2O emission (> 10 Gg N yr 1), China was the only country where synthetic fertilizer was the dominant source. In other countries, background emission was the dominant source (Table 8, Fig. 2). On average, background emission contributed 43% of the total N emission. Fertilizer-induced N2O emission amounted to 356 Gg N, 30% of the total emitted N2O, or 0.93% of the total consumption of 38.6 Tg synthetic fertilizer N in the area. Animal manure used as fertilizer resulted in a N2O emission of 233 Gg N. Crop residue returned to fields and the cultivation of leguminous crops contributed only 3 and 5%, respectively, to the total N2O emission. Total NO emission from croplands in the study area in 1995 was estimated to be 591 Gg N, nearly half the level of N2O emission. About 40% of the emission was from background emission, 32% was as a result of the application of synthetic N fertilizer and 21% was as a result of the use of animal manure as fertilizer. On average, fertilizer-induced NO emission accounted for 0.48% of the total synthetic N consumed in the area. ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

N I T R O G E N O U S G A S E M I S S I O N S F R O M C R O P L A N D S 1089 Table 8 Emissions of N2O, NO and NH3 from croplands in East, Southeast, and South Asian countries, estimated for 1995, by source (Gg N)

*Background emission of NH3 includes emission from decomposition of returned crop residue.

Emission of NH3 from croplands was 12 Tg N in 1995. In contrast to N2O and NO, background emission of NH3 contributed only 5% to the total emission; synthetic fertilizer and animal manure contributed 54 and 40%, respectively. This indicates that NH3 volatilization, a physical±chemical process, is more sensitive to the addition of external N than are the emission of N2O and NO, as microbiological processes. The overall NH3 emission ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

2.5

NO

(b)

NH3

(c)

2.0 1.5 1.0 0.5 0.0 60 50 40 30 20 10

Average

Others

Myanmar

Philippines

Japan

Thailand

0 Vietnam

NH3 emission 3559.8 22.1 2036.1 5774.3 1449.0 18.8 1568.3 3290.4 353.8 2.9 146.5 548.5 254.2 0.6 203.9 491.1 192.1 0.3 135.8 340.4 158.9 0.4 79.6 249.1 78.5 0.8 73.1 183.0 59.2 0.3 69.1 136.1 50.2 0.1 51.2 116.3 23.2 0.9 67.6 107.0 132.6 1.1 245.0 426.4 6311.4 48.3 4676.3 11662.5

0

Bangladesh

242.8 198.9 31.9 29.0 16.2 9.2 9.2 8.1 7.5 7.1 31.0 590.8

1

Pakistan

China 156.3 India 254.3 Indonesia 45.3 Pakistan 32.3 Bangladesh 12.2 Viet Nam 10.1 Thailand 30.6 Japan 7.6 Philippines 14.9 Myanmar 15.2 Others 47.8 Total 626.5

106.3 45.5 12.1 5.7 2.3 1.9 2.2 0.2 1.8 2.5 5.9 186.4

2

India

China 59.4 India 96.6 Pakistan 12.3 Indonesia 17.2 Thailand 11.6 Philippines 5.6 Bangladesh 4.6 Myanmar 5.8 Viet Nam 3.9 Japan 2.9 Others 18.1 Total 238.1

NO emission 14.5 5.1 57.5 12.3 4.0 40.5 0.4 0.6 6.5 1.9 0.7 3.4 0.5 0.4 1.4 0.0 0.3 1.3 0.2 0.2 1.9 0.6 0.2 1.3 0.3 0.2 1.3 0.2 0.2 1.4 0.7 0.4 5.9 31.7 12.2 122.4

476.3 415.8 63.6 59.3 33.7 18.8 18.6 16.7 15.2 14.2 63.3 1195.4

3

Indonesia

N2O emission 202.4 27.7 9.7 109.4 86.6 23.5 21.8 77.1 23.1 0.8 1.1 12.4 10.9 3.6 1.4 6.6 4.5 0.9 0.8 2.6 3.7 0.1 0.5 2.5 4.2 0.4 0.4 3.6 0.4 1.2 0.4 2.4 3.5 0.6 0.4 2.5 4.7 0.3 0.3 2.7 11.1 1.3 0.7 11.3 355.0 60.4 37.5 233.0

(a)

Background FSN FBN FCR FAW

4

FBN FCR FAW Total

China 127.1 India 206.8 Pakistan 26.3 Indonesia 36.8 Thailand 24.9 Philippines 12.1 Bangladesh 9.9 Myanmar 12.3 Viet Nam 8.2 Japan 6.1 Others 38.8 Total 509.5

N 2O

China

Background* FSN

Annual emission (kg N ha−1)

Country

Sources

5

Fig. 2 Average annual emissions of N2O, NO and NH3 from croplands in East, Southeast and South Asian countries, estimated for 1995, by source. Background, FSN, FBN, FCR and FAW represent background emission, emissions resulted from the use of chemical fertilizer, biologically fixed nitrogen, crop residue and animal manure used as fertilizer, respectively.

