Taxonomy and Identification of the Genus Scolopendra in China Using Integrated Methods of External Morphology and Molecular Phylogenetics

The centipede Scolopendra has important medicinal value and high toxicity, making it to be an interesting subject for evolutionary studies. However, species identification in China is difficult because of limited resource exploration and lack of recent taxonomic revision. To improve the identification and taxonomy of the genus Scolopendra in China, an in-depth investigation was conducted, and an integrated method that combined morphological characteristics with molecular data was applied. The identification key was revised to show the main difference among species. Our results indicated that morphologically-delimited species were consistent with the molecular analysis inferred from the COI sequences with genetic distances and phylogenetic trees. Additional morphometrics of four characteristics provided criteria for shape variation. These results suggested that the members of the genus Scolopendra in China could be delineated as 14 separate species. A new species from Lufeng county, Yunnan province, was proposed according to its characteristics, which was named as S. lufengia sp. nov. Our results comprehensively ascertained the taxonomic status of Scolopendra species in China, explored their phylogenetic relationships, showed a high success in the identification of medicinal centipedes.


Results
Morphological identification. As described in the methods section, 39 batch samples were observed. The characteristics, i.e, number of antennal articles, presence of gonopods, tarsal spurs on 20th leg and spines on the ultimate leg prefemur, had been identified, and the subtle features were inspected by light microscopy. The diagnostic characteristics of these assigned species are recorded in Table 1. All samples examined had common characteristics of the genus Scolopendra. Based on morphological characteristics, the examined samples were divided into seven nominal species and one putative new species. The seven species are as follows: S. mutilans, S. multidens, S. dehaani, S. mojiangica, S. negrocapitis, S. subspinipes and S. hainanum. The specimens collected from Lufeng County, Yunnan province, have some characteristics different from others, the details are as follow: Dark green-colored cephalic plate and tergites; legs are yellow at the base and gradually become green at the end; body length is less than 50 mm. There are many small pits on the head, and four ocelli on each side of the head's front. The antenna have 18 articles, 6 of which are glabrous. The tooth plate is clearly separated into two parts, each part with 4 teeth. The paramedian sutures start on tergite 4, the complete and visible margination start at tergite 8. Tergite 21 has a complete margination without sutures or depression. Sternites 2 to 19 show nearly complete paramedian sutures. The coxopleural process is conically shaped and usually 3-tipped. A dense, small stripe-like pore area leads to the tip of the coxopleural process. All locomotory legs have 2 accessory claws, and legs 1-19 show 1 tarsal spine. The ultimate legs are strong and short; the prefemur shows 1 VL, 2 VM, 2 DM and 3 corner spines (see Fig. 1). These characteristics have never been reported in China and do not match the records of Lewis 7 . It was accordingly identified as a new species and named under the standard binomial nomenclature as S. lufengia sp. nov., which was registrated in Zoobank, and the registration number is urn:lsid:zoobank.org:act:AEAB0CFC-3AB6-48BB-86CB-C6A61A74D27B. The voucher specimens are preserved in the Chinese Medicine Resource Centre of Hubei University of Chinese Medicine. The holotype is registered as WGLF11-2-20151023.
Previously, researchers had mainly focused on the medicinal use of the centipedes, and some incomplete keys for these species in China were used for identification 8,32 . However, there were still many species which were not included in the keys. In this study, the morphological characteristics of 8 collected species were examined. SCIEnTIfIC RepoRtS | 7: 16032 | DOI: 10.1038/s41598-017-15242 -7 In combination with records of 6 other species in China, i.e., S. japonica, S. morsitans, S. amazonica, S. mazbii, S. calcarata and S. diaoluoensis 33 , we were able to systematically revise the identification key for the Scolopendra species in China. The key is given in Table 2.
