Molecular Identification of Sibling Species of Sclerodermus (Hymenoptera: Bethylidae) That Parasitize Buprestid and Cerambycid Beetles by Using Partial Sequences of Mitochondrial DNA Cytochrome Oxidase Subunit 1 and 28S Ribosomal RNA Gene

The species belonging to Sclerodermus (Hymenoptera: Bethylidae) are currently the most important insect natural enemies of wood borer pests, mainly buprestid and cerambycid beetles, in China. However, some sibling species of this genus are very difficult to distinguish because of their similar morphological features. To address this issue, we conducted phylogenetic and genetic analyses of cytochrome oxidase subunit I (COI) and 28S RNA gene sequences from eight species of Sclerodermus reared from different wood borer pests. The eight sibling species were as follows: S. guani Xiao et Wu, S. sichuanensis Xiao, S. pupariae Yang et Yao, and Sclerodermus spp. (Nos. 1–5). A 594-bp fragment of COI and 750-bp fragment of 28S were subsequently sequenced. For COI, the G-C content was found to be low in all the species, averaging to about 30.0%. Sequence divergences (Kimura-2-parameter distances) between congeneric species averaged to 4.5%, and intraspecific divergences averaged to about 0.09%. Further, the maximum sequence divergences between congeneric species and Sclerodermus sp. (No. 5) averaged to about 16.5%. All 136 samples analyzed were included in six reciprocally monophyletic clades in the COI neighbor-joining (NJ) tree. The NJ tree inferred from the 28S rRNA sequence yielded almost identical results, but the samples from S. guani, S. sichuanensis, S. pupariae, and Sclerodermus spp. (Nos. 1–4) clustered together and only Sclerodermus sp. (No. 5) clustered separately. Our findings indicate that the standard barcode region of COI can be efficiently used to distinguish morphologically similar Sclerodermus species. Further, we speculate that Sclerodermus sp. (No. 5) might be a new species of Sclerodermus.

These species were artificially reared in large numbers in the laboratory. All these Sclerodermus putative species were collected from different geographical or climatic environments. Furthermore, they were collected from different host plant species infected by different host pests. Therefore, we speculate that there might be some new species among these Sclerodermus putative species collected from the field. However, distinguishing these Sclerodermus putative species is difficult for many applied entomologists because most scale insects are small, and closely related species might be very similar morphologically.
At present, DNA barcoding is a new approach for species-level identification. Although considerable biological research depends on species diagnoses, taxonomic expertise is lacking. Sustainable identification ability involves the construction of systems that employ DNA sequences as taxon barcodes. The mitochondrial gene cytochrome c oxidase I (COI) could serve as the core of a global bioidentification system for animals. In some cases, this has shown to have a 100% success rate in correctly identifying specimens [14]. Thus, the COI identification system is a reliable, cost-effective, and accessible solution to the current problem of species identification.
The family Bethylidae belongs to the aculeate (stinging) Hymenoptera that includes familiar bees, social wasps, and ants [15]. It is placed in the superfamily Chrysidoidea, members of which are almost exclusively parasitoids [16]. Within the superfamily, the Bethylidae are possibly the most species-rich family, with over 2,000 species that have been described [17]. They are globally distributed and develop as ectoparasitoid on larvae, and occasionally, pupae, of Coleoptera and Lepidoptera species, which they permanently paralyze with a sting (idiobiosis) [18]; a few species are found on other hosts, including Hymenoptera species [19]. Gordh and Móczár [17] published a catalog of the world species of Bethylidae, including Sclerodermus. Evans [20,21] studied the North American Bethylidae, including Sclerodermus. Lanes and Azevedo [22] listed 71 species of Sclerodermus worldwide when they studied the phylogeny and taxonomy of Sclerodermini. Although the phylogeny of Bethylidae was constructed in 2010 [23], there were only two samples of Sclerodermini; in this study, we further identified several Sclerodermus species and putative species Sclerodermus nos. 1-5 as well as other species of Bethylidae, sequences of which were recently obtained from GenBank. The phylogeny of Bethylidae and taxonomic classification of Sclerodermus were investigated in this study. However, most importantly, we aimed to use molecular methods to identify Sclerodermus putative species that are difficult to distinguish morphologically. Further, we provided reliable evidence for traditional morphological identification.

