Report on improved and upscaled understanding of how behaviour and behavioural change can support transformation to more sustainable socio-ecological forest systems and better biodiversity status at multiple spatial levels. BioConsent project Deliverable 3.2 / 3.3.

Project number: BiodivRestore 559 Acronym: BIOCONSENT Project title: BIOCONSENT - Decision-making Support for Forest Biodiversity Conservation and Restoration Policy and Management in Europe: Trade-offs and Synergies at the ForestBiodiversity-Climate-Water Nexus Report on improved and upscaled understanding of how behaviour and behavioural change can support transformation to more sustainable socio-ecological forest systems and better biodiversity status at multiple spatial levels. (Deliverable 3.2 / 3.3) Authors: Verkerk, P.J., Aquilué, N., Beland-Lindahl, K., Felton, A., Holmtröm, H., Jo, H., Johnstone, C.P., Kindermann, G., Krasovskiy, A., Kraxner, F., Maximo, Y. I., Nilsson, J., Park, E., Pecurul, M., Shchepashchenko, D. Project coordinator: Project co-coordinator: M. Sotirov (ALU-FR, Germany) K. Beland-Lindahl (LTU, Sweden) SUGGESTED CITATION Verkerk, P.J., Aquilué, N., Beland-Lindahl, K., Felton, A., Holmtröm, H., Jo, H., Johnstone, C.P., Kindermann, G., Krasovskiy, A., Kraxner, F., Maximo, Y. I., Nilsson, J., Park, E., Pecurul, M., Shchepashchenko, D., 2025. Report on improved and upscaled understanding of how behaviour and behavioural change can support transformation to more sustainable socio-ecological forest systems and better biodiversity status at multiple spatial levels. BIOCONSENT Project deliverable report 3.2/3.3. Zenodo. https://doi.org/10.5281/zenodo.17987566 FUNDING ACKNOWLEDGEMENT The BIOCONSENT project was funded through the 2020-2021 Biodiversa+ and Water JPI joint call for research projects, under the BiodivRestore ERA-NET Cofund (GA N°101003777), with the EU and the funding organisations Academy of Finland (AKA, Decision number 351884), Agencia Estatal de Investigación Spain (AEI), Austrian Science Fund (FWF), Bulgarian National Science Fund (BNSF), German Federal Ministry of Education and Research (BMBF), and the Swedish Environmental Protection Agency (SEPA). Disclaimer The information in this document is provided “as is”, and no guarantee or warranty is given that the information is fit for any particular purpose. The above referenced authors shall have no liability for damages of any kind including without limitation direct, special, indirect, or consequential damages that may result from the use of these materials subject to any liability which is mandatory due to applicable law. The content of this deliverable does not reflect the official opinion of the European Union. Responsibility for the information and views expressed lies entirely with the author(s). Page 3 Index Index ........................................................................................................................................................ 3 1. Introduction ..................................................................................................................................... 4 2. Methodology ................................................................................................................................... 5 2.1. Models ................................................................................................................................... 5 2.2. Scenarios ................................................................................................................................ 5 3. Germany .......................................................................................................................................... 7 3.1. Methods ................................................................................................................................. 7 3.2. Scenario results and interpretation ..................................................................................... 11 3.3. Main findings ....................................................................................................................... 20 4. Spain .............................................................................................................................................. 21 4.1. Methods ............................................................................................................................... 21 4.2. Scenario results and interpretation ..................................................................................... 26 4.3. Main findings ....................................................................................................................... 31 5. Sweden .......................................................................................................................................... 33 5.1. Methods ............................................................................................................................... 33 5.2. Scenario results and interpretation ..................................................................................... 37 5.3. Main findings ....................................................................................................................... 41 6. European Union ............................................................................................................................. 43 6.1. Methods ............................................................................................................................... 43 6.2. Scenario results and interpretation ..................................................................................... 46 6.3. Main findings ....................................................................................................................... 54 7. Synthesis ........................................................................................................................................ 57 References ............................................................................................................................................. 58 Annex I: Scenario storylines for the Spanish case study ....................................................................... 61 Scenario 1 (Multifunctional / BAU) ................................................................................................... 61 Scenario 2 (Closer-to-nature management)...................................................................................... 61 Page 4 1. Introduction Despite ambitious policy targets at the global and EU levels, biodiversity is under increasing threat (Leclere et al. 2020; Pereira et al. 2024). Ambitious policy targets have been set to halt the loss of biodiversity. Still, effective implementation depends, among other factors, on supportive behavioural responses from forest owners and conservation managers who must respond to multiple policy and socio-economic drivers, forcing them to make decisions and trade-offs while dealing with complexity and uncertainty (Deuffic et al. 2018). Previous research suggests that cross-sectoral goal conflicts, as well as failures to understand behavioural responses, constitute major barriers to achieving desired forest biodiversity outcomes (Beland Lindahl et al. 2017; Sotirov et al. 2019). The complex nature of human decision-making has been the focus of studies for decades across several themes, including environmental modeling (Groeneveld et al., 2017). Many socio-ecological factors shape forest management decisions and lead to different behavioural responses. There are various types of forest owners and managers, hereafter collectively called ‘agents’, with a range of objectives, preferences, and behaviours that influence forest management decision-making. It is crucial to consider these differences in behaviour and behavioural change in forest management decision support tools when evaluating the impact of policies, market drivers, and conservation goals on forest environments (e.g., see Sotirov et al., 2019). Maximo et al. (2025) recently described how certain biophysical forest models have been improved in representing the behaviour and behavioural change of forest owners and managers. Rather than adding human agency to forest models through scenarios, Maximo et al. (2025) integrated human behaviour and behavioural change into four forest simulation models. They did this based on a survey conducted by Sotirov et al. (2025a) and by analyzing the main management practices used by different forest owner and manager typologies (agents) and the factors influencing their decision-making. Building on the work by Maximo et al. (2025), the current study aimed to examine how behaviour and behavioural change can support improvements in forest biodiversity status and, more generally, in more sustainable socio-ecological forest systems, including synergies and trade-offs, by quantitatively assessing the outcomes of selected policy and management scenarios. Page 5 2. Methodology 2.1. Models In this study, we build on four forest models (EFISCEN-space, FORMES, HEUREKA and G4M) that have been improved by Maximo et al. (2025) with regard to the representation of behaviour and behavioural change of forest owners and managers for case studies across Europe (Table 1.1). For a more detailed description of the models and the improvements made to them, we refer to Maximo et al. 2025 or the references given in Table 1.1. Table 1.1. Overview of models used in this study (updated from Maximo et al 2025). Germany Spain Sweden EU Model EFISCEN-space FORMES HEUREKA G4M + FLAM Reference Schelhaas et al. 2022 Trasobares et al. 2022 Lämås et al. 2023 Kindermann et al. 2013 Spatial resolution and coverage Plot / regional, BadenWürttemberg and North Rhein Westphalia Regional, Catalonia Regional, Norrbotten County Grid, EU-27 Temporal coverage ~2020-2100 2020-2050 ~2020-2120 (or shorter) ~2020-2050 Forest and ecosystem service indicators • growing stock • wood removals • carbon stocks and storage • growing stock • wood removals • carbon stocks and storage • growing stock • wood removals • carbon stocks and storage • annual growth • growing stock • wood removals • carbon stocks and storage • burned areas Biodiversity indicators* • Share of forest with unevenaged structure • Stock of organic carbon • Share of forest dominated by native tree species • Tree species diversity • Tree microhabitats • Deadwood • Share of forest dominated by more than one tree species • Tree species diversity • Share of forest with unevenage structure • Deadwood • Species richness • Broadleaf forests • Veteran trees • Standing deadwood • Lying deadwood • Share of forest dominated by native tree species • Tree species diversity *Biodiversity indicators have been identified based on indicators specified in the EU Nature Restoration regulation (article 12 and Annex V1). 