Review of the millipede family Opisotretidae (Diplopoda, Polydesmida), with descriptions of new species

Abstract The small, basically Oriental family Opisotretidae is rediagnosed, reclassified, and shown to comprise the following seven genera, all keyed: Carlotretus Hoffman, 1980, with two species, including Carlotretus triramus sp. n. from southern China; Corypholophus Attems, 1938, with two species, one in Vietnam, the other in the Ryukyus, Japan; Martensodesmus Golovatch, 1987, with eight species, all keyed, including Martensodesmus cattienensis sp. n. from southern Vietnam, as well as Martensodesmus bedosae sp. n. and Martensodesmus spiniger sp. n. from southern China; Opisotretus Attems, 1907, with seven species, all keyed, including Opisotretus beroni sp. n. and Opisotretus hagen sp. n., both from Papua New Guinea, Opisotretus deharvengi sp. n. from Sulawesi, Indonesia, and Opisotretus spinosus sp. n. from Nusakambangan Island, off Java, Indonesia; Opisthoporodesmus Silvestri, 1899, with six nominate species; Retrodesmus Chamberlin, 1945, with two species, i.e. the type-species Retrodesmus dammermani Chamberlin, 1945, from Java, Indonesia, revised from the holotype, and Retrodesmus cavernicola sp. n., from Papua New Guinea; and Solaenaulus Attems, 1940, with two species. Comments are presented on the family’s possible relationships and palaeogeographic history. Instead of being considered as the sole component of the superfamily Opisotretoidea, the Opisotretidae is believed here to form one of the families of the diverse superfamily Trichopolydesmoidea, perhaps the sister-group to, if not immediately derived from, the pantropical family Fuhrmannodesmidae. The origin of Opisotretidae, previously dated as far back as the Triassic (220 Ma) in relation to the fragmentation of eastern Gondwanaland, mainly in the region of present-day Indonesia, could have had nothing to do with Gondwanaland. Opisotretids might have originated in mainland Southeast Asia well within the Cenozoic, with subsequent dispersals along the Himalayas in the West and across Indonesia (including New Guinea) in the East, also reaching as far north as the Ryukyus, Japan and Guangxi, southern China.

Apparently, introduced to both latter localities.
As one can see from the above list, several species have been described too poorly to realistically become recognized. This holds especially true of what Chamberlin (1945) described in Opisthoporodesmus, making the compilation of even a superficial key to Opisthoporodesmus species impossible. The few he described from ♂ material must be revised, whereas the identities of the species which were based on ♀ and/or juvenile samples are bound to remain enigmatic until ♂ topotypes have been obtained and properly described. Since the main objective of the present paper is to address the generic classification of Opisotretidae in order to identify and name a number of fresh samples ranging from continental southern China, through Indochina and Indonesia, to Papua New Guinea, only Retrodesmus dammermani, fortunately an intact ♂ holotype, has been revised here.

Abbreviations used AMNH
American Museum of Natural History, New York, U.S.A.

Material and methods
The bulk of the material treated below was taken by Louis Deharveng and Anne Bedos (MNHN) in Indonesia and China, as well as nearly entirely by Petar Beron (NMNHS) in Papua New Guinea. A few samples derive from ZMUM. The holotype of Retrodesmus dammermani was received on loan from AMNH. The holotypes from Indonesia have been housed in MZB, those from China in the collection of IZAS, whereas a few paratypes from China have been deposited in SCAU. Much of the material has been kept at MNHN, a few duplicates have also been donated to ZMUC and NMNHS, as indicated below.
SEM micrographs were taken using a JEOL JSM-6480LV scanning electron microscope.
After examination, SEM material was removed from stubs and returned to alcohol, all such samples from Papua New Guinea being kept in NMNHS, from the remaining places in MNHN.

The main characters used in the classification of Opisotretidae
The following characters have been used for defining the genera in Opisotretidae, the only family in the entire order Polydesmida in which the gonopods are directed dorsolaterad, curving very strongly around coxae 8 along the sides of segment 7:

Number of body segments.
Like in most other families in Polydesmida, the number varies from 19 to 20, mostly being sex-characteristic. Thus, in Carlotretus setosus, Corypholophus minutus and Opisotretus kraepelini, the type species of their respective genera, the ♂ has 19 segments, whereas ♀♀ are unknown. Regrettably, Chamberlin (1945) did not care to mention the number of body segments in the ♀♀ of his Opisotretus mimus and O. euthus. Opisthoporodesmus species, perhaps including also O. bacillifer which was described from presumably subadult ♀♀ with 19 segments (Carl 1912), show equally 20 segments in both sexes. All of the remaining genera and species seem to have 19 segments in the ♂ versus 20 segments in the ♀.

