ArticlePDF Available

A third species of glassfrog in the genus Chimerella (Anura, Centrolenidae) from central Peru, discovered by an integrative taxonomic approach

  • Hessisches Landesmuseum Darmstadt


We studied the taxonomic status of glassfrogs collected in Departamento Huánuco, central Peru, which in the field were tentatively allocated to Chimerella, one of the twelve genera currently recognized in the family Centrolenidae. Detailed analyses of their morphology, bioacoustics, and molecular genetics supported their generic allocation and provided evidence for them representing a divergent and unnamed evolutionary lineage within Chimerella. We herein describe this lineage as a new species, being mainly distinguished from the two other known congeners, C. corleone and C. mariaelenae, by details of colouration in life and preservative, substantial differences in advertisement call, and differentiation in mitochondrial markers (12S rRNA, 16S rRNA, cytochrome b) and a nuclear-encoded marker (Rag-1). The new species is the southernmost distributed species in the genus and was found in a swampy habitat at the bank of the Río Patay Rondos, a tributary of the Río Monzon, in rainforest at the Andean-Amazon foothills at 798 m above sea level. Aspects of species delimitation within Chimerella and related future research are briefly addressed and discussed.
A third species of glassfrog in the genus Chimerella
(Anura, Centrolenidae) from central Peru, discovered
by an integrative taxonomic approach
Jörn Köhler1, Pablo J. Venegas2,3, Ernesto Castillo-Urbina4, Frank Glaw5, César Aguilar-Puntriano4, Miguel Vences6
1 Hessisches Landesmuseum Darmstadt (HLMD), Friedensplatz 1, 64283 Darmstadt, Germany
2 Rainforest Partnership, 4005 Guadalupe St., Austin, TX 78751, USA
3 Instituto Peruano de Herpetología (IPH), Augusto Salazar Bondy 136, Urb. Higuereta, Surco, Lima, Peru
4 Universidad Nacional Mayor de San Marcos, Museo de Historia Natural (MUSM), Departamento de Herpetología, Av. Arenales 1256, Lima 11, Peru
5 Zoologische Staatssammlung München (ZSM-SNSB), Münchhausenstr. 21, 81247 München, Germany
6 Zoological Institute, Technische Universität Braunschweig, Mendelssohnstr. 4, 38106 Braunschweig, Germany
Corresponding author: Jörn Köhler (
Academic editor: Alexander Haas
3 March 2023
9 May 2023
16 May 2023
We studied the taxonomic status of glassfrogs collected in Departamento Huánuco, central Peru, which in the eld were tentative-
ly allocated to Chimerella, one of the twelve genera currently recognized in the family Centrolenidae. Detailed analyses of their
morphology, bioacoustics, and molecular genetics supported their generic allocation and provided evidence for them representing
a divergent and unnamed evolutionary lineage within Chimerella. We herein describe this lineage as a new species, being mainly
distinguished from the two other known congeners, C. corleone and C. mariaelenae, by details of colouration in life and preservative,
substantial dierences in advertisement call, and dierentiation in mitochondrial markers (12S rRNA, 16S rRNA, cytochrome b) and
a nuclear-encoded marker (Rag-1). The new species is the southernmost distributed species in the genus and was found in a swampy
habitat at the bank of the Río Patay Rondos, a tributary of the Río Monzon, in rainforest at the Andean-Amazon foothills at 798 m
above sea level. Aspects of species delimitation within Chimerella and related future research are briey addressed and discussed.
Key Words
Amphibia, Chimerella mira, new species, bioacoustics, molecular genetics, morphology
Glassfrogs in the family Centrolenidae, being distributed
from Mexico southward to Argentina and southeastern
Brazil (Frost 2023), constitute one of the most peculiar
groups of Neotropical frogs, not only for their (partly)
transparent ventral skin, but also for their complex ecolo-
gy, behaviour and evolutionary history (e.g., Guayasamin
et al. 2008, 2009, 2020; Delia et al. 2017; Taboada et al.
2022). Currently, 163 species of centrolenids are recog-
nized, allocated to twelve genera (Frost 2023). Among
these is the genus Chimerella, erected by Guayasamin et
al. (2009) to accommodate Centrolene mariaelenae Cis-
neros-Heredia and McDiarmid, 2006 from Ecuador, which
is phylogenetically sister to all other genera of the tribe
Cochranellini. At the time of its description, Chimerella
was monotypic containing Chimerella mariaelenae only.
Later, Twomey et al. (2014) described a second species
in the genus, Chimerella corleone, originating from the
Cainarachi valley near Tarapoto (610 m a.s.l.) in the De-
partamento San Martín, north-eastern Peru.
Knowledge about Chimerella glassfrogs remains
limited. However, since its description, C. mariaelenae
has been recorded from northern Peru (Catenazzi and
Evolutionary Systematics. 7 2023, 195–209 | DOI 10.3897/evolsyst.7.102950
Copyright Jörn Köhler et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Jörn Köhler et al.: New species of Chimerella glassfrog
Venegas 2012) and numerous additional localities in
Ecuador (Cisneros-Heredia and Guayasamin 2007; Cis-
neros-Heredia and McDiarmid 2007; Cisneros-Heredia
2009; Guayasamin et al. 2020). Its larval morphology
(Terán-Valdez and Guayasamin 2014) and calls (Batallas
and Brito 2016; Guayasamin et al. 2020) have also been
described. Knowledge about the second species, C. cor-
leone, so far is restricted to information provided in the
original species description (Twomey et al. 2014).
During eldwork in November 2019 in central Peru (see
also Castillo-Urbina et al. 2021; Köhler et al. 2022), we
collected glassfrogs near the town of Tingo Maria, Depar-
tamento Huánuco, among which we tentatively identied
two individuals, according to their supercial morpholog-
ical similarity, as Chimerella corleone. However, the sub-
sequent study of the collected specimens, the analysis of
molecular markers and call recordings revealed substantial
dierences to C. corleone and provided dierent lines of
evidence for the presence of a third undescribed species in
the genus Chimerella, which we describe and name herein.
Materials and methods
Fieldwork was conducted in dierent areas of north-east-
ern and central Peru. Specimens were observed and
collected during opportunistic searching at night using
torchlights and headlamps. Geographic position was
recorded using a handheld GPS receiver set to WGS84
datum. Collected specimens were euthanised with an
overdose of 5% lidocaine or benzocaine gel applied on
the ventral surfaces of individuals (McDiarmid 1994).
