As a solution, rather than employing proximity as a primary
selection criterion, we echo the recommendation of (Stewart-
Oaten, 1996) that, in the design of impact studies, control sites
should be chosen that are similar to the impacted sites in “features
judged to be important” natural determinants of the species’
distribution. This in itself is no easy task as (Ronnback et al.,
1999) point out (but did not empirically determine) that the
nekton species composition of a mangrove intertidal site may be
influenced by many habitat features such as “spatial variations
in elevation, topography, proximity to open water, structurally
complexity of the root system and, sediment characteristics”.
Our trimming study did suggest that the shoreline slope might
also play a strong role in determining local fish abundance and
community composition (Ellis, unpublished data), as it does in
other coastal systems (Allen et al., 2007). Consequently, shoreline
slope might be considered as a possible criterion for future
selection of control sites in intertidal fish research. Such selection
will limit the inference space of the studies to sites that share the
“important” features, but potentially improve the tests’ sensitivity.
Alternatively, site-specific characteristics can be included as
covariates in the analysis of impact data improving the tests’ ability
to distinguish the effect of anthropogenic stress from natural
variability (Parker and Wiens, 2005). Clearly, more research needs
to be directed at identifying the primary determinants of fish
distribution within mangrove-forested systems