University of Nebraska - Lincoln University of Nebraska - Lincoln
DigitalCommons@University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln
Agronomy & Horticulture -- Faculty Publications Agronomy and Horticulture Department
3-2002
Agroecosystems, Nitrogen-use E<ciency, and Nitrogen Agroecosystems, Nitrogen-use E<ciency, and Nitrogen
Management Management
Kenneth G. Cassman
University of Nebraska-Lincoln
, kcassman1@unl.edu
Achim R. Dobermann
University of Nebraska-Lincoln
, adobermann2@unl.edu
Daniel T. Walters
University of Nebraska-Lincoln
, dwalters1@unl.edu
Follow this and additional works at: https://digitalcommons.unl.edu/agronomyfacpub
Part of the Plant Sciences Commons
Cassman, Kenneth G.; Dobermann, Achim R.; and Walters, Daniel T., "Agroecosystems, Nitrogen-use
E<ciency, and Nitrogen Management" (2002).
Agronomy & Horticulture -- Faculty Publications
. 356.
https://digitalcommons.unl.edu/agronomyfacpub/356
This Article is brought to you for free and open access by the Agronomy and Horticulture Department at
DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Agronomy & Horticulture --
Faculty Publications by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln.
132
© Royal Swedish Academy of Sciences 2002 Ambio Vol. 31 No. 2, March 2002
http://www.ambio.kva.se
INTRODUCTION
The focus of this paper is on nitrogen-use efficiency (NUE) in
cereal production systems because maize (Zea mays L.), rice
(Oryza sativa L.), and wheat (Triticum aestivum L.) provide
more than 60% of human dietary calories either as cereals for
direct human consumption or embodied in livestock products
produced from animals fed with feed grains and their by-prod-
ucts (http:/apps.fao.org/, agricultural production). It is likely that
these same cereal crops will continue to account for the bulk of
the future human food supply because they produce greater
yields of human-edible food, are easily grown, stored, and trans-
ported, and require less fuel and labor for processing and cook-
ing than other food crops. Our analysis will examine the NUE
of these primary cereals in the world’s major cropping systems,
which also account for the majority of global N fertilizer use.
We define the NUE of a cropping system as the proportion of
all N inputs that are removed in harvested crop biomass, con-
tained in recycled crop residues, and incorporated into soil or-
ganic matter and inorganic N pools. Nitrogen not recovered in
these N sinks is lost from the cropping system and thus contrib-
utes to the reactive N (Nr) (1) load that cascades through envi-
ronments external to the agroecosystem.
Our evaluation will focus on NUE in on-farm settings because
estimates of NUE from experimental plots do not accurately rep-
resent the efficiencies achieved in farmers’ fields. This lack of
agreement results from differences in the scale of farming op-
erations and differences in N-management practices—some of
which are only feasible in small research plots. The effect of
scale not only influences N fertilizer application, but all other
management operations such as tillage, seeding, weed and pest
management, irrigation, and harvest, which also affect efficiency.
As a result, N-fertilizer efficiency in well-managed research ex-
periments is generally greater than the efficiency of the same
practices applied by farmers in production fields. For example,
the average N-fertilizer uptake efficiency (defined as the percent-
age of fertilizer-N recovered in aboveground plant biomass dur-
ing the growing season—hereafter called the N-fertilizer recov-
ery efficiency – RE
N
), achieved by rice farmers is 31% of ap-
plied N based upon on-farm measurements in the major rice-pro-
duction regions of four Asian countries (Table 1). In contrast,
RE
N
for rice in well-managed field experiments typically range
from 50–80% (3–5). In the authors’ experience, similar overes-
timation of RE
N
in small plot experiments occurs for irrigated
and rain-fed maize in the North-Central USA and for irrigated
wheat in California.
