A mountain of millipedes VII: The genus Eviulisoma Silvestri, 1910, in the Udzungwa Mountains, Tanzania, and related species from other Eastern Arc Mountains. With notes on Eoseviulisoma Brolemann, 1920, and Suohelisoma Hoffman, 1963 (Diplopoda, Polydesmida, Paradoxosomatidae)

Twenty-two new species of the genus Eviulisoma Silvestri, 1910, from the Eastern Arc Mountains, Tanzania, are described: E. acaciae sp. nov., E. aequilobatum sp. nov., E. akkariae sp. nov., E. angulatum sp. nov., E. articulatum sp. nov., E. biquintum sp. nov., E. breviscutum sp. nov., E. cetafi sp. nov., E. chitense sp. nov., E. commelina sp. nov., E. coxale sp. nov., E. ejti sp. nov., E. grumslingslak sp. nov., E. kalimbasiense sp. nov., E. navuncus sp. nov., E. nessiteras sp. nov., E. ottokrausi sp. nov., E. paradisiacum sp. nov., E. sternale sp. nov. and E. zebra sp. nov. from the Udzungwa Mts, E. culter sp. nov. from the Rubeho Mts and E. kangense sp. nov. from the Kanga Mts. Eviulisoma kwabuniense Kraus, 1958, and E. dabagaense Kraus, 1958, both from the Udzungwa Mts, are redesribed based on new material. Notes are provided on E. iuloideum (Verhoeff, 1941) based on type material. Eoseviulisoma Brolemann, 1920, is synonymized under Eviulisoma, based on newly collected material of E. julinum (Attems, 1909), type species of Eoseviulisoma. New material of Suohelisoma ulugurense Hoffman, 1964, type species of Suohelisoma Hoffman, 1964, has revealed that the gonopod structure is more similar to that of Eviulisoma than originally thought, but Suohelisoma is retained as a valid genus. Four species groups are recognized among Eviulisoma species from the Udzungwa Mts, but the need for a revision of the entire genus is emphasized. Two types of epizootic fungi are recorded from Eviulisoma spp., and an enigmatic amorphous mass, which may be a kind of plugging substance, is recorded from the gonopod tips and excavated sixth sternum of several species.


Introduction
This is the seventh in a series of articles dealing with the millipede fauna of the Udzungwa Mts. For general information on the Udzungwa Mts, see Enghoff (2014) and Scharff et al. (2015). ENGHOFF H., Eviulisoma millipedes from the Udzungwa Mountains 3 The previous six articles in the series (Enghoff 2014(Enghoff , 2016a(Enghoff , 2016b(Enghoff , 2016c(Enghoff , 2018Enghoff & Frederiksen 2015) all deal with the family Odontopygidae, but the present contribution concerns the genus Eviulisoma Silvestri, 1910, of the very large family Paradoxosomatidae, which is distributed over all zoogeographical regions, except the Nearctic .
The large genus Eviulisoma, as understood by recent workers (Jeekel 2003;Nguyen & Sierwald 2013;VandenSpiegel & Golovatch 2014), is endemic to the African continent. Eviulisoma belongs to the ʻecarinateʼ paradoxomatids in which the paranota are very strongly or even completely reduced, resulting in a virtually cylindrical body (cf. the name which alludes to the ʻjulidʼ appearance of these millipedes). The original description (Silvestri 1910), although possibly based on a misidentifi ed type species (Jeekel 2003), mentions several important characters and is accompanied by quite a nice gonopod drawing.
So far, only four species of Eviulisoma have been described from the Eastern Arc Mts, viz., E. dabagaense , and E. kwabuniense Kraus, 1958, from the Udzungwa Mts, and E. taita VandenSpiegel & Golovatch, 2014, and E. taitaorum VandenSpiegel & Golovatch, 2014, from the northernmost part of the Eastern Arcs, the Taita Hills in southern Kenya.
Twenty new species of Eviulisoma from the Udzungwa Mts are described here, as are two additional species, obviously closely related to Udzungwan species, but coming from other Eastern Arc mountain blocks (Rubeho Mts and Kanga Mts). For the sake of simplicity these two species are included wherever "Udzungwan species of Eviulisoma" are referred to. When only species from the Udzungwa Mts themselves are meant, the term "Udzungwan s. str. species" is used. Figure 1 shows one of the more strikingly coloured new species described here. European Journal of Taxonomy 445: 1-90 (2018) 4

Material and methods
The vast majority of material comes from the zoological collections of the Natural History Museum of Denmark, University of Copenhagen (ZMUC). Much of the material was collected during several fi eld trips to the Udzungwa Mts by ZMUC staff and students, but a very substantial part of it was collected by the NGO Frontier Tanzania (https://frontiergap.com/About-Us/Background-Mission.aspx) and was subsequently deposited in ZMUC. Additional material derives from the collections of the Virginia Museum of Natural History (VMNH), where the very large collection of Tanzanian millipedes accumulated by Richard L. Hoffman (1927Hoffman ( -2012 is housed. Marshall et al. (2001a) described how the material from West Kilombero FR was collected: "Grounddwelling millipedes were primarily sampled by timed searching of 3 m × 3 m quadrats. [...] The leaf litter and topsoil within quadrats was searched thoroughly by hand for a total of eight person hours per quadrat at all trapsites. In addition, the general proximity of all trapsites was searched for two person hours. During such searches, particular attention was paid to rotting logs, the underside of rocks and other such microhabitats, which may have been missed by the quadrat samples." The collecting procedure in New Dabaga / Ulangambi FR was the same (Marshall et al. 2001b). Andrew R. Marshall (pers. comm.) further informs that during these campaigns the collectors had to dig down to around 15 cm to fi nd millipedes after initially fruitless searches in the topsoil alone. Figure 2 shows the areas from where the specimens derive. All studied specimens are from Tanzania, Udzungwa (sometimes spelled Uzungwa) Mts.
A total of 155 male specimens were examined. All samples are kept in 70% alcohol.
Specimens were examined in alcohol under a stereo microscope. Specimens for scanning electron microscopy (SEM) were cleaned with ultrasound, transferred to 96% ethanol, then to acetone, air dried, mounted on aluminium stubs or on triangles of fl exible aluminium tape in turn mounted on stubs, coated with platinum / palladium and studied in a JEOL JSM-6335F scanning electron microscope.
Quite often, more than one species was present and represented by males in a sample. Under ʻDistribution and habitatʼ for each species such co-occurences are summarized in statements starting with "Collected together with …"

