Nubeam Hellinger distances correlate well with composition-based dissimilarities among synthetic communities. We generated four sets of synthetic communities; each set contains 15 samples composed of 10 taxonomic units. The relative abundances of taxonomic units in each sample follow Dir(α1×10), with α being 1 and 10 in A and B, respectively, controlling for the complexity of the community. The relationship among the taxonomic units is shown in the star tree, with branch lengths of either 1 × 10−4 or 2 × 10−4. Each sample has 50 million 75-bp reads. The 105 pair-wise Nubeam distances were calculated among the 15 samples. The significance of linear relationship is measured by R2 and P-value for regression coefficient.
