Application of self-organizing neural networks to classification of plant communities in Pangquangou Nature Reserve, North China

Abstract

Vegetation classification is an important topic in plant ecology and many quantitative techniques for classification have been developed in the field. The artificial neural network is a comparatively new tool for data analysis. The self-organizing feature map (SOFM) is powerful tool for clustering analysis. SOFM has been applied to many research fields and it was applied to the classification of plant communities in the Pangquangou Nature Reserve in the present work. Pangquangou Nature Reserve, located at 37°20′–38°20′ N, 110°18′–111°18′ E, is a part of the Luliang Mountain range. Eighty-nine samples (quadrats) of 10m × 10m for forest, 4m × 4 m for shrubland and 1m × 1m for grassland along an elevation gradient, were set up and species data was recorded in each sample. After discussion of the mathematical algorism, clustering technique and the procedure of SOFM, the classification was carried out by using NNTool box in MATLAB (6.5). As a result, the 89 samples were clustered into 13 groups representing 13 types of plant communities. The characteristics of each community were described. The result of SOFM classification was identical to the result of fuzzy c-mean clustering and consistent with the distribution patterns of vegetation in the study area and shows significant ecological meanings. This suggests that SOFM may clearly describe the ecological relationships between plant communities and it is a very effective quantitative technique in plant ecology research.