2.2. Scenarios Three contrasting explorative “what-if” future policy scenarios at the EU-level were developed (Sotirov et al., 2025b). The policy scenarios envisioned how forest policy could evolve by 2050 and described both end-states and the processes leading to them. Each scenario describes alternative developments in social, technological, environmental, economic, and policy factors, which resulted in policy pathways with different priorities for regulatory vs. economic instruments (Sotirov et.al., 2025b). The three scenarios can be summarized as follows (Haase et al. in prep.): Page 6 • Biodiversity First: The EU prioritizes biodiversity conservation and ecosystem restoration with strict forest protection laws and a shift towards close-to-nature forestry. 30% of EU forests are protected and clear-cutting is banned on all lands, promoting natural regeneration and increased forest connectivity. Carbon sequestration focuses on storing carbon in forests rather than harvested wood products. Timber production is reduced, but alternative payments for ecosystem services (e.g., carbon storage, recreation) support forest owners. • Multifunctionality: The EU balances economic growth, biodiversity protection, and carbon neutrality, creating forests that provide multiple ecosystem services. Sustainable timber production, selective logging, and reforestation coexist with protected areas and carbon sequestration goals. Policies mix regulations with market incentives, supporting both forest conservation and renewable energy from bio-based materials. Bioenergy, timber, and biodiversity goals are pursued simultaneously, aiming for economic stability while maintaining ecological integrity. • Wood Bioeconomy First: The EU prioritises economic growth and energy security with forests managed intensively for timber, bioenergy, and bioplastics. Clear-cutting, fast-growing monocultures, and genetically modified trees are widely used to meet increasing global demand. EU policies focus on boosting domestic timber production, reducing imports, and expanding the biomass energy sector aimed to drive the forestbased economy. These policy scenarios have been interpreted and implemented in the four models to quantify. and assess impacts on key forest variables, biodiversity and ecosystem services (see Table 1.1). The implementation of the scenarios in each forest model is described in chapters 3-6. The forest dynamics were simulated until 2050 and 2100 for two climate scenarios, i.e., Representative Concentration Pathways 4.5 and 8.5 (van Vuuren et al. 2011). RCP4.5 is considered an intermediate scenario of the expected changes in radiative forcing, leading to moderate greenhouse gas emissions and therefore a slight increase in global surface temperatures and the side-effects of climate change on ecosystems. RCP8.5 is the most extreme scenario as it is assumed that greenhouse gas emissions will continue to steadily rise throughout all the 21st century. It leads to the major increases in temperatures. Page 7 3. Germany 3.1. Methods 3.1.1. Scenarios and implementation The scenario implementation for the German case study was built on the approach described by Maximo et al. (2025) for Baden-Württemberg and North Rhine-Westphalia, summarized in Figure 3.1. Maximo et al. (2025) developed (i) a model to spatially allocate agent typologies across the forest landscape for the two regions; (ii) a model to predict current forest management; (iii) a model to predict changes in forest management in favour of biodiversity; and (iv) a simplified equation to predict the occurrence of behavioural change. Models ii and iii rely on forest structure as input parameters, but also consider information on forest ownership and forest owner typology. Together, these components provided the stepping stones for predicting forest management practices, which served as the basis for simulating future forest resources using the EFISCEN-Space model. Figure 3.1. Process flow diagram for the German case-study (Maximo et al., 2025). In this case study, we combined EFISCEN-Space with the models by Maximo et al. (2025) to assess the outcomes of the Biodiversity First policy scenario (see Chapter 2). The Multifunctionality and Wood Bioeconomy First scenarios were not considered, as survey data on owner responses were not available to parameterize these scenarios. Instead, we Page 8 compared the Biodiversity First policy scenario with a Baseline management scenario, which is based on the continuation of harvest practices according to repeated forest inventory observations. In addition, we included a Current management scenario, which is based on stated harvest practices and tree species selection, as derived from the survey by Sotirov et al. 2025a. The three scenarios are described below and a summary is provided in Table 3.1. Table 3.1. Summary of the scenarios covered in the German case study. Scenario label Climatechange scenario Cutting regime Tree species selection / regeneration method Baseline 4.5 RCP4.5 Observed harvest regime retrieved from EFISCENSpace Management Rule Database. The harvesting regime is defined by forest structure, with clear-cutting or partial cutting (i.e., group or single-tree selection). Forest regeneration is only based on natural regeneration. Baseline 8.5 RCP8.5 Current 4.5 RCP4.5 Stated harvest management, according to survey responses for current forest management practices (Figure 2, A). Harvest is primarily based on single-tree selection, targeting trees with larger diameters. Forest regeneration is mostly based on natural regeneration. Additional tree species are introduced into the stand in accordance with the stated planting activities. Current 8.5 RCP8.5 Biodiversity 4.5 RCP4.5 Stated management, according to survey responses for management practices to improve biodiversity (Figure 2, B). Harvest is mainly based on single tree (targeting bigger trees) and group selection (no specific diameter target). Forest regeneration is mostly based on natural regeneration. Additional tree species are introduced into the stand in accordance with the stated planting activities. Biodiversity 8.5 RCP8.5 To implement the Biodiversity First policy scenario, the drivers defined in this scenario were compared to the surveyed 26 socio-economic factor influences decision-making for forest management (see chapter 3 Maximo et al. 2025 for details). Table 3.2 shows how the drivers in the “Biodiversity First” policy scenario were linked with corresponding socio-economic factors affecting the behaviour of agents. Climate change was also considered a driver in the Biodiversity First policy scenario, being its effects considered separately by applying EFISCEN-Space for two climate scenarios, as described in Chapter 2. Page 9 Table 3.2. Biodiversity First policy scenario (Sotirov et al. 2025b) and interpreted socio-economic factors (Sotirov et al. 2025a). Drivers Driver setting Socio-economic factors Societal changes Environmentally concerned stagnating society with reduced consumption NA Economic changes Low traditional economic growth with low energy and low material (biomass) consumption Item 8: Timber prices Policy changes Priority for nature conservation and climate protection related forest policy at EU and national level mainly governed by regulatory tools, and partial support by subsidies Item 2: Regulatory biodiversity policy (objectives, targets, standards in law and bylaw) Item 5: Economic instruments (subsidies, compensation payments, taxes) Forest management prescriptions Promotion of forest biodiversity conservation/restoration and carbon forest sinks objectives Item 6: Informational instruments (advisory services, knowledge, research, know-how transfer) Item 17: Advice from a consultant, managing company or forest owner association that I am member of Based on the mapping in Table 3.2, the behaviour change equation was used to estimate the probability that an agent, belonging to a specific typology group, decides to implement changes in their forest management according to the policy driver scenario. This equation relies on a combined probability distribution, utilizing the survey’s average responses by agent typology regarding the importance of each socio-economic item for forest management decisions (see chapter 3 by Maximo et al., 2025, for details). In practice, applying the behaviour change equation in this study serves as a theoretical approach, as the connection between socioeconomic factors, policy drivers, and forest management forms a complex system. The developed function only offers a simple mathematical representation of this reality. Additionally, since survey respondents indicated high scores for most of the socio-economic factors covered, the probability of targeting a change is always very high. When applying the behaviour change equation to the Biodiversity First policy scenario and the five socio-economic factors shown in Table 3.2, it resulted in approximately a 99.9% probability of change for all agent typologies. Figure 3.2 shows the proportion of assigned harvesting practices among the simulated plots. On the left-hand (A) the stated management prescriptions for current conditions are shown and on the right-hand (B) the potential changes toward biodiversity gathered in the survey. Page 16 conduct single-tree selection cutting in their forest (Maximo et al., 2025; Figure 3.2). Fourthly, respondents may not have fully understood the forest-related terminology used in the survey. Overall, we observed similar results for the Current and Biodiversity scenarios, both of which are based on stated management, as indicated by the survey results. Our findings for these scenarios generally indicated an improvement in biodiversity indicators, except for the forest dominated by native tree species. This indicator measures the share of stands where the dominant species is considered native for Germany. The dominant species was identified as the species with the highest summed basal areas on the plot. Although most trees in the plots are native, this likely explains why, during the early years of the simulation, the Current and Biodiversity scenarios experienced a sudden decline in plots dominated by native trees. These scenarios used more group selection harvesting, which does not target specific diameters and therefore increases the likelihood of removing native trees compared to exotic ones. Conversely, the Baseline scenario did not exhibit such a sharp decline, likely because it employed more intensive harvesting methods, such as clear-cutting, primarily in plots dominated by exotic species. This indicates the importance of the type of harvests (and the species and tree size that are harvested), and not only the amount of harvesting. For some indicators (tree species diversity and the share of veteran trees), differences between scenarios were time dependent. Tree species diversity was lowest in the Baseline scenario for most of the simulation period. Only by the end of the 2100 simulation period did tree species diversity in the Baseline scenario exceed that in the Current and Biodiversity scenarios. This result may be explained by larger harvest levels in the Baseline scenario, which stimulate regeneration and allow ingrowth by a larger set of tree species. The opposite effect was found for the share of veteran trees, which was generally highest in the Baseline scenario and decreased towards the end of the simulation period. Apart from the indicators related to forest ecosystems and biodiversity, we observed significant differences in forest structures between the scenarios (Figure 3.8). Similar patterns appeared for both climate change scenarios, where the number of trees per hectare in the Baseline scenario is noticeably lower compared to the other scenarios. Likewise, the median diameter of the forests seems to decrease due to initially higher harvest levels. However, while the median decreases for the Baseline, it increases for the Biodiversity scenario and remains unchanged for the Current scenario. This may be explained by the reduction in clear-cutting activities between the two specified scenarios. These differences in forest structure characteristics affect the performance of the different indicators. For example, according to Courbaud et al. (2022), the likelihood of some microhabitats (e.g., breeding-woodpecker-hole, rot-hole, or root-concavity) occurring in a tree increases with its diameter at breast height. Therefore, this group of tree microhabitats is more common in the Current and Biodiversity scenarios. Bark loss or dendrotelm are other tree microhabitats, but they are more common on smaller-diameter trees. It is important to note that stands with more trees have a higher probability of supporting tree microhabitats. However, the difference in this indicator's performance across the three scenarios is very small. Page 17 Figure 3.6. Forest structure across simulated years for both regions. The values shown in Tree per hectare graph represents the mean of all simulated plots results, while the second graph represents the median of DBH distribution. The diameter distribution of the stands also appears to influence harvesting performance. In the initial years, harvesting in the Baseline scenario is relatively higher than in the other two; however, over time, they become more similar (Figures 3.4-3.7 and 3.9). Although stands in the Baseline scenario are generally more likely to undergo clear-cutting, the overall probability (averaged across species and biogeographical regions) of such events remains relatively low at approximately 6%. Since thinning, single-tree selection, and group selection follow the same management rules and differ only in their target DBH, the volume of wood removed seems similar. However, a significant difference can be observed in the number of trees harvested, which is notably higher in the Baseline and lower in the Current and Biodiversity scenarios. This difference is likely because the Current and Biodiversity management options typically target larger diameters compared to the Baseline scenario. Additionally, the reduction in diameter distribution over the years in the Baseline scenario means that more trees need to be harvested to meet the same harvest volume. This characteristic may influence the share of forest with an uneven-aged structure indicator, which performs better (approximately 10%) in the Baseline scenario. This is probably due to the higher harvesting intensity in the Baseline scenario, where harvesting more trees promotes stands with greater DBH diversity and facilitates regeneration, whereas low harvesting levels can lead to stagnation of the stand structure. Page 18 Figure 3.9. Average of trees harvested across all simulated plot over the years for both regions. Regarding the share of veteran trees, the Baseline scenario has also performed better compared to the Current and Biodiversity scenarios. This could possibly be explained by the fact that, while the Baseline harvesting regime typically targets all DBH trees, the Current and Biodiversity targets mainly focus on larger DBH trees. The higher harvesting intensity in the Baseline scenario can also promote the development of remaining trees, helping median trees grow larger faster (i.e., veteran trees). By the end of the simulation years, it is apparent that the stated management scenarios reach or surpass the observed Baseline. In Figure 3.107, the spatial distribution of the growing stock is shown. As shown in Figures 3.43.7, the growing stock is higher on the Current and Biodiversity scenarios, also it is possible to observe that over the different scenarios the same pixels seem to be following the same relative tendency. No distinct spatial trends can be identified at this resolution across the regions, although North-Rhine Westphalia presents greater variability in growing stock, whereas Baden-Württemberg seems more homogeneous. Page 19 Figure 3.10. Maps showing growing stock for RCP4.5 in the year 2100. The three panels show wood volume average by the plots included in the pixel size of 0.1°.7 Page 20 3.3. Main findings In this case study, we integrated forest owner typologies, socio-economic drivers, and management practices to estimate the future development of forest ecosystems in North-Rhine Westphalia and Baden-Württemberg. Forest owners and managers did not appear to make major changes in their management in our Biodiversity scenario compared to their current management in the Current scenario. This is likely because they already reported implementing biodiversity-friendly practices in their current management. Overall, the results for our scenarios suggested potential future improvements in biodiversity, but the extent of these improvements depended on the specific scenario and timing considered. This indicates the importance of the type of harvest (and the species and tree size harvested), rather than the amount of harvesting alone. Moreover, we generally found a stronger effect of management on the indicators included in this case study than of climate. Given the relatively significant differences observed between the Baseline (i.e., observed management from repeated forest inventory measurements) and Current (i.e., stated management in the survey) results on forest structure variables, it is important to investigate the underlying causes. Several explanations were proposed in this study; however, further research is needed to identify and address the issue properly. Page 21 4. Spain 4.1. Methods 4.1.1. Scenarios and implementation In this case study, we used the FORMES multi-objective forest planning system to evaluate the results of the Biodiversity First policy scenario described in Chapter 2. We did not include the Multifunctionality and Wood Bioeconomy First scenarios because data on owner responses for these policy types were unavailable to parameterize them. The qualitative scenarios by Sotirov et al. (2025b; see section 2.2) have been updated based on the survey results (Sotirov et al. 2025a) concerning the key factors for forest owners when adopting different silvicultural practices (Maximo et al. 2025). The revised scenario description is available in Annex 1. Table 4.1. Matching between drivers and their settings in the Biodiversity First policy scenario (Sotirov et al. 2025b) and socio-economic factors (Sotirov et al. 2025a). Drivers Driver setting Socio-economic factors Societal changes Environmentally concerned stagnating society with reduced consumption NA Economic changes Low traditional economic growth with low energy and low material (biomass) consumption NA Policy changes Priority for nature conservation and climate protection related forest policy at EU and national level mainly governed by regulatory tools. partial support by subsidies Subsidies for natural evolution Subsidies for forest management Human and technological changes Availability of labour Joint technical management and forest improvement plans Innovations that facilitate forest exploitation For each agent typology, we explore whether each of socio-economic factors had an influence on the probability of changing forest management, represented in an additional question posed in the survey (Q8.2) – where we asked respondents their willingness to change their management practices to improve biodiversity in their forests following technical advice. There were no significant results based on the chi-square test or the Fisher test. These results lead us to change the original plan on applying (Table 4.1. and (Sotirov et al. 2025b) where socio-economic factors related to the scenarios. Instead, new scenarios were defined only responding to survey (Q8.2), willingness to change their management practices to improve biodiversity in their forests following technical advice. Again, the results revealed that willingness to change is not strictly determined by typology. All four owner groups (i.e., environmentalists, traditionalists, optimizers, and multifunctionalists) include individuals open Page 22 to modifying their practices (see next section). As a result, the simulation was done for each forest inventory plot covered in the survey using the following implementation logic (Figure 4.1.) Figure 4.1. Forest management scenarios and implementation SCENARIO BAU includes all the respondents of the survey. This scenario assumes that their management follows the silvicultural prescriptions are based on the Guidelines for Sustainable Forest Management in Catalonia (ORGEST) developed for the main tree species present in the region 1 . The ORGEST series constitute a set of technical tools to help forest management planning. They collect a series of decision elements, models and management recommendations, adjusted to Catalan conditions, which constitute a body of practical and upto-date information on forest management, as described by Maximo et al. (2025) and their Table 16. They aim to support the manager in the decision-making process regarding the allocation of preferential objectives and the planning and execution of management actions. SCENARIO C2N includes the respondents that indicated the willingness to adopt Closer to nature practices (52.05%) and the rest, which continue with BAU. For this 52.05% a new prescription (BIO-ORGEST) to adopt close-to-nature principles was applied (Beltrán et al. 2020, Baiges et al. 2023). These BIO-ORGEST are widely accepted by technicians and researchers, as described by Maximo et al. (2025) and their Table 17. From the management decisions, we implemented changes in the harvesting methods to link a specific agent (regardless of its type) to a change in prescription in their associated inventory plot. SCENARIO NOMAN represents a no-management scenario. Here, we hypothesize that all forests in Catalonia were left to natural evolution. This last scenario is relevant for at least three reasons: i) to contrast the indicators from the previous scenarios to a no-management option; ii) natural evolution as an alternative to active-management approaches existing in conservation policies and advocated for re-wilding; iii) forest and land abandonment as one of the challenges in Catalonia. Currently, only a 30% of forests account with a formal forest management plan 2 . In this case, the simulations no not contemplate any management prescriptions. For all scenarios, forest dynamics were simulated from 2021 to 2050 (3 decades) for the climatic scenarios Representative Concentration Pathways 4.5 and 8.5. We relied in the 1 https://cpf.gencat.cat/es/cpf_actualitat/cpf_publicacions/cpf_colleccions/cpf_orientacions_gestio_forestal_sostenible/ 2 https://www.observatoriforestal.cat/planificacio-forestal/Idescat 2023 Page 23 climatic projections of the global circulation model MPI‐ESM‐LR downscaled with the RCA4 regional model for Spain. We had access to daily climatic projections for the meteorological stations of the Meteorological Service of Catalonia, that were originally obtained from the CORDEX program of COPERNICUS (https://climate.copernicus.eu/climate-projections). We applied a spatial interpolation approach to obtain the future series of climate variables required by FORMES, based on climate data from COPERNICUS, the topographic variables from PNOA, and using the methods included in the R-package meteoland (De Cáceres et al. 2018). 4.1.2. Agent typology and willingness to change. Table 4.2 shows, by agent typology, the willingness to adjust practices such as thinning, species selection, regeneration, or specific biodiversity measures. For instance, 51% of environmentalists and 45% of optimizers reported a willingness to change their harvesting methods. Biodiversity measures also receive strong support, particularly among environmentalists and traditionalists. These results underscore that, despite their different profiles, all typologies are inclined to reconsider certain aspects of their management, especially those related to ecological resilience and stand-level improvements. This overall openness provides a solid foundation for extension services to promote closer-to-nature management practices. Table 4.2. Percentage of respondents for each agent typology willing to change their management to improve biodiversity. Agent typology Change in cutting method Change in thinning Change in species selection Change in regeneration Change in biodiversity TRAD 48.2 62.5 50.1 48.2 57.1 OPTI 45.8 70.8 54.2 56.2 43.8 MULTI 43.6 62.8 52.1 53.2 51.1 ENVI 51.0 69.4 57.1 59.2 55.1 We also check if there was a significant relation between the change of cutting method and any of the socio-economic factors we asked about in the survey. Figure 4.2 shows the plots for this analysis and we can see how those forest owners with a non-inherited property, lower studies, members of a forest owners’ association and younger want to change their cutting method than the other respondents. Although these small differences, there were not significative based on chi square test or fisher test. Page 24 Figure 4.2. Characterization of the willingness to change depending on respondents’ socio-economic factors Page 25 4.1.3. Model initialisation data We aimed at simulating forest evolution for each owner answering the survey under a climate change scenario and according to different alternative management scenarios. To simulate forest dynamics adopting different management rationalities. we used the FORMES projection system for multi-objective forest planning (see D3.1). FORMES is based on a series of empirical, climate-sensitive, individual-tree models to simulate forest stands dynamics (Trasobares et al. 2022). The forest inputs of the FORMES models are NFI-like data. That is, a set of forest stands or plots for each of them describing their composition and structure via four variables: tree species, diameter at the breast height (cm), height (m), and density (trees/ha). Additional plot-level data required by the FORMES model are topography (elevation (m), slope (º), and aspect (º)), area in hectares (optional parameter), climatic variables (mean annual temperature (ºC), annual precipitation (mm), potential evapotranspiration (mm), and solar radiation (MJ/m2)). and soil water holding capacity (mm). We initialized as follows each of these plotand stand-level data. Using the publicly available coordinates of the Spanish NFI plots, we extracted the three topographic variables required from the Digital Terrain Model at 25 m of spatial resolution (DTM25). The DTM25 has been obtained by interpolating 5-m resolution digital terrain models derived from the interpolation of the terrain class from PNOA LiDAR flights (https://pnoa.ign.es/web/portal/pnoa-lidar/modelo-digital-del-terreno).Topographical variables are used by FORMES model as explanatory variables in the individual-tree models of growth, mortality, and ingrowth. Finally, the soil water holding capacity was described based on soil physical properties. such as texture (i.e., volume percent of sand. silt and clay), organic matter content, bulk density and rock fragment content corresponding to plot coordinates obtained from the SoilGrids global database at 250 m resolution (Hengl et al. 2017). 4.1.4. Agent allocation To initialize the forest data for each owner with complete survey responses (234), we assigned one or more plots from the 4th Spanish National Forest Inventory (sampled between 2016 and 2017 in Catalonia) based on the ownership location (either the municipality or, if not available, the county), the property right type (public or private), whether it has a protection status, whether the forest is considered mono-species or multi-species, and the main tree species (up to three). This process resulted in 1107 NFI plots used to describe the current composition and structure of the forest stands analyzed. Examining the distribution of dominant species at the ownership level (Figure 4.1), we observe that pine species are the most common (as dominant species). They range from 40% of the OPTI ownerships to 62% of the ENVI. Overall, Quercus species are the second most common dominant species, especially notable among optimizers, with 49% of these forests dominated by Quercus, compared to 40% by pine. Minor species or genera in Mediterranean forests, such as birch, Douglas fir, or balsam fir, are not even present as dominant species in the forests of optimizers. Forests belonging to multifunctionalists are the most diverse, represented by up to seven different genera or 13 dominant species (Figure 4.3). Management is simulated in FORMES based on tree-specific silvicultural prescriptions that are applied according to rules dictating when thinning occurs, what type of thinning is used, how much wood is removed, and the minimum DBH for final cuts during regular management. More details on how management is implemented in FORMES are provided by Maximo et al. (2025). Page 32 friendly practices. This is followed closely by optimizers (OPTI) owners, especially in thinning and regeneration. These behavioural patterns are influenced by key drivers such as infrastructure and transport, profitability of forest management, and forest-specific factors like age structure, productivity, and natural disturbance risks. Looking ahead to 2100, the scenario analysis reveals important trade-offs. Business-as-usual management (BAU) cannot sustain initial levels of wood removals and shows a lower increase in carbon stocks. Annual growth patterns are negative over the 80-year periods regardless of the climatic scenario analysed, resulting in less diversity in both forest structure and tree community composition. By contrast, no-management (NOMAN) maximizes carbon stocks and growing stock, and enhances uneven-aged structures, though it provides no timber output and its annual growth volume is systematically lower than in managed scenarios. The mixed approach, BAU-C2N, where close-to-nature practices coexist with conventional management, achieves the best balance: carbon stock increases beyond business-as-usual, growing stock shows the best trends compared to the other two management approaches, species diversity reaches levels similar to the NOMAN scenario, and structural heterogeneity is promoted—all while increasing wood supply. It is important to note that within an 80-year period (between 2020 and 2100), it is already possible observe major changes in forest indicators, such as changes in forest composition or structure, since processes like species diversification, regeneration, and structural transformation unfold slowly over time. However, the management strategies applied do not explicitly target creating new mixtures, planting new species or combinations of them. The steady increase of carbon and growing stocks has to be read with caution, as the performed simulations do not include natural disturbances (such as fires or pest outbreaks) that could largely modify the stocks of living biomass, even more in a climate change context that entails increasing risk and impact of disturbances. Overall, the findings suggest that encouraging greater adoption of close-to-nature management (especially among owners already inclined toward environmental or optimization goals) could improve biodiversity and carbon storage without fully sacrificing economic returns. This offers a promising way for sustainable forest management. Page 33 5. Sweden 5.1. Methods 5.1.1. Scenarios and implementation In Sweden, the leading national forest management decision support system Heureka (Lämås et al. 2023) was used to simulate and evaluate the expected shortand long-term consequences of forest owners’ behavioral adaptations to three different policy scenarios. First, a reference scenario (referred to as BAU) was derived, simulating the ongoing forestry in the current study area and applying the climate scenario RCP4.5 (as in the three alternative policy scenarios). Other climate scenarios, like RCP8.5, are available in Heureka. However, applying these climate scenarios comes with a disclaimer since the downsides with a warmer climate, mainly comprehensive forest damages like fires and floodings (currently followed by landslides), are not modelled or available for simulation in Heureka. Common for all simulated scenarios are also: • The use of a discount rate of 2%. • A time horizon of 100 years represented by 20 5-year periods where forest management activities (harvest operations etc.) are simulated in the middle of each 5-year period. • The use of a model optimizing the net present value with respect to even-flow constraints limiting the total harvest volumes not to vary with more than 15% from one 5-year period to the next 5-year period. The BAU scenario was compared with three alternative policy scenarios (Chapter 2): • Scenario 1: Biodiversity first. • Scenario 2: Multifunctionality. • Scenario 3: Wood bioeconomy first. As described in Sotirov et al. (2025b), the EU-level scenarios were adapted to fit the Swedish context. Initially, a survey was distributed to forest owners and managers in Norrbotten County. Based on the survey results, three agent types were identified in the county: Optimizers, Multifunctionalists, and Environmentalists (Maximo et al. 2025). The data on agent types was combined with survey data on owner categories (state and other public owners; corporations and church owners; non-industrial private owners) enabling an assessment of the relative prevalence of each agent type within each forest owner category. The adapted scenarios where then presented to forest-owners in a series of local workshops, each workshop targeting owners of a specific agent type. Before the workshops, the forest owners responded to selected questions in the above-mentioned survey and based on their responses they could be divided into distinct agent types – and corresponding workshop groups. In the workshops, each group of owners, representing one agent type, described how they would change their management relative to the BAU scenario. When simulating the scenarios in Heureka, this information was used to inform the regional-level survey results so that distinct forest management responses could be constructed for each owner category in ways that reflected the behaviour of the three forest owner types included in the category: • State and other public owners. • Corporations and church owners. • Non-industrial private forest owners. Page 34 The somewhat weak correlations between these forest owner categories and three agent typologies (Optimizers, Multi-functionalists, and Environmentalists) are discussed by Maximo et al. (2025). As described above, any behavioural adaptations were investigated by questionnaires and group discussions held at workshops with representatives of the forest owner categories before the simulations. Forest management prescriptions for each scenario was developed based on the distribution of each agent typology within each owner category, as described in Table 6.1 by Maximo et al. (2025). Agent type behaviours are thus considered and incorporated into the responses of each owner category. The Table shows the aggregated impacts of the responses of all owner categories. The different adaptations for each forest owner category and policy scenario were then translated into specific settings in Heureka PlanWise to simulate a particular scenario, as described in Tables 5.1-5.3. Table 5.1. Simulated forest management in scenario 1: Biodiversity first. The primary management adaptations (see “adapted even-aged forestry” in the table below), relative to the current even-aged rotation management informing the reference (or BAU) scenario (see above), include prolonged rotation periods, promoting of broadleaves, extensive extraction of wood and no extraction of harvest residuals. The management changes and adaptations described below are based on the forest owner’s responses to the conditions stipulated in the scenario description. Forest domain* Area (ha) Share Forest management Protected areas (wildlife preserves etc.) 549340 14.1% Free development/Unmanaged Informally protected areas (volunteer pres.) 183245 4.7% Free development/Unmanaged Informally protected areas (set aside patches) 272936 7.0% Free development/Unmanaged Biodiversity and/or societal high value areas 269508 6.9% Continuous cover forestry and enhanced preservations (30% set aside) Contorta forests 88988 2.3% Contorta liquidation. followed by adapted even-aged forestry Public forests, old growth etc. 364955 9.4% Free development/Unmanaged Public forests with broadleaves etc. 475849 12.2% Broadleaf promoting continuous cover forestry and general preservations (15% set aside) Public forests 311541 8.0% Adapted even-aged forestry and general preservations (15% set aside) Corporate forests, old growth etc. 128362 3.3% Free development/Unmanaged Corporate forests 222450 5.7% Adapted even-aged forestry and general preservations (15% set aside) Private, small-scale forests 1029452 26.4% Adapted even-aged forestry and general preservations (15% set aside) Sum: 3896626 100.0 % *) Subareas, with specific characteristics, of the analysis area (described in section 5.1.2). Table 5.2. Simulated forest management in scenario 2: Multifunctionality. Forest domain* Area (ha) Share Forest management Protected areas (wildlife preserves etc.) 549340 14.1% Free development/Unmanaged Informally protected areas (volunteer preserves) 183245 4.7% Free development/Unmanaged Informally protected areas (set aside patches) 272936 7.0% Free development/Unmanaged Biodiversity and/or societal high value areas 269508 6.9% Continuous cover forestry and enhanced preservations (30% set aside) Public forests, old growth etc. 337095 8.7% Free development/Unmanaged Page 35 Public forests, spruce dominated etc. 78693 2.0% Adapted even-aged forestry with fertilization (according to Sveaskog) and general preservations (7.5% set aside) Public forests, areas suitable for exotic tree species 81265 2.1% Adapted even-aged forestry with exotic tree species (according to Sveaskog) and general preservations (7.5% set aside) Public forests, lichens and low productive areas 159153 4.1% Reindeer herding-adapted even-aged forestry and general preservations (7.5% set aside) Public forests, middle-aged areas 185591 4.8% Continuous cover forestry and general preservations (7.5% set aside) Public forests 347331 8.9% Productivity-adapted even-aged forestry (according to Sveaskog) and general preservations (7.5% set aside) Corporate forests, lichens and low productive areas 84923 2.2% Reindeer herding-adapted even-aged forestry and general preservations (7.5% set aside) Corporate forests, high productive areas 207602 5.3% Productivity-adapted even-aged forestry (according to the corporations) and general preservations (7.5% set aside) Corporate forests 96361 2.5% Even-aged forestry and general preservations (7.5% set aside) Private, small-scale forests, pine dominated and low productive areas 206701 5.3% Adapted even-aged forestry with exotic tree species and general preservations (7.5% set aside) Private, small-scale forests, spruce and broadleaf dominated areas 208139 5.3% Adapted even-aged forestry with fertilization and general preservations (7.5% set aside) Private, small-scale forests 628744 16.1% Even-aged forestry and general preservations (7.5% set aside) Sum: 389662 6 100.0 % *) Subareas, with specific characteristics, of the analysis area (described in section 5.1.2). Table 5.3. Simulated forest management in scenario 3: Wood and bioeconomy first. Here, the adaptations of the even-aged forestry imply production enhancements, shortened rotation periods, promoting of conifers and intensive extraction of wood and harvest residuals. Forest domain* Area (ha) Share Forest management Protected areas (wildlife preserves etc.) 549340 14.1% Free development/Unmanaged Informally protected areas (volunteer preserves) 183245 4.7% Free development/Unmanaged Informally protected areas (set aside patches) 272936 7.0% Free development/Unmanaged Biodiversity and/or societal high value areas 269508 6.9% Continuous cover forestry and enhanced preservations (30% set aside) Public forests, spruce dominated etc. 103047 2.6% Adapted even-aged forestry with intensive fertilization and general preservations (5% set aside) Public forests, areas suitable for exotic tree species 129752 3.3% Adapted even-aged forestry with exotic tree species and general preservations (5% set aside) Public forests 956328 24.5% Adapted even-aged forestry (according to Sveaskog) and general preservations (5% set aside) Corporate forests, spruce dominated etc. 44186 1.1% Adapted even-aged forestry with intensive fertilization and general preservations (5% set aside) Corporate forests, areas suitable for exotic tree species 58152 1.5% Adapted even-aged forestry with exotic tree species and general preservations (5% set aside) Page 36 Corporate forests 286547 7.4% Adapted even-aged forestry (according to the corporations) and general preservations (7.5% set aside) Private, small-scale forests 104358 4 26.8% Adapted even-aged forestry and general preservations (7.5% set aside) Sum: 389662 6 100.0% *) Subareas, with specific characteristics, of the analysis area (described in section 5.1.2). 5.1.2. Model initialisation data The initial state of the forest in the current analysis area, the county of Norrbotten with 3,896,626 hectares of productive forest land, was described using data from the National Forest Inventory (Fridman et al. 2014) and its 3470 sample plots surveyed in the area during 2016-2020. Norrbotten is the northernmost county in Sweden, with latitudes between 65 and 69° N, where the pine-dominated forest (57% of standing stock), 25% spruce, and the remaining share mostly birch are described in Table 5.4 and Figure 5.1. Table 5.4. Characteristics of the initial state of the forest in the analysis area (3 896 626 ha). Average standing stock (m3/ha) 95.8 Average stand age (yrs) 80.1 Average wood productivity potential (m3/ha, yr) 2.9 Figure 5.1. Age class distribution of the initial state of the forest in the analysis area (3 896 626 ha). Note that old forests, which consist of 679,735 hectares (17.4%) of forests with a basal areaweighted stand age over 135 years, are largely either formally protected or designated as informal set-asides to be left unmanaged; both in reality and in the simulations. The area distribution in all four scenario analyses, when simulating different forest management regimes, can be seen in Table 5.5. Table 5.5. Initial subarea distribution of the analysis area (3 896 626 ha). Area (ha) Share Forest management Protected areas (wildlife preserves etc.) 549340 14.1% Free development/Unmanaged Informally protected areas (volunteer preserves) 183245 4.7% Free development/Unmanaged Page 37 Informally protected areas (set aside patches) 272936 7.0% Free development/Unmanaged Biodiversity and/or societal high value areas 269508 6.9% Continuous cover forestry and enhanced preservations (30% set aside) Other, managed areas 2621596 67.3% Forestry adapted to each forest owner group and policy scenario, respectively Sum: 3896626 100.0% Note that in the simulations of forest management tailored to the behaviour of the three forest owner groups across the three alternative policy scenarios, the differentiation was made for the 2,621,596 hectares of managed forest land. The distribution of the area in the three policy scenario analyses concerning the three forest owner groups can be seen in Table 5.6. Table 5.6. Forest owner distribution of the managed part of the analysis area (2 621 596 ha). Forest owner group Area (ha) Share State and other public 1189127 45.4% Corporations and church 388885 14.8% Non-industrial private (small scale) 1043584 39.8% Sum: 2621596 100.0% In Heureka, the effects of global warming on forest growth and similar factors are simulated using a model by Freeman (2009). The climate change scenario corresponding to RCP4.5 was used in all scenario analyses. 5.2. Scenario results and interpretation The results from the simulations of the three alternative scenarios are summarized in Tables 5.7-5.9. Note that year 0, or period 0, is a “starting point” from which the simulations begin. Heureka simulates the development of the forest in five-year steps, considering the forestry activities (like harvest operations) during these periods. This means that period 1 covers years 1-5, period 2 covers years 6-10, and so on, up to period 20, which corresponds to years 96100. When estimating by period, this approach is valid for years 1-5, 6-10, and so forth. This contrasts with point estimates, which are valid for years 2.5, 7.5, and so on—that is, the middle of each period. Additionally, all simulations start at year 0. Year 2.5, which falls in the middle of the first five-year period, is the first “stop" point when looking into the future with Heureka. All forestry activities are simulated at the midpoint of each five-year period. Correspondingly, all point estimates are valid immediately before the simulation of any forestry activity at the midpoint of each period. The results, usually average values over the entire 100-year time horizon but sometimes average values for the first ten (or 20) years for each scenario, are compared to the corresponding results for the BAU scenario. Important forest and ecosystem service indicators, along with biodiversity indicators, are highlighted. Overall, the presented resultscan be interpreted as describing the forest management simulated in each scenario. Page 38 Table 5.7. Simulation results from scenario 1: Biodiversity first, compared to results from the reference scenario (BAU), for the analysis area of Norrbottens county (3 896 626 ha). Indicator variable* (unit) BAU Scenario 1 Rel. Difference Average harvest/extracted roundwood (m3/ha, yr): 1.54 1.20 -22.2% Average standing/growing stock (m3/ha): 152 175 15.6% Average standing/growing stock of broadleaves (m3/ha): 21.5 27.2 26.8% Average current annual increment, net (m3/ha, yr): 2.88 2.90 0.6% Average tree biomass carbon stock (ton C/ha): 58.1 66.7 15.0% Average tree biomass carbon flux (ton C/ha, yr): 0.35 0.48 37.0% Average soil carbon flux (ton C/ha, yr): 0.03 0.05 46.8% Average soil carbon flux year 1-20 (ton C/ha, yr): -0.09 -0.11 -22.5% Average share of old broadleaf forests: 8.7% 12.1% 38.9% Average deadwood formation (m3/ha, yr): 0.13 0.22 66.8% Average total deadwood formation (m3/ha, yr): 0.13 0.22 66.8% Average amount of veteran trees (m3/ha): 16.8 21.1 25.4% Average amount of species rich forests (ha): 375484 502683 33.9% *) For the 100-year planning horizon, if nothing else stated. Table 5.8. Simulation results from scenario 2: Multifunctionality, compared to results from the reference scenario (BAU), for the analysis area of Norrbottens county (3 896 626 ha). Indicator variable* (unit) BAU Scenario 2 Rel. Difference Average harvest/extracted roundwood (m3sub/ha, yr): 1.54 1.55 1.0% Average standing/growing stock (m3/ha): 152 159 4.7% Average standing/growing stock of broadleaves (m3/ha): 21.5 21.2 -1.5% Average current annual increment, net (m3/ha, yr): 2.88 3.03 4.9% Average tree biomass carbon stock (ton C/ha): 58.1 60.4 4.1% Average tree biomass carbon flux (ton C/ha, yr): 0.35 0.40 14.5% Average soil carbon flux (ton C/ha, yr): 0.03 0.05 36.8% Average soil carbon flux year 1-20 (ton C/ha, yr): -0.09 -0.09 -1.9% Average share of old broadleaf forests: 8.7% 8.7% -0.1% Average deadwood formation (m3/ha, yr): 0.13 0.15 14.0% Average total deadwood formation (m3/ha, yr): 0.13 0.15 14.0% Average amount of veteran trees (m3/ha): 16.8 17.7 5.3% Average amount of species rich forests (ha): 375484 401508 6.9% *) For the 100-year planning horizon, if nothing else stated. Page 39 Table 5.9. Simulation results from scenario 3: Wood Bioeconomy first, compared to results from the reference scenario (BAU), for the analysis area of Norrbottens county (3 896 626 ha). Indicator variable* (unit) BAU Scenario 3 Rel. Difference Average harvest/extracted roundwood (m3sub/ha, yr): 1.54 1.75 13.5% Average standing/growing stock (m3/ha): 152 147 -3.3% Average standing/growing stock of broadleaves (m3/ha): 21.5 19.3 -10.2% Average current annual increment, net (m3/ha, yr): 2.88 3.09 7.1% Average tree biomass carbon stock (ton C/ha): 58.1 56.1 -3.3% Average tree biomass carbon flux (ton C/ha, yr): 0.35 0.34 -2.4% Average soil carbon flux (ton C/ha, yr): 0.03 0.04 23.8% Average soil carbon flux year 1-20 (ton C/ha, yr): -0.09 -0.08 -8.0% Average share of old broadleaf forests: 8.7% 8.2% -6.0% Average deadwood formation (m3/ha, yr): 0.13 0.12 -12.4% Average total deadwood formation (m3/ha, yr): 0.13 0.12 -12.4% Average amount of veteran trees (m3/ha): 16.8 14.2 -15.7% Average amount of species rich forests (ha): 375484 347432 -7.5% *) For the 100-year planning horizon, if nothing else stated. Examples of period-specific results from the policy scenario analyses are shown in Figures 5.2 to 5.7. In some cases, for certain indicator variables, significant differences only appear after a relatively long time, in the second half of the 100-year planning horizon. However, in other cases, rapid changes are visible, such as for the carbon flux, which relates to climate change mitigation. A key finding from the scenario analyses is that forest owner behaviour matters, sometimes in the short term and other times in the long term. Overall, the differences between the various scenarios are minor. This appears true even if forest owners or managers anticipate significant behavioural changes in the different policy scenarios. Therefore, more than path dependencies, the results are likely influenced by limited flexibility due to relatively intensive management of the forest landscape over several decades. Figure 5.2. Average standing stock, in cubic meters per hectare, for policy scenario 1: Biodiversity first, 2: Multifunctionality and 3: Wood and bioeconomy first, respectively, during a planning horizon of 100 years for Norrbotten. Page 40 Figure 5.3. Average standing stock of broadleaves, in cubic meters per hectare, for policy scenario 1: Biodiversity first, 2: Multifunctionality and 3: Wood and bioeconomy first, respectively, during a planning horizon of 100 years for Norrbotten. Figure 5.4. Average harvest, in cubic meters per hectare and year, for policy scenario 1: Biodiversity first, 2: Multifunctionality and 3: Wood and bioeconomy first, respectively, during a planning horizon of 100 years for Norrbotten. Figure 5.5. Average growth, in cubic meters per hectare and year, for policy scenario 1: Biodiversity first, 2: Multifunctionality and 3: Wood and bioeconomy first, respectively, during a planning horizon of 100 years for Norrbotten. Figure 5.6. Average carbon stock, in tonnes per hectare, for policy scenario 1: Biodiversity first, 2: Multifunctionality and 3: Wood and bioeconomy first, respectively, during a planning horizon of 100 years for Norrbotten. Page 41 Figure 5.7. Average carbon flux, in tonnes per hectare and year, for policy scenario 1: Biodiversity first, 2: Multifunctionality and 3: Wood and bioeconomy first, respectively, during a planning horizon of 100 years for Norrbotten. Across the three scenarios, the Biodiversity-First strategy produced the most significant deviations from business-as-usual, creating major benefits for biodiversity and carbon-related metrics but causing notable trade-offs in timber production and economic returns. The biggest positive changes include sharp increases in rotation length (+84%) and deadwood formation (+67%), along with substantial gains in soil and total carbon flux (+37–47%), species-rich forest area (+34%), and veteran tree abundance (+25–26%). These improvements come at the cost of sharp declines in harvest outputs (−20 to −25%), net revenue (−18%), and net present value (−22%), indicating a system strongly shifted toward structural complexity and long-term carbon storage rather than wood production. In contrast, the Multifunctionality scenario shows moderate positive benefits across most ecosystem services, with small to medium gains in biodiversity (e.g., deadwood +14%) and carbon flux (+16–37%), coupled with slight improvements in timber yield and economic indicators (around +1%), while avoiding major trade-offs. The Bioeconomy-First scenario prioritizes timber production and revenue, demonstrating the largest increases in harvested stemwood and roundwood (+13–14%), net revenue (+7%), and total carbon flux (+24% for soil carbon flux; +3–23% in other carbon variables), but at the expense of declines in most biodiversity indicators—including veteran trees (−13 to −16%), species-rich forests (−8%), and deadwood (−5%)—reflecting a focus on wood provisioning over ecological complexity. Overall, these results highlight strong synergies between biodiversity conservation and carbon storage, moderate compatibility among ecosystem services under multifunctionality, and clear trade-offs between bioeconomic objectives and biodiversity-related ecosystem services. Importantly, caution should be exercised when interpreting these results due to limitations in Heureka’s ability to project outcomes for non-standard forestry practices in Sweden (e.g., selective logging), as well as its inclusion of only positive (growth increases) but not negative (disturbance increases) impacts of human-driven climate change in the projections. 5.3. Main findings Across the three policy pathways, the Biodiversity-First strategy causes the greatest shift from current practices by increasing habitat indicators and carbon storage while significantly reducing timber outputs and economic returns. The Multifunctionality approach achieves more balanced results, providing moderate improvements in both habitat for biodiversity and carbon metrics, along with slight gains in timber production and revenue, with few notable trade-offs. The Bioeconomy-First strategy, on the other hand, focuses on wood supply and financial performance, raising harvest levels and revenues but leading to declines in multiple biodiversity indicators. Overall, the results show that biodiversity and carbon storage goals Page 48 Figure 6.4 Stemwood carbon density and its projected changes. Upper row: Maps showing stemwood carbon densities for SSP245 in the year 2050. The three panels show biomass density for the current management scenarios (left panel), and the differences between the two other scenarios and the current scenario (middle and right panels). Note that in these panels, in order to make the distribution most visible, the colour scheme does not show the entire range of values. Lower row: Forest area weighted histograms of the ratio of biomass density in the bio management scenario to biomass density in the current management scenario for the years 2050 and 2100. Figure 6.5 Stemwood carbon density in the living forest in the bio management scenario against the value in the current management scenario broken down by pixels where the productivity increase happens in the current scenario only (red), the bio scenario only (blue), both scenarios (green), and neither scenarios (grey). This is for the SSP245 climate scenario and the multifunctionalist management type. This shows that the main regeneration method management decision is primarily responsible for differences in stock between the management scenarios. Page 49 The differences between individual agents within a scenario are also small, as shown for the SSP245 climate and current management scenario in Figure 6.6, which displays the development of average biomass density and total cumulative removals broken down by agent type. One possible reason why the total stock development is similar across all scenarios is that the main factors affecting stock development are harvest and thinning intensity, which were not examined in the survey used for our management model. As a result, we only observe shifts between the binary options of harvest/no-harvest and thinning/no-thinning. Since our management model reasonably predicts that harvest and thinning almost always occur, this results in similar total stocks in all cases. Any changes in rotation time or harvest amounts that might occur due to different management objectives are not represented in our model. Figure 6.6 Development of living stemwood carbon density for the SSP245 climate scenario and the current management scenario broken down by agent types (left panel) and cumulative average removals for each case (right panel). The removals include harvests, thinning activities, and disturbances. Although the total European stock is similar across different scenarios and agents, larger variations become evident when examining the spatial distribution of stock. Figure X illustrates the distributions of the differences between management scenarios and the histograms of the ratio of biomass density in the bio management scenario compared to the current management scenario. Some regions show higher stock in the current scenario than in the bio scenario, while others are lower, with differences reaching up to 20% in either direction. In Figure 6.7, we show the development of the percentage of forested area covered by broadleaf and needleleaf forests for each climate and management scenario considered, along with the breakdown of these percentages by agent type. In all cases, there is a general trend toward more broadleaf forest. The bio management scenario results in the highest broadleaf fraction, while the current management scenario results in the lowest. This trend varies significantly between the agent types. For the optimizers, the broadleaf fraction by the end of the century is much higher under the bio management scenario than under the current scenario. Spatial distributions of these changes in the current and bio cases are examined in Figure 6.8. While the overall trend leans toward more broadleaf forest, significant areas of western Europe, especially in France, show a trend toward increased needleleaf forest. The shift toward broadleaf forest is most prominent in northern Europe, although in Scandinavia, the largest changes are only observed in the bio case. The notable shift in the far north is partly due to Page 50 the low initial broadleaf share. In Alpine regions, despite the high needleleaf fraction at the start of the models, the broadleaf fraction can either increase or decrease, with no clear trend. However, it is important to note that the group of respondents our management model is based on does not include anyone responsible for managing Alpine forests, meaning the management decisions are entirely derived from extrapolations and interpolations from other regions. 6.1.2 Deadwood, litter and burned area In Figure 6.7, we show the fractional changes in the total European carbon stocks of litter, standing, and lying deadwood between 2020 and 2100 for all three management scenarios and both climate scenarios. In all cases, we observe a strong decrease in litter stock during this period due to rising temperatures causing more rapid decomposition. This is evident from the lower litter stocks found in the more extreme climate scenario. The management scenario has a much smaller impact on litter stocks compared to climate change effects. The current scenario has a slightly higher stock than the other two, which is related to the management decision on the main regeneration method. This method involves increasing growth productivity when regeneration uses material obtained from tree breeding or enrichment planting, which is more common in the current scenario. Higher growth productivity leads to increased litterfall and, consequently, a higher litter stock. Page 51 Figure 6.7 Development of total carbon stocks in European forests in the litter (upper-left panel), the lying deadwood (upper-right panel), and in the standing deadwood (lowerleft panel) components for the three management scenarios and both climate scenarios considered. In our model, the litterfall rate at any location depends on that year’s increment, not the amount of biomass present. Therefore, an increase in stock does not necessarily lead to higher inputs into the litter pool. For deadwood, inputs are residuals from natural mortality, disturbances, and harvests. The input rate of deadwood then depends on the total forest biomass; thus, any management decisions affecting the stock indirectly influence the amount of deadwood present. Some parts of the management model directly influence deadwood levels and are both significant. The decision regarding disturbance residuals is crucial because mortality caused by disturbances from insects and diseases is the largest source of standing deadwood and disturbance residuals from windthrow is an important source of lying deadwood. Whether these residuals are removed or left in the forest impacts the local amount of standing deadwood. Additionally, the main diversity improvement method allows some pixels to add an extra 10% of harvest residuals to lying deadwood. One of the key management decisions that directly affects deadwood is whether to perform thinning. If thinning occurs in a pixel, residuals are added to the lying deadwood pool. If not, increased natural mortality adds more carbon to the standing deadwood pool. Since we assume no decomposition of standing deadwood, the climate-related effects only indirectly influence the carbon stocks. This results in a relatively small difference between climate scenarios and a larger impact of management decisions. Our models show that on local scales, the choice to remove disturbance residuals or not is significant. Pixels have notably lower amounts of standing deadwood if disturbance residuals are removed. On a European scale, this management decision affects both the current and bio management scenarios equally, but in different locations, and therefore does not change the overall differences in the total European stock of standing deadwood shown in Figure 6.7. The Page 52 difference in the total Europe-wide stocks between these scenarios is primarily due to the decision of whether to perform thinning. Figure 6.8 Upper row: Spatial distributions of standing and lying deadwood in 2050 for the current management scenario and SSP245. Middle row: Change in standing deadwood between 2020 and 2050 for the current (left) and bio (middle) management scenarios, and the difference between the two scenarios (right) for SSP245. Lower row: The same as the middle row but for lying deadwood. For lying deadwood, climate driven changes in the decomposition rates have a much larger effect on total carbon stocks. In our models, the dominant source of lying deadwood is harvest (including thinning) residuals. The management scenario also has an effect similar to that of Page 53 the standing deadwood described above. While the decisions to leave or remove disturbance residuals and to whether to perform thinning influence the local amount of lying deadwood, the total European stocks are the same in the current and bio management scenarios. The 2050 spatial distributions of standing and lying deadwood in the current management model for SSP245, along with comparisons of the changes in both of these quantities between 2020 and 2050 for the current and bio management scenarios, are shown in Figure 6.8. The largest increases in both components are typically observed in Norway, Sweden, and Finland, although lying deadwood decreases in much of southern Sweden and Finland. The differences between these scenarios on a pixel-by-pixel basis are shown more clearly in Figure 6.9, showing that locally standing deadwood varies substantially more than lying deadwood between the two scenarios. Figure 6.9 Pixel by pixel comparison of deadwood amounts in the bio and current management scenarios separated for standing lying deadwood. In Figure 6.10, we present the total integrated burned areas calculated with our litter and deadwood data and the FLAM model. We show both annual burned area and cumulative burned area for each management and climate scenario. Cumulative burned area is useful because the stochastic nature of annual burned areas makes comparing different scenarios difficult. The management decisions from our model only affect the fuel stock in forested areas of each pixel, primarily consisting of litter and lying deadwood. Since these two components are similar across different management scenarios, we observe only a small effect of management on the total burned area. The impact of the climate scenario is much more significant because the more extreme climate conditions provide a better environment for fire ignition and spread. Page 54 Figure 6.10 Yearly (upper panel) and cumulative (lower panel) burned area for each management and climate scenario calculated using the litter and deadwood estimates from Section 6.2.2 and FLAM. 6.3. Main findings The modeling results for this case study provide European-wide predictions of how forest structure, carbon dynamics, and disturbance risks change under different forest management strategies and climate scenarios throughout the 21st century. Overall, the simulations show a continued increase in total carbon stocks across Europe, mainly driven by forest growth, moderate increases in the share of broadleaf species, decreases in litter stocks, increases in deadwood stocks, and a rise in the expected burned area from wildfires. Differences between management scenarios (current, biodiversity-focused ‘bio’, and climateadaptation ‘clim’) are relatively small at the scale of the entire European stock. This is likely influenced by the survey-derived management model (Maximo et al., 2025), which only includes binary decisions about whether to harvest or not and whether to thin or not. Both decisions are chosen to be implemented almost everywhere in each scenario. Management choices that forest managers are more likely to change and that could significantly impact carbon stocks, such as harvest or thinning intensities, were not included in the survey or the management model. Future surveys designed to inform forest models should explicitly Page 55 examine variations in harvest and thinning intensities, as these factors are crucial for determining long-term carbon dynamics in large-scale forest models like G4M. The main factor in our management model that creates differences in carbon stocks is the regeneration method. When managers keep harvest intensity the same but aim for biodiversity-friendly replanting, the total carbon stock in Europe increases less than with current management practices. This is because the choice of regeneration method in biodiversity-friendly replanting often favours techniques like planting with regular planting material and natural regeneration over more productive methods, such as using material from tree breeding. If this biodiversity-friendly replanting was part of a broader biodiversity improvement management strategy that also involved reducing harvest intensity, we would expect higher carbon stocks in Europe in the bio management scenario. These results suggest that the carbon-sequestration benefits of biodiversity-oriented management may be more limited than expected unless combined with adjustments in harvest intensity. The results indicate that forest managers tend to stick with existing practices even when faced with different objectives and that any changes they are likely to implement will not significantly alter the fundamental properties of forests across Europe. Policy-driven incentives will be crucial to encourage shifts in forest management so managers are motivated to adopt new practices. Future research combining surveys of forest managers with large-scale forest models could also help frame potential management changes in terms of incentives for making beneficial decisions. A proactive range of management strategies might include introducing non-native or highly productive tree species, speeding up the transition from coniferous to broadleaved forests, and designating set-aside conservation areas, among other options. On local scales within Europe, management effects are more apparent. Comparisons of individual pixels reveal localized differences of up to ±20% in living forest biomass between scenarios. While all scenarios show a general shift toward more broadleaf cover, the bio scenario accelerates this trend most strongly, with significant increases in northern Europe and more modest or even opposite trends in western Europe, where needleleaf may expand under some management-climate combinations. Agent-type differences are also notable: multifunctionalists and optimizers display the largest shift toward broadleaf forests, with the latter also showing the greatest divergence in broadleaf expansion between the current and bio scenarios. For deadwood and litter, climate has a stronger influence than management. Rising temperatures accelerate decomposition, leading to significant declines in litter stocks, especially under the SSP585 climate scenario, while management has a smaller impact, mainly driven by differences in growth productivity. Note that changes in total stock are less relevant for litter stocks. We expect that harvest intensity would be important if included in our models since it affects forest age and, consequently, growth rates. Standing deadwood responds more strongly to management, particularly through thinning decisions and whether deadwood caused by disturbance-induced mortality is removed (salvage logging). The absence of thinning results in higher natural mortality and, therefore, more standing deadwood, although these effects vary across locations. Lying deadwood is most influenced by changes in decomposition and harvest residuals, with climate parameters and management both playing significant roles. Despite these structural changes, wildfire outcomes show only limited sensitivity to management. Because fuel loads are mainly influenced by litter, which is mostly determined by forest growth productivity and climate factors, the burned area remains nearly unchanged across management scenarios. The climate scenario is influential with two opposing effects: higher temperatures create conditions that favour fires but also speed up decomposition, Page 56 reducing fuel availability. Overall, we find that more extreme climate scenarios lead to an increase in the total area affected by fire. However, our model does not account for changes in fire suppression efficiency. The geographic distribution of survey respondents also limits our ability to interpret the management model. Responses were collected from northern Sweden, parts of Germany, Poland, Slovenia, and Catalonia, leaving large areas of Europe, such as the Alpine region, unrepresented. As a result, the spatial patterns of management decisions predicted by our model and the forest properties that result from these decisions are, in much of Europe, based on extrapolations of management behaviour into areas where we lack direct information. Future work of this type could benefit from a more comprehensive geographic distribution of survey respondents when it is practically feasible. Page 57 7. Synthesis This study aimed to examine how behaviour and behavioural change can support improvements in forest biodiversity and promote more sustainable socio-ecological forest systems, including synergies and trade-offs. We evaluated the results of selected policy and management scenarios using forest simulation models that were recently expanded to incorporate human behaviour and behavioural change. We conducted these assessments for three regional case studies in Germany, Spain, and Sweden, as well as an EU-level case study. Across all four case studies, we found that changes in forest management behaviour can enhance biodiversity, but the extent of improvement depends heavily on the magnitude of management change and the practices modified. Additionally, we observed that biodiversity-friendly management practices are positively associated with carbon storage in forests, particularly when management (particularly harvest) intensity is reduced. In the German case studies, the biodiversity scenario resulted in the largest carbon accumulation in forests, regardless of the climate scenarios considered. In the Swedish case study, the biodiversity-first and multifunctional scenarios simultaneously improved habitat indicators and forest carbon stocks. Similarly, in the Spanish case study, the no-management and close-to-nature scenarios enhanced carbon stocks, structural heterogeneity, and species diversity. At the European level, all scenarios show increasing total carbon stocks over time, and biodiversity-oriented regeneration methods contribute to structural changes such as increased broadleaf cover and deadwood accumulation. Our results also emphasize the importance of the type and intensity of management practices. The German case study showed that biodiversity outcomes are affected by what is harvested (tree species, size, and harvest type) not just by the quantity harvested. This aligns with the findings from the Spanish and Swedish case studies, where close-to-nature or multifunctional approaches can provide balanced outcomes for biodiversity, carbon, and timber supply. In Sweden, biodiversity-first management clearly reduced timber supply and revenues, while multifunctional management minimized these trade-offs. In the Spanish case study, the nomanagement scenario maximized biodiversity and carbon storage but eliminated timber production, whereas closer-to-nature management served as a compromise that still increased wood supply. Across the four cases, management effects often had a greater influence than climate effects on biodiversity-related and other indicators included in our study, especially forest structure, forest carbon, and deadwood. In the German and Spanish case studies, management had a stronger impact than climate on most indicators examined. In the EU case study, several indicators (e.g., standing deadwood) were more strongly affected by management decisions (e.g., thinning and salvage logging) than by climate, but, for example, litter stocks mainly were affected by climate. These findings suggest that behavioural changes among forest owners and managers can significantly influence biodiversity outcomes even amid climate change. All case studies highlight behavioural inertia as a key limiting factor for significant changes in forest structure and composition, as well as biodiversity indicators. Policy-driven incentives can help encourage shifts in forest management, but forest managers and owners often indicated or exhibited limited deviations from current practices in favour of forest biodiversity. Results from the German case studies suggested that many owners already perceive themselves as applying biodiversity-friendly practices. Future research combining surveys of forest managers with large-scale forest models could also help frame potential management changes in terms of incentives for making beneficial decisions.