Broken stick eigenvalues indicate that the first principal component (PC1) captured more variance
than expected by chance and selection of the 25 habitat heterogeneity variables was based on the variable
weightings for the first two component axes.
For MDS the habitat variables selected from the PCA accounted for 65.5% of variance in the data as
compared to 38.7% when all the species were examined. This suggests that MDS are more influenced
by habitat heterogeneity than other categories of bird species (Table 2). The number of mangroves zones
at a site was an important habitat heterogeneity variable and was included among the most influential
variables for all categories of bird species, except MDS (the latter are widely distributed in the
mangroves) (Table 3). The number of mangrove zones was correlated with mangrove plant species
richness (excluding the rainforest species) (F1, 12 = 8.8, p = 0.01, R2 = 0.45).