New species of Prosopodesmus Silvestri, 1910 (Diplopoda, Polydesmida, Haplodesmidae) from Queensland, Australia

Abstract Prosopodesmus crater sp. n., Prosopodesmus kirrama sp. n. and Prosopodesmus monteithi sp. n. are described from the Wet Tropics of north Queensland. The hothouse species Prosopodesmus panporus Blower & Rundle, 1980 is recorded from rainforest on Queensland’s Cape York Peninsula, where it is likely to be native.


Introduction
Prosopodesmus Silvestri, 1910 is a small genus of tropical haplodesmid Polydesmida with three transverse rows of large tubercles on the metatergites and downward-bending paranota (Golovatch et al. 2009). The unbranched, sickle-shaped gonopod telopodites normally lie parallel to the main body axis in a hollow formed by ventromedial concavities on the gonocoxae. Two species have been described in Prosopodesmus. The type species P. jacobsoni Silvestri, 1910 is a pantropical tramp first collected at Jakarta ('Batavia') on Java. It has since been found in Florida and Puerto Rico (USA), Haiti (as the synonymous Homodesmus parvus Chamberlin, 1918), St Eustatius (Leeward Islands, West Indies), Panama, Brazil, Zanzibar, India, New Caledonia and the Galapagos Islands (Hoffman 1999). It has recently been confirmed as present on Taiwan (Golovatch et al. 2011) and there are P. jacobsoni records from Louisiana (USA) , Christmas Island (Indian Ocean) (Jeekel 2006) and Fiji (Akkari and Enghoff 2011). The native range of P. jacobsoni is unknown.
P. panporus Blower & Rundle, 1980 was described from tropical plant hothouses at the Royal Botanic Gardens at Kew, England (Blower and Rundle 1980). The describers did not suggest a native range for the species: "Presumably P. panporus lurks in some unworked tropical habitat." (Blower and Rundle 1980, p. 32).
Two more species were added to Prosopodesmus when the Japanese genus Rhipidopeltis Miyosi, 1958 was made a junior synonym (Golovatch et al. 2009, Golovatch et al. 2010. P. sinuatus (Miyosi, 1958) is known only from far southern Honshu and P. similis (Haga, 1968) from Kyushu, to the south of Honshu across the Kanmon Strait.
While sorting museum samples of Pyrgodesmidae from eastern Australia, I found four species of pyrgodesmid-like Prosopodesmus in Berlese samples from rainforest in tropical north Queensland. Three of the species are new, while the fourth is P. panporus.

Methods
'Male' and 'female' in the text refer to adult individuals. All specimens are stored in 75-80% ethanol in their respective repositories.
Whole, selected specimens and rings 7 of selected males were cleared in 80% lactic acid and temporarily mounted in 60% lactic acid for optical microscopy. Preliminary gonopod drawings were traced from digital images acquired with an eyepiece video camera temporarily fitted to an optical microscope. Body parts of uncleared specimens were temporarily mounted in a 1:1 glycerol:water mixture for examination and measurement. Photomicrographs were taken with a Canon EOS 1000D digital SLR camera mounted on a Nikon SMZ800 binocular dissecting microscope equipped with a beam splitter. Specimens for scanning electron microscopy were briefly air-dried, temporarily fixed to a stub with double-sided adhesive tape or a sticky carbon pad, examined uncoated with a FEI Quanta 600 operated in low-vacuum mode, then returned to alcohol. Images and drawings were prepared for publication using GIMP 2.6. The map figure was generated using ArcView GIS 3.2.
The specimen localities given below are also listed in Appendix 2 in table form. Latitude and longitude (WGS84 datum) are from museum collection databases. The uncertainty for each locality (in square brackets) is the radius of a circle around the stated position, and is my own estimate.
Spiracles not evident. Telson (as for P. monteithi sp. n in Fig. 1B) facing ground; preanal ring with 5 lobes on each side and 1 larger lobe (epiproct) apically. Spinnerets recessed in individual chambers, basal sheaths with unnotched distal edges; setal shafts smooth. Paraprocts more or less flat, rounded-rectangular; paraproctal setae close to and equidistant from margin. Hypoproct trapezoidal, hypoproctal setae not on raised tubercles.