Tissue samples were taken prior to xation and stored in
99% ethanol, while specimens were xed using 96% eth-
anol or formalin and subsequently stored in 70% ethanol.
Specimens were deposited in the herpetological collec-
tions of the Museo de Historia Natural, Universidad Na-
cional Mayor de San Marcos (MUSM), Lima, Peru, and
the Centro de Ornitología y Biodiversidad (CORBIDI),
Lima, Peru. FGZC refers to Frank Glaw eld numbers.
Morphometric measurements (in millimetres) were taken
by ECU with a digital calliper to the nearest 0.1 mm. For
proper comparison, denition of morphological character
states, diagnostic and descriptive schemes follow Cis-
neros-Heredia and McDiarmid (2007) and Guayasamin
et al. (2020). Measurements taken and used throughout
the text are: SVL, snout–vent length; HL, head length
(straight line distance from posterior corner of mouth to
the tip of the snout); HW, head width (measured at lev-
el of angle of jaws); TD, tympanum diameter (measured
horizontally); IND, internarial distance; IOD, interorbital
(distance between anterior margins of orbits); ED, hori-
zontal eye diameter; EW, upper eyelid width; END, eye–
nostril distance (from anterior margin of orbit to centre of
nostril); HaL, hand length (from proximal edge of inner
metacarpal tubercle to tip of third nger); TL, tibia length
(from the femur-tibia articulation to the tibia-heel proxi-
mal articulation); THL, thigh length (from the middle of
the cloacal slit to the proximal part of the femur-tibia ar-
ticulation); FL, foot length (distance from proximal mar-
gin of inner metatarsal tubercle to tip of toe IV). Colour in
life was described using digital photographs.
Vocalizations were recorded using an Olympus LS-5 dig-
ital recorder with built-in microphones, at a sampling rate
of 44.1 kHz and saved as uncompressed les. Recordings
were analysed using the software CoolEdit Pro 2.0 (Syntril-
lium Software Corp.). Frequency information was obtained
through Fast Fourier Transformation (FFT, width 1024
points) with Hanning window function; the audiospectro-
grams were obtained with Blackman window function at
256 bands resolution. Temporal measurements are given
in milliseconds (ms) or seconds (s), as range, with mean
± standard deviation in parentheses. Analysis of calls and
terminology in call descriptions follows the recommenda-
tions of Köhler et al. (2017), using the note-centered termi-
nological scheme. The recording is provided at the Zenodo
repository (
Molecular genetics
Our genetic analyses aimed at identifying divergence
among lineages of Chimerella. In addition to Chimerel-
la samples obtained by our own eldwork, we searched
for available GenBank sequences of Chimerella, and also
included a limited set of sequences representing all gen-
era currently recognized in the family Centrolenidae for
a representative taxon sampling. Allophryne ruthveni,
family Allophrynidae, the sister taxon of Centrolenidae
(Guayasamin et al. 2009), was chosen as the outgroup.
To reconstruct the phylogenetic relationships among
Chimerella samples, we combined sequences of three mi-
tochondrial genes: one fragment of the 12S rRNA gene
(12S), two fragments of the 16S rRNA gene (16S), and
one fragment of the cytochrome b gene (cob). We ex-
tracted DNA from tissue samples using a standard salt
protocol and PCR-amplied the gene fragments with the
following primers: 12SAL (AAACTGGGATTAGATAC-
GTAC) of Kocher et al. (1989) and Hrbek and Larson
(1999) with PCR protocol 94 °C (90 s), [94 °C (45 s),
52 °C (45 s), 72 °C (90 s) × 33], 72 °C (300 s); 16SL3
es et al. (2003) with PCR protocol 94 °C (90 s), [94 °C
(45 s), 52 °C (45 s), 72 °C (90 s) × 33], 72 °C (300 s);
Evolutionary Systematics 7 2023, 195–209
of Palumbi et al. (1991) with PCR protocol: 94 °C (90 s),
[94 °C (45 s), 50–53 °C (45 s), 72 °C (90 s) × 36‒40],
72 °C (300 s); and Cytb-a (CCATGAGGACAAATAT-
CGATTCATGT) of Bossuyt and Milinkovitch (2000)
with PCR protocol 94 °C (90 s), [94 °C (30 s), 53 °C
(45 s), 72 °C (90 s) × 35], 72 °C (600 s). Furthermore,
we sequenced fragments of a single-copy protein-cod-
ing nuclear-encoded gene, the recombination-activating
gene 1 (Rag-1) with a nested PCR approach, rst using
the primers Rag1-Mart Fl1 (AGCTGGAGYCARTAY-
CARTGRTGYTT), modied from Martin (1999), and
TCAT of Chiari et al. (2004), with PCR protocol 94 °C
(240 s), [94 °C (45 s), 45 °C (40 s), 72 °C (120 s) × 45],
72 °C (600 s) for both PCR rounds.
PCR products were puried with Exonuclease I and
Shrimp Alkaline Phosphatase digestion, and the puried
products along with sequencing primers were shipped to
LGC Genomics (Berlin) for sequencing on automated
capillary sequencing instruments. Chromatograms were
checked for base-calling errors and edited with Codon-
Code Aligner 6.0.2 (Codon Code Corporation, Dedham,
MA, USA). Newly generated sequences were submitted
to GenBank (accession numbers OQ877056OQ877069,
OQ888203OQ888210, and OQ888212OQ888219). A
table with all samples used, the associated GenBank ac-
cession numbers and sequences, as well as voucher num-
ber and locality, is available from the Zenodo repository
( along with the
alignment les.
Sequences of the focal Chimerella samples were com-
bined with sequences obtained from GenBank. For the
cob gene, very few comparative sequences were available
and the dataset therefore mostly consists of sequences ob-
tained in our own study for samples of Chimerella. The
combined sequences of the mitochondrial genes for 12S,
16S (two fragments), and cob were aligned with MAFFT
(Katoh and Standley 2013) as implemented in Concate-
nator (Vences et al. 2022). We then used Concatenator to
assemble a combined alignment partitioned by gene, and
analysed it in IQ-Tree 1.6.12 (Nguyen et al. 2015). We
used Modelnder (Kalyaanamoorthy et al. 2017) in IQ-
Tree with the MFP+MERGE setting to determine the best
partition and substitution models and subsequently ran
tree inference under the maximum likelihood optimality
criterion, with 1000 standard bootstrap replicates to test
robustness of nodes. Based on the Modelnder results, the
analysis was run with a partition of two character subsets:
(i) 12S and the two 16S fragments, with a TIM2+F+I+G4
model; and (ii) cob, with a TIM2+F+G4 model. We fur-
thermore performed unpartitioned ML analyses of the
same data set with MEGA7 (Kumar et al. 2016) under
a GTR+G substitution model selected by the Bayesian
Information Criterion in MEGA. To quantify genetic di-
vergences, we calculated uncorrected pairwise distances
among the 16S sequences (p-distances) in MEGA7.
