Target-enriched DNA sequencing from historical type material enables a partial revision of the Madagascar giant stream frogs (genus Mantidactylus)

Abstract

The subgenus Mantidactylus is a group of frogs endemic to Madagascar, including the largest anuran species on the island. Although these frogs are common and widely distributed, their taxonomy remains unclear. Two species are currently recognised, M. grandidieri and M. guttulatus, with another available name, Rana pigra, considered to be a synonym of M. grandidieri. However, molecular studies have suggested the presence of several cryptic species within the group. Additionally, due to the lack of prominent morphological features, allocating the available names to evolutionary lineages has proven challenging. In the present study, we take a first step towards solving these problems by using fragments of the 16S mitochondrial gene and RAG1 nuclear gene from all over the range of the subgenus to describe its genetic diversity. We also use a newly designed target enrichment laboratory protocol to sequence three mitochondrial fragments from five name-holding museum specimens (as old as 120 years) in order to determine to which lineages the existing names should be applied. The study of the 16S mitochondrial gene revealed 7 geographically separated lineages, distinct enough to be considered candidate species. Out of the five museum specimens analysed, four successfully yielded DNA sequences and could be attributed to one of the aforementioned lineages. Therefore, the name Mantidactylus grandidieri should be applied to the populations from North-Eastern Madagascar, while M. guttulatus refers to populations from inland localities along the Eastern coast of the island. On the other hand, the holotype of Rana pigra did not yield enough genetic material to allow definitive identification. While our data were not sufficient to assess the status of the four lineages distributed along the Eastern coast, the populations from North-Western Madagascar were highly distinct on both the mitochondrial and nuclear markers. We thus describe them as a new species, M. radaka sp. nov.

Full text

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=tnah20
Journal of Natural History
ISSN: 0022-2933 (Print) 1464-5262 (Online) Journal homepage: https://www.tandfonline.com/loi/tnah20
Target-enriched DNA sequencing from historical
type material enables a partial revision of
the Madagascar giant stream frogs (genus
Mantidactylus)
Loïs Rancilhac, Teddy Bruy, Mark D. Scherz, Elvis Almeida Pereira, Michaela
Preick, Nicolas Straube, Mariana L. Lyra, Annemarie Ohler, Jeffrey W.
Streicher, Franco Andreone, Angelica Crottini, Carl R. Hutter, J. Christian
Randrianantoandro, Andolalao Rakotoarison, Frank Glaw, Michael Hofreiter
& Miguel Vences
To cite this article: Loïs Rancilhac, Teddy Bruy, Mark D. Scherz, Elvis Almeida Pereira, Michaela
Preick, Nicolas Straube, Mariana L. Lyra, Annemarie Ohler, Jeffrey W. Streicher, Franco Andreone,
Angelica Crottini, Carl R. Hutter, J. Christian Randrianantoandro, Andolalao Rakotoarison,
Frank Glaw, Michael Hofreiter & Miguel Vences (2020): Target-enriched DNA sequencing from
historical type material enables a partial revision of the Madagascar giant stream frogs (genus
Mantidactylus), Journal of Natural History
To link to this article: https://doi.org/10.1080/00222933.2020.1748243
View supplementary material Published online: 06 May 2020.
Submit your article to this journal View related articles
View Crossmark data

Target-enriched DNA sequencing from historical type
material enables a partial revision of the Madagascar giant
stream frogs (genus Mantidactylus)
Loïs Rancilhac
a
, Teddy Bruy
b,c
, Mark D. Scherz
a,b
, Elvis Almeida Pereira
a,d
,
Michaela Preick
e
, Nicolas Straube
f
, Mariana L. Lyra
g
, Annemarie Ohler
c
,
Jeffrey W. Streicher
h
, Franco Andreone
i
, Angelica Crottini
j
, Carl R. Hutter
k
,
J. Christian Randrianantoandro
l
, Andolalao Rakotoarison
l
, Frank Glaw
b
,
Michael Hofreiter
e
and Miguel Vences
a
a
Zoological Institute, Braunschweig University of Technology, Braunschweig, Germany;
b
Sektion
Herpetologie, Zoologische Staatssammlung München (ZSM-SNSB), München, Germany;
c
Institut de
Systématique, Évolution, Biodiversité (ISYEB), Sorbonne Université, Paris, France;
d
Programa de Pós-
Graduação em Biologia Animal, Departamento de Biologia Animal, Laboratório de Herpetologia,
Universidade Federal Rural do Rio de Janeiro, Seropédica, Brazil;
e
Faculty of Mathematics and Natural
Sciences, Institute for Biochemistry and Biology, University of Potsdam, Potsdam, Germany;
f
Department of
Natural History, University Museum of Bergen, Bergen, Norway;
g
Depto de Zoologia, Instituto de Biologia,
Universidade Estadual Paulista - UNESP, Rio Claro, Brazil;
h
Department of Life Sciences, The Natural History
Museum, Cromwell Road, London, UK;
i
Sezione zoologia, Museo Regionale di Scienze Naturali, Torino, Italy;
j
Cibio, Research Centre in Biodiversity, Genetics and Evolution, InBio, Universidade do Porto, Vairão,
Portugal;
k
Biodiversity Institute and Department of Ecology and Evolutionary Biology, University of Kansas,
Lawrence, KS, USA;
l
Mention Zoologie et Biodiversité Animale, Faculté des Sciences, Université
d’Antananarivo, Antananarivo, Madagascar
ABSTRACT
The subgenus Mantidactylus is a group of frogs endemic to
Madagascar, including the largest anuran species on the island.
Although these frogs are common and widely distributed, their tax-
onomy remains unclear. Two species are currently recognised, M.
grandidieri and M. guttulatus, with another available name, Rana
pigra,consideredtobeasynonymofM. grandidieri.However,mole-
cular studies have suggested the presence of several cryptic species
within the group. Additionally, due to the lack of prominent morpho-
logical features, allocating the available names to evolutionary lineages
has proven challenging. In the present study, we take a first step
towards solving these problems by using fragments of the 16S mito-
chondrial gene and RAG1 nuclear gene from all over the range of the
subgenus to describe its genetic diversity. We also use a newly
designed target enrichment laboratory protocol to sequence three
mitochondrial fragments from five name-holding museum specimens
(as old as 120 years) in order to determine to which lineages the
existing names should be applied. The study of the 16S mitochondrial
gene revealed 7 geographically separated lineages, distinct enough to
be considered candidate species. Out of the five museum specimens
analysed, four successfully yielded DNA sequences and could be
ARTICLE HISTORY
Received 15 October 2019
Accepted 20 March 2020
KEYWORDS
Anura; Mantidactylus
guttulatus;Mantidactylus
grandidieri;Mantidactylus
radaka sp. nov; Madagascar;
Target enrichment
CONTACT Loïs Rancilhac loisrancilhac@gmail.com
Supplemental data for this article can be accessed here.
JOURNAL OF NATURAL HISTORY
https://doi.org/10.1080/00222933.2020.1748243
© 2020 Informa UK Limited, trading as Taylor & Francis Group
Published online 06 May 2020

attributed to one of the aforementioned lineages. Therefore, the name
Mantidactylus grandidieri should be applied to the populations from
North-Eastern Madagascar, while M. guttulatus refers to populations
from inland localities along the Eastern coast of the island. On the
other hand, the holotype of Rana pigra did not yield enough genetic
material to allow definitive identification. While our data were not
sufficient to assess the status of the four lineages distributed along
the Eastern coast, the populations from North-Western Madagascar
were highly distinct on both the mitochondrial and nuclear markers.
We thus describe them as a new species, M. radaka sp. nov.
http://www.zoobank.org/urn:lsid:zoobank.org:pub:1EDDAF0D-FE37-490A-B09E-E136A0C5CB35
Introduction
Type specimens are an important source of information for taxonomists as they allow
name allocation to taxa (Banks et al. 1993). They provide an indication of morphological
features of the species described, and are linked to valuable information, such as the date
and location of collection. In taxonomic revisions of species complexes, attributing a type
specimen –and thereby a scientific name –to the correct taxon can prove challenging,
especially with cryptic species that show poor morphological divergence. This is particu-
larly true in the case of species without clear diagnostic morphological features, or when
the state of specimen preservation does not permit the evaluation of morphological or
other diagnostic characteristics. Information that can be helpful in these cases, such as
collection locality, is often missing or imprecise for old type specimens, making their
identity even more difficult to disentangle (Cong et al. 2019).
In this context, DNA data can be extremely valuable, assuming that the focal species are
genetically distinct, and that genetic data are available to compare with the type specimens.
However, sequencing molecular markers from historical specimens can prove challenging,
due to degradation of DNA over time or to the specimen preservation techniques used
(Wandeler et al. 2007). Specific laboratory workflows have been developed to overcome this
issue (e.g. Gansauge and Meyer 2013), which have helped to resolve many taxonomic
problems (e.g. Hind et al. 2014;Shokralaetal.2015; Prosser et al. 2016;Congetal.2019)as
well as other biological questions (reviewed in Burrell et al. 2015). However, these approaches
have so far rarely been successful when applied to specimens preserved in wet collections (i.e.
most amphibians, reptiles and fishes). An important aspect is that the chemical composition of
the preservatives used in these collections over time is often unknown. For instance, many
specimens collected since around 1895 have been fixed with formaldehyde before being
stored in ethanol for long-term preservation (Musiałet al. 2016). While this method enables
long-term conservation of morphological features, formaldehyde alters the DNA by cross-
linking other macromolecules (e.g. proteins) to it such that its extraction, amplification and
sequencing become difficult (Tang 2006; Gilbert et al. 2007).
While some authors have successfully sequenced mtDNA fragments (Friedman and
DeSalle 2008; Ruane and Austin 2017; McGuire et al. 2018), and even complete mitochon-
drial genomes (Hykin et al. 2015; Li et al. 2015,2016; Evans et al. 2019) from specimens in
wet collections, standard workflows are still rare. In this context, short-read sequencing
2L. RANCILHAC ET AL.

through Massively Parallel Sequencing (MPS) platforms is a very promising approach to
apply ‘museomics’to wet collections. Targeting of DNA fragments commonly used as
barcodes for lineage characterisation via DNA hybridisation capture could be an efficient
way to identify enigmatic museum specimens and solve taxonomic problems. As an
attempt to provide a workflow combining MPS and targeted DNA barcode sequences
to be applied to alcohol-preserved anuran amphibians, we here revise the taxonomy of
the Madagascar giant stream frogs (genus Mantidactylus, subgenus Mantidactylus). Using
an MPS-based approach, we aim to assign the existing type specimens, preserved in wet
collections for up to 136 years, to genetic clades represented by more recently collected
material in order to correctly apply nomenclature.
The genus Mantidactylus Boulenger, 1895 (family Mantellidae Laurent, 1946)isan
apparently diverse, but largely understudied group of frogs, endemic to Madagascar.
The current taxonomy accepts 32 species, divided into six subgenera. Although recent
DNA barcoding surveys revealed the existence of almost 70 undescribed candidate
species (Vieites et al. 2009; Randrianiaina et al. 2011; Rosa et al. 2012; Perl et al. 2014)
characterised by high mitochondrial DNA divergences, only limited taxonomic work has
focused on this genus in the past 20 years (Glaw and Vences 1999,2004,2006; Vences and
Glaw 1999,2004; Mercurio and Andreone 2007; Vences et al. 2018; Scherz et al. 2019).
Within this genus, the species of the subgenus Mantidactylus stand out because of their
very large size, being Madagascar’s largest native frogs. Two species are currently accepted:
Mantidactylus grandidieri Mocquard, 1895 and M. guttulatus (Boulenger, 1881) (Glaw and
Vences 2007; AmphibiaWeb 2019). One more taxon, Rana pigra Mocquard, 1900,was
described, but the name was later synonymised with M. guttulatus by Boulenger (1918),
and then with M. grandidieri by Blommers-Schlösser (1979). Molecular studies have revealed
that several lineages within the genus may deserve species status (Vieites et al. 2009;Perlet
al. 2014), but due to the lack of interspecific morphological differentiation combined with
intraspecificvariation, their status is unclarified, and it is uncertain to which populations the
described names should be applied. We here use a comprehensive molecular dataset
covering all the known range of the subgenus Mantidactylus to delimit candidate species
within this subgenus, and include samples of type specimens of M. grandidieri, M. guttulatus
and R. pigra, housed in the British Museum of Natural History (BMNH) and the Muséum
National d’Histoire Naturelle (MNHN), in the genetic analysis. Doing so, we aim to (i) clarify
species boundaries within the subgenus Mantidactylus, (ii) determine to which lineages the
existing names should be applied and (iii) allocate new names to undescribed lineages
when relevant. This study is also intended to serve as a pilot to assess the efficiency of
genetic analysis of historical wet collection material for taxonomic revision.
Materials and methods
Sample collection
Frog individuals were collected during day or night along streams in both rainforest and
secondary habitat, over multiple expeditions to Madagascar from 2000 to 2016.
Specimens were anaesthetised and then euthanised by an overdose of MS222 or chlor-
obutanol, fixed in 95% ethanol, and thereafter transferred to 75% ethanol for long-term
JOURNAL OF NATURAL HISTORY 3