Telopodite base in shallow recess at posteroventral corner of gonocoxa. Telopodites (Figs 3A, 5A) slender, straight, more or less uniformly wide, parallel, just touching when retracted, slightly convex anteriorly, slightly concave ventrally; bent posteriorly at tip; with thin, rounded tabs projecting posteriorly from lateral edge at between 1/4 and 1/2 the telopodite length and just basal to the apical bend; with thin, rounded tab similarly extending the medial edge just distal to the bend. Telopodites with a few larger setae to about 1/2 telopodite length on posterior surface laterally; with numerous fine setae around and just inside basal concavity into which the prominent cannula inserts. Prostatic groove running straight to small, low mound on middle of posterior telopodite surface at about 3/4 telopodite length; mound covered with 'hairpad' of slender, pointed villi.
Female with head + 20 rings; a little larger than male, ca 8 mm long. Epigyne ca 1/3 ring width, slightly raised, rounded rectangular with straight distal edge; cyphopods not examined.
Distribution. Rainforest on the Atherton Tableland southwest from Cairns, Queensland, with a known north-south range of ca 40 km (Fig. 8). There is also one questionable record from the Cammoo Caves area in central coastal Queensland. (See Remarks.) Etymology. Latin crater, 'cup'. The type locality surrounds Lake Eacham, a crater lake on the Atherton Tableland in far north Queensland.
Remarks. The male in 'Other material' is from a site nearly 800 km south and east from the nearest other P. crater locality. This specimen is from the same ANIC berlesate that indicated a similar 800+ km disjunction in the range of the unrelated millipede Asphalidesmus magnus Mesibov, 2011(Mesibov 2011. Although the Cammoo Caves P. crater is very slightly different in gonopod details from those collected on the Atherton Tableland, the most likely explanation is that the Cammoo Caves specimens of P. crater and A. magnus were actually collected in the Wet Tropics, and that the locality labels are incorrect. Figure 5. Medial views of right gonopod of Prosopodesmus species, not to same scale. A P. crater sp. n., paratype, ANIC 64-000214. B P. kirrama sp. n., paratype, QM S91626. C P. monteithi sp. n., QM 91635. D P. sinuatus (Miyosi, 1958), holotype, drawing scanned and modified from Fig. 1G in Miyosi (1958). e P. similis (Haga, 1968), holotype, drawing scanned and modified from Fig. 12B in Haga (1968 Diagnosis. Males and females with head + 20 rings; adults 8-9 mm long; midbody metatergites typically with 3 transverse rows of 12 large tubercles; posterior portion of prozonite ( Fig. 2B) with small disks, no microtubercles; ozopores not on porosteles; gonopod telopodite with two posterior bends, one at about midlength and one near tip, the more basal bend marked by strong anterior production of telopodite.
Gonocoxae ( Fig. 3B) massive, strongly tapering anteroventrally and posteroventrally. Telopodite (Figs 3B, 5B) strongly produced anteriorly at about midlength, bending posteriorly there and again near tip. Hairpad mound medial on posterior surface midway between second bend in telopodite and tip. Thin, rounded tab on medial edge of telopodite just below level of hairpad; thin, rounded tabs on lateral edge at level of first bend and just distal to hairpad; narrow longitudinal ridge medially on posterior telopodite surface from near base to near first bend.
Distribution. Rainforest in the mountains southwest from Tully and northwest from Ingham, Queensland, with a known north-south range of ca 25 km (Fig. 8).
Telopodite (Figs 4, 5C) slender, smoothly curving in J-shape; hairpad mound at about midlength; a small triangular tab directed basally near curved-over tip; tip apically slightly excavate, the lateral side extended and terminating in 3 minute, finger-like processes.
Distribution. Rainforest from Daintree National Park west of Cape Tribulation to the Malbon Thompson Range on the coast southeast from Cairns in Queensland, a north-south range of ca 125 km (Fig. 8).
Etymology. For Geoff Monteith, former curator of insects at the Queensland Museum. Geoff and his colleagues collected most of the specimens of the three new Prosopodesmus species described in this paper.
