Rhombophryne regalis Scherz, Hawlitschek, Andreone, Rakotoarison, Vences & Glaw, 2017, sp. nov.

Rhombophryne regalis sp. nov.

Suggested common name: Regal saw-browed diamond frog (Figs 4, 5, 9, 10, S3–S7)

Naming remarks. This species has frequently been referred to as Plethodontohyla (after 2008, Rhombophryne) serratopalpebrosa (Vences & Glaw 2003; Glaw & Vences 2007; Wollenberg et al. 2008; Vieites et al. 2009). It was figured as such on page 119 of Glaw & Vences (2007). We have referred to it as R. sp. aff. serratopalpebrosa (Scherz et al. 2015b), R. sp. ‘Ambolokopatrika’ (Scherz et al. 2014, 2015a, 2016a, Lambert et al. 2017), and R. sp. CaNEW Ambolokopatrika (Scherz et al. 2016b). We consider it likely that some records from COMATSA (e.g. Rabearivony et al. 2015) refer to this species, but specimens and DNA samples are required to verify this.

Holotype. MRSN A4602 (FN 7292), adult male, captured 13 December 1997 by F. Andreone, J.E. Randrianirina, and G. Aprea at Camp 3 of Ambolokopatrika-Betaolana Forest, locally known as ‘ Antsinjorano’ (14.5433°S, 49.4300°E), around 980 m a.s.l., in the Sava Region, northeastern Madagascar (Figs 4, 5, S3).

Paratypes. MRSN A4603 (FN 7146), adult female, same collection data as the holotype; MRSN A4619, subadult, and MRSN A4620, adult female, collected 6 December 1997 by F. Andreone, J.E. Randrianirina, and G. Aprea at Camp 2 of Ambolokopatrika-Betaolana Forest, locally known as ‘ Andranomadio’ (14.5400°S, 49.4383°E), around 860 m a. s.l.; MRSN A4618, adult female, collected 26 June 1996 by F. Andreone and J.E. Randrianirina in Besariaka Forest, at the campsite locally known as ‘ Ambinanin’antsahamaloto’ (14.8283°S, 49.5958°E), around 800 m a.s.l.; MRSN A6058, adult female, collected January 1996 by F. Andreone, J.E.

Randrianirina, and H. Randriamahazo on the west slope of Anjanaharibe-Sud (Analabe) at ‘Camp W1’ (14.7783°S, 49.4634°E), around 1050 m a.s.l.

Diagnosis and comparisons. A species assigned to the genus Rhombophryne on the basis of molecular phylogenetic affinities (Fig. 3), and the possession of a clavicle combined with the absence of T- or Y-shaped terminal phalanges (vs. either absence of a clavicle or possession of a clavicle combined with T- or Y-shaped terminal phalanges in the morphologically similar Plethodontohyla). Within the genus Rhombophryne, it is assigned to the R. serratopalpebrosa species group on the basis of possessing superciliary spines and molecular phylogenetic affinities (Fig. 3).

Rhombophryne regalis sp. nov. is distinguished from all congeners by the following unique suite of characters: SVL 20.2–26.5 mm; loreal region strongly concave, possessing an S-shaped fold behind the naris; tympanum distinct, TDH/ED = 0.47–0.67; weak supratympanic fold extending from rear corner of eye over tympanum toward axillary region; three superciliary spines, the first of which is largest and pointed anteriorly, the second of which is smaller, and the third of which is diminutive; second finger slightly shorter than fourth, fifth toe distinctly shorter than third; tibiotarsal articulation reaching to or beyond the snout tip; TIBL/SVL = 0.47–0.56. Osteologically, it is characterised by an anteriorly broadening parasphenoid cultriform process, ventromedial contact of exoccipitals, straight postchoanal vomers, smoothly sigmoidal anterior edge of ventral ramus of squamosal, strong dorsal prominence on iliac shafts, and ossified pubis. Additionally, R. regalis is distinguished from all other Rhombophryne species for which molecular data are available by uncorrected pairwise distances of at least 6.1% in a segment of the 16S rRNA mitochondrial gene (Table 1).

