D7.2 - Quantified ex-ante global impacts from selected supply chains on biodiversity and other ecosystem services

Name of the Deliverable Quantified ex-ante global impacts from selected supply chains on biodiversity and other ecosystem services Deliverable 7.2 2 Summary Work Package 7 Deliverable No 7.2 Dissemination Level Public Type Report Lead Partner IIASA Due Date July 15th, 2025 Submission Date July 17th, 2025 Status Draft Authors David Leclère, Pekka Lauri, Bettina Meinhart, Amanda Palazzo, Clara Frezal, Fulvio di Fulvio, Niklas Hinkel, Sibylle Rouet-Pollakis, Andrey Lessa Derci Augustynczik 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 ....................................................................................................... 6 Overview of methods ........................................................................................................ 6 Scenario outcomes for soy supply chains ........................................................................... 6 Scenario outcomes for forestry supply chains .................................................................... 8 Scenario outcomes for crop aquafeed supply chains .......................................................... 9 Next steps ....................................................................................................................... 10 INTRODUCTION & OBJECTIVES ....................................................................................... 11 Soy supply chains ............................................................................................................ 11 Forest supply chains ........................................................................................................ 13 Aquaculture and aquafeed supply chains ........................................................................ 14 METHODS .......................................................................................................................... 16 Soy supply chains ............................................................................................................ 16 Overall approach ............................................................................................................................... 16 Scenario description ......................................................................................................................... 16 Quantification ................................................................................................................................... 18 Forest supply chains ........................................................................................................ 19 Overall approach ............................................................................................................................... 19 Scenario description ......................................................................................................................... 21 Quantification ................................................................................................................................... 25 Aquaculture and aquafeed supply chains ........................................................................ 26 Overall approach ............................................................................................................................... 26 Scenario description ......................................................................................................................... 27 Quantification ................................................................................................................................... 30 RESULTS AND DISCUSSION .............................................................................................. 31 Soy supply chains ............................................................................................................ 31 Results ............................................................................................................................................... 31 Discussion ......................................................................................................................................... 54 Forest supply chains ........................................................................................................ 59 Results ............................................................................................................................................... 59 Discussion ......................................................................................................................................... 71 Aquaculture & aquafeed supply chains ............................................................................ 72 Results ............................................................................................................................................... 72 5 Discussion ......................................................................................................................................... 78 PROJECT OUTPUTS ACHIEVED ........................................................................................ 81 REFERENCES ...................................................................................................................... 82 APPENDIX .......................................................................................................................... 89 Stakeholder workshop for soy supply chain scenarios ..................................................... 89 Stakeholder workshop for forestry supply chain scenarios .............................................. 90 Stakeholder workshop for aquaculture & aquafeed supply chain scenarios ..................... 91 Model improvements for soy supply chains ..................................................................... 92 Implementation of Brazilian environmental protection regulations and land-use restrictions in GLOBIOM .......................................................................................................................................... 92 Improvement of soy production system dynamics ........................................................................... 93 Model improvements for forest supply chains ................................................................. 97 Age-class dynamics and forest management ................................................................................... 97 HWP and BECCS carbon accounting ................................................................................................. 99 Deforestation and afforestation ..................................................................................................... 100 Other natural land division to abandoned land and natural land................................................... 101 Wood-based products trade and regional competitiveness ........................................................... 102 Model improvements for aquaculture and aquafeed supply chains ............................... 103 Projecting the capacity of unfed aquaculture beyond 2020 ........................................................... 103 Sensitivity analysis of the crop composition of crop aquafeed requirements ................................ 103 6 EXECUTIVE SUMMARY The CLEVER project sets out to explore the effectiveness of innovative international trade and supply chain governance interventions on non-food biomass supply chains and biodiversity through the use of models and scenarios. This deliverable constitutes the first of two scenario quantification deliverables (D7.2-3), focused on explorative scenarios for three non-biomass supply chains (soy, forestry and crop-based aquafeeds), as part of the Task 7.2. 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 and D6.4. Some of the modelling developments are still on-going: these are briefly described in the appendix, and will be described in more detail in the upcoming deliverable 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. Scenario outcomes 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 recent land use policies in Brazil. Under this business-as-usual scenario, the global demand for livestock products and crop 7 products keeps growing, and the exports of soy-based products from Brazil increase by 76% over the 2020-2050 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 recent decades for terrestrial biodiversity. Overall, this scenario 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 but increased level of imports from soy products to the EU. A first alternative scenario explores the impact of a global food systems sustainability transition towards bending global biodiversity loss. Based on the Bending the Curve study (Leclère et al., 2020), it entails a comprehensive package of increased conservation and restoration, 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 large decreases in the value of production for livestock products, and to a more moderate extent, soy. 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 to 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 may pose to the soy and livestock sector, globally and in particular for the soy sector in Brazil. A second alternative scenario explores 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 for the soy sector in Brazil. In contrary to the previous scenario, it does not assume any of the 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 value of agricultural production. However, it leads 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 some potential mutual interest in such a scenario, between producers in Brazil willing to pursue economic growth, and EU consumers not willing to reduce their consumption of animal products. A last alternative scenario explores the long-term impacts from the US-China trade dispute on the soy markets. Based on recent developments in soya exports from USA to China, it explores the implications of an assumed cap to USA soya exports to China to 25% below 2020 levels, 8 picturing long-lasting reduced market shares for USA soya exports to China through trust and reliability issues. In this scenario, we project China to turn to Brazil for compensating most of the import shortfall. This is projected to be achieved with limited impacts on other export destination of Brazil 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 is expected to remain similar to the baseline. Impacts on biodiversity in Brazil and elsewhere, which are significantly 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 much lower risk to Brazilian ecosystems than the growing demand for livestock products, in particular in China. Scenario outcomes 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 SSP2 socio-economic pathway 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 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 reduction is observed more 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 pressures on biodiversity. A first set of scenarios explores 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 as 90% of the new urban population is assumed to live in wooden buildings, with the related demand increase 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 9 management whereas higher demand for bioenergy has a limited impact as it mainly sourced from energy crops expansion. A second set of scenarios explores 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 plantation forests expansion 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-offs between wood and bioenergy production growth due to land competition between plantation forests and energy crops. Scenario outcomes for crop aquafeed supply chains The baseline scenario depicts a business-as-usual future, in which demand trends for blue food products follow historical trends, the level of wild catch remains constant, and technological progress leads to a higher share of waste in fish meal and fish oil reduction, a higher productivity of fed aquaculture systems, and an increased share of crops in aquafeed requirements. As a result, the consumption of blue food products, and in particular, products sourced from freshwater aquaculture (e.g., carps) and unfed aquaculture (e.g., crustaceans, fish and prawns), increase, primarily through fed and unfed aquaculture. While the demand for fish meal and fish oil decrease, the demand for crop aquafeed keep increasing, but at slower pace than the aquaculture supply, and at the same pace than the total demand for concerned crops. The share of aquafeed use in the total consumption of related crops therefore remains constant to a few percentage points. Further loss of terrestrial biodiversity and an increase in nutrient losses from freshwater fed aquaculture systems are projected. This points to the growth of demand of blue food products, primarily supplied via aquaculture, as contributor to global growth in the environmental impacts from raising food consumption. This trend is however moderated by changes in aquaculture feeding practices, that also redirect pressures from aquafeed demand from marine and pelagic fish stocks to terrestrial ecosystems. A second set of 4 sensitivity scenarios allows to explore the role of specific assumptions about the development of various forms of aquaculture and related feeding practices. In particular, potential increase in unfed aquaculture are required for future levels of demand for specific products (e.g., crustaceans) to be met, and it could also slightly alleviate the need for fed 16 METHODS Soy supply chains Overall approach Goal: For soy supply chains, the main objective is to analyse 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 focus on analysing the impact of stylized intervention scenarios capturing key sources of uncertainties for soy markets around a business-as-usual (BAU) future. An additional set of scenarios, exploring the impact of specific policies in Brazil and EU will be 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. 