A taxonomic review of the centipede genus Scolopendra Linnaeus, 1758 (Scolopendromorpha, Scolopendridae) in mainland Southeast Asia, with description of a new species from Laos

Abstract The centipede genus Scolopendra in mainland Southeast Asia is reviewed taxonomically based on morphological characters, informed by a molecular phylogenetic analysis using sequences from three mitochondrial and nuclear genes (COI, 16S rRNA and 28S rRNA). Eight nominal species of Scolopendra, namely Scolopendra morsitans Linnaeus, 1758, Scolopendra subspinipes Leach, 1816, Scolopendra dehaani Brandt, 1840, Scolopendra multidens Newport, 1844, Scolopendra calcarata Porat, 1876, Scolopendra japonica Koch, 1878, Scolopendra pinguis Pocock, 1891, and Scolopendra dawydoffi Kronmüller, 2012, are redescribed together with some revision of type materials. Geographical variation in each species has been compiled with reference to samples that span their distribution ranges in Southeast Asia and some parts of neighbouring areas such as East Asia, the Indian Ocean, and Africa. Comparative study of traditional taxonomic characters from external morphology provides further information to distinguish some closely related species. Scolopendra cataracta Siriwut, Edgecombe & Panha, sp. n., is described from the southern part of Laos, with additional records in Thailand and Vietnam. The phylogenetic framework for Southeast Asian Scolopendra recognizes Scolopendra calcarata + Scolopendra pinguis, Scolopendra morsitans, and a Scolopendra subspinipes group that unites the other six species as the main clades. Within the Scolopendra subspinipes group, two monophyletic groups can be distinguished by having either slender or short, thick ultimate leg prefemora and different numbers of apical spines on the coxopleuron. Scolopendra arborea Lewis, 1982, is placed in subjective synonymy with Scolopendra dehaani. A survey of external morphology of the genital segments confirms its potential for improving species identification in Scolopendra. Some observations on biology and behaviour are recorded based on field surveys in this area.


Introduction
The genus Scolopendra Linnaeus, 1758, is among the predominant centipede groups in tropical regions. These animals are generalist feeders that play an important role as one of the top carnivorous invertebrates in soil ecosystems. In several Asian countries, Scolopendra has symbolic status or figures in superstitions, and is used commercially in traditional medicine (Pemberton 1999). A few species have been proposed as model animals for medical and biological subjects (Minelli and Fusco 2004) but comprehen sive work on the regional biota has not been consolidated since the last monograph on Scolopendromorpha (Attems 1930b).
The taxonomic study of Scolopendra dates back to the late 19 th to mid20 th centu ries (Kohlrausch 1881, Pocock 1891a, Kraepelin 1903, Attems 1930b. Several names have fallen into synonymy through the course of taxonomic revisions (Kohl rausch 1881, Kraepelin 1903, Attems 1930b. Conversely, some populations that had been classified in geographically widespread species have recently been identified as distinct species e.g., S. antananarivoensis Kronmüller, 2010, versus S. morsitans Lin naeus, 1758, or S. subcrustalis Kronmüller, 2009, versus S. subspinipes Leach, 1816. The phylogenetic position of Scolopendra has been investigated in the context of broadscale phylogeny of Scolopendromorpha (Vahtera et al. 2012(Vahtera et al. , 2013. Combined molecular and morphological data supported a hypothesis that Old World species of Scolopendra can be distinguished from the New World species, and nested most Old World species in a clade with the genus Asanada Meinert, 1886. Morphological dis crimination between two regional groups within Scolopendra is made mainly based on a transverse suture on tergite 1 in New World Scolopendra that is absent in nearly all Old World species. Some Old World species are polymorphic with respect to external phenotypic characters. For instance, S. subspinipes and S. morsitans are cosmopoli tan species worldwide, and both of them include several colour variants (Kohlrausch 1881, Attems 1930a. Previous scolopendrid studies proposed that morphological variation within species is influenced by geographic distribution and ontogeny (Lewis 1968(Lewis , 1972. Ontogenetic variation in colouration patterns has been recorded in S. dehaani Brandt, 1840 from Southeast Asia, where the species has been investigated using both molecular and morphological data (Siriwut et al. 2015a).
To date, 99 described species of Scolopendra have been recorded (Bonato et al. 2016), of which fourteen species have been found in the Asian tropics (Schileyko 2007, Lewis 2010b, Kronmüller 2012. Taxonomic studies have been undertaken in the following regions of Asia: the Indian Subcontinent (Jangi and Dass 1984), Indochina including Burma (Pocock 1889, 1891b, Attems 1953, Schileyko 2007, the Malay and Philippine Archipelagos (Pocock 1894, Wang 1962, 1965b, 1967a, and the East coast of the China Sea (Takakuwa 1942a, Chamberlin and Wang 1952, Wang 1955a, b, 1956, 1957, including Taiwan (Chao 2008). In Southeast Asia, the following Scolopendra species are endemics: S. pinguis Pocock, 1891, S. gracillima Attems, 1898 Kraepelin, 1903 and S. arborea Lewis, 1982. There are also species which are widely distributed: S. subspinipes, S. morsitans, S. dehaani and S. japonica Koch, 1878; each of them extends into neighbouring territories such as the Indian subcontinent and the Asian temperate region (Koch 1878, Jangi andDass 1984). In mainland Southeast Asia, Scolopendra comprises ten nominal species. Most of them are cosmopolitan species found synanthropically. For several species, geographical variation has not previously been documented, but we now have access to molecular evidence by which such variability can be mapped to genetic structure among and between populations.
