Quadtree-based Seed Dispersal Simulations

Abstract

In this thesis we introduce a new model for simulating seed dispersal that makes use of quadtrees. We explain how our quadtree-based model works and what decisions were made when designing it. We evaluate our dispersal model by comparing the simulation outcome, the habitat encounter, the kin avoidance and the computation time to the discrete dispersal model from Treep et al. We evaluate how the threshold value, used to determine how the quadtree splits, affects the simulation. Subsequently, the models are evaluated and compared across various landscape sizes. In simulations where our dispersal model creates a quadtree with many cells larger than 1x1, the simulation outcomes will no longer match those of the discrete model. We conclude that the discrete model is faster than our model if it makes use of Fast-Fourier transform to convolve seed probabilities. But our quadtree-based dispersal model is faster than a naive implementation of a discrete dispersal model.