storage. Before fixation, tissue samples of thigh muscle or tongue were taken and
separately preserved in 99–100% ethanol.
Field numbers refer to the zoological collections of F. Andreone (FAZC, FA), A. Crottini
(ACZC,ACP),P.-S.Gehring(PSG),M.D.Scherz(MSZC),C.R.Hutter(CRH),D.R.Vieites(DRV),M.
Vences and F. Glaw (FGMV, ZCMV, FGZC) and the tissue collection of M. Vences (MVTIS).
MAVOA refers to tissues collected by the team of Madagasikara Voakajy. Preserved voucher
specimens studied are deposited in the Zoologische Staatssammlung München (ZSM),
Germany, and Museo Regionale di Scienze Naturali di Torino, Italy (MRSN). Additional speci-
mens (not analysed morphologically herein) were deposited at the Université d’Antananarivo,
Madagascar, Mention Zoologie et Biodiversité Animale (UADBA) and the University of Kansas
(KU) Biodiversity Institute. We took samples of historical type specimens preserved in the
Muséum National d’Histoire Naturelle (MNHN), Paris, France, and the Natural History Museum
(formerly British Museum of Natural History;BMNH),London,UK.Forsomesequences
retrieved from GenBank, voucher specimens are tagged with field numbers RAN (R. A.
Nussbaum), deposited in the Museum of Zoology, University of Michigan (UMMZ).
Target enrichment laboratory workflow for historical type specimens
DNA extractions from numerous specimens of historical wet collections, in many cases
formalin-fixed, showed that these often contain substantial amounts of highly fragmented
DNA, therefore not accessible using standard PCR-based amplification (e.g. Hykin et al. 2015;
Li et al. 2016; Evans et al. 2019). We therefore set out to develop a bait-capture approach,
specifically to obtain DNA information from historical collections of anurans from
Madagascar. Because of the existing dense sampling of ethanol-preserved tissues of almost
all described Malagasy anuran species, as well as numerous undescribed candidate species
(Vieites et al. 2009;Perletal.2014), we did not aim to obtain sequences of a large number of
markers from the historical specimens; instead, we followed a strategy to maximise the
probability of capture of three mitochondrial gene fragments that have been sequenced as
standard barcoding and phylogenetic markers from Malagasy frogs, i.e., 16S rRNA (16S),
cytochrome b (cob) and cytochrome oxidase subunit 1 (cox1). For these genes, we compiled
and aligned representative sequences for each species and candidate species as available
from Vences et al. (2005a,2005b), Vieites et al. (2009), Wollenberg et al. (2011), Perl et al.
(2014) and Scherz et al. (2016). Because our main target was Mantidactylus, we included 58
cob (415 bp) and 72 cox1 (510 bp) sequences of this genus only, as well as 610 16S (ca. 415
bp) sequences of all species of Malagasy anurans available. An additional 57 ND2 sequences
(389 bp) of chameleons were added to be able to capture sequences of this group for
another project. The 797 sequences in total were processed by Arbour Biosciences to
produce a bait set using the myBaits approach. For this, N bases were replaced with T.
RepeatMasker was used to soft mask simple repeats with a 0.1% threshold. Finally, 18,928
baits of 70 nucleotides in length with 5 x tiling density were designed. These were BLASTed
against seven complete or partial mitochondrial genomes of Malagasy frogs (from
Kurabayashi et al. 2008), and their hybridisation melting temperature estimated. Non-
specific baits were filtered out based on melting temperature criteria, and baits that were
99% identical were collapsed, leaving a final number of 5962 baits.
Tissue samples were taken in 2018 and 2019 from five museum specimens of the
holdings of BMNH and MNHN, corresponding to all name-bearing specimens in the
4L. RANCILHAC ET AL.

subgenus Mantidactylus. All specimens were collected at the end of the nineteenth
century and are currently stored in ethanol (ca. 70%), with an unknown history of fixation
and preservation. However, at the BMNH it is suspected that the specimens were directly
fixed in pure spirits of wine as this was the standard practice for preservation of ‘lower
vertebrates’during the nineteenth century, when the types of M. guttulatus were regis-
tered (Günther 1880). Specimens were taken out of their jars, and samples of muscle
tissue from the thigh (and additionally, liver in some specimens) were taken, using a clean
and DNA-free pair of scissors and forceps for each individual. Tissue samples were stored
and transported in clean vials filled with ca. 1 ml 100% ethanol, which had been filled in a
laboratory where no DNA work on Mantidactylus had previously been carried out.
DNA from historical samples was extracted, converted into a single-stranded library
and sequenced following a protocol developed by NS, ML, and collaborators: the samples
were taken to a clean lab and were washed prior to extraction with Qiagen® PE buffer.
DNA was then extracted from muscle tissue following Dabney et al. (2013). Extracts were
incorporated into a single-stranded DNA (ss-DNA) library, which has been shown to
increase library complexity for both ancient (Gansauge and Meyer 2013) as well as
Formalin-Fixed Paraffin-Embedded samples (Stiller et al. 2016). After adapter ligation
using custom adapter sequences from Gansauge and Meyer (2013), libraries were ampli-
fied and indexed with custom Illumina indexing primers.
Ss-DNA libraries were captured for aforementioned target sequences using the Arbour
Biosciences MyBaits kit as described in the following protocol: we used 14.5 μL of each
indexed library in 24-h reactions at a hybridisation temperature of 65°C following the
MyBaits target enrichment protocol, except for reducing the bait volume to 2.75 μL per
reaction and substituting the missing 2.75 μL with nuclease-free water. After hybridisa-
tion, the DNA-RNA hybrid molecules were bound to streptavidin-coated magnetic beads
and the reactions were washed and eluted according to the MyBaits protocol. Final
elution was in 30 μL of 10 mM Tris-Cl, 0.05% TWEEN-20 solution (pH 8.0).
To increase target capture success, the procedure was performed twice, i.e. the
amplified captured library was used as a template for repeating the process as in Li et
al. (2013,2015), Paijmans et al. (2016,2017), Springer et al. (2015) and Templeton et al.
(2013). Enriched libraries were pooled in equimolar amounts (10 nM) and sequenced on
an Illumina NextSeq500/550 platform using a 75 cycle SE high-output kit aiming at 3
million reads per sample.
Identification of type specimens using target enrichment data
The raw Illumina reads from the five museum specimens were trimmed and quality
checked using Trimmomatic 0.38 (Bolger et al. 2014) to remove Illumina adapters and
low-quality bases. Leading and trailing ends with a quality below 10 were removed, as
well as sliding windows of 2 bp with average quality below 25. Reads shorter than 36 bp
were filtered out. All bases with a quality score below 30 were masked with ‘N’using seqtk
(https://github.com/lh3/seqtk).
For the three targeted mitochondrial genes (16S, cox1 and cob), a library of reference
sequences was built, including references from the different candidate species identified
in the Mantidactylus subgenus using the 16S tree, as well as one sequence from each of
the following species: Mantidactylus curtus, Boophis goudotii and Stumpffia grandis. These
JOURNAL OF NATURAL HISTORY 5

last three sequences were added to make sure that the recovered reads were specifically
matching the sequences from the subgenus Mantidactylus, and not mitochondrial regions
conserved across a larger phylogenetic scale in frogs. The processed Illumina reads were
then aligned to the reference sequences with vsearch 2.3.4 (Rognes et al. 2016) using a
similarity threshold of 98%. The full pipeline was wrapped into a bash script available at
https://github.com/rancilhac/Museoscript. We then used the matching reads to assign the
museum specimens to a species using two approaches (see complementary Figure S1 for
a graphical summary of the pipeline): first, the reads producing unique matches with the
reference sequences (i.e. aligning to only one sequence in the vsearch analysis) were
extracted and counted, while those matching several sequences were ignored as they
would not allow us to discriminate between lineages. Secondly, all the reads that matched
one or more reference sequence were aligned to a reference sequence using CodonCode
Aligner 6.0.2 (CodonCode Corp.), using a majority-based alignment approach. For this, we
first considered that in some of our vsearch analyses, samples matched uniquely to
Mantidactylus sp. Ca56, a lineage known from Betampona on Madagascar’s east coast
only; although this assignment was unlikely, in a conservative approach to avoid refer-
ence bias, we used a sequence of Mantidactylus sp. Ca56 as reference for alignment. As
the obtained consensus sequences did not cluster with this species in preliminary
phylogenetic analyses, we then used reference sequences of the North-Eastern lineage
and Inland Eastern lineage, respectively, for building the final consensus sequence,
obtaining virtually identical consensus sequences as in the initial approach. After aligning
these consensus sequences with the 16S sequences obtained from fresh samples using
MEGA, we inferred a phylogenetic tree and assigned the museum specimens to the
lineages with which they clustered (c.f. details below).
A visual representation of per base coverage of the 16S fragment of archival type
specimens was computed in bedtools v.2.26.0 (Quinlan and Hall 2010). Mean, minimum
and maximum coverage values per nucleotide site were computed in Geneious R6
(https://www.geneious.com; Kearse et al. 2012).
Molecular phylogenetics
We extracted total genomic DNA by standard salt extraction from all available fresh
samples of the subgenus Mantidactylus and amplified fragments of DNA using polymer-
ase chain reactions (PCRs). DNA sequences of the 3ʹfragment of the mitochondrial 16S
rRNA gene (16S) were amplified and sequenced using previously established protocols
described elsewhere with primers 16SAr-L and 16SBr-H (e.g., Vences et al. 2003).
Furthermore, fragments of the nuclear recombination-activating gene 1 (RAG-1) were
obtained with primers Rag1-Manti-F1 (CGTGACAGAGTSAAAGGAGT) and Rag1-Manti-R1
(TCAATGATCTCTGGAACGTG) from Vences et al. (2018) with the following PCR protocol:
120 s at 94°C, followed by 35 cycles of (20 s at 94°C, 50 s at 53°C, 180 s at 72°C) and finally
600 s at 72°C, combined with a nested RAG-1 PCR to obtain additional sections of this
gene, with primers and protocols as in Rakotoarison et al. (2015). Further, one additional
sample (KU 340853 [CRH729]) had 16S data extracted from the raw Illumina reads from a
much larger sequence capture dataset (Hutter et al. 2019).
Sequences were resolved on automated capillary DNA sequencers and error-
checked with CodonCode Aligner (Codon Code Corp.). All newly obtained DNA
6L. RANCILHAC ET AL.

