GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 13, NO 2, PAGES 647-662, JUNE 1999
Nitrogen in crop production: An accountof global flows
Vaclav
Smil
Departmentof Geography,Universityof Manitoba,Winnipeg,Manitoba,Canada.
Abstract. Human activitieshaveroughlydoubledthe amountof reactiveN that entersthe
element's
biospheric
cycle.Cropproduction
is by far thesinglelargestcauseof thisanthropogenic
alteration.
Inorganic
fertilizers
nowprovide
80TgN yr-1(Tg= 1012g),managed
(symbiotic)
biofixation
addsabout20TgN yr-1, andbetween
28and36TgN yr-1arerecycled
in organic
wastes.Anthropogenic
inputs(includingN in seedsandirrigationwater)now supplyabout85%
of 170 (151-186) Tg N reachingthe world'scroplandevery year. About half of this input, 85
TgN yr-1, istakenupbyharvested
crops
andtheirresidues.
Quantification
ofN losses
fromcrop
fields is besetby major uncertainties.Lossesto the atmosphere(denitrificationand volatilization)
amount
to26-60TgN yr-1, whilewaters
receive
(fromleaching
anderosion)
32-45TgN yr-1
TheseN lossesare the major reasonbehindthe growingconcernsaboutthe enrichmentof the
biospherewith reactiveN. The bestevidencesuggests
thatin spiteof somesignificantlocal and
regionallosses,the world'sagriculturallandaccumulates
N. The additionof 3-4 billionpeople
beforetheyear2050 will requirefurthersubstantial
increases
of N inputin cropping,but a large
shareof this demandcancomefrom improvedefficiencyof N fertilizeruse.
1. Introduction
the nutrient suppliedby biofixation; outputsinclude, besidesthe
Human interferencein the global nitrogencycle hasbecomea
topic of increasingresearchattention[Smil, 1990, 1991, 1997a;
Kinzig and Socolow, 1994; Galloway et al., 1995' Jordan and
Weller, 1996; Vitousek et al., 1997]. Compared to the
preindustrial era, human activities have roughly doubled the
amountof reactive N that entersthe element'sbiosphericcycle
[Galloway et al., 1995; Smil, 1997a]. These compoundsinclude
NOx (NO andNO2) andotheroxygenated
single-Nspecies
(the
harvested
biomass,
thelosses
dueto gaseous
emissions
of NO, N2
andN20, volatilization
of NH3, leaching,
soilerosion,andlosses
from tops of plants. In conclusionI evaluate the N balanceof the
global cropping and offer a brief assessmentof ways to reduce
the existingN burden.
Many quantifications in this paper (i.e., nitrogen removed
from fields in crops and in crop residues and nitrogen lost in
leaching and in eroded soils) are my original calculationsbased
on the best available
twocategories
jointly
designated
asNOy),
NH3 andNH4
+(labeled detailed in
NHx), and organicN, and therehasbeenalsoan increasein
concentrations
of N20, a greenhouse
gas. Effectsof this N
statistics
and on other relevant
information
the text; others are derived from a wide-ranging
surveyof rapidlyincreasing
literatureon nitrogencycling.
Becauseof our poor understanding
of many fluxesof the
enrichment range from atmospheric changes (higher
complexnitrogen
cycleandbecause
of considerable
spatial,as
concentrations
of N Ox, NO3-,andNHx) to alterations
of terrestrial wellassecular,
variationof mostof theflows,quantifications
of
and aquatic ecosystems (caused by acidification, nutrient
leaching,and eutrophication)and to impactson carboncycle.
Crop production is by far the single largest cause of human
alteration of the global N cycle: Rhizobium bacteria symbiotic
with leguminouscropsfix much more N than would be the caseif
naturalplant communitieswere occupyingthe samespace,and N
global N fluxes shouldavoid an appearance
of unwarranted
accuracy.Consequently,
I will presentall inputsand losses
(Tables5 and6) asranges
ratherthanassinglevalues.
applied
to fieldsin synthetic
fertilizers
(nowabout80Tg N yr-l)
Paper number 1999GB900015.
The Food and Agriculture Organization's(FAO) worldwide
databaseis the most comprehensivesourceof statisticson arable
land, annual crop production, and yields (see FAOSTAT
Agriculture Data available as http://apps.fao.org,hereinafter
referredto asFAOSTAT Data). Accuracyof thesefiguresis high
(errorsmostly smallerthan 5%) for high-incomenations,but the
FAO must make many in-houseestimatesto fill numerousdata
gaps for low-income countries. FAO statisticsalso do not take
into account home gardensand backyard plots, which make
important nutritional contributions in most Asian, African, and
Latin American countries [Hoogerbrugge and Fresco, 1993].
Moreover,someofficial farmlandfiguressubmittedby the FAO's
member states are known to be substantial (more than 10%)
underestimates.Perhaps the most notable example of such
0886-6236/99/1999GB900015512.00
underestimates is the case of China's farmland: Its real total is
is about4 times as large as is the total amountof the elementthat
humans fix by burning all fossil fuels [Galloway et al., 1995;
Smil, 1997a].
This paper accountsfirst for nitrogentaken up by the global
harvest of agricultural crops and their residues, and then it
reviewsall major naturaland anthropogenic
flows of N reaching,
and leaving, the world's crop fields. Quantified inputs include
nitrogen in atmospheric deposition, seeds, irrigation water,
recycledcrop residues,animal manures,inorganicfertilizers,and
Copyright1999by theAmericanGeophyscial
Union.
647
2. Global Crop Harvest
648
SMIL: NITROGEN
IN CROP PRODUCTION
50% largerthanthe official claim, at least 140 Mha ratherthan95
Mha [State Statistical Bureau, 1997; National Intelligence
Council, 1997].
AN ACCOUNT OF GLOBAL FLOWS
Table 2. NitrogenIncorporatedin theWorld Crop Harvestof the
Mid-1990s
Crops
HarvestedCrops
Crop Residues
almost
certainly
largerthantheFAOtotalof 1.477Gha(1.362
Cereals
30
15
45
Gha of arable land and 115 Mha of land under permanentcrops)
and so is the aggregatecrop harvest.Increasesof up to 5% appear
to be highly plausible,but the absenceof detailed information
neededto quantify and to apportiontheseadditionalamountsof
farmland and harvestsprecludesany satisfactorycorrectionsof
FAO's statistics even on a continental basis. According to
uncorrectedFAO data, the mid-1990s global crop harvest,taken
as the mean of 3 years (1994-1996) and excluding the forages
Legumes
10
5
15
Sugarcrops
2
I
3
Roots, tubers
2
I
3
Vegetables,fruits
Othercrops
Forages
2
4
10
I
2
...
3
6
10
Total
60
Consequently,the actual area of globally cultivated land is
25
Total Harvest
85
All figures
arein Tg N yr-l. Freshweights
of crops
anddryweights
grownonfarmland,
averaged
almost
5.5Pg yr-1in freshweight
of cropresiduesfrom Table I aremultipliedby averageN contentsgiven
(Table 1).
The most accurateway to calculateN takenup by this output
is to use N contentsfor harvestedparts (grains, stalks, leaves,
by NRC [1972], Watt and Merrill [1975], Smil [1994], and Bath et al.
roots,andfruits) of everymajorcrop.I havedonethiscalculation
for 40 separatecrops aggregatedinto 8 principal categories.
BecauseN contentof cropsis commonlyexpressedin terms of
absolutelydry matter,appropriateconversions
mustbe donefirst.
I used average moisture and N values from National Research
Council (NRC) [1972] and Bath et al. [1997]. The mid-1990s
[1997].
ground
phytomass
[DonaldandHatnblin,1976;Gif•brdand
Evans,1981;Hay, 1995]. Residualbiomassexpressed
as a
multipleof harvested
yieldscanbe obtained
simplyasa ratioof
(1-HI)/HI, but this total will be somewhatlargerthan the
globalcropharvest
of about2.65Pgyr4 of absolutely
drymatter commonlyusedresidue/crop(R/C) ratios as the latter indices
may not includeplant stubble.
contained
50 Tg N yr-l (Table2). Cereals
accounted
for 60%of
Thestandard
practice
is to quotetheresidue
in termsof drytotal dry crop mass,and they also contained60% of all removed
N; sugarcropsranked secondin dry mass(16%), but legume
cropswere secondin assimilatedN (20%).
Calculations of N incorporatedby crop residuescannot be
done with a comparableaccuracyas no country keeps statistics
on their production.Publishedestimatesof national or global
totals of crop residue production were prepared in order to
evaluateoptionsfor agronomicmanagementand animal feeding
or to find potential contributionsof biomassenergiesor total
emissionsof greenhousegases [United States Department of
Agriculture (USDA), 1978; Smil, 1985, 1994; Kossila, 1985;
Andreae, 1991]. Many figures are available for harvestindices
(HI) of major crops, the ratios of crop yield and total above-
mattermassand the crop yield at field moisture.The R/C ratios
varyamongcultivarsas well asfor the samecultivargrownin
differentenvironments.
Agronomicfactors(plantingdate,
irrigationregime,andN applications)
can alsomakesubstantial
differences
[Priharand Stewart,1991;Robertset al., 1993].
Assumptions
concerningaveragestraw/grain(S/G) ratio of
cerealswill makea particularlylargedifferencein global
calculations.
UsingaverageS/Gratioof 1.3,ratherthan1.2,adds
almost200 Tg more straw, a total larger than all residual
phytomass
produced
bytheworld'stuberandrootcrops.I triedto
minimize
sucherrorsbycalculating
residue
outputseparately
for
40 different
cropsratherthanjustformajorcropcategories;
still,
the grand total could not be calculatedwith an error smaller than
5%.
Table 1. AnnualGlobalHarvestof CropsandCropResidues
in
the Mid- 1990s
Crops
HarvestedCrops
CropResidues TotalHarvest
Tg of
Tg of
Tg of
Tg of
freshweight dryweight dryweight
dryweight
Cereals
1900
1670
2500
Sugarcrops
1450
450
350
4170
800
Roots, tubers
650
130
200
330
Vegetables
600
60
100
160
Fruits
400
60
100
160
Legumes
200
Oil crops
150
190
110
200
100
390
210
Othercrops
Forages
100
2500
80
500
200
280
500
Total
7950
3250
3750
7000
Freshweightsof harvestedcropsare accordingto FAOSTAT Data;
dry weightis calculated
by usingstandard
moisture
contents
by Wattand
Merrill [ 1975]andBathet al. [ 1997];derivationof cropresiduetotalsare
described
in thetext.All figuresareroundedto avoidthe appearance
of
unwarrantedprecision.
I usedFAO's crop productionfigures,standardvaluesfor
water contentof harvestedparts,and fairly conservativeR/C
multipliers[Smil, 1994, 1999]. These calculations result in
outputs
ofbetween
3.5and4.0Pgyr-1ofcropresidues
during
the
mid-1990s;
themostlikelytotalof 3.75Pgis about1.4timesthe
aggregatecropharvest(Table 1). Cerealstraws,stalks,andleaves
accounted
for 2/3 of all residual
phytomass,
andsugarcanetops
andleaveswerethe secondlargestcontributor.
Justover60% of
all residualphytomass
were producedin low-incomecountries,
andcloseto 45% of it wereoriginated
in thetropics.Substantial
interspecificand intraspecificvariationsresultin a ratherbroad
range of N contentsof cereal straws(0.4-1.3%, valuesaround
0.6%arecommon);
onlyleguminous
straws
arerelatively
N-rich.
Thetotalremoved
in residues
amounted
to about25 Tg N yr-1
duringthemid-1990s
or 1/3of thetotaltakenupbycropbiomass
(Table 2).
Quantification
of annualharvests
of foragesgrownon arable
land (alfalfa, clovers,vetches,and variouslegume-grass
mixtures)canbe doneonlyapproximately.
Thesecropsinclude
all-legume
or legume-grass
swards
cultivated
intermittently
on
arableland,eitherfordirectforaging
or,moreoften,forroughage
SMIL: NITROGEN IN CROP PRODUCTION
feed (hay and silage), as well as a variety of greenmanures,i.e.,
leguminousspecieswhich are plowed under after 40-80 days of
growth.FAO doesnot keep track of thesecrops,but there is no
doubtthat their worldwide extent hasbeendeclining:diffusionof
intensivegrain monocultures
and wideruseof syntheticfertilizers
have been the main reasons for these reductions [Smil, 1993;
Wedinand Klopfenstein,1995].
I have consultedstatisticalor agriculturalyearbooksfor about
20 of the world's largest countriesin order to derive the most
likely range of the land under forages grown on cropland:
between 100-120 Mha in the mid- 1990s, with green manures
accounting for no more than 20% of the total area. Fairly
conservativeassumptionsabout meandry-matterharvests,5 Mg
AN ACCOUNT OF GLOBAL FLOWS
649
furtherdownwind
beforetheyareprecipitated
ordeposited
in dry
form. Dry and unpollutedregionshave the lowest ratesof N
deposition,while the annualinputsin the mostintensively
farmed, industrializedand denselypopulatedareasof North
America,EuropeandAsiaarecommonly
anorderof magnitude
higher(seetheNationalAtmospheric
Deposition
Program's
data
available as http://nadp.nrel.colostate.edu/NADP,
hereinafter
referredto asNADP Data,andErisman[ 1995]).