Descriptions
In light of the large amount of material at hand, and the modest (males) to very small / virtually nonexistent (females) number of non-gonopodal differences between the studied species, only adult males are considered. In those cases where females could with some confi dence be referred to a particular species, they are listed, but no separate description is given for them. Females are generally larger, especially thicker, than males.
Previous recent descriptions of species of Eviulisoma vary from extremely detailed (Jeekel 2003) to much briefer (VandenSpiegel & Golovatch 2014). I have chosen to follow the lead of the latter authors, and even to make the species description shorter still, because many of the characters specifi ed for each species by VandenSpiegel & Golovatch (2014) hardly vary between the species from the Udzungwa Mts. Instead, these characters are treated under the generic heading. ENGHOFF H., Eviulisoma millipedes from the Udzungwa Mountains 5 Characters treated at species level include: -Body length -Midbody width -Colour -Length of antennae -Paranota -Stricture -Pilosity of body rings -Length of legs, relative lengths of podomeres from prefemur to tarsus -Scopulae -Hypoproct -Modifi cation of sternum 5 -Modifi cation of sternum 6 -Gonopods For the relative lengths of podomeres, notations such as the following are used: femur > prefemur ≈ tarsus > tibia (>) postfemur. This means that femur is clearly longer than prefemur, prefemur is approximately as long as tarsus, tarsus is clearly longer than tibia, tibia is slightly longer than postfemur. Tribe Eviulisomatini Brölemann, 1916 Diagnosis Paradoxosomatidae in which paranota are very strongly reduced or virtually absent, and the ʻfemoriteʼ of the gonopod is strongly reduced, such that solenomere, solenophore and up to two acropodital processes seem to originate directly from the end of the prefemoral part.
In the key to Central African eviulisomatinine genera by Hoffman (1971), Eviulisoma keys out together with Suohelisoma in couplet 1 because of the ʻdorsalʼ (actually cranio-distal) lobe on the mesal side of the gonopod coxa. These two nominal genera also share the strongly excavated sternum 6 (absent in a few species of Eviulisoma, however), a character also present in Wubidesmus. Variously excavated sterna 6 have been described for certain other paradoxosomatids, e.g., Luzonomorpha pallidula Jeekel, 2000, and Montesecaria golovatchi Jeekel, 2002(Jeekel 2000, 2002, but as far as can be deduced from the descriptions, these excavations are quite different from those seen in Eviulisoma spp. Most eviulisomatinine genera share a strongly reduced body ring setation: two rows of very thin setae on the collum and only one row on each of the following body rings. The majority of paradoxosomatid genera have two or three rows of setae on postcollar body rings, and this character state is shared by the eviulisomatinine genera Boreviulisoma (Reboleira & Enghoff 2013) and Jeekelosoma (Mauriès 1985;Enghoff & Reboleira in prep.). Table 1 summarizes some important characters across the eviulisomatinine genera.

Included species
See Table 2. Table 1. Comparison of eviulisomatinine genera. 1 except for tiny keel on ring 2 2 but peritremata present on poriferous rings 3 Hoffman (1953) was in doubt about the existence of a separate solenomere; dissection of the gonopod of a ♂ of S. ventriconus (Attems, 1931) Mauriès (1985) described a subgenus of Eviulisoma, based on a species from Morocco. However, a study by Enghoff & Reboleira (in prep.) shows that Jeekelosoma Mauriès, 1985, should be upgraded to full generic status, and by this action Eviulisoma will again become an endemic Afrotropical genus.

Diagnosis
A succinct diagnosis of Eviulisoma was provided by Hoffman (1953). Supplemented with information from Hoffman (1964), Jeekel (2003, Vandenspiegel & Golovatch (2014) and the present study, and with an updated terminology (see below), Eviulisoma may be diagnosed as follows: A genus of Paradoxosomatidae in which: -paranota are missing, or at most present as tiny keels on ring 2 only -there is (usually) a process between the coxae of the fourth male legs -the sternum of body ring 6 is usually deeply excavated -the collum bears two transverse rows of thin setae, postcollar body rings bear only one such row -the gonopod coxa usually has a conspicuous meso-anterior lobe -the gonopod prefemur is shorter than the acropodite, usually less than half as long -the acropodite consists of at least three branches which seem to originate directly from the prefemur: -the fl agelliform solenomere -a mesal acropodital process which is often the longest of the acropodital branches -sometimes an intermediate acropodital process originating between the mesal acropodital process and the solenomere -a lateral solenophore which serves as protection of the solenomere   Silvestri, 1910. (continued) Gonopod terminology Jeekel (2003) gave detailed redescriptions of many species of Eviulisoma, as well as general comments on the gonopods. In Jeekel's terminology, the gonopod in Eviulisoma consists of a basal coxa, followed by a prefemur and a terminal acropodite. The ʻfemoriteʼ which usually forms a distinct basal ʻshaftʼ of the acropodite is extremely reduced in Eviulisoma, and "Moreover, it has made a torsion of 180° which is shown by the course of the spermal channel along the anterior side towards the lateral side. The result is that the solenomere arises from the lateral side of the femorite" (Jeekel 2003: 48).
On the acropodite Jeekel (2003) distinguished three elements: the "solenomerite", the "tibiotarsus" and a mesal process which he called the "postfemoral process". There has been a long tradition of attempting to homologise parts of millipede gonopods with podomeres of ordinary walking legs, from which the gonopods have evolved. In the colobognathan orders, where the gonopods retain an obviously leglike structure, this is no problem, but in the eugnathan groups (Nematophora, Merocheta and Juliformia) the homologisation is not straightforward. Eugnathan gonopods often show some more or less obvious articulations, sutures, or constrictions which might correspond to articulations between podomeres, but apart from the articulation between coxa and telopodite, these subdivisions of the gonopod probably have nothing to do with the original leg segmentation. Thus, developmental studies on polydesmid millipedes indicated that the entire telopodite corresponds to the prefemur of a walking leg (Petit 1976). It therefore makes sense to minimize use of a gonopod terminology suggesting homology with walking leg podomeres. VandenSpiegel & Golovatch (2014) already took such a step for Eviulisoma, using the functional term "solenophore" for what Jeekel (2003) called "tibiotarsus". I suggest a further step in the same direction and herewith propose the terms "mesal acropodital process" and "intermediate acropodital process" for what Jeekel (2003) and Vandenspiegel & Golovatch (2014) referred to as postfemoral processes.

