You are on page 1of 8

AJCN. First published ahead of print August 1, 2012 as doi: 10.3945/ajcn.112.038729.

Sustainable diets for the future: can we contribute to reducing greenhouse gas emissions by eating a healthy diet?13
Jennie I Macdiarmid, Janet Kyle, Graham W Horgan, Jennifer Loe, Claire Fyfe, Alexandra Johnstone, and Geraldine McNeill
ABSTRACT Background: Food systems account for 1820% of UK annual greenhouse gas emissions (GHGEs). Recommendations for improving food choices to reduce GHGEs must be balanced against dietary requirements for health. Objective: We assessed whether a reduction in GHGEs can be achieved while meeting dietary requirements for health. Design: A database was created that linked nutrient composition and GHGE data for 82 food groups. Linear programming was used iteratively to produce a diet that met the dietary requirements of an adult woman (1950 y old) while minimizing GHGEs. Acceptability constraints were added to the model to include foods commonly consumed in the United Kingdom in sensible quantities. A sample menu was created to ensure that the quantities and types of food generated from the model could be combined into a realistic 7-d diet. Reductions in GHGEs of the diets were set against 1990 emission values. Results: The rst model, without any acceptability constraints, produced a 90% reduction in GHGEs but included only 7 food items, all in unrealistic quantities. The addition of acceptability constraints gave a more realistic diet with 52 foods but reduced GHGEs by a lesser amount of 36%. This diet included meat products but in smaller amounts than in the current diet. The retail cost of the diet was comparable to the average UK expenditure on food. Conclusion: A sustainable diet that meets dietary requirements for health with lower GHGEs can be achieved without eliminating meat or dairy products or increasing the cost to the consumer. Am J Clin Nutr doi: 10.3945/ajcn.112.038729.


It is estimated that the food system accounts for 1820% of all GHGEs in the United Kingdom, and the percentage increases to 30% if the impact of land use change is taken into account (4). GHGEs occur at every stage in the life cycle of a food from primary production through to processing, packaging, distribution, consumption, and waste. As a result, GHGEs of different food groups vary widely, and meat and dairy make the greatest contribution to GHGEs in the diet (57). Many countries have made international (eg, the Kyoto protocol) and national (eg, Climate Change Act UK 2008) commitments to reduce GHGEs, and to achieve these, the food sector will have to play a significant role. This will require efciency savings in the primary production and processing of food and changes in consumer behavior to adopt a diet with lower GHGEs. The concept of a sustainable diet is not new (8, 9), but it is a complex issue, and there are still many gaps in our understanding of what a sustainable diet might comprise (10). The complexity is captured in the following recent FAO denition of a sustainable diet: those diets with low environmental impacts which contribute to food and nutrition security and to healthy life for present and future generations. Sustainable diets are protective and respectful of biodiversity and ecosystems, culturally acceptable, accessible, economically fair and affordable; nutritionally adequate, safe and healthy; while optimizing natural and human resources (11). The aim of this study was to take one element of environmental sustainability (ie, GHGEs) and test the compatibility of diets that meet dietary requirements for health with dietary changes needed to reduce GHGEs by using mathematical modeling. An important and, to our knowledge, novel element of the study was to go

Two challenges for public health and nutrition are climate change and obesity (1). It is recognized that current consumption patterns contribute to both of these issues and need to be tackled together to ensure consistent dietary advice for consumers while avoiding any unintended consequences by addressing them separately. In the United Kingdom, as in many other countries, the diet of the population fails to meet dietary recommendations by exceeding the maximum recommendations for saturated fatty acids, added sugars (ie, non-milk extrinsic sugars), and sodium and failing to achieve minimum recommendations for ber (2). These consumption patterns not only fuel a high prevalence of obesity but also contribute signicantly to climate change through high greenhouse gas emissions (GHGEs)4 (3).

1 From the Rowett Institute of Nutrition and Health (JIM, JL, CF, AJ, and GM) and the Division of Applied Health Sciences (JK and GM), University of Aberdeen, Aberdeen, United Kingdom, and Biomathematics and Statistics Scotland, Aberdeen, United Kingdom (GWH). 2 Supported by the World Wildlife Fund-United Kingdom & Scottish Government Rural and Environment Science and Analytical Services. 3 Address correspondence to Jennie I Macdiarmid, Public Health Nutrition Research Group, Polwarth Building (1.073), Foresterhill, University of Aberdeen, Aberdeen, AB25 2DZ, United Kingdom. E-mail: j.macdiarmid@abdn. 4 Abbreviations used: CO2e, carbon dioxide equivalents; GHGE, greenhouse gas emission; LCA, life cycle analysis; RDC, regional distribution center. Received March 12, 2012. Accepted for publication June 11, 2012. doi: 10.3945/ajcn.112.038729.

