A recent study from researchers at Carnegie Mellon University in Pittsburgh considered changes in energy usage, water usage and greenhouse gas emissions that could result from changing US food consumption patterns. This post focuses on the emissions aspect of that study. It uses emissions figures from the same source used by the study’s authors, and nutrient figures from the US Department of Agriculture’s National Nutrient Database for Standard Reference.

Some curious aspects of the university’s announcement

On 14th December, 2015, the university released an article regarding the study. Some points to note:

  • It said the study’s findings were contrary to what had been said by Arnold Schwarzenegger in a speech at the recent Paris Climate Summit, where he had called for a reduction in meat consumption. The article referred to him solely as an “actor”, and neglected to mention the seemingly relevant point that Schwarzenegger had served two terms as governor of California, with environmental issues high on his agenda. (I am not critiquing his record in that regard.)
  • The study finding that was said to be contrary to Schwarzenegger’s statements was that “eating a vegetarian diet could contribute to climate change”. However, contrary to that statement, the study did not consider vegetarian or vegan diets. It considered dietary scenarios based on the 2010 USDA Dietary Guidelines, which included seafood products. (Neither the study nor the university’s article referred to vegan diets, which exclude egg and dairy products.)
  • The finding that a vegetarian diet “could contribute to climate change” is hardly a revelation. The key point is that its impact is generally less than that of a diet that includes meat, while a vegan diet’s impact would be less again.
  • The article quoted a co-author of the study, Paul Fischbeck, saying, “Eating lettuce is over three times worse in greenhouse gas emissions than eating bacon”. Apart from the poor grammar, that widely circulated statement lacked a key point that was mentioned later in the article, which was that the study assessed emissions on a “per calorie” basis, rather than the conventional basis of “per kilogram of end product”. For reasons referred to below, the validity of the “per calorie” approach is extremely questionable.
  • Lettuce was not specifically referred to in the study, but was included in the “vegetables” category.

Not a valid comparison

We do not generally eat lettuce for calories, which are a measure of food’s energy content. Energy from food is essential for our survival, but is generally obtained from foods other than lettuce. (It is widely known that excessive calories can contribute to weight problems.)

The authors were investigating the environmental impacts of achieving a healthy diet in terms of calorie count. However, it seems to make little sense to compare a food high in calories, such as bacon, to one which we rely on for other benefits.

Cos (romaine) lettuce was the variety considered for the purpose of the study. It is a good source of riboflavin, vitamin B6, calcium, magnesium, phosphorus, copper, dietary fiber, vitamin A, vitamin C, vitamin K, thiamine, folate, iron, potassium and manganese, while being low in saturated fat, cholesterol and sodium. They are all good reasons to eat it, but we would not be doing so as an energy source.

Bacon, on the other hand, is a good source of protein, niacin, phosphorus and selenium, but has the disadvantages of being high in saturated fat (and related calories) and sodium. It has also been found by the World Health Organization and The World Cancer Research Fund to increase the risk of bowel cancer.

If Paul Fischbeck intended to comment on greenhouse gas emissions in relation to a particular nutrient or other feature, then it may have been beneficial to discuss a feature that was prominent in the foods being compared, so that we could choose a realistic option.

According to the USDA, cos lettuce has only 170 calories per kilogram, compared to bacon with 5,330. It would take around 48 average size heads of cos lettuce (weighing around 650 grams or 1 pound, 7 ounces each) to generate the same level of calories as 1 kilogram of bacon (comprising 25 to 30 thin or 15 to 20 thick slices).

Fischbeck also chose to comment on a type of meat with low emissions relative to meat from ruminant animals such as cows and sheep. Ruminants emit large amounts of methane, a potent greenhouse gas, and often graze widely, with implications for CO2 emissions through land clearing and soil carbon losses. Their impact should not be ignored in any discussion comparing greenhouse gas emissions of different foods.

Emissions per kilogram of product including alternative time horizons

A more valid measure than the one used in the study would seem to be the widely used greenhouse gas emissions per kilogram of end product.