General description of Udzungwan Eviulisoma
In order to minimize redundancy in the species descriptions, the following general description is presented. It is based on the Udzungwan species studied here, but when appropriate, additional literaturebased information from other species is added [in square brackets]. The description applies to adult males.
-Colour highly variable. Some species are uniform pale whitish (this seems not always to be a result of preservation), some are more or less uniform brownish or even black (E. biquintum sp. nov., partly), many are more or less strikingly ringed, with metazonites or a part thereof being brown or black, contrasting with pale prozonites (e.g., E. zebra sp. nov., Fig. 1; E. akkariae sp. nov., Fig. 3A), one colour form of E. coxale sp. nov. has large brownish dots at the ozopore level, contrasting with a pale background (Fig. 3B), and one colour form of E. biquintum sp. nov. is pitch black with contrasting white legs (Fig. 3C). -Lower part of head capsule ʻclypeo-labral regionʼ with numerous setae up to between antennal sockets; upper part (ʻvertigial regionʼ) with at most a few scattered setae, sometimes arranged in one or more pairs close to midline (Fig. 4). -Antennae (Fig. 4) reaching back to ring [2]3-5 when folded along the side of the body. Antennomeres 2-6 of roughly same length, 1 and 7 much shorter. -Collum (Fig. 4) unmodifi ed, with two transverse rows of thin setae, one row near anterior margin, one ca in the middle, laterally somewhat wrinkled.
-Body rings: -Paranota completely absent or present as inconspicuous, simple ridges on ring 2 only (Fig. 4). In some species, such ridges are seen in some specimens, not in others. -Pleurosternal keels (Fig. 5) simple, not prominent, not drawn out as posterior denticles, best developed on rings 2-5 (7), decreasing in size backwards, but in some species recognizable as far back as ring 17. -Surface smooth, sometimes with visible cellular structure, metazonites more or less longitudinally wrinkled, especially ventro-laterally. -Stricture between pro-and metazonite sometimes smooth, sometimes more or less conspicuously striolate (Fig. 5).  -Ozopores opening fl ush with metazonital surface at ca ⅔ of metazonital length behind stricture (Fig. 5). -A single transverse row of two (1+1) or four (2+2) thin setae often present on metazonites, sometimes only seen on ring 2, sometimes apparently missing (probably abraded in many cases). -Sterna sometimes with small cones at base of legs (Fig. 5), at least on ring 8. -Preanal ring with a long triangular epiproct; lateral setiferous tubercles usually poorly differentiated, sometimes virtually absent; apical tubercles also usually not proment. -Anal valves (paraprocts) with margin raised as narrow lips.
-Subanal scale (hypoproct) variable, sometimes semicircular, sometimes trapezoid, sometimes with three more or less conspicuous tubercles at distal margin. -Legs 0.8-1.5 × as long as body diameter. Relative lengths of podomeres variable; femur the longest in most cases, but in some short-legged species, prefemur is as long as femur, and in a few long-legged ones, tarsus is as long as or longer than femur. -Non-gonopodal sexual characters: -Sternum 5 usually with a tongue-shaped / subtriangular / subrectangular process ( Fig. 6) between anterior legs (pair 4). One species (E. breviscutum sp. nov.) without a process (Fig. 36) and one (E. biquintum sp. nov.) with a small tubercle between the 4 th leg pair and a similar one between the 5 th (Fig. 35). -Sternum 6 usually deeply excavated (Fig. 6), anterior margin of excavation curved, with a row of long setae. Fig. 6 shows a sternum 6 excavation as it appears in most species; deviating morphologies occur in certain species (Figs 26,28,30,34 -Scopulae (dense coverings of modifi ed setae) usually present on the ventral side of femur, postfemur, tibia and tarsus on anterior legs (Fig. 7), decreasing in size from anterior to posterior, often disappearing from (femur and) postfemur on posteriormost legs (scopulae only on tibia and tarsus in some non-Udzungwan species). E. kangense sp. nov. and E. sternale sp. nov. furthermore with the ventral surfaces of prefemora and femora fl attened and hairless on some leg pairs. No podomeres with swellings or processes. -Gonopods (Fig. 8): -Coxa (cx) with a rounded distomedial lobe of variable size.
-Acropodite consisting of a highly reduced basal part from which solenomere, solenophore, mesal acropodital and sometimes intermediate acropodital process arise at ca same level. -Solenomere (slm) originating near lateral side of gonopod, simple, whip-like, usually largely concealed within folded solenophore ( Fig. 8A-B). -Solenophore (sph) originating on lateral side of gonopod, very variable in shape, sometimes a simple rolled sheet, sometimes with a distinct ʻconductorʼ process guiding the solenomere; in species with an 'open' solenophore, an area with parallel ridges (e.g., 18 ) is sometimes seenwhether similar ridges are present in rolled-up solenophores is unknown. -Mesal acropodital process (map) originating on mesal side of gonopod, mostly but not always the longest part of the acropodite, very variable in shape. In species with an intermediate acropodital process (iap), map is basally closely contiguous with the solenomere and map might be interpreted as a solenomeral process (?parasolenomere). In species without iap, the solenomere articulates with the highly reduced femorite, close to the basis of map. -Intermediate acropodital process (iap) only present in some species, originating between map and slm, long, slender, often spinose.
Females are generally larger than males. Although females have not been considered in the present species descriptions, it is worth noting that the ventral part of the third body ring (ʻepigyneʼ) and the basal part of the second pair of legs show considerable variation among species (Brolemann 1920). Also within the Udzungwan species of Eviulisoma, several distinct types of ʻepigyneʼ and second legs exist, but often it was not possible to correlate a particular female morphotype with a particular species as defi ned by male characters.

Species groups
Four species groups can be recognized among the Eviulisoma species treated here ( Table 3). All groups include species with an excavated male sternum 6 and a ventral process / lobe on sternum 5. One group is characterized by the presence of an intermediate acropodital process (iap), one by having the margins of the sternum 6 excavation lobed, one by having a laterally compressed mesal acropodital process (map) and one by having neither of these characteristics, but a large, sheet-like, unrolled solenophore (sph) and, notably, a separate basal part (ʻfemoriteʼ) of the acropodite. Three species are left ungrouped, being not particularly similar to any other new or previously described species. Two of the ungrouped species lack the sternum 6 excavation and the sternum 5 lobe.
The kwabuniense group

Diagnosis
Differs from all other species of Eviulisoma by the presence of a basal acropodital process. Further differs from other Udzungwan members of the E. kwabuniense group by the combination of largely identical, smooth map and iap, and a relatively short solenophore with three lobes of approximately equal length.

Descriptive notes (male)
Information on the holotype, from   HYPOPROCT. Trapezoid, with three prominent apical tubercles.

Distribution and habitat
Known only from the New Dabaga / Ulamgambi FR. Altitudinal range 1800-2100 m a.s.l. (upper limit according to . Habitat: montane forest (studied specimen) and semi-rainforest, under leaf litter . Collected together with E. ottokrausi sp. nov.

Remarks
The studied specimen, which is a near-topotype, agrees completely with the original description ); a side-by-side comparison with the holotype was therefore deemed unnecessary.

Diagnosis
Differs from other species of the E. kwabuniense group by the combination of a map ending in two equal, parallel prongs, a spinose iap and a two-lobed solenophore with a dorsal lobe reaching tip of acropodital processes.

Etymology
This species is named after the dominant tree at the type locality.
Material studied (total: 5 ♂♂) BODY RINGS. Paranota represented by very faintly developed keels on body ring 2 (as in Fig. 4C), otherwise completely absent. Stricture between pro-and metazonite not striolate. A transverse row of setae on all body rings, but many setae abraded.

Diagnosis
Differs from other species of the E. kwabuniense group by having the solenophore almost as long as map and iap, which are both smooth, in combination with the equal length of the three apical lobes of the solenophore.

Distribution and habitat
Known only from the type locality, New Dabaga / Ulangambi FR, 1930 m a.s.l.

Diagnosis
Differs from other species of the E. kwabuniense group by being larger (width 2.6-3.3 mm vs 1.5-2.1 mm in other species), in having contrasting dark and pale transverse bands, and in the combination of a smooth intermediate acropodital process (iap) and a large, two-lobed solenophore with a dorsal lobe (sph-d) as long as acropodital processes and ending in a hook.

Etymology
This species is named after Nesrine Akkari, one of the very few myriapodologists from the African continent, author of several important papers on myriapods, now curator of the important myriapod collection in the Naturhistorisches Museum Wien, Austria, and always a dear friend.

Distribution and habitat
Known from two sites in the Udzungwa Mts National Park. Altitudinal range 850-1207 m a.s.l. Differs from other species of the E. kwabuniense group by having the solenophore two-lobed, its dorsal lobe developed as a very strong hook, hook much larger than those seen in certain other species (E. ejti sp. nov., E.akkariae sp. nov., E. nessiteras sp. nov.).

Etymology
The species name honours CETAF, Consortium of European Taxonomic Facilities, www.cetaf.org, in recognition of the immense importance of CETAF for natural history collections in Europe and for collections-based research. LEGS. Length 1.2 × body width. Relative lengths of podomeres: femur > prefemur > tarsus > tibia > postfemur. Scopulae (Fig. 7B, D) on femur, postfemur, tibia and tarsus until ca midbody, thereafter gradually disappearing.
STERNUM 5. A small trapezoid process between legs 4. STERNUM 6. Deeply excavated. Rim of excavation simple. GONOPODS (Fig. 13). Coxal lobe (cxl) moderate. Prefemoral part (prf) ca half as long as acropodite. Mesal acropodital process (map) a long, smooth, almost straight rod, apically pointed, subapically with small triangular side branch. Intermediate acropodital process (iap) a very slender, straight rod, densely covered in long spines on part of its surface, especially on apical part (Fig. E). Solenophore (sph) deeply split into two long lobes; dorsal lobe (sph-d) a very large, strong hook, almost reaching to tip of acropodital processes; ventral lobe (sph-v) much shorter, lanceolate; no intermediate process between sph-d and sph-v; internal surface of sph with a ridged area (ra, Fig. 13D, F).

Diagnosis
Differs from other species of the E. kwabuniense group by the combination of a map with a short subapical laterad side branch, a partly spinose iap, and a clearly three-lobed solenophore that is much shorter than map and iap, dorsal lobe of solenophore much longer than the others.

Etymology
The name is an adjective referring to the type locality.

Diagnosis
Differs from other species of the E. kwabuniense group by having the solenophore somewhat shorter than the smooth map and iap, in combination with the small size of the intermediate apical lobe of the solenophore, compared with the very long, slender dorsal and ventral lobes.

