D7.3 - Ex-ante impact assessment of supply chain governance initiatives on biodiversity considering leakage

Name of the Deliverable Ex-ante impact assessment of supply chain governance initiatives on biodiversity considering leakage Deliverable 7.3 2 Summary Work Package 7 Deliverable No 7.3 Dissemination Level Public Type Report Lead Partner IIASA Due Date September 30th, 2025 Submission Date October 8th, 2025 Status Version 1.0 Authors David Leclère, Clara Frezal, Pekka Lauri, Bettina Meinhart, Amanda Palazzo, Sibylle Rouet-Pollakis, Niklas Hinkel, Leopold Ringwal, Fulvio di Fulvio & Andrey Lessa Derci Augustynczik (IIASA), Rafaella Ferraz Ziegert (University of Bonn) About CLEVER Project Number 101060765 Project Title CLEVER: Creating leverage to enhance biodiversity outcomes of global biomass trade Topic HORIZON-CL6-2021-BIODIV-01-15 Start date 1st September 2022 End date 31st August 2025 Coordinator RHEINISCHE FRIEDRICH-WILHELMS-UNIVERSITAT BONN This document has been prepared in the framework of the project CLEVER. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Executive Agency. Neither the European Union nor the granting authority can be held responsible for them. 3 This project has received funding from the European Research Executive Agency under HORIZON Research and Innovation Actions, grant agreement no101060765; and from the UK Research and Innovation (UKRI) under the UK government's Horizon Europe funding guarantee [grant number 10038491] 4 Contents Table of Contents EXECUTIVE SUMMARY .................................................................................................................... 5 Overview of methods .................................................................................................................... 5 Summary for soy supply chains ...................................................................................................... 5 Summary for forestry supply chains ............................................................................................... 8 Summary for crop aquafeed supply chains .................................................................................... 11 INTRODUCTION & OBJECTIVES ..................................................................................................... 13 Soy supply chains .......................................................................................................................... 13 Forest supply chains ..................................................................................................................... 15 Aquaculture and aquafeed supply chains ...................................................................................... 17 METHODS ........................................................................................................................................ 19 Soy supply chains .......................................................................................................................... 19 Overall approach ............................................................................................................................................... 19 Scenario description .......................................................................................................................................... 20 Quantification .................................................................................................................................................... 25 Forest supply chains ..................................................................................................................... 26 Overall approach ............................................................................................................................................... 26 Scenario description .......................................................................................................................................... 28 Quantification .................................................................................................................................................... 32 Aquaculture and aquafeed supply chains ...................................................................................... 33 Overall approach ............................................................................................................................................... 33 Scenario description .......................................................................................................................................... 34 Quantification .................................................................................................................................................... 38 RESULTS AND DISCUSSION ............................................................................................................ 39 Soy supply chains .......................................................................................................................... 39 Results ............................................................................................................................................................... 39 Discussion .......................................................................................................................................................... 70 Forest supply chains ..................................................................................................................... 81 Results ............................................................................................................................................................... 81 Discussion .......................................................................................................................................................... 94 Aquaculture & aquafeed supply chains ......................................................................................... 96 Results ............................................................................................................................................................... 96 Discussion ........................................................................................................................................................ 105 PROJECT OUTPUTS ACHIEVED .................................................................................................... 108 REFERENCES .................................................................................................................................. 109 5 EXECUTIVE SUMMARY This deliverable contains the analysis of future scenarios to 2050 for three supply chains: soy, forest and aquafeeds. While it complements the analysis from an earlier deliverable (D7.2) with the analysis of additional scenarios, most of the information contained in D7.2 is included as well, at least in a summary form, to enable an analysis of the full scenario set. For the scenarios already covered in D7.2, the reader might however find more details in D7.2 than in this deliverable. Overview of methods The scenarios have been co-designed through literature review and dedicated stakeholder workshops, and quantified with the GLOBIOM land use model, which projects the dynamics of most important agricultural and forestry supply chains and their socio-economic and environmental impacts from the year 2000 into the future with a decadal time step (up to 2050 or even 2100). Impacts are analysed for various indicators related to market balance (demand, supply, trade) as well as environmental (e.g., land and water use, GHG emissions, biodiversity) and economic (e.g., food availability, value of production) outcomes, at global scale with a focus on the EU and Brazil. A particular effort is made to quantify impacts on biodiversity, with multiple ecosystems and impact pathways considered when possible. A brief introduction to the modelling framework is provided in the deliverable D7.1, while a detailed description of dedicated modelling improvements conducted as part of Work Package 6 activities can be found in the deliverables D6.1, D6.2, D6.3, D6.4 and D6.5. For all three supply chains, the scenario set contains a baseline scenario (MS15) depicting what these supply chains might look like if historical trends in demand, supply and trade are prolonged in the future. We then consider alternative futures, specific to each supply chains, and varying various aspects related to potential alternative sectorial developments, partly in response to the global climate and biodiversity crisis, but also investigating alternative assumptions about land management strategies, technological progress, demand preferences, and trade networks. Altogether, these scenarios provide contrasted explorative futures, providing a broad picture of what future sectorial trends might look like, and what the contribution of key underlying assumptions are. The literature review underpinning the preliminary scenarios (prior to the stakeholder workshops) are provided in the INTRODUCTION section, while the feedback from the stakeholders and the scenarios for each supply chains are detailed in the METHODS section. For each supply chain, datasets of the model outputs are provided as separate Zenodo records, listed in the PROJECTS OUTPUTS ACHIEVED section. Summary for soy supply chains The baseline scenario depicts a business-as-usual future, based on the ‘Middle of the Road’ Shared Socioeconomic Pathway (SSP2), complemented with key land use policies in Brazil. Under this business-as-usual scenario, the global demand for livestock and crop products keeps growing, and the exports of soy-based products from Brazil increase by 76% over the 2020-2050 6 period, with increases to the EU but primarily to Asia. While food availability and the value of agricultural production increase globally, further biodiversity losses from agricultural activities are projected. In Brazil, although limited by the Amazon Soy Moratorium and the Forest Code, additional deforestation is projected, primarily through conversion to pastures. The increase in soy production is achieved mainly through an expansion over pastures and yield gains. Biodiversity impacts from agriculture increase for both aquatic and terrestrial ecosystems, although at a slower pace than in recent decades for terrestrial biodiversity. Overall, this points to a clear trade-off between economic and environmental goals in the coming decades, with considerable development opportunities for the soy sector in Brazil, and a modest role of the EU with moderately increasing imports but a small and declining market share. A first alternative scenario in D7.2 explored the impact of a global food systems sustainability transition towards bending global biodiversity loss. Based on the Integrated Action Portfolio (IAP) scenario from the Bending the Curve study (Leclère et al., 2020), it entails a comprehensive package of conservation and restoration measures, demand-side measures (lower share of animal products in the diets, reduced waste and loss) and supply side measures (sustainably increased yields). It leads to similar outcomes in terms of food availability globally, but to large decreases in the value of production for livestock products and, to a lower extent, soy through a reduction in livestock feed demand. The production and trade of soy is projected increase slightly above 2020 levels, and exports of soy-based products from Brazil increase by about 20% over 2020-2050, while soya imports from the EU decrease below 2020 levels. As a result, biodiversity losses decrease significantly globally, as well as in Brazil through avoided land conversion and the restoration of low intensity pastures. Overall, this points to the risks efforts towards the global environmental crisis, and in particular diet shifts, may pose to the livestock and soy sector, globally and in particular in Brazil. A second alternative scenario in D7.2 explored the impact of an idealized ambitious conservation policy in Brazil. It entails a sensitivity experiment in which, on top of the baseline scenario no conversion of either forest or other natural land to agricultural land is allowed, to better understand the boundary conditions of the Brazilian soy sector. As opposed to the previous scenario, it does not assume any global food systems sustainability transition. This scenario leads to very similar outcomes than the baseline scenario, including a very large increase in soy production and exports from Brazil (including some increase to the EU), as well as in the value of agricultural production. However, it leads to a strong decrease in biodiversity impacts, even lower than in the previous scenario for land-use change-mediated impacts on terrestrial ecosystems, for one out of the two related indicators. Although this scenario is projected to lead to much lower levels of pasture restoration as compared to the previous scenario, this points to a large margin for achieving ambitious biodiversity outcomes in Brazil through halting land degradation without sacrificing potential economic opportunities. This indicates a potential interest from Brazil soy producers to support ambitious conservation in Brazil, in order to avoid a global shift in human diets. A last alternative scenario in D7.2 explored the long-term impacts of the US-China trade dispute on soy markets. Based on recent developments in soya exports from USA to China, it explores the implications of an assumed cap on USA soya exports to China set 25% below 2020 levels, picturing long-lasting reduced market shares for USA soya exports to China through trust 7 and reliability issues. In this scenario, we project China to turn to Brazil to compensate for most of the import shortfall. This is projected to be achieved with limited impacts on other export destination of Brazil’s soy-based products, while the export shortfall from the USA is not reallocated to other destinations, in a context of very competitive markets. Trends in the import of soya-based products to the EU from different sourcing regions are expected to remain similar to the baseline. Impacts on biodiversity in Brazil and elsewhere, which are significant increasing in the baseline, are not projected to be significantly higher in this scenario. Moderate losses in soy producer revenues are projected in the USA as compared to the baseline, as a result of both production and price declines. Under the condition that current key land use policies for the soy sector stay in place, this points to the US-China trade dispute being a very modest threat to Brazilian ecosystems, as compared to the growing demand for livestock products, in particular in China. The first two scenarios developed in D7.3 look at the impact of the implementation of new EU trade-related policies, namely the EU Deforestation Regulation (EUDR) and a combination of the EUDR and the EU-Mercosur trade agreement (EUM+EUDR). Based on econometrics modelling and a literature review, the EUDR and the EUM+EUDR are estimated to lead to a 3% decline (through EUDR compliance costs) and a 15% increase (through a combination of EUDR compliance costs and EUM-related increased competitiveness), respectively, in Brazil soy exports to the EU compared to the SSP2 baseline scenario by 2050. We assumed that the supply of deforestation-free soy in Brazil is sufficient to cover EU imports. The implementation of the EU policies leads to small reallocation in soy trade but do not affect overall Brazil soy export levels and production compared to the SSP2. Given that the EU policies do not affect Brazil soy production, they lead to similar outcomes for land use, AFOLU emissions and biodiversity loss in Brazil as the SSP2 scenario. This points to the limited impact of EU policies on global markets, but also on Brazil soy sector and ecosystems given the large availability of EUDR-compliant soy in Brazil and the small share of the EU in total soy exports from Brazil. The next two new scenarios in D7.3 consider the reaction of public and private stakeholders in Brazil to the new EU policies, either strengthening or weakening their land conservation efforts. It is first assumed that the implementation of the EU policies leads to zero absolute conversion of forest and other natural land to agriculture in Brazil due to incentives for stakeholders to maintain or increase their access to the EU market. This results in slightly lower post-2020 growth in Brazil soy production and exports compared to the SSP2 but in large reduction in AFOLU emissions and biodiversity impacts. This suggests that EU policies could have positive impacts on biodiversity in Brazil without compromising economic opportunities for farmers if they succeed to support increased domestic conservation efforts (‘Brussels effect’). The opposite assumption is then made, namely that the EU policies lead to a weakening of public and private land conservation efforts in Brazil. The latter is modelled as a removal of the Amazon Soy Moratorium production constraint together with a reduction in the probability of illegal deforestation controls as part of the Forest Code in both the Amazonia and Cerrado biomes. This leads to a reallocation of soy production (and other crops to a lower extent) to the Amazonia biome and mostly away from the Cerrado but to a limited expansion in overall soy production and soy area in Brazil compared to the SSP2. Given the limited overall land use impacts in Brazil, impacts on biodiversity by 2050 are small, and of inconsistent sign across metrics. However, cumulative biodiversity impacts associated with land use change over the 2020-50 period are 8 significantly worse than when considering the implementation of the EU policies on their own and the terrestrial ecosystems impact of the soy supply chain are quite higher than in the 2050 SSP2. Nevertheless, this suggests that the potential adverse effects of deregulation in Brazil in response to EU policies might be limited in absence of higher demand, and assuming limited ‘land grabbing’ dynamics. Finally, the last two scenarios of D7.3 explore the impact of alternative policies to the EUDR in reducing the environmental footprint of EU imports. First, the implementation of the EU border biodiversity adjustment tax on forest risk commodities (i.e., soy-based products, palm oil and beef) is considered. The tax leads to an overall decline in EU imports of palm oil and soy-based products as well as a substitution away from BRA and OSA soy imports towards imports from regions with lower biodiversity footprint for soy (mainly the US) and domestic soy production, to a lower extent. This leads to small decline in the production volume and value of soy in BRA and OSA, and an increase in the EU (although from a very low basis) and NAM, to a lower extent. However, given the limited impact of the tax on overall production levels and land use in the different regions, it does not impact terrestrial and aquatic ecosystems in the EU, and only leads to a small decline in impacts on freshwater ecosystems in Brazil and a small increase in NAM compared to the SSP2. An increase in EU demand-side sustainability efforts is also considered as an alternative to the EUDR, including both a diet shift away from animal products and waste reduction for farm to fork. The shift in EU diets leads to large reduction in the production volume and even larger decline in production value of livestock and crop products in the EU compared to the 2050 SSP2. This results in a drop in pastureland and cropland (to a lower extent) and corresponding increase in other natural lands. AFOLU emissions and biodiversity impacts are significantly reduced compared to the SSP2 (but remain higher than in the IAP scenario which also considers supplyside and conservation and restoration measures in addition to measures on the demand-side). EU imports of soy-based products also drop compared to the 2050 SSP2, mainly due to a decline in its imports from Brazil. However, this has limited impact on Brazil overall soy exports and production level, and therefore results in biodiversity impacts on freshwater and terrestrial ecosystems that are only slightly (1-2%) lower than the SSP2. This suggests that, despite larger decreases in EU soy imports and a low risk that the foregone Brazil soy exports is reallocated to other destinations, alternative policies do not lead to significantly better biodiversity outcomes in Brazil than the EUDR, due to the low EU share in total soy exports from Brazil. It should also be noted that all three options have contrasted impacts on soy producers in various regions, with EU demand-side measures leading to the most biodiversity impact reductions but also to the highest impact on soy producers overall. Summary for forestry supply chains The baseline scenario depicts a business-as-usual future where bioenergy demand is fixed at 2020 level, construction material demand is driven by population and GDP growth under the Share Socioeconomic Pathway (SSP) 2 and plantation forest area is fixed at 2020 level. An alternative business-as-usual demand scenario, where plantation forests can expand outside the natural forests area is also considered. Under the business-as-usual scenarios, roundwood harvest volumes are projected to moderately increase by 2100. Net exports of wood-based 9 products are relatively stable, with boreal/temperate regions losing competitiveness and tropical regions gaining some. Afforestation uptake leads to an increase in forest carbon storage in the coming decades, compensating for declining carbon storage in the existing natural/seminatural forest area. The global biodiversity impacts from forestry diminish over time, and slightly higher reduction is observed when allowing plantation forest expansion. This suggests that moderate growth in demand for wood-based products could be achieved with an increase in forest carbon storage and with less pressure on biodiversity. A first set of scenarios developed in D7.2 explored the impact of an increase in demand for construction materials and/or bioenergy without allowing for plantation forests expansion. The global demand for bioenergy is assumed to double between 2020 and 2100 and 90% of the new urban population is assumed to live in wooden buildings, with the related demand increase for wood-based products primarily located in Africa and Asia. Higher demand for construction materials leads to a 50% increase in harvest volumes compared to the baseline, mostly occurring in Asia, Africa and Latin America. Natural forest management is largely intensified, with up to 300Mha of natural forests being taken into production, and negative implications for both forest carbon balance and biodiversity. Higher bioenergy demand, on the other hand, has a limited impact on harvest volumes as it is mainly met through an expansion in energy crops production. Higher bioenergy demand results in an increase in forest carbon storage compared to the baseline due to growing carbon storage of BECSS, and a reduction in biodiversity impacts over time, although at a lower rate than in the BAU scenarios. This suggests that strong growth in demand for construction materials could exert significant pressure on biodiversity if met through intensified natural forests management whereas higher demand for bioenergy has a limited impact as it mainly sourced from energy crops expansion. A second set of scenarios developed in D7.2 explored the impact of a similar increase in demand for construction materials and/or bioenergy but with the possibility of expanding plantation forests outside the natural forests area. This leads to an additional increase in harvest volumes as compared to the baseline, due to lower roundwood prices as plantation forests are more efficient than natural/seminatural forests. This also leads to an increase in the share of wood sourced from tropical regions, where the productivity of forest plantations is higher. Importantly, plantation expansion enables the release of 200-300 million hectares of seminatural forests from production and their restoration into natural forests, but leads to some decline in (biodiversity poor) other natural land and agricultural land. Moreover, when considering a joint increase in construction materials and bioenergy demand, available plantation area and therefore spared natural forests area are lower (compared to the scenario with higher material demand only) as energy crops and plantation forests compete for the same land. But overall, plantation forest expansion leads to much better outcomes for forest carbon balance and biodiversity than an intensification in natural forest management (with impacts being closer to baseline levels). This highlights the potential of a large expansion in plantation forests in supporting the development of a wood-based bioeconomy while alleviating pressure on natural forests and biodiversity. However, this also points to a potential trade-off between wood and bioenergy production growth due to land competition between plantation forests and energy crops. 