Three levels of physical disturbance were applied to corals in permanent 10
x10 m quadrats along a section of fringing reef at Lizard Island on the Gre
at Barrier Reef to investigate the response of fish assemblages. Tabular an
d corymbose corals were overturned and left in situ, reducing total hard co
ral cover from similar to 55% to similar to 47%, similar to 43%, and simila
r to 34%. Despite pre-existing associations with benthic cover, all fish gr
oups examined (pomacentrids, labrids, chaetodontids, and acanthurids) were
resistent to benthic disturbances at the level and scale at which they were
applied. Partial Mantel's tests, in combination with partial Canonical Cor
respondence Analysis enabled spatial and temporal variation to be factored
out from experimental effects. Most of the variation in the fish community
could be assigned to spatio-temporal variables, indicating that spatial str
ucture over the reef landscape may moderate localised disturbance effects.
This study indicates that coral reef fish assemblages may be more resistant
to disturbance than many correlative studies would suggest, and highlights
a need for further information on levels and scales of natural habitat dis
turbance in order to apply a structured approach to the experimental invest
igation of the importance of habitat in structuring coral-reef fish assembl
ages. (C) 1998 Elsevier Science B.V. All rights reserved.