Relating spatial and temporal patterns in floristics with vegetation mapping: an example from Fraser Island, south-east Queensland, Australia

Abstract

The use of mapped polygons to represent vegetation patterns is becoming more widespread in conservation planning and natural resource management. The degree of homogeneity (i.e., one vegetation unit per polygon) of these polygons influences the efficacy of mapped patterns. It was observed that Fraser Island, south-east Queensland (SEQ) exhibited near complete map unit homogeneity relative to adjacent mainland areas. We sought to establish whether this homogeneity had an ecological or cartographic basis. Floristic presence-absence data from 50 quadrats, each sampled over three consecutive years, were analysed using cluster analysis (UPGMA) and ordination (SSH MDS) to define and compare eight numerically-derived plant community types with eight a priori regional ecosystem types derived from mapping of the island. Strong congruence between the two approaches was evident, suggesting that the mapping effectively portrayed ecological patterns. Floristic composition and species richness remained consistent at most quadrats for three consecutive years. Perennial woody species (including tree and shrub species) showed significant (P < 0.05) variation between regional ecosystem units and between plant community groups, but was stable through time. Conversely, the herb assemblage showed significant temporal variation and poor fidelity to vegetation classifications. High levels of map unit homogeneity were shown to extend to five barrier sand islands of SEQ across various map scales and map authors in these sandy environs. We conclude that homogeneity in mapped polygons on Fraser Island is a consequence of discrete boundaries and the associated distinctive floristics of the vegetation communities in this area. We suggest focussing limited mapping resources to areas of high polygon heterogeneity, instead of areas such as the SEQ sand islands which appear well represented by existing mapping.