The alignment of the nuclear Rag-1 gene was anal-
ysed separately from the mitochondrial sequences, with
the goal to assess concordance in the dierentiation of a
nuclear encoded and a mitochondrial gene. Sequences of
Rag-1 were aligned with the Muscle alignment option,
and the nuclear gene alignments trimmed in MEGA7
(Kumar et al. 2016). We then graphically visualized re-
lationships among alleles (haplotypes) of Rag-1 using a
haplotype network approach. Alleles (haplotypes) of the
nuclear gene were inferred using the PHASE algorithm
(Stephens et al. 2001) implemented in DnaSP (Version
5.10.3; Librado and Rozas 2009), a Maximum Likeli-
hood tree from the phased sequences inferred under the
Jukes-Cantor substitution model in MEGA7, and this
tree used along with the respective alignment as input for
Haploviewer (written by G. B. Ewing; http://www.cibiv.
at/~greg/haploviewer), a software that implements the
methodological approach of Salzburger et al. (2011).
For a formal species delimitation analysis, we used
ASAP (Puillandre et al. 2021) as implemented in
iTaxoTools (Vences et al. 2021) on a trimmed alignment
(475 bp) of the 16S 3’ fragment, which was available
from all Chimerella individuals.
Nomenclatural act
The electronic version of this article in Portable Docu-
ment Format (PDF) will represent a published work ac-
cording to the International Commission on Zoological
Nomenclature (ICZ), and hence the new name contained
in the electronic version is eectively published under that
Code from the electronic edition alone. This published
work and the nomenclatural acts it contains have been
registered in ZooBank, the online registration system for
the ICZN. The ZooBank LSIDs (Life Science Identiers)
can be resolved and the associated information viewed
through any standard web browser by appending the
LSID to the prex The LSID for this
publication is:
Molecular relationships
Our Maximum Likelihood tree (Fig. 1), based on 2326 nu-
cleotides of the combined mitochondrial DNA sequences
(12S and 16S rRNA and cob), grouped with high support
species into genera according to current centrolenid classi-
cation and recovered the two subfamilies Centroleninae and
Hyalinobatrachinae. However, many deep nodes in the tree
were poorly supported (bootstrap proportions often <50%),
indicating that the combined gene fragments contained in-
sucient phylogenetic information to reliably resolve in-
tergeneric relationships within the Centrolenidae. Our tree
contains a polytomy with respect to the genus Ikakogi and
the subfamily Hyalinobatrachinae, which is unsurprising
Jörn Köhler et al.: New species of Chimerella glassfrog
given that more comprehensive phylogenetic studies re-
vealed the uncertain relationships of Ikakogi within Centr-
olenidae (see Guayasamin et al. 2009; Hutter et al. 2013a).
With regard to the focal genus Chimerella, the newly
collected central Peruvian samples from west of Tingo Ma-
ria form a highly supported clade being sister to C. corle-
one from the type locality and nearby sites in the Cainara-
chi valley, plus three samples from higher elevations in the
Departamentos Amazonas and San Martín. We tentatively
refer to these high-elevation samples as Chimerella sp. and
not C. corleone, as they dier remarkably in morphology
and need further research. The clade containing C. corle-
one, C. sp. and the new samples from central Peru is sis-
ter to C. mariaelenae. In summary, our analysis reveals
three distinct and highly supported clades within the genus
Chimerella, one representing C. corleone (including high
elevation populations in need of taxonomic clarication),
one representing C. mariaelenae, and a third containing our
samples from central Peru. Furthermore, among the limit-
ed samples available from these three Chimerella clades,
no haplotype sharing was detected in the nuclear-encoded
Rag-1 gene fragment (1021 nucleotides; Fig. 2).
Figure 1. Maximum Likelihood phylogenetic tree of centrolenid frogs inferred from an alignment of 2326 nucleotides of the mito-
chondrial genes 12S and 16S rRNA, and cytochrome b. Allophryne ruthveni was used to root the tree (removed for better graphical
presentation). Numbers at nodes are bootstrap values in percent calculated with MEGA (500 replicates; not shown if <50%) and
IQ-Tree (1000 replicates; not shown if <50). Sequences from samples in bold font were newly obtained for this study. The taxon
name is followed by the sample locality and number of the voucher specimen (as provided in GenBank) in parentheses. Inset photos
depict the holotypes in life of Chimerella corleone (CORBIDI 10467) and C. mira sp. nov. (MUSM 40278), respectively.
Evolutionary Systematics 7 2023, 195–209
Uncorrected p-distances in the 16S rRNA gene (for
a fragment of 523 nucleotides at the 3’ terminus of the
gene) among studied samples of Chimerella are as fol-
lows: Between the new species and C. corleone, p-dis-
tances range from 3.6–4.0%; between the new species
and C. mariaelenae they range from 3.5–3.8%; between
C. corleone and C. mariaelenae they range from 3.7–
4.0%; between the new species and Chimerella sp. they
range from 3.2–3.8%; between C. mariaelenae and C. sp.
they range from 3.3–3.8%; and between C. corleone and
C. sp. they range from 0.2–0.6%. Apart from the low
genetic divergence observed between C. corleone and
C. sp., distances are in a similar or higher range when
compared to distances of congeneric species pairs of oth-
er centrolenids, as revealed by cross-checking available
GenBank sequences.
The best species partition suggested by ASAP, with a
score of 1.5, supported the presence of three subsets in
the 16S data, corresponding to (a) C. corleone plus the
three specimens from Nuevo Chirimoto, Posic, and Santo
Toribio, (b) C. mariaelenae, and (c) the focal specimens
from central Peru (graphic presentation of result avail-
able from the Zenodo repository, DOI: 10.5281/zeno-
Our examination of morphological character states of
newly collected specimens and their comparison with de-
scribed species of Chimerella revealed shared character
states conrming their allocation to Chimerella (humeral
spine in males, transparent ventral parietal peritoneum,
white pericardial, hepatic and visceral peritonea; see
Guayasamin et al. 2009). However, details of dorsal co-
louration in life, iris colouration, colour in preservative
and snout shape revealed a few constant qualitative dif-
ferences among the mitochondrial clades identied, pro-
viding further indication of these representing divergent
evolutionary lineages.
