Biodiversity is rapidly declining globally by about 5% every decade¹. Food systems are largely responsible for this loss. Directly, biodiversity is impacted by converting natural areas into agricultural land. Indirectly, food systems impact biodiversity loss via their increased contribution to e.g. climate change, pollution and eutrophication^2,3. Besides its environmental impact, current food systems are also responsible for the increased occurrence of non-communicable diseases such as coronary heart disease, diabetes, and cancer due to Western diets⁴. All these impacts are often justified in public debates under the argument that this intensive production configuration of the food system, and its respective diet, is the only way to feed global populations. Therefore, the major challenge for the coming decade is redesigning the food system to respect both human and planetary health^3,5.

One of the main underlying reasons for the large environmental impact of food systems in high-income regions, such as Europe, is the current linear extract-produce-consume-discard model⁶. Circularity is, therefore, increasingly seen as a key solution. A circular food system implicitly requires searching for practices and technologies capable of minimizing the need for finite resources (e.g. land, phosphate rock, fossil fuels), encourage the use of renewable ones (e.g. wind and/or solar energy), prevent the loss from the food system of natural resources (e.g. nitrogen (N) and/or phosphorus (P)), and stimulate reuse/recycling of unavoidable losses inherent to the system(e.g. human excreta) in a way that adds the highest value to the food system^7,8,9. Recent studies have explored different circularity scenarios and indeed showed that transitioning towards circular food systems reduces the use of finite resources such as mineral fertilization and reduces agricultural land use and GHG emissions while still producing enough healthy food for the European population^10,11.

But what about biodiversity? Can we enhance biodiversity in circular food systems without considerable trade-offs with other environmental goals (e.g. GHG emissions)? This question so far remains unanswered. Therefore, this paper aims to assess the potential of enhancing biodiversity in circular European food systems. Our results show that redesigning the European food system based on circularity principles can help enhance biodiversity compared to the current food system.

We developed two scenarios: one focusing on land sparing and the other on land sharing (Table 1). The land sharing and land sparing strategies follow different narratives. Land sparing focuses on sparing areas with the underlying assumption that this spared land would be converted into natural areas. Land sharing, on the other hand, focuses on adopting more biodiversity-friendly practices, e.g., the absence of pesticides or tilling within the agroecosystem (i.e., the land used within the food system) with the aim to enhance biodiversity while simultaneously producing food. To compare the land sharing and land sparing scenarios, we assessed the (agricultural) land use and the associated biodiversity score, linked to agricultural practices. The land use represents all agricultural land (considering the land uses and the farming practices applied) needed to produce enough healthy food to feed the EU27 + UK population assuming a self-sufficient European food system. The biodiversity score was determined by accounting for biodiversity within agricultural areas as follows: each agricultural practice known to enhance biodiversity for a certain taxonomic group obtained one point per ha of used land. The agricultural practices considered were no tillage, organic fertilizer, no pesticide use, cover crops, and natural buffer areas¹². According to Cozim-Melges et al.¹², these practices were the most consistent in their positive effects on biodiversity. The biodiversity score is depicted as a percentage of the value obtained when compared to the maximum score possible with all practices (so maximum is 100%). The taxonomic groups considered were arthropods, birds, mammals, bacteria, fungi, earthworms and nematodes. Not all practices enhanced each of the taxonomic groups, and not all practices can be applied in all regions (see methodology in supplementary information for more information). In other words, the practices are specific per taxonomic group and limited by the region where they are applied, e.g., cover crops are limited by the seasonality of chosen crops in a certain agroecological zone in Europe defined in the model.

A food systems optimization model called the Circular Food System (CiFoS) model (Van Zanten et al., 2023) was used to compare the Land sparing and Land sharing scenarios with a baseline scenario (Table 1). CiFoS is an interactive bio-physical data-driven optimization model that facilitates the selection of food system redesigns by minimizing environmental impacts while producing healthy diets. A healthy diet is assessed as the recommended range of intake levels per food group derived from the EAT-Lancet guidelines plus fulfilling the European Food Safety Authority (EFSA)¹³nutrient requirements. In the land sparing scenario, the objective function of the model was to minimize agricultural land. The objective function for the land sharing scenario was to maximize the biodiversity score. These two scenarios were compared to the baseline production scenario which matches empirical data related to, for example, current crop production systems for domestic use and export with the current areas they are located while minimizing the difference with the current food supply (objective function).

The baseline reaches 2% of the total biodiversity score (considered 100% if all practices were applied in the whole system). Compared to the baseline scenario, the land sparing scenarios reduced agricultural land use by 81% and the biodiversity score became 0%. (Fig. 1). The land sharing scenario, on the other hand, resulted in an increased biodiversity score, reaching 86% of the maximum score, and the agricultural area did not change (Fig. 2). Both scenarios, land sparing and sharing, produced healthy diets and reduced GHGe. The land sparing scenario reduced GHG emissions by 65% and the land sharing scenario reduced emissions by 44%. What is important to note is that both scenarios resulted in a complete redesign of the European food system. In the next sections we will explain the land sharing and land sparing scenarios and their food system reconfigurations (production and consumption) in more detail.

