The role of protein quality in designing sustainable food systems

Abstract

The influence of protein digestibility and the adequacy of indispensable amino acids (IAAs) on designing sustainable food systems is not yet fully understood. In this study, we used FOODSOM (FOOD Systems Optimization Model) to evaluate the effects of including total or digestible protein and IAA requirements on land use and dietary outcomes of the Dutch food system. Our results show that land use is minimal with 18 g of animal-source protein per person per day. We also demonstrate that diets that meet all nutrient requirements, including protein, automatically satisfy IAA requirements. However, in digestible form, IAA requirements are met only when digestible protein requirements are considered alongside other nutrient requirements. Depending on the amount of animal-source protein in the diet, nutrients other than digestible protein and IAAs can be limiting and therefore influence land use. To ensure nutrient adequacy of advised diets, therefore, future food system modelling studies should include protein digestibility in addition to all macro- and micronutrient requirements. Only then, such modelling studies can explore the land use consequences of diets varying in the amount of plant and animal-source food, while assuring nutrient adequacy.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-48122-0.

Introduction

It is widely recognized that the global food system has a significant impact on the environment¹. A transition to more circular food systems is increasingly seen as a promising pathway to produce and consume food within planetary boundaries^2,3. Circular food systems safeguard soil and water health, avoid the use of non-essential resources and waste of essential ones, prioritize biomass for humans needs, and recycle unavoidable losses to feed animals or fertilize soils⁴.

Food systems modelling studies in high-income countries demonstrate that increasing circularity can substantially reduce environmental impacts. However, this transition also requires a shift toward higher consumption of plant-source foods (PSF), such as vegetables, nuts and legumes, and lower consumption of animal-source foods (ASF)^5,6. This shift is primarily driven by land-use efficiency; arable land is used most efficiently for direct food crop production, whereas its use for feed production entails conversion losses when animals transform feed into ASF³.

Thus far, studies modelling alternative food systems have paid limited attention to qualitative aspects of food, such as protein quality^6–8. However, designing sustainable food systems requires consideration of not only the quantity but also the quality of nutrients in the diet⁹. This is particularly important when shifting toward more plant-based diets, as ASF are currently the only dietary sources of certain essential nutrients, like vitamin B_12, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA)¹⁰. Compared with PSF, ASF also provide several nutrients in higher bioavailable amounts, including protein (especially lysine) as well as iron and zinc^11,12. Furthermore, altering the share of ASF in the diet affects the intake of digestible macro- and micronutrients^13,14. Such changes may result in inadequate nutrient intake and should therefore be carefully considered when formulating diets with varying shares of ASF to ensure nutrient security.

Accordingly, these nutritional dimensions should be explicitly integrated into modelling studies that aim to identify healthy and sustainable diets. To date, most food systems modelling studies have generally defined protein adequacy solely in terms of daily dietary intake^5,7,8. In doing so, they implicitly assume that dietary protein is fully digestible, likely underestimating the protein amount needed to meet requirements. Moreover, although indispensable amino acids (IAAs) must be obtained from the diet, digestible IAA requirements have not been explicitly included in prior modelling studies. To ensure true protein adequacy, food systems modelling studies should therefore account for both protein digestibility and IAA requirements.

Beyond nutrient adequacy, food systems models allow flexibility in diets¹⁵, meaning that formulated diets may not reflect current or recommended consumption patterns. Some studies have considered lower and upper limits on food groups to approximate habitual or recommended diets^5,7,16. Diets that meet nutrient requirements without such constraints may differ in environmental outcomes but the effects of including food group constraints on both dietary and environmental outcomes remain largely underexplored.

Therefore, this study aims to explore the role of protein quality in designing sustainable food systems that meet human macro- and micronutrient requirements, while minimizing environmental impact, both with and without additional food group constraints. We choose land use as the measure of environmental impact because it is central to key planetary boundaries affected by food systems, including biodiversity integrity, land use change, freshwater change, climate change (via deforestation) and biogeochemical flows³. To achieve this, we compared land use across 12 scenarios differing in protein requirements (total protein, IAAs, digestible protein, digestible IAAs) and three dietary contexts (see Table 1). In the first context (Base), only calorie and protein requirements were considered; in the second (Nutr), all nutrient requirements were included, and in the third (FG), additional food group constraints were applied. In all scenarios, we evaluated the impact of different amounts of plant- and animal-source protein on land use (for more details, see “Methods” section).

