A comparison of Self-Organizing Feature Map clustering with TWINSPAN and fuzzy C-means clustering in the analysis of woodland communities in the Guancen Mts, China

摘要

SOFM (self-organizing feature map) clustering is powerful in analyzing and solving complicated and non-linear problems. This method was used and compared with fuzzy C-means clustering and TWINSPAN, the most common classification methods, in analysis of plant communities in the Guancen Mts., China. The dataset consisted of importance values of 112 species in 53 quadrats of 10 m x 20 m. All the three methods classified the 53 quadrats into eight groups, representing eight associations of vegetation. They were all effective in the analysis of ecological data. The consistency of SOFM clustering with fuzzy C-means clustering (FCM) and TWINSPAN classification was 81.1% and 94.3%, respectively. SOFM clustering has some advantages and more potentiality in application to studies of ecology.