The millipede family Polydesmidae in Taiwan, with descriptions of five new species (Polydesmida, Diplopoda)

Abstract Polydesmidae are represented in Taiwan by seven species in two genera. Neither of the genera is endemic to Taiwan, but six of the species are, including five new: Nipponesmus minor sp. n., Epanerchodus bispinosus sp. n., Epanerchodus curtigonopus sp. n., Epanerchodus flagellifer sp. n. and Epanerchodus pinguis sp. n. In addition, the diagnosis of the hitherto enigmatic genus Nipponesmus Chamberlin & Wang, 1953 is refined vis-à-vis the especially similar, Central Asian, Siberian and Eastern European genus Schizoturanius Verhoeff, 1931, chiefly based on new material of the type-species Nipponesmus shirinensis Chamberlin & Wang, 1953; this species is adequately redescribed and represents still another Taiwanese endemic. A key to all three currently known species of Nipponesmus Chamberlin & Wang, 1953 is given. The highly speciose Central to East Asian genus Epanerchodus Attems, 1901 is represented in Taiwan by five species, all keyed, including Epanerchodus orientalis Attems, 1901, which is long known to be highly variable in Japan and found particularly polymorphous and apparently allochthonous in Taiwan. The following synonymy is formalized: Epanerchodus orientalis orientalis Attems, 1901 = Epanerchodus orientalis takakuwai Verhoeff, 1913, syn. n. The genus Usbekodesmus Lohmander, 1932 is formally synonymized with Epanerchodus Attems, 1901, syn. n., resulting in the following new formal transfers: Epanerchodus redikorzevi (Lohmander, 1932), Epanerchodus swatensis (Golovatch, 1991), Epanerchodus varius (Geoffroy & Golovatch, 2004), Epanerchodus anachoretus (Golovatch, 1986), Epanerchodus buddhis (Golovatch, 1986), Epanerchodus occultus (Golovatch, 1986), Epanerchodus sacer (Golovatch, 1987), Epanerchodus theocraticus (Golovatch, 1990) and Epanerchodus theosophicus (Golovatch, 1986), all comb. n. ex Usbekodesmus. The distributions of all seven species of Polydesmidae occurring in Taiwan are mapped and discussed.

Keywords millipede, Polydesmidae, taxonomy, new species, new synonymy, key, Taiwan introduction The millipede fauna of Taiwan is still quite poorly explored (Korsós 2004), with fewer than 70 species currently reported from this large, subtropical to tropical island. Many more species may be expected to occur there, to judge from the recent increase in the list of Taiwanese Glomerida from one species to five ). This statement is easy to reconfirm with the large family Polydesmidae, order Polydesmida, taken as another example. Taiwan has hitherto been known to support only two polydesmid species: Epanerchodus orientalis Attems, 1901 and Nipponesmus shirinensis Chamberlin & Wang, 1953, both from a few localities only (see review: Korsós 2004).
Prompted by the abundant new material collected during the last couple of decades all over the island, below we offer a revision of the Taiwanese Polydesmidae. Already seven species appear to be involved, representing two genera. These new samples also allow for the diagnosis of one genus to be refined, and the synonymy of another one formalized.

Material and methods
Material serving as the basis for the present contribution was preserved in 75% alcohol and is currently shared between the collections of the National Museum of Natural Science, Taichung, Taiwan (NMNS), Taiwan Forest Research Institute, Taipei, Taiwan (TFRI), Department of Biological Sciences, National Sun Yat-Sen University, Kaohsiung, Taiwan (NSYSU), Zoological Museum, State University of Moscow, Russia (ZMUM), Natural History Museum of Denmark, University of Copenhagen, Denmark (ZMUC), Muséum national d'Histoire naturelle, Paris, France (MNHN), and Institute of Biology and Soil Science, Far Eastern Branch, Russian Academy of Sciences, Vladivostok, Russia (IBSS), as indicated hereafter. Specimens were studied and illustrated using standard stereomicroscopic, photographic and drawing equipment.
In the catalogue sections, which mostly refer to the fauna of Taiwan alone, D stands for the original description, N for additional descriptive notes, and R for a mere mention or record.  antennal socket (Fig. 5). Antennae rather long and only slightly clavate (Fig. 5), either slightly overreaching segment 3 dorsally (♂) or slightly shorter (♀); antennomere 3 longest (Figs 3-5); antennomeres 5 and 6 each with a small, compact, distodorsal group of bacilliform sensilla; antennomere 7 with a minute dorsoparabasal cone and a distodorsal group of microscopic sensilla.
