Article. - PDF Free Download

Zootaxa 3937 (1): 061 089 www.mapress.com/zootaxa/ Copyright 2015 Magnolia Press Article http://dx.doi.org/10.11646/zootaxa.3937.1.3 http://zoobank.org/urn:lsid:zoobank.org:pub:5b7b0508-0a35-4566-b979-90febf12e2a1 ISSN 1175-5326 (print edition) ZOOTAXA ISSN 1175-5334 (online edition) Revision and phylogeny of narrow-mouthed treefrogs (Cophyla) from northern Madagascar: integration of molecular, osteological, and bioacoustic data reveals three new species ANDOLALAO RAKOTOARISON 1,2, ANGELICA CROTTINI 3, JOHANNES MÜLLER 4, MARK-OLIVER RÖDEL 4, FRANK GLAW 5 & MIGUEL VENCES 2,6 1 Département de Biologie Animale, Université d Antananarivo, BP 906, Antananarivo 101, Madagascar 2 Zoological Institute, Technische Universität Braunschweig, Mendelssohnstr. 4, 38106 Braunschweig, Germany 3 CIBIO, Research Centre in Biodiversity and Genetic Resources, InBIO, Universidade do Porto, Campus Agrário de Vairão, Rua Padre Armando Quintas, Nº 7, 4485-661 Vairão, Vila do Conde, Portugal 4 Museum für Naturkunde, Leibniz Institute for Evolution and Biodiversity Science, Invalidenstr. 43, 10115 Berlin, Germany 5 Zoologische Staatssammlung München, Münchhausenstr. 21, 81247 München, Germany 6 Corresponding author. E-mail: m.vences@tu-bs.de Abstract We provide a revision of microhylid treefrogs of the genus Cophyla, the type genus of the subfamily Cophylinae. A phylogeny inferred from DNA sequences of multiple mitochondrial and nuclear genes, with representatives of all cophyline genera except Madecassophryne and including representatives of the two most divergent intrageneric lineages within Cophyla, placed Cophyla as sister group of Platypelis and confirmed both genera as reciprocally monophyletic. We describe three new Cophyla species based on osteological, morphological and bioacoustic characters as well as genetic differentiation in one nuclear and several mitochondrial markers. As in the vast majority of cophylines, all species of Cophyla emit long, stereotyped repetitions of a single tonal note, and we here consider one of these notes as a call; call duration thus equals note duration and the intervals between calls are named inter-call intervals. Cophyla maharipeo sp. nov. collected in Joffreville and Forêt d Ambre Special Reserve (adult SVL 22 27 mm) is characterized by having long calls (1166 1346 ms) with long inter-call intervals (2154 3881 ms). Cophyla noromalalae sp. nov. collected in Montagne d Ambre National Park (adult SVL 22 29 mm) is characterized by having short calls (662 821 ms) and short inter-call intervals (874 1882 ms). Cophyla puellarum sp. nov., also from Montagne d Ambre National Park, is larger than the other two species (adult SVL 27.3 33.6 mm) and characterized by the shortest calls (326 390 ms) and long inter-call intervals (1961 3996 ms). Osteological analyses based on micro-ct scans and cleared and stained specimens confirms that the shape of the posterior vomer (centrally divided vs. undivided) may be a useful character to diagnose most species as belonging to either Platypelis and Cophyla, and suggest the absence of clavicles (present in Platypelis) is a derived character of most Cophyla. However, clavicles were present in C. puellarum, the only known Cophyla occurring at relatively high elevations (1250 1300 m a.s.l.) while otherwise in northern Madagascar, forests at higher elevations up to 2700 m a.s.l. are occupied by Platypelis species. Cophyla maharipeo was found at relatively low elevations (630 720 m a.s.l.), similar to the three previously known congeners (C. berara, C. occultans, C. phyllodactyla). Cophyla noromalalae occurs at intermediate elevations (900 1050 m a.s.l.). The molecular phylogeny inferred herein suggests that the ancestor of a clade containing all Cophyla species except C. puellarum evolved a modified shoulder girdle structure without ossified clavicles, and adapted to low-elevation habitats. Key words: Amphibia, Anura, Microhylidae, Platypelis, conservation status, integrative taxonomy, Montagne d'ambre, micro-ct scanning Introduction Compared to other tropical regions, the amphibians of Madagascar are relatively well studied. Due to the intensive studies of the last 20 years significant gaps in species inventory have become obvious (Vieites et al. 2009). These Accepted by J. Rowley: 18 Feb. 2015; published: 24 Mar. 2015 61

gaps may be due to multiple causes including, for seasonal breeders, reduced detectability outside of the reproductive period, reduced detectability due to minute body size (Vences et al. 2010), or genuinely highly restricted distributions. Species with one or more of these characteristics often remain overlooked until intense herpetological surveys take place in the right season (e.g. Rosa et al. 2012). This applies especially to many species of the family Microhylidae, a group represented in Madagascar by three subfamilies: the Dyscophinae Boulenger, the Scaphiophryninae Laurent and the Cophylinae Cope, all three endemic to the island (van der Meijden et al. 2007). Among Madagascar's microhylids, the cophylines are the most species rich subfamily, and exhibit the highest diversity of ecological and morphological adaptations (Andreone et al. 2005b). All species for which reproductive behavior has been documented are nidicolous. Their non-feeding, endotrophic larvae (depending on their yolk reserves only) develop in water-filled tree holes or similar cavities, or in terrestrial jelly or foam nests (Blommers- Schlösser 1975; Blommers-Schlösser & Blanc 1991; Glaw & Vences 2007). The type genus of cophylines is the genus Cophyla, originally described by Boettger (1880) to accommodate an arboreal species of microhylid from the northwestern Malagasy offshore islet Nosy Bé. Later, Vences et al. (2005) described a second species, C. berara, and Wollenberg et al. (2008) transferred a third species (Platypelis occultans Glaw & Vences) to the genus. The genera Cophyla and Platypelis Boulenger both contain arboreal frogs that, as far as is known, cannot be reliably assigned to either genus based on external morphology alone (Glaw & Vences 2007; Rakotoarison et al. 2012). Generic attribution and species identification might be possible based on skeletal features (Blommers-Schlösser & Blanc 1991), but only a small set of species has so far been examined osteologically (Parker 1934; Blommers-Schlösser & Blanc 1991). Andreone et al. (2005b) and Wollenberg et al. (2008) established the reciprocal monophyly of the two genera by phylogenetic analysis of mitochondrial DNA sequences. These two studies also recovered the sister group relationship between Cophyla and Platypelis, a result, however, not confirmed by a subsequent supermatrix analysis (Pyron & Wiens 2011). The genus Cophyla has been poorly studied to date. In his monograph of microhylid frogs, Parker (1934) provided osteological data of the genus and noted the absence of a clavicle and the unusual condition of the vomer region. He also included illustrations of the shoulder girdle and the vomerine teeth. Guibé (1978) provided additional osteological data for a number of arboreal microhylids from Madagascar, mostly species of the genus Platypelis, and still recognized the monotypic genus Paracophyla (today considered a synonym of Platypelis barbouri), which was diagnosed in his key from Cophyla solely by the absence of vomerine teeth. Later Blommers-Schlösser & Blanc (1991) summarized the osteology of Malagasy microhylids, emphasizing in their osteological account the missing clavicle, and using in their key the character state of the posterior vomer (called prevomer by these authors), represented by a single unpaired element, as a character distinguishing Cophyla from Platypelis. The genus Cophyla currently contains three species (Fig. 1): C. phyllodactyla Boettger, 1880; C. occultans (Glaw & Vences, 1992); and C. berara Vences, Andreone & Glaw, 2005, but additional candidate species have already been identified on the basis of genetic differentiation (Vieites et al. 2009; Perl et al. 2014). The three nominal species of Cophyla (as well as the three new species described herein) are characterized by stereotyped advertisement calls, which are typical for the vast majority of cophylines (Glaw & Vences 1994, 2007; Vences et al. 2006), and are emitted at night from perches, mostly leaves, in the vegetation. As argued in Materials & Methods below, these vocalizations consist of a tonal single-note call repeated after regular inter-call intervals. The duration of calls and inter-call intervals are diagnostic for the three nominal species of Cophyla (e.g., Vences et al. 2005). In this paper we undertake a revision of the genus and describe three new species of Cophyla collected at three different elevations in the Montagne d Ambre massif. We base our study on the integration of data from external morphology, CT-scans of skeletons, natural history, DNA barcoding, congruence between mitochondrial and nuclear DNA differentiation and bioacoustic analyses. We first define candidate species based on genetic distances, then discuss the genetic, morphological and bioacoustic evidence for their distinct taxonomic status. Finally, the three species are formally described and illustrated. Material and methods Study sites, fieldwork and external morphology. The three species described herein all originate from rainforests of the Montagne d'ambre massif in northern Madagascar, comprising Montagne d'ambre National Park (MANP), 62 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

