A revision of dragon millipedes I: genus Desmoxytes Chamberlin, 1923, with the description of eight new species (Diplopoda, Polydesmida, Paradoxosomatidae)

Abstract The dragon millipede genus Desmoxytes s.l. is split into five genera, based on morphological characters and preliminary molecular phylogenetic analyses. The present article includes a review of Desmoxytes s.s., while future articles will deal with Hylomus Cook and Loomis, 1924 and three new genera which preliminarily are referred to as the ‘acantherpestes’, ‘gigas’, and ‘spiny’ groups. Diagnostic morphological characters of each group are discussed. Hylomus is resurrected as a valid genus and the following 33 species are assigned to it: H. asper (Attems, 1937), comb. n., H. cattienensis (Nguyen, Golovatch & Anichkin, 2005), comb. n., H. cervarius (Attems, 1953), comb. n., H. cornutus (Zhang & Li, 1982), comb. n., H. draco Cook & Loomis, 1924, stat. rev., H. enghoffi (Nguyen, Golovatch & Anichkin, 2005), comb. n., H. eupterygotus (Golovatch, Li, Liu & Geoffroy, 2012), comb. n., H. getuhensis (Liu, Golovatch & Tian, 2014), comb. n., H. grandis (Golovatch, VandenSpiegel & Semenyuk, 2016), comb. n., H. hostilis (Golovatch & Enghoff, 1994), comb. n., H. jeekeli (Golovatch & Enghoff, 1994), comb. n., H. lingulatus (Liu, Golovatch & Tian, 2014), comb. n., H. laticollis (Liu, Golovatch & Tian, 2016), comb. n., H. longispinus (Loksa, 1960), comb. n., H. lui (Golovatch, Li, Liu & Geoffroy, 2012), comb. n., H. minutuberculus (Zhang, 1986), comb. n., H. nodulosus (Liu, Golovatch & Tian, 2014), comb. n., H. parvulus (Liu, Golovatch & Tian, 2014), comb. n., H. phasmoides (Liu, Golovatch & Tian, 2016), comb. n., H. pilosus (Attems, 1937), comb. n., H. proximus (Nguyen, Golovatch & Anichkin, 2005), comb. n., H. rhinoceros (Likhitrakarn, Golovatch & Panha, 2015), comb. n., H. rhinoparvus (Likhitrakarn, Golovatch & Panha, 2015), comb. n., H. scolopendroides (Golovatch, Geoffroy & Mauriès, 2010), comb. n., H. scutigeroides (Golovatch, Geoffroy & Mauriès, 2010), comb. n., H. similis (Liu, Golovatch & Tian, 2016), comb. n., H. simplex (Golovatch, VandenSpiegel & Semenyuk, 2016), comb. n., H. simplipodus (Liu, Golovatch & Tian, 2016), comb. n., H. specialis (Nguyen, Golovatch & Anichkin, 2005), comb. n., H. spectabilis (Attems, 1937), comb. n., H. spinitergus (Liu, Golovatch & Tian, 2016), comb. n., H. spinissimus (Golovatch, Li, Liu & Geoffroy, 2012), comb. n. and H. variabilis (Liu, Golovatch & Tian, 2016), comb. n. Desmoxytes s.s. includes the following species: D. breviverpa Srisonchai, Enghoff & Panha, 2016; D. cervina (Pocock,1895); D. delfae (Jeekel, 1964); D. des Srisonchai, Enghoff & Panha, 2016; D. pinnasquali Srisonchai, Enghoff & Panha, 2016; D. planata (Pocock, 1895); D. purpurosea Enghoff, Sutcharit & Panha, 2007; D. takensis Srisonchai, Enghoff & Panha, 2016; D. taurina (Pocock, 1895); D. terae (Jeekel, 1964), all of which are re-described based mainly on type material. Two new synonyms are proposed: Desmoxytes pterygota Golovatch & Enghoff, 1994, syn. n. (= Desmoxytes cervina (Pocock, 1895)), Desmoxytes rubra Golovatch & Enghoff, 1994, syn. n. (= Desmoxytes delfae (Jeekel, 1964)). Six new species are described from Thailand: D. aurata Srisonchai, Enghoff & Panha, sp. n., D. corythosaurus Srisonchai, Enghoff & Panha, sp. n., D. euros Srisonchai, Enghoff & Panha, sp. n., D. flabella Srisonchai, Enghoff & Panha, sp. n., D. golovatchi Srisonchai, Enghoff & Panha, sp. n., D. octoconigera Srisonchai, Enghoff & Panha, sp. n., as well as one from Malaysia: D. perakensis Srisonchai, Enghoff & Panha, sp. n., and one from Myanmar: D. waepyanensis Srisonchai, Enghoff & Panha, sp. n. The species can mostly be easily distinguished by gonopod structure in combination with other external characters; some cases of particularly similar congeners are discussed. All species of Desmoxytes s.s. seem to be endemic to continental Southeast Asia (except the ‘tramp’ species D. planata). Some biological observations (relationship with mites, moulting) are recorded for the first time. Complete illustrations of external morphological characters, an identification key, and distribution maps of all species are provided.

Intensive field surveys focusing on this genus were made by our team (ASRU), mainly in Southeast Asia (Malaysia, Myanmar, Laos, Thailand). After examination of newly collected specimens, and comparison with type material of all congeners, we found distinctive morphological characters, mainly in gonopods and paraterga, indicating heterogeneity of Desmoxytes s.s. A preliminary study on phylogeny of dragon millipedes based on mtDNA and nuclear DNA shows a perfect congruence with morphology  and further indicates that Desmoxytes as hitherto understood (i. e., sensu Golovatch and Enghoff 1994) is not a monophyletic taxon. Therefore, we find it necessary to subdivide the dragon millipedes into five genera. In the present study, the first in a series of articles about dragon millipedes, Desmoxytes is re-diagnosed, and the four other genera (Hylomus Cook & Loomis 1924, and three new genera yet to be named) are outlined. Ten species of Desmoxytes in the new, restricted sense are revised, eight new species are described, a new identification key to Desmoxytes species is provided, and 33 Desmoxytes species are assigned to the reinstated genus Hylomus.

Specimen collecting and preservation
The specimens were collected by hand from different localities in China, Laos, Malaysia, Myanmar, and Thailand during the rainy season. The GPS coordinates were recorded by using the Garmin GPSMAP 60CSx and the elevation was obtained by checking in Google Earth.
Intensive surveys in several parts of those countries, especially in Thailand, have been made since 2007 by staff and students from Animal Systematics Research Unit, Department of Biology, Faculty of Science, Chulalongkorn University, referred to as "ASRU members" in the lists of material. Specimens were preserved in 70% ethanol for morphological study and partly in 95% ethanol for molecular study.

Illustrations
Photos of living specimens were taken in the field and after collecting using a Nikon 700D+AFS VR 105 mm lens. The gonopods were illustrated with a scanning electron microscope (JEOL, JSM-5410 LV); gonopods were coated with gold and mounted on aluminium stubs, and after imaging the gonopod was removed from the stub to be kept in dry condition. Drawings were made using dot-line techniques under a stereo microscope. Table 1. Gonopod structures in Desmoxytes s.s., and their abbreviations (in Bold: structure only occurring in certain species).

