Introduction

Forests cover about 30% of the Earth’s land surface, store almost half of the terrestrial carbon, constitute a net carbon sink, supply important resources to billions of people, and host over half of Earth’s biodiversity (Pan et al. 2011; Jenkins et al. 2013; Vira et al. 2015; Ramage et al. 2017). Yet, ongoing and future environmental changes put forests at risk. This rises the demand for a more detailed understanding of forest dynamics and for assessing the future of forest ecosystems to continuously update our knowledge base and support decision-makers (United Nations 2014; Mouquet et al. 2015; IPBES 2016; Mori 2017). Forest ecology is however confronted with the challenge of investigating complex systems that are characterized by long-term dynamics over large spatial scales, and therefore many questions remain unresolved (Sutherland et al. 2013).
During the past decades an increasing amount of field and remote sensing data has been made available, providing valuable information on forest ecological systems at various spatial and temporal scales and resolutions. However, their integration into a coherent picture remains a considerable challenge (Levin 1992; Chave 2013; Estes et al. 2018). In parallel, a variety of vegetation and forest models have been continuously developed by different scientific communities and for different purposes. Orchestrating the interplay of various data with forest modelling has been identified as a promising approach to tackle current research challenges (Zuidema et al. 2013; Shugart et al. 2015; van der Sande et al. 2017). The availability of various forest modelling approaches and decades of experience in assimilating observational knowledge offer invaluable tools to address key applied and fundamental ecological questions on forests.
Our contribution draws on the simultaneous trajectories of development that different modelling communities have followed until today to evidence the powerful capabilities models offer to understand forest ecology. In the first part, we present three widely used, but contrasting, modelling approaches to simulate forests, namely species distribution models (SDMs), individual-based forest models (IBMs) and dynamic global vegetation models (DGVMs). Our aim is to illustrate the diversity of modelling approaches to address key ecological questions for forests. In doing so, we outline the development of these different modelling approaches over the past decades. In a second part, we discuss how scientific and technical advancements have been alleviating the main constraints that initially restricted applications of forest modelling, and show how model development have progressively allowed models to tackle questions beyond their historical objectives. Finally, we will sketch out how forest models, singly and in combination, could take on an increasing role in addressing a variety of key ecological questions in the future.