The first representatives of the millipede family Glomeridellidae (Diplopoda, Glomerida) recorded from China and Indochina

Abstract A new species of glomeridellid millipede is described from Guizhou Province, southern China: Tonkinomeris huzhengkunisp. nov. This new epigean species differs very clearly in many structural details, being sufficiently distinct morphologically and disjunct geographically from T. napoensis Nguyen, Sierwald & Marek, 2019, the type and sole species of Tonkinomeris Nguyen, Sierwald & Marek, 2019, which was described recently from northern Vietnam. The genus Tonkinomeris is formally relegated from Glomeridae and assigned to the family Glomeridellidae, which has hitherto been considered strictly Euro-Mediterranean in distribution and is thus new to the diplopod faunas of China and Indochina. Tonkinomeris is re-diagnosed and shown to have perhaps the basalmost position in the family Glomeridellidae. Its relationships are discussed, both morphological and zoogeographical, within and outside the Glomeridellidae, which can now be considered as relict and basically Oriental in origin. Because of the still highly limited array of DNA-barcoding sequences of the COI mitochondrial gene available in the GenBank, the first molecular phylogenetic analysis of Glomerida attempted here shows our phylogram to be too deficient to consider meaningful.


Introduction
The chiefly Holarctic millipede order Glomerida (Shelley and Golovatch 2011) is currently known to comprise only three families: Glomeridellidae Cook, 1896, Protoglomeridae Brölemann, 1913, and Glomeridae Leach, 1815(Enghoff et al. 2015. The family Glomeridellidae presently contains only two accepted genera: Glomeridella Brölemann, 1913, with seven or eight species from Spain, France, the eastern Alps, and the northwestern Balkans, and Typhloglomeris Verhoeff, 1898 (= Albanoglomus Attems, 1926, synonymized by Golovatch (2003), with ca 15 species, many troglobionts, from the Balkans, Caucasus, and Near East. The family has hitherto been considered strictly Euro-Mediterranean (Enghoff et al. 2015). The ranges of both Glomeridella and Typhloglomeris overlap only marginally in the northern Dinaric Mountains, Balkans. Makarov et al. (2003) delimited several species groups within Typhloglomeris and, based on morphological evidence alone, outlined the main trends of their evolution, both morphological and ecological (= shifts to geo-or cavernicoly).
Continental China, unlike the Nearctic + Southeast Asia + Taiwan which contain several genera of Glomeridae (11) and Protoglomeridae (1) (Enghoff et al. 2015;Nguyen et al. 2019), has heretofore been known to support only numerous (32) species of a single genus, Hyleoglomeris Verhoeff, 1910, family Glomeridae (Golovatch and Liu 2020). This genus presently contains 100+ species ranging from the Balkans in the west, through Greece, Anatolia, Caucasus, and Central Asia, to Korea, Japan, and Taiwan in the east and to Indochina, Indonesia (Sumatra, Java, Borneo, and Sulawesi) and the Philippines in the southeast (Golovatch et al. 2006;Enghoff et al. 2015;Nguyen et al. 2019). One species has recently been described from Eocene Baltic amber (Wesener 2019).
All the more interesting is the discovery of a new species of Tonkinomeris in southern China. Moreover, this genus appears to actually belong to the family Glomeridellidae, being formally transferred therein from Glomeridae where it was originally placed. This represents the first formal records of glomeridellids not only in China, but also in entire Asia east of Hyrcania (the Republic of Azerbaijan and northeastern Iran near the Caspian Sea coast). The present paper is devoted to a description of our new species and to a discussion of its morphological, molecular and zoogeographical affinities.

Morphological analysis
The underlying material was taken from leaf litter in a protected forest patch and preserved in 95% ethanol. The types are deposited in the Zoological Collection of the South China Agricultural University (SCAU), Guangzhou, Guangdong Province, China. A detailed examination of characters and dissections were performed using a Leica S8 APO stereomicroscope. Line drawings were prepared with a Zeiss Axi-oskop40 microscope with an attached camera lucida. Photographs of specimens were taken with a Keyence VHX-5000 digital microscope and edited using Adobe Photoshop CS6. The terminology used here largely follows that of Golovatch and Turbanov (2018), with only a few modifications.