Coxa cx
The basal part of the gonopod, connecting to body ring, attached to the apertural rim dorsally; rather stout; ca. half as long as femur, sometimes quite short (equal in length with prefemur); with distoanterior group of setae. Broad lobe of lamina medialis blm A broad lobe originating from lamina medialis, lamella-like; normally broad with thick edge.
Cannula ca A short tube, lever-like, curved, long and slender; originating from coxa, tip inserted into concavity in prefemur. Distal lobe of lamina medialis dlm A lamella-like process, situated on the top of lamina medialis, consisting one or two small lobe(s)/ lamella(e). Femur fe The longest part of the gonopod, straight; accommodates the seminal groove. Lamina medialis lm A large part distally on the gonopod, normally with one process and one/two lobe(s) Lamina lateralis ll A distinct lobe in the distal part of gonopod; sometimes comprising a ridge and/ or lobe Lateral sulcus ls A distinct sulcus distally on femur, visible in lateral view Mesal sulcus ms A distinct sulcus distally on femur, usually seen in mesal view.
Postfemur pof A short part of telopodite, supporting solenophore and solenomere, demarcated from femur by lateral sulcus and mesal sulcus. Prefemur pfe The basal portion of the telopodite, with densely setose. Process of lamina medialis plm A protruding process originating from lamina medialis Seminal groove sg A conspicuous groove, similar to a tunnel, seen as a transparent line, visible on femur in mesal view. Solenomere sl A usually long, flagella-like appendage, originating on base of solenophore. Solenophore (= tibiotarsus) sph Apical part of telopodite, consisting of lamina lateralis and lamina medialis.
Dorsal refers to a position on the side nearest to the body ring. Ventral refers to a position on the side farthest away from the body ring. Mesal refers to a position on the side nearest to the midline. Lateral refers to a position on the side away from the midline.
Dorsad refers to a direction towards the body ring. Ventrad refers to a direction away from the body ring. Mesad refers to a direction towards the midline. Laterad refers to a direction away from the midline.
"Sub-" is used as a prefix to indicate positions and directions slightly different from the ones given above. For example, "submesal" means a position close to, but not quite on the mesal side. Tribe Orthomorphini Brölemann, 1916 The tribe Orthomorphini was established by Brölemann (1916), as belonging to family Strongylosomidae. Later, the correct name for this family was found to be Paradoxosomatidae (Jeekel 1963). The tribe Orthomorphini is characterised by the following gonopodal characters:
The solenophore consists of a lamina lateralis and a lamina medialis. 4. Most species without a femoral process and a tibiotarsal process.
The tribe Orthomorphini currently contains 22 genera (Nguyen and Sierwald 2013). Around 200 named species have been recorded so far. The tribe is broadly distributed mainly in Southeast Asia (Cambodia, Indonesia, Laos, Malaysia, Myanmar, Thailand, Vietnam, a few synanthropic species more or less globally widespread).

Subdivision of Desmoxytes s.l.
It is here proposed to split Desmoxytes s.l. into five groups. These groups are morphologically distinct, as detailed below, and are all supported by a molecular phylogeny (work ongoing). The molecular phylogenetic tree yields five groups of Desmoxytes s.l., and Orthomorpha Bollman, 1893 and Antheromorpha Jeekel, 1968 appear as ingroups. Desmoxytes s.l. thus comes out as non-monophyletic .
The present paper deals with one of the five groups, viz., Desmoxytes s.s. The remaining species (see Table 2) will be assigned to their proper genus in forthcoming papers. For one of the groups, a genus name is already available, viz, Hylomus Cook & Loomis, 1924, and this name is herewith re-instated as a valid genus name.
The five groups of Desmoxytes s.l. are characterised as follows (see Table 3 for all comparable characters, and Figs 2, 3): Desmoxytes s.s. (Figs 2A, can be easily distinguished from others by the following combination of characters: wing-shaped paraterga; male femora 5 and 6 swollen or humped (except D. terae without modification); lamina lateralis obviously demarcated from lamina medialis; lamina medialis and lamina lateralis equal in size, with process and lobes.
Two groups, the 'acantherpestes' group and the 'gigas' group are characterised by having subspiniform paraterga and tegument of metaterga microgranulate.
The fifth group, corresponding to genus Hylomus and including numerous species, has wing-like, antler-like or spiniform paraterga, but is well-defined by gonopod characters: postfemur inconspicuous or absent (mesal and lateral sulci shallow or absent), lamina medialis short (except in a few species), lamina medialis mostly without lobe and process (except some species which have a process, a spine or a hook) (Figs 2E, F, 3H-P).
As Figs 2 and 3 clearly show, the differences in morphology (paraterga and gonopod) easily allow separation of the five groups of Desmoxytes s.l. species. Table 2. Species assigned to Desmoxytes s.l. and their placement according to our analysis.
General description of the genus Desmoxytes s.s. The description applies to adult males and females, except for the gonopods section and when "male" is specified (Figs 5,6,7,8). The description hereunder is mainly based on illustrations of D. planata.
COLOUR: Most species in life with aposematic colouration: purplish pink, red, orange, brown, black, brownish black (piceous), brownish red (testaceous). Colour in alcohol: all specimens partly faded after one year's preservation in alcohol; specimens kept in darkness faded more slowly.
METATERGA (Figs 5A, 6A, C, D, G): With 2 transverse rows of setae/tubercles/ cones/spines, number in each row varying between species, lateral cones/spines of posterior row often longer and larger than inner and/or mesal ones. Suture (transverse sulcus) on metaterga present, usually conspicuous on body rings 5-17 in all species. Mid-dorsal (axial) line missing. Pleurosternal carinae forming complete, tooth-liked crests on ring 2, small ridges on ring 3, missing on remaining body rings.
Distribution and habitat. From south China to Malaysia. Most species seem to be local endemics (only D. planata is dispersed, certainly by anthropochory, through mainland Southeast Asia and in many islands). Desmoxytes specimens were usually found by ASRU personnel in limestone habitats or on granitic mountains, and some were seen crawling on rocks or vegetation or tree branches ( Figure 4).
Key to species of Desmoxytes s.s.

1
Male femora 5 and 6 without modification (Fig. 90H   Diagnosis. Body bright orange, low degree of elevation of paraterga, femora 5 and 6 strongly humped ventrally in middle part, collum with row of 3+3 anterior setae and metaterga with rows of 2+2 anterior and 2+2 posterior small tubercles. Similar in these respects to D. delfae and D. perakensis sp. n., but differs from those by having paraterga of collum quite long; lateral sulcus (ls) quite shallow; lamina lateralis (ll) stout and compact, ventral ridge (vrl) short; process (plm) of lamina medialis crenate; sternal lobe between male coxae 4 thin when seen in lateral view.

