Description of a new microhylid frog species of the genus Xenorhina ( Amphibia : Anura : Microhylidae ) from the Fakfak Mountains , far western New Guinea

According to the results of recent molecular studies by Frost et al. (2006) and Köhler & Günther (2008), the genus name Xenobatrachus has to be considered as a junior synonym of Xenorhina because no stringency was found for a separation of both groups into different monophyletic taxa. There is one important feature to discriminate between Xenorhina and the former members of Xenobatrachus which can be still used in determination practice: the latter posseses vomero-palatine spikes (lacking in Xenorhina old sense). Xenorhina species without vomero-palatine spikes are Description of a new microhylid frog species of the genus Xenorhina (Amphibia: Anura: Microhylidae) from the Fakfak Mountains, far western New Guinea


Introduction
According to the results of recent molecular studies by Frost et al. (2006) and Köhler & Günther (2008), the genus name Xenobatrachus has to be considered as a junior synonym of Xenorhina because no stringency was found for a separation of both groups into different monophyletic taxa.There is one important feature to discriminate between Xenorhina and the former members of Xenobatrachus which can be still used in determination practice: the latter posseses vomero-palatine spikes (lacking in Xenorhina old sense).Xenorhina species without vomero-palatine spikes are Description of a new microhylid frog species of the genus Xenorhina (Amphibia: Anura: Microhylidae) from the Fakfak Mountains, far western New Guinea Guinea, and only the offshore island Yapen is presently known to be settled by three species (Price, 1994;Günther & Richards, 2005;Günther & Knop, 2006).
Together with native friends I collected some specimens of a hitherto unknown Xenorhina species in September 2008 in the Fakfak Mountains, located on the "throat" of the Vogelkop Peninsula in western Indonesian New Guinea, which is scientifi cally described in this paper.

Materials and methods
Frogs were collected at night after locating them by their advertisement calls.All specimens were photographed in life the next day, anaesthetised with chlorobutanol and subsequently fi xed in 2 % formalin.Tissue probes from the thigh muscle were taken from one specimen (ZMB 74631) and stored in 96 % ethanol to enable later DNA sequencing, before fi xing the animal in formalin.All specimens were later transferred to 75 % ethanol in the museum collection.One specimen (ZMB 74631) was cleared and stained as an osteological preparation according to a method modifi ed from Dingerkus & Uhler (1977).
The following measurements were taken with a digital calliper (> 10 mm) or with a binocular dissecting microscope fi tted with an ocular micrometer (< 10 mm) to the nearest 0.1 mm from preserved specimens only: SUL snout-urostyle length from tip of snout to distal tip of urostyle bone; SUL is about one to two mm shorter than the snout-vent length (SVL).As the measurement error is higher in the latter, I prefer to use the former.In general, both measurements are more or less identical and are used interchangeably in this paper; TL tibia length: external distance between knee and ankle; TaL length of tarsus: external distance, tarsal and ankle joints held at right angles; T4L length of 4th toe: from tip of toe to proximal end of inner metatarsal tubercle; T4D transversal diameter of disc of 4th toe; F3L length of third fi nger; F3D transversal diameter of disc of 3rd fi nger; T1L length of fi rst toe, distal of the inner metatarsal tubercle; MTL length of the inner metatarsal tubercle; HL head length, from tip of snout to posterior margin of tympanum; HW head width, taken in the region of the tympana; SL snout length, from an imaginary line connecting the centres of the eyes to tip of snout; END distance from anterior corner of orbital opening to centre of naris; IND internarial distance between centres of nares; ED eye diameter, from anterior to posterior corner of orbital opening; TyD horizontal diameter of tympanum.
Advertisement calls were recorded under natural conditions with a Sony Digital Audio Tape (DAT) Walkman TCD-D 100 and a Sennheiser microphone MKE 300 and analysed with Avisoft-SAS Lab Pro software.All specimens are currently stored in the Museum für Naturkunde Berlin (ZMB) and bear registration numbers of this institution.Part of the type series will be transferred to the Museum Zoologicum Bogoriense (MZB) after completion of my studies.
Material compared: one paratype of Xenorhina eiponis stored in the American Museum of Natural History (AMNH 128234); two paratypes of X. arboricola stored in the Bernice P. Bishop Museum Honolulu (BPBM 13745 and 13747); two syntypes of X. oxycephala from the National Museum of Natural History, Leiden (RMNH 2280A and 2280B); the holotype of X. bouwensi stored in the collection of the Institut Royal des Sciences Naturales de Belgique, Brussels (IRSNB 1019); the type series of X. varia (Museum für Naturkunde Berlin, ZMB); the holotype of X. macrodisca (Museum Zoologicum Bogoriense, MZB Amph.10916); and some specimens of X. bouwensi and X. oxycephala collected by R. Günther in recent years and stored in the ZMB collection.Moreover, morphometric and other data, published by the authors mentioned above, were appraised.
Figure 7 is by Björn Stelbrink (ZMB), all others are by the author.with circum-marginal grooves, tips as wide as, or even narrower, than corresponding penultimate phalanges; all toe tips with discs wider than corresponding penultimate phalanges; relative length of fi ngers 3 > 2 > 4 > 1; relative length of toes 4 > 3 > 5 > 2 > 1; except for a rather large metatarsal tubercle, no distinct subarticular, plantar or palmar tubercles present.Body sides in life and in preservative with some distinct tubercles, a few tubercles also on dorsum and on dorsal surfaces of extremities; all ventral surfaces smooth.Tip of snout smooth and not pustulose, as, for example, in Xenorhina lanthanites.

