Dietary greenhouse gas emissions and resource use among Bavarian adults: associations with sociodemographics and food choices

Abstract

Background:

This study assessed dietary greenhouse gas emission (GHGE), land use (LU), and water footprint (WFP) among Bavarian residents while exploring sociodemographic characteristics, food consumption patterns, sustainability beliefs, and behaviors across GHGE quintiles.

Methods and design:

The 3rd Bavarian Food Consumption Survey (BVS III) was conducted from October 2021 to January 2023, involving participants aged 18–75 years. The study employed demographic weighting to represent the Bavarian population. Dietary data (N = 1,100) were linked to sustainability databases.

Results:

In Bavaria, the average dietary GHGE is 6.14 kg CO₂eq, with LU at 7.50 m²*yr. and WFP at 4.39 kiloliters per 2,500 kcal. Multivariate regression analyses indicated that females had significantly higher GHGE (β = 0.204, p = 0.023) and WFP (β = 0.466, p < 0.001) compared to males. Waist circumference was positively associated with GHGE (β = 0.011, p < 0.001) and LU (β = 0.035, p < 0.001). Participants following vegetarian or vegan diets show significantly lower GHGE, LU, and WFP than omnivores. High CO₂eq emitters also consumed more coffee, tea, and most foods of animal origin. Lowest CO₂eq emitters are more inclined to reduce meat consumption (91% vs. 61–77%, p = 0.012), while higher emitters focused on purchasing regional foods (93–95% vs. 80%, p = 0.041).

Conclusion:

This study provided a view of dietary sustainability metrics in Bavaria. Considering energy-adjusted diets, higher emissions are associated with being female, having a higher waist circumference, and following an omnivorous diet. Increased consumption of animal products, coffee, and tea contributed to greater environmental impacts.

1 Introduction

One of the greatest challenges of the 21st century is to feed a growing world population with healthy food without exceeding planetary boundaries (1). Our diets and food systems are currently significant drivers of climate change – accounting for 34% of global greenhouse gas emissions (GHGE) (2). In Germany and overall Europe, diets have significant and growing environmental impact, exceeding a fair share of global ecological resources. This can largely be attributed to the high consumption of animal products, which are associated with a greater carbon footprint, land use (LU) and water footprint (WFP) (3–6). Recent studies have highlighted the necessity of dietary shifts toward more sustainable, less resource-intensive options to mitigate climate change and ameliorate environmental pressures (1, 7, 8). Evidence from systematic reviews showed that transitioning from current Western diets with high proportions of animal-derived products to more plant-oriented diets can significantly lower environmental impacts. For instance, adopting more sustainable dietary patterns could reduce GHGE and land use by up to 50% (9), or even 70–80%, and water use by as much as 50%, with median reductions of 20–30% across studies (7). A recent large-scale modeling study from the UK underpins this finding, demonstrating that the dietary environmental impacts of vegans are only a quarter compared to high meat-eaters for GHGE and LU, and used less than half of water. Eutrophication and biodiversity loss were also notably lower. All environmental indicators increased with higher consumption of animal-derived foods, independent of production methods and sourcing (10). These environmental benefits were accompanied by improvements in public health outcomes (7, 9). Based on data from the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort, Laine et al. (11) highlighted the inherent benefit of sustainable diets for both health and environmental outcomes. Among 443,991 EPIC-participants, higher dietary GHGE and LU were associated with increased risks of all-cause mortality (adjusted hazard ratio 1.13 [95% CI 1.10–1.16] and 1.18 [95% CI 1.15–1.21] for GHGE and LU, respectively) and cancer incidence (adjusted hazard ratio 1.11 [95% CI 1.09–1.14] and 1.13 [95% CI 1.10–1.15] for GHGE and LU, respectively) (11). In this study, dietary shifts were modeled toward a plant-oriented diet, the Planetary Health Diet, which was introduced by the EAT-Lancet Commission (1). A better alignment with the reference diet led to an estimated prevention of up to 19–63% deaths and 10–39% incidences of cancers over 20 years, while simultaneously reducing dietary GHGE by up to 50% and LU by up to 62% (11).

In Europe, dietary emissions vary widely due to differences in food consumption habits and dietary patterns (12). Men typically emit more GHGE per day than women, largely due to higher energy and meat intake (12–16). Stratified by sex, women emitted 2.9 kg CO₂eq per day and men 3.6 kg in Sweden (15), while in the Netherlands, emissions were 3.7 kg CO₂eq for women and 4.8 kg for men (16). Across other European countries, the reported dietary GHGE values were, on average, 6.0 kg CO₂eq per day in France (17), 4.3 kg CO₂eq in Ireland (18), 5.2 kg CO₂eq in Italy (17), and 5.4 kg CO₂eq in Denmark (17), while Switzerland had a median of 3.25 kg CO₂eq per day (13). Comparing these figures is challenging due to differences in methodologies. Results are substantially affected by variations in dietary assessment tools (e.g., 24-h recalls or food frequency questionnaires [FFQs]), sampling units (individual, household, or region), measurement units (daily, annual, or energy-adjusted), system boundaries, and the sustainability databases used. Research on dietary emissions in Germany remains limited, with only a few studies providing insights into the environmental impact of food consumption. Koelman et al. (19) reported energy-adjusted emissions of 6.6 kg CO₂eq per 2000 kcal for men and 7.0 kg CO₂eq for women in Eastern Germany, with dairy, meat, and beverages as the primary contributors. Notably, energy-adjusted emissions were higher in women. Against this, Meier and Christen used data from the 2nd German Nutritional Food Consumption Survey from 2006 and found that men’s diets had 25% higher emissions and 24% higher LU than women’s per 2000 kcal, primarily driven by greater meat intakes. The WFP was 11% higher in women (20). Treu et al. (21) compared average conventional and organic diets in Germany, observing similar GHGE but 40% higher land use for organic diets, which contained less meat and more plant-based foods. Eberle and Fels (22) modeled German food consumption mainly using national statistics, identifying 2.7 t CO₂eq per person per year as the environmental burden, with animal products and food losses as key contributors.

These studies highlight the dual benefits of dietary shifts. Nevertheless, the implementation of dietary changes at individual level remains challenging due to the emotional, cultural, and habitual nature of eating (23). Knowledge about the current dietary situation is essential for achieving effective dietary changes. For example, when using a mathematical optimization model to create food-based dietary guidelines, actual intake data are needed to develop a deviation constraint, ensuring that recommended diets remain practical and acceptable to the population. This was demonstrated with the recently published food-based dietary guidelines in Germany (24). Furthermore, there is often a discrepancy between an individual’s awareness of the environmental impact of dietary choices and their actual dietary behaviors (25). This paper describes the status of environment-related sustainability parameters, particularly greenhouse gas emissions (GHGE) but also land use (LU) and water footprint (WFP), using data from the 3rd Bavarian Food Consumption Survey (BVS III). The focus is on comparing lower vs. higher GHGE groups for differences in food choices, and the influence of sociodemographic factors, behaviors, and attitudes toward (environmental) sustainability.

2 Methods

2.1 Study description

2.1.1 Sampling

The 3rd Bavarian Food Consumption Survey (BVS III) was designed as a statewide representative survey conducted in Bavaria, Germany, between October 2021 and January 2023. The target population comprised all persons aged 18 to 75 years living Bavaria and in private households with sufficient knowledge of the German language. The selection was based on a stratified, multi-stage randomization procedure.

In the first sampling step, a random sample of 60 municipalities was drawn from the entirety of Bavarian municipalities. They were representatively chosen according to the stratification characteristics of administrative district, province (German: Kreis), and region type (German: BIK-Kennzahl). Each municipality could potentially be sampled multiple times as a sample point (i.e., random sampling with replacement). The BVS III finally comprised 80 sample points of 60 municipalities. A total of 11,000 addresses were requested via the residents’ registration offices. Initially, 100 addresses were obtained per sample point. During the study, the number of addresses was increased to 150 due to low participation rates.

In the second step, 90, respectively 100, addresses per sample point were randomly drawn from the requested addresses. To reach the target sample size of approx. 1,500 participants, a total of 7,449 people were contacted.