Although most published emissions figures are based on a 100-year time horizon, it is also important to consider a 20-year period. The reason is that methane breaks down in the atmosphere to a large extent within that time frame. As a result, the 100-year measure (showing the average impact of a gas over the longer period) understates methane’s shorter term impacts, as the gas would be almost non-existent over the final eighty years. Those impacts will be critical as we try to avoid near-term acceleration of climate change, influenced by significant feedback mechanisms, potentially causing us to lose any ability to influence the climate system in favourable ways. The multiplier used to convert a gas’s warming impact to a “CO2-equivalent” (CO2-e) figure is known as the “global warming potential” or “GWP”.

Figures 2 and 3 show emissions of bacon, lettuce and beef with 100-year and 20-year GWPs.

As indicated earlier, the 100-year emissions figures used throughout this article (and used as the basis for calculating the 20-year figures) are from the data source utilised by the Carnegie Mellon researchers. It was a 2014 review by Martin Heller and Gregory Keoleian from the University of Michigan of life cycle assessment studies (LCAs) relating to US food consumption. The average figures from that review have been used. The LCAs it used were from the US and “other developed countries”. As a result, the figures for animal-based products, in particular, may be conservative relative to the global average. Emissions vary by region, and are influenced by factors such as feed digestibility, livestock management practices, reproduction performance and land use.

For the purpose of the 20-year comparison, the figures for beef and pork have been adjusted based on global average apportionment of emissions categories, as estimated by the Food and Agriculture Organization of the United Nations. As such, in the context of US emissions, the 20-year figures are approximations only.

The comparison is based on 1 kilogram servings of each product, with depictions of the estimated quantities shown in Figure 1. The depiction of eighteen bacon slices is based on an approximate average weight of thick rindless back bacon slices. The lettuces shown here are relatively small heads of cos lettuce, with some of the outer leaves removed. A large cos can weigh around 800 grams, while smaller heads with some leaves removed would typically weigh around 500 grams. A regular steak can weigh around 250 grams.

Figure 1: Estimated 1 kilogram (2.2 pound) servings (not to scale)

Slide1 (1)

Figure 2: Kilograms of greenhouse gas emissions per kilogram of product (100-year GWP)


Figure 3: Kilograms of greenhouse gas emissions per kilogram of product (20-year GWP)


Other products and emissions per kilogram of protein

It is possible to compare emissions per kilogram based on any common nutrient. As protein is abundant in animal and plant products, and is often the focus of attention in terms of nutrition, a comparison based on emissions per kilogram of protein may be useful. The protein content of various products is shown in figure 4.

Figure 4: Protein content (grams per kilogram of product)


Notes: 1. The average of soybeans (365), lentils (258) and chickpeas (193) is allowed for in the “legumes” figures below; 2. The legume figures are based on raw product. Due to increased water content, soaking or boiling reduces protein content per kilogram.

100-year Global Warming Potential

The charts below show emissions per kilogram of: (a) product; and (b) protein; based on a 100-year time horizon.

Figure 5(a): Kilograms of greenhouse gas emissions per kilogram of product (GWP100)


Figure 5(b): Kilograms of greenhouse gas emissions per kilogram of protein (GWP100)


20-year Global Warming Potential

The charts below show emissions per kilogram of: (a) product; and (b) protein; based on a 20-year time horizon.

Figure 6(a): Kilograms of greenhouse gas emissions per kilogram of product (GWP20)


Figure 6(b): Kilograms of greenhouse gas emissions per kilogram of protein (GWP20)


Alternative multiples

Based on the preceding analysis, I argue that the following multiples of emissions from beef and bacon relative to legumes are more valid comparisons than Paul Fischbeck’s comparison of lettuce to bacon.

100-year Time Horizon

Kilograms of CO2-e greenhouse gas per kilogram of product relative to legumes (100-year GWP):

  • Beef: 34 times
  • Bacon: 9 times

Kilograms of CO2-e greenhouse gas per kilogram of protein relative to legumes (100-year GWP):

  • Beef: 34 times
  • Bacon: 6 times

20-year Time Horizon

Kilograms of CO2-e greenhouse gas per kilogram of product relative to legumes (20-year GWP)

  • Beef: 69 times
  • Bacon: 14 times

Kilograms of CO2-e greenhouse gas per kilogram of protein relative to legumes (20-year GWP)

  • Beef: 69 times
  • Bacon: 10 times

Other considerations and conclusion

The emissions intensity of different food products, as utilised in the Carnegie Mellon study and this article, are important factors in helping to identify opportunities for a low emissions diet. Although they almost invariably favour plant-based over animal-based foods, we must also consider other critical problems in animal-based food production.

