Abstract Climate change is expected to alter species assemblages by affecting the outcome of competition between species. Investigating processes of competition remains challenging in tree communities, as they unfold over extensive spatio-temporal scales. Here, we used a deep learning-based meta-model trained on 135 million simulated tree responses to climate across Europe to investigate changes in the competitiveness of nine major tree species under future climate. We harnessed projections from local process models to train a Deep Neural Network of forest state transitions to investigate climate-induced changes in competition at continental scale. We found decreasing competitive strength for evergreen conifers across their distribution, while deciduous broadleaved species increased in competitiveness. Most investigated species lost competitive strength at their warm range edges. Consequently, up to 25% of Europe’s forests could experience a change in the dominant tree species until the end of the 21st century, suggesting a profound climate-induced reassembly of Europe’s forests.

Wildfires, insect outbreaks, and storms cause large pulses of tree mortality. Climate change amplifies these forest disturbances, yet their future magnitude and extent remain uncertain. Here, we simulated future forest disturbance regimes at 100-meter resolution across Europe using a deep learning-based simulation framework. Our results show that forest disturbances will continue to increase throughout the 21st century, with disturbed areas more than doubling relative to the recent past under an unabated continuation of climate change. Wildfires are the main agent driving future disturbance change. Changing disturbances result in an increase in young forests, substantially altering Europe's forest demography. Because of their profound implications for forest carbon storage and the habitat value of forest ecosystems, disturbances should be a priority of forest policy and management.

When an even-aged forest stand reaches maturity, it can be renewed within a limited period to maintain the even-aged structure or gradually transformed to an uneven-aged stand. However, there is still debate as to which silvicultural approach is more profitable, conducive to carbon storage, favourable to biodiversity or resilient. In this study, we simulated the evolution of fifteen stands representative of the Walloon forest, whose initial structure was even-aged. The stands were managed according to two contrasting silvicultural approaches (continuation of the even-aged system vs transformation to an uneven-aged one), and the simulations were run with the SSP3-7.0 climate projections produced by five global circulation models. Our simulations indicate that even-aged and uneven-aged silviculture yield similar outcomes in terms of carbon storage. Financial indicators were likewise largely unaffected, except in the oak-beech mixture, where uneven-aged silviculture increased profitability through the substitution of oak by beech. This shift reduced tree species diversity in uneven-aged oak-beech stands. Tree microhabitats, except in beech stands, were more abundant under uneven-aged silviculture. While mean values of forest ecosystem functioning indicators are largely comparable between the two approaches, uneven-aged stands exhibit higher temporal stability. Uneven-aged silviculture produces stands that are more wind-resistant and avoid periods of extreme vulnerability. Our study shows that the most appropriate silviculture may vary depending on the aspect considered. Uneven-aged silviculture has a definite advantage in terms of stability, risk management, and tree microhabitats. However, maintaining even-aged patches at the landscape scale remains important to facilitate the regeneration of shade-intolerant species.

Climate change impacts forest functioning and dynamics, but large uncertainties remain regarding the interactions between species composition, demographic processes and environmental drivers. While the effects of changing climates on individual plant processes are well studied, few tools dynamically integrate them, which precludes accurate projections and recommendations for long-term sustainable forest management. Forest gap models present a balance between complexity and generality and are widely used in predictive forest ecology, but their lack of explicit representation of some of the processes most sensitive to climate changes, like plant phenology and water use, puts into question the relevance of their predictions. Therefore, integrating trait- and process-based representations of climate-sensitive processes is key to improving predictions of forest dynamics under climate change. In this study, we describe the PHOREAU model, a new semi-empirical forest dynamic model resulting from the coupling of a gap model (FORCEEPS), with two process-based models: a phenology-based species distribution model (PHENOFIT) and a plant hydraulics model (SurEAU), each parametrized for the main European species. The performance of the resulting PHOREAU model was then evaluated over many processes, metrics and time-scales, from the ecophysiology of individuals to the biogeography of species. PHOREAU reliably predicted fine hydraulic processes at both the forest and stand scale for a variety of species and forest types. This, alongside an improved capacity to predict stand leaf areas from inventories, resulted in better annual growth compared to ForCEEPS, and a strong ability to predict potential community compositions. By integrating recent advancements in plant hydraulic, phenology, and competition for light and water into a dynamic, individual-based framework, the PHOREAU model, developed on the Capsis platform, can be used to understand complex emergent properties and trade-offs linked to diversity-effects effects under extreme climatic events, with implications for sustainable forest management strategies.