2.1.2 Design

The BVS III comprises multiple components. First, the participants were visited at home. The visits consisted of a standardized and interviewer-led computer-assisted personal interview (CAPI), followed by a self-administered computer-assisted self-interview (CASI), Body mass and weight were self-reported. Waist circumference was measured. Dried bloodspots samples (DBS) and blood glucose measures were also taken during the home visits. The CAPI and CASI questionnaires encompassed inquiries regarding various aspects of dietary and health behavior, demographic information, and medical history and included various validated questionnaires (see Supplementary methods). Sustainability was another focal point, with a dedicated section of questions. The selection of questions was based on various publications (26–28) and mainly on Tobler et al. (27).

After home visits, participants completed up to three 24-h dietary recalls via telephone (CATI) over 6 weeks, using GloboDiet^©, a computer-based survey program derived from EPIC-SOFT^© and developed by the International Agency for Research on Cancer (Lyon, France) (29). The Max Rubner-Institute (Karlsruhe, Germany) provided the German version. Food and supplement intakes were recorded on two weekdays and one weekend day. A photo book aided portion size estimations. Data collection was managed by KANTAR (Munich, Germany), and DBS samples were analyzed by Vitas AS (Oslo, Norway). Home visits were conducted from October 2021 to November 2022. Dietary 24-h recalls ended in January 2023. The study design is described in detail in the Supplementary methods and by Rohm et al. (30).

2.1.3 Data quality assurance

To ensure high data quality, attention was devoted to quality assurance procedures. Multiple training sessions and refresher courses were conducted for the interviewers for both home visits and 24-h recalls. A preceding pilot study for the home visits and a pre-test for the 24-h recalls were conducted in August 2020 to facilitate a flawless procedure.

The completeness and plausibility of the collected data from the CAPI and CASI were continuously assessed during the field phase. Due to the computer guidance through and programming of the questionnaire, permissible value ranges and answer options were predefined, and the filter guidance was predetermined, minimizing the possibility of incorrect entries from the start. Minor plausibility corrections were made as needed. The dataset for CAPI and CASI was additionally examined regarding completeness of variables, variable and value labels, value ranges, distribution, and plausibility of values at the beginning, midpoint, and end of the field phase. Minor necessary corrections were made as needed. The nutritional data collection via GloboDiet^© enabled a preliminary assessment of completeness and plausibility per participant, facilitated by predefined value ranges for food and recipe portions. GloboDiet^© issued alerts during data entry for missing or unusually high inputs and reminders for commonly forgotten items during recall interviews. Before concluding the interview, summary data, including energy and macronutrient intake, were presented, juxtaposed with long-term energy requirements. Warning messages were prompted for intake deviations from these requirements. During the CATI data collection, an interim dataset was reviewed to assess the plausibility of the data. After the final data handover, refined data validation and preparation were conducted, including correcting erroneous inputs and content-based plausibility checks. All corrections were documented.

2.1.4 Data analysis

2.1.4.1 Weighting factors

To ensure representativeness for the Bavarian population, weighting factors for the data from the home visits and dietary assessments were generated based on several criteria. These included administrative district, education level, and combinations of education and age as well as gender and age. Education was classified into three levels: (1) no qualification, lower secondary, or elementary school; (2) secondary education without a high school diploma (“Abitur”) or currently in school; and (3) high school diploma (“Abitur”), university entrance qualification, or university degree. Cases with no information or other forms of education were assigned a factor of 1 (i.e., they were not weighted). The reference was the German 2020 micro census, i.e., an extrapolation of the data for Bavaria (31).

2.1.4.2 Estimates of energy, nutrient, and food intake

Upon completion of the dietary surveys and to assess nutrient intakes, reported foods were matched with the German Nutrient Database (BLS 3.02; German: Bundeslebensmittelschlüssel 3.02) (32). Composite dishes and recipes were deconstructed using their individual ingredients based on the recipes from the BLS 3.02. Subsequently, all food items were categorized into food groups on two levels of detail (Table 1).

Only participants who completed at least two 24-h dietary recalls were included in the food intake analyses for this paper. Questionnaire-based variables were analyzed using the total study population (N = 1,503), with occasionally missing data points.

2.2 Measures

Data on sex, age, body height and mass, waist circumference, education level, net household income, and attitudes toward sustainability were used, alongside dietary intake data of the 24-h recalls. Participants self-reported their diet type as vegetarian, vegan, no special diet, or other. Vegans and vegetarians were collectively categorized as plant-based diets, while those with no special diet or other classifications were grouped as omnivorous. For the sociodemographic description of the study population, the variables civil status, living situation, employment, smoking, and physical activity were included, Physical activity groups were determined as described by Gerrior et al. (33).

The equivalized net income was computed using the modified OECD equivalence scale (34). The net household income data were assessed using predefined ranges. The median of the range was used as the metric value for calculation. The maximum value of the monthly net household income was presented in text form as “7,000 Euro and more.” This was interpreted as 7,500 Euro/month. Similarly, the minimum was described as “below 500 Euro” and set to 250 Euro/month.

The dietary data were converted into weekday-weighted averages to account for variations in diet across different days. Individuals with a total energy intake to basal metabolic rate ratio below 0.6 were identified as underreporters and excluded. As there were no cases with a ratio of total energy intake to basal metabolic rate greater than 2.4, overreporters were not excluded. The basal metabolic rate was calculated using World Health Organization (WHO) equations (35). Dietary data were linked to two sustainability metric databases: SHARP-ID (36) for dietary GHGE (in kg carbon dioxide equivalents [CO₂eq]) and LU (in m²*yr), and SuEatableLife (37) for WFP (in kiloliters).

The SHARP-ID database estimates GHGE and LU per kilogram of food as consumed based on dietary data from four European countries (Denmark, Czech Republic, Italy and France). The system boundaries applied within the SHARP-ID encompass the entire life cycle of food products, including primary production, processing, packaging, transport, and consumption (36); and were supplemented with the recorded cooking method by the participants of the BVS III. For animal-derived foods, GHGE and LU were estimated up to the farm gate using the CAPRI model, which provides a cradle-to-gate life-cycle assessment (LCA) of livestock products (meat, milk, and eggs) (5). The EU-average was primarily used as source for animal-derived foods within the SHARP-ID (36). Other sources for primary production impacts for other foods were the Agri-Footprint 2.0 (38, 39) and Ecoinvent 3.3 (40). More detail of the development of the SHARP-ID are comprehensibly described elsewhere by Mertens et al. (36).

The SuEATableLife database provides the WFP for food, based on literature, public reports, and Environmental Product Declarations. The system boundaries applied within the database include the life cycle up to the distribution center (primary production, processing, packaging, and transport), excluding post-market stages, i.e., consumption (37). Hence, the recorded cooking method by BVS III participants was not included in WFP calculations. More details are described by Petersson et al. (37).

Participants were categorized into quintiles based on their GHGE per 2,500 kcal.

2.3 Statistical analysis

Descriptive statistics were used for sociodemographic characteristics, with standard deviations (SD) for unweighted and standard errors (SE) for weighted data. Wilcoxon’s rank-sum test was used for two-group comparisons, and Kruskal–Wallis rank-sum test for more than two groups, adjusting for survey samples. Categorical variables were compared using the chi-squared test with Rao and Scott’s correction for weighted data or Pearson’s chi-squared and Fisher’s exact tests for unweighted data. Generalized linear models were used for linear trend analyses with ordered factors. Food group intakes, GHGE, LU, and WFP were adjusted to 2,500 kcal. Weighted linear regression was performed to assess the association of sociodemographic factors and diet type with dietary GHGE, LU, and WFP. Heteroscedasticity-consistent standard errors type 3 were used when necessary. Outliers, defined as residuals exceeding two standard deviations, were removed to maintain normality in the models. Marginal means were calculated from regression models. The estimate for the marginal mean represents the predicted average for each group while averaging over the other independent variables. This approach accounts for confounding factors for adjusted comparisons between groups. Data were weighted to reflect the Bavarian population. All statistical analyses and graphical depictions were performed using the statistical software R version 4.4.0 (41).