Etymology
The name is a noun in apposition, referring to the one short and two long apical lobes of the solenophore. Commelina L. is a genus of plants (ʻdayfl owersʼ) with fl owers characterized by one small and two large petals. Linnaeus (1737: 79) dedicated this genus to three members of the family Commelijn, two of whom were well-known botanists, while the third accomplished nothing (at least not in botany); see also Wijnand (1983: 11).
COLOUR. Completely pallid after 17 years in alcohol, possibly not due to fading, cf. remarks under E. ottokrausi sp. nov. BODY RINGS. Paranota completely absent. Stricture between pro-and metazonite striolate. A transverse row of setae on pre-gonopodal rings, a few scattered setae seen on post-gonopodal rings.

Distribution and habitat
Known only from the New Dabaga / Ulangambi FR. Altitude 1800-1900 m a.s.l.

Diagnosis
Differs from other species of the E. kwabuniense group by map being apically shaped like a narrow, slightly hooked spoon and at ca ¾ of its length having a side branch, in combination with a spinose iap and a solenophore with a large, hooked dorsal lobe.

Etymology
The species name honours the European Journal of Taxonomy (EJT), in recognition of its immense importance for the dissemination of taxonomic research in Europe, and beyond.
Material studied (total: 5 ♂♂)  BODY RINGS. Paranota completely absent. Stricture between pro-and metazonite clearly striolate. A transverse row of setae on all body rings, but many setae abraded.

Distribution and habitat
Known only from the Udzungwa Scarp FR, 11 km SE of Masisiwe Village, Kihanga Stream, 1800 m a.s.l.
Eviulisoma kalimbasiense sp. nov. urn:lsid:zoobank.org:act:347BA562-0339-4D55-992C-4E14B6FEAB93 Fig. 17 Diagnosis Differs from other species of the E. kwabuniense group by having a smooth iap much shorter than map and by having the large dorsal lobe of the solenomere bifi d.

Distribution and habitat
Known only from Kalimbasi Mountain, S of Mazombe town, 2000-2100 m a.s.l. The site is located in the Kisinga-Rugaro FR.
Eviulisoma navuncus sp. nov. urn:lsid:zoobank.org:act:91A53055-C236-4F4E-A32A-172D2289DE25 Fig.18 Diagnosis Differs from other species of the E. kwabuniense group by having a long, curved side branch from map fi tting over the semicircular dorsal lobe of the solenophore, in combination with an extremely slender, spinose iap.

Etymology
The name is a noun in apposition, from the Latin navis (ʻboatʼ) and uncus (ʻhookʼ), referring to the boathook shape of the mesal acropodital process. Referred non-type material TANZANIA: 6 ♀♀, tentatively referred to this species, same collection data as for paratypes (VMNH). BODY RINGS. Paranota indicated by very faint keels on body rings 2 (as in Fig. 4C), otherwise completely absent. Stricture between pro-and metazonite striolate. A transverse row of setae on all body rings.

Distribution and habitat
Known from two sites, one in Kitungulu / Kiranzi FR, the other in Udzungwa Scarp FR. Altitudinal range includes 1500 m. Collected together with E. dabagaense Kraus, 1958 and E. nessiteras sp. nov. in Kitungulu / Kiranzi FR.

Remarks
Interpreting the locality name "Bomalamzinga" caused a lot of problems until Andy Marshall (pers. comm.) informed me that it refers to a place at the northern end of the Udzungwa Scarp FR.
Eviulisoma nessiteras sp. nov. urn:lsid:zoobank.org:act:7442C1C4-8C4F-4CCB-BD53-64347CB6185A Fig. 19 Diagnosis Differs from other species of the E. kwabuniense group by having map and the spinose iap extremely slender, in combination with a solenophore with very to extremely slender dorsal and ventral processes, such that the acropodite seems to consist of four very slender branches (in addition to the solenomere).

Etymology
The specifi c epithet is a noun in apposition. Nessiteras is the genus name given to the famous Loch Ness Monster, and the dorsal lobe of the solenophore of E. nessiteras sp. nov. resembles the most famous photograph of the alleged monster sticking its long neck out from the lake surface. GONOPODS (Fig. 19). Coxal lobe (cxl) large. Prefemoral part (prf) ca 0.3 × as long as acropodite. Mesal acropodital process (map) a very slender, smooth, slightly arched rod with a short triangular subapical side branch. Intermediate acropodital process (iap) as slender and ca as long as map, slightly arched, densely covered in long spines along dorsal side (Fig. 19F). Solenophore (sph) deeply split into two long lobes; dorsal lobe (sph-d) almost as long as acropodital processes, stouter than these, apically bent at right angles and ending in two triangular lobes; ventral lobe (sph-v) ca as long and stout as sph-d, somewhat sinuous (in Fig. 19C the sph-v looks strongly twisted, but this is due to distortion during preparation of the SEM mount), apically pointed; a tiny intermediate lobe (sph-i) between sph-d and sph-v; internal surface of sph with a ridged area (ra).

Diagnosis
Differs from other species of the E. kwabuniense group by having a short subterminal side branch on the map, in combination with having spines on the basal part of the long, slender iap and having the dorsal lobe of the solenophore much longer than the ventral and intermediate lobes.

Etymology
The name honours Otto Kraus (1930Kraus ( -2017 who described the fi rst species of Eviulisoma from the Udzungwa Mountains and authored numerous other papers on myriapod (and arachnid) taxonomy.

Paratypes
ANTENNAE. Reaching back to middle of ring 4. BODY RINGS. Paranota seen as extremely inconspicuous ridges on ring 2 in one specimen (as Fig. 4C), completely absent in others. Stricture between pro-and metazonite striolate. A transverse row of setae on all body rings back to ring 9 in one specimen, only on ring 2 in another.

Distribution and habitat
Known only from the New Dabaga / Ulangambi FR. Altitudinal range 1800-1962 m a.s.l. Habitats include montane forest and scrub / thicket / bush. Collected together with E. kwabuniense .

Remarks
One male infested with nematodes, one worm sticking out from each gonopore.

Diagnosis
Differs from other species of the E. kwabuniense group by having a strong hook emerging from the concave side of the solenophore and, together with the ventral lobe of the solenophore, delimiting a narrow slit.

Etymology
The name is an adjective referring to the name Frontier Tanzania gave to the montane forest plot where the species was found. LEGS. Length 0.9 × body width. Relative lengths of podomeres: femur > prefemur > tarsus > postfemur = tibia. Scopulae (Fig. 7A) on femur, postfemur, tibia and tarsus, diminishing towards posterior, especially on femur.
STERNUM 5. A bell-shaped process between legs 4. STERNUM 6. Deeply excavated. Rim of excavation simple. GONOPODS (Fig. 21). Coxal lobe (cxl) moderately large. Prefemoral part (prf) ca ⅔ as long as acropodite. Mesal acropodital process (map) a long, straight, stout, pointed rod with a subapical lateral fl ange. Intermediate acropodital process (iap) a more slender, straight, non-spinose pointed rod, as long as map. Solenophore (sph) a massive sheet, almost as long as map, with a large, hook-shaped process (sph-h) on the hollow side, basal part of sph-h separated from small ventral, ʻsolenomere-conductingʼ lobe (sph-v) by narrow slit, dorso-lateral part of sph apically with a fi nger-shaped dorsal lobe (sph-d) separated by U-shaped incision from multi-cusped intermediate lobe (sph-i). The dabagaense group Diagnosis Species of Eviulisoma in which sternum 6 is deeply excavated, with unlobed margins, there is no intermediate acropodital process, the solenophore is a tightly rolled-up sheet of a tube-like appearance (with a few small processes) and the mesal acropodital process is strongly laterally compressed. No further species can be assigned to this group based on the existing literature. Kraus, 1958 Figs 6E, 22 Eviulisoma dabagaense : 2 (holotype (not studied) in the Überseemuseum Bremen).