Although call recordings are sparse, our analysis of
recordings and published call descriptions (see be-
low) revealed qualitative and quantitative dierences
among the calls of individuals assigned to the three mi-
tochondrial clades. Calls of C. corleone and C. mari-
aelenae dier from those of the population from cen-
tral Peru by containing simple single pulse ‘Tic’ notes
versus multi-pulsed ‘Trii’ notes (sensu Duarte-Marín
et al. 2022). Moreover, note duration in calls of C. cor-
leone and C. mariaelenae is much shorter (4–7 and
10–15 ms, respectively) when compared to calls of the
central Peruvian population (42–85 ms). These nd-
ings constitute a very strong indication of respective
lineage divergence, as the dierences observed are far
beyond those to be expected from intra-specic call
variation (see Köhler et al. 2017), particularly in cen-
trolenids which commonly exhibit rather similar calls
among dierent species (e.g., Guayasamin et al. 2020;
Duarte-Marín et al. 2022).
In summary, our results from the analyses of mo-
lecular genetics, morphology and bioacoustics provide
independent lines of evidence for the central Peruvian
samples of Chimerella representing a distinct divergent
evolutionary lineage which so far remains undescribed
and is herein named:
Chimerella mira sp. nov.
Type material. Holotype. MUSM 40278 (FGZC
6233), adult male (Fig. 3a–c), from a point approxi-
mately 16 km airline west of Tingo Maria (09°18.09'S,
76°08.71'W, 798 m above sea level), close to the settle-
ment “Corvina Colorada”, at the bank of the Río Patay
Rondos, Provincia Leoncio Prado, Departamento Huánu-
co, Peru, collected on 6 November 2019 by Ernesto Cas-
tillo-Urbina, Frank Glaw and Jörn Köhler.
Figure 2. Haplotype network based on 1021 nucleotides of the nuclear-encoded Rag-1 gene from seven specimens of Chimerella
(based on phased alleles, each specimen is therefore represented twice in the network). Size of circles represents number of times
the allele was observed. Colours chosen correspond to those of mitochondrial lineages in Fig. 1.
Jörn Köhler et al.: New species of Chimerella glassfrog
Paratype. MUSM 40264 (FGZC 6215), adult male
(Fig. 3), same data as holotype, but collected on 5 No-
vember 2019.
Etymology. The specic epithet is a Latin adjective
(feminine form) meaning ‘surprising’. It refers to the
fact that this species surprisingly turned out to be unde-
scribed, after at rst impression in the eld having been
tentatively identied as C. corleone.
Denition. A species in the genus Chimerella, based on
molecular relationships and shared morphological traits,
characterized by the following combination of characters:
(1) dentigerous processes of vomer and vomerine teeth
absent; (2) snout truncate in dorsal view, truncate in
lateral prole; canthus rostralis straight in dorsal view,
rounded in cross-section; nostrils ush with surround-
ing skin; (3) tympanum and tympanic annulus evident,
round, its diameter about 25% of eye diameter
; su-
pratympanic fold weakly dened, not concealing upper
tympanum; (4) dorsal skin nely shagreened, with few
minute scattered dorsal tubercles; skin on venter and ven-
tral surfaces of thighs granular; (5) a pair of enlarged sub-
cloacal warts; (6) ventral parietal peritoneum
(condition P0 sensu Cisneros-Heredia and McDiarmid
2007); iridophores in pericardium and peritonea cover-
ing digestive tract
; kidneys and urinary bladder lacking
iridophores (condition V5); (7) liver with two broadly
rounded right/left lobes, not forming free aps, covered
by iridophores (condition H1); (8) humeral spine and sin-
gle subgular vocal sac present in adult males; (9) webbing
absent between ngers I and II, basal webbing between
ngers II and III; webbing formula II2-–3-III2+–2+IV
(10) webbing between toes extensive
webbing formu-
enamelled fringe
present on postaxial edge of nger IV;
ulnar fold diuse;
tarsal fold
absent; enlarged tubercles on ventrolateral edg-
es of arm and tarsus absent; (12) concealed prepollex, not
enlarged, prepollical spine not projecting; nuptial pad ab-
sent, but diuse nuptial excrescence formed by glandular
clusters (Type V); (13) nger I slightly longer than n-
ger II; (14) diameter of eye three times wider than width
of disc on nger III; (15) in life, dorsum yellow-green
with small round scattered pale-yellowish ecks; venter
transparent white; bones green; (16) in preservative, dor-
sum lavender with small scattered round cream ecks;
dorsal surfaces of limbs yellowish cream, with scattered
melanophores; ventral surfaces yellowish cream; (17)
in life, iris silvery white with ne black spotting, and a
dark brown median streak formed by ne spots; circum-
pupillary ring absent; (18) dorsal surfaces of ngers and
toes lacking melanophores, except for toes IV and V; (19)
males call from the upper surface of leaves; calls consist
of 2–3 high-pitched pulsed notes (‘Trii’ calls sensu Duar-
te-Marín et al. 2022), 42–85 ms note duration, 160–239
ms inter-note interval duration within calls; dominant fre-
quency 5543–6135 Hz; (20) ghting behavior unknown
(but probably present in males; see below); (21) egg
clutches unknown; (22) tadpoles unknown; (23) minute
body size (sensu Guayasamin et al. 2020), SVL in adult
males 18.1–19.6 mm (n = 2); females unknown.
