MEPS 247:225-235 (2003)  -  doi:10.3354/meps247225

ABSTRACT: Inferences about the effects of habitat loss may be derived from comparisons of different places when pre-impact data are unavailable. We compared the fish faunas of sandy substrata off beaches with and without seagrass in Cockburn Sound (Western Australia), a bay where >3000 ha of seagrass was lost from the 1950s to the 1970s, and for which there are no pre-impact data. Numbers of species and individuals varied considerably in space and time, and this variation obscured patterns that might have been associated with the presence or absence of seagrass. Multivariate tests indicated some significant differences in species composition between beaches with and without seagrass; however, patterns among species were complex. Catches were strongly influenced by depth and the quantity of drifting vegetation. In general, fewer species were captured at shallower beaches. Hauls with a lot of drift captured more species, more individuals and higher biomasses of fishes that eat invertebrates and are associated with plants. When widespread seagrass loss occurs, resultant decreases in the supply of drift are likely to negatively affect densities of some fauna. We can make no firm conclusions about the effects of seagrass loss in Cockburn Sound, but suggest that declines in the production of drift have caused changes in fish assemblages.

KEY WORDS: Australia · Cockburn Sound · Drift · Guilds · Inference · Spatial subsidy