News & Views | Published:


Food choices for health and planet

Nature volume 515, pages 501502 (27 November 2014) | Download Citation

Are you wondering what to prepare for dinner tonight? Data analyses reveal that certain food choices greatly benefit both your health and the environment. But what to do with this evidence remains a challenge to society. See Article p.518

Food production has a strong effect on the environment — it is responsible for about 25% of global greenhouse-gas emissions1, and biodiversity is greatly affected by agricultural land and water use, nutrient loss and fisheries. Within the agricultural sector, livestock farming has the largest environmental footprint2, and this impact is increasing as traditional diets around the world are being rapidly replaced by diets that are higher in meat, refined sugar and fat. As the scientific basis for the link between diet and the environment grows stronger, the idea has emerged that global dietary changes may contribute to climate-change mitigation3. In response, campaigns to promote meat-free days have been launched, such as 'Meatless Monday' in the United States and United Kingdom and 'Veggie Thursday' in Germany and Belgium. In this issue, Tilman and Clark4 (page 518) show that dietary adjustments would not only reduce greenhouse-gas emissions and agricultural land use, but also greatly reduce individual health risks.

The main novelty of Tilman and Clark's study is that it summarizes strong empirical evidence for the effect of diet on both health and the environment in one publication. For the link between diet and health, the authors compiled data from 18 papers, comprising 8 study cohorts and 10 million person-years of observations, to compare reference diets (including all food groups) to three alternatives: a Mediterranean diet (rich in vegetables, fruit and seafood, but including other foods); a pescetarian diet (including fish and almost no meat); and a vegetarian diet (including dairy products and eggs but almost no meat or fish).

Their review finds a substantial reduction in several negative health indicators for each of these alternative diets compared with the reference diet, including type II diabetes incidence (16–41%, depending on the diet), cancer incidence (7–13%), mortality due to heart disease (20–26%) and overall mortality (0–18%). These effects relate to the fact that the alternative diets contain higher amounts of fruits, vegetables and nuts, and fewer 'empty' calories (energy-containing products that have little other nutritional value, such as alcohol and added sugars) and less meat.

To evaluate the environmental effects of food consumption, the authors assessed two aspects: agricultural greenhouse-gas emissions and changes in land use for agricultural purposes. For agricultural emissions, the authors compiled 120 publications containing 555 life-cycle assessment (LCA) analyses of 82 types of food, which quantify direct emissions along the production chain, including livestock farming, feed production, crop growth, fertilizer application and farm operations. To address land-use change, which has implications for deforestation-associated emissions and biodiversity, they built a simple, transparent model, mostly extrapolating from historical trends. In general, both effects are particularly strong for animal-based food, because livestock, especially ruminants, have a low feed-conversion efficiency, and because ruminants emit methane, a potent greenhouse gas.

Tilman and Clark evaluated both environmental aspects for the year 2050 by comparing a predicted reference diet (based on observed relationships between changes in gross domestic product and food-consumption patterns) with the three alternative diets used for the health analysis (Fig. 1). They find that agricultural emissions would be reduced by 1.2–2.3 gigatonnes of carbon-dioxide carbon equivalents per year (translating to around 30–60% of the projected 2050 emissions from agriculture under the reference diet), and cropland requirements would be reduced by 450 million to 600 million hectares (about 20–30% of the projected 2050 cropland area for the reference diet) if any one of these alternative diets were adopted by the world's population. The authors conclude that “the implementation of dietary solutions ... is a global challenge, and opportunity, of great environmental and public health importance”.

Figure 1: Dietary transitions.
Figure 1

Tilman and Clark4 find that the global average diet in 2050 ('reference diet'; estimated on the basis of current income-dependent trends in food consumption) will have a high content of 'empty' calories (such as from alcohol and added sugar), meat and other animal products, similar to diets now prevalent in high-income countries. They show that this dietary transition will greatly increase the incidence of non-communicable diseases such as type II diabetes and coronary heart disease, and will be a major contributor to agricultural greenhouse-gas emissions and land clearing. By contrast, their analysis suggests that alternative diets that are richer in fruits, vegetables and pulses could both reduce these environmental effects and offer substantial health benefits.

