A mountain of millipedes X: Species of Pyrgodesmidae and Cryptodesmidae in the Udzungwa Mountains, Tanzania (Diplopoda, Polydesmida)

. Three new species of the family Pyrgodesmidae Silvestri, 1896 from the Udzungwa Mountains are described: Cryptocorypha geminiramus sp. nov., Cryptocorypha cactifer sp. nov., and Cryptocorypha exovo sp. nov. Several additional, unidentifiable pyrgodesmids from the Udzungwa Mountains are recorded. Elythesmus enghoffi Hoffman, 1978 (Cryptodesmidae Karsch, 1880), is recorded from the East Usambara (type locality), West Usambara, Uluguru, Nguru (Kanga) and Udzungwa Mountains, and its variation in size and body ring numbers is analyzed. The surface sculpture of pro- and metazonites is discussed. and Cryptodesmidae in the Udzungwa Mountains, Tanzania (Diplopoda, Polydesmida). European Journal of Taxonomy 845: 66–100.

Transforming this extensive characterization into a differential diagnosis of Cryptocorypha is not possible, not the least considering the high degree of variability in gonopod structure among the included species. The genus obviously belongs to the "Pyrgodesmidae cryptodesmiformes", an informal grouping named by Mauriès & Maurin (1981) for several genera of Pyrgodesmidae with a superficial resemblance to species of Cryptodesmidae. Neither Mauriès & Maurin (1981) nor any subsequent author has provided a more strict diagnosis of this grouping or an account of its content. As examples of cryptodesmiform pyrgodesmids Mauriès & Maurin (1981) mentioned the genera Dusboscquiellina Brolemann, 1926 andCachania Schubart, 1955. Based on available literature (Brolemann 1926, Schubart 1955, Golovatch & VandenSpiegel 2014, Cryptocorypha differs from Dusboscquiellina by the unmodified third male femur (tuberculate in Dusboscquiellina) and from Cachania by the lobulated / striate collum (neither lobulated nor striate in Cachania).
While Cryptocorypha thus remains poorly diagnosed, it serves as a useful temporary 'home' for several Afrotropical and Oriental cryptodesmiform pyrgodesmids. Future revisionary and phylogenetic studies will probably reshuffle the taxonomy of the species currently ascribed to the genus, as well as pyrgodesmid (and cryptodesmid, see Discussion) taxonomy in general. Golovatch (2019) gave a provisional key to the 25 Cryptocorypha species recognized at that time, and Likhitrakarn et al. (2019) gave details about the species known to them.

Diagnosis
Differs from all other species of Cryptocorypha Attems, 1807, except C. chenovi Golovatch et al., 2013 from Vanuatu, by its small size (max. diameter of males 0.72 mm), vs 0.78-0.92 mm in the congeners from Udzungwa and min. 1.00 mm in other non-Udzungwan species (the diameter of C. kumamotensis (Murakami, 1966) is unknown but the published male body length, 5 mm, suggests a diameter of > 0.8 mm). Differs from all other species of Cryptocorypha, except C. cactifer sp. nov., C. bocal Golovatch, Nzoko Fiemapong &VandenSpiegel, 2017 from D.R. Congo, andC. leia Chamberlin, 1945 from Java, by the virtually identical, very slender solenomere and exomere. Differs from C. cactifer sp. nov. by the absence of setae in podous body rings and from C. bocal and C. leia by the very poorly developed velum on the gonopod telopodite.

Etymology
The species epithet (noun in apposition) refers to the virtually identical pair of slender branches (solenomere and exomere) of the gonopod telopodite.