Within the genus Rhombophryne, R. regalis sp. nov. can be distinguished from all species except members of the R. serratopalpebrosa group by the possession of superciliary spines. Within the R. serratopalpebrosa group, R. regalis sp. nov. may be distinguished from R. serratopalpebrosa, which it most strongly resembles, by its smaller size (adult female SVL 20.2–26.5 mm vs. 28.5 mm; males unavailable from R. serratopalpebrosa), weak supratympanic fold (vs. strong supratympanic fold), possession of three superciliary spines: 1st large, 2nd medium, 3rd diminutive (vs. four: 1st large, 2nd medium, 3rd small, 4th diminutive), smaller relative tympanum size (TDH/ED = 0.47–0.67 vs. 0.72), shorter relative forelimb length (FORL/SVL = 0.59–0.70 vs. 0.80), parasphenoid cultriform process broadening anteriorly (vs. not broadening), exoccipitals in ventromedial contact (vs. not in contact), broad (vs. narrow) quadratojugal–squamosal contact, anterior edge of squamosal ventral ramus smoothly sigmoidal (vs. stepped), and ossified (vs. unossified) pubis; from R. vaventy by its much smaller size (SVL 20.2–26.5 vs. 51.9 mm), weak supratympanic fold extending from posterior of eye over tympanum toward axillary region (vs. distinct supratympanic fold, curved over and behind the tympanum but not extending anterior to the tympanum), possession of three superciliary spines: 1st large, 2nd medium, 3rd diminutive (vs. four, anterior two most distinct), shorter relative forelimb length (FORL/SVL = 0.59–0.70 vs. 0.76), and absence (vs. presence) of enlarged inner metatarsal tubercles; from R. guentherpetersi by its smaller size (SVL 20.2–26.5 vs. 27.3–35.7 mm), possession of three superciliary spines: 1st large, 2nd medium, 3rd diminutive (vs. 2–3 small superciliary spines), tibiotarsal articulation reaching at least the nostril (vs. reaching the insertion of the arms), longer relative tibia length (TIBL/ SVL = 0.47–0.56 vs. 0.35–39), and absence of a dorsal tibial gland (vs. presence); from R. ornata by its smaller size (single known male SVL 22.4 vs. 33.0 mm), weak supratympanic fold extending from rear corner of eye over tympanum toward axillary region (vs. distinct supratympanic fold extending from rear corner of eye over and behind tympanum toward axilla; see Fig. 5), possession of three superciliary spines: 1st large, 2nd medium, 3rd diminutive (vs. two of roughly equal size), thinner thighs (THIW/THIL = 0.26–0.34 vs. 0.36–0.40), slightly longer relative tibia length (TIBL/SVL = 0.47–0.56 vs. 0.38–0.46), tibiotarsal articulation reaching at least the nostril (vs. reaching between the tympanum and eye), and absence of reddish colour on the hidden portions of the legs (vs. presence); from R. tany by its weaker supratympanic fold (see Fig. 5), possession of three superciliary spines: 1st large, 2nd medium, 3rd diminutive (vs. two of roughly equal size), considerably thinner thighs (THIW/THIL 0.26– 0.34 vs. 0.41), longer relative tibia length (TIBL/SVL = 0.47–0.56 vs. 0.43), and tibiotarsal articulation reaching at least the nostril (vs. reaching between the tympanum and eye); and from R. coronata by possession of three superciliary spines differing in size: 1st large, 2nd medium, 3rd diminutive (vs. three of roughly equal size), thinner thighs (THIW/THIL = 0.26–0.34 vs. 0.36–0.44), thinner shanks (TIBW/TIBL = 0.18–0.24 vs. 0.31–0.36), longer relative tibia length (TIBL/SVL = 0.47–0.56 vs. 0.35–0.39), and first toe unreduced (vs. sometimes reduced to a short nub; see Fig. 5).

Description of the holotype. (Figs 4, 5, 9 a, b) An adult male specimen in a moderately good state of preservation, though quite soft. Tissue samples taken from left thigh and tongue for sequencing. A small incision is present in the right side. The testes are well developed and mature, and can be easily identified.

Body robust. Head wider than long (HW/HL = 1.46). Pupils tiny and horizontally oblong (possibly deformed by preservation). Snout rounded in dorsal and lateral view. Canthus rostralis concave. Loreal region concave, with a subtle S-shaped fold that surmounts the nostril anteriorly (otherwise known only from R. serratopalpebrosa). Nostrils nearer to snout tip than to eye (END/NSD = 1.22), directed laterally, slightly protuberant. Tympanum distinct, TDH/ED = 0.47. Weak supratympanic fold, hardly noticeable anterior to tympanum but extending posterior to it toward the axillary region. Three superciliary spines above each eye, the first large and anterior to the eye, the second smaller and above the eye, the third diminutive and over the posterior half of the eye (hardly perceptible without the aid of magnification). Vomerine teeth present in a straight row either side of the palate, approaching each other medially but separated by a small gap.

Arms slender. Fingers without webbing, relative lengths 1<2<4<3; fourth finger distinctly longer than second; finger tips not expanded; fingers not reduced (Fig. 5); nuptial pads absent; inner metacarpal tubercle present, outer metacarpal tubercle absent; subarticular tubercles weak, undivided. Hindlimbs fairly slender; tibiotarsal articulation reaches the snout tip; TIBL/SVL = 0.48. Inner metatarsal tubercle present, not enlarged, outer metatarsal tubercle absent, subarticular tubercles indistinct. Toes unwebbed; relative lengths 1<2<5<3<4, fifth toe distinctly shorter than third. Toe tips not expanded, third toe tip pointed.