16 management and wood-based products supply chains, but does not include the agriculture sector (Azuero-Pedraza et al., 2024; Fulvio et al., 2025a; Lauri et al., 2021; Schulte et al., 2025). Wood-based products have been offered as a solution to reduce CO2 emissions in the energy and material sectors. Woody biomass can be used to produce different final energy carriers such as heat, power and transport fuels as well as different materials such as bioplastic, textiles, packaging and construction materials. Moreover, about 25 % (1150 Mha) of total global forest area (4060 Mha) is currently used for production, so it would be possible to increase woody biomass production by mobilizing the remaining 75% (2910 Mha) of forest area for production (FAO, 2020). However, the remaining forest area consists of biodiversity-rich ecosystems that provide important ecosystem services such as carbon sequestration, soil protection or groundwater filtration. These services are unlikely to be maintained if the production forest area was to be considerably increased from its current level. Therefore, there is a need for new types of forest supply chain solutions, which do not compromise the supply of other forest-based ecosystem services. Traditionally woody biomass (or roundwood) has been produced in natural or seminatural forests, but during the last 30 years an increasing share of roundwood production has moved to plantation forests. Plantation forests are “intensively managed planted forests, which specifically include short rotation plantations and exclude forests planted for protection or restoration”(FAO, 2020). Currently they cover about 10% of the global production forest area and account for about 30% of the global roundwood supply (Mishra et al., 2021). This highlights the importance of plantation forests with respect to meeting increasing demand of woody biomass for bioenergy and construction materials. Plantation forests could also play an important role in the sustainability of forest management, as roundwood productivity is 2-5 times higher than in natural or semi-natural forests. Therefore, moving roundwood production from natural/seminatural forest to plantation forests would free up more area for the supply of other forest-based ecosystem services. However, plantation forests expansion have also raised concern about losing out old-growth grassland and other biodiversity rich natural vegetation areas (Bond, 2016). This may occur directly if plantation forests develop over such ecosystems, but may also occur indirectly if plantation forests develop over agricultural land and trigger the conversion of natural ecosystems into agricultural land elsewhere. To address this issue, in our model development we separate “other natural land” to “abandoned land” and “natural land”, and limit “natural land” conversion to plantations, afforestation or agricultural land. A circular bioeconomy can also play a role in improving the sustainability of forest management. A circular bioeconomy is an economic system that combines the principles of the circular economy (i.e., waste reduction, recycling and reuse) with the ones of the bioeconomy (i.e., use of renewable biological resources)(Tan & Lamers, 2021). It has the potential to reduce pressure on forest carbon balance and biodiversity by reducing harvest volumes through increasing wood-based products recycling and reuse and cascading use of biomass. 17 The objective of WP7 for forest supply chains is to analyze the impacts of the wood-based bioeconomy on the forest sector, carbon sequestration, and biodiversity loss. To conduct this analysis, new features of forest supply chains such as plantation forests, wood-based final products demand, forest management, age-class dynamics and harvested wood products (HWP) carbon accounting, substitution effects between wood-based products and fossil fuel-based products, and plantation carbon uptake have been included in the GLOBIOM-forest model. The role of plantation forests and circularity in the wood-based bioeconomy, and the associated potential benefits and trade-offs for both climate mitigation and biodiversity, remain largely unexplored in the literature. As part of CLEVER WP7 Task 7.3, new scenarios were designed to explore the role of a circular bioeconomy (i.e., increased wood recycling and reuse, and cascading use of woody-biomass) and plantation forests in meeting increasing global demand for construction materials and bioenergy. D7.3 also includes additional scenarios exploring the potential impact of the EU Biodiversity Strategy for 2030 and of different types of land restoration policies in Brazil on the forest sector, forest carbon balance and biodiversity. Aquaculture and aquafeed supply chains The demand for aquatic food products is expected to increase in the coming decades, as result of increasing population and per capita incomes (Naylor et al., 2021), as well as efforts to reduce the environmental impact of food consumption (Halpern et al., 2022), and improve food security (Gephart & Golden, 2022). As marine catch stagnates and may already be beyond sustainable potentials, a large share of the future growth in aquatic food demand is expected to be met by aquaculture (FAO, 2018). Yet, about half of today’s total aquaculture requires feed inputs that were traditionally composed of fish meal and fish oil aquafeeds. These are produced from the reduction of pelagic fish catch, and margins to increase the supply of traditional aquafeeds may be limited by the ecological impacts of the fish reduction sector (Froehlich, Jacobsen, et al., 2018). While improvements in aquaculture feed conversion efficiencies (Gephart et al., 2021) or the uptake of novel aquafeeds (Cottrell et al., 2020) may contribute to alleviating such a pressure, crop-based aquafeed are increasingly being used as an alternative to traditional aquafeeds (Tacon & Metian, 2015). Future scenarios about dietary changes, food demand and aquaculture and aquafeed developments indicate potentially important land use implications. Blue food demand and crop aquafeeds are therefore considered as one of the main interactions between land and sea supply chains (Cottrell et al., 2018). For example, (Froehlich, Runge, et al., 2018) estimated that the future increase in land requirements for crop-based aquafeed to sustain demand for blue food products could be significant, although dwarfed by the land requirements for crop-based feed to sustain future demand for livestock products. The study also estimated aquaculture production to be more efficient in crop feed use than livestock production, despite the expected increased reliance of aquaculture on crop feed. The authors found that scenarios assuming a substitution of livestock products by aquaculture products in future food demand might decrease the overall land use pressure from food demand. 18 Such estimates, however, rely on static representations of the agricultural sector, thereby ignoring potential adjustments in markets and in land use systems expected in such type of scenarios. They also lack a translation of the projected land use impacts into biodiversity impacts. Exploring these questions requires dynamic and integrated modelling of the blue food and agricultural sectors, and might be facilitated by a better integration of blue food systems into integrated modelling tools. As described in CLEVER deliverable D6.2, the GLOBIOM global land use model has recently been extended to cover key blue food supply chains, including an endogenous representation of demand, trade and production for fish products (Spillias et al., 2025). Fish production covers both the catch from marine fisheries and the supply from aquaculture production systems, these two forms of production being connected through the fish reduction sector (i.e., processing marine catch into fish meal and fish oil, used as aquaculture feed). The aquaculture sector is connected to the GLOBIOM crop sector through crop aquafeed, while the demand for blue food final products is explicitly modelled and scenarios of dietary substitutions between blue food and other food products can be considered. The objective of deliverable 7.3 for aquaculture and aquafeed supply chains is to complement the analyses contained in deliverable 7.2 of potential future evolution of blue food demand and supply, alternative aquaculture feeding strategies, and related impacts on land use and terrestrial biodiversity. D7.2 relied on new scenarios designed to explore these potential interactions, and a quantitative assessment of these at the global scale with the GLOBIOM model improved in WP6 of the CLEVER project. In this deliverable, we specifically add two additional scenarios expanding the spectrum of alternative futures explored, by considering a change in the blue food dietary preferences of consumers, and a more sustainable management of fisheries at the global level (see also CLEVER Deliverable D6.5 for related model developments). 19 METHODS Soy supply chains Overall approach Goal: For soy supply chains, the main objective is to analyze the potential evolution of soy markets and to quantify the potential impact of various governance mechanisms on soy production, trade and associated socio-economic and environmental outcomes. In D7.2, we focused on analyzing the impact of stylized intervention scenarios capturing key sources of uncertainties for soy markets around a business-as-usual future (Middle of the Road Shared Socioeconomic Pathway 2; SSP2). An additional set of scenarios, exploring the impact of specific policies in Brazil and EU has been developed in D7.3. Method summary: The scenarios’ design relies on the Story-and-simulation approach (Alcamo, 2001) with an iterative process between expert-led storyline development, quantification with the GLOBIOM model and feedback from stakeholders. The scenarios were primarily developed through a literature review, as well as expert feedback on scenario ideas and key assumptions (including preliminary quantification), gathered during an online workshop and through written feedback. For the new scenarios developed in D7.3, econometrics modelling and a literature review have been used to estimate the magnitude of the impacts of the EUDR and EU-Mercosur on Brazil’s soy exports to the EU and to parametrize the new scenarios in GLOBIOM. More details on the methodology are provided in D6.5. Literature review: As summarized in the introduction, the literature review suggests that the development of future global demand for animal products and vegetable oils, and well as system-wide interventions across land use and food systems towards global sustainability goals are likely to increasingly shape soy supply chains in the future. Recent trade tensions and tariff escalation between the United States and China are also expected to affect soy trade and environmental outcomes in the coming decades. Finally, new EU trade-related policies (e.g., EUDR, EU-Mercosur), as well domestic policies (e.g., Forest Code) and private sector initiatives (e.g., ASM) in Brazil as well as Brazilian stakeholders’ response to EU policies, are expected to impact Brazil-EU soy trade and associated biodiversity impacts in the next decades. Insights from the stakeholder workshop: A two-hour online workshop took place on 14 May 2025. It gathered 23 soy and agriculture experts, from the private sector, public sector, international organizations and NGOs from both Brazil and Europe. The workshop included: 1) a 20 min presentation from IIASA on model improvements, potential scenario options and preliminary results, followed by a Q&A; 2) a first 20 min breakout session where experts were asked to share their vision for the soy sector by 2050 and a reporting back in plenary; 3) a second 20 min breakout session where experts were asked to identify their preferred scenario options and discuss the likely impact of the selected policies/stylized interventions on soy exports and biodiversity, followed by a reporting back in plenary. Experts mentioned population dynamics, the evolution in China’s soy demand, dietary shifts away from meat towards plant-based products, deforestation regulations and the United States-China trade dispute as main factors 20 likely to impact the soy sector in the coming decades. Several of these elements are considered in the stylized scenarios developed in D7.2. Stakeholders also shared feedback on scenarios considering specific policy interventions in Brazil and EU, individually and in various combinations (e.g., ASM, Forest Code, EUDR, and EU-Mercosur) that have been used to inform D7.3. Experts first shared their interest in having a scenario looking at the impact of the EUDR on its own given that it is a new type of policy for which there is high uncertainty about potential effect, including interactions with the EU-MERCOSUR RTA. They also mentioned the importance of the econometric estimate of the trade costs/tariff equivalent of the EUDR as it will drive the model results and suggested that it could be complemented with estimates from the literature (which has been taken into account - see D6.5 for more details). Stakeholders also highlighted the importance of looking at the impacts of both a weakening (including a removal of the ASM which is currently under threat) and a strengthening of Brazil’s land and forest conservation policies as potential alternative responses from Brazilian actors to EU policies. As a follow up from the online workshop, deliverable D7.2 and a draft of deliverable D7.3 were sent to soy stakeholders by email, with a request for written feedback. Feedback received included suggestion to add a discussion on indirect soy-driven deforestation (i.e., pasture is converted to soy, and forest to pasture), as well as clarification that the EU policy scenarios (EUDR and EUM+EUDR) only target soy trade between the EU and Brazil and therefore do not capture the whole impacts on these policies. Scenario description Based on insights from the literature and the stakeholder workshop, we explore the socioeconomic and environmental impacts of a SSP2 scenario and nine intervention scenarios. The first three intervention scenarios are stylized scenarios that have been developed in the context of D7.2 and aim to capture key sources of uncertainties for soy markets. The next six scenarios are new scenarios developed for D7.3. These mainly aim to explore the impacts of new EU traderelated policies (i.e., EUDR and EU-Mercosur) as well as to capture the potential reaction of public and private stakeholders in Brazil to these new policies (i.e., either reinforcing or weaking their land conservation efforts). The last two scenarios aim to provide insights into the impact of alternative policy options in the EU, namely the implementation of an EU biodiversity border adjustment mechanism and increasing demand-side sustainability efforts (as compared to regulatory approach such as the EUDR). The main assumptions behind each scenario are summarized in Table 1 while a more detailed description of the new scenarios is provided below (See D7.2 for a detailed description of the remaining scenarios). Table 1 - Summary of assumptions for soy supply chain scenarios Scenario name Assumptions Questions explored Middle of the Road Shared Socioeconomic Pathway SSP2 (included in D7.2) Prolongation of historical trends (no policy change) What does a SSP2 future mean for soy markets and biodiversity? Integrated Action Portfolio IAP scenario from the Bending the Conservation and restoration, supply and demand-side efforts aligned with the KMGBF goal of What does reaching ambitious biodiversity goals mean for soy markets? 21 curve study (included in D7.2) reversing global biodiversity loss from land use change by 2050 Tr.Dis (included in D7.2) China’s imports of US soy-based products capped at 75% of 2020 value What could be the long-term consequences of the US-China trade dispute on soy markets and biodiversity? ZNLBra (included in D7.2) Zero absolute conversion of forest and other natural lands to agricultural land use in Brazil after 2020 What strong land conservation in Brazil means for soy production and trade? EUDR (new scenario) After 2020, 3% decline in Brazil exports of soy-based products to the EU compared to SSP2, following the implementation of the EUDR What are the potential impacts of new EU policies on soy trade and biodiversity? EUM+EUDR (new scenario) After 2020, 15% increase in Brazil exports of soy-based products to the EU compared to SSP2, following the implementation of both the EUDR and the EUM EUM+EUDR_ZNLBra (new scenario) EUM+EUDR scenario + ZNLBra scenario (i.e., zero absolute conversion of forest and other natural lands to agricultural land use in Brazil after 2020) i.e., implementation of EU policies, and assumed subsequent strengthening of conservation policies in Brazil What are the impacts of EU policies, if they lead to increased conservation efforts in Brazil? EUM+EUDR_WeakLUR (new scenario) EUM+EUDR scenario combined with a removal of the ASM and a weakening of the illegal deforestation control in the Cerrado and Amazon biomes What are the impacts of EU policies, if they lead to declining conservation efforts in Brazil? EUM+EUBBAM (new scenario) On top of EU-MERCOSUR, a biodiversity border adjustment mechanism for imports to the EU, for high-risk commodities (including soy-based products) is implemented as a substitute for the EUDR. The biodiversity border adjustment mechanism is modeled as a tax on imports volumes, with a tax level and What could be the impact of replacing the EUDR by a biodiversity-focused and pricebased intervention to reduce biodiversity impacts imported to the EU? 22 proportional to the difference in land occupation-related biodiversity impacts per ton of product, between imported and domestically produced. EUM+EUDemSide On top of EU-MERCOSUR, the EUDR is replaced by demandside efforts in the EU. The efforts are similar to that pictured in the IAP scenario (reduced waste and dietary shift from animal to plantbased products), but restricted to the EU. What could be the impact of replacing the EUDR by demandside interventions to reduce overall environmental impacts from consumption in the EU? As illustrated in Table 1, in addition to the SSP2, five new scenarios are considered, namely: • EU Deforestation Regulation (EUDR) scenario: this scenario assumes that, as a result of the EUDR implementation, Brazil’s exports of soy-based products (i.e., soybean, soy oil and soy protein meal) to the EU decline by 3% compared to what is projected in the SSP2 baseline scenario. The effect of the EUDR on Brazil-EU soy trade assumes full compliance is reached (i.e., imports are sourced from land not deforested after 2020), with compliance costs leading a slight reduction in trade volumes, estimated through econometrics modelling and a literature review (see D6.5). This reduction in soy trade between Brazil and EU is implemented as a tax on EU imports of Brazil’s soy-based products from 2020 onwards, equivalent to 3% of the EU price in 2020. The underlying assumption is that a sufficient share of soy production in Brazil can comply with the EUDR requirements 2 at a moderate compliance cost. This scenario enables us to isolate the impact of the EUDR on Brazil’s soy exports to the EU. • EU Deforestation Regulation + EU-Mercosur RTA (EUDR+EUM) scenario: this scenario aims to provide insights into the cumulative effect of the EUDR and the recently agreed (but not yet ratified) RTA between the EU and Mercosur States on Brazil-EU soy trade. While the EUDR restricts market access for soy and other commodities into the EU market, the EU-Mercosur RTA is expected to increase it though enhanced trade integration (mainly through indirect effects, as the tariff level is already low). Based on a combination of literature review and econometric modeling (see D6.5), we estimate plausible that i) the magnitude of the resulting increase in trade might be in the order of +20% (with large uncertainty), and ii) it might slightly exacerbate the negative impact of the EUDR on Brazil’s soy exports to the EU (i.e., with reduction around -5% compared to -3% assumed in the EUDR scenario). Taken together, the two EU policies are 2 The EUDR stipulates that commodities in scope placed on the EU market, or exported from the EU, should be produced on land that was not subject to deforestation (after 31 December 2020) and in accordance with the laws applicable in the country of production. 