Published April, 1990

Nitrous Oxide Emissions from Fertilized

Soils:

Summaryof Available

Data

Melissa J. Eichner*
ABSTRACT
the global source assumingthat 0.5 to 1.5%of N ferDirectmeasurements
of fertilizer-derived
N20emission
datafrom tilizer is emitted as N20(McElroyand Woofsy,1985).
104field experiments
reported
in agriculture
andsoil sciencelitSeveral investigators have estimated the annual emiseraturethatwereobtained
between
1979and1987weresummarized sions of fertilizer-derived N20assumingvarious annual fertilizer consumption (6-20 Tg N20-Nyr
andusedto estimateworldwide
fertilizer-derived
N20emissions. -~,
Although
without
statisticaldetermination,
thereappears
to bea
Hahn and Junge, 1977; <3 Tg N20-Nyr-~, Crutzen,
trendbetween
emissions
andtypeandquantity
of fertilizerapplied; 1983,in Bolle et al., 1986; 0.6-2.3 Tg N20-N
yr-~, with
the available
datadoesnotindicatea trendbetween
emissions
and an average of 1.5 Tg N20-Nyr-t, Seiler, 1986, in Bolle
a particular
soiltypeoragriculture
system.
Using
thefraction
of the
et al., 1986; 0.2-0.6 Tg N20-Nyr -t, McElroy and
Nfertilizerevolved
asN20andfertilizerconsumption
estimates
for
Woofsy, 1985).
fivefertilizertypes,0.1to 1.0TgN20-N
(avg.0.3;median
0.2)were
The Food and Agriculture Organization of the
estimated
to bereleased
during
the"sampling
period."
If theseesUnited Nations has estimated that 70.5 Tg N equivtimates
aredoubled
to account
for emissions
afterthesampling
pealent fertilizer were consumed worldwide in 1984
riodandemissions
fromfertilizerlost in drainage
water
andground- (FAO, 1985). China, the Soviet Union, and the USA
water,the expected
rangewould
be0.2 to 2.1 TgN20-N
(avg.0.7;
together consume approximately one-half of the
median
0.5) emitted
into theatmosphere
in 1984.Themagnitude
of
worlds fertilizer; India, France, the United Kingdom,
this estimate
is in agreement
withrecentglobalestimates.
If 100Tg
West
Germany, Poland, Canada, Indonesia, Mexico,
Nfertilizerareconsumed
worldwide
in the year2000,the global
and
Italy
consume approximately 25%. The World
releaseof fertilizer-derived
emissions
intotheatmosphere
willprobBankestimates that 93.5 Tg N fertilizer will be conablynotexceed3 TgN20-N
in theyear2000.It is estimated
that
sumed in 1996/1997-a 34%increase over consump23 to 315GgN20-N
wereemittedinto the atmosphere
fromfields
of cultivated
leguminous
cropsin 1986.Future
research
needswere tion in 1985/1986. Seventy percent of the growth
between 1985/1986 and 1996/1997 will occur in
suggested.
developing nations (World Bank, 1987, unpublished
data).
Emissions from Legumes
ITROUS
oxide, implicated for its possible deleteNitrogen-fixing
leguminouscrops that add N to the
rious effects on global warming(Wanget al.,
soil can also increase the level of N20emitted from
1976) and the catalytic destruction of stratospheric
soils. Three studies have sampled emissions from unozone (Crutzen, 1976; Liu et al., 1976), is naturally
fertilized sites cropped with legumes. Bremneret al.
produced in soils by microbial processes during de(1980) sampled sites cropped with soybean [Glycine
nitrification (Bremnerand Blackmer, 1978; Rydenand
max(L.) Merr.] in six different soil types. Emissions
Lund, 1980) and nitrification reactions (Smith and
ranged from 0.34 to 1.97 kg N20-Nha-t yr -~ with an
Chalk, 1980; Bremner and Blackmer, 1978; Cochran
average of 1.2 kg N20-Nha-l yr-~. Comparisons to
et al., 1981). Theproduction and evolution of natural
nonlegumecrops or to uncultivated soil were not inand fertilizer-derived N20from the soil surface are
cluded in the study. Duxburyand Bouldin (I 982) samdependenton agriculture practices, biogenic processes,
pled emissions from sites cropped with alfalfa (Medsoil properties, and climatic conditions. By providing
icago sativa L.). Emissions in 1979-1980were 4.2 kg
an additional Nsource, mineral fertilizers increase the
N20-Nha-~ yr -~, and in 1980-1981emissions were 2.3
emissions of N20from the soil. The impact of inkg N20-Nha-x yr-~. Emissions from a weedypasture
creased usage of N fertilizer on the atmosphericburof timothy grass (Phleumpratense L.) ranged from 1.7
den of N20 prompted experiments to measure emisand 0.9 kg of N20-Nha-~ yr-~, respectively. The varsions from fertilized
soils. Field experiments
iability of precipitation during the 2 yr mightaccount
documented
in agriculture, soil science, and other scifor the difference. Duxbury (1984) sampled sites
entific journals and publications were reviewed and
cropped with alfalfa. Emissions ranged from 2 to 4.6
emission measurements compiled as a preliminary
kg N20-Nha-~ yr-~. Emissions from a weedypasture
step in assessing the atmospheric burden of N20from
of timothy grass ranged from 0.7 to 1.9 kg of N20-N
Nfertilizer.
ha-l yr-~, respectively. Theweedypasture probably
BACKGROUND
contained legumes.
Based on the above three experiments, the level of
Anthropogenic sources of N20have been estimated
emissions from agricultural soils cropped with leto account for 29 to 46%of the total source of N20;
gumesranged from 0.34 to 4.6 kg N20-Nha-t yr-~;
excluding emissions from manure, fertilizer-derived
this range includes natural emissions, emissions asemissions are estimated to account for 2.3 to 2.7%of
sociated with cultivation and emissions from N fixed
ScienceandPolicyAssociates,
Inc., WestTower,Suite400,1333H
by the legumecrop. Theeffects of residual N fixed by
St. NW,
Washington,
DC20005(formerlyStrategicStudiesStaff,
Officeof PolicyAnalysis,USEPA,
PM220,
401MSt. SW,Wash- the previous seasons legumecrop, prior cropping hisington,DC20460).
Thispaperdoesnotnecessarily
reflectthe official
tory, and other managementfactors are not accounted
position of the USEPA.
Received19 Jan. 1989.*Corresponding for. The Food and Agriculture Organization of the
author.
United Nations has estimated that 68 403 000 ha of
leguminous crops (pulses) were cultivated in 1986
Published
in J. Environ.Qual.19:272-280
(1990).