The validity of various scolopendrid species has been ambiguous because their dis tributions have not been comprehensively documented and/or because the diagnostic value of particular taxonomic characters has been unclear (Lewis 1978, Lewis et al. 2006, Edgecombe 2007. Infraspecific variation within Scolopendra has long been noted as a fundamental problem for distinguishing between similar species (Newport 1845, Attems 1930a, Lewis 2010b. Exclusive reliance on the traditional ex ternal morphological characters may not be sufficient to resolve some of these questions, not the least those involving problems of colour variation (Koch 1982, 1983a, Shelley 2005. The phylogenetic relationships of Scolopendra to other scolopendrid genera have also been contentious, as is the monophyly of the genus (Vahtera et al. 2013). Recently, evidence has been presented that morphological identification, molecular phylogeny, and geometric morphometric analyses congruently support the traditional delimitation of Scolopendra species in mainland Southeast Asia (Pocock 1891b, Flower 1901, Schi leyko 1992, 1995, Kronmüller 2012. Molecular sequence analyses of Southeast Asian species indicated previously unrecognized groups within nominal species (Siriwut et al. 2015a). These results may indicate that even within a morphologically conservative centipede group, cryptic species can potentially be identified.
In this work, we review Scolopendra species in mainland Southeast Asia. The type material of some species has been redescribed and, where available, type material is photographed and illustrated. All species are compared with the most closely allied congeners to provide distinguishing taxonomic characters. Variability in morphologi cal characters is recorded in order to document geographical variation. The description of a new species is based on specimens from three SE Asian countries. Molecular phy logeny of three standard genes is analyzed, adding new samples to previous work on Scolopendra, to test the monophyly of each species and to determine the phylogenetic position of new species. An identification key to Scolopendra is presented and distribu tion ranges for species are updated.

Methodology
Material examined. Specimens were collected mainly throughout mainland Southeast Asia, principally from Thailand, Laos, Cambodia, Myanmar, Singapore and Malay sia and kept at Chulalongkorn University Museum of Zoology, Bangkok, Thailand. Examination of additional Southeast Asian and other Oriental regional collections including available type material was based on both identified and previously undeter mined specimens in several museums. All specimens were observed by using either a LEICA MZ 16A, Nikon SMZ25 or Olympus stereomicroscope. Morphological char acters were photographed using montaged image stacks. Each morphological feature was serially captured with a Canon 700d linked to an automated calibration program, either Cell'D imaging or Helicon Focus on a desktop PC. In addition, illustration of some morphological variation was made by freehand drawings. Behaviour, biology and distribution. Feeding behaviour and brooding of eggs and hatchlings was observed and photographed both in the field and the laboratory. Characteristics of habitats and brood chambers are discussed in Lewis (1981), Mitić et al. (2012) and Siriwut et al. (2014).
The distributional ranges of all SE Asian Scolopendra species were reinvestigated based on field sampling, museum collections, and literature records from this region. Localities cited in the descriptions are arranged geographically and are separated in two sections: I. A determined locality refers to the corrected name of a locality. In cases of inac curate spelling and outdated names on old labels, we provide the corrected name in square parentheses based on resources from the internet and/or historical notes. Spellings of new collection localities in Thailand were transcribed by the Thai Ro manization program (Wirote 2001). Latitude and longitude coordinates are given for each new collecting locality, tracked by a GPS conductor via a Garmin GPS travelling device. II. An undetermined locality refers to a name that is localized at only a regional scale such as by region or country.
Distribution maps for each species include the records from recent field surveys and specimens from museum collections that provide sufficiently detailed locations. Each of those localities is marked by a filled symbol. Some localities from previous taxonomic work are included using a blank symbol. All undetermined localities and some specimens which were attributed only to a region, island or country have been excluded from the distribution maps.
The list of synonyms for each Scolopendra species follows Chilobase (http://chilobase. biologia.unipd.it/). Diagnoses are revised from Siriwut et al. (2015a), and the range of geographical variation from the type, voucher specimens and previous surveys.   Phylogenetic reconstruction. Southeast Asian and some temperate Asian Scolopendra sequences were obtained from GenBank, based on previous phylogenetic analy ses (Joshi and Karanth 2011, Vahtera et al. 2013, Siriwut et al. 2015a. We add more Scolopendra sequences from additional specimens collected during 2014 from various parts of the region (Table 1). DNA extraction methods follow Siriwut et al. (2015a). Three standard genes for centipede phylogeny (cytochrome c oxidase subunit I, 16S rRNA and 28S rRNA) were used to reconstruct phylogenetic trees. Maximum likelihood and Bayesian inference approaches were employed, using RAxML (Stama takis 2006) and MrBayes Ronquist 2001, Ronquist et al. 2012), respectively. Standard statistical tests were applied to evaluate branch support (boot strap support and posterior probability). Algorithms and parameter settings for both analyses followed protocols detailed previously by Siriwut et al. (2015a).

Phylogenetic relationships of mainland Southeast Asian Scolopendra and the position of Scolopendra cataracta sp. n.
The phylogenetic tree from the updated concatenated DNA dataset aggregates studied specimens into eleven monophyletic groups within Scolopendrinae that are compat ible with morphological identification (Fig. 1: Clade A). The phylogeny supports the monophyly of the genus Scolopendra, in contrast to a previous analysis in which a sampled species of Cormocephalus fell within Scolopendra (Siriwut et al. 2015a). The sequence annotation for each partial marker is given in Table 2. Genetic divergence was calculated by pairwise comparison of k2 parameter distance under one thousand bootstrap replicates (Table 3). The genetic distance among Scolopendra species ranges from 15.9-24.4% in .0% for European Scolopendra by Oeyen et al. (2014)). Comparing with different genera from the same/another subfamily, the distances are between 21.6-28.  in 16S, respectively. Within populations, intraspecific variation is between 8.3-18.4% in COI and 5.2-11.3% in 16S (Table 4).
