Home Search Collections Journals About Contact us My IOPscience

A tradeoff frontier for global nitrogen use and cereal production

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2014 Environ. Res. Lett. 9 054002

(http://iopscience.iop.org/1748-9326/9/5/054002)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 72.93.11.41
This content was downloaded on 01/02/2015 at 16:12

Please note that terms and conditions apply.

 Environmental Research Letters

Environ. Res. Lett. 9 (2014) 054002 (8pp) doi:10.1088/1748-9326/9/5/054002

A tradeoff frontier for global nitrogen use

and cereal production
Nathaniel D Mueller1,2, Paul C West2, James S Gerber2,
Graham K MacDonald2, Stephen Polasky3 and Jonathan A Foley2
1
Center for the Environment, Harvard University, 24 Oxford Street, Cambridge, MA 02138, USA
2
Institute on the Environment (IonE), University of Minnesota, 325 Learning and Environmental Sciences,
1954 Buford Avenue, St Paul, MN 55108, USA
3
Department of Applied Economics, 1994 Buford Avenue St. Paul, MN 55108, USA

E-mail: [email protected]

Received 17 February 2014, revised 7 April 2014

Accepted for publication 8 April 2014
Published 15 May 2014

Abstract
Nitrogen fertilizer use across the worlds croplands enables high-yielding agricultural
production, but does so at considerable environmental cost. Imbalances between nitrogen applied
and nitrogen used by crops contributes to excess nitrogen in the environment, with negative
consequences for water quality, air quality, and climate change. Here we utilize crop input-yield
models to investigate how to minimize nitrogen application while achieving crop production
targets. We construct a tradeoff frontier that estimates the minimum nitrogen fertilizer needed to
produce a range of maize, wheat, and rice production levels. Additionally, we explore potential
environmental consequences by calculating excess nitrogen along the frontier using a soil
surface nitrogen balance model. We nd considerable opportunity to achieve greater production
and decrease both nitrogen application and post-harvest excess nitrogen. Our results suggest that
current (circa 2000) levels of cereal production could be achieved with 50% less nitrogen
application and 60% less excess nitrogen. If current global nitrogen application were held
constant but spatially redistributed, production could increase 30%. If current excess nitrogen
were held constant, production could increase 40%. Efcient spatial patterns of nitrogen use on
the frontier involve substantial reductions in many high-use areas and moderate increases in
many low-use areas. Such changes may be difcult to achieve in practice due to infrastructure,
economic, or political constraints. Increases in agronomic efciency would expand the frontier to
allow greater production and environmental gains.

S Online supplementary data available from stacks.iop.org/ERL/9/054002/mmedia

Keywords: agriculture, nitrogen, fertilizer, crop yield

1. Introduction century. While this technology has enabled large increases in

food production on existing croplands, it has also come at
Improving nitrogen (N) management across global croplands substantial costs to ecosystem and human health.
is crucial to increasing agricultural productivity and human Excess reactive nitrogen (hereafter referred to as excess
well-being [17]. Ammonia synthesis from the HaberBosch nitrogen) in the form of nitrate readily leaches into water-
process allowed a dramatic acceleration of reactive nitrogen ways. Nitrate can contaminate groundwater and contributes to
(Nr) use for agriculture during the latter half of the 20th eutrophication of surface waters, decreasing water clarity as
well as sh abundance [1, 2, 5, 6, 8, 9]. Nitrous oxide (N2O),
released from agricultural soils through nitrication and
Content from this work may be used under the terms of the denitrication, is a powerful greenhouse gas that also depletes
Creative Commons Attribution 3.0 licence. Any further
distribution of this work must maintain attribution to the author(s) and the stratospheric ozone [5, 1012]. Volatilizing ammonia and
title of the work, journal citation and DOI. nitric oxide (NO) emissions from agricultural soils can

