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 diet–environment–health trilemma is a global challenge, and opportunity, of great environmental and public health importance.
Agriculture is having increasingly strong global impacts on both the environment1, 2, 3, 4, 5 and human health, often driven by dietary changes6, 7, 8, 9. Global agriculture and food production release more than 25% of all greenhouse gases (GHGs)2, 3, 4, pollute fresh and marine waters with agrochemicals1, 5, and use as cropland or pastureland about half of the ice-free land area of Earth10. Despite the intensity and impacts of global agriculture, almost a billion people still suffer from inadequate diets and insecure food supplies11, 12, 13. Moreover, the global transition towards diets high in processed foods, refined sugars, refined fats, oils and meats has contributed to 2.1 billion people becoming overweight or obese6, 14. These dietary shifts and resulting increases in body mass indices (BMI) are associated with increased global incidences of chronic non-communicable diseases, especially type II diabetes, coronary heart disease and some cancers7, 8, 9, 15, 16, 17, 18, 19, 20, 21, 22, which are predicted to become two-thirds of the global burden of disease if dietary trends continue9, 16, 17. In China, for instance, as incomes increased and diets changed20, the incidence of type II diabetes increased from <1% of its population in 1980 to 10% in 2008, partly because type II diabetes occurs at lower BMI levels and earlier in an individual’s life in Asian than in western populations9. Moreover, diet-driven increases in global food demand7, 8, 12, 23 and increases in population are leading to clearing of tropical forests, savannas and grasslands1, 5, 23, which threatens species with extinction1, 3, 4, 5, 23, 24, 25.
Because it directly links and negatively affects human and environmental health, the global dietary transition is one of the great challenges facing humanity. Meaningful solutions will not be easily achieved. Solutions will require analyses of the quantitative linkages between diets, the environment and human health, on which we focus here, and the efforts of nutritionists, agriculturists, public health professionals, educators, policy makers and food industries.
Here we compile and analyse global-level data to quantify relationships among diet, environmental sustainability and human health, evaluate potential future environmental impacts of the global dietary transition and explore some possible solutions to the diet–environment–health trilemma (Methods and Supplementary Information). To do so, we first expand on earlier food lifecycle analyses24, 25 (LCAs) by searching for all published LCAs of GHG emissions of food crop, livestock, fishery and aquaculture production systems that delimited the full ‘cradle to farm gate’ portion of the food/crop lifecycle. Next we use about 50 years of data for 100 of the world’s 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.
Lifecycle environmental impacts of foods
Dietary composition strongly influences GHG emissions2, 24, 25, 26, 27. The 120 LCA publications that met our criteria report a total of 555 LCA analyses on 82 types of crops and animal products, allowing us to calculate diet-related GHG emissions per gram protein, per kilocalorie and per serving from ‘cradle to farm gate’ (Fig. 1; Methods, Extended Data Tables 1, 2, 3). We express emissions as CO2 warming equivalents, in grams (g) or gigatonnes (Gt) of CO2 carbon equivalents (CO2-Ceq).
GHG emissions vary widely among foods (Fig. 1; Extended Data Table 3 lists means, s.e.m. and number of data points). As is well known, relative to animal-based foods, plant-based foods have lower GHG emissions. This difference can be large; the largest we found was that ruminant meats (beef and lamb) have emissions per gram of protein that are about 250 times those of legumes (Extended Data Table 3; Student’s t-test comparison of means: P < 0.0001). Eggs, dairy, non-trawling seafood, traditional (non-recirculating) aquaculture, poultry and pork all have much lower emissions per gram of protein than ruminant meats (Tukey range test comparing ruminant meats with each other item: P < 0.0001 for each comparison). However, when sustainably grazed on lands unsuitable for cropping and fed crop residues, ruminant dairy and meat production can increase food security, dietary quality, and provide environmental benefits via nutrient cycling28, 29. How a given food is produced can also affect emissions. Seafood caught by trawling, in which nets are often dragged across the ocean floor, has emissions per gram of protein about 3 times those of non-trawling seafood (Fig. 1; Extended Data Table 3; t-test mean comparison: P = 0.017). Items within the same food group can also differ. For instance, among cereal grains, wheat has a fifth the GHG emissions per g protein of rice (t-test comparison: P = 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.
Global dietary change
Although diets differ within and among nations and regions for a variety of climatic, cultural and historic reasons, diets have been changing in fairly consistent ways as incomes and urbanization have increased globally during the past five decades6, 7, 8, 9. This dietary transition has many components, but, in broad outline, its magnitude and global nature are illustrated by trends in per capita demand for meat, empty calories and total calories (Fig. 2), where demand is defined as food brought into a household.