23 estimated to lead, after 2020, to a 15% increase in Brazil’s exports of soy-based products to the EU compared to the baseline. This is implemented as a subsidy on EU imports of Brazil’s soy-based products from 2020 onwards, equivalent to 17% of the EU price in 2020. Similarly to the EUDR scenario, we assume that a sufficient share of Brazil soy production can comply with the EUDR requirements. This scenario enables us to capture the interactions between different EU trade-related policies and their combined effects on soy trade and biodiversity. • EU Deforestation Regulation + EU-Mercosur RTA + Zero natural land loss in Brazil (EUDR+EUM_ZNLBra) scenario: this scenario explores the potential impacts of new EU trade-related policies (i.e., EU-Mercosur and EUDR) on Brazil-EU soy trade and associated biodiversity impacts, under the assumption that they would foster increased land conservation efforts in Brazil. This is done by simultaneously implementing the EUDR+EUM scenario and the ZNLBra scenario (developed in D7.2), which assumes zero absolute conversion from forest and non-forest natural land to agricultural land in Brazil after 2020 (see D7.2). This scenario assumes that international supply chains and the Brazilian society react to the new EU policies by ambitious supply chain traceability and land conservation efforts (well beyond that of current policies, or even pursued by the EUDR) in Brazil. This represents the manifestation in Brazil of a maximalist version of a ‘Brussels effect’, by which EU actions lead to the adoption of new, ambitious standards. • EU Deforestation Regulation + EU-Mercosur RTA + Weak Land Use Regulation (EUM+EUDR_WeakLUR) scenario: this scenario explores the potential impact of an opposite assumption to the EUDR+EUM_ZNLBra scenario, in which the new EU traderelated policies (i.e., EU-Mercosur and EUDR) are assumed to lead to a decrease in supply chain and land conservation efforts in Brazil. This is done by simultaneously implementing the EUDR+EUM scenario and a weakening of key policies included in the SSP2 and all other scenarios (see details of Forest Code representation in D6.5). This includes from 2030 onwards a removal of the Amazon Soy Moratorium and a weakening of the Forest Code. From 2030 onwards, an increase in land area dedicated to soy production is allowed in the Amazone biome, and the enforcement probabilities for illegal deforestation control as part of the Forest Code are reduced by a factor of 3 in both the Amazonia and Cerrado biomes. However, as in all EUDR-based scenarios, it is assumed that a sufficient share of Brazil’s soy production can comply with the EUDR requirements, and that any soy produced from newly deforested land in Brazil is destined to other markets. This scenario provides insights into the implications of a weakening of public and private land conservation efforts in Brazil as a potential adverse consequence of unilateral interventions from the EU, like the EUDR. • EU-Mercosur RTA + EU biodiversity border adjustment mechanism (EUM+EUBBAM) scenario: this scenario explores the impacts of substituting the EUDR by a biodiversity border adjustment mechanism applied to selected high-risk commodities. This is done by implementing, from 2020 onwards, a tax on the volume of EU imports of selected 24 commodities (soya bean, soya oil, soya cake, palm oil, beef). The level of the tax is proportional to the difference between the exporting region and the EU in the marginal land occupation-related biodiversity impact 3 embedded in the supply of a commodity (including local direct impact but also remote indirect impact, for example through impacts from the production in another of feed imported to produce livestock 4 ), and increases over time to reach by 2050 a value based on a medium-range estimate of the global economic value of nature (Costanza et al., 2014) 5 . For soya beans exported from Brazil to EU, this implies tax values of about 2, 113 and 221 USD2000 per t by 2030, 2040 and 2050, respectively. This scenario provides insights into the implications of EU mobilizing an alternative instrument to the EUDR for reducing the footprint of its imports, based on a biodiversity-focused price-based approach. • EU-Mercosur RTA + EU demand-side measures (EUM+EUDemSide) scenario: this scenario explores the impacts of substituting the EUDR by demand-side sustainability efforts similar to those assumed applied globally in the IAP scenario (waste reduction, shift from animal-based to plant-based products). This is done by implementing in the EU27 a progressive transition from 2030 onwards, to reaching by 2050 a 50% reduction in food waste and a substitution of 50% of the demand in animal-based products (meat and dairy) by plant-based products. This scenario provides insights into the implications of EU focusing on demand-side efforts as an alternative instrument to the EUDR for reducing the footprint of its imports. Preliminary versions of these scenarios were discussed during the stakeholder workshop and met with interest, as they were considered to well capture the key sources of uncertainties for the future evolution of soy markets as well as the main EU and Brazil policies with implications for soy trade and biodiversity. Indeed, soy experts considered the future evolution in China’s soy demand, dietary shifts away from meat towards plant-based products, deforestation regulations and the United States-China trade dispute as key factors likely to impact the soy sector in the coming decades. Several of these elements have been captured in scenarios developed for D7.2. Moreover, they considered that key policies in both EU and Brazil with potential effects on soy markets have been identified and provided feedback on relevant combinations between 3 Land occupation impacts compare the biodiversity impact of current land use to an undisturbed natural land baseline at the same location, and can be interpreted as the opportunity cost for biodiversity of keeping that land in use, as compared to restoring it. 4 See Deliverable 6.5 for a description of the underlying footprinting module. 5 Reflecting an instrumental human-nature relationship perspective, we set a long-term target for the tax value that reflects a medium estimate of the monetary value of nature, and assume that this long-term tax value is only fully reached by 2050 (5% and 50% of this long-term tax value apply by 2030 and 2040). For the long-term target, we choose a value of 125 trillion USD2000 per year for a potentially disappeared fraction of 1 (i.e, all terrestrial biodiversity lost), based on (Costanza et al., 2014). This is below the 179 trillion USD2025 estimate for the value of all ecosystem services provided by nature (Boston Consulting Group, (Oğuz, 2025)), and above the World Economic Forum estimate of 44 trillion USD/year for the annual economic value generation of sectors moderately or highly dependent on nature and its services (World Economic Forum, 2020). 25 different policies and stylized interventions as well as on their likely impact on soy trade and biodiversity. This feedback has been used to inform the development of new scenarios for D7.3. As part of an effort to improve the representation of land use dynamics in Brazil, key policy and private sector interventions including the Forest Code and the Amazon Soy Moratorium were parameterized in the model. As further detailed in D6.5, the parameterization of the Forest Code assumes restrictions on illegal deforestation in the Amazonia and Cerrado biomes from 2020 onwards, adjusted by enforcement probabilities, together with restoration efforts in all biomes from 2030 onwards modelled as a conversion of cropland and pasture into protected forest. For the Amazon Soy Moratorium no expansion in the land area dedicated to soy in the Amazon biome is assumed. These features apply both to the SSP2 scenario and the intervention scenarios (except for the EUM+EUDR_WeakLUR scenario where the ASM is removed from 2020 onwards and the enforcement probabilities for illegal deforestation control under the Forest Code are reduced). Quantification The GLOBIOM economic model is used to generate projections for market development indicators and environmental and socio-economic indicators from the initial year (2000) until 2050 (with a 10-year time step) for the baseline and the explorative scenarios. Results from 2020 onwards for main global regions, and at subnational scale for Brazil, are analyzed in the results section of this deliverable. Outcomes under the baseline and the stylized policy scenarios are analyzed for different indicators of food consumption, production and trade and several environmental indicators including land use, water use, GHG emissions and biodiversity loss. Biodiversity loss indicators are based on life cycle assessment (LCA) methods and have been developed in the context of D6.3. These indicators have been linked to the GLOBIOM model, and their value has already been projected for a SSP2 scenario in D6.4. The analysis for the SSP2 and the policy scenarios is carried out at the global and regional levels (for five aggregated regions and the EU) 6 as well as at the country and biome level for Brazil. For soy in Brazil, in addition to the biodiversity impact of farming quantified with GLOBIOM, the model was expanded to also include upstream and downstream supply chain impacts based on the results of LCA (see D6.4). These include biodiversity impacts associated with input production (e.g., fertilizer), soy processing and soy transportation, both domestically and abroad. This enables to get a more complete picture of the full range of impact associated with soy supply chains in Brazil. 6 The five aggregated regions considered are Northern America (NAM), Other South America (OSA) (excl Brazil), Africa and Middle East (AME), South Asia (SAS – including China), and the Rest of the World (ROW). 32 periodic restoration and deforestation area constraints for the periods 2030-2100. Periodic areas are based on the G4M model RPC1p9 afforestation and deforestation areas. Plantation forest area can expand freely in Brazil and is fixed at 2020 level in the rest of the world. Bioenergy demand is fixed at 2020 level, demand for construction materials is driven by population and GDP growth from the SSP2 scenario. Quantification The scenarios are quantified with a refined version of the GLOBIOM economic model that includes an enhance representation of the forest sector and is used to generate projections for relevant outcomes until 2100 under the different scenarios. More details on the modelling framework can be found in the CLEVER Deliverables D6.2 and D6.5. Outcomes projected under the various scenarios are analyzed for different indicators related to forest management area, roundwood harvest volume, net trade in wood-based products, forest carbon balance and biodiversity loss, based on indicators presented in D6.2. The analysis of projected outcomes is carried out at the level of 7 world regions, and includes outcomes projected for individual scenarios, as well as differences to a business usual scenario (BASE). 33 Aquaculture and aquafeed supply chains Overall approach Goal: For aquaculture and aquafeed supply chains, the main goal is to quantify potential developments in the demand for blue food products, aquaculture blue food production, fishbased vs crop-based aquafeed requirements, and related risks to the environment. In D7.2, we presented and assessed explorative scenarios picturing plausible future developments around a business-as-usual future (‘Middle of the Road’ Share Socioeconomic Pathway SSP2). In this deliverable, we expand the analysis to include two additional scenarios, exploring interventions related to additional actors i.e., consumers (via alternative dietary preferences) and fisheries (via a transition to more sustainable fisheries management). Methods summary: To design and quantify the scenarios, we rely on the Story-and-simulation approach (Alcamo, 2001) with an iterative process between expert-led storyline development, quantification with the GLOBIOM model and feedback from stakeholders. The scenarios were primarily developed through literature review, as well as feedback from stakeholders on scenario ideas and key assumptions (including preliminary quantification), gathered during an online workshop. Insights from the literature review: As summarized in the introduction, the literature review highlighted future population and dietary preferences as key aspects leading to future demand growth, a stagnation or moderate increase in wild catch levels depending on the extent to which fisheries management limits overfishing, and the growing role of aquaculture to sustain recent and future increases in blue food consumption. It also highlighted efficiency gains and a substitution of traditional fish-based aquafeed (fish meal and fish oil) by crop-based aquafeeds as key recent trends mediating the demand for various aquafeed products. Insights from the stakeholder workshop: A 90-minute online workshop was conducted on December 4th 2024, involving 16 stakeholders from NGOs, universities, and the private sector, predominantly from Europe but with participants from Northern and Latin America. The workshop included an introduction session on the scope of the work, a session dedicated to gathering stakeholder views on their vision for the sector by 2050, and a session dedicated to collecting feedback on the main scenario dimensions, related assumptions and quantified outcomes. More details on the workshop organization can be found in appendix of D7.2. The stakeholders found the initial scenario scope relevant but also expressed interest in alternative sustainable demand (e.g., towards lower trophic species and bivalves) and supply (e.g., novel feeds, integrated and circular systems) scenarios, which were partially addressed in D7.2, and further addressed in D7.3. They also pointed to the need to vary some of the baseline assumptions (e.g., growth in non-fed aquaculture, maximum substitution potential for fish oil), which was incorporated in the design of scenarios in D7.2 and D7.3. They identified some additional markets (e.g., seaweed) and sectoral drivers (e.g., legislation to regulate aquaculture development, geopolitics and trade disruption, land competition) that could be important to 34 capture. Doing so would, however, go beyond the scope of what can be modelled within CLEVER. In terms of outcomes quantified, stakeholders expressed interest in GHG emissions impacts from various blue food sourcing strategies, and in social and economic outcomes related to production (e.g., revenue, human rights), consumption (e.g., food security) and nutrient cycle (e.g., nutrient losses from crop aquafeed production). Except for nutrient losses from crop aquafeed (aquaculture on-farm losses are already reported upon, but not agriculture on-farm losses), the model developments required to quantify these aspects are beyond the scope of CLEVER. Scenario description Based on insights from the literature review and the stakeholder workshop, we focused on exploring the biodiversity impacts from crop aquafeed under a business-as-usual (SSP2) future, together with the impact of key underlying assumptions. As illustrated in Table 5, this led to the creation of eight scenarios (including 2 new scenarios that have been developed in D7.3, and 6 scenarios that were developed in D7.2). These scenarios differ along four main dimensions: i) future trends in the demand for blue food products, ii) the contribution of wild catch to blue food production growth, iii) the contribution of non-fed aquaculture to future blue food production growth, and iv) changes to the aquafeed requirements from fed aquaculture (including feed conversion efficiency, substitutions between fish-based and crop-based aquafeeds, and possible substitutions between aquafeed crops). More details about the assumptions behind each of the scenarios are provided below. Table 5 - Summary of assumptions for aquaculture and aquafeed scenarios Scenario name Scenario rationale Demand assumptions Wild catch assumption Non-fed aquaculture assumptions Aquafeed requirement assumptions SSP2 (included in D7.2) Baseline scenario, prolongation of recent historical trends to 2050 SSP2 trends beyond 2020 2020 levels SSP2 trends beyond 2020 SSP2 trends beyond 2020 BFS20 (included in D7.2) Counterfactual scenario, blue food sector constant to 2020 2020 levels 2020 levels 2020 levels 2020 levels AF20 (included in D7.2) Sensitivity analysis, SSP2 trends beyond 2020 except for aquafeed requirements SSP2 trends 2020 levels SSP2 trends 2020 levels for both feed conversion efficiency and share various input products 35 AFCOMPO20 (included in D7.2) Sensitivity analysis, SSP2 trends beyond 2020 except for substitution between fishbased and cropbased aquafeeds SSP2 trends 2020 levels SSP2 trends SSP2 trends for fish conversion efficiency, but constant level for the relative shares of fishand crop-based aquafeeds in total aquafeed input AFCOMPOCROPMIX (included in D7.2) Sensitivity analysis, SSP2 trends beyond 2020 except for the share of crop composition of crop-based aquafeeds SSP2 trends 2020 levels SSP2 trends SSP2 trends for fish conversion efficiency and fishbased aquafeed, but partial replacement of soy and corn by wheat in crops cropbased aquafeed input for freshwater products UNFEDAC20 (included in D7.2) Sensitivity analysis, SSP2 trends beyond 2020 except for non-fed aquaculture SSP2 trends 2020 levels 2020 levels SSP2 trends SUSFISH (new scenario) Sensitivity analysis, SSP2 trends beyond 2020 except for wild catch, which moderately increase as a result of a global uptake of sustainable fisheries management SSP2 trends 2020 levels + 15% globally, assuming effective fisheries management at Maximum Sustainable Yield (MSY) SSP2 trends SSP2 trends 36 BFDIET (new scenario) Sensitivity analysis, SSP2 trends beyond 2020 except for blue food products dietary patterns, for which a partial substitution of freshwater species-based products by mollusks-based products is assumed by 2050 SSP2 trends + substitution of 20% of freshwater speciesbased food products by mollusksbased products by 2050 in regions if high consumption of freshwater species based products (EAS, SEA, SAS) 2020 levels SSP2 trends SSP2 trends Baseline scenario (SSP2) This baseline scenario relies on a prolongation of historical trends for various components of blue food supply chains. While a more detailed description can be found in CLEVER deliverable D6.2, the main assumptions are summarised below: • Demand trends: we assume future demand trends to follow i) regional scale future population trends from the shared socioeconomic pathway (SSP) 2, and ii) regional scale future trends in dietary preferences from the SSP2 scenario. The latter reflects SSP2 projections of economic wealth (measured in GDP per capita) combined with a relationship between GDP per capita and the demand for individual blue food products estimated over the historical period. • Capacity of various blue food supply sources: to reflect recent historical trends, we assume that after 2020 the future growth in blue food demand is supplied by both unfed and fed aquaculture, while the levels of wild catches remain at 2020 levels. For unfed aquaculture, the regional capacity cannot exceed a projected regional capacity, which is estimated from historical trends and projected into the future (this was introduced after the stakeholder workshop, see D6.5 for model improvements for aquaculture supply chains). For fed aquaculture, capacity is not constrained but the contribution of individual regions to global fed aquaculture supply (i.e., regional market shares) must remain within 20% of 2020 levels. • Technological progress in the blue food sector: we assume both a moderate increase in the share of fish processing waste in fish meal and fish oil production (from about 50% in 2020 to up to 60% by 2050), and a prolongation of recent trends in aquafeed 37 requirements. The latter assume i) further decreases in the economic feed conversion ratio (from 1.3-1.8 in 2020 to 1.1-1.4 in 2050, resulting in less aquafeed requirement per unit of aquaculture output), and ii) further reductions in the share of fish meal and fish oil in aquafeed (from 0.02-0.14 in 2020 to 0.00-0.01 in 2050, resulting in a higher share of crop-based aquafeed in total aquafeed requirements). The share of various crops in total crop-based aquafeed requirements differs across regions but is assumed to remain constant until 2050. Counterfactual scenario (BFS20) This scenario provides a counterfactual, designed to differentiate the land use and biodiversity impacts from the blue food sector development from that of other sectors covered in the model (i.e., agriculture, forestry, bioenergy). In this hypothetical scenario, all components of the blue food sector (demand, supply, processing, trade) remain fixed at 2020 levels, while the other sectors follow SSP2 projections (without climate mitigation effort, i.e., low bioenergy demand). Sensitivity scenario (AF20, AFCOMPO20, AFCOMPOCROPMIX, UNFEDAC20, SUSFISH, BFDIET) These four scenarios are designed to isolate the impact of specific assumptions from the baseline scenario, but with variation in the assumed post-2020 trends in related parameters: • AF20: decreases in total aquafeed requirements per unit of fish product projected in the SSP2 scenario after 2020 are disregarded (i.e., economic feed conversion ratio in each fed aquaculture production system remains constant at 2020 levels), as well as SSP2projected post-2020 further substitutions between crop-based and fish-based aquafeed (i.e., the share of individual crops in total aquafeed requirements remains constant at 2020 levels for each fed aquaculture production system). This scenario isolates the role of future technological change in fed aquaculture, and emerged from initial model developments (see CLEVER Deliverable D6.2). • AFCOMPO20: similar to the AF20 scenario, except that decreases in total aquafeed requirements per unit of fish product projected in the SSP2 scenario after 2020 are included (i.e., the share of individual crops in total aquafeed requirements remains constant at 2020 levels, but the amount of feed requirements per unit of output decreases). This scenario emerged from the stakeholder workshop and complements the AF20 scenario by isolating, within future technological trends in fed aquaculture, the specific role of additional substitution between fish-based and crop-based aquafeeds. • AFCOMPOCROPMIX: similar to the SSP2 scenario, except that 50% of the crude protein content from corn and soya feed requirements per unit of aquaculture output for fed aquaculture in China are replaced by a similar crude protein content from wheat. This scenario explores the impact of assuming that aquafeed from crops commonly grown in biodiversity-rich tropical areas is replaced by domestically produced temperate crops in the largest fed aquaculture producing region. This scenario emerged from the discussion of initial model developments (see CLEVER Deliverable D6.2, and Deliverable D6.5 for parameterization). 38 • UNFEDAC20: similar to the SSP2 scenario, except that the supply of unfed aquaculture at regional level cannot exceed 2020 levels. This scenario emerged from the stakeholder workshop discussions. • SUSFISH: similar to SSP2 scenario, except that the supply of wild catch is increased globally by 15% as a result of an assumed global uptake of effective sustainable fisheries management aiming at avoiding overfishing. Underlying parameterization is based on estimates from (Elleby et al., 2025), that relied on fish stock modeling (see Deliverable D6.5 for details). • BFDIET: similar to SSP2 scenario, except that a partial substitution between blue food products is assumed after 2020. For the three regions with the highest level of consumption of freshwater fish by 2050 (EAS – except for Japan and South Korea, where freshwater fish consumption is comparatively lower – SAS and SEA), we assume that 20% of the food consumption level of freshwater species-based products projected by 2050 is replaced by mollusks-based products. This rests on the assumption that consumers switch to products that score high in nutritional quality and low in environmental impacts (based on (Gephart & Golden, 2022)). This represents more than a doubling of mollusks species-based products globally, and the substitution occurs gradually after 2020 (see also deliverable D6.5). Quantification The scenarios are quantified with a version of the GLOBIOM economic model that includes a fish module, and is used to generate projections for relevant outcomes until 2050 for the different scenarios. More details on the modelling framework, the baseline scenario parameterization and illustrative model outputs can be found in the CLEVER Deliverable D6.2, and additional model improvements can be found in CLEVER Deliverable D6.5. Outcomes projected under the various scenarios are analyzed for different indicators related to the demand and supply of various blue food products, aquafeed requirements, land use changes and biodiversity impacts from land use, water use, GHG emissions and biodiversity loss. Biodiversity loss indicators are based on life cycle assessment (LCA) methods and have been developed in the context of D6.3. The analysis of projected outcomes is carried out at the level of 10 world regions 8 , and includes outcomes projected for individual scenarios, as well as differences to the counterfactual scenario. 8 The ten world regions considered Northern America (NAM), Latin America and Caribbean (LAC), Sub-Saharan Africa (SSA), Middle East and Northern Africa (MEN), Europe (EUR), Former Soviet Union (FSU), Eastern Asia (EAS, including China), RESULTS AND DISCUSSION Soy supply chains Results Results at the global and regional levels Food availability As illustrated in Figure 1, food consumption patterns differ across aggregated regions both in terms of total consumption and diet composition. In the SSP2 scenario, per capita food availability is projected to increase between 2020 and 2050 in BRA and OSA, ROW, SAS (incl. China) and AME and to stay broadly stable in NAM and EUE where diets have stabilized. The consumption of animal products and their shares in diets are expected to grow in most regions due to growth in per capita incomes, except in NAM and EUE as consumption is already at high levels. As discussed in D7.2, in the IAP scenario, which assumes the replacement of half of animal calories by plant-based calories in regions with high consumption, the share of plant-based products in diets, but also total food availability in 2050 are higher than in the baseline. Per capita availability of plant-based products increases due to the assumed dietary shift but also because of a decline in crop prices due to a large drop in feed demand and reduced land scarcity. The Tr.Dis scenario, which assumes a cap on China’s imports of US soy-based products from 2020 onwards, has no visible impact on per capita food availability at the regional level, which suggests that this scenario does not generate significant impacts on global markets. Zero forest and other natural lands conversion to agriculture in Brazil after 2020, as assumed in the ZNLBra scenario, is expected to have a limited impact on food availability, with only a small decline in per capita availability of animal products in Brazil compared to the 2050 SSP2. The EUDR, EUM+EUDR and EUM+EUDR_WeakLUR scenarios also have no impact on calorie availability in Brazil, EU or any other aggregated regions, indicating that they do not generate significant impacts on markets. The EUM+EUDR_ZNLBra scenario has a similar impact on food availability in Brazil than the ZNLBra scenario i.e., it leads to a small decline in in per capita availability of animal products compared to the baseline. The EUM+EUDemSide scenario leads to similar outcomes for food availability in the EU as the IAP scenario i.e., an increase in the share of plant-based products in diets, but also in total food availability in 2050 compared to the baseline. As explained above, this is due to the assumed shift in EU diets away from animal products but also because of a decline in crop prices due to a large drop in feed demand and reduced land scarcity. 40 The EUM+EUBBAM scenario leads to a small decline in calorie availability in the EU compared to the 2050 SSP2 scenario (-1.5%), mainly due to slightly lower per capita availability of plantbased products – following the implementation of an EU border biodiversity adjustment tax on selected high-risk commodities. Figure 1. Projected trends in food availability per capita (in energetic content, kcal/cap/day) for five aggregated and two zoomed regions (Brazil BRA and European Union EUE). The colors differentiate animal-based (turquoise) vs plant-based food products (yellow). Production and net trade Figure 2 displays trends in production and net trade by aggregated regions for livestock products, soybean and other crops while Figure 3 shows net bilateral trade in all soy-based products (i.e., soybean, soy oil and soybean meal - in soybean equivalent) between aggregated regions. In the SSP2 scenario, agricultural production is projected to increase globally between 2020 and 2050, with most output growth occurring in SAS, AME and OSA. Soy production is projected to grow in all producing regions, but more strongly in Brazil, which is expected to account for over half of global output growth by 2050. Net exporters of livestock products are expected to increase their trade surpluses between 2020 and 2050 while for other crops, net exports are projected to increase in Brazil but decline in NAM and OSA. Net exporters of soy-based products are also projected to increase their trade surpluses, with Brazil consolidating its position as the world’s largest exporter and increasing its 41 net exports by 76% between 2020 and 2050. AME, in turn, is expected to see its agricultural trade deficit widen whereas SAS is projected to reduce its net imports of both livestock and crop products (excl. soy-based products) due to strong production growth. In contrast, SAS imports of soy-based products represent most of the projected global increases in net imports. As discussed in D7.2, in the IAP scenario, livestock production drops in all regions compared to the 2050 SSP2 scenario, together with global soy production - for which feed use accounts for the largest share of total demand. The production of other crops also declines slightly from the 2050 SSP2 - although at a lower rate than soy production - as the reduction in demand for feed crops offsets the increase in demand for plant-based food. Net exports of soy-based products from Brazil increase at a far lower rate than in the SSP2 (20% compared to 76%), and its exports to the EU even decline. In line with the observed trends in production, net trade in livestock and other crops also increase at a much slower rate than in the baseline in all regions. In the Tr.Dis scenario, NAM soy production and net exports of soy-based products decline from 2050 SSP2 level, with a corresponding increase in production and exports in Brazil (+5% and 6%, respectively, compared to the 2050 SSP2) and OSA, to a lower extent. This is in line with what has been observed during the recent US-China trade disputes, where Brazil absorbs China’s soy demand that is not met by the US, and limited re-allocation of US exports to other markets. The ZNLBra scenario only has a small impact on agricultural production in the country; with soy and livestock output slightly declining from 2050 baseline levels (but increasing from 2020 levels). Accordingly, Brazil net exports of livestock and soy-based products increase slightly less than in the baseline (-3% for net exports of soy-based products compared to the 2050 SSP2), which benefit other soy exporters (NAM and OSA). As assumed in the EUDR scenario, the implementation of the EUDR leads to a 3% decline in Brazil’s exports of soy-based products to the EU compared to the SSP2 by 2050. The decline in Brazil’s net exports to the EU (-0.6 Mt compared to the 2050 SSP2) is very moderately offset by an increase in its exports to other regions (+0.2 Mt) (OSA, SAS – incl. China, ROW and AME) – indicating limited evasion to other markets. Similarly, EU net imports from OSA and NAM slightly increase from the 2050 SSP2 (+0.3Mt) but less than the decline in its imports from Brazil. Overall, the EUDR implementation leads to a marginal reallocation in soy trade but does not affect the level of soy production in Brazil (or elsewhere) nor its overall export level. As assumed in the EUM+EUDR scenario, by 2050 Brazil’s exports of soy-based products to the EU increase by about 15% compared to what is projected in the SSP2 scenario (+3 Mt). Half of the increase in EU net imports of soy-based products from Brazil is offset by lower imports from other regions (i.e., OSA, NAM and ROW), as the EUM trade agreement increases the preference for Brazilian soy exports on the EU market (compared to exports from other regions). Similarly, the increase in Brazil’s net exports to the EU is partly offset by lower exports to other regions (mainly SAS but also ROW, AME and OSA). This leads to small overall increase in total EU net imports (+5%) and Brazil net exports (+1%) of soy-based products compared to the 2050 SSP2, but does not significantly impact the level of soy production in Brazil - which is only 1% higher than 2050 SSP2 level. However, it leads to lower growth in EU soy production than what is projected in the SSP2 (-3.5% from 2050 SSP2 levels). Value of production The projected value of production (USD 2000) for primary agricultural products (i.e., excluding secondary products such as vegetable oils and protein meals) is displayed in Figure 5. It is calculated as production volume multiplied by producer prices, and provides an understanding of changes in the economic value of agricultural production. In the SSP2 scenario, the total value of agricultural production is expected to increase in all regions between 2020 and 2050, with strong increase projected in AME and SAS. The value of soy production is projected to decline in NAM (due to a projected price decrease) and to increase in BRA and OSA. As discussed in D7.2, in the IAP scenario, the value of production drops in all regions compared to the 2050 SSP2, mainly driven by a strong decline in the value of livestock production, which decreases at a higher rate than physical volume due to a drop in market prices. The value of soy production also declines in BRA and OSA (at a similar rate than physical volume) as well as in NAM, despite an increase in production in this region. In the Tr.Dis scenario, the value of soy production declines at a higher rate than physical volume in NAM as soy prices drop due to a reduction in Chinese import demand. Soy production value, in turn, increases in BRA and OSA – broadly in line with changes in physical volume, indicating limited price effects on global markets. In the ZNLBra scenario, agricultural production value in Brazil slightly increases from the 2050 SSP2 due to an increase in the value of livestock production (+8%) as price increases more than production declines. This highlights that the overall economic impact of such intervention could even be positive at the sectorial level. The EUDR scenario has no significant impact on the value of agricultural production in Brazil nor in other aggregated regions, as it does not really impact soy production volume nor market prices. The EUM+EUDR scenario has limited impact on the value of soy production in Brazil but leads to a 11% decline in the value of EU soy production compared to the 2050 SSP2. Soy production value in the EU declines at a stronger rate than physical volume (-11% vs -3.5%) indicating a decline in EU soy prices following the reduction in domestic demand. The EUM+EUDR_ZNLBra scenario has the same impact on the value of agricultural production in Brazil than the ZNLBra scenario i.e., it leads to some increase in the value of livestock production compared to the 2050 SSP2 as price increases more than production declines. In the EUM+EUDR_WeakLUR scenario, the value of agricultural production slightly increases in Brazil compared to the 2050 SSP2, due to a small increase in the value of soy production (+3.5%). The value of soy production increases at a similar rate as physical volume, which indicates limited price effects on markets from a weakening of land conservation efforts in Brazil, but a small increase in production volumes. 49 In the EUM+EUBBAM scenario, the value of agricultural production in the EU increases by 4% compared to the 2050 SSP2 scenario, due to increase in the value of production of other crops (+11%) and soy (+32% but from a very low basis). The values of soy and other crop production increase at a much higher rate than physical volume, indicating a rise in market prices, following the increase in domestic demand. The value of soy production also increases at a higher rate than production volume in NAM (8% vs 1.3%) as the increase in EU demand for NAM soy-based products pushed prices up. In BRA and OSA, on the other hand, the value of soy production declines at a similar rate as production volume indicating limited price effects. Overall, this suggests that an EU border adjustment tax on biodiversity would have positive economic impact for EU and NAM producers but slightly negative implications for BRA and OSA soy producers (due to relatively higher biodiversity footprint for soy), as well as for palm oil producers in SAS. The EUM+EUDemSide scenario leads to important decline in the value of agricultural production in the EU (-44% compared to 2050 SSP2) mainly due to drop in the value of livestock production (-62%) as well as other crops (-11%) and soy production (-41% but from a low basis). The production values of livestock and soy decline at higher rate than physical volume indicating a drop in market prices following the reduction in demand for animal products and feed crops. This suggests that a shift in EU diet could generate trade-offs in terms of economic impact on EU agricultural producers. In BRA and OSA, the value of soy production declines at a similar rate to production volume indicating limited price effects. Figure 5. Projected trends in value of production (in billions of USD2000) for five aggregated regions and two zoomed regions (Brazil BRA and European Union EUE). The colors differentiate livestock products (turquoise, LSP) from soy (purple, SOY) and other primary crops (yellow, OCROP). 50 Land use As shown in Figure 6, the projected changes in land use from 2020 to 2050 vary across regions, with relatively stable land use in NAM, EUE and ROW and further conversion of forest and other land to cropland and pasture in other regions. In a SSP2 future, BRA is expected to have a relatively high and increasing share of land dedicated to soy (including both single cropping and soy-corn double cropping), and a relatively high forest loss rate. However, from 2030 onwards, there is an increase in restoration land following the implementation of the Forest Codes’s Legal Reserve requirements (see D6.5 for more details). As discussed in D7.2, in the IAP scenario, pasture drops in all regions from both 2050 and 2020 SSP2 levels following the shift in diets away from animal products. Cropland also declines slightly from the 2050 SSP2 in all regions (and from 2020 in most regions), with relatively larger drop in land dedicated to soy in Brazil due to the reduced demand for feed crops and the assumed additional yield growth. Freed up agricultural land is converted to restoration land in all regions (as assumed as part of the policy package) 10 , while losses of forest and other natural lands are largely mitigated (preventing the conversion of 80Mha of forest and of 56Mha of other natural land compared to the SSP2). In the Tr.Dis scenario, soy area slightly increases in Brazil from the 2050 SSP2 (+4%) to meet the increase in soy import demand from China, with corresponding decline in pasture and forest to a lower extent. In the ZNLBra scenario, pasture – soy and other cropland to a lower extent - decline in Brazil compared to the 2050 SSP2 scenario, preventing the conversion of 50 Mha of forest and 16 Mha of other natural lands. The EUDR and EUM+EUDR scenarios have no impact on land use in Brazil or in other aggregated regions as they have no impact on agricultural production. The EUM+EUDR_ZNLBra scenario has the same impact on land use as the ZNLBra scenario i.e., it leads to large avoided loss of forests and other natural lands. This suggests that the EU policies could contribute to reduction in natural lands conversion if they succeeded to foster increased land conservation efforts in Brazil. In the EUM+EUDR_WeakLUR scenario, cropland (mainly for soy production) and other natural lands slightly increase in Brazil compared to the 2050 SSP2 (+ 27Mha and 38Mha, respectively), with corresponding decline in forest. This suggests that a weakening of land conservation efforts in Brazil following the implementation of the EU trade-related policies could slightly exacerbate conversion of forests to agricultural land compared to what is projected in the SSP2. This moderate effect is consistent with the fact that in our projections, deforestation remains driven by pasture expansion, and that new soy production is sourced primarily from pasture with limited leakage to forests. 10 In Brazil, restoration efforts as part of the IAP are additional to those as part of the Forest Code (with large increase in restoration land compared to the baseline). 51 In the EUM+EUBBAM scenario, soy area in Brazil drops by 2% compared to the 2050 SSP2 following the reduction in EU import demand, with corresponding increase in other natural lands, and pasture (to a lower extent). In the EU, on the other hand, cropland slightly increases compared to the 2050 SSP2 (+3%), with corresponding decline in other natural lands. The EUM+EUDemSide scenario leads to strong decline in pasture (-34%) and cropland to lower extent in the EU due to a reduction in demand for animal products and feed crops, and corresponding increase in other natural lands. It also leads to a 3% decline in soy area in Brazil compared to the 2050 SSP2 and corresponding increase in other natural lands and pasture, to a lower extent. Figure 6. Projected trends in land use (in million hectares) for five aggregated regions and two zoomed regions (Brazil BRA and European Union EUE). The colors differentiate pasture (turquoise) from soy crop physical area (purple) and other primary crop physical area (yellow), as well as forest (red), other land (blue) and restoration land (orange). GHG emissions Figure 7 shows projected GHG emissions from livestock (through enteric fermentation and manure management), soy and other crops (through cropland soil N2O emissions and CH4 emissions from rice cultivation), and LUC (through changes in above ground carbon stocks) in the different scenarios. Overall, livestock accounts for the largest share of AFOLU emissions in most regions, except in BRA and AME where LUC emissions dominate. 