The spatio temporal forest landscape model TreeMig is presented. It is based on a forest dynamics model that incorporates spatial variability by frequency distributions for tree densities and light intensities. It also takes into account seed production, intra-specific density regulation and seed dispersal. As a case study, the change of the tree species patterns in the Central-Alpine region of Valais was simulated with the model, using climate anomaly and immigration scenarios. The simulations were run on a grid with 1km x 1km resolution and with a yearly time step during the Holocene in the highly structured and heterogeneous environment of the valley of Valais in the Alps. As input, a scenario of temperature anomalies in the Holocene, spatially interpolated climate data, and times of species immigration into the simulation area were used. The results show a vivid pattern of species spread, changes of dominance, and up and down shifts of the timberline, which are triggered by the variability of the external factors but exhibit endogenous dynamics, i.e. migration and succession, after drastic changes of the boundary conditions, such as immigration of species into the simulation area or strong climate changes. However, on the observation scale no purely endogenous effects such as pattern formation can be observed.