Diagnosis
Differs from other species of the E. dabagaense group by the combination of a gonopod coxa without a distolateral process and a parallel-margined map with two strong apical denticles and several smaller ones on the dorsal margin.

Remarks
The studied near-topotype agrees completely with the original description ); a side-by-side comparison with the holotype was therefore deemed unnecessary. Several juveniles with a colour pattern like that of E. dabagaense were found in samples from New Dabaga-Ulangambi FR containing no males of E. dabagaense, but one or several males of the entirely pallid E. ottokrausi sp. nov. Although a safe identifi cation of these juveniles cannot be made, they probably belong to E. dabagaense.

Diagnosis
Differs from other species of the E. dabagaense group, except E. culter sp. nov., by having a distolateral process on the gonopod coxa. Differs from E. culter sp. nov. by having map with two strong apical denticles and several small denticles along the dorsal margin.

Etymology
The name, an adjective, refers to the peculiar modifi cation of the gonopod coxa.  BODY RINGS. Paranota absent, except for barely discernible keels on ring 2 (as in Fig. 4C). Stricture between pro-and metazonite indistinctly striolate. A transverse row of setae on most body rings (probably abraded where absent).
STERNUM 5. A rounded-trapezoidal process between legs 4 (Fig. 6F). STERNUM 6. Deeply excavated, rim simple. ENGHOFF H., Eviulisoma millipedes from the Udzungwa Mountains 51 GONOPODS (Fig. 23). Coxa with a long, digitiform distolateral process (cxp). Other gonopodal characters as in E. dabagaense: coxal lobe (cxl) moderate; prefemoral part (prf) ca 0.25 × as long as acropodite; mesal acropodital process (map) lamelloid, straight, with parallel margins, tip of process with two strong denticles, a row of smaller denticles along apical part of dorsal margin; solenophore (sph) a rolled sheet, separated from map by a distinct gap, less than ⅔ × as long as map, forming a tube with two small apical denticles and a slender process ca at mid-length.

Specimens from Udzungwa Mts National Park
Similar to type specimens, except as follows: SIZE. Length 25 mm, max. width 2.5 mm. COLOUR. Overall pallid after 3 years in alcohol; only head and collum slightly ochre-yellow; a small brown patch between antennae.

Distribution and habitat
Known from two sites in the Udzungwa Mts National Park. Altitudinal range 1400-1850 m a.s.l. Habitat: montane rain forest. Collected in litter and pitfall traps, together with E. breviscutum sp. nov. in Mwanihana Forest.

Remarks
This species is very similar to E. dabagaense, but differs in the presence of a distolateral fi nger-formed process on the gonopod coxa, the latter character being shared with E. culter sp. nov. There are several females and juveniles in the same sample as the holotype, in addition to the listed females. They agree with the holotype in all non-sexual characters except that they are uniformly pale. Whether this is due to fading of the large lateral spots seen in the holotype, or to the possibility that they represent a different species, is diffi cult to say. Several specimens from Udzungwa Mountains National Park carry large lumps of an amorphous mass (see Discussion) on the tip of process map (Fig. 23E). These specimens are slightly larger than the holotype and they do not have large lateral spots on the body rings, but these differences hardly warrant the recognition of a separate species.

Diagnosis
Differs from other species of the E. dabagaense group by lacking denticles on the dorsal and ventral margins of map.

Etymology
The name is a noun in apposition, meaning ʻknifeʼ in Latin and referring to the shape of the mesal acropodital process (map).
Material studied (total: 5 ♂♂) BODY RINGS. Paranota absent, except for barely distinguishable keels on ring 2 (as in Fig. 4C). Stricture between pro-and metazonite smooth. A transverse row of setae on all rings.
LEGS. Length 1.2 × body width. Relative lengths of podomeres: femur > prefemur > tarsus > tibia > postfemur. Scopulae on anterior legs on femur, postfemur, tibia and tarsus; those on femur missing from posterior legs, those on postfemur, tibia and tarsus present almost until end of body.

Distribution and habitat
Known only from Mangalisa Peak in the Rubeho Mts, 2100 m a.s.l.

Diagnosis
Differs from other species of the E. dabagaense group by its larger size (width 2.5-3.4 mm vs 1.8-2.4 mm in other species) and by having the ventral margin of map strongly convex.

Etymology
The name, here to be regarded as a noun in apposition, means ʻpregnantʼ in the now extinct Danish cryptolect ʻrotvaelskʼ and refers to the shape of the gonopodal postfemoral process which (somewhat) resembles the profi le of a pregnant woman's torso.
Material studied (total: 9 ♂♂)  STERNUM 5. A small semicircular process between legs 4. STERNUM 6. Deeply excavated, rim simple. GONOPODS (Fig. 25). Coxa without a lateral process; coxal lobe (cxl) not very prominent. Prefemoral part (prf) ca 0.3 × as long as acropodite; mesal acropodital process (map) large, lamelloid, ventral margin strongly convex, dorsal margin concave, in profi le the process vaguely resembles the torso of a pregnant woman (sway-backed, big belly), tip of process with two strong denticles, a few smaller denticles along apical part of ventral and especially dorsal margin; solenophore (sph) a simple rolled sheet, less than ⅔ × as long as map, forming a tube with two small apical denticles.

Specimens from West Kilombero Scarp FR, including Ndundulu Forest:
Similar to type specimens, except as follows: SIZE. Max. width 2.5-2.9 mm.
COLOUR. After 10 years in alcohol more or less faded, but differs from type specimens in having the collum pale with dark margins.
ANTENNAE. Reaching middle of ring 3.

LEGS. Length 1.4 × body diameter.
GONOPODS. Denticles on ventral and dorsal margin of map in part indistinct / missing.

Distribution and habitat
Known from several sites in the Udzungwa Mts National Park and West Kilombero FR. Altitudinal range 1452-1880 m a.s.l. Habitat: mainly found in tropical semi-evergreen forest, but also in (open) Acacia woodland (cf. Frontier Tanzania 2001: 27, 174). Collected together with E. acaciae sp. nov. in West Kilombero FR, Plot Acacia, and with E. sternale sp. nov. in Ndundulu Forest.

Remarks
The differences between the type specimens and those from West Kilombero FR seem to be constant, but they are not regarded as signifi cant enough to distinguish the two populations taxonomically.

The sternale group Diagnosis
Species of Eviulisoma in which sternum 6 is deeply excavated, there is no intermediate acropodital process, the solenophore is a tightly rolled-up sheet of a tube-like appearance (with a few small processes), the mesal acropodital process is not particularly compressed, but is apically furcate and longer than the solenophore, and the rim of sternum 6 forms four lobes.
No further species can be assigned to this group based on the existing literature. VandenSpiegel & Golovatch (2014: fi g. 6e) illustrated the sternum 6 excavation in E. taita VandenSpiegel & Golovatch, 2014, and it appears that there are small lobes at the base of legs 6-7. However, in E. taita the mesal acropodital has no apical furcation and is shorter than the solenophore.

Diagnosis
Differs from other species of the E. sternale group by its smaller size (width 2.0 mm, vs 3.0-3.2 mm in other species) and by having an apical dentate ridge on map.

Etymology
The name is an adjective referring to the modifi ed 6 th male sternum.
ANTENNAE. Reaching back to middle of ring 4. BODY RINGS. Paranota completely absent. Stricture deep, broad, smooth. Setae absent, except for a middorsal pair on ring 2.

Distribution and habitat
Known only from the West Kilombero Scarp FR, Ndundulu Forest, 1550 m a.s.l. Habitat: tropical semievergreen forest. Collected together with E. grumslingslak sp. nov.

