Dynamics and stability of spatially structured ecosystems

Overview & Synopsis

In the context of global changes, it is crucial to understand how ecosystems react to anthropic pressures. Theoretical ecology has already largely studied how populations recover from perturbations, but it still relies on relatively simple systems, which are typically isolated in space and species-poor. This thesis tackles this issue by investigating whether considering space explicitly affects our current understanding of ecosystem stability. More specifically, we explore the interplay between the spatial structure of perturbations and the spatial structure of the landscape and its effects on the response of meta-populations and meta-communities to disturbances.In the first chapter, we use laboratory microcosm experiments coupled with a mathematical model to study how the number and spatial clustering of local extinctions affects the dynamics of metacommunities. In particular, we focus on patterns of species diversity and show that local extinctions can increase both local and inter-patch diversity, but that the contribution to each level of diversity depends largely on the spatial structure of extinctions. In the second chapter, we investigate the impact of a metapopulation's spatial structure on its stability. We focus on spatially explicit metapopulations in which local dynamics are bistable and explore the resulting landscape-scale stability. We show that local catastrophic shifts can spread in space until they affect the whole metapopulation, and that local bistability results in landscape-scale bistability with a strong hysteresis. We highlight that the size of the hysteresis depends strongly on the metapopulation spatial structure and that previous studies using a simple one-dimensional spatial structure had underestimated the hysteresis than can be observed in more realistic landscapes. In the third chapter, we reflect on which layers of complexity should be considered to better understand catastrophic shifts by reviewing existing work and highlighting promising approaches for future developments. In the fourth chapter, we present such an approach that allows to reduce the dynamics of a spatially-structured metapopulation to a single dimension.We investigate how well this reduction works in metapopulations with different structures (random, modular, dendritic), and show that the landscape-scale equilibria and temporal variability of complex spatial networks can effectively be predicted by a one-dimensional equation.