Circular food system redesign in the land sparing scenario

In the baseline scenario, 171.7 Mha of agricultural land is used (Fig. 2, panel A). Of the agricultural land 104.6 Mha is cropland and 67.1 Mha is grassland (permanent grasslands and rangelands). Croplands are reduced to 43.9 Mha (reduced by 58%) and grasslands to 2.3 Mha (reduced by 97%). If we zoom into the different European regions, a clear trend can be seen in which grassland reduction is favoured over reduction of cropland. Nevertheless, in Eastern Europe we see the opposite trend 96% reduction of cropland and 81% reduction in grassland (Fig. 2, panel B). This shows the importance of accounting for regional differences. To reduce land use the most intensive agricultural practices were used– increasing the yield per ha and therefore decreasing land use. Alternative agricultural practices applied in the baseline, e.g. no pesticide use, were therefore no longer applied. This resulted in a decrease of the biodiversity score in each region, and a shift in practices used (Fig. 3, panel A) when compared to the overall composition (Fig. 3, panel B).

The reduction in land use in the land sparing scenarios greatly impacted the crop and livestock production systems (Supplementary Figure A 2.1). Regarding crop production systems, the area for most crops was reduced, except for vegetables. Production of vegetables increased, and this can be explained by the underconsumption of vegetables in current European diets. Transitioning towards a healthy diet, therefore, requires an increase in the production of vegetables. If we zoom in on the different regions, we see small changes in crop production systems: southern Europe increase their production of soybean while especially western Europe increased the production of vegetables (supplementary figure A 2.1.2.3). With regard to the animal production systems, a clear reduction in each of the systems is seen. Especially the production of broilers (reduced by 98%), layers (reduced by 95%), and beef (reduced by 87%) largely reduced, while dairy (0%), fish (reduced by 36%) and pig (reduced by 29%) production systems faced no to less reductions (supplementary figure A 4.1.1).

Besides changes in production, the land sparing scenario also required dietary changes. Overall protein intake was reduced by 25% from 80g of protein per person per day in the current diet to 60g of protein per day in the spared diet. The share of animal versus plant protein changed from 60% animal-source food to 40% plant source food in the current diet to 20% animal source food and 80% plant source food. Especially the consumption of red meat largely decreased from 6% in the current diet to 2% in the land spared diet (Fig. 4). Where in the current diet we see an overconsumption of energy and cholesterol, all nutritional requirements are met in the land spared diet (supplementary figure A 1.2-3).

Circular food system redesign in the e land sharing scenario

In the land sharing scenario, the area of agricultural land is maintained (Fig. 1). In other words, the land sharing scenario offers the possibility to farm less intensive and increase biodiversity within the current agricultural land. To increase the biodiversity score, alternative agricultural practices were applied, which required more land use than land sparing option, but still equal to baseline (Fig. 2, panel A). Practices such as ‘no pesticide use’ result in lower yields and therefore more agricultural land is needed to produce a healthy diet for the European population compared to the land sparing scenario. The biodiversity score per regions were very similar: 83% for Southern Europe, 90% for Western Europe, 83% for Northern Europe, and 86% for Eastern Europe. Focusing on the practices, not all practices are applied simultaneously everywhere. Natural buffer areas, no tillage and no pesticide use are the most applied alternative practices. Although organic fertilizer application was also widely used, there were still many occasions where artificial fertilizers were applied as well to meet crop nutrient demands. As a result, the maximum biodiversity score for the alternative practice of fertilizer application was not achieved. Similarly, for cover crops, the achieved biodiversity score was lower than the potential maximum because they could only be applied to agroecological zones in temperate regions and under the cultivation of winter crops. The seasonality circumstances limited the use of leguminous cover crops to temperate Northern regions, making cover crops the main constraint, followed by organic fertilizer (Fig. 3, panel C).

The land sharing scenario, just like the land sparing scenario, largely required a redesign of the crop- and livestock production systems. The area for vegetables (8,6 mi ha, + 265%), fruits (7,8 mi ha, + 31%) and forage crops (87.8 mi ha, + 3%) increased compared to the baseline scenario, while the production of cereals (46.9 mi ha, -14%) and oil crops (15.2 mi ha, -7%) decreased (supplementary figure A 2.1.1.2). Furthermore, amount of tubers largely decreased to zero due to the influence of the practice no tillage. However, large differences were observed among regions, e.g., wheat production increased in northern and western Europe while Eastern and Southern Europe increased their production of fruits and vegetables (supplementary figure A 2.2.2.1.). With regards to the animal production systems, the reduction in animal numbers was lower compared to the land sparing scenario. Pigs reduced by 94%, followed by layers (reduced by 71%), broilers (reduced by 64%), beef (reduced by 56%), fish (reduced by 16%) and dairy (reduced by 15%) (supplementary figure A 4.1.1).

