Many fishes on coral reefs are known to associate with particula~ microhabitats If these associations help determine population dynamics then w e would expect (1) a close assoclation between the abundances of these fishes and the abundances of the most frequently used mlcrohabitats and (2) changes in the abundance of microhabitats would result in a corresponding change In fish population sizes We examined habitat associations among obligate coral-dwelling g o b~e s (genus Goblodon) and then investigated relationships between the spatial and temporal ava~labillty of habitats and the abundances of Goblodon species among locations and anlong zoncs on the leef at Lizard Island (Great Barrler Reef) Out of a total of 11 Acropora species found to be used by Gobiodon, each specic3s of Goblodon occupied 1 01 2 species of Acropora significantly more often than expected from the availability of these corals on the reef Across reef zones, the abundance of most species of Gobiodon was closely correlated with the abundance of coral species most frequently inhabited However, the abundance of 1 species G a x~l l a r~s .iiras not conelated with the availability of most frequently used corals aci oss reef zones or among locations Foi G axlllarls the1 efoi e factors other than habitat availab~lity weie more important In determining differences In abundance among zones and locations Also corielations between the abundance of 2 species, G histrlo and G quinquestrigatus, and the coral species they most frequently used was less cons~stent anlong locations than among zones This suggests that the relative importance of habitat availability in determining the abundance of these species of Goblodon is reduced at the spatial scale of locations around Lizard Island The population dynamics of Gob~odon during this study were closely linked to the population dynamics of host corals Because of a crown of thorns starfish (Acanthaster plancl) outbreak the abundance of corals declined in paiticular reef zones Cor~esponding with t h~s loss of h a b~t a t was a decline in Goblodon abundance This study demonstrated that the abundance of habitat specialised fishes can be closely correlated with the spatlal and temporal availab~lity of suitable habitats among and w~t h i n reefs

Studies of coexistence in biotic communities have focused largely on local ecological processes. As a result, effects of regionally varying processes on community structure and their interactions with other processes operating locally have received considerably less attention. Here I investigate variation in predator abundance and species richness at a large spatial scale and associated comm u n~t y -w~d e patterns of species nchness and abundance of troplcal fishes on Australia's Great Barrier Reef. I constructed reefs from natural substrata to standard~ze the11 structure, isolation and history, w h~c h could otherwise be confounded between locations. Recruitment, both total abundance and species richness of recruits, to these reefs was greater at the northern location than at the southern one. In contrast, communities of resident fishes that developed on these reefs showed the oppos~te pattern; species richness and abundance were greater on the southern reefs. Piscivorous fishes were proportionately more abundant on the northern reefs. Therefore, predators were more abundant a t the location with greater recruitment but lower abundance and species richness of resident fishes. Also, the declines in species richness and abundance of fishes from the observed maximum in one year to the following observed minimum were related to average predator densities among reefs. These results suggest that one ecological process that varies between distant locations, in this case predation estimated by predator abundance, may override the effects of other ecolog~cal processes, in this case recruitment, In determining local patterns of coexistence. Furthermore, they suggest that understanding causes of local patterns of species richness and abundance may require information about processes that determine regional variation in ecological interactions.

Latitudinal variation in the dynamics of biological communities is among the least addressed and understood topics of modern ecology. Here, I examine the dynamics of coral-reef fish communities inhabiting small rubble patches at 2 locations separated by approximately 9" of latitude (One Tree Island and Lizard Island) on Australia's Great Barrier Reef. This sampling program ran simultaneously at the 2 locations using identical census methods, sampled over 16 mo and included 2 summer recruitment seasons. Total abundance of all fish species pooled fluctuated seasonally, peaking in summer Total abundance at sites wlthin locations differed, but sites at different locations overlapped In abundance. The dynamics of these communities were apparently unrelated to family membership among species, intraspeciflc abundances or diet. Rank abundances of species present at both locations did not differ, and only a slngle negative correlation of abundances between specles was detected. Changes in abundance, however, were density dependent. Maximum abundance of fishes observed in the first summer explained between 42 and 93% of the variation in subsequent per capita declines in abundance by the following winter and before recruitment had commenced in the second year. These density-dependent effects were observed across families and across widely separated locations. These results suggest that the dynamics of these communities were not structured by strong pairwise interspecific interactions, but, instead, may have been structured by some density-dependent process(es) that affected a broad cross sectlon of species and that operated irrespective of locality.