Comparison of larval fish assemblages in three large estuarine systems, KwaZulu-Natal, South Africa

 The structure of the larval fish assemblages in three large estuarine systems on the KwaZulu-Natal coast of South Africa was examined using a combination of univariate, distributional and multivariate techniques. The database was comprised of a full annual set of larval fish samples taken from each estuarine system: Durban Harbour, Richards Bay Harbour and St Lucia Estuary. The mean monthly densities of each species in each system were used in the species matrix, and the mean monthly values of salinity, temperature and turbidity were used in the environmental matrix. The mean species diversity and evenness index were significantly higher in Durban Harbour (H′ = 1.03, J′ = 0.65) than in the other two systems. The cumulative dominance curve showed that St Lucia Estuary has a high dominance of a few species, with Richards Bay Harbour intermediate and Durban Harbour being the most diverse. Classification and MDS (multiple-dimensional scaling) analyses of larval fish densities in all three systems grouped together into three main clusters on the basis of system. The species similarity matrix (inverse analysis) clustered into five groups at the 25% similarity level. The MDS analysis of the same matrix showed that the groups separated out according to the degree of estuarine association of a species and hence habitat type. The species most responsible for system groupings were: Glossogobius callidus, Gilchristella aestuaria, Stolephorus holodon, Croilia mossambica and Gobiid 12. The “best fitting” of the environmental variables to explain the larval fish community patterns in each system was turbidity on its own (weighted Spearman's rank correlation, ρ_w = 0.55). The relationship of larval densities to environmental conditions was shown to be species-specific with estuarine species (e.g. G. callidus and G. aestuaria), having a strong positive correlation with temperature and turbidity but negative correlations with salinity. In summary, much longer term studies with more sites within each system are needed to assess whether the larval fish assemblages are stable or at an equilibrium (both spatially and temporally) and whether these assemblages are indicative of the relative “health” and nursery function of the system. Received: 28 April 1999 / Accepted: 30 May 2000