The need to improve RE
N
will be emphasized because N fer-
tilizer is the largest source of N input to and losses from cereal
cropping systems. A recent study estimates total N input to the
world’s cropland at 169 Tg N yr
–1
(6). Inorganic N fertilizer, bio-
logical N fixation from legumes and other N-fixing organisms,
atmospheric deposition, animal manures, and crop residues ac-
count for 46%, 20%, 12%, 11%, and 7%, respectively, of this
total. Hence, crop-management practices that increase RE
N
have
a substantial impact on the amount of Nr that escapes from ce-
real production systems. While we recognize that solutions to
global concerns about effects of Nr on the environment must in-
volve integrated management of both inorganic and organic N
sources to maximize NUE, other papers in this issue of Ambio
and elsewhere address issues of N efficiency in livestock pro-
duction systems and the contributions of organic N sources such
as legume crops and green manures (6, 7).
NITROGEN-USE EFFICIENCY TODAY
Applied N not taken up by the crop or immobilized in soil or-
ganic N pools-which include both microbial biomass and soil
organic matter—is vulnerable to losses from volatilization,
denitrification, and leaching. The overall NUE of a cropping sys-
tem can therefore be increased by achieving greater uptake effi-
ciency from applied N inputs, by reducing the amount of N lost
from soil organic and inorganic N pools, or both. In many crop-
ping systems, the size of the organic and inorganic N pools has
reached steady-state or is changing very slowly, and the N in-
puts from biological N
2
fixation and atmospheric deposition are
relatively constant. For example, analysis of the N balance in
long-term experiments on irrigated rice in Asia suggests that
many of these systems have reached steady-state (8), and simi-
lar evidence suggests that some maize-based cropping systems
in the USA corn belt are also near steady-state (9). In these cases,
the overall NUE of a cereal cropping system is equal to the RE
N
.
In contrast to systems at steady-state, adoption of new man-
agement practices or crop rotations that affect the soil carbon
(C) balance will also affect the N balance because the C/N ratio
of soil organic matter is relatively constant. In such cropping sys-
tems, the overall NUE of the cropping system must include
changes in the size of soil organic and inorganic N pools in ad-
dition to the RE
N
. When soil-N content is increasing, the amount
of sequestered N contributes to a higher NUE of the cropping
system, and the amount of sequestered N derived from applied
N contributes to a higher RE
N
. Conversely, any decrease in soil-
N stocks will reduce NUE and RE
N
.
Unfortunately, there is a paucity of reliable data on RE
N
based
The global challenge of meeting increased food demand
and protecting environmental quality will be won or lost in
cropping systems that produce maize, rice, and wheat.
Achieving synchrony between N supply and crop demand
without excess or deficiency is the key to optimizing trade-
offs amongst yield, profit, and environmental protection in
both large-scale systems in developed countries and small-
scale systems in developing countries. Setting the research
agenda and developing effective policies to meet this
challenge requires quantitative understanding of current
levels of N-use efficiency and losses in these systems, the
biophysical controls on these factors, and the economic
returns from adoption of improved management practices.
Although advances in basic biology, ecology, and biogeo-
chemistry can provide answers, the magnitude of the
scientific challenge should not be underestimated because
it becomes increasingly difficult to control the fate of N in
cropping systems that must sustain yield increases on the
world’s limited supply of productive farm land.
Agroecosystems, Nitrogen-use Efficiency,
and Nitrogen Management
Kenneth G. Cassman, Achim Dobermann and Daniel T. Walters
133Ambio Vol. 31 No. 2, March 2002 © Royal Swedish Academy of Sciences 2002
http://www.ambio.kva.se
on measurements from on-farm studies in the major cereal pro-
duction systems. Likewise, we are not aware of measurements
of on-farm NUE that include the contributions from both RE
N
and changes in soil-N reserves. This shortage of information re-
flects the logistical difficulty and high cost of obtaining direct
on-farm measurements and the lack of funding for what appear
to be routine on-farm evaluations. Available data indicate a very
low mean RE
N
of 31% in continuous irrigated rice systems in
Asia (2, 10), and somewhat higher efficiency of 37% for maize
in the major maize-producing states of the USA (Table 1). In
contrast, mean RE
N
for wheat in rice-wheat systems of India was
18% in one year and 49% the next. This difference was associ-
ated with low grain yields in the first year caused by unfavorable
weather, and highlights the importance of robust crop growth
and yield to greater RE
N
. Good crop management and high yields
of rain-fed wheat in northwestern Europe also contribute to rela-
tively high RE
N
in those systems (11). Most other estimates of
RE
N
in the literature are obtained from experimental plots at re-
search stations, which tend to overestimate RE
N
for the reasons
previously described.