Description
Size. Males: Length 3.2-3.8 mm; maximal body width (across metazona) 0.59-0.72 mm in males from Kidatu; males from Mito Mitatu up to 4.6 mm long. 17 podous + 1 apodous ring + telson. Females: Length 4.6-6.1 mm. One specimen (from Kidatu, Plot 9) 6.6 mm long (0.92 mm wide). Width 0.83-1.10 mm. One specimen (from Kidatu, Plot 5) 1.36 mm wide. 18 podous + 1 apodous ring + telson. Colour. Entirely pallid or with faint brown pigment, especially on paranotal margins and on level 1 tubercles. SurfaCe SCulpture (Figs 1, 3). Entire exposed dorsal surface covered in a three-level ornamentation: Level 1: Rounded, hemispherical to subspherical tubercles of ca 35 μm diameter arranged in two or three transverse rows on all body rings. These tubercles are covered with level 3 spherules. Level 2: Very small, smooth, circular, flattened 'buttons' of ca 8 μm diameter, covering entire dorsal surface except level 1 tubercles and lateral parts of paraterga. Level 3: A dense cover of minute stalked spherules of ca 1.5 μm diameter over entire dorsal surface, including level 1 tubercles, mostly arranged in neat lines (which possibly correspond to the limits between individual cuticular microscutes), but denser and more irregularly arranged on lateral margins of paraterga. These are the "microvilli" of Likhitrakarn et al. (2019).

Females
With a low epigynal ridge (ep) on ring three (Fig. 7A). Non-sexual characters as in males. A subadult female (from Kidatu, Plot 6) is 4.5 mm long, 0.72 mm wide, 17 podous + 1 apodous ring + telson. This is the only non-adult specimen of this species seen.

Remarks
A key to species of Cryptocorypha was provided by Golovatch (2019). In the first couplet of the key Afrotropical and Asian species are separated, but the mentioned criteria concerning ring numbers are non-exclusive. If the Afrotropical alternative is chosen, C. geminiramus sp. nov. runs to C. diffusa (Brolemann, 1920) from which it differs by several characters, especially by strongly different gonopods. If the Asian alternative in the first couplet is chosen, C. geminiramus sp. nov. runs to couplet 17 (18) where it fits neither alternative.
Cryptocorypha cactifer sp. nov. urn:lsid:zoobank.org:act:F640EE01-ADBC-46DD-BF9B-0688B8A554E9 Figs 8-9 Diagnosis Identical with C. geminiramus sp. nov. in almost all characters, but differs from the smaller C. geminiramus and from all other species of Cryptocorypha by the presence of setae on podous body rings (Fig. 8). Further differs from C. geminiramus by the smooth part of head capsule extending well above antennal sockets (Fig. 8D, compare with Fig. 2A), by having the gonopod coxa (cx) more regularly hemispherical

Etymology
The species epithet (noun in apposition) means 'bearer of cacti' and refers to the some cactus-shaped metazonital setae. ( otHer CHaraCterS. Virtually identical to the smaller C. geminiramus sp. nov. with the following exceptions: All level 1 tubercles and all lateral paratergal lobes on podous and apodous rings with a club-shaped, ornamented club-shaped, somewhat cactus-like seta (Fig. 8C); smooth part of head capsule extending well above antennal sockets (Fig. 8D); gonopod coxa (cx) more regularly hemisphaerical (Fig. 9A, C); basal part of gonopod telopodite with a jagged semicircular ridge (velum?, ve), on the basal part of the gonopod telopodite (Fig. 9B, D).

Remarks
Almost all specimens of C. cactifer sp. nov. were found in mixed samples with C. geminiramus sp. nov. The single female from Udzungwa Scarp Forest Reserve fully matches the females from Udzungwa Moutains National Park, but in the light of the considerable geographical distance, combined with the absence of a male, the specimen from Udzungwa Scarp Forest Reserve is not included in the type series.
If the Afrotropical alternative in the first couplet of the key provided by Golovatch (2019) is chosen, C. cactifer sp. nov. runs to C. diffusa (Brolemann, 1920)

Diagnosis
Differs from all congeneric species by the unique shape of the 'snakelike' gonopod telopodite. Agrees with C. geminiramus sp. nov. in non-gonopodal characters.

Etymology
The species epithet means 'out of the egg' and refers to the shape of the gonopod which somewhat resembles a snake (telopodite) hatching from the egg (coxa). otHer non-gonopodal CHaraCterS (Fig. 10). Virtually identical to C. geminiramus sp. nov. gonopodS (Fig. 11). Coxa (cx) moderately globose, its surface scaly and with scattered level 3 spherules, mesally forming a shallow gonocoel. Cannula (Fig. 11C, E: ca, not seen in the other two species) with a few tiny denticles (de).Telopodite projecting strongly from gonocoel, very evident in lateral and ventral views. Basal ('prefemoral') part of telopodite (bt) massive, with several distolateral setae, giving rise to a very long, sinous shaft which slightly beyond its midlength divides into a slightly curved branch, probably corresponding to an exomere (ex), and a strongly S-shaped branch, probably the solenomere (slm). A velum is not evident.

