Thanks to new technologies, it is possible to make an automatic robotic treatment of plants for the mildew in greenhouses. The optimization of the scheduling of this robotic treatment presents a real challenge due to the continues evolution of disease level. The conventional optimization methods can not provide an accurate scheduling able to eliminate the disease from the greenhouse. This paper proposes a solution to provide a dynamic scheduling problem of evolutionary tasks in horticulture. We first developed a genetic algorithm (GA) for a static model. Then we improved it for the dynamic case where a dynamic genetic algorithm (DGA) based on the prediction of the task amount is developed. To test the performance of our algorithms, especially for the dynamic case, we integrated our algorithms in a simulator.