Three main clades are identified in mainland Asian Scolopendra ( Fig. 1: Clade B), corresponding to pinguiscalcarata, subspinipes and morsitans groups ( Fig. 1: Clade C, D and E). The highest intraspecific variation is observed in S. pinguis, a species native to this region, and the lowest variation is in S. dehaani, which is widespread and the dominant species in the region. The high measure of genetic divergence among S. pinguis populations in our previous study prompted a reexamination of those specimens and additional ones that were added in this study. The updated phylogenetic tree revealed that the former S. pinguis clade (Siriwut et al., 2015a: fig. 1, clade C) can be divided into two species, S. pinguis and S. calcarata ( Fig. 1: Clade C), and this separa tion can also be supported by diagnostic morphological characters. These two species are distributed along the montane areas of Burma and Thailand and occur eastward to the Indochina subregion in mountain ranges between Laos and Vietnam. Synapo morphic characters shared by these two species are the comparatively robust, vaulted shape of their body segments, the tergite of the ultimate legbearing segment being acute posteriorly, four glabrous antennal articles, and the dichromatic colouration on the cephalic plate in all S. calcarata specimens and most S. pinguis populations. In ad dition, the phylogeny indicates that even after recategorising a closely related species (S. calcarata) that had previously been classified as an aberrant clade within S. pinguis, the genetic distance within S. pinguis is still considerable. This distance might suggest cryptic speciation among different geographical populations.  The remaining Scolopendra species may be divided into two groups, one consisting of S. morsitans and another including former subspecies of S. subspinipes sensu Kronmüller (2012). In the case of S. morsitans, monophyly is corroborated with high bootstrap support and posterior probability in ML and BI analyses, respectively ( Fig. 1: Clade D). Within the S. subspinipes group ( Fig. 1: Clade E), a clade uniting S. cingulata Latreille, 1829, S. japonica, S. dawydoffi and S. multidens ( Fig. 1: Clade F) differs from a clade composed of S. subspinipes, S. cataracta and S. dehaani ( Fig. 1: Clade I). A morphological feature shared by S. cingulata, S. japonica, S. dawydoffi and S. multidens is the cingulata-like ultimate legs, which have a dorsally flattened prefemur and femur and are much shorter and stouter than in the subspinipes clade ( Fig. 1: Clade I). Attems (1938) referred to similar groupings based on form of the ultimate legs when describing S. dawydoffi. The number of apical spines on the coxopleural process might also be useful for discrimination of these two groups (one or two versus more than two spines in the subspinipes and cingulata groups, respectively). A monophyletic group composed of S. subspinipes and the two allied species (S. cataracta and S. dehaani) received statistical support both in ML and BI ( Fig. 1: Clade I).
All samples of S. cataracta united as a clade with S. dehaani to the exclusion of S. subspinipes ( Fig. 1: Clade J). This new Scolopendra species shares the following mor phological similarities with various species of the S. subspinipes group: presence of two ventrolateral spines on the ultimate leg prefemur (as in S. subspinipes), long and slen der ultimate legs (like S. dehaani), and incomplete paramedian sutures on the sternites (like S. dawydoffi). However, S. cataracta is clearly distinguished from all of them by extremely short tergal paramedian sutures. Interspecific variation of DNA sequences ranges between 15.9-19.4% and 13.8-16.4% in COI and 16S, respectively, among these three related species.

Species diversity of Scolopendra in mainland Southeast Asia
In this region, nine species are identified from our survey. The taxomomic boundaries between species were based on information from both morphology and molecular analysis. Two other species of Scolopendra were not included in this paper, namely S. mirabilis (Porat, 1876), and S. hardwickei Newport, 1845. In the case of S. mirabilis, an Africancentral Asian species, the single known specimen in SE Asia may be in troduced, being found on an island in a coastal area of northern Vietnam (Schileyko 1995). Likewise S. hardwickei was reported from Singapore, the largest port in South east Asia (Decker 2013). This species has been documented from India and the Nico bar and Andaman islands, and it probably occurs in Sumatra and Java (Khanna 2001, Lewis 2010b, Decker 2013. Without further material from Singapore or neighbour ing areas, the status of this recorded species in the mainland SE Asian fauna is ques tionable. S. gracillima sternostriata Schileyko, 1995, from Vietnam (Schileyko 1995, 1997, is similar to S. pinguis in most respects. We include it in the key below but have no new material of this subspecies, and accordingly have not revised it. The following key to native species of Scolopendra excludes only the two first aforementioned species.  Aldaima, Maldives, 17/12/1952. NHMUK 1952.247, one spm., Astove Island, Indian   Anterior margin of T1 underlying cephalic plate (Fig. 3E). Complete paramedian sutures from TT4-5; margination typically starting on T14 (one spm., with margina tion restricted restricted to last two tergites). Tergite surface (Figs 3G, 5B) smooth. Tergite of ultimate legbearing segment (Figs 4C, 6A) curved posteriorly, with median suture; ratio of width: length of tergite of ultimate legbearing segment 1.34:1. Ster nites (Figs 4A, 5E) with complete paramedian sutures. Sternites without depressions. Sternite of ultimate legbearing segment ( Fig. 4E) with sides converging posteriorly; surface without depression. Porefield on coxopleuron terminating well beneath mar gin of tergite of ultimate legbearing segment, pore area slightly widened anteriorly.

Key to species of Scolopendra in mainland Southeast Asia
Coxopleural process moderately long or short, usually with 4-5 apical and 0-1 lateral spines (Fig. 4E); porefree area extending 40-50% length from distal part of coxopleural process to margin of sternite of ultimate legbearing segment (Fig. 6B).
Genital segments well developed, reaching longer than the distance between posterior margin of sternite of ultimate legbearing segment and distal part of coxopleural process (Fig. 7C). Sternite of genital segment 1 round and convex posteriorly, with median su ture. In male, sternite of genital segment 2 attached to penis. Tergites of genital segments without small setae. Gonopods with small setae in male. Penis with fine posterior seta.