U.S. regionalmeansof N depositedin precipitationare an
excellentillustrationof thesedifferences
(seeNADP Data).Most
of thecountrywestof the Mississippi
annuallyreceives
just 1-2
kgN ha-1inwetdeposition,
withNH4+inputs
negligible
nearly
everywhere west of the Rocky Mountains, while total N in
ha-1foralfalfa(planted
onabout30%of cropland
pastures)
and4 precipitation
averages
over7 kg N ha-1in thecoastal
Northeast,
Mg ha-i for otherleguminous
andmixedlegume-grass
forages, andit would
bemuch
higher
if asmuch
75%ofthearea's
NOx
resultin the total yield of about500 Tg containing10 Tg N. This
would increasethe mid-1990s global dry-matter crop harvestto
emissions
werenot exportedoutto the AtlanticOcean[ Jaworski
et al., 1997]. Europehas even greaterregionaldifferences:At
about7 Pg yr-1, withabout85 Tg N yr-1incorporated
in this
over
20kgN ha-1, theannual
rates
ofNH4+inprecipitation
inthe
phytomass(Tables 1 and 2).
Netherlands,
northeastern
France,andsouthern
Englandaremore
than20 timesashighas thosein southern
Spainor Italy [Van
3. Nitrogen Inputs
Leeuwen et al., 1996].
I used
separate
estimates
ofcontinental
averages
ofwetNOy
A smallpart of harvestedN returnsto fieldsin plantedseeds andNH x deposition
rates,derivedfrom long-termdeposition
and tubers.All fields receiveN in dry and wet depositionof
measurements(see NADP Data and Erisman [1995]), from
airbornecompounds
andin recycledrootsandstubble.Irrigated emissioninventories[Bouwmanet al., 1997] and from
fields receiveadditionalN, largely from nitratesdissolvedin
approximate
globalbalances[BernerandBerner, 1996],and
water. More intensiverecyclingof strawsand stalksis common FAOSTATData's
farmland
statistics
to calculate
thetotalinput
in manyagroecosystems
andin all regions,withmixedfarming
of about13(11-15)Tg N yr-l. Aswehavenocomparatively
animal manuresproduced'
in confinement
also recycled. widespread
andreliablemeasurements
of dry deposition,
all
Biofixationcontributesrelativelysmallamountsof N in dry generalizations
are highly uncertain.Minimum estimatescan be
fieldsplantedto nonleguminous
crops,butit is a majorsourceof
derived
byassuming
thatdryNHx deposition
isequal
toatleast
the nutrientin leguminous
culturesandin rice fields.Although 1/3ofthewetfluxand
thatdryNOyisequal
toatleast
3/4ofthe
many fields, especiallyin Africa, have never receivedany wetinput[Warneck,
1988].Theactual
rates,
particularly
fordry
inorganicfertilizer,syntheticN compounds
are nowthe single NHx deposition,
maybemuchhigher[Sutton
etal., 1993].
largestinputof thenutrientin theworld'scropproduction.
Theseadjustments
wouldresultin totaldeposition
of about20
(18-22) Tg N on the world'sagriculturalland. As the total
3.1. Planted Seeds and Tubers
Thisis a minorinput,whichis fairly easyto quantify.Many
agriculturaldata sourcescarry informationon seedingrates,
mostlyin termsof massunits per area, sometimesas sharesof
harvestedcrops. Mean seeding rates for more than a dozen
terrestrial
deposition
ofNOyandNHx isnowabout
60TgN yrq
[Gallowayet al., 1995;Bernerand Berner, 1996],this would
meanthatat least1/3 of all reactiveN deposited
on landsettles
on the farmlandwhichaccounts
only for some11% of ice-free
surfaces.
Thishigherlevelof enrichment
is expected
owingto a
regionsand nonagricultural
principalcropsarelistedin detailedfoodbalancesheets
prepared closeproximityof agricultural
sources
of
NO
x
emissions
(particularly
in Europe,eastern
North
by the FAOSTAT Data and multipliedby respectiveplanted
andeastAsia)andto a large
share
ofNOxemissions
areastabulatedin the FAOSTAT Data. In turn, theseproducts America,
were multiplied by averageN contentsof seedsand tubersused
to calculateN incorporatedin the globalcropharvest(Table 2).
emanatingfrom agriculturalactivities.
This procedureresultedin an annualreturnof about2 Tg N in
3.3. Irrigation Water
seeds and tubers.
About17%of theworld's
cropland
(250Mha)wereirrigated
in 1995,with nearly2/3 of the globaltotalin Asia(FAOSTAT
3.2. Atmospheric Deposition
Data). Irrigation waters always contain some N, and its
Higherproductionof nitrates (generatedmainlyby oxidation concentrations
arerelatively
high(morethan5 mgN kgq) in all
of NOx released
by combustion
of fossilfuels)andvolatilization intensively
cultivated
regions(wherethe leaching
of N from
aswell asin
of NH 3 (from animal wastes,soils receivingammoniacal fertilizersandanimalwastesarethemainsources)
fertilizers,andplanttops)havebeenthe maincausesof steadily densely
populated
andheavilyindustrialized
areas(wherehuman
increasing wet and dry deposition of N compoundson wastes
andindustrial
pollutants
arethemaincontributors).
agriculturalland.Becauseof thebrief atmospheric
residence
time
of ammonia
compounds,
largeshares
of NHx originating
from
Annual
inputs
ofthenutrient
inirrigation
watermaybehigher
than
20kgN haq indouble-cropped
rice[Wetselaar
etal.,1981
],
ortheymaybebelow5 kgN ha-l inthose
predominantly
rain-fed
agriculturalactivitieseither fall back on the agriculturalland
from which they originatedor are depositedcloseto the areasof
cropsthat receiveonly somesupplementary
irrigation.With
theiremissions;
incontrast,
NOycompounds
arecarried
much
average water applicationsof at least 9000-10,000 m3 haq
650
SMIL: NITROGEN IN CROP PRODUCTION
[Shiklomanov,
1993; Postel et al.,
1996] and with N
AN ACCOUNT OF GLOBAL FLOWS
high-income and 40% in low-income countries, would return
concentrations
mostlybetween
1-2mgN kg-l, theannual
global about 12 Tg N yr
-1
inputsof the nutrientwouldbe about4 (3-5) Tg N.
3.5.
3.4. Recycling of Crop Residues
Recyclingof crop re sidues,directlyby leavingthem after the
harvestto decay on field surfacesor by incorporatingthem into
soil by ploughing,discing,or chiselling,and indirectlyby using
them in mulches and compostsor returning them to fields in
animalwastes,hasbeenpracticedvigorouslyby everytraditional
agriculture, and it remains an essentialpart of modern field
management [Smil, 1999]. Besides the nutrient recycling,
protection against water and wind erosion, enhanced water
storage capacity of soils, and their enrichmentwith organic
matterare otherprincipalbenefits[Barreveld, 1989].
Crop residueshave many competinguses:They have been an
importantsourceof householdfuel andbuildingmaterialin many
low-income countries;they provide indispensablebeddingand
feed for animals, particularly ruminants,of all continents;they
offer an excellent substrate for cultivation of mushrooms; and
they have been used for making paper and as feedstocksfor
extractingorganiccompounds.Unfortunately,a significantshare
of crop residues is still burnt in fields, an agronomically
undesirablepracticewhich also generatesgreenhouse,and other
trace, gases[Andreae, 1991].
No country keeps comprehensivestatisticsof crop residue
uses. Fairly reliable information on major usesis available for
some affluent countries, but elsewhere we have to rely on
fragmentarydata and expert estimates.IntergovernmentalPanel
on Climate Changeand Organizationfor EconomicCooperation
and Development( IPCC and OECD) [ 1997] estimatedthat about
25% of all residues are burnt in low-income
countries and that the
correspondingsharein affluent nationsis just 10%. The first rate
is almostcertainlytoo conservative,especiallywhen one includes
the use of residues for fuel: Widespread energy shortagesin
deforested rural areas make the residues the only accessible
householdfuel for hundredsof millionsof poorpeasants.
Chinese
estimatesindicatethat at least3/5 of all cropresiduesareburntby
rural households[Smil, 1993; Sinton, 1996].
The estimated mean for residue burning in high-income
countriesis also most likely too low, as somefairly reliable data
on average burn fractions indicate regionally much higher rates
both for field and orchardcrops[Jenkinset al., 1992]. Adding up
field and householdcombustion,a more realistic assumption
would
be that 35% of all residues in low-income
countries
and
15% of all residuesin affluent nationsare burnt.Burningof about
1 Pg of crop residuescontainingroughly 6 Tg N would release,
assuming 80% combustion efficiency, almost 5 Tg N. These
emissions
wouldnotbemadeuponlyof NOx andNH3:30-40%
of the element present in the phytomassis convertedduring
flamingcombustion
directlyintoN 2 [Kuhlbusch
etal., 1991].
Feed and bedding account for anywhere between 20-25% of
the remainingtotal in high-incomenationsand for up to 1/3 in
low-income countries with large numbersof domesticanimals.
Other usesare negligible,particularlywhen comparedto inherent
errors in estimating the productionof crop residuesin the first
place. Consequently,the highestplausiblerecyclingrates,about
70% in high-income and 60% in low-income countries,would
return annually about 16 Tg N to the world's croplands.The
lowest plausibleestimates,with averagerecyclingratesof 60% in
Animal
Manures
Manure production,and its N content,dependson breed,sex,
age, health, feeding, and water intake of animals [S•nil, 1985].
Freshwasteoutputper headcan be estimatedfairly reliably as a
shareof animal's live weight, but large variationsin body mass
and feeding, especially the differences between confined and
well-fed animals in modern settingsand traditional,free-ranging
and poorly fed (and often still also hard-working)breeds,make
even national
means uncertain.
Choosing the best means of N content of manures is even
more difficult as various reports list twofold to threefold ranges
for dairy, pig, and poultry manures[Misra and Hesse, 1983;
Whiteheadand Raistrick, 1993; Choudharyet al., 1996]. Powers
et al. [1975] found an even greaterrange (0.6-4.9% N) for beef
manures.Using the meanslisted in Table 3 and multiplying them
by the FAOSTAT Data animal head counts of the mid-1990s
resultsin annual voiding of about75 (70-80) Tg N. This is less
than severalother recently publishedestimatesby Mosier et al.
[ 1998]; Bouwmanet al. [ 1997] and Nevisonet al. [ 1996].
The main reason for this difference is that I am assuming
lower averages of body weights and poorer feeds for cattle,
sheep,and goatsin low-income countries(they now have about
75% of the world's bovines and almost 80% of all sheep and
goats).Even so, my estimatefor nitrogenin cattlemanuremay be
still be too large as some sourcecredit Indian cattle manurewith
just 0.7% N on a dry-weight basis[Singhand Balasubramanian,
1980].
Quantifying the amount of N returned to cropland is even
more
uncertain
because
of
the
need
for
concatenated
assumptions.Only animals grazing on cropland pastures,
harvestedfields, or cover cropsin tree plantationswill deposit
their wastesdirectlyon the farmland:This contributionis only a
small shareof all N recycledin manure.Wastesproducedby
Table 3. AverageAnnualProduction
of AnimalWasteSolidsand
TheirTypicalNitrogenContent
Animals
Waste Solids,
N,
N Output,
kghead-l
%
kghead-l Producedin Losses,
Manure NH3-N
Confinement,
%
%
Dairy cattle
Modem
Traditional
2000
1200
4
3
80
45
60
65
36
36
Modem
Traditional
Waterbuffalo
Horses,mules
Pigs
1200
900
900
1200
300
4
3
3
3
3
50
30
30
35
10
35
25
35
50
100
36
36
28
28
36
Sheep,goats
Poultry
150
8
3
4
10
100
28
36
Nondairy cattle
5
0.3
Thefollowing
average
liveweights
areassumed:
modern
dairycattle,
500kg;modem
nondairy
cattle
andhorses,
400kg;traditional
dairyand
nondairy
cattleandwaterbuffaloes,
300kg;pigs,100kg;sheep
and
goats,
40 kg;andpoultry,1.5kg [MisraandHesse,1983,Smil,1983,
Nordstedt,1992,andBouwman
etal., 1997].
SMIL: NITROGENIN CROPPRODUCTIONAN ACCOUNTOF GLOBALFLOWS
animals grazing on permanentpastureswill be unavailablefor
recycling to cropland; only manure producedin confinementon
mixed farms and in feeding facilities locatedin farming areascan
be economicallydistributedto nearbyfields.
Recyclingratesrangefrom only about30% in the USA, where
about 40%
of all animal
wastes are voided
in confinement
40% of the nutrient
in all animal
Table 4. Ranges of Published Biofixation Estimates and the
Mean Values Used for CalculatingGlobal Biofixationby Major
LeguminousSpecies
Crops
and
about 75% of this output are actually returnedto fields [USDA,
1978], to more than 90% in several small European countries
[Rulkensand ten Have, 1994]. Even greaterdifferencesare seen
in Asia: Hardly any manureis usedas fertilizer in the continent's
interior; a large shareof cattle manurein the Indian subcontinent
is gatheredfor fuel, while about 80% of China's pig manureis
recycled.
Using average shares of manure produced in confinement
(Table 3) that are nearly identical to those of Bouwman et al.