Gonopods (Figs 4,5,(11)(12)(13) with large, subquadrate, medially fused coxae carrying a few long setae ventrally. Telopodite elongated, slender, falcate, prefemoral (densely setose) portion almost half as long as entire telopodite; seminal groove running mesally over most of its extent, only distally moving frontally to recurve first laterad and then a little basad at base of both a somewhat shorter, more complex endomere (en) and a longer, simpler exomere (ex); en beset with long, bacilliform setae distally and supplied with an evident spine (s) mesally, as well as a prominent pulvillus near base, this pulvillus being likewise beset with bacilliform setae and marking the end of seminal groove devoid of any accessory seminal chamber; ex strongly unciform apically, with an additional lateral tooth distally to subapically.
Remarks. This species is the largest, as well as one of the most conspicuous and common among the polydesmids in Taiwan. It inhabits a wide range of habitats, mainly montane woodlands above 1,200 m a.s.l. (Map 1).   Name: To emphasize the smaller body size and the shorter gonopod telopodite. Diagnosis: Differs from N. shirinensis, the only other congener known from Taiwan, in the smaller size, as well as in the collum being narrower than the head and in the gonopod telopodite being stouter and shorter (see also Key below).

Nipponesmus minor
Description: Length of both sexes ca 12-16 mm; width of pro-and metazona varying between specimens from 0.8-1.3 to 1.5-2.0 mm, respectively. Holotype ca 12 mm long, and 0.8 and 1.5 mm wide on pro-and metazona, respectively. Usually ♂♂ Map 1. Distribution of Nipponesmus species in Taiwan. Borderlines show borders between the counties. N. minor sp. n.: filled red triangles; N. shirinensis: filled black circles. somewhat smaller than ♀♀. Coloration in alcohol from uniformly pallid (faded?) to yellowish to reddish-brown; in the latter case, sides, venter and legs light grey-brown .
Gonopod telopodite (Figs 20-23) much stouter; endomere (en) slightly longer than exomere (ex), beset with long, bacilliform setae nearly throughout, usually supplied with an evident spine (s) or tooth mesally, as well as a usually somewhat less prominent pulvillus near base; ex often not so strongly unciform apically, often with an additional lateral tooth in distal part.
Remarks. This species is not so widely distributed in Taiwan. Allopatry is prevailing, as it is only at Mei-Feng that both the congeners co-occur (Map 1). However, N. minor sp. n. quite often lives even syntopically together with Epanerchodus orientalis (see below).
A study of the ample material representing not only the type species N. shirinensis, but also the above new congener from Taiwan allows for the identity of Nipponesmus Chamberlin & Wang, 1953 to finally become clarified.
This genus has hitherto remained enigmatic, originally described too poorly (Chamberlin and Wang 1953) to shed any significant light on its affinities. It was only based on the gonopod conformation of the beautifully described N. tangonis (Murakami, 1973), a species from Honshu, Japan definitely most similar to N. shirinensis, that Golovatch (1991) suggested the relationships of Nipponesmus as a genus possibly somewhat intermediate between Schizoturanius Verhoeff, 1931 andEpanerchodus Attems, 1901. In N. shirinensis, the gonopod telopodite (Figs 11-13) is indeed biramous only distally, being divided into subequally prominent endo-and exomere. Furthermore, the endomere (en) is beset with a characteristically bacilliform trichome, whereas the exomere (ex) is simple. The seminal groove runs mostly mesally to recurve neatly between ex and en and then to debauch somewhat basally into a prominent hairy pulvillus which is also beset with the same peculiar trichome, and is devoid of an accessory seminal chamber. The same general pattern is observed both in N. tangonis and N. minor . What we term here as endomere is the branch which Murakami (1973) erroneously referred to as solenomere in his N. tangonis. Yet it can hardly be called such, because the seminal groove ends somewhat basally of it and thus fails to support it at all. This is where one of the basic distinctions between Nipponesmus and Schizoturanius seems to lie, because in Schizoturanius the recurvature point of the seminal groove between both distal branches ex and en is about level to the pulvillus (Golovatch 1979). So we can speak in this case about a true solenomere. In addition, in Schizoturanius an accessory seminal chamber, however small, is present. Another characteristic feature of Nipponesmus vis-à-vis not only Schizoturanius, but also all