DNA extraction and sequencing
Genomic DNA was extracted from legs and thoracic tissue of the paratype with Qiagen DNeasy Blood and Tissue kit following the manufacturer's extraction protocol. Fragments of the COI gene were amplified using the degenerate primer pair HCO2198-JJ (AWACTTCVGGRTGVCCAAARAATCA) / LCO1490-JJ (CHACWAAYCATAAA-GATATYGG) (Astrin and Stüben 2008). The PCR amplification was performed using a T100™ thermal cycler (BIO-RAD) with a final reaction volume of 25 μL. In addition to the new nucleotide sequence in this study, MT522013, 34 Glomerida and nine non-Glomerida sequences (consisting of four Sphaerotheriida, three Polyxenida and two Polydesmida species) as outgroups were downloaded from the GenBank. All analysed species, Genbank accession numbers and voucher numbers/taxonomy ID were listed in Figure 5.

Phylogenetic analyses
The sequences were aligned using Clustal W and edited in Bioedit (Hall 1999). The final aligned dataset included 44 COI sequences with 656 positions. Bayesian Inference (BI) analysis was implemented through the on-line CIPRES Science Gateway V.3.3 (Miller et al. 2010). The BI analysis was performed by MrBayes 3.2.6 using the Markov chain Monte Carlo technique (MCMC) (Ronquist et al. 2012). The numbers of generations used amounted to 5,000,000 in the parameters for MCMC. The type of a consensus tree was chosen for all compatible groups. Maximum likelihood (ML) analysis was conducted using IQ-TREE web server (Trifinopoulos et al. 2016) with 1,000 bootstrap replications and under the GTR+G+I model (Pende et al. 2014).

Taxonomy
Considering the new species described below, the following amended diagnosis of Tonkinomeris can be proposed.  (Nguyen et al. 2019).

New diagnosis.
A genus of Glomeridellidae with the caudal margins of several ♂ tergites sometimes modified into small lobes drawn posteriad into small lobes; the caudal margin of the ♂ pygidium is clearly emarginate centrally; the anterior telopods are flattened sagittally, somewhat incrassate, with evident mesal outgrowths on either T3 alone or both T2 and T3; posterior telopods with a trichotele (sometimes rudimentary) on T1, each of T2 and T3 with a caudal process and both forming a rather indistinct apical pincer.  (Nguyen et al. 2019), by the larger body size (> 11 mm vs 9.6 mm), the more numerous ommatidia (at least 6+1 vs 5+1), the glabrous, but caudomedially produced posterior margins of ♂ tergites 8-11 (vs unmodified in both sexes), the vivid, peculiar, sexually dimorphic colour pattern (vs even more vivid, but the same in both sexes), and some structural details of the telopods: the much higher central lobe and the much shorter horns of ♂ syncoxite 19 (vs the opposite), the shape and armament of both telopod pairs, the less strongly reduced, 4-segmented ♂ telopodites 17 (vs 2-segmented), etc. (see Discussion below).