[1997], I calculated the total amount of N voided annually in
stables,barns, sheds, and corrals at 30 (25-35) Tg N or about
manures. Even if this manure
were to be completely recycled its initial N content would be
greatly diminished before these organic wastes could enrich
agriculturalsoils:with more than 70% of urine N voidedas urea,
there are large lossesdue to rapid hydrolysisof the compound
and subsequentvolatilizationof NH 3 during collection,storage,
composting,and handlingof wastes[SubbaRao, 1988; Eghball
651
Rangesof Published
Rangesof Means
Estimates*,
Used in Calculations,
kgN ha-•
kgN ha-I
Seedlegumes
Beans
Broad beans
Chickpeas
3-160
45-300
30-40-50
80-100-120
3-141
40-50-60
Lentils
10-192
Peas
10-244
30-40-50
30-40-50
Peanuts
37-206
60-80-100
Soybeans
Otherpulses
15-450
7-235
60-80-100
40-60-80
Forages
Alfalfa
65-600
150-200-250
Clovers
28-300
130-150-170
Otherforages
9-180
80-100-120
* Data are from LaRue and Patterson [ 1981], Heichel [1987], Giller
and Wilson[1991],Peopleset al. [1995],and Smil[1997b]
et al., 1997; Bouwman et al., 1997].
These losses,together with denitrification and leaching, are
especiallyhigh in traditionalsettingswith prolongedstorageof
manuresin the open. Depending on the kind of handling and
concentrationsof inorganic forms of N, shortagesof essential
micronutrients(particularly Mo, V, and Fe presentin the redox
centerof nitrogenase[Eady, 1995]), and low soil pH.
storage,mostor nearlyall of NH3-N (up to 50% of initially
Dozens of N fixation
rates from a number of different
voidedN), aswell asappreciable
amounts
of NO3-N andorganic agroecosystemshave been publishedfor such widely cultivated
N, may be lost before the applicationto fields. Assumingthat
90% of manureproducedin confinementis eventuallyrecycledto
fields (the restis dumpedor recycledto pastures)andreducingits
leguminousfood and feed crops as common beans (Phaseolus
vulgaris) and soybeans(Glycine max) as well as for suchleading
forage species as alfalfa (Medicago) and various clovers
initialN content
by average
NH3-N losses
proposed
by Bouwman (Trifolium). Almost all publishedvalues have at least three-fold
et al. [1997] and listed in Table 3, ! calculated that recycled
ranges and much larger differences are common (Table 4).
animal manurescontribute about 18 Tg N (14-22) Tg N. I will
Particularly notable are low-fixation rates in common beans in
estimate
specific
post-application
NH 3 losses
in section
4.
Latin America, the crop'sregionof origin [Henson, 1993; Tsai et
al., 1993]. For lesscommonleguminousspecies,we havetoo few
3.6. Biofixation
reliable figures; our knowledgeof fixation rates is particularly
Reduction of atmosphericN2 to NH 3 is performed inadequate as far as tropical legumes, be they used for food,
enzymaticallyby at least60 generaof cyanobacteria(blue-green fodder, or as green manure, are concerned [Blair et al., 1990;
Peoplesand Herridge, 1990].
algae), 15 genera of symbiotic actinomycetes,and, above all,
In order to make the best possible estimate of symbiotic
some25 generaof free-livingandsymbioticbacteria.By far, the
most important symbionts belong to the genus Rhizobium biofixation, I have usedspecificrangesfor eight differentkinds
associatedwith leguminous plants. Nearly a generation ago,
of seedlegumeslistedin Table 4 (beans,broadbeans,chickpeas,
LaRue and Patterson [1981] noted that there was not a single
lentils, peas, peanuts, soybeans, and other pulses) whose
cultivated areas were taken from FAOSTAT
Data. This resulted in
legume crop for which we had valid estimatesof N fixation.
Rhizobium
fixationof 10Tg N yrq
Although many new estimateshave been publishedsince that themostlikelytotalof global
time, we are still unableto offer reliable, representativevaluesof
(rangeof 8-12Tg N yr-1), whileleguminous
cropsremoved15
average annual fixation rates even for the most important Tg N yr-1 (Table2). Thismeans
thatthesecropsderived
2/3 of
leguminouscultivars.This is not surprisingwhenconsidering
the their N from biofixation, an averagesharein excellentagreement
with numerous
studies of biofixation's
contribution
to N
large natural variability of symbiotic fixation, as well as the
errorsinherentin techniquescommonlyusedto measurethe rates requirementsof legumes[Hardarson, 1993; Peopleset al., 1995].
This does not mean that there is no transfer of N from seed
of N fixation [ Hardarson and Danso, 1993; Danso et al., 1993].
Fixation rates vary enormously with both abundance and
legumes to subsequent nonleguminous crops. This eventual
persistence
of specificRhizobiumstrainsin soils,with the vigor
enrichmentdependson the degreeto which legumescan satisfy
their total N needs from the biofixation
and on the share of the
of host plants and with environmental stresses[Ayanaba and
Dart, 1977; Broughton and Puhler, 1986]. Major factors
fixed N taken up by the harvestedlegume seeds [Hardarson,
reducing,or even inhibiting, N fixation includesoil temperature, 1993; Giller et al., 1994]. The latter sharerangesfrom as little as
bothtoolittle andtoo muchmoisture,
low soil0 2 levels,high 30% for common beans to more than 80% for soybeans.
652
SMIL: NITROGEN
IN CROP PRODUCTION
Completerecyclingof beanplantresidueswill thustransfermost
of the fixed N for eventual use by subsequentcrops;however,
soybeans,althoughthey are much more prolific N fixers than
beans,may not be able to provideall of the needednutrients,and
the crop will have to claim considerableamountsof soil N
[Heichel , 1987].
Estimatingbiofixation by leguminouscover cropsis a much
greaterchallenge.As alreadynoted,foragesand greenmanures
occupy100-120 Mha of farmland.Averagefixationratesof 200
AN ACCOUNT
OF GLOBAL FLOWS
3.7. Inorganic Fertilizers
This is the mostaccuratelyknown,andnow alsothe single
largest,
inputintotheglobal
N cycle.
Mineral
compounds
(KNO3
andNaNO3)arestillused,buttheirapplications
aredwarfedby
the use of syntheticfertilizers.Their productionnow begins
invariably
withtheHaber-Bosch
process
of NH3 synthesis
first
introducedcommercially in 1913 [ United Nations Industrial
Development Organization and International Fertilizer
Development
Center(UNIDO andIFDC), 1998].Duringthemid-
kg N ha-1for alfalfaand150kg N ha-1forclovers
andvetches 1990sthenominalcapacityof theworld'sammoniaplants,now
are well supportedby numerous measurements[Frame and
Newbould, 1986; Dovrat, 1993; Meetu et at., 1994], but fixation
by legume-grass
mixturesis muchmorevariable,rangingfrom a
justa fewkilograms
to morethan250kg N ha-1 [Heicheland
Henjum, 1991; Farnham and George, 1994]; I will assumejust
usinglargelyCH4 bothasthefeedstock
(H source;
N is separated
fromthe air) andfuel, wasabout115Tg N yr-l [Foodand
AgricultureOrganization(FAO), 1997]. Actual synthesis
was
about95 Tg N yr-1, of whichalmost80 Tg N yr-•was
incorporatedin syntheticfertilizerswith the remainderconsumed
50kgN ha-1. A fairlyconservative
totalof N fixationin 100-120 by chemical industries and lost during processingand
Mha of croplandforagesand green manures(averaging100 kg
N ha-1) isthenabout12TgN yr-I (range
of 10-14TgN yr4 ).
As thesecropsremoveannuallyonly about10 Tg N, therecan
be no doubt that they leave behind large amountsof N for
subsequentnonleguminouscrops. The actual total is almost
certainly somewhat higher as it includes biofixation by
leguminouscropsgrown as soil coversin manytree plantations
as well fixation by trees and shrubsused in alley croppingand
planted along field boundaries and house gardens for feed,
fuelwood, and shade[Blair et at., 1990; Gutteridgeand Shelton,
1994]. Biofixation by non-Rhizobiumdiazotrophsis of lesser
importance.Some studiesreport valuesfor biofixation by freeliving bacteria and cyanobacteria in cereal fields in humid
environments
in excessof 20 kg N ha-1 duringthe growing
season[Neyra and Dobereiner, 1977], but typical rates in drier
environments
aremostlylessthan5 kg N ha-1or nomorethan4
(2-6)TgN yr-1fortheworld's
cereal,
tuber,andoilcropfields.
transportation.
Consumption
data,global,regional,andnational,arereported
annuallyby the FAOSTAT Data as well as by the International
Fertilizer Association (see the World Nitrogen Fertilizer
Consumption
reportavailableashttp://www.fertilizer.org).
After
a slightdeclineduringtheearly 1990s(to 72.5 Tg N by 1993),
globalconsumption
roseagainto almost79 Tg N in 1995andto
nearly 83 Tg N in 1996. Theseconsumption
figuresdo not
account for a variety of preapplication losses which are
particularly
largeforthehighlyvolatile
NH4HCO3,stilla major
fertilizerin China [Smil, 1993]. In addition,someN fertilizersare
usedon permanentpasturesand in forests,and increasing
quantitiesare alsoappliedto lawns.
Ammonia is either used directly as the most highly
concentrated
(82% N) andthe cheapest
N fertilizer(appliedas
gas or in aqueoussolutions)or convertedto urea, the most
concentrated
solidfertilizer(46% N) nowaccounting
for nearly
40% of globalN consumption
andgreatlypreferredin Asia,the
ricefieldscanfix 20-30kg N ha4 duringthegrowingseason. continentwhichnow usesnearly3/5 of the world'sfertilizerN.
When including the contributions by Anabaena azollae, a
Variousnitratesor compound
fertilizers(combinations
withP, K,
symbiont ofAzotla pinnata, a small floating fern cultivated in
andmicronutrients)
are alsoused.About60% of synthetic
N
In contrast, cyanobacteria(mainly Anabaena andNostoe) in
Asia'spaddies,
whichcanproduce
asmuchas50-90kg N ha-I
fertilizers are now consumed in the low- and middle-income
during 40-60 days [Meelu et aL, 1994], and when taking into
accountthe rate of rice multicropping(mean global ratio of about
1.25), cyanobacterialfixation may have contributedbetween4-6
countries,
andthenutrientis indispensable
for maintaining
the
Tg N yr-1in 150Mhaof theworld's
paddies.
A recentdiscovery
recentlyachievedstaplefoodself-sufficiency
in theworld'smost
populous
nations(China,India,andIndonesia).
Annual applicationmeansof the mid-1990s were around50
of endophytic diazotrophs (Acetobacter and Herbaspiritlum)
kgN ha-1of arable
landfortheworld,
andtheyranged
between
living inside sugar cane roots, stems, and leaves explains why
10-100
kgN ha-1 forcontinents
(AfricaandEurope
being
the
unfertilized crops maintain high yields even after years of
extremes).
Nationalapplication
meanshideenormous
regional
consecutivecultivation[Boddeyet at., 1995]. Theseendophytes variations:
Forexample,
theU.S.meanof about50 kg N ha-l is
fix annually
at least50 kg N ha-I , maxima
maybehigherthan madeupof ratesbelow20 kgN ha-1in NorthDakotawheatfields
150kg N ha-1, resulting
in therangeof 1-3Tg N fortheworld's towellover200kgN ha4 in Iowacornfields.
Application
rates
18 Mha of sugarcane.
shouldbe alsoadjusted
for variousdegrees
of multicropping:
As calculated above, the total biofixation in crop fields,
plantations,and croplandpastures,a combinationof contributions
by Rhizobium-legume
symbioses
in food,feed,forage,andgreen
manure crops; symbiotic Azotta-Anabaena fixation in paddies;
free-living cyanobacteriaand bacteriain wet and dry fields; and
endophyticbacteria in sugar cane, would have been 33 (25-41)
High fertilizationratesin Asia'smonsoonal
agricultures,
where
double-andeventriple-cropping
is common,are thenhalvedor
cut by 2/3 in order to make propercomparisons
with those
temperate
countries
whereonlya singlecropisgrownperyear.
Duringthe cominggenerationeverypopulouslow-income
country,and particularlyIndia, China,Pakistan,Nigeria,and
Tg N yr-1 duringthemid-1990s.
Thiswaslessthanhalfthetotal Indonesia,the five nationsthat will accountfor morethan40% of
massof N appliedin inorganicN fertilizers,but more than all N
recycledin crop residuesand animal manures.
theworld'spopulation
increase
by theyear2020[ UnitedNations
Organization(UNO), 1998], will have to increaseits N fertilizer
SMIL: NITROGEN
IN CROP PRODUCTION
AN ACCOUNT OF GLOBAL FLOWS
653
applications. In contrast, European applications have been
declining,and they will remain stableor will declinefurther with
the removal of excessive farming subsidies and with limits
imposed due to environmental concerns (see Jongbloed and
Henkens [1996] and European Fertilizer Manufacturers
Association's Code of Best Agriculture Practice-Nitrogen
availableashttp://www.efma.org;referredto asEFMA data).
biomass,
we havemuchlessconfidence
aboutapportioning
this
3.8. Fertilizer Recovery Rates
huge N loss amongthe major processes
that removeN from
agroecosystems
to the atmosphere
andto bothunderground
and
Calculated inputs sum up to 151-186 Tg N, with additions
climates,
55%forcropsin humidclimates,
65%forlegumes,
and
75% for croplandforages,by the FAOSTAT Data areassownto
thesecropsresultsin a weighted
recovery
average
of about50%,
anexcellent
confirmation
of thecalculated
mean.Whilewemay
concludewith a highdegreeof confidencethat 1/2 of all N added
annuallyto theworld'scroplands
is notincorporated
in harvested
surface waters.
closeto 170Tg N yr-1beingmostlikely(Table5). Dividingthe
element removed in crops by the estimatesof total N inputs
resultsin averageglobal recovery rates of 46-56% (mean 50%).