sequences were submitted to GenBank (accession numbers MT187991–MT188029,
MT188687–MT188690, MT193311–MT193398). For the 3ʹ16S fragment, the newly
determined sequences were combined with all homologous sequences from
GenBank for the subgenus Mantidactylus. These sequences were combined with the
consensus sequences obtained from the museum specimens (c.f. details above) and
aligned. The final 16S alignment (514 bp) was used to infer a maximum likelihood tree
using IQTREE (Nguyen et al. 2014) under a K2P+I substitution model. Branch support
was assessed using 1000 ultrafast bootstrap replicates (Hoang et al. 2017). The tree
was rooted using a sequence of Mantidactylus (Brygoomantis) ulcerosus (Boettger 1880)
as the outgroup. Mega X (Kumar et al. 2018) was used to calculate the average p-
distances between the lineages delimitated using the phylogenetic tree. The museum
specimens were not considered for this last analysis.
We inferred alleles (haplotypes) of the nuclear RAG-1 gene using the PHASE algorithm
(Stephens et al. 2001) in DnaSP (version 5.10.3; Librado and Rozas 2009). We then
constructed a ML tree from the phased and unpartitioned RAG-1 sequences using the
Jukes-Cantor substitution model in MEGA X, and entered this tree with the phased
alignment in Haploviewer, written by G. B. Ewing (http://www.cibiv.at/~greg/haplo
viewer), to build a network following the methodological approach of Salzburger et al.
(2011).
Morphometric measurements of preserved specimens
Morphological measurements of type specimens (Table 1) were taken in 2018 and 2019
using a digital calliper to 0.1 mm by MV at MNHN and BMNH. The following measure-
ments were taken: snout-vent length (SVL), maximum head width (HW), head length from
posterior maxillary commissure to snout tip (HL), horizontal eye diameter (ED), horizontal
tympanum diameter (TD), distance from eye to nostril (END), distance from nostril to
snout tip (NSD), distance between nostrils (NND), foot length (FOL), foot length including
tarsus (FOTL), tibia length (TIBL), hindlimb length from cloaca to tip of longest toe (HIL),
forelimb length from axilla to tip of longest finger (FORL), hand length (HAL), length and
width of the femoral gland (FGL, FGW), inner metatarsal tubercle length and height (IMTL,
IMTH), diameter of the third finger disc (F3D) and diameter of the fourth toe disc (T4D).
These features were also measured in 17 specimens attributed to either M. grandidieri or
M. guttulatus at ZSM in 2019 by LR and TB. Webbing formulae follow Blommers-Schlösser
(1979); femoral gland terminology follows Glaw et al. (2000).
Results
Mitochondrial and nuclear sequences reveal several distinct lineages of
Madagascar giant stream frogs
The phylogenetic tree reconstructed from 514 bp of the mitochondrial gene for 16S rRNA
(Figure 1) shows seven highly supported divergent lineages, all corresponding to geo-
graphically distinct groups of populations. Uncorrected 16S pairwise distances between
these lineages (Figure 2) range between 1.9% and 4.9%; hence, in many cases, they differ
by less than 3%, the threshold used by Vieites et al. (2009) for this gene fragment to define
JOURNAL OF NATURAL HISTORY 7

Table 1. Morphometric measurements of specimens of the two nominal species Mantidactylus guttulatus and M. grandidieri, and the newly described species M.
radaka sp. nov. (all in mm). Asterisks indicate name-bearing type specimens: Holotype of M. radaka sp. nov., syntypes of M. grandidieri and M. guttulatus, as well as
the holotype of Rana pigra (here considered as junior synonym of M. guttulatus). Acronyms: M = male, F = female, SA = Sub-adult.
Museum number Field number Sex SVL HW HL TD ED END NSD NND FORL HAL HIL FOTL FOL FGL FGW TIBL IMTL IMTH F3D T4D
M. radaka sp. nov.
* ZSM 644/2001 FGMV 2001.132 M 93.0 42.4 38.8 3.9 11.4 7.5 5.5 8.5 49.3 24.9 140.1 63.4 45.5 18.1 9.1 46.8 7.3 3.0 2.6 2.8
ZSM 180/2010 ZCMV 12567 M 88.3 39.1 35.1 3.3 9.1 6.2 4.9 8.7 45.5 23.4 141.6 61.1 45.2 11.7 4.9 44.6 5.5 2.8 2.0 2.4
ZSM 581/2014 DRV 06072 F 84.9 34.6 32.0 2.7 8.9 6.8 5.1 6.5 45.5 21.8 132.5 57.1 42.1 ––40.1 6.1 2.5 2.3 3.2
ZSM 1800/2010 ZCMV 12345 F 98.1 39.2 37.4 3.8 10.8 6.2 6.6 7.6 47.8 23.9 146.6 64.7 46.7 ––45.6 7.0 2.6 2.5 3.2
ZSM 1802/2010 ZCMV 12596 F 102.8 46.8 40.6 3.5 9.2 7.4 6.7 9.1 51.3 25.3 150.1 66.3 49.2 ––47.0 6.4 3.1 2.2 3.2
ZSM 97/2016 MSZC 80 F 102.3 43.5 41.9 4.4 11.1 7.1 6.0 9.1 52.7 24.6 159.7 70.2 49.3 ––48.9 6.2 3.1 2.9 3.1
M. grandidieri
* MNHN 1895.255 M 88.1 38.9 36.1 4.9 10.9 6.4 7.2 7.7 55.4 24.3 131.4 62.0 43.2 7.5 6.0 –––––
ZSM 5076/2005 ZCMV 2158 M 80.9 33.6 32.6 4.5 10.3 6.9 4.0 7.2 52.1 21.4 124.0 54.3 40.4 7.3 5.6 37.6 5.7 2.5 1.8 2.4
* MNHN 1883.580 F 90.4 37.1 37.6 5.0 10.6 6.5 7.3 8.5 54.0 26.0 151.0 67.1 43.7 4.1 3.7 –––––
ZSM 5077/2005 ZCMV 2159 F 90.7 37.2 33.2 5.0 9.1 8.9 3.8 8.0 60.6 24.6 143.8 64.6 44.8 6.9 4.2 46.4 5.3 2.1 2.0 2.5
ZSM 276/2005 FGZC 2682 F 112.5 47.9 45.2 6.7 14.5 9.7 4.9 10.6 55.9 29.0 152.3 72.1 52.4 6.3 4.8 50.2 7.7 3.4 3.0 4.0
ZSM 262/2005 FGZC 2114 F 116.3 45.0 41.1 6.9 10.8 8.6 5.3 9.9 59.0 28.6 157.0 73.8 54.9 4.0 2.5 50.5 7.5 2.7 1.9 3.1
ZSM 263/2005 FGZC 2119 F 109.8 42.5 42.1 6.9 11.9 8.5 7.4 9.8 61.9 31.1 153.9 70.1 50.9 4.4 3.8 49.0 6.3 2.7 2.1 3.1
ZSM 264/2005 FGZC 2120 F 117.5 46.2 43.5 6.8 11.8 7.2 5.3 9.6 61.6 31.6 158.0 74.8 56.2 4.4 3.0 49.9 8.7 3.4 2.4 3.0
M. guttulatus
ZSM 714/2003 FGMV 2002.362 M 89.8 36.5 35.1 5.3 8.6 7.6 4.5 8.2 –22.1 115.0 58.2 39.3 5.3 3.9 43.0 4.9 2.4 1.7 2.6
ZSM 266/2005 FGZC 2660 M 97.5 42.9 42.7 8.0 13.1 7.5 5.2 8.1 58.1 24.1 136.6 63.4 44.5 7.7 5.3 45.6 6.3 3.5 1.8 2.5
ZSM 1013/2003 FGMV 2002.438 F 109.0 43.0 39.6 6.2 11.0 7.2 5.9 9.5 51.5 27.2 146.8 68.7 48.0 5.5 4.2 47.7 6.4 3.2 2.2 2.9
ZSM 300/2000 SA F 87.0 34.5 33.5 4.5 10.9 6.2 5.1 7.6 47.4 23.8 129.9 56.5 39.1 5.8 3.4 43.9 4.5 2.0 1.2 2.2
ZSM 474/2006 ZCMV 2342 F 115.1 43.5 40.1 6.2 14.7 8.4 6.5 8.5 78.4 25.7 153.9 66.0 47.0 5.9 4.8 49.5 6.3 3.4 2.4 2.9
* BMNH 1947.2.25.48 F 122.3 50.3 50.4 4.3 13.0 7.7 10.3 9.0 64.3 34.7 171.7 79.6 54.4 4.2 3.7 51.1 11.8 2.9 1.6 2.4
* BMNH 1947.2.25.49 F 123.7 50.3 51.6 5.1 14.4 8.0 7.7 10.9 63.1 31.3 166.0 73.0 52.5 4.7 3.0 50.6 6.3 3.4 1.8 2.1
* BMNH 1947.2.25.50 F 125.6 51.1 49.0 4.6 13.5 8.0 7.9 10.5 62.8 33.2 179.1 77.3 54.1 5.0 4.1 52.8 7.9 3.1 2.5 2.7
* BMNH 1947.2.25.51 F 116.5 44.2 46.4 4.6 13.7 8.0 8.0 10.3 59.8 30.3 151.7 69.4 48.2 5.0 5.0 47.7 7.6 3.2 2.1 2.2
* BMNH 1947.2.25.52 SA 95.8 37.6 36.8 4.7 10.4 5.6 6.6 8.0 50.6 23.6 130.6 57.7 40.7 3.4 4.7 41.0 5.9 2.2 1.3 1.5
MNHN 1899.410 (R. pigra) M 118.8 46.8 49.7 6.0 15.0 7.8 8.2 10.5 71.2 33.2 162.5 72.0 51.7 6.9 6.0 –––––
8L. RANCILHAC ET AL.

Figure 1. Maximum likelihood phylogenetic tree obtained from 514 bp of the mitochondrial 16S rRNA
gene. The values at the nodes are the bootstrap supports (not given for intra-lineage nodes for
improved clarity). The type specimens of M. guttulatus and M. grandidieri from the London and Paris
museum collections are highlighted in red and brown, respectively.
JOURNAL OF NATURAL HISTORY 9

candidate species. Yet, the syntopic occurrence of some of these lineages, with apparently
limited admixture in mtDNA and one nuclear gene, RAG-1 (cf. Figures 2 and 3, discussed in
more detail below), suggests that at least some of them may represent valid species. To
allow an easier discussion, we here opt for a consistent definition of all lineages differing
by 1.9% or more as candidate species; yet, we emphasise that this does not imply all of
them merit future recognition as distinct species, and once hybrid zones are studied in
more detail, several of them may be relegated to deep conspecific lineages.
To describe the distribution of the various lineages (species and candidate species), we
here follow the geographical zonation scheme for Madagascar of Boumans et al. (2007). In
the South-East, extending slightly northwards in the Southern Central East, we found
populations belonging to a mitochondrial clade, composed of two lineages that so far
have not received candidate species numbers in the scheme established for Malagasy
frogs by Vieites et al. (2009). Following that numbering system, the last named candidate
species in Mantidactylus is Mantidactylus sp. Ca65 in Cocca et al. (2018), and we therefore
use the next consecutive numbers to refer to the South-East lineages here newly
Figure 2. Diagonal matrix visualising the mean uncorrected genetic distances (p-distances) in the
mitochondrial 16S rRNA gene between the different lineages in the subgenus Mantidactylus, calcu-
lated from 514 bp of the 16S mitochondrial gene.
10 L. RANCILHAC ET AL.

identified in the subgenus Mantidactylus. One of these lineages is restricted to the
extreme South East (locality Manantantely and near Tolagnaro) and is here called
Mantidactylus sp. Ca66; the second lineage appears to have a wider distribution, from
Tsitongambarika in the South East northwards to Ambohitsara in the Southern Central
East, and is here called Mantidactylus sp. Ca67. These are the two least divergent lineages
in 16S (pairwise uncorrected distance of 1.9%) and it is possible that future research will
reveal them to be conspecific. All known localities of these two candidate species are in
the lowlands (<600 m above sea level), relatively close to the east coast.
Two additional geographically restricted lowland lineages are found in the Northern
Central East, Mantidactylus sp. Ca55 and Mantidactylus sp. Ca56. Both of these were found
in sympatry in Betampona, and our analysis also provides evidence for occurrence of the
former lineage further north, at a site named Ambatoroma. These two sympatric candi-
date species differ by 3.4% pairwise 16S distance.
The highlands of the Southern Central East and the Northern Central East, from Ivohibe
to Fierenana, are occupied by one rather widespread lineage that, based on analysis of
type material (see below), corresponds to the nominal species, M. guttulatus. In a small
area of the Northern Central East around Moramanga (localities Fierenana, Mangabe,
Vohidrazana), M. guttulatus co-occurs with another lineage that spreads northwards into
the Northern Central East and westwards into the highlands in Ambohitantely but also
occurs at various coastal lowland sites, coming into sympatry with Mantidactylus sp. Ca55
in Ambatoroma. This lineage has previously been dubbed Mantidactylus sp. Ca57 (e.g.,
Figure 3. Haplotype network of the subgenus Mantidactylus based on 1227 bp of the nuclear RAG-1
gene from 39 samples. Small black dots represent additional mutational steps.
JOURNAL OF NATURAL HISTORY 11