other genera of Polydesmidae is an abundant bacilliform trichome on the endomere. Although a similar trichome is known to occur in some European Polydesmus species as well, it is always located either on the exomere or on a solenomere. These three apomorphies distinguish Nipponesmus from the obviously most similar Schizoturanius, a genus encompassing several species in Central Asia, Siberia and the southern part of the Eastern European Plain (Golovatch 1979(Golovatch , 1991. As regards Golovatch's (1991) placement of Nipponesmus also near Epanerchodus, following Hoffman (1980) who had synonymized these genera, this idea is false, as they appear to show too many profound differences in gonopod structure. Thus, in Epanerchodus the endomere is mostly absent, rarely present as only a rudimentary structure, while the seminal groove after the recurvature point still makes a long way basad to debauch into a prominent, simple-haired, accessory seminal chamber placed at the bottom of a profound parabasal cavity in the telopodite (see also below). Based on the gonopod conformation of one of the new Taiwanese species (see below), Usbekodesmus Lohmander, 1932, differing from Epanerchodus only in a somewhat better developed exomere, albeit also simple and more or less spiniform, is to be regarded as another junior synonym of Epanerchodus, syn. n. Arguments for synonymizing both these genera have long been put forth (Geoffroy and Golovatch 2004), but until now no formal synonymy has been advanced.
The following key can serve to separate all three currently known species of Nipponesmus. Diagnosis: Differs from the other Epanerchodus species known from Taiwan in the variable, mostly medium size, coupled with the caudal corner of most of the paraterga being pointed, and in the gonopod telopodite being relatively slender, complex, highly variable in shape and armature (see also Key below).
Gonopods (Figs 32-36) with large, subquadrate, medially fused coxae carrying a few long setae ventrally. Telopodite stout to rather slender, subfalcate, prefemoral (densely setose) portion one-third to half as long as entire telopodite; seminal groove running mesally over much of its extent, only distally moving frontally to recurve first laterad and then mesad, squeezing neatly between a simple, more or less rudimentary to completely reduced exomere (ex) and a more complex, branching and always welldeveloped endomere (en), then groove continuing to a considerable extent basad to end into a large accessory seminal chamber lying near base of a prominent excavation, the latter carrying an evident hairy pulvillus; en often but not always with a strong mesal process (s), either slenderer and simple or larger and more complex in shape, as well as usually with a distinct, often again branching, laterobasal outgrowth (p); distal remainder of en elongate, often branching and enlarged, sometimes fringed with short setae or spines. Remarks. This species seems to be the most widespread and variable among congeners. It has heretofore been accepted as being split into two nominal subspecies: E. orientalis orientalis and E. o. takakuwai Verhoeff, 1913, but, given a similarly profound variation range in body size and shape, and, especially, in gonopod structure as observed in Japan alone, the subspecific status of takakuwai has long been questioned (e.g. Miyosi 1959, Murakami 1969, Nishikawa and Murakami 1993. , in the latest checklist of the Japanese Diplopoda, treats E. orientalis without subspecies. Moreover, numerous samples from Hokkaido, Honshu and Shikoku, Japan, reveal that variation in gonopod conformation appears to be purely individual, failing to demonstrate any meaningful geographical patterns (Nishikawa and Murakami 1993). The same concerns the available samples from Taiwan. So we formalize here the long suspected synonymy: E. orientalis orientalis Attems, 1901 = E. orientalis takakuwai Verhoeff, 1913, syn. n.
It is noteworthy that, formally, E. orientalis fails to occur in southernmost Japan, i.e. on Kyushu Island and in the Ryukyus (Nishikawa andMurakami 1993, Nakamura andKorsós 2010), to reappear further south only in Taiwan. In northern Honshu, Japan, it is known to reach 26 mm in length and 3.4 mm in width, with the collum being slightly broader than the head (Murakami 1969). In Taiwan, as proven by our study of several abundant syntopic samples, the animals tend to be smaller, the collum is invariably narrower than the head, while rather often the gonopods are totally devoid of both an exomere remnant and process s, frequently with the distal, longest part of their endomere being slender, not expanded.