from synthetic fertilizer was 6.5 Tg N, 16.8% of the total synthetic N consumption in the area. Emission resulting from biological N fixation was negligible. Average fluxes of N2O, NO and NH3 from croplands in the study area were 2.86, 1.41 and 28.3 kg N ha 1 yr 1, respectively. As shown in Fig. 2, the highest fluxes for each gas occurred in China because of the heavy use of chemical N fertilizer (194 kg N ha 1 yr 1) and the large animal populations there. India had more cropland area than China, and accordingly had higher background emissions. However, total N2O and NH3 emissions from cropland in China were 15 and 75% higher than the corresponding emissions in India. Uplands had a higher potential for N2O and NO emissions, whereas rice fields

1090 X . Y A N et al. had a higher potential for NH3 emission; therefore, in Bangladesh and Vietnam, where rice fields account for a large proportion of the croplands, NH3 emission fluxes were as high as 43 and 38 kg N ha 1 yr 1, respectively, whereas the N2O and NO emission fluxes were relatively small (Fig. 2). In order to prepare input data for atmospheric transport and deposition models, it is necessary to spatially locate the emissions. In the current estimation, version 2.0 of the Global Land Cover Characteristics Database of the USGS was used in order to translate the emissions into grid format (USGS, 2001). In order to distribute the emissions of each country, as well as those of each province in China and state in India, pure crop ecosystems were weighted as 1 and mixed ecosystems of crops and others were weighted as 0.5. Emissions were then averaged per 0.58 by 0.58 grid cell. The results are shown in Fig. 3. The highest N2O emissions occurred in the North China Plain, the main agricultural region of China, where upland crops are the dominant agro-ecosystem. Other relatively high flux regions included East China and North India. Flux in the rice-cultivating region in South China is usually below 1 kg N ha 1 yr 1. The distribution of NO emissions is similar to that of N2O. High NH3 emissions occurred in the lower reaches of the Yangtse River, in North India and in some patches of Central China and Bangladesh.

Discussion N2O emission The current study considered background emissions in the process of estimating direct N2O emission from croplands; this approach differs from the IPCC methodology, which emphasizes anthropogenic effects. Background emission values were derived from measurements in studies that compared emissions from fertilized and unfertilized fields. As pointed out by Mosier et al. (1998), this background emission could differ from historical natural emissions as a result of the enhanced mineralization of soil organic matter or to the depletion of soil organic matter. In addition, residual effects of N fertilizer applied prior to the experiments could have had an impact, as most of the experiments were not long-term studies. Nevertheless, background emissions should be included in estimates of total N2O emission from croplands. Applying the current method on a global scale would result in an estimate of approximately 3.6 Tg N2O±N for global croplands. As background emission is included, this estimate is 70% higher than the estimate of Mosier et al. (1998), who estimated 2.1 Tg N by using the IPCC methodology. Background emission was validated

running a process model, such as DNDC (Li et al., 2001; Brown et al., 2002). Our estimated background emission of 1.22 kg N ha 1 yr 1 is much less than the modeled emission of about 1.66 kg N ha 1 yr 1 for China (Li et al., 2001). As mentioned above, however, background emission rates could be affected by residual N. These rates could also be affected by N deposition. It is likely, therefore, that for countries with little synthetic N input, such as Afghanistan, assigning such a background emission rate would overestimate the total N2O emission. As background emission is the largest source of total N2O emission, the background emission rate becomes one of the most sensitive influencing factors in determining total emission. Unfortunately, there is much uncertainty involved in the estimation of background emission; the current background emission rate was derived from seven measurements, with only two from Asia, whereas the others were from Europe and the USA. Influencing factors, such as climate and soil properties, could be quite different for Europe and the USA than for Asia. Varying the background emission rate within the 95% confidence interval of the mean (from 0.62 to 1.81 kg N ha 1 yr 1) would result in a 21% variation in the total emission. The current study distinguished between N inputs to uplands and those to paddy fields; this constitutes another important difference between the present study's process of estimation and the IPCC methodology. The overall fertilizer-induced N2O emission factor was thereby reduced from the IPCC value of 1.25±0.93%. The overall fertilizer-induced emission factor for China was 1.04%; the simulated value from the DNDC model was 0.8% (Li et al., 2001). If upland and paddy fields were not distinguished, the total emission would be 18% higher than the current estimate. The fertilizer-induced N2O emission factor was the most influential parameter in determining total N2O emission; a 50% change in its value would result in a 28% change in the total emission of N2O. Another relatively important parameter was the ratio of animal manure used as fertilizer, which is the net result of FracGRAZ, FracFUEL and FzracGASM. In this estimation, on average, 42% of all animal manure was presumably applied to croplands as fertilizer; an assumption of 20% was made by Li et al. (2001). A 10% variation in the ratio would result in a 4% change in the total emission. Comparing the current emission estimate to existing country estimates is difficult, because most existing estimates do not include background N2O emissions; some even consider N2O emissions from paddy fields to be zero. Two estimates from China, however, are comparable to the current estimate. By regionalizing croplands in China and scaling up representative field fluxes to a regional scale, Xing (1998) estimated N2O emission from ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