Diagnosis. The new species is morphologically
most similar to C. corleone. However, it diers from
C. corleone by ne dark spots in the iris in life (versus
dark reticulation; Figs 4, 5), a truncate snout in lateral
prole (versus slightly rounded; Fig. 6), tarsal fold ab-
sent (versus present as a line of low white warts on the
lateral edge of tarsus), greyish-lavender dorsal colour
in preservative (versus greyish-green), a dispersed net-
work of melanophores on dorsal surfaces resulting in
a light yellow-green colour in life (versus a very dense
network of melanophores on dorsal surfaces, resulting in
dark green life colouration; Figs 3, 4), a call consisting
of pulsed ‘Trii’ notes (sensu Duarte-Marín et al. 2022)
with 42–85 ms duration (versus simple ‘Tic’ notes of 10–
15 ms duration), and substantial dierentiation in certain
molecular markers. The new species mainly diers from
C. mariaelenae by a yellow-green dorsum with small
round scattered yellowish ecks (versus green dorsum
with black ecks and punctuation; Fig. 7), greyish-lav-
ender dorsum with small round cream ecks in preser-
vative (versus pale lavender with dark lavender ecks),
silvery white iris in life (versus orange to reddish iris), an
advertisement call consisting of pulsed ‘Trii’ notes with
42–85 ms duration (versus simple ‘Tic’ notes of 4–7 ms
duration; Guayasamin et al. 2020), and substantial dier-
entiation in certain molecular markers.
Description of the holotype. Adult male, SVL 19.6 mm,
in good state of preservation (Fig. 8). HW about 1/5 wider
than body; HW 29% of SVL; HW 1.15 times HL. Snout
truncate in dorsal view, truncate in lateral prole (Fig.
6a); END/ED 0.65; END/IOD 0.55. Loreal region con-
cave, nostrils ush with surrounding skin, round; internar-
ial region concave anterodorsally; canthus rostralis well
straight in dorsal view, rounded in cross-sec-
. Eyes directed anterolaterally, angled 51° relative to
midline of body (where anteriorly facing eyes would be
90° relative to midline); ED 3.0 times wider than width
of disc on nger III; ED 41% of HL and 100% of IOD.
Tympanum noticeable with tympanic annulus visible,
more evidently ventrally than dorsally, annulus and mem-
brane coloured as dorsum; supratympanic fold weakly de-
ned leaving entire tympanum visible, tympanum round
with slight dorsal inclination. Dentigerous processes on
vomers absent; dentigerous process on premaxillae and
maxillae present; choanae large, circular, separated more
widely than nostrils; tongue removed for tissue sample;
vocal slits present, wide, oblique and lateral to the tongue.
Forelimbs moderately robust, with forearm attened and
roughly 1.4 times as wide as arm; ulnar fold present, low
diuse, white; tubercles on ventrolateral edge of arm
absent; humeral spine externally visible as an elongated
bump, slightly less dened in preservative than in life.
Relative length of ngers: II < I < IV < III; nger discs
distinctly expanded, those on ngers I and II rounded, on
ngers III and IV slightly truncate, larger than toe discs;
Evolutionary Systematics 7 2023, 195–209
Figure 3. Chimerella mira sp. nov. from west of Tingo Maria in life: male holotype (MUSM 40278, FGZC 6233) in a frontal, b dor-
solateral, and c ventral views; male paratype (MUSM 40264, FGZC 6215) in d lateral and e ventral views.
Jörn Köhler et al.: New species of Chimerella glassfrog
width of disc on nger III 32% of ED; webbing absent
between ngers I and II, basal webbing between ngers
II and III, webbing formula II2-–3-III2+–2+IV. Prepollex
concealed; subarticular tubercles round, evident; supernu-
merary tubercles absent, palmar tubercle round and small,
thenar tubercle barely distinct, minute, ovoid; nuptial pads
absent, but diuse nuptial excrescence formed by glandu-
lar clusters (Type V sensu Guayasamin et al. 2020). Hind
limbs slender, TL 51% of SVL; tarsal fold absent; tuber-
cles on ventrolateral edge of tarsus absent. Relative length
of toes: I < II < III < V < IV; toe discs slightly expanded,
round; inner metatarsal tubercle narrow, elongated, ovoid,
slightly protruding; outer metatarsal tubercle not visible.
Webbing formula of feet: I1+–2+II1+–2.5III1+–3-IV3-–1+V.
Dorsal skin nely shagreened, with few small scattered
cream coloured tubercles on dorsum and dorsal surfaces
Figure 4. Chimerella corleone in life: a dorsolateral view of amplectant couple from the type locality photographed at night (note
the striking dark reticulation of the iris); b dorsolateral and c ventral views of the male holotype (CORBIDI 10467). Courtesy of J.
Delia and E. Twomey.
Evolutionary Systematics 7 2023, 195–209
of limbs; skin on venter and ventral sides of thighs gran-
ular, skin on throat smooth; cloacal opening at level of
upper thighs, concealed by faint superior dermal fold; a
pair of round, low, unpigmented subcloacal warts present
on ventral side; crenulated aps absent.
Measurements (in mm). SVL 19.6, HL 5.8, HW 6.7,
TD 0.7, IND 1.5, IOD 2.7, ED 2.7, EW 1.2, END 1.5,
HaL 5.4, TL 9.9, THL 10.8, FL 7.9.
In life (Fig. 3), dorsal surfaces translucent yel-
low-green, with small, round, widely scattered yellowish
cream ecks on dorsum and dorsal surfaces of thighs;
dorsal surfaces of hands and feet yellow-green; dor-
sal surfaces of nger and toe discs orange-yellow; area
around nostrils, dorsal surfaces of arms and thighs dusted
with minute dark melanophores; upper lip with a narrow
tan line anteriorly that vanishes posteriorly; venter trans-
parent, whitish; throat transparent with a turquoise tint;
ventral surfaces of limbs lemon green; ventral surfac-
es of nger and toe discs orange-yellow; greyish-white
diuse line along lateral edge of proximal ulna; diuse
white spots in cloacal area; subcloacal wart ornamenta-
tion transparent; parietal peritoneum transparent; peri-
cardium, hepatic peritonea and visceral peritonea white,
urinary bladder transparent; iris silvery white with ne
black spotting, median brown streak formed by densely
spaced ne spots, circumpupillary ring absent, posterior
iris periphery white.
After three years in preservative, dorsum greyish-lav-
ender with scattered small cream ecks; dorsal surfaces
of limbs yellowish cream with minute scattered mela-
nophores; dorsal surfaces of hand and ngers yellowish
cream; dorsal surfaces of feet and toes yellowish cream,
with scattered melanophores extending on dorsal surfaces
of toes IV and V; posterior surfaces of thighs yellowish
cream; venter and ventral surfaces of arms and legs yel-
lowish cream, throat cream (Fig. 8).