Forêt d'ambre Special Reserve, and environments of the village Joffreville. Collections from these sites were assembled during various field expeditions from 2000 to 2013, but most specimens and data were collected in January 2012 and December 2012. Most collections were made at five sites at different elevations (Latitude, Longitude, elevation in m a.s.l.): (1) Joffreville village, -12.49397, 49.20504, 634 m; (2) forest and forest edges around the entrance of MANP, - 12.616625, 49.150129, 900 m; (3) Lake Maudit within MANP, -12.5856, 49.1457, 1306 m; (4) a site locally called "Point de Vue du Grand Moulin within MANP, -12.54077, 49.15424, 1246 m; (5) Fôret d'ambre Special Reserve, approximately -12.47, 49.22, 550 m. Specimens were collected either during the day by searching in tree holes and bamboo holes, or at night using torches and headlamps and guided by the male advertisement call. Representative voucher specimens were euthanized in a chlorobutanol solution, fixed in 90% ethanol and preserved in 70% ethanol. Vouchers were deposited in the Université d Antananarivo, Département de Biologie Animale, Madagascar (UADBA), Zoologisches Forschungsmuseum Alexander Koenig, Bonn (ZFMK), and the Zoologische Staatssammlung München (ZSM). ACZC, FGZC, DRV, NSH, RDR, and ZCMV refer to A. Crottini, F. Glaw, D.R. Vieites, S. Ndriantsoa, R.D. Randrianiaina, and M. Vences field numbers, respectively. Tissue samples were taken by cutting pieces of muscle of the leg, and preserved separately in 99% ethanol. Males and females were distinguished based on the size of the prepollical tubercle which is bigger in the males than in the females, and presence of a vocal sac in males. The following morphological measurements on preserved specimens were taken by the first author with digital callipers to the nearest 0.1 mm: snout-vent length (SVL), maximum head width (HW), HL (head length), ED (horizontal eye diameter), END (eye-nostril distance), NSD (nostril-snout tip distance), NND (nostril-nostril distance), TD (horizontal tympanum diameter), HAL (hand length), FL (foot length), FOTL (foot length including tarsus), FORL (forelimb length), tibia length (TIBL), hindlimb length (HIL), and length and width of prepollical tubercle (PREPL, PREPW), and RHL (relative hindlimb length). Terminology and description scheme follow Vences et al. (2010) and Glaw et al. (2012a). Micro-computed tomographic analysis, and clearing and staining. Selected specimens were scanned at the Museum für Naturkunde Berlin using a Phoenix x-ray nanotom (GE Sensing & Inspection Technologies GmbH, Wunstorf, Germany) equipped with a 180 kv high-power nanofocus tube and a tungsten target. Reconstructions were performed in datos x-reconstruction software (GE Sensing & Inspection Technologies GmbH phoenix x-ray) and data were visualised in VGStudio Max 2.0 (Volume Graphics GmbH, Heidelberg, Germany). In addition, a few specimens were cleared and alizarin/alcian blue stained according to Dingerkus & Uhler (1977) to verify differences observed in the micro-ct scans. For the micro-ct scans we used the following specimens: C. maharipeo sp. nov., ZSM 3252/2012 (ZCMV 12194); C. noromalalae sp. nov., ZSM 3273/2012 (FGZC 4902); C. puellarum sp. nov., UADBA-A 60237 (FGZC 4903); Platypelis cf. barbouri, NSH 2587; P. pollicaris, NSH 2419, and for clearing and staining, we used C. noromalalae sp. nov., ZSM 3273/2012 (FGZC 4902); P. pollicaris, ZSM 1732/2012 (RDR 1066); and P. tsaratananaensis, ZSM 1643/2010 (DRV 06129). Bioacoustics. Advertisement calls were recorded in the field using a Tascam DR07 digital recorder with builtin microphone at a sampling rate of 44.1 khz and 24-bit resolution and saved as uncompressed files. Recordings were resampled at 22.05 khz and 16-bit resolution and computer-analysed using the software Adobe Audition (CoolEdit Pro) version 2000. Frequency information was obtained through Fast Fourier Transformation (FFT, width 1024 points); the audiospectrogram was obtained at Hanning window function with 256 bands resolution. Temporal measurements are given in milliseconds (ms) or seconds (s), as range, with mean ± standard deviation in parentheses. In most cases (except where otherwise stated) our call descriptions refer to the entire available recording (or a large part of it) from a single individual (the call voucher). Given the difficulty and time investment necessary for locating calling males of these small frogs, we preferred obtaining recordings of good quality of an unambiguously identified calling male (i.e., vocal sac seen moving during call emission) than recording multiple individuals. However, for all species, additional individuals were heard in the field and did not differ in the relevant call characteristics (given that call repetition rate is distinct enough among species to be discerned by the human ear). Air temperature was recorded using different digital thermometers (i.e., temperature information refers to air temperature at the day and time of recording, not body temperature of the calling specimens). REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 63

The advertisement calls of most cophyline species consist of potentially endless series of regularly repeated tonal notes. These general characteristics also apply to the calls of all species of Cophyla. Typically their vocalizations go on for indeterminate periods of at least many minutes, but usually they end only when the calling specimen is disturbed. As argued by Glaw & Vences (1994) it is not biologically meaningful to analyze and compare the number of units in these series, and the vast majority of recordings do not contain entire series. This distinctly differs from other Malagasy anurans such as Gephyromantis eiselti, which emit discrete series with a clearly defined number of whistles. We here follow Glaw & Vences (1994) by defining the tonal units in cophyline calls as single-note calls. Therefore, call duration equals note duration, and inter-note interval equals inter-call interval. In this paper, we use consistently the terms call duration and inter-call interval duration to refer to these variables. Molecular data sets and analyses. Total genomic DNA was extracted from tissue samples using Proteinase K (10mg/ml) digestion followed by a standard salt-extraction protocol (Bruford et al. 1992). In order to provide information on phylogeny and the genetic distinctness of species, we assembled three different molecular datasets: (1) We sequenced segments of the mitochondrial 16S rrna (16S), 12S rrna (12S), cytochrome b (COB), and cytochrome oxidase subunit I (COX1) genes, and the nuclear recombination-activating gene I (RAG1) gene from a large series of samples comprising 5 outgroup samples (Stumpffia and Platypelis) and 46 ingroup samples (Cophyla); however, not all samples were sequenced for all genes. (2) From the first data set we separately analyzed the RAG1 sequences of the ingroup samples (31 sequences) to assess whether mtdna differentiation among Cophyla is paralleled by nuclear DNA variation and whether species-level lineages have exclusive haplotypes in this nuclear gene. Only a small number of heterozygotes were observed (represented by IUPAC ambiguity codes) and we therefore did not undertake a separation of haplotypes by phasing prior to phylogenetic analysis. (3) Because the first data set did not support the reciprocal monophyly of Platypelis vs. Cophyla, we assembled a more comprehensive data set of four nuclear and four mitochondrial genes for a reduced set of taxa. These were selected to represent all genera in the Cophylinae (except Madecassophryne for which no tissue sample was available), and within Cophyla and Platypelis species were selected that belong to the main intrageneric clades as found in our first analysis, and in Andreone et al. (2005b) and Wollenberg et al. (2008). The gene segments used were the same as in analysis 1 (i.e., 12S, 16S, COB, COX1, RAG1) with longer sequences of RAG1, plus segments of three further nuclear genes, leucine-rich repeat and WD repeat-containing protein (KIAA1239), sacsin (SACS), and titin (TTN). Primers targeting the selected markers were employed in standard polymerase chain reactions (PCR) for amplification. Polymerase chain reactions were performed in a final volume of 10 µl using 0.3 µm of each primer, 0.25 mm of dntps, 0.4 U GoTaq and 1.25 x reaction buffer (Promega). Primer characteristics, sources, and specific thermal cycling schemes are given in Table 1. PCR products were purified using Exonuclease I and Shrimp Alkaline Phosphatase (SAP) or Antarctic Phosphatase (AP) according to the manufacturer s instructions (NEB). Purified PCR templates were directly sequenced using dye-labeled dideoxy terminator chemistry on an ABI 3130 automated DNA sequencer (Applied Biosystems). Chromatograms were checked and sequences manually corrected, if necessary, in CodonCode Aligner 3.5.6 (CodonCode Corporation). Newly obtained sequences were submitted to GenBank (accession numbers: KP764200-KP764393). Alignments in interleaved Nexus format are available from the authors. Using the software MEGA 5 (Tamura et al. 2011), protein-coding sequences were aligned by hand and translated into amino acids for authentication. Non-coding sequences were aligned with the MUSCLE algorithm under default settings implemented in MEGA. Indels in 12S and 16S sequences were included as missing data in the analysis. We used PartitionFinder 1.0.1 software (Lanfear et al. 2012) to infer the best-fitting models of molecular evolution and partition scheme applying to the three data sets. The best-fitting partition/substitution model scheme, as determined by the AICc, was implemented in a Bayesian inference (BI) analysis with MrBayes 3.2 (Ronquist et al. 2012). Results of two independent runs of 10 million generations, each comprising four Markov Chains (three heated and one cold), were sampled every 1,000 generations. Chain mixing and stationary was assessed by examining the standard deviation of split frequencies and by plotting the -lnl per generation using Tracer 1.5 software (Rambaut & Drummond 2007). 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 burnin (all compatible nodes with probabilities <0.5 kept). 64 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