Am J Clin Nutr doi: 10.3945/ajcn.112.038729. Printed in USA. 2012 American Society for Nutrition

1 of 8

Copyright (C) 2012 by the American Society for Nutrition

2 of 8


beyond the mathematical modeling, which simply generated combinations of quantities of different food, to create and prepare a sample menu to test whether the food lists were realistic, acceptable, and affordable for the general population.

Nutrient composition and GHGE data A database of 82 food groups was created that included the nutrient composition and GHGEs of each food in the form in which the foods would be consumed (eg, cooked weights and edible portion of the food). Nutrient data Nutrient composition data were based on data from the UK National Diet and Nutrition Survey databank (12), and the composition of each food group was calculated from the average of several similar food items. Soya or mycroprotein products as alternatives for animal-based products were not included because they are not commonly consumed in the average UK diet, and the purpose was to create an acceptable diet for the majority of the population. GHGE data Life cycle analysis (LCA) is an approach used for the estimation of GHGEs throughout the stages of production and consumption of a food (13). A full LCA is complex and involves a very detailed assessment of a product from emissions from agriculture and primary production through processing, transportation, retail, home use, and waste disposal. Because of the challenge of the assessment of GHGEs at all these stages for individual products, there are relatively few published data for the full LCA of individual food items. At the time of this study, the most comprehensive list of GHGE data for a range of primary food commodities consumed in the United Kingdom was published by Audsley et al (4). These data do not represent the full life cycle of food items but, rather, average GHGEs for the production of primary food commodities up to the regional distribution center (RDC). The RDC is described as a nominal boundary of primary production up to the point of distribution; therefore, it is based predominantly on GHGEs of production of primary commodities (eg, wheat, sugar, potatoes, beef, and milk) rather than the later stages of processing the ingredients into different food products (eg, bread, biscuits, crisps, pizza). With the use of data for all foods consumed in the United Kingdom, Audsley et al (4) estimate that w56% of total GHGEs are generated up to the point of the RDC. For the purpose of the current study, GHGEs of processed foods were estimated by using the ingredients that make up the foods consistent with work by Wallen et al (6) and using recipes from the sixth edition of McCance and Widdowsons The Composition of Foods and Supplements (14). GHGE data were initially based on the weight of the raw product, and thus, these had to be adjusted to represent the food as eaten (eg, the edible portion of food and weight after cooking) to be compatible with the nutrient-composition data. Therefore, GHGE values were increased to account for weight losses during cooking (eg, for meat and sh) or only the edible proportion of food items (eg, for fruit and vegetables) and decreased to account for any weight

increase through hydration when cooked (eg, for rice and pasta). For example, the GHGE for uncooked rice is estimated to be 3.5 kg carbon dioxide equivalent (CO2e)/kg raw rice (4), but with hydration, it was recalculated to be 1.4 kg CO2e/kg for the equivalent weight of cooked rice. GHGEs can also vary for the same food item depending on how and where it is produced; eg, it has been estimated that the average GHGE for beef consumed in the United Kingdom can vary from 12.1 to 32.0 kg CO2e/kg depending on whether it was produced in the United Kingdom or outside of Europe, respectively (4). The higher emissions are the result not of transportation but of different farming methods because higher GHGEs tend to be associated with less intensive farming methods (15). Food consumed in the United Kingdom comes from a variety of different countries, and therefore, it was important that this variety was accounted for in the GHGE data used in this study. The average GHGE for each food was calculated to reect the proportion of the food imported and domestically produced for UK consumption by using data from HM Revenue & Customs UK Trade Statistics (16). Foods with high, medium, and low GHGEs are shown in Table 1. Dietary requirements and GHGE target The diet for this study was modeled on the UK dietary requirements of an adult woman (1950 y old) by using the current UK dietary recommendation for health (17). The diet was based on the requirements of a woman because women have higher iron but lower energy requirements than men do. It was important that any new diet contained sufcient iron because recent national surveys showed that, at a population level, many women and older girls have low intakes of iron (2, 18), and foods such as red meat, which has a high heme iron content, were likely to be restricted in a sustainable diet to reduce GHGEs. Energy and nutrient requirements for the diet are listed in Table 2. The minimum requirement of protein was increased from 45 to 53 g/d to account for a higher proportion of protein that came from plant sources in the new diet because the digestibility of the protein from plant sources is only w85% compared with .95% from meat (17, 19). The reduction in GHGEs for the diet was set against the UK 1990 GHGE gure because this is typically used as the baseline for GHGE-reduction targets (20, 21). The GHGE of food supplied and consumed in the United Kingdom in 1990 has been estimated to be w152 Metric ton CO2e/y (4), which, by simply dividing by the size of the population, is equivalent to 7.28 kg CO2e/d. However, this gure assumes equal energy consumption across the population. With adjustment for the different energy requirements of the population age and sex structure, the total GHGE for the diet of an average woman (aged 1950 y) was estimated to be 6.74 kg CO2e/d or 3.77 kg CO2e/d for GHGEs up to the RDC (ie, 56% of the total GHGE). The gure of 3.77 kg CO2e/d was used as the baseline for the reduction in GHGEs in the diets in this study. Linear programming optimization Linear programming is a mathematical modeling technique (22) used in other studies to construct nutritionally complete diets while optimizing another outcomes, such as the cost of the