Looking at such a profound variation as observed in a number of peripheral and gonopod characters in E. orientalis, one cannot ignore several further nominal congeners described from Japan, including Kyushu, in which the gonopods look especially similar to those of E. orientalis: E. inferus Verhoeff, 1941, E. lobatus Verhoeff, 1941, E. satoi Takakuwa, 1954, E. tenuis Takakuwa, 1954, E. aculeatus Miyosi, 1954, E. etoi Miyosi, 1955, E. chichibensis Haga, in Takashima and Haga 1956, E. yoshidai Haga, in Takashima and Haga 1956, E. lacteus Shinohara, 1958, etc. (e.g. Takashima and Haga 1956, Shinohara 1958, Miyosi 1959. We suspect that some of them might well prove to represent junior synonyms of E. orientalis. Only future in-depth biological observations and such genetic investigations as bar-coding of such particularly similar forms can shed light on their true identities and statuses, because polymorphic variation has long been known in Epanerchodus. Thus, E. polymorphus Mikhaljova & Golovatch, 1981, widespread in the southern part of the Russian Far East and in northern Korea, shows two morphologically distinct morphs both in gonopod and peripheral structure in males, but a complete, overlapping range of the same somatic characters in females. Both male morphs invariably co-occur syntopically and either mates with any female variety (Mikhaljova andGolovatch 1981, Mikhaljova 2004). A similar situation is found in the nominate species E. acuticlivus Murakami, 1970 and E. aster Murakami, 1970, both described from the same cave in Shikoku, Japan. Murakami (1970) explicitly admitted that they were very close, with their females being indistinguishable, while the males showed small but stable differences in gonopod telopodite armature. Based on the great variation observed in the populations of E. orientalis in Japan and, especially, Taiwan, the statuses of E. polymorphus, E. acuticlivus and E. aster as further congeners highly similar to E. orientalis are likewise to be questioned, as at least some of them might also prove to be the latter's junior synonyms. Yet no formal synonymies are advanced here as obviously being too premature at this purely descriptive taxonomic stage.
That not all of the congeners similar in gonopod structure to E. orientalis are synonyms of the latter is proven at least by E. koreanus Verhoeff, 1937, a species common throughout Korea, in Kyushu, Japan and in the southern Russian Far East. At least in Russia, it often occurs syntopically together with E. polymorphus. Yet E. koreanus, despite its minor variations in gonopod conformation, is easily recognizable even superficially through its much wider paraterga (3.3-3.8 versus 2.6-3.0 mm); in addition, these species never mate (Mikhaljova 2004). As another proof to the above statement may also serve E. pinguis sp. n., described below and representing still one more congener quite similar in gonopod structure to E. orientalis. In Taiwan, both are at least partly sympatric, but E. pinguis sp. n. differs markedly enough in a number of peripheral and gonopod characters to warrant recognition of a distinct species (see below).
In other words, several, but definitely not all, of the nominate species of Epanerchodus that show their gonopods particularly similar in structure to those of E. orientalis are jeopardized as potential junior synonyms of the latter species.
In Taiwan Name: To emphasize the stout gonopod telopodite. Diagnosis: Differs from the other Epanerchodus species, in particular from the apparently especially similar E. orientalis, in the mostly square, broader and slightly upturned paraterga, coupled with the gonopod showing an unusually densely setose coxa and a remarkably stout telopodite (see also Key below).
All characters as in E. orientalis except as follows. Antennae rather long, slender, only slightly clavate, reaching behind segment 3 dorsally; antennomere 3 longest, clearly longer than highest 5 th ; antennomeres 5 and 6 each with an evident, compact, distodorsal group of bacilliform sensilla; antennomere 7 with a minute dorsoparabasal cone and a distodorsal group of microscopic sensilla.
Remarks. This species is apparently very local in distribution (Map 2) and seems to be allopatric with E. orientalis. Name: To emphasize the distal half of the gonopod telopodite being like two long spines.

Epanerchodus bispinosus
Diagnosis: Due to the presence of a strong and spiniform exomere, this new species joins the few congeners hitherto referred to the erstwhile genus Usbekodesmus (see above the synonymy with Epanerchodus), but differs in the distal half of the gonopod telopodite being represented by only two spiniform branches showing no additional outgrowths (see also Key below).