Remarks
A male of C. geminiramus sp. nov. was found in the same pitfall trap as the unique holotype of C. exovo sp. nov.
If the Afrotropical alternative in the first couplet of the key provided by Golovatch (2019) is chosen, C. exovo sp. nov. runs to C. diffusa (Brolemann, 1920)
otHer CHaraCterS. As in the three species of Cryptocorypha described above. Notably, some of the specimens have metatergal setae, like C. cactifer sp. nov. (but the setae are smaller and less numerous), whereas others seem to lack such setae.

Remarks
In the absence of males, these specimens, which come from an area where none of the described species have been collected, are not associated with any species. In body size they best match C. cactifer sp. nov., with which some of them also share the metazonital setation.

Remarks
In the absence of a male this specimen cannot even be referred to a genus. Cryptocorypha geminiramus sp. nov. was found in the same pitfall trap.

Remarks
In the absence of a male this specimen cannot even be referred to a genus.

Other included species
None.

Diagnosis
A genus of Thelydesminae Cook, 1896 in which the prefemora are unmodified in both sexes and the surface of the metaterga is covered by a dark secretion. Hoffman, 1978 Figs 12-17

Descriptive notes
The original description (Hoffman 1978) is exhaustive. The large material studied here, including by scanning electron microscopy (SEM), allows some supplementary notes as well as an account of interspecific variation. See Figs 12-16.
Size. Males: Length 6.6-13.2 mm; maximal body width (across metazona) 1.32-2.35 mm; 17 or 18 podous rings + 1 apodous ring + telson. One male possible with only 16 podous rings, see remarks. Females: Length 9.6-16 mm, maximal body width (across metazona) 1.57-2.69 mm; 18 podous rings + 1 apodous ring + telson. The considerable size variation is in part geographical, specimens from the East and West Usambara Mountains being larger than those from elsewhere. The very extensive material from the Udzungwa Mountains suggests a correlation between altitude and body size, cf. Remarks. body ringS. Prozonites divided into two zones (Figs 13C, 14C). The anterior ca ⅔ is covered in a sculptural pattern consisting of tiny depressions delimited by fine walls, the caudal part of the walls being drawn out, each with ca five micro-ridges, each micro-ridge ending in a fine tooth (Fig. 14C). Posterior ca ⅓ unevenly covered by circular 'buttons' (Akkari & Enghoff 2011: fig . 2) resembling the level 2 structural elements described for Cryptocorypha. Metazonites covered by a brown to blackish layer or secretion which can be removed partially using a needle, but which is resistant to ultrasonic cleaning as well as digestion with trypsin or commercial enzymatic detergent. Under the secretion, the metazonital tubercles are covered by rows of hairlike processes which presumably help to keep the secretion layer in place (Fig. 14A). The long setae of which each metazonital tubercle carries one are in high magnification seen to be segmented (Fig. 14D), their apical part is flattened (Fig. 14E). The limbus consists of tongue-shaped, apically sometimes finely serrate lobes, overlaid by a row of spines of ca the same length (Fig. 14B, see also Akkari & Enghoff 2011: fig. 26). gonopodS (Fig. 16). Exactly as described by Hoffman (1978) in all studied males, irrespective of geographical origin, body size or number of podous rings.