How certain are these effects? The link found by Tilman and Clark between diets and health is astonishingly strong, and they used only data that had been corrected for other lifestyle factors. However, as the authors rightly stress, their data are not meant to compare the alternative diets with each other, nor to imply that other diets might not show even higher health benefits. Future research should aim to expand the empirical basis for the connection between diet and health, and to further investigate the mechanisms behind it.

Consistent with other LCA review studies5, the authors' data analysis shows that greenhouse-gas emissions are highest for ruminant meat, followed by other animal products, and lowest for most cereals, fruits, vegetables and pulses. However, LCA data purely reflect the current state of production systems, and cannot take into account potential efficiency improvements6. Other uncertainties arise from the limited scalability of LCA data and agricultural systems in general. For example, it is not clear if vegetable production can be scaled up while maintaining low greenhouse-gas emissions (for example, because there are higher emissions from growth in greenhouses), and changes in livestock consumption and production will also lead to nonlinear effects, feedback and leakages7 not captured by LCA factors. Marine biologists will also question the scalability of fish production; current levels already overexploit natural stocks, and any increase beyond the sustainable global fisheries catch will have to come from aquaculture, as Tilman and Clark suggest — but expansion of aquaculture will have to rely mostly on land-based feed. However, although such scalability issues should receive further attention, the overall advantage of the alternative diets compared with the reference diet, in terms of emissions, is probably a robust finding.

Predictions of future land requirements for food products and the corresponding environmental consequences are much more uncertain than the emissions predictions, and strongly depend on assumptions about future crop yields, livestock technology and trade. The projected change in global cropland found by Tilman and Clark for the ir 2050 reference diet is higher than the highest estimates in a recent comparison of global agro-economic models8. However, the authors find that the reduction in global cropland that would be achieved by the alternative diets is rather constant when varying the most uncertain determinants of their projection in a sensitivity analysis.

With such clear health and environmental benefits of alternative diets, what could be done with this knowledge? First, it can be used by everybody to make informed consumption choices. But individual choices are strongly influenced by the 'food environment' — factors such as shop proximity, food prices, food and nutrition programmes, labelling schemes and community characteristics. Governments and other agencies play a part in shaping these environments to support healthier and more-sustainable food choices, and increased efforts to include both health and environmental factors in dietary guidelines will be key to promoting behavioural change.

Furthermore, addressing consumption should be accompanied by measures on the production side, because regulations at the source of a problem are often the most effective. For example, agriculture and land-use change should be subject to targets and regulations similar to those for the energy and industry sectors. Such interventions will also help to include environmental costs in the price of resource-intensive food products and would therefore further influence individual choices.


  1. 1.

    , & Annu. Rev. Environ. Resour. 37, 195–222 (2012).

  2. 2.

    et al. Livestock's Long Shadow: Environmental Issues and Options (FAO, 2006).

  3. 3.

    et al. Clim. Change 95, 83–102 (2009).

  4. 4.

    & Nature 515, 518–522 (2014).

  5. 5.

    , & Food Policy 37, 760–770 (2012).

  6. 6.

    et al. Proc. Natl Acad. Sci. USA 111, 3709–3714 (2014).

  7. 7.

    et al. Agric. Syst. 114, 38–53 (2013).

  8. 8.

    et al. Agric. Econ. 45, 3–20 (2014).

Download references

Author information


  1. Elke Stehfest is in the Department of Climate, Air and Energy, PBL Netherlands Environmental Assessment Agency, Bilthoven, 3720 AH, the Netherlands.

    • Elke Stehfest


  1. Search for Elke Stehfest in:

Corresponding author

Correspondence to Elke Stehfest.

About this article

Publication history



Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing