Article: Global Diets Link Environmental Sustainability and Human Health
Article: Global Diets Link Environmental Sustainability and Human Health
Article: Global Diets Link Environmental Sustainability and Human Health
doi:10.1038/nature13959
Diets link environmental and human health. Rising incomes and urbanization are driving a global dietary transition in
which traditional diets are replaced by diets higher in refined sugars, refined fats, oils and meats. By 2050 these dietary
trends, if unchecked, would be a major contributor to an estimated 80 per cent increase in global agricultural greenhouse
gas emissions from food production and to global land clearing. Moreover, these dietary shifts are greatly increasing the
incidence of type II diabetes, coronary heart disease and other chronic non-communicable diseases that lower global life
expectancies. Alternative diets that offer substantial health benefits could, if widely adopted, reduce global agricultural
greenhouse gas emissions, reduce land clearing and resultant species extinctions, and help prevent such diet-related
chronic non-communicable diseases. The implementation of dietary solutions to the tightly linked dietenvironment
health trilemma is a global challenge, and opportunity, of great environmental and public health importance.
for 100 of the worlds more populous nations to analyse global dietary
trends and their drivers, then use this information to forecast future
diets should past trends continue. To quantify effects of alternative diets
on mortality and on type II diabetes, cancer and chronic coronary heart
disease, we compile and summarize results of studies encompassing ten
million person-years of observations on diet and health. Finally, we
combine these relationships with projected increases in global population to forecast global environmental implications of current dietary
trajectories and to calculate the environmental benefits of diets associated with lower incidences of chronic non-communicable diseases.
Department of Ecology, Evolution and Behavior, University of Minnesota, St Paul, Minnesota 55108, USA. 2Bren School of Environmental Science and Management, University of California Santa Barbara,
California 93106, USA.
0 0 M O N T H 2 0 1 4 | VO L 0 0 0 | N AT U R E | 1
RESEARCH ARTICLE
6
5
4
Mediterranean diet
Pescetarian diet
Sugar+oil
Fruit+vegetable Dairy+egg
Fish
Livestock
Cereals
Sugar+oil
Fruit+vegetable Dairy+egg
Fish
Livestock
300
250
200
150
100
50
70
60
50
40
30
20
10
0
NA NA NA NA NA NA
NA
M
ai
W ze
he
O
a
th
er Ri t
ce ce
re
a
Su ls
ga
Te
r
m Oil Oil
pe cr s
Tr rate ops
op f
ic rui
Ve al fr ts
ge uit
ta s
bl
R es
Le oo
gu ts
m
e
Bu s
N
N
tte
o
non
Eg r
tra
-r
ec
gs
w
Re ircu Tra ling Da
ci lat wl fis iry
rc in in h
ul g g er
at aq fis y
in u h
g a er
aq cu y
ua ltu
cu re
lt
Po ure
Ru
ul
tr
m
in P y
an o
t m rk
ea
t
Cereals
350
Vegetarian diet
1
0
Omnivorous diet
also differ. For instance, among cereal grains, wheat has a fifth the GHG
emissions per g protein of rice (t-test comparison: P 5 0.002).
Finally, to understand its environmental impacts, it is important to
know the nutritional needs that a food meets and how much is consumed to do so. Fruits and vegetables are important sources of micronutrients, antioxidants and fibre. Unlike root crops and legumes, which
are calorie-dense or protein-dense, most vegetables are not primarily
consumed for calories or protein and should be evaluated by emissions
per serving. For instance, 20 servings of vegetables have less GHG emissions than one serving of beef (Fig. 1b). However, fish and meats, which
are high in protein, are also nutritionally dense foods that provide essential fatty acids, minerals and vitamins28,29, and can have relatively low
GHG emissions if eaten in moderation. Finally, the nutritional value of
some foods can depend on how they are produced. For instance, in comparison to grain-fed cattle, grass-fed beef and dairy have nutritionally
superior fatty acid and vitamin content30.
2 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 4
a
2009
25
1961
20
15
1961
10
1961
2009
2009
Economic group
A
B
C
D
E
F
India
China
India
0
5,000
15,000
25,000
1,400
2009
1,200
1961
1,000
800
2009
China
600 1961
400
2009
200
0
1961
5,000
15,000
30
ARTICLE RESEARCH
3,500
2009
3,000
1961
2009
2,500
2,000
1,500
25,000
1961
5,000
15,000
25,000
1961 to 2009 for India, China, and six economic groups containing 98 other
nations (Extended Data Table 4). Fitted curves were used to forecast 2050
income-dependent demand.
have different compositions. A vegetarian diet consists of grains, vegetables, fruits, sugars, oils, eggs and dairy, and generally not more than
one serving per month of meat or seafood. A pescetarian diet is a vegetarian diet that includes seafood. A Mediterranean diet is rich in vegetables, fruit and seafood and includes grains, sugars, oils, eggs, dairy
and moderate amounts of poultry, pork, lamb and beef. Omnivorous
diets, such as the 2009 global-average diet and the income-dependent
2050 diet, include all food groups.