Remarks on variation
There is a considerable size variation in the large material of E. enghoffi studied. The width of adult males varies from 1.32 mm in the smallest male (from Udzungwa Mountains) to 2.35 mm in the largest males (from East Usambara Mountains). On the whole, specimens from the Udzungwa Mountains are smaller (male width 1.32-2.04 mm), compared to 2.20-2.35 mm in males from East and West Usambara Mountains and 1.55-2.11 mm in males from Uluguru Mountains (Table 1).
Even more remarkable is the variation in ring number in adult males. Whereas adult females always have 18 podous rings (plus one apodous ring plus telson, adding up to the traditional "20 segments"), adult males may have 18 or 17, possibly even 16 podous rings. In the Udzungwa Mountains there is a geographical pattern: all examined males from the SW part of the Udzungwa (Udzungwa Scarp Catchment Forest Reserve, 659-1400 m a.s.l.) have 18 podous rings. In contrast, among males from the NE part of the mountains (Udzungwa Mountains National Park), males with 18 podous rings were found only at 1482+ m a.s.l.; all males from lower altitudes (down to 250 m a.s.l.) had 17 podous rings, but such males also occurred at altitudes as high as 1527 m a.s.l.
The sample from Kidatu, plot 9 (NHMD 621790) is particularly intriguing: out of four males, one has 18 podous rings, two seem to have 17, and one seems to have only 16 podous rings. However, this apparent case of very local variation in ring number needs verification because the males were unfortunately dissected for gonopod study before the rings were counted, so although deemed improbable, it cannot be completely excluded that one or two rings have become lost.  Figure 17 shows the relationship between podous ring number, body width and altitude in the Udzungwa Mountains.
The much smaller material from the Usambara Mountains only includes males with 18 podous rings, whereas males with 17 and 18 podous rings are present in the also quite small material from the Uluguru Mountains.

Remarks
These specimens are extremely similar to Elysthemus enghoffi, with one notable exception: the paranota are shining white, completely devoid of the dark layer which is so characteristic of E. enghoffi, and on the central part of the metazonae, there is only a faint brownish colouration. In the absence of males, the significance of this difference remains unclear.

Discussion
The Udzungwa species of Cryptocorypha Attems, 1907 The species of Cryptocorypha described here from the Udzungwa Mountains constitute a remarkable trio and highlight some general problems in pyrgodesmid / cryptodesmoid / polydesmidean taxonomy. All three species are tiny, and all share the same non-gonopodal morphology, except for the metatergal setation in C. cactifer sp. nov. The gonopods in C. geminiramus sp. nov. and C. cactifer sp. nov. are virtually identical (and differ significantly from those of all other known congeners), whereas the gonopods of C. exovo sp. nov. are radically different. Under a splitter approach, the gonopods of C. exovo sp. nov. would easily warrant the erection of a new genus, and the same might be said about the presence of metatergal setae on podous body rings in C. cactifer sp. nov.
According to the informal classification of Mauriès & Maurin (1981), Cryptocorypha belongs to the cryptodesmoid pyrgodesmids. However, C. cactifer sp. nov., might as well be grouped with the trichopolydesmid pyrgodesmids because of its extensive dorsal setation, cf. the treatment of the genus Cachania Schubart, 1955by Golovatch & VandenSpiegel (2014. The absence of metatergal setae was part of the diagnosis of Cryptocorypha by Golovatch (2019) which therefore has been modified (see under the heading "Genus Cryptocorypha Attems, 1907"). To add to the confusion, C. geminiramus sp. nov. and C. exovo sp. nov. both have a few setae on the apodous ring in front of the telson, and the females here identified as Cryptocorypha (?) sp. also have metatergal setae on podous rings.
The prozonital sculpture of Elythesmus enghoffi as described here is identical to that described for the same species by Akkari & Enghoff (2011). An almost identical pattern was found in Thelydesmus dispar Cook, 1896 ( Fig. 18) belonging to the same subfamily, Thelydesminae Cook, 1896, as Elythesmus.
The only difference between the two thelydesmines is that in Thelydesmus the caudal part of the walls separating the tiny depressions is not ridged, and the buttons on the posterior ⅓ are somewhat larger. Of other cryptodesmids of which SEM images are available, Aporodesmus sp. (Pterodesminae) and Astrolabius hoffmani Golovatch et al. 2010 (Otodesminae), lack buttons on the posterior part of the prozonite while the micro-sculpture on the anterior part resembles that seen in Thelydesmus (Akkari & Enghoff 2011: fig. 52;Golovatch et al. 2010: fig. 36). Trichopeltis spp. (Otodesminae) do have the buttons, but the sculpture on the anterior ⅔ is different (Golovatch et al. 2010: figs. 15, 33). (Also, the limbus of T. dispar is similar to that of E. enghoffi, see Fig. 18C) Summarizing, our findings underline the 'phylogenetic potential' of the prozonital surface sculpture; however, the pattern is not at all clear-cut.
"Cerotegument" and "microvilli" Likhitrakarn et al. (2019) described the tegument of Cryptocorypha enghoffi as "encrusted with a microspiculate cerotegument, dull, beset with microvilli". A cerotegument is "a thick cement layer built after moulting … an additional layer of hardening secretion covering the epicuticle" (Wolff et al. 2017). The fine SEM images of Likhitrakarn et al. (2019) in my opinion show no indication of a cerotegument in this sense, and likewise, in the species of Cryptocorypha described here there is nothing that looks like a cerotegument. On the other hand, the dark layer covering the dorsal side of Elythesmus enghoffi might well be termed a cerotegument although Hoffman (1978) did not use this term.
The "microvilli" referred to by Likhitrakarn et al. (2019) would correspond to what I have called "level three spherules" (Fig. 3). Because "Microvilli are finger-like membrane protrusions, supported by the actin cytoskeleton, and found on almost all cell types" (Orbach & Su 2020), I would recommend not using this term for the minute cuticular outgrowths found on different arthropods, including the species of Cryptocorypha described here (Fig. 3A).