Variation. Overall, the male paratype MUSM 40264
is rather similar to the holotype. In life, it had a slight-
ly paler green dorsal colouration, with fewer and less
distinct minute yellowish ecks on dorsum. The throat
lacked the turquoise tint and the brown median streak in
the iris was less distinctly expressed when compared to
Figure 5. Comparison of eye colouration in life: a male holo-
type (MUSM 40278) and b male paratype (MUSM 40264) of
Chimerella mira sp. nov.; c male holotype (CORBIDI 10467)
and d female paratype (CORBIDI 10465) of Chimerella cor-
leone (courtesy of J. Delia and E. Twomey). Note the ne dark
spotting versus dark reticulation in the iris.
Figure 6. Lateral views of heads of preserved holotypes of
a Chimerella mira sp. nov. (MUSM 40278) and b Chimerella
corleone (CORBIDI 10467; courtesy of E. Twomey). Orange
lines indicate outline of snout shape in lateral prole. Not to scale.
Figure 7. Amplectant couple of Chimerella mariaelenae from
the Cordillera de Kampankis, 1100 m a.s.l., Departamento Am-
azonas, Peru, in life.
Jörn Köhler et al.: New species of Chimerella glassfrog
the holotype (Fig. 3). Measurements (in mm) of the para-
type are as follows: SVL 18.1, HL 5.5, HW 6.7, TD 0.6,
IND 1.7, IOD 2.4, ED 2.4, EW 1.3, END 1.4, HaL 5.2, TL
10.7, THL 10.7, FL 7.6.
The male paratype was dissected for inspection of
internal organs which appear as follows: liver with two
broadly rounded right/left lobes sagittally divided, not
forming free aps, completely covered in iridophores
(white), corresponding to state H1 sensu Cisneros-He-
redia and McDiarmid (2007). Gall bladder, pericardium,
liver and gastrointestinal peritoneum covered in irido-
phores (white). Kidneys and urinal bladder are tan in co-
lour and thus not covered by iridophores. Testes ovoid
and partially covered in a white iridophore reticulum.
Distribution of iridophores in visceral peritonea falls into
state V5 sensu Cisneros-Heredia and McDiarmid (2007).
Natural history, distribution, and threat status. Both
males were collected at the stream bank of the Río Patay
Rondos, a medium-sized tributary of the Río Monzon,
which itself is part of the Huallaga river system. The hab-
itat consisted of a swampy area, apparently temporarily
ooded by the river, with small lentic waterbodies, emerg-
ing shrub vegetation and younger trees (Fig. 9). Shortly af-
ter dusk, male individuals were sitting on upper surfaces of
leaves approximately 0.5 to 2.5 m above the ground while
calling during light rainfall. Some ne transverse scratches
were visible on the anterior dorsum of the paratype MUSM
40264, arguing for the occurrence of male-male ghting
behaviour (e.g., Hutter et al. 2013b). Egg clutches and
larvae are unknown. Anuran species found in sympatry
were Boana lanciformis, Leptodactylus griseigularis, and
Adenomera sp. The glassfrog Hyalinobatrachium carles-
vilai occurred at nearby sites within a few hundred metres
distance. So far, the species is only known from the type
locality at an elevation of 798 m, but might be more wide-
spread in the Huallaga River basin at similar elevations.
Because population size, actual range, and thus potential
threats are unknown, we propose the IUCN Red List status
‘Data Decient’ for C. mira (see also Scherz et al. 2019).
Figure 8. Preserved male holotype of Chimerella mira sp. nov. (MUSM 40278, FGZC 6233) in a dorsal and b ventral views.
Figure 9. Type locality and habitat of Chimerella mira sp. nov.
on the bank of the Río Patay Rondos, a tributary of the Río
Monzon: a view to the east along the river bed. The yellow
arrow indicates the area where both reported specimens were
collected; b night view of the swampy habitat at the edge of the
river showing shrub vegetation from which males were calling.
Evolutionary Systematics 7 2023, 195–209
Advertisement call. Calls were emitted at somewhat
irregular intervals and occurred in ‘waves’ with several
males calling nearly synchronously. The advertisement
calls recorded on 5 November 2019 at the type locality
(estimated air temperature ca. 25 °C; recording distance
approximately 1.5 m) consist of 2 to 3 high-pitched,
pulsed notes of short duration (Fig. 10a). Notes exhibit
considerable amplitude modulation, with maximum call
energy present at the beginning of the note, continuously
decreasing towards its end. Pulse structure is rather irreg-
ular within notes, with pulses being partly fused and of
diering amplitude. As a consequence, the total number
of pulses per note is not reliably countable, but in most
cases 3 to 4 distinctly separated pulses are evident at the
beginning of each note, followed by about 10–12 less dis-
tinctly separated pulses. We observed pulse rate within
notes to range approximately around 200 pulses/second.
Other numerical parameters of 12 analysed calls from 4
individuals are as follows: number of notes per call 2–3
(2.8 ± 0.5); call duration 322–707 ms (552.1 ± 141.8 ms);
note duration 42–85 ms (64.6 ± 11.7 ms); inter-note inter-
val within calls 160–239 ms (197.0 ± 26.4 ms); call repe-
tition rate approximately 1.5–1.9 calls/minute; dominant
frequency 5543–6135 Hz (5897 ± 148 Hz); prevalent
bandwidth 4500–7500 Hz, with weak call energy present
up to 21 kHz. Among the notes within a call, dominant
frequency is highest in the rst note and slightly decreas-
es in subsequent notes. The character of this call would
qualify as a ‘Trii’ call according to the denition of Duar-
te-Marín et al. (2022).
Comparative call data. The only available call record-
ing of Chimerella corleone is that described by Twomey
et al. (2014) recorded from a topotypic male in captivity
after dislodging it from the female with which it has been
in amplexus. These conditions argue for the call repre-
senting a mating call, not an advertisement call (J. Delia,
pers. comm.) and thus comparison should be regarded
with some reservation. However, as mentioned by Twom-
ey et al. (2014), advertisement calls heard in the eld were
rather similar in character. We re-analysed the available
recording of the single call using the methodology de-
scribed above (for recording equipment used see Twomey
et al. 2014). The call (Fig. 10b) has the following numer-
ical parameters: number of notes per call 2; call duration
521 ms; note duration 10 and 15 ms; inter-note interval
493 ms; dominant frequency of the rst note 6485 Hz, and
6526 Hz in the second note; prevalent bandwidth dicult
to determine due to oversaturated recording level, but call
energy is apparently present up to 20 kHz. The character
of this call would qualify as a ‘Tic’ call according to the
denition of Duarte-Marín et al. (2022).