Table 1.

Overview of scenarios studied.

Dietary requirements
Protein requirement	Base—meets only calorie and protein requirements	Nutr—adds nutrient requirements	FG—adds food group constrains
Total protein (TP)	Base-TP	Nutr-TP	FG-TP
Total indispensable amino acids (IAAs)	Base-IAAs	Nutr-IAAs	FG-IAAs
Digestible protein (DP)	Base-DP	Nutr-DP	FG-DP
Digestible IAAs (Dig-IAAs)	Base-Dig-IAAs	Nutr-Dig-IAAs	FG-Dig-IAAs

Results

Base scenarios: calorie and protein requirements only

Figure 1 shows the change in land use in response to varying amounts of animal-source protein (ASP) in the diet across the four Base scenarios. Minimum land use occurred when 12 g of ASP per day (20% of protein requirement) was included in all Base scenarios. Land use was lowest (i.e., 226 m² per person per year) when protein and IAAs digestibility were not considered (Base-TP and Base-IAAs). Including digestibility of protein (Base-DP) increased minimal land use by 2.7%, from 226 to 232 m² per person per year (see Fig. 1). Including digestible IAA requirements further increased minimal land use by 14%, to 265 m² per person per year. This difference likely reflects the fact that digestibility of individual IAAs can differ from that of total protein and is often notably lower. For example, total protein digestibility of oats is 92% whereas lysine digestibility in oats is only 83%¹⁷. Accounting for IAAs digestibility, therefore, required higher inclusion of protein sources, leading to increased land use.

Fig. 1.

Land use for four Base scenarios (i.e. only calorie and protein requirements), where TP is a diet with a requirement for total protein content only, IAAs is a diet with requirements for total protein and indispensable amino acids content, DP is a diet with a requirement for digestible protein content, and Dig-IAAs is a diet with requirements for digestible protein and indispensable amino acids content.

Deviating from 12 g ASP per person per day led to increased land use in all the Base scenarios. Land use increased proportionally as ASP in the diet decreased from 12 to 0 g (Fig. 1). This increase has two causes. First, in diets with 0–12 g ASP, (some) losses and wastes from the production and consumption of PSF cannot be valorized in ASF, requiring additional PSF to fulfil protein and calorie requirements. Second, protein and IAAs digestibility are lower in PSF than in ASF, necessitating greater PSF inclusion in the diet to meet protein and IAAs requirement, further increasing land use.

For diets with 15–30 g ASP, land use increased when accounting for protein digestibility, but not when including IAAs requirements. This is because the proportion of IAAs per gram protein in most food items, particularly in ASP, exceeds dietary requirements¹⁸. Finally, when diets include more than 30 g of ASP per person per day, land use increased with increasing ASP, and this trend was similar across all four Base scenarios.

Figure 2 shows the adequacy of the individual nutrients for each of the Base scenarios. Nutrients, including the aromatic amino acids (AAA), histidine, threonine and tryptophan, that were consistently adequate across all Base scenarios are not included in Fig. 2 but are shown in supplementary Fig. S1. Results show that considering only total protein requirements, without accounting for digestible protein (Base-TP), led to insufficient digestible protein in diets containing up to 48 g of ASP per day. Additionally, the amounts of lysine and leucine were inadequate when ASP was below 9 g per person per day (see supplementary Fig. S1). As expected, including IAA requirements (Base-IAAs) ensured all IAA requirements were met. However, digestible protein requirements were only fulfilled with more than 40 g of ASP per person per day. Including protein digestibility (Base-DP) ensured adequacy of all IAA requirements, except for digestible lysine, leucine, isoleucine, sulphur-containing amino acids (SAA) and valine (see Fig. 2). Digestible lysine, in particular, was adequate only in diets with more than 21 g of ASP per person per day in all Base scenarios, except Base-Dig-IAAs. Lysine is typically the most limiting IAAs in PSF, especially in cereals¹¹. Hence, plant-based diets relying heavily on cereals can become lysine-deficient. Consequently, including both digestible protein and digestible IAA requirements is necessary to meet all protein and amino acid requirements. While protein and IAA requirements are met, other nutrient requirements, except for carbohydrate, phosphorous and zinc, were not met in at least one scenario. Notably, alpha-linolenic acid and linoleic acid were the only nutrients consistently unmet across all scenarios, regardless of ASP intake (Fig. 2).

