• Biofilm
• Critical slowing down
• Early warning signals
• Epilithic Microphytobenthos
• Intertidal rocky shore
• Macroalgal canopy
• Regime-shift
• Remote sensing
• Tipping-point

The prospect to anticipating abrupt shifts in ecosystem has called for the development of early warning signals able to predict the proximity to a tipping point. There is a growing awareness that increasing anthropogenic pressure on natural systems may trigger a regime shift at the global scale. Early warming signals of regime shifts would be a powerful tool to anticipate undesirable transition, but experimental tests of their validity in natural fluctuating conditions remain rare. My thesis aims at understanding the effects of regime shifts and loss of resilience in rocky shore ecosystems, and to identify suitable and effective early warning indicators under naturally fluctuating conditions. I focused on two rocky intertidal systems: epilithic microphytobenthos (EMPB) and highly productive assemblages dominated by the canopy alga Cystoseira amentacea var. stricta.
In Chapter 2 and 3 I combined modelling with experiments to explore the spatio-temporal dynamics of rocky intertidal biofilm in response to warming. In chapter 2, I developed a simple mathematical model showing that biofilm exhibits a regime shift and alternative states in response to an increase in air temperature. In chapter 3, I experimentally evaluated this prediction by exposing rocky intertidal biofilms to a gradient of increasing temperature to induce systemic regime shifts and to assess the performance of temporal warning signals of impending regime shifts in natural conditions. My results showed that recovery time from small perturbations increased along the gradient of increasing temperature. However, early warning indicators failed to signal this pattern. These results provide experimental evidence that recovery time is more reliable than indirect early warning signals in signalling tipping points.
In Chapter 4, I used rocky intertidal biofilms to evaluate the performances of spatial and temporal early warning signals under an intensifying gradient of grazing pressure. Results supported the existence two alternative states in response to grazing pressure: a high- and low-biomass state. Empirical reconstruction of basin of attractions indicated that the resilience of the high biomass state lowered at increasing levels of grazing pressure. Moreover, this experiment provided some support to for theoretical prediction of increasing variance in time series prior to a tipping point.
In Chapter 5 I used rocky intertidal communities of algae and invertebrates with well-characterized alternative states as model systems to experimentally test spatial early warning signals in naturally fluctuating environments. These systems consist of highly productive algal canopies that maintain species rich understory assemblages of sessile organisms. I tested a novel spatial indicator, the recovery length, defined as the distance effect of a spatial perturbation vanishes. To this aim, I conducted a canopy perturbation experiment to test the hypothesis that the spatial scale at which a system recovers from a pulse perturbation in space should decrease as the system approaches the tipping point. Empirical estimates of recovery length, a recently proposed spatial indicator of critical slowing down, were obtained by comparing the spatial scale at which algal turfs propagate from low quality patches to high quality regions with decreasing canopy cover. I found that recovery length and spatial early warning indicators increased markedly along the gradient in canopy degradations, providing field-based evidence of spatial signatures of critical slowing down in naturally fluctuating conditions.