272

273

EICHNER: N20 EMISSIONS FROM SOIL

(FAO,1986). Based on the annual emission rates and

caveats above, total N20 emissions from fields of cultivated legumesare estimated to range from 23 to 315
Gg NEO-N in 1986. This does not include emissions
from legumes that are not harvested, emissions from
seedlings, nor emissions fromfertilizer applied to the
legumefield before planting. If leguminousfields are
found to emit more N20 than fertilized fields, dual
croppingwith legumes, or rotating hectares under cultivation with legumesas an alternative to fertilizer use
could raise the atmospheric burden of N20.

Table 1. Keyfactors affecting fertilizer-derived

N20emissions.

Factors Affecting Emissions

Manyfactors affect the occurrence and rate of denitrification and nitrification reactions that produce
N20and its evolution from the soil surface. The factors in Table 1 influence the biological processes and
environmentof the soil microorganismsthat regulate
N20emissions. Utilizing these factors to predict natural and fertilizer emissionsis complicatedby the interaction amongthe factors and the variability of their
influence on emissions. Sahrawat and Keeney(1986)
provide a summaryof the effects of these factors on
N20 emissions.

duction is debatedin the literature. For a regional and

global assessment of emissions, determining the reaction source is not essential. Likewise,predicting the
spontaneousflux is not critical to obtaining daily and
annual emissions. The total amount evolved during
the specified sampling period, in theory, should capture and average out the flux resulting from episodic
events during the sampling period. Althoughthe soil
mayat times be a sink for N20, over all, the soil system
is a source of N20.The spatial, diurnal, seasonal, and
annual variability of emissions complicates obtaining
baseline estimates and the assessmentof fertilizer-derived N20 emissions.

Emissions Sampling Methodology

Emissioncoefficients can be obtained in two ways:
from fractional accountability of the fertilizer and
from direct measurementof N20evolution from the
soil. Fractional accountability entails determiningthe
various avenuesfertilizer maytake after it is applied.
This methodrequires an estimate of the amountof N
fertilizer volatilized fromthe soil surface; extracted by
the crop; trapped in the organic soil fraction and in
the biomass of the microorganisms,both of whichare
not readily accessible to the crop; dissolved in water
and leached deepinto the soil system; or released into
the atmosphereas N2. TheN fertilizer that is leached,
immobilizedin the biomassand in the organic fraction
of the soil could eventually enter NEO-producing
pathways. After these fractions have been determined, the
remaining portion is assumedto be emitted ,as N20
or a ratio of NE/N20 is assumed.
The total loss of N fertilizer as gaseous N and the
relative proportion of N2 to N20are variable. However, N2 is the dominantgaseous product in all soils,
usually accounting for more than 80%of the gaseous
N evolved from the soil surface (Duxbury, 1984).
Forty to 75%of the N fertilizer applied is recovered
by the crop (Allison, 1955). Becausethe fractional accountability of the fertilizer varies with soil type, climatic conditions, and fertilizer type, for example,obtaining emissions coefficients by this methodwould
be extremely difficult and possibly highly variable
from region to region.
The direct measurementof the N~Oevolving from
the soil surface is the methodutilized in the experiments surveyed. Although the N source mayno longer
resemble its applied form, the subsequent N20 emissions are considered fertilizer-derived if emissions
from a fertilized site are greater than emissions from
a similar unfertilized control site. Therelative contribution of denitrification and nitrification to N20 pro-

RESULTS AND DISCUSSION

Emissions data from 24 primary articles are summarized in Tables 2 and 3. The 104 experiments were
conductedat nine locations in five countries, and all
the experiments were conducted in temperate regions.
The experiments are field studies conducted between
1979and 1987by soil and agricultural scientists. The
experiments sampled emissions during the specified
sampling period and then interpolated the amount
evolved betweensamples. The fertilizer-derived emissions were sampled by closed chambertechniques. Refer to the individual articles for the details of the sampling procedure. Only variables published consistently
in the individual experiments were recorded. Several
valuable experiments were excluded because the experimental conditions were unnatural, the sampling
periods were too short, or the summaryinformation
was not published in the article. Experiments from
flooded agriculture such as rice (Oryzasativa L.) were
excluded.
In Tables 2 and 3, the emissionof fertilizer-derived
N20 iS expressed four ways: total emission, the total
quantity of N~O-N
emitted over the specified sampling
period; average daily fertilizer-derived emission, the
quantity of fertilizer-derived NEO-Nemitted during
the given sampling period, divided by the numberof
days in the sampling period; average daily emission
per kilogramof fertilizer applied, the quantity of fertilizer-derived NEO-Nemitted during the sampling period divided by the numberof days in the sampling
period, divided by the amountof fertilizer applied;
and percent of fertilizer evolved, the amountof fertilizer-derived N20-Nevolved over the specified sampling period, divided by the amountof fertilizer applied.
Of the 104 experimentsfound in the literature survey, 49 experiments did not sample emissions from

Management
practices

Environmental
factors

Fertilizertype
Applicationrate
Applicationtechnique
Timingof application
Tillage practices
Use of other chemicals
Crop type
Irrigation
Residual N and C from crops and
fertilizer