Number of rows of setae on body metaterga.
Only two known species show two transverse rows on the metaterga: Carlotretus setosus and Corypholophus ryukyuensis, as opposed to the other species which clearly have three transverse rows of bacilliform setae, these sometimes being evidently shifted caudad. However, this character appears to be only species-specific, as one of the new species of Martensodesmus described below also has only two rows of tergal setae.
The pattern of metatergal sculture is that typical of the Polydesmidea, i.e. three transverse rows of polygonal bosses, with a more or less deep sulcus separating the first row from the two following ones. Each boss is typically surmounted by a seta sometimes borne on a small knob, the pattern being 3+3 per row (see above). In Opisotretidae, only few species show very distinct bosses, like those observed in Solaenaulus butteli (Figs 4,5) or Opisotretus beroni sp. n. (Figs 17A-F), whereas in most species the bosses tend to be poorly visible to virtually untraceable (Figs 24A-F), whereas the transverse sulcus is largely superficial.
Location of ozopores.
The location of ozopores is often quite peculiar in species of Opisotretidae. The pore formula always being normal, 5, 7, 9, 10, 12, 13, 15-18(19), the ozopores are normally placed near the caudolateral corner of paraterga, very to quite close to the caudal margin of the tergite (Figs 1A, 5B, G, 8A, B, 11A, B, 14B, G, K, 17C, L, 22C, E, M, 24B, C, K, 26B, C, 30B, C, E, F, K, 31B, 33B, C, F, 34B, 36, 37B, E, F, 39B-E, 40B). Chamberlin (1945) paid special attention to this character when assigning his species to either Opisotretus or Opisthoporodesmus, or Retrodesmus. In particular, he tended to treat all species with ozopores placed especially close to the caudal metatergal margin in Opisotretus, apparently following Attems (1907), whereas the species with the ozopores slightly more strongly removed forward from the caudal margin he placed in Opisthoporodesmus. In the type species of Retrodesmus, the ozopore is well removed from the caudal corner of paraterga, lying closer to their lateral margin, but in a new congener described below the ozopores lie just at the caudal margin, this being more typical of the family.
However, this feature must be admitted as not being unique to and characteristic of some Opisotretidae alone. Thus, many species of Fuhrmannodesmidae possess ozopores which also flush open at the caudolateral corner of poriferous paraterga quite close to very close to the caudal margin, in South America (Golovatch 1994) and Vietnam (Figs 42C,F). In general, this condition strongly depends on the degree of development of paraterga which varies between species, as well as between segments. The more strongly the caudal corner of a paratergum is drawn caudad, this being increasingly marked towards the telson, the closer the ozopore to the caudal margin.
In other words, the distinctions based on ozopore location are species-specific at most and clearly fail to characterize opisotretid genera.
The presence of especially prominent shoulders seems to correlate positively with a shift caudad of the transverse rows of tergal setae. In such species, the frontal row of setae is situated close to the metatergite's midway sulcus, whereas both following rows are considerably shortened in extent, strongly shifted to the caudal margin of the tergite and placed very close to each other (Figs 11A, B).
The ♂ vertex of several Opisotretidae is modified. In particular, Martensodesmus, among other things, was first distinguished by ♂ vertigial modifications usually traceable as humps or tubercles above the antennal sockets (Golovatch 1987). Later this feature had to be abandoned as a generic-level character after the discovery of M. excornis which lacks any such modifications. Moreover, sexual dimorphism in M. bicuspidatus was found to concern not only a complex structure (a fossa with two cusps of filaments) on the ♂ vertex, but also the shape of the collum (Golovatch 1988). ♂ vertigial modifications are also known in Corypholophus minutus (a hump with a tuft, Fig. 2A) and two new Opisotretus described below.
However, like in the case of ozopore location (see above), similar vertigial modifications in the ♂ concern numerous species of Fuhrmannodesmidae, including those occurring in South America (Golovatch 1994) and Vietnam (Figs 42D, G).