Xenorhina arndti sp. nov.
Colour in life.Dorsal surface of head, body and extremities uniform brown, lower fl anks white with a brownish network.Conspicuous is a dark brown supratympanic ridge (Fig. 4).A whitish mid-dorsal line extends from eye level to urostyle and further as a broken line along the posterior thighs.Anal region blackish; lower tarsi, anterior and posterior tibiae, knee-region, and dorsal surfaces of fi ngers and toes exhibit a pattern of irregular dark brown and whitish spots.Throat brown with irregular darker and lighter spots; snout tip from below dark brown; chest and belly creamcoloured with some irregular white spots (Fig. 5).Iris nearly circular, blackish with a few golden speckles and a golden inner margin.Colour of the animal in preservative differs only slightly from that in life.

Morphological characters of the paratypes.
For measurements of the paratypes see Table 1.While paratype ZMB 74630 exhibits a light brown dorsal colour and a similar arrangement of the colour pattern as the holotype, the dorsal surfaces of paratype ZMB 74631, as in the holotype, are darkish brown, but in life this brown colour as well as all dorsal and lateral light areas were more or less covered by blue (Fig. 6).With exception of the brownish margin of the throat, all other ventral surfaces in ZMB 74630 were whitish.The distribution of darker pigments on ventral surfaces of ZMB 74631 was more like that in the holotype, but basic colour was not cream-coloured but a light orange.Both paratypes have a fi ne yellowish middorsal line.
Osteology.Unfortunately, the bone-and-cartilage preparation was only partly successful.Consequently, various osteological characters cannot now be properly appraised.However, the genus traits and a lack of vomero-palatine spikes could clearly be assessed.

Molecular evidence.
According to B. Stelbrink and T. von Rintelen (pers.comm., September 2010) DNA isolation and PCR were done using the protocol of Köhler & Günther (2008).Forward and reverse strands were aligned using Codon-Code Aligner v.3.0.3.(CodonCode Corporation, Dedham, MA, Diagnosis.A typical representative of its genus, with a small head, acuminate snout, small eyes (average ED/ SUL 0.077), and a rather uniform colour.With a snouturostyle length of between 23.8 and 27.8 mm in three males, the new species belongs to the small members of its genus.Circum-marginal grooves present on all fi nger and toe discs, discs of all toes wider than corresponding penultimate phalanges, discs of all fi ngers not wider than penultimate phalanges; lacking vomero-palatine spikes.Of those species lacking vomeropalatine spikes, only Xenorhina adisca, X. bou wensi, and X. minima are of a size similar to the new species.Xenorhina adisca has no discs on fi ngers and toes; X. bouwensi has shorter tibiae (mean of TL/SUL 0.44, mean in X. arndti 0.49), a higher ratio of T4D/F3D (mean in X. bouwensi 2.06, mean in X. arndti 1.64), and different advertisement calls; X. minima has no discs on fi ngers and toes and shorter legs than the new species.Xenorhina arndti seems to spend most of its life underground, and its advertisement call consists of a chain of musical notes and is uttered from small holes in the ground.
Description of the holotype.For measurements see Tab. 1. Head in the region of the tympana broader than long (HL/HW 0.82) and merging seamlessly into the wider body; snout acuminate from above and slightly protruding in profi le; loreal region oblique, no canthus rostralis, nostrils near tip of snout, directed dorsolaterally, visible from above but not from below; internarial distance smaller than eye diameter; tympanum clearly visible, nearly as large as the eye (TyD/ED 0.95); supratympanic fold extends from posterior corner of eye, touches the dorsal margin of tympanum and reaches up to insertion of foreleg; tibia very long for the genus (ratio TL/SUL 0.51); fi ngers rather short; no webbing between fi ngers and toes (Figs. 2 and 3); all fi nger tips answer the question of whether X. oxycephala in the present sense shows an extreme polymorphism in its mitochondrial genes or whether it consists of several taxa.
The analysis of 480 base pairs of the 16S rRNA gene and 680 bp of the 12S rRNA gene revealed that the new species (ZMB 74631) is a sister taxon of Xenorhina cf.oxycephala (ZMB 74628 and ZMB 74640) that occurs syntopically in the Fakfak Mountains.Xenorhina cf.oxycephala (ZMB 69562) from the Wondiwoi Mountains, located about 260 km W of the Fakfak Mountains, is obviously a sister taxon of X. varia (ZMB 65133, ZMB 65136 and ZMB 65137) from Yapen Island, while X. cf.oxycephala from Yapen Island is a sister taxon of X. varia + X. cf.oxycephala from the Wondiwoi Mountains (Fig. 7).All X. cf.oxycephala specimens were determined on the basis of external morphology, using the holotypes and the key by Zweifel (1972) as a reference.According to the molecular characters, shown above, it is very doubtful that X. oxycephala inhabits areas in New Guinea as large as, for example, suspected by Zweifel (1972) and Menzies (2006).More detailed molecular and behavioural (mating calls) studies should be conducted to     sistently to the end (Fig. 8, above).The fi rst note is the shortest and lowest of all notes.There is an increase in note length and audibility from note one to note fi ve or six, then note length remains more or less stable.The audiospectrogram (Fig. 8, below) shows six harmonics with weak frequency modulation.According to the power spectrum (Fig. 9), the fundamental frequency band centres around 0.80 kHz und exhibits by far the most energy (dominant frequency).The following frequency bands decrease in energy and their main frequencies are around 1.50 kHz, 2.25 kHz, 3.0 kHz, 3.60 kHz and 4.40 kHz.
Comparison with other species.Xenorhina arndti sp.nov.differs from the majority of its congeners by lacking vomero-palatine spikes.From the congeners without vomero-palatine spikes, the species X. arboricola (SVL 38 -41 mm), X. eiponis (SVL 35 mm), X. macrodisca (SVL 41 mm), X. oxycephala (SVL 40 -50 mm), X. parkerorum (SVL 70 mm), X. similis (SVL about 50 mm), and X. varia (SVL 39 mm) are larger than the new species.In X. eiponis, which possibly overlaps in size, the discs of fi ngers and toes are of a similar size while in X. arndti toe discs are ince, Indonesia (Fig. 1).Besides the specimens that were collected, we heard only a few more calling in a sector between 400 m and 860 m a.s.l.These latter frogs were found at greater distances from one another, and calls were uttered rather infrequently.This situation could mean that the species is indeed rare, but it is also not impossible that we visited the area outside of the mating season.All three specimens uttered their calls from small holes at 5 -15 cm beneath the ground surface.They inhabited primary and secondary rain forest with a greater or lesser developed understory.
Vocalisation.All calls were heard at between 6 and 10 p.m. Three complete calls of one male were analysed.All three have exactly 12 notes.Call duration was from 5.69 s to 5.80 s, mean 5.73 s.Note length was from 57 ms to 88 ms, and the mean of 31 notes was 77 ± 7.1 ms.Twenty-eight internote intervals ranged from 408 ms to 501 ms, mean 438 ± 26.2 ms.Mean note repetition rate was 2.1 notes per s, range 2.07 -2.12 notes per s.
Notes are melodious and unpulsed.They start with maximum amplitude and decrease more or less con-

