Bioclimatic class (BCC) GeoTiff grids for the 1990-2009 baseline, 2020-2039 and 2060-2079 NARCliM climate projections.

BIAP employed BCCs as biological surrogates. BCCs are coarse communities (or assemblages). They were derived from generalised dissimilarity models by using an unsupervised hierarchical clustering of biological data to define 250 BCCs. A nearest neighbour algorithm was used to assign each cell in the analysis region to a BCC. BCCs were represented spatially as both stacks of probability surfaces and as single surfaces where each cell was assigned the class of highest probability.

Initially the baseline GDM was used to develop a classification of 250 BCCs. To generate inputs for this classification, spatial environmental predictors (Table 1) were transformed by a set of coefficients that best fit environmental data to biological data. Transformed predictors were used to derive an ecological distance between pairs of sites by calculating the sum of absolute distances across all GDM dimensions. This ‘GDM distance’ functions as a compositional dissimilarity metric for the classification function.

To assess the spatial shift in BCCs over the 12 climate futures, the transformation coefficients derived from the 2000 baseline GDM were applied to the corresponding environmental predictors for all 12 climate futures at both 2030 and 2070 centred epochs. Transformed environmental predictors were then used to generate BCC classification grids.

For presentation in maps, using colour variation to represent the number of species shared between classes, the training data were run through a classical metric multi-dimensional scaling process to scale the dissimilarities among BCCs into a three dimensional space. After rescaling between zero and 255 these axes were used as RGB values for colouring the final classification grid.

This classification showed a good visual representation of compositional dissimilarity using BCC distributions. The dissimilarity between the individual classes is shown by their relative RGB colouring.

[Report section 2.3.1 - 2.3.2] [Report Appendix A]