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. 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 20min presentation from IIASA on model improvements, potential scenario options and preliminary results, followed by a Q&A; 2) a first 20min 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 20min 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 war as main factors 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 will be used to inform D7.3. Scenario description Based on insights from the literature and the stakeholder workshop, we explore the socioeconomic and environmental impacts of a BAU future and of three stylised scenarios capturing 17 key sources of uncertainties for soy markets. The main assumptions behind these scenarios are summarized in Table 1 while a more detailed description is provided below. Table 1 - Summary of assumptions for soy supply chain scenarios Scenario name Assumptions Questions explored SSP2 Prolongation of historical trends (no policy change) What does a BAU future mean for soy markets and biodiversity? IAP Conservation and restoration, supply and demandside efforts aligned with the KMGBF goal of reversing global biodiversity loss from land use change by 2050 What does reaching ambitious biodiversity goals means for soy markets? Tr.Dis 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 war on soy markets and biodiversity? ZNL_Bra 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? As illustrated in Table 1, four scenarios are considered, namely: • BAU scenario (SSP2): The ‘Middle of the Road’ Shared Socioeconomic Pathway (SSP) 2 is used as a baseline scenario. It pictures a prolongation of historical trends in population, dietary preferences, trade and agricultural productivity into the future. More specifically, SSP2 represents a world in which the human population peaks at 9.4 billion by 2070, economic growth is moderate and uneven, and globalization continues with slow socioeconomic convergence between countries. Crop yields increase globally with a partial convergence between highand low-yield regions, while no additional effort are made towards mitigating the triple planetary crisis (i.e., climate change, biodiversity loss, pollution) beyond a reduction in deforestation from other sources than agricultural land use expansion (Popp et al., 2017). • Integrated action portfolio (IAP) scenario: this scenario considers the global implementation of food and land use systems-wide interventions which enables to reverse the global terrestrial biodiversity trends caused by habitat conversion by 2050 while feeding the growing human population (Leclère et al., 2020). The integrated action portfolio includes a mix of supply-side (sustainable increase in crop yields, following the ‘Green road’ SSP1 scenario instead of SSP2), demand-side (diet shift towards lower share of animal calories than in the baseline in regions of high consumption1, and waste 1 It is assumed that the consumption of animal products is decreased by half as compared to the levels assumed in the baseline for 2050, in all regions but the Middle-East and Sub-Saharan Africa (part of the AME aggregated region), India and Pacific islands (part of SAS aggregated region), as well as Southeast Asia (SEA aggregated region), and replaced by an increased consumption of plant-based products to reach the same level of consumption in calorific terms. 18 reduction from farm to fork) and land conservation and restoration measures (increased extent of protected areas to about 40% of terrestrial areas by 2030, increased effectiveness of protected areas preventing any biodiversity-detrimental additional land use conversion within those after 2020, progressive restoration reaching about 450 million hectares by 2050, and pervasive consideration of biodiversity in land use planning). • Trade disruption (Tr.Dis.) scenario: this scenario caps China’s imports of soy-based products from the United States at 75% of 2020 value from 2030 onwards. It represents a plausible assumption about the expected long-term effects of the recent US-China trade dispute, which is expected to erode the competitiveness of US soy exports to China. This represents more than a 25% decline in US soy exports to China after 2020, compared to the BAU scenario. This could be considered as quite pessimistic, given that China’s soy import demand is projected to continue increasing in the coming decades. • Zero natural land loss (ZNL_Bra) scenario: this scenario assumes zero absolute conversion from forest and non-forest natural land to agricultural land in Brazil after 2020. It is used to assess the production and export expansion potential of Brazil’s soy sector if very ambitious land conservation ambitions were met. It is not meant to represent a realistic policy development but is rather intended to capture the boundary conditions of Brazil’s soy sector. These four scenarios were discussed during the stakeholder workshop and met with interest, as they were considered to capture well the key sources of uncertainties for the future evolution of soy markets. 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 war as key factors likely to impact the soy sector in the coming decades. 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 appendix, 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 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 BAU scenario and the three explorative scenarios. 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 19 onwards for main global regions, and at subnational scale for Brazil, are analysed in the results section of this deliverable. Outcomes under the baseline and the stylised policy scenarios are analysed 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 BAU SSP2 scenario in D6.4. The analysis for the BAU and the policy scenarios is carried out at the global and regional levels (for five aggregated regions and the EU)2 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. Forest supply chains Overall approach Goal: For forest-based supply chains, the main goal is to quantify how the potential future developments of wood demands will impact on roundwood harvest volumes, forest management areas, land use changes, forest carbon balance and biodiversity. This includes also understanding how the forest sector competitiveness will develop across different regions and how this would affect global sustainability. In this deliverable, we focus on explorative scenarios picturing plausible future developments in the sectorial demands and land uses impacts, as a complement to more preliminary scenarios presented in D6.2. Methods summary: To design and quantify the scenarios, we rely on a series of scenarios narratives, designed by sectorial experts and quantified with the GLOBIOM model, including the feedback from stakeholders. The scenarios were firstly developed through literature review and sectorial experts, afterwards they were further refined by integrating stakeholders’ input on scenario ideas and key assumptions (including preliminary quantification), gathered during an online workshop. The GLOBIOM model was further improved for accommodating the scenarios specific needs related to carbon dynamics and biodiversity loss. Model improvements, including the addition of forest age-class dynamics, harvested wood products (HWP) and Bioenergy with Carbon Capture and Storage (BECCS) carbon accountings, are described in annex to this report. 2 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). 20 Insights from the literature review: A literature review was used for identifying future major developments that will shape the future bioeconomy in terms of wood demands and carbon storage potentials. The most interesting emerging developments include the growth of wood demands in the building sector, disclosing the potential for storing carbon in long lived harvested wood products (HWP). Accordingly, we developed alternative scenarios for wood use in buildings, one more conservative where timber demands are aligned to GDP/POP growth, and one where we considered also future “timber cities” development and aligned to exploratory scenarios in (Mishra et al., 2022), driven by urbanization in different global regions. Additionally, we considered future alternative growths in demand for BEECS associated with bioenergy scenarios (IIASA, 2021), given that bioenergy demand is another relevant driver of wood-based mitigation. Insights from the stakeholder workshop: A 90-minute online workshop was conducted on December 5th 2024, involving 10 stakeholders in roughly equal proportions from NGOs, universities and industry & private sector, predominantly based Europe but with participants also based in Latin America. The workshop included an introduction session on the scope of the work, a session dedicated to gathering stakeholder views on the 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. The stakeholders found the initial scenario scope relevant but also expressed interest in considering more “multi purpose” managements in forests ecosystems together with scenarios that could address protection of primary forests and restoration ambitions in natural ecosystems. Multifunctional forests are included in the management options considered in the modelling of our scenarios. For further expanding the conservation and multipurpose value of natural/seminatural forests, we have refined our preliminary scenarios (D6.2), by focusing on forest plantations expansion outside forest area and establishing new plantations only on abandoned agricultural land (i.e. excluding natural land). This approach is able of incentivizing a win-win solutions able of reducing pressure on primary and seminatural forests without depleting other valuable ecosystems. At the same time, we have considered the expansion of afforestation for carbon and biodiversity as a restoration option. These different assumptions together go in the direction of carbon-oriented forest managements and bio-based solutions as suggested by stakeholders. Interest in more socioeconomic indicators was also raised by stakeholders, while our modelling approach allows to disclose some macroeconomic trends, like the ones related to forest sector competitiveness across different global regions, including wood products production and trade, this approach does not allow to cover in more detailed indicators related to local welfare (i.e. local employment, wages, health), however, the overall regional sectorial development can be used as an indicator for the forest sector economic growth. Suggestions included also to consider native species and local communities; however, this aspect is hardly captured without a substantial refinement of the modelling framework in specific regions and would go beyond the scope of the scenarios presented in this deliverable. 21 Scenario description We consider a total of 8 scenarios (Table 2), which differ in terms of the assumptions made regarding forest management and wood demand. All scenarios are consistent with the RCP1p9 climate mitigation target and the SSP2 socio-economic development pathway. The BASE scenario is a “business-as-usual” scenario where forest management is based on current management practices and there is no development for energy use and limited development of material use. Table 2. Summary of assumption for forest supply chains scenarios Scenario name Storyline Bioenergy Construction materials Forest plantations BASE “Business as usual” Fixed 2020 level SSP2 demand Fixed 2020 level BIO High bioenergy with current management RCP1p9 net zero SSP2 demand Fixed 2020 level CON High construction with current management Fixed 2020 level Timber cities Fixed 2020 level BIO&CON High bioenergy and construction with current management RCP1p9 net zero Timber cities Fixed 2020 level PLA Business as usual with forest plantations Fixed 2020 level SSP2 demand Expansion outside current forest area PLA BIO High bioenergy with forest plantations RCP1p9 net zero SSP2 demand Expansion outside current forest area PLA CON High construction with forest plantations Fixed 2020 level Timber cities Expansion outside current forest area PLA BIO&CON High bioenergy and construction with forest plantations RCP1p9 net zero Timber cities Expansion outside current forest area The scenarios are based on combinations of three dimensions, including the bioenergy demand (2 variations = fixed 2020 level, RCP1p9 net zero), construction material demand (2 variations = SSP2 demand, Timber cities) and forest plantation area expansion (2 variations = fixed 2020 level, Expansion outside current forest area), as described in the following sections. Under the BASE scenario, we assume a business at usual development which considers bioenergy demand fixed to 2020 level, construction materials demand following SSP2 demand 22 development (i.e. “middle of the road” socioeconomic growth scenario) and forest plantation area is fixed at 2020 level. Bioenergy demand Standard high mitigation bioenergy demand in GLOBIOM is based on the RCP1p9 overshoot scenario, where global bioenergy demand increase from 57 exajoules (EJ) in 2020 to 226 EJ in 2100 (IIASA 2018). This scenario was calculated about 10 years ago under minor sustainability constraints on the land-use sector. Nowadays the general view of the integrated assessment model (IAM) community is that this level of bioenergy demand cannot be reached if stricter land-use sustainability constraints are applied (Wu et al., 2019). Therefore, we apply a lower level bioenergy demand based on the RCP1p9 net-zero scenario. Specifically, two scenarios are considered for bioenergy demand (Figure 1). In the first scenario (RCP1p9 fix2020) bioenergy demand is fixed at 2020 level 57 EJ. This scenario implicitly assumes that additional renewable energy needed for the RCP1p9 mitigation target comes from other renewable energy sources such as solar and wind power instead of bioenergy. The second scenario (RCP1p9 net zero) is the MESSAGE RCP1p9 net-zero scenario (EN_NPi2050_500) where global bioenergy demand increase from 57 EJ in 2020 to 116 EJ in 2100 (IIASA, 2021). Figure 1. Global bioenergy demand scenarios Note: 1 EJ primary energy ≈140 Mm3 woody biomass In 2020, the majority of bioenergy demand is located in Asia and Africa where bioenergy provides the major energy source for heating and cooking for households (Figure 2). In the net zero scenario, bioenergy demand increases most in Asia and Latin America which have the largest biomass potentials for bioenergy production. 