The model is based on the Swiss context, since it is an exemplary market for early adoption of nutritional and dietary trends (high income, innovative, well interconnected, high awareness of environmental issues and animal welfare, good data availability, large share of imported foods). Only data on already purchasable alternative products in Switzerland were included in the investigation. To ensure relevance and avoid assumptions that could decrease the accuracy of our results, no preliminary or laboratory data were considered. The products were compared directly based on weight (as described in the literature12).
Datasets on alternatives were selected based on availability, as described below. A market analysis confirmed that the included datasets reflected the range of products available in Switzerland (SI2-T5). Weighted averages were not calculated, as the analysis captured product availability but not sales volumes.
Meat and dairy, and alternatives at the product level – Grouping
The products were grouped to facilitate analysis. The groups were based on their reference product, main ingredient, and type of processing, which we defined as the main method used to transform the raw ingredients into the final product. Each group contained between 1 and 20 individual datasets for the nutrient composition and the environmental assessment, respectively. Generally, more nutrient data than environmental data were available. We did not use common categorisation systems, such as NOVA5 or the Good Food Institute (www.gfi.org/investment) classification, since they would categorise most products into the same group, preventing a detailed analysis.
Single datasets that could not be associated with an ingredient-specific group were included in two broader groups: (1) plant-based and mechanically texturised meat alternatives, and (2) milk alternatives. Due to limited data, yoghurt, cheese, and cream alternatives could not be grouped based on raw material or processing type. The average nutrient content and environmental impacts were calculated for each group as unweighted mean values of the available datasets.
Meat and dairy, and alternatives at the product level – Nutrient data
Nutrient data at the product level were retrieved from the EuroFIR platform (www.eurofir.org), which provides harmonised access to national food composition databases. Through the platform, data from France, Greece, Portugal, Slovenia, Spain, Switzerland, and the United Kingdom were accessed. These countries were selected based on the availability of data for alternatives and to complement the Swiss database in that regard. Meat and dairy products were taken exclusively from the Swiss database. This ensured sufficient data for the relevant products while maintaining regional focus. Suitable datasets were identified by searching for appropriate terms and filtering the metadata (SI1-1.3.1). The process was repeated twice to ensure completeness. Raw products were selected for consistency, since the preparation of the products can have large consequences for nutrient composition as well as environmental impacts and is highly dependent on individual consumer food preparation habits57,58,59. The inclusion of the preparation stage would likely have added large uncertainty to the results. Moreover, energy use during preparation is typically higher for meat, which requires more thorough cooking, than for many alternatives that are often highly convenient and require minimal heating. Thus, excluding preparation yields a conservative comparison that avoids overstating the benefits of alternatives due to assumptions about cooking practices. For nutrients, the effects of cooking are highly variable, depending on both the method and the nutrient in question. Given this complexity and the lack of standardised preparation practices, systematic integration was not feasible.
Some of the selected datasets were incomplete. In these cases, they were complemented using literature values, average values from similar datasets, or informed assumptions (see SI1-1.3.4 and 1.3.5). Due to a lack of data on insect-based, pea-based, and mycoprotein-based alternatives, new datasets were created from the literature and data provided by the U.S. Department of Agriculture (fdc.nal.usda.gov) (SI1-1.3.6).
Nutrient density was assessed using the NRF index, a widely applied method well suited for LCAs that integrate the nutritional dimension. Originally developed by Fulgoni et al.30, the NRF has since been adapted and applied in various food system studies23,60,61,62,63. In this study, we applied the NRF11.3 variant, which offered a balance between comprehensiveness and data availability for our dataset and objectives. The nutrient selection was tailored to reflect Swiss dietary concerns and the specific context of meat and dairy alternatives. Iodine was added due to its known risk of deficiency in Switzerland64,65, while vitamin B12 and calcium were included for their relevance in comparing animal-based and alternative products. Sugar and sodium were included as disqualifying nutrients, consistent with rising intake concerns and their public health significance5,6. Nutrient contributions were expressed relative to dietary reference values consistent with those used for the diet-level analysis, with sex-specific recommendations averaged to represent population-level requirements (SI2-T6).
Meat and dairy, and alternatives at the product level – Environmental data
To assess the environmental impacts of meat and dairy and their alternatives at the product level, life cycle inventories were taken from different databases, including Agribalyse v3.1, AgriFootprint v6.3, ecoinvent v3.9.1, SALCA (Swiss Agricultural LCA) v3.4 and v4.0, and World Food LCA Database (WFLDB) v3.5. The life cycle inventory selection process included a term search comparable to the one for nutrient data (SI1-1.4.1), as well as filtering based on metadata. The process was repeated independently by two researchers to ensure completeness. To complement the selection of products, additional inventories were created based on the literature and publicly available product data (SI1-1.4.2). Most of the final selections of inventories were based on European background data. The system boundaries were set from cradle to production gate, since packaging, retail, and consumption were assumed to be similar between alternative and reference products, hence not affecting the comparison between them. Environmental impacts were calculated in SimaPro v9.6 by applying the SALCA v2.01 method66. Group averages were obtained using unweighted mean values. The results for additional impact categories are presented in SI2-T3.
Diet level – Food consumption data
Two diets were chosen as references: the Swiss self-selected diet and the Swiss recommended diet for the adult population. The self-selected diet was based on 24-h food consumption recall questionnaires31. Given the large number of food items reported, a selection procedure was applied to reduce complexity while maintaining representativeness, resulting in 104 food items used for modelling. The remaining food items were accounted for via proxies (see SI1-1.2.1).
Although the consumption data were based on the most recent national nutrition survey available for Switzerland, we acknowledge its age as a limitation. To improve validity, meat consumption was updated based on reported sales trends67, adjusting product proportions while keeping the total intake constant. Dairy consumption data were cross-checked with national purchase statistics68. These comparisons showed stable overall dairy intake, with a gradual decline in milk intake but a recent increase in cheese consumption. However, due to inconsistencies between purchase and consumption data, no adjustments to product proportions were made, as such changes would have been uncertain.
The recommended diet was based on Swiss dietary recommendations, which are visualised in the ‘Swiss food pyramid’32. Explicit recommendations were translated into consumption amounts. If the recommendations were only food group-specific, the composition of the food group was assumed to be identical to the self-selected diet. If this was not possible, the amounts were divided among the available food items, following the principle of a balanced diet (SI1-1.2.2).
Alternative diets were based on the total replacement of meat or both meat and dairy. This approach was chosen to illustrate the largest possible adoption of alternative products, as the consequence of any partial substitution can then be derived from linear equations. To limit the variety of diets presented, the substitution of meat is shown individually but not the substitution of dairy products. Moreover, the reduction of meat consumption is advisable from both a health and an environmental perspective, while the reduction of dairy is mainly advisable from an environmental perspective. The incorporation of meat and dairy alternatives into the diets assumed a 1:1 replacement by weight12,27.
To calculate the nutrient composition and environmental impacts of the alternative diets, the average product groups were used as a standard for replacement, that is, plant-based and mechanically texturised meat alternatives, milk alternatives, cheese alternatives, cream alternatives, and yoghurt alternatives.
Diet level – Nutrient data
The nutrient composition data for food products other than the alternatives were taken from the EuroFIR dataset for Switzerland. Datasets that best represented the relevant foods we previously identified were selected through targeted searches. All food items were considered raw to minimise uncertainty due to individual food preparation practices. The selection of nutrient thresholds depended on the availability of reference values. Population Reference Intakes (PRIs) and Average Requirements were used as reference values for nutrient adequacy where available, while Adequate Intakes (AIs) were applied when no PRI was defined. For nutrients recommended to be limited, Recommended Upper Intake Levels were used where available. Otherwise, the AI served as a pragmatic reference (SI2-T6). When reference values differed by sex, the average of female and male recommendations was used to represent population-level requirements.
Diet level – Environmental data
Environmental data were extracted from several sources to ensure suitability. Swiss raw material production was modelled based on internal data69 and ecoinvent inventories. If no Swiss inventory was available, a suitable proxy was identified. The processing steps for food items in Switzerland were modelled by adapting suitable inventories from external databases—that is, Agribalyse, ecoinvent, and WFLDB—to Swiss production conditions (SI1-1.4.5). For each food item, the self-sufficiency of production (i.e., the ratio between Swiss production and total consumption) was considered70. Hence, import inventories were created. Here, the import mixes by country of origin were accounted for (SI1-1.4.6). All available environmental data within the Agribalyse, Agri-Footprint, ecoinvent, SALCA, and WFLDB databases were considered to create the import inventories. Within inventories for processes taking place outside of Switzerland, background data and elemental flows were regionalised according to the country of origin of the food item. The system boundaries were set at the cradle-to-production gate for Swiss production and at entry into Switzerland for imported food. Since the diets were based on consumption data, food losses and waste occurring between the production gate and the consumption were accounted for, as they increased the required production volume71. Additionally, weight adjustments for water absorption during cooking, such as for rice and pasta, were applied to align the reported consumption weights with the raw material weights used in environmental modelling. Economic allocation was the preferred measure for dividing environmental impacts into multifunctional processes.
For the analysis, the environmental impacts of the diets were compared to those of the reference diet. Currently, there are no established reference values for maximum dietary environmental impacts comparable to dietary reference intakes. Nonetheless, approaches to defining absolute environmental sustainability have been proposed72. The selection of the presented impact categories was based on their relevance in the food Life Cycle Assessment literature1. The results for additional impact categories are provided in SI2-T4.
Sensitivity analyses
The robustness of the findings was evaluated through sensitivity analyses. To assess the influence of product choice, the nutrient content and environmental impacts of the alternative diets were calculated individually, assuming total replacement with only one product group at a time. The observed sensitivity of the diet-level results is depicted graphically as ranges in Fig. 3. The origin of the raw materials as an uncertainty factor was additionally investigated for an exemplary product. The product was chosen based on the availability of matching nutrient and environmental data, as well as a high content of the main ingredient. The input flow of the main ingredient was replaced with the same raw material from six different locations on four continents. Lastly, the effect of the products’ composition was investigated. For that purpose, individual products from the largest product group, the mechanically texturised soy-based alternatives, were analysed for their ingredient composition and put into direct comparison with the group’s average performance. The results of the sensitivity analyses are provided in SI2-T8 and T9.