An example is the precarious position of the Amazon rainforest, which results primarily from inherently inefficient animal-based food production, including livestock grazing and soybean plantations feeding billions of chickens, pigs and cows. We have virtually no buffer available in our efforts to avoid catastrophic climate change, and it is essential that we remove the pressure that currently exists on the Amazon and other critical ecosystems.

Another example is the dramatic release of carbon from ocean vegetation and sediment, along with loss of carbon sequestration capacity, due to industrial and recreational fishing.

In a nation considered to be the home of free enterprise, indirect subsidies to US animal-based food producers should be considered an anathema. Those subsidies are created by the fact that the true cost of animal-based food production is not accounted for in the consumer price. Rather, such costs currently represent externalities, borne by the community as a whole. If they were incorporated in the end price, the market for the more environmentally harmful products would contract, with major overall benefits.

The authors of the Carnegie Mellon study may be concerned about Americans’ ability to achieve a healthier diet, while also reducing greenhouse gas emissions. However, with the huge array of plant-based options available, which on any reasonable comparison offer enormous climate change benefits, it is clear that a general transition away from animal-based food products is essential and achievable.


Paul Mahony (also on Twitter, Facebook, Scribd, Slideshare, New Matilda, Rabble and Viva la Vegan)


Tom, M.S., Fischbeck, P.S., Hendrickson, C.T., “Energy use, blue water footprint, and greenhouse gas emissions for current food consumption patterns and dietary recommendations in the US”, Environment Systems and Decisions, published online 24th November, 2015, http://link.springer.com/article/10.1007/s10669-015-9577-y

Rea, S., “Vegetarian and ‘healthy’ diets could be more harmful to the environment”, Carnegie Mellon University News, 14th December, 2015, http://www.cmu.edu/news/stories/archives/2015/december/diet-and-environment.html

Harrabin, R., BBC Science and Environment, “COP21: Arnold Schwarzenegger: ‘Go part-time vegetarian to protect the planet'”, 8th December, 2015, http://www.bbc.com/news/science-environment-35039465

USDA National Nutrient Database for Standard Reference http://ndb.nal.usda.gov/ via Nutrition Data http://www.nutritiondata.com

Heller M., Keoleian G. (2014) “Greenhouse gas emission estimates of US dietary choices and food loss”, J Ind Ecol 19(3):391–401. doi:10.1111/jiec.12174, http://onlinelibrary.wiley.com/doi/10.1111/jiec.12174/abstract

Brake Bros Ltd, Bacon Buying Guide, http://www.brake.co.uk/_assets/Buying%20Guides_BACON.pdf

Food and Agriculture Organization of the United Nations, “Tackling climate change through livestock: A global assessment of emissions and mitigation opportunities”, Nov 2013, http://www.fao.org/ag/againfo/resources/en/publications/tackling_climate_change/index.htm

Mahony, P., “Chickens, pigs and the Amazon tipping point”, Terrastendo, 5th October, 2015, https://terrastendo.net/2015/10/05/chickens-pigs-and-the-amazon-tipping-point/

Mahony, P., “Seafood and climate change: The surprising link”, New Matilda, 23rd November, 2015, https://newmatilda.com/2015/11/23/seafood-and-climate-change-the-surprising-link/


Cos Lettuce, Romaine Lettuce © Penchan Pumila | Dreamstime.com

Four Striploin Steaks On White Photo © Paulcowan | Dreamstime.com

Raw Bacon Rashers Photo © Philkinsey | Dreamstime.com

Green cos salad © Sasinun Poolpermboonkusol | Dreamstime.com

Please Note

None of the information contained in this article is intended to represent nutritional, dietary, medical, health or similar advice.

Additional comments regarding the Heller and Keoleian paper and the protein content of blade loin roast pork were added on 29th December, 2015, and the main image updated. Various figures were updated on 3rd January, 2016, and notes to figure 4 amended.