Relative to conventional omnivorous diets, across the three alternative diets incidence rates of type II diabetes were reduced by 16%41%
and of cancer by 7%13%, while relative mortality rates from coronary
heart disease were 20%26% lower and overall mortality rates for all causes
combined were 0%18% lower (Fig. 3). This summary illustrates the
magnitudes of the health benefits associated with some widely adopted
alternative diets. The alternative diets tend to have higher consumption
of fruits, vegetables, nuts and pulses and lower empty calorie and meat
consumption than the 2009 average global diet and the 2050 incomedependent diet (Extended Data Fig. 1). Our analyses are not designed
to compare the health impacts of the three alternative diets with each
other, nor to imply that other diets might not provide health benefits
superior to these three diets. Indeed, the reported impacts of individual
foods, such as deleterious impacts from sugars40 and processed meats19,22,
and benefits from nuts and olive oil41, suggest that variants of these three
diets may offer added health benefits, as may other diets.
50
Type II diabetes
Cancer
Coronary
mortality
All-cause
mortality
40
30
20
10
ed
ite
rra
Pe nea
n
sc
et
ar
ia
Ve
n
ge
M
t
ar
ed
ite ian
rra
Pe nea
n
sc
et
ar
ia
Ve
n
ge
M
t
a
ed
ria
ite
n
rra
n
ea
Pe
n
sc
et
ar
ia
Ve
n
ge
M
t
a
ed
ria
ite
n
rra
n
ea
Pe
n
sc
et
ar
ia
Ve
n
ge
ta
ria
n
400 4
200
0
200
500
400
300
200
100
ed
i
In
co
m
ed
ep
en
d
600
en
t
20
50
te
rra
ne
an
Pe
sc
et
ar
ia
n
Ve
ge
ta
ria
n
ed
i
ep
en
d
ed
co
m
In
Fruits/vegetables
Nuts/pulses
Roots
Cereals
Empty calories
en
t
ep
2
en
d e 009
nt
M
2
ed
ite 050
rra
Pe nea
n
sc
et
ar
ia
Ve
n
ge
ta
ria
n
Ruminant
Poultry
Pork
Seafood
Dairy/eggs
600 3
700
Pe
sc
et
ar
ia
n
Ve
ge
ta
ria
n
0.5
0.0
te
rra
ne
an
0.5
800 2
20
50
te
rra
ne
an
Pe
sc
et
ar
ia
n
Ve
ge
ta
ria
n
1.0
1,000
ed
i
100
2050 production
emissions relative
to 2009
200
1.5
300
co
m
In
2.0
400
ed
RESEARCH ARTICLE
such as lines 1-4, have fairly parallel trends) (c). d, 2050 global cropland
reductions from alternative diets relative to income-dependent diet. The box
and whisker plots (c, d) show mean (centre line) and percentiles below (2.5th,
10th, 25th) and above it (75th, 90th, and 97.5th) based on 243 scenarios.
and of each alternative diet (Fig. 4d). Across these scenarios, the incomedependent diet requires from 370 to 740 million hectares more cropland than the alternative diets, and averages 540 million hectares more
(Fig. 4d). These results suggest that shifts towards healthier diets could
substantially decrease future agricultural land demand and clearing, as
could improvements in the five factors (Extended Data Table 6). Land
clearing also leads to GHG emissions. Clearing 540 million hectares
from 2010 to 2050 would release about 0.6 Gt yr21 of CO2-Ceq.
In addition to dietary shifts, other changes will be needed for agriculture to become environmentally sustainable13,23,2831,4749. If, by 2050,
all forms of crop and food wastage13,31 were globally reduced by 50%,
food production emissions could be reduced by about 0.5 Gt yr21 of
CO2-Ceq relative to the 2050 income-dependent diet. Increases in use
efficiencies of animal feeds (from those of Extended Data Table 7), fertilizer and irrigation, and improvements in pasture management and
aquaculture would increase food production, decrease GHG emissions
and improve water quality28,29,4749. Increases in yields of under-yielding
nations could also reduce emissions23. Climate change, though, can affect
yields50, which could in turn have an impact on agricultural GHG emissions and land clearing.
burden, often affecting even the poorer members of poorer nations for
whom appropriate health care is unavailable16,17.
The dietary choices that individuals make are influenced by culture,
nutritional knowledge, price, availability, taste and convenience, all of
which must be considered if the dietary transition that is taking place
is to be counteracted. The evaluation and implementation of dietary
solutions to the tightly linked dietenvironmenthealth trilemma is a
global challenge, and opportunity, of great environmental and public
health importance.
