Abstract
The debate on the role of stochasticity is central to evolutionary biology, often summarized by the question of whether evolution is predictable or repeatable. Yet this " repeatability " or parallel evolution has been used as " evidence " of contradictory claims : that Darwinism is wrong (repetition indicates a form of intelligent design), that selection is all-powerful (in the modern synthesis), that the modern synthesis is incomplete (evo-devo), that chance matters (parallel evolution reveals that mutation is limiting), or that chance doesn't matter (parallel evolution reveals that evolution is repeatable). In this seminar, I will return to this question, showing what parallel evolution (or its absence) tells us about evolutionary processes. I will distinguish three types of stochasticity : the stochasticity of mutations and variations, of individual life histories and of environmental changes. I will then show how stochasticity can be important in evolution, distinguishing four main situations. (1) Stochasticity contributes to maladaptation or limits adaptation ; (2) it drives evolution on flat fitness landscapes (" freedom " evolutionary) ; (3) it can promote jumps from one fitness peak to another (" revolutions " evolutionary) ; (4) and it can shape selection pressures themselves. Stochasticity, by directly guiding evolution, has in fact become an essential ingredient of evolutionary theory, beyond the classic Wright-Fisher debates or that between neutralists and selectionists.