Climate change impacts forest functioning and dynamics, yet significant uncertainties persist regarding the interactions between species composition, demographic processes, and environmental drivers. While the effects of climate change on individual plant ecophysiology are better understood, few robust tools integrate these processes dynamically, hindering accurate projections and recommendations for long-term sustainable forest management. Forest gap models strike a balance between complexity and generality and are widely used in predictive forest ecology. However, their lack of explicit representation of critical processes, such as plant phenology and water use, limits their ability to fully capture tree sensitivity to climate change, calling into question the robustness of their future predictions. Therefore, incorporating trait- and process-based representations of climate-sensitive processes within gap models is a crucial step toward generating realistic predictions of forest evolution under climate change. In this study, we coupled the ForCEEPS gap model, validated across a broad range of forest types and environmental conditions in Europe, with two process-based models: a plant phenology model (PHENOFIT) and a plant hydraulics model (SurEAU), each parameterized for the main European tree species. We then evaluated the performance of the resulting PHOREAU model across multiple processes, metrics, and time- and spatial-scales, thereby minimizing the risk of equifinality. PHOREAU demonstrated robust capabilities in predicting fine hydraulic processes at both the forest and stand scales for various species and forest types. This, combined with its enhanced ability to predict stand leaf areas from inventories, led to modest improvements in annual growth predictions compared to the original ForCEEPS model and a strong capacity to predict potential community compositions. By integrating recent advancements in plant hydraulics, phenology, and competition for light and water into a dynamic, individual-based framework, the PHOREAU model bridges the gap between trait diversity and long-term forest productivity and resilience. It offers insights into complex emergent properties and trade-offs linked to diversity effects under extreme climatic events, with significant implications for sustainable forest management strategies. ### Competing Interest Statement The authors have declared no competing interest.

Questions have been raised about the ability of long‐lived organisms, such as trees, to adapt to rapid climate change, and to what extent forest management actions influence the evolutionary responses of tree species. Given the life history of trees and the time scales involved, these questions are often addressed through modeling approaches. Yet, most of these studies focus on single‐species case studies. The main objective and originality of our work is to explore the evolutionary responses of tree species to climate change using a process‐based model, in a multi‐specific context. This approach allows us to investigate the conditions necessary for evolutionary rescue in a mixed beech–fir forest. Furthermore, we explored how climate change adaptation and mitigation solutions, such as assisted gene flow and assisted migration, affect the conditions for evolutionary rescue in this forest type. To achieve these objectives, we integrated a quantitative genetic module into a process‐based forest gap model, enabling species‐specific parameters to evolve as quantitative traits under selective pressure and drift. Our results show that increased trait variability and heritability reduce the loss of forest cover following climatic warming in the short term (over a century). We also found that assisted gene flow had the expected effect of aiding species adapt to climate change. Finally, our study suggests that introducing new pre‐adapted species into the forest could improve recovery after climate change but could also hinder the evolutionary rescue of local species. We conclude that integrating evolutionary dynamics into process‐based models significantly enhances their predictive power by incorporating genetic adaptation scenarios that would otherwise be overlooked. This approach also allows us to test eco‐evolutionary hypotheses and better understand the potential consequences of adaptation measures to climate change for tree species.