Dataset 3 was subjected to a bootstrap analysis under the maximum parsimony (MP) optimality criterion, to test whether reciprocal monophyly of Platypelis vs. Cophyla remains stable also under this analysis method. We conducted MP bootstrap analyses with 2,000 pseudoreplicates in PAUP* v4.10 (Swofford 2002), using a treebisection-reconnection (TBR) branch-swapping algorithm and 10 random addition sequence replicates. This additional analysis method was included because relationships of Platypelis vs. Cophyla have been contentious in previous studies (e.g., they were not recovered as sister groups in the study of Pyron & Wiens 2011), and the goal of dataset 3 was specifically to test whether monophyly of both genera would hold also under non-likelihood based analysis methods such as they are used by some amphibian systematists (e.g., Frost et al. 2006). Results Molecular diversity and phylogenetic relationships within Cophyla DNA sequences of all mitochondrial genes provided evidence for the presence of seven main clusters of individuals within Cophyla, all separated by substantial genetic divergences of >5% in the 16S gene. These clusters corresponded to the three described species (Cophyla berara, C. phyllodactyla, C. occultans), three candidate species from the Montagne d'ambre region (named Cophyla sp. Ca1, Ca2 and Ca3 in Perl et al. 2014) and one divergent individual from Marojejy (previously considered as C. sp. 2 in Andreone et al. 2005b; here considered as unconfirmed candidate species Cophyla sp. Ca4). We will herein describe C. sp. Ca1, Ca2, Ca3, respectively, as C. maharipeo sp. nov., C. noromalalae sp. nov., and C. puellarum sp. nov., and for convenience we will use these names (without the suffix sp. nov. in most of the sections) throughout the manuscript, anticipating their formal description below. Phylogenetic analysis (Fig. 2) of the concatenated mitochondrial data plus RAG1 confirmed reciprocal monophyly of these seven clusters, and resolved the relationships among them with maximum posterior probabilities. Cophyla phyllodactyla was sister taxon to a clade containing C. noromalalae and C. sp. Ca4, and these three taxa formed the sister group of C. maharipeo. A further monophyletic group contained C. berara and C. occultans, and this group was sister to a clade containing all previously mentioned species. The basal-most node in a monophyletic Cophyla separated C. puellarum from its sister clade which contained all other Cophyla species. However, the genus Platypelis was recovered as paraphyletic (Fig. 2). An analysis based on the nuclear RAG1 gene alone (Fig. 3) did not reliably resolve relationships among most species, but it confirmed the species-level clustering by recovering all included species and candidate species as reciprocally monophyletic, except C. occultans and C. berara which however, did not share the same alleles. In this analysis, C. noromalalae rather than C. puellarum was reconstructed as sister to all other Cophyla, but this alternative placement was not statistically supported by significant posterior probabilities. To understand whether Cophyla and Platypelis constitute reciprocally monophyletic sister genera as suggested by previous studies (Andreone et al. 2005b; Wollenberg et al. 2008), or Platypelis is paraphyletic as in our initial analysis (Fig. 2), we assembled a more extensive matrix containing DNA sequences of a larger number of gene fragments, especially of nuclear genes, for selected representatives of all cophyline genera (except Madecassophryne) (Fig. 4). For this analysis we chose Cophyla phyllodactyla and C. puellarum as representatives of the deepest clades within Cophyla, and three species representing the main intrageneric clades of Platypelis. The analysis recovered with high posterior probability and high maximum parsimony bootstrap values a monophyletic Cophyla (1.0/100), a monophyletic Platypelis (0.99/83), and sister-group relationship among these two genera (1.0/ 83). This topology was inferred from a larger number of characters (6244 bp, vs. 2627 bp in dataset 1), mainly from nuclear genes that are known to resolve more reliably deeper evolutionary relationships, as compared to the tree suggesting Platypelis paraphyly (Fig. 2), and we are therefore more confident in this second analysis. Consequently, the molecular data support the hypotheses that Cophyla (i) is monophyletic, (ii) is the sister group of Platypelis, (iii) contains seven species and candidate species concordantly differentiated in mitochondrial and nuclear genes, and (iv) C. puellarum is the sister taxon to all other species of Cophyla. REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 65

TABLE 1. Approximate fragment sizes (bp), primers and thermal cycling profiles used for the amplification of DNA fragments. The two different fragments of RAG1 refer to two alignments used in data sets 1 and 3, respectively (see Materials and Methods). Thermal cycling schemes start with temperature (in C) of each step, followed by the time in seconds between parentheses; cycling repetitions are indicated within brackets. Marker Primer name Primer sequence PCR conditions Reference 12S (635 bp) 12SAL AAACTGGGATTAGATACCCCACTAT 94(90), [94(45), 52(45), 72(90) x 33], 72(300) Kocher et al. (1989) / Hrbek and Larson (1999) 16SR3 TTTCATCTTTCCCTTGCGGTAC 12SAL AAACTGGGATTAGATACCCCACTAT 12SBH GAGGGTGACGGGCGGTGTGT 16S (609/731 bp) 16SL3 AGCAAAGAHYWWACCTCGTACCTTTTGCAT 94(90), [94(45), 52(45), 72(90) x 33], 72(300) Vences et al. (2003) 16SAH ATGTTTTTGATAAACAGGCG Vences et al. (2003) COB (350/615 bp) Cytb-a CCATGAGGACAAATATCATTYTGRGG 94(90), [94(30), 53(45), 72(90) x 35], 72(600) Bossuyt & Milinkovitch (2000) Cytb-c CTACTGGTTGTCCTCCGATTCATGT Bossuyt & Milinkovitch (2000) COX1 (530/622 bp) LCO1490 GGTCAACAAATCATAAAGATATTGG 94(90), [94(30), 49(45), 72(90) x 35] Folmer et al. (1994) HCO2198 TAAACTTCAGGGTGACCAAAAAATCA Folmer et al. (1994) RAG1 (503 bp) Rag1_Coph_F1 CGTGATCGGGTAAAAGGTGT 94(120), [94(20), 53(50), 72(180) x 35], 72(600) Newly developed Rag1_Coph_R1 TCGATGATCTCTGGAACGTG Newly developed RAG1 (1295 bp) Rag1-Mart Fl1 AGCTGGAGYCARTAYCAYAARATG 94(240), [94(45), 45(40), 72(120) x 45], 72(600) modified from Martin (1999) Rag-1Mart R6 GTGTAGAGCCARTGRTGYTT modified from Martin (1999) Rag-1AmpF2 ACNGGNMGICARATCTTYCARCC Chiari et al. (2004) Rag1-UC-R TTGGACTGCCTGGCATTCAT provided by D.R. Vieites KIAA1239 (836 bp) KIAA1239F1 CARCCTTGGGTNTTYCA 94(240), [94(45), 45(40), 72(120) x 45], 72(600) Shen et al. (2012) KIAA1239R1 CMACAAAYTGGTCRTTR Shen et al. (2012) KIAA1239NF1 GAGCCNGAYATHTTYTTYG Shen et al. (2012) KIAA1239NR1 TTCACRAANCCMCCNG Shen et al. (2012) SACS (865 bp) SACSF2 AAYATHACNAAYGCNTGYTAYAA 94(240), [94(45), 45(40), 72(120) x 45], 72(600) Shen et al. (2012) SACSR2 GCRAARTGNCCRTTNACRTGRAA Shen et al. (2012) SACSNF2 TGYTAYAAYGAYTGYCCNTGGAT Shen et al. (2012) SACSNR2 CKGTGRGGYTTYTTRTARTTRTG Shen et al. (2012) TTN (671 bp) TTNF1 TATGCTGARAAYATNGCNGGNAT 94(240), [94(45), 45(40), 72(120) x 35], 72(600) Shen et al. (2012) TTNR1 CCMCCRTCAAAYARNGGYTT Shen et al. (2012) TTNNF1 GATGGNMGKTGGYTNAARTGYAA Shen et al. (2012) TTNNR1 AGRTCRTANACNGGYTTYTTRTT Shen et al. (2012) 66 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

FIGURE 1. The three nominal species of Cophyla known to date: (a) C. phyllodactyla calling male, Nosy Be (voucher number unknown); (b) C. phyllodactyla, Nosy Be (voucher number unknown); (c) C. phyllodactyla, Manongarivo (ZSM 842/2003); (d) C. berara, Berara (holotype, ZSM 410/2000); (e) C. occultans, Nosy Be (voucher number unknown). REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 67

FIGURE 2. Phylogenetic tree of species and candidate species in the genus Cophyla. Majority-rule consensus tree derived from a partitioned Bayesian inference analysis of concatenated DNA sequences of the 12S, 16S, COX1, COB, and RAG1 genes (2627 bp), showing relationships among species of Cophyla. Numbers at nodes are posterior probabilities (only values >0.9 shown). For the three new species described herein the tree also shows their candidate species names according to Perl et al. (2014). 68 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

