Technical ReportPDF Available

World Resources Report: Creating a Sustainable Food Future


Abstract and Figures

By 2050, nearly 10 billion people will live on the planet. Can we produce enough food sustainably? The synthesis report of the World Resources Report: Creating a Sustainable Food Future shows that it is possible – but there is no silver bullet. This report offers a five-course menu of solutions to ensure we can feed everyone without increasing emissions, fueling deforestation or exacerbating poverty. Intensive research and modeling examining the nexus of the food system, economic development, and the environment show why each of the 22 items on the menu is important and quantifies how far each solution can get us. WRI produced the report in partnership with the World Bank Group, UN Environment, UN Development Programme, and the French agricultural research agencies CIRAD and INRA.
Content may be subject to copyright.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 i
A Menu of Solutions to Feed Nearly 10 Billion People by 2050
Tim Searchinger, Richard Waite, and Tim Beringer (Humboldt University at
Berlin) contributed to development of the GlobAgri-WRR model, as did a number
of researchers from the Centre de coopération internationale en recherche
agronomique pour le développement, and the Institut national de la recherche
agronomique, including Agneta Forslund, Hervé Guyomard, Chantal Le Mouël,
Stéphane Manceron, and Elodie Marajo-Petitzon.
Major GlobAgri-WRR model subcomponents include a livestock model with lead
developers Mario Herrero (Commonwealth Scientific and Industrial Research
Organisation) and Petr Havlík (IIASA), with additional contributions from Stefan
Wirsenius (Chalmers University of Technology); a land-use model with lead developer
Fabien Ramos (European Commission Joint Research Centre); a rice model with lead
developer Xiaoyuan Yan (Chinese Institute for Soil Science); a nitrogen emissions
model with lead developer Xin Zhang (Princeton University); and an aquaculture
model with lead developers Mike Phillips (WorldFish) and Rat tanawan Mungkung
(Kasetsar t University).
A number of individuals were coauthors on working papers that serve as the
foundation for the full repor t, this synthesis report, and many of the menu items
profiled therein. They include Tapan K. Adhya (KIIT University, India), Tamara Ben Ari
(INRA), Maryline Boval (INRA), Tim Beringer (Humboldt University at Berlin), Malcolm
Beveridge (WorldFish), Randall Brummet t (World Bank), Sarah Castine (WorldFish),
Philippe Chemineau (INRA), Nuttapon Chaiyawannakarn (Kasetsart Universit y), Ayesha
Dinshaw (WRI), Patrice Dumas (CIRAD), Dennis Garrity (World Agroforestry Centre),
Jerry Glover (U.S. Agency for International Development), Sarah Harper (Oxford Institute
of Population Ageing and University of Oxford), Ralph Heimlich (Agricultural Conservation
Economics), Debbie Hellums (International Fertilizer Development Center), Norbert
Henninger (WRI), Sadasivam Kaushik (INR A), Lisa Kitinoja (The Postharvest Education
Foundation), Jean-Marc Lacape (CIRAD), George Leeson (Oxford Institute of Population
Ageing and University of Oxford), Bruce Linquist (University of California at Davis), Brian
Lipinski (WRI), David Makowski (INRA), Mike McGahuey (U.S. Agency for International
Development), Ra ttanawan Mungkung (Kasetsart Univer sity), Supawat Nawapakpilai
(Kasetsart University), Michael Phillips (WorldFish), Chris Reij (WRI), Katie Reytar
(WRI), Sara Scherr (EcoAgiculture Partners), Daniel Vennard (WRI), Reiner Wassmann
(International Rice Research Institute, Philippines), Rober t Winterbottom (WRI), and
Xiaoyuan Yan (Institute for Soil Science, Chinese Academy of Sciences).
All unreferenced numbers are results from the
GlobAgri-WRR model.
All dollars are U.S. dollars unless otherwise indicated.
All tons are metric tons unless otherwise indicated.
All general references to greenhouse gas emissions
are in carbon dioxide equivalents using a 100-year
global warming potential unless otherwise indicated.
“Kcal” = kilocalorie, also referred to as simply
This synthesis report summarizes the findings of the World Resources Report Creating a Sustainable Food Future, a multiyear partnership
between World Resources Institute, the World Bank Group, United Nations Environment, the United Nations Development Programme,
the Centre de coopération internationale en recherche agronomique pour le développement, and the Institut national de la recherche
agronomique. The full report will be published in the spring of 2019. Previously published installments analyzing many of the issues covered
in this report in greater detail are available at
The report focuses on technical opportunities and policies for cost-eective scenarios for meeting food, land use, and greenhouse gas
emissions goals in 2050 in ways that can also help to alleviate poverty and do not exacerbate water challenges. It is primarily global in
focus. As with any report, it cannot address all issues related to the global food system, such as many ethical, cultural, and socioeconomic
factors or remedies for tackling acute food shortages in the short term. Future research may pursue quantitative estimates of agricultural
freshwater use.
Tim Searchinger (WRI and Princeton University)
Richard Waite (WRI)
Craig Hanson (WRI)
Janet Ranganathan (WRI)
Patrice Dumas (CIRAD)
Emily Matthews
1 Executive Summary
5 Scope of the Challenge and Menu of
Possible Solutions
13 Course 1: Reduce Growth in Demand for
Food and Other Agricultural Products
21 Course 2: Increase Food Production
Without Expanding Agricultural Land
31 Course 3: Protect and Restore Natural
Ecosystems and Limit Agricultural Land-
39 Course 4: Increase Fish Supply
43 Course 5: Reduce Greenhouse Gas
Emissions from Agricultural Production
55 The Complete Menu: Creating a
Sustainable Food Future
65 Cross-Cutting Policies for a Sustainable
Food Future
75 Conclusions
78 Endnotes
83 References
89 Acknowledgments
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 1
As the global population grows from 7 billion
in 2010 to a projected 9.8 billion in 2050, and
incomes grow across the developing world, overall
food demand is on course to increase by more than
50 percent, and demand for animal-based foods by
nearly 70 percent. Yet today, hundreds of millions
of people remain hungry, agriculture already uses
almost half of the world’s vegetated land, and agri-
culture and related land-use change generate one-
quarter of annual greenhouse gas (GHG) emissions.
This synthesis report proposes a menu of options
that could allow the world to achieve a sustainable
food future by meeting growing demands for food,
avoiding deforestation, and reforesting or restoring
abandoned and unproductive land—and in ways
that help stabilize the climate, promote economic
development, and reduce poverty.
Achieving these goals requires closing three great
“gaps” by 2050:
The food gap—the dierence between the
amount of food produced in 2010 and the
amount necessary to meet likely demand in
2050. We estimate this gap to be 7,400 trillion
calories, or 56 percent more crop calories than
were produced in 2010.
The land gap—the dierence between global
agricultural land area in 2010 and the area
required in 2050 even if crop and pasture yields
continue to grow at past rates. We estimate this
gap to be 593 million hectares (Mha), an area
nearly twice the size of India.
The GHG mitigation gap—the dierence
between the annual GHG emissions likely from
agriculture and land-use change in 2050, which
we estimate to be 15 gigatons of carbon dioxide
equivalent (Gt CO2e), and a target of 4 Gt that
represents agriculture’s proportional contribu-
tion to holding global warming below 2°C above
pre-industrial temperatures. We therefore
estimate this gap to be 11 Gt. Holding warming
below a 1.5°C increase would require meeting
the 4 Gt target plus reforesting hundreds of mil-
lions of hectares of liberated agricultural land.
This report explores a 22-item “menu for a sus-
tainable food future,” which is divided into ve
“courses” that together could close these gaps: (1)
reduce growth in demand for food and agricultural
products; (2) increase food production without
expanding agricultural land; (3) exploit reduced
demand on agricultural land to protect and restore
forests, savannas, and peatlands; (4) increase sh
supply through improved wild sheries manage-
ment and aquaculture; and (5) reduce greenhouse
gas emissions from agricultural production.
On the one hand, the challenge of simultaneously
closing these three gaps is harder than often recog-
nized. Some prior analyses overestimate potential
crop yield growth, underestimate or even ignore the
challenge of pastureland expansion, and “double
count” land by assuming that land is available for
reforestation or bioenergy without accounting for
the world’s growing need to produce more food,
protect biodiversity, and maintain existing carbon
storage. Signicant progress in all 22 menu items is
necessary to close the three gaps, requiring action
by many millions of farmers, businesses, consum-
ers, and all governments.
On the other hand, the scope of potential solutions
is often underestimated. Prior analyses have gener-
ally not focused on the promising opportunities for
technological innovation and have often underes-
timated the large social, economic, and environ-
mental cobenets. Our menu is detailed but several
themes stand out:
Raise productivity. Increased eciency of
natural resource use is the single most impor-
tant step toward meeting both food production
and environmental goals. This means increas-
ing crop yields at higher than historical (linear)
rates, and dramatically increasing output of
milk and meat per hectare of pasture, per ani-
mal—particularly cattle—and per kilogram of
fertilizer. If today’s levels of production ecien-
cy were to remain constant through 2050, then
feeding the planet would entail clearing most of
the world’s remaining forests, wiping out thou-
sands more species, and releasing enough GHG
emissions to exceed the 1.5°C and 2°C warming
targets enshrined in the Paris Agreement—even
if emissions from all other human activities
were entirely eliminated.
Manage demand. Closing the food gap will
be far more dicult if we cannot slow the rate
of growth in demand. Slowing demand growth
requires reducing food loss and waste, shifting
the diets of high meat consumers toward plant-
based foods, avoiding any further expansion
of biofuel production, and improving women’s
access to education and healthcare in Africa
to accelerate voluntary reductions in fertility
Link agricultural intensication
with natural ecosystems protection.
Agricultural land area is not merely expanding
but shifting from one region to another
(e.g., from temperate areas to the tropics)
and within regions. The resulting land-use
changes increase GHG emissions and loss of
biodiversity. To ensure that food production is
increased through yield growth (intensication)
and not expansion, and productivity gains do
not encourage more shifting, governments must
explicitly link eorts to boost crop and pasture
yields with legal protection of forests, savannas,
and peatlands from conversion to agriculture.
Moderate ruminant meat consumption.
Ruminant livestock (cattle, sheep, and goats)
use two-thirds of global agricultural land and
contribute roughly half of agriculture’s produc-
tion-related emissions. Ruminant meat demand
is projected to grow by 88 percent between
2010 and 2050. Yet, even in the United States,
ruminant meats (mostly beef) provide only 3
percent of calories. Closing the land and GHG
mitigation gaps requires that, by 2050, the 20
percent of the world’s population who would
otherwise be high ruminant-meat consumers
reduce their average consumption by 40 per-
cent relative to their consumption in 2010.
Target reforestation and peatland
restoration. Rewetting lightly farmed,
drained peatlands that occupy only around 0.3
percent of global agricultural lands provides
a necessary and cost-eective step toward
climate change mitigation, as does reforesting
some marginal and hard-to-improve grazing
land. Reforestation at a scale necessary to hold
temperature rise below 1.5 degrees Celsius (i.e.,
hundreds of millions of hectares) is potentially
achievable but only if the world succeeds in
reducing projected growth in demand for
resource-intensive agricultural products and
boosting crop and livestock yields.
Require production-related climate
mitigation. Management measures exist to
signicantly reduce GHG emissions from agri-
cultural production sources, particularly enteric
fermentation by ruminants, manure, nitrogen
fertilizers, and energy use. These measures
require a variety of incentives and regulations,
deployed at scale. Implementation will require
far more detailed analysis and tracking of agri-
cultural production systems within countries.
Spur technological innovation. Fully clos-
ing the gaps requires many innovations. For-
tunately, researchers have demonstrated good
potential in every necessary area. Opportunities
include crop traits or additives that reduce meth-
ane emissions from rice and cattle, improved
fertilizer forms and crop properties that reduce
nitrogen runo, solar-based processes for mak-
ing fertilizers, organic sprays that preserve fresh
food for longer periods, and plant-based beef
substitutes. A revolution in molecular biology
opens up new opportunities for crop breeding.
Progress at the necessary scale requires large
increases in R&D funding, and exible regula-
tions that encourage private industry to develop
and market new technologies.
Using a new model called GlobAgri-WRR, we
estimate how three scenarios we call “Coordinated
Eort,” “Highly Ambitious,” and “Breakthrough
Technologies” can narrow and ultimately fully close
our three gaps. Figure ES-1 illustrates how our ve
courses of action could feed the world and hold
down global temperature rise. Although a formi-
dable challenge, a sustainable food future is achiev-
able if governments, the private sector, and civil
society act quickly, creatively, and with conviction.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 3
Figure ES-1 | Ambitious eorts across all menu items will be necessary to feed 10 billion people while keeping
global temperature rise well below 2 degrees Celsius
Crop production (trillion calories per year)
Agricultural GHGs (production + land-use change), Gt CO2e/year
2050 (Baseline)2010 (Base year) Reduce growth in demand
for food and other
agricultural products
Increase food production
without expanding
agricultural land
emissions were
12 Gt/yr in
…but emissions
triple by 2050
historical rates
of productivity
…to 15 Gt/yr by
2050 (our
Slowing and
shifting growth
in food demand
…as do
emissions from
cattle, fertilizers,
rice, and on-
farm energy use
trims emissions
forests and
peatlands could
oset remaining
…to achieve 4
Gt/yr (2° C
target) or even
0 Gt/yr (1.5° C
gains reduces
Boosting fish
supply reduces
slightly (but is
important for
Note: These charts show the mos t ambitious “Break through Technologies ” scenario. “Res tore forests and pe atlands” item includes full ref orestation of at least 80 million hectares
of liberated agricult ural land, in order to reach t he 4 Gt CO2e/year target by 2050 for limiting global temperature rise to 2°C. As an even more ambitious option, in order to limit
warming to 1. 5°C, full refore station of at leas t 585 million hectares of libera ted agricultural land could oset global agricultural production emissions for many years .
Source: GlobAgri-WRR model.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 5
This World Resources Report addresses a fundamental
question: How can the world adequately feed nearly 10
billion people by the year 2050 in ways that help combat
poverty, allow the world to meet climate goals, and reduce
pressures on the broader environment?
A Recipe for Change
The challenge of creating a sustainable
food future involves balancing many
competing needs. By 2050, the world
must feed many more people, more
nutritiously, and ensure that agriculture
contributes to poverty reduction through
inclusive economic and social develop-
ment, all while reducing greenhouse
gas (GHG) emissions, loss of habitat,
freshwater depletion and pollution, and
other environmental impacts of farming.
Pursuing any one of these goals to the
exclusion of the others will likely result
in failure to achieve any of them.
We quantify the core of the challenge in
terms of the need to close three “gaps”:
in food production, agricultural land
area, and greenhouse gas (GHG) mitiga-
tion. To measure the size of these gaps,
we use a new model, GlobAgri-WRR,
developed in a partnership between Le
Centre de coopération internationale en
recherche agronomique pour le dével-
oppement (CIRAD), L’Institut national
de la recherche agronomique (INRA),
World Resources Institute (WRI), and
Princeton University (Box 1).
This global accounting and biophysical model quantifies food
production and consumption from national diets and populations,
as well as land-use demands. The model also estimates GHG
emissions from agriculture, including emissions from production
(primarily methane and nitrous oxide), carbon dioxide emissions
from the energy used to produce fertilizers and pesticides or
to run farm machinery, and emissions from land-use change.
Emissions modeled include everything up to the farm gate but do
not include those from food processing, transportation, retail, or
cooking. GlobAgri-WRR is designed to estimate land use and GHG
emissions with specified levels of population, diets and other
crop demands, specific trade patterns, and specified agricultural
production systems for crops and livestock in dierent countries.
The model by itself does not attempt to analyze what policies
and practices will achieve those systems; that is the focus of this
synthesis report and the full report. For this reason, GlobAgri-
WRR does not attempt to analyze economic feedback eects
but concentrates on more biophysical detail. A strength of the
GlobAgri-WRR model is that it incorporates other biophysical
submodels that estimate GHG emissions or land-use demands in
specific agricultural sectors, benefitting from the detail available
from other researchers’ work.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 7
The Food Gap
The food gap is the increase above the amount of
food (measured as crop calories)1 produced in 2010,
the base year for our analysis, to the amount that
the world will require in 2050, based on projected
demand (Figure 1). Rising food demand over this
period—leading to this 56 percent food gap—will be
driven by population growth (from 7 billion to 9.8
billion people)2 and by increasing demand for more
resource-intensive foods, particularly animal-based
foods, as incomes grow.3 Consumption of milk and
meat—foods that rely heavily on pasture for their
production—is likely to grow by 68 percent. These
rates of growth exceed those that prevailed from
1962 to 2010.
The food gap can be closed both through measures
that decrease the rate of unnecessary demand
growth and measures that increase supply. The
more the gap can be closed through demand-
reduction measures, the smaller will be the
challenge of increasing food production. And as
that challenge decreases, so does the risk that the
world will fail to meet food needs, which would
most harshly aect the poor.
Frequent claims that the world already has an
overabundance of food and could meet future
needs without producing more food4 are based on
an unrealistic, even if desirable, hypothetical. It
presumes that the world not only consumes fewer
animal products per person, as this report encour-
ages, but by 2050 eliminates nearly all meat con-
sumption; that people shift from meat to vegetables
and legumes and consume the same high-yield
crops now used for animal feed; that all food loss
and waste is eliminated; and that food is distributed
just enough and no more than to meet nutritional
needs of every person in the world.
Figure 1 | The world needs to close a food gap of 56 percent by 2050
Note: Include s all crops intended for direct human consumption, animal feed , industrial uses, s eeds, and biofuels.
Source: WRI analysis based on FAO (2017a); UNDE SA (2017); and Alexandratos and Bruinsma (2012).
2050 (Baseline)2010 (Base year)
Crop production (trillion calories per year)
The Land Gap
One strategy to close the food gap could be to clear
more land for agriculture—but at the cost of great
harm to forests and other ecosystems and the
people who depend on them, and large releases of
stored carbon from vegetation and soils. Today,
croplands and pasture occupy roughly half of all
vegetated land.5 Between 1962 and 2010 alone,
almost 500 million hectares (Mha) of forests and
woody savannas were cleared for agriculture.6 More
land clearing would exacerbate a biodiversity crisis
driven heavily by land-use change. And virtually all
strategies for stabilizing the climate assume no net
releases of carbon from land clearing between now
and 2050, while many require net reforestation.
Our target is to hold agricultural land area—crop-
land and pastureland—to the area used in 2010.
The land gap is thus the dierence between the
projected area of land needed to meet global food
demand in 2050 and the amount of land in agricul-
tural use in 2010.
The size of the land gap depends on how quickly
crop and livestock yields can be improved. If the
world were to experience no gains in crop and pas-
ture yields and no moderation in demand for food
(what we call our “no productivity gains after 2010”
scenario), agricultural land would expand by 3.3
billion hectares, virtually eliminating the world’s
forests and savannas. In our baseline projection, we
use estimated yields from the Food and Agriculture
Organization of the United Nations (FAO), which
projects that crop yields will increase, on average, at
roughly the same rate as they did between 1961 and
2010. Livestock and pasture productivity gains are
from the GlobAgri-WRR model. These gains hold
down the expansion of agricultural areas to 593
Mha (Figure 2). However, if future crop yields grow
at the somewhat slower rates experienced more
recently (1989–2008), and pasture and livestock
productivity also grow more slowly than in our
baseline scenario, agricultural areas could instead
expand by 855 Mha by 2050.
Future yield growth is uncertain, but the key lesson
is that the world faces an unprecedented challenge.
Crop and pasture yields must increase at rates
even faster than those achieved between 1961
and 2010—a period that included the widespread
synthetic fertilizer and scientically bred seeds and
a doubling of irrigated area—to fully meet expected
food demand and to avoid massive additional
clearing of forests and woody savannas.
2050 (Baseline) 2050 (Target)
Agricultural land expansion, 2010–50 (Mha)
593 Mha
0 Mha
192 Mha
401 Mha
Figure 2 | The world needs to close a land gap of 593 million hectares to avoid further agricultural expansion
Note: “Cropland” increase includes aquaculture ponds.
Source: GlobAgri-WRR model.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 9
Figure 3 | Agricultural emissions are likely to be ~70 percent of total allowable emissions for all sectors by
2050, creating an 11 gigaton mitigation gap
The Greenhouse Gas Mitigation Gap
The GHG mitigation gap is the dierence between
agriculture-related GHG emissions projected for
2050 and an agricultural emissions target for 2050
that is necessary to help stabilize the climate at
globally agreed targets.7
Agriculture and land-use change contributed one-
quarter of total human-caused GHG emissions in
2010—roughly 12 gigatons (Gt) measured as carbon
dioxide equivalent (CO2e).8
Of this total, a little
more than half resulted from agricultural produc-
tion, including such sources as methane from
livestock production and rice cultivation, nitrous
oxide from nitrogen fertilizer, and carbon dioxide
released by fossil fuels used in agricultural produc-
tion.9 A little less than half of the emissions resulted
from land-use change (vegetation clearing and soil
plowing) as agriculture expanded. The land-use
category includes 1.1 Gt released annually by the
ongoing degradation of cleared peatlands, which
are carbon-rich soils that decompose and some-
times catch re once drained for agriculture.10
Using the GlobAgri-WRR model, we project total agri-
cultural GHGs to be roughly 15 Gt per year in 2050–
9 Gt of annual emissions from agricultural production
and an annual average of 6 Gt between 2010 and
2050 from agricultural expansion and drained peat-
lands.11 What are the implications of this estimate?
Modeled strategies for holding climate warming to
the global target of 2 degrees Celsius (2°C)
(3.6 degrees Fahrenheit) above preindustrial levels
typically require that total emissions from all
human sources in 2050 amount to no more than
around 21 Gt and decrease rapidly thereafter.12
Although agriculture is likely to generate less than
2 percent of global GDP, it alone would ll about 70
percent of the allowable “emissions budget” in 2050
(15 of 21 Gt), leaving almost no space for emissions
from other economic sectors and making achieve-
ment of even the 2°C target impossible (Figure 3).
Reecting this dilemma, we dene a GHG mitiga-
tion gap of 11 Gt: the dierence between the 15 Gt
of likely annual emissions in 2050 and a target of 4
Gt. The gap represents a nearly 75 percent reduction
from the projected level—a reduction in line with the
principle of “equal sharing” required from all sources
to keep global warming to well below 2°C.
To limit warming to 1.5°C (2.7 degrees Fahrenheit),
typical scenarios contemplate similar levels of
emissions from agricultural production but require
extensive reforestation to oset other emissions.
We therefore also explore options for liberating
agricultural land to provide such osets.
2010 (Base year) 2050 (Baseline) 2050 (Target)
12 15 4
Agricultural and land-use
change emissions
Nonagricultural emissions
Gt CO2e/year
Agricultural emissions
in 2050 baseline equal 70%
of total allowable emissions
for all sectors in 2050
Sources: GlobAgri-WRR model, WRI anal ysis based on IEA (20 12); EIA (2012); Houghton (2 008); OECD (2012); and UNEP (20 13).
A Menu of Solutions
To close these three gaps, we develop a “menu for
a sustainable food future”—a menu of actions that
can meet these challenges if implemented in time,
at scale, and with sucient public and private sec-
tor dedication (Table 1). We analyze the potential
of the menu items to sustainably close the food,
land, and GHG mitigation gaps by 2050. They are
organized into ve “courses”:
1. Reduce growth in demand for food and other
agricultural products
2. Increase food production without expanding
agricultural land
3. Protect and restore natural ecosystems and
limit agricultural land-shifting
4. Increase sh supply
5. Reduce GHG emissions from agricultural
A dominant theme of all menu items is the need to
increase the eciency in use of resources, whether
through changes in consumption patterns or uses of
land, animals, and other agricultural inputs.
In addition to helping close the three gaps, we
impose three additional sustainability criteria on
the menu items:
To reduce poverty and hunger, the menu must
neither inate food prices signicantly nor deny
agricultural opportunities for small and poor
farmers, even as they transition to alternative
employment as economies develop.
Because women’s gains in income dispropor-
tionately reduce hunger for the entire house-
hold, the menu must provide opportunities for
women farmers, who contribute the majority of
agricultural labor in many countries and whose
productivity has been hampered by unequal ac-
cess to resources.
To avoid further overuse and pollution of fresh
water, the menu must contribute to pollution
control, avoid increases in large-scale irriga-
tion, and conserve or make more ecient use of
water wherever possible. Agriculture accounts
for roughly 70 percent of global fresh water
withdrawals and is the primary source of nutri-
ent runo from farm elds.13
Table 1 | The menu for a sustainable food future: five courses
Course 1: Reduce growth in demand for food and other agricultural products
Reduce food loss and waste Reduce the loss and waste of food intended for human consumption between the farm and the
Shift to healthier and more sustainable
Change diets particularly by reducing ruminant meat consumption to reduce the three gaps in
ways that contribute to better nutrition.
Avoid competition from bioenergy for
food crops and land
Avoid the diversion of both edible crops and land into bioenergy production.
Achieve replacement-level fertility rates Encourage voluntary reductions in fertility levels by educating girls, reducing child mortality, and
providing access to reproductive health services.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 11
Course 2: Increase food production without expanding agricultural land
Increase livestock and pasture
Increase yields of meat and milk per hectare and per animal through improved feed quality, grazing
management, and related practices.
Improve crop breeding to boost yields Accelerate crop yield improvements through improved breeding.
Improve soil and water management Boost yields on drylands through improved soil and water management practices such as
agroforestry and water harvesting.
Plant existing cropland more frequently Boost crop production by getting more than one crop harvest per year from existing croplands or
by leaving cropland fallow less of ten where conditions are suitable.
Adapt to climate change Employ all menu items and additional targeted interventions to avoid adverse eects of climate
change on crop yields and farming viability.
Course 3: Protect and restore natural ecosystems and limit agricultural land-shifting
Link productivity gains with protection of
natural ecosystems
Protect ecosystems by legally and programmatically linking productivity gains in agriculture to
governance that avoids agricultural expansion.
Limit inevitable cropland expansion
to lands with low environmental
opportunity costs
Where expansion seems inevitable—such as for local food production in Africa—limit expansion to
lands with the lowest carbon and other environmental costs per ton of crop.
Reforest abandoned, unproductive, and
liberated agricultural lands
Protect the world’s remaining native landscapes; reforest abandoned, unproductive, and
unimprovable agricultural lands as well as lands potentially “liberated” by highly successful
reductions in food demand or increases in agricultural productivity.
Conserve and restore peatlands Avoid any further conversion of peatlands into agriculture and restore little-used, drained
peatlands by rewetting them.
Course 4: Increase fish supply
Improve wild fisheries management Stabilize the annual size of the wild fish catch over the long term by reducing overfishing.
Improve productivity and environmental
performance of aquaculture
Increase aquaculture production through improvements in breeding, feeds, disease control, and
changes in production systems.
Course 5: Reduce greenhouse gas emissions from agricultural production
Reduce enteric fermentation through
new technologies
Develop and deploy feed additives to reduce methane releases from ruminant animals.
Reduce emissions through improved
manure management
Use and advance dierent technologies to reduce emissions from the management of manure in
concentrated animal production systems.
Reduce emissions from manure left on
Develop and deploy nitrification inhibitors (spread on pastures and/or fed to animals) or through
breeding biological nitrogen inhibition traits into pasture grasses.
Reduce emissions from fertilizers by
increasing nitrogen use eiciency
Reduce overapplication of fertilizer and increase plant absorption of fertilizer through management
changes and changes in fertilizer compounds, or breeding biological nitrification inhibition into
Adopt emissions-reducing rice
management and varieties
Reduce methane emissions from rice paddies via variety selection and improved water and straw
Increase agricultural energy eiciency
and shift to nonfossil energy sources
Reduce energy-generated emissions by increasing eiciency measures and shifting energy
sources to solar and wind.
Focus on realistic options to sequester
carbon in soils
Concentrate eorts to sequester carbon in agricultural soils on practices that have the primary
benefit of higher crop and/or pasture productivity and do not sacrifice carbon storage elsewhere.
Table 1 | The menu for a sustainable food future: five courses (continued)
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 13
The size of the food challenge—and the associated environmental
and economic challenges—depends on the scale of the increase
in demand for crops and animal-based foods by midcentury. The
food, land, and GHG mitigation gaps are derived from reasonable
estimates of business-as-usual growth in demand for crops
and livestock. Yet such levels of growth are not inevitable.
Course 1 menu items explore ways to reduce this projected
growth in socially and economically beneficial ways.
MENU ITEM: Reduce Food Loss and Waste
Of all the food produced in the world each year,
approximately one-third by weight and one-quarter
by calories is lost or wasted at various stages between
the farm and the fork (Figure 4).14 Globally, food loss
and waste results in nearly $1 trillion in economic
losses,15 contributes to food insecurity in some devel-
oping countries, squanders agricultural land and
water resources, and generates roughly one-quarter
of all agricultural GHG emissions.16
Reducing food loss and waste in developed countries
relies heavily on subtle “nudges” to change consumer
behavior, such as eliminating the use of trays in
cafeterias or streamlining product date labels. Many
retail operations can reduce waste through improved
inventory management and purchasing agreements
that allow suppliers to plan better. Such strategies
enabled the United Kingdom to reduce retail and
consumer food waste by 21 percent between 2007
and 2012 (and overall food loss and waste by 14
percent).17 In developing countries, better harvest-
ing equipment can reduce losses, as can agricultural
practices that ripen crops for harvesting at more
consistent times or produce food with more consis-
tent qualities. Low-technology systems also exist to
improve storage, including evaporative coolers and
specially designed, low-cost plastic storage bags.
Despite these opportunities, large reductions
globally are challenging because food loss and
waste arises at so many dierent stages in the food
chain, each one contributing only a small fraction
of the whole. The complexity of food loss and waste
sources leads us to propose three basic strategies:
Target. Governments and companies should
adopt food loss and waste reduction targets
aligned with Sustainable Development Goal
Target 12.3, which calls for reducing food loss
and waste by 50 percent by 2030.
Measure. Major actors in the food supply
chain should more carefully measure sources of
food loss and waste to identify hotspots, devise
actions to reduce them, and assess progress.
Act and Innovate. Many food producers,
processors, and vast numbers of consumers will
need to take a variety of actions. Many tech-
nological innovations will be needed, such as
new methods that slow food degradation even
without refrigeration and improved handling
equipment that reduces damage.
Reducing food loss and waste by 25 percent globally
would reduce the food calorie gap by 12 percent, the
land use gap by 27 percent, and the GHG mitigation
gap by 15 percent.
North America
and Oceania
Europe North Africa,
West and
Central Asia
Latin America South and
Southeast Asia
 
Share of total food available that is lost or wasted
25% 22% 19% 15% 17% 23%
and Market
and Storage
Percentage of calories lost and wasted
Figure 4 | Food loss and waste primarily occurs closer to the consumer in developed regions and closer to the
farmer in developing regions
Source: WRI analysis based on FAO (2011b).
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 15
MENU ITEM: Shift to Healthier and
More Sustainable Diets
We project consumption of animal-based foods to
rise 68 percent between 2010 and 2050, with an
88 percent increase in consumption of ruminant
meat (meat from cattle, sheep, and goats). These
trends are a major driver of the food, land, and
GHG mitigation gaps. For every food calorie gener-
ated, animal-based foods—and ruminant meats in
particular—require many times more feed and land
inputs, and emit far more greenhouse gases, than
plant-based foods (Figure 5).
As nations urbanize and incomes rise above poverty
levels, diets tend to become more varied and “West-
ern”—high in sugar, fats, rened carbohydrates,
meat, and dairy. Although modest consumption of
meat and dairy by the world’s poor supplies critical
micronutrients, the large global rise in consump-
tion of animal-based foods is both unnecessary and
unhealthy. Half of the world’s population already
consumes 50 percent more protein than needed18
and, contrary to popular understanding, plant
proteins can readily meet protein requirements in
balanced diets that contain enough calories.19 New
research downplays health risks from cholesterol
and other saturated fats but has now identied
processed meats as carcinogenic and red meat as
probably carcinogenic.20
Researchers have long presented the environmental
case for shifting high-meat diets toward plant-
based foods, but achieving large global benets is
harder than often suggested, for two reasons. First,
a common assumption is that, if people reduce
meat consumption, they will instead consume much
of the food formerly fed to animals (feed grains and
oilseeds).21 However, in practice, people often shift
from meat to dairy products, legumes, and vegeta-
bles.22 As shown in Figure 5, the land use and GHG
emissions impacts of dairy products actually match
or exceed those of pork and chicken and, while
beans and vegetables are more environmentally e-
cient than meat, they are not as ecient as animal
feeds. Second, a 10 percent global cut in consump-
tion of all animal-based foods relative to the 2050
baseline, achieved by reducing consumption in
wealthy regions, would be necessary just to allow
6 billion people across Asia and Africa to consume
even half of Europe’s present consumption of such
foods while staying within total consumption levels
estimated in our baseline projection.23
Despite these cautions, by properly factoring in
the consequences of diets on land use we nd the
potential of shifting diets to be even more conse-
quential for GHG mitigation than commonly esti-
mated. In a world where population and demand
for food are growing, and yield gains are not keep-
ing pace, agricultural land is expanding. Each per-
son’s diet requires additional land-use change equal
to the total land area needed to produce that diet,
requiring conversion of forests and woody savannas
to croplands and pasture. The eects on carbon are
typically ignored. By counting the carbon dioxide
released by that land-use change, and amortizing
that amount over 20 years, we estimate that the
average U.S. diet causes emissions of nearly 17 tons
of CO2e per year—an amount on par with per capita
emissions from energy use in the United States.24
Beef accounts for roughly half of land use and
emissions associated with U.S. diets, but it provides
just 3 percent of the calories. Major environmental
benets would therefore result simply from shift-
ing from beef toward chicken or pork (Figure 5).