Fig. 3 .
Fig. 3. Ventral view of right foot of the holotype of Xenorhina arndti sp.nov.

Fig. 2 .
Fig. 2. Ventral view of right hand of the holotype of Xenorhina arndti sp.nov.

Fig. 7 .
Fig. 7. Bayesian phylogram of the concatenated data set (12S rRNA and 16S rRNA).The numbers on branches are maximum parsimony bootstrap values, maximum likelihood bootstrap values, and posterior probabilities of Bayesian inference (from top to bottom).

Fig. 8 .
Fig. 8. Wave form (above) and spectrogram (below) of three notes from the middle section of an advertisement call from Xenorhina arndti sp.nov.
Body measurements and body ratios of the type series of Xenorhina arndti sp.nov.ZMB-No are the inventory numbers of the Museum für Naturkunde Berlin, FN are the author's fi eld numbers.ZMB 74629 is the holotype, ZMB 74631 is now an osteological preparation.All three specimens are adult males.All measurements are in mm; abbreviations are explained in "Materials and methods".
(Günther & Knop, 2006;Menzies, 2006) legs in the genus(Günther & Knop, 2006;Menzies, 2006).Besides shorter tibiae, X. bouwensi also exhibits a greater ratio T4D/F3D (1.67 -2.66, mean 2.06) than X. arndti (1.60 -1.71, mean 1.64), has another structure of the advertisement call notes, has a more than 10 % genetic difference in the 12S gene, and has a more than 15 % difference in the 16S gene.Etymology.The specifi c epithet is a patronym in genitive singular.With this specifi c name I acknowledge certain fi nancial support and editorial assistance by my dear colleague Dr. Rudolf G. Arndt of The Richard Stockton College of New Jersey, Pomona (New Jersey/USA) for my studies on the herpetofauna of New Guinea in recent years.clearlywider than fi nger discs.Xenorhina adisca is smaller (SVL 18 -24 mm) than X. arndti and has no discs on fi ngers and toes.Xenorhina minima has very short legs (TL/SVL < 0.40) and fi ngers and toes without discs.Xenorhina bouwensi differs from the new species by shorter tibiae.In 14 specimens of X. bouwensi, the ratio TL/SUL varies from 0.40 -0.50, mean 0.44; X. arndti with a TL/SUL ratio of 0.47 -0.51, mean 0.49, has, together with X. eiponis, X. ophio-Tab.1.