The specific noun “atsingy” (pronounced: “a-tseen-jě”) is a Malagasy word. The terms “atsingy” or “tsingy” are the common names used to refer to the pointed and sharp calcareous lime stone formations and pinnacles originated through rainfall erosion. Although present in several other localities in western Madagascar (e.g.: Ankarana), the outcrops of Bemaraha are typical of this area and the specific name is therefore associated with the locality of provenience of the types.

This species has been referred to as Gephyromantis sp. aff. corvus “Bemaraha” by Glaw and Vences (2007), as Gephyromantis sp. 10 “Bemaraha” by Vieites et al. (2009), and Gephyromantis sp. aff. corvus by Bora et al. (2010).

MRSN A5487 (NFN), subadult male, collected at Tsingy de Bemahara National Park, western Madagascar, Andamozavaky (Bekopaka commune, Antsalova district, Melaky region, Mahajanga province), 19°01.86'S, 44°46.80'E; 122 m a.s.l., collected by J. E. Randrianirina on 23 May 2003.

MRSN A5486 (BMR 001), subadult male without evident femoral glands, MRSN A5484 (NFN), adult female, MRSN A5482 (BMR 008), MRSN A5483 (BMR 031), MRSN A5485 (BMR 002), three juveniles (sex unknown) sampled from the same locality, collector and date of the holotype (tissue sample taken for genetical analysis for all individuals); ZSM 23/2006 (FGZC 0715), adult female, from Grotte Crystal, close to Andranopasazy, Tsingy de Bemaraha National Park (Antsalova commune, Antsalova district, Melaky region, Mahajanga province), 18°42'31"S, 44°43'08"E, 146 m a.s.l., collected by F. Glaw, J. Köhler, P. Bora and H. Enting on 19 March 2006, fixed in ethanol (tissue sample taken for genetical analysis), individual found at night on limestone cliffs, close to the entrance of the cave; ZSM 37/2006 (FGZC 0746), juvenile (unknown sex) from Grotte Anjohimbazimba, Tsingy de Bemaraha National Park (Antsalova commune, Antsalova district, Melaky region, Mahajanga province), 18°41'34"S, 44°42'36"E, 160 m a.s.l., collected by F. Glaw, J. Köhler, P. Bora and H. Enting on 20 March 2006 (tissue sample taken for genetical analysis), individual found in the cave; ZSM 107/2006 (FGZC 0886), juvenile (sex unknown) from Bendrao Forest (“Camp 3”), Tsingy de Bemaraha National Park (Antsalova commune, Antsalova district, Melaky region, Mahajanga province), 18°47'04"S, 44°51'37"E, 427 m a.s.l., collected by F. Glaw, J. Köhler, P. Bora and H. Enting on 26–27 March 2006; (tissue sample taken for genetical analysis). All these specimens were fixed in 90% ethanol and preserved in 70% ethanol. UADBA 28112 (RBJ 708), female from Ranotsara (Bekopaka commune Antsalova district, Melaky region, Mahajanga province), 19°02'08"S, 44°46'29"E, 65 m a.s.l., collected by R. Andriantsimanarilafy on 18 November 2006; UADBA 28116 (RBJ 792), female from Ankilogoa (Bekopaka commune, Antsalova district, Melaky region, Mahajanga province), 19°07'52"S, 44°48'32"E, 57 m a.s.l., collected by R. Randrianavelona on 13 December 2006; UADBA 28120 (RBJ 791), female from Ankilogoa (Bekopaka commune, Antsalova district, Melaky region, Mahajanga province), 19°07'52"S, 44°48'32"E, 57 m a.s.l., collected by R. Randrianavelona on 13 December 2006; UADBA 28127 (RBJ 718), female from Ranotsara (Bekopaka commune, Antsalova district, Melaky region, Mahajanga province), 19°02'08"S, 44°46'29"E, 65 m a.s.l., collected by R. Randrianavelona on 19 November 2006; UADBA 39057 (RBJ 660), female from Anjaha (Antsalova commune, Antsalova district, Melaky region, Mahajanga province), 18°39'43"S, 44°49'33"E, 403 m a.s.l., collected by J.C. Randrianantoandro, R. Randrianavelona, R.K.B. Jenkins, R.R. Andriantsimanarilafy and E.F. Hantalalaina and Madagascar National Parks personnel on 15–24 February 2006; UADBA 39081 (RBJ 609), female from Andranopasazy (Melaky region, Mahajanga province), 18°42'31"S, 44°43'02"E, 146 m a.s.l. collected by J.C. Randrianantoandro, R. Randrianavelona, R.K.B. Jenkins, R.R. Andriantsimanarilafy and E.F. Hantalalaina and Madagascar National Parks personnel on 13–30 January 2006; UADBA 39082 (RBJ 630), female from Andranopasazy (Antsalova commune, Antsalova district, Melaky region, Mahajanga province), 18°42'31"S, 44°43'02"E, 146 m a.s.l. collected by J.C. Randrianantoandro, R. Randrianavelona, R.K.B. Jenkins, R.R. Andriantsimanarilafy and E.F. Hantalalaina and Madagascar National Parks personnel on 13–30 January 2006; UADBA 39099 (RBJ 627), adult male (with developed glands) from Andranopasazy (Antsalova commune, Antsalova district, Melaky region, Mahajanga province), 18°42'31"S, 44°43'02"E, 146 m a.s.l. collected by J.C. Randrianantoandro, R. Randrianavelona, R.K.B. Jenkins, R.R. Andriantsimanarilafy and E.F. Hantalalaina and Madagascar National Parks personnel on 13–30 January 2006; UADBA 39100 (RBJ 658), female from Anjaha (Antsalova commune, Antsalova district, Melaky region, Mahajanga province), 18°39'43"S, 44°49'33"E, 403 m a.s.l., collected by J.C. Randrianantoandro, R. Randrianavelona, R.K.B. Jenkins, R.R. Andriantsimanarilafy and E.F. Hantalalaina and Madagascar National Parks personnel on 15–24 February 2006.