Two methods are commonly used for direct measurement of
RE
N
, and both have inherent weaknesses (12). The ‘N-difference’
method is based on the difference in N uptake between a crop
that receives a given amount of applied N and N uptake in a ref-
erence plot without applied N. Another technique uses
15
N-
labeled fertilizer to estimate crop recovery of applied N. Each
of these methods can be confounded by ‘added-N effects’ when
the applied N alters the ability of the plant root system to ac-
quire N from soil, the rate of net N mineralization from organic
N pools, or both. In addition, the
15
N-fertilizer technique can also
be confounded by ‘pool substitution’ whereby N from applied
15
N-fertilizer replaces N in the various soil N pools during the
processes of N immobilization-mineralization turnover from or-
ganic matter and microbial biomass. Because estimates of RE
N
by the N-difference method are influenced by fewer confound-
ing factors, we believe it is preferable to the
15
N-fertilizer tech-
nique. The data in Table 1 and cited throughout this paper are
based on this method.
The NUE of agricultural systems also have been calculated
using aggregate databases on crop production statistics and lit-
erature-based assumptions about N cycling to estimate N inputs
and outputs on a regional or global basis. For example, Smil’s
(6) elegant global N balance for crop production estimates an
average N recovery efficiency in crop biomass of 50% from all
sources of N input—including fertilizers, atmospheric deposi-
tion, biological N
2
fixation, recycled crop residues, and manures.
However, N recovery efficiencies can differ substantially from
each of these N sources, and therefore it is not possible to esti-
mate RE
N
by this approach. The much lower estimates of RE
N
based upon direct on-farm measurements for rice in Asia and
maize in the North-Central USA (Table 1) may reflect higher N
uptake efficiency from indigenous N sources than from applied
fertilizer. Moreover, the overall NUE of these systems would be
higher or lower depending on whether soil N reserves are in-
creasing or decreasing over time.
In recent years, significant strides towards increasing RE
N
are
suggested from aggregate data of fertilizer use and crop yields.
Since the early 1980s, the ratio of crop yield per unit of applied
N fertilizer (called the partial factor productivity for N ferti-
lizer—PFP
N
) has increased in Japan (13), and the USA (14). For
USA maize, PFP
N
increased by 36% in the last 21 years, from
42 kg kg
–1
in 1980 to 57 kg kg
–
1 in 2000 (Fig. 1). Because crop
dry matter accumulation and grain yield are closely correlated
with N uptake, the increase in PFP
N
since 1980 suggests an as-
sociated increase in RE
N
—assuming the indigenous N supply
from net mineralization of soil organic matter, atmospheric N
Table 1. Nitrogen fertilizer-uptake efficiency
+
(or recovery efficiency, RE
N
) by
maize, rice, and wheat crops based on data obtained from on-farm
measurements in their major cropping systems.
Crop Region/Countries Number N fertilizer (kg ha
–1
)RE
N
(% of applied)
(cropping system) of farms mean (+/- SD) mean (+/- SD)
Maize
++
North-central USA 55 103 ( 85) 37 ( 30)
(various rotations)
Rice
+++
China, India, 179 117 ( 39) 31 ( 18)
Indonesia, 179 112 ( 28) 40 ( 18)
Phillipines,
Thailand, Vietnam
(rice-rice)
Wheat
++++
India 23 145 ( 31) 18 ( 11)
(rice-wheat) 21 123 ( 30) 49 ( 10)
+ Recovery efficiency is the proportion of applied N fertilizer that is taken up by the
crop. It is determined by the difference in the total amount of N measured in
aboveground biomass at physiological maturity between replicated plots that receive
N fertilizer and control plots without applied N. Except for the omission of N in control
plots, all crop-management practices are determined and applied by the farmer of
each field.