Ethics statement
Our study did not involve the use of any non-human primates. Further, we confirm that our field studies did not involve endangered or protected species (i.e., the specimens sampled were Sclerodermus wasps (Hymenoptera: Bethylidae)). The putative species of Sclerodermus spp. (Nos. [1][2][3][4][5] were collected from Tongchuan, Ningbo, Kuandian, Bayanhaote, and Kunming, respectively, China. All the fields from where the samples were collected were managed by the local Forest Pest and Disease Control Stations, and the secretaries of the respective stations provided permission to conduct field studies. Finally, no permits were required for the described study, which complied with all the relevant regulations.

Specimen sampling
In total, 136 individuals were collected from different progeny, and 15 individuals of each putative species were reared at the Chinese Academy of Forestry, Beijing, China. Further, S. guani, S. sichuanensis, and S. pupariae were collected from Beijing Xishan Forest Farm. Some S. sichuanensis samples were obtained from Chongqing Academy of Forestry. Table 1 provides an overview of the origin hosts, origin place from where the specimens were collected, and rearing conditions for each putative species, as well as the number of species that have been named and published or are unnamed. In this study, all putative species were reared artificially in the laboratory and were descendants of origin species. For each species, the genomic DNA was extracted from an adult female by using a DNA extraction kit (Tiangen Biotech, Beijing, China). The specimens for phylogenetic analysis of 28S ribosomal RNA gene and thirteen species of other three closely related families were obtained as an outgroup from GenBank; these are listed in Table 2.

DNA sequence analyses
In this study, 136 individuals from eight putative species of the genus Sclerodermus were used. The samples per putative species ranged from one to twenty-four, with an average (mode) of fifteen. The sequence of a 594-bp COI fragment was obtained from all Sclerodermus individuals after deleting the terminal ambiguous part of the aligned data. There was 100% sequence identity between S. guani and S. sichuanensis. For COI data, low G-C content was found in all species, averaging to about 30.0%. Sequence divergences (K2P) between congeneric species averaged 4.5%, whereas intraspecific divergences averaged 0.09%. Furthermore, no insertions or deletions were found in DNA sequences, and no stop codons were found when amino acids were translated, implying that all amplified sequences coded for functional COI. The 594-bp nucleotide fragment of COI corresponded to 198 codons. Interestingly, there was 3.0% interspecific divergence between the COI regions of S. guani and S. pupariae, whereas the translated amino acid sequences of the two species were not different. Further, the 1.4% divergence (K2P) between S. guani and Sclerodermus sp. (No. 3) did not lead to a difference in amino acid sequences. Generally, the maximum interspecific divergence between S. guani and Sclerodermus sp. (No. 5) was 16.5%; this translated into a difference of two amino acids.   Comparison of the sequences of five unknown putative species with those of three known species (S. guani, S. sichuanensis, and S. pupariae) revealed that the sequence similarity of Sclerodermus sp. (No. 4) and S. pupariae was 100%. However, the lowest sequence similarity of about 85% was found between Sclerodermus sp. (No. 5) and the three known species (Table 3).
In the COI region, the averaged frequency of adenine (A) and thymine (T) was high (A = 30.9%, C = 18.4%, G = 11.6%, T = 39.1%). The G-C frequency was lower at the third codon position (mean, 14.8% for all individuals) than that at the first and second positions (mean, 36.4% and 38.9%, respectively). The nucleotide frequency of the three codon positions in the eight species is shown in Table 4. The G-C frequency at the third codon position (mean, 18.2% for all individuals) was the highest for Sclerodermus sp. (No. 5), whereas that of S. pupariae and Sclerodermus spp. (Nos. 3 and 4) was 15.1%, which was higher than that of the others (14.1%). Further, the G-C frequency at the first codon position of Sclerodermus sp. (No. 3) was slightly higher than that of the others. The 594-bp region of COI mtDNA was obtained for the 136 individuals; it included 491 conserved sites, 103 variable sites, 102 parsimony-informative sites, one singleton site, 375 zero-fold degenerate sites, 115 two-fold degenerate sites, and 81 fourfold degenerate sites. The COI sequences showed interspecific divergence of 0-17.4%, with a mean divergence of 4.5%. Intraspecific COI sequence divergences were 0-0.1%, with a mean divergence of 0.075% (Fig. 1). The maximum K2P distance was found between Sclerodermus sp. (No. 5) and Sclerodermus sp. (No. 3; 17.4%). Unexpectedly, the minimum distance was

