The concept of the fundamental niche of a species has so far been tackled in a rather descriptive way, leading to the widely used correlative niche models (or correlative species distribution models). I will present how we can move towards a more analytic and systemic approach of the niche, leading to the development of process-based niche models. I will present a process-based species distribution model we are developing for forest tree species, and some key results, among which a validation of the model by hindcating. I will present also briefly some new applications of the model to study the genetic evolution of the niche.