1748-9326/14/054002+08$33.00 1 2014 IOP Publishing Ltd

Environ. Res. Lett. 9 (2014) 054002 N D Mueller et al

negatively impact regional air quality. Excess nitrogen can fertilizer application [4], irrigated area [27], and climate [28].
also acidify soil [13], and atmospheric deposition of reactive Additional bootstrap analyses are performed to provide a 95%
nitrogen can negatively impact terrestrial biodiversity [5]. The CI on the MB nitrogen response coefcients. We utilize
HaberBosch process itself also has substantial environmental 1000 bootstrap samples per climate zone per crop, and sam-
impacts; the process consumes 2% of global energy [5]. pling of grid cells is performed in proportion to harvested
The nonlinear response of crop yields to nitrogen appli- area. We note that the bootstrapping does not account for
cation rates affords an opportunity to improve crop yields spatial autocorrelation, and thus may underestimate uncer-
while greatly reducing environmental pollution from excess tainty. Some crop-climate zone combinations in the Mueller
nitrogen application. Crop yield response to nitrogen is rela- et al [4] model lack a unique nitrogen response coefcient;
tively steep at low application rates and shallow to at at high this occurred when adding nitrogen as an explanatory variable
application rates [4, 14]. Imbalances between nitrogen applied did not minimize root-mean square error of the model t (for
and the ability of crops to utilize nitrogen leads to extreme example, nitrogen can have weak explanatory power in cases
amounts of excess nitrogen in some areas and nitrogen de-
where yields are primarily limited by irrigation or other fer-
cits in others [3, 15, 16]. Efforts to balance patterns of
tilizer nutrients). In these climate zones the model utilizes
nitrogen surplus and decit could reduce post-harvest nitro-
average MB response coefcients across all climates (and
gen losses and can have the additional benet of improving
upper and lower bound response coefcients from the boot-
net production outcomes. For example, reduction of nitrogen
use in parts of the Peoples Republic of China would have strapping results) in conjunction with climate-specic mini-
little or no impact on crop yields [17, 18]. In the European mum and maximum yields to estimate an expected nitrogen-
Union, it is estimated that current wheat nitrogen application yield response.
rates are approximately 50 kg ha1 higher than what would be Using the crop- and climate-specic nitrogen response
socially optimal, given both the costs and the benets of curves, we build a tradeoff frontier relating total global pro-
nitrogen application [19]. Elsewhere, cropping systems in duction to total global nitrogen fertilizer use. To do this, for
some developing countries suffer from too-little nitrogen use, each crop we nd levels of nitrogen application across all
which may have detrimental long-term effects on soil quality climate zones for each crop that give the same marginal yield
and yields [5, 20]. In areas of low nutrient application response (dY/dN), which simulates maximum production for
places at the steep end of the nitrogen-yield curvesmall a total amount of nitrogen. We use 30 dY/dN values from
increases in fertilizer use can help prevent mining of soil 0.0001 to 2 Mg yield kg1 N. This exercise results in a series
nutrients while dramatically increasing yields [5, 2123]. of crop-specic production and nitrogen application maps
Opportunities to simultaneously increase food production (simulated at 5 arc min 5 arc min resolution) for every point
and minimize environmental harm can be mapped using a on the tradeoff frontier, which are then summed across the
tradeoff frontier that describes the minimum level of applied globe (using xed crop areas [26]) to determine total global
nitrogen necessary to achieve various crop production levels. production and nitrogen consumption per crop. Maximum
Tradeoff frontiers are a useful framework for determining yields differ between irrigated and rainfed systems, and
potential win-win scenarios between endpoints seemingly in expansion of irrigation may not be feasible in areas with
conict [e.g. 24, 25]. limited water supplies. Thus, similar calculations are made for
Here we provide a rst estimate of the tradeoff frontier a scenario with no changes to irrigated areas, where nitrogen
for global nitrogen fertilizer use and production of three major application rate increases stop on rainfed areas when yields
cereals: maize, wheat, and rice. In this analysis, we assume reach the rainfed yield ceilings of Mueller et al [4]. These
other nutrients and management practices are not limiting, frontier calculations assume other inputs or management
and we utilize agronomic efciencies (i.e. crop production
practices (e.g. phosphorus and potassium application, cultivar
realized for a given application rate) implicit in a suite of
choice, etc) are not limiting yields. Organic and biologically-
crop- and climate-specic nitrogen-yield curves para-
derived nitrogen sources are not included as drivers of pro-
meterized using circa 2000 datasets [4]. We additionally
duction due to data limitations (see section 4.4), although they
explore potential environmental consequences by calculating
can be important sources of fertility.
excess nitrogen using a soil surface nutrient balance model for
each point along the frontier. To aggregate tradeoff frontiers across multiple crops, we
keep the observed proportion of nitrogen allocation between
crops (circa 2000) xed, and then add up modeled production
at different levels of total nitrogen use. Spline interpolation in
2. Methods MATLAB was used to interpolate between points on the
crop-specic curves in order to construct the aggregate curve.
2.1. Building the nitrogen consumption-production tradeoff
Finally, we overlay 19972003 nitrogen consumption by crop
frontier
[4] and 19972003 cereal production [26] to assess the cur-
Mueller et al [4] estimated MitscherlichBaule (MB) nitro- rent nitrogen use and production situation relative to the
gen-yield curves and rainfed maximum yields for crop-spe- modeled frontier. All calculations are performed over the
cic climate zones (100 zones per crop). These curves were 95% of crop harvested area [26] encompassed by the crop-
parameterized using global data on crop areas and yields [26], specic climate zones.