TABLE 1 Greenhouse gas emissions of different food groups on the basis of food supplied to and produced within the United Kingdom1 Low GHGEs (,1.0 kg CO2e/kg edible weight) Potatoes Pasta, noodles Bread Oats Vegetables (eg, onions, peas, carrots, sweet corn, brassicas) Fruits (eg, apples, pears, citrus fruit, plums, grapes) Beans, lentils Confectionery, sugar Savory snacks Medium GHGEs (1.04.0 kg CO2e/kg edible weight) Chicken Milk, butter, yogurt Eggs Rice Breakfast cereal Spreads Nuts, seeds Biscuits, cakes, desserts Fruits (eg, berries, banana, melons) Salad vegetables Vegetables (eg, mushrooms, green beans, cauliower, broccoli, squash)

3 of 8

High GHGEs (.4.0 kg CO2e/kg edible weight) Beef Lamb Pork Turkey Fish Cheese

1 All GHGE values were adjusted to represent the edible weight as cooked and/or the edible portion of each product and adjusted to reect the import: domestic production ratio of produce consumed in the United Kingdom. GHGEs were based on the preregion distribution center rather than a full life cycle analysis. CO2e, carbon dioxide equivalent; GHGEs, greenhouse gas emissions.

diet (2325). In this study, the outcome was to minimize the environmental impact of the diet (ie, GHGEs). To construct a diet that consisted of amounts of n food groups (x1, x2, ., xn), each food group was associated with the GHGE per unit weight of an amount ei as follows: GHGE of the diet e1 x1 e2 x2 . . . : : en xn 1

and with each food group i contributing aij per unit weight to requirement j, a set of j dietary constraints were established such that: a1j x1 a2j x2 . . . : : anj xn  bj 2

The diet had to provide sufcient energy and meet the nutrient requirements listed in Table 2. With macronutrient and micronutrient requirement limits, which were denoted as b1, b2, ., bm,

These constraints were of the following 3 types: constraints with a lower limit (protein, ber, complex carbohydrates, vitamins, and minerals), constraints with an upper limit (sodium, total fat, saturated fatty acids, and nonmilk extrinsic sugars), and a constraint for which there was exact equality (energy). The model

TABLE 2 Energy, nutrient, and food intakes for the nal sustainable diet (with acceptability constraints) compared with dietary requirements and the average dietary intake of a woman in the United Kingdom from the NDNS1 Daily requirements Energy (MJ) Total fat (percentage of food energy) Saturated fat (percentage of food energy) Total carbohydrate (percentage of food energy) Non-milk extrinsic sugar (percentage of food energy) Nonstarch polysaccharide (g) Protein (g) Iron (mg) Folate (mg) Vitamin B-12 (mg) Zinc (mg) Calcium (mg) Sodium (mg) Food groups Fish (g/wk) Red and processed meat (g/wk)5 Fruit and vegetables, including fruit juice g/d Portions/d
1 2

Sustainable diet 8.1 28.4 10.4 54.9 8.4 23 80 14.7 327 5.7 10.0 1015 2400 223 190 555 6.9

NDNS 20082010 (n = 461) 6.88 6 2.02 34.4 6 6.9 12.6 6 3.4 48.3 6 7.5 12.2 6 7.0 12.8 6 4.5 65.4 6 18.1 9.8 6 3.0 232 6 83 4.7 6 3.5 7.7 6 2.4 740 6 254 20294 161 392 330 4.1

8.1 ,35 ,11 .50 ,11 .18 $53 $14.83 $2003 $1.53 $7.03 $7003 1600 (maximum 2400)3 2 portions ,500 $400 5.0

From Bates et al (2). NDNS, National Diet and Nutrition Survey. Mean 6 SD (all such values). 3 Reference nutrient intakes [the amount to meet the needs of 97.5% of the population (15)]. 4 Does not include salt added to meals. 5 Red meat includes beef, lamb, and pork.