3 Results

3.1 Study characteristics

3.1.1 Response

Out of the 7,449 individuals contacted, the cleaned gross sample, excluding quality-neutral dropouts (QND), consisted of 5,770 individuals. From this, 1,503 (26.0% response rate) participated in the CAPI/CASI during home visits. During these visits, waist circumference was measured for 1,448 participants, blood glucose levels for 1,346; HbA1c for 1,294; and total cholesterol for 1,260. Of the 1,503 participants, 1,239 completed at least one 24-h recall, and 1,148 (76.4%) at least two 24-h recalls. Only the latter were considered for the estimation of dietary intake. Of these, 48 people were excluded as underreporters, resulting in a final sample size for the dietary assessment of 1,100 participants (Figure 1). 17.6% of the participants who completed the house visit were not reached for the dietary assessment. Some variables had missing data points. The sample sizes for each variable were as follows: waist circumference (N = 1,448), education (N = 1,502), equivalized net household income (N = 1,380), civil status (N = 1,501), employment (N = 1,502), and smoking status (N = 1,502). Absolute numbers may differ when weighting is applied. Further details are shown in the Supplementary results.

Figure 1

3.1.2 Demographics

The total BVS III study sample is described in Table 2 without having applied weighting factors. The sample consisted of 1,503 participants, with 46% male (n = 688) and 54% female (n = 815). The mean age of the participants was 48.0 years, with no significant difference between males and females (p = 0.484). Age distribution across the defined age groups (18–24, 25–34, 35–50, 51–64, ≥65 years) was similar between sexes (p = 0.920).

Most participants had a higher education (48%). Education levels differed significantly by sex (p = 0.004). A higher proportion of males (52%) had a higher education than females (46%). Conversely, a greater proportion of females had midlevel education.

Married individuals made up 61% of the sample. A larger proportion of males were single (17% vs. 13% of females), while more females were divorced (9% vs. 6% of males) or widowed (4% vs. 1% of males) (p < 0.001).

Most participants lived in private households with family or friends (83%), with a small proportion living alone (16%) or with other arrangements (1%).

Two-thirds of the subjects were employed. Employment status showed significant differences between sexes (p < 0.001). A higher percentage of males were employed (69% vs. 63% in females), while more females were marginally, occasionally, or irregularly employed (4% vs. 1% in males). Additionally, more females were currently unemployed, including those on parental leave (7% vs. 4% in males).

The demographic description of the study sample that completed the dietary assessment with at least two 24-h recalls is shown in Supplementary Table 1 and was comparable to the complete study sample with only minor deviations. Weighting factors were applied to ensure further results were representative of the Bavarian population. Consequently, further analyses used weighted data to make inferences about the Bavarian population.

The average BMI in the Bavarian population was 25.9 kg/m² (Table 3). Half of the population was overweight, i.e., they had a BMI ≥25 kg/m². The proportion of overweight people was lower in females (41%) than in males (59%, p < 0.001). The average BMI in females was 25.4 kg/m² and significantly lower than the males’ average of 26.5 kg/m² (p < 0.001). Nearly a quarter of the population were smokers during the survey. Around 50% of the individuals were physically active or very active (Table 3 and Supplementary Table 2).

3.2 Dietary sustainability

3.2.1 Dietary greenhouse gas emissions, land, and water use in Bavaria

Table 4 presents data on GHGE, LU, and the WFP differentiated by various demographic and dietary groups.

In Bavaria, the average environmental impact per 2,500 kcal of dietary intake was 6.14 kg CO₂eq for GHGE, 7.50 m²*yr. for LU, and 4.39 kiloliters for the WFP. There were notable differences in environmental impacts between the sexes. While GHGE per 2,500 kcal was similar between males (6.17 kg CO₂eq) and females (6.11 kg CO₂eq), LU was significantly higher (p < 0.001) in males (7.91 m²*yr. per 2,500 kcal) compared to females (7.08 m²*yr. per 2,500 kcal). However, the WFP was higher in females, with 4.67 kiloliters per 2,500 kcal compared to 4.13 kiloliters for males.

Plant-based diets resulted in significantly lower GHGE (4.72 kg CO₂eq per 2,500 kcal) and LU (5.07 m²*yr. per 2,500 kcal) compared to omnivore diets (GHGE: 6.24 kg CO₂eq per 2,500 kcal; LU: 7.67 m²*yr. per 2,500 kcal) (p < 0.001). The WFP was marginally lower in people following such plant-based diets (4.23 kiloliters per 2,500 kcal) compared to their omnivorous counterparts (4.40 kiloliters per 2,500 kcal), albeit not significantly (p = 0.286).

Mean energy intake differed between BMI groups with 1,575 kcal for underweight, 1,797 kcal for normal weight, 1,857 kcal for pre-obese, and 1,854 kcal for obese participant. Unadjusted, all environmental metrics showed higher values in the higher BMI groups, with significant p-trends (all p-trend ≤0.031, see Table 4). Adjusted to 2,500 kcal, obese individuals had the highest GHGE per 2,500 kcal (6.51 kg CO₂eq) compared to other BMI groups, with the difference being significant (p < 0.001). However, there was no statistically substantiated positive trend (p-trend = 0.169) across the BMI groups. LU was also significantly higher (p < 0.001) in obese (8.39 m²*yr. per 2,500 kcal), and a positive trend was observed (p-trend = 0.005). However, no differences were observed across the groups with respect to the WFP.

Differences by education level showed a decreasing trend, with higher education associated with lower GHGE (5.96 kg CO₂eq; p-trend = 0.015) and LU (7.15 m²*yr.; p-trend = 0.004).

To further analyze the contributing factors to dietary emissions, regression analyses were performed to control for potential confounders. Table 5 displays the result of three weighted linear regression analyses examining the association of sociodemographic factors and diet type with dietary GHGE, LU, and the WFP, all adjusted to 2,500 kcal. The results refined the results of the previous trend analyses in Table 4. For group comparisons, marginal means were used to adjust for other independent variables in the model. The regression models indicated that females had higher GHGE and WFP compared to males when controlling for the other independent variables. The marginal means for GHGE was 5.53 kg CO₂eq for females and 5.33 kg CO₂eq for males, with a significant difference (β = 0.204, p = 0.023). Similarly, the marginal mean for the WFP was 4.33 kiloliters for females and 3.87 kiloliters for males with a significant difference (β = 0.466, p < 0.001). However, there was no significant difference in LU.

Waist circumference was positively associated with GHGE (β = 0.011, p < 0.001) and LU (β = 0.035, p < 0.001). Plant-based diets were linked to significantly and considerably lower environmental impact. The marginal means for GHGE for participants adhering to a plant-based diet was 4.74 kg CO₂eq, compared to omnivores with 6.12 kg CO₂eq (β = −1.38, p < 0.001). Likewise, LU was 5.56 m²*yr. for plant-based diets compared to omnivores with 7.39 m²*yr. (β = −1.83, p < 0.001), and WFP was 3.93 kiloliters against 4.27 kiloliters (β = −0.335, p = 0.003). Education and the equalized net income did not have a statistically substantiated effect on any of the outcomes.

3.2.2 Characterization of the GHGE quintiles

Emphasis was placed on GHGE. Five GHGE groups were built by stratifying the participants based on quintiles of their CO₂eq emissions per 2,500 kcal: Lowest, low, medium, high, and highest. The quintiles are characterized by their environmental sustainability metrics in Table 6 and by their sociodemographic characteristics in Supplementary Table 3. Expectedly, GHGE showed a positive trend across the emission groups, as they were constructed based on the GHGE per 2,500 kcal, from 4.24 to 8.52 kg CO₂eq per 2,500 kcal, but also LU from 5.31 to 10.52 m²*yr. per 2,500 kcal and WU from 3.62 to 5.44 kiloliters per 2,500 kcal (all p-trend <0.001).

Figure 2 shows the perceived environmental impact of various nutritional behaviors across the GHGE quintiles. Figure 3 complements these findings by depicting the self-reported implementation of these behaviors. The lowest GHGE quintile perceived the environmental impact of exclusive consumption of seasonal fruits and vegetables as less impactful (33% vs. 37–55%, p = 0.002, Figure 2) and were less likely to intend or actually practice purchasing regional food (80% vs. 93–95%, p = 0.041, Figure 3). However, they were more inclined to reduce their meat consumption, with 91% either practicing or having the concrete intention to show this behavior, compared to 61–77% among higher emitters (p = 0.012, Figure 3). In contrast, the highest emitters placed greater importance on reducing food waste (71% vs. 48–67%, p = 0.050, Figure 2) and predominately purchased or intended to buy regional foods (94%, p = 0.041, Figure 3).