Calls of C. mariaelenae from Pangayaku Creek
(929 m a.s.l.), Provincia Napo, Ecuador, have been de-
scribed by Guayasamin et al. (2020). The calls (call du-
ration 231–1761 ms) contain 2–10 unpulsed high-pitched
notes of very short duration (4–7 ms), repeated at compar-
atively short intervals. Dominant frequency ranged from
6718–8010 Hz (Guayasamin et al. 2020). Also, Batallas
and Brito (2016) described the call of C. mariaelenae,
from Sangay National Park (1750 m a.s.l.), Provincia
Morona Santiago, Ecuador. Their analysis described
call parameters quite dierent from those reported by
Guayasamin et al. (2020), with calls always containing 3
notes (call duration 668–808 ms) and much longer note
durations of 54–116 ms. The dierences among these two
call descriptions for calls of C. mariaelenae are beyond
Figure 10. Audiospectrograms and corresponding oscillograms of calls of Chimerella: a Chimerella mira sp. nov. (call voucher
MUSM 40264) from west of Tingo Maria, Departamento Huánuco, Peru. Below an expanded oscillogram depicting the rst note of
the call; b Chimerella corleone from the type locality, Departamento San Martín, Peru. Below an expanded oscillogram depicting
the rst note of the call. Both recordings high-pass ltered at 1000 Hz.
Jörn Köhler et al.: New species of Chimerella glassfrog
those usually considered to represent inter-specic call
variation (see Köhler et al. 2017) and would argue for the
calls belonging to dierent species. However, although
apparently rare in centrolenids (see Duarte-Marín et al.
2022), the calls described could also refer to two dierent
call types of C. mariaelenae. Batallas and Brito (2016)
described the males calling in dense choruses containing
numerous males, thus territorial and/or aggressive func-
tion of the calls recorded could be an explanation. With
the data at hand, we are unable to clarify the reason for
these call dierences described for C. mariaelenae, but
for our diagnosis above we here relied on the description
provided by Guayasamin et al. (2020). However, if con-
sidering the calls described by Batallas and Brito (2016)
to represent the advertisement call of C. mariaelenae,
there would be some overlap in numerical parameters
with calls of C. mira, but the distinct pulsed amplitude
structure present in C. mira has not been observed in calls
described by the mentioned authors.
Our study of Chimerella glassfrogs in Peru revealed
some surprising results. When discovering the specimens
west of Tingo Maria, we tentatively identied them as
C. corleone according to an overall morphological simi-
larity and the fact that the locality is part of the Huallaga
River system, which includes the type locality of C. cor-
leone, in the Cainarachi valley, further north. On the oth-
er hand, specimens collected in northern Peru, occurring
at higher elevations in the Departamentos Amazonas and
San Martín, are morphologically dierent when com-
pared to C. mariaelenae and C. corleone and therefore
are believed to represent an undescribed species. Molec-
ular studies, including mitochondrial and nuclear mark-
ers, revealed an unexpected picture. The specimens from
west of Tingo Maria, at rst impression morphologically
cryptic to topotypic C. corleone, turned out to represent
a rather divergent lineage distinguished also by consid-
erable dierences in the advertisement call and details
of morphology. Although morphologically considerably
dierent, the specimens from higher elevations in Am-
azonas and San Martín, approximately 160–180 km east
of the type locality of C. corleone (Fig. 11), were recov-
ered to be part of a clade containing topotypic C. corle-
one and showed only very little genetic dierentiation to
them in the 16S gene fragment (p-distances 0.2–0.6%),
ranging among values typical for intraspecic variation.
However, there is apparently no haplotype sharing of
these populations with C. corleone in the nuclear encod-
ing Rag-1 gene.
As a rst consequence of our ndings, we here de-
scribed the clade from west of Tingo Maria as a new
species, Chimerella mira. Our molecular, morphological,
and bioacoustic results provided independent lines of
evidence for this population representing a third species
in the genus. The genetic divergence between this new
species and the two known congeneric species is rather
pronounced, being at a similar level of other congener-
ic species pairs within the Centrolenidae, or even great-
er, as revealed by cross-checking available GenBank
Figure 11. Schematic map of north-western South America indicating the approximate known distribution of Chimerella by co-
loured dots (data for C. mariaelenae partly taken from Guayasamin et al. 2020).
Evolutionary Systematics 7 2023, 195–209
sequences. Among the morphological dierences found
between the three species of Chimerella, details in dor-
sal colour pattern are evident, but most striking are the
dierences in iris colouration in life, being silvery white
with ne dark spotting in C. mira, versus silvery grey
with dark reticulation in C. corleone, and orange to red-
dish in C. mariaelenae. Although these dierences may
appear negligibly small, iris colouration has proven to
be a very reliable diagnostic character in many groups
of frogs to distinguish among species (Glaw and Venc-
es 1997), including centrolenids (e.g., Cisneros-Heredia
and McDiarmid 2007; Guayasamin et al. 2020). Further-
more, although calls of centrolenid species often share
high dominant frequency and short note duration and thus
may sound rather similar (Duarte-Marín et al. 2022), the
call dierences revealed by our call analyses are clearly
beyond those considered to represent intra-specic vari-
ation (see Köhler et al. 2017). Although the allocation of
calls to C. mariaelenae includes some uncertainties (see
above) and the available call recordings of C. corleone
may possibly represent a mating call (although similar in
character to an advertisement call; see above), calls of
C. mira dier signicantly in qualitative and quantitative
traits. This mainly is the presence of multi-pulsed notes
versus notes consisting of a single pulse only (‘Trii’ calls
versus ‘Tic’ calls sensu Duarte-Marín et al. 2022), and the
much longer note duration (42–85 ms versus 10–15 ms in
C. corleone and 4–7 ms in C. mariaelenae). In summary,
these dierences provide multiple lines of evidence for
evolutionary lineage divergence and justication for the
description of a new species.
Remarkably, the habitat at the type locality of C. mira
diers considerably from that of C. corleone, which has
been described as vegetation on vertical rock walls with-
in the spray zone at the edge of waterfalls, where all in-
dividuals were exclusively found (Twomey et al. 2014).