A medium sized frog species (adult SVL 35–43 mm), assigned to the genus Gephyromantis (sensu Glaw and Vences 2006), subgenus Phylacomantis, according to genetic, phenetic and morphological similarities to the other known species (Gephyromantis azzurrae, Gephyromantis corvus, and Gephyromantis pseudoasper), and recognizable by the presence of the following characters:(a) femoral glands of “Type 2” (sensu Glaw et al. 2000), (b) webbing between toes present, (c) inner and outer metatarsal tubercles present, (d) tongue bifid, (e) lateral metatarsalia partly connected, (f) enlarged triangular finger tips, (g) not evident paired subgular vocal sacs, (h) crepuscular/nocturnal activity, (i) occurrence in limestone caves and deciduous forest habitat of dry western Madagascar.

Subadult male in mediocre state of preservation, with the belly opened for gonadal inspection and part of the ventral surface of thighs cut and opened to check the glands. SVL 34.8 mm; for other measurements see Tab. 1. Body slender; head longer than wide, in line with the body; snout slightly pointed in dorsal view, rather rounded in lateral view; nostrils directed laterally, much nearer to tip of snout than to eye; canthus rostralis well defined; tympanum distinct, rounded, its horizontal diameter about 50% of eye diameter; supratympanic fold well distinct, regularly curved; tongue distinctly bifid posteriorly. Arms slender; subarticular tubercles single; outer and inner metacarpal tubercles paired; fingers without webbing; finger disks triangular distinctly enlarged; nuptial pads absent. Hind limbs slender; tibiotarsal articulation reaching the nostril when hindlimbs are adpressed along body; lateral metatarsalia partly connected; inner metatarsal tubercle distinct, outer metatarsal tubercle small but recognizable; webbing of foot 1(1), 2i(1), 2e(1), 3i(2), 3e(1), 4i(2), 4e(2), 5(1). Skin slightly granular on dorsum and belly, ventral skin smooth on throat and chest. Femoral glands cluster ("Type 2", according to Glaw et al. 2000) hardly recognizable from external view, but with an overall granular structure and with 4–6 single whitish granular glands of ca. 1 mm diameter scattered on thighs. The vocal sacs in the male holotype are indistinct. The live colouration, based upon the photograph taken by J.E. Randrianirina is light brownish with darker dots and marbling (Fig. 2; A). The finger and toe tips are lighter than the remnant parts of fore- and hindlegs. After about seven years of preservation in ethanol the holotype still conserves the original marbled-brownish colour patterns, although it showed a slight loss of colour (Fig. 2; K–L). In particular, the belly became much whitish and inconspicuous. A rather characteristic and darker X-shaped marking is visible on the shoulder region, as well as a diffuse marbling darker pattern on the back and head. The tympanum is whitish. Limbs are brownish, with dark brown cross-bands: 3 on femur, 3 on tibia, 5–6 on tarsus and foot, 4 on lower arm and hand. On the flanks, the dorsal colour fades into the whitish ventral colour. The ventral side is uniformly cream-whitish on forelimbs and belly, while the throat is very lightly pigmented.