FIGURE 3. Majority-rule consensus tree derived from a partitioned Bayesian inference analysis of DNA sequences of the nuclear RAG1 gene (503 bp), showing relationships among species of Cophyla. Numbers at nodes are posterior probabilities (only values >0.9 shown). The tree was rooted with the same outgroups as in Fig. 2 (removed for better graphical representation). Bioacoustic, morphological, and osteological differentiation within Cophyla The molecular data indicate that the Montagne d Ambre massif is inhabited by three different species of Cophyla (C. maharipeo, C. noromalalae, C. puellarum) which are segregated by altitude (Figs. 2 3). Cophyla maharipeo was collected from Joffreville and Forêt d Ambre (630 720 m a.s.l.), C. noromalalae around the MANP park entrance (900 1050 m a.s.l.), and C. puellarum at Lac Maudit and Point de Vue du Grand Moulin sites within MANP (1246 1306 m a.s.l.). Although the three species occur in the Montagne d'ambre massif (Fig. 5), we did not record them in syntopy. They might be restricted to specific elevations but we cannot exclude the possibility that they may occur syntopically at some sites within the massif. Advertisement calls, as shown in oscillograms in Fig. 6, are structurally similar in Cophyla (a melodious single-note call repeated after regular intervals), but distinct differences are recognizable in call duration and intercall interval duration among most species and candidate species, including the three new species from Montagne d'ambre (Fig. 6; for detailed call descriptions see the species accounts below). This results in differences in call repetition rate among species. Summarizing the most important differences (Table 2), C. phyllodactyla has the fastest call repetition rate with 10 11 calls per 10 s, followed by C. berara, C. occultans, and C. noromalalae (all 5 6 calls per 10 seconds). Cophyla maharipeo and C. puellarum have the slowest call repetition (3 4 calls per 10 seconds). While the repetition rates might be influenced by different recording temperatures (i.e., the slowest call repetition rates REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 69

correspond to the lowest recording temperatures; Table 2), this is not true for some of the comparisons (e.g., C. berara, C. phyllodactyla, C. occultans, C. maharipeo recorded at rather similar temperatures). In addition, call duration (less affected by temperature differences than inter-call interval duration) also varies drastically: in particular, C. puellarum has the shortest calls despite having been recorded at the lowest temperature, and C. maharipeo has the longest calls although recorded at rather high temperatures. Hence, in particular these two species can easily be distinguished by advertisement call patterns from each other and from all other Cophyla, while C. noromalalae has values rather similar to C. berara. TABLE 2. Summary of call data of Cophyla species. Data for C. berara, C. occultans and C. phyllodactyla taken from Vences et al. (2005), except for call repetition rate (measured from Fig. 6). Data for C. occultans only from type locality (Nosy Be). Data from a population in Marojejy tentatively assigned to C. sp. Ca4. Recording temperature Call duration (ms) Inter-call interval duration (ms) Approximate call repetition rate (calls / 10 s) Spectral frequency range (Hz) C. berara 23.5 C 774 824 986 2272 5 2800 3000 C. occultans 26 C 500 550 1210 1360 6 ca. 4000 C. phyllodactyla 26 C 360 450 555 605 11 2300 2500 C. sp. Ca4 25 C 411 475 1126 1994 6 3450 3950 C. maharipeo ca. 25 C 1166 1346 2154 3881 3 2627 2842 C. noromalalae ca. 20 C 662 821 874 1882 5 2250 2484 C. puellarum ca. 17 C 326 390 1961 3996 4 2282 2411 FIGURE 4. Majority-rule consensus tree derived from a partitioned Bayesian inference analysis of concatenated DNA sequences of the 12S, 16S, COX1, COB, RAG1, KIAA1239, SACS, and TTN genes (6244 bp), showing relationships among species of the Cophylinae. Numbers at nodes are posterior probabilities (first number; values >0.95 bold) and maximum parsimony bootstrap values in percent (second value; >70% bold). The grey box highlights the included species of the genera Cophyla and Platypelis, which form two highly supported and reciprocally monophyletic groups. 70 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

FIGURE 5. Maps of northern Madagascar showing all known Cophyla localities verified by molecular data (or bioacoustic evidence: C. phyllodactyla at Benavony). The left map shows elevation whereas the right map shows remaining primary vegetation from the Madagascar vegetation mapping project (http://www.vegmad.org), with green colours symbolizing humid rainforest and montane forest and orange indicating dry forest. Morphologically we did not find any obvious external characters to differentiate species of Cophyla from each other, except for differences in body size among some lineages (Vences et al. 2005 and Table 3); C. berara: SVL of adult males 23 26 mm; C. occultans: SVL of adult males 16 21 mm; C. phyllodactyla: adult males 27 29 mm; C. maharipeo: adult males 23 mm, max. SVL 27 mm; C. noromalalae: adult males 24 25 mm, max. SVL 29 mm; C. puellarum: adult males 27 29 mm, max. SVL 34 mm; C. sp. Ca 4: adult male 20 mm. Because sizes are largely overlapping SVL is of limited use as a diagnostic character although C. phyllodactyla, C. noromalalae and C. puellarum are generally the largest and C. occultans and C. sp. Ca4 are the smallest species. Our osteological examination (no data for C. berara, C. occultans, and C. sp. Ca4; Figs. 9 10) confirmed the osteological character states described by Parker (1934) for Cophyla and Platypelis. All Platypelis examined (using both CT scans and clearing-and-staining) had a rudimentary ossified clavicle. Parker (1934) documented this character state in Platypelis cowanii, and we verified it in specimens of P. cf. barbouri, P. pollicaris, and P. tsaratananaensis. The posterior vomer (i.e., the post-choanal portion of the vomer) is not clearly recognizable in P. cf. barbouri, and we hypothesize it may be largely reduced in this species, or incompletely ossified and therefore not visible in the CT scan. In Platypelis cowanii (as reported by Parker 1934) and in P. pollicaris and P. tsaratananaensis (according to our data) the posterior vomer is centrally divided; it consists, on each side of the skull, of an elongated element with a central, enlarged and dentigerous part, and a lateral process ventrally running along the neopalatine and overlapping with it. As described by Parker (1934) for C. phyllodactyla, all species of Cophyla examined by us (C. maharipeo, C. noromalalae, C. phyllodactyla, C. puellarum) had the posterior vomer medially fused and not overlapping with neopalatines. This fused posterior vomer had no clearly recognizable vomerine teeth in C. noromalalae, probably weakly developed vomerine teeth in C. phyllodactyla and C. maharipeo (poorly recognizable in the CT scans), and more strongly developed, clearly recognizable vomerine teeth in C. puellarum. In agreement with Parker (1934), the clavicle was completely absent (no ossified element recognizable) in C. maharipeo, C. noromalalae, and C. phyllodactyla; however, this element was present and well recognizable in C. puellarum. These osteological differences were visible not only in the CT-scans (Figs. 9 10) but also in cleared and stained specimens: we REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 71

observed a clearly ossified clavicle in Platypelis pollicaris (RDR 1066 / ZSM 1732/2012), while no ossified or cartilaginous clavicle was visible in Cophyla (except C. puellarum). FIGURE 6. Comparative call data of partial call series of six species of Cophyla. C. occultans recorded in February 1995, 25 C; C. berara recorded in February 2000, 23.5 C; C. phyllodactyla, recorded in February 1992, 26 C; C. maharipeo sp. nov., UADBA-A 60232, recorded in January 2012, ca 25 C; C. noromalalae sp. nov., ZSM 3273/2012, recorded in November 2012, ca 20 C; C. puellarum sp. nov., UADBA-A 60237, recorded in November 2012, ca 17 C. Recordings of C. berara, C. occultans, and C. phyllodactyla are from Vences et al. (2005). The initial intensity peaks at each call of C. maharipeo and C. puellarum probably represent an artefact of the recording device, or an abnormal sound emission of the recorded specimens. To conclude, the three Cophyla candidate species from Montagne d'ambre are distinct from all nominal species of the genus, and from each other, by a substantial genetic differentiation in mitochondrial genes and in one nuclear gene. Further important diagnostic characteristics are the long call duration of C. maharipeo, the short call duration of C. puellarum, and the presence of a clavicle in C. puellarum. Additional minor differences in 72 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