Figure 2

Figure 3

3.2.3 Contribution of food groups to GHGE and food consumption patterns by GHGE quintiles

Figure 4 displays the contribution of different food groups to GHGE. In Figure 4A, the proportional GHGE were standardized to 2,500 kcal to reflect the relative dietary composition, while Figure 4B illustrates GHGE per day without energy adjustment. The overall distribution of GHGE across food groups did not differ greatly between the two approaches and was also similar when stratified by sex (Supplementary Figures 1, 2) or by BMI group (Supplementary Tables 4, 5). Animal-derived foods, i.e., meat, sausages, and meat products; fish and fish products; eggs; milk and dairy products; and butter, accounted for almost two-thirds of dietary GHGE. Among these, meat, sausages, and meat products contribute about 50% to the emissions, regardless of energy-adjustment. Stratified by sex and adjusted to 2,500 kcal (Figure 5), the overall GHGE distribution by sex was generally similar compared to the full study sample. Between the sexes, some differences were, however, notable. Meat, sausages, and meat products contributed to a lower degree to emissions in females (26.1%) compared to males (36.4%), whereas milk and dairy products accounted for a higher proportion in females (21.1%; males: 17.7%). Additionally, the contribution of tea and other infusions, vegetables, fruits, and, to a lesser extent, coffee was higher in females, while the proportion of non-alcoholic and alcoholic beverages (excluding coffee and tea) to GHGE was greater in males. Stratified by BMI groups and adjusted to 2,500 kcal, the consumption of meat, sausage and meat products was strikingly higher in pre-obese (36.0%) and obese (38.7%) participants compared to those with normal weight (24.9%, Supplementary Tables 4, 5).

Figure 4

Figure 5

Visual depictions of the food and beverage consumption adjusted to 2,500 kcal across the GHGE quintiles are presented in Figure 6 for foods of animal origin and in Figure 7 for plant-based foods and beverages. High emitters consumed more foods of animal origin, including butter, fish and fish products, meat and sausage products, meat, i.e., beef and veal, pork, poultry, and other meat (all p-trend ≤0.001), as well as milk and dairy products (p-trend = 0.049). They also had a noticeable higher intake of roasted coffee and tea and other infusion (all p-trend <0.001). Low emitters consumed significantly more grain-based staple foods, substitute products, nuts, kernels, and seeds (p-trend ≤0.007), and indicatively more legumes and pulses (p-trend = 0.064). Alcoholic and non-alcoholic beverages, eggs, fruit, potatoes, vegetables, and whole-grain products showed no significant trends (p-trend ≥0.140). A detailed tabular depiction of all food group intakes can be seen in Supplementary Table 6.

Figure 6

Figure 7

4 Discussion

4.1 Response

While modest, the BVS III’s overall response rate of 26.0% reflects both the anticipated challenges of modern survey research and careful planning. While the response rate of the predecessor study BVS II in 2002/2003 was substantially higher with 51.3% (70.9% of cleaned gross sample) (42), low response rates are typical in contemporary survey environments (43–46). Postema et al. (46) described that the response rate in the PIENTER studies, three nationwide serological surveys in the Netherlands, declined from 55.0% in PIENTER 1 (1995/1996) to 33.9% in PIENTER 2 (2006/2007) to 15.8% in the most recent PIENTER 3 (2016/2017). Similarly, the response in the National Maternity Surveys in England dropped from 67% in 1995 to 29% in 2018. Definite reasons for the decline in response rates remain unclear. However, a variety of factors are deemed important, including survey fatigue (particularly during the COVID-19 pandemic) (47, 48), privacy concerns, topic sensitivity (49), participation effort, and the logistical challenges of reaching diverse population groups. These factors often introduce non-response biases, skewing the study sample toward specific characteristics. The BVS III participants reflect this common profile seen in many surveys and studies, typically being more often female, older, married, and higher educated (43, 50).

As these trends are anticipated challenges of modern survey research, they necessitate preventive or corrective measures. Accordingly, more addresses from the residents’ registration offices were requested for subject acquisition, ensuring that the final sample size would reach the targeted 1,500 participants even with anticipated dropouts. This approach resulted in a final sample of 1,503 participants, providing a solid basis for the study’s conclusions. Moreover, the use of weighting factors helped to counterbalance non-response biases, making the findings more reflective of the general Bavarian population.

The COVID-19 pandemic introduced additional challenges before and during the survey’s field phase. Notably, 718 individuals cited quarantine as their reason for non-participation, the largest share of the QNDs. This may reflect both the genuine impact of the pandemic and the possibility that some potential subjects might have used the pandemic as a convenient excuse to opt out. Albeit COVID-19 quarantine was categorized as QND, this dropout might have disproportionately affected specific groups. Within this study, the reason for non-participation was not further analyzed. To the authors’ knowledge, there is no evidence in the literature suggesting that COVID-19-related measures, i.e., quarantine or “lockdown,” were wrongfully used as justification for non-participation in other studies. Gao et al. reported in 2023 that COVID-19 pandemic-related variables (i.e., previous COVID-19 infection or job loss due to the COVID-19 pandemic) positively correlated with the dropout rate among psychosomatic rehabilitation patients in a web-based study (51). Krieger et al. compared the response rates of six major surveys in the US before and during the onset of the COVID-19 pandemic. All but one ongoing surveys reported a relative decline in response rates of approximately 29%, with the largest declines seen among individuals with lower income and education levels (52).

4.2 Dietary sustainability

This work used established databases that provided sustainability metrics for the foods consumed in the BVS III. The choice of underlying databases influences environmental sustainability estimates. LCA methodologies differ and include different system boundaries, data sources, and assumptions. For example, the SHARP-ID and SuEATableLife databases, while both providing valuable insights, differ in their scope and approach. SHARP-ID gives estimates on dietary GHGE and LU based on EU-averages and encompasses the entire life cycle of food products, beginning with primary production and includes processing, packaging, transport, and post-market processing (36). However, the use of EU-averages in SHARP-ID means that the environmental impacts of imported foods, such as beef from non-EU countries, are not fully captured, potentially underestimating the footprint of diets reliant on such imports. In contrast, the SuEATableLife database was used to assess the WFP in this work. SuEATableLife also gives estimates for GHGE, which were not used in this work. System boundaries extend only to the distribution center, excluding post-market processing such as cooking (37). These differences lead to variations in environmental impact estimates. This was highlighted by Carvalho et al. by comparing commonly used dietary sustainability databases, i.e., SHARP-ID (36), SuEATableLife (37), and values provided by Poore and Nemecek (53). Same food consumption data were linked to these databases and dietary GHGE were compared, which led to notable discrepancies in GHGE estimates with medians of 4.42 kg CO₂eq for SHARP-ID, 5.64 kg CO₂eq for SuEATableLife, and 6.17 kg CO₂eq for Carvalho et al. (54). These variations reflect the importance of carefully selecting databases for dietary sustainability assessments. The highlighted differences in system boundaries, data sources, and overall purpose of selected database significantly influence results (12). For instance, as SHARP-ID uses EU-averages (36), it is suitable for intra-EU comparisons, while global databases like Poore and Nemecek (53) serve broader, potentially cross-regional intent. Additionally, SuEATableLife provides WFP data but does not differentiate between the type of water usage (37), i.e., green, blue, and gray water (55), which may limit its applicability for certain analyses. SHARP-ID was chosen in this study to ensure comparability within the EU and to align with other, EU-based research. SuEATableLife was selected because it provides comprehensive WFP data for a wide range of food items to complement GHGE and LU estimates by SHARP-ID.