Discussion
5.
6.
Online Content Methods, along with any additional Extended Data display items
and Source Data, are available in the online version of the paper; references unique
to these sections appear only in the online paper.
Received 20 April; accepted 13 October 2014.
Published online 12 November 2014.
1.
2.
3.
4.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
Matson, P. A., Parton, W. J., Power, A. G. & Swift, M. J. Agricultural intensification and
ecosystem properties. Science 277, 504509 (1997).
Steinfeld, H. et al. Livestocks Long Shadow (FAO, 2006).
Edenhofer, O. et al. Climate Change 2014: Mitigation of Climate Change Technical
Summary (Intergovernmental Panel on Climate Change, 2014).
Tubielle, F. N. et al. Agriculture, Forestry and other Land Use Emissions by Sources and
Removals by Sinks (FAO Statistics Division, ESS/14-02, 2014).
Tilman, D. et al. Forecasting agriculturally driven global environmental change.
Science 292, 281284 (2001).
Popkin, B. M., Adair, L. S. & Ng, S. W. Global nutrition transition and the pandemic of
obesity in developing countries. Nutr. Rev. 70, 321 (2012).
Popkin, B. M. The nutrition transition in low-income countries: an emerging crisis.
Nutr. Rev. 52, 285298 (1994).
Drewnowski, A. & Popkin, B. M. The nutrition transition: new trends in the global
diet. Nutr. Rev. 55, 3143 (1997).
Hu, F. B. Globalization of diabetes: the role of diet, lifestyle, and genes. Diabetes
Care 34, 12491257 (2011).
Food and Agriculture Organization of the United Nations. http://faostat.fao.org
(FAO, 2013).
Smil, V. Feeding the World: a Challenge for the Twenty-First Century (MIT Press,
2000).
FAO. Global agriculture towards 2050. In How to Feed the World 2050 110 (FAO,
2009).
Godfray, H. C. J. et al. Food security: the challenge of feeding 9 billion people.
Science 327, 812818 (2010).
Ng, M. et al. Global, regional, and national prevalence of overweight and obesity in
children and adults during 19802013: a systematic analysis for the Global
Burden of Disease Study 2013. Lancet 6736, 116 (2014).
Willett, W. C. et al. Mediterranean diet pyramid: a cultural model for healthy eating.
Am. J. Clin. Nutr. 61, 1402S1406S (1995).
Chopra, M., Galbraith, S. & Darnton-Hill, I. A global response to a global problem:
the epidemic of overnutrition. Bull. World Health Organ. 80, 952958 (2002).
Nishida, C. et al. Diet, nutrition and the prevention of chronic diseases: report of a
joint WHO/FAO expert consultation. Public Health Nutr. 7, 245250 (2004).
Singh, P. N., Sabate, J. & Fraser, G. E. Does low meat consumption increase life
expectancy in humans? Am. J. Clin. Nutr. 78, 526S532S (2003).
4 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 4
ARTICLE RESEARCH
19. Aune, D., Ursin, G. & Veierd, M. B. Meat consumption and the risk of type 2
diabetes: a systematic review and meta-analysis of cohort studies. Diabetologia 52,
22772287 (2009).
20. Kearney, J. Food consumption trends and drivers. Phil. Trans. R. Soc. B 365,
27932807 (2010).
21. Huang, T. et al. Coronary heart disease mortality and cancer incidence in
vegetarians: a meta-analysis and systematic review. Ann. Nutr. Metab. 60,
233240 (2012).
22. Pan, A. et al. Red meat consumption and mortality: results from 2 prospective
cohort studies. Arch. Intern. Med. 172, 555563 (2012).
23. Tilman, D., Balzer, C., Hill, J. & Befort, B. L. Global food demand and the sustainable
intensification of agriculture. Proc. Natl Acad. Sci. USA 108, 2026020264 (2011).
24. de Vries, M. & de Boer, I. J. M. Comparing environmental impacts for livestock
products: a review of life cycle assessments. Livest. Sci. 128, 111 (2010).
25. Nijdam, D., Rood, T. & Westhoek, H. The price of protein: review of land use and
carbon footprints from life cycle assessments of animal food products and their
substitutes. Food Policy 37, 760770 (2012).
26. Eshel, G. & Martin, P. A. Diet, energy, and global warming. Earth Interact. 10, 117
(2006).
27. Marlow, H. J. et al. Diet and the environment: does what you eat matter? Am. J. Clin.
Nutr. 89, 16991703 (2009).
28. Eisler, M. C. et al. Steps to sustainable livestock. Nature 507, 3234 (2014).
29. Smith, J. et al. Beyond milk, meat and eggs: role of livestock in food and nutrition
security. Anim. Front. 3, 613 (2013).