Temperature
Precipitation
Soil moisturecontent
OrganicC content
Oxygen
availability
Porosity
pH
Freeze andthawcycle
Microorganisms

274

J. ENVIRON.QUAL., VOL. 19, APRIL-JUNE 1990

Table 2. Fertilizer-derived

Fert. Fert.
type used
-~
kg Nha

N20emissions from agricultural

-~
N~O-N
ha

soils for the controlled experiments surveyed/f

Ratio
Fert.Total Control
Average
total derived
emission (base. Fert.-derived Average dallyfert.emission emission
for
emission)emissiondaily fert.derived Percent
to
as a %
Sample sample for sample(total minasderived emissionper of fert. control of total
period period
period
control) emission kg Nased evolved emission emission Region System Soil C pH Reference
kg

mg

AA

180

140

2.17

0.51

1.66

11.86

65.87

0.92

4.25

0.76

IA

Soil

2.7 6.9

AA

180

140

4.40

0.65

3.75

26.79

148.81

2.08

6.77

0.85

IA

Soil

2.5

7.7

AA

180

140

1.92

0.38

1.54

11.00

61.11

0.86

5.05

0.80

IA

Soil

4.6

7.9

AA

225

116

3.17

0.45

2.72

23.45

104.21

1.21

7.04

0.86

IA

Soil

ra

3.8

6.9

AA

150

116

2.58

0.45

2.13

18.36

122.41

1.42

5.73

0.83

IA

Soil

3.8

6.9

AA

75

116

!.67

0.45

1.22

10.52

140.23

1.63

3.71

0.73

IA

Soil

3.8

6.9

AA

375

116

4.26

0.45

3.81

32.84

87.59

1.02

9.47

0.89

1A

Soil

3.8

6.9

AA

450

!!6

4.54

0.45

4.09

35.26

78.35

0.91

10.09

0.90

1A

Soil

3.8

6.9

AA

300

116

3.75

0.45

3.30

28.45

94.83

I.I 0

8.33

0.88

IA

Soil

3.8

6.9

AA

250

139

15.00

1.70

13.30

95.68

382.73

5.32

8.82

0.89

IA

Soil

4.6

7.9

AA

250

139

12.10

2.00

10.10

72.66

290.65

4.04

6.05

0.83

IA

Soil

2.5

7.7

AA

250

139

19.60

2.50

17.10

123.02

492.09

6.84

7.84

0.87

1A

Soil

2.7

6.9

180

140

0.45

0.38

0.07

0.50

2.78

0.04

1.18

0.16

IA

Soil

4.6

7.9

180

140

0.86

0.65

0.21

1.50

8.33

0.12

1.32

0.24

IA

Soil

2.5

7.7

180

140

0.62

0.51

0.11

0.79

4.37

0.06

1.22

0.18

IA

Soil

2.7

6.9

AC
AC
AC
AC
AC
AC
AC
AC

100
100
100
100
100
100
100
100

61
92
31
31
61
31
61
72

0.03
0.70
0.22
0.07
0.05
0.38
0.15
0.09

0.41
7.61
6.97
2.10
0.87
12.13
2.51
1.26

4.10
76.09
69.68
20.97
8.69
121.29
25.08
12.56

0.03
0.70
0.22
0.07
0.05
0.38
0.15
0.09

GR
GR
GR
GR
GR
GR
GR
GR

Grass
Grass
Grass
Grass
Grass
Grass
Grass
Plant

m
m
m
m
m
m
m
c

2.0
2.6
0.5
2.0
0.8
2.0
0.5

7.3
7.3
7.1
7.3
7.4
7.3
7.1

AC

100

32

0.30

9.38

93.75

0.30

GR

Grass m

AC

100

72

0.90

12.50

125.00

0.90

GR

Plant

AC

100

49

0.70

14.29

142.86

0.70

GR

Grass m

AS

250

96

0.61

0.33

0.28

2.95

11.79

0.11

1.86

0.46

IA

Soil

4.9

8.2

AS

125

96

0.56

0.33

0.23

2.36

18.92

0.18

1.69

0.41

IA

Soil

4.9

8.2

AS
AN
AN

100
100
100

92
98
72

2.38

0.67

0.08
1.71
0.05

0.85
17.45
0.63

8.48
174.49
6.33

0.08
1.71
0.05

3.55

0.72

GR
DN
GR

Grass m
Grass m
Plant c

2.6
1.9

7.3
5.3

AN

175

46

0.16

0.09

0.08

1.65

9.44

0.04

1.86

0.46

NY

Wheat

AN

175

46

0.30

0.09

0.22

4.67

26.71

0.12

3.44

0.71

NY

Wheat

AN
AN
AN
AN
CN

112
224
56
75
250

155
155
155
92
96

1.04
!.43
0.93

0.54
0.54
0.54

28.80
25.63
44.93
4.35
1.17

0.45
0.40
0.70
0.04
0.01

0.48
0.62
0.42
0.08

Barley
Barley
Barley
Grass
Soil

1.7
1.7
!.7

1.08

CO
CO
CO
SP
IA

m
m
m

0.33

3.23
5.74
2.52
0.33
0.29

1.93
2.65
1.72

0.36

0.50
0.89
0.39
0.03
0.03

4.9

8.1
8.1
8.1
7.4
8.2

CN

125

96

0.38

0.33

0.05

0.54

4.33

0.04

1.16

0.14

IA

Soil

4.9

8.2

CN

180

140

0.44

0.38

0.06

0.43

2.38

0.03

1.16

0.14

IA

Soil

4.6

7.9

CN

400

365

6.00

1.00

5.00

13.70

34.25

1.25

6.00

0.83

UK

Grass m

2.3

CN

400

365

8.00

1.00

7.00

19.18

47.95

1.75

8.00

0.88

UK

Grass m

4.0

Breitenbeck&
Bremner,1986a
Breitenbeck&
Bremner,1986a
Breitenbeck&