4 of 8


included constraints for 6 micronutrients (zinc, iron, calcium, vitamin B-12, folate, and sodium), which were selected either because they are largely derived from animal products or because a large proportion of the UK population does not currently meet recommended intakes. In addition, the diet also had to achieve UK recommendations for sh intake of 2 portions/wk, one of which should be oily sh (26). The linear programming was rst run with only energy and nutrient constraints, but as observed in previous studies (27), diets generated by using this methodology tend to include only a small number of food items, mostly in unrealistic quantities. To include more food items in the diet, lower or upper weight limits were placed on the quantity of each food that could be selected from the database. Upper limits restricted the maximum quantity of a food that could be included in the diet, whereas lower limits forced foods into the diet that would not have been selected because of high GHGEs or a low nutritional quality. These limits were referred to as acceptability constraints, which were based on foods consumed by $50% of the UK population (taken from intake data from the National Diet and Nutrition Survey) (28). The acceptability of a diet is subjective and varies between individuals, but for the purpose of this study, it was based on what is commonly consumed by the population. Throughout this article, the diets generated by the linear programming is referred to as a sustainable diet. Linear programming was carried out by using the GNU Linear Programming Kit as implemented in the Rglpk (0.35) package of the R (2.11) statistical software environment (R Foundation for Statistical Computing). Creating sample menus for a healthy and sustainable diet Linear programming mathematically optimizes a list of foods that met the constraints while minimizing GHGEs. The next step was to create a sample 7-d menu to test whether the types and quantities of the foods generated could be combined into a realistic diet for the UK population. To nalize the sample sustainable diet, an iterative process was used that required the program to be rerun by adjusting the upper and lower weight limits of individual foods until the quantities of food items could be combined into a weekly menu, while still meeting the dietary constraints. It was also important that perishable foods with a limited storage life, such as fruit, were only included in whole units to minimize waste. Once the sample 7-d menu had been created, all meals and recipes were prepared in the metabolic kitchen at the Human Nutrition Unit at the Rowett Institute of Nutrition and Health to ensure that the quantities of each food were appropriate. The nutrient content of the diet was also compared with the nutrient requirements for essential micronutrients that had not been specied as constraints in the model by using the WinDiets dietary assessment program (Univation Ltd, The Robert Gordon University). The diet was compared with the current average intake of women in the United Kingdom (2). For this calculation, the contribution of food groups for the 2 diets was determined by weight, with the weight of liquids (eg, milk or fruit juice) halved to adjust for the high water content compared with that of solid foods (29).

requirements with a 90% reduction in the GHGE (0.39 kg CO2e/ d) but consisted of only 7 foods (whole-grain breakfast cereal, pasta, peas, fried onions, brassicas, sesame seeds, and confectionery). The objective of this model was to minimize GHGEs, therefore foods high in GHGEs, such as meat and dairy products, were not selected when it was possible to get the required nutrients from other foods. Many of the micronutrient requirements were adequately met by breakfast cereals, which are usually fortied with vitamins and minerals. However, the amounts required were large and unrealistic to be included in the diet. This diet was unrealistic not only because of the limited range of foods but also the inappropriate combinations of foods (eg, a large quantity of breakfast cereal with no milk). Sustainable diet with acceptability constraints Types and quantities of foods for a week that were included in the nal version of the sustainable diet are shown in Table 3. The inclusion of acceptability constraints resulted in a diet that was much more realistic. This diet included 52 different food groups and met energy, nutrient, and sh recommendation specied in the model plus recommended minimum intakes for fruit and vegetables and did not exceed the maximum recommendation for red and processed meat (Table 2). The diet also met other micronutrient requirements that were not included as constraints in the model (ie, vitamins A, B-6, C, thiamine, riboavin, niacin, and magnesium). The nal diet included 372 g meat/wk, of which 190 g was red meat (of which 20 g was processed meat). This amount of meat provided enough meat for w4 main meals if used in quantities that matched typical meals in the United Kingdom today. Alternatively, meat could be included in 7 main meals by using smaller quantities and adding other ingredients such as vegetables or beans to bulk out the dish or by increasing the ratio of starchy food to meat in a meal. The quantity of milk in the diet allowed enough to be added to the breakfast cereal and for addition to w3 hot drinks/d. Sufcient fruit and vegetables were included to achieve the minimum recommendation of 5 portions/d (400 g/d). The sample 7-d menu showed that the quantities of each food could be combined into a realistic diet with appropriate links between amounts of key food groups (eg, milk and breakfast cereals). The menu was based on 3 meals/d (breakfast, lunch, and an evening meal) plus snacks that could be consumed at any time of day, including at mealtimes. Examples of 4 d from the 7-d menu to illustrate the types of meals that were included are shown in Table 4. However, the inclusion of a greater variety of foods in the nal sustainable diet was at the expense of the reduction in GHGEs. The GHGE of the nal diet was estimated to be 2.43 kg CO2e/d (a reduction of 36% from the 1990 baseline). This reduction, however, was based on emissions per person with the assumption of the population size in 1990, but current and future annual emissions that are based on this diet need to take into account current and projected population growths. With the use of the Ofce of National Statistics current population estimates and projected estimates for the United Kingdom, it was calculated that the reduction would be equivalent to 30%, 25%, and 14% in 2010, 2020, and 2050, respectively, after the increase in population was taken into account.