Theseratesserveas an importantmeansof checkingthe validity
of presentedcalculations.Data from numerousN intake studies
(mostreliablyfrom thoseusing15N)constrain
the rangeof
plausible recovery rates (their global mean could not be below
35% or above 65%), and they also help to delimit the most likely
rate. Analyses of 30 (mostly temperate zone) agroecosystems
showedthat 2/3 of them had N recoveryratesbelow 50%, while
the most efficient cropping absorbednearly 70% of applied N
4. Nitrogen Losses
Both nitrification and denitrification remove soil N as NO and
N20, anddenitrification,
the closingarm of the biospheric
N
cycle,restores
N2 to theatmosphere.
Volatilization
of NH3 is
responsiblefor large N lossesfrom both animal manuresand
fromall ammoniafertilizers.Leachingof highlysolublenitrates
andsoil erosiontransfers
oftenlargeamounts
of N to ground
water, and to streams, lakes and coastal waters where the nutrient
cancauseserious
eutrophication;
erosion
canalsoremovea great
dealof organicN. Therearealsovarious,andoftenconsiderable,
N
lossesfrom topsof plantsbeforetheharvest.
below150kgN ha-i , uptake
efficiencies
wereashighas60-65%
but with higher N application recovery rates scatteredaround
[Frissel and Kolenbrander, 1978]. Where fertilization rates were
50%.
An extensiveEuropeansurvey found the following N uptake
efficiencies:high-yieldingwheatsin France39-57% (from urea)
and 38-70% (from ammonia), in England 52-65%, and in the
Netherlands52-62%, but in Portugaljust 27-40%, and in Greece
only 18-37% [Jenkinson and Smith, 1989]. In Asian rice,
typicallyrecoveryratesare between30-35%; they may be as low
as 20% and only rarely do they exceed 40% of the applied
nutrient [De Datta, 1995; Cassmanet al., 1996]. North American
corn recovers between 40-60% [Reddy and Reddy, 1993], and
leguminous crops, particularly forages, incorporate generally
more of the available (and largely self-supplied) N, usually
anywhere between 50-90% [Peoples and Herridge, 1990].
Recovery ratesof N addedto soilsby atmosphericdepositionand
irrigationwater are similar to thoseof inorganicN.
Multiplying conservativemeansof N recoveryratesfor major
cropcategories,35% for rice andfor rainfedcropsgrownin drier
4.1. NOx andN 20 Emissions
Two key flows in the biosphericnitrogen cycle, bacterial
nitrification
(oxidation
of NH4+) anddenitrification
(reduction
of
NO3- or NO2-), arethemainsources
of NOx (mostlyNO) and
N20 emittedfromsoils.Chemicaldenitrification
andotherkinds
of bacterial metabolisminvolving oxidation or reductionof N
alsoyield trace amountsof the two gases.Nitrificationratesare
regulated
primarilyby theavailability
of NH4+;denitrification
is
controlledmainly by temperature,precipitation,and soil texture,
C availabilityand pH [Bouwmanet al., 1993].
Ratesof fertilization,crop varieties,and tillage practicesare
the most important management factors determining the
emissions.Water-filledporespace(WFPS) appearsto be the key
determinantof the relative fluxes of the two gases:NO fluxes
peakwith30-60%WFPS,whiletheN20 flowspeakwhenWFPS
is between 80-90% [Veldkamp et al., 1998]. Rates of NO and
N20 emissions
fromagricultural
soilsaresohighlyvariablethat
Table 5. Annual N Inputs to the World's CroplandsDuring the
annual fluxes (derived mostly from short-termmeasurements)
range over several orders of magnitude[ Williams et al., 1992;
Mid- 1990s
Bouwman, 1996].
Inputs
Seeds
Atmosphericdeposition
Irrigationwater
Cropresidues
2
2
2
18
3
12
20
4
14
22
5
16
Shepherdet al. [1991] measuredNO-N emissionsequal to
11% of added N in a fertilized soil in Ontario. Typical NO-N
lossesare considerablylower, rangingbetween0.02-5.7% of the
nutrientappliedin fertilizers[Harrisonet al., 1995; Veldkampet
al., 1998]. Consequently,annual means of NO emissionsfrom
agriculturalsoils derived from long-termmeasurements
range
frommeretraces
to nearly10kgN ha-1. Potteretal. [1996]used
Minimum
Mean
Maximum
Animal manures
16
18
20
Biofixation
25
33
41
the averagerates of 1.5 kg N/ha as NO in global simulations;
Inorganicfertilizers
75
78
80
Davidson
andKingerlee
[1997]foundthemeanof 3.6kg N ha-1
151
169
186
for temperatefields worldwide; Davidsonet al. [1998] usedthe
Total
All valuesarein Tg N yr-I. Derivations
of individual
inputratesis
describedin the text. All figuresare roundedto avoidthe appearance
of
unwarrantedprecision.
meanof 8.8 kg N ha-1for thesoilsof thesoutheastern
United
Statesbut admittedit may be somewhathigh. Given this great
uncertainty
I will assume
a conservative
rangeof 0.5-4kgN ha-1,
654
SMIL: NITROGENIN CROPPRODUCTION AN ACCOUNTOF GLOBALFLOWS
resulting in annual global releasesof 1-6 Tg NO-N from the
information, combined with other data, can be used to narrow
world's cultivated
down the uncertainty. Most of their ratios fit between 1-20;
simplemean is about8; and weightedmean(alter discardinga
handful of the most extreme outliers) is about 5. Taking the last
soils.
Recorded
N20 emissions
alsorangeoverseveralordersof
magnitude.Annual meansfor emissionsfrom soilsthat have not
beenrecently
fertilizedrangemostlybetween
1-5kgN ha-1, and value as the minimum ratio would, with annual emissions of at
theaverage
amountof fertilizer-induced
N20 emissions
increases least2 Tg N20-N, generate
nolessthan10Tg N orabout10%of
over time, and it ranges mostly between 0.5-2% of initial N
applications[Harrison et al., 1995]. Veldkampet al. [1998]
recordedrates as high as 6.8% in tropical pastureson soils
developedfrom volcanicash,and Shepherdet al. [1991] found
conversions equal to 5% of applied fertilizer. Manure
all N particularly susceptibleto relatively rapid denitrification
(that is N introducedin atmosphericdeposition,irrigation water,
animal manures,andinorganicfertilizers).
Similar rates have been reportedin relatively rare studiesof
total denitrification.
Rolston
et al.
[1978]
found
total
applications
canincrease
N20 emissions,
frombothnitrification denitrific•itionfluxes equal to 11-14% of applied N (inorganic
and manure)on croppedsites,and Rydenet al. [ 1979] measured
and denitriflcation, several-fold compared to fluxes from soil
an average loss of about 15% from a heavily fertilized and
receivingsome,or no,inorganicfertilizer[Li et al., 1994].
In their model of global N trace gas emissions,Potter et al.
irrigatedfarmland,withaverage
N2: N20 ratiosbetween
5.6-7.4.
Svenssonet al. [1991] found seasonallossesbetween 2-7% just
[1996]usedthe rateof 0.8 kg N/ha asN20 as an averageflux
for the appliedinorganicN. Lossesequalto 10-15% of N most
from cultivated soils; this would produceannually about 1.2 Tg
N. Bouwman [1996] recommended the mean of 1.25% of N
susceptibleto denitrification would produce annual fluxes
between 11-18 Tg N, and, uncertain as this estimate is, it is
fertilizer applications,in addition to 1 kg N20-N/ha of
backgroundemissions,to be used for large-scale,order-of- unlikelythat the actualflux wouldbe lowerthanthe meanof this
magnitude
quantifications;
thiswouldtranslate
to about2.5 Tg N
range.Obviously,muchmoreN2 is eventuallyreturnedfrom
of annual emissions.
agroecosystemsto the atmosphereby denitrification of N
compounds
whichwereremovedfrom fieldsby leachingandsoil
erosion,and carriedaway in harvestedfeed and food crops.
Li
et at [1996], using their detailed denitrification-
decomposition
model,estimated
totalN20 emissions
from the
U.S. croplands
at 0.5-0.74Tg N yr-1 or 3-4.6kg N ha-l;
extendingthe highestrate to the globalcroplandwouldresultin
4.3. Volatilizationof NH3
emissions
of upto 7 Tg N yr4 . | will usetherangeof 1-7Tg N
Relativelylargeamounts
of NH3-N thatcanbe lostduring
for globalN20-N emissions
fromagricultural
soils.Giventhe daysand weeksfollowing applicationsof ammoniafertilizersand
enormous
variabilityof NO andN20 fluxesfromagricultural animal manures have been a long-standing concern among
soilsand from appliedfertilizers,the rangeemergingfrom these
agronomists[Terman, 1979; dayaweera and Mikkelsen, 1991].
Reported sharesof initial N loss are as high as 46% within a
week for urea and 80% for animal urine in just 3 days [Hargrove,
4.2. Complete Denitrification
1988]. Dry, calcareous soils, surface applications of shallow
Estimatingannual return of N2 producedby complete incorporationof fertilizers, and manuresand high temperatures
promotethe lossin rain-fed fields.
alenitrification is by far the most elusive task in quantifying
Volatilization losses are particularly high when ammonia
principalfluxes of the global N cycle. Obviously,the processis
governedby the samefactorsas N20 generation,
with rates fertilizers are broadcastdirectly onto flooded soils [dayaweera
and Mikkelsen, 1991]. Principalfactorspromotingthe processare
highly dependent on soil water content (high WFPS) and
available
soilcarbon.
Highconcentrations
of soilNO3- inhibitthe shallow waters, their alkaline pH, high temperature and high
estimates,
2-13Tg yr-1, isnotexcessive.
final conversion
stepfrom N20 to N and lower the N2:N20
NH4+-Nconcentration,
andhigherwindspeeds.
Consequently,
ratios [Weier et al., 1993]. Reliable direct measurements of
completealenitrificationin farmlandsare rare, and alenitrification
fluxes in field or regional N balancesare commonly estimatedas
residualvaluesafter accountingfor othermajor N losses.
the highest volatilization lossesare in heavily fertilized paddy
fields in the tropics; dominanceof urea in rice farming makes
matters worse as the pH of noncalcareoussoils is temporarily
elevatedafter its application[ Hatgrove, 1988].
As with NO andN20 emissions,
theenormous
variabilityof
fluxes
from
different
soils
and under
different
environmental
Volatilization
losses of the order of 10% within
1-3 weeks of
N applicationsare common,and recordedmaxima are above60%
or even70%.Volatilization
fromnon-NH3 fertilizers
is minimal,
highlyquestionable.
However,withN 2 emissions
beingsomuch and large-scale (national or global) means used to calculate
largerthanthecombined
NO andN20 flux,thereis muchgreater annual losseshave been as low as 1% [ApSimonet al., 1987] for
conditions makes the choice of any typical or average value
opportunityfor underestimatingor exaggeratingthe overall flux
by relatively small shiftsin assumptions.Galloway et al. [1995]
basedtheirestimates
of totalglobalalenitrification
on N2:N20
ratios between 14-32, but, as explained in the following two
paragraphs,the most likely ratios for agricultural soils may be
much lower.
After 5 days of measurementsin four kinds of benchmark
soils,Weieret al. [1993]foundN 2 :N20 ratiosrangingfromas
low as 0.2 to as much as 245 and, not surprisingly,concludedthat
using an average ratio for estimating total alenitrificationfrom
N20 field measurements
cannotbe recommended.
Still, their
the UK and as high as 11% [Bouwmanet al., 1997] for the world.
Given the still increasinguse of urea and higher intensity of N
applicationsin humid and warm environmentsin generaland in
Asia'smonsoonalrice fields in particular,worldwide lossesof 812% (10%) of N applied in inorganicfertilizers would produce
annual
fluxes
of6-10(8)TgNH3-Nyr-1. Tothese
losses
must
be
addedvolatilization of animal wastesappliedto cropland.
Numerous experimentaland fields studiesshow lossesof at
least 10-15% of the initial N content within days or weeks after
applicationof animal manures,and with surfaceapplicationsof
fresh cattle or pig slurriesthe lossesmay be as much as 60-80%
SMIL: NITROGEN
IN CROP PRODUCTION
ofNH3-Nor some30-40%of totalN [Terman,1979;Hansen
and
Henriksen,
1989]. This means that in addition to the
conservativelyestimatedpreapplicationlosses,there will be
furthervolatilization
lossesof at least15-20%of N appliedin
animalmanures
or about2-4 Tg N. This wouldbringthe total
lossof thenutrientappliedin inorganic
andorganicfertilizersto
atleast11(8-14)Tg N orroughlybetween
5-10kgN ha-1
AN ACCOUNT OF GLOBAL FLOWS
655
applyingtypicalannualleachingrates(5-30 kg N ha-1) to the
sameregionalset,cameup withabout17 Tg N yr-1. To these
totals must be added the leaching lossesfrom animal manures:
with no less than 10-20% of applied N they would come to 1-3
Tg N yr-1.Annualgloballeaching
lossfromagricultural
land
would be then about 17 (14-20) Tg N, averagingbetween 10-15
kgN ha4 .
Neteffects
ofannual
deposition-volatilization
fluxes
ofNH3
overcroplandsarehighlyvariable.Changesbothin directionand
magnitudetake placeon daily as well as seasonal
basis,and in
4.5.