Bremner,1986a
Breitenbeck&
Breraner,1986b
Breitenbeck&
Bremner,1986b
Breitenbeck&
Bremner,1986b
Breitenbeck&
Bremner,1986b
Breitenbeck&
Bremner,1986b
Breitenbeck&
Bremner,1986b
Bremner
et al.,
1981
Bremner
et al.,
1981
Bremner
et al.,
1981
Breitenbeck&
Bremner,1986a
Breitenheek&
Bremner,1986a
Breitenbeck&
Bremner,1986a
Conrad
et al., 1983
Conradet al., 1983
Conrad
et al., 1983
Conradet al., 1983
Conradet al., 1983
Conradet al., 1983
Conrad
et al., 1983
Conrad&Seiler,
1980
Seiler &Conrad,
1981
Seiler &Conrad,
1981
Seiler &Conrad,
1981
Breitenbeck
et al.,
1980
Breitenbeck
et al.,
1980
Conrad
et al., 1983
Christensen,1983
Conrad&Seiler,
1980
Duxbury,
unpublishad,
1987
Duxbury,
unpublished,
1987
Mosleret al., 1982
Mosleret al., 1982
Mosleret al., 1982
Slemret al., 1984
Breitenbcck
et al.,
1980
Breitenbeck
et al.,
1980
Breitenbeck&
Bremner,1986a
Webster&
Dowdell,1982
Webster&
Dowdell,1982
(Continued)

275

EICHNER: N20 EMISSIONS FROM SOIL

Table 2. (Continued)
-t
N20-Nha

Fert. Fert.
type used

Total
Control
emission (base. Fert.-derivedAverage
for
emission)emissiondaily Cert.Sample sample for sample(total minus~rived
period period
period control) emission

Ratio Fert.total derived

Average
dailyCert.emissionemission
as a %
derived Percent to
emission
perof Cert. control of total
kg N used evolved emission emission Region System Soil C pH Reference

Webster&
Dowdell,1982
CN
400
365
6.00
0.80
5.20
14.25
1.30
7.50
0.87
UK Grass m 4.0
Webster&
35.62
Dowddl,1982
PN
100
92
0.02
0.18
1.85
0.02
GR Grass m 2.6 7.3 Conrad
et al., 1983
SN
100
31
0.07
22.90
0.07
GR Grass m 2.0 7.3 Conrad
et al., 1983
2.29
SN
100
61
0.02
2.95
0.02
GR
Soil m 0.5 7.1 Conrad
et al., 1983
0.30
SN
100
31
0.00
0.03
0.32
0.00
GR Grass m 2.0 7.3 Conrad
et al., 1983
SN
100
61
0.07
1.20
! 1.97
0.07
GR Grass m 0.8 7.4 Conrad
et al., 1983
SN
100
61
0.01
0.11
1.15
0.01
GR Grass m 2.0 7.3 Conrad
et al., 1983
SN
100
72
0.01
0.10
1.00
0.01
GR Plant c
Conrad
&Seiler,
1980
SN
100
72
0.10
1.39
13.89
0.10
GR Plant c
Seller &Conrad,
1981
Seiler &Conrad,
SN
100
49
0.50
10.20
102.04
0.50
GR Grass m
1981
SN
100
32
Seiler &Conrad,
0.10
3.13
31.25
0.10
GR Grass m
1981
125
96
U
0.50
0.33
0.17
1.80
14.42
0.14
1.52
0.34
IA
Soil m 4.9 8.2 Breitenbeck
et al.,
1980
U
250
96
0.62
0.33
0.29
2.99
11.96
0.11
1.87
0.47
IA
Soil m 4.9 8.2 Breitenbeck
et al.,
1980
U
180
140
0.64
0.51
0.13
0.93
5.16
0.07
1.25
0.20
IA
Soil m 2.7 6.9 Breitenbeck&
Bremner,
1986a
U
180
140
0.77
0.65
0.12
0.86
4.76
1.18
0.16
IA
Soil m 2.5 7.7 Breitenbeck&
0.07
Bremner,
1986a
U
180
140
0.57
1.36
IA
0.38
0.19
7.54
0.11
1.50
0.33
Soil f
4.6 7.9 Breitenbeck&
Bremner,
1986a
450
62
0.49
U
0.14
0.35
5.65
12.54
0.08
3.45
0.71
CO Grass m
Mosier
et al., 1981
U
!00
92
0.18
!.96
19.57
0.18
SP
Grass
Slemr
et al., 1984
U/AN 140
85
2.50
0.30
2.20
25.88
184.87
!.57
8.33
0.88
NY COm m 1.0 6.9 Duxbury&
McConnaughey,
1986
CN/
140
85
NY COrn m 1.0 6.9 Duxbury&
0.30
0.30
1.00
AN
McConnaughey,
1986
MAN/ 181
365
3.60
0.34
3.26
8.93
49.35
1.80
10.59
0.91
WI
Corn f
0.1 6.7 Cates&Keeney,
AN
1987
MAN/ 273
365
5.20
4.86
0.34
13.32
48.77
1.78
15.29
0.93
WI
Com f
0.1 6.7 Cates&Keeney,
AN/U
1987
Fertilizer type: A ~ ammonium
type, AAffi anhydrous ammonium,AC= ammonium
chloride, Amend= green manure, AN= ammonium
nitrate, AS ffi
ammonium
sulfate, CN= calcium nitrate, MAN
= manure, PN= potassiumnitrate, Sludge = sludge, SN= sodiumnitrate, U = urea. Soil type (based on
FAOtextural classes): f = fine, m = medium,c = coarse. Region: CA= California, CO= Colorado, GR= The Republic of WestGermany,DN= Denmark,
IA ~ Iowa, NY= NewYork, SP ffi Spain, WI= Wisconsin, UK= the United Kingdom.Blanks indicate unpublisheddata. The quantity of N20emitted
into the atmosphereis written as N20-N,the quantityof the N component
in the N20.
CN

