|
Describing, classifying and quantifying vegetation communities is fundamental for understanding their current distribution, rarity, interrelationships and ecosystem functions. In the present study, we apply a consistent objective classification system for ephemeral wetlands of arid and semi-arid areas of New South Wales (NSW), Australia. Our approach uses a two-step statistically based, hierarchical, multiscale classification of environmental data at broad scales and floristics data at intermediate scales. At broad scales, ecoregionalisation methods were used to describe three wetland macrogroups. Within these groups, we performed unsupervised analyses of 640 floristic survey plots using the Bray-Curtis algorithm, clustering by group averaging and testing of clusters using similarity profile analysis (SIMPROF). From this we delineated 18 vegetation groups with class definition based on a combination of diagnostic and nondiagnostic similarity percentage analysis (SIMPER) outputs and dominant taxa. We show that a consistent classification system can be effectively created for subsets of vegetation that have adequate plot data within a general matrix that is poorly sampled if outputs are restricted to appropriate scales of resolution. We suggest that our approach provides a stable and robust classification system that can be added to as more data become available. |
|