Colouration. Scolopendra morsitans demonstrates colour variation among its populations in SE Asia. Previously, colour variation has been recorded in African, Australian and Taiwanese populations (Lewis 1968, Koch 1983a, Chao 2008, those studies proposing that latitude and habitat composition might affect this variability. Recent molecular analyses of ThaiCambodian S. morsitans suggested that some colour morphs may be specific to local populations (Siriwut et al. 2015a), although similar patterns occur in each of three different continental faunas. We have recorded the col ouration pattern in juvenile and adult specimens ( Fig. 2A    Discussion. Scolopendra morsitans is morphologically varied and subsumes many synonyms that are now attributed to geographical and/or ontogenetic variation. Intraspecific variation has been studied in Africa (Lewis 1969), India (Jangi 1955(Jangi , 1959, and Australia (Koch 1983a), revealing that some diagnostic characters are inconsistent within its populations. These include: number of glabrous antennal ar ticles, number of teeth on the forcipular toothplates, number of tergites that are marginated, and the number of legs with tarsal spurs. This species also demonstrates differences in colour patterns that might be correlated with its geographical distri bution. Lewis (1969) noted that a population of S. morsitans from Bihe, Angola, demonstrated a dark body with red legs whereas specimens from Sudan were straw coloured. Here we record two colouration patterns in Thai populations that do not occur sympatrically. In addition, some morphological characters might be restricted to certain geographical populations, such as a tarsal spur on leg 20, which has been reported from India and in some African populations. For this reason, the utility of this character for defining boundaries between S. morsitans and other Scolopendra species that share some morphological characters with it, such as S. laeta Haase, 1887 and S. antananarivoensis, is not absolutely clear. Our survey of geographic variation in S. morsitans is presented in Table 5.
Our phylogenetic analysis corroborates the monophyly of SE Asian populations of this species (Fig. 1). Previous molecular phylogenetic analyses of the S. morsitans complex in India suggested that S. morsitans was paraphyetic with respect to specimens that were determined as S. amazonica, the latter name being used for specimens with tarsal spurs on leg 20 (Joshi and Karanth 2011). From these results, it seems that mo lecular phylogenetics of this species complex throughout its geographic range may be necessary to clarify the taxonomic value of some variable morphological characters and to more confidently determine the taxonomic status of some phenotypically similar species. It is likely that some names currently treated as junior subjective synonyms of S. morsitans may be found to be applicable to cryptic species.
Genital segments well developed, reaching longer than distance between posterior margin of sternite of ultimate legbearing segment and distal part of coxopleural pro cess. Sternite of genital segment 1 round and convex posteriorly, with median suture. In male, sternite of genital segment 2 attached to penis. Tergite of genital segment without small setae. Gonopods with small setae in male. Penis with apical bristle.
Discussion. Recently, the taxonomic validity of S. subspinipes and its former sub species has been evaluated both by morphology (Kronmüller 2012) and molecular methods (Chao et al. 2011, Siriwut et al. 2015a. Three former subspecies of S. subspinipes, namely S. s. japonica, S. s. dehaani and S. s. cingulatoides (= S. dawydoffi), have been raised to species rank (Kronmüller 2012), whereas the remaining four subspecies (in the classification of Attems (1930b)) have been synonymized with the nominotypi cal subspecies. However, some subspecies still remain of ambiguous status. Notably,  (20) 1-19 1-19 (20) 1-19 (20) 1-20  (Schileyko 2007, Kronmüller 2012, whereas molecular analyses based on four loci found it to either resolve as sister taxon to S. subspinipes s.str. or to group more closely with other species (Vahtera et al. 2013). In this study, we document a syntype of S. mutilans Koch, 1878 in the NHMW collection (Figs 12-13) and reconfirmed its taxonomic status by using molecular analysis from the concatenated DNA dataset of S. subspinipes s.str. and S. mutilans Koch, 1878. The phylogenetic tree supports the proposition that this subspecies cannot be distinguished taxonomically from S. subspinipes. According to genetic divergence among examined populations, S. mutilans Koch, 1878 should be regarded as a geographical variant of S. subspinipes, as was suggested in other recent taxonomic studies (Schileyko 1995(Schileyko , 2007. Some morphological comparisons of several populations from Southeast and East Asia are provided in Table 6. S. subspinipes piceoflava, another former subspecies of S. subspinipes from Sulawesi, Indonesia, is currently treated as a synonym of S. subspinipes (see Kronmüller 2012), but may upon further study prove to be a valid species. Attems (1934) stated that it could be distinguished from other forms of S. subspinipes by yellowish colouration on the posterior part of its tergites. He also referred to the weakness or near absence of tergal paramedian sutures, which also occurs in some other SE Asian Scolopendra species (see Kronmül ler 2012 and discussion of S. cataracta in this study). Reexamining the syntypes of S. subspinipes piceoflava leads us to dispute the taxonomic validity of this character because paramedian sutures are visible on the tergites in all syntypes. In order to provide a more complete evaluation of its taxonomic status, a redescription of its syntypes is as follows: Description. Body length 16.7 cm in male and 17.1 and 16.5 cm in female syn types. Preserved male still exhibiting traces of its colouration pattern: cephalic plate and segments dark greenish or brown. Antenna yellowish. Tergites with yellowish or pale colour on posterior margin. All legs bluegreenish, distal part yellow. Cephalic plate without small punctae on anterior part, median sulcus present. Posterior part of cephalic plate without paramedian sulci.
Coxopleural process long ( Fig. 15C-D) with 1-2 apical spine(s) and absence of lateral and dorsal spines; porefree area extending 30-50% length from distal part of coxopleural process to margin of sternite of ultimate legbearing segment.
Sternite of genital segment 1 round and convex posteriorly, with median suture. In male, sternite of genital segment 2 attached to penis. Tergite of genital segment without small setae. Gonopods present in male.
Discussion. Based on examination of the syntypes, we corroborate the assignment of this nominal subspecies to the S. subspinipes group. Some morphological characters that appear, however, not to be identical with S. subspinipes are the sharpness and length of the coxopleural process, which bears one or two strong apical spines, the ratio of ultimate leg podomeres, and the colouration pattern on the tergites that is clearly distinct from other geographical populations of S. subspinipes (the posterior part of the tergites exhibiting a yellowish colouration). On the other hand, the syntypes of S. subspinipes piceoflava also display morphological variation between each other with respect to the number of prefemoral spines on the ultimate legs: a male specimen has 4-6 spines on the prefemoral process whereas VL, M and DM spines are absent in one female specimen. The latter is similar to S. dehaani but it is possible that this absence may be due to regeneration in this individual. However, without additional mate rial and lacking molecular data with which to test relationships among morphological similar species, we tentatively accept S. subspinipes piceoflava as a junior synonym of S. subspinipes as proposed by Kronmüller (2012).
Composite description. Body length up to more than 25 cm (collections from Java, NHMW). Varied colouration; cephalic plate and segments monochromatic. Tergites usually brownishorange with or without dark band on posterior border of tergites. Cephalic plate with small punctate on anterior part; median sulcus present. Posterior part of cephalic plate without paramedian sulci.
Genital segments well developed, reaching longer than distance between posterior margin of sternite of ultimate legbearing segment and distal part of coxopleural process. Sternite of genital segment 1 round and convex posteriorly, with median suture (Figs 7A, 25A). In male, sternite of genital segment 2 attached to penis. Tergites of genital segments without small setae. Gonopods with or without small setae. Penis with apical bristle.
Colouration. This species is among those that exhibited the most varied colouration patterns. Chao (2008) suggested two major types of Scolopendra colouration, referred to as monochromatic and dichromatic. The variability in colouration of S. dehaani has    been reported in Southeast Asia over the past century (Flower 1901). In the present study, the colour patterns of growth stages have been photographed (Fig. 19). Recently, ontogenetic variation including specific changes in colouration has been discussed with reference to geographical distribution (Siriwut et al. 2015a). These findings indicated that the colouration change might involve a heritable component among populations.
A descriptive classification of colouration in S. dehaani is as follows: Colour morph 1: Dichromatic. All segments including cephalic plate dark brownish.
Posterior border of tergites with a dark band. Antenna reddish or orange. Pleuron with grey integument, pleurites orange. All legs reddish.
Discussion. This is the largest centipede in Asia. A consistent character that is treated as diagnostic for this species is the absence of ventral spines on the ultimate leg prefemur. Scolopendra dehaani possesses characters of the S. subspinipes s.l. sensu Lewis (2010) but after morphological survey (Kronmüller 2012) and molecular delimitation (Siriwut et al. 2015b) a species rank has been conferred to this name. The morphologi cal variability in the extent of paramedian sutures recorded in some specimens might demand further examination. Three morphotypes of S. dehaani within the examined collections can de delimited as follow: Morphotype 1: Complete paramedian sutures on tergites and sternites. This is the typical form of S. dehaani, according to Attems (1930b) and Jangi and Dass (1984), observed throughout the geographic range of the species. Morphotype 2: Paramedian sutures complete on tergites, confined to 20-30% length of sternites. This morphotype is observed in a specimen apparently from Bangladesh.

Morphotype 3:
Paramedian sutures lacking on tergites on all segments (Fig. 24D) and confined to 10-20% length of sternites (Fig. 23F). This morphotype has been observed in spec. from Hong Kong and northern India and is common in Java.
Based on these morphotypes, we surmise that this variability might have a ge ographic basis and could suggest evidence of cryptic speciation. Molecular data are presently lacking for morphotype 3 in particular and, presently, we apply the specific name S. dehaani Brandt, 1840, throughout the entire geographic range. A morpho logical comparison through the species' geographical range is given in Table 7. In addi tion, we recorded some brooding and feeding behaviour of individuals in their natural habitat. Scolopendra dehaani exhibited double coiling when guarding offspring in the brood chamber (Fig. 20A), similar to S. morsitans (Fig. 20B). Two stages of feeding behaviour have been observed, which may be described as late foraging (Fig. 20D) and consumption stages (Fig. 20C). The centipede attached itself to the posterior part of the body of the snaileating snake Pareas carinatus Boie, 1828 and then advanced to the anterior part of the prey's body, stabbing the snake several times. The posterior part of the centipede hung twisted with a palm trunk by using the locomotory and ultimate legs. The period of consumption of the prey lasted approximately one to two hours after the initial recording; all somatic tissue was completely eaten by the centipede.
One new subjective synonym is proposed for S. dehaani. The holotype and sole known specimen of S. arborea Lewis, 1982(NHMUK 1952) is approximately 40 mm in length, from an elevation of 2,000 ft. at Koyan Forest, Sarawak. It is here regarded as an immature specimen of S. dehaani. Lewis (1982Lewis ( , 2010 noted that mor phological characteristics of S. arborea are similar to some other Asian and one Pacific island species, namely S. dehaani, S. puensis Jangi & Dass, 1984, S. gracillima, S. hardwickei Newport, 1844, and S. metuenda Pocock, 1895. In its original description, S. arborea was noted to have a colouration pattern similar to juveniles of S. dehaani, i.e., the cephalic plate and last two segments black, the other tergites bright orange, the legs bright blue with blackish lateral lines. Its taxonomic characters appear very close to S. dehaani, the apparent differences being the number of spines on the coxopleural process (one in S. arborea versus two in S. dehaani) and tergite margination starting from T20 or only T21 margination (reaching more forward to anterior segments in S. dehaani). According to our study, the number of apical spines on the coxopleural process of adults of S. dehaani is quite strictly two, but on some occasions a single minute apical spine is present in small subadult and juvenile specimens. Morover, our reexamination of the holotype of S. arborea indicated that tergite margination starts from T15, which decreases the distinction from S. dehaani. An unusual morphologi cal feature of S. arborea is the expansion of the peritrema on the spiracle of segment 3 to cover part of the tricrescentic flaps. This atypical feature may be ontogenetic variation or the effect of muscle contraction or extension around the spiracle margin which might be affected by fixation. Without other significant diagnostic characters to distinguish this species from S. dehaani, which is abundant on the mainland and on some islands in Southeast Asia, we synonymise S. arborea with S. dehaani.  (20) 1-19 (20) 1-20    Distribution. Widespread species in the Southeast Asian mainland and some is lands (Fig. 18). The first occurrence was reported from Java, Indonesia. In this study, we provide all recorded localities of this species in Asian territory, together with some localities in which the species might be introduced, as follows: Southeast Asia: prob ably this is the native distribution area of this species according to population abun dance and genetic structure. Population density is high throughout mainland territory especially in synanthropic areas but populations are more scattered in montane areas. The currently known distribution is as follows: Thailand ( Meinert, 1886: 202. Scolopendra subspinipes multidens -Kraepelin 1903: 264. Attems 1907: 81, 1914a: 107, 1914b: 568, 1930b: 31. Muralewicz 1913: 201. Takakuwa 1942: 359, 1943: 171. Takashima 1949: 11. Takashima and Shinohara 1952: 4. Wang 1955: 16, 1956: 158, 1962: 101. Zhang 1992 fig. 1.
Composite description. Body length 11.4 cm in syntype. Dried holotype brown ish on entire body. Cephalic plate with small punctae; median sulcus present. Posterior part of cephalic plate without paramedian sulci.
Genital segments well developed, reaching longer than distance between poste rior margin of sternite of ultimate legbearing segment and distal part of coxopleural process (Fig. 28A). Sternite of genital segment 1 round and convex posteriorly, with median suture (Fig. 28B). In male, sternite of genital segment 2 welldeveloped. Ter gite of genital segment without small setae. Gonopods and penis absent (Fig. 28A-B).
Colouration. According to Chao (2008), juvenile specimens have a reddish orange cephalic plate and T1. Basal part of antenna reddish orange, distal part greenish. The re maining tergites dark green. All legs reddish orange. In adult, all tergites reddish orange.
Discussion. Morphological characters are similar to S. subspinipes sensu Chao, 2008. The validity of S. multidens as a separate species was reestablished by the ab sence of gonopods on the first genital segment in males (Chao 2008). Two species of Scolopendra in the region, S. hainanum Kronmüller, 2012 from Hainan Island, China, Figure 28. Genital segment(s) of A-B Scolopendra multidens (male; lateral and ventral views, respec tively) C-D Scolopendra cataracta (female; lateral and ventral views, respectively). and S. multidens, distributed in eastern coastal Asia, have been reported to lack gonop ods. Study of the type specimens of both S. multidens and S. dawydoffi indicates a close relationship between these two species. The lack of gonopods in males of S. dawydoffi from Thailand was found in the present study, which might provide a synapomorphic character for a clade composed of these species. However, the species boundaries be tween these taxa are complicated by disjunct distributional data, with previous records indicating that S. multidens mostly occurs in the East China Sea and possibly ranges as far as Japan. Vahtera et al. (2013) reported the occurrence of S. multidens from northern Vietnam and provided DNA sequences for a specimen. In this present study, we analyzed molecular data for numerous specimens of S. dawydoffi as well as the Vi etnamese specimen of S. multidens to explore their relationships. The phylogenetic tree based on mitochondrial and nuclear genes indicated that S. multidens is sister taxon to S. dawydoffi and both are genetically distinct from other members of the S. subspinipes group (Fig. 1). Based on to these results, an absence of gonopods is corroborated as a synapomorphy for this clade. Only geographical distribution and molecular data (i.e., branch length) can be used to distinguish these two species. COI is the only molecular marker available for S. multidens from East Asia (unpublished results from sequences in GenBank from Taiwan), and analyses of our COI data with these included groups the Vietnamese specimen together with other S. multidens. Because of their genetic distinctness and reciprocal monophyly, we regard S. multidens and S. dawydoffi as valid species until further morphological examination throughout their distribution ranges has been done to clarify species boundaries.

Scolopendra calcarata
Composite description. Body length up to 5.3 cm. Blackish colouration on most of dorsal part of body. Cephalic plate dichromatic; anterior part of cephalic plate dark blue or black, posterior margin green yellowish (Fig. 30B). Antenna dark blue. Ter gites dark blue or nearly blackish. All legs light blue, their basal part yellowish. Cephal ic plate with median sulcus on anterior part. Posterior part of cephalic plate without paramedian sulci.
Genital segments well developed, reaching longer than distance between posterior margin of sternite of ultimate legbearing segment and distal part of coxopleural pro cess. Sternite of genital segment 1 round and convex posteriorly, with median suture (Fig. 25B). In male, sternite of genital segment 2 well developed. Tergites of genital segments with small setae.
Discussion. This montane species was sometimes collected together with other scolopendrids such as species of Otostigmus and Rhysida. External phenotypic charac ters are similar to S. pinguis but the unique, diagnostic character that permits species identification is the presence of a tarsal spur on the ultimate legs, which is atypical for Scolopendra. However, this character has been reported in some individuals of a few other Scolopendra species in Southeast Asia, notably S. subcrustalis (see Kronmüller 2009, Lewis 2010b. We also recorded the occurrence of a tarsal spur on the ultimate leg in two juveniles of S. subspinipes from Yokohama, Japan (NHMW 758). For this reason, variation in tarsal spurs on legs needs to be used cautiously for justification of species boundaries when sample size is limited. However, S. calcarata is readily distin guished morphologically from S. subspinipes s.l. and S. subcrustalis by the number of glabrous antennal articles (four glabrous dorsally), only the tergite of the ultimate leg bearing segment showing margination, sternites with incomplete paramedian sutures, and 4-5 apical and subapical spines on the coxopleural process. All specimens from Thailand and the description of Vietnamese populations by Schileyko (1995) consist ently exhibited a tarsal spur on the ultimate legs. However, some characteristics seem to be variable between these two populations, such as a count of six glabrous antennal articles in Vietnam versus four in Thailand (four in the original description), the last    4-6 tergites marginated versus only the tergite of the ultimate legbearing segment (but in original description, margination starting from TT12 (13)), the sternite of the ultimate legbearing segment with a median depression versus its absence, and the ar rangement of spines on the ultimate leg prefemur. With respect to the latter, Vietnam ese populations exhibited 9-12 VL, 11-12 VM, 2-3 both M and DM, and 2 spines on the prefemoral process versus 4-7 VL, 0-3 VM, 1-2 M, 1-2 DM, and 2-4 spines on the prefemoral process in Thai populations.
Morphological similarity between S. calcarata and S. pinguis is indicated by several characteristics, including the number of antennal articles, the shape of teeth on the forcipular toothplates (these being in the form of minute denticles), and the number of spines on the coxopleural process. In addition, the habitat preferences of these two species resemble each other, both of them being found only in montane territory, and they also show similar dichromatic colouration patterns. There is no evidence from our survey that these two species are distributed sympatrically. These characters are consist ent with the molecular phylogeny, which resolves these two species as sister taxa.
Composite description. Body length up to 12.9 cm. Two colour morphs; morph 1 with antenna and legs 1-20 yellowish, morph 2 with antenna and legs 1-20 reddish. All tergites greenish brown. Cephalic plate with median sulcus. Paramedian sulci or sutures absent on posterior part.
Coxopleural process moderately long, usually with three apical spines (Fig. 39A); porefree area extending 70-90% length from distal part of coxopleural process to margin of sternite of ultimate legbearing segment.
Genital segments well developed, reaching longer than distance between posterior margin of sternite of ultimate legbearing segment and distal part of coxopleural pro cess. Sternite of genital segment 1 round and convex posteriorly, with median suture (Fig. 7D). Tergites of genital segment without small setae. Gonopod present in male.
Discussion. The validity of Scolopendra japonica at the species level was defended by Kronmüller (2012) from morphological surveys of former subspecies of S. subspinipes. Diagnostic characters of this species by comparison to the other former subspe cies are the number of apical spines on the coxopleural process (S. subspinipes s.str. having two apical spines versus S. japonica having three spines, including a subapical spine), and the number of ventral spines on the prefemur of the ultimate legs. With respect to the latter, S. japonica has three spines whereas S. subspinipes s.str. has only two, although in this study an asymmetrical number of spines in S. japonica was also found in syntypes (Fig. 39C-F). Four specimens among the japonica syntypes have been recorded as presenting only two or three ventral spines on the prefemur of the ultimate leg either on only one side or on both (six spines is the observed maximum). This variability might represent ontogenetic variation more so than geographical vari ation. Chao (2008) cited the lack of a tarsal spur on leg 20 as an additional diagnostic character for this species but from our study one specimen of S. japonica from the Izu Peninsula, Japan (NHMUK 1912.12.12.914), also exhibits a tarsal spur on leg 20. As such, the occurrence of a spur on leg 20 and number of ventral spines on the ultimate leg prefemur are not completely reliable for diagnosing S. japonica.
The sympatric distribution of this species and former subspecies of S. subspinipes as well as other Asian temperate Scolopendra complicates morphological delimitation of species boundaries except using the three phenotypic characters discussed above. In this paper, we compared taxonomic characters based on collections in the NHMUK and NHMW (Table 8). Additional characters of these two species that might be use ful for species identification are the proportions of the ultimate leg podomeres and the number of spines on the prefemoral process on the ultimate leg. In S. subspinipes s.str., two spines are usually present on the prefemoral process whereas in S. japonica there are typically three. The length of the antenna also permits a distinction berween these two species; the antenna extends backwards only as far as TT2-3 in S. japonica whereas it can reach to 14B and 36A). Moreover, molecular analysis of three combined genes (COI, 16S and 28S) indicated a genetic distinction between S. japonica and S. subspinipes (Fig. 1), and the validity of these species has been corroborated herein. Ratios of ultimate leg podomeres have been      used as diagnostic characters of some putative species of Scolopendra in Asia, such as S. negrocapitis Zhang and Wang, 1999 from Jingshan (northeast coastal area of China). The authors mentioned the close similarity between that species and S. japonica but the Chinese species can be distinguished from the latter only by the width:length ratio of the ultimate leg prefemur, which is twice as long as broad. The lack of further informa tion from fresh material from the type locality and molecular data from S. negrocapitis renders the status of these two closely related species questionable.
In the current phylogenetic framework of Scolopendra, S. japonica is resolved in the same clade as S. cingulata Latreille, 1829. The two species are morphologically similar despite their markedly disjunct distributions, i.e., S. cingulata in the Mediterranean versus S. japonica in East Asia (Table 8). However, exploration of microrefugia of populations of S. cingulata during glacial maxima in Europe (Simaiakis et al. 2012, Oeyen et al. 2014) and a record of S. japonica in the northern part of Laos could indi cate that these two species may be more widespread than previously recognised. How ever, distributional data for S. japonica are patchy due to incomplete faunistic surveys in several parts in Asia. For these reasons, the relationship between these two species warrants further scrutiny in both morphological and molecular studies.
Colouration. According to Siriwut et al. (2015a), S. pinguis exhibits four colour morphs which can classified as either monochromatic or dichromatic ( Fig. 42A-B). All of these patterns are specific to populations, with no mixing or sympatry found in our surveys. The full description of each colour morph is given below:   Discussion. Scolopendra pinguis has not been revised since Pocock (1891) described the holotype from Burma. Two additional records from BataviaBuitenzorg (Bogor, Java) expanded its geographical distribution across Southeast Asia (Pocock 1894). Kraepelin (1903) confirmed additional material from Buitenzorg; Bogor, Java. Attems' (1930b) monograph followed Kraepelin's description and argued that this species is similar to another Javan species, S. gracillima. This argument has been followed in sev eral subsequent taxonomic reviews (Schileyko 1995, Lewis 2010b. Comparative taxonomic characters of these two species (Table 9) can, in the present state of knowl edge, be used to defend species validity until further data (e.g., molecular data for both species from Java) can be considered. Another related species from northwestern India, Scolopendra ellorensis Jangi & Dass, 1984, was also noted to be morphologically similar to S. pinguis. However, this Indian species was decribed from one juvenile specimen (31 mm) that might not permit confident comparison (Lewis 2010b). Molecular phyloge netic analysis of S. pinguis revealed a high level of genetic divergence among populations that might suggest regional endemism and the possibility of cryptic species. The latter would be consistent with the marked degree of colour polymorphism noted above.
Composite description. Body length up to 16.2 cm (14.7 and 15.1 cm in syn types). Reddish colouration on entire body. Cephalic plate and tergites dichromatic. Cephalic plate and tergites reddish orange; posterior border of tergites with dark band. Cephalic plate with small punctae; median sulcus present. Posterior part of cephalic plate without paramedian sulci.
Coxopleural process moderately long or short with two apical spines and one subapi cal spine (atypically only two apical spines; Fig. 53F); porefree area extending 65-90% length from distal part of coxopleural process to margin of sternite of ultimate legbear ing segment (Figs 49D, 53D).
All legs without setae and tibial spur. One tarsal spur on legs 1-19. Ultimate legs: thick and moderately long (Figs 49C, 53D), with ratios of lengths of prefemur        (Figs 49G, 53B). Posterior margin of prefemur with shallow median groove Genital segments well developed, reaching longer than distance between posterior margin of sternite of ultimate legbearing segment and distal part of coxopleural pro cess. Sternite of genital segment 1 round and convex posteriorly, with median suture (Fig. 28C-D). Sternite of genital segment 2 developed. Gonopod absent in male. Lamina subanalis between genitalia and anal valve; lamina analis between anal valve and tergite of genital segment. Tergite and sternite of genital segments with small se tae. Penis with apical bristle.
Discussion. This species is distinguished from S. subspinipes by its short, robust ul timate legs and three apical/subapical spines on the coxopleural process. The character istic of incomplete paramedian sutures on the sternites further distinguishes it from S. subspinipes and S. japonica (which have complete paramedian sutures on the sternites). However, S. dawydoffi is similar to S. multidens in the absence of gonopods in the male. The distribution of S. dawydoffi is restricted to mainland Southeast Asia whereas S. multidens occurs in temperate regions of Asia, including both inland and insular parts. A specimen identified as S. multidens from Vietnam is genetically differentiated from Thai populations (see discussion of S. multidens above for molecular arguments in favour of the two taxa being separate species). Moreover, to test the hypothesis that characteristics of S. dawydoffi might indicate affinities to the cingulata group (with reference to the Mediterranean species S. cingulata; Attems 1930a), as implied by the original "cingulatoides" name for S. dawydoffi, our phylogenetic analysis included S. cingulata sequences from Spain. The result (Fig. 1) demonstrated that S. dawydoffi was not grouped together with S. cingulata but should be recognized as a distinct species based on its genetic distance and geographical distribution. A morphological compari son between these two species is presented in Table 10.
Etymology. From "cataract", meaning waterfall, for the type locality at Tad Etu Waterfall.
Holotype description (variation of paratypes is given in parentheses). Body length 12.8 cm (up to ca. 20 cm long in paratype NHMUK 1928.v.30.6). Blackish colouration on entire body. Cephalic plate and segments monochromatic. Tergites dark greenish or black. Cephalic plate without small punctae on anterior part; median sulcus present. Posterior part of cephalic plate without paramedian sulci.
Genital segments well developed (Figs 28C-D, 54C), reaching longer than distance between posterior margin of sternite of ultimate legbearing segment and distal part of coxopleural process. Sternite of genital segment 1 round and convex posteriorly, with median suture. Tergites of genital segment lacking small setae. Presence of gonopods and penis uncertain because genitalia are mostly retracted in holotype and paratypes; only two female specimens show genital segments (NHMUK 010305528 and 1928.5.30.6).
Remarks. The paratype collected in Thailand in 2001 (NHMUK 010305528) was observed to display apparent amphibious habits. The following account is based on observations by G. Beccaloni (pers. comm., Jan. 2016). The centipede was initially observed under a rock slab beside a stream ca. 1.5 m wide and 20 cm deep. It escaped into the stream and concealed itself under a rock. After extraction from the stream it was placed in a glass container of water, in which it swam powerfully on the bottom of the container with vigorous horizontal undulating motions.
Discussion. This species exhibits an atypical characteristic for scolopendromorphs, namely incomplete paramedian sutures on the tergites. Very few members in only three genera share this character, these being within Scolopocryptops Newport, 1844, Scolopendra and Rhysida Wood, 1862. Within Scolopendra, only two described spe cies, S. hainanum from Hainan Island, China, and "S. subspinipes piceoflava" (treated above as a synonym of S. subspinipes following Kronmüller (2012)), from Sulawesi, Indonesia, have been reported to lack or have nearly absent paramedian sutures on the tergites. Scolopendra cataracta and "S. subspinipes piceoflava" can be distinguished from each other by: paramedian sutures confined to the posterior part of tergites in S. cataracta vs complete sutures in "S. subspinipes piceoflava", very short paramedian sutures on the sternites vs sutures extending along 80-100% the length of the sternites, and three apical/subapical spines on the coxopleural process vs one or two.
S. cataracta differs from S. hainanum by the short paramedian sutures on the ster nites versus being nearly complete in S. hainanum, the number of spines on the coxo pleural process (three vs one or two), and the number of VL spines on the ultimate leg prefemora. The two species can also be distinguished by their colouration patterns and their distributions, though the latter are closely associated. For these reasons, we regard S. cataracta as distinct from S. hainanum, and its sampled populations group as mono phyletic for each partial gene analysis (see Table 11 for morphological comparison). It is likewise morphologically distinct from S. subspinipes, S. multidens and S. dawydoffi. DNA sequences are not available for S. hainanum but would be useful to study the relationship between these two apparently related species. Distribution. All localities are in mainland territory. The currently known distri bution (Fig. 29) is as follows: Southeast Asia: Laos (Champasak and Luang Namtha), Thailand (Surat Thani) and Vietnam (Bac Kan and DacTo).