Dorsal and ventral skin smooth in preservative, but quite bumpy in life (Fig. 9 a). Dorsolateral folds absent.

Colouration of holotype: After 19 years in preservative, specimen dorsally brown, with two darker spots above the suprascapulae, anteriorly bordered by cream lines. A darker brown chevron extends from the inguinal regions on either side of the body to the mid-dorsum, bordered anteriorly and posteriorly by thin cream lines. Another cream line extends from the middle lateral side of the frog anteriorly over the axial pit, dorsally encircling the insertion of the arm. Over the suprascapular region, a clover-shaped dark brown marking is present, outlined with a thin cream border (much more distinct in life; see Fig. 9 a). A dark brown interocular bar is also present, likewise bordered with cream. Laterally lighter brown, with a cream line extending from posterior of eye along anterior edge of tympanum to corner of mouth. Legs with one darker brown crossband each on the thigh, lower leg, tarsal, and metatarsal regions, that on the thigh with thin cream borders. Venter cream, with a darker chin. Colouration in life as in preservative, but more vibrant and clear, and somewhat more reddish (Fig. 9 a).

Variation. All paratypes have been more strongly fixed than the holotype. Snout more or less truncate and squared. For variation in measurements, see Appendix 1. TDH larger in all paratypes than in the holotype (full range of TDH/ED = 0.47–0.67). Leg length varies, but the tibiotarsal articulation extends to or beyond the nostril. Supratympanic fold and superciliary spines generally consistent, but influenced by quality of preservation. The tongues of the paratypes are broad, unlobed, and attached anteriorly (that of the holotype has been removed). No sexual size dimorphism is apparent. The paratypes are more uniformly brown dorsally than the holotype, mostly lacking dorsal patterns and cream colouration; ventrally similar, but can be more mottled with light brown. The dorsal skin of all paratypes is rougher than the holotype, both in preservative and in life (Fig. 9). Notes on osteological variability are given in the Osteology section, below.

Natural history. Several of the paratypes contain developing eggs, revealed by their slightly higher X-ray absorbance than the surrounding tissue in the micro-CT scans; MRSN A4620 had 11, and A6058 had 17. This suggests relatively small clutch sizes. Little is known about the species’ life history. The individuals were usually found active on the ground during rainy weather at night. The type series was collected while installing basecamp and drift fences in primary, mid-altitude rainforest, confirming the species as at least partly fossorial.

Etymology. The species epithet is the Latin nominative singular two-ending adjective regalis, meaning ‘kingly’ or ‘regal’, and refers to the crown-like superciliary spines that individuals of this species, and all other members of the R. serratopalpebrosa group, possess.

Distribution and conservation status. Rhombophryne regalis sp. nov. is reliably known from Anjanaharibe- Sud, Ambolokopatrika, and Besariaka, over an altitudinal range from 980 to 1050 m a.s.l. (Fig. 8). It may also be distributed south into Makira, northwest into COMATSA, southeast into Masoala, or northeast into Marojejy—a more detailed sampling including genetic material will be needed to assess its full distribution. Its currently known distribution encompasses an area of 213 km 2 over three threat-defined locations (TDLs). A TDL is defined by the IUCN (2012) as a ‘distinct area in which a single threatening event can rapidly affect all individuals of the taxon present’; here, we consider each of the three localities from which the new species is known to be separate TDLs, as deforestation, the greatest threat to these frogs, is generally localized, and disappearance of habitat in one of the locations will not affect the status of the others.

Records of ‘ R. serratopalpebrosa ’ from COMATSA by Rabearivony et al. (2015) probably refer to this species as well, but need to be verified with specimens. However, we hypothesise that it may in fact be distributed over an area of> 7000 km 2 encompassing adjacent, mid-altitude rainforest in northern Madagascar. The habitat in all three of the confirmed locations where this species occurs is being degraded, despite partial protection—cattle grazing, slash-and-burn, and selective logging are widely practiced, leading to on-going decline in the extent and quality of available habitat. Due to its restricted known range but probable wider distribution, this species is a borderline case between Endangered and Vulnerable using the IUCN classification system. To avoid being inflationary, we conservatively propose a status of Vulnerable for this species according to the IUCN Red List criterion B1ab(iii) (as above, but extent of occurrence <20,000 km 2 and known from 10 or fewer locations) and recommend research to quantify its distribution in northern Madagascar’s rainforests.