Soil Erosion
Recent concernsabout soil erosion and its effects on crop
spiteof oftensubstantial
NH3deposition
somesitesmayactually productivity[Pimentel, 1993; Pimentelet al., 1995; Agassi,1995]
show
small
(<1kgN ha-1yr-1) netNH3emissions
[Sutton
etal.,
have not been matchedby reliable information about the actual
1993].Moreimportantly,
thereis a sizeable
NHqfluxfromtops extentand intensityof the processor, moreaccurately,aboutsoil
of maturing plants, and I will accountfor theseemissionsat the
end of this surveyof N losses.
losses in excess of natural denudation. Global assessment of soil
degradation[ Oldeman et al., 1990] estimatedthat about750 Mha
of continentalsurfacesare moderatelyto excessivelyaffectedby
4.4. Leaching
water erosion and 280 Mha by wind erosion,with deforestation
Leaching rates depend on levels of fertilization, compounds and overgrazingbeing major causes;mismanagementof arable
used(NH3 leachesverylittle in comparison
to readilysoluble land was estimatedto be responsiblefor excessiveerosion on
NO3), soil thicknessand permeability,temperature,and some 180 Mha of cropland, but no global quantification of
precipitation. The single most important land management
excessiveerosionwas attempted.
determinant of the intensity of leaching is the presenceand
Becauseerosionratesvary widely even within a singlefield,
quality of the groundcover [Hill, 1991]. Everything else being
any large-scalegeneralizationsare merely order-of-magnitude
equal, leaching from bare, fallow, or freshly ploughedsoils is
indicators.They alsoignorethe complexityof the process:much
much higher than beneath row crops, which, in turn, is
of the eroded soil is not lost to food production.Larson et al.
considerablyhigher than from soilsunder suchdensecover crops
[1983] found that very little or no eroded sediment leaves the
as legume-grassmixtureswhoserootscan take up large amounts cultivatedland in areaswith gentle relief without any major
of added N.
surfaceoutlet,the landscapecommonin the north-centralUnited
Relatively abundantinformation on nitrate exportsin streams States;andwhenthe sedimentleavesthe land,mostof it maybe
and groundwaters[Cole et al., 1993; Canter, 1997] is of little use
depositeddownstreamas colluviumor alluvium [Trimble, 1975].
in assessingN leaching from agricultural land as there is no
Consequently,
evenif croplandswerethe only sourcesof eroded
reliable way to separate the contributions of atmospheric
soil, the amount of N exported from the fields could not be
deposition, sewage and industrial processes from the flux
reliably quantifiedby using the relatively abundantdata on the
originating in inorganic fertilizers and animal manures.Because transportof suspendedsedimentto the ocean.
of its effects on the quality of surfaceand groundwaters,nitrate
Short-termratesof croplanderosionhave been measuredand
leachinghasreceiveda greatdeal of researchattention[Addiscott estimatedwith varying degreesof accuracyfor many locales
et al., 1991; Hill, 1991; Burt et al., 1993], but, as with many other
aroundthe world. Their extremesrangefrom negligiblelossesin
N fluxes, leaching rates have been measuredmostly over short
rice paddiesto morethan200 metrictons(t) ha-1 on steep
tropical slopes and in the world's most erodible soils of China's
periodsof time (days to monthsafter fertilizer application)and
proratingthesefiguresmay underestimatelong-termthroughputs LoessPlateau[Liu, 1988;HamiltonandLuk, 1993].Averagesoil
[Valiela et al., 1997].
erosionestimatesfor largerareasare rarely available.Nationwide
Annual leachinglossesrange from negligible amountsin arid
U.S. inventories,begunin 1977, are a major exception:Recent
and semi-aridfieldsto 20-30 kg N ha-1 in rainytemperate annualmeanswere just above 15 t ha-l , with about60%
regions.
Maximaof over50kgN ha-I arenotunusual
in themost contributedby water erosion,and the preliminaryvalue for 1997
heavily fertilized crop fields of northwesternEuropeand the U.S.
is about14t ha-• [Lee,1990](seealsoCropland
acreage,
soil
Midwest. Even higher leachinglosseshave beenreportedin some
irrigatedcrops[Diez et al., 1997; Prunty and Greenland., 1997],
but lossesin many Asian paddy fields are very low. A surveyof
40 agroecosystemson three continents indicated that with
erosion, and installation of conservation buffer strips:
Preliminaryestimatesof the 1997NationalResources
Inventory
available as http://www.nhg.nres.usda.gov/land).The first
nationwide approximationfor India indicateslossesof at least 13
applications
of lessthan150kg N ha-1 leaching
equaled
about t ha-• forwatererosion
alone[Singh
etal., 1992],andregional
10% of fertilizer N, while with additionsof more than 150 kg N
estimatesfor China [Zhu, 1997] imply a nationalmeanof about
ha-1 about20% of addedN waslost[FrisselandKolenbrander, 30 t ha-1.
1978]. Losses of 15-25% of initial N were also measured with
repeatedapplicationsof cattle feedlot manure[ Changand Entz,
1996]. Highest leaching rates, on sandy soils, may remove over
60% of N appliedin manure[Hansenand Djurhuus, 1996].
I prepared two kinds of global estimates.The first one,
assigningspecificaverageleachingshares(5-25%) to inorganic
N applicationsdisaggregatedby 11 major agriculturalregions,
Globalmeanof 20 t ha-1impliesannual
removal
of about30
Pg of soil, a conservativetotal given the high erosionratesin
partsof Africa, Latin America, and Asia [Oldeman et al., 1990;
Pimentel,1993]. A disaggregated
calculationusingspecificrates
forNorthAmerica(17t ha-l),Europe
(15t ha-1), Asia(35t ha-I
exceptfor rice paddieswhereerosionis generallynegligible),
Africa,andLatinAmerica
(30t ha4 ) ended
upwithabout
35Pg.
endedup with a flux of about13 Tg N yr-1;thesecond
one, With at least a quarter of this soil redepositedon adjacent
656
SMIL: NITROGENIN CROPPRODUCTION AN ACCOUNTOF GLOBALFLOWS
croplandor on moredistantalluvia,the losswouldbe 22-25 Pg
Indeed, the very notion of any direct, immediatebalancingof
yr-1. SoilN content
ishighlyvariable
evenwithina single
field, N inputsand removalson an annualbasisis misleadingas most
ranging
overanorderof magnitude
(1.5-17t N ha-i ) in cropped of the addedN is not useddirectly by crops(or weeds):The bulk
soils[Stevenson,1986]. A valuejust shortof 0.1% N is closeto
the currentChinesemean [ Lindeft and Wu, 1996], while the U.S.
of the nutrient recoveredby plantsbecomesavailable only after
extensive turnover through the two opposing processesof
averageis around0.125% [Sprent,1987].A conservative
range
of 0.08-0.1% N would meanthat 22-25 Pg of erodedsoil would
immobilization
and mineralization.
Experiments
using15N-
carryawayabout20 (18-25)Tg N yr-1.
labeled fertilizers are the best way to reveal this complexity:At
the growing season'send, a large share (commonly up to 40%,
sometimes more than 50%) of the fertilizer 1Nis immobilized in
4.6. LossesFrom Tops of Plants
the soil's biomass,and the amount of (previously immobilized
and then mineralized)soil N utilized by cropscan be 3-6 timesas
Althoughrarelydiscussed,
theselossesaddup to oneof the
largestglobal N fluxes. They occur mostly within 2 or 3 weeks high as the nutrient drawn directly from the applied fertilizer
after anthesis(full bloom), and at harvesttime the cropsmay [Stevenson,1986; Reddyand Reddy, 1993].
High C/N ratios of most crop residues,commonly above 50
commonlyhavebetween15 and30% lessN thanwastheirpeak
and
as high as 150, are particularly conducive to rapid 1N
content.Neithertranslocation
to rootsnor rootexudatesaregood
immobilization. Some residual 1Nwill be bound in persistent
explanations.A major part of the lossesmust be due to the
sheddingof variousplant parts(pollen, flowers,and leaves), humus compoundsand be unavailableto plants for decadesor
leachingof N from senescingleaves, and to heterotrophic centuries.How fast the short-livedfractionwill cycle dependson
(microorganismic,
insectand bird) grazing.All of theselosses the activity of microbial decomposers (which is highly
(exceptfor windbornepollen) can be seenas merelyinternal temperature-and moisture-dependent)and on the availability of
other sources of N.
redistributions
of N, asthelitterfall, leachingandherbivorywill
returnthe nutrientto soils,and do not haveto be quantified.
5.1. Gains or Losses?
Although
bothreduced
andoxidized
N compounds
areemitted
by
plant tops,mostof the N lossfrom topsof plantsis due to
volatilization
of NH3 [Francis
etal., 1993].
Compai'isons
ofmyminimum
input
andoutput
estimates
show
an
As with most other N losses, there is a considerable
appreciablenet gain (8 Tg N); thoseof the two maximumvalues
variability:Measuredpostanthesis
lossesin wheatrangedfrom
lessthan6 to 80kg N ha-l or fromlessthan8% to almost60%of
showa smallnet loss(4 Tg N); cross-comparisons
of minimum
andmaximumestimates
indicaterelativelylargegainsof 43 Tg N
N presentat anthesis[Daiggeret al., 1976;Harper et al, 1987;
yr-1(about
28kgNha-1) oranalmost
aslarge
loss
of39TgN yr4
Kanampiuet al., 1997].Postanthesis
declinesrangingmostly
(about25 kg N ha-•). Thetwoextreme
comparisons
arenot
Are then the world'scroplandsgainingor losingnitrogen'?
between
20-50kgN ha-1 weremeasured
in other
cereal
crops plausible:We have no indicationthat the world'sagricultural
(rice,sorghum,
andbarley),withdailylosses
ashighas2 kg N
soilswouldbeexperiencing
a netgainor lossashighas40 Tg N
eitherof thefirsttwopossibilities
isentirely
ha-1[Wetselaar
andFarquhar,
1980],
andFrancis
etal. [1993] yr4. In contrast,
astheyimplyannualincrements
or removals
of onlya
foundratesashighas45-81kgN ha-1 fromcorn,equalto 52- plausible,
wecannot
exclude
thepossibility
of a
73% of all N unaccounted
for by standardbalancecalculations. fewkgN ha-1. Although
by improper
agronomic
practices
Evenif thetypicallosswouldbejust 10kg N ha-i , theworld's smallbutchronicN losscaused
naturalsoil degradation
in manyfarming
croplandswouldloseannuallysome15 Tg N, a flux similarto and by accelerated
regions,
thelikelihood
of smallgainsappears
to bemuchhigher.
NO3-Nlosses
dueto leaching.
At20kgN ha-i, theglobal
lossofaround
30TgN yr-1would An international comparisonof N balancesin almost 40
agroecosystems
on threecontinents
foundgainsin soilorganicN
maketheN lossfromtopsof plantsaboutaslargea routeof N
egress
fromcroplands
asdenitrification.
Giventhemagnitude
of
this flux, it is imperative that all 1Nbalance studiesshould
considerthis neglectedvariable before attributingany
Table
6. Annual
Balances
ofN Flows
intheWorld's
Croplands
Duringthe Mid- 1990s
unaccounted
lossesto unknownfactorsor to higherratesof
denitrification
or leaching[Kanampiuet al., 1997].Becausewe
Flows
Minimum
Mean
Maximum
donotknowto whatextentthegaseous
N losses
fromtopsof
Inputs
l 51
143
85
169
165
85
186
190
85
I
I
11
8
14
18
5
4
4
14
11
17
20
10
6
7
18
14
20
25
15
plantsmayhavebeencaptured
by previously
outlinedestimates Outputs
ofNH3andNOx fluxes,I will assume
thatnomorethan10(5-15)
TgN yr-• arelostinthisway.
5. Nitrogen Balances
Tabularrecapitulation
of the estimatedinputsand losses
(Table6) shows
thatglobalcroplands
receive169(151-186)
Tg
N yr-1andthatbetween
143-190
(mean
ofabout
165)TgN yr-1
areremovedin harvested
cropsanddueto a varietyof N losses.
Harvested
plants
Losses
NO emissions
N2¸ emissions
N 2emissions
NH3volatilization
NO3-leaching
Soilerosion
Losses
fromtops
of plants
Figure1 showsthe fluxesinvolvedin globalcropproduction
withina generalN cyclemodelcentered
onplants.Thefactthat Balance
+8
+4
-4
themeansof my conservatively
estimated
inputsandoutputs All values
arein Tg N yr-l. Inputs
arefromTable5;derivations
of
(169and165TgN) aremerely2%apartisnota strong
indication individual
output
rates
aredescribed
inthetext.Allfigures
arerounded
to
thatglobalagroecosystem
is in N equilibrium.
avoidtheappearance
of unwarranted
precision.
SMIL: NITROGEN IN CROP PRODUCTION AN ACCOUNT OF GLOBAL FLOWS
657
NO, NO2
LU
LIGHTNING••
N2
FOSSIL
FUELS
uJ
•r•
N20
NH3, NH•
NO•
FIXATION
115
FOOD
DOMESTIC
23
ANIMALS
PEOPLE
FERTILIZERS
FEEDING
6
78
23
17
DIGESTION
78
LOSSES FROM
PLANT TOPS
10
CROP
z
ORGANIC
RESIDUES
WASTES
25
75
HARVESTED
CROPS
85
RECYCLING
14
100
RECYCLING
18
5
ORGANIC
MA'n'ER
NO
•
IMMOBI
LIZATIO
N
NITRIFICATION
ONIFIC
ATI
ON
NH3, NH4
.,..........••LEAC
HI
NG
NO•
•TRIFICATION
[•
EROSION
17 20
NO• ••
•
•,
J ORGANIC
'•
MA'I-rER
Figure 1. Simplifiedgraphof the biospheric
nitrogencyclecenteredon agricultural
crops.Only the storages
(rectangles)
andflows(valves)discussed
in thispaperarequantified
(all valuesarein Tg N yr-1). Thicklines
identifyfluxesdirectlyaffectedby humanactions.