morphology (i.e., body size and coloration) and bioacoustics will be discussed in the taxonomic accounts below. We conclude that the three candidate species represent independent evolutionary lineages and describe them as new species below. Taxonomic accounts Cophyla maharipeo sp. nov. (Figs. 7a, 11 & 12a) Remarks.This new species was listed as confirmed candidate species Cophyla sp. 1 by Vieites et al. (2009). Not included in Glaw & Vences (2007) and Wollenberg et al. (2008). Referred to as Cophyla sp. Ca1 by Perl et al. (2014). D Cruze et al. (2008) did not distinguish the three new species described herein; their Cophyla sp. nov. therefore includes records and altitudinal ranges of C. maharipeo, C. noromalalae, and C. puellarum. Holotype. ZSM 3251/2012 (ZCMV 12195), adult male, collected in a small patch of bamboo in the nun garden of Joffreville, -12.49397, 49.20504, 634 m a.s.l., on 17 January 2012, by A. Rakotoarison and A. Razafimanantsoa. Paratypes. UADBA-A 60231 (ZCMV 12193, female), ZSM 3252/2012 (ZCMV 12194, female), UADBA-A 60232 (ZCMV 12196, male), ZSM 3253/2012 (ZCMV 12200), UADBA 60233 (ZCMV 13501), same locality and collectors as the holotype. ZSM 1659/2008 (FGZC 1866), juvenile, collected in the Fontenay Private Nature Park near Joffreville (-12.495278, 49.200833, 720 m a.s.l.), on 26 February 2008, by J. Köhler & M. Franzen. Diagnosis and comparisons. Assigned to the genus Cophyla in the microhylid subfamily Cophylinae based on enlarged terminal discs on fingers and toes, absence of nuptial pads, absence of clavicle, posterior vomer undivided and not overlapping with neopalatines, and molecular phylogenetic relationships. From other arboreal species of cophylines the species can be distinguished by a combination of the following character states: small to medium body size (adult SVL 22 27 mm); usually with yellowish and white ventral colour; third toe length equals fifth toe; posterior vomer probably with weakly developed vomerine teeth; males with prepollical tubercle but lacking a finger-like prepollex as typical for Anodonthyla. Distinguished from all other species of Cophyla by advertisement call characteristics (long call duration 1166 1346 ms). Further distinguished from C. phyllodactyla by smaller body size (adult SVL 22 26 mm vs. 27 29 mm); from C. occultans by larger size (SVL 22 26 mm vs. 16 21 mm); and from C. berara by the fourth finger being equally long as third (vs. fourth finger slightly longer). For the distinction from the other two new Cophyla species described herein, see their respective diagnoses below. Description of the holotype. Adult male in good state of preservation, some muscle tissue removed from right thigh; snout-vent length 22.6 mm; body slender; head wider than long, not wider than body; snout bluntly rounded in dorsal and lateral views; nostrils not protuberant, nearer to tip of snout than to eye; canthus rostralis distinct and somewhat rounded, loreal region slightly concave to straight; tympanum moderately distinct, 47% of eye diameter; supratympanic fold starting at the posterior border of the eye and ending at the base of the forearm (less distinct on left side); tongue broadly rounded, not bifid or notched, no lingual papilla; maxillary teeth present; vomerine teeth small, rudimentary. Forelimbs slender; subarticular tubercles on all fingers single, protuberant; outer metacarpal tubercle relatively large and flat; inner metacarpal tubercle oblong, forming small protuberance at base of first finger; hand without webbing; fingers distinctly flattened and broad along entire length; relative length of fingers 1<2=4<3; finger discs distinctly enlarged, broadly rounded to slightly bilobate, with lateral fringes; nuptial pads absent. Hindlimbs slender; tibiotarsal articulation reaching between forelimb and tympanum when hindlimb adpressed along body; tibia length 39% of SVL; lateral metatarsalia strongly connected; inner metatarsal tubercle small; outer metatarsal tubercle small and flat, difficult to recognize; webbing between toes weakly developed, restricted to traces of webbing between third and fourth toe; subarticular tubercles on toes present, single, toe discs flattened and relatively broad, broadly rounded to slightly bilobate, smaller than discs of fingers; relative length of toes 1<2<5=3<4. Dorsal skin smooth, without dorsolateral folds. Ventral skin slightly granular on throat and chest and granular on belly. Skin on throat extended, marking the presence of vocal sac. REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 73

TABLE 3. Morphometric measurements (all in mm) of Cophyla form Joffreville and Montagne d Ambre and of three specimens of C. phyllodactyla from Nosy Be, one specimens from Manongarivo and one specimen of C. occultans from Nosy Be, all from the north of Madagascar. For abbreviations of measured variables, see Materials and Methods. Further abbreviations used: RHL, relative hindlimb length; (1) reaches between forelimb and tympanum, (2) reaches tympanum. MANP, Montagne d'ambre National Park. Museum number Field number C. maharipeo ZSM 3251/2012 ZCMV 12195 Sex Locality SVL HW HL TD ED END NSD NND FORL HAL HIL FOTL FOL TIL RHL HT male Joffreville 22.6 6.6 5.8 1.5 2.9 1.8 1.3 1.5 9.8 6.0 28.2 12.6 8.0 8.9 1 UADBA A60232 ZCMV 12196 PT male Joffreville 23.2 6.7 5.9 1.1 3.1 1.3 0.9 2.1 10.5 5.3 29.1 14.2 8.9 9.2 1 UADBA-A60233 ZCMV 13501 PT?? Joffreville 25 6.8 6.6 1.1 3.0 1.9 1.1 2.0 11.9 5.6 30.0 14.0 9.5 9.9 1 UADBA-A60231 ZCMV 12193 PT female?? Joffreville 24.3 7.9 7.4 1.2 2.9 1.8 1.4 1.9 14.8 6.3 31.2 15.4 9.2 11.1 1 ZSM 3252/2012 ZCMV 12194 PT female?? Joffreville 26.5 7.5 6.1 1.1 2.9 1.3 1.0 1.7 13.9 5.5 29.3 14.7 9.7 10.2 1 ZSM 3253/2012 ZCMV 12200 PT?? Joffreville 26.5 7.9 7.0 1.2 2.9 1.5 0.9 1.7 12.6 6.0 33.7 15.3 10.1 10.9 1 ZSM 1659/2008 FGZC 1866 PT subadult? Forêt d Ambre 17.3 6.2 5.6 1.2 2.3 1.7 1.0 1.9 11.7 4.0 11.3 11.3 7.0 8.3 1 C. noromalalae ZSM 3250/2012 ZCMV 13518 HT male MANP entrance 21.6 6.2 5.9 1.2 2.8 1.6 1.2 1.9 12.9 6.2 25.2 12.8 8.5 9 1 UADBA-A60235 ZCMV 13519 PT?? MANP entrance 23.4 7 6.3 1.2 3.3 1.8 1.1 1.8 11.2 5.8 28.5 12.7 8.5 9.7 1 ZSM 2070/2007 FGZC 1013 PT?? Montagne d Ambre 28.9 9.8 7.9 1.4 3.5 1.6 1.2 2.2 15.1 7.4 37.4 16.4 11.5 12.2 1 UADBA-A 60236 FGZC 4901 PT male Montagne d Ambre 25.4 7.8 6.7 1.6 3.2 2.2 1.6 1.6 13.1 6.4 31.4 14.6 9.7 10.2 1 ZSM 3273/2012 FGZC 4902 PT male Montagne d Ambre 24.3 6.8 6.6 1.6 2.4 2.3 1.2 1.7 11.7 5.3 27.5 13.6 8.5 9.7 1 continued on the next page 74 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

TABLE 3. (Continued) Museum number Field number C. puellarum ZSM 3249/2012 ZCMV 13521 ZSM 0782/2009 ZCMV 13005 ZSM 0781/2009 ZCMV 13004 UADBA-A 60237 FGZC 4903 ZSM 3274/2012 FGZC 4904 ZSM 3275/2012 FGZC 4905 UADBA-A60238 FGZC 4906 C. phyllodactyla --- ZCMV 12169 ZSM 0842/2003 FGZC 808 ZSM 460/2000 FGMV 2000-225 C. occultans ZSM 446/2010 FGZC 4690 Sex Locality SVL HW HL TD ED END NSD NND FORL HAL HIL FOTL FOL TIL RHL HT male MANP- Point de Vue du Grand Moulin 27.3 8.8 8 1.5 3.3 1.9 1.6 3.2 16 8.5 33.5 17.3 10.7 11.6 1 PT male MANP- Lac Maudit 28 11.1 7.9 2.1 3.6 2.2 1.3 2.1 15.8 8.5 34.3 17.2 11.0 11.6 1 PT juvenile MANP- Lac Maudit 21.2 7.5 5.3 1.5 1.9 1.6 1.5 2.2 12.5 6.3 25.2 12.5 8.4 8.6 1 PT male MANP- Point de Vue du Grand Moulin 28.2 9.9 8.1 1.7 3.8 3.1 1.6 1.8 15.5 7.4 35.9 16.7 11.1 11.3 1 PT male MANP- Point de Vue du Grand Moulin 28.1 9.4 8.2 1.9 3.8 2.0 1.3 2.0 16.26 7.9 34.6 17.2 11.0 10.7 1 PT female MANP- Point de Vue du Grand Moulin 33.6 10.3 9.7 2.5 5.2 2.1 1.4 2.5 19.5 9.8 42.6 19.3 13.3 13.7 1 PT male MANP- Point de Vue du Grand Moulin 28.8 9.8 8.6 1.9 3.8 2.3 1.3 1.9 14 8.6 36.8 17.2 11.4 11.6 1 -- male Nosy Be 28.4 11.0 8.5 1.4 3.1 1.9 1.6 2.0 16.6 6.9 35.2 17.16 11 11.8 1 -- male Manongarivo 27.4 9.3 7.9 2.0 3.4 2.1 1.2 2.1 14.2 6.4 36.0 16.6 11.1 11.0 1 -- Nosy Be 27.7 8.2 7.6 1.7 3.4 2.3 1.4 2.3 15.9 7.4 35.4 17.5 11.3 10.7 2 -- male Nosy Be 20.1 6.2 5.9 1.8 2.7 1.7 1.2 1.5 11.0 4.8 25.1 12.3 8.0 8.4 1 REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 75