Legs.
Variation in leg length and armament in Opisotretidae is pronounced, ranging from short and stout, sometimes also supplied with special ventral trichomes in the ♂, e.g. in Solaenaulus butteli (Fig. 5H), to extremely long and slender, e.g. in Retrodesmus cavernicola sp. n. (Figs 15B, 16A), but most species show medium-sized, moderately stout legs which are usually devoid of special trichomes in the ♂ and thus fail to differ much between the sexes. Claw length seems to vary proportionately to leg length (Figs 5H, I, 8C, 15B, 16A, 27K, 28D).
As usual in the systematics of any subgroup of Polydesmida, the gonopods offer most of the characters deemed useful, if not crucial, for the discrimination of genera and species. This fully applies to Opisotretidae as well.
As noted above, the gonopods in Opisotretidae are really unique in obviously having rather small, subglobose, medially fused and nearly fully exposed coxae, these being only very poorly sunken into an unusually small gonocoel. The gonopod aperture is invariably obcordate (Fig. 6B). The coxae support the usual cannulae medially and elongated, sometimes strongly curved telopodites laterally. The telopodites are directed dorsolaterad, curving, often very strongly, around coxae 8 along the sides of segment 7. The seminal groove runs along most of the telopodite's extent to termi-nate distally either on a special branch or tooth (= solenomere), or flush open on the surface, or debauch inside an accessory seminal chamber which normally is supplied with a hairy pulvillus.
Against this general pattern, various species and genera show several important modifications. Species in two of the genera are recognized for the presence of a very peculiar frontobasal process placed on the ventral side of the femorite: in Corypholophus minutus, this process (p) is neatly attached to a rather slender, unipartite and suberect gonotelopodite, and it carries an additional groove (Fig. 2B), whereas p in Solaenaulus is devoid of a groove, being well separated from a strongly unciform, bipartite telopodite beset with bacilliform structures distally (Figs 6A, C-F, 7). Because Corypholophus ryukyuensis has no process p (Fig. 3), its generic assignment has been questioned (Golovatch 1987). However, since a well-developed p also occurs in a new species of Martensodesmus described below, this character must be regarded as being only speciesspecific. This clearly supports maintaining Corypholophus ryukyuensis in Corypholophus, even though C. minutus has a p (Fig. 2B).
Carlotretus seems to be the only genus in Opisotretidae in which the distal part of a unipartite gonotelopodite is totally free from fringes or bacilli, including a long, erect and simple solenomere (sl) (Figs 1B,C,41). Solaenaulus, in addition to p, also shows a prominent solenomere (sl) attached closely to a similarly long exomere (ex), both the branches being curved and abundantly ornamented with bacilli (Figs 6, 7). In all other opisotretid genera and species, the solenomere is a rather small denticle or lobule at most. The gonopods of Opisotretus, of O. kraepelini at least (Fig. 8D), look very similar to those of Solaenaulus, especially as regards the unciform appearance and the distal ornamentation of the telopodite, but the latter in Opisotretus is unipartite, sometimes being also devoid even of a vestigial solenomere (Figs 26F,G). Opisthoporodesmus, at least O. obtectus, Martensodesmus and Retrodesmus share the gonopod telopodite being rather short, poorly curved and, in the former two genera, modestly ornamented distally. The gonotelopodite in Martensodesmus species is often more or less hollow or flattened on the caudal face and carries considerable lobes or processes, sometimes including p. In Opisthoporodesmus obtectus, the gonopod telopodite ( Fig. 11C) is very simple, attenuating distad and virtually fully devoid of a trichome other than the one on a subterminal hairy pulvillus. In contrast, Retrodesmus has an enlarged, bifid and elaborate tip of the gonotelopodite, one of its apical branches being beset with bacilliform ornamentations, but showing neither a solenomere nor an accessory seminal chamber, nor a hairy pulvillus.

Vulva.
No special studies have been conducted on the conformation of the vulva in Opisotretidae. Only Chamberlin (1945) depicted the vulva of Opisthoporodesmus silvestri as showing a remarkable subelliptic lobe. The epigynal crest has never been described either. Because these structures are too small and inconspicuous in the samples we have examined, they have been omitted from the descriptions.

Generic reclassification
Based on the above information, as well as facing the need to properly allocate several new species described below, we propose the following new classification of Opisotretidae.

Diagnosis.
A family of the suborder Polydesmidea with 19 (♂) or 20 (♂, ♀) segments. Body small to very small (3-16 mm long). Tegument microalveolate, limbus microspiculate. ♂ head with or without vertigial modifications. Antennae geniculate between segments 5 and 6, antennomeres 5 and 6 each with a compact group of bacilliform sensilla distodorsally. Metaterga with 2 or 3 regular, transverse rows of bacilliform, longitudinally ribbed setae sometimes borne on minute knobs; frontal margin of midbody paraterga only seldom forming clear-cut shoulders; side margin of paraterga often slightly incised, with 2 or 3 bacilliform setae. Pore formula normal, ozopores flush open dorsally, usually near to very near caudolateral corner of paraterga, only seldom clearly removed from caudal margin. Legs rather short to long, ♂ ones often stouter and longer, sometimes with peculiar, bi-or trifid ventral trichomes, but sphaerotrichomes missing.
Gonopods peculiar in having rather small, subglobose, medially fused coxae nearly fully exposed in a small gonocoel; coxae at most only slightly setose ventrally, supporting usual cannulae medially and elongated, sometimes strongly curved telopodites laterally; the latter directed dorsolaterad, curving, often very strongly, around coxae 8 along sides of segment 7; seminal groove running along most of telopodite on caudal face to terminate distally either on a special branch or tooth (= solenomere), or flush open on caudal surface, or debauching inside an accessory seminal chamber which normally, but not always, is supplied with a hairy pulvillus.
Type genus. Opisotretus Attems, 1907. Remarks. The above somatic features of Opisotretidae are in no way unique to the family, at least sometimes being also encountered, in various combinations, in the other families of the micropolydesmoid superfamily Trichopolydesmoidea, such as Fuhrmannodesmidae, Trichopolydesmidae, Macrosternodesmidae, Mastigonodesmidae and Nearctodesmidae (e.g. Golovatch 1994Golovatch , 2011. This can also be said about the basically finely microspiculate limbus obviously characteristic of most of the Polydesmidea. It is only the gonopod structure that is truly characteristic of Opisotretidae, the family formally representing a superfamily of its own, the Opisotretoidea (Simonsen 1990). Superficially, female and/or juvenile Opisotretidae are not or only barely distinguishable from the often sympatric or even syntopic female or juvenile Fuhr-mannodesmidae, the latter family dominating most of the tropical micropolydesmoid faunas. In this connection, we rather believe that Opisotretidae is also a family of Trichopolydesmoidea, probably the closest to Fuhrmannodesmidae (see below).
Gonopod telopodite rather stout, unipartite, at best slightly hollow on caudal face; distal part devoid of ornamentations (spines or setae), being a long and simple solenomere (sl) supplied with lobes or processes either subtending it or lying at its base. Neither an accessory seminal chamber nor a hairy pulvillus (Figs 1B, C, 38B-E, 40D-H, 41).