Desmoxytes aurata
Etymology. The name is Latin adjective and refers to the lamina lateralis (ll) of the gonopod which bears some resemblance to the "hooded" head of the oranda breed of goldfish (Carassius auratus).
Distribution and habitat. Known only from Surat Thani and Nakhon Si Thammarat Provinces. All specimens were collected in limestone mountains (on the mainland and on two islands) (Fig. 9E). We also surveyed other islands in Mo Ko Ang Thong National Marine Park, but found no specimens of D. aurata sp. n. The new species is probably distributed along the islands in the Gulf of Thailand and also on the mainland near the type locality. We regard it as endemic for Thailand.
Desmoxytes aurata sp. n. is morphologically similar to D. delfae and D. perakensis sp. n. in the remarkable orange colouration, as well as some morphological characters      (except characters in diagnosis). These three species show allopatric distribution ranges, and the big mountain ranges known as the Nakhon Si Thammarat and Sunkala Khiri mountains possibly act as dispersal barriers.
Remarks. The bright orange colouration is without doubt aposematic. There is some distinct variation within populations in the sternal lobe between male coxae 4, especially its shape: in most specimens the lobe is subrectangular, in others subtrapeziform, and its tip also varies -subtruncate/ subemarginate/ round. The shape of the sternal lobe of the new species is similar to that seen in D. delfae and D. perakensis sp. n., however, it looks thinner than those when seen in lateral view.
COLOUR: In life with body shocking pink to purple (some female specimens brownish pink); paraterga vivid pink; metaterga and surface below paraterga brownish pink to brownish purple; head brown; antenna blackish brown (except distal part of antennomere 7 and antennomere 8 whitish); legs, sterna and epiproct pink; a few basal podomeres whitish pink. Colour in alcohol: after two years changed to pale brown.
STERNA: Cross-impressions shallow. Sternal lobe between male coxae 4 swollen, usually subrectangular (in some specimens subtrapeziform), quite long and slightly thick when seen in lateral view, tip usually emarginate (some specimens subtruncate).
Distribution and habitat. Known only from the type locality and nearby areas. Desmoxytes breviverpa was collected from limestone forest, crawling on logs and litter (Srisonchai et al. 2016). We believe it may be distributed though the central and southern parts of north Thailand. This species has been reported from Phrae, Lampang and Uttaradit Provinces. Therefore, D. breviverpa should be regarded as endemic for Thailand.
Remarks. Specimens collected from Uttaradit Province south of the type locality showed the same morphological characters as the type specimens. No variation was found between populations although most limestone areas in north Thailand are geographically isolated by several big mountain ranges and quite far from each other. Srisonchai et al. (2016) discussed variation of the tip of sternal lobe between male coxae 4; we found additional variation within populations as follows: -Tip of sternal lobe in some specimens almost truncate (=subtruncate), albeit other specimens show an emarginate tip (slightly or deeply emarginate). -Tip of hypoproct in some individuals subsemicircular, in others subtrapeziform.
-Tip of process (plm) of lamina medialis emarginate in most specimens, but almost blunt in some.
Coexisting species. None known. Srisonchai et al. (2016, pp. 99-103) wrote in the description of this species that the paraterga (including paraterga of collum) are directed dorsolaterad. In fact they are directed caudolaterad as described above.
Distribution and habitat. Desmoxytes cervina is known from Myanmar (Lenya National Park) and Thailand (Chumphon, Krabi, Nakhon Si Thammarat, Phang Nga, Phuket, Ranong, and Surat Thani Provinces). The type locality is in south of Tenasserim, but we do not know the exact location, probably somewhere near Taninthayi township. The locality of the paralectotype (see below) "Malewoon, Tenasserim", is currently known as Maliwan Village.
Almost all specimens were collected from limestone habitats, a few specimens were collected from granitic areas. Interestingly, D. cervina was also found in some islands in both the Andaman Sea and the Gulf of Thailand (Mu Koh Surin National Park, Phuket, and Ko Samui).
Decker (2010) classified "D. pterygota" as endemic for Thailand due to its narrow range near the type locality. After intensive collecting and synonymising of D. cervina and "D. pterygota", the known distribution range has expanded to southern Myanmar and southern Thailand. This species was found living together with D. delfae and D. corythosaurus sp. n. According to our observations, it is probable that they may even share microhabitat: humid rocks, branches of trees and rock walls. Note on material. This species was described based on two males, one of which was collected by L. Fea (Malewoon, Tenasserim -MSNG) and the other collected by E. W. Oates (south Tenasserim -NHMUK). Jeekel (1964) revised this species by examining the male in MSNG; he stated that "in anticipation of the designation as holotype of the specimen collected by Oates in the British Museum, I have labelled this specimen as paratype". However, the holotype has never been designated in the original description. R.L. Hoffman visited NHMUK and labelled the specimen collected by Oates as paratype, but he did not publish this designation, which is in conflict with Jeekel's (1964) paratype designation. Therefore, we here designate the male collected by Oates and belonging to NHMUK as lectotype to stabilise the name. The male in MSNG (which we have not examined) is considered to be a paralectotype.  Some specimens kept in MHNG and identified as "D. pterygota" by Decker (2010), are probably D. cervina because the localities fall within the distribution range of this species.
Remarks. This is the first report of the colour of living specimens for this species; the brownish red colour is apparently aposematic. Pocock (1895) and Golovatch and Enghoff (1994) did not mention the colour of living specimens. We found two colour morphs of D. cervina: brownish red and brown. The majority of specimens are brownish red, and the minority are brown; the latter colour can be found in a few specimens within a population.
Interestingly, brownish red and brown morphs occur in the same habitat in the valley behind Tiger Cave and Ban Song Phi Nong. The brown morph was found at Wat Tham Kanlaya Namit, Wat Tham Thong Phannara, Tham Nam Pud, Phung Chang Cave and Wat Suwan Khuha (Monkey Cave). Specimens from the remaining localities are of the brownish red morph. We examined the morphological characters of all specimens of both colour morphs; all specimens exhibit the same morphology, especially in the gonopods which are identical. Perhaps the difference in colour is caused by environmental factors and/or genetic variation.
We assume that the type material of both D. cervina, collected by Oates and Fea a hundred years ago, and "D. pterygota" collected by M. Andersen and A.R. Rasmussen 17 years ago, did probably exhibit brownish red colour because all specimens near the type localities are brownish red. We found additionally that the colour of some females is pale brownish red or pale brown, and the colour of juveniles is pale brown. Jeekel (1964) wrote in his redescription that D. cervina shows collum without setae, metaterga 2-19 with 1+1 anterior tubercles and 2+2 posterior tubercles. Golovatch and Enghoff (1994) distinguished "D. pterygota" from D. cervina by having smooth metaterga. They also described "D. pterygota" as having no pleurosternal carinae, antenna reaching to ring 4 in male, collum with 2 rows of 3+3 anterior setae and 1+1 intermediate setae, metaterga 2-19 with 2+2 anterior cones and 2+2 posterior cones. After examination of all type material and newly collected specimens of D. cervina and "D. pterygota", we found that: -all specimens of D. cervina and "D. pterygota" display fine microgranulation on the metaterga. -all specimens have pleurosternal carinae, in body ring 2 very distinct and crest-like, in ring 3 very small, thereafter absent. -antenna reaches to ring 6-7 in male and to ring 5-6 in female of both D. cervina and "D. pterygota". -collum has 3+3 anterior setae in all specimens -because Jeekel studied old preserved specimens, the setae may have been lost over time.
Several other characters show variability, as follows:

I. variation within populations
-size of tubercles on metaterga: tubercles conspicuous in some specimens, inconspicuous in the others (bigger and more obvious in the holotype of "D. pterygota" than in the lectotype of D. cervina). -tip of process (plm) of lamina medialis: in some specimens terminating in one blunt process, in others terminating in a sharp spine. -shape of sternal lobe between male coxae 4: in some individuals subtrapeziform, in others subsemicircular. -tip of sternal lobe between male coxae 4: in some specimens round, in some subtruncate, in others emarginate. -cross-impressions on sternum: in some individuals shallow and faint, in others slightly deep. -tip of epiproct: in some individuals truncate, in others slightly emarginate.
-caudal margin of hypoproct: in some specimens slightly round, in others subtruncate.
-size of sternal lobe between male coxae 4 when seen in lateral view in specimens from Wat Satit Khirirom: in some specimens thick, in others thin.

II. variation between populations
-colour: all individuals in the same population usually have the same colour: brown or brownish red. However, in some populations (valley behind Tiger Cave and Ban Song Phi Nong) brownish red and brown individuals coexist.
Although the male paralectotype of D. cervina (in MSNG) has not been examined by us, the morphological characters for this specimen as redescribed by Jeekel (1964) perfectly match the morphology of the numerous other specimens we have seen. Based on our analysis of morphology and variation of these specimens we have synonymised "D. pterygota" under D. cervina.
Distribution data support the synonymisation: the type localities of "D. pterygota" (Ranong Province) and D. cervina (Tenasserim = Taninthayi) are very close to each other. Our intensive surveys prove that this species is distributed quite widely, but nevertheless is found in south Myanmar and south Thailand only.
During the field survey, we noticed several adult males of D. cervina which were infested with red mites. The mites are probably larvae of the genus Leptus Latreille, 1796 (Prostigmata, family Erythraeidae). Associations between mites and millipedes may be of a phoretic or a parasitic nature (Gerdeman et al. 2000, Swafford and Bond 2010, Farfan and Klompen 2012, Mwabvu 2014). In Fig. 17C and D, several engorged mites are seen along with a few small, non-engorged ones, and we therefore assume that the mite species found in D. cervina is parasitic, like other Leptus larvae. Southcott (1992) described Leptus millipedius from julid millipedes, but this is the first record of a parasitic prostigmatan mite from a paradoxosomatid millipede. The only record of a mite from Paradoxosomatidae concerns the widespread Oxidus gracilis (C. L. Koch, 1847) which was reported as associated with Cosmolaelaps hortensis Ishikawa, 1986 (Mesostigmata, family Laelapidae) (Ishikawa 1986). Golovatch and Enghoff (1994) reported one broken male of "Desmoxytes sp." from Phuket Province, Thailand. According to the remarks of Golovatch and Enghoff (1994), the gonopod characters of this specimen were similar to D. delfae, but the paraterga showed a higher degree of elevation. We examined this specimen again and found it to share gonopod and other characters with D. cervina. We therefore treat "Desmoxytes sp." as D. cervina. Etymology. The name is a Latin noun in apposition, referring to the similarity of the lamina lateralis (ll) to the crest-liked structure on the head of the dinosaur genus Corythosaurus.
Distribution and habitat. This species is known only from a narrow distribution range in Phanom district. We consider D. corythosaurus sp. n. to be endemic for Surat Thani Province, Thailand.
Remarks. The shape of the sternal lobe between male coxae 4 is variable: the tip is emarginate in the populations from Wat Tham Wararam and Tham Nam Lod, whereas specimens from the type locality have a truncate tip. The new species shares a similar shape of gonopodal solenophore with D. terae.
This species is almost impossible to find at first glance. We collected all specimens that were found on the humid rock walls by using flashlight. It blended in perfectly with the brown or black rock, this way probably avoiding being detected by predators.
Coexisting species. Desmoxytes cervina (brown morph) at Ban Song Phi Nong, was collected from rock walls, same habitat as the new species.
The redescription hereunder is modified from Jeekel (1964), and Golovatch and Enghoff (1994). We 'harmonised' descriptions of all morphological characters and added some morphological characteristics from additional collected specimens.

Figure 33. Desmoxytes delfae
Distribution and habitat. Desmoxytes delfae is known from several provinces in southern Thailand. All new specimens were collected by us from limestone habitats (the recorded locations in previous papers are also limestone); most specimens were seen crawling on leaf litter and climbing on branches of trees (Fig. 30D).  The type locality, Bukit Besar, may be part of what is now Thale Ban National Park in Khuan Don District, Satun Province, and refers to the big mountain which is probably Khao Chin (ca. 2500 ft. or 756 m). Specimens collected by us from Thale Ban National Park may thus be topotypes. The other paratype locality (Bukit Besar, Nawnchila) has not been exactly located, but it is probably near Thale Ban National Park (Thailand-Malaysia border), possibly close to the type locality.
We assume that D. delfae is distributed mainly in Thailand and possibly south to Malaysia near the Thailand-Malaysia border.
This species was reported as endemic for Thailand (Decker 2010), and we would agree that it should be regarded as endemic for the southern part of Thailand due to its narrow distribution. However, two males of D. delfae were reported from Khaosok National Park in west of Surat Thani Province by Decker (2010); it is probable that the specimens from this location belong to another species because all the specimens of D. delfae which we have seen are from an area in south Thailand (Krabi, east of Surat Thaini, Nakhon Si Thammarat, Phatthalung, Satun, Songkhla, and Yala Provinces). Furthermore, Decker (2010) also identified one male, which was collected from Nakhon Si Thammarat Province (Lan Saka District, Khao Luang National Park, near Karom Waterfall) as D. rubra. Re-examination of those specimens in MHNG is necessary to evaluate the distribution of this species.
Remarks. The remarkable and vivid bright orange colouration is clearly aposematic. Golovatch and Enghoff (1994) distinguished D. rubra from D. delfae based on D. rubra having 1+1 anterior and 2+2 posterior metatergal tubercles, the mid-dorsal (axial) line traceable, the sternal lobe between male coxae 4 roundly subtriangular and lamina medialis (lm) more strongly produced mesally than dorsally. After examination of all type material and new specimens of both D. delfae and "D. rubra" collected by us, we found that: -There is a variation in the size of tubercles on metaterga (metaterga with two rows of 2+2 anterior and 2+2 posterior tubercles, lateral tubercles of anterior row in some specimens (tiny and very inconspicuous). -All specimens are without mid-dorsal line.
-The sternal lobe between male coxae 4 is highly variable in shape, even within populations, as round/ subtrapeziform/ subrectangular; we found this variation in both D. delfae and "D. rubra" specimens. Its tip also varies as subtruncate/ round/ emarginate. -SEM images clearly show that specimens of these two nominal species have identical gonopods, especially in details of lamina lateralis and lamina medialis.
Due to this variation, we have synonymised D. rubra under D. delfae. Jeekel (1964) described this species as lacking a tiny denticle near the tip of paraterga on the lateral margin of rings 9, 10, 12, 13, 15-18, and collum as having 2 rows of 3+3 conspicuous setae (anterior row) and 1?+1? inconspicuous tubercles (posterior row). All specimens studied by us have a tiny denticle near the tip (conspicuous in some specimens, inconspicuous in others), and we regard this character as variable within populations.
For D. rubra, Golovatch and Enghoff (1994) described the colour of alcoholpreserved specimens as bright pinkish, that of living specimens as bright red, collum with 3 rows of setae (anterior conspicuous, intermediate and posterior inconspicuous), mid-dorsal line traceable. Based on the re-examination of type material of D. rubra and examination of newly collected specimens we have found that D. rubra (= D. delfae) exhibits: -Specimens in life with bright orange colouration, newly moulted adult stage pinkish or pinkish orange, late adult stage reddish orange or dark orange. As Golovatch and Enghoff (1994) reported that living specimens have a bright red colour, it is possible that the type specimens of D. rubra were collected at late adult stage (red = reddish orange?). -Collum with one row of setae (3+3 anterior setae), intermediate and posterior rows absent. Therefore, we here report collum with only one row of setae (3+3 anterior setae). -All specimens without mid-dorsal line.
As we mentioned above, this species shows high variability in morphology, e.g., colour, rows of setae on collum, size of metatergal tubercles, occurrence of a tiny denticle near tip of paraterga, shape of sternal lobe between male coxae 4. All variations are typically present within a population. Although there are deviations in several morphological characters, interestingly, gonopod characters of all specimens are quite stable, looking exactly the same in details.
Coexisting species. Desmoxytes cervina in several places, D. terae at Tham Tone Din, D. flabella sp. n. at Tham Khao Ting and Tham Khan Ti Phol. Diagnosis. Differs from all other Desmoxytes species by the combination of the following characters: paraterga knife-like; lateral sulcus (ls) of gonopod shallow; lamina lateralis (ll) separated into two ridges by a deep and wide furrow; process (plm) of lamina medialis long and thin, lamellar, tip dentate or crenate; distal lobe (dlm) of lamina medialis quite long, tip directed ventroanteriad; caudal margin of hypoproct concave or truncate.
Distribution and habitat. Known only from the type locality and nearby areas. This species seems to be rare because we made intensive surveys again in 2015 and 2016, no further specimens were found. As mentioned by Srisonchai et al. (2016), D. des is known only from two localities (Doi Angkhang and Wat Tham Krab), and we consider it to be endemic to Thailand.
Remarks. This species exhibits some variation in the gonopods: the tip of process (plm) of lamina lateralis is dentate in some specimens, crenate in others. Desmoxytes des is easy to discriminate from other dragon millipedes by the distinct shape of paraterga and unique gonopod characters.
Etymology. "Euros" (noun in apposition) is the name of the ancient Greek god of the east wind; the name refers to the occurrence of this species in the eastern part of Thailand.
Distribution and habitat. Known only from east Thailand (Chantaburi, Chonburi, Sa Kaeo and Rayong Provinces). Interestingly, this species exists in both limestone areas and granitic mountains. It was seen crawling on rocks during the rainy season and occurs only in natural habitat inside primary forest. On the basis of current data, the distribution area is quite narrow, and the species seems to be restricted to the eastern part of Thailand. We thus regard D. euros sp. n. to be an endemic for the Thai fauna.
Remarks. The observation of all living specimens reveals variation on colour of paraterga within a population; yellow in some specimens, yellowish orange to orange in others.
Desmoxytes euros sp. n. strongly resembles D. planata in several morphological characters (except for the characters mentioned in the diagnosis); notably the gonopod characters are identical. However, our initial study on DNA barcoding gene (COI) revealed that D. planata and the new species are separated enough to support the suggestion that D. planata and the new species are indeed different species (paper in preparation).
Because of the similarity in gonopod morphology, it is difficult to discriminate old material of D. planata and D. euros sp. n. However, they can be distinguished by colour of paraterga (yellow or yellow orange in D. euros sp. n., pink in D. planata), and by characters of hypoproct (subtriangular in D. euros sp. n., subtrapeziform in D. planata).
We kept several adults in an acrylic box with litter at room temperature. Two weeks later, we found a nest with eggs and a cluster of stadium 1 juveniles at ca. 2 cm depth in the soil and leaf litter (Fig. 37E, F).

Desmoxytes flabella
Etymology. The name is a Latin noun, referring to the shape of process (plm) and the distal lobe (dlm) on lamina medialis which somewhat resemble a handheld fan or flyswatter.
Distribution and habitat. Known only from the type locality and a few nearby localities. The new species is restricted to limestone habitats (Fig. 43D), and it is sympatric with D. delfae at Tham Khao Ting and Tham Khan Ti Phol. Unlike the bright orange D. delfae, which was easily spotted crawling on branches of shrubs and on rocks, the new species, blackish brown, was found on rocks where it was quite hard to see. This suggests that these two species, although sharing the same habitat, may show microhabitat differences, but this has not yet been studied in detail. We assume that the new species is distributed along limestone mountain ranges in a narrow area at the  border between Trang and Satun Provinces. This species should be regarded as endemic Description. SIZE: Length 27-31 mm (male), 32 mm (female); width of midbody metazona ca. 1.9 mm (male), 2.3 mm (female). Width of head < collum < body ring 2 = 3 = 4 < 5-16, thereafter body gradually tapering towards telson. COLOUR (Fig. 49A-C): In life with body vivid pink; paraterga, metaterga, surface below paraterga and epiproct pink; head brown; antenna black, except distal part of antennomere 7 and antennomere 8 whitish; sterna and legs pinkish brown; a few basal podomeres brown to whitish. ANTENNAE (Fig. 50D): Very long and slender, reaching to body ring 7 (male) and 5 (female) when stretched dorsally. COLLUM (Fig. 50A) GONOPODS (Figs 53,54): Coxa (cx) longer than prefemur. Cannula (ca) slender. Prefemur (pfe) ca. 2/3 as long as femur. Femur (fe) long, slightly curved. Mesal sulcus (ms) and lateral sulcus (ls) conspicuous, deep. Postfemur (pof ) conspicuous, ventrally wide. Solenophore (sph) well-developed: lamina lateralis (ll) round and compact (obvious when seen in ventral view): lamina medialis (lm) well-developed; process (plm) somewhat short, tip terminating in a sharp spine (in some specimens with a tiny spine-like process situated between process (plm) and distal lobe (dlm)); distal lobe (dlm) quite long, distally with two indistinctly separated lamellae; broad lobe (blm) dorsally expanded, indentation between broad lobe (blm) and distal love (dlm) very wide and shallow. Solenomere (sl) relatively long.    Distribution and habitat. Known from the type locality and nearby areas in Kanchanaburi Province only. The type locality is situated on a small, isolated limestone mountain near Khwae Noi River. All specimens were found in limestone habitats (Fig. 49D).
This species is distributed along the limestone mountain ranges in Sai Yok and Thong Pha Phum districts. Based on many intensive surveys, the current distribution of the new species is evidently quite narrow, ca. 100 km 2 . Thus, D. golovatchi sp. n. should be regarded as a Thai endemic.
Remarks. Desmoxytes golovatchi sp. n. is aposematic in its vivid pink body. During the field trips, this animal was noticeable by the contrast of its bright colour to green leaves or brown rocks, it thus was easy to see and collect after rain.
This species is morphologically similar to D. breviverpa, D. purpurosea, D. takensis, and D. waepyanensis sp. n. with which it shares colourful pink or red body colour, as well as further characters, viz., the same patterns of row of cones on metaterga (metaterga 2-8 with 2+2 cones in anterior row).
Coexisting species. The new species was found in one place together with D. octoconigera sp. n. (see detail in D. octoconigera sp. n.), with D. planata at Wat Huay Charoen Srattha Tham and Tham Khao Noi Bureau of Monks, and with D. purpurosea at Daowadueng Cave. This species and D. purpurosea were hand-collected after rain  when lots of them were climbing on vegetation and limestone rocks. Microhabitat differences have not yet been observed. Moreover, D. planata also occurs near the new species, but the habitats of these two species are clearly different: D. planata was found on humid cement and on construction materials whereas D. golovatchi sp. n. was seen crawling on limestone rock. Diagnosis. Differs from all other Desmoxytes species by the combination of the following characters; collum with three rows of 5+5 anterior, 1(2)+1(2) intermediate, and 3+3 posterior setiferous tubercles; metaterga 2-8 with two rows of 3+3 (anterior row) and 3+3 (posterior row) setiferous cones; metaterga 9-17 with two rows of 4(3)+4(3) (anterior row) and 4(5)+4(5) (posterior row) setiferous cones; sternal cone between male coxae 4 incompletely bilobed, cordiform; lateral lamella of distal lobe (dlm) on lamina medialis broad and thin, demarcated from broad lobe (blm) by a deep indentation.

Desmoxytes octoconigera
Etymology. The name is a Latin adjective, referring to the two rows each with eight setiferous cones on metaterga 9-17.
Distribution and habitat. D. octoconigera sp. n. is known only from Kanchanaburi Province. At the type locality where the holotype and the majority of the paratypes were collected, the animals were crawling on litter inside rock holes; some specimens were seen crawling on rock walls near the cave.
Interestingly, some specimens of D. planata were also found near the cave, however, according to our surveys this species is usually found in places with human activity. We assume that it was probably accidentally introduced to the cave by human actions. Therefore, D. planata and the new species might not share microhabitat, although they live syntopically.   The new species and D. golovatchi sp. n. are sympatric in one location at Tham Khao Noi Bureau of Monks (Wat Tham Khao Noi), and they both have narrow distribution ranges (<100 km 2 ). Currently, the type locality of D. octoconigera sp. n. is situated in an area with considerable human activity (Bureau of Monks and tourist cave), where the forest habitat is cut every year.
We have tried in vain to find this species in another area nearby. Given the narrow distribution range, the new species is probably endemic to Thailand.
Remarks. There is considerable variation in tip of process (plm) of lamina medialis within populations. The process tip in some specimens is bifurcate as two small spines whereas in other specimens it may be relatively blunt.
Coexisting species. Desmoxytes golovatchi sp. n. and D. planata. Diagnosis. Differing from all other species, except D. delfae and D. aurata sp. n., by having a low degree of elevation of paraterga, femora 5 and 6 strongly humped ventrally in middle part, collum with a row of 3+3 anterior setae and metaterga 2-18 with rows of 2+2 anterior and 2+2 posterior small tubercles. Differs from D. delfae and D. aurata sp. n. by having paraterga wider than those species; lamina lateralis (ll) with two distinct furrows ventrolaterally; process (plm) of lamina medialis lamellar, tip blunt.
COLOUR: In life with bright orange. Colour in alcohol: after 10 years in alcohol pale yellow to whitish. ANTENNAE (Fig. 61D): Moderately long and slender, reaching to body ring 6 (male) and 5 (female) when stretched dorsally. COLLUM (Fig. 61A): With 1 transverse anterior row of 3+3 setae; paraterga of collum low, almost horizontal, directed caudolaterad, with two inconspicuous setiferous notches on lateral margin.

Distribution and habitat.
Known only from the type locality. Currently, the habitats at this site are being destroyed and disturbed by human activities, e.g. deforestation for tourist attractions. Our extensive surveys in 2007 allow us to consider this species to be endemic to Malaysia.
Remarks. D. perakensis sp. n. is morphologically similar to D. delfae and D. aurata sp. n. Unfortunately, we did not photograph a living specimen during the field trip; however, one collector noticed orange colouration similar to D. aurata sp. n. and D. delfae.
Coexisting species. None known. Diagnosis. Differs from all other Desmoxytes species by the combination of the following characters; sternal lobe between male coxae 4 subrectangular or subtrapeziform, flattened when seen in lateral view; apical tubercles of epiproct conspicuous, long, digitiform; process (plm) of lamina medialis sharkfin-like, long.
COLOUR: In life with body vivid pink or brownish pink; paraterga vivid pink; metaterga and surface below paraterga brownish pink; head and antenna blackish brown (except distal part of antennomere 7 and antennomere 8 whitish); legs pink or brownish pink; a few basal podomeres whitish pink; sterna brown or pinkish brown; epiproct pink. Colour in alcohol: after two years changed to pale brown.
Distribution and habitat. Known only from the type locality and nearby area. It was found in limestone habitats. Given the narrow distribution in the small limestone area in the west of Thung Salaeng Luang National Park (Srisonchai et al. 2016), we regard this species as endemic for Thailand.
Remarks. We collected additional topotypes during the rainy season. Some morphological variation was found. Thus, in some individuals, the sternal lobe between male coxae 4 is subrectangular, in others subtrapeziform. Gonopod variation is also found in this species; the tip of process (plm) of lamina medialis seems to be slightly blunt in some specimens, slightly sharp in others.
Distribution and habitat. This species was collected at several places together with D. octoconigera sp. n., D. golovatchi sp. n. and D. purpurosea, but these species apparently occupy different microhabitats. D. planata can be found all year round in humid places in environments strongly influenced by humans. It was very easy to find, mostly in plant farms or shaded gardens. Although D. planata was sometimes seen in areas close to another Desmoxytes species, the habitats where it was collected are clearly different: D. planata is often found in human-influenced habitats while the other ones live in natural habitats.
However, some specimens from Tham Khoa Ma Rong, Khoa Ta Mong Lai, Ban Yang Chum (all in Prachuap Khiri Khan Province) seem to be indigenous as they were found in limestone habitats, although not too far from human-influenced habitats. Certain introduced paradoxosomatid species seem to have penetrated natural habitats and have become more dominant and abundant than the native ones (Jeekel 1980b), and D. planata at the above-mentioned locations may similarly have dispersed into the forest. This species has been reported from Hawaii (Chamberlin 1923(Chamberlin , 1941. Because the only reliable records concern specimens taken from quarantine in Honolulu, it seems reasonable to delete D. planata from the Hawaii fauna. We believe that this species is transported easily by global commerce, most probably in soil-containing or plant-associated materials. The origin of D. planata was assumed to be Burma or Malaya by Jeekel (1980a). Later, Shelley and Lehtinen (1998) regarded it more probable that this species occurs naturally in Thailand and China. Based on all recent data analysed by us, D. planata may originally be native to Thailand or Myanmar. Surprisingly, however, no specimen of D. planata has yet been found in Cambodia, Laos, or Malaysia, possibly owing to insufficient surveys in these areas. Decker (2010) reported D. planata from eastern Thailand, viz., Namtok Phliu (Chantaburi Province), Khao Chamoa National Park (Rayong Province) and Ko   Chang National Park (Trat Province). However, these localities all lie in the distribution range of the very similar D. euros sp. n., and Decker's specimens may well belong to the latter species.
Note on type material. The lectotype was designated by Mauriès (1980). In the jar which contains lectotype and paralectotypes in NHMUK, there are three small vials: -one vial contains the male lectotype with a label "Pratinus planatus (Poc.) male lectotype vid. Mauriès (Paris)", pinned through the body; collum and telson were broken off. -a second vial with one male and two female paralectotypes (one female with only rings 10-20), all specimens pinned through the body. -a third vial with broken specimens (>5 specimens).
There is also one more vial containing many broken and mixed specimens (>15 specimens).
Remarks. The vivid pink paraterga are obviously aposematic. According to many previous works, as well as our own results, this species seems to be almost pantropical. In order to assess morphological variability, we compared the gonopods of several specimens (including illustrations) reported from different localities: D. planata from Myanmar (lectotype and paralectotypes); D. coniger - (Jeekel 1980b) from Hawaii taken from Java (Bogor); D. planata -(Mauries 1980) from Seychelles; D. planata - (Shelley and Lehtinen 1998) and (Golovatch and Enghoff 1994) from Fiji; Pratinus rastrituberus - (Zhang, 1986) from China. Combining the examination of previous works and the newly collected specimens from China, Myanmar, and Thailand, variation in morphological characters was as follows. (Fig. 71) -anterior row of tubercles on collum (usually with 4+4 anterior setae): in some specimens lateral setae located in anterior margin (conspicuous tubercles), in others lateral setae located almost at base of paraterga (inconspicuous tubercles). -type of metatergal projections (anterior row and posterior row) on body rings 2-17:

I. Variation within populations
anterior tubercles/cones in some specimens, posterior cones/spines in others. -tip of sternal lobe between male coxae 4: truncate in some specimens, quite round in others. -process (plm) of lamina medialis: short in some individuals, long in others.
-tip of process (plm) of lamina medialis of specimens from Great Cocos Island: slightly emarginate in some, sharp in others. -epiproct tip: in some specimens subtruncate, in others slightly emarginate.

II. Variation between populations
-Fiji population: posterior tubercles on collum seem to be bigger than in other populations.
-Fiji population: sternal lobe between male coxae 4 more round than in others.
-Great Cocos Island population: size seems to be smaller than in others (16-20 mm in male, 20-23 mm in female). -Great Cocos Island population: metatergal tubercles shorter than in others.
On this basis, we strongly agree with Jeekel (1980a) and Golovatch and Enghoff (1994) with the synonymy of D. coniger, E. greeni, E. vector and P. rastrituberus under Desmoxytes planata. Although the type material of D. coniger in MCZ has not been examined by us, its identity with D. planata is clear from the photo and remarks given by Jeekel (1980a). The morphological characters of this specimen match perfectly with the others.
Desmoxytes planata is morphologically similar to D. euros sp. n. with which it shares the metaterga with 2+2 anterior tubercles/cones and 2+2 posterior cones/spines. Moreover, gonopod characters of these species are very similar in shape and processes. Based on morphological characters only, they could be supposed to be the same species. However, the colour of living specimens (paraterga) and shape of hypoproct are totally different, as well as ongoing COI analysis supports to separate them as different species.
Distribution and habitat. Known from several places in many provinces (Kampaeng Phet, Kanchanaburi, Lamphun, Suphan Buri, and Uthai Thani). All specimens were collected from limestone habitats (Fig. 75E); they were very easy to collect because of their aposematic colouration. We noticed that August-September is an annual peak period for adult swarming.
Although this species has been found in several places, it is distributed only in central, west, and north Thailand. Hence, D. purpurosea is regarded as endemic for the Thai fauna.
A specimen from Nakhon Sawan Province (Mae Wong National Park, near the type locality of D. purpurosea) which really looks very much like D. purpurosea is  shown on YouTube ("shocking pink dragon millipede -living treasure of the forest at Mae Wong National Park" (https://youtu.be/jQsn6rOrlA8 -in Thai)). Although we cannot confirm this record because we did not examine specimens from this location, according to the known distribution of D. purpurosea, the specimen from Nakhon Sawan Province may possibly be this species.
Remarks. Interestingly, all adult specimens in all populations show exactly the same colour as reported in the original description: vivid pink to purple. However, we found morphological variation between four main populations delimited as follows: -Size: Specimens from the Lamphun and Uthai Thani populations are larger than others (length 28-34 mm in male, 32-38 mm in female), whereas specimens from   the Kanchanaburi B population seems to be smaller than others (length 22-26 mm in male, 26-28 mm). -Sternal lobe between male coxae 4: The only studied specimen from the Lamphun population has a subsemicircular and quite short lobe, while in others the lobe is trapeziform. -Apical tubercles of epiproct: Distinctly longer in the Kanchanaburi B population than in others. -Process (plm) of lamina lateralis: A bifurcate tip, as two conspicuous spines, in the Lamphun and Uthai Thani populations, but terminating in several spines in the Kanchanaburi A and B populations.
The shape of the hypoproct varies within populations: in some specimens it is trapeziform, in others it is subsemicircular.

Diagnosis.
Differs from all other Desmoxytes species by the combination of the following characters; lamina lateralis (ll) subtriangular; ventral lobe (vll) of lamina lateralis thumb-like, large and long; broad lobe (blm) of lamina medialis indistinctly demarcated from distal lobe (dlm) of lamina medialis by very shallow or slightly deep indentation.
Distribution and habitat. Known only from Tak and Kamphaeng Phet Provinces. This species is restricted to limestone habitats and was seen crawling on litter and decaying bark (Fig. 82F). Srisonchai et al. (2016) reported that D. takensis was found on humid plastic garbage, a sign that the type locality is clearly under human influence. There is a long, broad concrete natural trail into the waterfall, and lots of garbage littered the type locality. However, the species has also been found in other natural habitats.
During our intensive surveys in western Thailand, we found this species in many places. However, it has a narrow range and occurs in only two provinces. Thus, D. takensis should be regarded to be endemic to the Thai fauna.
Remarks. Based on morphology, we divided our material specimens into two main populations: Population A includes specimens from Nangkruen Waterfall (type locality), Tham Takhobi The two populations differ in some characters as follows: -Colour: The remarkable body colour of the two populations apparently differs: bright red in population A, vivid pink in population B (some old females strongly pinkish to reddish). -Size: Population B individuals seem to be bigger than population A ones in both width and length (see size description). -Hypoproct: The shape of the hypoproct in population A is subsemicircular whereas it is subtrapeziform in population B. -Gonopods: The ventral lobe (vll) of lamina lateralis of population B specimens is large, thumb-like, longer, and more slender than that of population A ones. The distal lobe (dlm) of lamina medialis in population A specimens consists of one lamella while population B specimens have two lamellae distally.
Although the two populations vary in some morphological characters, they show an overall gonopodal resemblance. According to the differences in morphology of the two populations, this might an example of ongoing speciation in allopatry, supported by the confinement of the two populations to two large isolated limestone regions located in the northern (Population B) and southern (Population A) parts of the distribution area.
We collected some juveniles during the field trip and kept them with litter until they moulted. Interestingly, the juveniles made a moulting chamber which was apparently produced by fecal material and silk; it is probable that the building process is the same as in the families Polydesmidae, Pyrgodesmidae and in order Callipodida (Adis et al. 2000, Youngsteadt 2009, Reboleira and Enghoff 2016. This is the first observation of moulting in dragon millipedes; however, we did not keep an eye on them in detail. After moulting and emerging from the chamber, the specimens were in an early adult stage showing a pale whitish colouration. Nearly 2 weeks later, they became vivid pink (Fig. 82D Diagnosis. Metaterga 9-19 usually with 2+2 cones/spines (anterior row) and 3+3 cones/spines (posterior row). Similar in this respect to D. breviverpa, D. purpurosea and D. takensis. Differs from these species by the following combination of characters; process (plm) of lamina medialis short, thick and broad, directed mesad, tip blunt; distal lobe (dlm) apically with two distinct lamellae, mesal and lateral lamellae equal in size, very broad and thick; epiproct short; male femora 5 and 6 slightly humped ventrally.
Distribution and habitat. This species is known from Myanmar -Rangoon and Pegu (Taikkyii and Palon). [Rangoon is currently Yangon. Pegu is presently Bago   township (Hanthawaddy), Bago region. Taikkyii is Taikkyi township, Yangon region. Palon is probably a small village located in the area north to Taikkyii in the west of Pegu, Yangon region]. Taikkyii and Palon were formerly parts of Pegu region, now they belong to Yangon region. Therefore, the label of specimens collected by Fea gives the locality as Pegu (Taikkyii and Palon).
Habitat details for this species have never been reported; however, all locations are supposed to be granitic and limestone mountain ranges based on geological data, and the two locations were approximately 20 km apart.
We assume that D. taurina is distributed in a narrow range. A field survey near Yangon in 2015 revealed no further specimens of D. taurina. Therefore, we regard this species as endemic to Myanmar.
Note on material. In the original description Pocock (1895), wrote that all specimens were collected from Rangoon by Oates and from Pegu (Taikkyii and Palon) by Fea. Two females collected by Oates are now in NHMUK, one specimen collected by Fea in ZMH and two males collected by Fea in ZMUC.
The two males in ZMUC collected by Fea are labelled "cotypes", and only "Palon" is given as locality whereas in the original description Pocock gave "Pegu (Taikkyii and Palon)". We assume that these two males were probably collected from different locations, one from Taikkyii and one from Palon. Weidner (1960) classified a specimen (unknown sex, not studied by us) of Prionopeltis taurinus (= Desmoxytes taurina) in ZMH as a "paratypoid". However, Pocock (1895) and the following authors did not designate a holotype or lectotype for this species, thus, all specimens are syntypes.
The lectotype chosen is the ZMUC male with one remaining gonopod. The other ZMUC, ZMH and NHMUK specimens are designated here as paralectotypes.
Colour of type specimens: the lectotype is brown without metallic oxidation of the pin while the paralectotypes in NHMUK have become greenish black with metallic oxidation of the pin.
Remarks. This species has not been revised since Golovatch and Enghoff gave a good description in 1994. Golovatch and Enghoff (1994) described the collum with rows of 3(4?)+3(4?) anterior tubercles, suture between prozona and metazona distinctly beaded, pleurosternal carinae absent. After examining all known specimens except the one in ZMH, we found: -collum with rows of 4+4 anterior tubercles (lateral tubercles near base of paraterga).
-suture between prozona and metazona not beaded, but with very small ridges of irregular shape. -pleurosternal carinae of all specimens conspicuously present on body ring 2, very small ridges on body ring 3, thereafter missing.
We noticed that the number of cones (posterior row) on metaterga varies between individuals. Most specimens have metaterga 8 with 2+2 tubercles in the posterior row, but some have 3+2 tubercles. Metaterga 9-19 usually have 3+3 tubercles in the posterior row, whereas some individuals have with 3+4 or 4+3 tubercles.
The length of antenna in male could not be examined (antennae missing in both males), but the antennae are supposed to reach to ring 5 (Pocock 1895).

Figure 92. Desmoxytes terae
Distribution and habitat. This species is known only from the Malaysia-Thailand borderland (Kaki Bukit and Thale Ban National Park). We noticed that it prefers to live on humid mosses, logs or litter in limestone habitats (Fig. 88C). The localities are all on limestone mountains. The species is probably distributed in a narrow range in limestone areas near the Thai-Malay border. D. terae has so far been recorded from three localities (near Kaki Bukit, Thale Ban National Park, and Tham Tone Din) which are located only 10-15 km apart. Despite several attempts by us, D. terae was not found in other areas. Hence, this species should be regarded as endemic to the Malaysia and Thailand faunas.
Remarks. In the recent field surveys we noticed that the colour of living specimens is black or brownish black with contrasting white paraterga as reported earlier by Jeekel (1964) and Golovatch and Enghoff (1994). This species blends so perfectly with its environment that it is difficult to collect specimens without a flashlight.
In the original description, Jeekel (1964) stated about the collum: "near the anterior margin a transverse row of six hairs, which may be present partly rubbed off, the lateral pair placed on the low tubercles". This means collum with one row of 3+3 anterior setae/tubercles. Moreover, Jeekel also described paraterga without a tiny denticle near the tip. After we examined all specimens, it is clear that: -Collum with 3 rows of 3+3 anterior setae, 1+1 intermediate tubercles and 2+2 posterior tubercles (tubercles without setae, quite small but conspicuous).
The sternal lobe between male coxae 4 shows some variation within populations; the lobe of some specimens is trapeziform whereas in others it is semicircular. We also found some variability on the telson: tip of epiproct subtruncate in some individuals, in others slightly emarginate; lateral setiferous tubercles conspicuous in some specimens, inconspicuous in others; caudal margin of hypoproct truncate in some individuals, slightly round in others.
Distribution and habitat. Known only from the type locality in a limestone area (Fig. 93E). We assume that it is distributed in Kayin State. We made an intensive search for this species in September 2016 in Mawlamyine State (very close to Kayin State), however, no further specimens were found. Since it is so far known only from the type locality, D. waepyanensis sp. n. should be regarded as endemic to Myanmar.
Remarks. The tip of process (plm) on lamina medialis is quite variable, having one, two, or three small spine(s), or being almost blunt in some specimens.
This species seems to be aposematic, to judge by the remarkable pink paraterga and pinkish brown body.

Discussion
Our analyses of morphology, as well as our preliminary molecular phylogeny, supports the subdivision of Desmoxytes s.l. into five groups (Desmoxytes s.s., the 'acantherpestes'group, the 'gigas'group, the 'spiny'group and Hylomus). All groups are clearly defined on morphological characters, especially of the gonopods and paraterga (Figs 2, 3). The distribution areas of the groups seem to be clear in their boundaries.
The Hylomus group is more diverse in shape of paraterga, however, its members show a notable similarity in gonopod shape (Fig. 2E, F). Desmoxytes s.s., the focus of the present paper, is well-defined based on gonopod characters especially the solenophore (lamina lateralis (ll) and lamina medialis (lm)). Species of Desmoxytes s.s. share several morphological similarities including wing-like paraterga; lamina lateralis (ll) swollen; lamina medialis (lm) comprised of process (plm), distal lobe (dlm) and broad lobe (blm); and the modification of male femora 5 and 6 only (exception: D. terae). Certain morphological characters show intra-and inter-population variations within the same species. The most variable characters within populations are: Inter-population variation was also found in some species as follows: -size: In D. purpurosea, specimens in the two main northern populations are clearly bigger than those from the two main southern populations. Specimens in a population of D. planata from Great Cocos Island seem to be smaller than others. -colour: colour variation of living specimens of dragon millipedes is reported here for the first time. D. cervina includes brownish red as well as brown individuals. Specimens from the northern populations of D. takensis are red and those from southern populations are pink; however, the other morphological characters are identical.
Variation of colour and size within and between populations may at least in part be due to quality and quantity of food, differences in the physical environment (temperature, soil, humidity), like in other arthropods. Hagen (1881) and Buckton (1879) found that the colour of some arthropods was affected by nutrients and temperature.
Many studies, e.g. Baker (1989) and Juliano (1986), have shown that food is one of the main factors controlling growth rate, body size, etc. in arthropods. For millipedes, there is the study by Berns and Keeton (1968) on Narceus annularis (order Spirobolida) showing that semi-starved individuals attained smaller body sizes than well-fed ones. David and Célérier (1997) showed that individuals of Polydesmus angustus kept on a diet of leaf litter plus yeast attained larger body sizes than individuals fed on leaf litter alone. On this background we assume that food and the physical environment may affect colour and size in Desmoxytes. Another possible factor controlling differences in colour and size might be the genetic variation within and between populations.
There are some species showing great resemblance in gonopod characters. In particular, D. planata and D. euros sp. n. are remarkable in having identical gonopods. Nevertheless, the yellow paraterga, shape of hypoproct and the initial study on mitochondrial COI gene supports to separate them as different species. The D. planata-D. euros sp. n. case reminds of what Pimvichai et al. (2011) found for Thyropygus induratus Attems, 1936vs T. quietus Attems, 1938: a pair of species with virtually identical gonopods but significant genetic and non-gonopodal morphological differences.
Desmoxytes (and other dragon millipedes) are particularly attractive animals because of the peculiar paraterga, in combination with the unusual vivid colour in some species. The bright colour probably is a warning signal (Svadova et al. 2009, Shear 2015, Marek and Moore 2015. Many mating couples were found during our field surveys, and we collected some representative couples and reared a few specimens of D. euros sp. n. and D. takensis in acrylic boxes, feeding them with natural litter. We observed that the millipedes made moulting chambers using fecal material and silk as found in other polydesmidan families such as Polydesmidae and Pyrgodesmidae, as well as in the order Callipodida (Adis et al. 2000, Youngsteadt 2009, Reboleira and Enghoff 2016. We also observed a host-parasitic association between millipedes and mites in D. cervina (probably the mites belong in genus Leptus Latreille, 1976). We assume that the mite species is a parasite, like in other Leptus spp. (Southcott 1992) which use the millipede host for nourishment and dispersal purposes.
Figs 99 and 100 clearly show that all Desmoxytes species (except D. planata) are narrowly distributed, and all are restricted to limestone habitats or granitic mountains. The narrow distributional ranges of Desmoxytes species are perhaps the result of their poor dispersal capacities.
Desmoxytes planata, a pantropical species, has been recorded from widely scattered places. According to our survey we suspect that D. planata is probably originally native to Thailand or Myanmar. Especially in Thailand, we noticed that it ranges from Chiang Rai (northern end) to Chumphon Province (middle) (Fig. 100).
The species diversity of dragon millipedes is impressive. At the moment, Thailand, Malaysia, and Myanmar contain 18 species of Desmoxytes. We believe that many more new species remain to be discovered, especially in Thailand, Myanmar, Malaysia, Cambodia, and Laos.