N I T R O G E N O U S G A S E M I S S I O N S F R O M C R O P L A N D S 1091 (a) 50 N

N 2O

45 N kg N ha−1 yr−1

40 N

Degrees

35 N 30 N

3

25 N

2 1

20 N

0.5

15 N

0.1

10 N 5N EQ 5S 10 S 70 E

80 E

90 E

100 E Degrees

110 E

120 E

130 E

140 E

150 E

(b) 50 N

NO

45 N kg N ha−1 yr−1

40 N

Degrees

35 N 30 N

1.5

25 N

1 0.5

20 N

0.2

15 N

0.1

10 N 5N EQ 5S 10 S 70 E

80 E

90 E

100 E Degrees

110 E

120 E

130 E

140 E

150 E

(c) 50 N 45 N

NH3

40 N

kg N ha−1 yr−1

Degrees

35 N 30 N

30

25 N

20 10

20 N

5

15 N

1 10 N 5N EQ 5S 10 S 70 E

80 E

90 E

100 E Degrees

110E

120 E

130 E

140 E

150 E

Fig. 3 Distribution of N2O (a), NO (b) and NH3 (c) emissions from croplands in East, Southeast and South Asia at a 0.58 by 0.58 resolution. Emission per grid cell is an average of the emissions over the entire grid cell area.


1092 X . Y A N et al. croplands in China in 1995 to be 398 Gg N, about 15% lower than the current estimate. There are primarily two reasons for the difference. Firstly, in Xing's estimate, field fluxes were measured during the cropping season; the background emission in the current estimate, however, included measurements from the fallow season. Secondly, Xing (1998) did not consider permanent croplands, such as tea plantations and fruit tree orchards, which account for 10% of the total cropland area. If these two factors were taken into account, Xing's estimate would be very close to the current estimate. The DNDC simulated N2O emission from arable land in China was 340 Gg N for 1995 (Li et al., 2001). Similarly, this estimate did not consider permanent croplands. As mentioned above, other discrepancies, including a lower fertilizerinduced N2O emission factor and a lower ratio of animal manure used as fertilizer, have been noted. However, the simulated background emission in the DNDC model was higher than the background emission estimate in the current study partially because N2O emissions from biological N fixation and returned crop residue were not considered separately in the DNDC, as they were in the current study.

NO emission Because the N source for NO and N2O production is the same, the method for the estimation of N2O was also used in order to estimate NO emission from croplands, with the background NO emission flux and fertilizerinduced NO emission factor developed from field measurements. Background emission accounted for 40% of the total NO emission from croplands; therefore, background emission flux was one of the most sensitive parameters for determining the total NO emission. However, the background emission rate was extremely uncertain. Measured emission flux was from 0.89 to 12.6 mg NO±N m 2 h 1. More year-round measurements are needed in order to improve the accuracy of estimations. The fertilizer-induced NO emission factor for uplands adopted in this study was 0.66%, which is close to the estimate of Veldkamp & Keller (1997) but much lower than the value of 2.5% used by Yienger & Levy II (1995), and almost double the estimate of Skiba et al. (1997). The rationality of the current emission factor is discussed in Yan et al. (submitted). In the study area, paddy fields accounted for a large proportion of the cropland and approximately one-third of all N fertilizer was applied to these fields. However, no field measurements of NO emission from paddy fields in this area were available and the emission estimate was based on assumptions. Field measurements of NO emission from paddy fields in this area are, therefore, urgently needed.

