The spatial distributions of marine fauna and of pollution are both highly structured, and thus the
resulting high levels of autocorrelation may invalidate conclusions based on classical statistical
approaches. Here we analyse the close correlation observed between proxies for the disturbance associated
with gas extraction activities and amphipod distribution patterns around four hydrocarbon
platforms.We quantified the amount of variation independently accounted for by natural environmental
variables, proxies for the disturbance caused by platforms, and spatial autocorrelation. This allowed us to
demonstrate how each of these three factors significantly affects the community structure of amphipods.
Sophisticated statistical techniques are required when taking into account spatial autocorrelation;
nevertheless our data demonstrate that this approach not only enables the formulation of robust
statistical inferences but also provides a much deeper understanding of the subtle interactions between
human disturbance and natural factors affecting the structure of marine invertebrates communities.