Colour of holotype in preservation. After three years in 70% ethanol (Fig. 7a), anterior part of dorsum brown with a darker X-shaped pattern. Forelimbs and hindlimbs dorsally brown and beige. Ventrally white on throat with dark dots, brown and yellowish with dark dots on chest, dark brown and whitish with dark dots on belly. FIGURE 7. Comparative photographs of the three preserved holotypes (scale bar 5 mm): (a) C. maharipeo sp. nov., ZSM 3251/2012, (b) Cophyla noromalalae sp. nov., ZSM 3250/2012; (c) C. puellarum sp. nov., ZSM 3249/2012. Colour of holotype in life. Rather similar to colour in preservative (Fig. 11a). The beige part of the body was more yellowish. Ventrally (Fig. 12a) yellow on throat and chest, whitish on belly. Variation. Most of the paratypes are morphologically similar to the holotype (Table 3) except ZSM 1659/2008 (FGZC 1866) which is smaller than the other specimens. In five of them the third and fifth toes are of equal length and in three of them the third is slightly shorter than the fifth. The tibiotarsal articulation in all paratypes reaches between forelimb and tympanum. Males are smaller than females (Table 3). The dorsum is yellow with a brown patch and line (Fig. 11a), or greyish with a yellowish patch (Fig. 11c) for adult and subadult specimens. In one juvenile specimen the dorsum is dark beige with a dark brown line and patches (Fig. 11d). One unsexed specimen (ZSM 233/2004) from low elevations (650 m) of Montagne d Ambre with uncertain identity, but possibly referable to C. maharipeo, has a SVL of 31.4 mm, suggesting that this species might reach a larger size than is suggested by the type specimens. Etymology. The name is derived from Malagasy mahari-peo (equalling maharitra feo): long voice. This species has the longest call duration among all Cophyla species. Natural history. The holotype was found at night, calling on a bamboo tree. About eight calling males (including the holotype) were heard from the same small patch of bamboo in the garden of an abbey in Joffreville. During the day, at the same site, one female with oocytes and one male were found in a bamboo trunk, together with five juveniles. No further information is available on the natural history and reproduction of this species. Advertisement call. Calls were recorded from paratype UADBA-A 60232 (ZCMV 12196) on 17 January 2012, at an air temperature of ca. 25 C (Fig. 13). The following call description refers to the calls of this single specimen. As with all other Cophyla and most cophylines, single tonal calls were repeated in long series (continued at least for several minutes but probably much longer in undisturbed specimens), with regular inter-call intervals. Dominant spectral call frequency ranged from 2600 2800 Hz (N= 28). Temporal call parameters were as follows: call duration 1166 1346 ms (1257±40 ms; N= 28); duration of inter-call intervals 2154 3881 ms (2552±387 ms; N= 27). In the recorded call sequence, each call starts with a very short intensity peak visible in the oscillograms and of slightly higher frequency than the rest of the call. While we cannot exclude that this peak is a genuine part of the vocalization of this species, we find it more likely that it represents either an artefact of the recording equipment, or an abnormal feature in sound emission. 76 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

Cophyla noromalalae sp. nov. (Figs. 7b, 12b &14) Remark. Referred to as Cophyla sp. Ca2 by Perl et al. (2014). Holotype. ZSM 3250/2012 (ZCMV 13518), adult male, collected in a bamboo patch near the entrance office of Montagne d Ambre National Park (MANP), -12.616625, 49.150129, 900 m a.s.l., on 24 January 2012, by A. Rakotoarison and A. Razafimanantsoa. Paratypes. UADBA-A 60235 (ZCMV 13519), same locality, collection date and collector as the holotype; UADBA-A 60236 (FGZC 4901), calling male, collected on 10 December 2012 around MANP entrance by F. Glaw, F. Ratsoavina, O. Hawlitschek and T. Rajoafiarison; ZSM 3273/2012 (FGZC 4902), male, same locality, collection date and collector as the calling male, ZSM 2070/2007 (FGZC 1013), sex unknown, collected at MANP (Voie des milles arbres, -12.516667, 49.176667, 1050 m a.s.l.), on 23 February 2007 by P. Bora, H. Enting, F. Glaw, J. Köhler & A. Knoll. Diagnosis and comparisons. Assigned to the genus Cophyla in the microhylid subfamily Cophylinae based on enlarged terminal discs on fingers and toes, absence of clavicle, posterior vomer undivided and not overlapping with neopalatines, absence of nuptial pads, and molecular phylogenetic relationships. Among other arboreal species of cophylines the species can be identified by a combination of the following character states: small to moderate body size (adult SVL 22 29 mm; verified adult males 22 25 mm); most specimens uniformly yellowish ventrally; discs of fingers and toes pinkish in life; third toe at least slightly but often distinctly shorter than fifth; posterior vomer without clearly recognizable vomerine teeth (rudiments are probably present); males with prepollical tubercule but lacking a finger-like prepollex as typical for Anodonthyla. Distinguished from C. maharipeo by distinctly shorter calls, and from C. phyllodactyla by a distinctly slower call repetition rate (Table 2). Further distinguished from C. phyllodactyla by smaller body size (verified adult male SVL 24 25 mm vs. 27 29); from C. occultans by larger size (verified adult male SVL 22 25 mm vs. 16 21 mm). Most similar by morphology and bioacoustics to C. berara which however has strong and concordant differences in mitochondrial and nuclear DNA (Figs. 2 3). For the distinction from C. puellarum, see the respective diagnosis of that species below. Description of the holotype. Adult male in good state of preservation, some muscle tissue removed from right thigh. Snout-vent length 21.6 mm; body slender; head slightly longer than wide, not wider than body; snout bluntly rounded in dorsal and lateral views; nostrils not protuberant, nearer to tip of snout than to eye; canthus rostralis broadly rounded, loreal region concave; tympanum moderately distinct, 41% of eye diameter; supratympanic fold distinct, almost straight, running from posterior edge of eye to forelimb insertion; tongue broadly rounded, without lingual papilla, not bifid or notched; maxillary teeth present; vomerine teeth not visible. Forelimbs slender; subarticular tubercles on all fingers single, curved; outer metacarpal tubercle relatively small and flat; inner metacarpal tubercle oblong, forming large and distinct protuberance at base of first finger; hand without webbing; fingers distinctly broadly rounded to slightly bilobate, with lateral fringes, smaller than discs of fingers; relative length of fingers 1<2=4<3; finger discs broadly rounded to slightly bilobate; nuptial pads absent. Hindlimbs slender; tibiotarsal articulation reaching between forelimb and tympanum when hindlimb adpressed along body; tibia length 41% of SVL; inner metatarsal tubercle small and flat, outer metatarsal tubercle absent; webbing between toes weakly developed, with traces of webbing between third and fourth toe; subarticular tubercles on toes single; toes flattened and their discs relatively broad, broadly rounded to slightly bilobate; relative length of toes 1<2<3<5<4; third toe slightly (left) to distinctly (right) shorter than fifth. Dorsal skin smooth, without dorsolateral folds. Ventral skin smooth on throat and chest and moderately granular on belly. Colour of holotype in preservation. After several years in 70% ethanol (Fig. 7b), dorsum almost uniformly greyish with a few fine and small dark dots, and faint dark crossbands on toes and lower arm. Forelimbs greyish with moderate brown band; hindlimbs greyish with yellow dots. Finger and toe discs pinkish. Ventrally, brown and white with dark dots on throat, dark brown on chest, dark brown and whitish with dark dots on belly, and brown and yellowish on limbs. Colour of holotype in life. Dorsum almost uniformly beige with some light yellow patches (Fig. 14a). Finger and toe discs light orange. Ventrally almost uniformly whitish with a pinkish shade (Fig. 12b). Variation. The paratypes are of larger body size than the holotype (Table 3). In all paratypes, the third toe is slightly shorter than the fifth toe. Sexual dimorphism uncertain due to lack of data on females. Coloration varies among specimens. Dorsum can be beige with yellowish patch (Fig. 14a-b), or almost uniformly greyish with some yellow patches (Fig. 14c), or dark grey with pinkish and whitish patches on the flank (Fig.14d). REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 77

Etymology. This species is dedicated to Noromalala R. Raminosoa, the Malagasy supervisor of the first author, to whom we are immensely grateful for her support over the past years. Natural history. The holotype was found during the day in a bamboo patch near the entrance office of Montagne d Ambre National Park. The calling male (paratype 3273/2012) was found on a bamboo stand at night. No further information is available on the natural history and reproduction of this species. Advertisement call. Recorded from paratype ZSM 3273/2012 (FGZC 4902), around the park entrance of MANP, on 10 December 2012, at an air temperature of ca. 20 C (Fig. 15). The following call description refers to the calls of this single specimen. As with all other Cophyla and most cophylines, single tonal calls were repeated in long series (continued at least for several minutes but probably much longer in undisturbed specimens). Dominant frequency ranged from 2300 2500 Hz (N= 64). Temporal call parameters were as follows: call duration 662 821 ms (716±38 ms; N= 64); duration of inter-call intervals 874 1882 ms (1217±239 ms; N= 63). Cophyla puellarum sp. nov. (Figs. 7c, 12c & 16) Remark. This species (ZFMK 57465) has been figured in Glaw & Vences (1994: Fig. 361) as Platypelis sp. and in Glaw & Vences (2007: 135, Fig. 5d) as unidentified Platypelis from Montagne d Ambre. Referred to as Cophyla sp. Ca3 by Perl et al. (2014). FIGURE 8. Skull and shoulder girdle in ventral view of: (a) Cophyla puellarum sp. nov., UADBA-A 60237 and (b) Platypelis pollicaris (NSH 2419) (all scale bars 1 mm); Abbreviations: AVom Anterior vomer (=Pre-choanal portion of vomer); Cl Clavicle; Col Columella; Co Coracoid; Eo Exoccipital; Max Maxilla; Na Nasal; Npl Neopalatinum; ParSph parasphenoid; P-Co Precoracoid; Po Prootic; PrMax premaxilla; PVom Posterior vomer (=Post-choanal portion of vomer; coloured purple); Pt Pterygoid; Qj Quadratojugale; S Scapula; Spheth Sphenethmoid; S.Sc Supra-scapula. 78 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

