Leaping towards a saltatorial lifestyle? An unusually long-legged new species of Rhombophryne (Anura, Microhylidae) from the Sorata massif in northern Madagascar

The Madagascar-endemic microhylid genus Rhombophryne consists of a range of partly or completely fossorial frog species. They lead a poorly known, secretive lifestyle, and may be more diverse than previously thought. We describe a new species from the high altitude forests of the Sorata massif in north Madagascar with unusual characteristics for this genus; R. longicrus sp. n. has long, slender legs, unlike most of its fossorial or semi-fossorial congeners. The new species is closely related to R. minuta, a much smaller frog from the Marojejy massif to the southeast of Sorata with similarly long legs. We discuss the morphology of these species relative to the rest of the genus, and argue that it suggests adaptation away from burrowing and toward a more saltatorial locomotion and an accordingly more terrestrial lifestyle. If this is the case, then these frogs represent yet more ecological diversity within the already diverse Cophylinae. We recommend an IUCN Red List status of Endangered B1ab(iii) for R. longicrus sp. n., because it is known only from a single site in a forested area of roughly 250 km2, which is not yet incorporated into any protected area.


Introduction
The microhylid frog genus Rhombophryne (Anura, Microhylidae, Cophylinae) is receiving renewed taxonomic attention in the wake of recent genetic barcoding efforts on the frogs of Madagascar.These DNA barcode studies revealed that less than 60% of the island's amphibian diversity might so far have been formally described (Vieites et al. 2009, Perl et al. 2014).The taxonomic gap between what has been described and the total diversity that exists is extreme in the cophyline microhylids, a Madagascar-endemic radiation of narrow-mouthed frogs.The cophyline genus Rhombophryne typifies the taxonomic gap: In 2007, it contained eight described species (Glaw and Vences 2007).Thirteen candidate species were identified by integrative inventories is morphologically very similar to Plethodontohyla (Andreone et al. 2005).Rhombophryne contains obligate and facultative burrowing frogs (Andreone et al. 2005, Wollenberg et al. 2008), and the enigmatic R. minuta, the ecology of which we discuss below.Stumpffia, a genus in need of revision, consists of mostly terrestrial dwarf frogs, some of which number among the smallest frogs in the world (Glaw and Vences 2007).Recently, Peloso et al. (2015) proposed the synonymy of Stumpffia and Rhombophryne.However, we here refrain from adopting their changes pending further investigation, as these genera are morphologically distinct.Plethodontohyla species resemble Rhombophryne more closely in size, but consist of a mixture of probably facultative burrowers and arboreal or semi-arboreal climbers.Reasons for this ecological flexibility, and the ancestral states of the genera, have so far been poorly explored (Andreone et al. 2005).
Here, we describe a new Rhombophryne clearly distinct from all other known species, although phylogenetically nested.We describe its skeleton by means of micro-computed tomography (micro-CT), revealing additional differences to the other Rhombophryne species for which skeletal data are available (Scherz et al. 2014(Scherz et al. , 2015, unpublished results), unpublished results).We discuss the implications of this potentially novel morphology for the evolution of the genus Rhombophryne, and highlight the need for herpetological survey work in less accessible areas of Madagascar.

Materials and methods
Specimens were collected, euthanized, fixed in 90% ethanol and transferred to 70% ethanol for long-term storage.One specimen was deposited in the Université d'Antananarivo Département de Biologie Animale (UADBA), and the other in the Zoologische Staatssammlung München (ZSM).
In this work we refer to the species R. ornata and R. tany.Their descriptions are in press (Scherz et al. 2015), and we therefore give the disclaimer that the use of these two names here should be considered conditional, and these two names herein are not made nomenclaturally available, in accordance with Articles 8.3 and 15.1 of the International Code of Zoological Nomenclature (ICZN 1999).
