A mountain of millipedes III: A new genus for three new species from the Udzungwa Mountains and surroundings, Tanzania, as well as several ‘orphaned’ species previously assigned to Odontopyge Brandt, 1841 (Diplopoda, Spirostreptida, Odontopygidae)

. The new genus Geotypodon gen. nov. is described. It includes two species from the Udzungwa Mountains: G. millemanus gen. et sp. nov. (type species) and G. submontanus gen. et sp. nov., one species from nearby Iringa: G. iringensis gen. et sp. nov., and 18 previously described species hitherto incorrectly assigned to Odontopyge Brandt, 1841.


Introduction
This is the third in a series of articles about the millipedes, especially the endemic Afrotropical family Odontopygidae, of the Udzungwa Mountains, Tanzania.For general information on the Odontopygidae and the Udzungwa Mountains see the fi rst article in the series (Enghoff 2014); see also Enghoff & Frederiksen (2015).
In the present article three further new odontopygid species are described from the Udzungwa Mountains and their immediate surroundings (Fig. 1).The new species resemble several species which were assigned to the genus Odontopyge Brandt, 1841 by Kraus (1960) and authors before him.However, Hoffman (1991) showed that Odontopyge had been misinterpreted by virtually everybody dealing with the genus after its original description and that Odontopyge is the correct name for a genus until then known as Omopyge Manfredi, 1941. Hoffman (1991) rectifi ed the situation and formally synonymized Omopyge under Odontopyge, but by this action he left a large number of species originally described in or subsequently transferred to Odontopyge 'orphaned', i.e., without a valid genus name.Some of these have been transferred to other genera by Hoffman & Howell (1981, 2012), Hoffman (2000Hoffman ( , 2002) ) and Frederiksen & Enghoff (2015), but very many remain orphaned.Kraus (1960), with updates as indicated.Part 1 of the list contains species placed by Kraus and subsequent authors in Odontopyge.Part 2 contains the "nomina dubia" listed by Kraus (1960).Numerous additional species originally described in Odontopyge and transferred to other genera by Kraus and his predecessors are not listed.
Part 1. Species assigned to 'Odontopyge sensu auctorum' by Kraus and subsequent authors.
Specimens were examined in alcohol under a stereo microscope.Specimens for scanning electron microscopy (SEM) were transferred to 96% ethanol, then to acetone, air-dried, mounted on aluminium stubs or on pieces of fl exible aluminium tape and in turn mounted on stubs, coated with platinumpalladium and studied in a JEOL JSM-6335F scanning electron microscope.
See Enghoff (2014) for the description standards used.

Diagnosis
A genus of Odontopygidae-Prionopetalini characterized by: a long basad metaplical spine on the anterior side of the coxa, a compact torsotope, lack of pretorsal or torsal spines/processes, a pronounced posttorsal narrowing without spines, a division of the telopodite into solenomere and telomere immediately distal to posttorsal narrowing, a spine emerging from the base of the telomere and curving more or less parallel to the basal parts of the solenomere, a slender, whip-like, smooth solenomere without any outgrowths or appendages (except sometimes a tiny subapical spine), a highly three-dimensional telomere consisting of various lobes and lamellae with largely smooth margins.

Etymology
An anagram of Odontopyge.To be regarded as masculine in analogy with other names ending in -odon.

Descriptive notes
Species of Geotypodon gen.nov.have a typical odontopygid habitus (Fig. 2) and are mostly mediumsized by odontopygid standards (cf.Kraus 1966).Published numbers of podous rings range from 49 (G.ollieri) to 72 (G.multianulatus) (one has been subtracted from the published numbers because these include the telson), and published male body diameters range from 1.7 mm (G.sennae) to 5.5 mm (G.specularis).The majority of species have 54-66 podous rings and a diameter of 4.3-5.5 mm, but there is a group of smaller species (G.angolanus, G. erratus, G. heteromodestus, G. ollieri) with 49-54 podous rings and a diameter of 2.2-3.2mm.G. sennae is a particularly slender species: 66-68 podous rings and 1.7 mm diameter, and G. multianulatus, true to its name (at least as misspelled by Kraus (1960): multiannulatus), stands out with 72 podous rings and 4.8 mm diameter.Of the new species described here, G. millemanus gen.et sp.nov.falls neatly in the main group, G. submontanus gen.et sp.nov. is slightly more slender than the main group, whereas G. iringensis gen.et sp.nov.has a larger body diameter than any other described congener (Fig. 3).