Description: Length of both sexes ca 8-11 mm; width of pro-and metazona varying between specimens from 0.8-1.3 to 1.1-1.7 mm, respectively. Holotype ca 9 mm long, and 0.8 and 1.1 mm wide on pro-and metazona, respectively. Coloration in alcohol from pallid to uniformly light grey, yellow or very light red-brown, sometimes head faintly marbled light red-brown; venter and legs yellowish to greyish (Figs 45-49). All characters as in E. orientalis except as follows. Antennae rather long and evidently clavate, reaching behind end (♂) or midway (♀) of segment 3 dorsally; antennomere 3 longest, considerably longer than a relatively stout, yet highest, 5 th (Figs 45); antennomeres 5 and 6 each with a small, compact, distodorsal group of bacilliform sensilla; antennomere 7 with a minute dorsoparabasal cone and a distodorsal group of microscopic sensilla.
In width, collum < segment 2 = 3 < 4 < head = 5-15 (Fig. 46), thereafter body gradually tapering towards telson (♂, ♀) (Fig. 48). Paraterga moderately developed, starting from collum, set high but invariably lying slightly below a faintly convex dorsum; paraterga on collum small, subtriangular, with a small lateral incision in front of a rounded caudal corner; front shoulders drawn forward and slightly convex only paraterga 2-4, thereafter straight, increasingly well rounded and directed increasingly caudolaterad; caudal corner on pataterga 2-5 subrectangular and narrowly rounded, starting from 6 th increasingly acutangular and beak-shaped, starting from segment 9 first faintly and then extending increasingly beyond rear tergal contour . All poreless segments with three, all pore-bearing ones with four, small but evident incisions, each usually bearing a small seta on top at lateral margin. Metatergal sculpture typical, moderately developed, with three indistinct transverse rows of setiferous, polygonal bosses . Tergal setae very short, mostly retained, a little longer only on collum and in rear row on metatergum 19. Stricture between pro-and metazona wide and smooth. Limbus very thin, microdenticulate. Epiproct rather short, conical (Fig. 48), only slightly bent ventrad, preapical papillae evident. Hypoproct semi-circular; caudal, paramedian, setiferous papillae evident and well-separated.
Remarks. This species is remarkable in showing an exomere, albeit usual and spiniform, almost as long as a particularly simple endomere. This condition nicely bridges the weak distinction which has hitherto remained to formally keep Usbekodesmus and Epanerchodus as independent genera (Geoffroy and Golovatch 2004), with the gonopod structure in E. bispinosus sp. n. providing the final evidence to formally synonymize these genera (see also above).
E. bispinosus sp. n. in Taiwan is apparently very local in distribution, having been encountered only at a single locality (Map 2).
Description: Length ca 15 mm; width of pro-and metazona 1.3 and 2.5 mm, respectively. Coloration in alcohol pallid.
Superficially, also very similar to E. pinguis sp. n., except as follows.
Remarks. This species is remarkable in showing an extremely short, simple, spiniform endomere (en), coupled with a single, even more rudimentary process (p) at en base.
In Taiwan Diagnosis: Differs from other Epanerchodus species in the distal half of the gonopod telopodite being particularly long, flagelliform, coupled with the presence of two rounded teeth at the base of the endomere (see also Key below). From the other congeners known from Taiwan, this new species differs also in the absence of sphaerotrichomes on ♂ legs.
Description: Length of both sexes ca 8-11 mm; width of pro-and metazona varying between specimens from 0.8-1.0 to 1.4-1.5 mm, respectively. Holotype ca 9 mm long, and 0.9 and 1.4 mm wide on pro-and metazona, respectively. Coloration in alcohol pallid to light grey-to red-brown (Figs 61-65).
All characters as in E. orientalis except as follows. Antennae rather long and evidently clavate (Figs 61, 64), reaching behind segment 3 dorsally.
Remarks. This small species is remarkable in showing a particularly long, flagelliform endomere, coupled with only two short outgrowths at its base.
In Taiwan, E. flagellifer sp. n. occurs very locally, having been encountered only at a single locality (Map 2). variable and lowland-dwelling across at least most of Japan, seems to be the only allochthonous element in the fauna of Polydesmidae of Taiwan, likely a later colonizer from Japan.