In the land sharing scenario the intake of protein increased by 4% (83 g per person per day). Despite this increase in protein, the results do show a trend towards more plant-based proteins, with only 25% of the proteins coming from animals and 75% from plants (Fig. 4). All nutritional requirements were fulfilled (supplementary figure A 1.2-3).

We have modelled two different pathways for enhancing biodiversity within circular food systems, land sparing and land sharing, and assessed their impact on land use, GHGe emissions and dietary needs. We found that both land sharing and sparing can be applied to circular food systems with distinct but potentially positive effects on biodiversity. Both scenarios also have the potential to synergistically reduce environmental impacts, such as reducing GHG emissions. However, they require different redesigns of the food system, both in terms of production and consumption patterns. Below, we discuss some of the implications of our results for biodiversity and the food systems as well as some of the limitations of our methodology and future research.

Both land sharing and sparing scenarios resulted in a reduction of animal source food and an increase in plant source foods, compared to the current food system (baseline scenario), with both scenarios showing an overall reduction in animal numbers. Nevertheless, the protein intake between these scenarios in terms of quantity was different. The overall protein intake was largely reduced (to almost the minimum protein requirement) in the land sparing scenario, while within the land sharing scenario protein intake remained similar to today’s situation. The land sparing scenario’s diet is more in line with results found in previous literature of the use of circular food systems in Europe^10,11, depicting that land sparing will bring more intense dietary changes. The findings of higher protein (and food overall) intake in the land sharing scenario is contrasting to earlier work, which showed that reduced efficiency and lower yields in land sharing systems might increase costs and reduce food accessibility^14,15. Our findings might point out that, while land sharing strategies reduce yield, they also help diversify crops, which might help nutrition in circular food systems, as suggested by Jenkins et al. ¹⁶.

The land use changes in both scenarios, but particularly in the land sparing scenario, were variable across the EU27 + UK. In general, our results indicated that agricultural production systems mainly take place in northern and western Europe and near removal of agriculture in southern and eastern Europe, which would affect biodiversity differently across regions as well^18,19,20. Additionally, that could shift economic activities and socio-economic dynamics^15,21,22, which are important factors for the implementation of such a pathway. While not the scope of this study, it should be an important aspect to consider in future research.

The biodiversity score was 86% out of the 100% in the land sharing scenario and 0% for land sparing. Our results showed that current agricultural land is enough to apply most alternative practices while producing enough healthy food within a self-sufficient European circular food system. The missing 14% of the total 100% can mainly be explained by the low inclusion of the cover crops and the need for artificial fertilizer. In practice, our findings could even underestimate the potential that more diverse cover crops could have in terms of biodiversity and nutrition if implemented in food systems^23,24, with only legumes being available as cover crops in the model. More diverse cropping systems, e.g. living mulches, agroforestry, offer the possibility to cover the soil whole year round to enhance biodiversity²⁵ and also adds additional nutritional potential²⁶.

Within the land sharing scenario, it was fertilization that mostly limited the biodiversity score. Applying organic fertilizers are quite well known to have benefits for soils and agroecosystems^27,28, but our results showed that they are limited in availability in line with Simon et al.¹¹, making the model require external input of synthetic fertilizers and reducing the score. In other words, although circularity enhance the recycling of biomass, still not enough nutrients are available to fill the nutrient requirements of all crops needed to produce enough healthy food for Europe¹¹. Additionally, there are arguments that the implementation of more extensive practices, used in land sharing, would compromise productivity or increase land use²⁹, which is contradicted by the findings in this paper. Nevertheless, current literature suggests that combining sparing and sharing is needed to enhance biodiversity holistically^14,17,30.

Our findings indicate that we could almost reach full potential of an agricultural practices enhancing strategy focussing on species level, from a qualitative instead of a quantitative perspective accounting for, e.g. species abundance or richness. Note, this means biodiversity enhancement of agroecosystems encompassing both above- and below-ground taxonomic groups - a usually ignored group in models (e.g. GLOBIO³¹, GLOBIOM³², ReCiPe 2016³³). Our study therefore does not indicate the number of species ‘affected’ and if applying alternative practices would safeguard biodiversity in agricultural areas. Instead, our score should be used to indicate how to enhance biodiversity across multiple different taxa with the underlying aim to contribute to reversing biodiversity loss.

To interpret our results, it is important to consider some of the methodological choices made. It is important to note that the biodiversity score developed in this study does not distinguish between types of land use, i.e. we assumed that practices had the same effect in both croplands and grasslands. Grasslands are known to be usually more biodiverse than croplands^19,34, and hence the reduction of grassland in the models could also have effects on biodiversity not considered in this model³⁵. Additionally, the biodiversity score could be enhanced by also integrating geographic information accounting for areas with high biodiversity or high proportion of endemism (biodiversity hotspots). The main limitation to such inclusion comes from the absence of data^37,38,39,40. Future research could benefit and require collaboration and alignment in data collection initiatives on species presence and distribution as well as in identifying species and indicators of particular interest^40,41 and their response to agricultural practices. That would help quantify and prioritize areas to be spared and areas to be shared in terms of biodiversity and production combined.