One further example of problems of logical conclusions from sampling at only
one or two spatial scales will be considered. Suppose that dredging is being done
at several places in an estuary, to keep channels open for shipping. The sediments
in shipping channels are contaminated by heavy metals. Sampling is required in
order to detect any impact on benthic infauna in areas around the sites being
dredged (not in the sites being dredged – they are not just impacted, but the habitat
is actually removed). The concern is that fine sediments with associated heavy
metals may wash from the dredged sites to areas up to 100 m away. A replicated
study can be designed (as in Fig. 1.2), with several dredged areas and several
controls being sampled. In each area, replicated patches of sediment are sampled.
Natural spatial variation from patch to patch and area to area is estimated and
analysis can reveal any systematic difference between sites near dredging and
control sites.