COMPARISON OF RSN METHOD WITH CORRESPONDE NCE-ANALYSIS BASED ON ACTUAL DATA ON BENTHIC MACROFAUNA

We have indicated some advantages of the Rsn method over correspondenc e analysis (CA) by numerical experiments on simulated data. In order t o confirm these advantages over CA with actual fauna, we used publishe d data on macrobenthos. For the comparison between CA and Rsn ordinati on, fuzzy cluster analysis was applied to each result of the ordinatio n. Using the result of the fuzzy clustering of sites and species, we p roposed some practical methods for the evaluation of the correspondenc e between site groups and species ones. With these methods, it was sho wn that the clustering of sites based on Rsn ordination corresponded b etter with that of species and gave more interpretable results than th ose based on CA. This means that Rsn ordination reflected more sensibl y the species distribution patterns relatively free from noise. It was also shown that the fuzzy clustering of sites and species could provi de useful tools to evaluate the correspondence between site groups and species ones. Therefore, the Rsn method together with fuzzy cluster a nalysis of sites and species is applicable to actual data such as bent hic macrofauna.