If global consumers shifted 30 percent of their
expected consumption of ruminant meat in 2050
to plant-based proteins, the shift would, by itself,
close half the GHG mitigation gap and nearly all
of the land gap. Such a shift would require roughly
2 billion people in countries that today eat high
amounts of ruminant meats to reduce their con-
sumption, on average, by 40 percent below 2010
levels to 1.5 servings per person per week—equiva-
lent to 2010 consumption levels in the Middle East
and North Africa (Figure 6). In China, the challenge
would be to moderate the growth of ruminant meat
consumption. The substantial shifts from beef
toward chicken that have already occurred in U.S.
and European diets since the 1970s suggest that
such shifts are feasible.25 This shift would still allow
global consumption of ruminant meats to grow by
one-third (instead of the 88 percent growth in the
baseline scenario) between 2010 and 2050.
Note: Data pr esented are global means. Indicators for animal-based f oods include resourc e use to produce feed , including pasture. Tons of harvested produc ts were converted to
quantities of calories and prote in using the global average e dible calorie and prote in contents of food types as reported in FAO (2017a). “Fish” includes all aquatic animal-based
foods. Estimates are based on a marginal analys is of additional agricultural land us e and emissions per additional million c alories consumed. Based on the approach take n by the
European Union for estimating emis sions from land-use change for biofuels, land-use-change impac ts are amortized over a period of 20 years and th en shown as annual impacts.
Estima tes of land use and greenhouse gas emiss ions for beef produc tion are based on dedic ated beef produc tion, not beef that is a coproduc t of dairy. Dairy figures are lower in
GlobAgri-WRR than some other models because G lobAgri-WRR assumes that bee f produced by dairy systems disp laces beef produced by dedica ted beef-produc tion systems.
Source: GlobAgri-WRR model.
Figure 5 | Animal-based foods are more resource-intensive than plant-based foods
Sugar Palm
Maize Soybean
Wheat Fruits
Pulses Pork Eggs Fish
Poultry Dairy Sheep
Land-use change
Agricultural production
Land use (ha) per million
calories consumed (2010)
GHG emissions (t CO
per million
calories consumed (2010)
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 17
Source: GlobAgri-WRR model, wit h source data from FAO (2017a); UNDESA (2017); FAO (2011b); and Alexandratos and Bruinsma (2012).
Figure 6 | Limiting ruminant meat consumption to 52 calories per person per day in all regions reduces the
greenhouse gas mitigation gap by half and nearly closes the land gap
Three strategies will be necessary to shift consump-
tion toward healthier and lower-impact diets:
Product innovation. Businesses should con-
tinue to increase investment in development of
meat substitutes (e.g., plant-based meats) and
blended meat-plant products until they satisfy
consumers who still want to enjoy the taste and
experience of eating meat at less cost.
Promotion and marketing. Businesses,
government, and civil society need to move
beyond relying solely on information and
education campaigns to shift diets. Rather,
they should improve marketing of plant-based
foods and plant-rich dishes. A suite of more
0 1 2 3 4 5 6 7 8 9 10
2050 population (billions)
2010 consumption (base year)
2050 additional consumption (baseline)
Threshold to equitably reduce global
consumption by 30% relative to baseline
2050 reduction in consumption (baseline)
(excl. China
and India) Sub-Saharan Africa
Middle East
North Africa
European Union
Former Soviet Union
U.S. and Canada
OECD (other)
Latin America (excl. Brazil)
sophisticated behavior-change strategies,
including minimizing disruption to consumers,
selling a compelling benet, maximizing
awareness, and evolving social norms, has
proven successful in shifting consumption
patterns in other food and nonfood products.26
Policy and pricing. Governments can sup-
port diet shifts through their own food pro-
curement practices and policies that shape the
consumption environment (e.g., marketing,
display). Once the quality and price of nonmeat
alternatives rival that of meat, retail-level taxes
on meats or other animal-based foods might
become politically acceptable.
MENU ITEM: Avoid Competition from
Bioenergy for Food Crops and Land
Bioenergy is produced mainly from food and energy
crops grown on dedicated land, which increases
global competition for land and widens the food,
land, and GHG mitigation gaps. Our 2050 base-
line projection assumes that the share of biofuels
from crops in transportation fuel remains at 2010
levels, but many governments have adopted goals
to increase biofuel’s share fourfold or more.27 Such
an increase globally would supply about 2 percent
of total energy use in 2050 but would increase the
food gap from 56 to 78 percent. Still more ambi-
tious goals—to supply 20 percent of world energy
from bioenergy by 2050—would require a quantity
of biomass equal to all the world’s harvested crops,
crop residues, forage, and wood in 2000 (Figure 7).
Bioenergy creates so much potential competition
for food and carbon storage because bioenergy con-
verts only a fraction of 1 percent of energy from the
sun into usable energy. Food or energy crops also
require well-watered, productive land. By contrast,
solar photovoltaic (PV) cells today can use drylands
and they produce at least 100 times more useable
energy per hectare than energy crops are likely to
produce in the future, even when grown on well-
watered lands.28
Burning (and rening) biomass also emits more
carbon per unit of energy generated than burn-
ing fossil fuels. Claims that bioenergy reduces
GHG emissions rely on the assumption that this
carbon does not “count” because burning plants
only returns carbon to the air that growing plants
absorb. But diverting land to produce bioenergy
comes at the cost of not using this land and the
plants it grows for other purposes, including food
production and carbon storage. To provide bio-
energy without losing these other services, people
must either grow additional plants or use organic
waste as a feedstock. Some low-carbon bioenergy
is available from wastes and possibly from winter
cover crops. But claims of large bioenergy potential
to reduce GHG emissions ignore the alternative
uses of land and plants, in eect assuming they can
continue to serve other needs even when dedicated
to bioenergy.
Avoiding increased use of bioenergy from energy
and food crops is critical to a sustainable food
future. Phasing out existing levels of biofuel use
would reduce the crop calorie gap from 56 to 49
percent. Governments should phase out subsidies
currently in place for bioenergy that is grown on
dedicated land. Governments also need to correct
“awed accounting” in renewable energy directives
and emissions trading laws that treat bioenergy as
Note: Assum es primary to final en ergy conversion for biomass is 24% lower than for fos sil energy.
Source: Authors’ calculations based on Haberl et al. (200 7); IEA (2017); and JRC (2011).
Figure 7 | If the world’s entire harvest of crops, crop residues, grasses, and wood in 2000 were used for
bioenergy, it would provide only 20 percent of energy needs in 2050
IN 2050
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 19
Sources: UNDESA (2017 ); Harper (2012); World Bank (2 017a).
Figure 8 | Sub-Saharan Africa has the world’s lowest performance in key indicators of total fertility rate,
women’s education, and child mortality
MENU ITEM: Achieve Replacement-Level
Fertility Rates
Expected population growth of 2.8 billion people
between 2010 and 205029 drives the majority of
expected growth in food demand. Roughly half
of this population increase will occur in Africa,
and one-third will occur in Asia. Overall, most of
the world—including Asia—is close to achieving
replacement-level fertility (~2.1 children per woman)
and will achieve or even dip below it by 2050.30
Sub-Saharan Africa is the notable exception, with a
total fertility rate above 5 in 2010–15 and a pro-
jected rate of 3.2 in 2050. As a result, sub-Saharan
Africa’s population, which was 880 million in
2010, is projected to reach 2.2 billion by 2050 and
4 billion by 2100.31 This population growth risks
exacerbating food insecurity in a region that is
already home to 30 percent of the world’s chroni-
cally hungry people.32
Given the choice, people worldwide have voluntarily
chosen to greatly reduce their fertility rates—even
in extremely poor countries and across religions
and cultures—wherever countries have achieved
three forms of social progress:
Increased educational opportunities for
girls, ensuring they get at least a lower second-
ary education (i.e., some high school). The lon-
ger girls stay in school, the later they typically
start bearing children and the fewer children
they bear.
Increased access to reproductive health
services, including family planning, to ensure
couples can have the family sizes they desire
and reduce maternal mortality.
Reduce infant and child mortality, so
parents do not need to have many children to
ensure survival of the desired number.
Reducing fertility also tends to produce strong
economic dividends. Unfortunately, sub-Saharan
Africa lags behind in these measures (Figure 8).
Most African countries have adopted a goal of
reducing population growth, so the challenge is
to direct adequate resources to these strategies,
develop the necessary administrative and technical
capacity, and mobilize civil society.
If sub-Saharan Africa could move toward replace-
ment-level fertility rates by 2050, its population
would grow to only 1.8 billion. The regional growth
in crop demand would then decline by nearly one-
third relative to our baseline projection. The region’s
farmers would need to clear only 97 Mha of forests
and savannas for agriculture rather than the 260
Mha in our baseline projection, closing one-quarter
of the global land gap. The global GHG mitigation
gap would decline by 17 percent.
Total fertility rate (2010–15) Mortality of children under age 5
per 1,000 live births (2010–15)
Percent of women ages 20–39 with at least
a lower secondary education (2005–10)
2.20 43 5
N/A N/AN/A 10010050 15080100 4060 20 0
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 21
In addition to the demand-reduction measures addressed in
Course1, the world must boost the output of food on existing
agricultural land. To approach the goal of net-zero expansion
of agricultural land, under realistic scenarios, improvements in
crop and pasture productivity must exceed historical rates of
yield gains.
Assessing the Challenge of
Agricultural Land Expansion
The single most important need for a sustainable
food future is boosting the natural resource e-
ciency of agriculture, that is, producing more food
per hectare, per animal, per kilogram of fertilizer,
and per liter of water. Such productivity gains
reduce both the need for additional land and the
emissions from production processes. Without the
large crop and livestock productivity gains built into
our baseline (based roughly on trends since 1961),
land conversion would be ve times greater by 2050
and GHG emissions would be more than double the
level projected in our baseline (Figure 9).
In some mitigation analyses, including reports by the
Intergovernmental Panel on Climate Change (IPCC),
agricultural productivity gains are barely mentioned,
for reasons that are unclear. Even under our base-
line projection, with its large increases in crop and
livestock yields, we project that agricultural land will
expand by 593 Mha to meet expected food demand.
Unless projected growth in demand for food can be
moderated, to avoid land expansion both crop yields
and pasture-raised livestock yields will have to grow
even faster between 2010 and 2050 than they grew
in previous decades.
Arguments can be made for both pessimism and
Studies have projected that farmers could
achieve far higher yields than they do today.
However, methods for estimating these “yield
gaps” tend to exaggerate gap sizes and farm-
ers can rarely achieve more than 80 percent of
yield potential. The most comprehensive study
suggests that fully closing realistic yield gaps is
unlikely to be enough to meet all food needs.
The massive yield gains of the 50 years from 1960
to 2010 were achieved in large part by doubling
irrigated area and extending the use of scienti-
cally bred seeds and commercial fertilizer to most
of the world. Only limited further expansion of
these technologies remains possible.
Optimistically, farmers have so far continued
to steadily boost yields by farming smarter in a
variety of ways, and new technologies are open-
ing up new potential.
Whatever the degree of optimism, the policy implica-
tions are the same: Going forward, the world needs
to make even greater eorts to boost productivity
than in the past to achieve a sustainable food future.
Source: GlobAgri-WRR model.
Figure 9 | Improvements in crop and livestock productivity already built into the 2050 baseline close most of the
land and GHG mitigation gaps that would otherwise exist without any productivity gains after 2010
Million hectares
Net agricultural land expansion (2010–50) Agricultural GHG emissions
(production + land-use change)
in baseline
due to
2050 (No
gains after
2050 (No
gains after
in baseline
due to
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 23
Figure 10 | Ineicient beef production systems result in far higher greenhouse gas emissions per unit of meat
MENU ITEM: Increase Livestock and
Pasture Productivity
Demand for milk and meat from grazing rumi-
nants is likely to grow even more than demand for
crops. Because pasture makes up two-thirds of all
agricultural land, the productivity of livestock will
critically aect future land use and emissions. Large
productivity improvements for pork and poultry are
unlikely in developed countries because of biological
limits.33 In developing countries, because traditional
backyard systems make use of waste and scavenging,
shifts to modern systems increase output but do not
reduce land-use demands and emissions.
By contrast, ruminant systems have greater potential
to improve, as suggested by the wide range in pro-
ductivities across countries. The GHG emissions that
result from producing each kilogram of beef—a good
proxy for all aspects of productivity—are far higher
in some countries than in others (Figure 10). Land-
use requirements can be 100 times greater,34 and the
quantity of feed 20 times greater.35
Higher ruminant productivity can be achieved by
increasing output per animal through improved
food quality, breeding, and health care; and by
increasing feed output per hectare. Neither requires
a shift to feedlots. On pastures with good rainfall,
productivity can be increased by proper fertiliza-
tion, growing legumes, rotational grazing, and add-
ing supplemental feeds in dry seasons and during
the last few months of “nishing.” In the “cut and
carry” systems that predominate in Africa and Asia,
farmers can grow a wide variety of improved forage
grasses and shrubs with high protein leaves.
The real challenge lies in the scale of improvement
required. Because much grazing land is too dry or
too sloped to support large feed improvements,
almost every hectare of wetter, accessible, and
environmentally appropriate land would need to
achieve close to its maximum productive potential
to meet expected global demand without the need
for further land conversion.
Most ruminant farmers need to shift from low-
management operations, which take advantage of
cheap land, toward careful, intensive grazing and
forage management using more labor and inputs.
Governments in developing countries, which
are home to the great majority of ruminants,
should establish livestock productivity targets
and support them with greater nancial and
technical assistance.
Implementation of systems to analyze improve-
ment potential and track changes in dierent
areas and on dierent types of farms would help
guide these investments and monitor their eects.
Source: Herrero et al. (2013).
kg CO2e/kg protein (2000)
10025 50 100 250 500 1,000
MENU ITEM: Improve Crop
Breeding to Boost Yields
Breeding of improved crops is generally credited for
half of all historical yield gains. Breeding can both
increase the potential yield of crops under ideal
conditions and help farmers come closer to those
potential yields by better coping with environmen-
tal constraints. Countries that have invested more
in recent years in crop breeding, such as Brazil and
China, have seen vast improvements in their yields.
“Incremental” crop breeding has been the primary
driver of yield gains through assessment and
selection of the best performing existing crops, fol-
lowed by purication, rebreeding, production, and
distribution. In the United States, improved maize
varieties are released every three years. Speeding
new crop cycles would boost yield growth in many
countries such as Kenya and India, where new
grain varieties are released typically every 13 to 23
Much debate has focused on genetically modied
organisms (GMOs), which involve insertion of
genes from one plant into another. The debate has
centered overwhelmingly on two types of traits that
assist pest control through glyphosate resistance
and expression of Bt (Bacillus thuringiensis), a bio-
logical pesticide. Some bona de debate is appro-
priate about whether the ease of use and relatively
lower toxicity provided by these traits in the short
term, and their potential value to small farmers
without access to pesticides, justies the longer-
term risks of building resistance in weeds, worms,
and insects—potentially leading to more pesticide
use in the future. There is no evidence that GMOs
have directly harmed human health.37
Gene editing has far greater potential. Sometimes
new genes can provide the only viable mechanisms
for crops to survive new diseases. New genes may
also play a major role in combating environmen-
tal challenges by making crops more ecient at
absorbing nitrogen or suppressing methane or
nitrous oxide emissions.
The CRISPR-Cas938 revolution since 2013 dramati-
cally increases opportunities to improve breeding
through genetic manipulation. CRISPR enables
researchers to alter genetic codes cheaply and
quickly in precise locations, insert new genes, move
existing genes around, and control expression of
existing genes. CRISPR follows a related genomics
revolution, which makes it cheap to map the entire
genetic code of plants, test whether new plants have
the desired DNA without fully growing them, and
purify crop strains more rapidly.
According to the most recent assessments, global
public agricultural research is roughly $30 billion
per year for all purposes, and private crop-breeding
research is around $4 billion, which we consider
modest. The vast opportunities created by new
technologies warrant large and stable increases in
crop-breeding budgets.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 25
MENU ITEM: Improve Soil and
Water Management
Revitalizing degraded soils, which may aect one-
quarter of the world’s cropland,39 provides another
opportunity to boost crop yields. Degradation is
particularly severe in drylands, which cover much
of Africa and where low soil fertility is a direct
threat to food security. Loss of organic matter is a
special concern because soils then hold less water
and are less responsive to fertilizers, making fertil-
izer use less protable.
Agencies in recent years have encouraged African
farmers to adopt “conservation agriculture,” which
relies on no or reduced tillage (plowing) of soils and
preserving crop residues.40 These practices can limit
soil erosion and may help boost yields modestly in
particularly dry areas, but farmers are often reluc-
tant to avoid tillage because of the increased need
for weeding or herbicides, and because they often
need to use crop residues for livestock feed.41
Some of the more promising approaches involve
agroforestry, often using nitrogen-xing trees.