Morphological measurements were taken with a digital calliper to 0.01 mm and rounded to the nearest 0.1 mm for presentation here.Fig. 1 shows our measurement scheme except cumulative measures (HIL, FORL, and FOTL).Abbreviations are used as follows: SVL (snout-vent length), HW (maximum head width), HL (head length, from the maxillary commissure to the anterior-most point of the mouth), ED (horizontal eye diameter), END (eye-nostril distance), NSD (nostril-snout tip distance), NND (internarial distance), TDH (horizontal tympanum diameter), TDV (vertical tympanum diameter), HAL (hand length, from the metacarpal-radioulnar articulation to the tip of the longest finger), LAL (lower arm length, from the carpal-radioulnar articulation to the centre of the radioulna-humeral articulation), UAL (upper arm length, from the centre of the radioulna-humeral articulation to the trunk, measured along the posterior aspect of the arm), FORL (forelimb length, given by the sum of HAL, LAL, and UAL), FOL (foot length, from the tarsal-metatarsal articulation to the tip of the longest toe), TARL (tarsal length, from the tarsal-metatarsal articulation to the tarsal-tibiofibular articulation), FOTL (foot length including tarsus, from the tibiotarsal articulation to the tip of the longest toe, given by the sum of FOL and TARL), TIBL (tibiofibula length), TIBW (tibiofibula width at thickest point, measured in dorsal aspect), THIL (thigh length, from the vent to the femoral-tibiofibular articulation), THIW (thigh width at thickest point, measured in supine position), HIL (hindlimb length, given by the sum FOL, TARL, TIBL, and THIL), IMCL (maximum length of inner metacarpal tubercle), IMTL (maximum length of the inner metatarsal tubercle).
Micro-CT scanning was carried out on a nanotom m (GE Measurement & Control, Wunstorf, Germany).The holotype of the new species, ZSM 1630/2012, was mounted on a polystyrene board inside a sealed polyethylene vessel, and secured in place using small wooden struts and additional polystyrene.A small volume of 80% ethanol was added to achieve air saturation, preventing desiccation of the specimen.The vessel was mounted on a polyvinylchloride tube, and placed inside the micro-CT scanner.Scanning was conducted at a voltage of 140 kV and a current of 80 mA, with a timing of 500 ms for 20 minutes (2440 projections).Scan data were assembled in phoenix DATOS|X 2 RECON-STRUCTION CT software (GE Measurement & Control, Wunstorf, Germany) and visualised in VG STUDIO MAX 2.2 (Volume Graphics GmbH, Heidelberg, Germany) and subsequently processed into a 3D surface render in AMIRA 5.4.5 (FEI Visualization Sciences Group, Burlington MA, USA).Skeletal description is based on both surface and volume renderings of micro-CT data, due to artefacts produced in surface rendering.Readers are advised that micro-CT scanning does not render poorly calcified structures, especially cartilage.Cartilages are therefore omitted from the osteological description below and the respective figures; additional specimens will need to be cleared and stained in order to assess cartilaginous characters of these frogs.We provide a PDF-embedded interactive 3D model of the skeleton of the holotype as Suppl.material 1. Osteological terminology follows Trueb (1968Trueb ( , 1993)).Skull ratio measurements were calculated from high resolution TIFF images of prepared models in ImageJ 1.48v (Schneider et al. 2012).
We extracted total genomic DNA from ethanol-preserved tissue samples using proteinase K digestion (final concentration 1 mg/mL) and a standard salt extraction protocol (Bruford et al. 1992).We amplified a fragment of the mitochondrial large ribosomal subunit or 16S rRNA (16S) via polymerase chain reaction (PCR) using the primers 16Sar-L and 16Sbr-H (Palumbi et al. 1991).Sequences were resolved on an ABI 3130xl automated DNA sequencer (Applied Biosystems) and newly determined sequences were deposited in GenBank (accession numbers KR025897 and KR025898).
We calculated a phylogenetic tree from the 16S sequences by Bayesian inference (BI) with MRBAYES 3.2 (Ronquist et al. 2012) after determining a GTR+I+G substitution model as best fitting the data using JMODEL-TEST (Darriba et al. 2012).We considered all sections of the 16S gene with more than two consecutive gaps in one or more sequences as ambiguous and excluded these sections from the alignment (total alignment length after exclusion: 397 nucleotides).Explorative analyses including these stretches resulted in an identical topology and similar support values for all nodes supported with PP > 0.9.We ran two independent analyses for 20 million generations, each comprising four Markov Chains (three heated and one cold), and sampled every 10,000 generations.Chain mixing and stationarity was assessed by examining the standard deviation of split frequencies and plotting the -lnL per generation using TRACER 1.6 software (Rambaut et al. 2014).Results were combined to obtain a 50%-majority rule consensus tree and the respective posterior probabilities of nodes, after discarding 25% of the generations as burn-in (all compatible nodes with probabilities <0.5 kept).