Remarks
The ventrally smooth prefemora and femora is an unusual feature of this species, shared only with E. kangense sp. nov. (see Discussion).

Diagnosis
Differs from other species of the E. sternale group by having the solenophore almost as long as map.
Differs from E. sternale sp. nov. by its larger size (width 3.2 mm vs 2.0 mm) and by having a slenderer map without an apical dentate ridge. Differs from E. zebra sp. nov. by having a fl attened, almost hairless area on the ventral side of the prefemora and femora of the legs of a number of post-gonopodal body rings.

Etymology
The name is an adjective referring to the type locality, the Kanga Mts.
COLOUR. Overall impression ringed. Head, collum, posterior ⅔ of metazonites, epiproct and paraprocts dark brown, rest of body and appendages pale yellowish.
ANTENNAE. Reaching back to middle of ring 4. BODY RINGS. Paranota sometimes visible as a very faint line / keel on ring 2 (as Fig. 4C), otherwise completely absent. Stricture between pro-and metazonite striolate. A transverse row of setae on all rings, but many setae abraded.

Distribution and habitat
Known only from the Kanga Mts, Kanga FR, 400-500 m a.s.l. Habitat: lowland rainforest. The Kanga Mts belong to the Nguru massif, one of the smaller blocks of the Eastern Arc Mountains, situated some 150 km N of the Udzungwa Mts.

Remarks
Although females have in general not been considered in this paper, it is worth noting that in this species, female legs are without dense setation on any podomere, and without modifi ed prefemora and femora. See also Discussion.

Etymology
A noun in apposition alluding to the striped appearance of this species.

Holotype
ANTENNAE. Reaching back to anterior part of ring 5. BODY RINGS. Paranota absent, except for tiny keels on ring 2 (Fig. 4C). Stricture between pro-and metazonite clearly striolate (Fig. 30A). A transverse row of setae on ring 2, ring 3 sometimes with a few tiny setae, following rings without setae.
STERNUM 5. A rounded-trapezoid process between legs 4. STERNUM 6 (Fig. 30B). Deeply excavated. Lateral rims of excavation produced into two lobes on each side, one lobe at basis of each leg. GONOPODS (Fig. 31). Coxal lobe (cxl) prominent. Prefemoral part (prf) ca 0.3 × as long as acropodite. Mesal acropodital process (map) long, slender, slightly curved, apically divided into three pointed tines (tn1, tn2, tn3); ventral tine (tn1) process largest, separated from the two others by a V-shaped incision; lumps of an amorphous mass (am) adhering to tip of map. Solenophore (sph) ca 0.7 × as long as map, a simple rolled sheet, ending in two small processes, one slender, the other broadly triangular.

Distribution and habitat
Known only from the Udzungwa Mts National Park, 1450 m a.s.l.

Remarks
Although this species is very similar to the two other species in the E. sternale group, especially to E. kangense sp. nov., it differs by lacking the peculiar modifi cations of the prefemur and femur of at least some post-gonopodal legs.

Diagnosis
Species of Eviulisoma in which sternum 6 is deeply excavated with unlobed margins, the gonopod acropodite has a separate basal ʻfemoriteʼ part, there is no intermediate acropodital process and the solenophore is a large, unrolled sheet without apical lobes.
This group corresponds to the genus Himatiopus Verhoeff, 1941. The description of E. iuloideum by Verhoeff (1941) is quite good, and the similarity with E. articulatum sp. nov. is obvious. An examination of the type material of Himatiopus iuloideus confi rmed the similarity but also confi rmed that the two species are indeed different. Attems, 1953 (Congo) also seems to have an articulated acropodite, but the solenophore is quite different (Jeekel 2003); hence, E. laceolatum is not regarded as a member of the iuloideum group.

Diagnosis
Differs from all other Eviulisoma species, except E. articulatum sp. nov. and E. lanceolatum, by the obviously articulated gonopod acropodite. Very similar to E. articulatum sp. nov., with which it shares the very large, subrectangular solenophore, but differs from that species by having a cylindrical mesal acropodite process (map) without a subapical side branch (vs map tapering, with a side branch), as well as by having a triangular hypoproct (vs a rounded trapezoid, tritubercuate hypoproct).

Remarks
As far as can be seen from the available material, E. iuloideum shares the general morphology of the Udzungwan Eviulisoma species. It is especially similar to E. articulatum sp. nov., from which it differs gonopod-wise as specifi ed in the diagnosis. Figure 32 (from Verhoeff 1941) shows the gonopod telopodite of E. iuloideum; a restudy of the gonopod slide confi rmed that Verhoeff's drawing is accurate. In particular, the mesal acropodital process (map) does not have a subapical side branch. The two species also differ in the shape of the hypoproct, which in E. iuloideum is triangular vs rounded trapezoid, with three tubercles in E. articulatum sp. nov. According to Verhoeff (1941), E. iuloideum, unlike E. articulatum sp. nov., has no setae on the collum and body rings ("spärliche Beborstung nur am Telson"), but considering how delicate and hard to see these setae are, this difference is hardly signifi cant. The type locality of E. iuloideum, "Tanganjikasee bei Mufundi-Iringa", probably refers to Mufi ndi SW of Iringa City, some 100 km SW of the type locality of E. articulatum sp. nov., i.e., outside the Udzungwa Mts (and, as is the case with several other localities at the "Tanganjikasee" mentioned by Verhoeff 1941, very far from Lake Tanganjika [did Verhoeff misread "tea nursery"?]).
Eviulisoma articulatum sp. nov. urn:lsid:zoobank.org:act:FD505E16-0A0E-47FB-AB2D-5200D8346CF2 Fig. 33 Diagnosis Differs from all other Eviulisoma species, except E. iuloideum and E. lanceolatum, by the obviously articulated gonopod acropodite. Very similar to E. iuloideum, with which it shares the very large, subrectangular solenophore, but differs from that species by having a pointed, tapering mesal acropodite process (map) with a subapical side branch (vs map cylindrical, not tapering and without a side branch), as well as by having a rounded trapezoid, tritubercuate hypoproct (vs a triangular hypoproct).

Distribution and habitat
Known only from Kalimbasi Mountain, S of Mazombe town, 2000-2100 m a.s.l. The site is located in the Kisinga-Rugaro FR.

Ungrouped species
Three species have no very similar congeners, neither among the species treated her, nor, as far as can be deduced, among previously described species. Rather than erecting a group for each of these, they are left ungrouped.

Diagnosis
Differs from all other Eviulisoma species by the angular rim of the sternum 6 excavation and by the opposing hooks of the mesal acropodital process (map) and the solenophore (sph).

Etymology
The specifi c epithet is an adjective referring to the angular shape of the rim of the excavation of the male sternum 6.

Distribution and habitat
Known only from the Udzungwa Mts National Park, Kidatu, 1448 m a.s.l.

Diagnosis
Differs from all other species of Eviulisoma by having two knobs on sternum 5, one between coxae 4 and one between coxae 6. Differs from most species of Eviulisoma by lacking an excavation of sternum 6.

Etymology
The name is composed of the Latin bi, meaning ʻtwoʼ, and quintum, meaning ʻfi fthʼ, and refers to the two knobs on sternum 5. HYPOPROCT. Semicircular, no marginal tubercles.

Distribution and habitat
Known from two parts of the Udzungwa Mts: Udzungwa Mountains National Park (incl. Mwanihana FR) and Nyambanike Mts in West Kilombero Scarp FR.

Remarks
The newly collected specimen from the Udzungwa National Park is strikingly coloured -black body and white legs (Fig. 3C) -whereas the one from Nyambenike seems to be less remarkable in this respect. Fading with age of the specimen may explain the difference, but more specimens from each of these sites would be highly desirable in order to ascertain whether we are really dealing with only one species.
The gonopods resemble those of E. taitaorum VandenSpiegel &Golovatch, 2014, andE. taita VandenSpiegel &Golovatch, 2014. These two species, however, have an excavated sternum 6, and E. taitaorum lacks a lobe on sternum 5. No other species has two processes behind each other on sternum 5, according to available descriptions.

