A dynamical system to model land use interactions by using population dynamics: application on vulnerable grasslands

We introduce a deterministic set of equations, the Lotka–Volterra (LV) generalized equations, to explain or predict the temporal dynamics and competition between land use and land cover (LULC) classes. We evaluated this new approach in two regions of Uruguay where human activities have reduced natural grasslands. We infer the interaction coefficients from the LV model through the pairwise maximum entropy method using the spatial covariance matrices obtained from annual remote-sensing LULC maps. The model performed well in predicting short-term temporal changes, with errors similar to those obtained using traditional Markov chains. This model introduces a novel avenue for assisting LULC modelling in improving the temporal dynamics.