52 In the SSP2 scenario, LUC emissions are projected to remain stable (when low in 2020) or decrease (when high in 2020) in all regions between 2020 and 2050, except in AME where an increase is projected due to the conversion of unmanaged land to agricultural land. Crop and livestock emissions, in turn, are projected to increase globally. Overall, total AFOLU GHG emissions are projected to increase in all regions except in Brazil, where the decline in LUC emissions (following the implementation of the Forest Code) more than offset the increase in direct emissions from agriculture. As discussed in D7.2, in the IAP scenario, GHG emissions decrease in all regions compared to the baseline, mainly due to a large drop in LUC emissions (which become a carbon sink in most regions) and in livestock emissions, and some decline in crop emissions. In the Tr.Dis scenario, no significant changes in GHG emissions are projected compared to the SSP2 scenario. In ZNLBra scenario, Brazil’s AFOLU emissions decrease compared to the baseline (but less than in the IAP scenario), mainly due to a large drop in LUC emissions and a small decline in livestock emissions. The EUDR, EUM+EUDR scenarios have a very marginal impact on AFOLU emissions in Brazil nor in other aggregated regions as they barely affect agricultural production nor land use. The EUM+EUDR_WeakLUR scenario also has no clear impact on AFOLU emissions in Brazil by 2050, given its limited impact on land use and the small contribution of soy to total cropland soil emissions. It should however be noted that if measured accumulated over 2020-2050 (rather than in 2050), GHG emissions from land use change are slightly higher. The EUM+EUDR_ZNLBra scenario leads to similar reduction in AFOLU emissions in Brazil than the ZNLBra scenario. The EUM+EUBBAM scenario leads to a small increase in AFOLU emissions in the EU (+1.2% compared to the 2050 SSP2) due to some increase in crop emissions (incl. soy) – following some increase in domestic production. It also leads to a small increase in AFOLU GHG emissions in Brazil, due to a small increase in emissions from LUC (+1%) and other crops (+1.2%), in reaction to the small global markets readjustments occurring by 2050. In the EUM+EUDemSide scenario, EU AFOLU emissions drop by 27% compared to the SSP2 baseline scenario in 2050, mainly due to a strong decline in livestock emissions and some decrease in crop and land use change emissions. However, this decline is less pronounced than in the IAP scenario which also considers supply side and conservation and restoration measures in addition to demand side measures. This scenario also leads to small decline in LUC and soy emissions in Brazil (by 2% and 3%, respectively, compared to the 2050 SSP2) and OSA to a lower extent. 53 Figure 7. Projected trends in annual GHG emissions (in billion tons of CO2 equivalents per year) from the agriculture and land use (AFOLU) sectors for aggregated world regions and two zoomed regions (Brazil BRA and European Union EUE). The colors differentiate emissions from livestock production (turquoise, LSP), soy crop production (purple), other crop production (yellow), and land use change (red). Water use The projected changes in water use for irrigated crop production are displayed in Figure 8. For most regions, irrigation water use is projected to remain stable or slightly increase, with projected increases in water use efficiency mitigating the increases in irrigated crop production. Water use for soy production is only significant in NAM (as soy is not irrigated in BRA) and is projected to stay broadly stable over time. As discussed in D7.2, in the IAP scenario, global water use is projected to be broadly in line with the 2050 SSP2, with a decline in some regions (SAS, and AME) and an increase in others. Soy water use in NAM is projected to increase compared to the baseline due to an increase in soy production. In the Tr.Dis scenario, soy water use slightly declines from 2050 baseline levels in NAM due to a drop in soy production following a reduction in Chinese import demand. The ZNLBra scenario has no clear impact on water use in Brazil. The EUDR, EUM+EUDR, EUM+EUDR_ZNLBra and EUM+EUDR_WeakLUR scenarios also have no significant impact on water use in Brazil nor in other aggregated regions. The EUM+EUBBAM scenario has no significant impact on overall water use in the EU, BRA nor in other aggregated regions. Soy water use in the EU more than doubles following the increase in 54 domestic production but starting from a very low basis, therefore it has very limited impact on overall water use for irrigated crops. Soy water use, in turn, slightly declines in OSA but also starting from a low base while there is no impact on water use in Brazil as soy is not irrigated. The EUM+EUDemSide scenario leads to a decline in soy water use in the EU (which goes to zero) as well as in OSA. Here again, this has very limited impact on overall water use in these regions given the minor contribution of soy to total irrigation water use. Figure 8. Projected trends in water use for irrigation (in km3) for five aggregated regions and two zoomed regions (Brazil BRA and European Union EUE). The colors differentiate soy crops (purple) from other crops (yellow). Biodiversity impacts Biodiversity impacts on terrestrial ecosystems from local agriculture, measured in potentially disappeared fraction of global species integrated over time (pdf.year), are displayed in Figure 9 (see D6.3 for more details on the methodology). In absolute terms, they are higher in regions with a significant share of tropical ecosystems (e.g., SAS, BRA and OSA, and AME). Land occupation has the largest impact in all regions, followed by land transformation (except in NAM where climate change impacts are higher). It should be noted that only biodiversity impacts from the land use and agricultural sectors are accounted for, and total climate change impacts (from all human activities) are larger. In the SSP2 scenario, land transformation impacts are projected to decline between 2020 and 2050, especially in regions where they are currently large (e.g. BRA and OSA, AME, SAS). Land 55 occupation and climate impacts, in turn, are projected to remain broadly stable or increase over time, except for climate impacts in Brazil due the restoration foreseen in the Forest Code. It should be noted that, by assumption, restoration does not directly reduce land transformation impacts on biodiversity: land transformation measures for a conversion into a worse state – e.g., from forest to cropland – that will occur once the habitat is left to recover (i.e., the impacts associated to the recovery of the modified habitat after it is left to recover), while restoration is a transition from modified to pristine habitat and therefore not a transition to a worse state (see CLEVER deliverable D6.3). On the contrary, restoration leads to reduced biodiversity impacts from land occupation (as it counts as pristine land use) and climate change (as it generates negative GHG emissions). As discussed in D7.2, in the IAP scenario, biodiversity impacts on terrestrial ecosystems are expected to be significantly lower than in the SSP2 scenario in all regions, mainly due to a strong decline in climate change impacts but also in land occupation and land transformation impacts. In the ZNLBra scenario, biodiversity impacts on terrestrial ecosystems in Brazil are also below baseline levels (but higher than in the IAP scenario) mainly due to strong decline in land transformation impacts as well as some decline in land occupation and climate change impacts. The Tr.Dis scenario has no clear impact on terrestrial biodiversity. The EUDR and EUM+EUDR scenarios have no significant impact on terrestrial biodiversity in Brazil nor in other aggregated regions. This can be explained by the fact that these scenarios have no significant impact on either agricultural production, land use, or GHG emissions. The EUM+EUDR_WeakLUR scenario also has limited impact on terrestrial biodiversity in Brazil, with overall impact being 1% above the 2050 SSP2 due to slightly higher land occupation and land transformation impacts at the scale of Brazil (but see next sections for a detail of impacts within Brazil). This indicates that the potential impact on biodiversity of deregulation in Brazil in response to EU unilateral measures might be limited in the absence of higher demand. It should also be noted that the model primarily responds to agricultural market dynamics and likely underestimate ‘land grabbing’ dynamics (i.e., deforestation of public or untitled to claim land ownership, without significant agricultural activity for several years if not decades), which could be exacerbated in such a scenario. The EUM+EUDR_ZNLBra scenario leads to similar decline in terrestrial biodiversity impacts as the ZNLBra scenario, which suggests that EU policies could have positive impacts on biodiversity if they succeed to foster increased domestic conservation efforts in Brazil (‘Brussels effect’). The EUM+EUBBAM scenario has no significant impact on terrestrial biodiversity in the EU (nor elsewhere) given its limited impact on land use and GHG emissions. The EUM+EUDemSide scenario leads to 9% decline in terrestrial biodiversity impacts in the EU compared to the 2050 SSP2 scenario. It is mainly driven by a decline in climate change and land occupation impacts and some decline in land transformation impacts. However, biodiversity impacts on terrestrial ecosystems in the EU are higher than in the IAP scenario, which also considers supply side and conservation and restoration measures in addition to diet shift. 56 Figure 9. Projected trends in biodiversity impacts on terrestrial ecosystems (in PDF.year) from local production for five aggregated regions and two zoomed regions (Brazil BRA and European Union EUE). The colors differentiate impacts through climate change (blue), land occupation (red), and land transformation (green). The projected biodiversity impacts from local production on freshwater ecosystems also differ across regions (Figure 10). In absolute terms, it is higher in NAM, SAS and ROW than in tropical regions. In terms of pressures, impacts tend to be dominated by water stress (in particular for NAM), followed by eutrophication (which however is the most important pressure in Brazil) and climate change (which however is the most important pressure in EUE). In the SSP2 scenario, impacts on freshwater ecosystems are projected to remain broadly stable in some regions (e.g., NAM, EUE) or increase (e.g., in OSA and AME due to increased water stress, and in BRA due to eutrophication) between 2020 and 2050. As discussed in D7.2, in the IAP scenario, impacts on freshwater ecosystems drop in most regions compared to the 2050 SSP2, mainly due to a strong decline in climate change impacts. In the Tr.Dis scenario, impacts on freshwater ecosystems decline slightly in NAM compared to the 2050 SSP2 due to a decrease in water stress associated with lower water use for soy production, while they marginally increase in BRA mainly due to a small increase in freshwater eutrophication. In the ZNLBra scenario, biodiversity impacts on freshwater ecosystems in Brazil are below 2050 baseline levels (but higher than in the IAP scenario), mainly due to a decline in climate change and freshwater eutrophication impacts. 57 The EUDR and EUM+EUDR scenarios have no significant impact on freshwater ecosystems in either Brazil or in other aggregated regions as they do not affect water use, AFOLU emissions nor input use. The EUM+EUDR_ZNLBra scenario leads to similar reduction in freshwater biodiversity impacts as the ZNLBra scenario, indicating that EU policies could help reduce pressure on Brazilian aquatic ecosystems if they succeed to foster increased domestic land conservation efforts. The EUM+EUDR_WeakLUR scenario leads to a 3% decline in freshwater biodiversity impacts in Brazil compared to 2050 SSP2 mainly due to lower eutrophication impacts, due to the slight increase in cropland for soy and other crops. The EUM+EUBBAM scenario has no significant impact on freshwater ecosystems in the EU. However, as compared to the 2050 SSP2, it leads to a 2% increase in biodiversity impacts on freshwater ecosystems in NAM (due to increase in water stress following increase in EU import demand for soy), and a 1% increase in freshwater biodiversity impacts in Brazil due to lower freshwater eutrophication. The EUM+EUDemSide scenario leads to an 8% decline in freshwater ecosystems impacts in the EU compared to the 2050 SSP2, mainly due to a decrease in climate change impact and some decline in freshwater eutrophication and water stress. It also leads to a 2% decline in biodiversity impact on freshwater ecosystems in Brazil, mainly due to lower freshwater eutrophication following a small decline in soy production. 64 Biodiversity impacts Figure 15 shows the impacts on terrestrial ecosystems from farming activities in the six Brazilian biomes, based on two different LCA methodologies. In the figures in the top panel, biodiversity impacts are calculated based on the LC-IMPACT average characterization factors (CFs) (Verones et al., 2020) while those in the bottom panel are calculated based on the new CFs developed for South America in the context of D2.4 (hereafter referred to as ‘CLEVER CFs’) (Oliveira et al., 2019; Oliveira & Pacheco, 2024) It should be noted that results for this figure omit climate change impacts, which are available for the LC-IMPACT CFs but not for the CLEVER CFs. As discussed in D7.2, while both methods tend to predict similar overall trends, the main differences lie in the scale of the impacts (in particular, for transformation impacts – which are about two times higher with the CLEVER CFs) and the relative contribution of different biomes to the overall impacts. These figures vary due to differences in approaches, especially as regards the pristine ecosystem baseline, as described in D6.3. In the SSP2 scenario, with both approaches, the total land impacts on terrestrial ecosystems are projected to decline over time, due to a drop in land transformation (i.e., LUC) impacts and despite an increase in land occupation impacts. As discussed in D7.2, in the IAP scenario, total impact on terrestrial ecosystems decreases compared to the baseline, due to a decline in both land occupation and land transformation impacts. In the ZNLBra scenario, the total impact on terrestrial ecosystems also declines compared to the SSP2 mainly due to a drop in land transformation and some decline in land occupation impacts. However, which of the IAP or ZNLBra scenarios lead to the best outcome for terrestrial biodiversity depends on the method, with the IAP scenario performing better when using LC-IMPACT CFs and the ZNLBra scenario when using the CLEVER CFs. 12 The Tr.Dis scenario, in turn, has no significant impact on terrestrial biodiversity with both metrics. The EUDR and EUM+EUDR scenarios have no impact on terrestrial ecosystems in Brazil with both methods as they do not impact agriculture production and land use nor GHG emissions. It should be noted, however, that in these scenarios, the EU policies only apply to soy trade between Brazil and EU, and not to other commodities (i.e., beef, palm oil, cocoa, coffee, timber and rubber) nor imports from other countries. Therefore, it does not capture the full impact of the EUDR and EUM on biodiversity, and it should also noted that our estimates rely on the assumed effects of these policies (see discussion). As a reminder, the EUM is assumed to increase exports to the EU by 20%, while a full compliance to EUDR is assumed for soy exports to the EU (given the limited share of EU exports in total Brazil production), with compliance costs leading to a reduction of exports to the EU by 3% (or an increase of 15% when combined with the EUM). Different assumptions here could lead to slightly different estimates. In addition, although doing so was not feasible within the frame of CLEVER (see D6.5), including available information (e.g., TARSE dataset) about the subnational distribution of market shares for various soy exports to 12 The avoided loss of forest and other natural land in the ZNLBra scenario (as compared to the SSP2 scenario) seem to be equally beneficial to biodiversity with both methods (close to full elimination of transformation impacts), while the land use change patterns of the IAP scenario as compared to other scenario seems to result in much higher benefits for biodiversity when using the LC-IMPACT CFs (close to full elimination of transformation impacts) than when using the CLEVER CFs (very limited decrease in transformation impacts). 65 the EU could for example allow for a finer impact on subnational deforestation and soy production patterns; this would however not dramatically change the picture. As the ZNLBra scenario, the EUM+EUDR_ZNLBra scenario leads to large decline in impacts on terrestrial ecosystems compared to the SSP2, which suggests that EU policies could have positive impacts on biodiversity if public and private stakeholders in Brazil react by increasing their land conservation efforts in order to maintain or increase their access to the EU market. In the EUM+EUDR_WeakLUR scenario, when using the LC-IMPACT CFs, the impacts on terrestrial ecosystems increase by 10% in the Amazonia biome compared to the 2050 SSP2 due to an increase in land transformation impacts and some increase in land occupation impacts. Terrestrial ecosystems impacts decline in all other biomes (mainly Mata Atlantica and Cerrado) – mainly due to a decrease in land transformation impacts in Mata Atlantica and in land occupation impacts in the Cerrado. This leads to overall biodiversity impact on terrestrial ecosystems that are 1% higher than the 2050 SSP2. One the one hand, as noted above when discussing land use impacts, this suggests that in the absence of both higher levels of demand for soy products and ‘land grabing’ dynamics, biodiversity impacts of a deregulation might be limited, as measured by comparing the sum of 2050 values for land occupation and transformation impacts. On the other hand, as also noted when discussing land use impacts, accumulated land use change over 2020-2050 is higher in EUM+EUDR_WeakLUR than in EUM+EUDR, and accumulated land transformation impacts from 2020 to 2050 (rather than 2050 value) are projected to be 10% higher over Brazil in the EUM+EUDR_WeakLUR scenario as compared to the EUM+EUDR scenario. A similar pattern occurs for accumulated LUC-related GHG emissions and related biodiversity impacts, suggesting overall that biodiversity impacts from the EUM+EUDR_WeakLUR scenario are clearly larger than that of the EUM+EUDR scenario, if accounting for accumulated impacts of climate change and transformation over the 2020-2050 period. When using the CLEVER CFs, the impacts in 2050 on terrestrial ecosystems are lower in most biomes – including in Amazonia – in the EUM+EUDR_WeakLUR scenario than in the EUM+EUDR scenario. In Amazonia, land occupation impacts increase by 5% compared to the 2050 SSP2 but this is more than offset by a decline in land transformation impacts. In most other biomes (and mainly Cerrado and Mata Atlantica) both land transformation and land occupation impacts decline from the 2050 SSP2. This leads to an overall 7% decrease in terrestrial biodiversity impacts compared to the 2050 SSP2. However, for transformation impacts, similar to impacts estimated with the LC-IMPACT characterization factors, the difference between scenarios is less favorable to the EUM+EUDR_WeakLUR scenario when looking at cumulated impacts between 2020 and 2050 (1% higher in EUM+EUDR_WeakLUR) than when looking at 2050 values (20% lower in EUM+EUDR_WeakLUR). Overall, this points to uncertainties in biodiversity impacts, and limited potential biodiversity impacts from a weaker land use regulation in Brazil in our projections (with compensations across biomes). The EUM+EUBBAM and EUM+EUDemSide scenarios have no significant impact on terrestrial ecosystems in Brazil when using both the LC-IMPACT and CLEVER CFs as they have no significant impact on land use nor AFOLU emissions. 66 Figure 15. Projected trends in biodiversity impact for terrestrial ecosystems (Potentially disappeared fraction, PDF·y) for the six biomes and split between the different pressures covered in GLOBIOM. The colors differentiate namely Amazonia (blue), Caatinga (yellow), Cerrado (red), MataAtlantica (turquoise), Pampa (green) and Pantanal (pink). The figures in the top panel are based on LC-IMPACT characterisation factors (CFS) while the figures in the bottom panel are based on spatially-explicit CFs developed for South America in CELEVER deliverable D2.4. As shown in Figure 16, freshwater ecosystems are most affected by farming activities in the Mata Atlantica, Cerrado, and Amazonia biomes. In the SSP2 scenario, the total impact on freshwater ecosystems in Brazil is projected to increase between 2020 and 2050, due to a rise in freshwater eutrophication in all biomes, and some increase in water stress in some biomes (mainly Mata Atlantica). Climate change impacts, however, are projected to decline, due to declining LUC emissions. As discussed in D7.2, in the IAP scenario, the impacts on freshwater ecosystems are projected to be lower than in the SSP2 and become negative, mainly due to strong decline in climate change impacts and some decrease in freshwater eutrophication in all biomes. However, as discussed above, water stress slightly increases from the SSP2 due to a small increase in water use for irrigated crops. In the Tr.Dis scenario, biodiversity impacts on freshwater ecosystems 67 slightly increase from the 2050 SSP2, mainly due to an increase in freshwater eutrophication. In the ZNLBra scenario, the total freshwater ecosystems’ impacts are projected to be lower than in the SSP2 scenario (but higher than in the IAP scenario), mainly due to lower climate change and eutrophication impacts. The EUDR and EUM+EUDR scenarios have no clear impact on freshwater ecosystems in Brazil as they have no impact on GHG emissions, water use nor input use. The EUM+EUDR_ZNLBra scenario leads to similar reduction in freshwater ecosystems impacts as the ZNLBra scenario. In the EUM+EUDR_WeakLUR scenario, the biodiversity impacts on freshwater ecosystems increase in the Amazonia biome compared to the 2050 SSP2 (+18%) due to a strong increase in eutrophication impacts. This is more than offset by a decline in eutrophication impacts in other biomes (mainly Cerrado, and Mata Atlantica), leading to a 3% decrease in total impacts on freshwater ecosystems compared to the 2050 SSP2. In the EUM+EUBBAM scenario, the biodiversity impacts on freshwater ecosystems in Brazil decline by 1% compared to the 2050 SSP2. This is mainly due to slightly lower freshwater ecosystems impacts in all biomes (except Mata Atlantica) following a small reduction in soy production. In the EUM+EUDemSide scenario, the biodiversity impacts on freshwater ecosystems in Brazil decline by 2% compared to the 2050 SSP2. This is mainly due to lower freshwater ecosystems impacts in Cerrado and Mata Atlantica, as well as slightly lower climate change (mainly in Amazonia and Mata Atlantica) and water stress impacts (mainly in Cerrado). Figure 16. Projected trends in biodiversity impact for freshwater ecosystems (Potentially disappeared fraction, PDF·y) for the six biomes and split between the different pressures covered in GLOBIOM. The colors differentiate the biomes, namely Amazonia (blue), Caatinga (yellow), Cerrado (red), MataAtlantica (turquoise), Pampa (green) and Pantanal (pink). 