Although it is acknowledged that the spatial scale of sampling and experime
ntation mentation is important in assessing the degree of stability of a co
mmunity, multiple scales are rarely incorporated into the same study. We co
mpared the persistence and resilience of a temperate blennioid fish guild a
t three scales spanning an order of magnitude. Replicate permanent square q
uadrats of three sizes (4 m(2), 25 m(2), 100 m(2)) were delimited within co
ntiguous tracts of reef in two habitats: broken reef and macroalgal forest.
Two forms of disturbance were applied to the treatment quadrats. Direct di
sturbance (fish removal) was applied to treatments in the broken reef habit
at. Indirect disturbance (macroalgal removal) was applied to treatments in
the macroalgal forest. Both experiments identified several scale-dependent
effects. Large quadrats were less variable than smaller quadrats, probably
due to reduced edge effects due to movement of fish in and out of the quadr
ats. The smallest quadrats did not adequately represent the available subst
ratum types within the reef zone. This interacted with species-specific beh
avioral factors such as mobility and habitat association to consistently bi
as the blennioid composition within the smallest quadrats toward dominance
of the most sedentary blennioid species. Fish responded to disturbance in d
ifferent ways depending on the stage of their life cycle. Adult fish respon
ded at scales within their motility range of only a few meters. In contrast
, newly settling fish responded to the habitat at larger scales. Quadrat si
ze also determined which assemblage dynamics could be measured. Within-year
changes in guild composition due to species-specific settlement timing and
survivorship could be measured only at larger scales. We conclude that dif
ferent scales of monitoring and manipulation will determine what can be per
ceived by the researcher as well as how the organism will respond to distur
bance.