400

365

4.00

0.80

3.20

8.77

21.92

an unfertilized control site. Consequently,

the fertilizer-derived emissions can not be distinguished from
the natural sources of N20.Of the controlled experiments,26 experimentsapplied fertilizer to bare soil
systems,25 experiments
appliedfertilizer to grass systems, and14 experimentsapplied fertilizer to agricultural systems.Thelater includes four sites cropped
with corn (Zea mays L.); two with wheat (Triticum
aestivumL.); three with barley (Hordeum
vulgate L.);
andfive sparsely coveredwith small plants. Eight of
the controlled studies hadfertilizer applicationrates
>250 kg N ha-~. These experiments have been excludedfromthe summary
tables becausethe fertilizer
application rates are muchgreater than typically used
in agriculturesystems.
In all but oneof the controlledfield studies, fertilization enhancedN20emissions fromthe soil surface;
that is, emissionsfromthe fertilized site exceededthe
emissionsfromthe unfertilized control site. Figures1
and2 showthe level of emissionsfor differentfertilizer
types. Althoughincreasing the fertilization rate gen-

0.80

5.00

0.80

UK

Grass m

2.3

erally correspondswith greater emissions, this does

not alwayshold true. Also, emissions varied significantly whenthe samefertilizer application rates were
used.
Toanalyze emissions trends, the controlled studies
weresorted by fertilizer type andby soil systemand
summarized
in Tables 4, 5, and6. Table 4 provides a
summary
of the influence of fertilizer type on emissions amongthe experimentsconsidered in the data
set. Fertilizer can increase emissions appreciably.
However,the impactis highly variable. Theresults of
Table 4 could be misleading because the amountof
fertilizer appliedis not considerednor is the length of
the samplingperiod. Table5 showsthat fertilizer type
appears to be an importantfactor influencing emissions. The summary
estimates in Table 5 differ appreciablyfromthe percentageof fertilizer evolvedestimated by Galbally (1985); 0.5% of anhydrous
ammonia;0.1% of ammonium
nitrate; 0.1% of ammoniumtype; 0.5%of urea, and 0.05%of nitrate.
FromTables 2 and 3, it is not clear that emissions

276

J. ENVIRON.QUAL., VOL. 19, APRIL-JUNE 1990

Table 3. Nitrous oxide emissions from agriculture

Fert. type

A
A
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AN
AS
AS
CN
CN
N
AN/MAN
AN/MAN
AN/AC/AN
AN/AC/AN
AN/amend
AN/amend
AN/amend
AN/amend
AN/amend
AN/amend
AN/amend
AN/CN/AN
AN/U/AA
AN/U/AA
AS/U/AA
AS/U/AA
AS/U/AA
AS/U/AA
AS/U/AA
AS/U/AA
AS/U/AA
Sludge
Sludge

Fert. used
-1
kg N ha
200
200
170
170
1"70
170
80
80
80
80
200
80
200
80
80
80
200
200
80
80
250
200
200
200
200
75
410
410
130
130
410
245
410
410
245
245
245
130
176
336
335
335
336
176
528
336
528
440
440

See Table 2 footnote.

Alliumcereale
cepa L.L.
Secale
Nicotiana tabacumL.
Betavulgaris L.

Sample
period
d
125
125
365
365
365
365
160
160
210
191
147
202
147
206
210
160
191
191
151
151
365
120
100
135
135
125
151
151
365
365
253
249
257
160
202
253
191
365
365
365
123
123
365
365
365
365
365
153
175

soils for uncontrolled experiments.~

Total emission
for sample
period
Region

System

Soil

pH

Reference

f
f

7.8
7.8

Hutchinson & Mosier, 1979

Mosier & Hutchinson, 1981
Duxbury& Bouldin, 1982
Duxbury& Bouldin, 1982
Duxbury & Bouldin, 1982
Duxbury & Bouldin, 1982
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroad & Keeney, 1985
Goodroad & Keeney, 1985
Ryden, 1981
Mosier et al., 1986
Mosieret al., 1986
Armstrong, 1983
Armstrong, 1983
Mosier & Hutchinson, 1981
Goodrnad & Keeney, 1985
Goodroad & Keeney, 1985
Duxbury& Bouldin, 1982
Duxbury& Bouldin, 1982
Goodroadet al., 1984
Goodroadet ai., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Goodroadet al., 1984
Duxbury& Bouldin, 1982
Ryden & Lend, 1980
Ryden& Lurid, 1980
Rydenet al., 1979
Rydenet al., 1979
Ryden & Lund, 1980
Ryden & Lend, 1980
Ryden & Lend, 1980
Ryden & Lend, 1980
Ryden & Lund, 1980
Goodroadet al., 1984
Goodroadet al., 1984

kg N20-N
-~
ha
2.60
2.50
76.00
152.00
85.00
72.00
0.20
0.40
0.90
0.30
0.30
1.60
0.60
1.50
1.00
0.20
3.50
6.30
0.22
3.23
3.25
3.00
0.80
0.22
0.30
2.00
10.31
0.48
1.60
2.90
2.70
2.20
6.10
1.20
3.20
0.70
0.30
2.20
19.60
20.20
9.22
6.14
41.80
26.90
26.60
26.40
29.20
0.20
1.60

CO
CO
NY
NY
NY
NY
WI
WI
WI
WI
WI
WI
WI
Wl
WI
WI
WI
WI
WI
WI
UK
CO
CO
UK
UK
CO
WI
WI
NY
NY
WI
WI
Wl
WI
WI
WI
WI
NY
CA
CA
CA
CA
CA
CA
CA
CA
CA
WI
WI