2
Environ. Res. Lett. 9 (2014) 054002 N D Mueller et al

2.2. Soil surface nitrogen balances

Nitrogen balance calculations are carried out using the

approach of Foley et al [3], which builds on earlier nutrient
balance work for both nitrogen [16, 29] and phosphorus [30].
This method tracks all of the nitrogen inputs (fertilizer [4],
manure [31], atmospheric deposition [32], and legume xa-
tion [3]) and outputs (removal of nitrogen in harvested
material) spatially for each grid cell. We modify the approach
of Foley et al [3] to assess crop-specic nitrogen balances by
distributing manure nitrogen between the crops within each
grid cell in proportion to the harvested area of each crop.
Legume xation rates are not included, as we only examine
cereal crops in this study.
As we build the nitrogen consumption and production
frontier, we generate a new nitrogen application rate map and
a new yield map for each crop at each point on the curve.
These maps are used as inputs to the nitrogen balance model
to generate a map of excess nitrogen at every point, which is
summed to create the excess nitrogen and production tradeoff
frontier.

3. Results

Across major cereals, we nd large inefciencies in the cur-

rent (circa 2000) distribution of nitrogen use and crop pro-
duction. These inefciencies are illustrated graphically as the
distance between the current state and the tradeoff frontiers in
gure 1 and online gures S1S3 available at stacks.iop.org/
ERL/9/054002/mmedia. Specically, we nd that production
could be held constant and nitrogen use could drop
approximately 48% (lower- and upper-bound results with
95% CIs on nitrogen response coefcients: 3755%). Alter-
natively, nitrogen use could be held constant and global
production of major cereals could increase approximately Figure 1. Global-scale tradeoff frontiers for (a) major cereal
28% (2034%) via changes in the geographic intensity of production and fertilizer nitrogen consumption and (b) major cereal
nitrogen use (and associated changes in correlated manage- production and excess nitrogen as dened by a soil surface nitrogen
balance model. Excess nitrogen considers manure application and
ment practices). Assuming no expansion of irrigated area
atmospheric deposition in addition to fertilizer nitrogen. Shaded
lowers production potential at high nitrogen consumption areas represent uncertainty ranges calculated with a 95% condence
values (gure 1). interval on the nitrogen-yield response coefcients.
With the excess nitrogen-production tradeoff frontier, we
nd that production could be held constant and excess
nitrogen could decrease approximately 61% (4967%). moderate application rates. Lower nitrogen application rates
Excess nitrogen could be held constant and production could translate into substantially reduced rates of mass balance
increase approximately 39% (2946%). excess nitrogen, with only a few locations showing noticeable
Spatial patterns of nitrogen application, excess nitrogen, increases due to higher application rates (gure 3). Some
and cereal yield change with efcient nitrogen allocation. To locations with high atmospheric deposition and manure
illustrate these changes, we provide a series of maps for a nitrogen remain excess nitrogen hotspots despite lower
scenario where cereal production and irrigated areas are held application rates (parts of the US, Mexico, northern Europe,
constant, and nitrogen fertilizer is applied spatially to mini- the Peoples Republic of China, and South Korea).
mize total nitrogen use (gures 24). We nd substantial While production is held constant under this scenario,
decreases in nitrogen application rates across the Peoples there are some changes to the spatial patterns of cereal yield
Republic of China, the United States, India, and parts of (gure 4). In general, there is also an evening out of yield
Western Europe (gure 2). In contrast, much of Eastern patterns that accompanies the more homogenous nitrogen
Europe and Africa show increases in nitrogen application application rates. However, the relatively high productivity of
rates. While total nitrogen consumption lowers, the remaining some major agricultural regions with high yield potential (e.g.
nitrogen consumption occurs at more homogenous, relatively the US Midwest) remains visible. As the nitrogen-yield

