Sewage and environmental impacts on rocky shores: necessity of identifying relevant spatial scales

The concentration of contaminants usually decreases with increasing distance from a point-source disturbance, so sampling to detect ecological impacts is usually done at 1 spatial scale, often at regular intervals from the point of discharge. There is, however, concern that the choice of an inappropriate scale will cause failure to detect impacts or failure to identify and estimate the size of impacts. In this study, the putative impact of a shoreline sewage outfall on the abundance of green ephemeral algae and gastropods was sampled at 2 spatial scales (tens of metres and several kilometres from the point of discharge) in order to determine whether the ecological impact of effluent was comparable across these, as would be expected if the abundance of species follows the gradient of contaminants. Such sampling also enabled the putative impact of this outfall on the spatial variability of taxa to be examined at 3 spatial scales: (1) among quadrats in the site with the outfall compared to variance among quadrats in other sites on the shore with the outfall; (2) among quadrats in non-outfall sites on the shore with the outfall compared with variance among quadrats in sites on control shores; (3) between non-outfall sites on the shore with the outfall in comparison to among sites on the control shores. A greater abundance of Enteromorpha spp. was found close to the outfall than further away at both spatial scales. Patterns in the abundance of many other taxa differed between the 2 spatial scales of sampling. The density of the limpet Patelloida latistrigata was much greater close to than far from the outfall, when considered on a large spatial scale. At the smaller scale among sites on a single shore, the impact was completely reversed-densities were much smaller close to than away from the outfall. Variances, like abundances, did not always follow the gradient of contaminants and different patterns were often seen at different spatial scales. Thus, putative impacts should be sampled on multiple spatial scales using nested sampling designs. Where this is not possible, the spatial scale at which an impact might be detected or interpreted needs to be clearly stated because the generalisation that a disturbance has a similar impact at all spatial scales relevant to the population being studied cannot be made without explicit tests.