Vieites et al. 2009), but based on analysis of type material (see below), it corresponds to
the nominal species, M. grandidieri. Pairwise distances between the locally sympatric
species M. grandidieri and M. guttulatus, and M. grandidieri and Mantidactylus sp. Ca55,
are 3.0% and 3.4%, respectively.
In the locality Besariaka, M. grandidieri occurs sympatrically with a further species that is
restricted to mid-altitude and highland localities of the North East and Sambirano regions.
This species, previously referred to as M. guttulatus (e.g., Vences et al. 2004; Glaw and
Vences 2007; Schulze et al. 2016) is the genetically most divergent lineage in the
subgenus Mantidactylus (Figure 2), differing by 3.2–4.9% from all other lineages, and by
3.8% from the locally sympatric M. grandidieri.
The haplotype network inferred from 1227 bp of the nuclear RAG-1 gene (Figure 3)
revealed a large number of haplotypes, which, however, were in most cases only con-
nected by single mutational steps. Except for one allele shared between M. guttulatus and
Mantidactylus sp. Ca66 we did not find allele sharing between main lineages, but the
alleles of most lineages did not form well-defined clusters: most individuals of the South
East lineages, Mantidactylus sp. Ca66 and Mantidactylus sp. Ca67, formed a cluster of
seven alleles, separated from other lineages by at least one step; and all sequences of the
North East/Sambirano lineage formed a cluster of three alleles, separated by a minimum
of five steps from all other alleles.
As a conclusion, the subgenus Mantidactylus possibly contains as many as seven
distinct species, corresponding to the mitochondrial lineages. However, the lack of well-
preserved voucher specimens for many of these, and the small number of samples
available for analysis especially from sympatric areas, do not allow for a conclusive and
comprehensive delimitation of all of these. We suspect that such an analysis would also
reveal that some of these lineages hybridise more widely and may constitute deep
conspecific lineages. However, our data do provide good integrative evidence for dis-
tinctness of the North East/Sambirano lineage which has (i) the highest and most
consistent divergence in the mitochondrial (16S) gene (Figure 2), (ii) occurs sympatrically
with M. grandidieri in Besariaka, (iii) has a distinct divergence in the nuclear RAG-1 gene
without allele sharing with any other lineage, and (iv) has consistent morphological
differences in femoral gland size and in some other morphological characters, as dis-
cussed below. We therefore conclude that this lineage corresponds to a distinct species
that we describe herein as Mantidactylus radaka sp. nov.
Target enrichment DNA sequencing allocates historical types to lineages
The existence of more than one species of giant Stream Frog in the subgenus
Mantidactylus has been assumed for many years (e.g., Blommers-Schlösser and Blanc
1991; Glaw and Vences 1994,2007; Vieites et al. 2009) but allocating existing scientific
names (nomina) to the various populations has proven extremely difficult due to an
apparent combination of high morphological variation within and lack of morphological
differentiation among several of the lineages. In the previous section, for ease of reading,
we have already assigned these names to lineages, anticipating our results reported in the
following, based on DNA sequences obtained from historical type material.
As summarised in Blommers-Schlösser and Blanc (1991) and Frost (2019), three species-
level nomina exist for the subgenus Mantidactylus: (i) Rana guttulata Boulenger, 1881,
12 L. RANCILHAC ET AL.

originally based on five syntypes: BMNH 1947.2.25.48–52 (formerly 1880.7.15.10–12 and
1881.1.3.29.15–16) from ‘Betsileo’(later changed to ‘S. E. Betsileo’by Boulenger 1882), of
which BMNH 1947.2.25.51 was designated lectotype by Blommers-Schlösser and Blanc
(1991); (ii) Mantidactylus grandidieri Mocquard, 1895, based on two syntypes MNHN
1883.580 and 1895.255 from ‘Madagascar . . . côte Est’; and (iii) Rana pigra,1900, based
on holotype MNHN 1899.410, from ‘forêt d’Ikongo’.
Our sampling included all name-bearing type specimens of these nomina plus one
additional paralectotype of Rana guttulata. Using target enrichment, we successfully
recovered DNA from these five museum specimens, the total number of Illumina reads
ranging from 677,442 to 1,658,433. After processing, the number of reads ranged from
159,890 to 928,481, representing 24–56% of the raw reads. The length of the processed
reads ranged from 36 to 76 bp (mean 49.4 bp). Of the five samples, four had several
thousand reads mapping to our reference sequences with a similarity threshold of 98%,
while the type specimen of Rana pigra only yielded a few reads. This last sample was
therefore removed from most downstream analyses. The details for each sample are given
in Table 2.
The patterns of unique matches to the reference sequences varied depending on the
sample considered (Figure 4). For all samples, the number of reads matching the cox1
gene in the target species was notably lower than for the other two genes, and matching
reads for cob were overall less frequent than those for 16S. For the BMNH samples, the
reads consistently matched best to the highland lineage from the Southern and Northern
Central East in all three studied mitochondrial genes (16S, cob, cox1). For the MNHN
samples, the results are less clear. In both samples, more than 3000 reads matched the M.
radaka sp. nov. reference sequence vs, respectively, 1300 and 600 matching the M.
grandidieri sequence. Very few reads matched to the cox1 reference sequences, and
only to M. grandidieri. Finally, some reads matched to the cob sequences of both Ca56
and M. guttulatus.
Table 2. Summary of the number of reads recovered in the different museum samples, and the 16S
consensus sequences obtained from them. Coverage (per nucleotide of the consensus sequence) is
given as mean ± SD, followed by range in parentheses. The sixth column gives the complete length of
the assembled fragment for each sample, followed by the number of nucleotides that could not be
reliably reconstructed (region covered either by no reads or by low quality reads).
Catalogue number
Number of
raw reads
Number of pro-
cessed reads
16S fragment –reads
matching reference
16S fragment –
length/missing
16S frag-
ment –
coverage
M. guttulatus
BMNH 1947.2.25.51
(lectotype)
1,658,433 928,481 (56%) 42,228 457 bp/2 bp 4537 ± 4923
(1–18,535)
BMNH 1947.2.25.48
(paralectotype)
1,066,500 516,158 (48%) 80,448 431 bp/0 bp 8572 ± 9672
(5–38,086)
MNHN 1899.410
(holotype of Rana
pigra)
677,442 159,890 (24%) NA NA NA
M. grandidieri
MNHN 1883.580
(syntype)
1,437,053 554,256 (39%) 66,902 451 bp/8 bp 6549 ± 9372
(1–35,403)
MNHN 1895.255
(syntype)
1,121,605 493,613 (44%) 76,569 414 bp/19 bp 8297 ± 11,825
(58–45,765)
JOURNAL OF NATURAL HISTORY 13

From these reads, we assembled consensus sequences, which for 16S had an average
coverage of 4537 to 8572 (the details for each sample are given in Table 2 and Figure 5).
The consensus sequences obtained by aligning reads to different reference sequences
were almost identical, except for stretches where the reference sequences differed by
indels, and in exploratory phylogenetic analyses clustered in the same clades in the tree.
We therefore used our preferred consensus sequences (obtained by alignment to those
reference sequences closest to their preliminary phylogenetic clustering) for final analysis.
These sequences were aligned to the main 16S alignment and could be successfully
placed in the phylogenetic tree (Figure 1). Additional alignments of the consensus
sequences for cob, and, where available, cox1 sequences of the historical types, with
single representatives of the various Mantidactylus lineages, yielded results fully concor-
dant with those of the 16S gene (Supplementary Figures S2 and S3).
We found that the lectotype and paralectotype of Rana guttulata from the London
museum clustered with the lineage occurring predominantly in central eastern
Madagascar to which we therefore apply the name M. guttulatus herein (Figure 1), as
anticipated in the previous section. The two analysed syntype specimens of M. grandidieri
from the Paris museum instead clustered with the lineage from the Northern Central East
Figure 4. Stacked barplots showing the number of reads uniquely matching different reference
sequences for the three targeted mitochondrial genes with a similarity threshold of 98%. The Rana
pigra type was not included because the number of reads was too low.
14 L. RANCILHAC ET AL.

and North East that previously had been named Mantidactylus sp. Ca57, which is herein
renamed accordingly as M. grandidieri.
The holotype of the third nomen Rana pigra (MNHN 1899.410) yielded no reads matching
Mantidactylus 16S or cox1 sequences. However, a total of 16 reads of this sample matched
the terminal stretch of the target cob sequence (Supplementary Figure S3). In their majority,
they agreed with a sequence of the lineage assigned to M. guttulatus.Thisassignment
would also agree with the type locality (Ikongo forest), which is located roughly between
Ranomafana and Ivohibe, within the distribution area of this lineage. However, given the
very low number of reads, this result clearly remains in need of confirmation.
Lineages of Madagascar giant stream frogs show weak morphological
differentiation
The morphometric comparison between lineages in the subgenus Mantidactylus is cur-
rently hampered by a lack of genetically identified adult voucher specimens, and no
satisfying quantitative analysis was therefore possible. Morphometric data for all mea-
sured individuals in Table 1, and photographs in Figures 6–9revealed only little morpho-
logical differentiation among lineages. Previous studies proposed to distinguish M.
grandidieri and M. guttulatus by a more granular dorsal skin texture, wider terminal
Figure 5. Per-base coverage plots for the 16S fragment in four Mantidactylus type specimens from the
MNHN and BMNH collections. (a) BMNH 1947.2.25.48 (paralectotype of Rana guttulata); (b) BMNH
1947.2.25.51 (paralectotype of Rana guttulata); (c) MNHN 1895.255 (syntype of M. grandidieri); (d)
MNHN 1883.520 (syntype of M. grandidieri).
JOURNAL OF NATURAL HISTORY 15

discs of fingers and toes, and wider head of the latter (e.g., Blommers-Schlösser and Blanc
1991), but based on our measurements, examination of available vouchers, and photo-
graphs, these differences do not seem to be consistent. In fact, to judge from photographs
(Figures 6 and 7) and the type specimens, and contrary to previous assumptions, most
specimens of M. guttulatus appear to have a rather smooth dorsal skin, whereas some
individuals of M. grandidieri are rather wide-headed and moderately granular on the back.
Based on comparison of one female (ZSM 1013/2003) and one male (ZSM 266/2005)
whose sex we confirmed by dissection, the male’s tympanum appeared to be larger (ratio
TD/SVL 0.082 vs 0.057), and its femoral glands were roughly of equal size to those of the
female, but more distinct and prominent.
Figure 6. Photographs of living specimens of Mantidactylus radaka sp. nov. (a, b) Male holotype ZSM
644/2001 (field number FGMV 2001.132) from Manarikoba forest, Tsaratanana Massif. (c–f) Female
paratype ZSM 1800/2010 (ZCMV 12345) from Camp 1 (Antevialambazaha), Tsaratanana Massif. (g, h)
Female paratype ZSM 97/2016 (MSZC 0080) from Ampotsidy. (i, j) Male paratype MSZC 0120
(uncatalogued in UADBA) from Ampotsidy. (k) Unidentified specimen from Camp 0 (Ankijagna
Lagnana), Tsaratanana Massif. (l) Paratype ZSM 582/2014 (DRV 6073) from Camp 0 (Ankijagna
Lagnana). (m, n) Unidentified female specimen from Manongarivo (Camp 0), probably preserved in
UADBA collection.
16 L. RANCILHAC ET AL.

