Detecting ecological groups from traits: a classification of subtropical tree species based on ecological strategies

We evaluated the frequently assumed but rarely tested hypothesis that in tropical and subtropical forests species form discontinuities along gradients in trait variation, which can be detected to classify species into ecologically meaningful and statistically defined groups. We also tested the hypothesis that the dominant conifer Araucaria angustifolia have a contrasting trait syndrome from the pool of angiosperms. Data were collected in subtropical mixed conifer-hardwood forests in southern Brazil. Eleven trait variables (relative growth rates, growth rates under favourable conditions, annual mortality rates, seed length, seed dispersal mode, wood density, maximum height, crown depth, stem slenderness, specific leaf area, and leaf blade length) were measured for 66 large tree species. The conifer Araucaria angustifolia differed significantly in most traits and vital rates from the angiosperm distributions. The Simple Structure Index criterion was used to identify the optimum number of non-hierarchical k-means groups. This index was largest for a solution with five groups. Non-hierarchical groups were more strongly related with the resource capture and height gradients than groups formed by hierarchical clustering. We propose the recognition of seven ecological species groups in the studied forests depending on growth rates, leaf size, wood density, height, stem slenderness, dispersal, and crown depth: Conifers, Palms, Pioneers, Large-seeded Pioneers, Wind-dispersed Large trees, Large shade-tolerant trees, and Small Shade-tolerant trees. The classification produced was coherent with the ecological strategies present in the community and represents a subdivision of Westoby's leaf-height-seed plant strategy scheme.