We based the current description of variability upon some specimens (paratypes and complementary individuals), part of which (ZSM 23/2006, 37/2006, 107/2006, MRSN A5486 and MRSN A5483) were also photographed in nature, and thus provided more diagnostic characters. The female ZSM 23/2006 (Fig. 2; C–E), shows a back with sparse larger warts. Its colouration appears light brown with greyish shadings, darker dots and transversal bands on the back and legs. These are more evident in the preserved individual, where a pattern of darker spots is visible on the back, suggesting the presence of a darker X-shaped drawing. These spots are visible in two other individuals, MRSN A5484 (a female) and in the holotype MRSN A5487 (Fig. 2; A, K), although for the former specimen we do not have photographs taken in life. The tympanum is uniformly brownish, and the iris is yellowish with darker reticulations. The belly is comparatively smooth, with fewer warts on its lateral parts. The throat is quite smooth. The central part of the belly is lighter than the flanks and the ventral sides of thighs, whitish on breast and thorax, with sparse darker spots. The inguinal part appears yellowish. The throat is darker than the belly, with a median lighter (although not so contrasted) line. The lateral borders of the lower jaw bear darker spots. After preservation, the colouration appears substantially similar, although faded. The juvenile ZSM 37/2006 (Fig. 2; E) presents a rather smooth back and flanks with sparse and barely evident warts. The colouration is brownish shading to the grey on the flanks and lateral parts of the back, with darker spots, extending around the flanks. The central part of the back is crossed by a longitudinal light (almost beige) band which enlarges on the head to cover the upper eyelids. The posterior part of such a band narrows to shade almost totally at the level of the vent. A thin, almost continuous whitish longitudinal line runs from the tip of the snout until the groin. The juvenile ZSM 107/2006 (Fig. 2; I) also shows a rather smooth back. The colouration is much darker, and the markings and spots are less visible. The tympanum is lighter than the surrounding areas, and the upper ridge is entoured by black pigment. Both these juveniles after about four years of preservation present a similar pattern of colouration as in life. In ZSM 107/2006 the central part of the back appears quite lighter than the surrounding areas, with a sort of arrow pattern. An interesting comparison is with the only mature available male (SVL 36.6 mm) photographed in life, the individual labelled UADBA 39099 (Fig. 2; F–H). This male appears quite slim in the photographs (either in dorsal or ventral view), with rather uniform light brown shading to greenish in life, and a moderately glandular skin texture (Fig. 2; F–G). The belly appears rather smooth in life, with the whole venter and thorax whitish (Fig. 2; H). The throat is darker with a rather indistinct central whitish band and vocal sacs are not recognizable. Lower parts of arms and thighs are pinkish, while tibiae are more whitish pigmented. The plantar surfaces are also reddish-pink. In this male, the glands are well visible and yellowish, and appear similar to those observed in Gephyromantis azzurrae, Gephyromantis corvus and Gephyromantis pseudoasper. In particular, they clearly belong to the gland “Type 2”, sensu Glaw et al. (2000), with 70 granules counted from the inner side of the right gland itself (whose external measure is 7.5*3.1 mm). In MRSN A5484 (a female) we notice a dark bar between the eyes, and an X-shaped darker spot at mid-dorsum; quite large and isolated dark spots are visible in the posterior part of the back. The belly is uniformly whitish and smooth. The three juveniles MRSN A5483, MRSN A5482, and MRSN A5485, are similar in colouration (excepting for MRSN A5483 exhibiting a light mid-dorsal line), with dark back with sparse lighter spots and shading, and almost whitish bellies. Of MRSN A5483 we also dispose of a photo taken in life, where the longitudinal light line is evident (Fig. 2; J).

According to our observations, the species lives in habitats that retain some humidity, such as rock cavities and along the walls of the canyon-like formations. One important notation comes from the fact that several of the collectors, independently (JER, FG, JCR) found this species within the caves which are typical of the area. We suspect that the species uses caves because these sites presumably have a higher humidity than the surrounding areas. In such a sense it behaves similarly to Gephyromantis corvus at Isalo, which is known to frequent narrow canyons and cave-like canyons (Mercurio et al. 2008). Apparently, the new species (both adults and juveniles) is not confined to the proximity of water, and it has been observed jumping among the tsingy pinnacles also far from water bodies. All the individuals were active at night on tsingy rocks or during the day in caves. No data are available about mating behaviour, advertisement calls and tadpole morphology.

Only known from the localities of the type specimens within the Tsingy de Bemaraha National Park.

Gephyromantis atsingy sp. n. differs from Gephyromantis pseudoasper, Gephyromantis azzurrae and Gephyromantis corvus by the lack of paired blackish skin folds (vocal sacs) along the lower jaws in adult males, and from Gephyromantis azzurrae also by details of colouration (see below). Following our measurements, adult males of Gephyromantis atsingy can also be differentiated among each other by the number of granules in the femoral glands: 70 granules in Gephyromantis atsingy; 96 granules in Gephyromantis corvus; 38–45 granules in Gephyromantis azzurrae and 39–43 granules in Gephyromantis pseudoasper. In addition, the new species differs from all three species by substantial genetic differentiation (see below).