Figure 7. Photographs of living specimens of Mantidactylus (Mantidactylus) guttulatus, M. (M.) grand-
idieri, and of three candidate species. (a, b) M. (M.) guttulatus, female ZSM 1013/2003 (FGMV 2002.438)
from Ranomafana. (c) Unidentified specimen from Ranomafana, assigned tentatively to M. (M.)
guttulatus (no genetic evidence). (d, e) M. (M.) guttulatus, specimen KU 340853 (CRH729) from
Ranomafana. (f) M. (M.) grandidieri, specimen ZSM 5077/2005 (ZCMV 2159) from Nosy Mangabe. (g)
M. (M.) grandidieri, specimen ZSM 276/2005 (FGZC 2682) from Vohidrazana. (h) M. (M.) grandidieri,
unidentified specimen (probably subadult) from Andranofotsy. (i, j) M. (M.) grandidieri, specimen KU
JOURNAL OF NATURAL HISTORY 17

The available photographs (Figure 7) indicate that the candidate species Mantidactylus
sp. Ca66 and Ca67 from the South East appear to have a rather narrow head and an
Figure 8. Lateral views of the heads of preserved adult males of Mantidactylus (Mantidactylus) radaka
sp. nov. in comparison with M. (M.) guttulatus and M. (M.) grandidieri. Note the more distinct and larger
tympanum (indicated by yellow arrows) in the latter two species. Not to scale.
340659 (CRH307) from Vohidrazana. (k) M. (M.) sp. Ca66, unidentified specimen from Manantantely
(belonging to the series included in molecular analysis, voucher probably uncatalogued in UADBA
collection). (l) M. (M.) sp. Ca66, juvenile from Nahampoana, tentatively assigned to this lineage (no
genetic evidence). (m) M. (M.) sp. Ca67, unidentified specimen from Vevembe forest. (n) M. (M.) sp.
Ca67, FAZC 15,332 from Tsitongambarika. (o) M. (M.) sp. Ca55, specimen FAZC 13,944 from
Betampona. (p, q) M. (M.) sp. Ca55, unidentified specimen from Betampona. (r) M. (M.) sp. Ca56,
specimen FAZC 13,579 from Betampona. Pictures: M, from J. E. Randrianirina; O-R, from G. M. Rosa.
18 L. RANCILHAC ET AL.

especially smooth dorsal skin, which may constitute a diagnostic character for future
taxonomic revision of these lineages.
The most obvious morphological difference encountered was the shape and size of the
femoral glands in the lineage from the Sambirano and North East regions, which is
described as a new species in the following: in this lineage, the males have large and
ovoid femoral glands, while they are smaller and rounded in males of M. guttulatus and M.
grandidieri. On the other hand, the femoral glands in the females of the new species are
indistinct (only tiny rudiments recognisable upon dissection), while always recognisable
Figure 9. Preserved type specimens of the four nomina in the Mantidactylus subgenus Mantidactylus
and one of the paralectotypes of Rana guttulata.
JOURNAL OF NATURAL HISTORY 19

externally in the two other species. Furthermore, the tympanum of the new species
appears to be smaller than in most specimens of M. grandidieri and M. guttulatus, and
often hidden. As an exception, the measured tympanum diameter of the lectotype and
paralectotypes of M. guttulatus was also rather low, which however may be a preservation
artefact, or due to measuring differences in different years of examination (2018 vs 2019).
Systematics
Mantidactylus (Mantidactylus) guttulatus (Boulenger, 1881)
Rana guttulata Boulenger, 1881
Lectotype. BMNH 1947.2.25.51, designated lectotype by Blommers-Schlösser and Blanc
(1991), from the region of Betsileo (S.E. Betsileo), collected by Bartlett.
Paralectotypes. Four specimens, BMNH 1947.2.25.48–50, BMNH 1947.2.25.52, with same
collection locality and data as lectotype.
Junior synonym. Rana pigra Mocquard, 1900. Holotype: MNHN 1899.410, from ‘forêt
d’Ikongo’.
Referred material. For field numbers of additional specimens referred to M. guttulat us
genetically, see Figure 1. For morphological measurements of types and five additional
specimens in the ZSM collection, see Table 1.
Remarks. Mantidactylus (M.) guttulatus is a large nocturnal stream-dwelling frog, distrib-
uted at elevations from 810 m a.s.l. (Vohidrazana) to ca. 1500 m a.s.l. (Antoetra). It is typically
found in slow-moving parts of small streams in rainforest, and almost, nothing is known
about its natural history. Based on genetic data herein, confirmed localities are (from north
to south) Fierenana, Andasibe, Maromizaha, Mangabe region, An’Ala and Vohidrazana in
the Northern Central East, and Antoetra, Vohiparara, Ranomafana and Ivohibe in the
Southern Central East of Madagascar (map in Figure 1). If Rana pigra is correctly assigned
as a junior synonym to M. guttulatus, then a further locality would be Ikongo Forest.
The definition of this species has had a very convoluted history, and many populations
and specimens have intermittently been named M. guttulatus. A complete revision of all
these uses in the literature is beyond the scope of this paper. Glaw and Vences (2007)defined
populations from Tsaratanana as M. guttulatus, and calls and tadpoles of the lineage occur-
ring at Tsaratanana were also described under this name (Vences et al. 2004; Schulze et al.
2016). However, this population corresponds to the new species M. radaka sp. nov. described
below. The tadpole described by Altig and McDiarmid (2006)asM. guttulatus actually belongs
to M. majori, a representative of the subgenus Hylobatrachus (Randrianiaina et al. 2011).
20 L. RANCILHAC ET AL.

Mantidactylus (Mantidactylus) grandidieri Mocquard, 1895
Syntypes. Two specimens, MNHN 1883.580 and MNHN 1895.255 collected by Humblot
and Grandidier, from ‘Madagascar . . . côte Est’.
Referred material. For field numbers of additional specimens referred to M. grandidieri
genetically, see Figure 1. For morphological measurements of types and additional six
specimens in the ZSM collection, see Table 1.
Remarks. Mantidactylus (M.) grandidieri is a large nocturnal stream-dwelling frog, dis-
tributed at elevations from near sea level (Nosy Mangabe) to ca. 1500 m a.s.l.
(Ambohitantely), but appears to be more common at low elevations <900 m a.s.l. It is
found in slow-moving parts of streams in rainforest, often among rocks and boulders, and
almost nothing is known about its natural history. Based on genetic data herein, con-
firmed localities are (from north to south) Marojejy, Sambava, Besariaka, Tsararano,
Antalaha, Masoala, Ilampy, Ambodivoangy, Andranofotsy, Nosy Mangabe, Angozongahy
(west slope of Makira Reserve), Antsahataloka, Ambatoroma, Ambohitantely,
Ambatodisakoana, Fierenana, Vohidrazana, Moramanga and the Mangabe region.
As with M. guttulatus, the definition of this species has changed multiple times in the
past. A complete revision of all these uses in the literature is beyond the scope of this
paper. Glaw and Vences (2007) partially following Blommers-Schlösser and Blanc (1991)
used the name M. grandidieri primarily to refer to highland populations in the Northern
Central East and Southern Central East, which, according to the present revision, are to be
referred to as M. guttulatus. Instead, populations of M. grandidieri were named
Mantidactylus sp. aff.grandidieri ‘North’by Glaw and Vences (2007), Mantidactylus sp. 57
by Vieites et al. (2009) and Mantidactylus sp. Ca57 by Perl et al. (2014).
Mantidactylus (Mantidactylus) radaka new species
Holotype. ZSM 644/2001 (field number FGMV 2001.132), adult male, collected on the
Tsaratanana massif, Manarikoba forest, Antsahamanara, ‘Camp I’(14.04500°S, 48.78528°E,
ca. 1000 m above sea level), Diana Region, former Antsiranana province, northern
Madagascar on 4–9 February 2001 by F. Andreone, F. Mattioli, J. Randrianirina & M. Vences.
Paratypes. ZSM 1800/2010 (field number ZCMV 12,345, female), collected on 10 June 2010
on the Tsaratanana massif, Camp 1 (14.17413°S, 48.94521°E, 1589 m above sea level), Sofia
Region, former Mahajanga province, northwestern Madagascar, by M. Vences, D. Vieites, R. D.
Randrianiaina,F.Ratsoavina,S.Rasamison,A.Rakotoarison, F. Randrianasolo, E. Rajeriarison &
T. Rajoafiarison. ZSM 1801/2010 (field number ZCMV 12,567, male), collected on 29 June 2010
at Bemanevika river (14.48251°S, 48.62723°E, 1109 m above sea level), Sofia Region, former
Mahajanga province, Madagascar, by M. Vences, D. Vieites, R. D. Randrianiaina, F. Ratsoavina, S.
Rasamison, A. Rakotoarison, F. Randrianasolo, E. Rajeriarison & T. Rajoafiarison. ZSM 1802/2010
(field number ZCMV 12,596, female), collected on 28 June 2010 at Bemanevika, Camp 2
JOURNAL OF NATURAL HISTORY 21

(14.35991°S, 48.59022°E, 1538 m above sea level), Sofia Region, former Mahajanga province,
Madagascar, by M. Vences, D. Vieites, R. D. Randrianiaina,F.Ratsoavina,S.Rasamison,A.
Rakotoarison, F. Randrianasolo, E. Rajeriarison & T. Rajoafiarison. ZSM 581/2014 (field number
DRV 6072, female), ZSM 582/2014 (field number DRV 6073, male), ZSM 583/2014 (DRV 6074,
male) and ZSM 584/2014 (field number DRV 6097, male), collected on 9 June 2010 at
Tsaratanana massif, camp 0 (Ankijagna Lagnana, 14.23989°S, 48.97208°E, 1162 m above sea
level), Sofia Region, former Mahajanga province, Madagascar,byD.Vieites,M.Vences,R.D.
Randrianiaina, S. Rasamison, A. Rakotoarison, F. Randrianasolo, E. Rajeriarison & T.
Rajoafiarison. ZSM 95/2016 (field number MSZC 0189), collected on 13 January 2016 at
Ampandrana forest (26.6 km SW of Bealanana on the RN31, 14.71788°S, 48.57548°E, 1038 m
above sea level), Sofia Region, former Mahajanga province, Madagascar, by M. D. Scherz & M.
Rakotondratsima. ZSM 96/2016 (field number MSZC 0205), collected on 14 January 2016 at
Andranonafindra forest (30 km SW of Bealanana on the RN31, 14.73654°S, 48.54846°E, 1204 m
above sea level), Sofia Region, former Mahajanga province, Madagascar, by M. D. Scherz & M.
Rakotondratsima. ZSM 97/2016 (field number MSZC 0080, female), collected on 22 December
2015 on Ampotsidy mountains, 15.7 km NNW of Bealanana (8.7 km NNW of Beandrarezona,
14.41456°S, 48.71148°E, 1400 m above sea level), Sofia Region, former Mahajanga province,
Madagascar, by M. D. Scherz, J. Borrell, L. Ball, T. Starnes, E. Razafimandimby,D.H.
Nomenjanahary & J. Rabearivony.
Diagnosis
The new species is assigned to the genus Mantidactylus based on the presence of an
intercalary element between terminal and subterminal phalanges of fingers and toes
(verified by external observation only), of a central depression in femoral glands and of a
(very) rudimentary femoral gland in the female. Within Mantidactylus, it is assigned to the
nominal subgenus Mantidactylus by combination of (1) very large body size (male SVL 88–
93 mm, female SVL 85–103 mm), (2) absence of dorsolateral colour border, (3) absence of a
distinct frenal stripe, (4) absence of large yellowish patches or stripes in the inguinal region
or between colouration of flanks and belly, (5) fully webbed feet and (6) riparian habits,
living very close to or in streams. The assignment of the species to this group is also
supported by its molecular phylogenetic relationships.
From the two described species in the subgenus Mantidactylus, the new species is
distinguished as follows:
From M. guttulatus by a longer relative hindlimb length (HIL/SVL 1.46–1.60 versus 1.28–
1.43); larger femoral glands in males (FGL/SVL 0.13–0.20 versus 0.06–0.08), which are ovoid
(versus round in M. guttulatus), and indistinct femoral glands in females (versus always
recognisable in M. guttulatus); and possibly by an often-hidden tympanum and smaller
relative tympanum size (TD/SVL 0.032–0.047 versus 0.052–0.082) [however not applying
to the M. guttulatus type specimens (0.035–0.049), possibly due to preservation artefacts].
From M. grandidieri by an often-hidden tympanum and smaller relative tympanum size
(TD/SVL 0.032–0.047 versus 0.055–0.063); larger femoral glands in males (FGL/SVL 0.13–
0.20 versus 0.09); which are ovoid (versus round in M. grandidieri); and indistinct femoral
glands in females (versus always recognisable in M. grandidieri).
Description of the holotype. Adult male in excellent state of preservation. For measure-
ments, see Table 1. Body rather stout. Head wider than long, slightly wider than body.
22 L. RANCILHAC ET AL.