Sustainable diet without acceptability constraints The program was run rst without the acceptability constraints, which generated a diet that met all energy and nutrient


TABLE 3 A weekly food list for the nal sustainable diet with acceptability constraints and the contribution of food groups to GHGEs of the diet1
Food groups GHGEs Percentage of total 9.4 14.4 12.8 25.2 17.8 11.5 0.2 1.4 2.0 5.4 2.43 Quantity of food g/wk 539 189 553 458 147 70 168 301 511 140 434 175 112 98 28 112 133 112 217 252 497 217 154 301 203 630 1813 182 294 77 20 85 85 182 1472 49 91 7 28 70 259 56 119 77 98 203 98 147 35 35 42 42

5 of 8

Starchy foods, including potatoes Pasta, noodles (cooked) Rice (cooked) White bread Whole-grain bread Whole-grain, high-ber breakfast cereals Other breakfast cereals Porridge oats Non-fried potato products (eg, oven chips) Potatoes (no added fat) Vegetables Carrots, turnips (cooked) Tomatoes Peas Brassicas Cauliower, broccoli, spinach Sweet corn Cucumber Lettuce Mushrooms (fried) Onions (fried) Peppers Fruit Apples, pears Bananas Grapes, kiwi, cherries Peaches, nectarines, apricots Raspberries, strawberries, blueberries Fruit juice Dairy products Semi-skimmed milk Other cheese (reduced fat) Yogurt (low fat) Ice cream Meat Ham Beef (cooked) Pork (cooked) Chicken (cooked) Fish White sh (coated) Shellsh Oily sh Nuts and seeds Sesame seeds Mixed nuts Beans and legumes Beans Baked beans Lentils (cooked) Eggs High-fat and sweet foods Biscuits (sweet and savory) Buns, cakes, pastries Desserts (eg, rice pudding) Low-fat spread Fried, roasted potatoes Crisps, savory snacks Sugar Chocolate confectionery Preserves Total GHGEs (kg CO2e/d)

All weights were based on the food as consumed (eg, cooked weight and edible portion). CO2e, carbon dioxide equivalent; GHGEs, greenhouse gas emissions. 2 Approximately 56% of the weight is sh.

6 of 8


TABLE 4 Examples taken from 7-d sample menus for the nal version of the healthy and sustainable diet with acceptability constraints Breakfast Day 1 Whole-grain, high-ber cereal and semiskimmed milk White toast and jam Porridge Whole-meal toast and low-fat spread Fruit juice Whole-grain/high ber cereal and semiskimmed milk Whole-meal toast and low-fat spread Fruit juice Lunch Vegetable and lentil soup Prawn sandwich Evening meal Chicken curry and rice Pita bread

Snacks Fruit (berries, apple) Biscuit Milk for hot drinks 1 teaspoon sugar2 Fruit (banana, peach) Scone and jam Milk for hot drinks 1 teaspoon sugar2 Fruit (pear, grapes) biscuit Small packet of crisps Milk for hot drinks 1 teaspoon sugar2

Day 2

Egg salad sandwich Yogurt

Chili beef and kidney bean tortillas3 Salad

Day 3

Tomato and red pepper soup Whole-meal roll

Salmon with cream cheese topping New potatoes, broccoli, and carrots Yogurt

Chicken curry consisted of 91 g cooked chicken (plus mushrooms, onions, and peas). To add to drinks (or equivalent to one glass of diluting juice). 3 Chili beef consisted of 85 g cooked minced beef (plus onions, peppers, kidney beans, and tinned tomatoes).

The nal sustainable diet described in Table 3 is only one example of a diet that meets both dietary requirements and a reduction in GHGEs. It could be remodeled to give many different food combinations and quantities of food by applying different acceptability constraints and allowing for variation in individual preferences or cultural practices around food. Comparison of the sustainable diet with the current UK dietary intake A comparison of the nal version of the sustainable diet (with acceptability constraints) with the most recent assessment of dietary intakes of adult women in the United Kingdom (2) is shown in Figure 1. The comparison illustrates the extent to which the diet needs to shift away from meat and high fat and/or sweet foods and toward fruit and vegetables and starchy foods (eg, bread, pasta, rice, and potatoes) to meet both dietary recommendations and reductions in GHGEs. The amount of meat in the nal sustainable diet was 60% of the current intake of all meat for women in the United Kingdom and 48% of the intake of red meat (2). The proportion of dairy products in the sustainable diet was similar to current dietary intakes, but the type of dairy products were lower in fat; in the current UK diet, 24.5% of dairy products came from ice cream, butter, and cream

compared with only 5.3% in the sustainable diet, which included more milk than the current diet (62% compared with 42% of the dairy products). Cost of the sustainable diet On the basis of the cost of midrange supermarket products in the United Kingdom in September 2010, the cost of the food in the sustainable diet was estimated to be w29.00 ($46.00)/wk, which is equivalent to w89% of the current average UK expenditure on food. This cost could be reduced further if based on cheaper unbranded products. Included within the cost were small quantities of some basic store-cupboard ingredients, such as herbs and spices. Other products such as tea and coffee were not included in the diet, but the additions of these products would still be within the average UK expenditure on food per week of 33.04 (with the exclusion of alcohol) in 2010 (30).

