No abstract

On the basis of repeated censuses of unmanipulated assemblages of fish on 20 lagoonal patch reefs. Sale & Douglas (1984) proposed a simple random placement model to account for the structure of such reef fish assemblages. The present paper uses their data augmented by additional censuses and a series of simulation models to explore the adequacy of their proposal. The distributions of a small proportion (about 8 species) of the 143 species recorded on the reefs, but including some of the most abundant species, could not be explained by a process of random colonization. These species vary greatly in abundance among censuses, and tend consistently to prefer certain reefs. The remaining species, while clearly possessing some microhabitat requirements, and perhaps responding to each other's presence, show distribution across patch reefs which are satisfactorily explained by a model of random placement of juveniles from a pool of potential colonists.

Data on the distribution of 25 common species of fish over 20 patch reefs located in a coral reef lagoon (Great Barrier Reef) were collected by 20 visual censuses over 10 yr. Associations among the 25 species were examined by cluster analysis. The overall pattern of association changed substantially from census to census. So too &d the relationships among small subsets of the 25 species which were tightly associated at one or more censuses. The degree of association among groups of species was compared with that expected under a null model of non-interactive recruitment of fish to patch reefs. Species in many apparently closely associated small groups were less strongly associated than expected under the null model, suggesting that species react differently to the habitats offered by the patch reefs, or respond negatively to each other's presence. Results are discussed in the light of other information on the structure of these assemblages of fish.

No abstract