As annual incomes (per capita real gross domestic product, GDP) increased from 1961 to 2009, there were concomitant increases in per capita daily demand for meat protein (Fig. 2a) within and among eight economically based groups of nations23 (Extended Data Table 4). In 2009, the richest 15 nations (Group A; Fig. 2a) had a 750% greater per capita demand for meat protein from ruminants, seafood, poultry and pork than the 24 poorest nations (Group F). Total protein demand also increased with income, but legume protein demand decreased as animal protein demand increased. India, a nation with low rates of meat consumption, is the major exception to an otherwise global trend in the income-dependence of demand for meat protein (Fig. 2a). China initially had meat demand increase more rapidly with income than Groups A–F, but was similar to them by 2009.
A second trend within and among economic groups is the income-dependent increase in demand for ‘empty calories’, here defined as calories from refined fats, refined sugars, alcohols and oils (Fig. 2b). In 2009, Group A nations had an average per capita empty calorie demand of 1,400 kcal per day, whereas demand was 285 kcal per day for Group F. The exception, China, is on an increasing but lower trajectory (Fig. 2b).
A third trend is that total per capita caloric demand also increased with income (Fig. 2c), with China falling below the fitted trend, and Group A being above it. Because some food brought into homes (demand) is wasted13, and the proportion wasted tends to increase with per capita GDP31, actual per capita consumption of meat, empty calories and total calories may be about 20%–25% lower than demand for the Group A nations and about 5% lower in Group F nations. This suggests that, in nations with per capita GDP above approximately $12,000 per year (in 1990$), per capita total caloric consumption may be about 500 kcal per day greater than needed nutritionally.
In total, annual data for 1961 to 2009 for China, India and six income-based groups of nations show that global dietary changes are associated with increased income (Fig. 2), which is itself associated with urbanization and industrial food production20. When these trends are combined with forecasts of per capita income for the coming decades, we estimate that, relative to the average global diet of 2009, the 2050 global-average per capita income-dependent diet would have 15% more total calories and 11% more total protein, with dietary composition shifting to having 61% more empty calories, 18% fewer servings of fruits and vegetables, 2.7% less plant protein, 23% more pork and poultry, 31% more ruminant meat, 58% more dairy and egg and 82% more fish and seafood.
Diet and human health
Diet is an important determinant of human health. Many of the world’s poorest people have inadequate diets, and would have improved health were their diets to include more essential fatty acids, minerals, vitamins and protein from fish and meats and added calories and protein from other nutritionally appropriate sources12, 29. In contrast, diets of many people with moderate and higher incomes are shifting in ways (Fig. 2) associated with increases in non-communicable diseases6, 7 including type II diabetes9, 19, coronary heart disease21 and cancer21, and with higher all-cause mortality rates18, 22.
A point of contrast to the detrimental health impacts of emerging global diets is provided by the benefits reported for three well-studied alternative diets. Here we summarize results from ten million person-years of observations across eight study cohorts32, 33, 34, 35, 36, 37, 38, 39 (Methods; Extended Data Table 5). For each cohort we use reported health outcome effect sizes that had been calculated after statistical control for potentially confounding variables to compare disease incidence rates of individuals who consumed typical omnivorous diets with those who had diets classified as Mediterranean, pescetarian or vegetarian (Fig. 1a). These diets 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 income-dependent 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.
Environmental impacts of diets
GHG emissions are highly dependent on diet24, 25, 26, 27, 42, 43, 44. Even foods that provide similar nutrition and have similar impacts on health can have markedly different lifecycle environmental impacts. Using LCA emission data, we calculated annual per capita GHG emissions from food production (‘cradle to farm gate’) for the 2009 global-average diet, for the global-average income-dependent diet projected for 2050, and for Mediterranean, pescetarian and vegetarian diets (Fig. 4a). Global-average per capita dietary GHG emissions from crop and livestock production would increase 32% from 2009 to 2050 if global diets changed in the income-dependent ways illustrated in Fig. 2. All three alternative diets could reduce emissions from food production below those of the projected 2050 income-dependent diet (Fig. 4a), with per capita reductions being 30%, 45% and 55% for the Mediterranean, pescetarian and vegetarian diets, respectively. However, minimizing environmental impacts does not necessarily maximize human health. Prepared items high in sugars, fats or carbohydrates can have low GHG emissions (Fig. 1) but be less healthy than foods they displace20. Solutions to the diet–environment–health trilemma should seek healthier diets that have low GHG emissions rather than diets that might minimize GHG emissions.