Fig. 2.

Nutritional compliance of diet for different Base scenario (i.e. only calorie and protein requirements), where TP is a diet with a requirement for total protein content only, IAAs is a diet with requirements for total protein and indispensable amino acids content, DP is a diet with a requirement for digestible protein content, and Dig-IAAs is a diet with requirements for digestible protein and indispensable amino acids content; EPA-DHA: Eicosapentaenoic acid and docosahexaenoic acid; SAA: Sulphur containing amino acid.

Nutr scenarios: all nutrient requirements

In the Nutr scenario, which included all nutrients requirements, the model could only find a solution when the diet contained at least 9 g of ASP per person per day. Moreover, including IAA requirements did not increase land use compared to the protein-only scenario (see Fig. 3). Therefore, the following paragraphs focus on total and digestible protein requirement scenarios.

Fig. 3.

Land use demand for different Nutr scenarios (i.e. all nutrient requirements need to be met), where TP is a diet with requirement for total protein content only, IAAs is a diet with requirements for total protein and indispensable amino acids content, DP is a diet with requirements for protein in digestible form, and Dig-IAAs is a diet with requirements for protein and indispensable amino acids in digestible form.

In the Nutr-TP scenario, minimal land use was 296 m² per person per year, achieved with 18 g of ASP in the diet. Thus, accounting for all nutrient requirements, rather than only calories and protein (Base), increased land use by 20% and ASP per person per day by 6 g. Considering protein digestibility (Nutr-DP) further increased minimal land use from 296 to 312 m² per person per year. This higher land use due to accounting for protein digestibility was observed for diets containing up to 33 g of ASP per person per day.

Deviating from 18 g of ASP per person per day again led to higher land use. When dietary ASP was below 15 g per person per day, land use increased sharply, by up to 30% (Fig. 3). A similar pattern was observed when protein digestibility was considered. Conversely, when dietary APS exceeded 18 g per person per day, land use increased with increasing ASP intake. Figure 3 also shows that the effect of protein digestibility on land use was apparent only up to 33 g of ASP per person per day. The difference in land use between the Nutr-TP and Nutr-DP scenarios decreased as dietary ASSP increased, reflecting the higher amounts of digestible protein and IAAs derived from ASF. No effect of protein digestibility on land use was observed when dietary ASP exceeded 33 g per person per day (Fig. 3).

Figure 4 shows the adequacy of each modelled nutrient across the different Nutr scenarios. Results show that protein, IAAs, vitamin B₂, fat, calcium, vitamin A and vitamin B₁₂ were limiting in at least one Nutr scenario, demonstrating that land use is driven by multiple nutrient requirements. Interestingly, calcium was limiting in the Nutr-TP scenario, but not in the Nutr-DP, likely due to the higher dairy content in diets when protein digestibility was considered. In Nutr-TP scenario, lysine, SAA, and valine were not met in digestible form. Accounting for protein digestibility, however, ensured that all IAAs requirements were satisfied in digestible form. Furthermore, limiting nutrients varied across ASP levels. For example, digestible lysine was limiting up to 21 g of ASP, whereas digestible leucine was limiting up to 12 g ASP (Fig. 4). Similarly, copper requirements were adequately met when ASP intake was below 30 g per person per day but became limiting above this level. Adequacy profiles for all modelled nutrients in Nutr scenario are provided in Supplementary Fig. S2.

Fig. 4.