Morphometric analysis. Overall, 124 representative specimens from eight species were selected for analysis. The samples were as follows: 42 specimens of S. mutilans, 25 of S. multidens, 15 of S. dehaani, 16 of S. mojiangica, 13 of S. negrocapitis, 8 of S. subspinipes, 4 of S. hainanum and 2 of S. lufengia sp.nov. The length and width of four parts that were measured in previous morphometric studies of Scolopendra 2,9 , including the body, cephalic plate, tergite 21 and ultimate leg prefemur, were likewise measured in our study. The body length and the length-width ratios of the cephalic plate, tergite 21, and prefemur of ultimate leg were obtained. The shape variation was analyzed using the Kruskal-Wallis test in SPSS Statistics Version 23. The results are given in Fig. 2. Based on the statistical results, there are significant differences among groups of the four characteristics for P < 0.05 on the whole. Body length. The length ranges from 4.80 cm to 23.80 cm, the total mean is 13.36 cm, and the mean for each species is 16.64, 14.50, 7.96, 9.66, 14.52, 13.95, 14.04 and 4.85 cm respectively (see Fig. 2a). There are significant differences between S. mojiangica and the other four species of S. multidens, S. mutilans, S. dehaani and S. subspinipes (P = 0.000 to 0.025, P < 0.05); S. mutilans has significant difference from S. multidens and S. negrocapitis (P = 0.003 to 0.035, P < 0.05); and S. multidens has significant difference from S. negrocapitis and S. lufengia (P = 0.000 to 0.005, P < 0.05). Other species have no significant difference (P > 0.05). According to the means of the measured data, the species can be divided into two groups, namely, large-or small-size species. The large-size group mainly contains S. multidens, S. dehaani, S. hainanum and S. subspinipes. Their mean lengths are more than the total mean of 13.36 cm. The small-size group includes S. mojiangica, S. negrocapitis and S. lufengia, and their mean lengths are less than 13.36 cm. Although the measured mean length of S. mutilans is more than 13.36 cm, its length is usually between 10 and 13 cm, for few individuals are longer than 13 cm, thus S. mutilans is treated as a small-size species. L/W ratio of cephalic plate. The length-width ratio of the cephalic plate ranges from 0.87 to 1.51. There is significant difference between S. dehaani and S. mutilans or S. negrocapitis (P = 0.000 to 0.006, P < 0.05), and between S. mutilans and S. mojiangica (P = 0.000, P < 0.05). However, there is no significant difference among the others (P > 0.05) (see Fig. 2b).

Molecular analysis. Sequence annotation.
A total of 59 raw nucleotide sequences from partial gene targets for COI, representing 8 nominal species, were successfully amplified and sequenced. All raw sequences were verified with other available Scolopendromorph sequences in GenBank using the BLASTn algorithm. The results showed that all the sequences belonged to the homologous sequences of the genus Scolopendra, and the outgroup contamination did not affect the genomic DNA. The final aligned sequences obtained by sequence editing and the alignment program consisted of 647 bp. The G+C content is 29.9% ~ 38.0%. The contents of A+T are obviously higher than those of G+C. The average percentages of the G+C content of species are given in Table 3. The sequences of COI gene fragments consists of 327 variable sites and 320 conservative sites. Corrected genetic distances were calculated by the Kimura-2-Parameter (K2P) model for DNA sequence alignment. The intraspecific maximum K2P distance is 0.000 to 0.129, and the interspecific minimum K2P distance is 0.000 to 0.187. The intraspecific maximum distance of each species is less than the interspecific minimum distance. The interspecific and intraspecific genetic distances are summarized in Table 4.