All the described species of Gephyromantis, subgenus Phylacomantis, show similarities with Gephyromantis atsingy (Tab. 2). The dorsal pattern is similar in all species, showing an assemblage of darker spots and reticulations on the lighter background, and barred legs and arms. The dorsal colouration in Gephyromantis atsingy is usually light brown-beige, with a somehow greenish shading, while in Gephyromantis corvus it is uniformly grey or dark grey with sparse darker (uniformly-sized) warts and dots. Notwithstanding, the examined specimens of Gephyromantis atsingy have a much more contrasted X-shaped dark spot on the back. This is less evident in Gephyromantis corvus, where the dark-light pattern is more confuse and irregular. We observed a longitudinal repetition of lighter elements, a longitudinal light band or a middorsal light line only in Gephyromantis atsingy. The belly in both species is light, but in Gephyromantis atsingy we detected more frequently the darker drawing with a lighter central area on the throat and chest. According to the original description and subsequent papers (Mercurio and Andreone 2007, Mercurioet al. 2008), Gephyromantis azzurrae has a quite variable dorsal colouration. The holotype of the species, as depicted by Glaw and Vences (2007), has a wide lighter dorsal band upon a darker dorsal colouration, and the belly is reddish. Other examined specimens of Gephyromantis azzurrae present more uniform dorsal colouration. In both species the dorsal skin is featured by the presence of similar larger warts. In comparison to Gephyromantis atsingy, the Gephyromantis pseudoasper specimens are smaller and have a more warty back. The colouration in Gephyromantis pseudoasper is much darker and the belly is much more pigmented: the throat, the thorax and the anterior part of the belly are heavily spotted in dark, with a clear median light line on the throat. The posterior parts of the belly and parts of the ventral side of the legs in Gephyromantis pseudoasper are often orange. The external vocal sacs are evident and well developed, while these are not visible in Gephyromantis atsingy.

The molecular data confirm the attribution of Gephyromantis atsingy to the subgenus Phylacomantis (Glaw and Vences 2006, Vieites et al. 2009). The analyzed specimens of Gephyromantis atsingy, Gephyromantis azzurrae, Gephyromantis corvus and Gephyromantis pseudoasper appear genetically very uniform and show an intraspecific uncorrected divergence of 0.5%, 0.4%, 0.1% and 0.1% respectively, in the 16S rRNA gene sequences. The genetic distance between Gephyromantis atsingy and the three other Phylacomantis species ranges between 10.2% (comparison between Gephyromantis atsingy and both Gephyromantis corvus and Gephyromantis pseudoasper) and 11.2% (comparison between Gephyromantis atsingy and Gephyromantis azzurrae). Among the genus Gephyromantis the smallest genetic distance is observed between Gephyromantis corvus and Gephyromantis azzurrae (9.1%) and the highest uncorrected divergence between Gephyromantis azzurrae and Gephyromantis pseudoasper (13.1%). Gephyromantis corvus and Gephyromantis pseudoasper have a genetic distance of 12%. These divergences are comparatively high among mantelline species (see Vences et al. 2005, Vieites et al. 2009), and corroborate the species status of Gephyromantis atsingy. The phylogenetic relationships between the species of the Phylacomantis subgenus have been resolved recently (N. Kaffenberger et al., in preparation). These analyses confirm the monophyly of the subgenus, provide evidence for the basal position of Gephyromantis pseudoasper and uncover the sister relationship between Gephyromantis atsingy and the complex made of Gephyromantis corvus and Gephyromantis azzurrae.

This species appears to be restricted to the Bemaraha Plateau, where it has been found in seven localities within the Tsingy de Bemaraha National Park. It may also occur in the Réserve Naturelle Intégrale, which forms the northerly limit of the Bemaraha Plateau, but survey data are lacking from this site. Within the national park, some areas of forest are damaged by conversion to agriculture and charcoal production, but the humid canyons where G. atsingy occur are generally well protected. We therefore recommend assigning a category of Near Threatened because the species nearly qualifies for listing as Vulnerable under D2: the species is confined to a single site, the Bemaraha Massif (1, 577 km2), with a plausible threat that could impact the species in the near future. If the threat became operational, the species would be eligible for listing as Endangered since its extent of occurrence is well within the 5, 000km2 threshold under the B criterion and it would occur at a single location (where the threat is habitat loss from agricultural activities and charcoal production) and there would be a continuing decline in the quality and area of habitat, qualifying the species for the criteria B1ab(iii).