Forests provide many ecosystem services that strongly depend on species diversity, as illustrated by the repeatedly observed diversity-productivity relationships (DPRs). These forest DPRs are assumed to result mostly from complementarity between species at the tree level whilst emerging community-level processes remain poorly explored. In this study, we propose that the 'tree packing effect' (TPE), where species diversity promotes productivity by positively impacting maximum stand density, is an important determinant of DPRs. We tested the two components of TPE: (i) whether maximum stand density increases with species richness and (ii) whether this higher stand density allowed by species richness promotes forest productivity. First, relying on national forest inventories of six European countries (NFIs, totaling 2,367,776 trees), we fitted self-thinning lines to examine whether these lines were influenced by plot species richness. We showed that maximum stand density increases with tree species richness in Europe, in all but one country. This trend was notably stronger in extreme climates. Second, we ran a large simulation-based experiment (including 7,024,815 simulations) with an individual-based forest dynamics model able to control for stand-density effects, to quantify DPRs for more than 1000 sites in Europe. Relying on an original method to quantify DPRs at the site level, we compared the strength of DPRs simulated with and without control for stand density. We found positive DPRs up to 10-times stronger when TPE is at play than when stand density is controlled. This positive effect of diversity on forest productivity through tree packing is also stronger in extreme climates, especially in warm and dry conditions. Synthesis. Highlighting the generality of the TPE in European forests, our results reveal that the effect of diversity on forest functioning is partly mediated by diversity-driven changes in stand density. This mechanism has been long overlooked in biodiversity-ecosystem functioning studies, but our findings strongly call for its reconsideration, especially in natural forests. It also opens key perspectives for management and climate change mitigation programmes. Les forê ts fournissent de nombreux services é cosysté miques aux populations, et ceux-ci dé pendent fortement de la diversité en espè ces, comme cela a é té fré quemment montré. Ces relations diversité-fonctionnement sont supposé es ré sulter principalement de la complé mentarité entre espè ces d'arbres au niveau individuel, alors que les processus é mergents à l'é chelle des communauté s restent peu é tudié s. Dans cette é tude, nous proposons qu'un effet de << densification des arbres >> (<< Tree packing effect >>, TPE), selon lequel la diversité des espè ces favourise la productivité en ayant un effet positif sur la densité maximale du peuplement, soit un dé terminant important des relations entre diversité et productivité forestiè re. Nous avons ici testé les deux composantes du << TPE >>: (i) si la densité maximale du peuplement augmente avec la richesse en espè ces, et (ii) si cette densité de peuplement plus é levé e grâ ce à la richesse en espè ces favourise la productivité forestiè re. Tout d'abord, en nous appuyant sur les inventaires forestiers nationaux de six pays europé ens (totalisant 2,367,776 arbres), nous avons estimé les lignes d'auto-é claircie afin d'examiner si ces lignes é taient influencé es par la richesse en espè ces. Nous avons ainsi montré que la densité maximale des peuplements augmente avec la richesse en espè ces d'arbres en Europe, dans tous les pays sauf un. Cette tendance é tait notamment plus forte dans les climats plus extrê mes. Deuxiè mement, nous avons mené une vaste expé rience virtuelle (comprenant 7,024,815 simulations indé pendantes) avec un modè le de dynamique forestiè re individu-centré capable de contrô ler les effets de la densité des peuplements, afin de quantifier les relations diversité-productivité pour plus de 1000 sites en Europe. En nous appuyant sur une mé thode originale pour quantifier ces relations au niveau du site, nous avons comparé la force des effets diversité simulé s avec et sans contrô le de la densité des peuplements. Synthè se. Soulignant le caractè re gé né ral du << TPE >> dans les forê ts europé ennes, nos ré sultats montrent que l'effet de la diversité sur le fonctionnement forestier est. en partie contrô lé par les changements de densité de peuplement induits par la diversité. Ce mé canisme a longtemps é té né gligé dans les é tudes liant biodiversité et fonctionnement des é cosystè mes, mais nos ré sultats invitent fortement à le reconsidé rer, en particulier dans les forê ts naturelles. Cette é tude ouvre é galement des perspectives importantes pour les programmes de gestion et d'atté nuation du changement climatique.

