A review of dynamics in spatial savanna models

Spatially explicit ecosystem models are often used to study changing savanna ecosystems. Many different models have been made with different research goals. An overview of all these existing spatial savanna models is currently lacking. Therefore, we performed a systematic search to gather the literature so far on spatial savanna models. We grouped these models into three different categories of model types: partial differential equations, integrodifferential equations, and cellular automata. We also determined whether the model contained stochastic elements. In addition, we identified the presence of four different ecosystem mechanisms in each model: water, shading, herbivory, and fire. Finally, we determined which types of spatiotemporal dynamics the models generated, where we focused on pattern formation, temporal oscillations, and abrupt changes. We show that some ecosystem mechanisms frequently co-occur with each other, which is mainly based on the type of savanna along a rainfall gradient that is modelled. Moreover, we show that some ecosystem mechanisms frequently co-occur with certain spatiotemporal dynamics. Based on these overviews, we propose that future research should focus on developing a unified spatial savanna model, which considers all of the relevant ecosystem mechanisms along a rainfall gradient. Such a model could be used to predict the response of savanna ecosystems to a changing climate. • Spatial savanna models frequently include water, shading, herbivory, and fire. • Spatial savanna models can generate different types of spatiotemporal dynamics. • Some ecosystem mechanisms and spatiotemporal dynamics frequently co-occur. • A unified spatial savanna model is needed.