indicate that unless samples are collected at the same point in the
tide cycle at all sites, small scale temporal variability will add substantially
to the “noise” over which an impact “signal” must be
detected (Grippo et al., 2007). Thus, attempts to increase the statistical
power of an experimental design by increasing the number of
replicate samples collected during a sampling interval will not be as
helpful as desired if samples are not collected at the same point in
the tidal cycle. Commonly, as was the case for our study, sampling
is performed sequentially at several study sites within a specified
time interval around the desired tide stage (e.g. ±1 h of high tide).
This practice, although often necessary for logistical reasons, may
be counterproductive as non-simultaneous sampling will add to the
observed variation in fish abundance and biomass, making detection
of subtle impacts even less likely. Thus, in tidally influenced
areas, imprecise sample timing can conflate spatial variability and
further impede detection of impacts (Ellis and Bell, 2008).