In this study, the sustainability metrics were standardized to a fixed energy intake of 2,500 kcal. Also, emission groups were defined based on quintiles of the standardized GHGE per 2,500 kcal. These methodological choices allowed for an energy-independent comparison of dietary environmental impact of the dietary composition rather than absolute food consumption. In contrast, assessing GHGE per day without adjustment reflects the total environmental burden of an individual’s diet but may be influenced by variations in energy intake and is susceptible to known biases in dietary surveys based on 24-h recalls, such as social desirability bias (56) or recall bias (57), where participants consciously or unconsciously over- or underreport their dietary intakes. By standardizing GHGE to 2,500 kcal, we aimed to mitigate the influence of these biases and provide a more accurate comparison of dietary sustainability across participants. When using GHGE per day, individuals with a high total energy intake, possibly due to high energy requirements, may be placed in higher GHGE groups, even if their diet has a relatively low environmental impact when standardized. Conversely, while standardizing to 2,500 kcal accounts for differences in energy intake and focuses on the dietary composition itself, it can, however, lead to shifts in the classification. Participants consuming large amounts of foods with low energy-density or having an overall low caloric intake could possibly have low GHGE per day but may be placed in higher GHGE groups per 2,500 kcal. While the food group contribution to GHGE differed only marginally between the energy-adjusted versus unadjusted/per day approach, these methodological differences highlight the importance of carefully interpreting results and comparing studies. These methodological considerations align with findings from a recent study, which demonstrated a strong correlation between energy intake and GHGE per day. The study also showed that the assessment of micronutrient adequacy varied depending on whether energy intake was standardized. However, micronutrient intake was not examined in our study, and our methodological approach remains valid within its intended framework. Rather, these findings underscore the importance of carefully considering whether energy standardization is appropriate based on the specific research question (58). Several studies have applied similar energy standardization when assessing the environmental impact of diets, enabling more consistent cross-study comparisons (15, 17, 19). The EAT-Lancet Commission provided dietary recommendations based on an intake of 2,500 kcal per day, known as the Planetary Health Diet (1). Standardizing GHGE to 2,500 kcal allowed for direct comparisons with this benchmark while controlling for variations in energy intake.

In 2012, the FAO emphasized assessing diets for environmental sustainability and nutritional value, defining sustainable diets as those minimizing environmental impact, supporting food security, and promoting health for current and future generations (59). It is now generally accepted that sustainable nutrition should not only encompass the environmental impact but also be nutritionally adequate, socially viable, and economically performative (60–62). In line with this, the EAT-Lancet Commission (1) provided the Planetary Health Diet for an intake of 2,500 kcal/d, which integrated all facets of sustainable diets to facilitate sustainable food systems by 2050 (1). Based on this framework, the assumption of a global population of 10 billion people in 2050 and a GHGE boundary for the food production of 5 Gt CO₂eq per year (1), dietary GHGE should not exceed 1.37 kg CO₂eq per person and day with an intake of 2,500 kcal. The average dietary GHGE in Bavaria amounts to 6.14 kg CO₂eq per 2,500 kcal, which is approximately 4.5 times higher than the defined boundary by the EAT-Lancet Commission. The lowest GHGE quintile in Bavaria still emitted 4.24 kg CO₂eq per 2,500 kcal, exceeding the planetary boundary more than three times. This considerable over-emission can be largely attributed to the high consumption of animal-derived foods, which accounted for almost two-thirds of dietary GHGE in the BVS III. These products were disproportionately consumed in higher amounts among high emitters. Due to their high environmental impact and potentially health-impairing effects, animal-derived foods are not emphasized in the Planetary Health Diet (1). Conversely, foods emphasized in the Planetary Health Diet, such as legumes, nuts, and potatoes (1), were not considerably consumed overall, except for legumes and meat substitutes in the lowest emitter group. Substitute products for meat and dairy were also not considerably consumed. Due to the low consumption, these foods had little to no contribution to GHGE. This disparity highlights the need for dietary shifts toward more sustainable food choices to align with planetary health goals.

Comparable data in Germany are limited. Koelman et al. (19) reported notably higher dietary emissions compared to the BVS III of 6.6 kg CO₂eq for men and 7.0 kg CO₂eq per 2,000 kcal in an Eastern German population, which extrapolates to 8.25 and 8.75 kg CO₂eq per 2,500 kcal for men and women, respectively. Comparably, women in the BVS III emitted 6.11 kg CO₂eq per 2,500 kcal, while men’s emissions were 6.17 kg CO₂eq. However, the study’s design and population, primarily aged 35–65 years, limit comparisons. The study population was comprised of a subsample of the EPIC (1994–1998) study cohort members who were still actively participating in the follow-up (2010–2012). It relied on FFQs to assess dietary intake, unlike the 24-h dietary recalls used in the BVS III (19).

Similarly, the Bavarian dietary GHGE of 6.14 kg CO₂eq per 2,500 kcal and LU of 7.50 m²yr per 2,500 kcal were generally comparable to those reported in other European studies by Mertens et al. (17). Mertens et al. (17) energy-adjusted GHGE and LU values to 2,000 kcal. These values were projected to be 2,500 kcal for comparison. Bavarian food-related emissions resembled those in Denmark (6.25 kg CO₂eq and LU of 7.88 m²*yr. per 2,500 kcal) and Italy (6.13 kg CO₂eq and LU of 7.88 m²*yr. per 2,500 kcal) while superseding those in the Czech Republic (5.5 kg CO₂eq and LU of 7.13 m²*yr. per 2,500 kcal). France reported the highest values, with an average GHGE of 8.0 kg CO₂eq and LU of 10.0 m²*yr. per 2,500 kcal. Again, these comparisons are constrained by differences in study populations, design, and data collection periods (2003–2008) (17). Dietary data were, however, linked to the same sustainability metric database, SHARP-ID, for all studies mentioned.

The current study found that high emitters per 2,500 kcal were more likely to be female, indicatively older, and have a higher BMI or waist circumference, which aligns with findings from Mertens et al. (17), where energy-adjusted GHGE increased with age in Denmark and France. Additionally, women in the Czech Republic and France exhibit higher GHGE, supporting the depicted sex differences. Also, Koelman et al. (19) reported higher energy-adjusted GHGE in women and with higher BMI. The current study found no significant effect of education when controlling for other factors. This was confirmed to a certain extent: Education levels positively correlated with GHGE in the Czech Republic (17) and Northern Sweden (15) but negatively correlated in France (17). However, sex-specific differences in dietary sustainability metrics are controversially discussed. While the sex-specific GHGE and WFP increases in females were highly significant in this study (p ≤ 0.023), the regression models offered low explanation of variance in dietary GHGE (adjusted R² = 0.10) and WFP (adjusted R² = 0.08). This indicates that the statistically significant sex differences explained only a modest proportion of the variance in dietary GHGE and WFP. Other factors not included in the models may contribute substantially to the observed variability. However, other studies reported an inverse relationship. Exemplarily, men showed higher energy-adjusted GHGE in the USA (63). Likewise, in a young Spanish study population, females consumed diets with lower GHGE density significantly more often (64). Against this, Koelman et al. found higher energy-adjusted emissions for Eastern German women (19). Here, by adding the food groups “tea and other infusion,” “roasted coffee” and “milk and dairy products” to the regression analysis as additional predictors, men and women no longer differed in terms of CO₂eq emissions (β = 0.022, p = 0.827) (data not shown). This suggests that the previously observed sex-specific differences in CO2eq emissions might be driven primarily by differences in the consumption patterns of these specific high-impact food groups and can largely be attributed to the differences in the consumption of these food groups. Likewise, the sex-specific differences in the WFP were reduced but not entirely resolved by controlling for these food groups (β = 0.159, p = 0.001) (data not shown). This indicates that while these food groups play a significant role in explaining sex-specific differences in WFP, other food groups or factors may also contribute to the remaining variance. Rippin et al. (14) analyzed the GHGE of individual diets in the UK and found that beverages, particularly tea, coffee, and alcoholic drinks, were major contributors. These items, along with cakes, biscuits, and confectionery, accounted for a quarter of the diet-related GHGE. The authors suggested that targeting these food and drink categories could offer alternative strategies for reducing diet-related GHGE, which is supported by the results depicted in this study. Despite limited information on the direct impact of coffee and tea on dietary GHGE at the individual level, the issue has been widely discussed (65–68). As tropical and subtropical plants, coffee and tea are commonly grown in monocultures, negatively affecting biodiversity (68) and fostering deforestation, erosion, and river pollution (69–71) in these regions. The high reported consumption of these beverages in Bavaria, especially in the high and highest GHGE quintiles, consequently, exerts additional environmental stress. Largely, previous studies lack a comprehensive analysis of the LU, GHGE, and WFP associated with the consumption of coffee and tea (14, 67), leaving gaps in understanding their full environmental effects, and underscoring the need for further research on their ecological and climate-related effects. Furthermore, the positive association between waist circumference and GHGE observed in the regression model (β = 0.01, p < 0.001) was most probably driven by the disproportionately high contribution of meat, sausages, and meat products to dietary emissions in pre-obese and obese (Supplementary Tables 4, 5). This is consistent with additional data from the BVS III, which showed that obese and pre-obese individuals consumed larger quantities of meat compared to those with normal weight, regardless of energy-adjustment (data not shown). This is in line with a plethora of publication that associated higher BMI with greater meat consumption, particularly red and processed meats (72–74), which was also the case in the BVS III (data not shown). In line with the findings from the BVS III, Strid et al. (15) showed in a Northern Swedish population that pre-obese and obese participants had the highest energy-adjusted GHGE. Energy was, however, adjusted to 1,000 kcal.