Corn
Corn
Corn
Corn
Onion:~
Onion
Plant
Plant
Plant
Rye
Corn
Rye
Corn
Plant
Plant
Plant
Corn
Corn
Tobacco
Tobacco
Grass
Corn
Barley
Soil
Soil
Beet#
Tobacco
Tobacco
Corn
Corn
Tobacco
Tobacco
Tobacco
Broccoli,S"
Tobacco
Tobacco
Tobacco
Corn
Artic
Celery**
Celery
Celery
Celery
Artic
Cauliflower
Celery
Cauliflower
Barley
Barley

m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
f
f
f
c
f
m
m
m
m
m
m
m
m
m
m
m
m
c
mf
m
m
mf
mf
m
c
m
m
m

5.1
4.7
5.1
4.7
6.8
4.7
6.8
4.7
6.7
6.7
6.8
6.8
4.7
4.7
6.25
7.3
7.3
7.6
4.7
4.7
4.7
4.7
4.7
4.7
4.7
4.7
4.7
7.9
7.6
7.6
7.6
6.5
7.6
7
7.7
5.8
5.8
5.8

~ Brassica oleracea (Botrytis Group).

~i* ApiumgraveolensL.
Brassicaoleracea(Botrytis Group).

froma fertilized crop systemare greater than emissions from a fertilized noncroppedsystem or vice
versa. Table6 indicates variability amongemissions
sorted by soil systemandfertilizer type. Becauseof
the small data set, the trend remainsuncertain. For
Tables4, 5, and6, other variables that are not consideredmightalso affect the emissioncoefficients, and
the relationship amongfertilizer types andcrop systems mayor maynot be an artifact of the data set.
The organic C content, porosity, pH, and location
variedacrossthe controlledgrass, soil, andagriculture
experiments. Overall trends between emissions and
each of these variables are not substantiated in the
dataset.
Althoughuncertaintyexists, the differences in emissions among
different fertilizer types is useful in developing preliminaryemission estimates. The varia-

bility between and amongthe experimental sites

precludecharacterizinga particular soil type or agriculture systemwith respect to emissions with the data
available. Althoughdiscrepancy exists betweenthe
studies surveyed,the individualstudies probablybetter characterize the relationship betweenemissions
andthe factors that affect emissions than the inferences of any columnin Tables 2 and 3.
Limitations of the Data Set for Estimating
Emissions
Severalimportantvariablesthat affect emissionsare
not summarized
or consideredin the experimentssurveyed. For example,the data set is unableto document
the temperature
andthe amountor intensity of rainfall
and/orirrigation at the experimental
sites, or the timing of these events with respect to emissions, although

277

EICHNER: N20 EMISSIONSFROMSOIL

4O
/% :AA
O:A
[] =CN
=AN
V=U

/%

0
=)

10

%
I

50

100

150

200
250
300
Fertilizer Used
(kg Nha-+)

350

400

450

Fig. 1. Relationship betweenaveragedaily fertilizer-derived NeOemissions and the quantity of fertilizer N applied regardless of the length
-I.
of the samplingperiods for controlled studies only. Excludesthree points for AA-derivedemissions: 72.7, 95.7, and 123.0 g N20-Nha
180
/% :AA

O=A

160

/%

:AN

~:N

40 I

2o
0
50

,
I
200
250
300
400
450
FertilizerUsed
(kgNha"+)
Fig. 2. Relationship betweenaveragedaily fertilizer-derived N20 emissionsper kilogramof fertilizer N applied and the quantity of fertilizer
N applied regardless of the length of the samplingperiods for controlled studies only. Excludesthree points for AA-dedved
emissions:
-~.
290.6, 382.7, and 492.1 mgN20-Nha
100

150

these are important determinants of emission. Many

studies did not specify the fertilizers physical state
whenapplied (dry or liquid) or the application procedure.
Manyof the controlled experiments had residual
plant material from the previous seasons harvest, including some with residual legumes. In someof the
studies surveyed, the unfertilized control sites might
not be comparableto the fertilized site becausethe soil
systemsweredifferent (i.e., a fertilized agriculture site
had a control site with a grass system), the sites had
different types and levels of residual plant material,
and the control sites werenot disturbed by cultivation.
Emissions from bare-soil and grass systems might
not be indicative of emissionsfromseedlings or sizable
plants becauseof the influence of the roots andfoliage.
Because the samplesites were not uniformly disturbed

Table 4. Comparisonof N20emissions from baseline (control) and

fertilizer-derived sources during the samplingperiod/re
Fert. No.
type sites

Ratioof total emissions

Fertilizeremissions
as a
to controlemissions percentageof total emissions
Range Median Avg.

Range Median Avg.

AA
9
3.7-8.8
6.0
6.1
73-89
83
82
AN 6
1.7-3.5 1.9/2.6 2.5
41-71
48/62 57
A
5
1.2-1.9
1.3
1.4
16--46 24
29
U
5
1.2-1.9
1.5
1.5
16-47
33
30
N
3
1.1-1.2
1.2
1.1
8-14
14
12
Controlledexperiments
only. Excluding
experiments
withfertilizer applications >250kg N ha-~. A = ammonium
type, ammonium
chloride, ammonium
sulfate; AA= anhydrousammonium;
AN= ammonium
nitrate;
N = calciumnitrate, potassiumnitrate, sodiumnitrate; U= urea. Experimentsusing mixedfertilizers, manure,g~*enmanure,andsludgeare not
included.
Regardless
of the amount
of fertilizer applied,the lengthof the sampling
period,the soil system,location,andothervariables.

278

J. ENVIRON.QUAL., VOL. 19, APRIL-JUNE 1990

Table 5. Fertilizer-derived

Fert.
type

No.
sites

N20emissions for five fertilizer

types during the samplingperiod only/f

Percentof Nfertilizer
evolved as N20
Range

Median

Daily averageemissions:~

Avg.