Etymology
The name is a noun in apposition meaning ʻshort shieldʼ and refers to the short, shield-like mesal acropodital process. COLOUR. After 3 months in alcohol dorsally dark brown to blackish brown, vertex and metazonites medium brown, rest of head, antennae and legs pale yellowish.

Material
ANTENNAE. Reaching back to middle of ring 3. BODY RINGS. Ring 2 with barely perceptible keels representing paranota (as in Fig. 4C). Stricture between pro-and metazonite striolate. No setae seen on body rings. HYPOPROCT. Triangular, with a distinct median tubercle.

Distribution and habitat
Known from two places in the Udzungwa Mountains National Park forest. Altitudinal range 1487-1800 m a.s.l. Collected together with E. coxale sp. nov. in Mwanihana Forest.

Remarks
The gonopods of this species bear some resemblance to those of E. somaliense Ceuca, 1971, sharing a short, shield-like mesal acropodital process ("ramo laterale" of Ceuca 1971) and an apically bifurcate solenomere ("ramo seminale" of Ceuca 1971). However, E. somaliense also has a large lobe ("placa allungata") between map and sph, and it has a lobe on sternum 5 as well as an excavated sternum 6. The shapes of map and slm are also reminiscent of those of E. julinum, the type species of Eoseviulisoma, and with this species E. breviscutum sp. nov. also shares the lack of a sternum 6 excavation and a sternum 5 lobe. However, in E. julinum the prefemoral part of the gonopod is very much larger than in E. breviscutum sp. nov.

Key to Udzungwan s. str. species of Eviulisoma
The key is based on adult males. Identifi cations should always be checked against the species descriptions and illustrations, as additional species of Eviulisoma are expected to exist in the Udzungwa Mts.

Distribution and habitat patterns
All Udzungwan s. str. species of Eviulisoma are endemic to the Udzungwa Mts. Figure 37 shows the distribution of the 22 species plotted on a diagrammatic map of the areas within the Udzungwas where the species have been collected. Only three species (E. dabagaense, E. navuncus sp. nov. and E. nessiteras sp. nov.) have been found in more than one area. The Udzungwa Mts National Park, where the highest number of species (nine) has been found, shares no species with other areas. Table 4 shows the altitudinal distribution of the Udzungwan s. str. species. Although some species occur as low as 800 m a.s.l., the highest diversity occurs from 1400 m a.s.l. upward. Eviulisoma covers a larger altitudinal spectrum than the Chaleponcus dabagaensis group, which is restricted to 1500-2200 m a.s.l. with a maximum at 1800-2000 m a.s.l. (Enghoff 2014(Enghoff , 2017. Habitat information is included on the labels of many specimens, and a small amount of habitat information can be extracted from the literature , Frontier Tanzania 2001. In by far the most cases, the habitat is indicated as forest, montane rain forest, tropical semi-evergreen forest, etc. Some species have been collected in open Acacia woodland (E. acaciae sp. nov., E. grumslingslak sp. nov.) or ʻscrub / thicket / bushʼ (E. ottokrausi sp. nov.), but the two last-mentioned have also been collected in proper forests.

The status of Eoseviulisoma Brolemann, 1920
Eoseviulisoma was proposed as a subgenus of Eviulisoma by Brolemann (1920) on the basis of the following characters: -No process between the 4 th male legs (vs a process present in Eviulisoma s. str.) -Male sternum 6 with a very shallow excavation (vs a pronounced excavation in Eviulisoma s. str.) -Gonopodal prefemur ("tronc du télopodite") longer than the distal processes ("les rameaux") (vs shorter in Eviulisoma s. str.) -Transverse suture of body rings striolate ("perlée") (vs smooth in Eviulisoma s. str.) The type and only species of Eoseviulisoma was Strongylosoma julinum Attems, 1909. Hoffman (1953 elevated Eoseviulisoma to a full genus and transferred Dyseviulisoma abnorme Attems, 1937 to it. In his key to African eviulisomatinine genera, Hoffman separated Eoseviulisoma from Eviulisoma in the fi rst couplet in which genera with a large mesapical lobe on the gonopod coxa (Eviulisoma and Suohelisoma) were separated from genera without such a lobe (Scolodesmus, Eoseviulisoma and Wubidesmus) (Hoffman 1971). A third species, Eviulisoma (Eoseviulisoma) rugegianum Attems, 1953, was regarded as "generic status uncertain" by Jeekel (1968), but judging from the original description it may be included in the present rather broad concept of Eviulisoma.
VandenSpiegel & Golovatch (2014) pointed out some diffi culties in distinguishing these two genera. Thus, Eviulisoma ngaia VandenSpiegel & Golovatch, 2014, has a process between the 4 th male legs, but no excavation on sternum 6. Eviulisoma taitaorum VandenSpiegel & Golovatch, 2014, has no process between the 4 th male legs, but is does have an excavated sternum 6 and a smooth suture. They therefore argued that the two genera "may well prove to be synonymous".
The Udzungwan species studied here offer further examples in favour of this suggestion. Thus, E. biquintum sp. nov. has two processes on sternum 5, one between the 4 th legs and one between the 5 th , but no excavation on sternum 6. Striolation of the transverse suture seems to be quite variable: in some Udzungwan species the suture is clearly striolate (Fig. 5A), in others the striolation is indistinct, and in still others the suture is virtually smooth. The relative length of the prefemur and acropodite is also variable; however, most species have the prefemur less than half the length of the acropodite.
Existing illustrations of the gonopod of E. julinum (Attems 1909;Brolemann 1920) are not very clear concerning the acropodital part. Thanks to Sara B. Frederiksen, I have been able to examine a male of Table 4. Altitudinal ranges of Udzungwan s. str. species of Eviulisoma Silvestri, 1910. The inferred ranges include all 100 m intervals between the lower and upper limits of the recorded range, although a species may not have been found in all 100 m intervals within the recorded range. Species recorded only from an altitude at exactly x hundred m a.s.l. have been assigned to the higher of the two 100 intervals including this altitude.  Figure 38 shows several images of the right gonopod of the specimen. There is a relatively pronounced ʻEviulisoma typeʼ coxal lobe (contra Hoffman 1971; the lobe is also evident in Brolemann 1920: fi gs 85-86). The prefemoral part is long -as long as or even longer than the acropodital part. The acropodite consists -in addition to the fl agelloid solenomere -of a relatively short, simple, lamelloid mesal acropodital process (map) and a large complicated solenophore which is split at ca half its length into a mesal process with a rounded, fi nger-shaped tip and a triangular expansion ca at mid-length, and a large lateral lamella which is rolled up to form a conductor for the solenomere. The entire gonopod telopodite is quite short and does not reach the unexcavated sternum 6.
The second species referred to Eoseviulisoma by Hoffman (1953), E. abnorme, does indeed, according to the original illustrations of Attems (1937), have the gonopod telopodite reminiscent of that in E. julinum, but according to Attems, the 6 th sternum is excavated. The same applies, more or less, to E. rugegeanum.
In conclusion, considering the wide variability in acropodite structure seen in ʻtypicalʼ species of Eviulisoma, there is no justifi cation for recognizing Eoseviulisoma as a separate taxon at the generic or subgeneric level. The suggestion by VandenSpiegel & Golovatch (2014) is endorsed, and Eoseviulisoma is herewith synonymized under Eviulisoma. Of course, the name will remain available if at some point it becomes desirable to recognize subgenera in Eviulisoma. Hoffman (1964) erected Suohelisoma (monotypic for S. ulugurense Hoffman, 1964) on the basis of an unbranched gonopod telopodite without "Femoralfortsätze". Examination of the gonopod of a specimen of S. ulugurense (Tanzania, Uluguru Mts, Lupanga, W, 1900 m a.s.l., 1 Jul. 1981, M. Stoltze and N. Scharff leg., ZMUC) with the SEM (Fig. 39) has, however, revealed a small separate sclerite at the base of the acropodite. This sclerite might be interpreted as a very small mesal acropodital process, but also as a separate ʻfemoriteʼ as seen in E. articulatum (Fig. 33). If the latter interpretation is correct, S. ulugurense has no mesal acropodital process.