Sustainable food choices are determined by a complex interplay between individual perceptions of the environmental impact of foods and their actual behaviors. The results of this study aligned with broader research that highlighted discrepancies between individuals’ awareness of the environmental impact of dietary choices, their intentions, and actual behaviors (25, 75). Notably, low emitters in this study demonstrated a more accurate perception of certain practices. For instance, they perceived the environmental impact of seasonal fruit and vegetable consumption as lower, while higher emitters tended to attribute a higher significance to this behavior. Additionally, low emitters in the study showed a greater awareness of the environmental impact of reduced meat consumption, which was also reflected in their behavior, indicating a closer alignment between perceptions and effective behaviors related to sustainability. This is consistent with research showing that consumers often overestimated the benefits of regional and seasonal products while simultaneously underestimating the impact of the reduction of meat (76, 77) and dairy consumption (78). Although higher emitters expressed intentions to engage in sustainable practices, their assessment of the relative impact of these habits appeared less accurate. These findings underscore the importance of enhancing consumer education regarding the environmental consequences of dietary choices to promote more sustainable eating behaviors.

4.3 Limitations

The BVS III was conducted partly during the COVID-19 pandemic, which potentially affected food availability and dietary behaviors. The unique circumstances of the pandemic, including lockdowns and changes in food supply chains, should be considered when interpreting the results. While the reliance on 24-h dietary recalls is prone to reporting bias, this limitation was mitigated by employing trained interviewers to conduct the recalls with participants, computation of the individual recalls into weighted averages to account for weekday-specific changes in the diet, omitting of extreme underreporters, and energy-adjusted data analyses. The survey employed a robust methodology, incorporating strategic oversampling and weighting factors to adjust for demographic imbalances and account for deviations from the underlying population, resulting in data closely aligned with Bavarian population characteristics. Nonetheless, the BVS III achieved a response rate of 26.0%, reflecting the broader trends in declining survey participation rates observed globally. The sustainability databases SHARP-ID and SuEATable Life were used in this work. The calculated dietary greenhouse gas emissions and resource use are highly dependent on the database used.

4.4 Conclusion

This study assessed the dietary GHGE, LU, and WFP in Bavaria, identifying significant differences across demographic and dietary intake groups. The average GHGE was 6.14 kg CO₂eq, LU was 7.50 m²*yr., and WFP was 4.39 kiloliters per 2,500 kcal. According to the Planetary Health Diet framework from the EAT-Lancet Commission (1), the average Bavarian exceeded the recommended GHGE boundary by 4.5 times, with even the lowest GHGE quintile surpassing it threefold, indicating an urgent call for action. Foods of animal origin have expectedly emerged as significant drivers of dietary GHGE. This, on the one hand, underscored the need to address their ecological impact, especially concerning the persistent underestimation of meat consumption on dietary GHGE and resource use among the highest emitters. Low emitters had a more accurate perception of the environmental benefits of reducing meat consumption, and their behavior aligned with this awareness, as a larger proportion actively reduced their meat intake. On the other hand, provision of information alone is unlike to drive meaningful behavior change, as education showed no association with GHGE when controlling for confounders. To support ecological sustainability and public acceptance, meat consumption should be critically addressed, realistically categorized, but not stigmatized or restricted. In addition, more sustainable food options, such as plant-based substitutes, or dietary behaviors, such as meat reduced diets (“flexitarian”) or meat consumption only for special occasions or festivities, should be made more attractive, accessible, and socially acceptable. Along with this, improving the quality of meat by improving the quality of livestock farming should be further pursued. These are all to a small extent the sole responsibility of the individual. Rather, it is the responsibility of the governing authorities to create framework conditions through financial support or incentives that ease sustainable food choices and improve the food environment. In this work, the equalized net household income had no association with GHGE. Creating financial incentives therefore does not exacerbate existing fiscal inequalities and could be a possible lever. Beyond products of animal origin, high emitters notably consumed more coffee and tea and other infusions, suggesting they may be overlooked contributors to dietary GHGE and resource use. Higher emissions per 2,500 kcal were associated with higher BMI or waist circumference, being female, and following an omnivorous diet. These findings point to a need for more nuanced discourse that addresses sustainable dietary shifts across relevant demographic groups with more targeted approaches.

References

• 1.

  WillettWRockströmJLokenBSpringmannMLangTVermeulenSet al. Food in the Anthropocene: the eat lancet commission on healthy diets from sustainable food systems. Lancet. (2019) 393:447–92. doi: 10.1016/S0140-6736(18)31788-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2.

  FAO. The future of food and agriculture—drivers and triggers for transformation. Rome, Italy: FAO (2022).

  • Google Scholar

• 3.

  MeierTChristenOSemlerEJahreisGVoget-KleschinLSchrodeAet al. Balancing virtual land imports by a shift in the diet. Using a land balance approach to assess the sustainability of food consumption. Germany as an example. Appetite. (2014) 74:20–34. doi: 10.1016/j.appet.2013.11.006

  • CrossRef
  • Google Scholar

• 4.

  SpringmannM. Towards healthy and sustainable diets in Germany. Dessau-Roßlau, Germany: Umweltbundesamt (2023).

  • Google Scholar

• 5.

  WeissFLeipA. Greenhouse gas emissions from the Eu livestock sector: a life cycle assessment carried out with the Capri model. Agric Ecosyst Environ. (2012) 149:124–34. doi: 10.1016/j.agee.2011.12.015

  • CrossRef
  • Google Scholar

• 6.

  TompaOLaknerZOláhJPoppJKissA. Is the sustainable choice a healthy choice?-water footprint consequence of changing dietary patterns. Nutrients. (2020) 12:2578. doi: 10.3390/nu12092578

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7.

  AleksandrowiczLGreenRJoyEJMSmithPHainesA. The impacts of dietary change on greenhouse gas emissions, land use, water use, and health: a systematic review. PLoS One. (2016) 11:e0165797. doi: 10.1371/journal.pone.0165797

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 8.

  PradhanPReusserDEKroppJP. Embodied Greenhouse Gas Emissions in Diets. PLoS One. (2013) 8:e62228. doi: 10.1371/journal.pone.0062228

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9.

  HallströmECarlsson-KanyamaABörjessonP. Environmental impact of dietary change: a systematic review. J Clean Prod. (2015) 91:1–11. doi: 10.1016/j.jclepro.2014.12.008

  • CrossRef
  • Google Scholar

• 10.

  ScarboroughPClarkMCobiacLPapierKKnuppelALynchJet al. Vegans, vegetarians, fish-eaters and meat-eaters in the Uk show discrepant environmental impacts. Nat Food. (2023) 4:565–74. doi: 10.1038/s43016-023-00795-w

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11.

  LaineJEHuybrechtsIGunterMJFerrariPWeiderpassETsilidisKet al. Co-benefits from sustainable dietary shifts for population and environmental health: an assessment from a large European cohort study. Lancet Planet Health. (2021) 5:e786–96. doi: 10.1016/S2542-5196(21)00250-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12.

  KliejunasECleghornCDrewJMMhurchuCNBradburyKE. The relationship between dietary greenhouse gas emissions and demographic characteristics in high-income countries. Proc Nutr Soc. (2024) 1–9. doi: 10.1017/S0029665124007523

  • CrossRef
  • Google Scholar

• 13.

  FrehnerAZantenHHEVSchaderCBoerIJMDPestoniGRohrmannSet al. How food choices link sociodemographic and lifestyle factors with sustainability impacts. J Clean Prod. (2021) 300:126896. doi: 10.1016/j.jclepro.2021.126896

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14.