Diagnosis.
A genus of Opisotretidae with 19 (♂) or 20 (♀) body segments. ♂ head often, but not always, with modifications on vertex, collum rarely enlarged. Most of metaterga with 2 (more rarely) or 3 (more usually) regular, transverse rows of bacilliform setae. Frontolateral margin of midbody paraterga usually without evident shoulders, in any event not so strongly developed as to cause a caudad shift of the rows of tergal setae. Ozopore lying from close to, to rather far in front of caudal margin of paratergite's caudolateral corner.
Gonopod telopodite rather stout, at least basal half so, unipartite, usually only faintly curved, slightly, more usually clearly, hollow/excavate/flattened on caudal face all along; parabasal and/or distal parts with lobes or processes, sometimes including p; both accessory seminal chamber and hairy pulvillus wanting, but a very short, dentiform solenomere usually ornamented with a few bacilli-or setiform structures nearby often present (Figs 9,10C,D,29,32,35).

Diagnosis.
A genus of Opisotretidae with 19 (♂) or 20 (♀) body segments. ♂ vertex with or without modifications. Metaterga with three regular, transverse rows of bacilliform setae. Frontal margin of midbody paraterga devoid of obvious shoulders. Ozopore usually lying close to very close to caudal margin of paratergite's caudolateral corner.  Carl, 1922; ♂ holotype from Sumatra, Indonesia; A midbody segments, dorsal view B, C right gonopod and its apical part, mesal and lateral views, respectively. Depicted not to scale. After Carl (1922).
Remarks. In addition to the type species, the genus currently contains two described congeners: Opisotretus euthus Chamberlin, 1945 andO. mimus Chamberlin, 1945. Because the gonopods of O. euthus are indeed very similar to those of O. kraepelini as depicted by Chamberlin (1945), the former species is definitely congeneric with the latter one. The identity of O. mimus, however, remains uncertain, but superficially it strongly reminds of Peronorchus parvicollis Attems, 1907, a species we think belongs in the family Fuhrmannodesmidae. It was originally described from Buitenzorg (= Bogor), Java, Indonesia   Murakami, 1975, ♂ paratype from Ryukyu Islands, Japan; Ae both gonopods in situ, ventral, caudal, subcaudal and caudal views, respectively F gonopod tip, dorsal view. Depicted not to scale. After Murakami (1975). (Attems 1907) and seems to be very similar to an opisotretid in showing long bacilliform tergal setae arranged in three transverse rows, notably reduced paraterga, and the ozopores located near the paratergite's caudal corner. Interestingly, Mauriès and Geoffroy (1999), when redescribing P. parvicollis from material taken on Mauritius, Indian Ocean, assigned  (Carl, 1922), ♂ from Lae, Papua New Guinea; A habitus, lateral view B, e, h anterior body part, lateral, dorsal and ventral views, respectively C, F, i midbody segments, lateral, dorsal and ventral views, respectively D, G, J posterior body part, lateral, dorsal and ventral views, respectively. -Scale bars: A 0.5 mm; B-I 0.1 mm; J, 0.12 mm. After Golovatch et al. (2010). this genus to the family Trichopolydesmidae, as opposed to Hoffman (1980) who had left Peronorchus among the genera of Polydesmidea of uncertain status and family position.
Four new species described below also belong in Opisotretus. A key to all seven Opisotretus species, including O. mimus, is given below.

Diagnosis.
A genus of Opisotretidae with 20 body segments (♂, ♀). ♂ vertex without modifications. Metaterga with three regular, transverse rows of bacilliform setae,  but, probably in conjunction with frontolateral margin of midbody paraterga bearing prominent shoulders, at least sometimes all three rows strongly shifted caudad, last two being also abbreviated. Ozopore usually lying close to caudal margin of paratergite's caudolateral corner. Gonopod telopodite elongate, subunciform, unipartite, markedly attenuating distad; distal part with only a few small outgrowths at best, devoid of both bacilliform ornamentations and a solenomere, but supplied with both a small accessory seminal chamber and a hairy pulvillus (Fig. 11C).
Remarks. In addition to the type species, the genus currently contains five formal congeners: Opisthoporodesmus anandrus Chamberlin, 1945, O. bacillifer Carl, 1912, O. conservandus Chamberlin, 1945, O. silvestri Chamberlin, 1945and O. simplex Chamberlin, 1945. As these five species require revision and their identities remain uncertain, no key to Opisthoporodesmus species is possible for the time being.

