Autocorrelation and variance of ecological processes as causes of change of biodiversity in marine coastal habitats

Disentangling the effects of the vast array of natural and human processes that drive variation in marine coastal biodiversity is a daunting task. Traditionally, ecologists have looked at the effects of individual drivers of change, although this approach seems not to be appropriate to explain the extreme variability of ecological systems over a broad range of spatial scales. The present thesis focuses attention on cross-scale interactions between multiple drivers, on the basis of previous research suggesting that the spectrum of environmental variance is a distinctive feature of processes generating variability at different spatial scales. Therefore, I propose that natural heterogeneity reflects interactive effects of small-scale processes characterised by white spectra of variability, embedded within processes dominated by red spectra, which display autocorrelation over large spatial or temporal scales. In addition, I examined the role of interactions among environmental forces in different marine systems, assessing whether processes operating at different scales are more likely to interact than processes operating at similar scales. These hypotheses were tested using a combination of long-term observations and large-scale field experiments. Observational data originated from publically available resources and from long-term observations in rocky-shore assemblages at Calafuria (south of Livorno). Experiments were repeated in three habitats (intertidal and subtidal algal stands and seagrasses) and consisted in disturbing algal canopies and entire assemblages along transects in factorial combinations, to simulate realistic spatial red and white noise processes, respectively. Statistical models that included interactions among predictor variables generally explained a larger amount of variability in response variables than additive models. On the contrary, interactions among processes with different spectra were not more likely than interactions among processes with similar spectra. Hence, the specific nature of potentially interacting environmental drivers must be taken into account when trying to explain the causes of variation in distribution and abundance of marine organisms, regardless of the scales at which these processes interact. Experiments revealed interactive effects of processes characterized by different levels of autocorrelation, indicating that the effects of environmental drivers operating at different scales are not additive and cannot be predicted by examining each driver in isolation. These results are relevant in the context of climate change, as an increase in the levels of autocorrelation is expected for several climate variables. The antagonistic nature of interactions among environmental variables with different variance spectra suggests that local scale processes can mitigate the effects due to increased autocorrelation in climate variables. In addition, the evidence of antagonist interactive effects of environmental processes suggests strategies for scaling-up local-scale observations to larger spatial extents