Applying the current estimation method on a global scale resulted in an estimate of approximately 1.6 Tg NO±N yr 1 for global croplands, of which nearly half was background emission. This estimate did not consider the scavenger effect of canopies. It was lower than the corresponding estimates from Yienger & Levy II (1995) of 2.98 Tg N yr 1 and from Davidson and Kingerlee (1997) of 5.4 Tg N yr 1. Explaining the discrepancy is difficult because of differences in the estimation methods. However, it is clear that one major difference lies in the manner in which the effect of fertilizer on emission was managed. For upland fields, Yienger & Levy II (1995) used a fertilizer-induced emission factor of 2.5%, which could be too high; this value was derived from only eight measurements and some were not representative (see Yan et al., submitted). Davidson & Kingerlee (1997) estimated the emission by multiplying the area of cropland by an average annual emission flux, which was 3.6 kg N ha 1 yr 1 for temperate croplands and 4.0 kg N ha 1 yr 1 for tropical croplands. Many field measurements lasted for only several days or for short periods and the annual fluxes in their data set were extrapolated based on judgments. For unfertilized plots, because the fluxes and variations were usually small, this extrapolation could work relatively well. The average annual flux from all the unfertilized croplands in their data set was 0.64 kg N ha 1 yr 1 (or 0.36 kg N ha 1 yr 1 if N-fixing crops were excluded), which is relatively close to our estimated value of 0.57 kg N ha 1 yr 1. For fertilized plots, because fluxes can vary greatly with time, extrapolation could lead to much uncertainty; as acknowledged by the authors; this considerable degree of uncertainty was inevitable. For example, for a corn plot fertilized with 30 kg N ha 1 and with an average flux of 340 mg NO±N m 2 h 1, as determined by the 9-day measurement (see Williams et al., 1988), Davidson & Kingerlee (1997) estimated an annual emission flux of 16 kg N ha 1 yr 1, which is more than half of the total fertilizer N applied to the plot. We have assembled the results of field experiments that measured NO and N2O simultaneously. The calculated NO/N2O ratio was 0.001±9.83, with an average of 1.15 (data not shown). However, of the 70 individual ratios, only 21 were greater than 1 and the median ratio was only 0.35. Thus we speculate that NO emission from agricultural soils is unlikely to be higher than direct N2O emission, which was estimated to be 2.1 Tg N yr 1, by Mosier et al. (1998), and 3.6 Tg N yr 1, in this study.

NH3 volatilization Nearly 55% of the NH3 emission from croplands originated from synthetic fertilizer applications. Owing to the use of urea and ammonium bicarbonate, and the ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

N I T R O G E N O U S G A S E M I S S I O N S F R O M C R O P L A N D S 1093 cultivation of rice in this region, the average NH3 loss ratio of synthetic N fertilizer in this region was 16.8%, a value much higher than that found in other regions. For example, the average NH3 loss from fertilizer applied to UK arable lands was only 2.2% (Pain et al., 1998). The highest NH3 loss ratios from fertilizer occurred in Bangladesh and Myanmar, where rice constitutes 74 and 53% of the total crop area, respectively. Ammonia emission from urea and ammonium bicarbonate accounted for 95% of the total fertilizer-induced emissions. The NH3 emission rates of urea and ammonium bicarbonate were derived from field experiments performed in Asia. Although uncertainties exist, because of the limited number of experiments and variations in fertilization times and methods, the average of all the reported emission rates in Asia (see Tables 5 and 6) amounted to 24%, with a standard deviation of 14%. Similarly high emission rates were also derived by Bouwman et al. (1997) for urea and ammonium bicarbonate. Uncertainty could also arise because of the possible temperature dependence of emission rates. Data required to determine clear temperature dependence were lacking, and data in Table 5 indicate that NH3 emission rates in tropical areas (Philippines) were not necessarily higher than those in subtropical areas (China). In many other NH3 emission inventories, emissions from animal manure used as fertilizer have been cataloged as animal production, rather than as soil emission. In the current estimate, in order to keep the sources consistent for both N2O and NO emission, this emission was separated from animal production. In order to estimate N2O emissions from agricultural soils, IPCC (1997) proposed a method in order to estimate the amount of animal manure used as fertilizer (FAW). As NH3 emission is from the same N source, this method was adopted in the current study. For emission rates, those from the European Emission Inventory Guide (EEA, 1999) were adopted, because of the lack of sufficient field measurements in the study area. Uncertainty could stem from differences in fodder composition and subsequent differences in the mineral N content of animal manure between developing Asian countries and countries in Europe. High temperatures in tropical Asia could cause higher NH3 volatilization rates from animal manure than has been observed in Europe.

Conclusion Background emission of N2O accounted for approximately 43% of the total N2O emission from croplands in the study area. As fertilizer use on croplands in this area is higher than the global average, background N2O emission is likely to be more important in global scale than in this area and should be included in emission estimates. ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

However, there is much uncertainty with respect to background flux estimates, especially from paddy fields, because background emission in the fallow season has not been well documented. More field measurements are needed in order to improve the accuracy of background flux estimates. By distinguishing the effects of fertilization on N2O emissions from upland fields and paddy fields, the average fertilizer-induced N2O emission factor was reduced from the default IPCC value of 1.25 to 0.93% for the study area. The estimated total N2O emission from croplands in China matches reasonably well with estimates from process models and up-scaling methods. The N2O estimation method was also used in order to estimate NO emission from croplands. The result was much lower than that of earlier estimations, primarily as a result of the small fertilizer-induced NO emission factor used; this result is highly uncertain and could be altered by only a few high measurements. Background emission levels of NO, as with N2O, are therefore important, but not, as yet, clearly identified. There is an urgent need to verify NO emission levels from rice fields, a principal ecosystem in the study area. Ammonia emission from croplands in the study area was estimated to be 11.8 Tg N yr 1, with 6.5 Tg from the use of chemical N fertilizer and 4.7 Tg from the use of animal manure as fertilizer. The average NH3 loss rate in the study area from chemical N fertilizer was 16.8%, which is much higher than in other areas; this was because of the widespread use of urea and ammonium bicarbonate and the large number of rice fields in the area. The large NH3 emission factors for fertilizer N were derived from a limited number of measurements and are, therefore, likely to change when more measurements become available. The highest emissions of N2O and NO occurred in the North China Plain and in North India. The highest NH3 volatilization occurred in the delta region of the Yangtse River in China, in the North China Plain, in Bangladesh and in North India.

