There is both theoretical and empirical evidence that in areas where savannas and forests coexist, the movement of fire through the landscape depends on the spatial configuration of these coexistence states (Schertzer et al 2015, Van Nes et al 2018, Newberry et al 2020). Because of the importance of fire for savanna and forest resilience (Staver et al 2011), this implies that forest-savanna patterning affects savanna and forest resilience via fire. Depending on the shape, size and distribution of forest and savanna patches associated with fuel load, fire may or may not percolate through the savanna and forest for certain levels of coexistence (Van Nes et al 2018). We will first use our spatially explicit savanna model to generate hypotheses of how different spatial patterns of fuel connectivity in theory encourage or evade tipping points via fire. Next, we will test these hypotheses empirically using satellite images of forest-savanna boundaries. We will compare the patterns of static forest-savanna boundaries with ones in areas where forest is encroaching into savanna due to fire reductions, as is observed in Africa (Venter et al 2018), and with ones in areas where savanna is expanding into forest due to fire increases, as is observed in the Amazon (Flores and Holmgren 2021). This allows us to understand savanna and forest resilience and predict the movement of forest-savanna boundaries based on snapshots of coexistence states.

Projects under this theme

The role of fire feedbacks in spatial resilience of forest-savanna boundaries

Max van der Ree

In this research project, we study the mechanisms through which savanna-forest spatial patterns come about and evolve through time. Such mechanisms not only include feedbacks between fire and tree cover, but also seed dispersal and underlying environmental heterogeneity such as soil conditions. We employ a wide range of techniques, from modelling to remote sensing and field-based measurements, in an attempt to create a synergy between theory and empirics. We’ve developed a model with a high spatial resolution, making it possible to represent individual trees and track demographic developments within the simulated forest, as well as fine-scale tree-fire interactions. The model incorporates spatially explicit fire spread and realistic dispersal by birds and wind, allowing us to study how dispersal modes and fire size and frequency affect savanna-forest development. Furthermore, we plan to extend the model to enable experiments with environmental heterogeneity, such as spatial variation in soil fertility. Moreover, we aim to validate the model using field measurements and remote sensing data. To conclude, by combining different techniques, we aim to increase our understanding of the scale as well as the drivers of savanna-forest bistability, enabling more accurate predictions of how these ecosystems may respond to global change.