The association with small, fast owing streams, provid-
ing high air humidity by a certain amount of spray, is a
common habitat for many of the stream-breeding centr-
olenids along the eastern Andean slopes. The swampy
habitat of C. mira at the edge of a comparatively large
river that we report here appears rather dierent partic-
ularly when compared to the habitat of C. corleone in
being much more exposed, lacking any rock walls and
water spray. This would be in agreement with the ob-
servations of Rivera and Folt (2018), who found con-
generic centrolenids to use dierent types of habitat.
However, as large streams tend to have high water level
and strong current during the rainy season, bearing the
danger of washing away even larvae with stream-adapt-
ed morphology downstream into unsuitable habitat, the
habitat of C. mira tadpoles remains unknown. It might
be hypothesized that either C. mira reproduces outside
the rainy season (individuals were found at the beginning
of the rainy seasons, but no clutches or larvae were ob-
served), or that conditions in the particular habitat, due
to the presence of small trees, may provide slow-running
lotic conditions at rising water levels suitable for larval
development. However, the habitat of C. mariaelenae
has been described as the edge of small streams and
ditches in cloud forest (Cisneros-Heredia and McDiar-
mid 2007; Guayasamin et al. 2020), and one of us (PJV)
observed the species to be abundant in swampy habitats
in Sangay National Park, Ecuador, and the Cordillera de
Kampankis, northern Peru. Chimerella sp. from higher
altitudes in Dapartamento Amazonas, Peru, occurred in
both types of habitat, namely swampy areas along larg-
er streams as described for C. mira and at the edges of
small torrential streams. These observations indicate that
certain glassfrog species are more exible with respect to
habitat choice when compared to others (see also Rivera
and Folt 2018).
Apart from the now three nominal species in the ge-
nus Chimerella, we are currently unable to clarify the
taxonomic status of populations occurring in montane
rainforest at around 1800–1900 m a.s.l. in the Departa-
mentos Amazonas and San Martín (Fig. 11), which here
were tentatively referred to as Chimerella sp. Analyses of
mitochondrial markers would argue for conspecity with
C. corleone, but on the other hand we found no haplo-
type sharing in the nuclear marker studied herein (Rag-
1). Moreover, the collected specimens exhibit numerous
qualitative morphological dierences (e.g., colouration,
snout shape, dermal fringes) when compared to the three
recognized species of Chimerella, and undoubtedly these
populations would have been described as a separate spe-
cies without the genetic data available and applying pure-
ly morphological species criteria. Because of the striking
dierences in morphology, we are reluctant to conclude
that these frogs represent C. corleone, as this would prob-
ably constitute a singular case of disproportional poly-
morphism within a single species of glassfrog. As a con-
sequence, further in-depth studies of these populations
are necessary and will be subject of a future contribution.
Our ndings demonstrate the need for future research to
evaluate the taxonomic status of numerous populations of
glassfrogs, particularly in Peru, where they remain mark-
edly understudied (see Twomey et al. 2014).
We are grateful to the Servicio Nacional Forestal y de
Fauna Silvestre (SERFOR) for issuing all necessary
scientic permits (RGD 071-2020-MINAGRI-SER-
FOR-DGGSPFFS). We are deeply indebted to Jesse
Delia and Evan Twomey for sharing their knowledge,
providing call recordings and photographs of C. corleone
for comparison. We thank Joke Evenblij and Carla Hüb-
ner for their help with laboratory work. Beatriz Alvarez
Dorda and Santiago Castroviejo-Fisher kindly helped to
provide DNA extractions from the MNCN collection. We
are furthermore grateful to Diego F. Cisneros-Heredia,
Brian Folt, Evan Twomey, and an anonymous reviewer
for their time and valuable comments on the manuscript.
Jörn Köhler et al.: New species of Chimerella glassfrog
Batallas D, Brito J (2016) Análisis bioacústico de las vocalizaciones de
seis especies de anuros de la laguna Cormorán, complejo lacustre
de Sardinayacu, Parque Nacional Sangay, Ecuador. Revista Mex-
icana de Biodiversidad 87: 1292–1300.
Bossuyt F, Milinkovitch MC (2000) Convergent adaptive radiations in
Madagascan and Asian ranid frogs reveal covariation between lar-
val and adult traits. Proceedings of the National Academy of Sci-
ences of the United States of America 97: 6585–6590. https://doi.
Castillo-Urbina E, Glaw F, Aguilar-Puntriano C, Vences M, Köhler J
(2021) Genetic and morphological evidence reveal another new toad
of the Rhinella festae species group (Anura: Bufonidae) from the
Cordillera Azul in central Peru. Salamandra 57: 181–195. https://
Catenazzi A, Venegas PJ (2012) Anbios y reptiles/Amphibians and
reptiles. In: Pitman N, Ruelas Inzunza E, Alvira Reyes D, Vriesen-
dorp C, Moskovitz D, del Campo A, Wachter T, Stotz DF, Noningo
S, Tuesta E, Smith RC (Eds) Perú: Cerros de Kampankis. Rapid
Biological and Social Inventories Report 24: 106–117. [Spanish]
[260–271 [English]]
Chiari Y, Vences M, Vieites DR, Rabemananjara F, Bora P, Ramilijao-
na Ravoahangimalala O, Meyer A (2004) New evidence for parallel
evolution of colour patterns in Malagasy poison frogs (Mantella).
Molecular Ecology 13: 3763–3774.
Cisneros-Heredia DF (2009) Amphibia, Anura, Centrolenidae,
Chimerella mariaelenae (Cisneros-Heredia and McDiarmid, 2006),
Rulyrana avopunctata (Lynch and Duellman, 1973), Teratohyla
pulverata (Peters, 1873), and Teratohyla spinosa (Taylor, 1949):
Historical records, distribution extension and new provincial record
in Ecuador. Check List 5: 912–916.
Cisneros-Heredia DF, Guayasamin JM (2007) Amphibia, Anura, Centr-
olenidae, Centrolene mariaelenae: distribution extension, Ecuador.
Check List 2: 93–95.
Cisneros-Heredia DF, McDiarmid RW (2006) A new species of the
genus Centrolene (Amphibia: Anura: Centrolenidae) from Ecuador
with comments on the taxonomy and biogeography of glassfrogs.
Zootaxa 1244: 1–32.
Cisneros-Heredia DF, McDiarmid RW (2007) Revision of the characters
of Centrolenidae (Amphibia: Anura: Athesphatanura), with com-
ments on its taxonomy and the description of new taxa of glassfrogs.
Zootaxa 1572: 1–82.
Delia J, Bravo‐Valencia L, