Quantitative approaches to identify floristic units and centres of species endemism in the Qinghai-Tibetan Plateau, south-western China

Aims To determine floristic units and centres of plant species endemism using quantitative methods and different taxonomic groups. Location The Qinghai-Tibetan Plateau in south-western China - the world's highest and geologically youngest plateau. Methods The plateau was divided into 260 grid cells, each 1 degrees x 1 degrees, and the numbers of different seed-plant taxa, including genera, total species, endemic species and alpine species, were recorded in each cell. Hierarchical clustering and non-metric multidimensional scaling ( NMDS) ordination were used to divide the flora quantitatively, and ANOSIM ( analysis of similarities) was used to evaluate differences between the derived groups in the composition of taxa. The stress value was used to assess the fitness of the NMDS ordination. Species endemism was measured using corrected weighted endemism, and centres of species endemism were delimited on the basis of floristic units. Results The plateau harbours a total of 12,058 seed-plant species and infraspecific taxa in 1619 genera; 38.2% of the species are endemic to the plateau and 27.7% are alpine plants. When two or four groups were considered, cluster analysis based on genera gave the best definition, whereas when 12 groups were considered, analysis of endemic species gave the best definition. The flora of the Qinghai-Tibetan Plateau represents two floristic subkingdoms, four floristic regions and 12 floristic subregions. Three centres of species endemism were identified, associated with five floristic subregions. Main conclusions Quantitative analysis can reveal associations within a flora, and partly confirm intuitive classifications. Genera provide clear divisions of higher floristic ranks, and endemic species give the clearest classifications of lower floristic ranks. Alpine species are important taxa for classifying the flora of mountainous regions. A centre of species endemism based on a floristic unit can be conveniently identified and delimited, and the spatial scale of this centre is determined by the taxonomic rank used.