Remarks
Although Geotypodon gen.nov. is proposed to accommodate several species of Odontopyge sensu Kraus (1960), its diagnosis and circumscription are narrower.Species of Odontopyge sensu Kraus (1960), may thus lack a long basad metaplical spine on the anterior side of the coxa (present in Geoptypodon), may have spines in the torsal region (absent in Geotypodon gen.nov.), and may lack a spine emerging from the base of the telomere and curving more or less parallel to the basal parts of the solenomere (present in Geotypodon gen.nov.).
A spine emerging from the base of the telomere is also found in several other 'Odontopyge' species, e.g., citernii Silvestri, 1898, diffi cilis Silvestri, 1895, and severini, Silvestri, 1897 -see Kraus (1960), where this spine is called "Tibialdorn".It is also found in Rhamphidarpoides species (Frederiksen & Enghoff 2015), in Calyptomastix kakandae (Kraus, 1958), as well as several Spinotarsus species, etc.In the Chaleponcus dabagaensis-group there is a spine at almost the same place, but emerging from the base of the solenomere instead of the base of the telomere (Enghoff 2014).
Considering the notorious mosaic-like distribution of morphological characters throughout the family, the genus Geotypodon gen.nov.as defi ned here is quite possibly not a monophyletic group, cf. the Discussion section (see below).Until a more satisfactory analysis of relationships within Odontopygidae becomes available, the new genus can, however, serve as a 'home' for several (but not all) 'orphaned' species hitherto classifi ed in Odontopyge.metaplica, the telomere is apically divided into two lobes of approximately equal size, and the limbus lobes are multi-cusped, hand-like.It shares these characters with G. submontanus gen.et sp.nov., but differs from that species in larger size, straight mesal margin of metaplical lamella and absence of a spine-like process on the posterior apical telomeral lobe.

Etymology
The name is a Latin noun in apposition, meaning "a thousand hands" and referring to the hand-like limbus lobes.A specimen with a diameter of 5 mm will have a body perimeter of approximately 15 mm.Assuming that the limbus covers only 10 mm, and taking into account that each "hand" occupies about 10 microns of the perimeter, each body ring will carry about a thousand "hands", and a specimen with 60 body rings will therefore have roughly 60,000 "hands".
COLOUR.After 15 years in alcohol somewhat faded, but pattern still evident.Upper part of head blackish, lower part yellowish.Overall colour of body ventrally and laterally yellowish.Posterior ca.40% of metazona amber, in front of amber zone a blackish zone; blackish zone narrow laterally, becoming wider dorsally where extending onto posterior ¼ -⅓ of prozona; resulting mid-dorsal dark band becoming narrower towards head.Telson blackish with yellowish margins, legs yellowish.
COLLUM.With a marginal and a submarginal furrow.BODY RINGS.Almost perfect cylinders, not vaulted; suture straight; ozopores ca.three diameters behind suture.
LIMBUS (Fig. 4).Consisting of isolated hand-like lobes of ca. 10 microns' length, each with 3-6 'fi ngers' and each with a ridge running along its length.
ANAL VALVES.Each with a well-developed dorsal spine and a smaller, yet distinctive ventral one; margin raised, with 3 setae not borne on tubercles.MALE LEGS.From 5 th pair with postfemoral and tibial pads on all legs, except last four pairs; tibial pads absent from several pairs in front of these.

Diagnosis
A species of Geotypodon gen.nov. in which the gonopod coxal metaplica is produced mesad in a thin lamella, a small longitudinal lamella, perpendicular to the other, is present on the anterior part of the metaplica, the telomere is apically divided into two lobes of approximately equal size, and the limbus lobes are multi-cusped, hand-like.Shares these characters with G. millemanus gen.et sp.nov., but differs from that species in smaller size, bicuspid mesal margin of metaplical lamella and presence of a spine-like process on the posterior apical telomeral lobe.

Etymology
The name is a Latin adjective referring to the habitat.

Distribution and habitat
Known only from West Kilombero FR.Altitude: 1145 m asl.Habitat: submontane forest.

Diagnosis
A species of Geotypodon gen.nov. in which the gonopod coxal metaplica is apically rounded, the telomere is apically divided into a small posterior lobe and a large anterior lobe, which gives rise to a strongly curved terminal lobe, and the limbus lobes are pointed triangular.

Etymology
The species is named after the type locality.
COLOUR.After 19 years in alcohol uniform greyish; posterior part of metazona amber, legs and antennae dark brownish.
COLLUM.With a marginal and a submarginal furrow.BODY RINGS.Almost perfect cylinders, not vaulted; suture straight; ozopores ca.three diameters behind suture.
ANAL VALVES.Each with a well-developed dorsal spine and a smaller, yet distinctive ventral one, margin raised, with 3 setae not borne on tubercles.MALE LEGS.From 4 th pair with postfemoral and tibial pads on all legs, except last few pairs where only postfemoral pads are present.GONOPOD COXA (Fig. 6).Basally parallel-sided, with antero-lateral concavity (lc) (Fig. 6A).Proplica ending in small proplical lobe (hidden behind anterior edge of metaplica on Fig. 6A).Metaplica with poorly developed basal anteriad fl ange (mf) (Fig. 6A), distally regularly rounded and projecting laterad as semicircular lobe; metaplica at level of proplical lobe, with a long, latero-basad process (msp) on anterior side of coxa (Fig. 6A-B); process straight and slender in anterior view, slightly curved and broader in lateral view.

Distribution and habitat
Known only from the environs of Iringa city.The altitude of the type locality will be at ca. 1600 m asl.