Nutritional compliance of diet for different Nutr scenario (i.e. scenarios where all nutrient requirements need to be met); where TP is a diet with requirement for total protein content only, IAAs is a diet with requirements for total protein and indispensable amino acids content, DP is a diet with requirements for protein in digestible form, and Dig-IAAs is a diet with requirements for protein and indispensable amino acids in digestible form; EPA-DHA: Eicosapentaenoic acid and docosahexaenoic acid; SAA Sulphur containing amino acid.

As mentioned previously, diets in the Nutr scenarios met all nutrient requirements only when at least 9 g of ASP was included in the diet. To examine land use of diets containing less than 9 g or no ASP, adjusted scenarios excluding vitamin B₁₂ and EPA-DHA requirements were developed (hereafter referred to as Nutr-Adj). Figure 5 shows that minimal land use in the Nutr-Adj scenario was 291 m² per person per year, achieved with 15 g of ASP per person per day. When protein and IAAs digestibility were considered and ASP intake per person per day was 0 g, land use increased to 370 m² per person per year, being 20% higher than the minimum observed at 15 g ASP. Overall, minimum land use in the Nutr scenarios was only 2–3% higher than in Nutr-Adj scenarios but required an additional 3 g ASP per person per day to meet all nutrient requirements. When ASP intake was lower than 15 g per person per day, the increase in land use with decreasing ASP was more pronounced in the Nutr scenarios (solid lines) than in the Nutr-Adj scenarios (dashed lines; Fig. 5). However, at ASP levels above 18 g per person per day, land use was identical in Nutr and Nutr-Adj scenarios. This indicates that vitamin B₁₂ and EPA-DHA requirements did not affect land use when diets contain at least 18 g of ASP per person per day. The adequacy of other nutrient requirements in the Nutr-Adj scenarios is presented in Supplementary Fig. S3. ASF, particularly dairy, meat and fish, increase with higher protein requirements whereas PSF such as grains, legumes and vegetables decrease, reflecting the focus on meeting total and digestible protein and IAAs targets. Details on each food group and absolute increase can be found in supplementary Figs. S4, S5.

Fig. 5.

Land use demand for different Nutr scenarios (i.e. all nutrient requirements need to be met) represented by solid lines and Nutr-Adj scenarios (i.e. all nutrient requirements need to be met except for vitamin B12, eicosapentaenoic acid, and docosahexaenoic acid) represented by dashed lines where TP/TP-Adj is a diet with requirement for total protein content only, DP/DP-Adj is a diet with requirements for protein in digestible form, Dig-IAAs/Dig-IAAs-Adj is a diet with requirements for protein and indispensable amino acids in digestible form.

FG scenarios: all nutrient requirements and food group constraints

The previously explored scenarios (Base and Nutr) did not impose restrictions on the quantity or diversity of dietary sources included per day. To assess the impact of such restrictions, we introduced constraints at the food group level. We evaluated a scenario in which dietary outcomes adhere to specified food group constraints, hereafter referred to as the FG scenario.

As in the Nutr scenarios, including IAAs requirements did not increase land use compared with considering protein requirements alone. Therefore, we focus on the FG-TP and FG-DP scenarios. In both, land use was minimal with 18 g of ASP per day (Fig. 6). Accounting for protein digestibility (FG-DP scenario) increased minimal land use by 3%, from 395 m² to 406 m² per person per year. The effect of digestibility on land use was observed only at ASP levels between 15 and 24 g of ASP per day. Reducing ASP intake from 15 to 9 g per day resulted in a sharp increase in land use (66%). Under these conditions, nutrient requirements need to be met primarily through PSF. However, meeting both nutrient requirements and food group constraints required greater diversity and higher quantities of PSF, including less land-efficient options, such as an eightfold increase in legumes. As in the Nutr scenario without food group constraints, a minimum of 9 g ASP was needed to meet all nutrient requirements; however, land use was 55% higher under the FG constraint (Fig. 3). This indicates that imposing food group constraints restricts the model’s solution space and necessitates a certain level of ASP intake to satisfy all nutrient requirements.