Remarks. P. monteithi is the largest known Prosopodesmus and the striking dorsal macrosculpture is easily visible to the unaided eye (Fig. 6A). One of the Queensland Museum samples contained a mating pair (Fig. 6B) which I partially dissected (Fig.  6C). As expected, the telopodites were rotated 90° out of the gonocoxal cavity in which they normally lie. The curved distal portion of each telopodite (Fig. 5C) was fully inserted into the cavity anterior to the epigyne, but how much of the curve was actually in contact with the cyphopod could not be seen, and would be better investigated with fixed, sectioned material.

Diagnosis.
Males with head + 19 rings, females with head + 20; adults ca 3.5-4 mm long; midbody metatergites typically with 3 transverse rows of 10-12 large tubercles; posterior portion of prozonite with small disks and microtubercles; ozopores on porosteles on all podous rings beginning with ring 5; gonopod telopodite bent posteriorly at midlength, lateral edge near tip with several rounded teeth.
Distribution. So far known in Queensland from a single site in rainforest in the Iron Range Resource Reserve, ca 10 km northwest of Lockhart on the Cape York Peninsula (Fig. 8).

Is P. panporus native to Queensland?
The Queensland P. panporus site is several kilometres inside a block of tropical rainforest several hundred square kilometres in extent. What is now the Old Coen Track, off Portland Roads Road, was used for vehicle access (Tom Weir, in litt., 24 April 2012). The site's remoteness is evidence that P. panporus is native there rather than introduced. A second reason to think that P. panporus is native to Queensland is that three of its congeners, described in this paper, were collected in little-disturbed tropical rainforest ca 450 km to the south.
The case for native status would be even stronger if the Kew hothouses where P. panporus was first collected in 1975-76 were known to have contained plants imported from the Cape York Peninsula. I queried the Gardens but was told that glasshouse records for that period were not detailed enough to answer this question (Roxana Glenn, in litt., 11 April 2012).
It seems very likely, but not certain, that P. panporus is native to the Cape York Peninsula. Sequencing of a marker such as mitochondrial COI, from the Kew population and from specimens collected across the range of the species in Queensland, might help to locate the source or sources of the hothouse millipedes in the wild.

Notes on morphology
The close resemblance of Rhipidopeltis sinuata to the Prosopodesmus species known at the time led Golovatch et al. (2009) to synonymise the two genera. They were aware that Miyosi (1958) had shown the prostatic groove terminating at the tip of the telopodite (Fig. 5D), rather than at a subapical hairpad. Golovatch et al. (2009) considered this difference to be only species-specific. Haga (1968) illustrated the gonopod of R. similis with the prostatic groove ending, again, at the telopodite tip (Fig. 5E), and with no hairpad or other projection, only a tooth-like tab at midlength on the telopodite's medial edge. The two species groups (with and without a hairpad termination for the prostatic groove) are clearly very closely related and I leave them here in Prosopodesmus.
The fine structure of the prozonite of P. panporus (Fig. 2E) corresponds exactly to that of P. jacobsoni as illustrated in Akkari and Enghoff (2011;Fig. 4, p. 7): anteriorly with cellular chambers whose walls are extended with microvilli, posteriorly with slightly elevated convex disks ('subspherical knobs' of Akkari and Enghoff) on a ground covered with microtubercles. These are not genus-level character states, since the other three Australian Prosopodesmus species lack microvillose extensions anteriorly and microtubercles posteriorly, and the posterior prozonite in P. monteithi lacks disks and is irregularly rugose with microvilli (Figs 2A-D).
I could not find spiracular openings above the legbases in any of the Australian Prosopodesmus species. Although this is consistent with the apparent lack of spiracles in two other Australian haplodesmids, Agathodesmus johnsi Mesibov, 2009and A. steeli Silvestri, 1910(Mesibov 2009, it is very puzzling. Spiracles are clearly visible in all known species of Asphalidesmus Silvestri, 1910 (Dalodesmidea) and in the Australian Pyrgodesmidae I have examined. Both groups are similar to Prosopodesmus species in body form, size range and cryptic habits in rainforest. Unless the two haplodesmid genera lack tracheae, which seems to me improbable, then their tracheal systems must have spiracular entrances. A histological study of freshly collected and suitably prepared Prosopodesmus specimens, e.g. the large P. monteithi, might solve this puzzle.
The following is a composite description based on all the material in our possession. Specific reference to one or another of the types is made in the legends to the figures.