FIGURE 1. Proportions (by weight) of food groups in the nal sustainable diet compared with the average current intake of women in the United Kingdom (National Diet and Nutrition Survey 20082010).

This study has shown that changing food choices to meet dietary requirements for health could also help toward mitigating climate change. We have shown that it is not necessary to eliminate meat and dairy products or for changes to be made that would result in an increased cost to the consumer. However, it cannot be assumed that all diets that meet the dietary requirements for health will necessarily have lower GHGEs; it is equally possible to create a healthy diet by using a different combination of foods that has a high GHGE. For this reason, it is important that these 2 issues are considered together in the development of dietary guidelines. There is no single sustainable diet, and the diet described in this study is only one example of an optimized diet. This study highlighted one of the main differences in the approach to recommendations for health and the environment. Dietary recommendations for health are based mainly on nutrient requirements (which can be achieved through many combinations of foods) (17), whereas the environmental impact is associated with the reduction of intakes of specic foods (eg, meat and dairy). There are synergies here; eg, a reduction in intakes of meat and dairy products could benet health because, together, these products contribute .50% of the total intake of saturated fatty


7 of 8

acids in the UK diet (2, 31), and a high consumption of red and processed meat has been linked to cancer (18, 32). However, any reduction in these foods need to be considered in the context of the whole diet to ensure that substitutions made in the diet are appropriate for health (ie, the replacement of them with lower-fat plant-based foods). If the focus is only on a reduction of GHGEs, animal based products could be replaced by cheaper high-fat and/or sugar products, which tend to have lower GHGEs than animalbased products (4, 6). This was recently illustrated in a study that compared a range of vegetarian and vegan diets, all of which had lower GHGEs than the average meat-based diet, but not all of which had an improved nutritional prole that would benet health (33). The total elimination of meat and dairy foods from the diet would have the greatest impact on reducing GHGEs, but this is likely to be unacceptable for a number of cultural, nutritional, and economic reasons for both consumers and producers. Only 2% of the UK population report being vegetarian, and the consumption of meat and meat dishes in adults in the United Kingdom have increased by almost one-third in the past decade (2). Therefore, the shift in the cultural norm required to achieve a reduction in meat consumption should not be underestimated. The sustainable diet (with acceptability constraints) contained w60% of the current amount of meat eaten by woman in the United Kingdom. It would be possible to increase the amount of meat in this diet, but this would have to be at the expense of the removal of other foods high in GHGEs. A reduction in the intake of animal products carries risk of a decrease in the intake of essential minerals in the population, if not managed appropriately. The bioavailability of some minerals is less from plant sources than from animal products, and some plants contain compounds that can inhibit the absorption of minerals such as zinc and non-heme iron (7, 34). Because there is evidence of suboptimal intakes of a number of vitamins and minerals, such as iron in women and older girls (2, 18, 28), any recommendation to reduce the consumption of animal products would need to specify appropriate alternative food groups and quantities to ensure the intakes of key nutrients are met. One of the most controversial areas in balancing health and environmental concerns is in relation to sh consumption. Consistent with US dietary recommendations (35) to increase the consumption of sh, in the United Kingdom, consumers are recommended to eat 2 portions of sh per week, one of which should be oily sh (26). Fish is recognized as a good source of protein, and oily sh is one of the few sources of n23 fatty acids, but these recommendations need to be considered in the context of whether there are sufcient global wild-sh stocks to meet these recommendations for the whole population (36). If the health and environmental advice on sh consumption is not harmonized, it will only create more confusion in consumers (37). Some countries have started to produce guides that combine dietary recommendations for health with a reduction in the environmental impact (3840). Although this is an important step forward, the guides focus on broad food groups and have not considered what every day menus might consist of. To our knowledge, this study is the rst to take the general principles for a healthy and sustainable diet to the endpoint of sample menus to show that it is possible to have realistic and affordable diets. In doing so, we have highlighted some of the practical challenges that exist in the combination of recommendations, such as linking food groups (eg, breakfast cereal and milk) and altering recipes for