(thepoolusuallycontaining
> 90% of soilN) in almost60% of
thecases,with a meanannualincrease
of 35 kg N ha-1 andno
of increasein the country's
averageN applications,
thereis no
Netherlands,
foundaverageannualaccumulations
of 47 and38
long-term balance sheetsfor the RothamstedContinuousWheat
doubtthatthesegainshavecontinued
duringthe 1990s'My
change
in about1/3of thebalances
[Floate,1978].In Europe, preliminary
estimates
putthemnowat about25 kgN ha-1 yr-I .
tworecentnationalagricultural
N balances,
for Germany
andthe Organicfarmingcanproduce
the sameresults:
Comparison
of
kgN ha-1, respectively
[vanDijk,1993].
Zhu[1997]hasshown Experiment
showthattheplotreceiving
35 t ha-I of farmyard
thatnitrogen
balance
in China'sagriculture
hasbeenpositive manuremorethandoubledits totalN soilcontentin 115years,
between
theearly1950sandthelate1980s.
Giventherapidrate gainingthenutrientat anannualrateof about33 kgN ha-1, and
658
SMIL: NITROGENIN CROPPRODUCTIONAN ACCOUNTOF GLOBALFLOWS
Suchgainsare not surprisinggiventhe oftenhigh ratesof N
applicationsand the well-knownsequestration
of a significant
shareof the addednutrientin long-livedfractionsof soil organic
matter. As many properly farmed soils aroundthe world are
Assumingan average body massof 45 kg (weighted mean
taking into account the age-sex structure of the world's
population)and using typical lipid, bone and protein sharesin
human bodies [Bailey, 1982] results in about 6 Tg N stored in
bodiesof the world's6 billion people.With the globalpopulation
growingannuallyby almost80 millionpeople,thenetincreaseof
gaining
annually
25-35kg N ha-1(thatis of theorderof 0.25-
thispoolis lessthan80 Gg N yr-1, or merely0.3%of 23 Tg N
0.5% of total soil N stores), these increments could outweigh
undoubtedN lossesfrom deterioratingsoils,particularlyin Africa
consumedin food. More than 99% of ingestedN is thusexcreted,
and advancingurbanization, as well as higher concentrationof
meat, egg, and milk production,meansthat increasingsharesof
this waste are released, directly or indirectly, into waters.
Subsequentfate of this aqueousN rangesfrom benign(relatively
rapid denitrification) to worrisome (nitrate contaminationof
aquifers,coastaleutrophication).
Another remarkable feature of anthropogenicN flows is the
extent to which the nutrient is now moved amongcountriesand
continentsby fertilizer and agriculturaltrade. During the mid1990s,about22 Tg N in inorganicfertilizers(more than a quarter
of the applied total) were sold annually on the international
market, and the traded farm products contained nearly 20%
even the plot that did not receiveany N fertilizershad no longterm N loss [ Jenkinson, 1982].
where nutrientmining is most widespreador in highly erodible
soilsof China'sLoessPlateau.Consequently,
it is quitelikely that
-1
the net resultis a nonnegligibleglobalN gain of 4-8 Tg N yr
While
this is welcome
as far as the maintenance
of the soil
qualityis concerned,
positiveN balancescarryrisksof increased
N losses.
Tracing major N inputs to the world's agroecosystems
and
subsequentN lossesassociatedwith food productionmakesit
easierto answera key questionposedby Gallowayet al. [1995]:
Where is N fixed by humansgoing?It also makesit possibleto
suggestthe most effective ways of reducing the inputs of
anthropogenic
N into the biosphere.
5.2. Nitrogen Flows in Food Production
My calculations show that almost half of all N received
annuallyby the world'scroplands(46%, range43-50%) comes
from syntheticfertilizers. This high dependenceis irreversible,
and it is boundto increaseeven if the global population,as some
forecastssuggest[UNO, 1998], were to stabilizeat a relatively
low level (lessthan 10 billion people)duringthe nexttwo to three
generations.While there are many energy sourcesthat can
replace fossil fuels, whose combustion is the main cause of
human alteration of carbon cycle, there appears to be no
imminent alternativesto our high, and increasing,relianceon the
nitrogenfixed by the Haber-Boschprocess[Smil, 1991].
Direct additionsof all newly fixed anthropogenicN, inorganic
fertilizers (75-80 Tg N) and symbioticbiofixationby food, feed,
andforagelegumes(22-32 Tg N), addedup to 105 (97-112) Tg N
(almost10 Tg N yrq) of all N incorporated
in foodandteed
crops, mostly in cereals, oilseed cakes, and soybeans
(FAOSTAT). The share of traded N will also increase in the
future, as will the fraction of all N traded intercontinentally.
5.3. What' Can be Done?
The best way to managethe continuinghuman interferencein
N cycle is to maximize the efficiencies of N used in crop
production.Fortunately,there are many realistic possibilitiesfor
lowering the input of anthropogenicN in agriculturalproduction
and for reducingits field and postharvestlosses[Munson and
Runge, 1990; Fragosoand van Beusichem,1993; Dudal and Roy,
1995; van der Voet et al., 1996; Prasad and Power, 1997]. Soil
testing,choiceof appropriate
fertilizingcompounds,
maintenance
of propernutrientratios,andattentionto timingandplacement
of
fertilizers are the most importantdirect measuresof universal
applicability.Indirect approacheswhich can either reducethe
yr-• duringthemid-1990s.
Asrecycling,
in seeds,
cropresidues, needfor syntheticfertilizersor increasetheefficiencyof theiruse
rely primarily on greater contributions by biofixation
and animal manures,and irrigation water returned38 (33-43) Tg
N yr-1, humanmanagement
wasresponsible
for inputstotaling accomplished
by morefrequentplantingof leguminous
cropsor
143(130-155)Tg N yr-1andequaltoabout85%of allN received by optimizingconditionsfor otherdiazotrophs.
by the world's agriculturalland. Natural atmosphericdeposition
Improvinga varietyof agronomic
practices,
includingthose
and no more than 1/3 of biofixation (by free-living bacteria and
cyanobacteria)are the only inputs not directly manipulatedby
humans. If the contributions from atmospheric deposition
not directly connected with fertilization, can further reduce N
attributable
to NOx andNH3 fromcombustion
andagriculture
are
integrated
useof organicandsynthetic
fertilizers,by minimizing
soil erosion,by assuringadequatemoisturesupply,and by
controllingpests.The valueof thisintegratedapproach
hasbeen
recognized
by Codesof BestAgriculturalPracticemandated
by
theEuropean
Union'sDirectiveof December1991anddesigned
included, then more than 90% of all N received by the world's
farmlandsoriginatein human activities.
With 50% (46-56%) of all N inputs assimilatedby harvested
crops, we have to accountfor 50% (44-54%) of addedN. Direct
losses.Overall N uptake can be improvedby plantingNresponsive
varieties,maximizingrecyclingof organicwastesand
losses
totheatmosphere
(asN2,NOy,andNHx)addupto26-60 to protect fresh, coastal, and marine waters against nitrate
Tg N, while 32-45 Tg N, or about 20-25% of all initial inputs,
entergroundand surfacewaters.About50 Tg N yr-1 were
removedfrom fields in harvestedcrops.SubtractingN in seeds(2
Tg N), postharvestlosses(5 Tg N), nonfooduses(1 Tg N), and
animal feed (23 Tg N) leaves 17 Tg N (after 2 Tg N of retail and
householdlosses)in plant foods.Adding 6 Tg N in animal foods
(whose N comesnot only from concentratefeeds and cropland
pasturesbut also from permanent grasslandsand, about 1 Tg,
pollutionfrom diffusesources.
Thesecodesrecommend
thatapplications
donotexceedcrop
needs(after the contributionst¾omorganicsourcesare taken
properly into account);thesecodesask that soils shouldnot be let
bareduringrainyperiodsandthatnitratespresentin soilbetween
cropsshouldbelimitedthroughplantingof trapcrops[seeIgnazi,
1996; and EFMA data].
fromaquatic
catches)
makesup23 Tg N yr-1 in theglobalfood
Effectsof theseimprovements
canbe impressive.
El-Fouly
and Fawzi [1996] concludedthat properN:P:K ratiosbasedon
consumption.
soil testingandplantanalysisandadjustedto the prevailing
SMIL: NITROGEN IN CROP PRODUCTION AN ACCOUNT OF GLOBAL FLOWS
cropping sequencecould raise typical Egyptian yields by 20%
withoutusing more nitrogen.Gainsin fertilizer efficiencycould
alreadybe seenin the U.S. recordsincethe early 1980s:Average
yields of principal field crops have continuedto increaseat a
fasterrate than the total applicationsof fertilizer N [Munsonand
Runge, 1990; USDA, 1998]. Good agronomicpracticesshould
raise the averageN use efficiency by at least 25-30% during the
next two generations,i.e., to averageuptakesof at least50-55%
in low-income
countries
and to around
65-70%
in affluent
nations.
Even if the utilization of nitrogen from other sourceswould
remain constant,suchhigher fertilizer efficiencieswould use 1012 Tg of the currently wasted N applications.In reality, an
effective supply of nitrogen from organic wastes,biofixation,
mineralization, and atmosphericdepositionshouldalso increase
because of reduced N losses in soil erosion and because of more
frequentrotationsand more vigorousrecycling.Again, relatively
modestimprovementswouldtranslateinto impressivetotalgains:
Reducingerosionallossesby 20% would saveroughly 4 Tg N
from nonfertilizer sources,and expandingbiofixation, and waste
recyclingby 10% would add another5 Mt N. Cumulativeeffects
of adoptingwell-provenand mostly low-costmeasuresaimed at
increasingefficiency of nutrient uptake would be then equal to
expanding
effective
N supply
tocrops
by20-22Tgyr-1.
These gains would make it possibleto satisfy virtually all
anticipatedN demandneededto produceimprovingfood supply
for the world's populationduring the next generationwithout
increasingthe input of anthropogenic
N in cropproduction.Such
a stabilization would be the first step toward easingthe human
impacton the biosphericN cycle.
6. Conclusions
Of all the N flows involved in the world's crop production,
only the inputs of the nutrient in inorganic fertilizers and N
uptakeby cropsand their residuescan be quantifiedwith a high
degreeof accuracy.For this reasonI presentedall N flows in the
global agroecosystemas ranges rather than as single values.
However, these extreme values do not represent the total
uncertaintyin the estimates:They merely indicate a most likely
range based on fairly conservativeassumptions.This makes it
unlikely that actual biospheric flows would be lower than the
lowestestimatespresentedhere. At the sametime, it is likely that
in a number of cases, the real fluxes may be higher than the
highest given estimates; indeed several published rates are
outsidethe rangespresentedin Tables 5 and 6.
With these caveats in mind, it appears that the world's
croplands
receiveabout170Tg of fixedN yr-l , withnearly9/10
of this total coming from managed, anthropogenic inputs.
Cropping thus represents by far the most important human
interference in the biospheric N cycle, far ahead of the
combustionof fossilfuels. Only abouthalf of all fixed N reaching
the fields is assimilated by crops and their residues; this
conclusionis well supportedby numerousstudiestracing the
recoveryof N in cropping.
Losses of agricultural N thus amount to about half of all
inputs, but their apportioningremains very uncertain.Complete
denitrification, volatilization, leaching, and soil erosion are the
principal routes of N removal, and the best available evidence
indicates that contributions of these four processesare rather
similar, with every one of them removing annually between 10-
659
20 Tg N from the world's crop fields. Given the enormousspatial
and temporal variability of these flows, it will not be easy to
comeup with reliable estimateson the global scale.
/
Fortunately, these large N losses introducing excessive
amounts of reactive N into the biosphere,can be significantly
reduced by better ways of fertilizing and by appropriate
agronomicpractices.While the world'sdependenceon the HaberBosch fixation of ammonia will have to increaseduring the first
half of the 21st century, impacts on the global nitrogen cycle
could be kept to tolerablelevels.
References
Addiscott,T.M., A.P. Whitmore,andD.S. Powlson,Farming,Fertilizers
and theNitrate Problem, 170 pp., CAB Int., Wallingford,U.K., 1991.
Agassi,M. (Ed.), SoilErosion,Conservation,
andRehabilitation,
424 pp.,
Marcel Dekker, New York, 1995.
Andreae,M.O., Biomassburning:Its history,use,and distributionand its
impact on environmentalquality and global climate, in Global
Biomass Burning: Atmospheric, Climatic, and Biospheric
Implications,editedby J.S.Levine,pp. 3-21, MIT Press,Cambridge,
Mss., 1991.
ApSimon, H.M., M. Kruse, and J.N.B. Bell, Ammonia emissions and
their role in acid deposition,Atmos.Environ., 21, 1939-1946, 1987.
Ayanaba,A., and P.J. Dart (Eds.), Biological Nitrogen Fixation in
Farming Systemsof the Tropics, 377 pp., John Wiley, New York,
1977.
Bailey, S.M., Absoluteandrelativesexdifferencesin bodycomposition,
in SexualDimorphismin HomoSapiens,editedby R.L. Hall, pp. 363390, Praeger,Westport,Conn., 1982.