++ Data obtained from on-farm experiments located in Illinois, Michigan, Minnesota,
Missouri, Nebraska, and Wisconsin. Experiments were conducted from 1995–1999
by researchers in the NC 218 Regional Research Project. At each site, replicated
plots received N-fertilizer across a wide range of N-application rates, including a
control without applied N. Management practices other than N-fertilizer rate were
imposed by the farmer. RE
N
was estimated as described above.
+++ Data from on-farm experiments conducted at 179 sites in major irrigated rice domains
of Asia from 1997 to 2000 with measurements taken in 4 consecutive rice crops at
each site (2). The first row of data were taken from the field-at-large where nutrient
management practices were applied by the farmer without guidance from
researchers, whereas the second row represents field-specific nutrient management
whereby the amount of applied fertilizer was adjusted to account for the balance
between soil nutrient supply capacity and crop demand.
++++ Data from on farm studies of rice-wheat systems in North India (A. Dobermann, C.
Witt, and B. Mishra, unpubl. data) following similar methods as for rice (2). Data in
the first row were from a year in which mean yields were relatively low because of
unfavorable weather (1998: average grain yield 2.3 Mg ha
–1
), whereas the second
row of data is for a favorable year with considerably larger mean yields (1999:
average grain yield 4.8 Mg ha
–1
).
inputs, and biological N fixation have re-
mained relatively constant during this pe-
riod. In contrast, there appears to have been
little improvement in RE
N
of irrigated rice
in tropical Asia; on-farm efficiencies meas-
ured in the late 1960s and early 1970s (15)
are comparable to estimates made in the
late 1990s as given in Table 1. Understand-
ing the reasons for these trends in PFP
N
and
RE
N
and the prognosis for improving them
depends on knowledge of the factors that
govern N demand and supply in cereal
cropping systems.
BIOPHYSICAL DETERMINANTS OF
CROP NITROGEN REQUIREMENTS
Crop-N demand is determined by biomass
yield and the physiological requirements
for tissue N. Crop-management practices
and climate have the greatest influence on
yield. Climate varies considerably from
year to year, which causes large differences
in yield potential. In irrigated systems, the
yield potential of a given crop cultivar is
largely governed by solar radiation and
temperature. In dryland systems, rainfall
amount and temporal distribution also have
a large influence on yield potential. While
solar radiation, temperature, and moisture
regimes determine the genetic yield ceiling,
actual crop yields achieved by farmers are
generally far below this threshold because
it is neither possible, nor economic, to re-
move all limitations to growth from sub-
134
© Royal Swedish Academy of Sciences 2002 Ambio Vol. 31 No. 2, March 2002
http://www.ambio.kva.se
Figure 1. Trends in maize grain yield, use of N fertilizer, and Partial Factor
Productivity from applied N fertilizer (PFP
N
, kg grain yield kg
–1
N applied) in
the USA.
C
3
crops [site-years] include lucerne [12], fescue [7], French
bean [1], potato [7], cabbage [1], wheat [2] and rape [4], n =
181. C
4
crops [site-years] include sorghum [10], maize [2] and
setaria [2], n = 75.
Sources of data: Mean annual maize yields, National Agricultural
Statistics Service, USDA http://www.usda.gov/nass/; mean
annual N fertilizer N use, USDA Annual Cropping Practices
Surveys (> 2000 farms representing 80 to 90% of the USA maize
area), Economic Research Service, USDA, http://
www.ers.usda.gov/
Figure 2. Relationships between dry matter yield and nitrogen
content of plant tissue for C
3
and C
4
crops. (Source: 17).