3
Environ. Res. Lett. 9 (2014) 054002 N D Mueller et al

Figure 2. Changes to nitrogen application rates under an efcient Figure 3. Changes to nitrogen excess rates under an efcient spatial
spatial nitrogen allocation scenario with constant total cereal nitrogen allocation scenario with constant total cereal production and
production and no changes to irrigated area. (a) Average nitrogen no changes to irrigated area. (a) Average nitrogen excess rates for
application rates for maize, wheat, and rice circa 2000. (b) Average maize, wheat, and rice circa 2000. (b) Average nitrogen excess rates
nitrogen application rates for major cereals under the modeled for major cereals under the modeled scenario. Integrated histograms
scenario. Integrated histograms visualize the amount of total nitrogen visualize the amount of total excess nitrogen within rate categories.
consumption within application rate categories, and show a shift
towards lower and more homogenous application rates.
We nd wheat nitrogen inputs of 14.7, 4.6, and 0.8 Mt
nitrogen from chemical fertilizer, manure, and atmospheric
deposition, respectively. Excess nitrogen for wheat is
curves utilized here represent the minimum nitrogen needed approximately 10.4 Mt, or 52% of total wheat nitrogen inputs.
to achieve a given yield, modeled frontier yields (as in We nd rice nitrogen inputs of 12.8, 5.4, and 0.6 Mt nitrogen
gure 4(b)) can be thought of as maximum possible yield from chemical fertilizer, manure, and atmospheric deposition,
given a certain level of nitrogen. This yield is modeled with respectively. Rice had the largest amount of excess nitrogen
circa year 2000 agronomic efciency, and it is assumed other of the three crops: 12.0 Mt nitrogen, or 64% of all rice
management practices, nutrients, and seed quality are not nitrogen inputs.
limiting.
Maize, wheat, and rice cultivation are responsible for
relatively similar amounts of excess nitrogen circa 2000. 4. Discussion
Across the top 95% of crop area modeled in this study, we
calculate maize nitrogen inputs to be 12.0 Mt chemical ferti-
4.1. Increasing production and decreasing excess nitrogen
lizer nitrogen, 4.0 Mt manure nitrogen, and 0.6 Mt nitrogen
from atmospheric deposition. Excess nitrogen for maize is The tradeoff frontier for nitrogen use and cereal production
approximately 8.6 Mt, or 51% of all maize nitrogen inputs. potential quanties the large potential for changes to global

4
Environ. Res. Lett. 9 (2014) 054002 N D Mueller et al

Moving the tradeoff frontier), as the models used to construct

the tradeoff frontier are calibrated to circa 2000 data.
Moving towards more efcient spatial allocation of
nitrogen would likely require substantial changes to infra-
structure and markets. For example, many regions in Sub-
Saharan Africa with low nutrient use suffer from a combi-
nation of high fertilizer prices due to overland transport costs
and low per-capita income. In addition, inadequate storage
facilities may prevent marketing of increased crop production,
further discouraging management for higher productivity
[33]. Likewise, reducing nitrogen use in areas of currently
high nitrogen use may be politically and economically chal-
lenging unless reductions can be done in a way that doesnt
hurt farm prots and productivity.

4.2. Impacts on human well-being

Reallocating nitrogen intensity to both increase production

and decrease global excess nitrogen should tend to increase
human well-being through several pathways. However, such
changes will depend strongly on local conditions and the
direction and magnitude of local changes to nitrogen use, and
areas with increases in excess nitrogen could see negative
impacts.
Increasing crop production in areas with currently low
yields could enhance food security and promote economic
development. However, while food availability (i.e. crop
production) is necessary for access to healthy, safe, and
nutritious food, it is not sufcient to ensure food security.
Chronic povertynot food availabilityis the primary cause
of food insecurity [34]. Income inequality, markets, infra-
structure, and institutions also play a major role [35].
Efforts to reallocate nitrogen fertilizer application inten-
sity would decrease N2O emissions disproportionately to any
reduction in nitrogen use, leading to large benets for climate
Figure 4. Changes to major cereal yield under an efcient spatial and stratospheric ozone. This result is due to nonlinear
nitrogen allocation scenario with constant total cereal production and
no changes to irrigated area. (a) Average cereal yield for maize, increases in N2O emissions at high nitrogen application rates,
wheat, and rice circa 2000. (b) Average cereal yield under the particularly when there are large imbalances between applied
modeled scenario. Integrated histograms visualize the amount of and removed nitrogen [36, 37].
total cereal production within yield categories. Potential human health benets from decreasing excess
nitrogen include decreases in nitrate-contaminated water and
improved air quality [5, 7, 8, 19]. Ecosystem recreation value,
nitrogen management that both decrease excess nitrogen and shing, and food provision could also benet from improved
increase crop production. While the analysis is not a pre- water quality and climate change mitigation [8, 38]. Biodi-
scription for any particular management change, it is a proof versity, which is important for underpinning myriad ecosys-
of concept that changes in the spatial distribution of nitrogen tem services, could benet from improved air and water
intensity could lead to large production and environmental quality, less severe climate change, and reductions in reactive
gains. Comparing spatial patterns of nitrogen application, nitrogen deposition [5, 8]. In contrast, areas where nitrogen
nitrogen excess, and yield on the frontier relative to the cur- use and excess nitrogen increases (such as Eastern Europe,
rent situation (gures 24) provides insight into the direction Russia, and Sub-Saharan Africa in the scenario presented by
and magnitude of changes possible to both application rates gures 24) could experience declines in human well-being
through negative impacts on water and air quality.
and production. Pathways to achieve these changes involve a
Though not explicitly modeled here, we expect that
reallocation of nitrogen use intensity: decreases in areas
efcient spatial allocation of nitrogen use would increase
where current nitrogen use is high, and increases in areas
aggregate human well-being by coupling relatively large
where nitrogen use is currently low. In addition, the whole
excess nitrogen reductions in areas of currently high nitrogen
system could and should improve by increasing the agro- use with small to moderate excess nitrogen increases in areas
nomic or technological efciency of N use (see section: with currently low nitrogen use. However, more information

5
Environ. Res. Lett. 9 (2014) 054002 N D Mueller et al

on the relationships between nitrogen-driven ecosystem improved seed can also improve eld-scale agronomic ef-
impacts and human well-being in different regions would ciency and production [14]. Such improvements, and the
substantially aid these analyses. For example, population continued development of crop varieties with higher yield
density in an area, the relationship between nitrogen pollution potentials, would contribute to future shifts in the frontier. In
and shing productivity, marginal benets of a reduction in addition, landscape management with perennial vegetation
air pollution, and other factors all inuence the complex can capture runoff, and winter cover crops can reduce reactive
relationships between nitrogen and human well-being in a nitrogen losses [40].
given region [8]. If future research is able to systematically Future work should quantify how the adoption of prac-
quantify the costs and benets (social and private) of agri- tices that improve agronomic efciency would change the
cultural nitrogen use across the globe [e.g. 5, 7, 19], this tradeoff frontier and the current nitrogen-production status
information would allow greater insight into nitrogen man- (i.e. changes to gure 1). Such efforts would benet from
agement scenarios that maximize well-being. consistent assessments of how global fertilizer use is allocated
between different crops over both time and space, detailed
4.3. Moving the tradeoff frontier
records of which are currently very limited.

The tradeoff frontier presented here should be viewed as a

4.4. Uncertainties and limitations
rst estimate of potential efciency gains at a snapshot in
time, characterizing production possibilities and tradeoffs All models are simplications of reality, and agricultural
with circa 2000 technology and average agronomic ef- management is a complex reality to simulate. Thus, there are
ciencies. Substantial changes to the frontier are possible with a number of limitations and caveats that should be considered
improved practices and technologies. when interpreting this work.
A variety of management techniques could increase on- The nitrogen-yield response curves utilized in this study
eld nitrogen use efciency and decrease excess nitrogen. estimate the minimum amount of nitrogen needed to deter-
Such strategies are characterized under the 4R Nutrient mine a given yield circa 2000, and do not represent theore-
Stewardship paradigm, which promotes use of the right tically optimal agronomic efciency [e.g. 45]. While the
fertilizer source with the right application rate, timing of curves were parameterized taking into account possible yield
application, and fertilizer placement in order to increase limitation by irrigation, phosphorus, and potassium, frontier
agronomic efciency [39, 40]. Integrated management sys- nitrogen use as estimated here assumes these inputs (and other
tems that incorporate these factors have provided dramatic management practices, including cultivar choice) are not
experimental evidence that substantial increases to agronomic limiting yields. The curves represent a general yield response
efciencies are possible [18, 41]. For example, Chen et al for a given range of growing conditions, and are not appro-
[42] were able to double maize yields in the Peoples priate for quantifying yield response for specic elds under
Republic of China while completely eliminating mass balance specic management regimes. Given the scale at which the
excess nitrogen by applying nitrogen in ve split doses with curves were parameterized, the curves do not capture varia-
soil testing guiding application rates. While such practices bility in agronomic efciency and yield that can occur at ne
may represent a substitution of labor and technology for spatial and temporal scales in response to eld-specic
nitrogen, increasing adoption of these sophisticated agro- management and soil conditions [46, 47]. Likewise, this
nomic practices could dramatically push up the tradeoff model only captures yield as a function of applied fertilizer
frontier as illustrated in gure 1. nitrogen, not plant-available soil nitrogen, so it does not
However, despite the large potential for improving simulate indigenous nitrogen supply or how that stock may
agronomic efciency, broad trends in observed nitrogen change between years or with specic cropping systems or
recovery efciency (a large-scale measure of agronomic soil conditions. Furthermore, we ignore the impact of biolo-
efciency) are relatively at over time in both developed and gically derived (i.e. organic) nitrogen on yield, focusing on
developing countries [43]. Developed country efciencies inorganic fertilizer in this initial study (further discussion
remain higher than developing country efciencies. So while below regarding manure).
there is considerable potential for progress, achieving gains in The bootstrapping analysis provides an estimate of
practice will require overcoming system inertia as well as uncertainty in the nitrogen-yield responses (specically, the
infrastructure, political, and economic obstacles. MB response coefcients), but it does not capture all of the
Increasing biologically-derived nitrogen inputs can model or data uncertainties. A comparison of median boot-
decrease dependency on chemical nitrogen for fertility, strap parameter estimates and mean estimates is presented in
although may not necessarily lead to decreases in excess gure S4. Yield y-intercepts and asymptotes were set in the
nitrogen [14]. However, organic inputs are especially critical model using the upper and lower tails of the yield distribution
for replenishing soil carbon [44] and ensuring responsive soils in each climate zone [4]. We note that if high-yielding areas
[20]. Appropriate crop rotations, mixed cropping systems, were not observed within a climate zone, the asymptote for
and agroforestry with leguminous trees can replenish soil the yield response could be articially low, which would tend
fertility [33]. to make our estimate of the tradeoff frontier conservative.
Other agronomic improvements, such as appropriate The soil surface nitrogen balance model used utilizes
plant population, protection from pests and diseases, and detailed spatial data on nitrogen inputs and outputs to estimate