Snout blunt. Nostrils directed dorsolaterally, slightly protuberant, nearer to tip of snout
than to eye. Canthus rostralis poorly distinct, loreal region concave. Upper part of the
tympanum hidden under the tympanic fold. Tympanum small, horizonal diameter of
tympanum 34% of horizontal eye diameter. Supratympanic fold distinct, straight above
the tympanum, curving gently towards the axilla. Tongue ovoid, distinctly bifid poster-
iorly. Vomerine teeth form two rounded aggregations, positioned posterolateral to choa-
nae. Choanae ovoid. Subarticular tubercles single. Outer metacarpal tubercle not
recognisable, inner metacarpal tubercle present and distinct. Fingers without webbing.
Relative length of fingers: I<II<IV<III. Finger discs enlarged. Nuptial pads absent. Foot
slightly shorter than tibia (97%). Lateral metatarsalia separated. Inner metatarsal tubercle
present. Outer metatarsal tubercle not recognisable. Webbing formula: 1(0), 2i(0.5), 2e(0),
3i(1), 3e(0), 4i(1), 4e(0.5), 5(0). Relative length of toes: I<II<III=V<IV. Skin on the upper
surface strongly rugose with numerous small warts, these becoming larger on the flanks.
Ventral side smooth. Femoral glands very large and distinct in external view.
Colour in preservative (after 18 years in preservative) dorsally brown, with tiny yellow-
ish dots on the warts. Venter beige, with indistinct mottling on the throat. The inner three
fingers are lighter than the fourth one. Colouration in life (Figure 6) similar to that in
preservative, with larger yellowish dots particularly on the flanks. Uniformly greenish-
brown iris. Light tips on the inner three fingers.
Etymology. The species name is a noun in apposition to the genus name, and is the
Malagasy word for large frogs in general (as opposed to ‘sahona’, which usually refers to
small frogs), and those of the subgenus Mantidactylus in particular.
Variation. No obvious difference in tympanum diameter between males and females
(TD/SVL female 0.032–0.043, male 0.037–0.047). The paratypes and additional photo-
graphed specimens in general agree with the holotype in morphology and colouration.
In preserved females, femoral glands are not recognisable externally. In life, their femoral
glands are visible as tiny black patches (cf. photos; Figure 6F, H). Some females (ZSM 1800/
2010 in particular) have very dark throat colouration. In a young male (ZSM 582/2014),
femoral pores are already distinct (both externally and internally) and ovoid.
Natural History. A large nocturnal frog typically found in relatively calm stretches of
streams, including small and shallow streams with sandy bottom in rainforest (e.g.,
Tsaratanana), and more rarely also in shallow stretches of wider streams with rocky bottom
in more open areas at the edge of rainforest (e.g., near Bemanevika). In Besariaka, individuals
of this species also showed an apparently distinct natural history and behaviour, preferring
running water (usually moving in the central parts ofstreams), while M. grandidieri appeared
associated more closely with stream banks. Mantidactylus radaka sp. nov. is the only species
of the subgenus for which some natural history information is available: calls and tadpoles
were described, from Tsaratanana and from a site between Bealanana and Antsohihy (close
to the collection locality of ZSM 96/2016), respectively, by Vences et al. (2004) and Schulze et
al. (2016), under the name M. guttulatus.
Distribution. Localities span an elevational range from 690 m a.s.l. (Manongarivo Special
Reserve, Camp 0) up to at least 1590 m a.s.l. (Tsaratanana Massif, Antevialambazaha/Camp 1).
JOURNAL OF NATURAL HISTORY 23

Based on genetic data herein, confirmed localities are (from north to south) the Tsaratanana
Massif (localities Manarikoba forest and Antevialambazaha/Camp 1), Manongarivo,
Ampotsidy, Bemanevika, several sites between Bealanana and Antsohihy (e.g. Ampandrana
and Andranonafindra), and Besariaka.
Conservation. During fieldwork in 2016, this species was found to be rather abundant,
even where the forest was heavily degraded, as long as some riparian habitat remained.
However, it is also actively hunted as a source of food by local people in the Bealanana
District (Figure 10). When asked, two locals stated that they no longer hunted these frogs
in their home village (Ampandrana, ca. 14.704°S, 48.589°E), because they are no longer
found in sufficient abundance in its vicinity. We managed to collect specimens from the
river that flows towards Ampandrana only by going upstream (14.71788°S, 48.57548°E),
and these frogs were considerably more abundant 5 km further SSW along the Route
National 31 (at Andranonafindra), where there are no further villages for another 15 km.
Thus, it seems evident that anthropogenic pressures are having impacts on local popula-
tions. However, the extent of those impacts is currently unclear, and given the apparently
very large clutches that these frogs have (see the egg-laden ovaries evident in Figure 10C)
Figure 10. Mantidactylus (Mantidactylus) radaka sp. nov. being prepared for human consumption. (a)
Frogs and crabs are collected from broad streams. Then (b) the frogs are gutted and skinned, and the
head, hands and feet removed. The frog is then rinsed in the stream, leaving (c) cleaned animals for
cooking in a stew. Note the ovaries full with hundreds of eggs.
24 L. RANCILHAC ET AL.

and their stream-dwelling tadpoles, it is possible that up-stream populations may seed
down-stream ones, and the species may thus be rather resilient to exploitation and able
to recover quickly if pressures cease.
At present, the distribution of this species appears to be rather large: a minimum
convex polygon of known localities has an area of 12,534 km
2
. Over that area, there is no
doubt that the habitat of the species is declining in extent and quality, and, although it is
found in numerous locations, its habitat is also severely fragmented. Therefore, this
species qualifies as Vulnerable under the International Union for Conservation of Nature
(IUCN) Red List of Threatened Species, criterion B1ab(iii) (IUCN 2012).
Candidate species Mantidactylus sp. Ca55, Ca56, Ca66, Ca67
Besides the three species now recognised, our analysis has identified four additional
lineages in the subgenus Mantidactylus (Figure 1), for which we have herein used the
candidate species names Mantidactylus sp. Ca55, Ca56, Ca66 and Ca67. At present, data
are clearly insufficient to take taxonomic decisions regarding these lineages.
Mantidactylus sp. Ca55 and Ca56 occur sympatrically in Betampona, but no specimens
of either candidate species were available for morphological examination in this study.
Both of these candidate species are crepuscular-nocturnal. In Betampona, Mantidactylus
sp. Ca56 seems to be both more abundant and more widespread, but precise data on its
fine-scale distribution within this reserve are thus far lacking. Mantidactylus sp. Ca56
inhabits all large and fast-flowing streams across the reserve, while Mantidactylus sp.
Ca55 seems to prefer lateral branches of the fast-flowing streams, where water is shal-
lower and runs much more slowly. Photos of genetically identified individuals of the two
candidate species (Figure 7) do not immediately reveal obvious morphological differences
between them. They do not appear to be sister species based on the mitochondrial data
(Figure 1), and in the nuclear RAG-1 gene, the two examined samples do not share alleles.
If confirmed with a larger number of samples, such a concordance between mitochondrial
and nuclear gene signals in sympatry would clearly be indicative of a status as separate
species (Padial et al. 2010).
In contrast, Mantidactylus sp. Ca66 and Ca67 appear to be allopatric sister lineages
distributed in the South East. Their mitochondrial divergence is low, and they may be
conspecific. However, their distinctness from other species and candidate species of the
subgenus Mantidactylus is supported by exclusive RAG-1 alleles of numerous individuals
sequenced, and by their apparently very smooth skin. Thus, they are likely to represent at
least one new and as yet undescribed species. In the forest at Manantantely, Mantidactylus
sp. Ca66 was nocturnal and living in rocky streams in rather pristine low-elevation
rainforest.
Discussion
Massively-Parallel-Sequencing-based method successfully allowed the molecular
barcoding of old museum specimens preserved in wet collections
Using a Massively Parallel Sequencing (MPS) approach, we attempted amplification and
sequencing of mitochondrial DNA from five historical frog specimens, all of them pre-
served for over 120 years in wet collections, with an unknown history of preservatives.
JOURNAL OF NATURAL HISTORY 25

After filtering, we ended up with hundreds of thousands of short reads for each specimen.
In four of these samples, several thousand of these reads could be mapped to reference
sequences of frogs from the subgenus Mantidactylus. However, for the fifth sample
(MNHN 1899.410, the holotype of Rana pigra), only a few reads could be identified as
Mantidactylus DNA, and thus we did not consider these data for further analyses. Why this
particular specimen did not work remains unclear. One possible explanation could be that
the specimen of R. pigra was the only one to be fixed with formaldehyde, since this
fixative was not yet in general use when the four other samples were collected. However,
formaldehyde fixation alone does not necessarily inhibit DNA isolation and subsequent
capture with adequate baits –instead this preservation method mainly leads to severe
fragmentation of DNA (Gilbert et al. 2007). Given the uncertainty regarding the fixation
and preservation history of the Rana pigra sample, further analyses will be required to
understand better why DNA sequencing failed.
We then used these reads to assign the four remaining specimens to a specific lineage
using two approaches. Firstly, we quantified the number of reads that had unique
matches to each sequence of a reference database representing the different lineages
of the Mantidactylus subgenera, as well as three other frog species as negative controls.
The underlying hypothesis is that the reference sequence with the most unique matches
should correspond to the lineage to which the specimen belongs (this method was also
used by Yuan et al. 2019). While this method worked well for the two M. guttulatus types,
clearly indicating their belonging to the northeastern lineage, the results for the M.
grandidieri types are less decisive. Indeed, the counts do not allow us to clearly distinguish
between the two northern lineages, and the total number of reads matching to the
reference sequences was around 10 times lower than in the M. guttulatus samples.
Secondly, for each of the four samples, we generated consensus sequences of the 16S,
cox1 and cob genes. These sequences were added to our phylogenetic tree, and the
specimens attributed to the lineage in which they clustered. Doing so, we could con-
fidently identify the type specimens, in concordance with the type localities when avail-
able. The two type specimens of M. guttulatus clustered within the inland eastern lineage,
while the type specimens of M. grandidieri were included in the northeastern group. This
result was consistently supported by the three targeted mitochondrial genes.
Although these two approaches yielded overall concordant results, interpreting them
was not equally straightforward in both cases. Indeed, the ‘phylogeny-based’approach
proved rather easy to interpret, as long as the consensus sequence obtained from the
reads was complete enough to ensure phylogenetic resolution. On the other hand, the
approach based on unique matches proved to be sensitive to the reference sequences
used. Using two very close references will reduce the number of unique matches
observed, since most reads will match both of them. The presence of ambiguous or
missing positions in the reference also proved to be a problem, by preventing the reads to
align to the reference when using a conservative similarity threshold. Finally, the number
of reads alone is not particularly valuable information, since a large number of reads may
cover only a small part of the sequence. Considering these observations, the ‘phylogeny-
based’approach seems to be the safest and most straightforward approach to assign a
specimen to a lineage. When using this approach however, aligning the reads to a set of
reference sequences including more distant species as negative controls is a good way to
ensure the absence of artefacts at the sequencing step, such as cross-contaminations.
26 L. RANCILHAC ET AL.