Range

Median
g N20-Nha-1 d-~

AA
9
0.86-6.84
1.63
2.70
10.5-123.0
AN
8
0.04-1.71 0.12/0.40
0.44
0.3-17.4
A
17
0.02-0.90
0.12
0.25
0.4-14.3
U
6
0.07-0.18 0.11/0.11
0.11
0.9-3.0
N
13
0.001-0.50
0.03
0.07
0.03-10.2
See Table 4 footnote t. Regardlessof soil system, location, and other variables.
Regardlessof the quantity of fertilizer applied.
Regardlessof the length of the samplingperiod.
Table 6. Fertilizer-derived

System

Fert.
type

No.
sites

A
AN
N
U
Total
AA
A
N
U
Total
A
AN
N
Total
AN
AA

10
2
7
1
20
9
5
4
5
23
2
1
2
5
5
4

N20emissions for soil systems and fertilizer

Percentof Nfertilizer
evolved as N20~
Range

Median

Soil

Plant

Grains
Corn

0.03-0.70 0.15/0.22
0.04-1.71
0.001-0.50
0.07
0.18
0.001-1.71 0.08/0.10
0.86-6.84
1.63
0.04-0.18
0.11
0.01-0.04 0.02/0.03
0.07-0.14
0.11
0.01-6.84
0.12
0.09-0.90
0.05
0.007-0.10
0.01-0.90
0.09
0.04-0.70
0.40
0.0-1.80 1.57/1.78

Avg.

Range

44.0
4.5
4.6
1.6
1.5

61.1-492.1
4.3-174.5
2.8-142.8
4.8-19.6
0.3-102.0

Avg.

Range

0.27
0.87
0.11

0.4-14.3
0.3-17.4
0.03-10.2
2.0
0.03-17.4
10.5-123.0
0.5-2.9
0.3-0.5
0.9-3.0
0.3-123.0
1.3-12.5
0.6
0.1-1.4
0.1-12.5
1.6-5.7
0.0-25.9

0.27
2.70
0.10
0.03
0.10
1.10
0.49
0.05
0.23
0.34
1.29

Median
Avg.
mg N20-N ha-~ d-~ kg N-t -140.0
25.6/26.7
18.9
7.5/12.0
2.9

200.9
40.1
44.4
10.6
15.2

types during the samplingperiod only.~"

Daily average emissions

Median
Avg.
-~
-~
g N20-Nha d -

%
Grass

23.4
2.5/3.2
2.4
1.4/1.8
0.4

Daily average emissions per kg N

applied per hectare

2.5/7.0
1.2
2.1/2.3
23.4
1.5
0.3/0.4
1.4
2.4

1.3
3.2
8.9/13.3

5.7
8.9
2.4
4.7
43.7
1.6
0.4
1.6
17.9
6.9
0.7
3.2
3.6
12.0

Daily average emissions per kg N

applied per hectare
Range

Median
Avg.
-~
-~
mg N20-N ha d
kg N-~ --

4.i-142.9
4.3-174.5
0.3-102.0
19.6
0.3-174.5
61.1-492.1
2.8-18.9
1.2-4.3
4.8-14.4
1.2-492.1
12.6-125.0
6.3
1.0-13.9
1.0-125.0
9.4-44.9
0.0-184.9

25.1/69.7
12.0

57.1
89.4
24.5

21.0/22.9
140.2
8.3
2.4/2.9
7.5
12.0

47.0
200.9
9.2
2.7
8.8
83.0
68.8

12.6
26.7
48.8/49.3

7.4
31.7
27.1
70.7

See Table 4 footnote 1". Regardlessof location and other variables.

Regardlessof the length of the samplingperiod.
Regardlessof the quantity of fertilizer applied.
-~.
Includes mixedN amendmentsand fertilizer application >250 kg N ha
Table 7. Fertilizer-derived N20emissions estimated for five fertilizer
types based on percent of fertilizer
evolved daring the sampling
period only.~"
Fert.
type

Fert.
used
Tg

AA
AN
A
U
N

8.5
16.2
6.3
24.7
14.8
Total

Emissions during sampling period*

Range

Avg.

Median

Gg N20-N
72.8-578.7
6.5-277.3
1.6-57.1
16.5-44.4
0.1-74.0
97.5-1031.5

228.5
71.3
15.9
27.1
11.1
353.9

137.6
64.4
7.4
28.4
4.9
242.7

1" See Table4 footnote 1..

:[:Basedon Table 5. Regardlessof soil system, location, length of sampling
period, and other variables.
Basedon estimates by Inez Funggroupedto fit this analysis. Personal communication, June 1988. Of the 70 503 197 MgNfertilizer consumedworldwide in 1984 (FAO, 1985), 0.09% was assumed to be A; 0.23% AN;0.35%
U; 0.12% AA; and 0.21% N.
Highmedian value used.

and sometimes the foliage and roots were removed

before the sampling procedure, the relationship betweenemissionsand the type of soil system is not well
documentedin the data set. Several grass system experimentsdid not disturb the site prior to fertilization;
due to the lack of cultivation, this data mightresemble
fertilizer-derived emissions from lawns.
Basedon the studies surveyed, mostof the fertilizer-

derived N20 from agriculture land has been found to

be released during the growing season, often shortly
after fertilization, although a significant amounthas
been found to be released during spring thaw. Because
the sampling procedures are expensive and time consuming, few studies have been conducted over a long
enoughtime period to capture annual emission trends.
Several experiments were concluded when the evolution of N20 at the fertilized sites reached the background levels at the corresponding control site. The
existence of emissions released after this time period
was not documented. In most of the experiments surveyed, the duration of the sampling period captures
the large, often episodic flux that occurs shortly after
fertilization. Thegradual loss of residual fertilizer N
as N20,whichis difficult to detect and quantify with
existing experimental designs, is largely unknown.
However,it is considered a negligible componentof
the total induced emission. Loss at spring thaw was
not considered in most of the studies.
The reliability of the emissiondata of each experiment depends on how well the sampled flux compare
to the mean flux rate during a given time period
(Goodroadet al., 1984), howwell the interpolations
between samples during the sampling period compare
to the actual emissions, and howmuchvariability exists betweenthe monitoring techniques of the studies
surveyed.

EICHNER: N20 EMISSIONSFROMSOIL

Theoretically, the variability amongthe emission

samplingtechniques--whichaffect the level of soil disturbance, air circulation, incomingradiation, gas mixing rate, and gas contamination--should account for
differences amongthe emission samples detected.
However,because the level of emissions is reasonably
similar across experiments, this variability does not
appearto greatly affect the reliability of the emissions
data.
Total emission and average daily emission do not
consider the quantity of fertilizer applied, and percent
of fertilizer applied does not considerthe length of the
sampling period. Averagedaily emission per kilogram
of N fertilizer applied mightbe a better overall indicator of fertilizer-derived emissions. However,this as
well as the average daily emission are influenced by
the variability of emissions throughout the sampling
period.
The emission coefficients provided in Tables 2 and
3 do not include N20emissions from fertilizer lost in
the drainage water, streams, and groundwater, nor the
fertilizer-derived N20emitted in addition to those
measured during the sampling period. Estimating
annual emissions from a given amountof fertilizer
requires an assumptionabout the level of fertilizerderived emissions beyond the sampling period.
Comparing emissions data from experiments with
sampling periods of various lengths, using different
types and amountsof fertilizer, and different application techniques should be done with caution.
Global AnnualEmission Estimates
Because of the complexity and variability of N20
emissions fromnatural and fertilized soil systems, extrapolating the available data to assess global emissions remainsdifficult. However,this data does reasonablyassure that the magnitudeof fertilizer-derived
emissionsis accurate.
To obtain an estimate of the total annual loss of
fertilizer-derived
N20, the annual consumption of
each fertilizer type is multiplied by the percent of the
fertilizer evolved during the sampling period as summarized in Table 5 and presented in Table 7. Emissions by fertilizer type are summedto obtain an estimate of the total annual atmospheric burden of
fertilizer-derived N20during the "samplingpe/-iod."
This procedureis not repeated for average daily emission or average daily emissionper kilogramof fertilizer applied, because this would require speculation
about the duration of expectedfertilizer-derived emissions. There was no attempt to calculate an estimate
of total annual emissions based on the soil system or
crop type becauseof the uncertainty of the data set.
Basedon the fertilizer use estimates, the average
total annual loss of fertilizer-derived N20was 354 Gg
N20-Nyr -~ during the sampling period; the median
total loss 243 GgN20-Nyr-~ during the sampling period. If these estimates are doubled to account for
emissions after the sampling period and emissions
fromfertilizer lost in drainage water and groundwater,
the expected range wouldbe 0.2 to 2.1 Tg N20-N(avg.
0.7 Tg N20-N; median 0.5 Tg N20-N) emitted into
the atmospherein 1984. The magnitude of this estimate is in agreementwith recent global estimates.
If 100 Tg N fertilizer are consumedworldwide in

279

the year 2000, the global release of fertilizer-derived

emissions could contribute 0.1 to 1.5 Tg N20-N(avg.
0.5 Tg N20-N;median 0.3 Tg N20-N)during the sampiing period and will probably not exceed 3 Tg N20N into the atmospherein the year 2000. However,this
will largely dependon the caveats above, agriculture
practices and policies, trends amongthe types of fertilizers used, natural processes, and the applicability
of the emission rates in the data set to N20emissions
fromfertilizer used in regions whereemissionsdata is
not available and from fertilizer used in different soil
systems such as flooded agriculture.
Future Research Needs
The existence of a historical changein baseline N20
emissionsfrom agriculture soils as well as the effects
of dry N precipitation and acid rain are unknown.
Natural emissions need to be monitored over longer
periods to obtain better baseline data. Ideally, future
studies should monitor manysites in a given area,
manytimes a day over an extended period of time.
The total annual emission trend needs to be explored
for different types of fertilized ecosystemsand for different crops, including emissions during winter
months and at spring thaw, using uniform sampling
techniques. Special emphasisshould be placed on the
relationship between agriculture managementpractices and environmental conditions and the level of
fertilizer-derived emissions.
Improvedsampling techniques and devices that allow for repeated sampling without significant perturbation of emission rates need to be developedas well
as sampling devices capable of monitoring low levels
of emissions and emissions from larger plants. The
amountof N20emitted from tall crops and crops fertilized after the plant has grownhave not been sampied.
The effect of agriculture and fertilizer management
practices on the production and release of both natural
and fertilizer-derived emissions need to be explored.
For example,the effect of dry verses liquid-applied;
broadcasting verses placement in concentrated areas;
more than one application during the growingseason;
the use of mixedfertilizers, mixedcropping, and agriculture chemicals; more than one harvest per growing season; and soil disturbance, different tillage practices, and different fertilizer application procedures.
The fertilizer-derived
emissions from flooded and
transplanted rice systems need further study. The N20
emissionsfrom alternative uses of fertilizer such as in
aquaculture are not known.
The fate of fertilizer leached into drainage waters
and subsequent N20release is not well understood
and warrants further study. Researchis neededto increase fertilizer efficiency, to minimizethe loss into
water systems, and to reduce the loss as N20.Learning
to distinguish whether N20 emissions are produced
from a nitrification or a denitrification process would
contribute to the understandingof emissions from soil
systems.
Better data on agriculture practices and fertilizer use
need to be gathered worldwide. The impact of the increasing use of mineral fertilizer on the use of animal
manure and the subsequent manure-derived N20 is
not known.

280

J. ENVIRON. QUAL., VOL. 19, APRIL-JUNE 1990

ACKNOWLEDGMENTS
The author is grateful to many soil scientists for helpful
discussions and to Lauretta Burke and Dennis Tirpak for
their assistance. This article does not necessarily reflect the
official position of the USEPA.