28S phylogeny of the Bethylidae species
Phylogenetic analysis was performed on a partial 28S ribosomal RNA D2 gene. The maximum likelihood (ML) tree indicated that Bethylidae are strongly monophyletic (90% ML bootstrap percentage (mlBP)). Of the subfamilies, Bethylinae are strongly monophyletic (100% mlBP) and sister to the other commonly recognized subfamilies (Fig. 3). Of the other subfamilies, Pristocerinae were found to be monophyletic (99% mlBP), with Foenobethylus clustering within the group [27,28], and Pristocerinae was sister to Epyrini [23]. Only Mesitiinae species were strongly included as clusters in a group with Sclerodermini. Sclerodermini and Cephalonomini [23] were strongly supported as monophyletic (98% mlBP). The relationships between the three clades Epyrini, Pristocerinae, and (Sclerodermini + Cephalonomini + Mesitiinae) were equivocal. Species in the same genus clustered together, although they were not always strongly supported to be monophyletic (e.g., Sclerodermus sp. MC-2010 and Sclerodermus in this study; Fig. 3).

Discussion
With the development of modern molecular biological techniques, specific DNA fragments can be used for species identification for determining the general phylogenetic relations and Table 4. Nucleotide frequency of the three codon positions in the eight species. Abbreviations are the same as those in Table 3 doi:10.1371/journal.pone.0119573.t004  taxonomic status. Gene fragments or site-specific information is the auxiliary tool to distinguish species and has been used in the identification of animal and plant species.
In the present study, COI barcode sequences were used for the identification of several Sclerodermus species for the first time, to our knowledge. Each of the eight Sclerodermus samples had a distinct COI sequence, and hence, they could be distinguished from conspecifics by using the DNA barcode method. The average intraspecies K2P distance was estimated to be only 0.09%, whereas it was about 4.5% among congeneric species. This distance is considerably lesser than that for the six Ceroplastes species (Hemiptera: Coccidae) from China [29]. This could be attributed to the fact that our intraspecies samples were reared in the laboratory and not collected from different locations. However, the sequence divergences between Sclerodermus sp. (No. 5) and any congeneric species was more than 16.0%, whereas the average intraspecies K2P distance for Sclerodermus sp. (No. 5) was only 0.1%. The COI gene marker could be used for phylogenetic analysis and for the identification of cryptic species. Different vertebrate species commonly show a value close to 3% K2P distance for COI, a threshold adopted for Lepidoptera [14]. Thus, more than 16% sequence divergences for Sclerodermus sp. (No. 5) might be strong evidence for a new Sclerodermus species. Similarly, the 594-bp nucleotide fragment of COI corresponding to 198 codons led to a difference of two different amino acids between Sclerodermus sp. (No. 5) and the others. Further, for the 28S gene, although all the other seven species were completely identical with each other, the sequence similarity between Sclerodermus sp. (No. 5) and others was 99.18%. The NJ tree of COI also indicated that the 15 individuals of Sclerodermus sp. (No. 5) were strongly monophyletic (100% (mlBP)). The 28S NJ tree also yielded the same result (99% mlBP). Both the molecular phylogenetic trees suggested that Sclerodermus sp. (No. 5) differentiated the earliest among the species belonging to Sclerodermini. This is also supported by the fact that there was 99.18% sequence similarity between Sclerodermus sp. (No. 5) and the remaining species. However, determining whether Sclerodermus sp. (No. 5) is a new species on the basis of the morphological characteristics identified by Prof. Zhongqi Yang [11] is important in order to develop a biological control method for the newly discovered parasitic natural enemy. Studies on the biological characteristics and behavior of the new species might help in determining whether it is an important parasitoid and a biological agent for the control of larvae and pupae of M. alternatus.
The COI gene sequences for S. guani and S. sichuanensis were exactly the same. Thus, we speculated that the samples for the individuals of the two species had been mixed with each other. Subsequently, we found a laboratory where S. guani and S. sichuanensis species were reared over a long time; S. guani and S. sichuanensis samples were then obtained from the Xishan Forest Farm in Beijing and Chongqing Academy of Forestry Labs, respectively. However, the results of gene sequence analysis were still the same. Further studies might be necessary to explain this phenomenon.