Barreveld,W.H., Rural UseqfLignocellulosic
Residues,89 pp.,Foodand
Agric. Org., Rome, 1989.
Bath,D., J. Dunbar,J. King, S. Bery, andS. OlbrichS., Byproducts
and
unusualfeedstuffs,Feedstuffy,69(30), 32-38, 1997.
Berner, E.K., and R.A. Berner, Global Environment,376 pp., PrenticeHall, EnglewoodCliffs, N.J., 1996.
Blair, G., D. Catchpoole,and P. Horne, Foragetree legumes:Their
managementandcontributionto the nitrogeneconomyof wet and
humidtropicalenvironments,
Adv.Agron.,44, 27-54, 1990.
Boddey,R.M., O.C. de Oliveira, S. Uequiaga,F.L. de Olivares,V.L.D.
Bladani, and J. Dobereiner,Biological nitrogenfixation associated
with sugar cane and rice: Contributions and prospects for
improvement,Plant Soil, 174, 195-209, 1995.
Bouwman,A.F., Direct emissionsof nitrousoxidefrom agriculturalsoils,
Nutrient CyclingAgroecosystems,
46, 53-70, 1996.
Bouwman,A.F., I. Fung,E. Matthews,andJ. John,Globalanalysisof
thepotential
for N 2¸ production
in naturalsoils,GlobalBiogeochem.
Cycles, 7, 557-597, 1993.
Bouwman, A.F., D.S. Lee, W.A.H. Asman, F.J. Dentener, K.W. Van Der
Hoek,andJ.G.J.Olivier,A globalhigh-resolution
emission
inventory
for ammonia,GlobalBiogeochem.
Cycles,11,561-587, 1997.
Broughton,W.J., and S. Puhler(Eds.), NitrogenFixation,348 pp.,
Clarendon,Oxford, England,1986.
Burt, T.P., A.L. Heathwaite, and S.T. Trudgill, Nitrate: Processes,
Patternsand Management,444 pp., JohnWiley, New York, 1993.
Canter,L. W., Nitratesin Groundwater,
263 pp.,CRC Press,BocaRaton,
Fa, 1997.
Cassman,K.G., S.K. De Datta, S.T. Amarante,S.P. Liboon,M.I. Samson,
andM.A. Dizon,Long-termcomparison
of the agronomic
efficiency
and residualbenefitsof organicand inorganicnitrogensourcesfor
tropicallowlandrice,Exp. Agric. 32, 927-944, 1996.
Chang,C., and T. Entz, Nitrate leachinglossesunderrepeatedcattle
feedlotmanureapplications
in southern
Alberta,J. Environ.Qual.,25,
145-153, 1996.
Choudhary,M., L.D. Bailey, and C.A. Grant, Review of the useof swine
manurein cropproduction:Effectson yield andcomposition
andon
soil andwaterquality, WasteManage.Res.,14, 581-595, 1996.
Cole,J.J.,B.L. Peierls,N.F. Caraco,andM.L. Pace,Nitrogenloadingof
riversas a human-driven
process,in Humansas Components
qf
Ecosystems,
editedby M.J. McDonnellandS.T.A. Pickett,pp. 141-.
157, Springer-Verlag,New York, 1993.
660
SMIL: NITROGEN IN CROP PRODUCTION
Daigger, L.A., D.H. Sander, and G.A. Peterson,Nitrogen content of
winter wheat duringgrowthand maturation,Agron. J., 68, 815-818,
1976.
Danso, S.K.A., G. Hardarson, and F. Zapata, Misconceptions and
practicalproblemsin the useof •SNsoilenrichment
techniques
for
estimatingN2 fixation,PlantSoil, 152, 25-52, 1993.
Davidson, E.A., and W. Kingerlee, A global inventory of nitric oxide
emissionsfrom soils,Nutrient CyclingAgroecos.,48, 37-50, 1997.
Davidson, E.A., C.S. Potter, P. Schlesinger,and S.A. Klooster, Model
estimates of regional nitric oxide emissions from soils of the
southeastern
United States,Ecol.Appl., 8, 748-759, 1998.
De Datta, S.K., Nitrogen transformationsin wetland rice ecosystems,
Fert. Res., 42, 193-203, 1995.
Diez, J.A., R. Roman, R. Caballero, and A. Caballero, Nitrate leaching
from soils under a maize-wheat-maize sequence, two irrigation
schedulesand three typesof fertilisers,Agric. Ecosyst.Environ., 65,
189-199, 1997.
Donald, C.M., and J. Hamblin, The biological yield and harvestindex of
cerealsasagronomicandplantbreedingcriteria,Adv.Agron.,28, 361405, 1976.
Dovrat, A., Irrigated Forage Production, 257 pp., Elsevier,New York,
1993.
Dudal, R., and R.N. Roy (Eds.), Integrated Plant Nutrition Systems,426
pp., Food and Agric. Org., Rome, 1995.
Eady, R.R., The enzymologyof biological nitrogenfixation, Sci. Prog.,
78, 1-17, 1995.
AN ACCOUNT OF GLOBAL FLOWS
Hardarson,G., Methods for enhancingsymbioticnitrogenfixation, Plant
Soil, 152, 1-17, 1993.
Hardarson, G., and S.K.A. Danso, Methods for measuringbiological
nitrogenfixationin grainlegumes,Plant Soil, 152, 19-23, 1993.
Hargrove, W.L., Soil, environmental, and management factors
influencing ammonia volatilization under field conditions, in
Ammonia Volatilization.fi'om Urea Fertilizers, edited by B.R. Bock
and D.E. Kissel, pp. 17-36, National Fertilizer DevelopmentCenter,
Muscle Shoals, Ala., 1988.
Harper, L.A., R.R. Sharpe,G.W. Langdale, and J.E. Giddens,Nitrogen
cycling in a wheat crop: Soil, plant and aerial nitrogen transport,
Agron.J., 79, 965-973, 1987.
Harrison,R.M., S. Yamulki, K.W.T. Goulding, and C.P. Webster,Effect
of fertilizerapplicationon NO andN 20 fluxesfromagriculturalfields,
J. Geophys.Res., 100, 25,923-25,931, 1995.
Hay, R.K.M., Harvest index: A review of its use in plant breedingand
cropphysiology,Ann. Appl. Biol., 126, 197-216, 1995.
Heichel, G.H., Legume nitrogen: Symbiotic fixation and recovery by
subsequentcrops, in Energy in Plant Nutrition and Pest Control,
editedby Z. Helsel,pp. 62-80, Elsevier,New York, 1987.
Heichel, G.H., and K.I. Henjum, Dinitrogen fixation, nitrogentransfer,
and productivityof foragelegume-grass
communities,Crop Sci., 31,
202-208, 1991.
Henson,R.A., Measurements
of N 2 fixationby commonbeanin Central
Brazil as affectedby differentreferencecrops,Plant Soil, 152, 53-58,
1993.
Eghball, B., J.F. Power, J.E. Gilley, and J.W. Doran, Nutrient, carbon,
and mass loss during cornpostingof beef cattle feedlot manure, J.
Hill, M. (Ed.), Nitrates and Nitrites in Food and Water, 196 pp., Ellis
Horwood, Chichester,England, 1991.
Environ. Qual ., 26, 189-193, 1997.
Hoogerbrugge,I.D., and L.O. Fresco,HomegardenSystems:Agricultural
E1-Fouly, M., and A.F.A. Fawzi, Higher and better yields with less
Characteristics and Challenges, 23 pp., International Institute for
environmentalpollutionin Egyptthroughbalancedfertilizeruse, Fert.
Environmentand Development(IIED), London, 1993.
Res., 43, 1-4, 1996.
Ignazi, I.C., Codeof bestagriculturalpractices,Fert. Res.,43, 241, 1996.
Erisman,J. W., AtmosphericDepositionin Relationto Acidt.)'ication
and
Intergovernmental
Panelon Climate Change(IPCC) and Organizationfor
Eutrophication,405 pp., Elsevier,New York, 1995.
Economic Cooperation and Development (OECD), Revised1996
Farnham, D.E., and J.R. George, Harvest management effects in
IPCC Guidelines.l'brNational GreenhouseGas Inventories,Org. for
dinitrogen fixation and nitrogen transferin red clover-orchardgrass
Econ. Coop. and Dev., Paris.
mixtures,J. Prod. Agric., 7, 360-364, 1994.
Jaworski,N.A., R.W. Howarth,andL.J. Hetling,Atmospheric
deposition
Floate, M.J.S., Changesin soil pools,in Cyclingq[Mineral Nutrientsin
of nitrogen oxides onto the landscape contributes to coastal
Agricultural Ecosystems., edited by M.J. Frissel, pp. 292-295,
eutrophicationin the NortheastUnited States,Environ. Sci. Technol.,
Elsevier, New York, 1978.
31, 1995-2004, 1997.
FoodandAgricultureOrganizationof the UnitedNations(FAO), Current
Jayaweera, G.R., and D.S. Mikkelsen, Assessment of ammonia
World Fertilizer
Situation and Outlook 1994/95-2000/2001,
Rome,
1997.
Fragoso,M.A., and M.L. van Beusichem(Eds.), Optimizationql"Plant
Nutrition, 682 pp., Kluwer, Norwell, Mass., 1993.
Frame,J., and O. Newbould,Agronomyof white clover,Adv. Agron., 40,
1-88, 1986.
Francis,D.D., J.S. Schepers,and M.F. Vigil, Post-anthesisnitrogenloss
from corn, Agron.J., 85, 659-663, 1993.
Frissel,M.J., and G.J. Kolenbrander,The nutrientbalances:Summarizing
graphsand tables, in Cycling ql"Mineral Nutrients in Agricultural
Ecosystems.,
editedby M.J. Frissel,pp. 273-292, Elsevier,New York,
1978.
Galloway, J.N., W.H. Schlesinger,H. Levy lI, A. Michaels, and J.L.
Schnoor, Nitrogen fixation: Anthropogenic enhancementenvironmentalresponse,Global Biogeochem.Cycles, 9, 235-252,
1995.
Gifford, R.M., and L.T. Evans,Photosynthesis,
carbonpartitioning,and
yield,Ann.Rev.Plant Physiol.,32,485-509, 1981.
Giller, K.E., and K.J. Wilson, Nitrogen Fixation in Tropical Cropping
Systems,200 pp., CAB Int., Wallingford,U.K., 1991.
Giller, K.E., J.F. McDonagh, and G. Cadisch, Can nitrogen fixation
sustainagriculturein the tropics?,in Soil Scienceand Sustainable
Land Managementin the Tropics, edited by J.K. Syers and D.L.
Rimmer,pp. 173-191,CAB Int., Wallingford,U.K., 1994.
Gutteridge,R.C., and H.N. Shelton,Forage Tree Legumesin Tropical
Agriculture,389 pp., CAB Int., Wallingford,U.K., 1994.
Hamilton, H., and S. Luk, Nitrogen transfers in a rapidly eroding
agroecosystem:
LoessPlateau,China, J. Environ.Qual., 22, 133-140,
1993.
Hansen,E.M., and J. Djurhuus,Nitrate leachingas affectedby long-term
N fertilization on a coarsesand, Soil Use Manage., 12, 199-204, 1996.
Hansen, J. A., and K. Henriksen (Eds.), Nitrogen in Organic Wastes
Appliedto Soils,381 pp., Academic,SanDiego,Calif., 1989.
volatilization from floodedsoil systems,Adv. Agron., 45, 303-356,
1991.
JenkinsB.M., S.Q. Turn,andR.B. Williams,Atmospheric emissions
from agriculturalburningin California: Determination of burn
Ëactions,distributionfactors,and crop-specificcontributions,Agric.
Ecosyst.Environ., 38, 313-330, 1992.
Jenkinson,D.S., The nitrogen cycle in long-term field experiments,
Philos. Trans. R. Soc. Lond., Set. B, 296,563-571,
1982.
Jenkinson,D.S., and K.A. Smith, Nitrogen Efficiency in Agricultural
Soils,450 pp., Elsever,New York, 1989.
Jongbloed,A.W., and C.H. Henkens,Environmentalconcernsof using
animal manure- The Dutch case,in Nutrient Managementof Food
Animals to Enhance and Protect Environment, edited by E.T.
Kornegay,pp. 315-332, Lewis Publ., Boca Raton,Fa, 1996.
Jordan,T.E., andD.E. Weller, Humancontributions
to terrestrialnitrogen
flux, BioScience, 46, 655-664, 1996.
Kanampiu,F.K., W.R. Raun,andG.V. Johnson,Effectof nitrogenrateon
plant nitrogenlossin winter wheat varieties, J. Plant Nutr., 20, 389404, 1997.
Kinzig, A.P., and R.H. Socolow,Humanimpactson the nitrogencycle,
Phys. Today, 47, 24-31, 1994.
Kossila, V.L., Global review of the potential use of crop residuesas
animalfeed,in Better Utilizationof CropResidues
andBy-products
in
Animal Feeding: ResearchGuidelines,editedby T.R. Prestonet al.,
pp. 1-13, Food andAgric. Org., Rome, 1985.
Kuhlbusch, T.A., J.M. Lobert, P.J. Crutzen, and P. Warneck, Molecular
nitrogenemissionsfrom denitrification during biomasscombustion,
Nature, 351, 135-137, 1991.
Larson,W.E., F.J. Dixon, and R.H. Dowdy, The threatof soil erosionto
long-termcropproduction,Science,219, 458-465, 1983.
LaRue, T.A., and T.G. Patterson,How much nitrogen do legumesfix?