Farmers have helped regenerate trees in farm elds
across 5 Mha in the Sahel, boosting yields.42 Com-
mitments to agroforestry made by many African
governments would benet from more systematic
evaluation of which systems work economically,
and where. Microdosing crops with small quantities
of fertilizer and trapping water on farms through
various blocking systems also shows promise in
Strategies to improve soils will need to address
the real obstacles facing farmers. Rebuilding soil
carbon may require diversion of land, labor, or
residues needed for food production and will
therefore need nancial support.44 Eorts to grow
more legumes to x nitrogen in African soils must
overcome high rates of disease, which requires
breeding plants with improved disease resistance.
Enhancing soil carbon also requires that farmers
add or x enough nitrogen to meet crop needs and
those of soil-building microbes, so cheaper fertil-
izers must be available.
In drylands like the Sahel, governments and
international aid agencies should increase
support for rainwater harvesting, agroforestry,
farmer-to-farmer education, and reform of
tree-ownership laws that can impede farmers’
adoption of agroforestry.
Elsewhere, governments and aid agencies need
to explore new models for regenerating soils.
One option may be to provide nancial help to
farmers to work incrementally on their farms,
improving one small piece of land at a time. If
one small area can be improved quickly to the
point where it generates large yield gains, the
economic return may come soon enough to
motivate farmer eorts.
MENU ITEM: Plant Existing
Cropland More Frequently
FAO data indicate that more than 400 Mha of
cropland go unharvested each year, suggesting that
this amount of land is left fallow.45 FAO data also
indicate that farmers plant roughly 150 Mha twice
or more each year (double cropping).46 The ratio
of harvests each year (harvested area) to quantity
of cropland is known as the “cropping intensity,”
a ratio that FAO currently estimates at 82 percent.
Planting and harvesting existing cropland more
frequently, either by reducing fallow land or by
increasing double cropping, could in theory boost
food production without requiring new cropland.
Some analysts have interpreted FAO data to suggest
a large recent increase in cropping intensity, but
these claims are mostly undercut by local satellite
studies. Using relatively crude criteria, other studies
have suggested a substantial theoretical potential
to increase double cropping on rainfed lands. But
roughly half of double-cropped land today is irri-
gated, and farmers probably plant two crops a year
on only 6 percent of rainfed area. Practically and
economically, the prospects for expanding double
cropping on rainfed lands must therefore be lim-
ited, as is expanding double cropping on irrigated
land because of water constraints.
In addition, there are signicant environmental
costs in some regions to planting fallow croplands
more frequently because some fallow lands are
either in very long-term rotations or are in the early
stages of abandonment. Typically, they will revert
to forest or grassland and help store carbon and
provide other ecosystem services. Planting them
more frequently sacrices these benets.
Despite diculties, there are opportunities for
progress. Raising cropping intensity is a promising
option, particularly in Latin America, where double
cropping has been growing. Our baseline assumes
a 5 percent increase in cropping intensity to 87 per-
cent. If cropping intensity were to increase another
5 percent, the land gap would shrink by 81 Mha, or
14 percent.
Strategies to encourage higher cropping intensity
require scientists to conduct more detailed and
spatially explicit analyses to determine realistic
potential increases in cropping intensity. Studies
should account for limitations on irrigation water
availability and build in at least some basic eco-
nomics. Governments and researchers will then be
better able to determine which improvements in
infrastructure or crop varieties can contribute to
economically viable increases in cropping intensity.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 27
MENU ITEM: Adapt to Climate Change
The global impacts of climate change on agriculture
are suciently uncertain that we did not attempt to
model them in our 2050 baseline. Although earlier
analyses suggested that eects on crop yields by
2050 might even be benecial, by the time of the
2014 IPCC report, models projected on average
that, without adaptation, global crop yields were
“more likely than not” to decline by at least 5 per-
cent by 2050—with even steeper declines by 2100.47
Many estimates are even larger, and uncertainty
should be a cause for greater concern because
“medium” impacts are not more likely.48 We mod-
eled one plausible estimate of a 10 percent decline in
crop yields due to climate change without adapta-
tion. Cropland would need to expand overall by 457
Mha (increasing the total land gap by 45 percent).
Climate change will benet some crops, at least in
the short term, as higher concentrations of carbon
dioxide increase the eciency of photosynthesis.
Warmer temperatures will extend the growing
season in colder countries and regional shifts in
rainfall patterns will make some locations wetter.49
But some areas will also become drier and hotter.
Higher temperatures will harm crops by drying
soils, accelerating water loss, and increasing pest
damage.50 Extreme heat events will harm maize,
wheat, coee, and many other crops by interfering
with reproduction.51 Growing seasons in parts of
sub-Saharan Africa could become too short or too
irregular to support crops (Figure 11), contributing
to major food security concerns.52
The evidence from crop models indicates signi-
cant but uncertain capacity to adapt using tailored
crop varieties. Uncertainties about local climate
change suggest broad “no regrets” strategies, many
of them already included in our other menu items.
For example, closing yield gaps in Africa and India
would increase incomes and provide a buer
against adverse climate impacts, forest protection
could increase resilience through improved local
hydrology, while safety net programs for the rural
poor will better equip small farmers to deal with
future variability.
Some climate eects, however, are suciently
clear to emphasize the need for new measures or
expanded eort on other menu items:
Farmers need eective regional crop-breeding
systems that enable them to select alterna-
tive crop varieties specically adapted to local
Small-scale irrigation and water conservation
systems will help farmers cope with rainfall
Research organizations and companies must
breed new traits to overcome highly likely big
climate challenges such as high temperature
eects on maize, wheat, rice, and coee.
Governments must help fund adaptation to
those major physical changes that are clearly
predictable, such as altering production systems
in areas that will be aected by sea level rise.
>20% loss
5–20% loss
No change
5–20% gain
>20% gain
Figure 11 | Climate change could shorten growing
seasons in much of sub-Saharan Africa
by more than 20 percent by 2100
Source: Verhage et al. (2018) using methods from Jone s and Thornton (2015).
Length of growing period in the 2090s
compared with the 2000s
How Much Could Boosting Crop and
Livestock Productivity Contribute to
Closing the Land and Greenhouse Gas
Mitigation Gaps?
The menu items in Course 2 are needed rst merely
to achieve our baseline. As Figure 9 and Table 2
show, the productivity gains assumed in our base-
line projection close more than 80 percent of the
land gap (and approximately two-thirds of the GHG
mitigation gap) that would result if agricultural
eciency did not improve at all after 2010. We also
modeled more optimistic scenarios to 2050, where,
relative to the baseline projection, we assume a 25
percent faster rate in ruminant livestock productiv-
ity gains, 20 and 50 percent faster rates of growth
in crop yield gains, and a 5 percent additional
increase in cropping intensity.
Even these additional improvements leave signi-
cant land and GHG mitigation gaps (Table 2). This
is why closing the land gap completely will require
demand-side measures (Course 1) and action to
protect and restore natural ecosystems (Course 3),
and why closing the GHG mitigation gap completely
will require action across all courses.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 29
Table 2 | Higher crop and livestock productivity could reduce agricultural land area and greenhouse gas
emissions in 2050
 MILLION HA
Gt COe GHG
GAP Gt COe
No productivity gains after 2010 2,199 1,066 3,265 11.3 26.9 38.2 34.2
2050 Baseline (crop yields grow
by 48%, cropping intensity by
5%, and output of meat or milk
per hectare of pasture by 53–71%
between 2010 and 2050)
401 192 593 9.0 6.0 15.1 11.1
Scenario variations relative to 2050 baseline
Failure to adapt to climate
change (10% decline in total crop
402 457 859 9.3 8.2 17.6 13.6
25% faster rate of output of meat
and milk per hectare of pasture
(58–76% growth between 2010
and 2050)
291 182 473 8.8 5.1 13.9 9.9
20% faster increase in crop
yields (crop yields grow by 56%
between 2010 and 2050)
401 100 501 8.9 5.3 14.3 10.3
5% additional increase in
cropping intensity (10% growth
between 2010 and 2050)
401 110 512 9.0 5.4 14.4 10.4
Notes: “Cropland” includes cropland and aquaculture ponds. Numbers not summed correctly are due to rounding.
Source: GlobAgri-WRR model.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 31
This course focuses on the land-management eorts that must
complement food demand-reduction eorts and productivity gains
to avoid the harms of agricultural land expansion. One guiding
principle is the need to make land-use decisions that enhance
eiciency for all purposes—not just agriculture but also carbon
storage and other ecosystem services. Another principle is the
need to explicitly link eorts to boost agricultural yield gains with
protection of natural lands.
The Causes and Consequences of
Shifting Agricultural Land
Merely eliminating the need for a net expansion
of agricultural land will not avoid all carbon and
ecosystem losses because agricultural land is not
merely expanding, it is also shifting. At a regional
level, agricultural land is shifting from developed to
developing countries.53 One reason is that growth
in population and food demand is mostly occurring
in developing countries. Rising food demand in
sub-Saharan Africa, for example, is likely to drive
cropland expansion of 100 Mha between 2010 and
2050, even allowing for high estimated yield gains
in the region and continued importation of roughly
one-fth of staple foods. Another reason is growing
global demand for some highly traded crops that
developing countries have learned to grow well,
such as soybeans and palm oil.
Agricultural land is also shifting within regions
and countries, particularly from less productive
and more sloped lands to atter, more productive,
more densely vegetated lands. These shifts result
in gross forest losses that are much larger than net
losses (Figure 12). Many abandoned agricultural
lands do reforest but, unfortunately, the trade-o
when native forests are replaced with planting or
regrowing forests elsewhere is not environmentally
neutral. Conversion of natural ecosystems, which
is occurring mostly in the tropics and neotropics,
tends to release more carbon per unit of food54
and harm more biodiversity than reforestation of
abandoned land osets elsewhere. The losses of
carbon during land conversion also occur quickly,
whereas rebuilding carbon in vegetation and soils
occurs gradually over longer time periods, exacer-
bating climate change in the interim.55 The common
tendency of countries to replant abandoned land as
forest plantations also reduces carbon and biodiver-
sity benets.
A sustainable food future therefore requires eorts
to reduce agricultural land-shifting, minimize the
environmental consequences of inevitable expan-
sion in some countries, and more actively reforest
abandoned agricultural land. Under the Bonn
Challenge—a global eort to bring 350 Mha into
restoration by 2030—47 national and subnational
actors have now committed to restore over 160
Europe Oceania South America
Gross gain Gross loss Net gain Net loss
Thousand ha/year
North and
Central America
Source: FAO (20 12a).
Figure 12 | Gross forest losses are far greater than net forest losses because agricultural lands are shifting
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 33
MENU ITEM: Link Productivity Gains
with Protection of Natural Ecosystems
Although yield gains are critical to achieving food
security and reducing the need for global agri-
cultural land expansion, yield gains may increase
protability locally, which may encourage conver-
sion of natural landscapes to increase export share.
New roads and other infrastructure can also make
it protable to convert new lands. Governments
today have plans for major new roads in Africa and
Latin America that would likely lead to extensive
conversion to agriculture and loss of habitat in
many biodiversity hotspots.57 If the world is to reap
the benets of productivity gains while protect-
ing natural ecosystems, eorts to do both must be
explicitly linked.
Experience has shown that, given political will and
sucient enforcement, governments can protect
forests and other natural landscapes. In many
countries, governments own the majority of natural
lands. They can control how and where private
parties may claim ownership or rights to develop
public lands though, in some cases, a dicult
balance must be struck between enforcement of
land-use restrictions and the needs of impoverished
smallholders.58 Where farmers have clear title to
their land, governments can combine enforcement
with support for agricultural improvement on
existing farmland to build social support. Modern
monitoring techniques can now provide a powerful
foundation for accountability and enforceability of
forest protection laws. Not least, governments can
designate natural and Indigenous Peoples’ pro-
tected areas that recognize local rights and protect
forests. Researchers have found that recognizing
indigenous lands has signicantly reduced forest
clearing and disturbance in the Amazon.59
Linking agricultural improvement and ecosystem
protection has potential to attract additional aid and
investment from parties interested in either goal.
Governments, nanciers, and other parties should
pursue such linkages and make them as explicit as
possible through a variety of mechanisms:
International nance. Development as-
sistance should explicitly link programs to
improve agriculture production with forest (or
other natural ecosystem) protection.
Also, international private nanciers should give
preferential access to nance for investments
that make the linkage to protection explicit.
Agricultural loans. National governments
should learn from Brazil’s example by legally
linking credit and other agricultural improve-
ment assistance to protection of native habitats.
Supply chain commitments. Buyers and
traders of agricultural commodities should set
purchasing contracts conditioned on the com-
modity not being linked to natural ecosystem
Land-use planning. International agencies
should help national governments develop
detailed spatial tools that guide agricultural
zoning and roadbuilding away from natural
All of the above could be integrated in a “jurisdic-
tional approach” at a national or subnational level,
motivated by REDD+ nance or other means. An
example from Brazil is Mato Grosso’s “Produce,
Conserve, and Include” development plan, which
aims to promote sustainable agriculture, eliminate
illegal deforestation, and reduce GHG emissions.60
MENU ITEM: Limit Inevitable Cropland
Expansion to Lands with Low
Environmental Opportunity Costs
In some countries, preventing all agricultural land
expansion is not feasible. For example, rising food
demand in Africa will realistically require some
land expansion, as will global demand for vegetable
oil in Southeast Asia. In countries where expansion
or shifts of agricultural land are inevitable, coun-
tries need to identify and facilitate expansion only
where it would cause less environmental harm.
The goal is to nd lands with relatively low environ-
mental (and other) opportunity costs but with good
productive potential. These opportunities involve
trade-os (Figure 13). Evaluation of land conver-
sion requires assessing not only the loss of existing
carbon but also the forgone carbon sequestration on
lands that would otherwise regenerate, for example,
on cut-over areas. It should also require analysis of
the carbon and biodiversity losses per ton of crop
(not per hectare) to minimize total environmental
costs while meeting food needs.
This menu item requires above all a commitment by
national governments, but it also requires tools that
international aid agencies should help fund and
that could also help link agricultural improvement
and natural landscape protection more generally:
Tools and models must estimate likely yields
and eects on biodiversity and carbon of
dierent development patterns, incorporate
information on various obstacles, and allow a
wide range of stakeholders to explore accept-
able alternatives. A tool developed for Zambia,
for example, showed how balancing production
and environmental goals could come close to
maximizing yield potential while holding down
transportation costs, carbon losses, and adverse
eects on biodiversity.61
Integrating such tools with analyses of agri-
cultural potential and current farming systems
on existing agricultural land could help target
use of agricultural improvement funds. Tools
that are useful at the farm level and then ag-
gregate to the regional and national level have
the greatest potential because their use should
improve the quality of analysis over time.
Governments will need to use such assessment
tools to guide land-use regulations, plan road
routes, and manage public lands.
Figure 13 | Screening out lands that do not meet environmental, economic, and legal criteria reduces the area
of land suitable for oil palm expansion in Kalimantan, Indonesia
Lands meeting the environmental criteria
for supporting sustainable oil palm
Not suitable Suitable
Lands meeting the environmental
and economic criteria for supporting
sustainable oil palm
Lands meeting the environmental, and
economic, and legal criteria for supporting
sustainable oil palm
Source: Gingold et al. (20 12).
MENU ITEM: Reforest Abandoned,
Unproductive, and Liberated
Agricultural Lands
Because some agricultural land will inevitably shift,
maintaining forest and savanna area will require
reforestation of abandoned agricultural land or
restoration to other natural or seminatural ecosys-
tems.62 History shows that regeneration typically
occurs naturally, though governments have assisted
the process by subsidizing tree planting. But plant-
ing often creates single-species forest plantations
with little biodiversity and less carbon than natu-
ral, diverse forests. Because of land-shifting, such
plantations can contribute to global loss of natural
forest cover.
The potential for reforestation is sometimes overes-
timated.63 For example, some studies assume that
wetter pasturelands, particularly those that were
originally forest, are simply available for planting
forests, without recognizing the important role they
play in producing milk or meat or the fact that their
intensication will be necessary just to keep pasture
area from expanding. Larger-scale reforestation
to mitigate climate change will be possible only if
agricultural land is “liberated” through highly suc-
cessful eorts to slow growth in food demand and
intensify production on existing land.
Pending such success, because of growing global
food needs, reforesting land in agricultural use for
climate purposes should generally be limited to
land that is producing little food and has low poten-
tial for agricultural improvement. Prime examples
are many of the degraded and low-productivity
pastures of Brazil’s Atlantic Forest region, which
are hard to improve because of steep slopes but
which could recover into carbon-rich and biologi-
cally diverse forests.64
Governments should commit more eorts to
natural reforestation of marginal or abandoned
agricultural land and should give greater emphasis
to establishing diverse natural forests. These eorts
will require new funding and they would be a good
use of international climate funds. Once com-
mitments in this direction are made, they should
take account of practical lessons that have already
become clear:
Governments and other actors can sometimes
keep costs down by pursuing “assisted natural
regeneration,” which involves keeping re, live-
stock grazing, or other disturbances away from
land targeted for reforestation.
Governments can provide lines of concessional
credit for replanting trees within traditional
agricultural loans.
Governments can help fund nurseries of native
tree species.
Governments can monitor progress, in part to
determine the need for midcourse corrections
and in part to enforce forest protection for
newly reforested areas as well as older forests.
