*2.4. Diversity Measures and Analyses*

Summary data of the numbers of species (SR), genera (GR), and families (FR) were generated for the Sunshine Coast heath community of 366 species. Each of the species from the Sunshine Coast heath taxa was represented by a barcode identifier displayed on the dated heath phylogeny and these were used to create community lists of species found in each regional ecosystem, based on the field work data. Species were assigned to each of the nine regional ecosystems on a presence or absence basis, to be able to make broad comparisons of the heath regional ecosystems in terms of diversity measures. Additionally, community lists were developed for each of the 80 sites on a presence or absence basis to enable diversity measures to be generated for each site, to be used to statistically investigate variation between sites grouped by Regional Ecosystems.

The dated phylogeny, the complete Sunshine Coast heath community file, the community lists for individual sites and the individual regional ecosystems were used to derive phylogenetic metrics, and all analyses used R software [76]. Phylogenetic diversity (PD), mean phylogenetic distance (MPD), and mean nearest taxon (MNTD) were calculated for each regional ecosystem, as well as for each site using PICANTE [24,75]. A randomised null model, using the whole Sunshine Coast heath taxa and shuffling the taxa labels across the tips of the phylogeny, was used to calculate the probability of the phylogenetic diversity measures deviating significantly from random distributions. PICANTE calculates a standardised size effect (ses) and this figure multiplied by −1 gives the net relatedness ness index (NRI) for MPD and a nearest taxon index (NTI) for MNTD. A NRI has a value of 0 for a completely random community, increases as the community becomes more clustered, and decreases as a community becomes more even, with the NTI following a similar pattern [23]. The NRI and NTI were tested for significance using a randomised null model in PICANTE [75]. All these diversity measures were obtained for each of the nine regional ecosystems and for each of the 80 sites.

Individual site diversity measures were used to test for differences in PD, SR, FR, GR, MNTD, MPD, NRI, and NTI as well as between structural data (maximum and minimum heights and percentage cover of vegetation layers) between sites grouped by regional ecosystems, using the Kruskal–Wallis test followed by a Dunn's post hoc test with a Bonferroni correction in the "stats" package and PMCNR packages [77]. Relationships between SR, GR, FR, and PD were tested using Spearman's rank correlation tests in the "stats" package [76]. The significance of the observed frequency of phylogenetically even and phylogenetically clustered sites in each regional ecosystem was tested using Pearson's chi-squared test in the gmodels package [78].

To investigate patterns and similarities among site communities, the presence/absence matrices of species composition were used to calculate pairwise dissimilarity matrices using Vegdist and the Bray-Curtis method in the Vegan package [79]. A dissimilarity matrix between sites was calculated based on PD using Unifrac, a measure of phylogenetic distance between sites, within PICANTE [80,81].

These distance matrices were used in non-metric multidimensional scaling (NMDS) to visualise relationships among sites using Vegan [79]. Northings and eastings data for each site were used to calculate geographic distance matrices using Vegdist and the euclidian method in Vegan. All the dissimilarity matrices were tested for correlation using the Mantel test and the Spearman method in Vegan [82].

To visualise and assist the interpretation of the regional ecosystem data, labelled phylogenetic trees were produced using the iTOL program [83].