Coexisting species
No other odontopygid species were found together with G. iringensis gen.et sp.nov.

Notes
Geotypodon iringensis gen.et sp.nov. is very similar to G. multianulatus from Kenya, but there are differences, including that in G. multianulatus the gonopod coxa has a much larger, narrower and less rounded lateral lobe, and the solenomere has a small subdistal spine (cf.Discussion section).

Discussion
Of the three species described here, G. millemanus gen.et sp.nov.and G. submontanus gen.et sp.nov., both from the Udzungwa Mountains proper, are particularly similar, notably sharing the lamellar structure of the mesal part of the coxal metaplica and the apically divided telomere with an intermediate lamella (itl) between the apical and posterior distal lobes (atl and ptl).G. iringensis gen.et sp.nov.does not have the metaplical lamella, but its telomere has the same distal elements (atl, itl, ptl) as the two other species.In most other species of Geotypodon gen.nov.as here defi ned, the telomere does not show a similar apical subdivision.Exceptions are G. multianulatus, G. sennae and G. specularis, where the telomere is divided into two large lobes (Attems 1914(Attems , 1927;;Brölemann 1903).
The unsatisfactory state of odontopygid taxonomy was characterized as follows by Hoffman & Howell (2012): "That classifi cation of odontopygid millipeds remains in a highly unsettled condition is due both to the inherent complexity of the male genitalia and the traditional reliance on a few obvious key-characters (a priori defi nition) instead of groupings made on the basis of overall similarity of the appendages.Either approach is further complicated by frequent contradictory states of characters as expressed in coxal or telopodital regions.Almost identical coxal forms may recur randomly amongst taxa defi ned on the basis of the telopodite and thought to be not closely related.Body form tends to be of monotonous similarity throughout the group, and female genitalia have so far provided very few insights into relationships."Hoffman & Howell went on to speculate that "The impression is thus gained of a group of organisms which have stabilized their general Gestalt whilst expressing innate genetic variability dominantly in permutations of male reproductive structures.A young evolutionary status is implied by the paucity of strong discontinuities in character systems and typical spectral expression of traits."Anybody who has tried to refer odontopygid specimens to a genus will agree that this is not easy.Whereas the gonopods of each species are often highly characteristic, when it comes to grouping the species, the problems are massive.
As an illustrative example, one may mention the record of Odontopyge cf.picea by Dieudonné (2014).The record is illustrated with a very nice optical photo of the gonopods, and they do indeed look very much like Geotypodon picea gen.et comb.nov.Examination of several specimens from among the material recorded by Dieudonné, kindly put at my disposal by Didier VandenSpiegel (Royal Museum for Central Africa, Tervuren, Belgium), did, however, reveal that whereas the specimens in most respects agree with the defi nition of Geotypodon gen.nov.given above, they differ in a striking detail: just distal to the posttorsal narrowing where the solenomere and the telomere begin, there is a peculiar sclerite protruding from the main telopodite axis.The basal telomeral spine sometimes curves between this sclerite and the main axis.Such a sclerite is absent from the true G. picea (Attems 1938: fi g. 32).Furthermore, the solenomere of these specimens has a tiny spine-like side branch near the tip, whereas the solenomere tip is not visible on Attems' fi gure and is not mentioned in the description.(As noted above, one of the few differences between G. iringensis gen.et sp.nov.and G. multianulatus is that the latter species has a similar small subapical spine.) An alternative approach to odontopygid classifi cation is clearly needed, and molecular characters are the obvious choice.This will, however, require a huge effort.A search for "Odontopygidae" in GenBank yields no results whatsoever, so there is scope for large-scale sampling and sequencing of well-vouchered odontopygids.
This way, a more satisfactory classifi cation may one day be obtained, but until then, in order to be able to handle the amazing diversity of odontopygids, several existing genera need to be revised and probably split into several smaller genera.This approach might seem to be at variance with the ideas expressed by Hoffman & Howell (2012), i.e., "traditional reliance on a few obvious key-characters (a priori defi nition) instead of groupings made on the basis of overall similarity", but at least as a temporary solution such a splitting approach will be useful.Taking the monographic work of Kraus (1960Kraus ( , 1966) ) as the starting point, several such splits have already been made by Demange (1981), Frederiksen & Enghoff (2015), Hoffman (2002) and Hoffman & Howell (2012).One further split is proposed in the present paper.
Around 400 species of Odontopygidae have been described (Enghoff 2014), but virtually every new collection of millipedes from the Afrotropical region contains undescribed species.The number of described odontopygid species can therefore potentially be multiplied by an unknown factor (5? 10?), and the number of genera will -at least temporarily -also need to be multiplied, hopefully to a somewhat smaller degree.

Fig. 3 .
Fig. 3. Body size of males of Geotypodon spp.Bold symbols indicate numbers of podous rings and midbody vertical diameter of the new species described here.Small circles and shaded areas indicate published measurements for other Geotypodon species.

Table 1 .
Species of Odontopyge sensu auctorum nec Brandt, 1841.The table is based on