6
Environ. Res. Lett. 9 (2014) 054002 N D Mueller et al

excess nitrogen. However, we stress that mass balance excess possibilities in agricultural systems that maximize human
nitrogen as calculated here is only a useful proxy for nitrogen well-being. These analyses could and should be carried out at
pollution. Even locations with little to no mass balance excess many different scales and over time, with the potential to
may still experience reactive nitrogen loss throughout a improve land-use and agricultural land management decision-
growing season. making.
We also note that we modied the nitrogen balance
model of Foley et al [3] by generating crop-specic nitrogen
balances. These crop-specic calculations required appor- Acknowledgements
tioning manure inputs between crops proportionally to the
area in each crop. The resulting manure use estimates are We are grateful for helpful conversations with K Brauman,
likely the most uncertain aspect of the spatial nitrogen balance E Butler, J Hill, P Reich, and D Tilman. Research support was
approach, given the relatively rough estimates about the provided by a National Science Foundation Graduate
proportion of manure generated that is applied [see 3, 16, 30] Research Fellowship and a Ziff Environmental Fellowship to
and how the manure is divided between crops (this study). NDM, a grant from the Gordon and Betty Moore Foundation
Given the lower condence in the manure application data, to JAF, and previous funding from NASAs Interdisciplinary
manure was left out of the yield response model, but included Earth Science program to JAF. Additional research support
in the nitrogen balance model. This allows us to roughly was from the University of Minnesotas Institute on the
characterize the important inuence of manure in determining Environment, the McKnight Foundation, the Grantham
excess nitrogen in areas with high livestock densities without Foundation, World Wildlife Fund, The Nature Conservancy,
introducing data artifacts into the yield model. and Stanford. Contributions by General Mills, Mosaic, Car-
When building the cereal production frontier (gure 1), gill, Pentair, Google, Kelloggs, Mars and PepsiCo supported
we aggregate across the crop-specic frontiers (online gures stakeholder outreach and public engagement. The funders had
S1S3) by keeping a xed proportion of nitrogen consump- no role in study design, data collection and analysis, decision
tion between the crops. One could also aggregate using xed to publish, or preparation of the manuscript.
production proportions, i.e. aggregating along the y-axis
instead of the x-axis. However, even without using xed
References
production proportions, the crop production mix varies only
slightly under our method of aggregation (gure S5). More
generally, we note that the aggregate (multi-crop) production [1] Tilman G D, Cassman K G, Matson P A, Naylor R L and
frontier could shift in the future with changes towards or away Polasky S 2002 Agricultural sustainability and intensive
production practices Nature 418 6717
more nitrogen-responsive crops. [2] Galloway J N, Townsend A R, Erisman J W, Bekunda M,
All of the global datasets utilized for this study capture Cai Z, Freney J R, Martinelli L A, Seitzinger S P and
the state of agriculture circa 2000, and we do not capture Sutton M A 2008 Transformation of the nitrogen cycle:
changes in yield potential and crop area that have occurred in recent trends, questions, and potential solutions Science 320
the recent past. Updates to the crop area, yield, fertilizer, and 88992
[3] Foley J A et al 2011 Solutions for a cultivated planet Nature
irrigation data would enable better understanding of global 478 33742
agricultural management and production. In addition to [4] Mueller N, Gerber J S, Johnston M, Ray D K,
greater temporal resolution, different methodological Ramankutty N and Foley J A 2012 Closing yield gaps
approaches could be useful as well. Future work could gen- through nutrient and water management Nature 490 2547
erate global production frontiers using process-based models [5] Sutton M A et al 2013 Our Nutrient World: the Challenge to
Produce More Food and Energy With Less Pollution
that may be more sensitive to ne-scale weather and soil (Edinburgh, UK: Centre for Ecology and Hydrology (CEH)
variations, and such an analysis would be a useful comple- on behalf of the Global Partnership on Nutrient Management
ment to our study. and the International Nitrogen Initiative)
[6] Diaz R J and Rosenberg R 2008 Spreading dead zones and
consequences for marine ecosystems Science 321 9269
4.5. Conclusions [7] Compton J E, Harrison J A, Dennis R L, Greaver T L,
Hill B H, Jordan S J, Walker H and Campbell H V 2011
This analysis quanties how substantial improvements are Ecosystem services altered by human changes in the
possible in crop production and reduction of both applied and nitrogen cycle: a new perspective for US decision making
excess nitrogen. Our results suggest global benet from Eco. Lett. 14 80415
reductions in nitrogen intensity in some regions and increases [8] Keeler B L, Polasky S, Brauman K A, Polasky S, Johnson K A,
Finlay J C, ONeill A, Kovacs K and Dalzell B 2012
in nitrogen intensity in others, along with wholesale increases Linking water quality and well-being for improved
in agronomic efciency across the entire system. While such assessment and valuation of ecosystem services Proc. Natl.
changes may be difcult to achieve, they would be important Acad. Sci. USA 109 1861924
steps towards meeting projected increases in food demand [9] Elser J J, Bracken M E S, Cleland E E, Gruner D S,
while improving environmental quality [48]. Harpole W S, Hillebrand H, Ngai J T, Seabloom E W,
Shurin J B and Smith J E 2007 Global analysis of nitrogen
Lastly, we suggest that the methodology adopted herea and phosphorus limitation of primary producers in
crop production and environmental outcome tradeoff frontier freshwater, marine and terrestrial ecosystems Ecol. Lett. 10
be utilized more widely to help determine production 113542