Fig. 6.

Land use demand for various FG scenarios (i.e. all nutrient requirements are needed to be met following food group constraints) where TP is a diet with a requirement for total protein content only, IAAs is a diet with requirements for total protein and indispensable amino acids content, DP is a diet with a requirement for digestible protein content, and Dig-IAAs is a diet with requirements for digestible protein and indispensable amino acids content.

Figure 7 confirms that all modelled nutritional requirements were met in the FG scenarios. However, in the FG-TP scenario with 15–21 g of ASP per day, digestible protein requirements were not met (Fig. 7), indicating that accounting for total protein alone is insufficient within this ASP range. Accordingly, total protein content was higher in the FG-DP diet than in the FG-TP diet, which contributed to the higher land use observed in the FD-DP scenario (Fig. 6). Protein and IAAs levels exceeded requirement most markedly at 9 g of ASP per day (see Fig. 7). This suggests that at low ASP levels, protein was not the limiting nutrient; rather food items selected to meet other limiting nutrients (e.g. calcium, and vitamin B₁₂) also provided substantial protein and IAAs. Similar to the Nutr scenarios, calcium, vitamin B₁₂, iron, carbohydrate and vitamin B₉ appeared limiting nutrients in the FG scenarios. Detailed nutrient adequacy results are provided in supplementary Fig. S6. Change in food group quantities with increasing ASP are less pronounced in the FG scenarios. This is especially prominent for PSF as food groups such as vegetables, tubers, and fruits reach the maximum allowed amounts based on dietary guidelines. Detailed changes for each food group, including absolute increase, are provided in supplementary Figs. S7, S8.

Fig. 7.

Nutritional compliance of diet for different FG scenario (i.e. scenarios where all nutrient requirements need to be met while following the food group constraints) where TP is a diet with a requirement for total protein content only, IAAs is a diet with requirements for total protein and indispensable amino acids content, DP is a diet with a requirement for digestible protein content, and Dig-IAAs is a diet with requirements for digestible protein and indispensable amino acids content; EPA-DHA: Eicosapentaenoic acid and docosahexaenoic acid; SAA: sulphur containing amino acids.

Discussion

Our results show that a moderate intake of ASP (12–18 g per day) is consistently associated with the lowest land use across all scenarios (Fig. 1, 3, 4, 6). An optimal level of ASP from a land use minimization perspective has been reported previously in food systems modelling studies^7,8. Van Kernebeek et al.⁸, who considered only protein and calorie requirements, similar to our Base scenario, concluded that including 12% of daily protein requirement as ASP (equivalent to 7 g) resulted in minimal land use. In contrast, our results show that land use in the Base scenario was minimized with 12 g of ASP per day (Fig. 1). However, Van Kernebeek et al.⁸ included only six crops with fixed rotations and two livestock production systems, whereas our study included 49 crops and five livestock production systems. The difference in results is likely attributable to the greater flexibility of our model, which offers more options to meet protein and calorie requirements.

Including all nutrient requirements, as done in the Nutr scenarios, increased overall land use compared with the Base scenarios. It also raised the amount of ASP in the diet at the point of minimal land use from 12 to 18 g per day (Fig. 3). Adding food group constraints further increased land use; however, the ASP level corresponding to minimal land use remained 18 g per day (Fig. 6). Simon et al.⁷ also identified 18 g ASP per day as optimal from a land use perspective, representing 40% of their total protein requirements. In our study, the same 18 g corresponds to only to 30% of the protein requirements, because the protein requirement used by Simon et al.⁷ was 23% lower than that applied here. Moreover, Simon et al.⁷ examined only 40:60 and 60:40 ratios of ASP to PSP in the diet, whereas we explored FG scenarios across the full range of 0–60 g ASP per day in 3 g increments and observed minimal land use at 18 g ASP per day.

Previous studies by Simon et al.⁷ and van Kernebeek et al.⁸ did not study the effect of including digestible protein requirements in the food systems model on land use. Our results show that including protein digestibility increased land use but did not change the ASP level at which minimal land use was observed (Figs. 3, 6). The magnitude of the land-use increase depended on other dietary factors, such as the level of ASP in the diet, overall nutrient requirements and food group constraints. The effect of including protein digestibility was most pronounced in diets with less than 33 g of ASP per day, except in the FG scenarios, where it was mainly observed between 15 and 24 g of ASP per day. This pattern can be explained by the lower digestibility of protein in PSF compared with ASF¹¹. Meeting all nutrient requirements also requires additional food items with higher land use. Furthermore, imposing food group constraints limited the food items that could be included in the diet, thereby restricting opportunities to optimize land use efficiency.

Despite this, including food group constraints is important. For example, diets in the Nutr scenarios with no food group constraints were inconsistent with dietary guidelines or current consumption patterns in the Dutch population, containing either no fruits or up to 2000 g of vegetables per day. In contrast, the FG scenarios met all nutrient requirements while adhering to the prescribed food group constraints, though at the cost of higher land use compared with other scenarios (Fig. 7).

We observed that when other nutrient requirements are accounted for, IAA requirements do not affect land use, regardless of IAA digestibility. Our study indicates that in food systems models designed to meet all nutrient requirements, digestible IAA requirements are automatically satisfied when the digestible protein requirements are met. Therefore, prioritizing the fulfillment of all nutrient requirements is more relevant than focusing solely on total protein adequacy in food systems modelling.

To meet all nutrient requirements, a minimum of 9 g of ASP was required in the diet (Figs. 4, 7). Diets containing less than 9 g of ASP could not satisfy vitamin B₁₂ and EPA-DHA requirements. This aligns with previous studies showing that both vegan (no ASP in diet) and vegetarian (some ASP in diet) diets often fail to meet vitamin B₁₂ requirements^19,20. Calcium was also identified as a limiting nutrient, significantly influencing model outcomes (Figs. 4, 7). Similarly, Simon et al.⁷ highlighted calcium as a nutrient of interest, noting that it approached the threshold of adequacy in their study.

Land use in the Nutr-Adj scenarios (excluding vitamin B₁₂ and EPA-DHA requirements) was lower than in the corresponding Nutr scenario. However, not all Nutr-Adj scenarios met vitamin B₁₂ and EPA-DHA requirements under these constraints. In our analysis, we did not account for additional dietary supplements or fortification. Therefore, when interpreting results from Nutr-Adj scenarios, it is important to recognize that supplements would be necessary to meet the requirements, and their inclusion would likely increase land use. The impact on land use depends on the choice of supplement production. For instance, depending on the substrate (e.g. glucose or beet molasses) and production method, 0.09–214 mg of vitamin B₁₂ can be produced per liter of substrate²¹, although the associated land use has not been established.

Our model included requirements for 23 nutrients, although more nutrients are essential for human nutrition. For instance, vitamin D was not included because, given adequate precursors, the body can synthesize sufficient amounts²². Furthermore, minimum requirements of sodium and cholesterol were not included,only the upper limits recommended by EFSA²² were considered. We were unable to include iodine and selenium, as the model could not be solved in multiple scenarios. Current guidelines indicate that iodine is fortified in food products²³. Since our model does not account for any fortification, excluding iodine is unlikely to have affected our outcomes. Selenium, in contrast, is not routinely fortified, and meeting its requirement depends entirely on the dietary intake. Therefore, no conclusions can be drawn regarding selenium inadequacies. Accounting for these nutrients via supplementations would further increase land use but would not alter our conclusions regarding the role of protein quality in designing sustainable food systems. Finally, some nutrients, such as vitamin K, phosphorus, and vitamin C, were consistently supplied well above minimum requirements across all scenarios. These nutrients can be consumed in excess, as no upper limits have been established to date²².

For this study, protein and IAAs digestibility values of individual dietary sources were averaged over food groups when item-level information was not available. A post hoc analysis shows that total protein intake across scenarios was dominated by sources with item-level digestibility, which contributed 33% of total protein, whereas 67% came from food-group-averaged values; foods with assumed digestibility of 1 did not contribute to dietary protein content (Supplementary File 4). Moreover, the model includes only raw dietary sources and excludes processed products. Processed foods often differ in protein and IAAs digestibility compared with unprocessed foods¹¹, which can alter the digestible protein IAAs content of the diet and consequently land use estimates. Digestibility of other micronutrients was not included in this study; however, attention should be given to their digestibility in future research. Furthermore, factors such as processing, packaging, and waste associated with processed foods also influence model outcomes, but these were not considered in our study.

This analysis assumes a closed food system without imports or exports. Allowing for imports and exports could decrease absolute land use of scenarios and change diet composition; however, incorporating such factors would require a fundamental restructuring of the model, including expanded system boundaries or alternative environmental data sources, and is therefore beyond the scope of this study. Such an improved model, however, is unlikely to alter the optimal amounts of PSF and ASF, as the current model already captures sufficient variation in these dietary sources.

Similar to our findings, Lam et al.²⁴ demonstrated that inclusion of ASF was necessary to achieve cost minimization. While their study focuses on food consumption and economic optimization and ours on production systems and land-use efficiency, both analyses independently indicate that moderate inclusion of ASF contributes to minimizing system-level objectives such as cost and land use. These findings indicate that complete exclusion of ASF may compromise the ability to meet nutritional requirements while simultaneously minimizing land use and meal cost²⁴.

Methods

The FOODSOM model

The FOODSOM (FOOD Systems Optimization Model) model⁵ was used to evaluate the effect of considering total and digestible amount of protein and IAAs on the diet composition and land use in food systems in the Netherlands. The general structure of the model is explained in “General structure of FOODSOM” section, whereas the procedure of incorporating protein quality in the model is presented in “Human requirement and nutrient content in FOODSOM” section. In “Scenarios” section, the scenarios explored to answer the research objective for this study are described.

General structure of FOODSOM

FOODSOM is an agro-ecological food system optimization model applied to the Dutch food system. It is created in GAMS 45 (GAMS Development Corporation. General Algebraic Modeling System (GAMS) Release 45.1.0, Fairfax, VA, USA, 2023) and described in detail by van Selm et al.⁵. In brief, the model uses linear programming to select the combination of crop and animal production systems to meet human dietary requirements, while minimizing environmental impacts (i.e., land use or greenhouse gas (GHG) emissions). Crop production in FOODSOM includes 49 representative crops, which can be grown on available arable land, grassland, and orchards or in mushroom sheds and greenhouses. The area of land available and crop yields are based on national statistics or survey data and vary depending on soil type. Crop residues, compost, artificial fertilizer, and animal manure are used to fertilize the crops.

Animal production in FOODSOM includes five common livestock systems, i.e., dairy cattle, beef cattle, broilers, laying hens, and pigs, and considers their entire life cycle. This implies that in addition to the yields and nutrient requirements of the food-producing animal (e.g., a dairy cow) those of the associated young, dry and parent stock are taken into account as well. For the food-producing animal, three productivity levels are included (low, medium, high). The high productivity level reflects current production averages in the Netherlands, while the low productivity level is based on more extensive farming systems found in other parts of Europe²⁵. To meet their nutrient requirements, livestock can consume feed-food crops (i.e., crops also suitable for human consumption), by-products (e.g., wheat bran), food losses and waste, grassland resources and synthetic amino acids. Additionally, marine fisheries based on current landings contribute to food supply.

Crop, livestock and fisheries products are processed into food suitable for human consumption and by-products. A portion of this food is lost or wasted as it moves along the supply chain, the exact fractions being dependent on the type of food. Food losses (i.e., post-harvest and processing and packaging losses) are assumed to be collected as individual products. Food waste from distribution and retail is assumed to be collected by food group, whereas household waste is assumed to be mixed together into a single product. Food loss and waste, and by-products can be fed to animals or composted in accordance with current legislation.

Total human nutrient requirements for gender and age category specific requirements were calculated from Dutch demographics²⁶. Although FOODSOM can allow for import and export of food and feed, for illustrative reasons we here considered a closed food system, with no import or export.

Human requirement and nutrient content in FOODSOM

The nutrient composition of the dietary sources included in the study was obtained from the Dutch food composition database (Nederlands voedingsstoffenbestand; NEVO)²⁷. As the NEVO database only includes total protein content and not amino acid composition, data were complemented with IAA composition for each dietary source. IAA composition was calculated by multiplying the IAA composition per gram protein from United States Department of Agriculture (USDA) database²⁸ with the protein content provided by NEVO²⁷. The concentrations of digestible protein and digestible IAAs were calculated by multiplying IAAs concentrations with their corresponding digestibility factors^11,29. Details on digestibility factors of protein and IAAs per dietary source are provided in Supplementary Table S1. For dietary sources with no data on the digestibility of protein and IAAs, we used the average digestibility of products with available data from the same food group.

Human nutrient requirements were categorized based on gender and age group (Supplementary File 1) and were based on recommendations by the European Food Safety Authority²². Where available, both minimum and maximum requirements for all nutrients were used. This data was complemented with IAAs requirements, based on FAO recommendations³⁰ and adjusted to fit gender and age group categorization (Supplementary File 1). Only upper limits were included for sodium and cholesterol²². Requirements for vitamin D, iodine, and selenium were not included. Therefore, throughout this study, the term “modelled nutrient requirements” and nutritional adequacy refers exclusively to diets that satisfy the nutrient requirements explicitly included in the model and does not imply biological completeness with respect to all essential nutrients.

Scenarios

To evaluate the effect of considering digestibility of protein and/or IAAs on agricultural land use of the food system, we compared 12 scenarios varying in protein quality components and dietary contexts (see Table 1). Four protein requirements were defined: meeting only total protein (TP), meeting total protein and total IAAs (IAAs), meeting digestible protein (DP) and meeting digestible protein and digestible IAAs (Dig-IAAs). For each of these four scenarios, three dietary contexts were considered. The first scenario followed food systems modelling studies that only considered minimum protein and energy requirements and maximum intake of 90 g of added sugar (hereon, referred to as Base)⁸. In line with more recent food systems modelling studies³¹, the second context added requirements for 23 other nutrients (Supplementary Table S2) to consider human nutrient requirements (hereafter, referred to as Nutr). The third context added the additional constraint to follow the food-based dietary guidelines at food groups level—as followed by van Selm et al.⁵ (Supplementary Table S3; this context is hereafter referred to as FG. A detailed overview of the scenarios is presented in Supplementary Table S4.

In all scenarios studied, we varied the amount of animal-source protein (ASP) and plant-source protein (PSP) in the diet. Given that the average protein requirement of the Dutch population is 59 g of protein per person per day, we explored the range from 0 to 60 g of ASP, at intervals of 3 g. To explore the full range of ASP in the Nutr-scenario, including diets with little or no ASP, we also defined an additional scenario (referred to as Nutr-Adj) where nutrients exclusively available from ASF (i.e., vitamin B₁₂ and EPA-DHA) were excluded from the nutrient requirements. The outcomes of this scenario (Nutr-Adj) were compared to the outcomes of the regular Nutr scenarios. The Nutr-Adj scenario represents a non-physiological configuration and was designed for model feasibility; therefore, this constraint should be consider when interpreting the results.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: Shiksha Adhikari, Ollie van Hal, Ben van Selm, Thom Huppertz, Imke de Boer; Methodology: Shiksha Adhikari, Ben van Selm (technical); Supervison: Ollie van Hal, Thom Huppertz, Imke de Boer; Formal analysis and investigation: Shiksha Adhikari; Writing—original draft preparation: Shiksha Adhikari; Writing—review and editing: Shiksha Adhikari, Ollie van Hal, Ben van Selm, Thom Huppertz, Imke de Boer; All authors have read and approved the final manuscript.

Data availability

Data, model codes and model descriptions are accessible at https://git.wur.nl/aps/foodsom-public/-/tree/2026_protein_quality?ref_type=heads.

Declarations

Competing interests

The authors declare no competing interests.