Conclusion

Forests play multiple important roles for the Earth system. Sound, quantitative knowledge of forest functioning, structure and diversity is therefore essential, especially in times of global change. However, many scientific questions regarding forest properties and dynamics remain partly unresolved, ranging from understanding tree community assembly and projecting forest responses to environmental changes, to assessing the management of forest ecosystems. We illustrated how different forest modelling approaches, due to their continuous development, their complementarity and mutual enrichment, represent an invaluable toolkit to address multiple ecological questions that require a renewed research effort.
The development of forest models crucially benefits from the interactions among scientists from various fields, within and across modelling communities, but also with field ecologists, physiologists, data scientists, computer engineers, remote-sensing researchers, and a variety of stakeholders. Owing to their long and successful history in integrating data and knowledge from these various sources, the models used to simulate forests have progressively reached maturity to tackle a broader array of ecological problems. For instance, forest models prove essential to understand the multiple drivers of forest productivity and biomass by combining field and remote-sensing data across space and time, and, as a result, provide informed quantification of carbon stocks and fluxes. Forest models also provide tractable platforms to perform virtual experiments still out of reach of empirical approaches on forest systems that are characterized by slow dynamics and large spatial extents. This notably allows to shed light on the complex links between forest biodiversity, functioning and resilience in the long term. Furthermore, forest models can disentangle the drivers of community assembly in forest communities, thus complementing theoretical approaches that typically remain limited to simplified systems. Last but not least, ongoing global change and the resulting biodiversity crisis, changing climate and disturbance regimes crucially increase the demand of informed projections on forest socio-ecosystems, for which forest models have a proven long history, while new developments allow for the integration of an increasing number of interacting factors.
We further demonstrated that the converging trajectories of the different modelling approaches used to simulate forests have provided new opportunities for comparisons among their outputs. This allows for the quantification of simulation uncertainties and the identification of their sources, and hence informs and fosters new model developments as well as empirical investigations. Overall, iterative model-data fusion approaches and the resulting cycles of simulation-assessment-improvement are continuously increasing the scope of model applications. Forest models will thus keep on contributing to a deeper understanding of forest structure and functioning, and they offer promising routes to fill remaining knowledge gaps and to take on future challenges of forest ecology.