Results
The new species described below is phylogenetically nested in the genus Rhombophryne (Fig. 2) and placed as sister to R. minuta with strong support.The relatively long branch length indicates, however, a strong genetic differentiation from its probable sister species.Uncorrected pairwise differences (p-distances) using the entire amplified 16S fragment (including hypervariable regions; alignment length 536 bp) was 6.8% to R. minuta and >8.6% to all other species of Rhombophryne (including R. guentherpetersi; M. Vences, analyses in progress).The concordance of such a high genetic divergence (>3% in the 16S fragment; see Vieites et al. 2009) with clear morphological differentiation is strong evidence for status as an independent evolutionary lineage, warranting its recognition as a distinct species.We therefore formally describe this new species here as: Paratype.UADBA-A 60271 (FGZC 3651), an adult male with the same collection data as the holotype.

Diagnosis.
A microhylid assigned to the genus Rhombophryne on the basis of overall morphology, including the possession of maxillary and vomerine teeth, absence of expanded toe pads, and absence of an enlarged prepollex.Confirmed as a member of the genus Rhombophryne on the basis of its phylogenetic relationships as assessed by mitochondrial DNA, as there are no known morphological characters by which Rhombophryne may be distinguished from Plethodontohyla.
Rhombophryne longicrus sp.n. is distinguished from all other Madagascan frog species by the following set of characters: SVL 23.8-27.9mm, head wider than long, horizontal tympanum diameter 47% of eye diameter, absence of superciliary spines, weak supratympanic fold, dark supratympanic region and nostril, tibiotarsal articulation reaching the nostril, total hindlimb length 183-185% of SVL, second finger shorter than fourth, and fifth toe shorter than third.It is also separated by a pairwise genetic distance of at least 6.8% in the 16S mitochondrial gene from all other known species of the genus Rhombophryne.
Rhombophryne species can be confused with Plethodontohyla species.Rhombophryne longicrus sp.n. differs from them in the following ways: absence of a sharp dorsolateral colour border and expanded finger and toe pads (versus presence in P. notosticta, P. guentheri, P. mihanika, and P. inguinalis), absence of inguinal spots (versus presence in P. mihanika, P. inguinalis, P. ocellata, and P. bipunctata), tibiotarsal articulation reaching the nostril (maximally reaching to the mid-eye in all Plethodontohyla except P. mihanika), absence of crossbands on legs (versus presence in P. fonetana, P. inguinalis, P. notosticta, P. guentheri, and P. mihanika), and smooth skin (versus granular to rough in P. tuberata).
Description of the holotype.Adult female in an excellent state of preservation.A ventral incision was made in order to check the sex and access the stomach contents.The incision runs laterally and posteroventrally anterior to the pubis and up the middle of the venter.
Colouration of the holotype: (Fig. 3a, c).In life, snout anterior to eyes, above eyes, side of head, and upper arms bronze to tan in colour; tip of snout darker, lightening posteriorly; area around nostril black; supratympanic region dark brown, fading below to the tan of the lateral side of the head.Body laterally light brown, becoming increasingly yellowish brown dorsolaterally until tan border with dark dorsal marking; this marking is flecked with additional tan spots, and extends from a black horizontal bar between the eyes to the legs, where dark ashy grey dominates; border between dorsal and dorsolateral colouration almost symmetrically emarginated.Hands and feet tan with dark flecks.Ventral skin pinkish and slightly translucent; chin dark relative to rest of venter, posteriorly lightening with few darker patches interspersed with translucent areas lacking pigment.Anteroventral surface of legs with dark pigment, becoming less pigmented more ventrally; ventral surface mottled pinkish and light brown.
After three years in 70% ethanol, all browns have faded to shades of grey.Dorsal areas that were lightest in shade Creative Commons Attribution 4.0 licence (CC-BY); original download https://pensoft.net/journalsare whitish, particularly between the eyes anterior to the dark inter-ocular bar.Ventrally, all areas that lack pigmentation and were pink in life are cream in preservation.Chin the same colour as the snout.