The status of Suohelisoma Hofmann, 1964
In other characters, S. ulugurense agrees with Eviulisoma spp., notably in the presence of a strongly excavated sternum 6 and only one transverse row of setae on postcollar body rings. The sternum 6 excavation in S. ulugurense is, however, smaller than that of Eviulisoma spp. Future analyses may show that S. ulugurense is nothing but a highly specialized Eviulisoma, but for the time being Suohelisoma is retained as a valid, separate genus.

Epizootic fungi
Millipedes are hosts to a number of fungi growing on their external cuticle. The best known of these are various species of the order Laboulbeniales Engler, which perform their entire life cycle on the host (Santamaria et al. 2014(Santamaria et al. , 2016Enghoff & Santamaria 2015). There are, however, several other groups of fungi which have at least part of their life cycle associated with millipede cuticle. What little is known about these fungi was summarized by Enghoff & Reboleira (2017), and two of the fungus types mentioned by these authors were found on one specimen of Eviulisoma chitense sp. nov. Figure 40A shows a secondary capilliconidium of the genus Basidiobolus Eidam, a fungus which has been found on several millipede species belonging to different orders and families, including Paradoxosomatidae. Figure 40B shows rows of minute spherules which seemingly come out between the cuticular scales of the millipede. Figure 40C shows similar structures on the ventral side of ring 6 in E. biquintum sp. nov. Enghoff & Reboleira (2017) hypothesized that these structures, which have been encountered on several  different millipedes and have been described as "intercalary cuticular microscutes", fi rst by Akkari & Enghoff (2011), may also be of fungal nature, although nothing defi nite can be said about this at present.

Discussion
With 22 species now known from the Udzungwa Mts, all apparently endemic, Eviulisoma matches the odontopygid genus Chaleponcus in terms of Udzungwan species richness. As in the case of Chaleponcus (Enghoff 2014) there doubtlessly are additional species of Eviulisoma to be found in the Udzungwa Mts, as indicated, i.a., by the fact that four out of 22 Udzungwan s. str. species are known only from one male (Fig. 41). Some of the species (E. coxale sp. nov., E. grumslingslak sp. nov. and E. biquintum sp. nov.) show geographical variation in colour and size and each may eventually prove to include several species. Whereas the Udzungwan species of Chaleponcus all belong to a morphologically well-defi ned group endemic to the Udzungwa and Rungwe Mts (Enghoff 2014(Enghoff , 2017, the Udzungwan species of Eviulisoma belong to several different species groups. At least three of the groups also have species outside the Udzungwas (the dabagaense group has E. culter sp. nov. in the Rubeho Mts, the sternale group has E. kangense sp. nov. in the Kanga Mts and the iuloideum group has E. iuloideum from Mufi ndi). Whether the kwabuniense group also has non-Udzungwan members it not certain, because as noted by Jeekel (2003): "Although the structure of the gonopods is basically simple, it is often quite diffi cult to identify the various branches." SEM, as fi rst applied to Eviulisoma by VandenSpiegel & Golovatch (2014), and also in the present contribution, greatly helps, but interpretation of earlier authors' drawings remains diffi cult, in some cases bordering on impossible. To quote Jeekel again: "an analysis of the genus based upon the available data in literature seems impossible on account of the [...] inadequacy of the published gonopod illustrations" (Jeekel 1968: 102). In other words, a full revision of Eviulisoma will require the restudy of a large number of type specimens as well as considerable new material of many species. I have seen several further species of Eviulisoma from other Eastern Arc blocks -Usambara Mts, Uluguru Mts, Rubeho Mts -but these belong to different species groups than the Udzungwan species and are not considered here.
In the material studied here I found no examples of polymorphism of sternum 6 and the gonopods, similar to what Carl (1909) observed. Fig. 41. The number of studied specimens (♂♂) per species of Udzungwan s. str. Eviulisoma Silvestri, 1910.

Excavation of sternum 6
The vast majority of species of Eviulisoma are characterized by a strongly excavated sternum 6 (e.g., Fig. 6). The margins of the excavation are sharp and usually carry a fringe of long setae. Carl (1909), in his description of E. fossiger, pointed out that this structure is unique among "Polydesmiden", i.e., polydesmidan millipedes. He suggested that the excavation protects the tip of the gonopods when the animal rolls up for protection. He also noticed that "among many specimens of various origin, a few are found here and there in which the gonopods are somewhat shorter and end more bluntly, and in these animals the mentioned cavity is completely missing or only very faintly indicated" (translated from German) and for this reason argued that there is only a "mechanical correlation" [between excavation and gonopods]. There may, however, be more than that.

A mysterious mass
Quite often, the excavation of sternum 6 is more or less fi lled up with an amorphous / microgranular mass (a small lump of this mass is seen in Fig. 6A), and just as often, a ʻhoodʼ of a similar amorphous mass is found attached to the tip of the mesal acropodital processes (Fig. 23E). The nature of the mass is unknown, but the shape of the ʻhoodʼ often corresponds to the shape of the sternal excavation, so a functional correlation between sternal excavation, gonopods and the amorphous mass is suggested. No spermatozoa could be distinguished in the mass even at high magnifi cation in the scanning electron microscope; since Akkari & Enghoff (2012) were able to see spermatozoa in a similarly amorphous matrix in species of the genus Ommatoiulus (Julida, Julidae), this observation speaks against interpreting the amorphous mass in Eviulisoma as a spermatophore. Possibly, the mass serves as a ʻcopulatory plugʼ preventing subsequent matings by the female. ʻCopulatory plugsʼ are common in insects, reptiles and certain groups of mammals (Stockley 1997). Amorphous masses more or less similar to those seen in Eviulisoma have been described, sometimes as spermatophores, from millipedes of the families Chordeumatidae Koch, 1847 and Tingupidae Loomis, 1966, order Chordeumatida Koch, 1847 (Verhoeff 1926(Verhoeff -1932Schubart 1939;Shear 2010), and have also been seen in the paradoxosomatid genus Desmoxytes Chamberlin, 1923 (R. Srisonchai, pers. comm.) the polydesmid genus Pseudopolydesmus Attems, 1898 (P. Sierwald, pers. comm.) and the julid genus Megaphyllum Verhoeff, 1894 (B. Vagalinski, pers. comm.).
Irrespective of whether the amorphous mass in Eviulisoma serves as a spermatophore, a copulatory plug, or both, one would expect to fi nd the mass attached to the female vulvae or their surroundings. Although females were not used for taxonomic purposes in the present study, numerous females were examined, but in no case could a mass similar to that found in males be seen on or near the vulvae. The nature and purpose of the mass remains fully enigmatic.

Modifi cations of non-gonopodal legs
Almost all species of Eviulisoma have scopulae on some or all of the walking legs of males -females have no scopulae. Scopulae (Fig. 7) are dense brushes of fl attened, often transversely ribbed setae and almost certainly play a role during copulation. In two of the species described here, E.sternale sp. nov. and E. kangense sp.nov., there is also another leg modifi cation in males: the prefemur and femur have a ventral smooth, apparently soft area (Figs 26B,28E), somewhat reminiscent of the ʻsoft padsʼ seen on the legs of many juliformian millipedes. Whether this modifi cation also exists in any previously described species is not clear.