Head almost completely covered by a fan shaped (sectorial) collum; metazonite of second ring broader than the metazonites of all subsequent rings, the paranota forwardly directed, embracing the head and collum, their lateral edges confluent with the sides of the collum and continuing the lines of same; the division between the collum and second ring in life is not immediately evident in dorsal view.
The antennae are inserted close together on the vertex; when at rest, they occupy deep grooves diverging dorsally and then passing posteriorly, their distal antennomeres overhung by the sides of the collum and the paranota of the second ring. The forehead triangle enclosed within the diverging antennal bases and the anterior edge of the collum is raised into seven tubercles, one at the apex, two in the next row and four forming the base beneath the collum edge. The sixth antennomere is the longest and broadest; the fifth is slightly shorter. Each of these two larger antennomeres carries a group of sensory cones on their distal dorsal extremities.
The dorsal surface of the collum is raised into an even pavement of tubercles; those of the outermost row are flatter than the remainder and form the incised anterior edge of twelve lobes; these are followed by four rows of more convex tubercles. The succeeding metazonites are strongly arched and are raised into three transverse rows of regular tubercles uniform with those of the collum. The metazonites of the third and fourth rings are much less broad than those of the second ring and slightly less broad than the fifth and all subsequent rings except the telson. The paranota are placed slightly below the mid-lateral level and are incised into three lobes; the middle lobe is slightly broader but not more protrusive; the posterior lobe is more clearly separated from the other two and is directed slightly postero-laterally.
Ozopores occur on all diplopodous segments, 5-17 in the male and 5-18 in the female and there is a pair of rudimentary pores on the single apodous ring (ring 18 of the male and 19 of the female). The pores are borne on the outermost pair of tubercles, of the middle row of the fifth and of the posterior row of the sixth to the last podous ring. The poriferous tubercles of the fifth metazonite obscure the middle lobe of the paranotal edge in dorsal view; those of the remaining diplopodous rings obscure the posterior paranotal lobes. The poriferous tubercles are directed slightly posteriorly, a tendency most obviously developed on the last two or three diplopodous rings where they project posterolaterally beyond the posterior limit of the metazonite.
The anterior row of tubercles on the metazonites slightly overhang the anterior edge. The third row of tubercles stop just short of the posterior edge which is incised into flat fields corresponding in number to the tubercles; these flat posterior portions of the metazonite could just conceivably be described as a fourth row of rather flat tubercles. There are eight tubercles in the first and second rows of the anterior rings, passing to ten on the sixth and eventually twelve from the eighth onwards. The posterior row on rings one to five include 9 tubercles increasing to 11 from the sixth to the last ring, the outermost pair carrying the ozopores, which on the fifth ring are borne on the outermost pair of tubercles of the second row. The telson carries the same three rows of tubercles although the numbers in the second and third rows naturally diminish towards the apex. The tubercles along the posterior edge of the telson overhang as 6+6 lobes.
The entire surface of the tubercles and the region between them is raised into microtubercles and micropapillae, as also is the surface of the prozonites (figs 2, 3). The only macrosetae are carried on the telson, anal valves and scale; there are 4+4 inserted just beneath the twelve-lobed posterior edge of the telson, level with the dorsalmost 3+3 of these lobes. A further four prominent setae are housed in a collar at the apex of the telson just below the medianmost lobes. Each of the anal valves carries 2+2 setae and a further pair are carried by the trapezoidal anal scale. None of these macrosetae is visible from the dorsal side due to the ventrally directed curvature of the telson.
The gonopods are shown in figs 6-8 along with those of P. jacobsoni. In P. panporus they are simple uniramous structures terminating in prominent teeth, one apical, two subapical arising at about the same level and a further two (sometimes a third rudimentary tooth) along the external edge of the ramus just below the first sub-apical pair. Figs. 6-8 indicate different appearances of these teeth when viewed from different aspects.  (Silvestri, 1910) 6 P. panporus n. sp. External profile view of right gonopod of paratype 2 (the specimen is mounted with left side uppermost, but the drawing is reversed to compare more easily with the other two) 7 P. panporus, gonopods of the holotype in situ, viewed from the right and slightly ventrally 8 P. panporus, gonopods of paratype 1 in situ, viewed from the right. [Reproduced with permission from Blower and Rundle (1980), p. 31]