A Formal Re-Description of the Cockroach Hebardina concinna Anchored on DNA Barcodes Confirms Wing Polymorphism and Identifies Morphological Characters for Field Identification

Background Hebardina concinna is a domestic pest and potential vector of pathogens throughout East and Southeast Asia, yet identification of this species has been difficult due to a lack of diagnostic morphological characters, and to uncertainty in the relationship between macroptyrous (long-winged) and brachypterous (small-winged) morphotypes. In insects male genital structures are typically species-specific and are frequently used to identify species. However, male genital structures in H. concinna had not previously been described, in part due to difficulty in identifying conspecifics. Methods/Principal Findings We collected 15 putative H. concinna individuals, from Chinese populations, of both wing morphotypes and both sexes and then generated mitochondrial COI (the standard barcode region) and COII sequences from five of these individuals. These confirmed that both morphotypes of both sexes are the same species. We then dissected male genitalia and compared genital structures from macropterous and brachypterous individuals, which we showed to be identical, and present here for the first time a detailed description of H. concinna male genital structures. We also present a complete re-description of the morphological characters of this species, including both wing morphs. Conclusions/Significance This work describes a practical application of DNA barcoding to confirm that putatively polymorphic insects are conspecific and then to identify species-specific characters that can be used in the field to identify individuals and to obviate the delay and cost of returning samples to a laboratory for DNA sequencing.


Introduction
The Blattaria (cockroaches) are a diverse order of some 4000-4,500 species, the majority of them denizens of tropical forests, but about 40-50 of all the known cockroach species are important domiciliary pests or house frequenting dwellers [1,2]. They carry numerous pathogens and could potentially transmit disease to humans [2][3][4][5]. Hebardina concinna is one of these pests. H. concinna is found in human dwellings and is believed to be a primary house pest [1], and could potentially transmit disease to humans; hence monitoring populations of this cockroach and identifying individuals within human dwellings is relevant for public health. However, due to wing-length plasticity within this species and related cockroaches, identification of individuals to the species level has been problematic.
The difficulty in describing H. concinna is reflected in the literature. De Haan first described this species as brachypterous [6] with the name of Blatta (Periplaneta) concinna. Shiraki later described females as brachypterous and males as macropterous [7]. Hebard erected a new genus of Blattina with Blatta (Periplaneta) concinna as type specimen and described the tegmina as moderately reduced, but without providing detailed descriptions of the tegmina or of male genital structures [8], although later work showed that the generic name of Blattina was preoccupied by Germar in 1842 [9]. Bey-Bienko replaced Blattina Hebard with Hebardina in order to memorialize Hebard and separated Hebardina Bey-Bienko from Periplaneta Burmeister on the basis of shortened elytra and hind wings [10], while Bruijning showed that the length of tegmina and the hind-wings of H. concinna varied widely, and listed all the tegminal lengths of the specimens that he examined [11]. Asahina then described macropterous specimens from Thailand [1], and most recently Roth examined H. concinna specimens collected from Krakatau, Sumatra, Java, India, and the Philippine Islands and noted extensive wing polymorphism [12].
In sum, most authors have focused on using wing morphology to describe this species but this is plastic in many cockroaches, including-as we show below-in H. concinna, and therefore of poor utility for species identification. Wing polymorphism exists in many insect orders, and individuals of the same species can be brachypterous or macropterous. In some insects these differences are sex-based (termed dimorphism), but in many species, including H. concinna, these differences occur in both males and females and are a response to varying environmental factors during nymphal stages, such as photoperiod, temperature, nutritional status, or population density. In general, good environmental conditions allow high populations whose individuals develop large wings and disperse [13].
To date wing polymorphism in Blattodea has rarely been studied, and is made difficult by their cryptic, nocturnal habits [14] and by the difficulties in identifying species by morphological examination [15][16][17][18]. In addition, identification based on morphology has some limitations, for instance it is difficult, even for specialists, to accurately identify females and immature stages [19].
DNA barcoding was developed by Paul Hebert and colleagues in 2003 [20,21], and in the decade since this method has become an important tool for the identification of insects, including cockroaches [22][23][24]. ''Integrated taxonomy'' [25,26] or more specifically ''Barcodes and morphology taxonomy'' or just ''B&M taxonomy'' [27] combines the power of DNA barcoding with traditional taxonomic methods and the authors above, and we ourselves, believe it should be integrated into the research framework. Nevertheless, many barcoding studies are not integrated with taxonomic research [28,29]. Here we present an integrated barcoding and morphological study of a cockroach.
There is a broad consensus that barcodes contain information relevant for species delimitation, although in some cases a single mitochondrial marker is insufficient as a sole criterion [30]. Notably, successful barcoding identification depends upon genetic diversity being markedly lower within than between species [31]. The commonly adopted standard 658 bp COI segment has proven to be highly informative and useful for species-level identification [27,32], including the matching of morphotypes in species with polymorphic forms. Despite increasing use of DNAbased methods morphology remains the most commonly used method in taxonomic research despite suggestions to abolish it altogether [33], although the future role of morphology in the age of genomes is anyone's guess [27]. The integrated B&M taxonomic method is potentially a very fruitful one [27], and our study proves its utility.
We describe, for the first time, a detailed morphology for male genital structures in H. concinna, compare the male genital structures of both macropterous and brachypterous specimens carefully, and compare these morphological observations with DNA barcode sequences (mitochondrial COI and COII) from males and females of H. concinna and related species. We also provide a complete morphological re-description of this species based on samples collected in China.

H. concinna specimens
Nine putative males and six females of H. concinna, as well as reference individuals from other blattotid species, were collected with a sweep net at night in the leaves and litter layer in woody habitats with the assistance of headlight. Specific permission was not required for collecting in these localities, and the GPS coordinates (latitude and longitude) were provided in Tables 1 and  2. No endangered or protected species were collected for this work.

Morphological study
General morphology. Our terminology follows McKittrick 1964 [34], Grandcolas 1996 [35], Anisyutkin 2010 [36] and Anisyutkin 2013 [37]. The genital segments of the examined specimens were macerated in 10% KOH and observed in glycerin with a Zeiss Discovery V12 stereomicroscope. Wings were floated in hot water until fully spread, embedded in neutral balsam, then mounted on slides and covered with coverslips. Drawings were made using a Zeiss Discovery V12 stereomicroscope fitted with a Canon PowerShot G1X digital camera and drawn using Adobe Illustrator CS6. All images of specimens were photographed using a Canon 60D plus a Canon EF 100 mm f/2.8L IS USM Macro lens combined with Helicon Focus software. All specimens studied were pinned in a natural posture and deposited in the medical vector collections of the Zhongshan Entry-Exit Inspection and Quarantine Bureau (ZSCIQ). The specimens we collected in China fully match other published morphological descriptions of H. concinna individuals from other geographic locations [6,38,39].
Quantitative morphology. Length measurements were taken from the specimens using vernier calipers. Three measurements were taken for each of the 15 specimens: tegmen length, pronotum length, and body length (excluding tegmen length). These are reported in Table 1 and plotted (tegmen vs. pronotum and tegmen vs. body length) in Figure 1. The plots suggested that these three measures do not vary independently, and their relationships were tested using a means-independent T-test (SPSS 19.0) on pronotum/tegmen length and body length/tegment length between sexes and within morphotypes.

Molecular methods
Sampled specimens. The sampled individuals were preserved in 8 mL 95% ethanol immediately after capture, this was replaced with fresh 95% ethanol twice the next day. Macropterous and brachypterous males and females were used for genomic DNA purification. Various species of Periplaneta, Blattella and Rhabdoblatta were also studied as references. Sampled species are summarized in Table 2.
Genomic DNA extraction. A single hind tibia and tarsus were removed from each specimen for DNA extraction. All instruments used to remove leg tissues were cleaned with 70% ethanol and flame sterilized between each specimen. Genomic DNA was purified with a TIANamp Genomic DNA Kit (DP304, TIANGEN). Voucher specimens were labeled uniquely and deposited in the Medical vector collections of the Zhongshan Entry-Exit Inspection and Quarantine Bureau.
PCR amplification. We amplified a 658 bp segment of the mitochondrial COI gene using the standard arthropod DNA barcoding primers [40] and a 601 bp segment across the mitochondrial COI and COII genes, also using previously published and widely used primers [41]  and female individuals were selected for DNA barcoding (Table 1). COI primers were LCO1490 (GGTCAACAAATCA-TAAAGATATTGG) and HCO2198 (TAAACTTCAGGGT-GACCAAAAAATCA). The COII primers were CO1DL (CCWCGWCGWTAYTCWGAYTAYCCWGA) and CO2DL (WGAATARRCATAWSWTCARTATCATTG).
Reaction conditions were 95uC 3 min; 95uC 45 s, 50uC 45 s, 72uC 1 min, 34 cycles; 72uC 10 min. PCR products are stored at 220uC at the Zhongshan Entry-Exit Inspection and Quarantine Bureau. PCR product purification and sequencing. PCR products were purified with a TIANgel Midi purification Kit (DP209-02), linked to a T-vector with the TIANgen pGEM-T ligation kit, at 16uC overnight, then transformed into DH5a competent cells (TakaRa Biotechnology (Dalian) Co., Ltd.) for white/blue selection. White clones on the LB-agar selection medium plate with Ampicillin (100 mg/mL), IPTG (1 mM) and X-Gal (20 mg/ mL) were selected for PCR Screening with a TIANgen pGEM-T recombinant colony identification Kit, then 3-5 randomly chosen positive colonies were cultured in LB medium with Ampicillin (100 mg/mL) at 37uC overnight, plasmid DNA was purified with TIANprep Mini Plasmid Kit and then sequenced commercially (Life Technologies Corporation). Plasmid DNA is stored at 2 20uC, and colonies in 20% glycerol at 280uC. Mutation rates are higher in sequencing directly from PCR products than sequencing colonies from a cloned PCR product sequencing [42,43], so we chose to sequence clones for this work. For each PCR product 3-5 sequenced clones were used for analysis: the maximum differences among different clones were 3 bp out of the 658 (0.46%). This is much lower than the species limitation proposed by Hebert in 2003 [21], so a consensus sequence for each clone was used for all analyses.
Phylogenetic analysis. Sequences from H. concinna and other species were submitted to the International Nucleotide Sequence Database Collaboration via NCBI GenBank. As noted above we used two primer sets: one amplified a section of the mitochondrial COI gene and a second amplified a short stretch of the 39 terminus of the COI gene, the tRNA-leu gene, and the 59 306 bp of the COII gene. For this second amplicon the tRNA-leu and 39 COI sequences have little variation, and we used only the COII sequences from this amplicon for our analysis.
We also identified nine additional cockroaches in GenBank for which either or both the COI and COII regions were available and created separate COI and COII data sets for phylogenetic analysis ( Table 2). Unfortunately there were not enough individuals for which both regions had been sequenced to assemble a combined dataset using both genes so we analyzed the two data sets separately.
We estimated maximum likelihood and neighbor-joining phylogentic trees for both the COI and COII data sets using Mega 5.2 [44], and tested robustness of the results using nonparametric bootstrapping. For both data sets the most appropriate maximum likelihood model (TN+I) was identified using the model testing function of Mega and this model was used to estimate the ML tree for each data set. Support for each branch was assessed using the same model for 1000 bootstrap replicates. Neighbor joining trees were constructed for each data set using the same Tamura-Nei model and also tested using 1000 non-parametric bootstrap replicates.

Generic diagnosis
Middle sized and uniformly dark colored cockroaches. Sexual dimorphism inconspicuous. Tegmen and wings fully developed or reduced. Front femur Type A. Tarsus with 2 rows of spines along lower margin; pulvilli and arolium present; post-tarsus claws symmetrical, unspecialized. First abdominal tergum of male specialized, with a densely setose medial tergal gland. Supra-anal plate and paraprocts symmetrical. Hypandrium slightly asymmetrical. Left phallomere with L2d large, occupied upper margin. L3d with a hook in the terminal. L2v elongated, plate-like and additional with a spiniform curved process. Right phallomer with caudal part of sclerite R1 plate-like, processes; sclerite R2 with groove at cranial part and right side; R4 palte-like and inset in the groove of sclerite R2. This species has also been described with the following Names [9] Hebardina concinna (de Haan, 1842

General Description
Sizes of the examined H. concinna specimens are summarized in Table 1.
Except for the differences in the length and the vein of the tegmen and hindwings, the morphology and structure of the male genitalia of the macropterous and brachypterous specimens are identical as illustrated in Figure 2, Figure 3 and Male genital structures of the macropterous and brachypterous specimens are the same.

Quantitative morphology
Tegmen length was significantly different between the macropterous and brachypterous H. concinna morphotypes (t-test, P,, 0.01): tegmen lengths of the macropterous individuals were greater than the brachypterous ones without any overlap. Moreover, the ratios of pronotum/tegmen length and body/tegmen length were significantly different (t-test, P,,0.01, Table 3). This relationship is visually apparent in Figure 1, where macropterous individuals and brachyperous individuals cluster separately when tegmen length is plotted against body length or pronotum length. There were no significant differences between the sexes in any characters (Table 3). Tegmen length, tegmen/pronotum length, and tegmen/body length were significantly different between the macropterous and brachypterous H. concinna morphotypes, but were not significantly different between males and females within the same morphotype (Table 3).

DNA barcoding
PCR products of H. concinna. Amplified COI sequences (not including primers) for all individuals were 658 bp, with no stop codons, insertions or deletions, and could be translated into 219 amino acids without any interruption, mean nucleotide content of COI sequences was A (31.9%), T (37.4%), G (15.7%) and C (15.0%). As reported for other insect mitochondrial sequences [46][47][48], A + T (69.4%) was in higher proportion than G + C (30.7%), and were comparable to those typical of insects in general for this COI gene region [46]. The 601 bp COII amplicon (after removing primer sequences) included 199 bp of the 39 terminus of the COI gene, the entire 96 bp tRNA-Leu gene, and the 306 59 bases of the COII gene. The mean nucleotide content of this amplicon was A (40.4%), T (36.0%), G (10.5%) and C (13.1%), again A + T (76.4%) was in much higher proportion than G + C (23.6%), as is usual for insects.
Phylogenetic analysis. Sequences of 30 individuals belonging to 9 species, 4 genera, and 3 families were analyzed. Phylogenetic trees were estimated using aligned COI or COII data sets as described in the methods and presented in  The aim of the phylogenetic analysis was to assess the relatedness of macropterous and brachypterous H. concinna individuals. The short mitochondrial sequences used for DNA barcoding are appropriate for this purpose. For deeper phylogenetic analyses such short mitochondrial DNA barcoding region(s) may be uninformative. Nevertheless, the data sets we have assembled do strongly support the three cockroach families represented: the Blattidae, Blaberidae, and Ectobiidae. This result is completely concordant with the morphological classification. Within each family each named genus; Hebardina, Periplaneta, Rhabdoblatta, and Blattella are also supported by at least 90% of bootstrap replicates.
The five H. concinna individuals shared identical COI sequences and differed only by one or two bases in the COII region: this phylogeny shows that these individuals are clearly distinct from other species within the Blattidae. There is no genetic distinction between macropterous and brachypterous individuals in either sex. This agrees with our observation that male genital morphology is identical in cockroaches with both wing morphs and we conclude, as hypothesized, that macroptery and brachyptery in both males and females of H. concinna are different ecotypes of the same species.

Discussion
Wing polymorphism exists in many insects, including the Coleoptera, Diptera, Hemiptera, Hymenoptera, Orthoptera, Lepidoptera, and Thysanoptera. The reasons for this polymorphism may relate to both population size and selection pressure. For instance, when population densities are high the proportion of macropterous brown plant hoppers Nilaparvata lugens is higher than when it is low [49]. This same phenomenon has also been described in Coleoptera, Diptera, Hemiptera, Hymenoptera, Orthoptera, Lepidoptera, and Thysanoptera [50]. When populations are low, nymphs develop into solitary brachypterous adults, the energy saved by not building large wings and flight muscles is instead directed into reproduction, thus allowing populations to increase rapidly [51]. Conversely, when populations are high, individuals develop into macropterous gregarious adults that are strong flyers who disperse to other habitats with lower population densities. There is a resource allocation trade-off relationship between the development and maintenance of flight muscles and reproductive capacity. Our results clearly demonstrate wing polymorphism in cockroaches that is not sex-based, and therefore is likely to have an environmental trigger. Further research using H. concinna or other Blattodea might reveal these triggers in cockroaches.
Wing polymorphism is rarely reported in Blattodea. This is likely due to their secretive, nocturnal habits, to the lack of molecular data, and the lack of description of male genital morphology: these factors together mean that collecting is difficult and identification of collected individuals is problematic. Male genital structures are the most important characters used for species identification in many insects, especially for species that exhibit polymorphism in other characters. In particular, McKittrick [34] first used male genital characters to describe cockroach species. Following this work male genital characters have been widely used by many taxonomists such as Anisyutkin, Roth, and Grandcolas, to distinguish different species of Blattodea [35,36,39,52]. Our results show that male genital structures are also useful for identification of H. concinna, but also demonstrate that it may be necessary to use other methods, such as DNA barcoding, to identify a cohort of conspecific animals in order to develop morphological keys to the genitalia.
Our results showed that, for population in China, both H. concinna males and females have macropterous and brachypterous morphotypes and allowed us to develop morphologically based criteria for identification of (males of) this species. This means that entomologists, and in this case pest control agents and public health officials, can be trained to identify H. concinna quickly in the field by picking up individuals and looking at their abdomens without the need to bring samples back to a laboratory for costly and time consuming DNA barcoding. We have no doubt that similar efforts will allow leveraging of additional up front investments in laboratory barcoding to develop field-friendly keys for morphological identification of previously problematic species across the Metazoa.
H. concinna is distributed throughout East and Southeast Asia, and the type location is in Malaysia. Our work is based on individuals collected in China and our results therefore remain to be confirmed for the rest of this species' range. In particular, our work represents the only report of mitchondrial sequences from this species, so further work is necessary to understand genetic variation in ''H. concinna'' across the rest of its range. However, our experience with numerous cockroach species strongly suggests that our observations on genital morphology will be confirmed across the entire range of the species, and we hope our work will stimulate further research on this topic.  . Distance matrix/neighbor joining phylogenetic tree based on 306 bp of aligned cockroach COII nucleotide sequences. The maximum likelihood tree was topologically identical, although with differing branch lengths. Numbers on branches represent support from 1000 non-parametric bootstrap replicates for distance matrix-NJ analysis and maximum likelihood analysis, respectively. Missing numbers indicate branches with less than 50% support. This analysis clearly supports grouping of the five H. concina individuals as a single species. doi:10.1371/journal.pone.0106789.g006

Author Contributions
Cockroach Barcoding and Morphological Identification