References Abao EB, Bronson KF Jr, Wassmann R et al. (2000) Simultaneous records of methane and nitrous oxide emissions in rice-based cropping systems under rainfed conditions. Nutrient Cycling in Agroecosystems, 58, 131±139. Anderson IC, Levine JS (1987) Simultaneous field measurements of biogenic emissions of nitric oxide and nitrous oxide. Journal of Geophysical Research, 92, 965±976. Arcara PG, Gamba C, Bidini D et al. (1999) The effect of urea and pig slurry fertilization on denitrification, direct nitrous oxide emission, volatile fatty acids, water-soluble carbon and anthrone-reactive carbon in maize-cropped soil from the Po plain (Modena, Italy). Biology and Fertility of Soils, 29, 270±276.

1094 X . Y A N et al. Asman WAH (1992) Ammonia emission in Europe: Updated Emission and Emission Variations RIVM report 228471008 RIVM, Bilthoven, the Netherlands. Baggs EM, Rees RM, Smith KA (2000) Nitrous oxide emission from soils after incorporating crop residues. Soil Use Management, 16, 82±87. Benton J, Fuhrer J, Gimeno BS (2000) An international cooperative programme indicates the widespread occurrence of ozone injury on crops. Agriculture, Ecosystems and Environment, 78, 19±30. Bian XJ, Wang WJ, Yang FC et al. (1997) Fate of fertilizer nitrogen applied to spring wheat growing on a meadow chestnut soil in plateau of Hebei Province. Acta Pedologica Sinica, 34, 60±65 (in Chinese). Bolle HJ, Seiler W, Bolin B (1986) Other greenhouse gases and aerosols: assessing their role for atmospheric radiative transfer. In: The Greenhouse Effect, Climatic Change, and Ecosystems SCOPE Report 29 (eds Bolin B et al.), pp. 157±203. Wiley, Chichester. Bouwman AF (1996) Direct emission of nitrous oxide from agricultural soils. Nutrient Cycling in Agroecosystems, 46, 53±70. Bouwman AF, Lee DS, Asman WAH et al. (1997) A global highresolution emission inventory for ammonia. Global Biogeochemistry Cycles, 11, 561±587. Brasseur GP, Orlando JJ, Tyndall GS (1999) Atmospheric Chemistry and Global Change. Oxford University Press, New York. Bronson KF, Mosier AR, Bishnoi SR (1992) Nitrous oxide emissions in irrigated corn as affected by nitrification inhibitors. Soil Science Society of America Journal, 56, 161±165. Bronson KF, Neue HU, Singh U et al. (1997a) Automated chamber measurements of methane and nitrous oxide flux in a flooded rice soil. I. Residue, nitrogen, and water management. Soil Science Society of America Journal, 61, 981±987. Bronson KF, Neue HU, Singh U et al. (1997b) Automated chamber measurements of methane and nitrous oxide flux in a flooded rice soil. II. Fallow period emissions. Soil Science Society of America Journal, 61, 988±993. Brown L, Syed B, Jarvis SC et al. (2002) Development and application of a mechanistic model to estimate emission of nitrous oxide from UK agriculture. Atmospheric Environment, 36, 917±928. Cai G, Peng G, Wang X et al. (1992b) Ammonia volatilization from urea applied to acid paddy soil in southern China and its control. Pedosphere, 2, 345±354. Cai Z, Xing G, Yan X et al. (1997) Methane and nitrous oxide emissions from rice paddy fields as affected by nitrogen fertilizers and water management. Plant and Soil, 196, 7±14. Cai G, Yang N, Lu W et al. (1992a) Gaseous loss of nitrogen from fertilizers applied to a paddy soil in southeastern China. Pedosphere, 2, 209±217. Cai GX, Zhu ZL, Trevitt ACF et al. (1986) Nitrogen loss from ammonium bicarbonate and urea fertilizers applied to flooded rice. Fertilizer Research, 10, 203±215. Chang C, Cho CM, Janzen HH (1998) Nitrous oxide emission from long-term manured soils. Soil Science Society of America Journal, 62, 677±682. Colbourn P, Harper IW (1987) Denitrification in drained and undrained arable clay soil. Journal of Soil Science, 35, 11±22.

