Kimoto's Cπ has frequently been used as the similarity index of the animal communities, but it had not been always applied to such instances as the analysis of the macrobenthos communities, which are usually composed of many species having various types of life form and large differences of occurrences. The following three indices, therefore, are newly introduced, instead of Cπ.
1) Cπ', the similarity index as a modification of Kimoto's Cπ.
2) Rs, the correlation index between samples, and 3) Rsn, the similarity index of the relationship between samples.
Using these indices, a new statistical method is also proposed for the analysis of the coastal environment. The procedure comprises the spectral decomposition of the matrix [Rsn] and the Fuzzy Cluster Analysis of the results of the spectral decomposition.
The method is verified by artificial data generated with random variables and by actual data of the benthic macrofauna, which were collected from an estuary region off the Inumarugawa river, Oita Prefecture. Consequently, results of the present “Rs” prove to be more effective than those of “Cπ” in both cases, and the analysis of “Rsn” shows satisfactory results on the environments of the research area.
Advantages claimed for this method may be the following points. All collected species are equivalently treated with one another, and many data are simply processed, not setting any artificial standard, such as the dominant or indicator species. Many useful informations about the environments can be drawn from some ecological knowledges peculiar to each species of the cluster species group. From them it is possible to infer some environmental factors of the research area.
As the result, the present method is suitable for the analysis of macrobenthos, and the effect may probably be similar to those of the Principal Component Analysis, which is, in general, difficult to apply to these data, because the sample size is not sufficiently large and also the data generally far depart from the multivariate normal distribution.
View full abstract