  RippinHLCadeJEBerrang-FordLBentonTGHancockNGreenwoodDC. Variations in greenhouse gas emissions of individual diets: associations between the greenhouse gas emissions and nutrient intake in the United Kingdom. PLoS One. (2021) 16:e0259418. doi: 10.1371/journal.pone.0259418

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15.

  StridAHallströmEHjorthTJohanssonILindahlBSonessonUet al. Climate impact from diet in relation to background and sociodemographic characteristics in the Västerbotten intervention Programme. Public Health Nutr. (2019) 22:3288–97. doi: 10.1017/S1368980019002131

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16.

  TemmeEHMToxopeusIBKramerGFHBrosensMCCDrijversJMMTyszlerMet al. Greenhouse gas emission of diets in the Netherlands and associations with food, energy and macronutrient intakes. Public Health Nutr. (2015) 18:2433–45. doi: 10.1017/S1368980014002821

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17.

  MertensEKuijstenAvan ZantenHHEKaptijnGDofkováMMisturaLet al. Dietary choices and environmental impact in four European countries. J Clean Prod. (2019) 237:117827. doi: 10.1016/j.jclepro.2019.117827

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18.

  KirwanLBWaltonJFlynnANugentAPKearneyJHoldenNMet al. Assessment of the environmental impact of food consumption in Ireland—informing a transition to sustainable diets. Nutrients. (2023) 15:981. doi: 10.3390/nu15040981

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19.

  KoelmanLHuybrechtsIBiesbroekSVan ‘t VeerPSchulzeMBAleksandrovaK. Dietary choices impact on greenhouse gas emissions: determinants and correlates in a sample of adults from eastern Germany. Sustain For. (2022) 14:3854. doi: 10.3390/su14073854

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20.

  MeierTChristenO. Gender as a factor in an environmental assessment of the consumption of animal and plant-based foods in Germany. Int J Life Cycle Assess. (2013) 17:550–64. doi: 10.1007/s11367-012-0387-x

  • CrossRef
  • Google Scholar

• 21.

  TreuHNordborgMCederbergCHeuerTClaupeinEHoffmannHet al. Carbon footprints and land use of conventional and organic diets in Germany. J Clean Prod. (2017) 161:127–42. doi: 10.1016/j.jclepro.2017.05.041

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22.

  EberleUFelsJ. Environmental impacts of German food consumption and food losses. Int J Life Cycle Assess. (2016) 21:759–72. doi: 10.1007/s11367-015-0983-7

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23.

  ShepherdR. Resistance to changes in diet. Proc Nutr Soc. (2002) 61:267–72. doi: 10.1079/PNS2002147

  • CrossRef
  • Google Scholar

• 24.

  SchäferABoeingHConradJWatzlB. Wissenschaftliche grundlagen der lebensmittelbezogenen Ernährungsempfehlungen für deutschland. Wiesbaden, Germany: Ernährungs Umschau (2024).

  • Google Scholar

• 25.

  AsvatourianVCraigTHorganGWKyleJMacdiarmidJI. Relationship between pro-environmental attitudes and behaviour and dietary intake patterns. Sustain Prod Consum. (2018) 16:216–26. doi: 10.1016/j.spc.2018.08.009

  • CrossRef
  • Google Scholar

• 26.

  BaroneBNogueiraRMGuimarãesKBehrensJH. Sustainable diet from the urban Brazilian consumer perspective. Food Res Int. (2019) 124:206–12. doi: 10.1016/j.foodres.2018.05.027

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27.

  ToblerCVisschersVHMSiegristM. Eating green. Consumers’ willingness to adopt ecological food consumption behaviors. Appetite. (2011) 57:674–82. doi: 10.1016/j.appet.2011.08.010

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28.

  WellerKEGreeneGWReddingCAPaivaALLofgrenINashJTet al. Development and validation of Green eating behaviors, stage of change, decisional balance, and self-efficacy scales in college students. J Nutr Educ Behav. (2014) 46:324–33. doi: 10.1016/j.jneb.2014.01.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29.

  VossSCharrondiereURSlimaniNKrokeARiboliEWahrendorfJet al. Epic-soft a European dietary assessment instrument for 24-H recalls. Z Ernahrungswiss. (1998) 37:227–33. doi: 10.1007/s003940050021

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30.

  RohmFWawroNGimpflSOhlhautNSengerMRögerCet al. Does the habitual dietary intake of adults in Bavaria, Germany, match dietary intake recommendations? Results of the 3rd Bavarian food consumption survey. Front Nutr. (2025) 11:11. doi: 10.3389/fnut.2024.1537637

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31.

  Bayerisches Landesamt für Statistik. Bevölkerungsstand Fürth (2023) [cited 2024 14.07.2024]. Available online at: https://www.statistikdaten.bayern.de/genesis/online?language=de&sequenz=tabellen&selectionname=12411*#abreadcrumb (Accessed July 14, 2024).

  • Google Scholar

• 32.

  Max Rubner-Insitut. Bundeslebensmittelschlüssel (2014) [cited 2024 16.07.2024]. Available online at: https://blsdb.de (Accessed July 16, 2024).

  • Google Scholar

• 33.

  GerriorSJuanWBasiotisP. An easy approach to calculating estimated energy requirements. Prev Chronic Dis. (2006) 3:A129.

  • Google Scholar

• 34.

  eurostat. Glossary:Equivalised Disposable Income: ec.europa.eu (2021) [cited 2024 08.10.2024]. Available online at: https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:Equivalised_disposable_income (Accessed October 08, 2024).

  • Google Scholar

• 35.

  Energy and Protein Requirements. Report of a joint Fao/who/Unu expert consultation. World Health Organ Tech Rep Ser. (1985) 724:1–206.

  • Google Scholar

• 36.

  MertensEKaptijnGKuijstenAvan ZantenHGeleijnseJMvan 't VeerP. Sharp-indicators database towards a public database for environmental sustainability. Data Brief. (2019) 27:104617. doi: 10.1016/j.dib.2019.104617

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37.

  PeterssonTSecondiLMagnaniAAntonelliMDembskaKValentiniRet al. A multilevel carbon and water footprint dataset of food commodities. Scientific Data. (2021) 8:127. doi: 10.1038/s41597-021-00909-8

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38.

  Agri-FootprintB. Agri-footprint 2.0: Part 1: Methodology and basic principles. Gouda, The Netherlands: Mérieux NutriSciences | Blonk (2015).

  • Google Scholar

• 39.

  Agri-FootprintB. Agri-footprint 2.0: Part 2: Description of data. Gouda, The Netherlands: Mérieux NutriSciences | Blonk (2015).

  • Google Scholar

• 40.

  WeidemaBPBauerCHischierRMutelCNemecekTReinhardJet al. Overview and methodology: Data quality guideline for the Ecoinvent database version 3. St. Gallen: The ecoinvent Centre (2013).

  • Google Scholar

• 41.

  R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing (2024).

  • Google Scholar

• 42.

  KargGWolframGGedrichKLinseisenJ. Bayerische Verzehrsstudie (Bvs) Ii Abschlussbericht. Stuttgart, Germany: Ernährungsdenkwerkstatt (2004).

  • Google Scholar

• 43.

  HarrisonSAlderdiceFHendersonJRedshawMQuigleyMA. Trends in response rates and respondent characteristics in five National Maternity Surveys in England during 1995-2018. Arch Public Health. (2020) 78:46. doi: 10.1186/s13690-020-00427-w

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 44.

  IshikawaMHemmiOTakimotoHMatsumotoMYokoyamaT. Trend of estimated participation rate by regional block, gender, and age group in the 1997-2019: National Health and nutrition survey in Japan. PLoS One. (2024) 19:e0286169. doi: 10.1371/journal.pone.0286169

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 45.

  KreuterF. Facing the nonresponse challenge. Ann Am Acad Pol Soc Sci. (2013) 645:23–35. doi: 10.1177/0002716212456815

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46.

  PostemaAFerreiraJAvan der KlisFde MelkerHMollemaL. Investigating sources of non-response Bias in a population-based Seroprevalence study of vaccine-preventable diseases in the Netherlands. BMC Infect Dis. (2024) 24:249. doi: 10.1186/s12879-024-09095-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 47.

  Ben-NunP. Respondent fatigue In: Encyclopedia of survey research methods. eds. Paul J. Lavrakas (Thousand Oaks, California, USA: Sage Publications), vol. 2 (2008). 742–3. doi: 10.4135/9781412963947.n480

  • CrossRef
  • Google Scholar

• 48.

  de KoningREgizAKotechaJCiuculeteACOoiSZYBankoleNDAet al. Survey fatigue during the COVID-19 pandemic: an analysis of neurosurgery survey response rates. Front Surg. (2021) 8:690680. doi: 10.3389/fsurg.2021.690680

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49.

  CouperMPSingerEConradFGGrovesRM. Risk of disclosure, perceptions of risk, and concerns about privacy and confidentiality as factors in survey participation. J Off Stat. (2008) 24:255–75. PMID:

  • Pubmed Abstract
  • Google Scholar

• 50.

  AndreevaVASalanaveBCastetbonKDeschampsVVernayMKesse-GuyotEet al. Comparison of the sociodemographic characteristics of the large Nutrinet-Santé E-cohort with French census data: the issue of volunteer Bias revisited. J Epidemiol Community Health. (2015) 69:893–8. doi: 10.1136/jech-2014-205263

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51.

  GaoLKellerFMBeckerPDahmenALippkeS. Predictors of dropout among psychosomatic rehabilitation patients during the Covid-19 pandemic: secondary analysis of a longitudinal study of digital training. J Med Internet Res. (2023) 25:e43584. doi: 10.2196/43584

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52.

  KriegerNLeBlancMWatermanPDReisnerSLTestaCChenJT. Decreasing survey response rates in the time of COVID-19: implications for analyses of population health and health inequities. Am J Public Health. (2023) 113:667–70. doi: 10.2105/AJPH.2023.307267

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53.

  PooreJNemecekT. Reducing Food's environmental impacts through producers and consumers. Science. (2018) 360:987–92. doi: 10.1126/science.aaq0216

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 54.

  CarvalhoCCorreiaDCostaSALopesCTorresD. Assessing the environmental impact of diet – influence of using different databases of Foods' environmental footprints. J Clean Prod. (2023) 416:137973. doi: 10.1016/j.jclepro.2023.137973

  • CrossRef
  • Google Scholar

• 55.

  HoekstraAYMekonnenMM. The water footprint of humanity. Proc Natl Acad Sci USA. (2012) 109:3232–7. doi: 10.1073/pnas.1109936109

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 56.

  FisherRJ. Social desirability Bias and the validity of indirect questioning. J Consum Res. (1993) 20:303–15. doi: 10.1086/209351

  • CrossRef
  • Google Scholar

• 57.

  CoughlinSS. Recall Bias in epidemiologic studies. J Clin Epidemiol. (1990) 43:87–91. doi: 10.1016/0895-4356(90)90060-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 58.

  StubbendorffAHallströmETomovaGBornéYJanziSSonestedtEet al. Greenhouse gas emissions in relation to micronutrient intake and implications of energy intake: a comparative analysis of different modelling approaches. Am J Clin Nutr. (2025). doi: 10.1016/j.ajcnut.2025.02.031

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 59.

  BurlingameBDerniniS. Sustainable diets and biodiversity: directions and solutions for policy, research and action. Rome, Italy: Food and Agriculture Organization of the United Nations (FAO) (2012).

  • Google Scholar

• 60.

  CampbellCGFeldpauschG. Teaching nutrition and sustainable food systems: justification and an applied approach. Front Nutr. (2023) 10:10. doi: 10.3389/fnut.2023.1167180

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 61.

  CareyCNPaquetteMSahye-PudaruthSDadvarADinhDKhodabandehlouKet al. The environmental sustainability of plant-based dietary patterns: a scoping review. J Nutr. (2023) 153:857–69. doi: 10.1016/j.tjnut.2023.02.001

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 62.

  MazzocchiADe CosmiVScaglioniSAgostoniC. Towards a more sustainable nutrition: complementary feeding and early taste experiences as a basis for future food choices. Nutrients. (2021) 13:2695. doi: 10.3390/nu13082695

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 63.

  RoseDHellerMCWillits-SmithAMMeyerRJ. Carbon footprint of self-selected us diets: nutritional, demographic, and behavioral correlates. Am J Clin Nutr. (2019) 109:526–34. doi: 10.1093/ajcn/nqy327

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 64.

  Telleria-AramburuNBermúdez-MarínNRocandioAMTelletxeaSBasabeNRebatoEet al. Nutritional quality and carbon footprint of university Students' diets: results from the Ehu12/24 study. Public Health Nutr. (2022) 25:183–95. doi: 10.1017/S1368980021002640

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 65.

  BawaKSLiuJ. Environmental impacts beyond land use and conservation risk hotspots: coffee and tea. Proc Natl Acad Sci USA. (2023) 120:e2314246120. doi: 10.1073/pnas.2314246120

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 66.

  HoangNTTaherzadehOOhashiHYonekuraYNishijimaSYamabeMet al. Mapping potential conflicts between global agriculture and terrestrial conservation. Proc Natl Acad Sci USA. (2023) 120:e2208376120. doi: 10.1073/pnas.2208376120

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 67.

  ChapagainAKHoekstraAY. The water footprint of coffee and tea consumption in the Netherlands. Ecol Econ. (2007) 64:109–18. doi: 10.1016/j.ecolecon.2007.02.022

  • CrossRef
  • Google Scholar

• 68.

  EnerothHKarlsson PotterHRöösE. Environmental effects of coffee, tea and cocoa -data collection for a consumer guide for plant-based foods. Uppsala, Sweden: Swedish University of Agricultural Sciences, Department of Energy and Technology (2022).

  • Google Scholar

• 69.

  HoyosN. Spatial modeling of soil Erosion potential in a tropical watershed of the Colombian Andes. Catena. (2005) 63:85–108. doi: 10.1016/j.catena.2005.05.012

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 70.

  D’haezeDDeckersJRaesDPhongTLoiH. Environmental and socio-economic impacts of institutional reforms on the agricultural sector of Vietnam: land suitability assessment for Robusta coffee in the Dak Gan region. Agric Ecosyst Environ. (2005) 105:59–76. doi: 10.1016/j.agee.2004.05.009

  • CrossRef
  • Google Scholar

• 71.

  AdamsMGhalyAE. Maximizing sustainability of the Costa Rican coffee industry. J Clean Prod. (2007) 15:1716–29. doi: 10.1016/j.jclepro.2006.08.013

  • CrossRef
  • Google Scholar

• 72.

  DaneshzadEAskariMMoradiMGhorabiSRouzitalabTHeshmatiJet al. Red meat, overweight and obesity: a systematic review and Meta-analysis of observational studies. Clin Nutr ESPEN. (2021) 45:66–74. doi: 10.1016/j.clnesp.2021.07.028

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 73.

  WangYBeydounMA. Meat consumption is associated with obesity and central obesity among us adults. Int J Obes. (2009) 33:621–8. doi: 10.1038/ijo.2009.45

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 74.

  YouWHennebergM. Meat consumption providing a surplus energy in modern diet contributes to obesity prevalence: an ecological analysis. BMC Nutr. (2016) 2:22. doi: 10.1186/s40795-016-0063-9

  • CrossRef
  • Google Scholar

• 75.

  FinkLStrassnerCPloegerA. Exploring external factors affecting the intention-behavior gap when trying to adopt a sustainable diet: a think aloud study. Front Nutr. (2021) 8:511412. doi: 10.3389/fnut.2021.511412

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 76.

  AustgulenMHSkulandSESchjøllAAlfnesF. Consumer readiness to reduce meat consumption for the purpose of environmental sustainability: insights from Norway. Sustain For. (2018) 10:3058. doi: 10.3390/su10093058

  • CrossRef
  • Google Scholar

• 77.

  GrothCWegmannCMeyerdingSGH. Perception of product sustainability: the case of processed tomatoes – a qualitative study in Germany. J Retail Consum Serv. (2023) 71:103214. doi: 10.1016/j.jretconser.2022.103214

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 78.

  MehlhoseCKühlS. Tierfreier Käse? Verbraucherstudie Zur Akzeptanz Von Präzisionsfermentation Zur Herstellung Von Lebensmitteln. Chart. (2024)

  • Google Scholar

• 79.

  World Health Organization. Obesity and overweight (2024) [cited 2024 04.11.2024]. Available online at: https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight (Accessed November 11, 2024).

  • Google Scholar