ANALISIS DE LA VEGETACION EMPLEANDO LA TEORIA DE CONJUNTOS DIFUSOS COMO BASE CONCEPTUAL.

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 gradients; thus, both overlapping and internal heterogeneity are important 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 limitations of conventional approaches. The fuzzy k-means algorithm was used to classify a vegetation sample from the region of the biosphere reserve 'La Michilía', in the state of Durango, México. This fuzzy classification was compared with a conventional classification produced by the TWlNSPAN program. It was reckoned that four groups were a suitable representation of the vegetation of the area. It was Interpreted 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 forests. All of the groups can be arranged along a main gradient of aridity. The results 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 that these groups can be analysed in terms of their association with external variables, providing a good insight into the ecology of the component species and the factors influencing the structure of each community. [ABSTRACT FROM AUTHOR]