Conrad R (1990) Flux of NOx between soil and atmosphere: importance and soil microbial metabolism. In: Denitrification in Soil and Sediment (eds Sorensen J, Revsbech NP), pp. 105±128. Plenum, New York. Crutzen PJ (1970) The influence of nitrogen oxides on the atmospheric ozone content. Quarterly Journal of Royal Meteorology Society, 96, 320±325. Crutzen PJ (1979) The role of NO and NO2 in the chemistry of the troposphere and stratosphere. Annual Review of Earth Planet Science, 7, 443±472. Davidson EA, Kingerlee W (1997) A global inventory of nitric oxide emissions from soils. Nutrient Cycling in Agroecosystems, 48, 37±50. Davidson E, Verchot LV (2000) Testing the hole-in-the-pipe model of nitric and nitrous oxide emissions from soils using the TRAGNET database. Global Biogeochemistry Cycles, 14, 1035±1043. De Datta SK, Buresh RJ, Samson MI et al. (1991) Direct measurement of ammonia and denitrification fluxes from urea applied to rice. Soil Science Society of America Journal, 55, 543±548. Denmead OT, Freney JR, Simpson JR (1979) Nitrous oxide emission during denitrification in a flooded field. Soil Science Society of America Journal, 43, 716±718. Dhyani BP, Mishra B (1992) Effect of nitrogen-application schedule on ammonia volatilization from field of rice (Oryza sativa). Indian Journal of Agricultural Sciences, 62, 73±74. Dobbie KE, McTaggart IP, Smith KA (1999) Nitrous oxide emissions from intensive agricultural systems: variations between crops and seasons, key driving variables, and mean emission factors. Journal of Geophysical Research, 104 (D21), 26891±26899. Dong Y, Scharffe D, Qi YC et al. (2001) Nitrous oxide emissions from cultivated soils in the North China plain. Tellus, 53B, 1±9. Duxbury JM, McConnaughey PK (1986) Effect of fertilizer source on denitrification and nitrous oxide emissions in a maize field. Soil Science Society of America Journal, 50, 644±648. Editorial Board of China Agriculture Yearbook (1996) China Agriculture Yearbook 1995. Agricultural Publishing House, Beijing, China. EEA (European Environment Agency) (1999) Joint EMEP/ CORINAIR Atmospheric Emission Inventory Guidebook, 2nd Edition, Copenhagen. EPA (2002) Inventory of US Greenhouse Gas Emissions and Sinks: 1990±2000 Office of Atmospheric Programs, US Environmental Protection Agency Available online at http:// wwwepagov/globalwarming/publications/emissions/ us2002/. FAOSTAT (2002) Statistical Satabase of the Food and Agricultural Organization of the United Nations. Available at: http://appsfaoorg/defaulthtm. Farquhar GD, Wetselaar R, Weir B (1983) Gaseous nitrogen losses from plants. In: Gaseous Loss of Nitrogen from Plant-Soil System (eds Freney JR, Simpson JR), pp. 159±180. Martinius Nijhoff/Dr. WJunk Publishers, The Hague. Fertiliser and Allied Agricultural Statistics 1995±96 (1996) The Fertiliser Association of India, New Delhi. Fillery IRP, De Datta SK (1986) Ammonia volatilization from nitrogen sources applied to rice fields. I. Methodology, ammonia fluxes, and nitrogen-15 loss. Soil Science Society of America Journal, 50, 80±86. ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

N I T R O G E N O U S G A S E M I S S I O N S F R O M C R O P L A N D S 1095 Fillery IRP, Simpson JR, De Datta SK (1984) Influence of field environment and fertilizer management on ammonia loss. Soil Science Society of America Journal, 48, 914±920. Fillery IRP, Simpson JR, De Datta SK (1986) Contribution of ammonia volatilization to total nitrogen loss after applications of urea to wetland rice fields. Fertilizer Research, 8, 193±202. Firestone MK, Davidson EA (1989) Microbiological basis of NO and N2O production and consumption in soil. In: Exchange of Trace Gases Between Terrestrial Ecosystems and the Atmosphere (eds Andreae MO, Schimel DS), pp. 7±21. John Wiley and Sons, New York. Freney JR, Denmead OT, Watanabe I et al. (1981) Ammonia and nitrous oxide losses following applications of ammonia sulfate to flooded rice. Australian Journal of Agricultural Research, 32, 37±45. Galbally IE, Freney JR, Muirhead WA et al. (1987) Emission of nitrogen oxides (NOx) from a flooded soil fertilized with urea: relation to other nitrogen loss processes. Journal of Atmospheric Chemistry, 5, 343±365. Galloway JN, Schesinger WH, Levy II H et al. (1995) Nitrogen fixation: anthropogenic enhancement-environmental response. Global Biogeochemistry Cycles, 9, 235±252. Harrison RM, Yamulki S, Gouldling KWT (1995) Effect of fertilizer application on NO and N2O fluxes from agricultural fields. Journal of Geophysical Research, 100 (D12), 25923±25931. Henault C, Devis X, Page S et al. (1998) Nitrous oxide emissions under different soil and land management conditions. Biology and Fertility of Soils, 26, 199±207. Holtan-Hartwig L, Bùckman OC (1994) Ammonia exchange between crops and air. Norwegian Journal of Agricultural Sciencesupplement, 14, 5±40. Humphreys E, Frency JR, Muirhead WA et al. 1988. Loss of ammonia after application of urea at different times to dryseeded, irrigated rice. Fertilizer Research, 16, 47±57. IFA/IFDC/FAO (1999) Fertilizer Use by Crop, 4th edn. Rome, 1999. Indian Agriculture in Brief, 27th edn (2000) Agriculture Statistics Division, Directorate of Economics and Statistics. Department of Agriculture and Cooperation, Ministry of Agriculture, Government of India. IPCC (1997) Greenhouse gas emissions from agricultural soils. In: Greenhouse Gas Inventory Reference Manual Revised 1996, IPCC Guidelines for National Greenhouse Gas Inventories, IPCC/OECD/ IGES, Bracknell, UK. IPCC (2000) Agriculture. In: Good Practice Guidance, Uncertainty Management in National Greenhouse Gas Inventories, IPCC/ IGES, Kanagawa, Japan. Johansson C, Granat L (1984) Emission of nitric oxide from arable land. Tellus, 36B, 25±37. Kessavalou A, Mosier AR, Doran JW et al. (1998) Fluxes of carbon dioxide, nitrous oxide, and methane in grass sod and winter wheat-fallow tillage management. Journal of Environmental Quality, 27, 1094±1104. Kodo M, de Dioz JL, Ito S (2002) N2O and NO emission from an andisol paddy soil as affected by soil moisture and nitrogen application method. In: Abstracts of the Annual Meeting, Japanese Society of Soil Science, Plant Nutrition, 48 (April, 2002), pp. 105. Kroeze C, Mosier A, Bouwman L (1999) Closing the N2O Budget: a retrospective analysis. Global Biogeochemistry Cycles, 13, 1±8. ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 1080±1096

Kumar U, Jain MC, Pathak H et al. (2000) Nitrous oxide emission from different fertilizers and its mitigation by nitrification inhibitors in irrigated rice. Biology and Fertility of Soils, 32, 474±478. Lardy G, Anderson B (1999) Alternative Feeds for Ruminants. Available at: http://wwwextnodakedu/extpubs/ansci/livestoc/as1182htm #Index. Lessard R, Rochette P, Gregorich EG et al. (1996) Nitrous oxide fluxes from manure-amended soil under maize. Journal of Environmental Quality, 25, 1371±1377. Li JJ, Bai JM, Ralph O (1998) Assessment on the Availability of Biomass Resources in China. China Environmental Science Press, Beijing, 270p. (in Chinese). Li C, Zhuang Y, Cao M et al. (2001) Comparing a national inventory of N2O emissions from arable lands in China developed with a process-based agroecosystem model to the IPCC methodology. Nutrient Cycling in Agroecosystems, 60, 159±170. McKenney DJ, Shuttleworth KF, Findlay WI (1980) Nitrous oxide evolution rates from fertilized soils: effects of applied nitrogen. Canadian Journal of Soil Science, 60, 429±438. Minami K (1990) Effect of nitrification inhibitors on emission of nitrous oxide from soils. Proceedings International Congress of the International Soil Science Society, Kyoto, Japan, August, 1990. Mosier AR, Guenzi WD, Schweizer EE (1986) Soil losses of dinitrogen and nitrous oxide from irrigated crops in Northeastern Colorado. Soil Science Society of America Journal, 50, 344±348. Mosier AR, Hutchinson GL, Sabey BR et al. (1982) Nitrous oxide emissions from barley plots treated with ammonium nitrate or sewage sludge. Journal of Environmental Quality, 11, 78±81. Mosier A, Kroeze C, Nevison C et al. (1998) Closing the global N2O budget: nitrous oxide emissions through the agricultural nitrogen cycle ± OECD/IPCC/IEA phase II development of IPCC guidelines for national greenhouse gas inventory methodology. Nutrient Cycling in Agroecosystems, 52, 225±248. Obcemea WN, Real JG, De Datta SK (1988) Effect of soil texture and nitrogen management on ammonia volatilization and total nitrogen loss. Philippine Journal of Crop Science, 13, 145±153. Olivier JGJ, Bouwman AF, Van der Hoek KW et al. (1998) Global Air Emission Inventories for Anthropogenic Sources of NOx, NH3 and N2O in 1990. Environmental Pollution, 102, 135±148. Pain BF, Van der Weerden TJ, Chambers BJ et al. (1998) A new inventory of ammonia emissions from UK agriculture. Atmospheric Environment, 32, 309±313. RoÈver M, Heinemeyer O, Kaiser E (1998) Microbial induced nitrous oxide emissions from an arable soil during winter. Soil Biology and Biochemistry, 30, 1859±1865. Schjùrring JK (1991) Ammonia emission from the foliage of growing plants. In: Trace Gas Emissions by Plants (eds Sharkey TD et al.), pp. 267±292. Academic Press, San Diego. Skiba U, Fowler D, Smith KA (1997) Nitric oxide emissions from agricultural soils in temperate and tropical climates: sources, controls and mitigation options. Nutrient Cycling in Agroecosystems, 48, 139±153. Smith CJ, Brandon M, Patrick WH Jr (1982) Nitrous oxide emission following urea-N fertilization of wetland rice. Soil Science and Plant Nutrition, 28, 161±171.

1096 X . Y A N et al. Song W, Wang S, Zeng J et al. (1997) Emission of nitrous oxide from dryland in northern China. Advances in Environmental Sciences, 5, 49±55 (in Chinese). Summerfield RJ, Roberts EH (1985) Grain Legume Crops. Collins, London. Suratno W, Murdiyarso D, Suratmo FG et al. (1998) Nitrous oxide flux from irrigated rice fields in West Java. Environmental Pollution, 102 (S1), 159±166. Sutton MA, Nemitz E, Fowler D et al. (1996) The EXAMINE project: exchange of atmoshperic ammonia with European ecosystems. In: Proceedings of the EUROTRAC Symposioum '96 (eds Borrel PM et al.), pp. 155±161. Computational Mechanics Publications, Southampton, UK. Taiwan Agriculture Yearbook (1997). Department of Agriculture and Forestry, Taiwan Provincial Government. Thornton FC, Bock BR, Tyler DD (1996) Soil emissions of nitric oxide and nitrous oxide from injected anhydrous ammonium and urea. Journal of Environmental Quality, 25, 1378±1384. Thornton FC, Valente R (1996) Soil emissions of nitric oxide and nitrous oxide from no-till corn. Soil Science Society of America Journal, 60, 1127±1133. Tsuruta H, Kanda K, Hirose T (1997) Nitrous oxide emission from a rice paddy field in Japan. Nutrient Cycling in Agroecosystems, 49, 51±58. USGS (United States Geology Survey) (2001) Global land cover characteristic data base. available online at http://edcdaacusgsgov/glcc/globdoc2 0html. Veldkamp E, Keller M (1997) Fertilizer-induced nitric oxide emissions from agricultural soils. Nutrient Cycling in Agroecosystems, 48, 69±77. Wagner-Riddle C, Thurtell GW, Kkidd G et al. (1997) Estimates of nitrous oxide emissions from agricultural fields over 28 months. Canadian Journal of Soil Science, 77, 135±144. Weitz AM, Linder E, Frolking S et al. (2001) N2O emissions from humid tropical agricultural soils: effects of soil moisture, texture and nitrogen availability. Soil Biology and Biochemistry, 33, 1077±1093. Whalen SC, Phillips RL, Fischer EN (2000) Nitrous oxide emission from an agricultural field fertilized with liquid lagoonal swine effluent. Global Biogeochemistry Cycles, 14, 545±558.

Whitehead DC, Lockyer DR (1989) Decomposing grass herbage as a source of ammonia in the atmosphere. Atmospheric Environment, 23, 1867±1869. Williams EJ, Parrish DD, Fehsenfeld FC et al. (1988) Measurement of soil NOx emissions in central Pennsylvania. Journal of Geophysical Research, 83, 9539±9546. Xi Z, Shi X, Huang W et al. (1987) Measuring urea loss from ammonia volatilization by a micro-meteorological method. Acta Agric Shanghai, 3, 47±56 (in Chinese). Xing GX (1998) N2O emission from cropland in China. Nutrient Cycling in Agroecosystems, 52, 249±254. Xing G, Yan X (1999) Direct nitrous oxide emissions from agricultural fields in China estimated by the revised 1996 IPCC guidelines for national greenhouse gases. Environmental Science and Policy, 2, 355±361. Xing G, Zhu ZL (2000) An assessment of N loss from agricultural fields to the environment in China. Nutrient Cycling in Agroecosystems, 57, 67±73. Yan X, Du L, Shi S et al. (2000) Nitrous oxide emission from wetland rice soil as affected by the application of controlledavailability fertilizers and mid-season aeration. Biology and Fertility of Soils, 32, 60±66. Yan X, Hosen Y, Yagi K (2001) Nitrous oxide and nitric oxide emissions from maize field plots as affected by N fertilizer type and application method. Biology and Fertility of Soils, 34, 297±303. Yan X, Shimizu K, Ohara T et al. (submitted) Determining fertilizer-induced NO emission ratio from soils by statistical distribution model. Biology and Fertility of Soils, in press:. Yienger JJ, Levy H II (1995). Empirical model of global soilbiogenic NOX emissions. Journal of Geophysical Research, 100, 11447±11464. Zhang S, Cai G, Wang X et al. (1992) Losses of urea-nitrogen applied to maize green on a calcareous fluvo-qauic soil in north China plain. Pedosphere, 2, 171±178. Zhu Z, Cai G, Simpson JR et al. (1989) Processes of nitrogen loss from fertilizers applied to flooded rice fields on a calcareous soil in north-central China. Fertilizer Research, 18, 101±115.
