WRI released Shifting Diets for a Sustainable Food Future in April 2016, finding that for people who consume high amounts of meat and dairy, shifting to diets with a greater share of plant-based foods could significantly reduce agriculture’s pressure on the environment. Below, we respond to some queries about the methods and findings.

1. The average per person daily protein consumption numbers in the paper seem high. Is this because you used global data based on food utilization rather than nationally collected data based on consumption?

Our estimates are based on food consumption, not food utilization (which is higher than consumption), using data from the United Nations Food and Agriculture Organization (FAO). For definitions of food utilization, food availability and food consumption, visit the FAO website.

Since our paper is a global study, we used FAO data, because it includes comparable data across countries and regions. We started with food availability data and converted this to consumption data, using FAO’s regional estimates of food loss and waste. This involved adjusting the food availability data downward by subtracting food waste at the consumption stage of the value chain (in households and dining out).

Using the global FAO dataset can lead to estimates of food consumption that are inconsistent with country-specific consumption estimates, such as the National Health and Nutrition Examination Survey (NHANES) in the United States. However, if you compare U.S. protein consumption based on FAO data with estimates derived using U.S. national survey data, the results are broadly consistent. For example, in the United States and Canada, the FAO data (adjusted downward as described above) suggest per capita consumption of 90 grams of protein per day (roughly two-thirds of which were from animal-based sources). As a comparison, national diet survey data from the United States show that the average person above age two consumed 80 grams of protein per day, and that the average adult consumed 83 grams of protein per day—63 percent above the average U.S. daily adult requirement. The NHANES diet survey data confirm that roughly two-thirds of protein consumption was from animal-based sources. In fact, since national dietary survey respondents may underreport their true food consumption, the difference between our FAO-derived estimates and the NHANES estimates may be even smaller.

2. The 2015 Dietary Guidelines Advisory Committee (Chart D1.20) reveals that Americans are hitting their recommended consumption targets for “protein foods.” Doesn’t that contradict your findings that Americans are overconsuming protein?

No, chart D1.20 only looks at a subset of proteins consumed by people. It should not be confused with total protein consumption. Our paper counts all plant- and animal-based sources of protein, not just the ones labeled “protein foods” in the U.S. Dietary Guidelines (which count meat, fish and nuts, but omit dairy, legumes and grains). The finding that Americans are overconsuming protein is not one solely identified by WRI—as noted above, the high level of protein consumption is apparent in both the FAO and the NHANES data.

While it is true that in overconsuming countries or regions there is a percentage of the population that underconsumes protein, consumption averages that are far above average requirements suggest that large portions of the population are overconsuming. In fact, if some people are underconsuming protein, then the data suggest that others are overconsuming protein even more.

3. If people reduce meat and dairy consumption, couldn’t they be at risk for not getting enough vitamins and minerals? (For example, as red meat intake has declined in the United States since the 1970s, consumption of “empty calories” like added sugars has grown—along with prevalence of obesity.)

The risk of nutrient deficiency resulting from our alternative diet scenarios is likely low. All eight of the scenarios modeled in our paper target overconsumers of calories and protein and do not reduce overall consumption levels below dietary requirements.

Only two of our meat reduction scenarios—the Ambitious Animal Protein Reduction and the Ambitious Beef Reduction scenarios—reduced meat and/or dairy consumption without replacing it with other foods. Even if we had added in an appropriate combination of plant-based foods (i.e., not just empty calorie foods) to replace the foregone meat and dairy consumption, the change would not have meaningfully altered the results of these scenarios. Because animal-based foods are generally more resource-intensive than plant-based foods, overall land use and GHG emissions would have still declined—by only a little less.

Also, others have indicated that you don’t have to eat meat to get all your dietary nutrients. The 2015–2020 Dietary Guidelines for Americans make it clear that there are multiple ways to achieve a healthy diet by eating appropriate amounts of nutrient-dense foods. The Guidelines detail several “healthy eating patterns,” including the Healthy Vegetarian Eating Pattern, which contains no meat or fish, but includes more legumes, soy products, nuts and seeds and whole grains than the Healthy U.S.-Style Eating Pattern.

4. Why did a paper published last December by Carnegie Mellon University (CMU) find that plant-based foods (e.g., lettuce) could actually be more resource-intensive than animal-based foods (e.g., bacon)?

The findings of the CMU paper do not contradict our finding that meat and dairy is generally more resource intensive and environmentally impactful to produce than plant-based foods. In fact, their findings are consistent with our paper’s findings. The confusion stems from CMU’s press release, which at first blush appeared to contradict our findings, stating that “vegetarian diets could be more harmful to the environment” and that “eating lettuce is more than three times worse in GHG emissions than eating bacon.” Unfortunately, the press release did not accurately represent the study’s findings. The study modeled alternative diet scenarios that were not vegetarian and actually increased overall animal-based food consumption relative to the current average U.S. diet. It is therefore not surprising that such scenarios also increased environmental impacts.