The constraint caused by wild ungulates on forest regeneration is increasing worldwide. Hypotheses for plant association effects predict that species susceptible to herbivory can gain protection from other neighbouring plant species. In theory, such interactions could help limit the impact of browsing on the regeneration of specific tree species. However, the presence of neighbouring species can also result in increasing competition for resources between species. The resultant effects on forest regeneration of these interactions, both positive (protection against herbivores) and negative (inter-specific competition) are still unclear. To gain insight, we coupled models of browsing by roe deer (Capreolus capreolus) and of forest dynamics to simulate trajectories of oak (Quercus petraea (Matt.) Liebl.) regeneration admixed with species of contrasted palatability and growth rate under different scenarios of browsing pressure and initial sapling density. We also investigated how releasing oak saplings from all or specific neighbours during the simulation affect regeneration. We found that admixed species composition had a relatively weak effect on the density of oak recruits, but a strong effect on the duration of the regeneration phase. Oak regenerated faster when admixed with species of intermediate growth and low palatability (Fagus sylvatica) than with species of fast growth and high palatability (Carpinus betulus L.), except at intermediate sapling density and high browsing pressure where we found the opposite. Releasing oak from all competitors was most effective in promoting oak regeneration when admixed with both species together, although the benefit of competition release was much weaker at high browsing pressure. Lastly, we found that at low initial sapling density (i.e., 10 saplings/m2), oak regeneration was driven only by browsing and the effect of admixing species became negligible. Our study showed that admixing oak with palatable neighbours impedes rather than improves oak regeneration due to increased competition for resources. As such, we suggest that the benefits of herbivore diversion can be off-set by increased inter-specific competition.

Process-based forest models combine biological, physical, and chemical process understanding to simulate forest dynamics as an emergent property of the system. As such, they are valuable tools to investigate the effects of climate change on forest ecosystems. Specifically, they allow testing of hypotheses regarding long-term ecosystem dynamics and provide means to assess the impacts of climate scenarios on future forest development. As a consequence, numerous local-scale simulation studies have been conducted over the past decades to assess the impacts of climate change on forests. These studies apply the best available models tailored to local conditions, parameterized and evaluated by local experts. However, this treasure trove of knowledge on climate change responses remains underexplored to date, as a consistent and harmonized dataset of local model simulations is missing.Here, our objectives were (i) to compile existing local simulations on forest development under climate change in Europe in a common database, (ii) to harmonize them to a common suite of output variables, and (iii) to provide a standardized vector of auxiliary environmental variables for each simulated location to aid subsequent investigations. Our dataset of European stand- and landscape-level forest simulations contains over 1.1 million simulation runs representing 135 million simulation years for more than 13,000 unique locations spread across Europe. The data were harmonized to consistently describe forest development in terms of stand structure (dominant height), composition (dominant species, admixed species), and functioning (leaf area index). Auxiliary variables provided include consistent daily climate information (temperature, precipitation, radiation, vapor pressure deficit) as well as information on local site conditions (soil depth, soil physical properties, soil water holding capacity, plant-available nitrogen). The present dataset facilitates analyses across models and locations, with the aim to better harness the valuable information contained in local simulations for large-scale policy support, and for fostering a deeper understanding of the effects of climate change on forest ecosystems in Europe.

Forests are expected to be strongly affected by modifications in climate and disturbance regimes, threatening their ability to sustain the provision of essential services. Promoting drought-tolerant species or functionally diverse stands have recently emerged as management options to cope with global change. Our study aimed at evaluating the impact of contrasting stand-level management scenarios on the resilience of temperate forests in eastern North America and central-western Europe using the individual process-based model HETEROFOR. We simulated the evolution of eight stands over 100 years under a future extreme climate according to four management scenarios (business as usual- BAU; climate change adaptation- CC; functional diversity approach- FD; no management- NM) while facing multiple disturbances, resulting in a total of 160 simulations. We found that FD demonstrated the greatest resilience regarding transpiration and tree biomass, followed by CC and then BAU, while these three scenarios were equivalent concerning the net primary production. These results were however dependent on forest type: increasing functional diversity was a powerful option to increase the resilience of coniferous plantations whereas no clear differences between BAU and adaptive management scenarios were detected in broadleaved and mixed stands. The FD promoted a higher level of tree species diversity than any other scenario, and all scenarios of management were similar regarding the amount of harvested wood. The NM always showed the lowest resilience, demonstrating that forest management could be an important tool to mitigate adverse effects of global change. Our study highlighted that tree-level process-based models are a relevant tool to identify suitable management options for adapting forests to global change provided that model limitations are considered, and that alternative management options, particularly those based on functional diversity, are promising and should be promoted from now on.