30. Daley, C. A., Abbott, A., Doyle, P. S., Nader, G. & Larson, S. A review of fatty acid
profiles and antioxidant content in grass-fed and grain-fed beef. Nutr. J. 9, 10
(2010).
31. Gustavsson, J., Cederberg, C., van Otterdiijk, R. & Meybeck, A. Global Food Losses
and Food Waste (FAO, 2011).
32. Snowdon, D. A., Phillips, R. L. & Fraser, G. E. Meat consumption and fatal ischemic
heart disease. Prev. Med. 13, 490500 (1984).
33. Key, T. J., Thorogood, M., Appleby, P. N. & Burr, M. L. Dietary habits and mortality in
11,000 vegetarians and health conscious people: results of a 17 year follow up.
Br. Med. J. 313, 775779 (1996).
34. Mann, J. I., Appleby, P. N., Key, T. J. & Thorogood, M. Dietary determinants of
ischaemic heart disease in health conscious individuals. Heart 78, 450455
(1997).
35. Lagiou, P. et al. Mediterranean dietary pattern and mortality among young women:
a cohort study in Sweden. Br. J. Nutr. 96, 384392 (2006).
36. Mitrou, P. N. et al. Mediterranean dietary pattern and prediction of all-cause
mortality in a US population. Arch. Intern. Med. 167, 24612468 (2007).
37. Brunner, E. J. et al. Dietary patterns and 15-y risks of major coronary events,
diabetes, and mortality. Am. J. Clin. Nutr. 87, 14141421 (2008).
0 0 M O N T H 2 0 1 4 | VO L 0 0 0 | N AT U R E | 5
RESEARCH ARTICLE
METHODS
Lifecycle compilation and analyses. We identified and used in our analyses a
total of 120 publications detailing 555 LCAs of GHG emissions from a total of 82
different food items. To find candidate publications, we searched for papers reporting LCAs for numerous food crops, livestock types, fishery species and aquaculture
species using Web of Knowledge, PubMed, AGRICOLA and Google Scholar. We
chose all published LCAs (Extended Data Table 1) that detailed the system boundaries of the study and that included and delimited the full cradle to farm gate
portion of the food/crop lifecycle GHG emissions, including emissions from prefarm activities such as fertilizer and feed production as well as infrastructure construction, but excluding emissions from land-use change. The better to compare the
emissions between different food groups, we calculated emissions per unit protein,
per kilocalorie, or per USDA serving using data from the USDAs Nutrient Database51
and the USDAs MyPlate52. Because few data were available on emissions from
post-farm-gate activities (processing, packaging and transportation to the household), they are not included in our analyses. However, on the basis of data for 21
crop/food production systems for which data were available, inclusion of postfarm-gate food emissions would increase our otherwise calculated 2050 total global
agricultural production emissions by about 20%.
For analysis, we aggregated food items into the 22 food groups shown in Fig. 1.
Extended Data Table 2 lists food items included in each food group. Extended
Data Table 3 has the number of data points as well as mean and standard error of
GHG emissions for each food group. To minimize bias, we do not use in our analyses
GHG emissions from uncommon ways of producing foods, such as greenhousegrown vegetables (33 g of CO2-Ceq per serving, versus 14 g of CO2-Ceq per serving
for field-grown vegetables), for an uncommon speciality food, lobsters caught via
trawling (690 g of CO2-Ceq per gram of protein), and one outlier, organic potatoes,
with estimated emissions 16 times that of conventional potatoes.
Economic groups. Our analyses of dietary trends and of environmental impacts
of alternative diets use data from the 100 most populous nations for which annual
data were available from 1961 to 2009. We analyse data from the two most populous nations, China and India, individually and use aggregated data for all other
nations, with these nations aggregated into six groups based on per capita GDP23.
Group A contains the 15 nations with the highest per capita GDP, Group B has the
next 15, and so on to Group F which contains the 24 nations with the lowest, except
for Group C which has 14 nations (Extended Data Table 4). These eight economic
groups/nations contain 89.9% of the 2009 global population. Nations that did not
have continuous data available from 19612009 were excluded from the study.
Health impacts of different diets. We used Web of Knowledge, PubMed and
Google Scholar to search for cohort-based peer-reviewed publications examining
the health outcomes of Mediterranean-like, pescetarian (fish consumption . once
per month but all other meats , once per month), or vegetarian (fish plus meat ,
once per month, except , once per week for one study32) diets relative to health
outcomes of typical omnivorous diets of individuals in the same cohort. We report
results for those cohort studies that followed more than 5,000 individuals for a
period of at least one year, and that detailed (1) the number of individuals following
each diet, including the average omnivorous diet in the studied cohort, (2) the
average number of years of observation and (3) the relative risk of one or more of
four medical conditions: type 2 diabetes incidence, all cancer incidence, heart
disease mortality or mortality from all causes. In total, our analysis contains results
presented in 18 publications3239,5362 that, in aggregate, summarize approximately
ten million person-years of observations, drawn from eight prospective health
study cohorts (Extended Data Table 5). We use only published incidences that had
been corrected, in the original publication, for effects of potential confounding
variables.