meat dishes to include more vegetables, beans, or pulses to allow meat-based dishes to be eaten several times a week. Because we took a holistic approach to the diet, we also highlighted where it was difcult to reduce amounts of certain foods. For example, despite the high GHGE associated with dairy products, it was difcult to reduce the amount of milk below the current intakes because, without the use of substitutes, the quantity was necessary to have with the amount of breakfast cereal in the diet (which provided other essential nutrients such as ber and vitamins). The high-fat dairy products were replaced with lower-fat versions for health benets, but this can potentially create an environmental problem in terms of waste with the byproducts (ie, cream). It would not be desirable for the cream to simply reenter the diet in the form of low-cost, high-fat foods. Another challenge was the restriction of the sodium content of the diet while trying to increase the amount of complex carbohydrates by using whole-grain breakfast cereals and bread because these products tend to be high in sodium. Linear programming provides a useful technique for the modeling of diets, but without acceptability constraints that take account of dietary habits and preferences, a realistic diet will not be created, as seen in other studies (27). In this study, an iterative process was used whereby the program was rerun with different upper and lower acceptability constraints for key foods to get the right quantities and combinations of foods to create a set of menus. Although the diet in this study was modeled around UK dietary intakes, the approach used is applicable for use in any country. With the use of the same methodology, diets could be created to meet different dietary or cultural needs or dietary preferences. One of the fundamental elements in the use of this methodologic approach was to create a database in which GHGE data and nutrient-composition data were standardized so that they both represented the food as consumed. This method involved the adjustment of published GHGE data to take into account differences between weights of foods as produced and as consumed (cooked, edible portions). In addition it was important to adjust the data to represent the origins of food consumed in the United Kingdom (imported or domestically produced) because GHGEs are heavily depending on production methods that can vary signicantly between countries. A limiting factor was the lack of GHGE data for food and drink products on the basis of a full LCA. In this study, GHGEs were based on emissions for the food production up to the RDC because, at the time of the study, this was the most comprehensive list of GHGEs for foods in the United Kingdom (4). GHGEs after this point in the life cycle of food will vary for different foods depending on the degree of processing, cooking, and waste associated with the product. Future work needs to focus on the contribution of GHGEs at these stages (ie. post RDC) of the life cycle to get data for total GHGEs. This study focused on climate change mitigation (ie, GHGEs), but this is only one element of sustainability. The impact of dietary choices on water use, land use, waste, and biodiversity, as well as social, ethical, and economic issues that threaten future food security, need to be considered in future research and when dietary guidelines are revised. To reect differences in GHGEs, some of the pictorial dietary guides, such as MyPlate (35) or Eatwell (41), could be subdivided. For example, the nondairy protein section currently includes meat, sh, eggs, beans, and nuts, but in terms of sustainability (in this case, GHGEs), it may be helpful for consumers to know what proportion should be meat.

8 of 8

17. Department of Health. Dietary reference values for food energy and nutrients for the United Kingdom. London, United Kingdom: HMSO, 1991. 18. Department of Health. Iron and health. Scientic Advisory Committee on Nutrition. London: TSO, 2010. 19. Kniskern MA, Johnston C. Protein dietary reference intakes may be inadequate for vegetarians if low amounts of animal protein are consumed. Nutrition 2011;27:72730. 20. United Nations. Kyoto protocol to the United Nations framework convention on climate change. United Nations, 1998. Available at 21. Government HM. Climate change act. London, United Kingdom: HMSO, 2008. 22. Dantzig G, Thapa M. Linear programming 1: introduction. New York, New York; Springer-Verlag, 1997. 23. Leung P, Wanitprapha K, Quinn L. A recipe-based, diet-planning modelling system. Br J Nutr 1995;74:15162. 24. Raffensperger JF. The least-cost low-carbohydrate diet is expensive. Nutr Res 2008;28:612. 25. Maillot M, Vieux F, Amiot M, Darmon N. Individual diet modelling translates nutrient recommendations into realistic and individual-specic food choices. Am J Clin Nutr 2010;91:42130. 26. Department of Health. Nutritional aspects of cardiovascular disease. Report of the cardiovascular Review Group Committee on Medical Aspects of Food Policy. Reports on Health and Social Subjects no. 46. London, United Kingdom: HMSO, 1994. 27. Conforti P, Damicis A. What is the cost of a healthy diet in terms of achieving RDAs? Public Health Nutr 2000;3:36773. 28. Henderson L, Gregory J, Irving K, Swan G. The National Diet and Nutrition Survey: adults aged 19 to 64 years. Vols 14. Norwich, United Kingdom: HMSO, 2003. 29. Gatenby S, Hunt P, Rayner M. The National Food Guide: development of the dietetic criteria and nutritional characterisics. J Hum Nutr Diet 1995;8:32334. 30. Horseld G. Family Spending: a report on the 2010 Living Costs and Food Survey. South Wales, United Kingdom: Ofce for National Statistics, 2011. 31. Friel S, Dangour A, Garnett T, Lock K, Chalabi Z, Roberts I, Butler A, Waag J, McMichael AJ, Haines A. Public health benets of strategies to reduce greenhouse-gas emissions: food and agriculture. Lancet 2009;374:201625. 32. World Cancer Research Fund International. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington, DC: American Institute for Cancer Research; 2007. 33. Berners-Lee M, Hoolohan C, Cammack H, Hewitt C. The relative greenhouse gas impacts of realistic dietary choices. Energy Policy 2012; 43:18490. 34. Hunt JR. Bioavailability of iron, zinc, and other trace minerals from vegetarian diets. Am J Clin Nutr 2003;78:633S9S. 35. United States Department of Agriculture. Choose my Plate. Available from: (cited 24 April 2012). 36. Food and Agriculture Organization of the United Nations. The state of world sheries and aquaculture. Rome, Italy: United Nations, 2006. 37. Clonan A, Holdsworth M, Swift J, Leibovici D, Wilson P. The dilemma of healthy eating and environmental sustainability: the case of sh. Public Health Nutr 2012;15:27784. 38. Barilla Center for Food and Nutrition. Double pyramid: healthy food for people, sustainable food for the planet. Parma, Italy: Barilla Center for Food and Nutrition, 2010. 39. Health Council of the Netherlands. Guidelines for a healthy diet: the ecological perspective. The Hague, Netherlands: Health Council of the Netherlands, 2011;publication no. 2011/08E. 40. U.S. Department of Health and Human Services, U.S. General Services Administration. Health and sustainability guidelines for federal concessions and vending operations. Atlanta, GA: Centers for Disease Control & Prevention, 2010. 41. Department of Health. The Eatwell plate. 18 August 2011. Available from: (cited 24 April 2012).

