Measuring environmental impacts affecting benthic habitats requires detection of
specific patterns of statistical interactions in data sampled before and after a potential
impact, in the potentially impacted place and in control or reference locations. This
is complex because ecological assemblages and populations vary at many spatial
and temporal scales. Here, we introduce methods to ensure appropriate, independent
replication of sampling at hierarchical scales in space and time. For statistical
analysis, the logic of sampling design is critical. Determining precision of estimates
and maximising power to detect impacts require care in the design, analysis and
interpretation of the relevant data.