In addition, CMU’s analysis grouped beef consumption with pork. It therefore could not calculate the true environmental benefits of reducing beef consumption. The analysis also did not factor in land use, which is the largest source of differences in GHG emissions between animal- and plant-based foods.

Furthermore, CMU’s headline comparing the GHG emissions per calorie of lettuce and bacon was misleading. Meat and lettuce serve different purposes in people’s overall diets. People do not consume lettuce for calories, but rather for its micronutrient content and taste (lettuce is high in vitamins A and K, among others). By contrast, people do consume bacon and other animal products for their calories and protein. In fact, CMU’s own data, taken from the NHANES survey of American consumers, shows that total U.S. vegetable consumption provided only 5 percent of calories in 2007-10, while 30 percent of calories were from animal-based products. A diet consisting entirely of lettuce would indeed be resource-intensive, but no one (except maybe a rabbit) would eat such a diet! Our paper found that the most effective way to reduce resource use and environmental impacts of the average U.S. diet would be to reduce consumption of meat (especially beef) and dairy—which together account for 80-90 percent of the land use and GHGs associated with the average American diet.

A detailed memo on the CMU paper is available here.

5. Couldn’t we just produce meat and dairy more efficiently? In the United States, for example, production is already more efficient than in developing countries.

Increasing productivity is an essential element of a sustainable food future, but given projected growth in global demand, reductions in consumption among high consumers are also needed to create planetary space to allow underconsumers to consume more. In general, creating a sustainable food future requires an “all of the above” strategy (see figure below).

World agricultural productivity has increased over time, and as discussed in other Creating a Sustainable Food Future installments, crop, livestock and aquaculture productivity can all be further increased without increasing inputs or shifting to the most concentrated feedlot systems. However, relying solely on increased production to close the food gap would exert pressure to clear additional natural ecosystems and would make it harder to reduce agriculture’s emissions and water use. For example, to increase food production by 70 percent by 2050 while avoiding further expansion of agricultural land, crop yields would need to grow one-third more quickly than they did during the Green Revolution. In short, yield increases alone will likely be insufficient to close the gap.

Reducing overconsumption of food is likely to generate environmental, resource-use, health and other benefits, regardless of the production systems employed. This is especially true for animal-based foods, which are resource intensive. In a world where meat and dairy demand is projected to increase by 80 percent between 2006 and 2050, and beef demand by 95 percent during that period, both production-focused and consumption-focused solutions need to be on the table.

6. Isn’t the real root of the sustainable food security issue population growth?

The number of people matters, but so too does the combination of foods people consume. Population growth and shifts to more resource-intensive diets (with rising incomes and urbanization) are two of the main drivers of the 70 percent food gap. Our paper uses the United Nations’ “medium growth scenario” that assumes that world population will grow to 9.7 billion people by 2050. Most of global population growth between now and 2050 will be in sub-Saharan Africa and Asia—the regions where consumption of calories, protein and animal-based foods is currently lowest in the world (and where, as a consequence, per person dietary resource use and environmental impacts are lowest).

We previously calculated that achieving replacement-level fertility in all the world’s regions by 2050—through approaches that empower girls and women, improve quality of life and save millions of lives—would close the global food gap by 9 percent (and close sub-Saharan Africa’s regional food gap by 25 percent). In contrast, we estimate that globally halving meat and dairy consumption (relative to 2050 “business as usual” projections) would close the global food gap by 30 percent. These results suggest that solutions to reduce high fertility rates and to shift diets of the world’s wealthier populations are both necessary ingredients of a sustainable food future.

7. Why are the land-use-change GHG emissions estimates in your paper higher than others’ estimates?

Our estimates are higher than others’ because the GlobAgri-WRR model provides a more complete picture of GHG impacts of different foods by quantifying these impacts at the margin. In other words, the model calculates how much additional land would be needed to produce another unit of protein (or another calorie) from any given food choice. It also estimates the GHGs that would be emitted from the release of stored carbon if that land were converted from native ecosystems (like forests). GlobAgri includes detailed estimates of the carbon content of the to-be-converted land, which vary by the type of land (e.g., shrubland, dense forest) and region.

This graph shows, for example, that converting 5 hectares to grow another ton of wheat-based protein would emit about 62 tons of carbon dioxide equivalent (CO2e)—or 1,245 tons of CO2e amortized over 20 years. However, converting 140 hectares to produce another ton of beef-based protein would emit about 2,230 tons of CO2e (or 44,600 tons of CO2e amortized over 20 years).

