Sustainability of Diets in Mexico: Diet Quality, Environmental Footprint, Diet Cost, and Sociodemographic Factors

Background Little is known about the current intake of sustainable diets globally and specifically in middle-income countries, considering nutritional, environmental and economic factors. Objective To assess and characterize the sustainability of Mexican diets and their association with sociodemographic factors. Design Dietary data of 2,438 adults within the National Health and Nutrition Survey 2012 by integrating diet quality measured by the Healthy Eating Index (HEI-2015), diet cost, and four environmental indicators were analyzed: land use (LU), biodiversity loss (BDL), carbon footprint (CFP), and blue water footprint (BWFP). We defined healthier more sustainable diets (MSD) as those with HEI-2015 above the overall median, and diet cost and environmental indicators below the median. Logistic regression was used to evaluate the association of sociodemographic factors with MSD. Results MSD were consumed by 10.2% of adults (4% of urban and 22% of rural), who had lower intake of animal-source foods, unhealthy foods (refined grains, added sugar and fats, mixed processed dishes and sweetened beverages), fruits, and vegetables, and higher intake of whole grains than non-MSD subjects. Characteristics of MSD vs. non-MSD (urban; rural) were: HEI-2015 (62.6 vs. 51.9; 66.8 vs. 57.6), diet-cost (1.9 vs. 2.8; 1.9 vs. 2.5 USD), LU (3.3 vs. 6.6; 3.2 vs. 5.9 m2), BDL (105 vs. 780; 87 vs. 586 species × 10−10), BWFP (244 vs. 403; 244 vs. 391 L), and CFP (1.6 vs. 4.4; 1.6 vs. 3.7 kg CO2eq). Adults from rural vs. urban (OR 2.7; 95% CI: 1.7, 4.1), or from the South (OR 2.1; 95% CI: 1.1, 3.9), Center (OR 2.3; 95% CI: 1.3, 4.4) vs. the North were more likely to consume MSD, while adults with high vs. low socioeconomic status were less likely (OR 0.17; 95% CI: 0.09, 0.3). Conclusions The MSD is a realistic diet pattern mainly found in disadvantaged populations, but diet quality is still sub-optimal. Increased consumption of legumes, fruits, and vegetables, and a reduction in unhealthy foods, is required to improve nutritional quality of diets while ensuring their environmental sustainability.

Saturated Fats 10 ≤8% of energy ≥16% of energy *The units for each dietary component are called unit equivalents, for example the unit equivalent for total fruits is the cup equivalent per 1,000 kcal, for whole grains it is the ounce equivalent per 1,000 kcal, etc. 1 Intakes between the minimum and maximum standards are scored proportionately. 2 Includes 100% fruit juice. 3 Includes all forms except juice. 4* Originally included legumes (beans and peas). In this study we included only vegetables and fresh green peas. 5* Originally included legumes (beans and peas). In this study we included only legumes (beans, lentils, chickpeas, etc.) 6 Includes all milk products, such as fluid milk, yogurt, and cheese, and fortified soy beverages. 7* Originally included legumes (beans and peas). In this study we included only protein from animal sources. 8* Originally included seafood, nuts, seeds, soy products (other than beverages), and legumes (beans and peas). In this study we excluded soy products and legumes. 9 Ratio of poly-and monounsaturated fatty acids (PUFAs and MUFAs) to saturated fatty acids (SFAs). Figure 3. Results of sensitivity analysis using the definition of more sustainable diets but excluding the diet cost. Relative difference (%) compared to average diet of HEI-2015 score, daily diet cost, and environmental footprint indicators (land use, blue water footprint, carbon footprint, and biodiversity loss), among the high-quality diets, and two definitions of more sustainable diets compared to the habitual diet by area of residence. Diets with HEI-2015 score above the median of the HEI-2015 of their respective reference group. 5 Diets that combine the criteria for high-quality, low-cost and low-environmental footprint diets. 6 Diets that combine the criteria for high-quality and low-environmental footprint diets.

