Summary of the results of the different species identification methods (per genetic marker) and the multi-locus species validation (BP&P).

The number of specimens is given per gene in the order LSU, SSU and H3. The uncorrected percentage of pairwise distances was calculated on the same reduced dataset as was used for the calculation of the parsimony networks; the lowest average percentage of pairwise distance between any two clades within the datasets versus the averaged intraspecific pairwise distance is shown. The parsimony networks were calculated with a threshold of 98% for LSU and SSU and 90% for H3; numbers specify the steps between the two clades in case they were recovered in one network, an asterisk indicates that a given clade was recovered in more than one network. Gene trees were calculated using RAxML and MrBayes, support is given in this order; non-monophyletic clades are indicated with a hash in superscript (#). The program BP&P relies on a true guide tree, which due to the low taxon sampling we could not provide. We therefore decided to test only species with at least three specimens, thus some of the clades in the dataset were not tested and will not be named in this paper (except for S. martindalei and S. papuensis, which were well supported and described without BP&P validation). The posterior probabilities of the three different BP&P analyses are shown in the table. Due to the high variability of S. variabilis clade in all analyses, the clusters within this clade were tested for all analyses with the H3 gene. However, as this part of the tree was not resolved in either phylogenetic analysis, so no BP&B validation could be performed. A dash indicates missing data, or not tested clades.Summary of the results of the different species identification methods (per genetic marker) and the multi-locus species validation (BP&P).