Semi-supervised delineation of riparian Macrogroups in plot deficient regions within eastern Australia using generalised dissimilarity modelling

Abstract

In areas of poor plot data, limited funding and expertise, alternate approaches are needed to create elements of a hierarchical classification schema to assist in landscape planning. This is especially important for vulnerable systems under pressure from human activities. Within this paper we introduce an approach to help create a consistent classification section for riparian vegetation at a subcontinental scale, within the context of low plot data availability. We collated occurrence data for selected dominant plants known to occur within riparian environments from electronic databases and our own unpublished survey data. We used generalised dissimilarity modelling (GDM), which models species turnover between pairs of 0.01° grid cells as a function of environmental differences between those cells. Eight climatic and landscape variables were derived for each grid cell. Average temperature and average rainfall had the greatest contribution to species turnover followed by elevation. A model incorporating eight climatic, physiognomic and spatial variables accounted for 48% of the turnover of species. Six ecoregions were defined and used to circumscribe the equivalent number of interim Macrogroups based on the GLM outputs and diagnostic species.