In the case of the pescetarian and vegetarian diets, individuals either followed
or did not follow these diets, based on the criteria described above. For the Mediterranean
diet, we compared the health outcomes for individuals with a Mediterranean diet
score of 69 (as defined by Trichopoulou63,64 and altered by individual studies) to
those with a score of 03.
Relative disease risk for each study of a particular dietdisease combination was
calculated as the reported risk of a particular medical condition for an alternative
diet (Mediterranean, pescetarian or vegetarian) divided by the risk for the same
condition for the cohort portion of that study consuming the local omnivorous
diet, then expressed as a percentage. To determine the average relative risk for each
disease and each alternative diet across all eight cohorts, we weighted the relative
risk we calculated for each instance of a diseasediet combination by the number of
person-years of observations for that particular medical condition and alternative
diet combination.
Per capita GDP forecasts. We forecast the 2050 per capita GDP for each economic/national group as described in ref. 23, by using a differential equation model
represented by a Kuznet-shaped curve fitted65 to the observed 19612009 dependence of (dP/dt)(1/P) on P, that is, the dependence of per capita real GDP growth
rates on real per capita GDP, where P is per capita GDP data from the Total
Economy Database of the Groningen Growth and Development Centre, New
York66.
Income-dependent diet. We use demand to refer to food brought into a household, which we do not call consumption because some portion of it is not eaten,
but rather is wasted13,31. We used data from the United Nations Food and Agriculture
Organization (FAO)10 in 2013 to calculate per capita daily demand for various
types of foods, for total dietary protein and for total calories for each year from 1961
to 2009 for each of the eight economic groups/nations. We then determined the
dependence of demand on per capita GDP using GDP data for 1961 to 2009 from
the Total Economy Database66.
Total demand, proportion of total demand from plants (barley, maize, wheat,
rice, other cereals, soybeans, oil crops other than soybeans, starchy roots and tubers,
legumes, fruits, vegetables and sugar crops), demand for meat (beef, lamb, mutton
and goat, pork, poultry, and seafood), demand for dairy and eggs, as well as demand
for empty calories (refined animal fats, oils, sugars and alcohols) were modelled
globally using a Gompertz 4p curve (Fig. 2). The Gompertz 4p is a logistic-like
function that has both an upper and a lower asymptote. The Gompertz 4p equation
is:
Y~az(b{a)(exp{exp{r(x{d))
where a is the lower asymptote, b is the upper asymptote, r is the growth rate, and d
is the inflection point. For each economic group, we assumed that the relative
consumption of foods within each of the modelled food groups remained constant
at 2009 proportions. Economic groups that followed a trend different from that of
the global trajectory (India for meats, China and economic Group D for dairy and
eggs, and China for empty calories) were modelled independently of the rest of the
economic groups using demand against the square root of per capita GDP (Fig. 2).
By combining the fitted dependence of demand on per capita GDP with countryspecific per capita GDP and United Nations population forecasts, we were able to
estimate total annual demand in 2050 for each food group within each economic
group.
Alternative diets. Per capita protein demand for the vegetarian diet is based on
the General Council of Seventh-day Adventists Nutrition Councils recommended
vegetarian diet67. The pescetarian diet was modified from the vegetarian diet, and
includes a single one-ounce serving of fish per day but reduced milk, egg and cereal
demand so that total per capita protein demand is 1.5 g less per day than that of the
vegetarian diet. The Mediterranean diet is derived from recommendations from
refs 68 and 64. Demand for 2010 through 2050 within each economic group was
then calculated using United Nations population forecasts10.
Marine fisheries and aquaculture. In this publication, we use FAO reported fisheries landings in 2009 (ref. 69) plus the increment in marine fisheries that is estimated to come from improved management70 as the global maximum fisheries
catch. For our projections of GHG emissions associated with alternative diets, we
assume that any fish consumption beyond this limit is produced in aquaculture
systems. Recent analyses discuss ways to more than double aquaculture protein
production by 2050 while minimizing the environmental impacts of this increased
production71,72. Global adoption of the Mediterranean or the pescetarian diet by
2050 would require 62% or 188% more seafood production, respectively. If wildcaught landings stayed at current levels, aquaculture, which grew at 6.1% per year
for 2002 to 2012 (ref. 71), would have to increase at 4.1% per year from 2010 to
2050 to meet the demand of the pescetarian diet.