73 (Acacia on peat)
55 (other plantation types on peat)
40 (oil palm on peat)
Unknown or zero emissions (peatland outside plantations)
Tons of carbon dioxide per hectare per year (2013–14)
MENU ITEM: Conserve and
Restore Peatlands
The highest priority for immediate restoration
should be the world’s 26 Mha of drained peatlands.
This small area is responsible for roughly 2 percent
of annual human-caused GHG emissions, accord-
ing to our estimates. The best evidence indicates
that roughly half of these peatlands have little or no
agricultural use or are used only for grazing.
Peatlands are wetlands that built up massive
carbon-rich soils over hundreds or thousands of
years.65 Their conversion for agriculture and planta-
tion forestry typically requires drainage, which
causes the soils to decompose and sometimes burn,
releasing large quantities of carbon into the atmo-
sphere (Figure 14). Rewetting peatlands by blocking
drainage ditches can typically eliminate emissions.
Peatlands appear to be far more extensive than
previously thought, suggesting high risk of further
losses. Researchers have recently discovered the
world’s largest tropical peatland in the heart of
the Congo rainforest in central Africa.66 It stores
an estimated 30 gigatons of carbon, equivalent to
roughly 20 years of U.S. fossil fuel emissions. Other
large peatlands probably exist in Latin America.
Modest eorts at restoration have occurred in
Russia.67 The president of Indonesia has announced
a broad restoration goal, and the country reported
200,000 hectares of peatland restoration between
2017 and 2018.68 Yet the global eort to restore
peatlands is minimal compared to the need. Elimi-
nating half of peatland emissions would close the
global GHG emissions gap by 5 percent, while
eliminating 75 percent would close the GHG mitiga-
tion gap by 7 percent. A series of actions is required:
Restoration eorts require more funding both
to perform the physical restoration and to
compensate farmers and communities who
must forgo other uses, even if relatively mod-
est. Ideally, assistance would be used to boost
productivity on farms outside peatlands.
Peatland conservation and restoration require
better mapping, especially because peatlands
cannot be identied from satellite imagery.
Mapping and data collection should be a prior-
ity for national governments, international
agencies, and even private parties.
Strong laws must protect peatlands to prevent
their conversion to agriculture.
Figure 14 | Greenhouse gas emissions from drained peatlands are ongoing in Indonesia and Malaysia
Source: WRI (2017).
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 37
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 39
Fish, including finfish and shellfish, provide only small
percentages of total global calories and protein, but they
contribute 17 percent of animal-based protein,69 and are
particularly important for more than 3 billion people in developing
countries.70 We project fish consumption to rise 58 percent
between 2010 and 2050, but the wild fish catch peaked at 94
million tons in the mid-1990s and has since stagnated or perhaps
declined.71 This course proposes ways to improve wild fisheries
management and raise the productivity and environmental
performance of aquaculture.
MENU ITEM: Improve Wild
Fisheries Management
According to FAO, 33 percent of marine stocks
were overshed in 2015, with another 60 percent
shed at maximum sustainable levels (Figure 15).
One World Bank study found that world shing
eort needs to decline by 5 percent per year over a
10-year period just to allow sh stocks to rebuild.72
Solutions to curb overshing are well-known and
documented. They focus on principles including
limiting catches to a level that allows the sh popu-
lation to reproduce, limiting the number of shers
to an economically sustainable level, protecting
habitat, and avoiding harvest during important
breeding times or in important breeding areas.
The challenges to implementing these solutions are
largely political and social. Wild sh are a public
resource that individual shers have incentives to
exploit before others can do so. Other challenges
reect power imbalances, where foreign eets from
richer countries often are able to obtain agreements
to sh in the waters of poorer countries with weaker
laws and enforcement capacity. Solutions require
mechanisms for persuading shers to support
reductions in sh catch levels:
Catch shares limit total sh catch and allocate
shares of the catch among shers, who then
have a long-term interest in preserving the
health of the shery.
Where oversight is weaker, community-based
comanagement systems may prove more eec-
tive. Such systems combine territorial shing
rights and no-take reserves designed and sup-
ported by coastal shing communities.
Removing perverse subsidies—estimated at $35
billion annually73—could dramatically reduce
Because reducing overshing is hard, we assume a
10 percent reduction in wild sh catch between 2010
and 2050 in our baseline scenario, and even that
goal requires major reforms. A scenario in which
sheries are rebuilt enough to maintain the 2010
level of sh catch in 2050 would have little eect on
our gaps but would supply an additional 9 Mt of sh
(relative to our 2050 baseline) and would avoid the
need to convert 5 Mha of land to supply the equiva-
lent amount of sh from aquaculture ponds.
Figure 15 | The percentage of overfished stocks has risen over the past 40 years
Source: FAO (2018).
Fully but sustainably fished
Not fully fished
1980 1990 2000 20101975 1985 1995 2005 2015
Percentage of marine fish stocks assessed
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 41
Figure 16 | Aquaculture production must continue to grow to meet world fish demand
MENU ITEM: Improve Productivity
and Environmental Performance of
Growth in world sh supply since the 1990s has
come from aquaculture (sh farming). Aquacul-
ture production would need to more than double
between 2010 and 2050 to meet projected sh
demand in our baseline (Figure 16).
Aquaculture is a relatively ecient means of sup-
plying animal-based protein. Although eciencies
vary by type of sh and production method, average
land-use demands are on par with poultry produc-
tion (Figure 5) and can even be zero for certain
species (e.g., bivalve mollusks). Greenhouse gas
emissions from aquaculture are similar to those of
poultry and pork production, and much less than
those of ruminant meats.
Yet aquaculture presents a range of environmental
challenges, which vary by production system. They
include conversion of valuable wetland habitats
(such as mangroves), use of wild-caught sh in
feeds, high freshwater demand, water pollution,
and eects of escaped farm sh on wild sh. Aqua-
culture ponds occupied an estimated 19 Mha in
2010, while an additional 27 Mha was used to grow
crop-based sh feed. The total land-use demands
roughly double in our 2050 baseline projection.
Aquaculture must become more land-ecient, espe-
cially because available land is constrained in Asia,
where nearly 90 percent of aquaculture production
occurs.74 Shifting to deeper ponds with water recircu-
lation will be necessary to increase production while
limiting land expansion. Opportunities also exist
to expand aquaculture in marine waters, possibly
further oshore.
Aquaculture growth will require development of
feed substitutes to replace oil from wild sh because
this source is already near or above ecological
limits. Promising alternatives include microalgae-
based feeds and uses of genetically engineered
yeasts or oilseeds bred to produce the omega-3 fatty
acids that characterize wild sh oil. Aquaculture
must also overcome signicant rates of sh disease.
Several strategies can help aquaculture grow sus-
tainably to help meet rising sh demand:
Selective breeding for improved sh growth
rates and conversion eciencies.
Technological developments in sh oil alterna-
tives, other feed improvements, and disease
Use of water recirculation and other pollution
Use of spatial planning to optimize aquaculture
Expansion of marine-based systems.
1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050
Wild (capture) fisheries
Source: Historical data, 1950–2016: FAO (201 7b) and FAO (2018). Projections to 2050: Calculated at WRI; assume s 10 percent reduction in wild fish catch from 2010 levels by 2050,
linear grow th of aquaculture production of 2 M t per year between 2010 and 205 0.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 43
Agricultural production emissions arise from livestock farming,
application of nitrogen fertilizers, rice cultivation, and energy use.
These production processes are traditionally regarded as hard
to control. In general, our estimates of mitigation potential in this
course are more optimistic than others’, partly because many
analyses have not fully captured the opportunities for productivity
gains and partly because we factor in promising potential for
technological innovations.
Even with Large Productivity Gains,
We Project Production Emissions to Rise
Annual emissions from agricultural production
processes (i.e., excluding emissions from land-use
change) reach 9 Gt in our 2050 baseline (Figure 17),
leaving a 5 Gt GHG mitigation gap relative to our
Figure 17 | Annual agricultural production emissions reach 9 gigatons in our 2050 baseline projection
Source: GlobAgri-WRR model.
target emissions level of 4 Gt. The baseline already
incorporates large productivity gains, without
which the gap would rise to 7 Gt. Most produc-
tion emissions take the form of two trace gases
with powerful warming eects, nitrous oxide and
2010 (Base year)
Million tons COe per year
2050 (Baseline) 2050 (No productivity
gains after 2010)
Ruminant wastes
on pastures
Soil fertilization
Rice methane
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 45
Emissions intensity (kg CO2e per kg milk)
kg milk per cow
01,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000
MENU ITEM: Reduce Enteric Fermentation
through New Technologies
Ruminant livestock (mainly cattle, sheep, and
goats) generate roughly half of all agricultural pro-
duction emissions. Of these emissions, the largest
source is “enteric methane,” generated by microbes
in ruminant stomachs.
The same measures needed to increase productiv-
ity of ruminants and reduce land-use demands will
also reduce methane emissions, mainly because
more milk and meat is produced per kilogram
of feed. Because the improvements are greatest
when moving from the worst-quality feeds to even
average-quality feeds, the greatest opportunities to
reduce emissions exist in poorer countries. Improv-
ing highly inecient systems causes emissions per
kilogram of meat or milk to fall very sharply at rst
as output per animal increases (Figure 18).
Other strategies to reduce enteric methane emis-
sions rely on manipulating microbiological commu-
nities in the ruminant stomach by using vaccines;
selectively breeding animals that naturally produce
fewer emissions; or incorporating special feeds,
drugs, or supplements into diets. These eorts have
mostly proved unsuccessful. For example, despite
testing thousands of compounds, researchers have
found that methane-producing microbes quickly
adapt to drugs that initially inhibited them.75
More recently, at least one highly promising option
has emerged that persistently reduces methane
emissions by 30 percent, and may also increase ani-
mal growth rates.76 So far, this compound—called
3-nitrooxypropan (3-NOP)—requires daily feeding
at a minimum, so it is not feasible today for most
grazing operations unless dosing can be rened.
Enteric fermentation, atypically, is receiving a
reasonable level of R&D funding. It is possible
that 3-NOP could eventually pay for itself through
reduced feed needs or increased productivity,
but these benets are not guaranteed. However,
because the compound will be highly cost-eective
for GHG mitigation, we recommend that govern-
ments consider three policies:
Provide incentives to the private sector by
promising to require use of 3-NOP or other
compounds if and when they are proven to
mitigate emissions at a reasonable cost.
Fund large-scale 3-NOP or related demonstra-
tion projects in the short term.
Maintain public research into compounds to
reduce methane from enteric fermentation.
Figure 18 | More eicient milk production reduces greenhouse gas emissions dramatically
Note: Dots represent country averages.
Source: Gerber et al. (2013).
MENU ITEM: Reduce Emissions through
Improved Manure Management
Manure is “managed” when animals are raised in
conned settings and farmers remove the manure
and dispose of it. (Manure that ruminants deposit
directly on elds is considered “unmanaged” and
is addressed in the next menu item.) Managed
manure generates both methane and nitrous oxide
emissions. Pigs generate roughly half of these emis-
sions, dairy cows just over one-third, and beef cows
roughly 15 percent.77
The majority of manure is managed in “dry”
systems, which account for 40 percent of total
managed manure emissions despite low emissions
rates.78 Estimates of mitigation potential tend to be
low, in part because management costs tend to be
high per ton of emissions. Even so, the use of dry
systems is desirable because emissions from “wet”
systems (where farmers do not attempt to dry out
manure before storage) can be 20 times higher per
ton of manure.79 We also believe that separating
liquids from solids has underappreciated potential
to reduce emissions. Separation technologies range
from simple gravity systems to sophisticated chemi-
cal treatments. They also reduce hauling costs and
make manure more valuable as fertilizer.
Per ton of GHGs reduced, existing wet systems are
easier to mitigate because the manure is generating
high levels of emissions. Even an extremely sophis-
ticated system for managing pig manure in North
Carolina (United States), using a series of treatment
tanks to eliminate virtually all air and water pollu-
tion, would cost only an estimated $22 per ton of
mitigation (CO2e).80 This system would add only
around 2 percent to the retail price of pork.81
Digesters, which convert manure into methane for
energy use, come in high-technology forms that
produce electricity in developed countries and
simpler household versions used extensively across
Asia. They can help reduce emissions but only if
manure would otherwise be managed in wet form,
and if strong safeguards are in place to keep meth-
ane leakage rates low.
Improving manure management will address a
range of environmental pollution, human health,
and nuisance concerns. Because measures to
mitigate emissions would typically contribute to
addressing these other concerns, the mitigation
may even be “free” from a socioeconomic perspec-
tive. Promising strategies are available:
Phased regulation of facilities, extending
from larger to smaller farms, to encourage
Government-funded programs, using
competition, to develop the most cost-eective
Establishment of government programs to
detect and remediate leakages from digesters.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 47
MENU ITEM: Reduce Emissions from
Manure Left on Pasture
According to standard emissions factors used by
the IPCC, nitrogen deposited in feces and urine
turns into nitrous oxide at roughly twice the rate of
nitrogen in fertilizer. Our 2050 projection already
incorporates productivity improvements that lower
emissions intensity from manure on pasture by 25
percent compared to 2010, but further increases
in feed eciency could lead to additional modest
emissions reductions.
Other studies typically estimate little to no global
potential to mitigate this diuse source of emis-
sions. We are cautiously optimistic, given further
development of two nascent technologies. Both
work by inhibiting the ability of microorganisms
to turn nitrogen from other molecular forms into
nitrate, whose further breakdown can release
nitrous oxide.
One method involves chemical nitrication inhibi-
tors, which have been found to be quite eective
when applied two or three times per year to pas-
tures in New Zealand82 and—in a very small number
of experiments—when ingested by cows. The other
involves biological nitrication inhibition, based on
ndings in Latin America that manure deposited
on one variety of the productive Brachiaria grass
generates almost no nitrous oxide emissions.83 To
exploit this property more broadly, breeders would
have to breed this trait into other planted grasses.
These techniques may be more practicable than
they appear. Newer research on nitrogen levels in
the atmosphere suggests that current emissions
factors for unmanaged manure may be too high and
that emissions rates are much higher in some elds
(usually in wetter areas) than others.84 This dier-
ence in emissions would allow more economical
targeting of these techniques toward wetter, more
intensively grazed elds.
Research funding to address this challenge is vanish-
ingly small. Far greater eorts and resources must be
devoted to exploring an array of potential solutions:
Governments and research agencies should
substantially increase research funding into
methods for reducing nitrication of manure.
Governments should commit in advance to
implement regulatory or nancial incentives
to implement these techniques when they do
become available, to encourage research and
development by the private sector.
reduce nitrous oxide emissions, and increase yields.
But they are used with only 2 percent of global
fertilizers,85 probably because of wide variability in
performance and because fertilizer manufacturers
today spend little money to improve them. Biologi-
cal nitrication inhibition is another promising
option for crops and pasture grasses but receives
little nancial support.
We modeled various scenarios of NUE improvement
and found that achieving an ambitious global aver-
age NUE of 71 percent by 2050 would reduce emis-
sions by 600 million tons, although that would only
keep nitrogen emissions close to their 2010 levels.
The scope of the nitrogen challenge requires govern-
ments to focus on innovative policy approaches:
Implement exible regulatory targets to push
fertilizer companies to develop improved fertil-
izers. India provides the closest example to date
with its New Urea Policy adopted in 2015.
Shift subsidies from fertilizers to support for
higher NUE, where nitrogen use is excessive.
Support critical research, particularly into bio-
logical nitrication inhibition.
Fund demonstration projects involving re-
searchers and high nitrogen-using farmers to
pursue higher NUE using inhibitors and other
innovative technologies.
MENU ITEM: Reduce Emissions
from Fertilizers by Increasing
Nitrogen Use Eciency
Fertilizers applied to crops and pastures (mostly
synthetic fertilizers but also manure and other
sources) were responsible for estimated emissions
of 1.3 Gt CO2e in 2010. Nearly all these emissions
result from the manufacture, transportation, and
application of nitrogen. We project that these
emissions will rise to 1.7 Gt by 2050 in our baseline
scenario. Globally, crops absorb less than half the
nitrogen applied to farm elds. The rest runs o
into ground or surface waters, causing pollution, or
escapes into the air as gases, including the potent
heat-trapping gas nitrous oxide. Countries, and
individual farms, vary greatly in their rates of nitro-
gen application per hectare and in the percentage
of nitrogen that is absorbed by crops rather than
lost to the environment (known as “nitrogen use
eciency,” or NUE) (Figure 19).
Mitigation strategies focus on changing agronomic
practices. Technically, extremely high rates of
eciency are possible if farmers are willing to
assess nitrogen needs and apply nitrogen frequently
in just the required amount over the course of a
growing season. The challenge is that such intensive
management is expensive while nitrogen fertil-
izer is cheap. Therefore, we believe innovations
are required. Nitrication inhibitors and other
“enhanced eciency” fertilizers can increase NUE,
Figure 19 | The percentage of applied nitrogen that is absorbed by crops varies widely across the world
Source: Zhang et al. (2015).