68 Additional impacts from upstream and downstream soy supply chain in Brazil In the previous sections, the projected environmental impacts of Brazilian agricultural activities have been illustrated, based on the outcome of GLOBIOM for a SSP2 scenario (SSP2) and eight stylized policy scenarios. However, parts of the supply chain of agricultural products are missing in GLOBIOM, the model primarily covering the impact of direct farming activities and related changes in resource use. Based on the results of the LCA developed in D6.4, the model was expanded to include upstream and downstream supply chain impacts of soy production in Brazil. These are estimated by combining LCA-based estimates of impacts per ton of soy produced with GLOBIOM projections for the amount of soy produced. In this section, the biodiversity impacts associated with all steps of the soy supply chains are compared across the SSP2 and the different scenarios. We analyze the differences in impact in terms of aquatic and terrestrial extinction risk, but also in terms of GHG emissions and freshwater consumption. Evolution of soy supply chain biodiversity impacts In the SSP2 scenario, impact on terrestrial ecosystems from the soy supply chains are projected to slightly increase overtime due to an increase in upstream (e.g. input production) and downstream (i.e., crushing, transport) supply chain impacts and despite some decline in farming impacts (more specifically land transformation) (Figure 17). The impact on aquatic ecosystems, in turn, is projected to increase strongly between 2020-50, due to an increase in all supply chain impacts (incl. Farming-GLOBIOM), and in particular impacts linked to input production and crushing. It should be noted that for other impacts than the farming impacts covered in GLOBIOM, the LCA coefficients used are estimated directly per ton of product and do not evolve dynamically with the scenarios in GLOBIOM to include potential reduction linked to technological progress – which might lead to an overestimation of these impacts in the projections. As discussed in D7.2, in the IAP scenario, where Brazil’s soy production drops compared to the SSP2 following a decline in global demand, the impacts of the soy supply chain on both terrestrial and aquatic ecosystems decline from the SSP2 levels mainly due to a decrease in farming impacts. Mid-points impacts (GHG emissions and freshwater consumption) also decline from the baseline, mainly due to lower impacts from upstream supply chain activities (i.e., input production) and crushing. In the Tr.Dis scenario, the impact on terrestrial and aquatic ecosystems slightly increase from 2050 baseline levels, mainly due to a rise in farming and upstream supply chain impacts. Mid-points impacts also increase, mainly due to higher impacts from upstream supply chain activities. In the ZNLBra scenario, the end-point impacts of the soy supply chain are below 2050 SSP2 levels (in particular for terrestrial ecosystems) mainly due to lower farming impacts. GHG emissions from the soy supply chain are also slightly below baseline levels, mainly due to lower impacts from farming and crushing, while freshwater consumption is broadly in line with the 2050 BAU. 69 In the EUDR and EUM+EUDR scenarios, the end-point impacts (aquatic and terrestrial biodiversity) and mid-points (GHG emissions and freshwater consumption) of the soy supply chain in Brazil are broadly in line with the 2050 SSP2, with all impacts being only 1% above 2050 BAU levels for the EUM+EUDR scenario. In the EUM+EUDR_ZNLBra, the end-point impacts of the soy supply chain are below 2050 SSP2 levels mainly due to lower farming impacts (by -13% and -5% for terrestrial and aquatic biodiversity, respectively), similarly to the ZNLBra scenario. GHG emissions of the soy supply chain are also slightly below 2050 baseline level (-3%) due to lower impacts from farming and crushing while freshwater consumption is in line with 2050 level. In the EUM+EUDR_WeakLUR scenario, the terrestrial ecosystems impact of the soy supply chain is 24% above 2050 SSP2 level mainly due to higher impacts from farming following some increase in soy production, while the impact on aquatic ecosystems is broadly aligned with the SSP2. GHG emissions from the soy supply chain are 2% above the 2050 SSP2 (mainly due to higher emissions from farming and domestic transport) while freshwater consumption impact is 3% below the 2050 SSP2 (due to lower impact from upstream supply chain activities) In the EUM+EUBBAM scenario, the aquatic ecosystems impacts of the Brazilian soy supply chain are 2% below 2050 SSP2 levels mainly due to lower impacts from upstream supply chain activities (i.e., input production) and farming while the impact on terrestrial ecosystems is 1% below 2050 SSP2 level mainly due to lower impact from farming. The mid-point impacts of the soy supply chain (i.e., GHG emissions and freshwater consumption) are both 2% below 2050 SSP2 levels, mainly due to lower impacts from upstream supply chain activities (i.e., input production) as well as from crushing in the case of GHG emissions. The EUM+EUDemSide scenario, the terrestrial and aquatic ecosystems impacts of the soy supply chain are 1% and 2.5% below 2050 SSP2 levels, respectively, mainly due to lower impacts from farming and from upstream supply chain activities (i.e., input production) associated with slightly lower soy production. The mid-point impacts of the soy supply chain (i.e., GHG emissions and freshwater consumption) are both about 3% below 2050 SSP2 levels, mainly due to lower impacts from upstream supply chain activities (i.e., input production) as well as from domestic transport and crushing in the case of GHG emissions. 70 Figure 17. Evolution of the mid-point impacts (GHG emissions and freshwater consumption) and end-point impacts (freshwater and terrestrial biodiversity) of the soy supply chain over time, differentiated per supply chain step. Discussion Soy is one of the most internationally traded agricultural commodities. Over the past decades, its production and trade expanded rapidly, mainly driven by strong import demand for protein meals in China and the European Union. These international demands have been associated with land use change, deforestation, and subsequent biodiversity loss in Brazil, the world’s largest producer and exporter of soybeans. In D7.2, we explored the impact of alternative future development in soy markets, capturing key sources of uncertainties. In D7.3, we explore the role of various governance initiatives in regulating Brazil-EU soy trade and associated biodiversity impact. This includes an analysis of the potential impact of new trade-related policies (i.e., the EUDR unilateral EU policy, and EUM trade agreement), as well as the potential reaction of public and private stakeholders in Brazil to these new policies – considering both a strengthening and weakening of their domestic land conservation efforts. The potential impacts of alternative policies in the EU that could be mobilized as a substitute for the EUDR to the environmental footprint of imports. These include the implementation of an EU border biodiversity adjustment mechanism and an increase in demand-side sustainability efforts. To ease a synthetic understanding, we cover here discussion elements related to scenarios from both D7.2 and D7.3. Trends in a business-as-usual future When considering the continuation of historical trends in population, diets, trade and productivity, and no change in current policies (SSP2 scenario), the global production, consumption and trade of livestock and crop products are projected to continue increasing by 2050. This includes continuing growth in soy production and trade in soy-based products, with Brazil projected to increased it exports by 76% from 2020 to 2050, and account for more than half of the global growth in output and net exports in that period. Most of the increase in exports are destinated to Asia, while EU also increases its imports level. These broad production and consumption patterns are associated with growing environmental pressures, leading to further climate change impacts and increasing extinction risk for aquatic and terrestrial ecosystems, in particular in Latin America, Asia and Africa. In Brazil, soy production increase is projected to occur primarily in the Cerrado biome, and to some extent, in the Mata Atlantica and Pampa biomes. This growth is achieved in large parts by a conversion of pastures and yield increases. Deforestation is projected to continue albeit at a slower pace than in previous decades. Large forest conversions to pasture are projected, in particular in the Amazon biome, large while moderate forest restoration efforts take place, in particular in the Mata Atlantica biome. It should be noted that the increase in soy area primarily occurs through expansion over pastures, and that the projected expansion of pasture is much larger in amplitude. While it may be considered that part of the pasture increase may be indirectly driven by pasture conversions to soy, it remains a small contributor to pasture-driven deforestation. It should be noted that these trends consider the effects of key domestic interventions in Brazil, such as the Amazon Soy Moratorium (limiting soy expansion in the Amazon Biome) and the Forest Code (limiting but not fully eliminating illegal deforestation, and 71 triggering moderate restoration efforts). A large share of the projected deforestation is considered legal under current Brazilian deforestation regulation. This leads to a small reduction in GHG emissions from the AFOLU sector, primarily driven by a decrease in land use change emissions partially compensated by an increase in GHG emissions from livestock production. Concomitantly, increases in the value of production are projected, in particular for soy, but these remain of a much lower amplitude than projected increases in production volume. Extinction risks from agricultural activities increase for aquatic ecosystems in Brazil, in particular due to eutrophication in the Cerrado biome. For terrestrial ecosystems, impacts from agriculture slightly decline for those associated with GHG emissions and land use change (reflecting the slowing rates of deforestation and increasing restoration efforts), but increase for impacts associated with land occupation (indicating that the land sector is still leading to net increases in land-use change mediated terrestrial biodiversity loss). Although a detailed comparison to the literature would be relevant for aspects such as China’s and EU’s future demand for soy-based products, these trends are broadly consistent with future projections for business as usual scenarios (e.g., for SSP2 scenario (Popp et al., 2017)) and medium-term agricultural outlook (OECD/FAO, 2024), and similar studies at the scale of Brazil (e.g., (Soterroni et al., 2018)). Although it might be relevant to consider additional global drivers, such as impacts from climate change, these are not expected to lead to large changes in Brazil’s soy export potential (Zilli et al., 2020). Summary. Our business-as-usual scenario entails a large expansion of soya production in Brazil beyond 2020 levels, associated to a global increase in demand for soy-based products (in particular in Asia) and a strengthening of Brazil dominance on the global market for soybased products. This contributes to a further increase in socio-economic activity at the expense of the environment (including biodiversity in Brazil), although in Brazil soy expansion generates limited deforestation (this remains a small contribution to Brazil deforestation), and land-use change-related environmental losses moderately decrease in speed as compared to previous decades. While Brazil exports of soy-based products to the EU increase, the EU market share in Brazil soya production keeps decreasing. Impact of alternative futures for soy markets Three explorative scenarios capturing key sources of uncertainty for soy markets have been designed in D7.2. This includes scenarios representing a) a global food system transformation (i.e., the IAP scenario), c) an idealized ambitious land conservation policy in Brazil (i.e., the ZNLBra scenario), and b) the potential long-term implications of the trade dispute between US and China (i.e., the Tr.Dis scenario). Global food system sustainability transition The IAP scenario assumes the global implementation of a mix of demand-side (incl. dietary shift, waste reduction), supply-side (e.g., sustainable yield increases) and conservation (increased 72 protected area extent and effectiveness, land use planning) and restoration measures. In this scenario, the projected global production and trade of livestock products and feed crops (incl. soy) by 2050 drops compared to the SSP2. This leads to a much more limited increase in net exports of soy-related products from Brazil (+20% instead of +76%). Net exports of soy-based products to the EU decrease compared to 2020 levels, while most other importing regions maintain total imports levels slightly above 2020 levels. This scenario is projected to lead to decline in all environmental impacts but also entail important socio-economic trade-offs, with large decreases in the value of livestock production, primarily as a result of changes in consumer preferences. In Brazil, a significant amount of pasture is restored, while forest and other natural land losses are partially mitigated, as compared to the SSP2 scenario. Land-use change becomes a large carbon sink, and the biodiversity impacts of local agriculture on both aquatic and terrestrial ecosystems are largely reduced. However, it leads to significant forgone economic opportunities for farmers, with the value of agricultural production declining from 2050 but also from 2020 levels. Decreases are particularly large (more than -50% as compared to 2020 levels) for livestock production. Decreases are very moderate for other crops except for OSA, while for soy, and in particular soy in Brazil, the value of production decrease compared to 2020 levels. Ambitious conservation in Brazil Enforcing zero conversion of forest and other natural lands to agriculture in Brazil from 2020 (as assumed in the ZNLBra scenario) only leads to a slightly lower post-2020 growth in the production and trade of livestock and soy products compared to the SSP2, and therefore on agricultural and land use trends in the rest of the world. The halting of land conversion leads to a drop in LUC emissions, and reduced biodiversity impacts on both aquatic and terrestrial ecosystems compared to the SSP2. However, biodiversity impacts remain higher than in the IAP scenario, which also considers land restoration and supply and demand-side measures. On the other hand, the value of production is projected to not be significantly affected in the ZNL scenario, in contrast to the stark declines projected in the IAP scenario, primarily in the livestock sector, but also in the soy sector. This suggests that Brazil could keep supplying domestic and world markets without clearing forest and other natural lands, thereby pursuing both environmental and economic goals, in particular through mobilizing pasture for soy production. However, the same pastureland is also projected to be restored in the scenario pursuing ambitious biodiversity goals, with clear benefits in terms of climate mitigation. In such a scenario, the fate of Brazilian pastures therefore appears at the center of potential conflicts between economic and environmental goals, while economic opportunities loss would emerge from a shift in global consumption patterns, with potentially large implications for livestock producers. There might be a room for scenarios with more moderate changes in consumption patterns in Brazil, and a mix of restoration and sustainable intensification of the livestock sector (Cohn et al., 2014; De Oliveira Silva et al., 2018). Long-term impacts of the US-China trade dispute 73 As compared to other alternative scenarios, the long-term impacts the US-China trade dispute might be limited. In this scenario we assume the trade dispute to result in a long-term decrease in US soy exports to China, capped at maximum 75% of their 2020 levels. This scenario leads to some increase in soy production and trade in Brazil compared to the 2050 SSP2, a pattern very similar to the short-term impacts observed in response to the trade shocks in the recent years. However, we projected this scenario to have no impact on deforestation, with an only slightly higher increase soy area (mostly at the expense of pasture in the Cerrado biome) over 20202050 as compared to the SSP2 scenario. We found no significant impact of this scenario on Brazil-level projected biodiversity impacts on aquatic and terrestrial ecosystems as compared to the SSP2 scenario. Interestingly, we found the shortfall in USA soya exports to China to be redistributed to other export destination only to a very limited extent, and the Brazil exports to other destinations than China to be very little affected. Although larger efforts from the USA to develop alternative soya export markets could be assumed, this reflects that overall, the international market for soybased products to become very competitive and although slightly increasing its surplus, the USA is projected to become a less important player. In such a context, it will be difficult for soy producers in the USA to recover from short-term drops in export opportunities, which might entail long-term risks. It should be acknowledged that there is considerable uncertainty in the long-term impacts from the US-China trade dispute, and a broader range of assumptions might be worth testing. It should also be noted that our results reflect two different aspects of the baseline scenario: first, soy exports from Brazil, in particular to China, are projected to already massively increase in the baseline scenario, while exports from the USA to China are projected to undergo limited increases. This makes both the assumed China soya import shortfall, and the potential additional export opportunities for Brazil soya bean to China, limited by 2050 in relative terms. Second, this assumes that that although only partially limiting deforestation, domestic interventions in Brazil such as the Amazon Soy Moratorium and the Forest Code are in place. Should this not be the case, the soy production and export development in Brazil lead to higher rates of deforestation in scenarios with high demand for soya exports. Summary. Our scenarios of alternative futures for soy markets provide two important insights. First, a long-term drop in US exports of soy-based products would have limited impacts outside of the USA. Assuming a permanent drop in US soy exports to China below 75% of 2020 levels is projected to have limited impacts on both global markets and Brazilian ecosystems. While most of the China import gap is covered by Brazil, the gap is represents a small addition to the increase in exports from Brazil to China projected between 2020 and 2050 in the SSP2 scenario (while that of US are projected to stabilize), and it is generated in Brazil by a slightly higher conversion of pastures without significant additional deforestation or biodiversity loss. Second, future economic opportunities for the Brazilian soy sector seem compatible with ambitious protection of ecosystems in Brazil, but at odd with a higher share of plant-based products in diets globally. The projected increases in the value and volume of production and exports for the soy sector in Brazil are not negatively affected if assuming conservation efforts in Brazil that are well beyond current ambition levels (i.e., no conversion of any natural ecosystem, be it forested or not, as pictured in the ZNL scenario), while under 80 chain greening in main soy importers like China. On the other hand, the EUDR could fail at elevating global soy supply chain greening ambitions and trigger a weakening of land use regulations in Brazil: we project such a scenario to generate contrasted land use trajectories in Brazil, including a redirection agricultural land pressure from the Cerrado to the Amazon biome, with moderately higher accumulated deforestation at country level, and uncertain but potentially negative biodiversity impacts (especially if phenomenon like land grabbing, underestimated by our model, would be exacerbated). These results reflect the fact that EU market shares in Brazil will continue declining, and the overall impact of EU interventions will primarily depend on how other key actors (like Brazil and China). Second, even if more effectively curbing EU imports of soy-based products from Brazil, alternative EU interventions may not necessarily achieve better biodiversity impact reductions in Brazil, and could have contrasted environmental and economic impacts. Our scenarios include substituting the EUDR with either an EU biodiversity border adjustment mechanism, or ambitious EU efforts to reduce waste and partially substitute animal-based products by plant-based products in human diets. Although all three interventions are mentioned as potential levers to reduce the environmental impacts from EU consumption, it should be noted that these interventions are not directly comparable in terms of goals and ambitions. We find that both the EU biodiversity border adjustment mechanism scenario and the EU demand-side efforts could lead to significant reductions (>30%) in the EU imports of soy from Brazil, and that even in such a case, soy leakage to other regions (e.g., reallocation foregone Brazil exports to other destination) to not be significant. The EU demand-side effort scenario would deliver reductions in environmental impacts in EU (moderate) and BRA (small) for both terrestrial and freshwater ecosystems, while the EU border adjustment mechanism scenario would lead to a small decrease in impacts on BRA freshwater ecosystems and, due to increases in EU imports of US soy-based products and global market reallocations (another form of leakage), no change in impacts on BRA terrestrial ecosystems and an increase impacts on NAM freshwater ecosystems. Gains in environmental impacts in Brazil for the EU demandside effort were however small as compared to potential gains from a ‘EUDR + global alignment of soy supply chains on ambitious conservation in Brazil’ scenario (see previous summary box). Impacts on future economic opportunities of soy producers were also contrasted across alternative interventions. On the one hand, the EU demand-side scenario pictures losses in future economic opportunities for soy producers in all major soya production regions (and livestock producers in the EU), although moderate due to the low soy market shares of the EU. On the other hand, the EU biodiversity border adjustment mechanism scenario benefits the value of production of EU and NAM soy producers while limiting losses in value of production for BRA producers, while the EUDR scenario favors BRA producers over other producers (including the EU). Overall, this implies that an effective reduction of biodiversity loss from EU imports of soy-based products might be challenging to achieve, that different approaches (e.g., EUDR, EU biodiversity border adjustment mechanism or EU demand-side measures) could have contrasted impacts on economic opportunities for soy producers of various regions, and that the most effective way to reduce imported biodiversity loss (demand-side measures) might have negative economic impacts for soy producers in all regions. 81 Forest supply chains Results Results for circular bioeconomy scenarios Round wood harvest volumes In the circular bioeconomy scenarios (CIR), harvest volumes are considerably lower than in scenarios without the circular bioeconomy (Figure 18). Hence, circular bioeconomy allows to match increased woody-biomass demand without the need to increase harvest volumes and forest resources use. Figure 18. Global roundwood harvest volumes in Mm3/year In the high demand scenario with plantations and circular bioeconomy (PLA&CIR BIO&CON), plantation forests expansion and harvests remain lower level than in the high demand scenario without circular economy (PLA BIO&CON) (Figure 19). This means that circular bioeconomy is a cost-efficient supply chain solution compared to plantation forests. In the high demand scenario with plantations (PLA BIO&CON), bioeconomy expansion mostly benefits Asia, Africa and Latin America, because they have higher plantation potential than traditional forest industry regions (i.e., EU, North America and FormerSovjetUnion). In the high demand scenarios with circular economy (CIR BIO&CON, PLA&CIR BIO&CON), we do not observe this effect. Hence, circular bioeconomy benefits relatively more traditional forest industry regions. 3100 3600 4100 4600 5100 5600 6100 6600 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 BASE BIO CON BIO&CON PLA BASE PLA_BIO PLA_CON PLA_BIO&CON CIR BASE CIR BIO CIR CON CIR BIO&CON PLA&CIR BIO&CON 82 0 1000 2000 3000 4000 5000 6000 7000 8000 20002010202020302040205020602070208020902100 CIR BIO&CON 0 1000 2000 3000 4000 5000 6000 7000 CIR BIO&CON 0 1000 2000 3000 4000 5000 6000 7000 8000 20002010202020302040205020602070208020902100 PLA&CIR BIO&CON 0 1000 2000 3000 4000 5000 6000 7000 PLA&CIR BIO&CON 83 Figure 19. Global roundwood harvest volumes by region and forest type (Mm3/yr) Wood-based products net exports Net exports of traditional forest industry regions (i.e., EU, North America and FormerSovjetUnion) are higher in the circular bioeconomy scenarios than in other scenarios (Figure 20). North America benefits less from the circular bioeconomy than other traditional forest industry regions, because it relies more on virgin fiber production than on recycling. The circular economy also benefits Asia and Africa by decreasing their net exports, but Latin America tends to lose its competitive advantage, which is based on plantations expansion. 0 1000 2000 3000 4000 5000 6000 7000 8000 20002010202020302040205020602070208020902100 PLA BIO&CON EU27 ASIA Africa FormerSovietUnion LatinAmericaRest NorthAmerica Brazil 0 1000 2000 3000 4000 5000 6000 7000 PLA BIO&CON Harvest natural/seminatural forests Harvest plantation forests -50 0 50 100 150 200 250 2020 2030 2040 2050 2060 2070 2080 2090 2100 EU27 0 50 100 150 200 250 300 350 400 450 500 2020 2030 2040 2050 2060 2070 2080 2090 2100 North America 84 Figure 20. Wood-based products net exports (Mm3/yr RWeq) Forest carbon balance The plantation forests and circular bioeconomy scenarios (PLA BIO&CON, CIR BIO&CON, PLA&CIR BIO&CON) lead to higher forest carbon stock and sink than scenarios based on natural 0 20 40 60 80 100 120 140 160 2020 2030 2040 2050 2060 2070 2080 2090 2100 FormerSovietUnion 0 100 200 300 400 500 600 2020 2030 2040 2050 2060 2070 2080 2090 2100 Latin America Rest -600 -500 -400 -300 -200 -100 0 2020 2030 2040 2050 2060 2070 2080 2090 2100 Africa -1200 -1000 -800 -600 -400 -200 0 2020 2030 2040 2050 2060 2070 2080 2090 2100 Asia 0 100 200 300 400 500 600 700 800 900 2020 2030 2040 2050 2060 2070 2080 2090 2100 Brazil BASE BIO CON BIO&CON PLA PLA_BIO PLA_CON PLA_BIO&CON CIR CIR BIO CIR CON CIR BIO&CON PLA&CIR BIO&CON 85 forests logging and linear supply chain solutions (BIO&CON). This is true regardless of whether we consider biogenic carbon in forests (Figure 21) or also include biogenic carbon outside forests (Figure 22). Circular bioeconomy leads to higher biogenic carbon storage than plantation forests, which is caused by lower overall harvests in the circular economy. The highest biogenic carbon storage is achieved in the scenario which combines circular bioeconomy and plantation forests (PLA&CIR BIO&CON). Figure 21. Forest carbon storage (PgC) and sink (MtCO2/yr) including only the biogenic carbon that stays in forests (F) Figure 22. Forest carbon storage (PgC) and sink (MtCO2/yr) including biogenic carbon that stays in forests (F) and biogenic carbon that is stored outside forests (HWP and BECCS) 290 300 310 320 330 340 350 360 370 2000 2020 2040 2060 2080 2100 F carbon storage -6000 -5000 -4000 -3000 -2000 -1000 0 2020 2040 2060 2080 2100 F carbon sink BIO&CON PLA_BIO& CON CIR BIO&CON PLA&CIR BIO&CON 290 310 330 350 370 390 410 430 2000 2020 2040 2060 2080 2100 F+HWP+BECCS carbon storage -7000 -6000 -5000 -4000 -3000 -2000 -1000 0 2020 2040 2060 2080 2100 F+HWP+BECCS carbon sink BIO&CON PLA_BIO& CON CIR BIO&CON PLA&CIR BIO&CON 86 Biodiversity Biodiversity loss is measured as in D6.2 by the Potential Disappeared Fraction of global species aggregated over all different taxa (PDF %). Biodiversity loss is lower in the plantation forests and circular bioeconomy scenarios than in scenarios based on natural forests logging and linear supply chain solutions scenarios (Figure 23). Circular bioeconomy leads to lower biodiversity loss than plantation forests, which is caused by lower overall harvests in the circular economy. Lower harvest volumes decrease the impact of natural forest and plantations production use on biodiversity loss (Figure 24). Figure 23. Global biodiversity loss in PDF% Figure 24. Global biodiversity loss in PDF% divided into different land-uses 14 15 16 17 18 19 20 21 22 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 BASE BIO CON BIO&CON PLA PLA_BIO PLA_CON PLA_BIO&CON CIR CIR BIO CIR CON CIR BIO&CON PLA&CIR BIO&CON 0 2 4 6 8 10 12 14 16 18 20 22 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 CIR BIO&CON 0 2 4 6 8 10 12 14 16 18 20 22 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 PLA BIO&CON Plantations Natural forests Grassland Cropland 87 Results for EU Biodiversity Strategy scenarios Round wood harvest volumes The EUBDS decreases EU27 harvest volumes by up to 34 Mm3/yr in 2100 (or -6%) compared to the business-as-usual scenario (Figure 25). Figure 25. EU27 roundwood harvest volumes in Mm3/year Wood-based products net exports The EUBDS also leads to decrease in EU27 net exports, which are 39 Mm3/yr below the baseline level in 2100 (corresponding to a 35% decline from BASE levels) (Figure 26). Figure 26. EU 27 net exports (Mm3/year RWeq) 480 500 520 540 560 580 600 2020 2030 2040 2050 2060 2070 2080 2090 2100 BASE EUBDS 0 20 40 60 80 100 120 140 2020 2030 2040 2050 2060 2070 2080 2090 2100 BASE EUBDS 88 Forest carbon balance The EUBDS increases EU27 forest carbon storage by 0.4 petagrams of carbon (PgC), but HWP and BECCS storage decreases by 0.1 PgC and reverse substitution storage by 0.25 PgC (Figure 27). Therefore, the overall impact of the EUBDS on EU27 total forest sector carbon balance is marginal, leading to an increase of 0.05 PgC compared to the business-as-usual scenario. Figure 27. EU27 forest sector carbon balance Biodiversity The EUBDS decreases EU27 total biodiversity loss, but this effect is small compared to the biodiversity benefits of cropland abandonment and afforestation (Figure 28). Moreover, the EUBDS is not able to stop the increase in biodiversity loss over time in natural forests, which is mainly caused by increased harvests and management intensification in the unprotected forest area. Consequently, the EUBDS would require additional measures such as restoration to stop biodiversity loss in natural forests. 10 11 12 13 14 15 16 2020 2040 2060 2080 2100 Forest carbon storage (PgC) BASE EUBDS 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 2020 2040 2060 2080 2100 HWP+BECCS carbon storage (PgC) BASE EUBDS -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 2020 2040 2060 2080 2100 Reverse substitution (PgC) BASE EUBDS 12 13 14 15 16 17 18 19 20 2020 2040 2060 2080 2100 Total carbon storage (PgC) BASE EUBDS 89 Figure 28. EU27 biodiversity loss in PDF% Results for Brazil restoration scenarios Land use change In the restoration scenarios (i.e., NAT RESTORE, MIX RESTORE, and PLA RESTORE), the restored area increases by up to 124 Mha in 2100 (Figure 29). Restoration leads to a corresponding decrease in cropland, grassland and abandoned land. In the non-restoration scenario (NO RESTORE), restored area remains at 39 Mha in 2100. Deforestation is higher in the nonrestoration scenario as there is no deforestation control (carbon price=0), which leads to an additional increase in cropland and grassland. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 2000 2020 2040 2060 2080 2100 Biodiversity loss BASE 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 2000 2020 2040 2060 2080 2100 Biodiversity loss EUBDS CropLand GrassLand NaturalForest Plantations 0.36 0.37 0.38 0.39 0.40 0.41 0.42 0.43 2020 2030 2040 2050 2060 2070 2080 2090 2100 Biodiversity loss total BASE EUBDS 96 30% protection target does not require much additional protection efforts, 2) the EU27 forest sector can easily adapt to the reduced domestic harvest potential by decreasing its net exports, replacing woody biomass use for energy by energy crops and intensifying forest management in the non-protected areas, and 3) the positive impact of the EUBDS on forest carbon stock is offset by reduced HWP carbon pool and reverse substitution effects. Moreover, the EUBDS is not able to stop the increase in biodiversity loss in natural forests, which calls for complementary measures such as restoration. Brazil land restoration policy Four scenarios have been designed to analyze the potential impacts of different types of land restoration policy in Brazil, namely: a) a baseline scenario considering the restoration of degraded agricultural land and abandoned land to natural forests, which are managed for carbon storage and biodiversity (i.e., NAT RESTORE), b) a counterfactual scenario considering no land restoration policy in Brazil (i.e., NO RESTORE), c) a scenario considering the restoration of degraded agricultural land and abandoned land to plantation forests, which are managed for roundwood production (i.e., PLA RESTORE), and d) a scenario considering the restoration of degraded agricultural land and abandoned land to plantation forests, which are managed for both roundwood production and carbon storage and biodiversity (i.e., MIX RESTORE). Our results suggest that restoration policy - independently of the restoration type - is always better than no restoration policy from a carbon storage and biodiversity perspective. From a biodiversity perspective, restoration based on natural forests is clearly a better option than restoration based on multifunctional or production plantations. On the other hand, from a carbon storage perspective, there is not much difference between different restoration policies. The reason is that the lower forest carbon storage in plantations scenarios is compensated by higher carbon storage outside forests (i.e., HWP and BECCS). Aquaculture & aquafeed supply chains Results Demand for final blue food products Projections for the final demand of blue food products are presented in Figure 34, with different time horizons for the SSP2 scenario (2010, 2020, 2030, 2040 and 2050), and 2050 values for other scenarios. In the SSP2 scenario, the global demand for fish final products remains relatively stable or slightly increases (i.e., salmonoids SALM, less than +15%) between 2020 and 2050 for most products, with a few exceptions. On the one hand, some products are projected to undergo a more significant increase in global demand, for example freshwater (FRSH, +51%) and shrimps & prawns (SHRI, +37%), even though their demand increase at lower speed than in recent decades (e.g., +51% over 2020-2050 vs +41% over 2010-2020). This is compatible with projections from other studies (e.g. (Naylor et al., 2021)), with the level of wild catch expected to stabilize. Countries from Eastern Asia (EAS, including China) are expected to remain the 97 largest consumers of freshwater and crustacean products and undergo moderate demand growth, while countries from South (SAS, e.g., India), South-East Asia (SEA) or Sub-Saharan Africa are projected to experience a large relative increase by 2050. For some wild catch-based fish products (e.g., pelagic fishes PELG, demersal fishes DMRS, marine fishes MARN, tunas TUNA), the globally stable demand hides more contrasted changes at the regional level, with demand often increasing in Sub-Saharan Africa (SSA) and decreasing in Eastern Asia (EAS). On the other hand, demand for fish meal and fish oil are expected to decrease substantially over the 20202050 period, by about 90% and 75%, respectively. This follows the assumed prolongation of recent trends in fed aquaculture in terms of feed conversion efficiency gains and substitution of fish meal and fish oil feed by crop-based feed. Most of the trends in final products are related to food use and to some extent other uses, except for fish meal and fish oil, that are used for feed. In other scenarios, the trends follow the SSP2 scenario, with a few exceptions: • First, in the BFS20 counterfactual scenario, demand remains by design constant at 2020 levels after 2020. • Second, in the UNFEDAC20 sensitivity scenario, the capacity of unfed aquaculture systems is assumed to remain constant at 2020 levels after 2020, leading to lower demand by 2050 for some products (freshwater fishes FRSH, crustaceans CRST, shrimps and prawns SHRI). The difference to the SSP2 demand by 2050 for these products reflects the degree to which they rely on unfed aquaculture systems: high for CRST (2050 demand is close to 2020 levels and well below SSP2 levels), moderate for FRSH and SHRI (2050 demand is close to SSP2 levels and well above 2020 levels). • Moreover, as opposed to the SSP2 scenario, the demand for fish meal (FSHM) and fish oil (FSHO) increases to higher levels than 2020 for the AF20 and AFCOMPO20 sensitivity scenarios. This results from the assumed constant share of fish meal and fish oil in aquafeed requirements per unit of output (fixed at 2020 levels, for both scenarios), and constant total aquafeed requirements per unit of output (fixed at 2020 levels, for the AF20 scenario). This highlights the importance of the expected prolongation of recent trends in aquaculture feeding practices (in terms of efficiency and aquafeed composition) in shaping future demands for fish-based aquafeed, in the context of growing demand for aquaculture products. • In the BFDIET scenario, the demand for mollusks-based products (MLSC) is close to about twice as large by 2050, while the demand for freshwater-based products (FRSH) is about 20% lower. These trends reflect scenario assumptions and represent the largest deviation from the BAU scenario across all scenarios (except for BF20), thereby highlighting the importance of dietary assumptions. • In the SUSFISH scenario, the demand for several products that are primarily supplied through fisheries (CEPH, TUNA, MARN, DMRS, MARN) increase by about 15%. These trends reflect scenario assumptions and represent the largest deviation from the BAU scenario across all scenarios. This implies that a larger supply of these products could easily be met by demand, and highlights the importance of sustainable fisheries management to avoid overfishing. 98 Figure 34 - Projections of demand of blue food final products for food, feed and other use (i.e., excluding input to reduction sector) by product, aggregated to ten world regions. Primary blue food product supply and aquafeed use Global-scale projections of primary fish products supply by source (catch, fed and unfed aquaculture) and crop aquafeed requirements (by crop) are presented in Figure 35. By assumption, most of the supply increase is projected to be sourced from fed aquaculture. In the SSP2 scenario, the total supply of primary blue food products is expected to continue increasing, but at a slower pace than in recent past. It increases from 169 million tons (Mt) in 2020 to 210 Mt by 2050 (+24%, roughly equal to the relative increase from 2010 to 2020). This translates into a 57% increase in fed aquaculture supply (from 50 Mt in 2020 to 78 Mt in 2050) and a 45% increase in unfed aquaculture (from 29 Mt in 2020 to 41 Mt in 2050). These trends remain below (for fed aquaculture) or nearing (for unfed aquaculture) 2010-2020 rates of increase (+87% for fed aquaculture, +18% for unfed aquaculture). These projections are qualitatively comparable to those from the OECD-FAO Outlook 2024-2033 (OECD/FAO, 2024). Projections of primary fish supply by 2050 for other scenarios are similar to the ones of the SSP2 scenario, except for four scenarios. For the BFS20 counterfactual scenario, supply remains, by assumption, constant at 2020 levels for all three sources. For the UNFEDAC20 sensitivity scenario, unfed aquaculture supply remains by assumption constant at 2020 levels, leading to a slightly higher increases in fed aquaculture (+59% over 2020-2050, instead of +57% in the SSP2 scenario), reflecting the limited substitution options for CRST and SHRI product groups, for which the demand is projected to be lower than in the SSP2 scenario (see Figure 34). For the BFDIET 99 scenario, the substitution in consumer preferences lead to a slightly lower (resp. higher) increase in fed (resp. unfed) aquaculture supply over the 2020-2050 period than in the baseline, and a lowest increase in fed aquaculture across all scenarios (except BF20). For the SUSFISH scenario, the assumed increased catch capacity leads to higher increases in catch level, and slightly lower increases in unfed aquaculture, and to some extent, fed aquaculture. As illustrated in Figure 35b, we project an increase in crop-based aquafeed requirements in the SSP2 scenario over 2020-2050, from 79 Mt in 2020 to about 100 Mt in 2050. This represents a 40% increase over this period, which is lower than the growth in aquaculture supply (+57%). This indicates that the assumed increase in feed conversion efficiency is projected to buffer the future increase in demand for aquafeed, despite the assumed further substitution of fishmeal and fish oil by crop-based aquafeed. The AF20 and AFCOMPO20 scenarios allow disentangling these two factors: the increase in total crop aquafeed requirements would more closely follow aquaculture supply increase if both the composition and the efficiency of feeding practices remained constant at 2020 levels (AF20 scenario, +53%), but would be slightly lower than in the SSP2 if the share of crops in aquafeed did not increase beyond 2020 levels but efficiency gains still occur (AFCOMPO20 scenario, +35%). It should also be noted that between 2020 and 2050, the projected increase in crop-aquafeed is of a lower amplitude than the projected decrease in fish meal and fish oil. The main reason for this is that crop aquafeed already represents a high share of total fed aquaculture requirements by 2020, and the impact of additional substitution is less strong than future trends in demand for aquaculture products and aquaculture feed conversion efficiencies. Except for the BF20 scenario, total crop feed requirements are lowest in the BFDIET scenario by 2050, and only slightly lower than in the BAU scenario for the SUSFISH scenario. This confirms the importance of dietary choices for the overall development of fed aquaculture, and a small potential for increased catch levels to moderate the future growth in aquaculture. By assumption, the proportion of various crops in total crop-based aquafeed differs across regions but remains constant over the different time horizons and scenarios, except for the AFCOMPOCROPMIX scenario. In this simple sensitivity scenario, the assumed substitution of 50% of corn and soya aquafeed requirements in China by wheat leads to a markedly lower increase in feed requirements (+28%, instead of +40% in the same period), primarily due to the higher crude protein content of wheat compared to corn. While this scenario was designed as a simplistic illustration of aquafeed crop mix change rather than as a realistic future evolution, it illustrates that related assumptions can be as important as that of other feeding practices aspects such as feed conversion efficiencies and the share of fish meal and fish oil. 100 Figure 35 - Projections at global scale of a) total blue food primary products supply by source and b) crop-based aquafeed used by crop. Land use and biodiversity impacts from crop aquafeed requirements As shown in Figure 36a, the crops used as aquafeed represent a small share of global crop production projected by 2050: up to 3% for wheat, 5% for corn and 6% for soya. This share is projected to remain relatively stable between 2020 and 2050 for most crops and across the different scenarios, except for the AFCROPOCROPMIX scenario (higher shares for wheat, and lower shares for corn and soya), the counterfactual BFS20 scenario (decrease for all crops, by design) and the BFDIET scenario (decrease for all crops, related to a moderation of fed aquaculture growth). This means that the demand for crop aquafeed products is not projected to increase faster than the demand for other uses of crops, and might even increase more moderately than other demands under scenarios with a higher share of unfed aquaculturebased products in diets. The AFCOMPOCROPMIX scenario is a simplified experiment picturing changes in the relative contribution of various crops to total crop-based aquafeed, only for one country (China). And yet, at the individual crop level, it differs markedly from other scenarios by 2050, which highlights the importance that crop aquafeed composition assumptions have on land use impacts from aquaculture development. As displayed in Figure 36b, the overall demand for agricultural products drives an expansion of agricultural land covers (cropland and grassland) from 2020 to 2050 of about 300 million hectares, at the expense of forest and – in larger proportion – other natural vegetation. Differences across scenarios, including for the counterfactual BFS20 scenario, are barely noticeable in Figure 36b, indicating that the impacts of further aquafeed demand play a very marginal role in projected land use expansion globally between 2020 and 2050. 