Osteology of the holotype: (Fig. 4, Suppl.material 1).All bones of the skull paired except the parasphenoid and sphenethmoid.Vomer divided into pre-and postchoanal portions; prechoanal part small, longer than broad, subtri-angular; postchoanal part overlapping neopalatine, bearing ventral serrations (vomerine teeth), separated medially from its counterpart by a gap of 0.7 mm.Postchoanal vomer+neopalatine in dorsal contact with anterior end of cultriform process of parasphenoid, and through it with the sphenethmoid; laterally not in contact with maxilla.Teeth present on maxilla and premaxilla.Premaxilla medially not fused to counterpart, anterodorsal alary processes Creative Commons Attribution 4.0 licence (CC-BY); original download https://pensoft.net/journalsrising dorsolaterally, pars palatina with two well-defined processes, the medial (palatine) process thin, lateral process broad; pars dentalis bearing small teeth.Septomaxilla roughly spiralling upward from posterior ramus to lateral ramus to anterior ramus to medial ramus.Nasal medially in contact with the sphenethmoid posteriorly, possessing a pointed maxillary process extending ventrolaterally toward the maxilla, lacking an anterolateral ramus.Maxilla long, bearing small, poorly resolved teeth, possessing a horizontal pars palatina along its lingual margin; in broad lateral contact with anterior ramus of pterygoid; posteriorly without clear distinction from quadratojugal.Pterygoid broad and triradiate, with anterior, medial, and posterior rami; the ventrolateral edge of its anterior ramus, posterior margin of its medial ramus, and lateral face of its posterior ramus sculpted inward; medial ramus much shorter than posterior ramus; posterior ramus not in contact with quadratojugal.Quadratojugal L-shaped, anterior process without clear distinction from posterior of maxilla; posteriorly possessing a ventral bulbous process with a concave posterior face; dorsally without clear distinction from squamosal.Squamosal thin and distally bifurcated, extending anterodorsomedially from quadratojugal to level of otic capsule passing anterior to columella; otic ramus longer than zygomatic ramus.Columella with a long shaft that exceeds the level of the squamosal; dorsal edge of columella straight even to end of footplate; columellar footplate broad and concave.Frontoparietal medial and lateral edges straight and parallel, lateral edge curved ventrally to form dorsolateral border of brain case; possessing paired bumps at the transverse level of the columellae; posterolateral sutures with prootics and exoccipitals not clear from micro-CT scans; anterior process contacting sphenethmoid.Parasphenoid T-shaped; cultriform process broadening anteriorly, contacting sphenethmoid at its anterior end; broad posterior alary processes perpendicular to cultriform process, in dorsal contact with prootics anteriorly and exoccipitals posteriorly.
Mandible slim, edentate.Mentomeckelian small, in narrow medial contact with counterpart (possibly artefactual), and in dorsolateral contact with dentary.Dentary long and thin with a sculpted outer face and smooth inner face, overlapping angulosplenial for much of its length.Angulosplenial broadening posteriorly, with a posterior dorsomedial crista; possessing a lateral channel running from the posterior into the sculpted outer edge of the dentary.
Posterolateral processes of hyoid shovel-like, a medial crista running along posteromedial process, the base of which is broad and flat with a rounded anteromedial edge and sharp anterolateral and posteromedial corners; parahyoid absent.
Pectoral girdle composed of paired coracoids, clavicles, scapulae, cleithra and suprascapulae.Sternal charac-ters not visible in CT render.Coracoids in medial contact; medially dorsoventrally flattened, laterally rounded, posterior surface straight, anterior surface strongly concave.Clavicle thin and curved approximately parallel to the anterior edge of the coracoid, its lateral end broadened, posteriorly in contact with ventral edge of scapular pars acromialis.Scapula thick, hourglass shaped, its posterior edge less strongly curved than its anterior edge, medioventrally bifurcated; pars acromialis distally rounded, in contact with the lateral end of the clavicle, its anterior surface concave; pars glenoidalis curved ventrally, in contact with lateral face of coracoid, posterior face concave; dorsal edge of scapula approaching cleithrum.Cleithrum thin and long, not possessing any cristae, anteriorly thicker than posteriorly.Suprascapula with highest X-ray absorption ventrally and posteriorly suggesting possible ossification in these areas.
Eight presacrals present; no vertebrae fused.Posterior articular processes round.Transverse processes of presacrals II-IV broader than those of V-VIII.Neural spines decreasing in size from presacral II to absent by V. Sacrum wide, with broad diapophyses articulating with the ilia; anterior edge of each diapophysis roughly perpendicular to body axis, posterior edge oblique.Urostyle long and thin, with a dorsal ridge along a third of its length, beginning at its anterior end; articulation with sacrum bicondylar.

Variation.