Percentage of nitrogen absorbed by crops
100N/A 25 50 100+
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 49
MENU ITEM: Adopt Emissions-Reducing
Rice Management and Varieties
The production of ooded or “paddy” rice contrib-
uted at least 10 percent of all global agricultural
production GHG emissions in 2010, primarily in the
form of methane.86 Available research suggests high
technical potential to mitigate rice emissions, and
most mitigation options also oer some prospect of
economic gains through higher yields and reduced
water consumption. We focus on four main options:
Increase rice yields. Because methane
emissions are tied more to the area than to
the quantity of production, exceeding FAO’s
forecast rate of yield growth would allow paddy
area to remain constant or decrease, reducing
Remove rice straw from paddies before
reooding to reduce methane produc-
tion. Straw may be used for other produc-
tive purposes such as growing mushrooms or
Reduce duration of ooding to reduce
growth of methane-producing bacteria.
Farmers can draw down water levels for a few
days during the middle of the growing sea-
sons, or plant rice initially into dry rather than
ooded land.
Breed lower-methane rice. A few existing
varieties emit less methane than others and re-
searchers have shown promising experimental
potential,87 but these traits have not been bred
into the most commercial varieties.88
A single drawdown reduces emissions, and multiple
drawdowns or dry planting plus one drawdown can
reduce methane emissions by up to 90 percent.89
In China and Japan, farmers practice at least one
drawdown because it increases yields,90 though
researchers do not nd those yield benets in the
United States. Reducing water levels also saves
irrigation water, at least at the farm scale.
Yet there are obstacles. Dry planting increases weed
growth. Farmers usually will not draw down water
unless they are sure they can replace the water and
elds are at enough to ensure no part dries out too
much. Another concern is that while drawdowns
decrease methane emissions, they tend to increase
emissions of nitrous oxide, another powerful
greenhouse gas, which encourages joint eorts to
use nitrication inhibitors. We propose the follow-
ing strategies:
Engineering analyses to determine which
farmers have irrigation systems that would
allow them to employ drawdowns, followed by
programs to reward farmers who practice draw-
downs where feasible.
A major breeding eort to shift to lower-meth-
ane varieties.
Greater eorts to boost rice yields through
breeding and management.
MENU ITEM: Increase Agricultural
Energy Eciency and Shift to Nonfossil
Energy Sources
Emissions from fossil energy use in agriculture will
remain at about 1.6 Gt CO2e/year in 2050 in our
baseline. Our assumption, based on past trends,
is that a 25 percent increase in energy eciency is
cancelled out by a 25 percent increase in energy
use. Mitigation options mirror those for reducing
energy emissions in other sectors: they rely on
increasing eciency and switching to renewable
energy sources.
Although studies of potential energy eciency
improvements in agriculture are limited, a small
number of country-level studies have found large
potential for eciency gains, for example, in alter-
native water pumps in India91 or cassava-drying
methods in Africa.92
Sixty-ve percent of expected agricultural energy
emissions in 2050 will result from on-farm energy
use. Heating and electrical power can often be pro-
vided by solar and wind energy sources, although
replacing on-farm coal will require innovative,
small-scale solar heating systems. Mitigating the
use of diesel fuel by tractors and other heavy equip-
ment will be more dicult and may need to rely on
transitions to fuel cells using hydrogen power gen-
erated with solar or wind power. Battery-powered
equipment and synthetic carbon-based fuels made
from renewable electricity may provide alternative
Renewable sources of hydrogen could also mitigate
85 percent of the emissions from the synthesis
of nitrogen fertilizer, currently a highly energy-
intensive process. Fortunately, because of the needs
of other sectors, substantial research is occurring
into production of hydrogen using electricity from
solar energy, and costs of solar electricity have been
declining rapidly.93
The eciency gains built into our baseline already
require signicant eort. We estimate that reducing
emissions per unit of energy used by 75 percent,
rather than the 25 percent in our baseline, would
reduce the GHG mitigation gap by 8 percent. To
achieve this goal:
Governments, aid agencies, and large food
purchasers should integrate low-carbon energy
sources and eciency programs into all devel-
opment eorts and supplier relationships with
Research agencies and private investors should
continue to fund research into production of
nitrogen from renewable electricity and sup-
port design of demonstration nitrogen fertil-
izer plants using renewable electricity. Such
research could be linked to ongoing work on
developing solar-based production of hydrogen.
Governments should commit to regulating the
emissions from fertilizer manufacturing once
viable, low-carbon technologies are available.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 51
MENU ITEM: Focus on Realistic Options
to Sequester Carbon in Soils
Because reducing agricultural production emissions
is challenging, much academic and policy attention
has focused on strategies to sequester carbon in
agricultural soils to oset those emissions. There
are only two ways to boost soil carbon: add more
or lose less. But recent scholarship and experience
indicate that soil carbon sequestration is harder to
achieve than previously thought.94
Changes in plowing practices, such as no-till, which
once appeared to avoid soil carbon losses, now
appear to provide only small carbon benets or no
benets when measured at deeper soil depths than
previously measured. No-till strategies must also
contend with adverse eects on yields on some
lands and the fact that many farmers who practice
no-till still plow up soils every few years, probably
releasing much of any carbon gain.95
Adding mulch or manure are proposed strategies to
add carbon to soils but, in eect, double-count their
carbon which would have contributed to carbon
storage elsewhere. Leaving crop residues otherwise
used for animal feed to become soil carbon requires
that the animals’ feed comes from other sources,
which has some carbon cost because it would often
require more agricultural land to grow that feed.
Building soil carbon also generally requires large
quantities of nitrogen, which is needed by the micro-
organisms that convert decaying organic matter to
soil organic carbon. Low nitrogen surely limits soil
carbon buildup in Africa (Figure 20), where nitrogen
additions are insucient even for crop needs, and
probably limits soil carbon buildup elsewhere.96
Scientists have come to realize that they do not well
understand the factors that lead carbon to remain
stored longer in soils rather than being consumed
and returned to the air by microorganisms. There
is some evidence that croplands are actually losing
soil carbon overall in ways neither we nor other
researchers count. For these reasons, we do not
include additional soil carbon sequestration as
a mitigation strategy. We believe eorts are best
directed toward stabilizing soil carbon, that is,
avoiding further losses, and focusing on no-regrets
strategies that provide additional benets:
Avoid conversion of carbon-rich ecosystems
(e.g., forests).
Increase productivity of grasslands and crop-
lands, which adds carbon in roots and residues.
Increase use of agroforestry, which builds
above-ground carbon.
Pursue eorts to build soil carbon, despite the
challenges, in areas where soil fertility is critical
for food security.
Figure 20 | Soils in Africa are relatively low in organic carbon
Source: Hengl and Reuter (2009).
Topsoil organic c arbon, percent mas s fraction
0.50N/A 11.5 22.5 510 40
The Need for Flexible Technology-
Forcing Regulations
Although many opportunities exist for developing
cost-eective, or even cost-neutral, GHG mitiga-
tion techniques to curb agricultural emissions, the
size of the GHG mitigation gap strongly suggests
that voluntary approaches will not be sucient.
We recommend a few forms of exible regula-
tions that should be designed to spur technological
In the case of fertilizer, we recommend that coun-
tries develop regulatory systems similar to those
developed in the United States that require auto
manufacturers to increase the fuel eciency of their
eets over time. Fertilizer manufacturers would
be required to sell increasing percentages of their
product in a form with “enhanced eciency,” such
as fertilizers incorporating nitrication inhibitors.
India has set an example by requiring that fertil-
izers be coated with neem, which slows nitrogen
release.97 Phased regulation would provide incen-
tives for manufacturers to develop better products,
identify the ideal uses, and market them appro-
priately to farmers who can most benet from
them. Regulating fertilizer is also consistent with
historical regulation of agricultural inputs, such as
Manure management is typically subject to weak
regulation. Governments should phase in require-
ments for pollution controls that tighten and reach
more sources over time, initially covering new and
large existing facilities and extending gradually to
medium-sized and smaller farms.
In areas where technologies are underdeveloped,
such as enteric methane inhibitors, governments
should commit in advance to requiring the use of
appropriate drugs or feed supplements if a com-
pany develops a system that achieves a certain level
of cost-eectiveness in mitigation—for example,
$25 per ton of CO2e. Greater market certainty
would provide incentives for the private sector to
develop needed innovations.
Many of these options may, at least initially, involve
additional costs, but they appear cost-eective
when compared to climate change mitigation strat-
egies in other sectors. Many options would have
large cobenets, such as reducing water and air
pollution, and controlling disease-bearing organ-
isms from poorly controlled livestock waste. Many
might eventually more than pay for themselves as
technologies evolve, which is likely true of nitri-
cation inhibitors and could be true of additives
to curb enteric methane. Yet these technologies
do not seem likely to evolve without either strong
incentives or some form of regulation designed to
advance their development and deployment.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 53
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 55
The individual menu items presented in Courses 1–5 can
each contribute to meeting global targets for increasing
food production, minimizing expansion of agricultural land
area, and reducing GHG emissions. In this section, we
use the GlobAgri-WRR model to examine some plausible
(or at least possible) combinations of menu items and
analyze how they could close the three gaps and achieve
a sustainable food future.
To assess the potential of the full menu to close the
food, land, and GHG mitigation gaps, we con-
structed three combination scenarios that reect
ascending levels of ambition (Table 3). They are
guided by the following criteria:
Coordinated Eort Scenario. Menu items
involve measures we are condent the world
could achieve with a strong, coordinated, global
commitment to action. The economic costs
would be limited or even positive. No funda-
mental breakthroughs in technology would be
Highly Ambitious Scenario. Menu items
involve measures at the outer range of what
might be technically achieved either with exist-
ing technology or with realistic improvements
to existing technology. Costs would likely be
Breakthrough Technologies Scenario.
Measures from the Highly Ambitious scenario
plus those that could be achieved with techno-
logical breakthroughs in elds where science
has shown signicant progress.
The size of the gap closure contributed by each
menu item does not necessarily reect the return
per unit of eort. It is more a measure of the
denitional scope of each menu item. For example,
large reductions in food loss and waste (aecting
24 percent of global calorie production) will, by
denition, contribute more than improving produc-
tivity of aquaculture, which only aects 1 percent of
global calorie consumption.
Table 3 | The GlobAgri-WRR 2050 baseline projection and three combination scenarios
Course 1: Reduce growth in demand for food and other agricultural products
Reduce food loss
and waste
Rate of food loss and
waste (24% of calories
globally) maintained in
each country and food
10% reduction in rate of
food loss and waste
25% reduction in rate of
food loss and waste
50% reduction in rate of
food loss and waste
Shift to healthier
and more
sustainable diets
88% increase in demand
for ruminant meat
between 2010 and 2050
as incomes grow across
the developing world
Ruminant meat demand
increases only 69% above
2010 levels, and calories
shift to pulses and soy. This
represents a 10% reduction
in ruminant meat demand
relative to baseline.
Ruminant meat demand
increases only 32% above
2010 levels, and calories
shift to pulses and soy. This
represents a 30% reduction
in ruminant meat demand
relative to baseline.
Same as Highly Ambitious
Avoid bioenergy
competition from
bioenergy for food
crops and land
Crop-based biofuels
maintained at 2010 share
of global transportation
fuel (2.5 percent)
Both food and energy crop-
based biofuels phased out Same as Coordinated Eort Same as Coordinated
fertility rates
UN medium fertility
estimate; global
population 9.8 billion in
UN low fertility estimate in
sub-Saharan Africa; global
population 9.5 billion in
Sub-Saharan Africa fertility
drops to replacement level
by 2050; global population
9.3 billion in 2050
Same as Highly Ambitious
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 57
Course 2. Increase food production without expanding agricultural land
Increase livestock
and pasture
62% growth in beef
output per hectare of
pastureland, 53% growth
in dairy output per
hectare, and 71% growth
in sheep and goat meat
output per hectare
Same as Baseline
Productivity growth is 25%
faster, resulting in 67%
growth in beef output per
hectare, 58% growth in
dairy output per hectare,
and 76% growth in sheep
and goat meat output per
Same as Highly Ambitious
Plant existing
cropland more
5% increase in cropping
intensity between 2010
and 2050 (to 89%)
10% increase in cropping
intensity between 2010 and
2050 (to 93%)
Same as Coordinated Eort Same as Coordinated
Improve crop
breeding to boost
yields 48% increase in crop
yields above 2010 levels
(similar to average linear
rates of yield growth from
1962 to 2006)
Same as Baseline
Crop yields rise to 56%
above 2010 levels (20%
improvement over baseline
growth rate)
Crop yields rise to 69%
above 2010 levels (50%
improvement over
baseline growth rate)
Improve soil
and water
Adapt to climate
Course 3. Protect and restore natural ecosystems and limit agricultural land-shifting
Link productivity
gains with
of natural
Linkage prevents most
agricultural land-shifting
due to yield gains
Same as Baseline Same as Baseline Same as Baseline
Limit inevitable
expansion to
lands with low
opportunity costs
Inevitable land expansion
is limited such that
carbon eects are oset
by the menu item below
Same as Baseline Same as Baseline Same as Baseline
and liberated
agricultural lands
Reforestation of lands
with little agricultural
potential osets carbon
eects of inevitable land
Same as Baseline Same as Baseline
80 Mha of liberated
land fully reforested (to
achieve 4 Gt CO2e/year
585 Mha of liberated land
fully reforested to oset
all remaining agricultural
production emissions
Conserve and
restore peatlands
Annual peatland
emissions stay constant
at 1.1 Gt CO2e between
2010 and 2050
50% reduction in annual
peatland emissions
75% reduction in annual
peatland emissions Same as Highly Ambitious
Table 3 | The GlobAgri-WRR 2050 baseline scenario and three combined scenarios (continued)
Course 4. Increase fish supply
Improve wild
10% decline in wild fish
catch between 2010 and
Wild fish catch stabilized at
2010 level by 2050 Same as Coordinated Eort Same as Coordinated
productivity and
performance of
10% increase in
aquaculture production
eiciencies between
2010 and 2050 across the
50% switch of extensive
pond production to
semi-intensive production,
and 50% switch of semi-
intensive to intensive
Same as Coordinated
Eort, plus 20% increase
in aquaculture production
eiciencies between 2010
and 2050 across the board
Same as Highly Ambitious
Course 5: Reduce greenhouse gas emissions from agricultural production
Reduce enteric
through new
Enteric methane
emissions of 3.4 Gt CO2e
in 2050 (51% above 2010
30% emissions reduction
from half of dairy cows, and
one-quarter of beef cows—
leading to a 9% reduction
in methane emissions from
ruminants (38% growth
above 2010 level)
30% emissions reduction
from all dairy cows, half of
beef cattle, and one-six th
of sheep—leading to an
18% methane emissions
reduction from ruminants
(24% growth above 2010
30% methane emissions
reduction from all
ruminants, including
those permanently grazed
(6% growth above 2010
Reduce emissions
improved manure
Managed manure
emissions of 770 Mt CO2e
in 2050 (31% above 2010
40% reduction in methane
emissions from wet
manure plus 20% reduction
in nitrous oxide emissions
from all manure (14%
growth above 2010 level)
80% reduction in wet
manure emissions plus
20% reduction of all nitrous
oxide emissions (17%
reduction below 2010 level)
Same as Highly Ambitious
Reduce emissions
from manure left
on pasture
Unmanaged manure
emissions from pasture of
653 Mt CO2e in 2050 (46%
above 2010 level)
Same as Baseline
20% reduction of nitrogen
left on pastures for non-
arid systems (31% growth
above 2010 level)
40% reduction in nitrogen
left on pastures for
nonarid systems (15%
growth above 2010 level)
Reduce emissions
from fertilizers
by increasing
nitrogen use
Nitrogen use eiciency
grows from 46% in 2010
to 48% in 2050
57% nitrogen use
eiciency due to a range of
management measures
61% nitrogen use
eiciency due to a range of
management measures
67% nitrogen use
eiciency due to improved
management plus new
Adopt emissions-
reducing rice
management and
Rice methane of 1.3 Gt
CO2e in 2050 (13% above
2010 level)
10% reduction in rice methane
emissions (17% below 2010
level) thanks to new water
management practices and
new rice breeds
Same as Coordinated Eort
Same as Highly Ambitious,
plus 20% faster rate of rice
yield gain (31% reduction
of rice methane below
2010 level)
energy eiciency
and shift to
non-fossil energy
25% decrease in energy
emissions per unit of
output for agriculture
between 2010 and 2050
Same as Baseline
50% decrease in energy
emissions per unit of
agricultural output
between 2010 and 2050
75% decrease in energy
emissions per unit of
agricultural output
between 2010 and 2050
Focus on realistic
options to
sequester carbon
in soils
Soil carbon gains
suicient to assure no
net loss of soil carbon
globally and contribute to
yield gains
Same as Baseline Same as Baseline Same as Baseline
Table 3 | The GlobAgri-WRR 2050 baseline scenario and three combined scenarios (continued)
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 59
Quantitative results of the three combination
scenarios are presented in Table 4. The contribu-
tions of specic menu items are shown in Figures
21–23 for the Breakthrough Technologies scenario
only. For each menu item, its contribution in the
combined scenarios is smaller than its “standalone”
contribution due to interaction between menu
items (e.g., land “savings” attributed to food waste
reductions are smaller if those reductions happen
simultaneously with additional crop yield growth).
All three scenarios substantially reduce the food
gap by reducing the rate of growth in demand for
food. The challenge of increasing crop production
by 56 percent between 2010 and 2050 (baseline) is
reduced to 43 percent, 35 percent, and 29 percent
in the three scenarios, respectively.
The Coordinated Eort scenario reduces agricul-
tural land expansion between 2010 and 2050 by 78
percent. The Highly Ambitious and Breakthrough
Technologies scenarios completely close the land
gap and create the opportunity for signicant refor-
estation on liberated agricultural land.
The hardest gap to close is the GHG mitigation gap
because it is dicult to reduce annual agricultural
production emissions to the 4 Gt CO2e target while
feeding everyone in 2050. Annual production emis-
sions remain at 4.4 Gt even in our Breakthrough
Technologies scenario (Figure 23). Reaching the 4
Gt goal would require major technological advances
as well as full reforestation on at least 80 Mha of
liberated agricultural land.
Furthermore, other analyses have suggested that
to meet the more ambitious 1.5°C warming target
in the Paris Agreement,98 the world will need to
use large quantities of land to oset other sources
of emissions. In our Breakthrough Technologies
scenario, it might be possible to liberate 585 Mha of
agricultural land—after accounting for some expan-
sion of timber plantations and human settlements—
which, if fully reforested, could oset around 4 Gt
of emissions per year for many years.
Plausible pathways toward a sustainable food
future exist, but they will require strong and almost
universal political and social eort. Achieving even
our Coordinated Eort scenario requires revers-
ing a wide range of current trends. Truly realizing
the environmental benets from food demand
reductions and crop and livestock yield gains also
depends on policies that greatly reduce agricultural
land-shifting and protect forests and other natural
Table 4 | Global eects of 2050 combination scenarios on the three gaps, agricultural land use,
and greenhouse gas emissions
Food Gap (2010–50) 62% 56% 43% 35% 29%
Change in
area, 2010–50
Pastureland 2,199 401 128 -390 -446
Cropland 1,066 192 4-180 -355
Total 3,265 593 132 -570 -801
Annual GHG
2050 (Gt CO2e)
production 11.3 9.0 7.4 5.5 4.4
Land-use change 26.9 6.0 1.7 0.3 0.3
Total 38.2 15.0 9.1 5.8 4.7
GHG Mitigation Gap (Gt CO2e) 34.2 11.1 5.1 1.8 0.6
Notes: Numbers may not sum correctly due to rounding. Under the Highly Ambitious and Break through Technologies sc enarios, 0.3 Gt C O2e of ongoing peatland emissions remain,
but total agricultural area declines between 2010 and 2050. We discus s the need to refore st “liberated” agricultural lands to oset agricul tural production emissions on page 59.
Source: GlobAgri-WRR model.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 61
We conclude that three categories of menu items
are particularly important at the global level:
Boosting productivity. The Coordinated Eort
scenario requires faster rates of crop yield growth
than historical rates since the 1960s. Recent yield
trend lines (since the 1980s) are actually slower
than those in our baseline, and far from the addi-
tional yield gains required. Ruminant meat and
milk yield gains in the Coordinated Eort scenario
require massive increases in output per hectare of
pastureland—far greater than the output gains pro-
jected by extending a linear trend from the 1960s.
Shifting diets to reduce demand for rumi-
nant meat. A reduction in ruminant meat
consumption by 30 percent relative to our 2050
baseline—which still results in a 32 percent increase
above 2010 levels—plays a major role in closing the
land and GHG mitigation gaps. We consider it emi-
nently practicable, but the cultural and behavioral
changes required will be challenging.
Reducing food loss and waste. Globally reduc-
ing the rate of food loss and waste by 10, 25, or 50
percent would signicantly close all three gaps.
However, there is little precedent for achieving
such large-scale reductions—particularly because as
countries’ economies develop, food waste near the
consumption side of the food supply chain tends
to grow even as food loss near the production side
Figure 21 | Under the Breakthrough Technologies scenario, the amount of additional food
needed to feed the world in 2050 could be cut by half
Note: Includes all crops intended for direct human consumption, animal feed, industrial uses, seeds, and biofuels.
Source: GlobAgri-WRR model.
Reduce growth in demand for food
and other agricultural products
Increase food
agricultural land
(Base year)
level fertility rates
Phase out crop-
based biofuels Shift diets
Reduce food
loss and waste
Crop production (trillion calories per year)
Figure 22 | Under the Breakthrough Technologies scenario, the area of land needed for agriculture could
shrink by 800 million hectares, which would be liberated for reforestation
Source: GlobAgri-WRR model.
Reduce growth in demand for
food and other agricultural products
Increase food production
without expanding
agricultural land
Net agricultural land expansion (Mha) (2010–50)
food loss
and waste
Phase out
level fertility rates
crop yields
Plant existing
cropland more
Improve wild
fish supply
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 63
Figure 23 | Under the Breakthrough Technologies scenario, agricultural greenhouse gas emissions
would fall dramatically but reforestation and peatland restoration would be necessary to
meet the target of 4 gigatons per year
Reduce food
loss and waste
Shift diets
Phase out crop-
based biofuels
level fertility rates
crop yields
Improve manure management
Improve rice management and breeds
Reduce energy emissions
Restore peatlands
80 Mha
of reforestation
fish supply
Reduce GHG
emissions from
Protect and
restore natural
Agricultural GHG emissions (production + land-use change), Gt CO2e/year (2050)
Reduce enteric fermentation
Reduce emissions
from manure
left on pasture
Increase nitrogen use eiciency
Increase pasture
Plant existing
cropland more
frequently Improve wild
management Increase
Reduce growth in
demand for
food and other
agricultural products
Increase food
agricultural land
585 Mha
of reforestation
Note: Solid areas represent agricultural production emissions. Hatched areas represent emissions from land-use change.
Source: GlobAgri-WRR model.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 65
The menu items for a sustainable food future, described
and analyzed in our five courses, focus heavily on technical
opportunities. However, menu items cannot be implemented in
isolation, and they are all subject to a variety of cross-cutting
public and private policies. In addition to reducing demand growth
and boosting productivity, policies must reduce rural poverty by
helping smallholder farmers become more market-oriented, even
as many of them inevitably shift toward alternative employment.
Boost Productivity and
Reduce Rural Poverty
Higher rural incomes will depend on raising the
productivity of smallholder farmers and linking
them to more lucrative markets. Progress will
require giving farmers greater security to invest in
producing marketable products where they have a
comparative advantage, securing their land rights,
and rectifying historical disadvantages confronting
women farmers.
Allow farms to grow based on market and
social forces, but avoid large or government-
facilitated acquisitions unless they involve
existing large farms
About 80 percent of the world’s farms are small (less
than 2 hectares), but they occupy only about 12 per-
cent of the world’s agricultural land.99 The number
of small farms is growing, as families subdivide their
land and some farms become too small to supply a
full-time livelihood. Small farmers face real obstacles
in the form of limited access to capital for productiv-
ity improvements, diculties in meeting the tight
sanitary and quality standards required by super-
market chains, and poverty traps that force farmers
to sell critical assets in tough times. Yet in Africa and
Asia, studies have consistently found small farms to
be more productive per hectare than larger farms100
(though sometimes not the largest farms).101 More
successful small farms tend to have market access
and opportunities for o-farm employment to
supplement income generated by the farm. Overall,
government policies should not force small farms
to consolidate or encourage large farms to take
over small farms, but neither should they ght the
autonomous growth of farms.
This less interventionist policy should also apply to
large land acquisitions, which some governments
have encouraged in recent years. The size and
number of such deals is hard to track, and some
early estimates were too high. The best information
is that international investors acquired ownership
or long-term leases for 44 Mha of land between
2000 and 2016 and are in some stage of agreement
to acquire another 18 Mha.102 Major domestic inves-
tors are also acquiring large tracts of land. Although
some acquisitions claim to be focused on supplying
food for domestic markets, bioenergy production
and to a lesser extent producing food destined for
foreign markets motivated many transactions.
Acquisitions of preexisting large farms, includ-
ing abandoned large farms in much of the former
Soviet Union, raise few social or environmental
concerns. In other parts of the world where small
farms predominate, careful analyses have shown
mostly adverse consequences for local people
despite a variety of claimed social protections.103
Large acquisitions often involve land that is not
intensively cropped but is used by local people for
grazing, shing, and long-rotation cropping. Many
acquisitions of forested or other wooded land and
wetlands are, in eect, forms of environmentally
harmful agricultural expansion.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 67
Move toward more equitable and secure property
rights, and facilitate cash crop production through
cooperative and contract farming
In much of the world, farmers and forest dwell-
ers lack the secure, registered titles to property
that are common in Europe and North America.
Many researchers, international aid agencies, and
nonprot organizations have long advocated for
stronger recognition of property rights to protect
farmer interests and sometimes increase access
to credit. However, as the World Bank found in
2008, eorts to shift to Western-style property
rights “were often adopted less to increase eciency
than to further interests of dominant groups” and
resulted in greater land consolidation and inequali-
ty.104 Many scholars also found that shifting to more
ocial land titling in Africa often did not result in
productivity increases, in part because customary
rights are more secure than previously thought.
Recent eorts have emphasized recognition of cus-
tomary rights, including shared rights to use land
and trees, and the need for formalization of rights
to correct historical inequities, such as the exclusion
of women from ownership of or decision-making
about land. Information technology has reduced the
physical diculties of mapping and registering land
but governments need to reduce the bureaucracy
and legal obstacles that still block the award of
community rights in many countries.105
Governments also should establish legal frame-
works and basic social security systems that make
it easier for small farmers to raise high-value cash
crops through contract or cooperative farming.106
Farmers can overcome the challenges of small size
by linking, in various forms, to larger organizations
through contracts or farmer associations. Such
arrangements oer potential advantages of brand-
ing, access to expertise, shared or lower costs for
inputs, and access to more specialized markets.
But the costs can sometimes include lower prices
imposed by a local monopoly, unfair or inecient
cooperative management, and potential cheating by
either party to the contract as prices uctuate. As
a result, these systems tend to focus on high-value
food or other cash crops, where high quality or
reliability is rewarded by the market.107 Developing
country climates often favor such crops, but, while
they are potentially more protable, specializa-
tion can also increase risk due to disease, changing
market conditions, or dishonest purchasers. Legal
frameworks that fairly enforce contract farming
and support basic incomes for rural workers can
help small farmers focus on growing more lucrative
products, raise incomes, and possibly help reduce
global land demands.
Climate Policy and Agriculture
Because no government can know the relative
feasibility of all actions to reduce GHG emissions,
sound, cost-eective climate policies typically
mandate outcomes rather than technologies or
The role of carbon pricing, emissions caps, and
carbon osets
Many governments and economists favor putting
a price on emissions via a carbon tax or cap-and-
trade system. Each emitting source then has an
incentive to reduce emissions as cheaply as pos-
sible. Unfortunately, applying this approach to
land-use and agriculture faces practical problems.
Monitoring emissions from diverse and diuse
sources is not feasible (unlike tracking the quantity
of coal or oil burned). Also, it is often not feasible
to determine which changes in soil or forest carbon
are caused by the landowner, the weather, or the
actions of others.
Creative pricing programs therefore have to be
designed for features of the agricultural system that
can be measured. For example, governments could
impose a tax on fertilizers that do not incorporate
a nitrication inhibitor or time-release mechanism
(assuming these alternative fertilizers are avail-
able to farmers). The tax level would be based on
the dierence in emissions expected from use of
conventional versus improved fertilizer. Dierent
forms of manure management could also be taxed
dierentially. Retail food taxes on high-emissions
foods would also help internalize climate costs and
could be oset with subsidies for low-emissions
For many years, climate mitigation policies for agri-
culture focused on their potential to provide osets
to mitigate fossil energy emissions. The hope was
that energy users could pay farmers to reduce emis-
sions more cheaply than they could themselves, and
that these practices would, in turn, help farmers
boost productivity and resilience. The Clean Devel-
opment Mechanism recognizes a small number of
agricultural practices as osets,108 and the Canadian
province of Alberta recognizes many more in a local
Unfortunately, agricultural osets mean that reduc-
tions in the agricultural sector are credited to the
energy sector and allow more emissions there, so
they cannot count as agricultural mitigation. How-
ever, stringent climate goals require sharp reduc-
tions of emissions in both sectors. Agricultural
osets also present large administrative challenges.
By denition, an oset must require “additional”
reductions in agricultural emissions, but additional-
ity is subject to uncertainty. The more cost-eective
the mitigation, the greater the likelihood that it
would occur anyway and therefore not be addi-
tional. Leakage and monitoring requirements are
signicant, and small farmers often cannot aord
to invest money to reduce emissions up-front while
waiting to be paid only after they have reduced
emissions or sequestered carbon. As discussed
above, soil carbon sequestration turns out to be
harder and more uncertain than expected. In the
short term, some osets could stimulate progress
in the land sector but, at best, osets have a limited
and temporary role to play in achieving a sustain-
able food future.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 69
Market price supports
(subsidies from consumers)
Safety, health, and inspection
Input subsidy
Conservation, production retirement,
and other public goods
Research, education, and
technical assistance
Other production support
Production payments
100% =
$598 Billion
Redirect government farm support and
attract climate funding
Farming, like any other large industry, requires
major investments. The direct investments made by
farmers with their own funds or by other domestic
private investors account for the overwhelming share
of agricultural investment.110 Policies that facilitate
and guide this private investment are therefore more
important than direct public funding.
Yet public funding is still important. In 2014–16,
public support for agriculture averaged $600 billion
per year in countries assessed by the OECD (Fig-
ure 24). Half of this total takes the form of market
interventions that raise prices to consumers, such
as import barriers, taris, or systems that limit
production by farmers to increase prices. Because
these supports are more prevalent in higher-income
countries, they oer little market protection for
the world’s poor. Few of these funds support the
menu items identied in this report. From a global
perspective, reducing or redirecting the costs of
these market interventions would reduce prices and
benet consumers. Many farmers in lower-income
countries do, however, benet from large fertilizer
subsidies. The environmental benets of reduc-
ing overapplication of cheap fertilizers are clear in
countries where fertilizers are heavily overused.
In much of Africa, where soils are nutrient-poor,
subsidized fertilizers have achieved only modest
yield gains or helped to reduce poverty, though at
signicant cost to government budgets.
Over the years, countries have reduced some
market barriers and linked subsidies to very modest
conservation requirements, but funding could do
much more to support the interventions needed
for a sustainable food future. The most promising
examples involve government support for multi-
partner research projects to promote new farming
Developed countries have promised to provide
$100 billion per year by 2020 to developing coun-
tries for climate mitigation and adaptation. To
date, they are not on track to meet this goal.111 For
example, as of September 2017, only $10 billion
had been committed to the Green Climate Fund.112
The money is for all climate-related work. If agri-
culture is to earn its fair share of climate funding,
countries are going to need to demonstrate they are
implementing detailed, scientically based plans
to mitigate emissions, identify the specic farming
changes that will be implemented on specic types
of farms, and the likely gains that will result.
Note: OECD as sessment of 51 countries excluding India.
Source: WRI analysis of OE CD (2016) data.
Figure 24 | The world’s leading agricultural producers provided nearly $600 billion in public funding to
support farms in 2015
Boost Research and Development
Total public and private spending on agricultural
research and development (R&D) in 2007–8 (the
latest year for which we found global data) was
roughly $50 billion.113 Today, the world probably
devotes only around 1.4–1.7 percent of agricultural
GDP to agricultural R&D.114 For a sector that is so
sensitive to constantly changing environmental
conditions and in which massive growth is required,
we consider this amount to be inadequate. We iden-
tify two key research and development themes.
Development: The world needs to commit far
more resources to the D in R&D for agricultural
emissions reduction. “Development” involves the
critical technical analyses concerning how to apply
research breakthroughs. Researchers know how
to draw down water in rice farms and can roughly
estimate the emissions reductions, but they do not
know which specic irrigation districts have suf-
ciently reliable water supplies to make drawdowns
feasible. Researchers know that improved feeds and
health care will increase the productivity of rumi-
nant livestock, but, outside of developed countries,
they have only the roughest proxy estimates for
how livestock systems work and how they can be
improved. Countries have not developed scienti-
cally based land-use plans for targeting agricultural
expansion where it is inevitable. Governments and
international institutions should fund this kind of
Indispensable innovations: We identify mul-
tiple examples of technological progress or break-
throughs that are either indispensable or would
be enormously helpful in achieving a sustainable
food future (Table 5). A few, such as the pursuit of
plant-based meat substitutes appetizing to meat
eaters, can probably be left mostly to the private
sector, but others will require public investment.
Governments can also spur innovation by funding
pilot projects—particularly large-scale pilots—to use
innovative technologies, and by enacting laws to
require the use of innovations if they prove eective
and cost-ecient.