0 50 100 150 200 250 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 EJ/yr Global bioenergy demand RCP1p9 fix 2020 RCP1p9 net zero RCP1p9 overshoot (not reachable) 23 Figure 2. Global bioenergy demand divided by region Wood-based construction materials demand Two demand scenarios are considered for wood-based construction materials (Figure 3). In the first scenario, the demand for wood-based construction materials is based on the demand for semifinished products, which is driven by population and GDP growth from the SSP2 scenario. The second scenario is the “Timber Cities” scenario from (Mishra et al., 2022), where 90% of the new urban population is assumed to live in wooden buildings. The number of new urban buildings is based on the amount of people moving from the countryside to cities and is approximated by the regional urbanization rates from to the SSP2 scenario. Figure 3. Global wood-based construction materials demand scenarios Note: 1 Mton construction materials ≈2 Mm3 woody biomass 0 20 40 60 80 100 120 EJ/yr RCP1p9 fix 2020 EU27 ASIA Africa FormerSovietUnion LatinAmericaRest NorthAmerica Brazil 0 20 40 60 80 100 120 EJ/yr RCP1p9 net zero EU27 ASIA Africa FormerSovietUnion LatinAmericaRest NorthAmerica Brazil 0 200 400 600 800 1000 1200 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Mton/yr Wood-based construction materials SSP2 baseline Timber cities (Mishra 2022) 24 As showed in Figure 4, in the “Timber Cities” scenario most new wooden buildings are built in Africa and Asia rather than in regions which traditionally have wooden buildings such as the European Union (EU) and North America. This is because urbanization rates are much higher in Asia and Africa than in the EU and North-America. Figure 4. Wood-based construction materials demand by global regions Plantation forests and energy crops plantations Two scenarios are considered for plantation forests (Figure 5). The difference to the scenarios presented in D6.2 is that natural forests are not allowed to be converted in plantations and plantations expansion is limited to other natural land. In the first scenario, plantation forests area is fixed at the 2020 level (132 Mha). In the second scenario, plantation forests area can be expanded under the specified land-use change constraints, which allows to increase plantation forests area up to 200-300 Mha in 2100 depending on the demand scenario. New plantations are mostly established in the tropical regions, because biomass yields are higher and land-use competition is lower in tropical zone than in boreal and temperate zones. 0 200 400 600 800 1000 1200 Mton/yr SSP2 BASE EU27 ASIA Africa FormerSovietUni LatinAmericaRest NorthAmerica Brazil 0 200 400 600 800 1000 1200 Mton/yr Timber Cities EU27 ASIA Africa FormerSovietUnion LatinAmericaRest NorthAmerica Brazil 25 Figure 5. Global plantation forests area according to different plantations scenarios Besides plantation forests, GLOBIOM also includes energy crops plantations, which compete with plantation forests for the same land areas (Figure 6). The difference between them is that plantations forests produce roundwood and are classified as forest area, while energy crops plantation produce bioenergy crops and are classified as agricultural land area Figure 6. Global energy crops plantation area development according to different bioenergy demand scenarios Quantification The scenarios are quantified with a refined version of the GLOBIOM economic model that includes an enhance representation of the forest sector and it 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 deliverable D6.2 and in the appendix. Additional model 0 50 100 150 200 250 300 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Mha Plantation forests Fix 2020 Free expansion (low construct demand) Free expansion (high construct demand) 0 100 200 300 400 500 600 700 800 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Mha Energy crops plantations RCP1p9 fix 2020 RCP1p9 net zero RCP1p9 overshoot (not reachable) 32 Figure 7. 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 8 displays trends in production and net trade by aggregated regions for livestock products, soy and other crops. In the BAU SSP2 scenario, agricultural production is projected to increase globally between 2020 and 2050, with most livestock and crop output growth occurring in SAS, AME and OSA. Soy production is projected to grow in all producing regions (i.e., NAM, BRA and OSA, SAS), but more strongly in Brazil - the world’s largest producer - which is expected to account for over half of global output growth. 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. As displayed in Figure 8 for net exports of soya bean only, and in Figure 9 for net bilateral trade in all soy-based products (in soybean equivalent) between aggregated regions, net exporters of soy and soya-based products are projected to increase their trade surpluses, with Brazil consolidating its position as the world’s largest exporter and increasing its net exports of soya-based products from 97 Mt in 2020 to over 171 Mt in 2050 (+76%). EUE maintains a surplus for livestock products, a relatively neutral trade position for other crops than soy, and an increased deficit for soy-related products (net imports increase from 25 Mt of soyaequivalent in 2020 to 33 Mt of soya-equivalent by 2050). AME, in turn, is expected to see its 33 agricultural trade deficit widen in the coming decades 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 soya-based products represent the majority of global increases in net imports of soya-based products, from 120 Mt of soya equivalents to about 200 Mt of soya equivalents in 2050. In the IAP scenario, livestock production drops in all regions compared to the 2050 BAU scenario, with strong decline in NAM, EUE, BRA and OSA, and ROW (all around 40%). Global soy production - which is mostly used as animal feed - also decrease compared to the baseline (with decline in BRA and OSA, and SAS but some increase in NAM, where soy largely contributes to the increase in vegetable oil consumption required to compensate for losses in animal calories, further incentivized by a decrease in vegetable oil price – see Figure 10). It should be noted that the decrease in soy production is lower than projected with GLOBIOM in (Leclère et al., 2020), due to the improved representation of oilseed supply chains in CLEVER. As soybean meal (used for livestock feed) and soy oil (used for other purposes) are jointly produced through crushing, reductions in feed demand following the assumed dietary shift translate into smaller reductions in soy production. In contrast to the large increase projected between 2020 and 2050 in the BAU scenario (+76%), total net exports of soy-based products from Brazil to other aggregated regions increase by only 20% over the same period in the IAP scenario, and even decrease for exports to the EU (- 34%). In the IAP scenario, the production of other crops also declines slightly from 2050 baseline levels - although at a lower rate than soy production - as the reduction in demand for feed crops (and associated reduction in production) offsets the increase in demand for plant-based food. In line with the observed trends in production, net trade in livestock and crop products (incl. soy) is below the 2050 BAU (in all regions). In the Trade disruption scenario, soy production in NAM declines from 2050 baseline level (- 15 Mt), with a corresponding increase in production in Brazil (+ 11 Mt) and OSA (+2.6 Mt). Accordingly, net soy exports from NAM decrease (due to the decline in exports to SAS, slightly compensated by increases in exports to AME, EUE and OSA) and those from Brazil and OSA increase. This is in line with what has been observed during the recent trade wars between the United States and China, where Brazil absorbs China’s soy demand that is not met by the United States by increasing its soy production and exports, and limited re-allocation of USA exports to other markets. This confirms the potential risks this trade dispute could have for biodiversityrich ecosystems in Brazil, and soya producers in the USA. However, it should be noted that the projected increase in soy exports from Brazil to China in the BAU is already very high, thereby highlighting that increased demand for soy in China constitutes a larger source of risk for Brazilian ecosystems. Zero conversion of forest and other natural lands in Brazil 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). According, Brazil net exports of soy and livestock products only slightly lower increases than over 2020-2050 than in the baseline, by 4.5 Mt and 162 thousand tons, respectively. This benefits other soy producing regions such as NAM and OSA, whose net exports moderately increase as compared to the 2050 BAU. 34 Figure 8. Projected trends in production and net trade (physical volume, in million tons of fresh matter) for five aggregated regions and two zoomed regions (Brazil BRA and European Union EUE). The colours differentiate livestock products (LSP, in turquoise) from soy (SOY, in purple) and other primary crops (OCRP, in yellow). Figure 9. Projected trends in net trade of soy-based products (physical volume, in million tons of primary products equivalent) for five aggregated importing regions and six exporting regions (incl. Brazil). 35 Substitutions in oilseed markets Figure 10 shows trends in the consumption of soy oil, other vegetable oils, soy protein meals and other protein meals. It enables us to see potential substitution effects between products. Vegetable oils: Soy oil is the main cooking oil consumed in NAM and BRA while other vegetable oils dominate in other regions (e.g., palm oil in SAS and AME, sunflower oil, and rapeseed oil in EUE). In a BAU future, global vegetable oils consumption is projected to increase between 2020 and 2050 – with strong growth expected in SAS and AME due to population growth and increase in per capita consumption. In most regions, the consumption of soy oil increases at a higher rate than that of other vegetable oils, with an increase over time in its share in total consumption (incl. in SAS, BRA and OSA, EUE, and ROW). This suggests that gains in soy competitiveness in vegetable oil markets will be a driver of future growth in soy production, partly driven by the demand for soy protein meal for animal feed. In the IAP scenario, global vegetable oil consumption increases slightly from the 2050 BAU to compensate for the decline in livestock calories, and as a result of an overall decrease in crop commodities prices. However, the share of soy oil in total vegetable oils consumption declines compared to the 2050 BAU, while the consumption of other vegetable oils increases. This substitution is due to the assumed shift in diet away from animal products, which leads to a reduction in demand for all feed products, including soybean meal. The ZNL and the trade disruption scenarios are projected to only marginally differ from the 2050 BAU in terms of the substitution between vegetable oils. Protein meals: Soybean meals account for about three quarter of global protein meal consumption. In the BAU scenario, the consumption of protein meals is projected to increase globally between 2020 and 2050, and most notably in SAS, OSA and AME due to an increase in livestock production and feed intensification in developing countries. About 80% of the global increase in protein meal consumption is expected to come from soybean meals. In the IAP scenario, protein meal consumption declines in all regions compared to the 2050 BAU (except in NAM) due to a shift in diets away from livestock products and associated decline in livestock production (see Figure 8). Most of the global decline in protein meals consumption comes from soybean meals while the consumption of other protein meals increases in some regions (e.g., SAS, OSA, and ROW). In NAM, however, the increase in soy oil consumption leads to an increase in the consumption of soybean meal (although reported as consumption in Figure 10, half of it remains unused due to the drop in domestic and global feed demand). The ZNL and the trade disruption scenarios are projected to only marginally differ from the 2050 BAU in terms of the substitution between protein meals. 36 Figure 10. Projected trends in consumption (physical volume, in million tons of fresh matter) for five aggregated regions and two zoomed regions (Brazil BRA and European Union EUE). The colors differentiate soy oil (SoyO, in turquoise), from other vegetable oils (OVegO, in purple) and soy protein meals (SoyProM, in yellow) from other protein meals (OProM, in red). 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 11. It is calculated as production volume multiplied by producer prices, and provides an understanding of changes in the economic value of agricultural production. In most regions, livestock products represent a larger share in the value of production than in physical volume, due to relatively higher market prices. In the BAU SSP2 scenario, the total value of agricultural production is expected to increase in all regions between 2020 and 2050, with strong increase in the production value of livestock and crops 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. In the IAP scenario, the value of production drops in all regions compared to the 2050 BAU mainly driven by a strong decline in the value of livestock production. Livestock production value decreases at a higher rate than physical volume (at around -60% in EUE, NAM, ROW, BRA and OSA) due to a drop in market prices. The value of soy production also declines in BRA and OSA (at a similar rate and slightly higher rate than physical volume, respectively) as well as in NAM, despite an increase in production in this region. 37 In the Trade disruption scenario, the value of soy production declines at a higher rate than physical volume in NAM (-15% vs -9%) as soy prices drop due to a reduction in Chinese import demand. This illustrates the vulnerability of US soy producers to trade dispute with China. Soy production value, in turn, increases in BRA (+6%) and OSA (+4%) – broadly in line with changes in physical volume, indicating limited price effects on global markets. In the ZNL scenario, agricultural production value in Brazil slightly increase from the 2050 BAU due to an increase in the value of livestock production as price increases more than production declines. This highlights that, in addition to the impacts on production levels, price reactions are important to consider when estimating the opportunity cost (in terms of sectorial economic output) of ambitious conservation efforts in Brazil. The overall impact in terms of economic output might even be positive at the sectorial level, with limited changes in the distribution of economic opportunities across agricultural sub-sectors. Figure 11. 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). Land use As shown in Figure 12, 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 (BRA and OSA, AME, SAS). This is an expected pattern for SSP2 (e.g., (Fricko et al., 2017; Stehfest et al., 2019)), with proportionally lower 38 increases in total cropland than in production due to projected yield growth (see deliverable D6.3). In a BAU 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, with only half allocated to soy for the latter), 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. In the IAP scenario, pasture drops in all regions from both 2050 and 2020 BAU levels following the shift in diets away from animal products. Cropland also declines slightly from the 2050 BAU in all regions (and from 2020 in most regions), with relatively larger decline 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)4, while losses of forest (in BRA, OSA and AME) and other natural lands (mainly in ROW and AME) largely decrease compared to the baseline (preventing the conversion of 80Mha of forest and of 56Mha of other natural land). A large share of the restoration land at global scale is projected to occur in Brazil, which is due to a different representation of pasture in Brazil as compared to other regions5. The projected restoration in the IAP scenario reach about 120 million ha, as compared to about 18 million ha restored as part of the Forest Code in 2030. In the Trade disruption scenario, pasture (and forest to a lower extent) decline slightly in Brazil compared to the 2050 BAU scenario while soy area increases (+4%) to meet the increase in soy import demand from China. In the ZNL scenario, pasture – soy and other cropland to a lower extent - decline in Brazil compared to the 2050 BAU scenario, preventing the conversion of 50 Mha of forest and 16Mha of other natural lands. However, the decline in agricultural land compared to the 2050 BAU is less pronounced than in the IAP scenario, which also assumes the implementation of other land saving measures (i.e., dietary shift, land restoration and sustainable increase in crop yields) in addition to conservation efforts. 4 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). 5 In the standard version of GLOBIOM, only pasture land required to feed ruminants (based on assumptions about the spatial distribution of ruminants, grassland productivity and grazing intensity) is labelled as such, while the remainder of non-forest and non-cropland vegetation is classified as other natural land. This leads to much lower amounts of pasture land than reported by FAOSTAT, and often assumed in other land use models (with a low opportunity cost, as assumed to have very productivity). As pasture is considered eligible for restoration and not other natural land, GLOBIOM usually project lower amounts of restoration than other models. is not considered eligible for restoration, the standard version of GLOBIOM usually projects lower amounts of restoration than most other models. As for CLEVER, the representation of pasture in made closer to that of other models (with a large area identified as pasture and with low opportunity cost for restoration), amounts of restoration are larger for Brazil than for other regions. 39 Figure 12. 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 13 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 dominates. 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. In the IAP scenario, GHG emissions decrease in all regions compared to the baseline, mainly due to a large drop in LUC and livestock emissions, and some decline in crop emissions. LUC becomes a carbon sink in most regions. This decline in AFOLU emissions is due to the assumed shift in diets away from animal products (which are more emission intensive than crop products), as well as the land restoration efforts, supported by supply-side and demand-side measures. 40 In the Trade disruption scenario, no significant changes in GHG emissions are projected compared to the BAU scenario. In ZNL scenario, Brazil’s AFOLU emissions decrease compared to the baseline, mainly due to a large drop in LUC emissions and a small decline in livestock emissions. However, AFOLU emissions are higher than in the IAP scenario, which considers additional measures on the demand and supply sides leading to lower GHG emissions from all sources. Figure 13. 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 colours 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 14. 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 (see deliverable D6.3). Water use for soy production is only significant in NAM (as soy in not irrigated in BRA) and is projected to stay broadly stable over time. In the IAP scenario, global water use is projected to be broadly in line with the 2050 BAU, 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 (see Figure 8). In the Trade disruption scenario, soy water use slightly declines from 2050 baseline levels in 41 NAM (-3%) due to a drop in soy production following a reduction in Chinese import demand (see Figure 8). The ZNL scenario has no clear impact on water use in Brazil. Figure 14. Projected trends in water use for irrigation (in km3) for five aggregated regions and two zoomed regions (Brazil BRA and European Union EUE). The colours 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 per year, are displayed in Figure 15 (see D6.3 for more details on the methodology). They are, in absolute terms, 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 this only accounts for biodiversity impacts from the land use and agricultural sectors. While for land occupation and land transformation impacts this may be close to total impacts, significantly larger climate change impacts on biodiversity would be expected if also accounting for biodiversity impacts from other economic sectors. In the BAU 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 occupation and climate impacts, in turn, are projected to remain broadly stable or increase over time, except for climate impacts in Brazil due the implementation of the Forest Code. It should be noted that, by assumption, restoration does not directly affect biodiversity, as only biodiversity-detrimental impacts are accounted for. 48 Although the Trade disruption scenario leads to a slight redistribution of water use from Pampa to Mata Atlantica and Cerrado, the projected level of water use in both the Trade disruption and ZNL scenarios do not differ significantly from the BAU scenario. Figure 20. Projected trends in Water consumption (water for irrigation, in km3) for the six Brazilian biomes. The colors differentiate the biomes, namely Amazonia (blue), Caatinga (yellow), Cerrado (red), MataAtlantica (turquoise), Pampa (green) and Pantanal (pink). Biodiversity impacts Figure 21 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 characterisation 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. Both methods predict similar overall trends across scenarios. In the BAU 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. 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 ZNL scenario, the total impact on terrestrial ecosystems also declines compared to the BAU mainly due to a drop in land transformation and some decline in land occupation 49 impacts. However, which of the IAP or ZNL scenarios lead to the best outcome for terrestrial biodiversity depends on the method, with the IAP scenario performing better when using LCIMPACT CFs and the ZNL scenario when using the CLEVER CFs. The Trade disruption scenario, in turn, has no significant impact on terrestrial biodiversity for both metrics. The main differences between the two methods lie in the scale of the impacts (in particular, for transformation impacts) and the relative contribution of different biomes to the overall impacts. First, the Brazil-level transformation impacts of farming activities on terrestrial ecosystems are more than two times higher when using the CLEVER CFs than when using the LC-IMPACT CFs (in line with what was discussed in D6.3). Then, when using the CLEVER CFs, Amazonia is the biome with the largest overall impact on terrestrial ecosystems (42% of the total in 2020), followed by Cerrado (26%) and Mata Atlantica (17%) while when using the LC-IMPACT CFs, Mata Atlantica has the largest impact (42%), followed by Amazonia (31%) and Cerrado (21%). Finally, the avoided loss of forest and other natural land in the ZNL scenario (as compared to the BAU 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). This provides important nuances regarding the best strategies for biodiversity protection in Brazil. Figure 21. 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. 50 As shown in Figure 22, freshwater ecosystems are most affected by farming activities in the Mata Atlantica, Cerrado, and Amazonia biomes. In the BAU 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. In the IAP scenario, the impacts on freshwater ecosystems are projected to be lower than in the BAU and become negative, mainly due to strong decline in climate change impacts (with LUC turning into a carbon sink) and some decrease in freshwater eutrophication in all biomes. However, as discussed above, water stress slightly increases from the BAU due to a small increase in water use for irrigated crops (see Figure 20). In the Trade disruption scenario, biodiversity impacts on freshwater ecosystems slightly increase from the 2050 BAU, mainly due to an increase in freshwater eutrophication (and water stress to a lower extent). In the ZNL scenario, the total freshwater ecosystems’ impacts are projected to be lower than in the BAU scenario, mainly due to lower climate change and eutrophication impacts. However, the total impact on freshwater ecosystems remains higher than in the IAP scenario. Figure 22. 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), Mata Atlantica (turquoise), Pampa (green) and Pantanal (pink). 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 BAU scenario (SSP2) and three stylised 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 51 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 additional impacts from these supply chain steps are presented for soy and compared to the footprint from farming presented above. We analyse the differences in impact in terms of aquatic and terrestrial extinction risk, but also in terms of GHG emissions and freshwater consumption. Comparison of farming impacts for all crop and livestock products to soy impact from farming and other supply chain steps The integration of all supply chains impacts into GLOBIOM for soy allows us to put these impacts into perspective by comparing them to the previously calculated Brazilian impacts from farming for all agricultural commodities (i.e., soy and other crops and livestock products) in 2020 (Figure 23). When looking only at the impacts of farming covered in GLOBIOM (in yellow in Figure 23), soy production alone accounts for 3% of GHG emissions and aquatic biodiversity impacts and 13% of terrestrial biodiversity impacts (soy has no impact of freshwater consumption it is not irrigated in Brazil). This highlights the relative importance of soy within all agricultural commodities for the farming impacts covered in GLOBIOM. However, they are different between biomes regarding the relative contribution of soy to the total farming impacts covered in GLOBIOM. For example, in the Cerrado - where 60% of Brazil’s soy production takes place - soy production accounts for a larger share of the total farming impacts than at national scale or than in any other biomes. Conversely, in Amazonia, where GHG emissions and terrestrial ecosystems impacts from farming are the largest, soy production accounts for a smaller portion of these impacts than at the national scale. Figure 23 also illustrates the contribution of different supply chain steps to soy overall impacts. For terrestrial ecosystem impacts, farming GLOBIOM accounts for 90% of the total impacts, while other supply chains steps are responsible for the remaining 10%. For aquatic ecosystems, farming GLOBIOM accounts for about half of the total impacts, while other supply chains steps (mainly crushing and input production) accounts for the other half. Farming GLOBIOM has no impact on freshwater consumption as soy is not irrigated in Brazil; most of the impacts coming from inputs production (~60%) and crushing (~30%). Finally, for GHG emissions, farming impacts covered in GLOBIOM accounts for about 25% of the total impact while other supply chain steps are responsible for the remaining 75%. This highlights the importance of considering additional supply chain impacts in the standard version of GLOBIOM, as including those for other crops would further increase the overall environmental impacts. 52 Figure 23. Soy supply chain footprint compared to total farming footprint for Brazil and four soy-producing biomes in 2020 (SSP2). In Yellow, described as “Farming - GLOBIOM” includes all already calculated impacts from GLOBIOM, which are related to direct agricultural and forestry activities. “Farming – other impacts” mainly includes impact associated with input production such as fertilizer. Evolution of soy supply chain biodiversity impacts When looking at the evolution of the biodiversity impact of soy production in Brazil (Figure 24), we observe a decline in impact on terrestrial ecosystems over time, due to a decrease in farming 53 impacts, more specifically in land transformation, while upstream and downstream supply chain impacts increase. The impact on aquatic ecosystems, in turn, increases over time, due to an increase in most supply chain impacts (excl. farming), more specifically GHG emissions. In the BAU scenario, there is a noticeable increase in the biodiversity impacts on aquatic ecosystems, and GHG emissions attributed to non-farming supply chain stages while farmingrelated impacts decline over time. 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 in impacts due to technological progress. By contrast, farming impacts covered in GLOBIOM used coefficient per hectare or per ton of input, and these impacts are affected by long-term increases in yield and input use efficiency. For other impacts than the farming GLOBIOM, and including aquatic biodiversity impacts, this could potentially lead to an overestimation of the impacts in the projections. In contrast, the impacts on terrestrial ecosystems, which are primarily driven by farming impacts covered in GLOBIOM, closely follow the trends observed for Brazil as a whole. In the IAP scenario, where Brazil’s soy production drops compared to the BAU (see Figure 17), the impacts of the soy supply chain on both terrestrial and aquatic ecosystems decline from the BAU levels (with the impact on terrestrial ecosystems becoming negative), mainly due to a decrease in farming impacts. Mid-points impacts (GHG emissions and freshwater consumption) also decline from the baseline, mainly due lower impacts from upstream supply chain activities (i.e., input production) and from crushing. In the Trade disruption scenario, where Brazil’s soy production slightly increases from the 2050 BAU, the impact on terrestrial ecosystems is comparable to 2050 baseline level while the impact on aquatic ecosystems increases 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 ZNL scenario, the end-point impacts (aquatic and terrestrial biodiversity) of the soy supply chain are below 2050 BAU levels mainly due to lower farming impacts following a small decline in Brazil soy production. Mid-points impacts are also slightly below baseline levels, mainly due to lower impacts from crushing for freshwater consumption, and from farming and crushing in the case of GHG emissions. 54 Figure 24. 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 this deliverable, we explored alternative future developments in the soy markets, with a focus on Brazil and the EU. 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, 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 account for more than half of the global growth in output and net exports. Most of the increase in exports are destinated to Asia, while EU also increases its level of imports. 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 until 2050 albeit at a slower pace than in previous decades. Large forest conversions to pasture are projected, in particular in the Amazon biome, while moderate forest restoration efforts take place, in particular in the Mata Atlantica biome. It should be noted that these trends consider the effects from 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 triggering moderate restoration efforts). This leads to a 55 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 the value of production are projected, in particular for soy, but these remain of a much lower amplitude than projected in increases in production volume. Extinction risks from agricultural activities increase for aquatic ecosystems of 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 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 at global scale with the literature of future projections for business as usual scenarios (e.g., for SSP2 scenario (Popp et al., 2017)) and medium-term agricultural outlooks (OECD/FAO, 2024), and similar studies at the scale of Brazil (e.g., (Soterroni et al., 2018)). It might be relevant to consider additional global drivers, such as impacts from climate change, but these are not expected to lead to large changes in Brazil’s soy export potential (Zilli et al., 2020). Impact of alternative futures We also designed explorative scenarios capturing key sources of uncertainty for soy markets (as identified in the literature and during a stakeholder workshop) and analysed their socioeconomic and environmental impacts around the BAU, with a focus on their implications for BRA-EU soy supply chains. This includes scenarios representing a) a global food system transformation (i.e., the IAP scenario), b) the potential long-term implications of the trade dispute between US and China (i.e., the Trade disruption scenario), and c) an idealised ambitious land conservation policy in Brazil (i.e., the ZNL 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 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 BAU. 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 BAU scenario. Land-use change becomes a large 56 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 BAU levels. Decreases are particularly large (more than -50% as compared to 2020 levels) for livestock production. Decrease 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 ZNL scenario) only leads to a slightly lower post-2020 growth in the production and trade of livestock and soy products compared to the BAU, 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 BAU. 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 pasture land 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 centre 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 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 BAU, 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 BAU scenario. We found no significant impact of this scenario on Brazillevel projected biodiversity impacts on aquatic and terrestrial ecosystems as compared to the BAU scenario. 57 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. A rich picture on uncertain future biodiversity impacts In this deliverable, we rely on multiple indicators to quantify biodiversity impacts, relying on connecting various LCIA-based characterization factors to various outputs the GLOBIOM model (see deliverables D6.3 and D6.4). This included pre-existing characterization factors from the LCIMPACT project, as well as new characterization factors generated in the CLEVER project and tailored for a more accurate representation of the Brazilian context. These indicators enable us to translate resource use (land, water), input use and GHG emissions (as quantified in GLOBIOM) into biodiversity impacts on both aquatic and terrestrial ecosystems, through multiple impact pathways (e.g., land use, water use, climate change, nutrient pollution). This provides a significant advancement to the literature, with the few land use models connected to biodiversity models often focusing on terrestrial ecosystems and land use change only (e.g., (Leclère et al., 2020)), and in rare cases including a broader set of impacts (e.g., (Schipper et al., 2020)). We for example explored the impact of future soy supply chains developments on freshwater ecosystems, with an important role of water use (for soy produced in the USA) and nutrient losses and GHG emissions (for soy produced in Brazil). This allowed us enable us to capture, for Brazil soy supply chains, the impacts of supply chain steps not covered by the GLOBIOM model, both upstream (e.g., GHG emissions from fertilizer production) and downstream (e.g., domestic and international transport of soy commodities) of the farming stage. For climate change, for instance, farming accounts for about 25% of the total impact while other supply chain steps (mainly input production and crushing) are responsible for the remaining 75%. Moreover, for aquatic ecosystems and climate change impacts, the share 64 land (AbdLnd) rather than biodiversity rich natural land (NatLnd) (Figure 30). This implies that the direct land-use change impact of plantations expansion on biodiversity remains small and the indirect land-use change impact is positive (i.e., spared natural forests area). Figure 29. Overall land-use change (Mha) Figure 30. Plantations (energy crops +forests) expansion and decrease in abandoned vs. natural land area (Mha) 0 500 1000 1500 2000 2500 3000 Mha BASE 0 500 1000 1500 2000 2500 3000 Mha PLA_BIO&CON 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 100 150 200 250 300 350 400 2020 2030 2040 2050 2060 2070 2080 2090 2100 Plantations 0 50 100 150 200 250 300 2020 2030 2040 2050 2060 2070 2080 2090 2100 Abandonend land 2800 2810 2820 2830 2840 2850 2860 2870 2880 2890 2900 2000 2020 2040 2060 2080 2100 Natural land BASE BIO CON BIO&CON PLA PLA_BIO PLA_CON PLA_BIO&CON 65 Forest carbon balance In the BASE scenario, forest carbon storage increases over time (Figure 31). The main reason is afforestation uptake, which compensates for carbon storage decrease in the existing forest area. The existing forest area is a sink until 2050, but after 2050 it turns to a source due to increased harvest volumes, natural mortality and the saturation of biomass growth. In addition to afforestation, the carbon changes in the existing forest area are compensated by HWP and BECCS. However, the impact of HWP and BECCS on overall carbon balance is relatively small in the BASE scenario due to “business as usual” bioeconomy development. Figure 31. Global forest carbon stock and carbon sink/source in the BASE scenario In the CON scenario, forest carbon storage remains lower than in the BASE scenario (Figure 32). The reason is higher harvest volumes, which decreases carbon in the existing forests more than is compensated by the increase in HWP carbon storage. 0 50 100 150 200 250 300 350 400 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Carbon storage (PgC) BECCS HWP Afforestation Existing forests -6000 -4000 -2000 0 2000 4000 6000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Carbon sink/source (MtCO2/yr) Deforestation Existing forest management Afforestation HWP BECCS total 66 Figure 32. Global forest carbon stock and sink/source in the CON scenario In the BIO scenario, forest carbon storage is higher than in the BASE and CON scenarios (Figure 33). There are two reasons for this. First, the higher bioenergy demand does not increase harvests and carbon emissions as much as the higher wood-based construction materials demand, because the majority of additional bioenergy demand is satisfied by energy crops. Second, the higher bioenergy demand increases carbon storage of BECSS considerably compared to “business-as-usual” bioenergy demand, especially after 2050 when large scale implementations of BECCS becomes feasible due to high carbon prices. 0 50 100 150 200 250 300 350 400 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Carbon storage (PgC) BECCS HWP Afforestation Existing forests -6000 -4000 -2000 0 2000 4000 6000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Carbon sink/source (MtCO2/yr) Deforestation Existing forest management Afforestation HWP BECCS total 67 Figure 33. Global forest carbon stock and sink/source in the BIO scenario In the PLA CON scenario, forest carbon storage is higher than in the BASE and CON scenarios but not as high as in the BIO scenario (Figure 34). There are two reasons for this. First, additional harvests come from forests plantations instead of natural/seminatural forests, which keeps the existing forest area carbon emissions at the BASE scenario level. Second, HWP storage is higher than in the BASE scenario. 0 50 100 150 200 250 300 350 400 450 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Carbon storage (PgC) BECCS HWP Afforestation Existing forests -6000 -4000 -2000 0 2000 4000 6000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Carbon sink/source (MtCO2/yr) Deforestation Existing forest management Afforestation HWP BECCS total 68 Figure 34. Global forest carbon stock and sink/source in PLA CON scenario To understand the role of HWP and BECCS in forest carbon balance, we compare forest carbon stock and sink with and without HWP and BECCS. In general, the forest carbon storage is increasing in all scenarios. Without HWP and BECCS, however, the increase starts to saturate after 2050 (Figure 35). Adding HWP and BECCS in the forest carbon storage improves forest carbon balance in all scenarios and tends to eliminate the saturation (Figure 36). The reason is that HWP and BECCS remove old trees and make space for new trees and/or remaining trees regrowth. This has basically a similar effect than natural mortality, but the difference is that HWP and BECCS store the carbon of the removed biomass instead of letting it to decompose and emit in the atmosphere. 0 100 200 300 400 500 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Carbon storage (PgC) BECCS HWP Afforestation Existing forests -8000 -6000 -4000 -2000 0 2000 4000 6000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Carbon sink/source (MtCO2/yr) Deforestation Existing forest management Afforestation HWP BECCS total 69 Figure 35. Forest carbon storage (PgC) and sink (MtCO2/yr) including only the biogenic carbon that stays in forests (F) Figure 36. 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) 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 PLA scenarios than in the “business as usual” forest management scenarios (). In particular, biodiversity loss is highest in the CON and BIO&CON scenarios, where increasing demand for wood-base construction materials leads to natural forests extraction for production. In Figure 38, biodiversity loss is divided into different land-uses. It shows the impact of natural forests in the BIO&CON scenario is clearly higher than in the PLA BIO&CON scenario. On the other hand, the plantation 290 310 330 350 370 390 410 2000 2020 2040 2060 2080 2100 Carbon storage F -6000 -5000 -4000 -3000 -2000 -1000 0 2020 2040 2060 2080 2100 Carbon sink F BASE BIO CON BIO&CON PLA PLA_BIO PLA_CON PLA_BIO&CON 290 310 330 350 370 390 410 2000 2020 2040 2060 2080 2100 Carbon storage F+HWP+BECCS -6000 -5000 -4000 -3000 -2000 -1000 0 2020 2040 2060 2080 2100 Carbon sink F+HWP+BECCS BASE BIO CON BIO&CON PLA PLA_BIO PLA_CON PLA_BIO&CON 70 expansions in the PLA BIO&CON scenario causes less biodiversity loss than natural forests harvests in the BIO&CON scenario. Figure 37. Global biodiversity loss Figure 38. Global biodiversity loss divided into different land-uses 0 2 4 6 8 10 12 14 16 18 20 22 BASE 0 2 4 6 8 10 12 14 16 18 20 22 BIO&CON 0 2 4 6 8 10 12 14 16 18 20 22 PLA BIO&CON Plantations Natural forests Grassland Cropland 14 15 16 17 18 19 20 21 22 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Total biodiversity BASE BIO CON BIO&CON PLA PLA_BIO PLA_CON PLA_BIO&CON 71 Discussion The problem of the wood-based bioeconomy is that it tends to increase harvests volumes considerably relative to current harvest volumes, which might have negative impacts on forest carbon balance and biodiversity. Here we consider the possibility of avoiding these effects by shifting logging from natural forests to plantations forests. Currently plantation forests cover about 10% of the global production forest area and account for about 30% of the global roundwood supply. According to our result, the negative effects of the bioeconomy on forest carbon balance and biodiversity could be avoided if we would double the plantation forests area and provide 60% of the global roundwood supply from these plantations. The question if we should produce woody biomass in natural/seminatural or plantation forests is analogues to the land-sharing vs. land-sparing discussion in conservation ecology and agricultural landscape management (Salles et al., 2017). Land-sharing combines production and nature conservation in the same area while land-sparing separates these objectives in different areas, which implies that our current management scenarios can be interpret as land-sharing while our plantation forests scenarios as land-sparing. According to previous studies focusing on forest landscape management, land-sharing tends to work better in regions with a long history of forest management, low biodiversity and slow biomass growth while land-sparing works better in evolving forest industry regions with high biodiversity and fast biomass growth (Betts et al., 2021). Our stakeholders emphasized the role of expanding multi-purpose forest managements, which would entail more natural/seminatural forest land managed for maximizing multiple ecosystem services rather than wood production, in the direction of a land-sharing approach. According to our scenario results, land-sharing implies that more natural/seminatural forests area is needed for production use, which tends to increase existing forest area carbon emissions and biodiversity loss. This effect is significantly larger in high construction materials demand scenario than in the high bioenergy demand scenario, because most of the bioenergy increase is met by an increase in energy crops production rather than in roundwood harvesting. Increased carbon emissions can be compensated by higher HWP and BECCS carbon storages. The BECCS carbon storage is typically higher than the HWP carbon storage, since HWP provides a temporary carbon storage while BECCS a permanent carbon storage. On the other hand, BECCS is a new and relatively expensive technology, and becomes economically available only after 2050 when carbon prices are high enough. Land sparing implies that less natural/seminatural forests area is needed for production use, which tends to decrease existing forest area carbon emissions and biodiversity loss. The spared natural forest area could be used for carbon sequestration and nature conservation, as also demanded by stakeholders. On the other hand, under the land sparing scenario, more land is needed for plantations, which could potentially decrease carbon sequestration and biodiversity, as plantations carbon storage and biodiversity are low. This problem is avoided by locating new plantations mostly in abandoned land rather than biodiversity and carbon rich natural forests or natural land. 72 Aquaculture & aquafeed supply chains Results Demand for final blue food products Projections for the final demand of blue food products are presented in Figure 39, with different time horizons for the BAU scenario (2010, 2020, 2030, 2040 and 2050), and 2050 values for other scenarios. In the BAU 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 others 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 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 seeing 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 2020-2050 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 BAU 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 BAU 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 BAU levels), moderate for FRSH and SHRI (2050 demand is close to BAU levels and well above 2020 levels). Lastly, as opposed to the BAU 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 (to 2020 levels, for both scenarios), and constant total aquafeed requirements per unit of output (to 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 a context of growing demand for aquaculture products. 73 Figure 39 - 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 40. By assumption, most of the supply increase is projected to be sourced from fed aquaculture. In the BAU 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) t 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 that of the BAU, except for two 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 BAU scenario), 80 production in 2020, and this share is not expected to increase by 2050, unless the future changes in feeding practices assumed in the BAU do not materialize. Over the same period, we project more than 300 million hectares of forest and non-forest natural land to be converted to agriculture, with very little variation across scenarios, suggesting that future trends in the blue food sector will have limited impact on land use change at the global scale, and that the specific impacts of future aquafeed requirements maybe challenging to capture. We could isolate these specific impacts by using a counterfactual scenario in which the blue food sector (demand, supply, feeding practices) remains constant at 2020 level, and investigate the role of various assumptions about future trends in the blue food sector by comparing impacts between the counterfactual and not only the baseline scenario, but also to the sensitivity scenarios. We found the impacts on global terrestrial species extinction risks from land occupation attributable to future crop aquafeed requirements to be negative but moderate, and our results illustrate at least four different sources of complexities when estimating those. First, assumptions about future trends in aquaculture feeding practices such as overall efficiency and share of cropvs. fish-based aquafeed will modulate the future increase in total crop aquafeed demand, with some of these factors (e.g., increased overall efficiency vs. increased share of crop-based products in total aquafeed) expected to play in opposite direction. Second, the relationship between natural land loss and biodiversity loss is variable, due to spatial differences in biodiversity patterns not only across but within regions. For example, the amount of global species extinction risks relative to natural land loss was systematically higher for Latin America and Caribbean, as well as South Asia and South-Eastern Asia, than for other world regions, and this link can vary across scenarios. Third, trade dependencies might displace the related increase in crop production to biodiversity hotspots, to an amplitude that may change with change with future trade patterns. For example, while less than 10% of the estimated future increases in crop aquafeed requirements is projected to be located in Latin America and the Caribbean, this region was found to host at least a quarter of global attributable natural land loss in most scenarios, in relation to the demand for crops like soya and corn. Fourth, assumptions about the composition of crop-based aquafeed requirements, whose future patterns are generally not well constrained, can have large applications. For example, we found that replacing half of corn and soya crop aquafeed demand by crops like wheat (with high protein content and grown in temperate regions) in China could significantly reduce risks of biodiversity loss in Latin America and the Caribbean. At last, we also found the projected developments of in crop aquafeed requirements to generate additional nutrient losses on finfish aquaculture farms, in particular for freshwater aquaculture in Eastern Asia. Similar patterns might be expected for non-finfish aquaculture, and losses are also expected for phosphorous nutrients, pointing to both increased eutrophication risks associated with aquaculture developments, and benefits of integrated aquaculture systems with nutrient recycling (e.g., (Xiao et al., 2017)). In addition, the increased demand for crop aquafeed is likely to generate additional nutrient losses within cropland farms, which might be possible to quantify with our modelling framework. 81 PROJECT OUTPUTS ACHIEVED List of Zenodo repositories containing model projections: • Zenodo record for soy supply chains: https://zenodo.org/records/15875871 • Zenodo record for forest supply chains: https://zenodo.org/records/15829433 • Zenodo record for aquaculture & aquafeed supply chains: https://zenodo.org/records/15875624 82 REFERENCES Azuero-Pedraza, C., Lauri, P., Lessa Derci Augustynczik, A., & Thomas, V. (2024). Managing Forests for Biodiversity Conservation and Climate Change Mitigation. https://pubs.acs.org/doi/10.1021/acs.est.3c07163 Besnard, S., Koirala, S., Santoro, M., Weber, U., Nelson, J., Gütter, J., Herault, B., Kassi, J., N’Guessan, A., Neigh, C., Poulter, B., Zhang, T., & Carvalhais, N. (2021). Mapping global forest age from forest inventories, biomass and climate data. Earth System Science Data, 13(10), 4881–4896. https://doi.org/10.5194/essd-13-4881-2021 Betts, M. G., Phalan, B. T., Wolf, C., Baker, S. C., Messier, C., Puettmann, K. J., Green, R., Harris, S. H., Edwards, D. P., Lindenmayer, D. B., & Balmford, A. (2021). Producing wood at least cost to biodiversity: Integrating Triad and sharing–sparing approaches to inform forest landscape management. Biological Reviews, 96(4), 1301–1317. https://doi.org/10.1111/brv.12703 Bond, W. (2016). Ancient grasslands at risk. https://www.science.org/doi/10.1126/science.aad5132 Brus, D. J., Hengeveld, G. M., Walvoort, D. J. J., Goedhart, P. W., Heidema, A. H., Nabuurs, G. J., & Gunia, K. (2012). Statistical mapping of tree species over Europe. European Journal of Forest Research, 131(1), 145–157. https://doi.org/10.1007/s10342-011-0513-5 Cabernard, L., Pfister, S., & Hellweg, S. (2024). Biodiversity impacts of recent land-use change driven by increases in agri-food imports. Nature Sustainability, 7(11), 1512–1524. https://doi.org/10.1038/s41893-024-01433-4 Choe, J., Hammer, A., & Montgomery, C. (2019). U.S. SOYBEAN EXPORTS TO CHINA CRUSHED AMID RISING TRADE TENSIONS. Cohn, A. S., Mosnier, A., Havlík, P., Valin, H., Herrero, M., Schmid, E., O’Hare, M., & Obersteiner, M. (2014). Cattle ranching intensification in Brazil can reduce global greenhouse gas emissions by sparing land from deforestation. Proceedings of the National Academy of Sciences, 111(20), 7236–7241. https://doi.org/10.1073/pnas.1307163111 Cottrell, R. S., Blanchard, J. L., Halpern, B. S., Metian, M., & Froehlich, H. E. (2020). Global adoption of novel aquaculture feeds could substantially reduce forage fish demand by 2030. Nature Food, 1(5), 301–308. https://doi.org/10.1038/s43016-020-0078-x Cottrell, R. S., Fleming, A., Fulton, E. A., Nash, K. L., Watson, R. A., & Blanchard, J. L. (2018). Considering land–sea interactions and trade-offs for food and biodiversity. Global Change Biology, 24(2), 580–596. https://doi.org/10.1111/gcb.13873 Cowley, C. (2020). Reshuffling in Soybean Markets following Chinese Tariffs. The Federal Reserve Bank of Kansas City Economic Review. https://doi.org/10.18651/er/v105n1cowley Curtis, P. G., Slay, C. M., Harris, N. L., Tyukavina, A., & Hansen, M. C. (2018). Classifying drivers of global forest loss. Science, 361(6407), 1108–1111. https://doi.org/10.1126/science.aau3445 83 De Maria, M., Robinson, E., Kangile, J. R., Kadigi, R. M., Dreoni, I., Couto, M., Howai, N., Peci, J., & Fiennes, S. (2020). Global Soybean Trade – The Geopolitics of a Bean. UN Environment Programme World Conservation Monitoring Centre (UNEP-WCMC). https://doi.org/10.34892/7YN1-K494 De Oliveira Silva, R., Barioni, L. G., Queiroz Pellegrino, G., & Moran, D. (2018). The role of agricultural intensification in Brazil’s Nationally Determined Contribution on emissions mitigation. Agricultural Systems, 161, 102–112. https://doi.org/10.1016/j.agsy.2018.01.003 Elleby, C., Domínguez, I. P., Nielsen, R., Nielsen, M., & Hoff, A. (2025). Introducing maximum sustainable yield targets in fisheries could enhance global food security. Communications Earth & Environment, 6(1), 33. https://doi.org/10.1038/s43247-02401851-4 FAO (Ed.). (2018). The State of World Fisheries and Aquaculture 2018—Meeting the sustainable development goals. FAO. (2020). Global Forest Resources Assessment 2020. FAO ; https://openknowledge.fao.org/handle/20.500.14283/ca9825en FAOSTAT. (2025). Forest Production and Trade. https://www.fao.org/faostat/en/#data/FO Frezal, C., & Deuss, A. (2025, February 11). Trade-related measures linked to the environmental sustainability of agriculture. OECD. https://www.oecd.org/en/publications/traderelated-measures-linked-to-the-environmental-sustainability-of-agriculture_ebfdca09en.html Fricko, O., Havlik, P., Rogelj, J., Klimont, Z., Gusti, M., Johnson, N., Kolp, P., Strubegger, M., Valin, H., Amann, M., Ermolieva, T., Forsell, N., Herrero, M., Heyes, C., Kindermann, G., Krey, V., McCollum, D. L., Obersteiner, M., Pachauri, S., … Riahi, K. (2017). The marker quantification of the Shared Socioeconomic Pathway 2: A middle-of-the-road scenario for the 21st century. Global Environmental Change, 42, 251–267. https://doi.org/10.1016/j.gloenvcha.2016.06.004 Froehlich, H. E., Jacobsen, N. S., Essington, T. E., Clavelle, T., & Halpern, B. S. (2018). Avoiding the ecological limits of forage fish for fed aquaculture. Nature Sustainability, 1(6), 298– 303. https://doi.org/10.1038/s41893-018-0077-1 Froehlich, H. E., Runge, C. A., Gentry, R. R., Gaines, S. D., & Halpern, B. S. (2018). Comparative terrestrial feed and land use of an aquaculture-dominant world. Proceedings of the National Academy of Sciences, 115(20), 5295–5300. https://doi.org/10.1073/pnas.1801692115 Fuller, M. R., Ganjam, M., Baker, J. S., & Abt, R. C. (2025). Advancing forest carbon projections requires improved convergence between ecological and economic models. Carbon Balance and Management, 20(1), 2. https://doi.org/10.1186/s13021-024-00290-0 Fulvio, F. D., Snäll, T., Lauri, P., Forsell, N., Mönkkönen, M., Burgas, D., Blattert, C., Eyvindson, K., Caicoya, A. T., Vergarechea, M., Antón-Fernández, C., Klein, J., Astrup, R., Lukkarinen, J., Pitzén, S., & Primmer, E. (2025). Impact of the EU biodiversity strategy 84 for 2030 on the EU wood-based bioeconomy. Global Environmental Change, 92, 102986. https://doi.org/10.1016/j.gloenvcha.2025.102986 Gephart, J. A., Henriksson, P. J. G., Parker, R. W. R., Shepon, A., Gorospe, K. D., Bergman, K., Eshel, G., Golden, C. D., Halpern, B. S., Hornborg, S., Jonell, M., Metian, M., Mifflin, K., Newton, R., Tyedmers, P., Zhang, W., Ziegler, F., & Troell, M. (2021). Environmental performance of blue foods. Nature, 597(7876), 360–365. https://doi.org/10.1038/s41586-021-03889-2 Guidotti, V., Freitas, F. L. M., Sparovek, G., Pinto, L. F. G., Hamamura, C., Carvalho, T., & Cerignoni, F. (2017). Números detalhados do novo Código Florestal e suas implicações para os PRAs. https://repositorio.usp.br/item/002937819 Hadjikakou, M., Bowles, N. I., Geyik, O., Conijn, S. J. G., Mogollón, J. M., Bodirsky, B. L., Muller, A., Weindl, I., Moallemi, E. A., Shaikh, M. A., Damerau, K., Davis, K. F., Pfister, S., Springmann, M., Clark, M., Metson, G. S., Röös, E., Bajzelj, B., Graham, N. T., … Bryan, B. A. (2025). Ambitious food system interventions required to mitigate the risk of exceeding Earth’s environmental limits. One Earth, 101351. https://doi.org/10.1016/j.oneear.2025.101351 Halpern, B. S., Frazier, M., Verstaen, J., Rayner, P.-E., Clawson, G., Blanchard, J. L., Cottrell, R. S., Froehlich, H. E., Gephart, J. A., Jacobsen, N. S., Kuempel, C. D., McIntyre, P. B., Metian, M., Moran, D., Nash, K. L., Többen, J., & Williams, D. R. (2022). The environmental footprint of global food production. Nature Sustainability, 5(12), 1027– 1039. https://doi.org/10.1038/s41893-022-00965-x Henders, S., Persson, U. M., & Kastner, T. (2015). Trading forests: Land-use change and carbon emissions embodied in production and exports of forest-risk commodities. Environmental Research Letters, 10(12), 125012. https://doi.org/10.1088/17489326/10/12/125012 Humpenöder, F., Popp, A., Dietrich, J. P., Klein, D., Lotze-Campen, H., Bonsch, M., Bodirsky, B. L., Weindl, I., Stevanovic, M., & Müller, C. (2014). Investigating afforestation and bioenergy CCS as climate change mitigation strategies. Environmental Research Letters, 9(6), 064029. https://doi.org/10.1088/1748-9326/9/6/064029 IBF-IIASA. (2023). Global Biosphere Management Model (GLOBIOM) Documentation 2023— Version 1.0. International Institute for Applied Systems Analysis. https://pure.iiasa.ac.at/18996 IFPRI. (2025a, April 8). How ‘reciprocal tariffs’ harm agricultural trade. IFPRI. https://www.ifpri.org/blog/how-reciprocal-tariffs-harm-agricultural-trade/ IFPRI. (2025b, April 21). U.S.-China trade war 2.0: What are the implications for global oilseed markets? IFPRI. https://www.ifpri.org/blog/u-s-china-trade-war-2-0-what-are-theimplications-for-global-oilseed-markets/ IIASA. (2018). SSP Database—Version 2.0. https://tntcat.iiasa.ac.at/SspDb/dsd?Action=htmlpage&page=welcome IIASA. (2021). ENGAGE Scenario Explorer. https://data.ece.iiasa.ac.at/engage/#/login 85 IIASA. (2025). GLOBIOM-forest documentation. GitHub. https://github.com/iiasa/GLOBIOM_forest/blob/main/GLOBIOM_forest_documentati on.pdf IPCC. (2019). Harvested wood products—2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories (Vol. 4). https://www.ipccnggip.iges.or.jp/public/2019rf/pdf/4_Volume4/19R_V4_Ch12_HarvestedWoodProduct s.pdf Johnston, C. M. T., & Radeloff, V. C. (2019). Global mitigation potential of carbon stored in harvested wood products. Proceedings of the National Academy of Sciences, 116(29), 14526–14531. https://doi.org/10.1073/pnas.1904231116 Lauri, P., Forsell, N., Di Fulvio, F., Snäll, T., & Havlik, P. (2021). Material substitution between coniferous, non-coniferous and recycled biomass – Impacts on forest industry raw material use and regional competitiveness. Forest Policy and Economics, 132, 102588. https://doi.org/10.1016/j.forpol.2021.102588 Leclère, D., Obersteiner, M., Barrett, M., Butchart, S. H. M., Chaudhary, A., De Palma, A., DeClerck, F. A. J., Di Marco, M., Doelman, J. C., Dürauer, M., Freeman, R., Harfoot, M., Hasegawa, T., Hellweg, S., Hilbers, J. P., Hill, S. L. L., Humpenöder, F., Jennings, N., Krisztin, T., … Young, L. (2020). Bending the curve of terrestrial biodiversity needs an integrated strategy. Nature, 585(7826), 551–556. https://doi.org/10.1038/s41586-0202705-y Lesiv, M., Schepaschenko, D., Buchhorn, M., See, L., Dürauer, M., Georgieva, I., Jung, M., Hofhansl, F., Schulze, K., Bilous, A., Blyshchyk, V., Mukhortova, L., Brenes, C. L. M., Krivobokov, L., Ntie, S., Tsogt, K., Pietsch, S. A., Tikhonova, E., Kim, M., … Fritz, S. (2022). Global forest management data for 2015 at a 100 m resolution. Scientific Data, 9(1), 199. https://doi.org/10.1038/s41597-022-01332-3 Mishra, A., Humpenöder, F., Churkina, G., Reyer, C. P. O., Beier, F., Bodirsky, B. L., Schellnhuber, H. J., Lotze-Campen, H., & Popp, A. (2022). Land use change and carbon emissions of a transformation to timber cities. Nature Communications, 13(1), 4889. https://doi.org/10.1038/s41467-022-32244-w Mishra, A., Humpenöder, F., Dietrich, J. P., Bodirsky, B. L., Sohngen, B., P. O. Reyer, C., LotzeCampen, H., & Popp, A. (2021). Estimating global land system impacts of timber plantations using MAgPIE 4.3.5. Geoscientific Model Development, 14(10), 6467–6494. https://doi.org/10.5194/gmd-14-6467-2021 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 86 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 Pereira, H., Martins, I., Rosa, I., Kim, H., & Leadley, P. (2024). Global trends and scenarios for terrestrial biodiversity and ecosystem services from 1900 to 2050. https://doi.org/10.1126/science.adn3441 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 protected planet. (2025). Protected Areas (WDPA). Protected Planet. https://www.protectedplanet.net/en/thematic-areas/wdpa 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 Salles, J.-M., Teillard, F., Tichit, M., & Zanella, M. (2017). Land sparing versus land sharing: An economist’s perspective. Regional Environmental Change, 17(5), 1455–1465. https://doi.org/10.1007/s10113-017-1142-4 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. Schipper, A. M., Hilbers, J. P., Meijer, J. R., Antão, L. H., Benítez-López, A., de Jonge, M. M. J., Leemans, L. H., Scheper, E., Alkemade, R., Doelman, J. C., Mylius, S., Stehfest, E., van Vuuren, D. P., van Zeist, W.-J., & Huijbregts, M. A. J. (2020). Projecting terrestrial biodiversity intactness with GLOBIO 4. Global Change Biology, 26(2), 760–771. https://doi.org/10.1111/gcb.14848 87 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). Slater, M., & James, P. (2023). Low trophic species in aquaculture—Growth and research challenges. Journal of the World Aquaculture Society, 54(1), 4–6. https://doi.org/10.1111/jwas.12944 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). Stehfest et al. (2019). Key determinants of global land-use projections. https://www.nature.com/articles/s41467-019-09945-w 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 United Nations. (2025). UN Comtrade. https://comtradeplus.un.org/ Valenti, W. C., & Ballester, E. L. (2024). Integrated Aquaculture and Monoculture of LowTrophic Species. Fishes, 9(11), 450. https://doi.org/10.3390/fishes9110450 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. (2018, May 18). Brazil’s forests could fall victim to the US-China trade war. World Economic Forum. https://www.weforum.org/stories/2018/05/brazilforests-could-be-casualties-in-u-s-china-trade-war/ 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/ Wu, W., Hasegawa, T., Ohashi, H., Hanasaki, N., Liu, J., Matsui, T., Fujimori, S., Masui, T., & Takahashi, K. (2019). Global advanced bioenergy potential under environmental protection policies and societal transformation measures. GCB Bioenergy, 11(9), 1041– 1055. https://doi.org/10.1111/gcbb.12614 88 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 Zhao, J., Wei, X., & Li, L. (2022). The potential for storing carbon by harvested wood products. Frontiers in Forests and Global Change, 5. https://doi.org/10.3389/ffgc.2022.1055410 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 APPENDIX Stakeholder workshop for soy supply chain scenarios To support the model and scenario work, a two-hour online stakeholder workshop took place on May 14th 2025. Potential participants were identified through the CLEVER stakeholder mapping, and 31 participants responded positively to the invitation, of which 23 participated. Of the 23 participating stakeholders, 11 were from NGOs, 6 from the public sector (including both national government and international organizations), and 6 from the private sector. The majority of participating stakeholders were based in Brazil and Europe, and 2 were based in Northern America. The goal of the workshop was to gather input from stakeholders on relevant scenarios and modelling features for the soy sector and its impact on biodiversity, with a specific focus on Brazil-EU soy supply chains. The workshop consisted of: a) a 10 min welcome and introduction session, b) a 20min presentation from IIASA on model improvements, potential scenario options and preliminary results, followed by a Q&A, c) a first breakout session where experts were asked to share their vision for the soy sector by 2050 (4 breakout groups for 10min followed by a 10min reporting back in plenary, with note taking on the whiteboard), d) a second 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 (4 breakout groups for 10min followed by a 10min reporting back in plenary, with note taking on the whiteboard), and e) a 5min closing session. All participants were provided with the presentation and the whiteboard, as well as a link to a form for any additional feedback they wish to provide. Figure 44. Linkedin post posted after the workshop, which includes a screenshot of the white board, of IIASA presentation and of the zoom participants. 96 Figure 48 - Comparison of observed (IBGE PAM) vs simulated (GLOBIOM) trends in various aspects (harvested area, production volume, yield and share of soy-corn double cropping in total harvested area) of soy production systems over 2000-2020 at the biome level. For each aspect and biome, values are time horizonand sourcespecific (i.e., observed vs simulated) values normalized by the value for year 2000 in the observation dataset. Model improvements for forest supply chains Details on the modelling framework for forest supply chains can be found in the deliverable D6.2 while additional model improvements supporting this deliverable (i.e. inclusion of forest age structure, HWP and BECCS accounting, natural land division) are detailed below. Age-class dynamics and forest management Forest age-class dynamics in GLOBIOM works similarly than in large-scale area-based matrix models such as the European Forestry Dynamics Model (EFDM) (Packalen et al. 2023) or the European Forest Information Scenario Model (EFISCEN) (Schelhaas et al. 2007). The main difference between GLOBIOM and these models is that harvest decisions and age-class dynamics are based on economic optimization instead of statistical transition probabilities. In GLOBIOM, the choice over different management systems and age-classes is made separately for each grid. Each grid consists of multiple even-aged stands which are divided to different agecohorts. Biomass growth happens through age-class dynamics where after each period the stand area is moved to the next age-cohort. The biomass in each age-cohort follows S-shaped growth curves which are determined by the Chapman-Richards biomass growth model, as in (Humpenöder et al., 2014) and (Mishra et al., 2021). If the stand area is harvested or damaged by natural mortality, then it is moved in the first age-cohort. Since the size of the stand is not defined explicitly, the age-class dynamics can be interpreted either as even-or uneven aged managements. Figure 49 displays an example of age-class dynamics in GLOBIOM. In 2020, the carbon stock of forests is calibrated to (FAO, 2020) data and downscaled to grid level by utilizing G4M data and global age-class database (Besnard et al., 2021) . After 2020, the carbon stock is determined endogenously by the age-class dynamics. The carbon stock of forests is higher in 2100 than in 2020, because the estimated biomass growth exceeds the estimated biomass removals from the forests (harvests, mortality) during this period. 98 Figure 49. Global carbon stock of forests divided by age-classes in the BASE scenario (excluding primary forests) In GLOBIOM, forests are divided into three classes: 1) Plantation forests, 2) Seminatural forests and 3) Natural forests. Within each forest class there are different management systems as showed in Table 4. The current formulation of the model includes 9 forest management systems. Table 4. Forest management systems in GLOBIOM Management Description Forest type Management intensity Calibration CurC_H Coniferous plantation forest Plantation forest 3PGmix short rotation FRA2020+Lesiv2022 plantation forest area CurNC_H Non-coniferous plantation forest Plantation forest 3PGmix short rotation FRA2020+Lesiv2022 plantation forest area CurC Coniferous production forest Semi-natural forest G4M EU 75-100% ROW 50-100 % FRA2020+Lesiv2022 production forest area CurNC Non-coniferous production forest Semi-natural forest G4M EU 75-100% ROW 50-100 % FRA2020+Lesiv2022 production forest area CurC_M Coniferous multifunctional forest Semi-natural forest G4M EU 50-75% ROW 25-50 % FRA2020+Lesiv2022 production forest area CurNC_M Non-coniferous multifunctional forest Semi-natural forest G4M EU 50-75% ROW 25-50 % FRA2020+Lesiv2022 production forest area CurC_L Coniferous close-tonature forest Natural forest G4M EU 0-50% ROW 0-25 % WDPA categories IV-VI CurNC_L Non-coniferous close-tonature forest Natural forest G4M EU 0-50% ROW 0-25 % WDPA categories IV-VI CurS Strictly protected secondary forest Natural forest 0% WDPA categories I-III Cur0 Unprotected secondary forest Natural forest 0% Residual forest area 0 5 10 15 20 25 30 35 40 0-10 10_20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150+ PgC BASE 2020 0 5 10 15 20 25 30 35 40 0-10 10_20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150+ PgC BASE 2100 99 PriFor Primary forest Natural forest 0% FRA2020+Lesiv2022 primary forest area, EU Sabatini map Note: 1) Management intensity=% share of G4M maximum sustainable harvest potential based on NNP maps, local tree species and rotation time that maximizes increment given than carbon stock stays at the current level. 2) The model can choose between coniferous and non-coniferous management option in the limits of tree species distribution based on FRA2020 data (FAO, 2020). In the EU, the model use additional grid level data on tree species from (Brus et al., 2012). The transition from natural or semi-natural forests to planted forest happens through intensive management, i.e., removing old trees by harvesting and planting new ones. The transition from planted forests to natural or semi-natural forests happens through extensive management, i.e., removing old trees by harvesting or through natural mortality and waiting for new trees to appear by natural regeneration. The spatially explicit biophysical data for each grid is based on the G4M forest management simulation model (Kinderman et al. 2006,2008). G4M provides GLOBIOM estimates on initial biomass stock, growing stock and sustainable level of harvests (=increment). Using the G4M data and global age-class database (Besnard et al. 2021), the model generates grid level biomass growth curves, i.e., biomass and growing stocks for different stands within each grid. Within each grid the model can have different management systems with different age-class dynamics, mortality and harvest intensity. The initial areas of different management systems are calibrated to the FRA (2020) data, the World Database of protect areas data (protected planet, 2025) and the Global Forest Management Map (Lesiv et al., 2022). Moreover, for the historical periods 2000-2020 the management areas are matched to FAOSTAT country-level harvest volumes data (FAOSTAT, 2025). After 2020, the age-class dynamics develops endogenously based on periodic harvests volumes, growth curves, mortality and forest area changes (deforestation/afforestation). To maintain the sustainability of harvest volumes over time in the recursive dynamics, the model includes two additional constraints. First, for each grid cell, the amount of harvested biomass cannot exceed biomass growth. Second, it is not allowed to harvest age-classes that are younger than the optimal rotation time. Optimal rotation times for natural/seminatural forests vary between 3080 years depending on the management system and climate zone while for plantation forests between 5-30 years. Finally, the initial calibration of the model for the historical period 2000-2020 has been improved to better match better harvest volumes data (FAOSTAT, 2025), forest biomass stocks and carbon emissions data (FAO, 2020) and forest areas and land-use changes data (FAO, 2020). HWP and BECCS carbon accounting Harvested wood products (HWP) carbon accounting follows the IPCC production approach (PA), which is the most commonly used methodology to calculate HWP carbon storage (IPCC, 2019). The initial HWP carbon pool is calculated by using the production and trade data of HWP from FAOSTAT database from 1961 to 2020 (FAOSTAT, 2025). Based on this method the global HWP 100 pool was a sink of about 300 MtCO2/yr in 2020, which is comparable to global HWP pool estimates by (Johnston & Radeloff, 2019). The HWP carbon storage and sink are considerable smaller than to the forest carbon storage and sink, typically HWP carbon storage is about 5% of forest above ground carbon storage and HWP carbon sink about 10% of forest carbon sink (Zhao et al., 2022). There are two reasons for this. First, about 75% of tree biomass is lost for harvest and forest industry residues when woody biomass is moved away from forests and converted to HWP. Second, the lifetime HPW (0-100 years depending on the product) is typically shorter than the lifetime of trees (10-500 years depending on the management). In addition to HWP carbon storage, we also calculate bioenergy with carbon capture and storage (BECCS) carbon storage. The difference between them is that a half-life HWP vary from 0 to 35 years while BECCS has an infinite half-life, i.e., HWP are temporary carbon storages while BECCS is a permanent carbon storage. The share of bioenergy with BECCS is provided by the MESSAGE model and increases from 5% in 2030 to 75% in 2100 (IIASA, 2018). The BECCS carbon capture rate typically varies between 25-75% depending on the final energy carrier. We assume that BECCS carbon capture rate is 63%, which is an average value estimated from the MESSAGE model outcome. Deforestation and afforestation In historical periods 2000-2020, deforestation and afforestation areas are based on (FAO, 2020) country level data, which is allocated in the grid level according to GLOBIOM land-use dynamics. The land-use dynamics is based on land-use change costs, land suitability maps, production potential maps and demand for land on food, feed and fiber production (IBF-IIASA, 2023). After 2020, deforestation area is endogenous and depends on food and feed demand as well as carbon tax on deforestation emissions. Because carbon prices are relatively high in RCP1p9 scenario, there is very little deforestation after 2020 (Figure 50). Remark that conversion of natural forests to plantations is excluded in the model, i.e., the motive for deforestation in the model is food and feed demand instead of timber demand or some other reason (FAO, 2020). After 2020, afforestation area is based on G4M SSP2 RCP1p9 country level scenario data, which is allocated in the grid level according to GLOBIOM land-use dynamics. Afforestation competes with energy crops plantations, forest plantations and food production on the same marginal land areas. The difference between afforestation and forest plantations is that afforested areas are managed for carbon sequestration and nature conservation while forest plantations area managed for production use. There are also some differences in afforestation and plantations land suitability maps, i.e., some areas are suitable for afforestation but not for plantations (low productivity areas, deep slopes etc.). 101 Figure 50. Global deforestation and afforestation Other natural land division to abandoned land and natural land In GLOBIOM, other natural land is divided in two land-cover classes: 1) Abandoned land (AbdLnd), and 2) Natural land (NatLnd). AbdLnd is abandoned agricultural land, which typically has low biodiversity value and degraded carbon stock. NatLnd is old-growth natural vegetation land such as savannas, prairies and open forests (not classified as forests), which have high biodiversity value and carbon stock. AbdLnd is allowed to be converted to other land-use according to GLOBIOM usual land-use change constraints (land conversion costs + maximum available land) while NatLnd conversion to other land-use is limited by applying higher land conversion costs for NatLnd than for AbdLnd. Higher land conversion costs eliminate most of NatLnd conversion to other land uses. Since no good map of historically abandoned lands is available, we let the model decide the level of available AbdLnd for each period. Based on this method, the AbdLnd area increase from zero in 2000 to 270 Mha in 2040, and then starts decreasing depending on the scenario (Figure 51). In general, AbdLnd decreases over time, because it is converted to plantations, forests or agricultural land. On the other hand, some agricultural land area is abandoned due to changes in demand and/or productivity, which increases AbdLnd area over time. 0 100 200 300 400 500 600 700 800 900 2000 2020 2040 2060 2080 2100 Mha DfrLnd 0 100 200 300 400 500 600 700 800 900 2000 2020 2040 2060 2080 2100 Mha AfrLnd 102 Figure 51. Global abandoned land (AbdLnd) development relative natural land (NatLnd) in the baseline scenario Wood-based products trade and regional competitiveness The GLOBIOM model includes 22 traded wood-based products, which differ in units and level of value-addition (i.e., primary, semifinished, final). Remark that energy crops are not included in the traded products in the model. Reason is that energy crops are low value-add products, which are usually not transported longer distances or traded between the countries. For simplicity, we do not consider here individual wood-based products bilateral trade flows, but only total net exports (i.e. difference between export volume and import volume aggregated over all products). The volume of total net exports reflects regional woody biomass resources, production costs and forest industry development state, and can be used as an indicator of regional economic competitiveness. Aggregation over different wood-based products requires that they are converted in comparable units. There are different methods to convert wood-based products to comparable units (Lauri et al. 2021). Here we use here so called roundwood equivalent (RWeq) units, which measure wood-based products according to the amount of primary biomass needed for their production. This method is standard in the forest sector analysis and recently applied also in the agricultural sector analysis (Zhao et al. 2025). 0 500 1000 1500 2000 2500 3000 3500 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Mha NatLnd AbdLnd 103 Model improvements for aquaculture and aquafeed supply chains Projecting the capacity of unfed aquaculture beyond 2020 As compared to the description of the BAU scenario in D6.2, and in accordance with the feedback from the stakeholder during the workshop, for all aquaculture and aquafeed scenarios in this deliverable (except for UNFED20), we assume that the capacity of unfed aquaculture can increase beyond 2020 levels. To constrain the post-2020 capacity of unfed aquaculture, we extrapolated at countryand product-level historical trends using a Generalized Additive Models. Future trends in capacity are provided in Figure 52. Figure 52 - Projections for the capacity of unfed aquaculture by region and product Sensitivity analysis of the crop composition of crop aquafeed requirements For all aquafeed and aquaculture scenarios except for the AFCOMPOCROPMIX scenario, we assumed the proportion of various crops (between soya, corn, rape and wheat) in total crop aquafeed demand to be country-specific (based on Frohlich et al 2018), but invariant across time horizons and scenarios. As this might not necessarily be a valid assumption, and might affect the potential of impact of crop aquafeed requirements on land use and biodiversity, we tested the impact of an alternative assumption in a simple scenario. In this alternative scenario (AFCOMPOCROPMIX), we focused on the dominant fed aquaculture production system (fed aquaculture in China), for which we assumed that 50% of the crude protein content from corn (which has relatively lower crude protein content) and soya (which is imported) feed requirements per unit of aquaculture output are replaced by a similar crude protein content from wheat. To do so, parameters for corn and soya aquafeed requirements where halved, while that of wheat was increased by the amount 104 necessary to compensate this loss in terms of crude protein intake (while accounting for differences across crops in crude protein content and apparent digestibility, see Table 5). Table 5 - Nutritional parameters of individual crops as aquafeed. Source: INRAE-CIRAD-AFZ Feed tables (https://www.feedtables.com/, accessed July 2024). CROP Crude protein content (gram crude protein per gram of freshmatter) Apparent digestibility (fraction) Corn 0.216 0.92 Soya 0.404 0.912 Wheat 0.404 0.95