Variation in assemblages of fish associated with deep and shallow margins of the seagrass Posidonia australis

Little is known about the fish fauna of deep Posidonia australis seagrass beds in Australia. We investigated associated with the deep and shallow margins of P. australis during two surveys in Jervis Bay, New South Wales. Fish associated with deep and shallow seagrass at each of two areas within a large bed of P. australis were compared over a period of 3 mo (November 1986 to January 1987). This was followed by a larger scale survey in which we compared fish found at the two depths within three large beds of P. australis in Jervis Bay over 2 yr (September 1988 to June 1990). In the smaller scale survey there were marked dissimilarities in relative abhundance and composition of species between samples from deep and shallow P. australis. There were also significant differences in abundance between deep and shallow seagrass on most occasions for the majority of the dominant fish species. As a result of the small-scale survey, we concluded that differences in species composition of fish associated with deep and shallow P. australis may have been due to the presence of species from nearby bare substrate among the sparser leaves of the deep seagrass. The results of the smaller scale survey were not confirmed by the larger survey. The relative abundances of species in samples from deep and shallow P. australis were not dissimilar for all beds. In addition, it was rare for any of the dominant species to have significantly different abundances in deep and shallow seagrass. There was little evidence that the species composition of fish associated with deep P. australis was strongly influenced by the fauna of deep bare substrate. Rather, samples from deep P. australis were far more similar to those from shallow P. australis than to a comparable set of samples collected from deep bare substrata. This study demonstrates how the results and conclusions of small-scale survey work, a feature of many environmental impact asseesments, can be misleading when applied to larger spatial and temporal scales, even within the same bay.