Phylogenetic relationships of subfamilies in the family Hesperiidae (Lepidoptera: Hesperioidea) from China

Hesperiidae is one of the largest families of butterflies. Our knowledge of the higher systematics on hesperiids from China is still very limited. We infer the phylogenetic relationships of the subfamilies of Chinese skippers based on three mitochondrial genes (cytochrome b (Cytb), the NADH dehydrogenase subunit 1 (ND1) and cytochrome oxidase I (COI)). In this study, 30 species in 23 genera were included in the Bayesian and maximum likelihood analyses. The subfamily Coeliadinae, Eudaminae, Pyrginae and Heteropterinae were recovered as a monophyletic clade with strong support. The subfamily Hesperiinae formed a clade, but support for monophyly was weak. Our results imply that the five subfamilies of Chinese Hesperiidae should be divided into: Coeliadinae, Eudaminae, Pyrginae, Heteropterinae and Hesperiinae. The relationships of the five subfamilies should be as follows: Coeliadinae + (Eudaminae + (Pyrginae + (Heteropterinae + Hesperiinae))).

Hesperiidae contains approximately 370 described species in 83 genera in China 15 and so far there has been no molecular study of the higher-level phylogeny of the Chinese skipper butterflies. Mitochondrial DNA (mt DNA) sequence has been widely used in phylogenetic studies of Lepidoptera [16][17][18] . In this study, we used DNA sequences from the mitochondrial cytochrome b (Cytb), the NADH dehydrogenase subunit 1 (ND1) and cytochrome oxidase I (COI) to analyze the phylogenetic relationships of the genera in Hesperiidae.

Materials and methods
Taxon sampling. The butterflies studied were either collected with aerial nets in the field or were specimens in the Entomological Museum of Northwest A&F University in China. The specimens sampled and their collection site are listed in Appendix 1. A total of 30 skipper butterfly species in 23 genera were used in this study. In addition, five outgroup species, Papilio protenor, Troides helena, Sericinus montelus (Papilionidae), Eurema andersoni, Pontia edusa (Pieridae), from the Papilionoidea, the putative sister clade to the Hesperioidea 19 , were collected and used in this study.
DNA extraction, PCR amplification and DNA sequencing. Total genomic DNA was extracted from a pair of legs of an adult specimen either dried or preserved in 95% ethanol, using the phenol-chloroform extraction protocol 20,21 . The genomic DNA prepared was dissolved in a 50 μ L TE buffer and stored in a freezer (− 20 °C).
PCR reactions were prepared in 50 μ L that included 5 μ L 10 × reaction buffer, 2.5 mM Mg 2+ , 0.6 mM primers, 4 μ L of DNA template, 0.25 mM dNTPs and 1.0 U Taq polymerase. For amplification of the fragment from Cytb, the PCR amplification was performed by an initial denaturation at 94 °C for 5 min, followed by 35 cycles of 1 min at 94 °C, 1 min at 50 °C and 2 min at 72 °C, and a final extension at 72 °C for 10 min. For amplification of the fragment from ND1, the PCR cycles included 0.5 min at 94 °C, 0.7 min at 49 °C and 2 min at 72 °C. For amplification of the fragment from COI, the PCR cycles included 0.5 min at 94 °C, 0.7 min at 50 °C and 0.5 min at 72 °C. All primers used in this study were listed in Table 1. The PCR products of the PCR reactions from individual specimens were examined by agarose gel electrophoresis to verify the specific amplification of the desired fragments and the PCR products were sequenced for strands by commercial service (GeneScript Biological Technology, Nanjing, China, and Aoke Biological Technology, Beijing, China). Phylogenetic analysis. The DNA sequences from the individuals were aligned using MAFFT v7.037 25 , and the parsimony informative sites, base frequencies and Kimura-2-parameter distances (K2P distance) were calculated using MEGA v5.05 26 . The alignment was evaluated by substitution saturation using DAMBE v5.3.74 27,28 . The combined sequence datasets of Cytb, ND1 and COI were used to construct phylogenetic trees.
Phylogenetic analysis by maximum likelihood (ML) model was conducted using jModelTest v 2.1.4 29 using the Akaike information criterion (AICc). The best-fitting model of nucleotide substitution was GTR + I + G for all genes, and the general ratchet analysis conditions were as following: Lset base = (0.3552 0.0751 0.0866), nst = 6, rmat = (2.7601 10.2085 4.5566 7.6424 44.6148), rates = gamma, shape = 0.4670, ncat = 4, pinvar = 0.3640. PAUP* v4.0b10 30 was used to calculate the ML analyses with 1000 bootstraps. Bayesian inference (BI) analysis was run in MrBayes 3.1.2 31 using the model generated in jModelTest. The partitioned analysis comprised two runs with four Markov chain Monte Carlo simulations (MCMC) each, with flat priors, dataset partitioned by one million generations, sampling every 100 generations with 25% of samples discarded as burn-in.