Changes towards healthier diets can have globally significant GHG benefits (Fig. 4b). From 2009 to 2050 global population is projected to increase by 36% (ref. 10). When combined with the projected 32% increase in per capita emissions from income-dependent global dietary shifts, the net effect is an estimated 80% increase in global GHG emissions from food production (from 2.27 to 4.1 Gt yr−1 of CO2-Ceq).This increase of 1.8 Gt yr−1 is equivalent to total 2010 global transportation emissions3. In contrast, there would be no net increase in food production emissions if by 2050 the global diet had become the average of the Mediterranean, pescetarian and vegetarian diets (Fig. 4b).
Future global land clearing for agriculture could threaten species with extinction1, 5 and release GHG beyond that from food production. However, the extent of such land clearing is uncertain, variously projected to total from 0 to 109 hectares5, 23, 45, 46 by 2050, perhaps because of uncertainties about the future values of five factors: crop yields, agricultural and food waste, livestock yields from pastures, animal feed use efficiency and agricultural trade. Here we focus not on forecasting the absolute amount of cropland needed in 2050, but on estimating across many scenarios (243 combinations of three values for each of the five factors; Methods) the differential impacts of diets on global cropland. The alternative scenarios forecast a range of changes in cropland from 2009 to 2050 for each diet (Fig. 4c). For each scenario we calculated the difference between projected 2050 land demands of the income-dependent diet and of each alternative diet (Fig. 4d). Across these scenarios, the income-dependent 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 yr−1 of CO2-Ceq.
In addition to dietary shifts, other changes will be needed for agriculture to become environmentally sustainable13, 23, 28, 29, 30, 31, 47, 48, 49. 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 yr−1 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, 47, 48, 49. 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.
Dietary choices link environmental sustainability and human health. Current dietary trajectories (Fig. 2) are greatly increasing global incidences of type II diabetes, cancer and coronary heart disease. These dietary changes are causing globally significant increases in GHG emissions and contributing to land clearing. Although this pattern does not mean that healthier diets are necessarily more environmentally beneficial, nor that more environmentally beneficial diets are necessarily healthier, there are many alternative dietary options that should substantially improve both human and environmental health.
Our analyses demonstrate that there are plausible solutions to the diet–environment–health trilemma, diets already chosen by many people that, if widely adopted, would offer global environmental and public health benefits. Clearly, to appeal to specific segments of the global population, other such diets should also be developed. The health benefits of adopting such diets could be substantial. Chronic diet-related non-communicable diseases are affecting an increasing number of children and adults in all but the poorest nations. Nations ranging from China and India to Mexico, Nigeria and Tunisia are in the midst of this increasing disease incidence17. Unless the nutrition transition that is under way is changed, diabetes, chronic heart disease and other diet-related chronic non-communicable diseases will become the dominant global disease 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 diet–environment–health trilemma is a global challenge, and opportunity, of great environmental and public health importance.
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 pre-farm 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 USDA’s Nutrient Database51 and the USDA’s 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 post-farm-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 greenhouse-grown 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.
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 1961–2009 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 publications32, 33, 34, 35, 36, 37, 38, 39, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62 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 6–9 (as defined by Trichopoulou63, 64 and altered by individual studies) to those with a score of 0–3.
Relative disease risk for each study of a particular diet–disease 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 disease–diet 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 1961–2009 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.
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:
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 country-specific 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.
Per capita protein demand for the vegetarian diet is based on the General Council of Seventh-day Adventists Nutrition Council’s 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 wild-caught 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 USDA’s Nutrient Database51.
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 diets74, 75, 76 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.
- Agricultural intensification and ecosystem properties. Science 277, 504–509 (1997) , , &
- 2006) et al. Livestock’s Long Shadow (FAO,
- 2014) et al. Climate Change 2014: Mitigation of Climate Change Technical Summary (Intergovernmental Panel on Climate Change,
- 2014) et al. Agriculture, Forestry and other Land Use Emissions by Sources and Removals by Sinks (FAO Statistics Division, ESS/14-02,
- Forecasting agriculturally driven global environmental change. Science 292, 281–284 (2001) et al.
- Global nutrition transition and the pandemic of obesity in developing countries. Nutr. Rev. 70, 3–21 (2012) , &
- The nutrition transition in low-income countries: an emerging crisis. Nutr. Rev. 52, 285–298 (1994)
- The nutrition transition: new trends in the global diet. Nutr. Rev. 55, 31–43 (1997) &
- Globalization of diabetes: the role of diet, lifestyle, and genes. Diabetes Care 34, 1249–1257 (2011)
- Food and Agriculture Organization of the United Nations. http://faostat.fao.org (FAO, 2013)
- 2000) Feeding the World: a Challenge for the Twenty-First Century (MIT Press,
- FAO. Global agriculture towards 2050. In How to Feed the World 2050 1–10 (FAO, 2009)
- Food security: the challenge of feeding 9 billion people. Science 327, 812–818 (2010) et al.
- Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 6736, 1–16 (2014) et al.
- Mediterranean diet pyramid: a cultural model for healthy eating. Am. J. Clin. Nutr. 61, 1402S–1406S (1995) et al.
- A global response to a global problem: the epidemic of overnutrition. Bull. World Health Organ. 80, 952–958 (2002) , &
- Diet, nutrition and the prevention of chronic diseases: report of a joint WHO/FAO expert consultation. Public Health Nutr. 7, 245–250 (2004) et al.
- Does low meat consumption increase life expectancy in humans? Am. J. Clin. Nutr. 78, 526S–532S (2003) , &
- Meat consumption and the risk of type 2 diabetes: a systematic review and meta-analysis of cohort studies. Diabetologia 52, 2277–2287 (2009) , &
- Food consumption trends and drivers. Phil. Trans. R. Soc. B 365, 2793–2807 (2010)
- Coronary heart disease mortality and cancer incidence in vegetarians: a meta-analysis and systematic review. Ann. Nutr. Metab. 60, 233–240 (2012) et al.
- Red meat consumption and mortality: results from 2 prospective cohort studies. Arch. Intern. Med. 172, 555–563 (2012) et al.
- Global food demand and the sustainable intensification of agriculture. Proc. Natl Acad. Sci. USA 108, 20260–20264 (2011) , , &
- Comparing environmental impacts for livestock products: a review of life cycle assessments. Livest. Sci. 128, 1–11 (2010) &
- 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, 760–770 (2012) , &
- Diet, energy, and global warming. Earth Interact. 10, 1–17 (2006) &
- Diet and the environment: does what you eat matter? Am. J. Clin. Nutr. 89, 1699–1703 (2009) et al.
- Steps to sustainable livestock. Nature 507, 32–34 (2014) et al.
- Beyond milk, meat and eggs: role of livestock in food and nutrition security. Anim. Front. 3, 6–13 (2013) et al.
- A review of fatty acid profiles and antioxidant content in grass-fed and grain-fed beef. Nutr. J. 9, 10 (2010) , , , &
- 2011) , , & Global Food Losses and Food Waste (FAO,
- Meat consumption and fatal ischemic heart disease. Prev. Med. 13, 490–500 (1984) , &
- Dietary habits and mortality in 11,000 vegetarians and health conscious people: results of a 17 year follow up. Br. Med. J. 313, 775–779 (1996) , , &
- Dietary determinants of ischaemic heart disease in health conscious individuals. Heart 78, 450–455 (1997) , , &
- Mediterranean dietary pattern and mortality among young women: a cohort study in Sweden. Br. J. Nutr. 96, 384–392 (2006) et al.
- Mediterranean dietary pattern and prediction of all-cause mortality in a US population. Arch. Intern. Med. 167, 2461–2468 (2007) et al.
- Dietary patterns and 15-y risks of major coronary events, diabetes, and mortality. Am. J. Clin. Nutr. 87, 1414–1421 (2008) et al.
- Adherence to Mediterranean diet and risk of developing diabetes: prospective cohort study. Br. Med. J. 336, 1348–1351 (2008) et al.
- Mediterranean diet and incidence of and mortality from coronary heart disease and stroke in women. Circulation 119, 1093–1100 (2009) et al.
- Added sugar intake and cardiovascular diseases mortality among US adults. JAMA Intern. Med. 174, 516–524 (2014) et al.
- Olive oil intake and mortality within the Spanish population (EPIC-Spain). Am. J. Clin. Nutr. 96, 142–149 (2012) et al.
- Climate benefits of changing diet. Clim. Change 95, 83–102 (2009) et al.
- Food consumption, diet shifts and associated non-CO2 greenhouse gases from agricultural production. Glob. Environ. Change 20, 451–462 (2010) , &
- Food choices, health and environment: effects of cutting Europe’s meat and dairy intake. Glob. Environ. Change 26, 196–205 (2014) et al.
- 2012) & World Agriculture Towards 2030/2050: The 2012 Revision Ch. 4 (ESA/12-03, FAO,
- Land-use change trajectories up to 2050: insights from a global agro-economic model comparison. Agric. Econ. 45, 69–84 (2014) et al.