Indeed, the baits we used are highly conserved, particularly for the 16S gene, and the risk
of capturing DNA from other organisms (e.g. Homo sapiens, or other historical specimens
that might have been conserved in the same jar) is substantial. In addition, such analysis
can also provide interesting insights on the origin of the reads. Thus, these two
approaches act complementarily, and in the present case allowed us to confidently
identify our samples.
In line with previous studies (Hykin et al. 2015; Li et al. 2016; Ruane and Austin 2017;
McGuire et al. 2018; Evans et al. 2019), we further demonstrate that MPS methods can be
successfully applied to historical specimens preserved in wet collections. As exemplified in
the present study, this approach can be very useful to solve complex taxonomic conun-
drums, particularly when morphologically cryptic lineages are involved. However, it could
be applied to other biological questions, such as resolving the phylogenetic placement of
rare or extinct taxa, or studying the dynamics of parasites through specimens of the host
species collected at different times, among others (reviewed in Burrell et al. 2015).
The subgenus Mantidactylus shows deep phylogenetic structure
As in other groups of Malagasy frogs, the study of genetic variation within the subgenus
Mantidactylus confirmed that the current taxonomy underestimates species richness. Using
mitochondrial sequences, we were able to delimit seven well-supported candidate species,
all of which are geographically separated, with little to no sympatry. In the present study, we
take a first step to revise the taxonomy of this group by determining to which lineages the
currently described names should be attributed, and describing a third lineage under a new
name. As discussed above, the study of molecular data from museum samples successfully
allowed us to attribute the type specimens of M. (M.) guttulatus and M. (M.) grandidieri.
Therefore, the former name should be used for the lineage distributed in inland localities of
Eastern Madagascar, from Ivohibe in the South to Fierenana in the North. On the other hand,
the name M. (M.) grandidieri should be applied to the populations from the northeast of
Madagascar, between Mangabe in the South and Marojejy in the North. Unfortunately, the
type specimen of Rana pigra, although here treated as junior synonym of M. (M.) guttulatus,
could not be identified with full certainty, and we cannot exclude that it may belong to the
Mantidactylus sp. Ca66/Ca67 clade at this time.
While the status of the four lineages distributed along the eastern coast is difficult to
assess due to our sparse sampling and few loci considered, the northwestern lineage
stands out by its high divergence from all the other lineages. The monophyly of this group
is fully supported on the 16S phylogenetic tree, and the p-distance matrix shows that it is
the most distinct lineage of the group. In addition, it does not share any RAG-1 haplotypes
with the other lineages. For all these reasons, as well as its geographic separation from the
other lineages, we have described it as a new species: Mantidactylus (M.) radaka. While the
molecular analysis of museum samples alone could not rule out that the name Rana pigra
refers to that lineage, we are confident that this is very unlikely. Indeed, the type locality
for R. pigra (‘forêt d’Ikongo’) is within the range of M. (M.) guttulatus whereas the range of
M. (M.) radaka to our knowledge had not been visited by zoologists at the time of the
description of R. pigra (absence of historical amphibian records from this area: see
Blommers-Schlösser and Blanc 1991; Glaw and Vences 1994). However, further studies
of the R. pigra type specimen will be needed to definitively confirm its synonymy with M.
JOURNAL OF NATURAL HISTORY 27

(M.) guttulatus. Despite their molecular distinctiveness, the lineages of Mantidactylus show
little morphological divergence. However, the new species tends to have a smaller
tympanum, as well as very large and ovoid femoral glands.
The combination of MPS-based molecular study of type specimens with traditional
phylogenetic and morphologic analyses proved to be an efficient and effective approach
to resolve the taxonomy of the Mantidactylus subgenus. Even if our pipeline remains to be
improved in many aspects, it could provide a standardised way to assess taxonomic
questions within complexes of cryptic species. Further development of our approach
could be the integration of coalescent-based species delimitation and discovery models
applied on multi-locus data.
Acknowledgements
We are grateful to numerous friends and colleagues for their help during fieldwork over the past 25
years, in particular to F. Mattioli, R. D. Randrianiaina, J. E. Randrianirina, F. M. Ratsoavina, G. M. Rosa
and D. R. Vieites. G. Keunecke, M. Kondermann and K. Warmuth assisted with laboratory procedures.
Fieldwork was carried out in the framework of collaboration accords of the authors’institutions with
the Parc Botanique et Zoologique de Tsimbazaza and the Université d’Antananarivo. We are grateful
to the Malagasy authorities, in particular to the Ministry of the Environment, Water and Forests and
Madagascar National Parks, for research, collection and export permits; in particular, most material
of the new species described herein was collected under permits 32-MEF/SG/DGEF/DGDRF/SCB (7.
XII.2000) and 156-MEF/SG/DGEF/DADF/SCB (12.XII.2002), 224/15/MEEMF/SG/DGF/DAPT/SCBT (31.
VIII.2015) and 64/10/MEF/SG/DGF/DCB.SAP/SLRSE (Tsaratanana 2010). LR, MV, NS and MH were
supported by the Deutsche Forschungsgemeinschaft (grant VE247/16-1 –HO 3492/6-1 and STR-
1429/1-1–HO-3492 7/1) in the framework of the ‘TaxonOmics’priority program. EAP thanks their
current scholarship supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico
(CNPq process n° 141718/2016-1). Portuguese National Funds through FCT (Foundation for Science
and Technology) supported the Investigador FCT grant to AC (IF/00209/2014).
Disclosure statement
The authors acknowledge no financial interest or benefits from the direct application of this
research.
ORCID
Mark D. Scherz http://orcid.org/0000-0002-4613-7761
Elvis Almeida Pereira http://orcid.org/0000-0002-8810-7878
Michaela Preick http://orcid.org/0000-0002-8014-1975
Nicolas Straube http://orcid.org/0000-0001-7047-1084
Mariana L. Lyra http://orcid.org/0000-0002-7863-4965
Annemarie Ohler http://orcid.org/https://orcid.org/0000-0001-6531-464X
Jeffrey W. Streicher http://orcid.org/0000-0002-3738-4162
Franco Andreone http://orcid.org/0000-0001-9809-5818
Angelica Crottini http://orcid.org/0000-0002-8505-3050
Michael Hofreiter http://orcid.org/0000-0003-0441-4705
Miguel Vences http://orcid.org/0000-0003-0747-0817
28 L. RANCILHAC ET AL.

References
Altig R, McDiarmid RW. 2006. Descriptions and biological notes on three unusual mantellid tadpoles
(Amphibia: Anura: Mantellidae) from southeastern Madagascar. Proc Biol Soc Washington.
119:418–426.
AmphibiaWeb. 2019. University of California, Berkeley, CA, USA; [accessed 2019 Sept 09]. https://
amphibiaweb.org/
Banks RC, Goodman SM, Lanyon SM, Schulenberg TS. 1993. Type specimens and basic principles of
avian taxonomy. Auk. 110:413–414.
Blommers-Schlösser RMA. 1979. Biosystematics of the Malagasy frogs. I. Mantellinae (Ranidae).
Beaufortia. 352:1–77.
Blommers-Schlösser RMA, Blanc CP. 1991. Amphibiens (première partie). Faune de Madagascar.
75:1–379.
Boettger O. 1880. Diagnoses reptilium et batrachiorum novorum a Carolo Ebenau in insula Nossi-bé
madagascariensi lectorum. Zoologischer Anzeiger. 3:279–283.
Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: a flexible trimmer for Illumina sequence data.
Bioinformatics. 30:2114–2120. doi:10.1093/bioinformatics/btu170.
Boulenger GA. 1881. Description of a new species of frog from Madagascar. Ann Mag Nat Hist.
5:360–361. doi:10.1080/00222938109459533.
Boulenger GA. 1882. Catalogue of the Batrachia Salientia s. Ecaudata in the collection of the British
Museum. 2nd ed. London: Taylor and Francis. p. 588.
Boulenger GA. 1895. On a genus of frog peculiar to Madagascar. Ann Mag Nat Hist. 6(15):450.
doi:10.1080/00222939508677910.
Boulenger GA. 1918. On the Madagascar frogs of the genus Mantidactylus. Proc Zoo Soc London.
88:257–261. doi:10.1111/j.1096-3642.1918.tb02096.x.
Boumans L, Vieites DR, Glaw F, Vences M. 2007. Geographical patterns of deep mitochondrial
differentiation in widespread Malagasy reptiles. Mol Phylogenet Evol. 45:822–839. doi:10.1016/
j.ympev.2007.05.028.
Burrell AS, Disotell TR, Bergey CM. 2015. The use of museum specimens with high-throughput DNA
sequencers. J Hum Evol. 79:35–44. doi:10.1016/j.jhevol.2014.10.015.
Cocca W, Rosa GM, Andreone F, Aprea G, Eusebio Bergò P, Mattioli F, Mercurio V, Randrianirina JE,
Rosado D, Vences M, et al. 2018. The herpetofauna (Amphibia, Crocodylia, Squamata, Testudines)
of the Isalo Massif, Southwest Madagascar: combining morphological, molecular and museum
data. Salamandra. 54:178–200.
Cong Q, Shen J, Zhang J, Li W, Kinch LN, Calhoun JV, Warren AD, Grishin NV. 2019. Genomics reveals
the origins of ancient specimens. bioRxiv. 752121.
Dabney J, Knapp M, Glock I, Gansauge M, Weihmann A, Nickel B, Valdioserad C, García N, Pääbo S,
Arsuag J, et al. 2013. Complete mitochondrial genome sequence of a Middle Pleistocene cave
bear reconstructed from ultrashort DNA fragments. Proc Natl Acad Sci USA. 110:15758–15763.
doi:10.1073/pnas.1314445110.
Evans BJ, Gansauge MT, Stanley EL, Furman BL, Cauret CM, Ofori-Boateng C, Gvoždík V, Streicher JW,
Greenbaum E, Tinsley RC, et al. 2019.Xenopus fraseri: mr. Fraser, where did your frog come from?
PLoS One. 14:e0220892. doi:10.1371/journal.pone.0220892.
Friedman M, DeSalle R. 2008. Mitochondrial DNA extraction and sequencing of formalin-fixed
archival snake tissue. DNA Sequence. 19:433–437. doi:10.1080/19401730802449170.
Frost DR. 2019. Amphibian species of the World: an Online Reference. New York (USA): American
Museum of Natural History. Version 6.0. Electronic Database; [accessed 2019 Sept 09]. http://
research.amnh.org/herpetology/amphibia/index.html
Gansauge MT, Meyer M. 2013. Single-stranded DNA library preparation for the sequencing of
ancient or damaged DNA. Nat Protoc. 8:737–748. doi:10.1038/nprot.2013.038.
Gilbert MTP, Haselkorn T, Bunce M, Sanchez JJ, Lucas SB, Jewell LD, Van Marck E, Worobey M. 2007.
The isolation of nucleic acids from fixed, paraffin-embedded tissues–which methods are useful
when? PLoS One. 2:e537. doi:10.1371/journal.pone.0000537.
JOURNAL OF NATURAL HISTORY 29

Glaw F, Vences M. 1994.Afieldguide to the amphibians and reptiles of Madagascar. 3nd ed. Köln:
Vences & Glaw Verlag; p. 480.
Glaw F, Vences M. 1999. Resurrection and redescription of Mantidactylus tricinctus (Guibé, 1947)
from eastern Madagascar (Anura: Ranidae: Mantellinae). J Herpetol. 33:639–647. doi:10.2307/
1565581.
Glaw F, Vences M. 2004. A preliminary review of cryptic diversity in frogs of the subgenus
Ochthomantis based on mtDNA sequence data and morphology. Spixiana. 27:83–91.
Glaw F, Vences M. 2006. Phylogeny and genus-level classification of mantellid frogs (Amphibia,
Anura). Org Div Evol. 6:236–253. doi:10.1016/j.ode.2005.12.001.
Glaw F, Vences M. 2007.Afield guide to the amphibians and reptiles of Madagascar. 3rd ed.
Cologne: Vences & Glaw Verlag; p. 496.
Glaw F, Vences M, Gossmann V. 2000. A new species of Mantidactylus (subgenus Guibemantis) from
Madagascar, with a comparative survey of internal femoral gland structure in the genus
(Amphibia: Ranidae: Mantellinae). J Nat Hist. 34:1135–1154. doi:10.1080/00222930050020140.
Günther A. 1880. An introduction to the study of fishes. Edinburgh: Adam and Charles Black; p. 743.
Hind KR, Gabrielson PW, Lindstrom SC, Martone PT. 2014. Misleading morphologies and the
importance of sequencing type specimens for resolving coralline taxonomy (Corallinales,
Rhodophyta): Pachyarthron cretaceum is Corallina officinalis. J Phycol. 50:760–764.
Hoang DT, Chernomor O, Von Haeseler A, Minh BQ, Vinh LS. 2017. UFBoot2: improving the ultrafast
bootstrap approximation. Mol Biol Evol. 35:518–522. doi:10.1093/molbev/msx281.
Hutter CR, Cobb KA, Portik D, Travers S, Wood PL Jr., Brown RM. 2019. FrogCap: A modular sequence
capture probe set for phylogenomics and population genetics for all frogs, assessed across
multiple phylogenetic scales. bioRxiv. 825307.
Hykin SM, Bi K, McGuire JA. 2015. Fixing formalin: a method to recover genomic-scale DNA sequence
data from formalin-fixed museum specimens using high-throughput sequencing. PLoS One. 10:
e0141579. doi:10.1371/journal.pone.0141579.
IUCN. 2012. IUCN Red List categories and criteria: version 3.1. 2nd. Gland (Switzerland and
Cambridge, UK): IUCN.
Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Sturrock S, Buxton S, Cooper A, Markowitz S,
Duran C, et al. 2012. Geneious Basic: an integrated and extendable desktop software platform for
the organization and analysis of sequence data. Bioinformatics. 28:1647–1649. doi:10.1093/
bioinformatics/bts199.
Kumar S, Stecher G, Li M, Knyaz C, Tamura K. 2018. MEGA X: molecular evolutionary genetics analysis
across computing platforms. Mol Biol Evol. 35:1547–1549. doi:10.1093/molbev/msy096.
Kurabayashi A, Sumida M, Yonekawa H, Glaw F, Vences M, Hasegawa M. 2008. Phylogeny, recombi-
nation, and mechanisms of stepwise mitochondrial genome reorganization in mantellid frogs
from Madagascar. Mol Biol Evol. 25:874–891.
Laurent RF. 1946. Mises au point dans la taxonomie des ranides. Revue de Zoologie et de Botanique
Africaines. 39:336–338.
Li C, Corrigan S, Yang L, Straube N, Harris M, Hofreiter M, White WT, Naylor GJP. 2015. DNA capture
reveals transoceanic gene flow in endangered river sharks. Proc Nat Acad Sci. 112:13302–13307.
doi:10.1073/pnas.1508735112.
Li C, Hofreiter M, Straube N, Corrigan S, Naylor GJ. 2013. Capturing protein-coding genes across
highly divergent species. Biotechniques. 54:321–326. doi:10.2144/000114039.
Li J, Kuang T, Li C. 2016. Determining mitochondrial genome sequences from formalin fixed
paddlefish (Polyodon spathula) samples. Journal of Shanghai Ocean University. 5:659–667.
Librado P, Rozas J. 2009. DnaSP v5: A software for comprehensive analysis of DNA polymorphism
data. Bioinformatics. 25:1451–1452. doi:10.1093/bioinformatics/btp187.
McGuire JA, Cotoras DD, O’Connell B, Lawalata SZ, Wang-Claypool CY, Stubbs A, Huang X, Wogan
GOU, Hykin SM, Reilly SB, et al. 2018. Squeezing water from a stone: high-throughput sequencing
from a 145-year old holotype resolves (barely) a cryptic species problem in flying lizards. PeerJ. 6:
e4470. doi:10.7717/peerj.4470.
30 L. RANCILHAC ET AL.

