See Fan et al. (2010) for materials for the morphological study. In order to examine the wing venation, wings were removed from thorax, cleaned with 95% ethanol, and dyed red with acetocarmine (Wang et al. 2011).

Twenty-three specimens including nine of the 15 valid species of Scobura and four outgroup species were included in the phylogenetic reconstruction. Detailed information on the specimens is provided in Table 1. Specimens used in this study were mainly deposited in the Insect Collection, Department of Entomology, South China Agriculture University (SCAU), except for some specimens in Kyushu University museum (KU) and Mr. Hiroaki Onodera’s private collection.

Genomic DNA was extracted from the thorax of specimens preserved in ethanol, or from legs of dried specimens, using Magen’s Blood/cell/tissue DNA extraction kit. One mitochondrial gene cytochrome c oxidase I (COI) and one nuclear gene elongation factor 1-α (EF-1α) were used as molecular phylogenetic markers. The following primers were used for amplification and sequencing in this study: for COI – primers LCO1490 and HCO2198 (Folmer et al. 1994); for EF-1α – primers ef44 and efrcM4 (Monteiro and Pierce 2001). Ploymerase Chain Reaction (PCR) were performed in 20 µl volumes containing 1 µl template DNA, 2 µl 10× buffer, 1.6 µl dNTPs (containing 2.5 mM of each dNTP), 0.8 µl of each primer (10 uM), 0.2 µl Taq Polymerase (2 U/µl), and 13.6 µl ddH₂O. The PCR Products were amplified using initial denaturation at 94 °C for 4 min, 35 cycles of denaturation at 94 °C for 30 s, annealing at 47 °C (COI) for 45 s, 55 °C (EF-1α) for 1 min, elongation at 72 °C for 1.5 min, and final elongation at 72 °C for 5 min.

Amplified DNA products were purified using an Agarose Gel Extraction kit (Magen Biotech), and directly sequenced, or cloned with pMD18-T vector (Takara Inc), and then sequenced. Sequencing was performed using the ABI 3730 automated sequencer. All sequences were submitted to the Genbank database (accession numbers are given in Table 1).

Alignment of the DNA sequences were performed in Clustal X (Thompson et al. 1997) and edited manually in MEGA 6.0 (Tamura et al. 2013). All base frequencies and molecular character statistics were calculated in MEGA 6.0. Phylogenetic trees were constructed under Maximum Likelihood (ML) and Bayesian inference (BI) criteria. For ML analysis, RAxML version 8 (Stamatakis et al. 2014) was used on a concatenated data set of two genes, with 1000 rapid bootstrap replicates using GTR+G substitution model on the CIPRES Science Gateway (Miller et al. 2010). BI was carried out using Markov Chain Monte Carlo (MCMC) randomization in MrBayes v3.2.3 (Ronquist et al. 2012). We used reversible-jump MCMC to allow for sampling across the entire substitution rate models. Four Markov chains (three heated chains, one cold) were run for 500, 000 generations, with the first 25% of sampled trees discarded as burn-in. The two independent runs were considered to have converged when the standard deviation of split frequencies value was <0.01. The convergence of the analysis was determined in Tracer v1.6 (Rambaut et al. 2014). Bayesian posterior probabilities (PP) and MLbootstrap values (BP) were used to evaluate branch support.

From a total of 23 samples, 22 sequences for COI and 21 for EF-1α were obtained. The alignment of the combined sequences consisted of a total of 1724 bp (658 bp of COI and 1066 bp of EF-1α genes, respectively), including 277 variable and 200 informative sites.

The pairwise P2K distances among the sequences were variable between genes. The ranges of sequence divergences for two loci and ingroup taxa are: COI (0–12.4%), EF-1α (0–5.0%). For COI, sequence divergence between conspecific individuals ranged from 0 to 0.6%; inter-specific genetic distances ranged from 3.6% to 12.4% with divergences among species averaging 7.9% (Table 2).

The two model-based analyses (BI and ML) revealed nearly identical topologies, differing mainly in branch support (Fig. 1). In both analyses, the monophyly of the genus Scobura is weakly supported (BP = 44, PP = 0.87). Within the genus, although support for the basal clades was low, the Scobura species included here are clearly distinguished from each other, and formed four clades: the S. isota clade (which only included two representative specimens), Clade A, the S. cephaloides clade (only with two representative specimens), and Clade B. Clade A is comprised by S. stellata + (S. parawoolletti + S. woolletti) and receive high bootstrap support and posterior probability (BP = 99, PP = 1.00). We hereafter called the clade S. woolletti group.

Figure 1. 

Majority-rule consensus tree from the Bayesian analysis (BI) of the concatenated COI and EF-1α sequences. Values at nodes represent the bootstrap support (BS) values of the maximum likelihood (ML) and the posterior probabilities (PP) of BI analyses, respectively (BP/PP).

Clade B is comprised by S. masutaroi and the representatives of S. coniata group (Devyatkin 2004): S. coniata, S. lyso and S. hainana, and the latter two are sister species with strong support (BP = 93, PP = 1.00). The monophyly of S. coniata group including S. masutaroi is strongly supported (BP = 100, PP = 1.00).

In all the analyses, S. cephaloides is sister to Clade B, with moderate support (BP = 63, PP = 0.92), whereas the relationships between S. isota and the other clades (Clade A, S. cephaloides and Clade B) remain unresolved.