Material and methods
We examined a specimen of D. oligopodus from Puerto Iguazù, Argentina [female, 15.xi.1980, L.A. Pereira lg; indicated in the original description as the allotype of the species; Pereira, 1984] and seven new specimens from La Plata, Argentina [two males and five females, 19.xii.1985, 14.iv-3.v.1986, 22-23.viii.2009, L.A. Pereira lg]. Specimens are preserved in the collections of the Museum of La Plata, Argentina, and in the Minelli-Bonato collection at the Department of Biology, University of Padova, Italy.
The specimens were examined with light microscopy (LM), with a Leica DMLB microscope equipped with a Leica DFC420 camera. A male and a female were also examined with scanning electron microscopy (SEM), using a Cambridge Stereoscan 260. For both LM and SEM, the head was detached from the trunk. For LM, the specimens were mounted in temporary slides, following standard protocols for geophilomorphs (Pereira, 2000). For SEM, the samples were gradually hydrated, post-fixed in 4% formaldehyde in water, rinsed with 0.5% Triton-X 100 in water, briefly sonicated, rinsed in water, cleaned with 3% H2 O2 , dehydrated in graded ethanol series, dried with hexamethyldisilazane (Sigma), and coated with gold.
The entire body of two specimens (collected in 2009 and fixed in absolute ethanol) were used for DNA extraction, with the aim to amplify and sequence the genes most commonly used in phylogenetic analyses in Chilopoda (CO1 and 16S rRNA from the mitochondrial genome; 18S and 28S rRNA from the nuclear genome; e.g., Murienne et al., 2010). We followed a protocol previously optimized in our laboratory for a broad sample of geophilomorphs (described in detail in Bonato et al., 2014). Because of difficulties due to limited mass and poor quality of preservation of the samples, we were successful in obtaining well readable sequences only for the three subunits of rRNA and only for one of the specimens.
In order to test the two competing hypotheses on the phylogenetic position of Dinogeophilus (within Geophilidae vs. within Schendylidae; see Introduction), the sequences of Dinogeophilus (16S, 18S, 28S rRNA) were compared with all homologous sequences available in GenBank for species of Geophilidae and Schendylidae. Following the cladistic revision proposed by Bonato et al. (2014), the two families are here intended in a broader sense than the traditional one, including subgroups that have been traditionally distinguished as distinct families: Geophilidae includes ‘Aphilodontidae’, ‘Dignathodontidae’ and ‘Linotaeniidae’; Schendylidae includes ‘Ballophilidae’. We considered all species of Geophilidae and Schendylidae for which sequence fragments were available and alignable for at least two of the three genes. Many of these sequences had been obtained directly in our laboratory using the same protocol (Bonato et al., 2014). Homologous sequences were aligned for the single genes by means of ClustalW implemented in MEGA 6.06 (Tamura et al., 2013).
We performed a similarity analysis of the molecular sequences by estimating alternative measures of pairwise distance (proportion of positions with different nucleotides, p-distance; distance according to the Kimura 2-parametres model, K2P) and clustering by the neighbour joining algorithm (NJ). Standard errors of the estimates were calculated by means of 1000 bootstrap replicates. This approach mirrors the common DNA-barcoding methodology for species identification, extended to above-species taxa (Wilson et al., 2011). We also performed the analysis by applying the minimum evolution (ME) criterion.
We performed a phylogenetic analysis of the molecular sequences, employing alternative criteria of optimization, including maximum likelihood (ML) and maximum parsimony (MP). The trees were rooted assuming the monophyly of the Schendylidae, which is supported by all previous molecular analyses (Edgecombe and Giribet, 2004; Murienne et al., 2010; Bonato et al., 2014). For the ML, the best-fit models of nucleotide substitution were selected according to the corrected Akaike information criterion (AICc) and the Bayesian information criterion (BIC). The statistical support of the nodes was tested by means of 1000 bootstrap replications. The MP tree was searched by 1000 replicates of random additions, using the Tree-Bisection-Reconnection algorithm.
For the terminology of the anatomical parts we follow Bonato et al. (2010). The analyses of the molecular sequences were performed with MEGA 6.06 (Tamura et al., 2013). The three genes were analysed both separately and concatenated. Differences of genetic distances between groups were tested for statistical significance with the Mann-Whitney U test (MW).