In conclusion, this study demonstrates that it is possible to create a healthy diet with lower GHGEs without the elimination of meat and dairy and at no additional cost to consumers. The study highlights some of the differences between dietary guidelines for health and environmental issues that need to be addressed to provide dietary recommendations that can be communicated in a consistent and coherent way to the public.
The authors responsibilities were as followsJIM, JK, GM, and GWH: designed the research; GWH, JL, CF, AJ, and JIM: conducted the research; JIM, GM, GWH and JK: wrote the article; and JIM: had primary responsibility for the nal content of the manuscript. None of the authors had a conict of interest.

1. UK Cabinet Ofce Strategy Unit. Food matters: towards a strategy for the 21st century. London, United Kingdom: HM Government, 2008. 2. Bates B, Lennox A, Bates C, Swan G. National Diet and Nutrition Survey: headline results from years 1 and 2 (combined) of the rolling programme 2008/09 - 2009/10. Department of Health, 2011. Available at PublicationsStatistics/DH_128166. 3. Garnett T. Cooking up a storm: food, greenhouse gas emissions and our changing climate. Research Network. Guildford, United Kingdom; Centre for Environmental Strategy, University of Surrey, 2008. 4. Audsley E, Brander M, Chatterton J, Murphy-Bokern D, Webster C, Williams A. How low can we go? An assessment of greenhouse gas emissions from the UK food system and the scope for reduction by 2050. WWF-UK, 2009. Available at low_can_we_go.pdf. 5. Carlsson-Kanyama A, Gonzales A. Potential contributions of food consumption patterns to climate change. Am J Clin Nutr 2009;89: 1704S9S. 6. Wallen A, Brandt N, Wennersten R. Does the Swedish consumers choice of food inuence greenhouse gas emissions? Environ Sci Policy 2004;7:52535. 7. Millward DJ, Garnett T. Plenary Lecture 3: Food and the planet: nutritional dilemmas of greenhouse gas emission reductions through reduced intakes of meat and dairy foods. Proc Nutr Soc 2010;69:10318. 8. Gussow J, Clancy K. Dietary guidelines for sustainability. J Nutr Educ 1986;18:15. 9. Coley D, Goodliffe E, Macdiarmid J. The embodied energy of food: the role of diet. Energy Policy 1998;26:4559. 10. Riley H, Buttriss J. A UK public health perspective: what is a healthy sustainable diet? Nutr Bull 2011;36:42631. 11. Food and Agriculture Organization of the United Nations. International Scientic Symposium. Biodiversity and sustainable diets - united against hunger. Rome, Italy: FAO headquarters, 2010. 12. Economic and Social Data Services. NDNS 2000-2001 nutrient databank. Available from: asp?sn=5140 (cited 2 February 2010). 13. British Standards Institute. Publicly available specication PAS 2050 specication for the assessment of the life cycle greenhouse gas emissions of goods and services. London, United Kingdom: British Standards Institute, 2008. 14. Foods Standards Agency. McCance and Widdowsons the composition of foods. 6th ed. Cambridge, United Kingdom: Royal Society of Chemistry, 2002. 15. Williams A, Pell P, Webb J, Moorhouse E, Audsley E. Comparative life cycle assessment of food commodities procured for UK consumption through a diversity of supply chains. DEFRA, 2008: project code Fo0103. Available at More&Location=None&Completed=0&ProjectID=15001. 16. HM Revenue & Customs. Overseas trade statistics (OTS) data. Available from: (cited 18 June 2010).