Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4378691 | Ecological Modelling | 2007 | 8 Pages |
From 1997 to 2002, benthic macroinvertebrates were collected with a Surber sampler from the 1970 sampling sites from clean to intermediately polluted areas in streams in Korea. The Self-Organizing Map (SOM) was used for patterning and visualizing community patterns of macroinvertebrates collected on the national scale. The SOM accordingly classified the samples to eight groups of communities, and the groups firstly corresponded to the geographical regions of the sampling sites, distinguishing the samples from mountain streams with high species richness from the samples from lowland areas with low species richness. Relations between species richness and abundance showed three patterns according to the SOM groups: (1) high species richness and high abundance in the clean area, (2) intermediate species richness and high abundance in intermediately polluted area, and (3) low species richness and low abundance in severely damaged area. Biological indices (Shannon diversity index, evenness, EPT richness, EPT abundance, and BMWP score) were also accordingly differentiated among the different clusters. The clustered communities were correspondingly related to the patterns of species richness and abundance, and the trained SOM appears to be efficient in defining the patterns of ecoregions.