Tridactylogonus obscurus Jeekel, 1982, by original designation.

T. rugosissimus Jeekel, 1982, T. warrenbenensis sp. n.

Male, Warrenben Conservation Park, SA, -35.1102, 137.0222 ±25 m, 30 m a.s.l., 16 August 2017, R. Mesibov and C. Arnold, open she-oak woodland, SAM OM2184.

3 males, 7 females, details as for holotype, SAM OM2185-OM2194.

1 male, 2 females, 1 juvenile, Warrenben Conservation Park, SA, -35.0926, 137.0121 ±100 m, 20 m a.s.l., 3 June 2016, R. Mesibov and T. Moule, degraded she-oak woodland, SAM OM2169; 1 juvenile, same details but -35.0922, 137.0464 ±100 m, 40 m a.s.l., 4 June 2016, burned she-oak and eucalypt woodland, SAM (not registered); 1 male, 2 juveniles, same locality but -35.1113, 137.0184 ±25 m, 30 m a.s.l., 15 August 2017, R. Mesibov, eucalypt and tea tree woodland, SAM OM2195-OM2197; 1 juvenile, same details but -35.1125, 137.0152 ±25 m, tea tree copse, SAM OM2198; 1 juvenile, same details but -35.1107, 137.0122 ±25 m, degraded she-oak woodland, SAM OM2199.

Differs from T. obscurus in having variably rugose rather than smooth metatergites, and in the anteromedial process of the gonopod telopodite being flattened rather than lanceolate. Differs from T. rugosissimus in having one process extending from the base of the solenomere rather than two. Differs from both T. obscurus and T. rugosissimus in lacking a femoral process or tubercle on male leg 1.

Male/female approximate measurements (all adults): length ca 12/14 mm, maximum midbody width 1.1/1.3 mm. Colour in alcohol (Fig. 2) light to medium brown, lightening ventrally, with yellowish paramedian bands dorsally, the bands on the prozonite closest together at the waist (Fig. 2C). Head yellowish laterally. Antennae brown, legs pale, in both cases darkening distally.

Male with vertex bare, frons sparsely setose, clypeus moderately setose; vertigial sulcus distinct, ending just above level of antennal sockets; post-antennal groove shallow; antennal sockets separated by ca 1 socket diameter. Antennae clavate, reaching dorsally to rear of ring 2; antennomeres with relative lengths 6>(2=3)>(4=5); 6 thickest. Collum (Fig. 2A) half moon-shaped, strongly convex, rear margin straight, corners rounded and slightly upturned. Head slightly wider than collum; collum to ring 18 nearly uniform in width but rings 2 and 3 slightly narrower. Ring 2 paranotum (Fig. 2A) thin, dorsally concave, set lower than collum corner and ring 3 paranotum, extending slightly past posterior ring margin. Paranota on rings 3 and 4 (Fig. 2A) similar but thicker. Paranota on rings 5–18 (Figs 2A–D, 4A) prominent, set at ca 1/2 ring height; in lateral view rounded anteriorly, bluntly pointed posteriorly, extending just past posterior ring margin; dorsally concave medial to thickened lateral margin (Fig. 3A). No pleural keels on anterior rings.

Waist (Figs 2C, 3A) distinct, the zone between suture and anterior metazonite margin longitudinally ridged. Prozonites (Fig. 3) with prominent cellular sculpture dorsally, laterally and ventrally. Metatergites with transverse row of 4 prominent setae anterior to transverse furrow, the setae often abraded; posterior rings (Fig. 2B) with transverse row of setae near rear margin of metatergite. Metazonites laterally and ventrally with very small, irregular ridges (Fig. 4A, E) with flattened tops and rounded edges. Metatergites sometimes also with low, irregular folds, giving metatergite a variably wrinkled appearance (Figs 2D, 3A). Transverse furrow (Figs 2C, 3A) on rings 3–18 at ca 1/2 metatergite length, deeply impressed with some longitudinal ridging. Limbus a short, thin, uniformly wide sheet. Pore formula normal; ozopore (Fig. 4A) round, opening laterally at anterior end of narrow, ovoid depression at rear of paranotal margin. Spiracles on diplosegments above and just anterior to leg bases, rims thin; anterior spiracle (Fig. 4D) with rim slightly elevated, filter slightly emergent dorsally, filter elements flattened and with forked tips. Midbody sternites about as long as wide, sparsely and finely setose, cross impressions about equal in depth; no cones or other projections on any sternites. Midbody legs (Fig. 2D) with relative podomere lengths (femur=tarsus)>prefemur>(postfemur= tibia). Pre-anal ring (Fig. 2B) with a few long setae; epiproct extending well past anal valves, in dorsal view slightly tapering and truncate, tip ca 1/3 width of pre-anal ring; hypoproct paraboloid. Spinnerets in square array, dorsal setae longer than ventral, each pair set in narrow, shallow, transverse concavity on posterior surface of epiproct.