Comparative morphology and systematics
Originally, Tonkinomeris was described in the family Glomeridae Leach, 1815, tentatively assigned to the subfamily Haploglomerinae Mauriès, 1971, and compared to the genus Peplomeris Silvestri, 1917, with two species from northern Vietnam (Golovatch 1983; Nguyen et al. 2019). Both Tonkinomeris napoensis and T. huzhengkuni sp. nov. are very similar and are also sufficiently close geographically. They can easily be distinguished by a good number of morphological characters: both show vivid colour patterns (apparently, because both are epigean and fairly large), the caudal margin of the ♂ pygidium is clearly emarginate centrally (yet with no evident paramedian tubercles), T3 is somewhat incrassate and sagittally flattened, the anterior telopod (♂ leg 18) is supplied with a blunt apico-mesal tooth, there is an elongate, subcylindrical, and suberect posterior telopod (♂ leg 19) which features T1 with a mesal trichostele, each of T2 and T3 have a distinct distocaudal process, etc. Furthermore, the telopodites of the posterior telopods are not only 4-segmented and supplied mesally with a trichostele on T1, but they also show a small caudal process on T3 in addition to a stronger caudal process on T2; thus, T3 is well developed, fully functional, and its apical part forms a kind of underdeveloped pincer together with T2. All this allows us to relegate Tonkinomeris from Glomeridae to the family Glomeridellidae. Moreover, as the apical pincer on the posterior telopod seems to be a little better developed in T. huzhengkuni sp. nov. than in T. napoensis, this pincer in the remaining Glomeridellidae may be considered as being clearly apomorphous. This contradicts the views of Oeyen and Wesener (2015) to regard the Glomeridellidae as the basal family of the order Glomerida, better agreeing instead with their later cladistic analysis (Oeyen and Wesener 2018).
In addition, like most species of Typhloglomeris, the caudal margins of a few ♂ tergites in front of the pygidium in Tonkinomeris huzhengkuni sp. nov. are modified, each drawn medially posteriad into a small, albeit glabrous, lobe (thus, clearly apomorphous), vs remaining simple and unmodified (plesiomorphous) in T. napoensis. In contrast, the particularly strongly reduced, 2-or 3-segmented ♂ telopodites 17 in T. napoensis definitely represent an apomorphous condition compared to the usual, 4-segmented ♂ telopodites 17 observed in T. huzhengkuni sp. nov. and most other Glom-erida. The presence of a sharp caudomesal tooth also on T2 of the anterior telopod, vs its absence from T. napoensis, is difficult to polarize in terms or apo-or plesiomorphy. However, the particularly strongly developed central syncoxital lobe and the especially small syncoxital horns, as well as the rudimentary trichostele on T1 of the posterior telopods, all observed in T. huzhengkuni sp. nov. as opposed to their more usual states in T. napoensis, seem to be apomorphous. Therefore, each of the species combines both apo-and plesiomorphies in a number of traits. Most of the characters seem to be more advanced (apomorphous) in T. huzhengkuni sp. nov. compared to T. napoensis, but a few others vice versa (e.g., the more strongly reduced ♂ legs 17). What appears evident in any case is, that overall Tonkinomeris seems to represent the most primitive, perhaps even the basalmost genus of Glomeridellidae. This is primarily because both T. napoensis and T. huzhengkuni sp. nov. still show very modest modifications of the ♂ tergites and pygidium, while their posterior telopods feature a trichostele on T1 and yet underdeveloped apical pincers formed by T2 and T3.
Glomeridae, in contrast to Glomeridellidae, are distinct in the posterior telopods (♂ legs 19) typically being stouter, clearly curved mesad, by themselves forming a strong pincer, some telopoditomeres before last one showing a mesal trichostele or its vestige, while each telopodite is devoid of clear-cut apical pincers. The Protoglomeridae seems to be a polyphyletic group (Oeyen and Wesener 2015), only superficially being similar to Glomeridae; sometimes their tergite 11 is fused to the pygidium (still retaining a suture), while the posterior telopods are even stouter, devoid of trichosteles, both T2 and T4 form a distinct pincer by themselves (T3 being strongly developed), while each telopodite is with an additional apical pincer due to T2 and T4 (e.g., Mauriès 1971). Among the Glomeridellidae, however, the posterior telopods are usually contrasting elongate, slender, suberect, each telopodite forming a more or less distinct pincer due to modified T2 and T4 or T2 and T3 (Attems 1926;Mauriès 1971;Enghoff et al. 2015). Within Typhloglomeris, the genus deemed both morphologically and geographically closest to Tonkinomeris, the pincers on the posterior telopods in most species are formed by T2 and T3=4, when the real T3 is completely suppressed, or by T2 and T4, when T3 is strongly reduced to a short, rudimentary, non-functional, but still visible, albeit sometimes incomplete, ring. In contrast, the pincers in Tonkinomeris tend to be somewhat underdeveloped and peculiar in showing a small caudal process on T3 in addition to a stronger caudal process on T2, with T3 being fully developed and functional. This definitely represents a plesiomorphy, perhaps even the basalmost state whence a gradual reduction of both T3 and a trichostele on T1 is traced within some more advanced Glomeridellidae like Typhloglomeris and Glomeridella (cf. Attems 1926;Mauriès 1971).

