Modelling plant community dynamics in changing forest ecosystems: a review

Modelling plant community dynamics in a context of changing environment is an old task but still a timely challenge, especially for the most complex of them, namely forest plant communities. Progress in mathematics and computer science has allowed to incorporate an increasing number of parameters and to make models more and more realistic, often to the detriment of their analytical tractability. Here, we successively review the key aspects of forest plant community dynamics, and how they are accounted for by demographic and spatial models. Demographic models address the dynamics of a community at a local scale, and can be either continuous or discrete in time, purely deterministic or stochastic. Scaling up the dynamics of local communities to a landscape scale supposes to use spatial models, which can incorporate spatial heterogeneity and species dispersal. Aside spatially-implicit models, spatially explicit models treat space either as a continuous or a discrete variable, leading to different mathematical formalisms. Compared to statistical models, mechanistic models have a higher predictive power under changing environment (i.e. model extrapolation as opposed to model interpolation for which statistical models perform better) and are thus increasingly popular. However, as their complexity increases, their mathematical tractability decreases and mechanistic models are chiefly used for simulations. In this review, we analyse the pros and cons of the most widely used modelling frameworks in plant ecology and finally provide some perspectives.