Mercurio V, Andreone F. 2007. Two new canyon-dwelling frogs from the arid sandstone Isalo Massif,
central-southern Madagascar (Mantellidae, Mantellinae). Zootaxa. 1574:31–47. doi:10.11646/
zootaxa.1574.1.2.
Mocquard F. 1895. Sur les reptiles recueillis à Madagascar de 1867 à 1885 par M. Grandidier. Bulletin
de la Société philomathique de Paris. 7:93–111.
Mocquard F. 1900. Nouvelle contribution a la faune herpétologique de Madagascar. Bulletin de la
Société Philomathique de Paris Series 9. 2:111.
MusiałA, Gryglewski RW, Kielczewski S, Loukas M, Wajda J. 2016. Formalin use in anatomical and
histological science in the 19th and 20th centuries. Folia Med Cracov. 56:31–40.
Nguyen LT, Schmidt HA, von Haeseler A, Minh BQ. 2014. IQ-TREE: a fast and effective stochastic
algorithm for estimating maximum-likelihood phylogenies. Mol Biol Evol. 32:268–274.
doi:10.1093/molbev/msu300.
Padial JM, Miralles A, De la Riva I, Vences M. 2010. The integrative future of taxonomy. Front Zool. 7
(1):16. doi:10.1186/1742-9994-7-16.
Paijmans JL, Baleka S, Henneberger K, Taron UH, Trinks A, Westbury MV, Barlow A. 2017. Sequencing
single-stranded libraries on the Illumina NextSeq 500 platform. ArXiv. 1711.11004.
Paijmans JL, Fickel J, Courtiol A, Hofreiter M, Förster DW. 2016. Impact of enrichment conditions on
cross-species capture of fresh and degraded DNA. Mol Ecol Resour. 16(1):42–55. doi:10.1111/
1755-0998.12420.
Perl RGB, Nagy ZT, Sonet G, Glaw F, Wollenberg KC, Vences M. 2014. DNA barcoding Madagascar’s
amphibian fauna. Amphibia-Reptilia. 35(2):197–206. doi:10.1163/15685381-00002942.
Prosser SW, deWaard JR, Miller SE, Hebert PD. 2016. DNA barcodes from century-old type specimens
using next-generation sequencing. Mol Ecol Resour. 16(2):487–497. doi:10.1111/1755-
0998.12474.
Quinlan AR, Hall IM. 2010. BEDTools: a flexible suite of utilities for comparing genomic features.
Bioinformatics. 26(6):841–842. doi:10.1093/bioinformatics/btq033.
Rakotoarison A, Crottini A, Müller J, Rödel M-O, Glaw F, Vences M. 2015. Revision and phylogeny of
narrow-mouthed treefrogs (Cophyla) from northern Madagascar: integration of molecular, osteo-
logical, and bioacoustic data reveals three new species. Zootaxa. 3937(1):61–89. doi:10.11646/
zootaxa.3937.1.3.
Randrianiaina R-D, Strauß A, Glos J, Glaw F, Vences M, Arntzen JW. 2011. Diversity, external
morphology and ‘reverse taxonomy’in the specialized tadpoles of Malagasy river bank frogs of
the subgenus Ochthomantis (genus Mantidactylus). Contrib Zoo. 80(1):17–65. doi:10.1163/
18759866-08001002.
Rognes T, Flouri T, Nichols B, Quince C, Mahé F. 2016. VSEARCH: a versatile open source tool for
metagenomics. PeerJ. 4:e2584. doi:10.7717/peerj.2584.
Rosa GM, Andreone F, Crottini A, Hauswaldt JS, Noël J, Rabibisoa NH, Randriambahiniarime MO,
Rebelo R, Raxworthy CJ. 2012. The amphibians of the relict Betampona low-elevation rainforest,
eastern Madagascar: an application of the integrative taxonomy approach to biodiversity assess-
ments. Biodivers Conserv. 21(6):1531–1559. doi:10.1007/s10531-012-0262-x.
Ruane S, Austin CC. 2017. Phylogenomics using formalin-fixed and 100+ year-old intractable natural
history specimens. Mol Ecol Resour. 17(5):1003–1008. doi:10.1111/1755-0998.12655.
Salzburger W, Ewing GB, von Haeseler A. 2011. The performance of phylogenetic algorithms in
estimating haplotype genealogies with migration. Mol Ecol. 20(9):1952–1963. doi:10.1111/j.1365-
294X.2011.05066.x.
Scherz MD, Glaw F, Hutter CR, Bletz MC, Rakotoarison A, Köhler J, Vences M, Lötters S. 2019. Species
complexes and the importance of Data Deficient classification in Red List assessments: the case of
Hylobatrachus frogs. PLoS One. 14(8):e0219437. doi:10.1371/journal.pone.0219437.
Scherz MD, Vences M, Rakotoarison A, Andreone F, Köhler J, Glaw F, Crottini A. 2016. Reconciling
molecular phylogeny, morphological divergence and classification of Madagascan narrow-
mouthed frogs (Amphibia: Microhylidae). Mol Phylogenet Evol. 100:372–381.
Schulze A, Randrianiaina RD, Perl B, Glaw F, Vences M. 2016. The unexpectedly dull tadpole of
Madagascar’s largest frog Mantidactylus guttulatus. Acta Herpetologica. 11:119–125.
JOURNAL OF NATURAL HISTORY 31

Shokralla S, Porter TM, Gibson JF, Dobosz R, Janzen DH, Hallwachs W, Golding GB, Hajibabaei M.
2015. Massively parallel multiplex DNA sequencing for specimen identification using an Illumina
MiSeq platform. Sci Rep. 5(1):9687. doi:10.1038/srep09687.
Springer MS, Signore AV, Paijmans JL, Vélez-Juarbe J, Domning DP, Bauer CE, He K, Crerar L, Campos
PF, Murphy WJ, et al. 2015. Interordinal gene capture, the phylogenetic position of Steller’s sea
cow based on molecular and morphological data, and the macroevolutionary history of Sirenia.
Mol Phylogenet Evol. 91:178–193. doi:10.1016/j.ympev.2015.05.022.
Stephens M, Smith NJ, Donnelly P. 2001. A new statistical method for haplotype reconstruction from
population data. The American Journal of Human Genetics. 68(4):978–989. doi:10.1086/319501.
Stiller M, Sucker A, Griewank K, Aust D, Baretton GB, Schadendorf D, Horn S. 2016. Single-strand DNA
library preparation improves sequencing of formalin-fixed and paraffin-embedded (FFPE) cancer
DNA. Oncotarget. 7:59115–59128.
Tang EPY (2006). Path to effective recovering of DNA from formalin-fixed biological samples in
natural history collections: workshop Summary [pdf file]; [accessed 2019 Oct 1]. http://www.nap.
edu/catalog/11712.html.
Templeton JE, Brotherton PM, Llamas B, Soubrier J, Haak W, Cooper A, Austin JJ. 2013. DNA capture
and next-generation sequencing can recover whole mitochondrial genomes from highly
degraded samples for human identification. Investig Genet. 4(1):26. doi:10.1186/2041-2223-4-26.
Vences M, Andreone F, Glaw F. 2004. Voice of a giant: bioacoustic data for Mantidactylus guttulatus
(Amphibia: Mantellidae). Amphibia Reptilia. 25:112–115.
Vences M, Glaw F. 1999. Variation in Mantidactylus madecassus Millot & Guibé, 1950, a little known
Malagasy frog, with resurrection of Mantidactylus pauliani Guibé, 1974. Herpetol J. 9:101–110.
Vences M, Glaw F. 2004. Revision of the subgenus Chonomantis (Anura: Mantellidae: Mantidactylus)
from Madagascar, with description of two new species. J Nat Hist. 38(1):77–118. doi:10.1080/
0022293021000010331.
Vences M, Hildenbrand A, Warmuth KM, Andreone F, Glaw F. 2018. A new riparian Mantidactylus
(Brygoomantis) frog from the Tsaratanana and Manongarivo Massifs in northern Madagascar.
Zootaxa. 4486(4):575–588. doi:10.11646/zootaxa.4486.4.10.
Vences M, Kosuch J, Glaw F, Böhme W, Veith M. 2003.Molecular phylogeny of hyperoliid treefrogs:
biogeographic origin of Malagasy and Seychellean taxa and re-analysis of familial paraphyly. J
Zoo Syst Evol Res. 41(3):205–215. doi:10.1046/j.1439-0469.2003.00205.x.
Vences M, Thomas M, Bonett RM, Vieites DR. 2005a. Deciphering amphibian diversity through DNA
barcoding: chances and challenges. Phil Trans Royal Soc B. 360(1462):1859–1868. doi:10.1098/
rstb.2005.1717.
Vences M, Thomas M, van der Meijden A, Chiari Y, Vieites DR. 2005b. Comparative performance of the
16S rRNA gene in DNA barcoding of amphibians. Front Zool. 2(1):5. doi:10.1186/1742-9994-2-5.
Vieites DR, Wollenberg KC, Andreone F, Köhler J, Glaw F, Vences M. 2009. Vast underestimation of
Madagascar’s biodiversity evidenced by an integrative amphibian inventory. Proc Nat Acad Sci
USA. 106(20):8267–8272. doi:10.1073/pnas.0810821106.
Wandeler P, Hoeck PE, Keller LF. 2007. Back to the future: museum specimens in population
genetics. Trends Ecol Evol. 22:634–642.
Wollenberg KC, Vieites DR, Glaw F, Vences M. 2011. Speciation in little: the role of range and body
size in the diversification of Malagasy mantellid frogs. BMC Evol Biol. 11:217.
Yuan J-X, Hou X-D, Barlow A, Preick M, Taron UH, Alberti F, Basler N, Deng T, Lai X-L, Hofreiter M, et al.
2019. Molecular identification of late and terminal Pleistocene Equus ovodovi from northeastern
China. PLoS One. 14(5):e0216883. doi:10.1371/journal.pone.0216883.
32 L. RANCILHAC ET AL.