Biological production systems and conservation programs benefit from and should care for evolutionary processes. Developing evolution-oriented strategies requires knowledge of the evolutionary consequences of management across timescales. Here, we used an individual-based demo-genetic modelling approach to study the interactions and feedback between tree thinning, genetic evolution, and forest stand dynamics. The model combines processes that jointly drive survival and mating success-tree growth, competition and regeneration-with genetic variation of quantitative traits related to these processes. In various management and disturbance scenarios, the evolutionary rates predicted by the coupled demo-genetic model for a growth-related trait, vigor, fit within the range of empirical estimates found in the literature for wild plant and animal populations. We used this model to simulate non-selective silviculture and disturbance scenarios over four generations of trees. We characterized and quantified the effect of thinning frequencies and intensities and length of the management cycle on viability selection driven by competition and fecundity selection. The thinning regimes had a drastic long-term effect on the evolutionary rate of vigor over generations, potentially reaching 84% reduction, depending on management intensity, cycle length and disturbance regime. The reduction of genetic variance by viability selection within each generation was driven by changes in genotypic frequencies rather than by gene diversity, resulting in low-long-term erosion of the variance across generations, despite short-term fluctuations within generations. The comparison among silviculture and disturbance scenarios was qualitatively robust to assumptions on the genetic architecture of the trait. Thus, the evolutionary consequences of management result from the interference between human interventions and natural evolutionary processes. Non-selective thinning, as considered here, reduces the intensity of natural selection, while selective thinning (on tree size or other criteria) might reduce or reinforce it depending on the forester's tree choice and thinning intensity.

Large ungulate populations are known to cause economic damage to agriculture and forestry. Bark damage is particularly detrimental to the timber production of certain species, including Picea abies (L.) Karst. (Norway spruce): after bark is wounded, rot often spreads in the trunk base, damaging the most valuable trunk section. Numerous studies have provided valuable information on various aspects of this process, but the financial consequences over a full timber production cycle remained poorly quantified and uncertain. To fill this gap, we coupled a forest dynamics model (GYMNOS) with models of damage occurrence and decay spread. We simulated the effect of ranging levels of bark-stripping damage on financial losses. The simulations were repeated for sites of ranging fertility and with different protection measures (fences or individual protections), in Southern Belgium. The net present values of these different simulations were estimated and compared to estimate the cost of the damage and the cost-effectiveness of the damage protections. Protecting plantations against bark-stripping damage with fences was found unlikely to be worthwhile. By contrast, individual protections placed on crop trees could be helpful, particularly in the most fertile stands. Loss of revenue depended greatly on the factors tested: we estimated that the average damage cost could be about 53€/ha/year, reducing timber yield by 19%. A model was built to predict the damage cost for different values of the discount rate, site index and bark-stripping rate. This model could help develop more effective management of Norway spruce plantations and deer populations.

Sexual selection has long been known to favor the evolution of mating behaviors such as mate preference and competitiveness, and to affect their genetic architecture, for instance by favoring genetic correlation between some traits. Reciprocally, genetic architecture can affect the expression and the evolvability of traits and mating behaviors. But sexual selection is highly context-dependent, making interactions between individuals a central process in evolution, governing the transmission of genotypes to the next generation. This loop between the genetic structure conditioning the expression and evolution of traits and behaviour, and the feedback of this phenotypic evolution on the architecture of the genome in the dynamic context of sexual selection, has yet to be thoroughly investigated. We argue that demogenetic agent-based models (DG-ABM) are especially suited to tackle such a challenge because they allow explicit modelling of both the genetic architecture of traits and the behavioural interactions in a dynamic population context. We here present a DG-ABM able to simultaneously track individual variation in traits (such as gametic investment, preference, competitiveness), fitness and genetic architecture throughout evolution. Using two simulation experiments, we compare various mating systems and show that behavioral interactions during mating triggered some complex feedback in our model, between fitness, population demography, and genetic architecture, placing interactions between individuals at the core of evolution through sexual selection. DG-ABMs can, therefore, relate to theoretical patterns expected at the population level from simpler analytical models in evolutionary biology, and at the same time provide a more comprehensive framework regarding individual trait and behaviour variation, that is usually envisioned separately from genome architecture in behavioural ecology.