Agricultural cropland use. We estimate 2050 land demand to see whether alternative diets have consistent differences in their land demands even when allowing for
a range of scenarios of future global agricultural development, as represented by
suites of values for future yields, food waste, pasture productivity, livestock efficiency, and agricultural trade. Specifically, we explore 243 different scenarios consisting of all combinations of three values for each of five factors (the 2050 percentage
increases in crop yields, in livestock productivity of pastures, in feed-use efficiency
of livestock and aquaculture, and in international agricultural trade, and the 2050
percentage decreases in food waste). Statistical analysis of the dependence of the
land needed in 2050 on diet type and on the values of each variable provides estimates
of the effect of a 1% change in each variable on 2050 land demand, and of each diet
on 2050 land demand (Fig. 4c, Extended Data Table 6). The values chosen for each
variable represent small (15%) to moderate (30%) changes that seem plausible
given past temporal trends.
Crop yield scenarios. We used crop production data as reported by the FAO10. We
calculated weighted average crop yields for each of the eight economic groups for
several crop groups (barley, maize, rice, wheat, other cereals, soybeans, other oil
crops, fruits, vegetables, pulses, roots and tubers, sugar crops and tree nuts). We then
converted the weighted average crop yields into nutritious yields (kcal per ha and
tonnes of protein per ha) using data from the USDAs Nutrient Database51.
ARTICLE RESEARCH
As our baseline scenario for future yields, we assume that crop yields will
continue increasing along the linear trajectory fitted to the past 25 years for each
crop group within each economic group. We use the previous 25 years of data, as
opposed to all of the data available, because recent analyses suggest that the trajectories of crop yield increases have slowed during this time frame73. The other
two yield scenarios are accelerated compared to past trajectories (that occur at
faster than historic rates), but limited so as not to exceed the 2009 yields of the
highest-yielding economic group (usually Group A or B, depending on crop type).
Accelerated yields increase from 2009 values linearly through the years such that
by 2050 they have closed either 15% more or 30% more of the 2009 yield gap
between an economic group and the highest-yielding economic group than would
have been closed by following their historic yield trends. Thus our three yield
scenarios accelerate the closing of the yield gap by 0%, 15% or 30%.
Food waste scenarios. We used available data31 to calculate food waste for different
sectors of the food production system (agricultural production, handling and
storage, processing and packaging, distribution and household consumption)
by crop groups (cereals, oil seeds and pulses, roots, fruits and vegetables, meats,
milk, and seafood) and by geographic region31.
When calculating the impacts of reduced food waste, we assume that food
consumption (what is actually eaten) remains constant. For our land-use forecasts, we keep waste at its current levels31 (0% reduction), or reduced waste in each
of the aforementioned sectors of the food production system by 15% or 30%.
Livestock feed-use efficiency scenarios. Livestock operations that use animal feeds
currently differ widely in their feed conversion efficiencies. Here we assume, as
our base case, that by 2050 all livestock operations in all economic groups will
achieve the feed conversion ratios of the best economic group in 2009 for each
type of livestock. The two accelerated scenarios assume that 2050 feed conversion
efficiencies of all economic groups are 15% or 30% greater than the efficiencies
observed in the best economic groups of 2009 (Extended Data Table 7).
Pasture livestock production scenarios. Our three scenarios for pasture productivity are that global livestock production from pastures will remain at its current
level (0% increase), will increase 15% by 2050, or will increase 30% by 2050.
Increased pasture livestock production is assumed to displace livestock production in animal feeding operations, thus decreasing the land area needed to grow
feed crops.
Agricultural trade scenarios. We define agricultural trade as the percentage of
demand within each economic group for a given crop group that is met through
international trade. For our scenarios we assume, for simplicity, that the magnitude of trade is the same for each crop group, with the exception of fruits and
vegetables, which are assumed not to be traded in our model. Trade is assumed to
be between a lower-yielding group and the economic group that has the highest
yield for each given crop group. Our three scenarios have international trade that
would provide 10%, 20% or 30% of domestic demand.
Cropland use forecasts. For each diet (income-dependent, Mediterranean, pescetarian and vegetarian), we forecast the cropland needed in 2050 for each scenario (each of the 243 combinations of three values for each of the five variables
discussed above).
Food demand and crop yields for each economic group were determined as
explained above. To forecast animal feed use, we used peer-reviewed publications
to perform an analysis of animal diets7476 and to calculate protein conversion
ratios (feed protein used to edible protein produced; Extended Data Table 7) and
the average animal feed composition for beef, mutton and goat, pork and poultry,
as well as for several aquaculture species. In combination with our food demand
projections, this analysis enabled us to estimate animal feed use.