FIGURE 9. Images of the ventral side of the skull, showing difference in shape and extension of posterior vomer (post-choanal portion of vomer; coloured purple) and its dentigerous part (vomerine teeth) in species of Cophyla and Platypelis, based on micro-ct scans (all scale bars 1 mm): (a) C. maharipeo sp. nov., ZSM 3252/2012 from Joffreville near Montagne d Ambre National Park; (a1) undivided posterior vomer, probably with weakly expressed vomerine teeth (poorly recognizable) of C. maharipeo sp. nov., ZSM 3252/2012; (b) C. noromalalae sp. nov., ZSM 3273/2012 from the entrance of Montagne d Ambre National Park; (b1) undivided posterior vomer with absence of or rudimentary vomerine teeth (not recognizable) of C. noromalalae sp. nov., ZSM 3273/2012; (c) C. puellarum sp. nov., UADBA-A 60237 from Point de Vue du Grand Moulin; (c1) undivided posterior vomer with well-developed vomerine teeth (clearly recognizable) of C. puellarum sp. nov., UADBA-A 60237; (d) P. pollicaris, NSH 2419, from Ranomafana National Park; (d1) posterior vomer divided, overlapping with neopalatines, and bearing vomerine teeth in P. pollicaris, NSH 2419; (e) P. cf. barbouri, NSH 2587, from Manombo Special Reserve; (e1) neopalatine of P. cf. barbouri, NSH 2587, posterior vomer not clearly recognizable, possibly reduced or not ossified. Holotype. ZSM 3249/2012 (ZCMV 13521), adult male, collected at Point de Vue du Grand Moulin in Montagne d Ambre National Park, -12.54077, 49.15424, 1246 m a.s.l., on 25 January 2012, by A. Rakotoarison and A. Razafimanantsoa. Paratypes. ZSM 0782/2009 (ZCMV 13005), male, and ZSM 0781/2009 (ZCMV 13004) juvenile, both collected at Lac Maudit, Montagne d Ambre National Park, -12.5856, 49.1457, 1306 m, on 19 and 20 November 2009 by A. Crottini, S. J. Hauswaldt, A. Lima, E. Rajeriarison, F. M. Ratsoavina and A. Razafimanantsoa; UADBA-A 60237 (FGZC 4903), calling male, ZSM 3274/2012 (FGZC 4904), calling male, ZSM 3275/2012 (FGZC 4905), gravid female, UADBA-A 60238 (FGZC 4906) male, all four collected at Point de Vue du Grand Moulin in Montagne d Ambre National Park, 12.54077, 49.15424, 1246 m, on 10 December 2012, by A. Rakotoarison and A. Razafimanantsoa; ZFMK 57465, adult female (SVL 33.3 mm), collected at high altitude of Montagne d Ambre National Park (no coordinates available) on 18 21 March 1994 by F. Glaw, N. Rabibisoa & O. Ramilison. REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 79

Diagnosis and comparisons. Assigned to the genus Cophyla in the microhylid subfamily Cophylinae based on enlarged terminal discs on fingers and toes, posterior vomer undivided and not overlapping with neopalatines, absence of nuptial pads and molecular phylogenetic relationships. Among other arboreal species of cophylines the species can be identified by combination of the following character states: body size comparatively large for Cophyla (SVL 27.3 33.6 mm); most of the specimens possess a uniformly yellow-greenish ventral colour; fifth toe slightly longer than third; males with prepollical tubercule but lacking a finger-like prepollex as typical for Anodonthyla; posterior vomer with well developed and clearly recognizable vomerine teeth; clavicle present. Distinguished from all other species of Cophyla by a unique advertisement call, characterized by a combination of short calls with long inter-call intervals; (Table 2). Further distinguished from C. phyllodactyla, C. maharipeo and C. noromalalae by the presence of a clavicle (no osteological data available for C. berara and C. occultans). Further distinguished from C. occultans and C. berara by larger body size (adult male SVL 27 29 mm vs. respectively 16 21 mm and 23 26 mm). FIGURE 10. Ventral view of shoulder girdle, showing differences in shoulder girdle structure between species and candidate species of Cophyla and Platypelis, based on micro-ct scans. Clavicle (missing in most Cophyla, present in in C. puellarum and as far as known in all Platypelis) coloured in red (all scale bars 1 mm). (a) C. maharipeo sp. nov., ZSM 3252/2012, from Joffreville; (b) C. noromalalae sp. nov., ZSM 3273/2012, from entrance of Montagne d Ambre National Park; (c) C. puellarum sp. nov., UADBA-A 60237, from Point de Vue du Grand Moulin, Montagne d Ambre National Park; (d) P. cf. barbouri, NSH 2587, from Manombo Special Reserve; (e) P. pollicaris, NSH 2419, from Ranomafana National Park. 80 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

FIGURE 11. Specimens of Cophyla maharipeo sp. nov. alive: (a) male holotype (ZSM 3251/2012); (b) calling male (holotype); (c) paratype specimen (assignment to voucher number not possible); (d) juvenile. Description of the holotype. Adult male in good state of preservation, some muscle tissue removed from right thigh. Snout-vent length 27.3 mm; body slender; head wider than long, not wider than body; snout rounded dorsally and laterally; nostrils not protuberant, nearer to tip of snout than to eye; canthus rostralis almost indistinct, loreal region very slightly concave; tympanum slightly distinct, 46% of eye diameter; supratympanic fold running from posterior edge of eye to forearm; tongue broadly rounded, without papilla, not bifid or notched; weakly developed maxillary teeth present; vomerine teeth not visible, but palpable. Forelimbs slender; subarticular tubercles on all fingers recognizable (single and rounded); outer metacarpal tubercle relatively large and flat; inner metacarpal tubercle oblong and large, forming distinct protuberance at base of first finger; hand without webbing; fingers distinctly flattened and broad along entire length; relative length of fingers 1<2=4<3, fourth finger slightly longer than second; finger discs broadly rounded to slightly bilobate, with lateral fringes; nuptial pads absent. Hindlimbs slender; tibiotarsal articulation reaching between forelimb and tympanum when hindlimb adpressed along body; tibia length, 42% of SVL; inner metatarsal tubercle oblong; outer metatarsal tubercle absent; webbing between toes very weakly developed, with traces of webbing between toes 3 4 and 4 5; subarticular tubercles on toes hardly recognizable; toes flattened and their discs relatively broad, broadly rounded to slightly bilobate, with lateral fringes, smaller than discs of fingers; relative length of toes 1<2<3<5<4 left toes and 1<2<3=5<4 right toes; fifth toe slightly (left) to distinctly (right) longer than third. Dorsal skin smooth, without dorsolateral folds. Ventral skin smooth on throat and chest and heavily granular on belly. Colour of holotype in preservation. After several years in 70% ethanol (Fig. 7c), dorsum brownish with irregular markings, including a darker W-shaped patch extending over the posterior dorsum. Forelimbs and hindlimbs dorsally lighter with dark crossbands. Ventrally, uniformly beige, with some darker colour on venter caused by inner organs shining through the ventral skin. Some dark colour along lower jaws. REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 81

Colour in life. Dorsum brownish with yellowish patches that have a greenish shade (Fig. 16b). Ventrally uniformly whitish (Fig. 12c). FIGURE 12. Ventral sides of the holotypes of the three newly described Cophyla species from Montagne d Ambre and Joffreville, photographed from living specimens. Inset photos show the same specimens in dorsolateral view. (a) C. maharipeo sp. nov., male holotype ZSM 3251/2012, Joffreville; (b) C. noromalalae sp. nov., male holotype ZSM 3250/2012, park entrance, male; (c) C. puellarum sp. nov., male holotype ZSM 3249/2012, Point de Vue du Grand Moulin. 82 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