Adv.Agron., 34, 15-38, 1981.
SMIL: NITROGEN IN CROP PRODUCTION AN ACCOUNT OF GLOBAL FLOWS
Lee, L.K., The dynamicsof decliningsoil erosionrates,J. Soil Water
Conserv.,45, 622-624, 1990.
661
Shepherd, M.F., S. Barzetti, and D.R. Hastie, The production of
atmospheric
NO x and N20 from a fertilizedagriculturalsoil, Atmos.
Environ., 25 Part A, 1961-1969, 1991.
Li, C., S.E. Frolking,R.C. Harriss,andR.E. Terry,Modelingnitrous
oxideemissions
fromagriculture:
A Floridacasestudy,Chemosphere, Shiklomanov, I.A., World fresh water resources, in Water in Crisis: A
28, 1401-1415, 1994.
Li, C., V. Narayanan,and R.C. Harriss,Model estimatesof nitrousoxide
emissions from agricultural lands in the United States, Global
Biogeochem.Cycles, 10, 297-306, 1996.
Lindert,P.H., andWu, W.L., Trendsin thesoilchemistry
of NorthChina
sincethe 1930s,J. Environ.Qual., 25, 1168-1178,1996.
Liu, T.S.,Loessin China,224 pp.,Springer-Verlag,
New York, 1988.
Meelu, O.P., Y. Singh, and B. Singh, Green Manuring.[br Soil
Productivity
Improvement,
FoodandAgric.Org.,Rome,1994.
Misra,R.V., andP.R.Hesse,Comparative
Analyses
qf OrganicManures,
97 pp.,FoodandAgric.Org.,,Rome,1983.
Mosier,A.R., et. al., Assessing
and mitigatingN20 emissions
from
agricultural
soils,Clim.Change,40, 7-38, 1998.
Munson, R.D. and C.F. Runge, ImprovingFertilizer and Chemical
EfficiencyThrough"HighPrecisionFarming",96 pp., Cent.for Int.
Foodand Agric. Policy, Univ. of Minn., St. Paul, 1990.
NationalIntelligenceCouncil, ChinaAgricultureCultivatedLandArea,
GrainProjections,
andImplications,
24 pp.,Nat. Intelligence
Council,
Washington,D.C., 1997.
NationalResearchCouncil(NRC), Atlas of NutritionalData on United
Statesand CanadianFeeds,772 pp.,Nat. Acad.of Sci.,Washington,
D.C., 1972.
Nevison,C.D., G. Esser,andE.A. Holland,A globalmodelof changing
N20 emissions
fromnaturalandperturbedsoils,Clim. Change,32,
327-378, 1996.
Neyra,C.A., andJ. Dobereiner,Nitrogenfixationin grasses,
Adv.Agron.,
30, 1-38, 1977.
Guide to the World's Fresh Water Resources,edited by P.H. Gleick,
pp. 13-24, Oxford Univ. Press,New York, 1993.
Singh, A., and V. Balasubramanian, Organic recycling in Asian
agriculture,in Organic Recyclingin Asia, p. 241-245, Food andAgric.
Org., Rome, 1980.
Singh,G., R. Babu, P. Narain, L.S. Bhushan, and I.P. Abrol, Soil erosion
rates in India, J. Soil Water Conse•a;.,47, 97-99, 1992.
Sinton,J.E. (Ed.), China EnergyDatabook, p. IV-76, LawrenceBerkeley
National Lab., Berkeley, Calif., 1996.
Smil, V., CarbonNitrogenSul.l'kr, 459 pp., Plenum,New York, 1985.
Smil, V., Nitrogen and phosphorus,in The Earth as TransJbrmedby
Human Action, edited by B.L. Turner II et al., pp. 423-436,
CambridgeUniv. Press,New York, 1990.
Smil, V., Population growth and nitrogen: An exploration of a critical
existentiallink, Pop. Dev. Rev., 17, 569-601,1991.
Smil, V., China's Environmental Crisis, 257 pp., M.E. Sharpe,Armonk,
N.Y., 1993.
Smil, V., Nutrient Flows in Agriculture, 54 pp., Rep. Project 2050,
Washington,D.C., 1994.
Smil, V., CyclesqfLO•k,221 pp., Sci. Am. Libr., New York, 1997a.
Smil, V., Some unorthodoxperspectiveson agriculturalbiodiversity:The
case of legume cultivation, Agric. Ecosyst. Environ., 62, 135-144,
1997b.
Smil, V., Crop residues:Agriculture's largest harvest,BioScience, 49,
299-308, 1999.
Sprent, J., The Ecology qf Nitrogen Cycle, 160 pp., Cambridge Univ.
Press, New York, 1987.
Nordstedt,
R.A., Selectinga wastemanagement
system,in LargeDais?
Herd Management,editedby H.H. Van Horn,andC.J.Wilcox,pp.
630-639,Am. Dairy Sci.Assoc.,Champaign,
I11.,1992.
State Statistical Bureau, China Statistical Yearbook I997, 830 pp.,
Beijing, 1997.
Stevenson,F.J., CyclesqfSoil, 300 pp., Wiley-lnterscience, New York,
Oldeman,L.R., et al., World Map of the Statusof Human-InducedSoil
Degradation: An Explanatory Note, United Nations Environment
Programme(UNEP), Nairobi, ! 990.
Subba Rao, N. S., Biq[krtilizers in Agriculture, 208 pp., Oxford Univ.
Peoples,M.B., and D.F. Herridge,Nitrogenfixation by legumesin
tropicalandsubtropical
agriculture,
Adv.Agron.,44, 155-223,1990.
Peoples,M.B., D.F. Herridge,and J.K. Ladha, Biologicalnitrogen
fixation:An efficientsourceof nitrogenfor sustainable
agricultural
production?,Plant Soil, 174, 3-28, 1995.
Pimentel,D. (Ed.), World Soil Erosionand Conservation,
349 pp.,
CambridgeUniv., New York, 1993.
Pimentel, D., et. al., Environmental and economic costs of soil erosion
and conservationbenefits,Science,267, 1117-1122, 1995.
Postel,S.L., G.C. Daily, and P.R. Ehrlich, Human appropriationof
renewable fresh water, Science,271,785-788, 1996.
Potter, C.S., P.A. Matson, P.M. Vitousek, and E.A. Davidson, Process
modeling of controlson nitrogen trace gas emissionsfrom soils
worldwide,J. Geophys.Res.,10I, 1361-1377, 1996.
Powers, W.L., G.W. Walingford, and L.S. Murphy, Formulas for
applyingorganicwastesto land, J. Soil Water Conserv.,30, 286-289,
1975.
Prasad,R., and J.F. Power,Soil Fertility Management.[brSustainable
Agriculture, 356 pp., CRC Press,BocaRaton,Fa., 1997.
Prihar,S.S,andB.A. Stewart,Sorghumharvestindexin relationto plant
size,environmentandcultivar,Agron.J., 83, 603-608, 1991.
Prunty,L., and R. Greenland,Nitrateleachingusingtwo potato-corn
Nfertilizerplanson sandysoil,Agric.Ecosyst.
Environ.,65, 1-13,1997.
Reddy,G.B., and K.R. Reddy,Fate of N-15 enrichedammoniumnitrate
appliedto corn, Soil Sci. Soc.Am. J., 57, 111-115, 1993.
Roberts, S.R., J.E. Hill, D.M. Brandon, B.C. Miller, S.C. Scarduci, C.M.
Wick, and J.F. Williams, Biological yield and harvestindex in rice:
Nitrogenresponseof tall and semidwarfcultivars,J. Prod. Agric., 6,
585-588, 1993.
Rolston, D.E., D.L. Hoffman, and D.T. Toy, Field measurementsof
denitrification,I, Flux of N 2 andN2O, Soil Sci. Soc.Am. J., 41,863869, 1978.
Rulkens,W.H., andP.J.W. ten Have, Centralprocessing
of pig manurein
the Netherlands, Water Sci. Technol., 30, 157-165, 1994.
Ryden, J.C., L.J. Lund, J. Lesy, and D.D. Focht, Direct measurementsof
denitrification from soils, II, Developmentand applicationof field
methods, Soil Sci. Soc. Am. J., 43, 110-118, 1979.
1986.
Press, New York, 1988.
Sutton, M.A., C.E.R. Pitcairn, and D. Fowler, The exchangeof ammonia
betweenthe atmosphereand plant communities,Adv. Ecol. Res., 24,
301-390, 1993.
Svensson, B.H., L. Klemendtsson, S. Simkins, K. Paustian, and T.
Rosswal, Soil denitrification in three cropping systemscharacterized
by differencesin nitrogenandcarbonsupply, Plant Soil, 138, 257271, 1991.
Terman, G.L., Volatilization lossesof nitrogenas ammoniafrom surfaceapplied fertilizers, organic amendments,and crop residues, Adv.
Agron., 31, 189-223, 1979.
Trimble, S.W., Denudation studies:Can we assumestreamsteadystate'?,
Science, 188, 1207-1208, 1975.
Tsai, S.M., P.M. da Silva, W.L. Cabezas,and R. Boneti, Variability in
nitrogen fixation in common bean (Phaseolus vulgaris L.)
intercroppedwith maize, Plant Soil, 152, 93-101, 1993.
United Nations Organization (UNO), World Population Prospects:The
1996 Revision, Annex I: Demographic Indicators, 253 pp., United
Nations, New York, 1998.
United Nations Industrial Development Organization (UNIDO) and
International Fertilizer Development Center (IFDC), Fertilizer
Manual, 615 pp., Kluwer Acad., Norwell, Mass., 1998.
United StatesDepartmentof Agriculture(USDA), ImprovingSoils with
Organic Wastes,157 pp., WashingtonD.C., 1978.
United StatesDepartmentof Agriculture(USDA), Agricultural Yearbook,
WashingtonD.C., 1998.
Valiela, I., G. Collins, J. Kremer, K. Lajtha, M. Gesit, B. Seely, J.
Brawley, and C. Sham, Nitrogen loading from coastalwatershedsto
receivingestuaries:New methodand application,Ecol. Appl., 7, 358380, 1997.
van der Voet, E., R. Kleijn, and H.A.U de Haes, Nitrogenpollutionin the
EuropeanUnion: Origins and proposedsolutions,Environ. Conserv.,
23, 120-132, 1996.
van Dijk, T.A., Appropriate managementpractices to minimize the
environmental impact of applied fertilizers, in Nitric Acid-Based
Fertilizers and the Environment,edited by R.G. Lee, pp. 349-358, Int.
Fertilizer Develop. Cent., Muscle Shoals,Ala., 1993.
Van Leeuwen, E.P., G.P.J. Draaijers, and J.W. Erisman,Mapping wet
662
SMIL: NITROGEN IN CROP PRODUCTION
AN ACCOUNT OF GLOBAL FLOWS
depositionof acidifying componentsand base cationsover Europe
usingmeasurements,
Atmos.Environ., 30, 2495-2511, 1996.
Veldkamp, E., M. Keller, and M. Nunez, Effectsof pasturemanagement
on N20 and NO emissionsfrom soilsin the humidtropicsof Costa
Rica, Global Biogeochem.Cycles,12, 71-79, 1998
Wetselaar,R. and G.D. Farquhar,Nitrogen lossesfrom tops of plants,
Adv.Agron.,33, 263-302, 1980.
Wetselaar,R., J.R. Simpson,and T. Roswall,(Eds.), NitrogenCyclingin
South-EastAsian WetMonsoonalEcosystems,216 pp., Aust. Acad. of
Vitousek, P.M., J.D. Aber, R.W. Howarth, G.E. Likens, P.A. Matson,
Watt, B.K., and A.L. Merrill, Handbook of the Nutritional Contents of
Whitehead,D.C., and N. Raistrick,Nitrogenin the excretaof dairy cattle:
Changesduringshort-termstorage,J. Agric. Sci, 121, 73-81, 1993.
Williams, E.J., G.L. Hutchinson,and F.C. Fehsenfeld,NO x and N20
emissionsfrom soil, Global Biogeochem.Cycles,6, 351-388, 1992.
Zhu, Z., Nitrogen balanceand cycling in agroecosystems
of China, in
Nitrogenin soilsq]"China,editedby Z. Zhu, Q. Wen, andJ.R. Freney,
pp. 323-338,KluwerAcad.,Norwell,Mass.,1997.
Foods,190 pp., Dover, Mineola, N.Y., 1975.
Wedin, W.F., and T.J. Klopfenstein,Croplandpasturesand cropresidues,
in Forages, vol. ii, edited by R.F. Barnes, D.A. Miller, and C.J.
Nelson,pp. 193-206,Iowa StateUniv. Press,Ames, 1995.
V. Smil, University of Manitoba, Department of Geography,
Winnipeg,ManitobaR3T 2N2, Canada. ([email protected])
D.W. Schindler, W.H. Schlesinger, and D.G. Tilman, Human
alteration of the global nitrogen cycle: Sourcesand consequences,
Ecol. Appl., 7, 737-750, 1997.
Warneck, P., Chemistryof the Natural Atmosphere,757 pp., Academic,
1988.
Sci., Canberra, 1981.
Weier, K.L., J.W. Doran, J.F. Power, and D.T. Walters, Denitrification
and the dinitrogen/nitrousoxide ratio as affected by soil water,
(ReceivedJuly 17, 1998; revisedFebruary18, 1999; acceptedFebruary
available carbon and nitrate, Soil Sci. Soc. Am. J., 57, 66-72, 1993.
25, 1999)