Phylogenetic analysis. Eighty sequences of 12 Scolopendra species (including 59 extracted sequences and 21 downloaded sequences) and two downloaded sequences of the scolopendrid subfamily Otostigminae, Otostigmus scaber Porat, 1876 and Rhysida longicornis Pocock, 1891 were used to construct the phylogenetic tree. Two optimality criteria of neighbor-joining (NJ) and maximum likelihood (ML) were applied for analysis. In the NJ tree, the members of Scolopendra are divided into two groups (see Fig. 3: clade A and clade B). The first group (clade A) is subdivided into two clades (clade C and clade D). Clade C is further divided into clade E and clade F. Clade E is mainly composed of S. mutilans, S. amazonica, S. morsitans and S. dehaani, whereas clade F includes S. subspinipes and S. hainanum. Clade D is further divided into clade G and clade H. Clade G is a single branch of S. lufengia sp. nov., whereas clade H consists of S. multidens, S. japonica, S. negrocapitis and S. mojinagica. The other group, clade B, is the species S. calcarata. In the topological structure, 10 morphological species of the genus Scolopendra are clustered into separate branches, and the taxonomy is supported by molecular data. The other two species, S. subspinipes and S. hainanum, are clustered into one branch together. They could not be differentiated from one another, and these were inconsistent with the morphological taxonomy 9 . In the ML tree, the structure shows separated groups (see Fig. 4), except S. subspinipes and S. hainanum overlaps with two sequences of S. mutilans and one sequence of S. multidens which may be suspicious sequences. The groups of S. amazonica, S. morsitans, S. subspinipes and S. hainanum are divided into different clades compared to the NJ tree. The putative species of S. lufengia sp. nov. is still a separate branch.

Discussion
Scolopendromorpha are widely distributed in the world, especially in tropical and subtropical territories. As the southern China is one of the world's biodiversity hotspots, there is an abundance of biological species, including centipedes 8 . However, comprehensive investigations of the genus Scolopendra in China have seldom been carried out, and the exact species and distribution range are still unknown. In our survey of many regions of China, we found that the animals of Scolopendromorpha are widely distributed across the country, but the species of the genus Scolopendra are mainly confined to the southern region of the Qinling-Huaihe isotherm (which is also the boundary between north and south China). Each species lives in relatively fixed distribution areas and under suitable geographical climate conditions 8 .
In recent years, many new species had been found in China, such as S. negrocapitis and S. mojiangica 34,35 . Many former subspecies, i.e, S. multidens, S. dehaani and S. japonica, were also elevated as valid species 8,9,32 . Although there are obviously differences in morphological characteristics between S. mutilans and S. subspinipes, Siriwut 1 failed to distinguish these two species in the phylogenetic tree. However, in our study, they clustered into two different branches in the tree, and the result is consistent with report of Vahtera 36 , which supports the status 1 Spines arranged in 1-2 rows on the ventral of ultimate leg prefemur 2 -Spines arranged in 3 rows on the ventral of ultimate leg prefemur 10 2 The color of the cephalic plate and the first tergite is nearly reddish, which is different from other tergites 3 -The color of the cephalic plate and the first tergite is similar with other tergites 5 3 The color of the cephalic plate and the first tergite is orange-reddish, other tergites dark green. 20th leg with tarsal spur, 2 spines on the ventral lateral of the ultimate leg prefemur S. mutilans L. Koch, 1878 -The color of the cephalic plate and the first tergite is reddish brown, other tergites brown. 20th leg without tarsal spur, 2-3 spines on the ventral lateral of ultimate leg prefemur 4 4 The color of the tergites is brown with red, the ultimate leg strong, male gonopods absent S. multidens Newport, 1844 -The color of the tergites is brown with dirty green, the ultimate leg slender and long, male with gonopods. S. japonica Koch, 1878

5
The size of individual is larger, the length of the biggest is over 200 mm 6 -The size of individual is smaller, the length is usually less than 100 mm 8 6 The ventral lateral of the ultimate leg prefemur without spines S. dehaani Brandt, 1840 -The ventral lateral of the ultimate leg prefemur with 1-3 spines 7   7 Legs with orange-brown stripes, The ventral lateral of the ultimate leg prefemur with 1 spine, 20th leg without tarsal spur, male gonopods absent S. hainanum Kronmüller, 2012 -The color of legs is yellow, red-brown or brown, The ventral lateral of the ultimate leg prefemur with 1-3 spines, mostly 2 spines, mostly 20th leg with tarsal spur, male with gonopods S. subspinipes Leach, 1814 8 The color of basal legs is yellow, gradually it become green at the end, the ventral lateral of the ultimate leg prefemur with 1 spines S. lufengia sp.nov.
-The color of legs is yellow to reddish brown, the ventral lateral of the ultimate leg prefemur with 2-3 spines 9 9 The color of cephalic plate and tergites is dark brown, the basal antennal articles are yellow brown, the ventral lateral of the ultimate leg prefemur with 2 spines, and 3 corner spines on prefemur  of separate species. Previously, S. subspinipes was considered to be a variable species widely distributed in the southern area, Kronmüller 9 explored the differences among individuals in morphology, and then S. hainanum was separated from S. subspinipes as a new species. Nevertheless, these two morph-species are not supported by our molecular information and morphometric data. The samples collected from Lufeng, Yunnan, have specific morphological characteristics as described above, which are obviously different from those of other species in China. The morphometric data show that this species has a small body length and strong ultimate leg prefemur with a ratio of approximately twice the length to width. The molecular analysis also indicates its independent taxa status, which is resolved as a separate branch in the phylogenetic tree. Its geographic distribution is near to that of S. mojiangica, but it is obviously different from S. mojiangica in both morphology and molecular features. As a result, we assumed it to be a new species and named it S. lufengia sp. nov. According to our description and sporadic reports 8,[32][33][34] , to date, there are 14 species of Scolopendra distributed in China. Previously, morphological examination was the main approach applied for taxonomy of the genus Scolopendra. However, there is much debate around the taxonomic status of some species, especially in terms of the taxonomy of the former S. subspinipes complex and S. morsitans complex. In our study, an integrated method of morphology-molecular analysis has been used to validate the taxonomy. Based on the morphological identification, the characteristics of the collected samples were redescribed. Combined with previous reports, the   identification key was revised to show the main differences among Scolopendra species in China. The morphometric analysis made the subjective characteristics digitized and more intuitive, which further highlighted the morphological variations among species, especially in the body length and the prefemur of ultimate leg. The COI barcoding both in NJ and ML analyses further confirmed the determination, and it provided a new approach to solve these difficulties in identification by using the genetic distance and phylogenetic trees, from which 10 of the 12 putative species were successfully classified. There are deficiencies in identifying damaged or powder samples by morphological methods, difficulties also existed in classifying species that are closely related using molecular methods, but the integrated approach can make these two methods complement each other, which showed good results in our identification of the Scolopendra species. In this way, we believe that the integrated approach gives evidence for the validation of the Scolopendra species in China to be separate species; thus, the former subspecies of S. subspinipes and S. morsitans in China are reasonably valid species. The morphological characteristics of S. lufengia do not match those of other Scolopendra species, and it can also be confirmed by molecular data to be separate taxon. In contrast, the characteristics of S. hainanum match the description given by Kronmüller 9 (see Table 1). However, the molecular data suggests that there is no obvious difference between S. hainanum and S. subspinipes in either the genetic distance or the phylogenetic tree, so these two morph-species would be considered one molecular-species for clustering into a single branch on the tree. We speculate that there would be a close relationship between these species, and that the unapparent differences of genetic distance in COI gene fragments from our limited samples and data from the same collected area of Pubei, Guangxi (Fig. 5e) would not be sufficient to be reflected on the tree. Thus, we consider the taxonomic status of S. hainanum to be still undetermined, and taxonomy for these two morph-species still warrants further study. In our study, the genetic relationships among most species are consistant in the NJ and ML trees, but the species of S. amazonica, S. morsitans, S. subspinipes and S. hainanum are clustered into different clades. Meanwhile, although the medical value of Scolopendra has gradually been recognized, the high risk of toxicity to humans is an inevitable topic 17,37 . Because there is difference in the toxicity and pharmacological effect between species 18 , accurately identifying Scolopendra species will help to reduce the risk, and their medicinal properties would be fully utilized. In this study, the authenticity of the Scolopendra species in China can be verified using the integrated method of morphology identification and COI barcoding, and S. mutilans can be distinguished accurately from the local medicinal species or other commonly confused species in markets.
In morphological examination, characteristics are usually used to evaluate identification and taxonomy. However, in fact, the importance of the characteristics in evaluation are not always equivalent, because there are flexible features, such as coloration which is one of the most variable characteristics. Usually the species are divided into monochromatic or dichromatic pattern according to the color of cephalic plate and tergites. Most individuals of S. dehaani have a monochromatic pattern for the brown body, but some of them also have reddish cephalic plates and brown or entirely black tergites, comprising a dichromatic pattern. Even the same color pattern maybe displays difference. S. subspinipes has a monochromatic pattern, and the coloration has a large range, which varies from reddish brown to blackish. The yellow legs of S. mutilans from Hubei province and the red legs from Zhejiang province are another obvious example in coloration. In immature stages or at ecdysis, the color is also relatively duller than in adults. The margination or paramedian suture on tergites and sternites is another unstable characteristic.The first tergite with suture or margination is occasionally inconspicuous in individuals, and can overlap with other related species. Abnormal structures will also cause errors in identification and taxonomy. Sometimes smaller or special regenerations are found in a damaged leg or antenna, which are obviously different from the protogenetic legs or antennae. Thus, it is not sufficient to differentiate species just based on these variable characteristics. However, some stable and specific features should be chosed as diagnostic characteristics. For example, S. mutilans has tarsal spurs on 20 legs, whereas the 20 legs of S. multidens lack tarsal spurs; the gonopods of S. multidens and S. hainanun are not visible; and ventral lateral spines are lacking on the ultimate leg prefemur of S. dehaani. In our morphometric study, four features are used to show the shape variation. In light of the body length and the length-width ratio of the ultimate leg prefemur, the samples can be easily divided to different types. Hence, we supposed that these critical and stable characteristics, including antennae, gonopods,  tarsal spurs on 20th leg and spines on the ultimate leg prefemur, as well as the length and length-width ratio of the ultimate leg prefemur (see Table 1 and Fig. 2), should be regarded as important taxonomical characteristics, which would play a very important role in identification and taxonomy. Reports have shown that some Scolopendra species have different patterns, in either external morphology or genetic materials, that are caused by their geographic environments. S. dehaani in Southeast Asia has five color patterns, and they are clustered into five small branches in the phylogenetic tree 2 . In our study, we found that the color of the legs in most individuals of S. mutilans was yellow in Hubei province, but red in Zhejiang province. In the phylogenetic tree, samples of S. mutilans from Hubei and the adjacent areas of Henan and Anhui are all clustered into one branch, whereas samples from Zhejiang are united into another branch (Fig. 5a). These conditions caused by distribution also exist in S. multidens and S. negrocapitis. The samples of S. multidens collected from Yulin and Pubei are clustered into separate branches (Fig. 5c), and samples of S. negrocapitis collected from Suizhou of Hubei and Chaohu of Anhui are also clustered into two separate branches (Fig. 5d). S. dehaani collected from Yunnan and Guangxi further confirmed these patterns (Fig. 5b). We consider that the geographical and ecological environments, such as mountains or plains, temperature and humidity, etc, might play an important role in difference of morphology or genetic material.
In our study, although as many as 39 batches of samples were collected, there are still limitations. First, it is difficult to obtain enough samples from all regions, some species can not be collected successfully, or the sample size of some species is small (e.g., only two specimen of S. lufengia were collected). In addition, certain samples were of poor quality after being stored for a long time, so their sequences could not be extracted during our molecular analysis. These factors may lead to uncertainty in the results. In future, further investigations and more representative samples would be needed to perfect the result and bring more accuracy to the identification and taxonomy of Scolopendra in China, especially in the putative species of S. lufengia. But this integrated method of morphology-molecular analysis would also provide a reference for the study of other species and medicinal materials.

Materials and Methods
Sample materials. A total of 39 batch samples were collected from the natural field during the course of surveys in southern China, commodity markets or companies since 2015. The sample information is given in Table 5. The living specimens were relaxed with 50% ethanol for 10~20 min and then transferred into 70% ethanol to settle their posture for photography. The samples for molecular analysis were kept in absolute ethanol at    antennal articles, and those that are sparsely hirsute or glabrous, number of teeth, spines on the coxopleuron, prefemur spines arrangement on the ultimate legs, presence or absence of tarsal spurs on legs 19 and 20, and presence or absence of a gonopod on the first genital segment of the male. The photos were taken with a Nikon D7000 digital camera assembled with a Nikonlens, and the fine features were observed using an Olympus optical stereomicroscope equipped with an imaging system (Olympus, Japan). All samples were classified into nominal species, and the identification key of species in China was revised.
Additional morphometrics. The representative specimens were chosen randomly from the samples of nominal species to avoid personal bias. In light of previous literatures, some characteristics have important value in identification and taxonomy 2,9 . The typical morphological features of four parts, including the body, cephalic plate, tergite 21 and ultimate leg prefemur, were chosed for this analysis. The length and width of these four parts were measured with a vernier caliper; and the length-width ratios of the cephalic plate, tergite 21 and ultimate leg prefemur were calculated. The data was analyzed using the SPSS Statistics Version 23 software. The Kruskal-Wallis test was used to determine significant difference. A P value less than 0.05 was considered to indicate statistical significance.
DNA extraction, amplification and sequencing. 40mg separated tissue from the locomotory legs of samples was disinfected with 75% ethanol and then dissected. The blood/cell/tissue TIANamp Genomic DNA Kit (Tiangen Biotech Co., China) was applied to extract total DNA according to the instructions. The PCR mixture consisted of the following: 2.0 μL of DNA template, 1.0 μL of forward and reverse primers, 12.5 μL of 2×Taq PCR Mix and 8.5 μL of ddH 2 O. COI gene fragments were chosen for detection, and the universal primer sequences used were 5′-GGTCAACAAATCATAAAGATATTGG-3′ (LCO1490, forward) and 5′-TAAACTTCAGGGTGACCAAAAAATCA-3′ (HCO2198, reverse) 39 . All PCR mixtures were activated using a PCR analyzer (Prime 5G, Techne British Co.). The COI gene amplification was performed under conditions of PCR reactions cycled at 94 °C for 1 min as an initial step, followed by 5 cycles of 94 °C for 1 min in a denaturation step, 45 °C for 1.5 min in an annealing step, and 72 °C for 1.5 min in an extension step. This process was followed by 35 cycles of 94 °C for 1 min in the denaturation step, an annealing step at 50 °C for 1.5 min, 72 °C for 1 min in an extension step, and then a final extension step at 72 °C for 5 min. The PCR cycler was programmed at a holding temperature of 4 °C as the final step. The PCR products were detected on 1% (w/v) agarose gel electrophoresis in a 0.5×TBE buffer. The fluorescence of PCR bands were enhanced with SYBR Safe illuminant and observed under UV light. The PCR products were directly cycle-sequenced using the Sanger method and original amplification primers at Sangon Biotech (Wuhan) Co., Ltd.
Phylogenetic reconstruction. DNA sequences were assembled in CodonCode Aligner V5.1.5 (CodonCode Co., USA). Double strand sequence comparisons were made using a shadow pairwise alignment function analysis to detect missing sites and gaps in nucleotide sequences based on the chromatograms for each sequence sample. The sequences (including the sequences from 8 collected species and 4 species from GenBank; 8 sequences of S. mutilans, S. dehaani and S. multidens were also from GenBank) were used to construct the phylogenetic tree. Moreover, two sequences of the scolopendrid subfamily Otostigminae from GenBank, Otostigmus scaber Porat, 1876 and Rhysida longicornis Pocock, 1891 were chosen as outgroup to root the trees. All of the final sequences were blasted in GenBank with the homologous sequences as a check to evaluate the efficiency and then submitted to GenBank. A summary of the sequences information is in Table 5. The DNA sequences were aligned using MUSCLE, and the genetic distances were computed using MEGA 6.06 according to the K2P model. The NJ and ML analysis were applied to construct phylogenetic trees. The NJ analysis was constructed in MEGA 6.06, the concatenated files were analyzed with K2P model, and the likelihood-based analyses were performed with 1,000 bootstrap replicates. Codon positions were combined as 1st+2nd+3rd+Noncoding. All positions containing gaps and missing data were eliminated. The ML analysis was conducted using PhyML 3.0 (Online execution) based on the General Time Reversible (GTR) model. Fast likelihood-based analyses were performed with 1000 bootstrap replicates, SPRs was applied for the tree search, and the starting tree was selected with BIONJ.