For each diet and each scenario (that is, each combination of values for waste,
yields increases, pastureland productivity, efficiency of feed conversion and international trade), cropland use in 2050 for a given crop group within a given
economic group is the total 2050 demand for the crop group (from both food
and animal feed demand from within that economic group and from trade)
divided by the 2050 crop yield for that economic. Global cropland demand for
each scenario and diet is the summation across all crop groups and all economic
groups of the land required for every crop group within every economic group.
51. United States Department of Agriculture. National Nutrient Database http://
ndb.nal.usda.gov/ (2013).
52. United States Department of Agriculture. Choose My Plate http://
www.choosemyplate.gov/ (2013).
53. Trichopoulou, A. et al. Modified Mediterranean diet and survival: EPIC-elderly
prospective cohort study. Br. Med. J. 330, 991 (2005).
54. Key, T. J. et al. Mortality in British vegetarians: results from the European
prospective investigation into cancer and nutrition (EPIC-Oxford). Am. J. Clin. Nutr.
89 (Suppl), 1613S1619S (2009).
55. Key, T. J. et al. Cancer incidence in British vegetarians. Br. J. Cancer 101, 192197
(2009).
56. Tonstad, S., Butler, T., Yan, R. & Fraser, G. E. Type of vegetarian diet, body weight,
and prevalence of type 2 diabetes. Diabetes Care 32, 791796 (2009).
57. Couto, E. et al. Mediterranean dietary pattern and cancer risk in the EPIC cohort. Br.
J. Cancer 104, 14931499 (2011).
58. The InterAct Consortium. Mediterranean diet and type 2 diabetes risk in the
European Prospective Investigation into Cancer and Nutrition (EPIC) study: the
InterAct project. Diabetes Care 34, 19131918 (2011).
59. Hoevenaar-Blom, M. P. et al. Mediterranean style diet and 12-year incidence of
cardiovascular diseases: the EPIC-NL cohort study. PLoS ONE 7, e45458 (2012).
60. Orlich, M. J. et al. Vegetarian dietary patterns and mortality in Adventist Health
Study 2. JAMA Intern. Med. 173, 12301238 (2013).
61. Tantamango-Bartley, Y., Jaceldo-Siegl, K., Fan, J. & Fraser, G. Vegetarian diets and
the incidence of cancer in a low-risk population. Cancer Epidemiol. Biomarkers Prev.
22, 286294 (2013).
62. Tonstad, S. et al. Vegetarian diets and incidence of diabetes in the Adventist Health
Study-2. Nutr. Metab. Cardiovasc. Dis. 23, 292299 (2013).
63. Trichopoulou, A. et al. Diet and overall survival in the elderly. Br. Med. J. 311,
14571460 (1995).
64. Trichopoulou, A. et al. Adherence to a Mediterranean diet and survival in a Greek
population. N. Engl. J. Med. 348, 25992608 (2003).
65. Kuznets, S. Economic growth and income inequality. Am. Econ. Rev. 45, 128
(1955).
66. Groningen Growth and Development Centre. Total Economy Database http://
www.conference-board.org/data/economydatabase (2013).
67. Seventh-day Adventist Diet. http://www.nutrition411.com/component/k2/item/
1458-seventh-day-adventist-diet (2013).
68. Bach-Faig, A. et al. Mediterranean diet pyramid today. Science and cultural
updates. Public Health Nutr. 14, 22742284 (2011).
69. FAO. FishStat http://www.fao.org/fishery/statistics/en (FAO Fisheries and
Aquaculture Department, 2013).
70. Costello, C. et al. Status and solutions for the worlds unassessed fisheries. Science
338, 517520 (2012).
71. Bartley, D. Bianchi, G., Soto, D. & Vannuccini, S. in The State of World Fisheries and
Aquaculture 199223 (FAO, 2014).
72. Waite, R. et al. Improving Productivity and Environmental Performance of Aquaculture
(World Resource Institute, 2014).
73. Grassini P., Eskridge, K. & Cassman, K. Distinguishing between yield advances and
yield plateaus in historical crop production trends. Nature Commun. 4, 2918
(2013).
74. Sundrum, A. et al. Effects of feed strategies, genotypes, sex, and birth weight on
carcass and meat quality traits under organic pig production conditions.
Wageningen J. Life Sci. 58, 163172 (2011).
75. OKiely, P. Intake, growth and feed conversion efficiency of finishing beef cattle
offered diets based on triticale, maize or grass silages, or ad libitum concentrate.
Ir. J. Agric. Food Res. 50, 189207 (2011).
76. Mathlouthi, N., Larbier, M., Mohamed, M. A. & Lessire, M. Performance of laying
hens fed wheat, wheat-barley or wheat-barley-wheat bran based diets
supplemented with xylanase. Can. J. Anim. Sci. 82, 193199 (2002).
