Numerical classification of associations in subtropical evergreen broad-leaved forest based on multivariate regression trees―a case study of 24 hm2 Gutianshan forest plot in China
Abstract:Aims A 24 hm2 permanent plot in Gutianshan National Nature Reserve provided a valuable case for association classification in evergreen broad-leaved forest. Our objectives were to divide the forest community into associations to provide a new classification of evergreen broad-leaved forest and introduce the algorithm of indicator value for species in associations. Indicator species previously could not be quantified. Methods We used multivariate regression trees, based on topographic factors and species composition, for association classification. An indicator value was introduced to quantify the indicator species of associations. We named associations after the dominant species in the tree layer, followed by indicator species of lower tree layers. Important findings The forest community was divided into three associations that not only reflect temporal and spatial disjunctions of the community, but also correspond with features of the basic unit of vegetation classification. The three associations are 1) Raphiolepis indica + Chimonanthus salicifolius―Eurya muricata + Syzygium buxifolium―Schima superba + Castanopsis eyrei Association; 2) Rhododendron simsii + Rhododendron mariesii― Quercus serrata var. brevipetiolata + Corylopsis glandulifera var. hypoglauca―Pinus massoniana + Castanopsis eyrei Association; and 3) Vaccinium carlesii + Camellia fraterna―Distylium myricoides + Neolitsea aurata―Schima superba + Castanopsis eyrei Association.