Diagnosis.
A genus of Opisotretidae with 19 (♂) or 20 (♀) body segments. ♂ vertex without modifications. Metaterga with three regular, transverse rows of bacilliform setae, but, in conjunction with frontolateral margin of midbody paraterga bearing evident shoulders, all three rows strongly shifted caudad, last two being also abbreviated. Ozopore from well removed from, to very near caudal margin of paratergite's caudolateral incision.
Gonopod telopodite rather stout, only slightly curved, unipartite, divided only distally into a frontal stump heavily beset with bacilliform ornamentions and a simple to complex caudal branch; seminal groove terminating near base of both these branches; neither a solenomere nor a hairy pulvillus (Figs 13B, 15C, D, 16B, C), only sometimes with a visible accessory seminal chamber.
Type species. Retrodesmus dammermani Chamberlin, 1945, by original designation.  Remarks. The holotype of this species has been examined in order to shed light on the identity of both the genus and species. A new species is added as well. The differences between both are clear from Figs 12, 13 and Figs 14-16, as well as from the diagnosis of R. cavernicola sp. n. Descriptive notes and remarks. The series also contains a microvial with several fragments of a presumed ♀ labeled "♀ allotype", but, having not been mentioned in the original description (Chamberlin 1945), this ♀ cannot be considered as part of the type series.

Retrodesmus dammermani
The holotype, an intact ♂, has been restudied, with several colour pictures taken to show the habitus (Fig. 12), and line drawings executed of a midbody paratergite and the gonopods in situ (Fig. 13).  Chamberlin's (1945) succinct description is basically correct in showing quite broad and mostly slightly upturned paraterga with 2 or 3 minute, lateral, setiferous incisions; the caudal corners of postcollum paraterga until the 17 th are produced increasingly well behind the rear tergal margin, roundly dentiform; the metaterga support three rather regular, transverse rows of short to medium-sized bacilliform setae; the ozopores are located rather close to the lateral margin of ozoporiferous paraterga, but quite far from the caudal corner (Fig. 13A). Body length ca 6 mm, width 0.55 mm.
As Hoffman (2005, p. 75) once put it quite sarcastically as regards the quality of Chamberlin's (1945) paper, "There is no evidence that Professor Chamberlin invested much time in consultation of available literature sources". Nevertheless, his Retrodesmus remains a valid genus sufficiently distinct from the other opisotretid genera.
In addition to the type species, Retrodesmus also includes R. cavernicola sp. n., a presumed troglobite from Papua New Guinea. Gonopod telopodite elongate, subunciform, bipartite; basal process (p) on frontoventral face of femorite prominent, removed from femorite proper; distal part of telopodite usually beset with bacilliform ornamentations both over a prominent solenomere (sl) and an even more prominent exomere (ex); a small accessory seminal chamber present, but a hairy pulvillus absent (Figs 6A, C-F, 7).
Remarks. In addition to the type species, the genus currently contains only one known species: S. birmanicus Carl, 1941(Golovatch et al. 2010, which, however, is sometimes treated as a variety of the type species (Jeekel 2006).  Diagnosis. Differs readily from R. dammermani Chamberlin, 1945, the only other known species of Retrodesmus, by the particularly broad paraterga, several clearly troglomorphic features such as especially long and slender antennae, legs and metatergal setae, the latter also being very dense, the subcaudal position of the ozopores, and the shape and ornamentation of the gonopod apex.

Descriptions of new species
Name. To emphasize the obvious troglomorphic traits strongly suggesting obligate cave-dwelling; a noun in apposition.
Gonopod aperture evident, transversely oblong-oval, taking up most of ventral part of metazonite 7. Gonopods (Figs 15C, D, 16B, C) with globose, medially fused coxae carrying a few setae on ventral face and a normal cannula mesally. Telopodite nearly straight, unipartite, rather short and stout. Distal part of telopodite split into a shorter frontal stump (s) (= solenomere?) beset with bacilliform ornamentations and  a quite complex, subtriangular, pointed, caudal branch (b) with a spine (sp) at base. Seminal groove terminating near base of both s and b, with neither a distinct solenomere nor a hairy pulvillus, but with a small accessory seminal chamber.

Remarks.
Because of several apparent troglomorphic traits, this species seems to be a troglobite. Surprisingly, it appears to be rather widespread in western Papua New Guinea, occurring in places like Finim tel and Goglme which are separated from each other by >200 km.
It is the gonopod conformation, not the location of the ozopores, that clearly indicates the true affinities of R. cavernicola sp. n. to R. dammermani, despite the great geographical gap between Java and New Guinea that also separates these species. Diagnosis. Differs readily from congeners by the shorter and bifid apical piece of the gonopod telopodite devoid of a spine level to a short solenomere, coupled with a deeper caudalmost incision of paraterga harbouring the ozopore in poriferous segments.
Gonopod telopodite (Figs 18B-D, 20C, D, 21) only slightly curved, unipartite, rather long and slender; apical piece (a) distal to a very short solenomere (sl) elongate, more strongly curved, clearly bifid, on caudal face with a few to several denti-or spiniform ornamentations, but devoid of a strong parabasal spine level to sl. An accessory seminal chamber at base of sl evident, crowned with a hairy pulvillus.
Remarks. The gonopod structure of this new species has already been illustrated by mistake elsewhere (Golovatch et al. 2010, figs 78 & 80), in connection with documenting the record of Solaenaulus butteli in Papua New Guinea. The same drawings (Fig. 21) are reproduced here again.  Diagnosis. Differs readily from congeners by a modified ♂ head (two prominent paramedian tubercles above the antennal sockets), coupled with a less strongly curved, nearly suberect gonopod telopodite, with its apical piece crowned with several peculiar, mostly digitiform outgrowths.
Name. Referring to the type locality; a noun in apposition. Description. Length of holotype (and of subadult ♀ paratypes) ca 9 mm, width of midbody pro-and metazona 0.8 and 1.15 mm, respectively. Coloration in alcohol from uniformly pallid to light yellowish.
Gonopod telopodite (Figs 23C, D) only very slightly curved, nearly suberect, unipartite, rather long and slender; apical piece (a) distal to a very short solenomere (sl) rather short, on caudal face with a few finger-shaped ornamentations and a strong, parabasal, subspiniform process (pr). An accessory seminal chamber at base of sl evident, crowned with a hairy pulvillus.
Remarks. This is the first Opisotretus showing ♂ head modifications, thus confirming the character as being only species-specific (Golovatch 1988). Paratype. 1 ♀ (SEM; MNHN JC 341), same locality, together with holotype. Diagnosis. Differs readily from congeners both by tergal sculpture and lateral paratergal incisions being rather poorly developed, coupled with the presence of a short solenomere and a peculiar ornamentation in the apical piece (a) of the gonopod telopodite.
Name. Honours Louis Deharveng, the collector. Description. Length of holotype ca 9 mm, width of midbody pro-and metazona 0.7 and 1.0 mm, respectively. Coloration in alcohol uniformly pallid.
Gonopod telopodite (Fig. 25) clearly curved, unipartite, long and slender; apical piece (a) distal to a short solenomere (sl) rather long due to a terminal uncus (u) bearing near its base a strong subcaudal spine (sp) and a short field of subspiniform, mostly curved ornamentations. An accessory seminal chamber at base of sl evident, crowned with a hairy pulvillus.
Remarks. This is the first formal encounter of an Opisotretus species in Sulawesi, Indonesia. Due to its long legs and uncoloured tegument, O. deharvengi sp. n. is likely to represent a troglobite. Opisthoporodesmus bacillifer, the only other opisotretid known from Sulawesi, differs readily in having only two, not three, lateral incisions on the paraterga (Carl 1912). Name. To emphasize the highly spinose apical piece of the gonopod telopodite. Description. Length of holotype ca 4 mm, width of midbody pro-and metazona 0.3 and 0.5 mm, respectively. Length of paratype ca 5 mm, width of midbody pro-and metazona 0.45 and 0.6 mm, respectively. Coloration in alcohol uniformly pallid.
Gonopod telopodite (Figs 26F, G) clearly curved, unipartite, long and slender; apical piece (a) distal to a vestigial solenomere strongly curved due to a terminal uncus (u) bearing near its base a strong subcaudal spine (sp) and a field of spiniform orna- mentations. An accessory seminal chamber at base of solenomere rather evident, but probably devoid of a hairy pulvillus.
Remarks. This new Opisotretus species has been taken together with several immature females (18 segments) of a different, somewhat larger and slightly pigmented (reddish metaterga and antennae) opisotretid with somewhat broader paraterga and a different location of the ozopores (these being placed close to the caudal tergal margin) which could not be identified in the absence of adult male material. Diagnosis. Differs readily from congeners by the missing modifications on the ♂ vertex, coupled with quite well developed shoulders on metaterga, the high and broad paraterga, as well as the presence near the gonopod telopodite's midpoint of three strong spines proximally to a considerably attenuating acropodite.

Martensodesmus cattienensis
Name. To emphasize the type locality. Description. Length of holotype ca 5.5 mm, width of midbody pro-and metazona ca 0.8 and 1.0 mm, respectively. Length of ♂ paratypes ca 5.5-6.0 mm, width of midbody pro-and metazona ca 0.75-0.8 and 1.0-1.1 mm, respectively. Length of ♀ paratypes ca 7.0-8.0 mm, width of midbody pro-and metazona ca 0.9-0.95 and 1.1-1.15 mm, respectively. Coloration in alcohol from nearly uniformly light yellowish to head and several anterior segments slightly infuscate, light yellow-brown, more rarely with a rusty reddish tint.
Gonopod telopodite (Figs 28E-I, 29) clearly curved, but stout, unipartite; basal half voluminous and supplied with three strong spines, distal half gradually attenuating, apical piece (a) distal to a vestigial solenomere with a number of short spinules. Neither bacilliform ornamentations nor an accessory seminal chamber, nor a hairy pulvillus.
Diagnosis. Differs readily from congeners by the presence of a well-developed frontobasal process p on the ventral side of the gonopod femorite, coupled with no modifications on the ♂ vertex.
Name. Honours Anne Bedos, one of the collectors. Description. Length of holotype and ♂ paratypes ca 4.0 mm, width of midbody pro-and metazona ca 0.4 and 0.6 mm, respectively. Length of ♀ paratypes ca 5.5 mm, width of midbody pro-and metazona ca 0.6-0.65 and 0.8-0.9 mm, respectively. Coloration in alcohol from nearly uniformly pallid to head, collum and following metazona (especially their caudal halves) light rusty brown, anterior body portion in ♂♂ being clearly more infuscate, rusty brown, compared to ♀♀.
Diagnosis. Differs readily from congeners by the presence of only two transverse rows of setae on metaterga, combined with five strong spines on the caudal face of a rather strongly curved gonopod telopodite which lacks even traces of a solenomere.
Name. To emphasize the highly spinose gonopod telopodite. Description. Length of holotype ca 4.0 mm, width of midbody pro-and metazona ca 0.35 and 0.5 mm, respectively. Length of paratype ♂ ca 4.5 mm, width of midbody pro-and metazona ca 0.4 and 0.6 mm, respectively. Length of adult ♀ ca 4.7 mm, width of midbody pro-and metazona ca 0.5 and 0.6 mm, respectively. Coloration in alcohol from uniformly pallid to head, several anterior segments and following metaterga clearly infuscate, rusty reddish, increasingly poorly pigmented towards telson.
Gonopod telopodite (Fig. 35) clearly curved, but stout, unipartite; apical piece (a) distal to orifice of seminal groove complex, consisting of a very strong apical spine (sp) protected in its basal half by a membranous ventral velum (ve) with a faintly fringed apical margin and a similarly membranous, apically spinigerous, dorsal lobe (k); caudal face below a with three distinct spines (x, y and z), z being longest. Neither bacilliform ornamentations, nor accessory seminal chamber, nor even traces of a solenomere, nor a hairy pulvillus.
Diagnosis. Differs readily from C. setosus, the only known congener, by the much longer, strong and totally unprotected solenomere branch, whereas the parabasal branches are slender and subunciform.  Name. To emphasize the clearly triramous midlength process of the gonopod telopodite.
Description. Length of holotype ca 4.3 mm, width of midbody pro-and metazona ca 0.4 and 0.55 mm, respectively. Length of paratype ♂ ca 4.6 mm, width of midbody pro-and metazona ca 0.45 and 0.6 mm, respectively. Length of adult ♀ ca 6.0 mm, width of midbody pro-and metazona ca 0.6 and 0.7 mm, respectively. Coloration in alcohol from uniformly pallid to head and metaterga faintly rusty reddish.
Gonopod telopodite (Figs 38B-E, 40D-H, 41) clearly curved, but its basal half quite stout, unipartite; solenomere (sl) very long, slender and simple, only faintly curved, orifice of seminal groove placed on a small subapical tooth, with neither bacilliform ornamentations, nor accessory seminal chamber, nor a hairy pulvillus. Two large, subunciform processes, m and n, at base of sl, process n lying more dorsally and being slightly larger than a ventral, very finely and densely microspinulate m.  -Gonopod telopodite usually either clearly curved when elongate and slender or nearly straight at least in basal half when thick and stout, often with various evident outgrowths, sometimes also with bacilliform ornamentations distally.......5 5 Gonopod telopodite (at least its basal half) rather stout, its basal and/or distal parts with lobes or processes, sometimes including a p; both accessory seminal chamber and hairy pulvillus wanting, but a very short, dentiform solenomere usually ornamented with a few bacilli-or setiform structures nearby often present (Figs 9, 10C

evolutionary and biogeographic implications
The family Opisotretidae appears to range from the Ryukyu Islands, Japan and southern China, through Indochina and Indonesia, to Papua New Guinea. Simonsen (1992) dates the origin of Opisotretidae back to the Triassic (220 Ma) in relation to the geological history and fragmentation of eastern Gondwanaland, placing it mainly in the region of present-day Indonesia. However, such a reconstruction fails to account for the presence of opisotretid species all over Indonesia, including Sulawesi and New Guinea, only parts of which might have been derived from Gondwanaland. Even less convincing is the attribution of such remote and outlying areas as the Ryukyus and southern China to Gondwanaland. Thus there is no compelling reason to suppose that the evolution of Opisotretidae is linked to Gondwanaland. It seems far more logical to consider Opisotretidae as a purely Oriental family, probably the sister-group to, or even a disjunct offshoot of, the pantropical family Fuhrmannodesmidae. Opisotretids might well have originated in mainland Southeast Asia, with subsequent dispersals along the Himalayas in the West and across Indonesia (including New Guinea) in the East. Such patterns do not need to be dated as far back as the early Mesozoic, but could have instead become established much more recently, well within the Cenozoic. Moreover, Southeast Asia hosts the largest number of higher diplopod taxa, including all sixteen Recent orders of the class, and it provides many other examples, at various taxonomic levels, of connections with the Himalayas and Southeast Asian expansions to Indonesia/ East Indies (Shelley & Golovatch 2011).
In the present work we place Opisotretidae in the superfamily Trichopolydesmoidea, as recently reviewed by Golovatch (2011), considering them to be close to, if not immediately derived from, the still poorly defined, highly diverse and mostly pantropical family Fuhrmannodesmidae. The reasons for this view of the relationships between them are as follows. Golovatch (1994) provided an evolutionary scenario for the genera of Fuhrmannodesmidae known from the Neotropical realm, accepting as the basalmost those genera showing rather small, subglobose gonopod coxae that form no significant gonocoel in which to hinge the largely exposed, usually rather simple and elongate telopodites. Moreover, as in some true Trichopolydesmoidea, the prefemoral (= setose) part of the gonopod is mostly orientated transversely to the body axis, extending mesally across the entire width of the coxae. Following a series of transitional states, such forms ultimately culminate in having the gonopod coxae strongly enlarged, forming a large gonocoel in which to conceal the clearly shortened, usually highly complex and deeply sunken telopodites. Their prefemoral parts already tend to be positioned increasingly parallel to the body's main axis, thus providing a transition between the usually small-sized Trichopolydesmoidea (= socalled "micropolydesmoids") to the normally medium-to large-sized Polydesmoidea (= so-called "macropolydesmoids"). In this respect, Fuhrmannodesmidae might well prove to be a paraphyletic group.
Naturally, similar general trends can be surmised to have occurred in the fuhrmannodesmids of the Afrotropical and, especially, Oriental realms, which support fairly diverse faunas of this family.
Based on the published record and the available collections of Fuhrmannodesmidae (largely kept at MNHN) from Southeast Asia and adjacent regions, including southern China, Indonesia and Melanesia (Golovatch et al., in preparation), most of the Oriental Fuhrmannodesmidae are indeed highly advanced, showing complex and strongly shortened gonopod telopodites deeply sunken inside the gonocoel formed by enlarged coxae. The orientation of the prefemoral part also varies, but it tends to be held parallel, not transversely, to the main axis of the body. However, amongst the Asian fuhrmannodesmids, there are certain genera and species that instead show quite primitive conditions, i.e. long, usually less complex gonopod telopodites that remain strongly exposed above a relatively small gonocoel. At least some of these have the prefemoral parts orientated strictly parallel to the main axis, a condition typical of the sister-superfamily Polydesmoidea. Moreover, one of these species (Figs 42, 43), from Vietnam, yet to be described, shows the distal parts of the gonopod telopodites elongated and directed laterad. This is in contrast to the much more frequent condition of the gonotelopodites in Fuhrmannodesmidae and most other groups of Polydesmida, which either cross mesally or are held parallel to each other. Based on this example, the evolution of the Opisotretidae might well be viewed as a case when the gonotelopodite of fuhrmannodesmids grows increasingly elongate and orientated laterally, while the coxae remain rather small and do not form a significant gonocoel. Against this back-ground, we would again emphasize that none of the peripheral, non-gonopod features of Opisotretidae seems to characterize this family alone.
To summarize, the Opisotretidae could have originated directly from a disjunct member of Fuhrmannodesmidae in which the gonopod coxae would have remained small and probably been held parallel to the main axis of the body, but the development of the telopodite would have culminated in its considerable elongation and fully dorsolaterad orientation. In addition, the gonopods of this stem species must have been equipped with an accessory seminal chamber and a hairy pulvillus, the apomorphies which are absent from present-day Fuhrmannodesmidae, but retained in most of the Polydesmoidea and a few Trichopolydesmoidea, including Opisotretidae. So the ancestor of Opisotretidae might have also been a species of Polydesmoidea or even a common stem member of the Polydesmoidea+Trichopolydesmoidea. A clearcut trend to having these apomorphies (as well as ♂ sphaerotrichomes) reduced is evident, apparently along with body miniaturization, not only in Opisotretidae, but also in some typical Polydesmidae, e.g. within the Siberian genus Uniramidesmus Golovatch, 1979(Mikhaljova 2004, and Trichopolydesmidae, e.g. within the western Mediterranean genus Cottodesmus Verhoeff, 1936(Verhoeff 1936, Mauriès and Vicente 1977, Mauriès 1983.

Conclusions
There can be no doubt that many more species of Opisotretidae await discovery and description. Representatives of this family seem to be rare, but it is likely that they are seriously under-collected due to their small size. Even in caves, they seem never to be abundant, making them easy to overlook, which is in strong contrast to most diplopod groups common in tropical caves, including those of the Oriental realm (Deharveng & Bedos 2012). Sympatry, even syntopy, of two different opisotretid species also seems to be common.