Table 5 | Critical research needs for breakthrough technologies
Course 1: Reduce growth in demand for food and other agricultural products
Reduce food loss and
Development of inexpensive methods
to prevent decomposition without
Companies are investigating a variety of compounds. For example,
Apeel Sciences, a small California start-up, has an array of
extremely thin spray-on films that inhibit bacterial growth and
hold water in.
Shift to healthier and
more sustainable diets
Development of inexpensive, plant-
based products that mimic the taste,
texture, and experience of consuming
beef or milk
The private sector is making significant investments in various
plant-based substitutes, including imitation beef using heme that
appears to bleed like real meat, and synthetic milk generated from
proteins produced by yeasts.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 71
Course 2: Increase food production without expanding agricultural land
Increase livestock and
pasture productivity
Breeding of better, high-yielding forage
grasses that can grow in “niche”
production areas
In much of Africa and Asia, with limited land available, quality
forage for cattle depends on producing high-quality grasses and
legumes in restricted land areas, such as underneath forest or
banana plantations.
Improve crop breeding to
boost yields
Breeding of cereals to withstand higher
peak temperatures
Recent research has shown that peak temperatures, particularly
at critical growth periods, can greatly restrict cereal yields, and
that climate change may push temperatures to exceed peak
Course 4: Increase fish supply
Improve productivity
and environmental
performance of
Development of fish oil substitutes from
microalgae, macroalgae (seaweeds), or
oil seeds for aquaculture feeds
Research groups have initial breeds of rapeseed containing
oils nutritionally equivalent to fish oils and promising seaweed
varieties. Work is also proceeding on producing algae more
Course 5: Reduce GHG emissions from agricultural production
Reduce enteric
fermentation through
new technologies
Finding feed compounds, drugs, or
breeds that lower methane emissions
from cows, sheep, and goats
Several research groups are working on feed compounds to
reduce methane emissions. After years without promising results,
a private company has claimed 30 percent reductions for a cheap
compound that does not appear to have significant health or
environmental side eects.
Reduce emissions
through improved manure
Development of lower-cost ways to
dry and consolidate manure, stabilize
nutrients to reduce methane and
nitrous oxide emissions, and make
them easier to use eiciently with crops
Technologies exist to dry manure and to turn it into energy, but
costs and leakage rates reduce viability and greenhouse gas
reduction benefits.
Reduce emissions from
manure left on pasture
Breeding of traits into pasture grasses
to inhibit formation of nitrous oxide or
developing safe, ingestible nitrification
inhibitors for livestock
Researchers have discovered one variety of Brachiaria that
significantly inhibits nitrification and thus nitrous oxide formation.
Reduce emissions from
fertilizers by increasing
nitrogen use eiciency
Development of more eective, lower-
cost, and integrated compounds, such
as improved nitrification inhibitors to
reduce nitrogen losses associated with
fertilizer use, and breeding nitrification
inhibition into crops
Various compounds exist and appear to be eective, but
improvements should be possible, including more tailored
understanding of which compounds are most eective under
which precise conditions. Moreover, researchers have now
identified traits to inhibit nitrification in some varieties of all major
grain crops that can be built upon through breeding.
Adopt emissions-
reducing rice
management and
Development of rice varieties that emit
less methane
Researchers have shown some common rice varieties emit less
methane than others and have bred one experimental rice that
reduces methane by 30 percent under scientifically controlled
conditions although its eects on yields are unknown.
Note: This table is not inte nded to be exhaustive and does not include all courses or menu items.
Source: Authors.
Table 5 | Critical research needs for breakthrough technologies (continued)
Avoid Double Counting of
Land and Biomass
Some of our menu items dier from other research-
ers’ recommendations where we believe those
recommendations are based on analyses that inap-
propriately count the same land or plant material
(biomass) twice. In other words, some other analy-
ses assume that the same land or biomass required
to meet one set of needs is simultaneously available
to meet another.
Prominent examples include bioenergy from food or
energy crops grown on dedicated land. Many analy-
ses assume that “potential cropland” or “marginal
cropland” can be used to produce bioenergy without
recognizing their current carbon storage and biodi-
versity values as forest or savannas, or their current
food production function as grazing land. Model-
ers who estimate large potential climate benets
from “bioenergy with carbon capture and storage”
(BECCS) rely on the same estimates of biomass
potential that are based on double counting.115 Other
analyses assume yield gains could be used to free up
land for bioenergy without clearing more forests and
savannas—even as those same yield gains are needed
just to meet rising food demand. In claiming GHG
savings from bioenergy, analyses often attribute the
carbon absorbed by plant growth as an oset for
burning biomass even when this plant growth would
otherwise have occurred and removed carbon from
the atmosphere anyway (Figure 25).
Large estimates of reforestation potential often
make similar mistakes in treating grazing land as
available to reforest without cost to food produc-
tion, or regarding potential increases in crop and
pasture productivity as automatically liberating
land for reforestation without recognizing that this
potential must rst be exploited just to meet grow-
ing food needs. Although important restoration
opportunities exist on peatlands and unimprovable
grazing land, large-scale reforestation will require
signicant reductions in demand growth and his-
torically unprecedented increases in yields.
Some soil carbon sequestration estimates also dou-
ble-count by assuming that biomass (as manure,
crop residues, or mulches) can be used as a soil
amendment when it is already in use—even if only
to store carbon—somewhere else. Other estimates
count the benets of reducing grazing pressure
without counting the costs of replacing the forgone
meat and milk.
A common theme appears to be a failure to rec-
ognize that land is a xed and therefore limited
resource. The only ways to meet growing human
demands for both food and carbon storage are to
use land more eciently and to consume agricul-
tural products more eciently.
Note: In scenario A, shifting f rom gasoline to ethanol use reduces emissions
through addi tional uptake of carb on on land that previously did not grow plants.
In scenario B, which is the t ypical bioenergy scenario, t he shift from gasoline to
ethanol does not reduc e emissions, as the demand for bioenergy merel y diverts
plant grow th (e.g., maize) that would have occurre d anyway.
Source: Searchinger and Heimlich (2015).
Figure 25 | Why greenhouse gas reductions from
bioenergy require additional biomass
Existing crop growth absorbs
carbon and is used for food...
Unproductive land
goes unused...
New crop growth absorbs carbon
and is converted to ethanol...
Existing crop growth absorbs carbon
and is converted to ethanol...
...ethanol is
used for car fuel
...ethanol is
used for car fuel
...while gasoline is used
for car fuel
...while gasoline is used
for car fuel
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 73
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 75
Creating a sustainable food future—simultaneously
feeding a more populous world, fostering development
and poverty reduction, and mitigating climate change and
other environmental damage—presents a set of deeply
intertwined challenges. Our report oers several insights
that dier in direction or emphasis from much prior work.
Productivity gains are critical. Productivity
gains in land, animals, and chemical inputs already
in our baseline are responsible for closing two-
thirds of the GHG mitigation gap and more than
80 percent of the land gap that would exist absent
productivity gains after 2010. When adding in the
various additional productivity gains required to
meet our 4 Gt CO2e/year emissions target by 2050,
the role of productivity gains grows even larger.
Productivity gains also provide the most important
potential synergy between income, food security,
and environmental goals. Thus, new molecular
crop-breeding methods will need to be exploited.
Moreover, every hectare of global pasture that is
capable of and appropriate for sustainable intensi-
cation must fully exploit its potential to increase
milk or meat output severalfold.
Slowing demand growth is critical too.
Despite the major contribution that productivity
gains can make to closing our three gaps, they will
not be enough. The largest diet-related opportunity
lies in limiting the global growth in demand for
beef, as well as sheep and goat meat. A 30 percent
global shift from ruminant meat to other foods—
achieved by the world’s highest consumers reducing
their consumption by roughly 40 percent relative to
2010 levels—would, by itself, nearly close the land
gap and halve the GHG mitigation gap. A 10 percent
shift from all animal-based foods by the world’s
wealthy would benet human health and open up
space for the great majority of poorer consumers to
modestly increase their consumption. Moving more
rapidly toward replacement-level fertility rates in
sub-Saharan Africa would greatly reduce the risks
of hunger in the region, provide multiple social
and economic benets, and reduce environmental
challenges. Global plans to greatly increase the use
of modern bioenergy derived from energy or food
crops grown on land dedicated to that purpose,
however, would make a sustainable food future
Innovation in farm management will also be
necessary to mitigate emissions. To imple-
ment management measures known to reduce
emissions, governments need to develop systems
to analyze mitigation potential and track progress
across their agriculture sectors, increase incentives,
and phase in mandatory performance standards.
To stimulate promising management innovations,
governments need to boost R&D, and encourage the
private sector by requiring that farms use innova-
tive technologies when those technologies demon-
strate cost-eective mitigation.
Productivity gains must be linked to protec-
tion of carbon-rich ecosystems. Shifting of
agricultural land both among and within regions
presents a major carbon and biodiversity challenge.
Governments therefore must make eorts to avoid
such shifts and place more emphasis on reforesting
abandoned agricultural land to natural forests when
shifts do occur. Because productivity gains can
sometimes encourage land-shifting, ensuring that
yield gains protect forests and other carbon-rich
and biodiverse ecosystems requires that govern-
ments and private parties explicitly link eorts to
boost yields with protection for those ecosystems
through nancing, lending conditions, supply
chain commitments, and public policies. The forest
frontier should be closed to agriculture wherever
feasible. New roads must also be located in ways
that minimize the incentives to convert natural
areas to agriculture.
Reforestation of some lands, and restora-
tion of peatlands, should proceed immedi-
ately, but larger-scale reforestation depends
on technological innovation and changes in
consumption patterns. Marginal agricultural
lands that cannot realistically be intensied are
appropriate for reforestation right now. However,
the scale of reforestation necessary to fully achieve
climate goals requires that much more land be
liberated from agriculture. Freeing up hundreds of
millions of hectares of land can only be achieved
through highly successful implementation of the
measures proposed in our demand-reducing and
productivity-boosting menu items.
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 77
Regulation and technological innovation
will be essential to achieve the most ambi-
tious levels of our menu items. Regulations
must be crafted to spur innovation while allowing
exibility to develop cost-eective solutions. They
should apply mostly to manufacturers of agri-
cultural inputs and to managers of concentrated
livestock facilities. Many categories of technological
innovations are needed but promising options have
emerged for menu items in all our courses. Govern-
ments will need to give far more weight to R&D and
encourage the private sector with a range of policy
We believe that the challenge of sustainably feed-
ing nearly 10 billion people by 2050 is greater than
commonly appreciated. Growth in food demand is
high due to population growth and the rapid rise of
a global middle class. The strength of competition
for land, particularly pastureland, has often been
underestimated. Proposed land-use solutions often
involve double counting, and the climate implica-
tions of land-shifting are not fully recognized.
Sub-Saharan Africa presents unique and formidable
challenges because of the region’s high population
growth and low agricultural yields.
Despite the challenges, we believe that a sustainable
food future is achievable. Our menu proposed in
this synthesis report can create a world with su-
cient, nutritious food for everyone. It also oers the
chance to generate the broader social, environmen-
tal, and economic benets that are the foundation
of sustainable development. But such a future will
only be achieved if governments, the private sector,
and civil society act upon the entire menu quickly
and with conviction.
1. We calculate the food gap measured in crop calories at 56%.
The growth in demand for animal-based foods is calculated
dierently (i.e., growth in demand for all food calories, includ-
ing animal- and plant-based foods) and estimated at 68%.
Overall, we estimate the total food gap at 55%. Because the
“crop calorie gap” and the “food gap” are so similar, we use the
terms interchangeably in this report.
2. UNDESA (2017). The figure of 9.8 billion people in 2050 reflects
the “medium fertility variant” or medium population growth
scenario (as opposed to the low-growth and high-growth
scenarios published by the United Nations Department of
Economic and Social Aairs).
3. To project food demands in 2050, we start with a 2012 FAO
projection of the diets that the average person in each country
will consume in that year (Alexandratos and Bruinsma [2012]).
FAO based its projections on economic growth and income
trends and culture in dierent countries. We adjust these
FAO projections moderately, adding fish consumption and
including enough additional calories in sub-Saharan Africa
and South Asia to ensure suicient nutrition for everyone, after
accounting for waste and unequal distribution. More specifi-
cally, we adjusted diets to assure food availability of 3,000
kcal per person per day in sub-Saharan Africa and South Asia
by proportionately scaling up all food items in the FAO 2050
projections. “Food availability” is food available to consumers
excluding postconsumer waste. The total quantity of calories
available must be adequate to feed all individuals after ac-
counting first for this food waste and second for the unequal
distribution of food, which means that many individuals will
consume less than the regional average. We based the 3,000
kcal/person/day on a recognition that once regions reach this
level of food availability, they have low levels of food insecurity.
Additionally, the United Nations has added more than half a
billion people to its medium-level estimate of the global popu-
lation in 2050 compared to the scenario used by FAO in 2012,
so we further adjust 2050 food demands upward to reflect the
new estimate of 9.8 billion people.
4. See, e.g., Holt-Gimenez (2012); Bittman (2013); and Berners-Lee
et al. (2018).
5. Figures exclude Antarctica. FAO (2011a).
6. Alexandratos and Bruinsma (2012), Table 4.8. FAO data estimate
an increase in arable land in use of 220 Mha from 1962 to 2006.
According to FAO (2017a), pasture area has increased by 270
Mha since 1962.
7. GlobAgri-WRR’s estimates of agricultural production emissions
in 2050 employ a variety of calculations and assumptions
based on our best estimates of trend factors wherever pos-
sible, which we describe more fully in Course 5. Some studies
include emissions from regular human burning of savannas
and grasslands, but we do not because these systems burn
naturally on occasion and we consider any increase in emis-
sions due to human eorts too uncertain. GlobAgri-WRR does,
however, consider a smaller set of emissions from the burning
of crop residues.
8. This estimate is based on the GlobAgri-WRR model.
9. See Figure 18 for a more detailed breakdown of production
emissions estimated by GlobAgri-WRR. It excludes down-
stream emissions from the entire food system in processing,
retailing, and cooking, which are overwhelmingly from energy
use and must be addressed primarily by a broader transforma-
tion of the energy sector.
10. This figure is based on an estimate of 5 Gt of CO2e emissions
per year from land-use change in recent years. It attempts
to count carbon losses from the conversion of other lands to
agriculture, or conversion of grasslands to cropland, the car-
bon gains from reversion of agricultural land to forest or other
uses, and the ongoing losses of carbon due to degradation of
peat. Because it is impossible to estimate land-use-change
emissions with data from a single year, we do not choose
to pinpoint a specific year for these emissions but instead
treat them as a typical rate from recent years. In reality, it is
not possible to generate a precise estimate of these num-
bers because it is not possible to track each hectare of land
globally and its carbon changes from year to year. There is a
large dierence between gross and net losses, and assump-
tions must be made about rates of carbon gain and loss from
land-use change. In addition, many of these data are based on
national reporting of net changes in forest area, which there-
fore assume carbon losses only on the net dierence in each
country where they occur and carbon gains from net gains in
forest where that occurs. This calculation cannot capture the
real net losses because the losses in areas losing forest are
unlikely to be dierent (and are often higher) than the gains
from regenerating forests.
In earlier reports in this series, we estimated emissions from
land-use change at 5.5 Gt CO2e on the basis of an average
from other estimates found in UNEP (2012); FAO (2012a); and
Houghton (2008). These estimates included losses from 2000
to 2005, a period for which FAO’s Forest Resources Assess-
ment (FRA) estimated heavy declines in forest. Several more
recent papers have reduced estimates of deforestation and
therefore emissions. Smith et al. (2014) estimate 3.2 Gt CO2e/
year in 2001–10 including deforestation (3.8 Gt CO2e/year),
forest degradation and forest management (-1.8 Gt CO2e/year),
biomass fires including peatland fires (0.3 Gt CO2e/year), and
drained peatlands (0.9 GtCO2e/year). Another paper estimates
3.3 Gt of CO2 equivalent from land-use change in 2011 but does
not include drained peatland (Le Quéré et al. 2012). Federici
et al. (2015), who based their estimates on FAO’s 2015 FRA,
calculated emissions from net deforestation at 2.904 Gt CO2e/
year from 2011 to 2015 but also suggested that this figure was
SYNTHESIS REPORT: Creating a Sustainable Food Future: A Menu of Solutions to Feed Nearly 10 Billion People by 2050 79
likely 30% too low due to failure to count carbon in some for-
est pools, which would increase the figure to 3.78 Gt CO2e/year.
FAO also estimated peatland emissions separately of 0.9 Gt
CO2e/year to the IPCC, leading to a recent FAO estimate of 4.7
Gt/year (Federici et al. 2015). Our peatland emissions estimate
of 1.1 Gt CO2e/year also includes fire. Federici et al. (2015) also
reported a large increase in “forest degradation,” which is due
principally to logging and other nonagricultural activities, and
which we do not discuss here. For a summary of the uncer-
tainties and methods, see Searchinger et al. (2013).
11. See Figure 17 for assumptions about changes in baseline emis-
sions from agricultural production,