The distribution and value of early warning signals in food webs under stress

Stress caused by human activity has many impacts on ecosystems, which may result in extinctions and tipping points to detrimental alternate stable states. Early warning signals (EWS) are a tool developed in simple theoretical systems, to predict when extinctions or tipping points are approaching. Current understanding of EWS in complex systems, which better represent natural ecosystems, is limited. Through a series of simulation experiments, this thesis provides insight into the expected strength and distribution of EWS in complex food webs, providing greater applicability for EWS empirically. Chapter 2 assesses the ability of EWS, biomass-based metrics and species’ topological properties to predict where and when extinctions will occur in a food web under stress. EWS were found to be the best predictors, indicating distinctively low resilience in species approaching extinction. This provides value for EWS in complex food webs. Chapter 3 uses a network’s topological properties to predict EWS distribution, finding that specialists of low trophic levels have the greatest capacity for EWS. This makes such species ideal candidates for monitoring, particularly under widespread stress. This is important where monitoring the entire network is impractical. Chapter 4 explores how the location of localised stress impacts EWS distribution, and how EWS propagate from species directly under stress. Here, changes in biomass flux were found to have a crucial role. Change in biomass flux is mediated by a species’ trophic relationship with directly stressed species. This provides crucial context for how a species may be being impacted by stress, based upon the observed EWS. In Chapter 5 I discuss how the key findings of my work, such EWS capacity and the role of biomass flux, can be utilised to improve our understanding of EWS in complex systems. Valuable future research directions include understanding EWS cascades and the impact of more complex stress regimes.