Supplementary methods: Environmental footprint indicators
To assess the environmental sustainability of the Mexican diets, we selected a set of commonly used environmental indicators, or environmental footprints (1). Below we justify the choice of indicators and briefly describe their limitations.
For estimating the contribution to climate change from the different diets, we used the carbon footprint (CFP) of foods, defined as estimated total amount of GHG emitted from a life cycle perspective from the product under study (2). CFP is an environmental indicator that shows emissions of greenhouse gases (GHG) including carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O) to the atmosphere. There are different metrics available to aggregate the different GHGs into one metric (3), here we used the commonly applied Global Warming Potential (GWP) (4). However, GWP has several limitations, e.g., it requires an arbitrary choice of time horizon, which heavily affects the results and underrepresents longer-term climate impacts. Therefore several alternative metrics have been suggested (3) and alternative approaches for aggregating emissions from agriculture to better represent how emissions affect temperature change are being discussed (5). However, all metrics have limitations and their suitability ultimately depends on the application (3,6). In this study, we used GWP over a 100-year perspective (GWP100) for the following reasons: 1) It is the most commonly used carbon footprint metric to date, thus enabling comparisons with other studies, 2) despite its limitations, it provides a valuable measure of the climate impact averaged over the coming hundred years, and 3) it is the metric for which data are available. The GLEAM-I estimates for animal-source food in Mexico use GWP factors from the AR5 IPCC report (7) (methane: 34, nitrous oxide: 298), while most of the studies estimating global average CFPs in the review by Clune et al. (8) probably used factors from the previous AR4 IPCC report (9) (methane: 25, nitrous oxide: 298) for plant-based foods. Thus, there is a difference in the methane factor used for animalsource foods and plant-based foods. However, rice is the only plant-based food associated with substantial methane emissions. Rice is not a staple in the Mexican diet to the same extent as maize and wheat (average daily intake of rice is 8.8 g per person, compared with 22.7 g tortilla and 23.1 g white bread). Using the lower methane factor for rice involved an underestimation of the CFP of the average Mexican diet of less than 0.5%, indicating that this inconsistency had no major impact on the results.
The blue water footprint (BWFP) as defined by Hoekstra et al. (2011)(10), was used to estimate the amount of water needed to produce the Mexican diets. BWFP is defined as "Volume of surface and groundwater consumed as a result of the production of a product. Consumption refers to the volume of freshwater used and then evaporated or incorporated into a product. It also includes water abstracted from surface or groundwater in a catchment and returned to another catchment or the sea. It is the amount of water abstracted from groundwater or surface water that does not return to the catchment from which it was withdrawn". To assess the impact on water resources associated with consumptive blue water use, the water scarcity in the region where different crops are grown needs to be considered (11). There are several methods to account for water scarcity (12,13). For example, the AWARE method (14) supplies region-and country-specific water scarcity factors that can be multiplied by consumptive blue water use to assess water impacts. However, for this study there was no information on where within a country the commodities were grown. Since water availability can vary considerably within a country, particularly large countries such as Mexico and the US, from which the vast majority of the Mexican diet originates, it was not possible to meaningfully apply water scarcity factors. Country-level factors could have been used, but considering the large uncertainties in country-level factors (especially for large countries such as Mexico and the US) we saw little value in using these.
Additionally, as the country-level AWARE factors for Mexico and the US are very similar, it would not have affected the results in this case. It should be stressed that the indicator for sustainability in relation to water use selected for this study does not capture the impact of water use, but it assesses the volume of blue water used to produce the diets. For a further discussion on benefits and drawbacks of different ways to assess water use, see e.g. (15).
Land use refers here to the total amount of agricultural land required for producing the Mexican diets, expressed in square meters. As agricultural land varies in productive capacity, different land use types should ideally be accounted for, most importantly land only suitable for pasture and land suitable for cropping (16). In this study, however, we used the total aggregated land use in order to simplify the analysis. This overestimates land use for diets high in ruminant products (16). The indicators used for land and water use in this study are both pressure indicators, i.e., they communicate the pressure human activities place on ecosystems rather than the potential impact due to such pressures (1).
The biodiversity loss associated with the use of land for producing the Mexican diets was assessed with the method developed by Chaudhary & Brooks (17), which accounts for regional species loss from land occupied by agriculture (reduction or removal of species that would otherwise exist on that land) by accounting for the relative abundance of those species within the region and the overall global threat level for the affected species. The method accounts for five broad land uses (managed forest, plantations, pasture, cropland, and urban), three land use intensities (minimal, light and intense), and five taxa (plants, mammals, birds, amphibians, and reptiles). There are still several limitations with this method, including poor resolution in land types and land management practices, and there are large gaps in current knowledge on how damaging a particular land use type is to different taxa in different parts of the world. The pathway from land use to biodiversity damage is highly complex, making it challenging to include all aspects in a single indicator (17).

Supplementary Figure 5
Methodological scheme for estimation of environmental footprint indicators of animal-based food using dietary data from SFFQ ¹Semi-quantitative food frequency questionnaire of the National Health and Nutrition Survey (ENSANUT-2012). ²Includes edible and non-edible portions (as purchased: with peel, seed, husk, etc.) ³Land use ⁴Environmental footprint indicators per food consumed per person. ⁵Biodiversity loss (potential number of species loss) (18). ⁶Blue water footprint (19). ⁷Greenhouse gas emissions. ⁸Carbon footprint (8).  Supplementary Table 3. Data used to calculate the environmental footprint indicators of processed animal and plant-based food in SFFQ-ENSANUT 2012 *The extraction ratio to estimate the equivalent primary crop of green coffee per kg of soluble coffee was based on the document conversions and statistics for world trade in coffee (http://www.intracen.org/guia-del-cafe/el-comercio-mundial-del-cafe/Conversiones-y-estadisticas/) (21). **ND. No data for water footprint in Mekonnen et al. (18), instead we estimated WFP based on comparison of tea and coffee (21% x WFP of coffee).

Chicken Pig
Source: GLEAM-interactive tool for México (23). Table 7. Parameters for cattle production used in estimation of environmental footprint indicators by orientation and production system, based on GLEAM-i data for Mexico Source: GLEAM-interactive tool for México (23).. Source: Calculation and parameters based on GLEAM-i emissions for Mexico.