The final concatenated sequence dataset contained a total of 96 sequences of 16S, including 14 Genbank sequences, and 93 COI sequences, including 12 Genbank sequences. Some samples have been represented by either only the COI or 16S sequence while the lacking sequence fragment was coded as missing data. However, we had usually at least one complete set of sequences for each species identified by comparative morphology. The final 16S alignment had a total length of 1,123 aligned nucleotide positions. This alignment was used to calculate uncorrected p-distances between all sequences. Of these 1,123 alignment positions, 859 were retained after removing ambiguous alignment positions with Gblocks for subsequent phylogenetic analysis.
The time-reversible model with invariant sites and a gamma distribution of rates (GTR+I+G; Tavaré, 1986) has been found to be the best-fit model of sequence evolution for both the 16S and COI sequence datasets by means of both the Bayesian and Akaike Information Criterion. Xia’s et al. (2003) test indicated no or little saturation in both mitochondrial fragments (Iss < Iss.c with P < 0.001).
The best maximum likelihood tree (Fig. 2) was rooted by using the designated out-group sequences, which were selected based on the phylogenetic tree presented by Hyman et al. (2007). Among the in-group, the southeastern Australian helicarionid radiation, some but not all genera as currently circumscribed have been retrieved as monophyletic groups. While the monophyly of Brevisentis, and Cucullarion as presently delimited has been confirmed with high nodal support, the genera Mysticarion, Peloparion, Parmavitrina and Desidarion have each been found to be polyphyletic with generally high bootstrapping support.
The analysis of the concatenated dataset revealed the sister pair of Helicarion spp. and ‘Peloparion’ iridis to form the most basal offshoot among the southeastern Australian radiation. Cucullarion was found to be the sister of all remaining in-group clades containing Parmavitrina+Desidarion, Brevisentis, Peloparion helenae, and Mysticarion, respectively (Fig. 2). However, there was usually only weak nodal support for the principal branching patterns in the phylogenetic tree. In order to explore this instability further, we have analysed the two sequence datasets separately. These analyses revealed that the two mitochondrial fragments supported vastly different topologies with respect to the relationships between the principal clades (Figs 3-4). In particular the COI tree is characterized by extremely low nodal support for the relationships among the principal clades. Moreover, the in-group was non-monophyletic as Helicarion clustered among the out-group taxa (Fig. 3). By contrast, in the 16S tree the monophyly of the in-group is well-supported while the relationships among its principal clades attracted weak nodal support only (Fig. 4).
By contrast to relationships among principal branches, the relationships among the tips of the tree are consistently well-supported and there has been little, if any, conflict between the topologies produced for different data sets. Some noteworthy findings are that among Brevisentis, specimens from Wollemi NP identified as B. atratus do not cluster with any of the three currently recognized species, and are thought to represent a yet undescribed species. However, because of the lack of reproductively mature specimens, we currently refrain from naming this species.
We also found that Mysticarion is monophyletic only if Fastosarion staffordorum is included. Sequences of this taxon group closely with M. hyalinus with very little genetic differentiation (Tables 1-2). The rather small amount of genetic differentiation confirms that M. porrectus, once suggested to be a species complex, is simply a widely distributed species. Specimens initially identified as M. insuetus occupied two highly distinct clades. This genetic differentiation is underpinned by subtle morphological differences that are in support of the taxonomic distinctiveness of these two previously unrecognized groups. Based on the specimens from the Hunter Valley, we describe a new species, M. obscurior sp. nov., below. Neither Desidarion nor Parmavitrina are monophyletic: Parmavitrina planilabris groups very closely with both D. rubricus and D. dispositus. The two Desidarion species also share a number of morphological characters with Parmavitrina; based on these similarities we treat both genus names as synonyms below. One population of D. rubricus from around Taree shows some genetic divergence. However, this is accompanied by insignificant anatomical differences. The two new species P. maculosa sp. nov. and P. flavocarinata sp. nov., recognized by their morphology, group closely with P. megastoma. The two species currently placed in Peloparion, P. helenae and P. iridis, do not group together; based on this and on considerable anatomical differences, we describe a new genus for P. iridis below.
The evolutionary divergence within and between species was estimated by calculating uncorrected pairwise distances across all sequences. We found that range of intraspecific distances in COI (0-0.04) and 16S (0-0.72) overlapped with the range of interspecific distances in both genes to some degree (COI: 0.029-0.157; 16S: 0.02-0.234) (Fig. 5, Tables 1-2). However, there has been no overlap between the average intraspecific distances per species (< 0.028, on average 0.011 in COI; < 0.022, on average 0.006 in 16S) and the average interspecific distance between any two congeners (0.036-0.095, on average 0.072 in COI; 0.029-0.114, on average 0.073 in 16S).
Fig. 5. Comparison of intra- and interspecific genetic p-distances for the two mitochondrial fragments analysed. A. Frequency distributions of distances in COI. B. Frequency distributions of distances in 16S.
Table 1. Average pairwise distances in 16S within and between species under pairwise removal of gaps.