Only two specimens are known.The paratype is male, and smaller than the holotype (SVL 23.8 mm).It agrees in all aspects of its morphology with the holotype, but differs strongly in colouration (see Fig. 3b, d).In life, the dorsum has a yellow-brown base colour, with distributed black or dark brown flecks.A black inter-ocular bar is present, behind which the skin fades from brown to the base colour; the back does not possess the dark marking of the holotype, but instead two darker areas with a few black Creative Commons Attribution 4.0 licence (CC-BY); original download https://pensoft.net/journalsflecks lie in the suprascapular region.The lateral skin fades to grey ventrally, also speckled with black.A dark line runs from the preocular region to the axial pit through the supraocular and supratympanic regions.The nostril is surrounded by black, and the tympanum has a dark fleck on it.The legs are grey at the hip, but this lightens to the yellowish brown of the dorsum further away; the arms are dorsally yellow, the hands possessing a few black spots.Ventrally, the pink areas of the holotype are orange in life in the paratype, particularly over the pectoral girdle and beneath the chin.The venter is marked with many irregular black flecks.The arms are ventrally orange, bordered posteriorly in black.
Etymology.The species epithet is an invariable noun in apposition to the genus name, derived from the Latin words longus (meaning long), and crus (meaning leg), and refers to the unusually long legs of this species.
Distribution.This species has only been found at high altitude in the montane forests of the Sorata massif in north Madagascar.Its distinctiveness leads us to hypothesize that it has never been found elsewhere and misidentified, so it may be microendemic to this small area.Additional surveys are required in areas in and around Sorata to identify its full distribution.
Ecology.Both specimens described here were captured in the early evening on the ground along a path through primary montane forest.The stomachs of both specimens contained remains of several small insects (mostly Coleoptera) and a spider (possibly belonging to the family Salticidae), mixed with moss, suggesting an opportunistic diet of arthropods.Calls of this species are unknown.The female holotype had more than twenty immature oocytes with the largest having diameters ranging from 1.3 to 1.6 mm.As a member of the Cophylinae, it is likely that R. longicrus lays its eggs away from running water or large water bodies, and has endotrophic tadpoles.
Conservation status.The forests of Sorata are currently unprotected.All locally endemic species are threatened by uninhibited deforestation and forest degradation.The greatest pressure on forests is at their edges.High altitude species like R. longicrus may therefore be the least threatened by this.However, a sustained rate of deforestation will increase the threat level to species at ever-higher altitudes.It is conceivable that a restriction of this species to high altitudes may mean that it is susceptible to climate change (Raxworthy 2008, Raxworthy et al. 2008).We consider this threat far less serious than that of deforestation.Batrachochytrium dendrobatidis has now been confirmed from numerous localities in Madagascar (Bletz et al. 2015).So far no negative impacts on native frogs have been observed.The water-independent lifestyle of Rhombophryne species suggests that they are probably at relatively low threat from chytridiomycosis.
While this species is, at present, known from just two specimens collected on one expedition, the fact that it has not been collected by previous expeditions suggests it may be scarce, seasonal, or have a scattered distribution.Even if it were distributed throughout the forests of the Sorata massif, its distribution would still only constitute an area of ~250 km 2 (as calculated in Google Earth® Pro 6.1.0.500,Google Inc., Mountain View, CA).Thus, because of its potentially limited range inside an unprotected forest, the on-going and intensifying threat of deforestation, potential threat by climate change, and potential scarcity or seasonality, it qualifies as Endangered B1ab(iii) under the IUCN Red List Criteria (2012).

Discussion
Hindlimb length is significantly associated with habitat and mode of life in frogs (Gomes et al. 2009).Longer legs relative to body length result in greater relative leaping performance (Zug 1978, Choi et al. 2003, Gomes et al. 2009).In general, fossorial frogs have the poorest jumping performance, while arboreal and semi-aquatic frogs are the strongest jumpers (Zug 1978).Terrestrial frogs are intermediate, but generally poor leapers, preferring to hop rather than leap.Emerson (1976) noted that adaptations involved in hopping in terrestrial frogs are similar to those required for burrowing.This can lead to difficulties disentangling the morphology involved in these two habits.We may also expect this to lead to frequent evolutionary transitions in preference between hopping and digging.