Crop biomass (Mg ha
–1
)
Biomass N concentration (g kg
–1
)
optimal nutrient supply, weed competition, and
damage from insects and diseases. Hence, the in-
teraction of climate and management causes tre-
mendous year-to-year variation in on-farm yields
and crop N requirements.
Crop physiological N requirements are con-
trolled by the efficiency with which N in the
plant is converted to biomass and grain yield.
Because cereal crops are harvested for grain, the
most relevant measure of physiological N effi-
ciency (PE
N
) is the change in grain yield per unit
change in N accumulation in aboveground
biomass. Crop-PE
N
is largely governed by 2 fac-
tors: i) the genetically determined mode of pho-
tosynthesis—either the C
3
or C
4
photosynthetic
pathway; and ii) the grain N concentration—also
under genetic control but affected by N supply
as well. Both rice and wheat are C
3
plants while
maize is a C
4
plant. The C
4
plants tend to have
greater PE
N
than C
3
plants because the C
4
path-
way has a higher photosynthetic rate per unit
leaf-N content (16), which results in greater
biomass production per unit of plant-N accumu-
lation (Fig. 2, ref. 17).
Large genetic variation in grain-N concentra-
tion within each of the major cereal species has
allowed plant breeders to develop cultivars with
the desired grain-N concentration for specific
end-use properties. Relatively low grain-N con-
tent of 10–12 g kg
–1
here and elsewhere, grain-
N concentration is given on a dry weight basis
desired in rice for optimal cooking and eating
quality. Maize-N content also is relatively low
(13–14 g kg
–1
) because maize products for hu-
man consumption or animal feed do not require
high protein. In contrast, the N concentration of
wheat must exceed 18 g kg
–1
to have acceptable
quality for bread or noodles. The relationship be-
tween grain yield and the N contained in
aboveground biomass at physiological maturity
provides a measure of PE
N
across a wide range
of production environments (Fig. 3). The line at
the upper boundary of data points in this Figure
provides an estimate of maximum N dilution in
plant biomass, which occurs when N is the most
limiting factor to plant growth. When N is no
longer the most limiting resource and other fac-
tors such as water supply, pest damage, or defi-
ciencies of other nutrients reduce crop growth,
the amount of grain produced per unit N uptake
decreases and moves off the line of maximum
N dilution.
Across a wide range of production environ-
ments and management practices, maize tends to
have a larger increase in grain yield per unit N
uptake than rice because it is a C
4
plant. This ad-
vantage in PE
N
is evident in the slopes of regres-
sion lines in Figure 3. Rice has a lower effi-
ciency than maize because it is a C
3
plant al-
though its lower grain N concentration partially
offsets this disadvantage. Wheat has the small-
est PE
N
of the 3-major cereals because it is a C
3
plant with high grain protein (data not shown).
Two additional points are noteworthy. First, the
lines defining both maximum N dilution and the
overall regression are curvilinear, which means
there is a diminishing return to the conversion
of plant N to grain as yields approach the yield
Fertilizer N (kg ha
–1
) Grain yield (Mg ha
–1
)
PFP
N
(kg grain kg
–1
N)
135Ambio Vol. 31 No. 2, March 2002 © Royal Swedish Academy of Sciences 2002
http://www.ambio.kva.se
Data sources:
i
) for rice, data obtained from on-
farm and research station experiments conducted
across a wide range of agroecological
environments in Asia from 1995 to 2000 (n = 1658);
ii
) for maize, data obtained from on-farm and
research station experiments conducted across a
wide range of agroecological environments in the
North-Central USA from 1995 to 2000 (n = 470).
Blue lines indicate the boundary of maximum
dilution of N in the plant (maximum physiological
efficiency), whereas the black lines depict the
average physiological efficiency as obtained from
nonlinear regression for the entire data set for
maize and rice.