Phylogeny
The phylogeny of Glomerida as recently recovered by Oeyen and Wesener (2018), based on morphological evidence alone, shows that both Glomeridella and Typhloglomeris cluster together with some Protoglomeridae and thus form no clear-cut family Glomeridellidae. Moreover, the joint Protoglomeridae + Glomeridellidae clade is not too basal on the tree and, thus, better agrees with our views that the Glomeridellidae is best considered as one of the relatively advanced groups of Glomerida.
The molecular sequences available in the GenBank and used in our phylogenetic analysis, which is apparently the first to be attempted for the entire order Glomerida, have allowed for two phylograms to be obtained. Since both BI and ML trees are similar and neither is congruent with the morphology-based phylogeny recovered by Oeyen and Wesener (2018), we present here only the BI tree (Fig. 5). Because the only genetic data available for Glomeridellidae in the GenBank are for a species of Glomeridella, and there is nothing yet for any Typhloglomeris sp., Tonkinomeris appears to cluster together with or close to two closer unidentified members of Glomeridae, one of which seems to be a Hyleoglomeris sp. (Wesener pers. comm.). Thus, there is no hint of a Glomeridellidae cluster. Instead, the whole tree (Fig. 5) is a rather random mixture of mostly genera and species of Glomeridae. This seems to indicate that any molecular analysis is bound to be too deficient and premature at this stage. It would seem especially interesting to compare Tonkinomeris to some other representatives of Glomeridellidae, especially the morphologically and geographically closest Typhloglomeris spp.

Key to genera of Glomeridellidae
The following key to the accepted genera of Glomeridellidae can be offered:

Zoogeography
Finding a glomeridellid genus in southern China and northern Vietnam is indeed remarkable, as the geographically closest record belongs to Typhloglomeris martensi (Golovatch, 1981), from Hyrcania, southwesternmost Caspian Sea coast within both the Republic of Azerbaijan and northwestern Iran (Golovatch 1981). As the huge gap between Hyrcania and Guizhou Province definitely reflects traces of former extinctions and dispersal events, this allows for the entire family Glomeridellidae to be considered both relict and of Oriental stock. Because on balance Tonkinomeris seems to be the most primitive among the glomeridellid genera, this also allows us to suggest some ancient, generally northwestward dispersal events from the Oriental realm to the Mediterranean area via southern China. Interestingly, in certain respects the relatively more advanced T. huzhengkuni sp. nov. looks like the remain of a stepping-stone in Guizhou Province, China; this is also quite far west of the overall more primitive T. napoensis from Vietnam, near the family's presumed Oriental roots. The above picture not only so considerably extends the known distribution area of Glomeridellidae to the east, but it also demonstrates the extent to which the millipede fauna of China is still understudied, as well as the possible roles that the Sino-Himalayan (= southern Chinese) and/or Oriental faunogenetic centres could have played in the origins of the Euro-Mediterranean diplopod fauna (Golovatch and Martens 2018;Golovatch and Liu 2020). Such a distribution pattern strongly resembles that of Hyleoglomeris, one of the largest, diverse, and widespread genera of Glomeridae and Glomerida (see above).
More information is necessary, especially phylogenetic reconstructions, in order to assess the remarkable disjunction of the Glomeridellidae and both its biological and spatial evolution. Further conclusions must be deferred until more evidence, both morphological and molecular, becomes available. New Glomerida are still being actively found and described from various places in Asia!