| Yhteenveto: | Most current techniques for classifying ecological community data are intended to typify discrete groups. However, ecological communities are not separated by distinct boundaries and some mixing between neighbouring communities occurs. Another aspect of these classifications is that they assume that the entities within each group are all equivalent, i.e. all of them will show the same characteristics or will have the same rank within the group. This is an over simplification since the structure of natural communities has been shown to vary as their component species respond more or less independently to environmental gradiente; thus, both overlapping and internal heterogeneity are importan! features of ecological communities that cannot be incorporated easily in conventional classification approaches. In this paper, it is proposed that fuzzy set theory provides a conceptual basis which overcomes the limitationsof conventional approaches.The fuzzy /c-means algorithm was used to classify a vegetation sample from the región of the biosphere reserve ‘La Michilía’, in the State of Durango, México. This fuzzy classification was compared with a conventional classification produced by the TWINSPAN program. It was reckoned that four groups were a suitable representaron of the vegetation of the area. It was interpretad that two of the groups were mixtures of deciduous tropical forest, xerophytic shrubland, grassland and oak-pine forest. The other two groups are oak-pine foreste. All of the groups can be arranged along a main gradient of aridity.The resulte suggest that a fuzzy set classification of vegetation data is appropriate and useful. It is shown that the groups formed give a sensible description of the ecological communities while also retaining the information on the natural variation and mixing between them. It is also demonstrated thatthese groups can be analysed in terms of their association with external variables, providing a good insight into the ecoiogy of the component species and the factors influencing the structure of each community.
|
|---|