Figure 3. Relationship between grain yield and plant-N accumulation in aboveground biomass at physiological maturity in maize and rice.
Plant-N accumulation (kg ha
–1
)
Grain yield (kg ha
–1
)
potential ceiling. Second, the N concentration of cereal straw
and stover is much smaller than in grain, and differences
among cereal crops or among cultivars of the same crop spe-
cies are relatively small. Therefore, the amount of N remain-
ing in straw or stover has a relatively small effect on PE
N
unless factors other than N are limiting crop growth and grain
yield.
DYNAMICS OF NITROGEN SUPPLY
Inorganic nitrate and ammonium ions are the primary source
of N taken up by plant roots. Both indigenous soil resources
and applied N inputs contribute to this plant-available N pool,
which represents a very small fraction of total soil-N. For
example, a typical irrigated rice soil in Asia contains about
2800 kg N ha
–1
in the top 20 cm of soil where roots derive
the majority of crop-N supply. Of this total, the amount of
N derived from indigenous resources during a single crop-
ping cycle typically ranges from 30–100 kg N ha
–1
(Fig. 4a),
which represents only 1–4% of total soil N. For cereal crops,
we define the indigenous soil-N supply as the amount of N
the crop obtains from the inorganic N pool, net N minerali-
zation from soil organic matter and incorporated crop
residues, biological N
2
fixation by soil microflora in the
rhizosphere and floodwater (in the case of irrigated rice), and
inputs of N from atmospheric deposition and irrigation wa-
ter. Similarly, total-N in the top 20 cm of a fertile prairie soil
in the USA corn belt is about 4000 kg N ha
–1
, and the indig-
enous N supply typically ranges from 80–240 kg N ha
–1
(Fig.
4b), which is 2–6% of total soil-N. Although small in size,
the indigenous N supply has a very high N-fertilizer substi-
tution value because of the relatively low RE
N
from applied
N fertilizer.
A maize crop that produces a grain yield of 10␣ 000 kg ha
–
1
requires total uptake of about 190 kg N ha
–1
(Fig. 3). The
indigenous N supply typically provides about 130 kg N ha
–1
(median value in Fig. 4b), which leaves 60 kg N ha
–1
that
must be provided by applied N. If RE
N
is 37%, which is typi-
cal of on-farm conditions (Table 1), then an N-fertilizer rate
of 162 kg N ha
–1
must be applied to meet crop-N demand.
If the indigenous N supply decreases from 130 to 100 kg N
ha
–1
(a 23% reduction), then the N-fertilizer requirement in-
creases by 50% to 243 kg N ha
–1
, assuming RE
N
remains con-
stant at this higher N fertilizer rate. However, RE
N
typically
I
N
for rice was measured at on-farm sites at 179 locations in South
and Southeast Asia (Source: C. Witt and A. Dobermann, Reversing
Trends of Declining Productivity in Intensive Irrigated Rice Sys-
tems, On-farm Monitoring Database, June 2000 release; IRRI, Los
Banos, Philippines). I
N
for maize was measured at 64 loca-
tions in 6 major maize-producing states in the North-Central USA
in replicated field experiments and on-farm trials (Source: D.
Walters, Univ. of Nebraska; North Central regional Research Pro-
ject NC-218). For both rice and maize, the I
N
at each site was mea-
sured as described in the footnotes to Table 1. For comparison,
mean total soil N content in the 0-20 cm topsoil layer was 1.4 ± 0.7
g kg
–1
at the rice sites in Asia and 1.6 ± 0.1 g kg
–1
at the maize sites
in North America.
Figure 4. Variation in the indigenous N supply (I
N
, plant N accumulation in
on-farm plots that did not receive N fertilizer). of a. rice fields in Asia; and
b. maize fields in the North-Central USA.
I
N
(kg ha
–1
)
No. of observationsNo. of observations
16 000
14 000
12 000
10 000
8 000
6 000
4 000
2 000
0
350300250200150100500