101 Altogether, results displayed in Figure 36 highlight that the projected future crop aquafeed requirements will occur concomitantly to much larger changes to the agricultural and land use systems. While future increases in crop aquafeed demand may lead to land use-mediated terrestrial biodiversity loss, it indicates that these may only represent a small share of future terrestrial biodiversity losses from the agricultural sector. It also points to the necessity of using dynamic modelling tools to understand such impacts Indeed, the impact of additional demand may be very different from those inferred from the current state of the agriculture and land use sectors, due to expected concomitant changes in the land use, productivity, production, demand and trade across regions. Using the GLOBIOM model enables to providing an estimate of future dynamics in agriculture and land uses systems, against which the specific impacts of increased crop aquafeed requirements can be estimated. In addition, by comparing production, land use or biodiversity impacts from various scenarios to a counterfactual scenario in which all components of the blue food sector remain fixed at 2020 levels (BFS20 scenario), we can diagnose precisely the impact of post-2020 changes in crop aquafeed requirements, on top of these broader trends in the agricultural and land use systems. Figure 36 - Projections at global scale of a) the share crop aquafeed represents in total crop supply (%) and b) the net difference to 2000 in the extent of various land covers. Figure 37 provides the difference between the counterfactual scenario BFS20 and other scenarios by 2050, for crop aquafeed requirements, changes to the natural land cover, and terrestrial biodiversity impacts from land occupation. When it comes to crop aquafeed requirements (left panel), Figure 37 confirms the differences across scenarios in total crop aquafeed requirements shown in Figure 35b, but also highlights that additional crop aquafeed demand occurs in EAS (including China), SAS and SEA regions, where additional fed aquaculture production takes place. However, as suggested by the middle panel, due to the trade dependencies of these regions, land use intensification and other market knock-on effects, Latin America and the Caribbean (LAC) appear as a hotspot of natural land loss associated with crop aquafeed production, in particular for soya and corn, to a lower extent. 102 Impacts in EAS and SAS remain high but represent a lower proportion of global impacts for natural land loss than for increased crop aquafeed demand, while to some extent NAM, MEN and EUR show the opposite trends. The spatial patterns of global natural land loss from crop aquafeed requirements are relatively similar across scenarios, except for the AFCOMPOCROPMIX and BFDIET scenarios, for which natural land losses are much lower (or even reversed, for SSA and SEA) in EAS where the reduced total crop aquafeed requirement occurs, but also in LAC where export demand for soya and corn, to some extent – is reduced, and SSA (where no additional aquafeed is required, but natural land losses to crop production for other purposes might be avoided via increased imports from other regions) and SEA (where aquafeed are produced). These trade-mediated effects, and the fact that the scenario that minimizes aquafeed crop requirements and related natural land outcomes differs across regions (e.g., AFCOMPOCROPMIX for LAC, BFDIET for SSA and SEA) highlights the complexities of international crop supply chains. The biodiversity impacts (right panel), however, illustrates that the translation of losses in natural land to terrestrial biodiversity impacts (in terms of global extinction risks) from land occupation is not direct, mainly due to variations at national to subnational scales in species richness and level of endemism, its sensitivity to various land uses, and in land use change patterns. For example, while the AFCOMPOCROPMIX scenario leads in LAC to comparable savings (as compared to the SSP2 scenario) in terms biodiversity loss and natural land loss, in the same region savings in the AFCOMPO20 scenario (as compared to the SSP2 scenario) are much higher for biodiversity than for natural land loss. While we found LAC to be a clear hotspot of biodiversity loss from future aquaculture development due to its export-oriented agricultural sector, actual impacts on biodiversity may be particularly sensitive to how additional crop production is achieved. In addition, the share of SAS and SEA regions combined in global additional losses of terrestrial biodiversity (as compared to the BFS20 scenario) is often close to 50% or even higher for the SSP2, AFCOMPO20, AF20 and UNFEDAC20 scenarios. Yet, for these scenarios, the share of these two regions combined in global additional natural land loss (as compared to the BFS20 scenario) is often less than 33%. Similarly, the relatively small natural land gains in SEA in the BFDIET scenario lead to comparatively large reductions in biodiversity impacts. This reinforces the importance of these regions as hotspots of biodiversity loss from future aquaculture development. Not only would the local demand for crop aquafeed increase in these regions (unless demand for freshwater products is moderated by diet shifts towards unfed aquaculture-based products), but the local ecosystems are particularly important for biodiversity (as measured in terms of global extinction risks). 103 Figure 37 - Net difference between the SSP2 and AF20 scenarios after 2020, in crop aquafeed use (left panel, million tons), extent of natural land cover (middle panel, million hectares) and total extinction risks impact on terrestrial ecosystems (right panel, potentially disappeared fraction). The stacked bars of different colours indicate the different world regions aggregated from original GLOBIOM regions. Nitrogen losses from finfish aquaculture Projected aquaculture on-farm nitrogen waste from the fed aquaculture of finfish species is presented in Figure 38. The projected increase in fed aquaculture production is expected to generate additional reactive nitrogen losses, primarily for freshwater species. In the SSP2 scenario, total on-site N waste in finfish aquaculture is expected to grow from 0.23 million tons of nitrogen (MtN) in 2020, to about 0.31 MtN by 2050 for the SSP2 scenario (i.e., +36% over 2020-2050). Similarly to total crop aquafeed requirements displayed in Figure 35b, values projected for the various sensitivity scenarios show that assuming constant feeding practices (in terms of efficiency and aquafeed composition) at 2020 levels would lead to significantly higher losses (+58% over 2020-2050 for the AF20 scenario), while changes to the proportion of various crops in total crop aquafeed requirements could also impact nitrogen waste (e.g., +22% over 2020-2050). Changes in consumer preferences towards unfed aquaculture-based products, in regions of high consumption of freshwater species-based products, would be the most impactful measure to mitigate future increases in aquaculture on-farm nitrogen waste. 104 Figure 38 - Projected nitrogen losses in finfish aquaculture at global scale. 105 Discussion The projections presented in this deliverable explore various factors that might affect the potential developments of the blue food sector, and their impacts on terrestrial biodiversity through increases in the use of crops as an input to aquaculture. The projections rely on a baseline scenario depicting a prolongation of historical trends, as well as additional scenarios designed to isolate the impact of various assumptions (sensitivity scenarios) or to provide a counterfactual in which the blue food sector remains constant at 2020 level while the land use and agricultural sector follow baseline trends. One important driver will be the projected future increase in the demand for blue food products. Our BAU projections for population and dietary preferences projects a slowed but prolonged increase in blue food products’ demand. The expected increase varies across products, partly driven by expected limitations in supply In particular, demand is projected to remain stable at the global scale for products groups based on wild catch (despite increases and decreases at regional scale) but to increase significantly for products based on aquaculture production systems (up to +51% over 2020-2050 for freshwater fish). The projected increase in demand is particularly important in Eastern Asia (which currently dominates freshwater fish consumption) and in regions with significant future increases in population, like South Asia and Sub-Saharan Africa. The demand for fish meal and fish oil is expected to continue decreasing, unless historical trends in aquaculture feeding practices (in terms of overall feeding efficiency and aquafeed composition) do not continue in the future. These trends are consistent with other studies (e.g., (Naylor et al., 2021; OECD/FAO, 2024)). As compared to the results of CLEVER Deliverable D7.2, the two additional scenarios considered in this deliverable (i.e., the BFDIET and SUSFISH scenarios) provide two key additional insights. First, any increase in catch levels is likely to be met with demand, and sustainable management of fisheries would allow to meet higher consumption levels with lower pressures on ocean ecosystems. Second, a transition in consumer preferences towards unfed aquaculture-based products of high nutritional profile (e.g., mollusks) might provide both food security and avoid environmental impacts such as crop aquafeed related biodiversity loss. Our BFDIET scenario, combining such an assumption with the future technological progress in feeding practices already considered in the baseline scenario, is the scenario with the least increases in fed aquaculture, and related environmental impacts. Another important determinant will be trends in the capacity of various sources and feeding practices in aquaculture. Reflecting historical trends, our baseline projections assume a stable capacity from wild catch, and a significant increase from unfed (+45% over 2020-2050) and fed (+57% over 2020-2050) aquaculture, although slower than in recent decades. Our baseline scenario also assumes further increases in efficiency (i.e., decreased in economic feed conversion ratio) and change in aquafeed composition (further replacement of fish meal and fish oil by crop-based aquafeeds) within fed aquaculture, as well as further increases in the use of fish waste in the reduction of fish into fish meal and fish oil. This leads to decreasing demand for fish meal and fish oil, a decrease in the fish reduction-related pressure on marine and pelagic species, and to moderate increases in crop aquafeed requirements compared to projected increases in feed aquaculture supply (e.g., respectively +40% and +57% over 2020-2050). As compared to results from the CLEVER Deliverable D7.2, the additional scenarios considered in this deliverable leads to one additional insight. Assuming that all fisheries are sustainably 112 Naylor, R. L., Kishore, A., Sumaila, U. R., Issifu, I., Hunter, B. P., Belton, B., Bush, S. R., Cao, L., Gelcich, S., Gephart, J. A., Golden, C. D., Jonell, M., Koehn, J. Z., Little, D. C., Thilsted, S. H., Tigchelaar, M., & Crona, B. (2021). Blue food demand across geographic and temporal scales. Nature Communications, 12(1), 5413. https://doi.org/10.1038/s41467-021-25516-4 OECD/FAO. (2024, July 2). OECD-FAO Agricultural Outlook 2024-2033. https://www.oecd.org/en/publications/oecd-fao-agricultural-outlook-20242033_4c5d2cfb-en/full-report.html Official Journal of the European Union. (2023). Regulation (EU) 2023/ of the European Parliament and of the Council of 31 May 2023 on the making available on the Union market and the export from the Union of certain commodities and products associated with deforestation and forest degradation and repealing Regulation (EU) No 995/2010. https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32023R1115 Oğuz, S. (2025). Charted: The Economic Value of Nature vs. Global GDP. https://www.visualcapitalist.com/sp/charted-the-economic-value-of-nature-vs-globalgdp/ Oliveira, U., & Pacheco, A. (2024). Deliverable 2.4: Biodiversity impact database including characterization factors and documentation ready for Module C. Zenodo. https://doi.org/10.5281/zenodo.13640620 Oliveira, U., Soares-Filho, B. S., Santos, A. J., Paglia, A. P., Brescovit, A. D., de Carvalho, C. J. B., Silva, D. P., Rezende, D. T., Leite, F. S. F., Batista, J. A. N., Barbosa, J. P. P. P., Stehmann, J. R., Ascher, J. S., Vasconcelos, M. F., Marco, P. D., Löwenberg-Neto, P., & Ferro, V. G. (2019). Modelling Highly Biodiverse Areas in Brazil. Scientific Reports, 9(1), 6355. https://doi.org/10.1038/s41598-019-42881-9 Pendrill, F., Persson, U. M., Godar, J., & Kastner, T. (2019). Deforestation displaced: Trade in forest-risk commodities and the prospects for a global forest transition. Environmental Research Letters, 14(5), 055003. https://doi.org/10.1088/1748-9326/ab0d41 Pendrill, F., Persson, U. M., Godar, J., Kastner, T., Moran, D., Schmidt, S., & Wood, R. (2019). Agricultural and forestry trade drives large share of tropical deforestation emissions. Global Environmental Change, 56, 1–10. https://doi.org/10.1016/j.gloenvcha.2019.03.002 Popp, A., Calvin, K., Fujimori, S., Havlik, P., Humpenöder, F., Stehfest, E., Bodirsky, B. L., Dietrich, J. P., Doelmann, J. C., Gusti, M., Hasegawa, T., Kyle, P., Obersteiner, M., Tabeau, A., Takahashi, K., Valin, H., Waldhoff, S., Weindl, I., Wise, M., … Vuuren, D. P. V. (2017). Land-use futures in the shared socio-economic pathways. Global Environmental Change, 42, 331–345. https://doi.org/10.1016/j.gloenvcha.2016.10.002 Riviere, M., Caurla, S., & Delacote, P. (2020). Evolving Integrated Models From Narrower Economic Tools: The Example of Forest Sector Models. Environmental Modeling & Assessment, 25(4), 453–469. https://doi.org/10.1007/s10666-020-09706-w Roberts, S., Jacquet, J., Majluf, P., & Hayek, M. N. (2024). Feeding global aquaculture. Science Advances, 10(42), eadn9698. https://doi.org/10.1126/sciadv.adn9698 Sabatini, F. M., Keeton, W. S., Lindner, M., Svoboda, M., Verkerk, P. J., Bauhus, J., Bruelheide, H., Burrascano, S., Debaive, N., Duarte, I., Garbarino, M., Grigoriadis, N., Lombardi, F., Mikoláš, M., Meyer, P., Motta, R., Mozgeris, G., Nunes, L., Ódor, P., … Kuemmerle, T. (2020). Protection gaps and restoration opportunities for primary forests in Europe. Diversity and Distributions, 26(12), 1646–1662. https://doi.org/10.1111/ddi.13158 113 Schilling-Vacaflor, A., & Lenschow, A. (n.d.). Bringing the State back in: Exploring new public environmental policy approaches for governing the Brazil-Europe soy chain. Schulte, M., Lauri, P., & Di Fulvio, F. (2025). Global forest carbon leakage and substitution effect potentials: The case of the Swedish forest sector. Carbon Balance and Management (in Review). Søndergaard, N., & Sá, C. D. D. (2023). Brazilian Stakeholder assessment of the European Deforestation Regulation. Soterroni, A. C., Mosnier, A., Carvalho, A. X. Y., Câmara, G., Obersteiner, M., Andrade, P. R., Souza, R. C., Brock, R., Pirker, J., Kraxner, F., Havlík, P., Kapos, V., Zu Ermgassen, E. K. H. J., Valin, H., & Ramos, F. M. (2018). Future environmental and agricultural impacts of Brazil’s Forest Code. Environmental Research Letters, 13(7), 074021. https://doi.org/10.1088/1748-9326/aaccbb Sotirov, M., Azevedo-Ramos, C., Rattis, L., & Berning, L. (2022). Policy options to regulate timber and agricultural supply-chains for legality and sustainability: The case of the EU and Brazil. Forest Policy and Economics, 144, 102818. https://doi.org/10.1016/j.forpol.2022.102818 Spillias, S., Blanchard, J. L., Castro-Cadenas, M. D., Cottrell, R. S., Depperman, A., Frank, S., Leclere, D., Tran, N., & Havlik, P. (2025). Integrating Blue Foods into Future Estimates of Land-Use Change (manuscript in preparation). Tacon, A. G. J., & Metian, M. (2015). Feed Matters: Satisfying the Feed Demand of Aquaculture. Reviews in Fisheries Science & Aquaculture, 23(1), 1–10. https://doi.org/10.1080/23308249.2014.987209 Tan, E. C. D., & Lamers, P. (2021). Circular Bioeconomy Concepts—A Perspective. Frontiers in Sustainability, 2. https://doi.org/10.3389/frsus.2021.701509 Tsiropoulos, I., Siskos, P., De Vita, A., Tasios, N., & Capros, P. (2022). Assessing the implications of bioenergy deployment in the EU in deep decarbonization and climate-neutrality context: A scenario-based analysis. Biofuels, Bioproducts and Biorefining, 16(5), 1196– 1213. https://doi.org/10.1002/bbb.2366 Valenti, W. C., & Ballester, E. L. (2024). Integrated Aquaculture and Monoculture of LowTrophic Species. Fishes, 9(11), 450. https://doi.org/10.3390/fishes9110450 Vasconcelos, A., & Hurd, J. (2025, May 15). Brazil and China are talking trade. Here’s how they can deliver on sustainability goals. World Economic Forum. https://www.weforum.org/stories/2025/05/brazil-and-china-can-transform-thesustainability-of-soy-supply-chains/ Verones, F., Hellweg, S., Antón, A., Azevedo, L. B., Chaudhary, A., Cosme, N., Cucurachi, S., de Baan, L., Dong, Y., Fantke, P., Golsteijn, L., Hauschild, M., Heijungs, R., Jolliet, O., Juraske, R., Larsen, H., Laurent, A., Mutel, C. L., Margni, M., … Huijbregts, M. A. J. (2020). LC-IMPACT: A regionalized life cycle damage assessment method. Journal of Industrial Ecology, 24(6), 1201–1219. https://doi.org/10.1111/jiec.13018 World Economic Forum. (2020). Nature Risk Rising: Why the Crisis Engulfing Nature Matters for Business and the Economy. https://www3.weforum.org/docs/WEF_New_Nature_Economy_Report_2020.pdf World Economic Forum. (2025, May 15). How Brazil and China can deliver on sustainability goals. World Economic Forum. https://www.weforum.org/stories/2025/05/brazil-andchina-can-transform-the-sustainability-of-soy-supply-chains/ 114 Xiao, X., Agusti, S., Lin, F., Li, K., Pan, Y., Yu, Y., Zheng, Y., Wu, J., & Duarte, C. M. (2017). Nutrient removal from Chinese coastal waters by large-scale seaweed aquaculture. Scientific Reports, 7(1), 46613. https://doi.org/10.1038/srep46613 Yarlagadda, B., Zhao, X., Iyer, G., Wild, T., Hultman, N., & Lamontagne, J. (2025). Emissions leakage and economic losses may undermine deforestation-linked oil crop import restrictions. Nature Communications, 16(1), 1520. https://doi.org/10.1038/s41467025-56693-1 Ziegert, R. F., & Sotirov, M. (2024). Regulatory politics and hybrid governance: The case of Brazil’s Amazon Soy Moratorium. Global Environmental Change, 88, 102916. https://doi.org/10.1016/j.gloenvcha.2024.102916 Zilli, M., Scarabello, M., Soterroni, A. C., Valin, H., Mosnier, A., Leclère, D., Havlík, P., Kraxner, F., Lopes, M. A., & Ramos, F. M. (2020). The impact of climate change on Brazil’s agriculture. Science of The Total Environment, 740, 139384. https://doi.org/10.1016/j.scitotenv.2020.139384