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.