Another common approach to estimating land-use-change emissions averages food-related emissions over total food production and observed land use change during a study’s time period. Under this approach, land-use-change emissions per unit of food produced are generally quite low—and can even be zero if land-use change did not occur during the study timeframe. For instance, if a study looks at soybean production in the United States over a one-year period, and soybean area did not expand during that year, then no land-use-change emissions are assigned to soybeans under this approach. The problem with this approach is that it can, in effect, assume that there is no land use cost to producing food.

However, in a world where population and food demand continue to grow, and where agriculture continues to expand into natural ecosystems, every additional amount of food demanded has a land use cost—and thus a related GHG consequence. Our paper examined the consequences of diet shifts while holding yields and trade patterns constant. GlobAgri’s marginal approach reveals that each person’s diet matters quite a lot for agricultural land use and the associated GHG emissions. Regardless of what anyone else does, individual dietary choices affect demand for land “at the margin,” and therefore have a significant effect on GHG emissions.

8. Emissions from livestock production are only 4 percent of total U.S. GHG emissions. Why worry about the GHG impacts of meat and dairy when energy use (for power generation, transportation, industrial use, domestic use, etc.) is a much bigger piece of the pie?

The 4 percent estimate from the U.S. Environmental Protection Agency does not count GHGs associated with agriculture-driven land use change. But because food is a global commodity, what is consumed in one country can drive land use impacts in another. For instance, with no change in productivity per hectare, a rise in beef demand in the United States (with beef demand in other countries held constant) could drive deforestation in Latin America to make way for additional pastureland.

When the land-use-change GHGs are accounted for, the average U.S. diet-related GHG impacts actually come close to energy-related emissions. Our paper found that supplying the average American-style diet over a year required approximately one hectare of land and generated around 1.4 t CO2e from production-related GHG emissions. However, if land-use-change-related GHGs are also taken into account (from clearing one hectare of forest at the beginning of a person’s lifetime to supply that person’s diet), the annual GHG impacts associated with the average U.S.-style diet rise to around 16 tons of CO2e , which is in the range of per capita annual energy-related emissions in the United States.

Globally, between 2006 and 2050, meat and dairy demand is projected to increase by 80 percent. However, during this period, the world also needs to sharply reduce GHG emissions in order to limit global warming to 1.5 or 2 degrees Celsius (2.7 and 3.6 degrees Fahrenheit). We estimate that, by 2050, if agriculture-related emissions from land-use change remain unchanged from present levels, and emissions from agricultural production grow under a “business as usual” path, food demand growth could lead agriculture to consume 70 percent of the total allowable global GHG “budget” consistent with limiting global warming to 2 degrees.

Because it will be an enormous challenge to achieve the necessary global GHG reductions, all solutions, including reducing the consumption of GHG-intensive foods like meat and dairy, must be on the table. Others have come to the same conclusion—a recent paper showed that reducing ruminant meat and dairy consumption among the world’s wealthier populations was a necessary strategy to limit global warming to 2 degrees.

9. If U.S. crop and livestock producers are more efficient than producers in developing countries, why do you use global average production efficiencies when estimating the environmental consequences of the U.S. diet in Figure ES-3?

We used global average production efficiencies in Figure ES-3 (even when examining the environmental impacts of the U.S. diet) because our paper had a global focus and we were interested in examining the consequences of the world’s changing preferences toward diets high in calories, protein and animal-based foods—of which the current U.S. diet shown in “U.S. reference” is a good example. Starting from that “U.S. reference” baseline, we were also interested in the per person consequences of shifting “high consumers” (like U.S. consumers) to diets lower in calories, protein and animal-based foods.

Even if we had used U.S./Canada average production efficiencies (instead of global averages) to calculate land use and GHG emissions in Figure ES-3, the results would be broadly similar. For the U.S. reference diet, overall land use would be 11 percent lower and overall production-related GHGs would be 14 percent lower, but both would still be much higher than the world reference diet. Animal-based foods would still account for 85 to 90 percent of total land use and production-related GHGs in the U.S. reference diet. And the gap between the efficiency of beef and other animal-based foods would be even larger—when using U.S./Canada production efficiencies, beef rises from about half to more than 60 percent of total land use related to the U.S. reference diet. The overall effect (when using the U.S./Canada production efficiencies) would be to make the land and GHG benefits of the three beef reduction scenarios even larger, and the Vegetarian Diet scenario would also have larger environmental benefits due to the greater efficiency of U.S./Canadian dairy production (relative to global average).

To be clear, though, our global analyses of diet scenarios (that shifted the diets of millions or billions of people, as summarized in Figure ES-4 and Tables 2-4) did use regional-level production numbers when applying the various scenarios to the affected regions and populations.