FIGURE 13. Audiospectrogram and oscillogram of two advertisement calls (part of a longer call series) of Cophyla maharipeo sp. nov. from paratype UADBA-A 60232, recorded at Joffreville on 17 January 2012, ca 25 C, number of recording DR089. The sounds above 3000 Hz represent other frogs or insects, and the initial intensity peaks at each call probably represent an artefact of the recording device or an abnormal sound emission of the recorded specimen. Variation. Most of the paratypes are morphologically similar to the holotype (Table 3) except the females ZSM 3275/2012 (FGZC 4905) and ZFMK 57465 which are larger than the males. The third toe is slightly shorter than the fifth toe in all specimens. The female is larger than the males (Table 3). There is large variation of coloration between the specimens. In life, dorsum is beige with yellow patches (Fig. 16a), or beige and greenish with yellow patches (Fig.16b), or brown with yellow patches (Fig. 16c), or yellow with brown patches (Fig. 16e) for adult and subadult specimens. The paler color pattern probably reflects coloration at night whereas the more contrasted pattern is probably typical for specimens during the day, but our fieldnotes do not allow to confirm unambiguously this hypothesized color change. Dorsum is dark brown with yellow patches and yellow vertebral line in a juvenile specimen (Fig. 16d). Etymology. The name is derived from Latin puellarum: girls, genitive plural of puella. This species is dedicated to the "girls team" of a joint TU Braunschweig and Université d'antananarivo expedition who discovered the first specimen in 2009 at Lac Maudit. Natural history. The holotype was found during the day with two metamorphosing tadpoles in a tree hole in highland forest. An ovigerous female (ZSM 3275/2012) contained 21 eggs. The recorded male (UADBA-A 60237) was found on a tree trunk in December 2012 in the same area as the holotype. A second calling male (ZSM 3274/ 2012) was found on a Pandanus leaf. More specimens were regularly observed calling. They were typically heard from low perches of ca. 100 150 cm on leaves, but in a few cases also from higher perches of above 2 m from the ground. Advertisement call. Recorded from paratype UADBA-A 60237 (FGZC 4903), at Point de Vue du Grand Moulin in MANP, on 10 December 2012, at an air temperature of 16.6 C (Fig. 17). The following call description refers to the calls of this single specimen. As with all other Cophyla and most cophylines, single tonal calls were repeated in long series (continued at least for several minutes but probably much longer in undisturbed specimens). The dominant spectral call frequency ranged from 2300 2400 Hz (N= 47). Temporal call parameters were as follows: call duration 326 390 ms (363±13 ms; N= 47); duration of inter-call intervals 1961 3996 ms (2640±594 REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 83

ms; N= 46). As in C. maharipeo, each call in the recorded sequence starts with a very short intensity peak, and we hypothesize this represents an artefact of the recording equipment or an abnormal call feature (see call description of C. maharipeo). FIGURE 14. Specimens of Cophyla noromalalae sp. nov. in life: (a) male holotype, ZSM 3250/2012; (b) male observed calling, ZSM 3273/2012; (c) male paratype, UADBA-A 60236; (d) paratype specimen, ZSM 2070/2007. Discussion Although only three nominal species of the Madagascan microhylid genus Cophyla have been described to date (Glaw & Vences 2007), the presence of additional candidate species has been recognised for some years (Wollenberg et al. 2008, Vieites et al. 2009, Perl et al. 2014). The present study doubles the number of species in the genus by describing three new species, which all occur in the Montagne d'ambre area: C. maharipeo, C. noromalalae and C. puellarum. These discoveries highlight the incomplete knowledge of the microhylid fauna of Madagascar (Andreone et al. 2005b), and confirm the remarkable degree of microendemism in Malagasy microhylids, with many species known from narrow ranges only (Wollenberg et al. 2008, Rakotoarison et al. 2012). The available data further confirm that the genus Cophyla has its highest species diversity in, and might be restricted to, northern Madagascar, further contributing to the known high species diversity of cophylines in this part of the island. It is likely that more species of Cophyla will be discovered in the future. For example molecular data suggest the existence of a candidate species (C. sp. Ca4) from the base of the Marojejy massif, previously considered as C. cf. occultans (e.g., Vences et al. 2005, Wollenberg et al. 2008, Vieites et al. 2009) or Cophyla sp. 2 Marojejy (Andreone et al. 2005b) but apparently more closely related to C. noromalalae (Fig. 2). Additional populations tentatively assignable to Cophyla due to their similarity with C. occultans are known from Sambava and Voloina, also situated in north-eastern Madagascar (Vences et al. 2005). However due to the lack of molecular data and the morphological similarity of Cophyla and Platypelis, these records need confirmation and further research before formal description. 84 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.

FIGURE 15. Audiospectrogram and oscillogram of three advertisement calls (part of a longer call series) of Cophyla noromalalae sp. nov. from paratype ZSM 3273/2012, recorded around the entrance of Montagne d'ambre National Park on 10 December 2012, ca. 20 C, number of recording DR90. The sounds above 2500 Hz represent other frogs or insects. It is worth noting that the majority of Cophyla records are from low elevations (between sea level and ca. 600 m), with only C. noromalalae and C. puellarum reaching moderately high elevations around 1050 m and 1300 m, respectively. High-elevation areas between 1300 2500 m in other massifs of northern Madagascar such as Marojejy, Tsaratanana, and Manongarivo, are instead populated by species of Platypelis, several of which are local or regional endemics (e.g., P. alticola, P. olgae, P. ravus, P. tsaratananaensis; Glaw & Vences 1994, Glaw et al. 2012a, Rakotoarison et al. 2012), and sometimes occur in remarkable population densities on these massifs (Rakotoarison et al. 2012). Our multi-gene molecular phylogeny (Fig. 4) provides the most compelling evidence to date that Platypelis and Cophyla are sister genera and reciprocally monophyletic. Our data also indicate that C. puellarum is sister to all other Cophyla, and is characterized by (i) a larger body size than other Cophyla, (ii) elevational occurrence >1000 m, and (iii) presence of a clavicle which is absent in at least three other species of the genus (and we hypothesize is absent also in the remaining ones), but is typical for Platypelis. A hypothetical evolutionary scenario would assume a Cophyla-Platypelis ancestor occurring in the highlands of northern Madagascar. After an initial split, one lineage (the Platypelis ancestor) diversified and the arising species adapted to a diverse set of habitats, with some expanding their ranges into southern Madagascar. The second lineage (the Cophyla ancestor) first continued dwelling in the highlands of northern Madagascar and then split into one species (C. puellarum) largely retaining the ancestral morphology and environmental adaptations, and a second species that reduced the clavicle and diversified in the lowlands of northern Madagascar, giving rise to the remaining Cophyla species. Once the species inventory of Cophyla is complete and their ranges have been sufficiently clarified, this scenario can be tested by reconstructing ancestral areas and ancestral character states. Similar to the situation in other cophylines, the new species described herein represent strongly differentiated genetic lineages, not indicative of taxonomic inflation or oversplitting. Indeed the genetic differentiation between them is deep, and also congruently found in a conserved gene such as RAG1 that is often used to decipher deep relationships among major clades of vertebrates. Furthermore, the bioacoustic differentiation is strong, with the call of C. maharipeo being different not only from that of all other nominal species of Cophyla, but also being the REVISION AND THREE NEW SPECIES OF COPHYLA Zootaxa 3937 (1) 2015 Magnolia Press 85

longest known call of all cophylines. The osteological differentiation of C. puellarum (presence of a clavicle), and the occurrence of the three new species in close geographic proximity (though not syntopy), without any signal of genetic admixture in mtdna and one nuclear gene, strongly suggests they represent not only independent evolutionary lineages, but would also be supported as species by the classical biological species criterion. FIGURE 16. Specimens of C. puellarum sp. nov. in life: (a) calling male paratype ZSM 3274/2012; (b) male holotype, ZSM 3249/2012; (c) male UADBA-A 60237; observed calling; (d) juvenile ZSM 0781/2009, (e) ovigerous female ZSM 3275/2012. Specimens in b, c, e from Point de Vue du Grand Moulin in Montagne d'ambre National Park. Specimen in d from Lac Maudit. The three Cophyla described here seem to be local endemics to the Joffreville, Forêt d'ambre Special Reserve and Montagne d Ambre area. Despite rather intensive surveys in other areas of northern Madagascar they have not yet been found elsewhere. One specimen of Cophyla (ZSM 1674/2008) from Montagne des Français, close to Montagne d'ambre (Fig. 5), might belong to C. noromalalae but we could not verify this due to the lack of a tissue sample for molecular analysis. Similarly, the identity of the Cophyla record from Ankarana (Raselimanana 2008) requires further study. Of the three new species, C. maharipeo is known from an unprotected area, i.e., a small patch of bamboo in a garden in Joffreville, and from Forêt d'ambre Special Reserve which despite being legally protected, is under high deforestation pressure (e.g., Glaw et al. 2012b). C. noromalalae and C. puellarum occur in Montagne d'ambre National Park in the core of this protected area. However, Montagne d'ambre is subject to limited anthropogenic pressure, especially at its edges. In general, the three new species are thus threatened by habitat loss from agricultural activities and logging. Following the criteria of the International Union for Conservation of Nature (IUCN 2001) and the Red List assessment of other amphibian and reptile species endemic to Montagne d'ambre and Forêt d'ambre (Andreone et al. 2005a, 2008; Glaw et al. 2012b) we propose for the three species a status as Near Threatened in the IUCN Red List, based on all three species nearly qualifying for listing as Vulnerable under criterion D2 (population with a restricted area of occupancy and less than five known locations, prone to the effects of human activities or stochastic events within a short time period). We do not 86 Zootaxa 3937 (1) 2015 Magnolia Press RAKOTOARISON ET AL.