Robust and at least partly burrowing frogs typify the genus Rhombophryne.Some species are specialised burrowers (R. matavy and R. testudo), while others are probably facultative burrowers (R. serratopalpebrosa group, and probably R. alluaudi, R. laevipes, and R. mangabensis).Two species however seem to have at least partly abandoned burrowing: R. minuta and R. longicrus.Rhombophryne minuta lives at high altitude (close to and above the tree line) on the Marojejy massif and calls from low bushes that make up the complex matrix of ground in its unusual habitat (Glaw and Vences 2007).Rhombophryne longicrus lives at high altitude on the Sorata Massif to the north of Marojejy; its ecology is more or less unknown.These species form a strongly supported sister group in our phylogeny (Fig. 2), and resemble each other in morphology.Most notably, both have exceptionally long, slender legs relative to those of their congeners: HILs of both species are between 178.5% and 183.8% of their SVLs, considerably above the genus mean of 158.0%; and relative to their lengths, the thighs of R. longicrus are thinner than any other Rhombophryne species (THIW 29.1-29.2% of THIL) except R. minuta (whose THIW/THIL ranges from 24.4-38.0%;Scherz et al. unpublished data).We therefore expect these frogs to be capable of leaps of relatively greater distance than their congeners.
In addition to leg length, several other characters are also associated with more saltatorial locomotion.Emer-Creative Commons Attribution 4.0 licence (CC-BY); original download https://pensoft.net/journalsson (1979) characterised three types of iliosacral articulation correlated with locomotion patterns: Type I, expanded sacral diapophyses without ligament attachment, allowing great anteroposterior freedom of movement, most common in aquatic frogs, but also found in burrowers and climbers (Reilly and Jorgensen 2011); Type IIA, broad sacral diapophyses and proximal attachment of a broad ligament, the most adaptable and widespread of the articulation types, typical of walking and hopping locomotion, common in burrowers; and Type IIB, distal attachment of a narrow ligament to thin and posteriorly pointed sacral diapophyses, typical of long-distance leapers (although frogs with this type of articulation are not necessarily better leapers; see Reilly and Jorgensen 2011).Reilly and Jorgensen (2011) expanded this classification into seven types, by incorporating also dorsal ridges of the iliac shafts, and the nature of the urostyle.The iliosacral articulation of R. longicrus is Type IIA sensu Emerson (1979), and the ilia possess no ridges, while the urostyle is bicondylar and bears a ridge on its anterior third (as in all other Rhombophryne spp.so far investigated).
The iliosacral articulation of Rhombophryne longicrus is almost the same as members of the R. serratopalpebrosa group (Scherz et al. 2014(Scherz et al. , 2015)), but is slightly modified such that its iliac shafts are closer together.This may produce faster launch speeds and thus greater leaping distances (Choi et al. 2003).This, coupled with its long, slim legs and lack of burrowing specialisations, such as enlarged internal metatarsal tubercles, suggests adaptation to saltatorial locomotion.In external morphology, and based on preliminary osteological data, its sister species, R. minuta, appears to share most of these characteristics.Therefore, we hypothesise that these two species constitute a divergent, ancestrally saltatorial lineage that diverged from possibly semi-fossorial ancestors.A thorough treatment of the osteology of Rhombophryne will shed light on this question.It is clear already, however, that this genus constitutes an osteologically and ecologically diverse group of frogs, rivalling the diversity seen in other cophylines.
The discovery of such a distinctive new species highlights the incompleteness and patchiness of herpetological survey work in Madagascar.Whilst some forests, particularly accessible, protected ones, are receiving a lot of research attention (e.g.Betampona: Andreone et al. 2010;Rosa et al. 2012Rosa et al. , 2014)), others, like the forests of the Sorata massif, are receiving little study.Sorata is part of a constellation of high-altitude massifs, linked to the massifs Tsaratanana to the west, and Marojejy to the east by narrow stretches of remaining forest.Further survey work will be needed to understand its diversity and role in this network of massifs.At the same time, however, its forests are unprotected and highly threatened by anthropogenic habitat destruction and modification.Protected status must therefore be pursued together with an enhanced knowledge of this area's flora and fauna.

Figure 1 .
Figure 1.Measurement scheme used to measure Rhombophryne longicrus sp.n. and congeners for this study.Abbreviations are explained in Materials and Methods, as are cumulative measures such as forelimb and hindlimb length.* indicates IMCL.

Figure 2 .
Figure 2. Majority-rule consensus tree derived from Bayesian inference analysis of the genus Rhombophryne based on the mitochondrial 16S rRNA gene.Numbers at nodes represent posterior probability (PP).PP values greater than 0.95 are bolded.Values lower than 0.8 are not shown.