7
Environ. Res. Lett. 9 (2014) 054002 N D Mueller et al

[10] Davidson E A, Keller M, Erickson H E, Verchot L V and agricultural production systems for the period 19702030
Veldkamp E 2000 Testing a conceptual model of soil Pedosphere 15 13755
emissions of nitrous and nitric oxides Biosci. 50 66780 [30] MacDonald G K, Bennett E M, Potter P A and Ramankutty N
[11] Schlesinger W H 2009 On the fate of anthropogenic nitrogen 2011 Agronomic phosphorus imbalances across the world's
Proc. Natl. Acad. Sci. 106 2038 croplands Proc. Natl. Acad. Sci. USA 108 308691
[12] Pinder R W, Davidson E A, Goodale C L, Greaver T L, [31] Potter P A, Ramankutty N, Bennett E M and Donner S D 2010
Herrick J D and Liu L 2012 Climate change impacts of US Characterizing the spatial patterns of global fertilizer
reactive nitrogen Proc. Natl. Acad. Sci. 109 76715 application and manure production Earth Interact. 14 122
[13] Guo J H, Liu X J, Zhang Y, Shen J L, Han W X, Zhang W F, [32] Dentener F et al 2006 Nitrogen and sulfur deposition on
Christie P, Goulding K, Vitousek P M and Zhang F S 2010 regional and global scales: a multimodel evaluation Global
Signicant acidication in major chinese croplands Science Biogeochem. Cycles 20 GB4003
327 100810 [33] Sanchez P A and Swaminathan M S 2005 Hunger in Africa:
[14] Cassman K G, Dobermann A, Walters D T and Yang H 2003 the link between unhealthy people and unhealthy soils
Meeting cereal demand while protecting natural resources Lancet 365 4424
and improving environmental quality Annu. Rev. Environ. [34] Barrett C B 2010 Measuring food insecurity Science 327 8258
Resour. 28 31558 [35] Food and Agriculture Organization of the United Nations 2012
[15] Vitousek P M et al 2009 Agriculture. Nutrient imbalances in The state of food insecurity in the world (Rome: Food and
agricultural development Science 324 151920 Agriculture Organization of the United Nations)
[16] Liu J, You L, Amini M, Obersteiner M, Herrero M, [36] Philibert A, Loyce C and Makowski D 2012 Quantifying
Zehnder A J B and Yang H 2010 A high-resolution uncertainties in N2O emission due to N fertilizer application
assessment on global nitrogen ows in cropland Proc. Natl. in cultivated areas PLoS One 7 e50950
Acad. Sci. USA 107 803540 [37] Van Groenigen J W, Velthof G L, Oenema O,
[17] Ju X-T et al 2009 Reducing environmental risk by improving Van Groenigen K J and van Kessel C 2010 Towards an
N management in intensive chinese agricultural systems agronomic assessment of N2O emissions: a case study for
Proc. Natl. Acad. Sci. USA 106 30416 arable crops Eur. J. Soil Sci. 61 90313
[18] Zhang F, Chen X and Vitousek P 2013 Chinese agriculture: an [38] Easterling W E et al 2007 Food, bre and forest products
experiment for the world Nature 497 335 Climate Change 2007: Impacts, Adaptation and
[19] Van Grinsven H J M, Holland M, Jacobsen B H, Klimont Z, Vulnerability ed M Parry, O Canziani, J Palutikof,
Sutton M A and Jaap Willems W 2013 Costs and benets of P Van der Linden and C Hanson (Cambridge: Cambridge
nitrogen for Europe and implications for mitigation Environ. University Press) pp 273313
Sci. Technol. 47 35719 [39] IFA Task Force on Fertilizer Best Management Practices 2009
[20] Tittonell P and Giller K E 2012 When yield gaps are poverty The global 4R nutrient stewardship framework (Paris:
traps: the paradigm of ecological intensication in African International Fertilizer Association)
smallholder agriculture Field Crop Res. 143 7690 [40] Robertson G P, Bruulsema T W, Gehl R J, Kanter D,
[21] Nziguheba G et al 2010 The African green revolution: results Mauzerall D L, Rotz C A and Williams C O 2012
from the millennium villages project Advances in Agronomy Nitrogenclimate interactions in US agriculture
vol 109 (Amsterdam: Elsevier) Biogeochemistry 114 4170
[22] Sanchez P A 2002 Ecologysoil fertility and hunger in Africa [41] Zhang F, Cui Z, Fan M, Zhang W, Chen X and Jiang R 2011
Science 295 201920 Integrated soilcrop system management: reducing
[23] Twomlow S, Rohrbach D, Dimes J, Rusike J, Mupangwa W, environmental risk while increasing crop productivity and
Ncube B, Hove L, Moyo M, Mashingaidze N and improving nutrient use efciency in China J. Environ. Qual.
Mahposa P 2010 Micro-dosing as a pathway to Africas 40 1051
green revolution: evidence from broad-scale on-farm trials [42] Chen X-P et al 2011 Integrated soil-crop system management
Nutr. Cycl. Agroecosys. 88 315 for food security Proc. Natl. Acad. Sci. USA 108 6399404
[24] Polasky S, Nelson E J, Lonsdorf E, Fackler P and Stareld A [43] Conant R T, Berdanier A B and Grace P R 2013 Patterns and
2005 Conserving species in a working landscape: land use trends in nitrogen use and nitrogen recovery efciency in
with biological and economic objectives Ecol. Appl. 15 world agriculture Global Biogeochem. Cycles 27 55866
1387401 [44] Sanchez P A 2010 Tripling crop yields in tropical Africa Nat.
[25] Polasky S et al 2008 Where to put things? spatial land Geosci. 3 299300
management to sustain biodiversity and economic returns [45] Carberry P S, Liang W-L, Twomlow S, Holzworth D P,
Biol. Cons. 141 150524 Dimes J P, McClelland T, Huth N I, Chen F,
[26] Monfreda C, Ramankutty N and Foley J A 2008 Farming the Hochman Z and Keating B A 2013 Scope for improved eco-
planet: 2. Geographic distribution of crop areas, yields, efciency varies among diverse cropping systems Proc.
physiological types, and net primary production in the year Natl. Acad. Sci. USA 110 83816
2000 Global Biogeochem. Cycles 22 GB1022 [46] Cassman K G, Dobermann A and Walters D T 2002
[27] Portmann F T, Siebert S and Doell P 2010 MIRCA2000-global Agroecosystems, nitrogen-use efciency, and nitrogen
monthly irrigated and rainfed crop areas around the year management Ambio 31 13240
2000: a new high-resolution data set for agricultural and [47] Cassman K G, Gines G, Dizon M, Samson M and Alcantara J
hydrological modeling Global Biogeochem. Cycles 24 1996 Nitrogen-use efciency in tropical lowland rice
GB1011 systems: contributions from indigenous and applied nitrogen
[28] Hijmans R J, Cameron S E, Parra J L, Jones P G and Jarvis A Field Crops Res. 47 112
2005 Very high resolution interpolated climate surfaces for [48] Tilman D, Balzer C, Hill J and Befort B L 2011 Global food
global land areas Int. J. Climatol. 25 196578 demand and the sustainable intensication of agriculture
[29] Bouwman A F, Van Drecht G and Van der Hoek K W 2005 Proc. Natl. Acad. Sci. USA 108 202604
Global and regional surface nitrogen balances in intensive