RESEARCH ARTICLE
ARTICLE RESEARCH
Extended Data Table 1 | Original data sources for LCAs in Fig. 1
Abbreviations for food types are: B 5 barley; BEER 5 beer; BU 5 butter; CER 5 cereals minus barley, maize, rice and wheat; DA 5 dairy; EGG 5 eggs; PUL 5 pulses; M 5 maize; NA 5 non-recirculating aquaculture;
NT 5 non-trawling fisheries; OC 5 oil crops; OIL 5 oils; PK 5 pork; PO 5 poultry; R 5 rice; SR 5 starchy roots; RA 5 recirculating aquaculture; RM 5 ruminant meat; SC 5 sugar crops; SOY 5 soybeans;
SUGAR 5 sugar; TF 5 trawling fisheries; V 5 vegetables; W 5 wheat; WINE 5 wine; TEF 5 temperate fruits; TRF 5 tropical fruits.
1
http://www.esu-services.ch/fileadmin/download/buchspies-2011-LCA-fish.pdf
2
https://elibrary.asabe.org/azdez.asp?AID519478&T52
3
http://www.sik.se/archive/pdf-filer-katalog/SR728(1).pdf
4
http://www.newbelgium.com/Files/the-carbon-footprint-of-fat-tire-amber-ale-2008-public-dist-rfs.pdf
5
http://www.google.com/url?sa5t&rct5j&q5energy%20consumed%20by%20north%20atlantic%20fisheries&source5web&cd51&ved50CCgQFjAA&url5https%3A%2F%2Fwwz.ifremer.fr%2Fpeche%
2Fcontent%2Fdownload%2F40520%2F552957%2Ffile%2FEnergie%2520consomm%25C3%25A9e%2520(GB).pdf&ei57OtCU7aJA6igyAHXm4DQBg&usg5AFQjCNEviRn3L9pf2Jkcn3-HVM_uWkAtVg&
sig25ff3wbuD91tSz6nwAmbWu6Q&bvm5bv.64363296,d.aWc
6
http://randd.defra.gov.uk/Default.aspx?Menu5Menu&Module5More&Location5None&Completed50&ProjectID511442
RESEARCH ARTICLE
Extended Data Table 2 | Food group composition
Specific food items included in each of the 22 food types detailed in Fig. 1 of the main text.
ARTICLE RESEARCH
Extended Data Table 3 | Mean food production emissions
Number of studies, mean CO2-Ceq emissions, and standard error of the mean associated with food production for each of the 22 food types.
RESEARCH ARTICLE
Extended Data Table 4 | Economic group country composition
List of nations and the percentage of the worlds population included in each economic group or nation in 2009. In parentheses are such percentages based on 2050 population forecasts. The 100 nations were
ranked by their 20002007 average per capita GDP (in 1990 international dollars), with the top 15 assigned to Group A, the next 15 to Group B, and so on, but with the last 25 assigned to Group F, except that China
(economically in Group C) and India (economically in Group E) were each designated as its own group because of their large population sizes.
ARTICLE RESEARCH
Extended Data Table 5 | Cohort studies and health conditions examined
For each of the publications3239,5362 used to quantify effects of three alternative diets on health conditions, its cohort, diet, person-years of data and health conditions examined are listed below. Abbreviations are:
AARP 5 American Association of Retired Persons; EPIC 5 European Prospective Investigation into Cancer; NHS 5 Nurses Health Study; AHS 5 Adventist Health Studies; SUN 5 Seguimento Universidad de
Navarra; UK Veg 5 Vegetarian Society of the United Kingdom: UHCR 5 Uppsala Health Care Region; Whitehall 5 Whitehall; C 5 Cancer; D 5 Diabetes, HD 5 Heart Disease; ACM 5 All Cause Mortality.
RESEARCH ARTICLE
Extended Data Table 6 | Effects of agricultural development variables on forecast 2050 cropland use
Increases in agricultural trade, pasture productivity, animal feed efficiency, waste reduction and accelerated yields consistently result in diet-dependent decreases in cropland requirements. a, Analysis of results
by diet type gives the reduction in 2050 global agricultural land use (million ha) associated with a 1% increase in each of the five variables, based on a separate multiple regression analysis for each diet type of the
forecasted 2050 cropland requirements for each of the 243 scenarios. b, Median forecasts of the additional cropland needed by each diet in 2050 relative to 2009 (Methods).
ARTICLE RESEARCH
Extended Data Table 7 | Protein conversion ratios of livestock
production systems
Number of studies and mean protein conversion ratios (feed protein used/edible animal protein
produced) in examined terrestrial and aquatic livestock production systems with high use efficiencies.
Data sources are listed in Extended Data Table 1.