- Biomass use, production, feed efficiencies, and greenhouse gas emissions from global livestock systems. Proc. Natl Acad. Sci. USA 110, 20888–20893 (2013) et al.
- Climate change mitigation through livestock system transitions. Proc. Natl Acad. Sci. USA 111, 3709–3714 (2014) et al.
- Integrated soil-crop system management for food security. Proc. Natl Acad. Sci. USA 108, 6399–6404 (2011) et al.
- Climate impacts on agriculture: implications for crop production. Agron. J. 103, 351–370 (2011) et al.
- United States Department of Agriculture. National Nutrient Database http://ndb.nal.usda.gov/ (2013)
- United States Department of Agriculture. Choose My Plate http://www.choosemyplate.gov/ (2013)
- Modified Mediterranean diet and survival: EPIC-elderly prospective cohort study. Br. Med. J. 330, 991 (2005) et al.
- Mortality in British vegetarians: results from the European prospective investigation into cancer and nutrition (EPIC-Oxford). Am. J. Clin. Nutr. 89 (Suppl). 1613S–1619S (2009) et al.
- Cancer incidence in British vegetarians. Br. J. Cancer 101, 192–197 (2009) et al.
- Type of vegetarian diet, body weight, and prevalence of type 2 diabetes. Diabetes Care 32, 791–796 (2009) , , &
- Mediterranean dietary pattern and cancer risk in the EPIC cohort. Br. J. Cancer 104, 1493–1499 (2011) et al.
- Mediterranean diet and type 2 diabetes risk in the European Prospective Investigation into Cancer and Nutrition (EPIC) study: the InterAct project. Diabetes Care 34, 1913–1918 (2011)
- Mediterranean style diet and 12-year incidence of cardiovascular diseases: the EPIC-NL cohort study. PLoS ONE 7, e45458 (2012) et al.
- Vegetarian dietary patterns and mortality in Adventist Health Study 2. JAMA Intern. Med. 173, 1230–1238 (2013) et al.
- Vegetarian diets and the incidence of cancer in a low-risk population. Cancer Epidemiol. Biomarkers Prev. 22, 286–294 (2013) , , &
- Vegetarian diets and incidence of diabetes in the Adventist Health Study-2. Nutr. Metab. Cardiovasc. Dis. 23, 292–299 (2013) et al.
- Diet and overall survival in the elderly. Br. Med. J. 311, 1457–1460 (1995) et al.
- Adherence to a Mediterranean diet and survival in a Greek population. N. Engl. J. Med. 348, 2599–2608 (2003) et al.
- Economic growth and income inequality. Am. Econ. Rev. 45, 1–28 (1955)
- Groningen Growth and Development Centre. Total Economy Database http://www.conference-board.org/data/economydatabase (2013)
- Seventh-day Adventist Diet. http://www.nutrition411.com/component/k2/item/1458-seventh-day-adventist-diet (2013)
- Mediterranean diet pyramid today. Science and cultural updates. Public Health Nutr. 14, 2274–2284 (2011) et al.
- FAO. FishStat http://www.fao.org/fishery/statistics/en (FAO Fisheries and Aquaculture Department, 2013)
- Status and solutions for the world’s unassessed fisheries. Science 338, 517–520 (2012) et al.
- 199–223 (FAO, 2014) , & in The State of World Fisheries and Aquaculture
- 2014) et al. Improving Productivity and Environmental Performance of Aquaculture (World Resource Institute,
- Distinguishing between yield advances and yield plateaus in historical crop production trends. Nature Commun. 4, 2918 (2013) , &
- 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, 163–172 (2011) et al.
- 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, 189–207 (2011)
- Performance of laying hens fed wheat, wheat-barley or wheat-barley-wheat bran based diets supplemented with xylanase. Can. J. Anim. Sci. 82, 193–199 (2002) , , &
We thank M. Burgess, A. Clark and E. Hallström for their comments, K. Thompson for assistance with data collection, editing, and creating figures, and the LTER programme of the US National Science Foundation and the University of Minnesota Foundation for support.
Extended data figures and tables
Extended Data Figures
- Extended Data Figure 1: Dietary composition. (120 KB)
The percentage of per capita total dietary protein (a) or calorie demand (b) that is met by each of ten food types is shown for each of five different diets: the global-average 2009 diet, the projected income-dependent diet for 2050, the Mediterranean diet, the pescetarian diet and the vegetarian diet.
Extended Data Tables
- Supplementary Data (215 KB)
This file contains numerous data sets used by Tilman and Clark. Each dataset is indicated by a letter (A to H) and has a descriptive title. See Methods for more details.