Infraspecific variation in body ring number, or several species of Elythesmus?
Elythemus enghoffi is remarkable in including males with 18 and 17 podous rings (maybe even 16). Intraspecific variation on ring number is very rare in the order Polydesmida. Apart from the genus Devillea Silvestri, 1903 (family Xystodesmidae), in which the normally strict regulation of ring numbers has obviously become relaxed (see Enghoff et al. 1993), and some laboratory-induced cases (David & Geoffroy 2011), the only known examples are from the family Ammodesmidae where several species of the genus Ammodesmus Cook, 1896 may have 16 or 17 podous rings in both sexes (VandenSpiegel & Golovatch 2012. Although the number of studied females of E. enghoffi is modest, nothing suggests that females may have other than 18 podous rings, since only females with this ring number have an epigynal crest (Fig. 15C-D).
David & Geoffroy (2011) mentioned Muyudesnus obliteratus Kraus, 1960 (now placed in the genus Poratia Cook & Cook, 1894, family Pyrgodesmidae) as another example of intraspecific variation in ring number, but actually the two 'morphs' in this case are now regarded as representing different species (Golovatch & Sierwald 2000). The obvious question is, whether the variation described here for E. enghoffi might similarly be due to the existence of two or more species of Elythesmus. This may very well be the case, but the pattern of variation is rather complicated (see "remarks on variation" under E. enghoffi): 1) males with 17 and 18 podous rings exist at least in two mountain blocks, Udzungwa and Uluguru, 2) in the Udzungwas, males with 18 podous rings are known only from the NE part (Udzungwa Mountains National Park), 3) there is considerable size variation which seems in part to be correlated with altitude, 4) no variation in the gonopods could be observed. For the time being, all Elythesmus samples, except the few females from West Kilombero Scarp Forest Reserve (see above under ?Elythesmus sp.) are referred to E. enghoffi, but an in-depth analysis applying molecular methods might possibly change the picture.

Sex ratio and juveniles in samples
The vast majority of material for the present study was obtained by pitfall trapping. Pitfall traps catch animals that move actively and pitfall catches therefore show a bias towards more active life stages. It is in full agreement with this that the material of the two most abundant species show a very strong preponderance of adult males over adult females, and an even stronger preponderance of adults over juveniles. Cryptocoryha geminiramus sp. nov.: 159 adult ♂♂, 19 adult ♀♀, 3 subadult ♀♀, Elythemus enghoffi: 124 adult ♂♂, 48 adult ♀♀, 10 subadults. These finding are in line with the general observation that adult millipedes are more active than juveniles and that male millipedes are more active than females. Examples of other studies on millipedes using pitfall trapping and showing 'superabundance' of adults, especially males, include Geoffroy & Célérier (1996) and Mesibov & Churchill (2003).