Sequence characterization. Alignment of the combined PCR fragment sequences from Cytb, ND1
and COI showed that in the 1458 bp combined DNA sequences there were 717 variable sites and 568 parsimony-informative characters. The base composition of the fragments showed a strong bias of A + T ( Table 2) as is commonly found in insect mitochondrial genomes 22  Phylogenetic relationships. The phylogenetic trees generated from the DNA sequence dataset by BI and ML methods trees are nearly identical in major clades and patterns of branching recovered. In BI analysis (Fig. 1), five of seven currently subfamilies of Hesperiidae were recovered as monophyletic clades with the following relationships: Coeliadinae + (Eudaminae + (Pyrginae + (Heteropterinae + Hes periinae))). The subfamily Coeliadinae (Clade I) was recovered as a monophyletic clade with strong support, although there were only three taxa (Burara miracula, Choaspes benjaminii and Choaspes hemixantha) included, and was placed in the basal position as the sister to the rest of the clades of the Hesperiidae. Although only two taxa within geneus Lobocla were included in our analysis from Eudaminae, its monophyly (Clade II) received strong support. The seven genera from Pyrginae formed a clade (Clade III) also with strong support. Furthermore, this clade split into three subclades: Abraximorpha + ((Daimio + (Capila + Coladenia)) + (Satarupa + (Celaenorrhinus + Sarangesa))).
The phylogenetic tree by ML analysis showed that four major clades of Hesperiidae were recovered, although the relationships between some nodes were not strongly supported (<50) (Fig. 2). Compared with the topology of the tree by BI method, the major difference in the tree by ML is that the genus Lobocla was placed into the Pyrginae group (Clade II in Fig. 2), but the support was weak.

Discussion
In the family Hesperiidae, Coeliadinae with morphological synapomorphies is relatively unique and easy to be distinguished from the remaining subfamilies 5,7 . The five genera (Bibasis, Burara, Hasora, Badamia, Choaspes) in this subfamily are distributed in China. In this phylogenetic study with two of these genera, the monophyly of Coeliadinae and its status as the sister of the rest of the Hesperiidae were confirmed, which is consistent with previous studies based on morphological and molecular data 13,14,32,33 . The morphological synapomorphy for Coeliadinae is the 3rd segment of labial palpi which is long, slender, cylindrical or awl-like 14 . Larvae generally feed on the plants of the class Dicotyledonopsida in China 15 .
The genus Lobocla was placed in the Celaenorrhinus group by Evans 7 ; while, Warren et al. 13 assigned it to the tribe Eudamini, which was then promoted to the subfamily of Eudaminae 14 . The result from the ML analysis of the mitochondrial DNA sequences in this study placed Lobocla in the subfamily of Pyrginae with weak support (Fig. 2). However, Lobocla became separated from the Pyrginae and formed an independent clade by BI analysis (Fig. 1). Given the higher value of posterior probabilities, we support the status of Eudaminae and that Lobocla as the only genus of Eudaminae occurring in the Oriental, Neotropical and Nearctic regions.   The subfamily Pyrginae has long been treated as a paraphyletic group 5,7,13,32 . Warren et al. 14 recovered the monophyly of Pyrginae with moderate support. Pyrginae has been divided into seven tribes (Pyrrhopygini, Achlyodini, Tagiadini, Celaenorrhinini, Carcharodini, Erynnini and Pyrgini), but no morphological synapomorphies have been known for the subfamily. In this study, seven genera were included in the analysis and they appeared to form a monophyletic group with moderate support. However, the status of many other tribes and genera and their relationships within Pyrginae (e.g., Caprona, Mooreana, Muschampia in China) remain unknown. Additional taxa with additional molecular makers will be needed to elucidate their phylogenetic positions at the level of tribe and genera. Morphological characters for this subfamily are the 3rd segment of labial palpi which is short and stout, and that the larvae generally feed on plants of the class Dicotyledonopsida in China 15 .

Number of sequences
The monophyly of the subfamily Heteropterinae was recovered with strong support (PP = 1.00) in the BI analysis (Fig. 1), which is consistent with the Warren et al. 14 study. Heteropterinae is grouped under two tribes (Heteropterini and Carterocephalini), but morphological synapomorphies could be difficult to identify. Morphological characters for Heteropterinae are that the abdomens are distinctly elongated, longer than the length of the hindwing dorsum. Female bursa copulatrix has an appendix bursa. Larvae feed on plants of the class Monocotyledonopsida in China 15 . Within this subfamily, our results indicate that the genus Carterocephalus is a monophyletic clade in both analyses (PP = 0.96, BS = 69). However, the taxonomic status of this group remains to be resolved.
The subfamily Hesperiinae, as the largest subfamily of Hesperiidae, has long been a controversial subfamily in the Hesperiidae. The monophyly Hesperiinae has been reasserted 13,14 and is also supported by the analysis in this study. Evans 7 split the subfamily Hesperiinae into eight groups, and Inoué and Kawazoé 34 reviewed Evans's system, i.e., defining the Halpe group to include Evans's Astictopterus group except for the genus Astictopterus. Chou 11,12 divided the Chinese Hesperiinae into ten tribes based on Evans's classification system. Warren et al. 14 reviewed and recognized eight tribes of Hesperiinae (Aeromachini, Baorini, Taractrocerini, Thymelicini, Calpodini, Anthoptini, Moncini, Hesperiini). The results form this study support the monophyly of Baorini and Hesperiini. However, the classification status of other tribes has yet to be established with more taxa to be added to the phylogenetic analysis. The morphological character for Hesperiinae is the terminal part of lower margin of discal cell in hindwing which is oblique upwards. Larvae of this subfamily generally feed on plants of the class Monocotyledonopsida in China 15 .
In this comprehensive phylogenetic analysis of Hesperiidae members from China at subfamily-level with 30 species in 23 genera, the monophyly of this family was demonstrated with strong support. This result is in agreement with the previous reports by Wahlberg et al. 19 and Warren et al. 13,14 , although higher level phylogenetic relationships remain challenging to decipher in Lepidoptera 3 . With strong posterior probability values, the results from BI analysis (Fig. 1) imply that the five subfamilies of Chinese Hesperiidae are under Coeliadinae, Eudaminae, Pyrginae, Heteropterinae and Hesperiinae. The relationships of the five subfamilies are Coeliadinae + (Eudaminae + (Pyrginae + (Heteropterinae + Hespe riinae))).