Leg 1 without ventral femoral process or tubercle (Fig. 4C). Gonopore small, round, opening on slight distomedial bulge of leg 2 coxa. Sternal lamella (Fig. 4B) between legs 5 ca 2/3 as wide as space between leg 5 coxae, short, distally with 2 bluntly rounded, setose projections. Sparse brush setae on leg 2 tarsus only. Anterior leg prefemora not swollen dorsally. Gonopod aperture (Fig. 3B) just wide enough to accommodate gonocoxae, ca 1/2 ring 7 prozonite width. Gonopod telopodites (Figs 3B, 5C, 5E) straight, parallel, reaching leg 6 bases when retracted; sternite between legpairs 6 and 7 slightly excavate. Gonocoxa short, truncate-conical, with a few long setae anterolaterally. Cannula prominent. Telopodite with prefemoral portion moderately setose medially, marked distally by very slight constriction and by obvious reduction in telopodite width. Femoral portion ca 1/2 acropodite length, subcylindrical, ending posteriorly in lip-like extension. Distal half of acropodite with two (not three) processes: broad, laminate, distally rounded-truncate solenomere, slightly concave anteriorly and broadly emarginate medially; and short anteromedial process, ca 1/2 solenomere length, directed slightly distomedially, laminate with rounded, sparsely microdentate distal margin. Prostatic groove prominent, running distally along anteromedial surface of femoral portion, then curving between bases of solenomere and anteromedial process and along anterior surface of solenomere, terminating as short, central projection on distal solenomere margin.

Female more robust than male; epigyne thickened but barely protruding; cyphopods not examined.

For the type locality, Warrenben Conservation Park.

Known from six localities over ca 4 km² in Warrenben Conservation Park at the southern end of the Yorke Peninsula, South Australia (Fig. 1A, B). Found in bark litter under dead she-oak trees (Allocasuarina stricta) and in Eucalyptus sp. bark litter in shrubby, mallee-type vegetation on limestone at 20–30 m elevation. The area has an annual rainfall of ca 440 mm (Bureau of Meteorology 2017).

Taxonomic affinities. Although its gonopod telopodite is “bidactylous” rather than “tridactylous”, T. warrenbenensis sp. n. closely resembles the other two Tridactylogonus species in its small size and gonopod form. The genus was thought by Jeekel (1982, p. 128) to “stand rather isolated” within the tribe Antichiropodini, and its discovery “might seem to narrow the taxonomic disjunction between this tribe and the Australiosomatini”. However, as in other Australian Antichiropodini, especially Aethalosoma Jeekel, 2006, Aulacoporus Verhoeff, 1924, Brochopeltis Verhoeff, 1924, Pseudostrongylosoma Verhoeff, 1924 and Walesoma Verhoeff, 1928, the gonopod telopodite in Tridactylogonus has a narrow, straight femoral portion arising from a small, setose prefemoral portion, with the femoral portion clearly demarcated from the solenomere and any other apical processes. Also, as in Aethalosoma, Dicranogonus Jeekel, 1982, Notodesmus Chamberlin, 1920 and Pogonosternum Jeekel, 1965 (fig. 2 in Mesibov 2009, fig. 4B in Decker et al. 2017), the spiracular filter elements in the new species are flattened and have forked tips.

Within the genus, T. warrenbenensis sp. n. is most similar to T. rugosissimus in gonopod form and in metazonite sculpturing (see below), which Jeekel (2002, p. 65) thought could distinguish the latter species “from all other known Australian Paradoxosomatidae”. The new species is remarkable in lacking a femoral process or tubercle on male leg 1, a character almost universally present in Australiosomatinae. Jeekel reported that the leg 1 femur in T. obscurus had “the usual ventral tubercle” (Jeekel 1982, p.130), and in T. rugosissimus “a small ventral tubercle” (Jeekel 2002, p. 64). Unfortunately I have not been able to examine the type and only known specimen of T. rugosissimus; it was not deposited in the South Australian Museum as proposed (Jeekel 2002, p. 60) and has not been found among material in the late Dr Jeekel’s study collection in the Naturalis Biodiversity Center in Leyden, the Netherlands (K. van Dorp, in litt., 17 September 2014).

Surface sculpture. Jeekel (1982, p. 129) noted “a fine, but quite conspicuous cellular structure” on the prozonites and “subgranulose” sides in T. obscurus. In T. rugosissimus the prozonites were said to have either “pronounced reticulate structure or minute and regular granulation”, while the metatergites were “coarsely and densely rugose, with irregular longitudinal or somewhat oblique or undulate rugae”, and the sides “coarsely granulose to subrugulose” (Jeekel 2002, p. 63). Cellular prozonite sculpturing seems to be a constant character state in T. warrenbenensis sp. n. and closely resembles the prozonite sculpturing in another Australian paradoxosomatid, Desmoxytoides hasenpuschorum Mesibov, 2006 (fig. 5A in Mesibov 2009). “Rugose” sculpturing of the metazonites is more variable. It is always present on the sides as irregular microscopic ridges, but is variably prominent on the metatergites as coarse, narrow wrinkles. Each microscopic ridge appears to project from one cuticular scale, but ridges on adjoining scales sometimes appear to be confluent (Fig. 4E), in contrast to the discretely spaced projections seen in some other recently examined Polydesmida (Mesibov 2009, Akkari and Enghoff 2011).

Biogeography and conservation. The three Tridactylogonus species occur around Spencer and St Vincent Gulfs in southern South Australia (Fig. 1B) and form an allopatric species mosaic. T. warrenbenensis sp. n. is so far known only from Warrenben Conservation Park (Fig. 1A), which covers ca 4000 ha and is managed for nature conservation. Over several days of searching in 2016 and 2017 I was unable to find any native millipedes in the nearby and larger Innes National Park (Fig. 1A), despite the presence there of apparently suitable woodland and scrub habitats. The National Park instead has a dense and almost ubiquitous population of the introduced Portugese millipede Ommatoiulus moreleti (Lucas, 1860). O. moreleti is also the most common millipede elsewhere on the Yorke Peninsula in both partly natural and entirely agricultural habitats. If O. moreleti has displaced native millipedes in Innes National Park, it may displace T. warrenbenensis sp. n. in future in Warrenben CP, which is linked to Innes NP by several large blocks of privately owned bushland. During my 2016 and 2017 visits I saw no O. moreleti in Warrenben CP.