Moreover, there was 100% sequence similarity between S. pupariae and Sclerodermus sp. (No. 4). That might suggest that Sclerodermus sp. (No. 4) was indeed S. pupariae. In 2009, Xiaoyi Wang had used S. pupariae to control Sphenoptera sp. in Bayanhaote, Inner Mongolia. We also collected Sphenoptera sp. from the same location in 2012 and reared Sclerodermus sp. (No. 4) from Sphenoptera sp. in our laboratory. Therefore, we could speculate that they are the same species. However, whether they should be considered as the same species needs further research. Furthermore, Sclerodermus sp. (No. 4) was found in Alashan, Inner Mongolia, suggesting that they could successfully survive and reproduce there. Thus, artificially enlarging their settlement range might contribute substantially to their biological control.
Our results that Bethylidae and Bethylinae as well as Pristocerinae are strongly monophyletic, and that Pristocerinae are clustered with Foenobethylus and Mesitiinae are clustered with Scleroderminae, are consistent with the findings of Martin Carr in 2010 [23]. Further, the relationships between the three clades Epyrini, Pristocerinae, and Mesitiinae + Scleroderminae are close, supporting the fact that they might be monophyletic. However, in our study, Sclerodermus was found to be more closely related to Cephalonomia than to Sclerodermus sp. MC-2010, which thus, was not considered as the sister of Sclerodermus. This discrepancy could be attributed to the geographical difference.
In summary, the Sclerodermus species reared in our laboratory belonged to three wellknown species S. guani, S. sichuanensis, and S. pupariae, as well as five unidentified Sclerodermus populations, all of which were small and very similar morphologically. Because distinguishing these Sclerodermus species is difficult for many applied entomologists, we researched the differences between the five unidentified Sclerodermus species and the three well-known species by comparing different gene sequences to verify whether they were in accordance with the existing species-level identification or they were a new species. Firstly, on the basis of traditional morphological identification, which provided reliable evidence and very similar gene sequences Sclerodermus sp. (No. 4) was identified as S. pupariae. Secondly, although S. guani and S. sichuanensis have been published as two different species, our results showed 100% sequence indetity providing reliable evidence for revising the two species as one in the future. Thirdly, this is the first record of Sclerodermus sp. (No. 5) parasitizing M. alternatus larvae and pupae in Pinus massoniana Lamb. Further 95% of the female adults reared from one M. alternatus larvae were alate females, and the sequence data confirmed that this is a new species. Sclerodermus sp. (No. 2) was also first recorded in Pinus massoniana Lamb but parasitizing the larvae of Chalcophora japonica. Further, nearly all reared females were apterous. Thus, although Sclerodermus sp. (No. 2) and (No. 5) were found in Pinus massoniana Lamb, their hosts and morphological features differed; we could not clarify whether they are morphospecies. Sclerodermus sp. (No. 3) was first recorded parasitizing the larvae of Chrysobothris sp., which attacks Larix gmelinii in the north of China. Lastly, the sequences of Sclerodermus sp. (No. 1) were the most similar to those of S. guani, which parasitizes the pupae and larvae of Semanotus sinoauster Gressitt and Saperda populnea L. However, Sclerodermus sp. (No. 1) was first recorded parasitizing the pupae and larvae of Agrilus mali, which mainly attacks Malus pumila Mill and Crataegi cuneatae. Therefore, this species might have a new host. To arrive at the above conclusions, we have considered the host species, the tree species that the host attacks and morphological features to differentiate the unidentified species from the well-known species. A similar method was used for the first record of Meteorus oviedoi (Hymenoptera: Braconidae), a hymenopteran parasitoid of Venadicodia caneti (Lepidoptera: Limacodidae) in Costa Rica [30]. A previous study showed that combining several genetic characters may be the most practical way for efficient species identification [31]. In addition, we used molecular identification to verify our results obtained by traditional morphological identification.
Thus, although the gene sequences prove that Sclerodermus sp. are similar to the three well-known species, in our parallel research, we have been conducting other molecular analyses, such as ITS-1 and ITS-2, Cytb, and ATP6 to verify whether there are any differences in the results.
Wang. Finally, we particularly thank the support received from the Key Laboratory of Forest Protection, China State Forestry Administration and Research Institute of Forest Ecology, Environment and Protection, Chinese Academy of Forestry, Beijing, China.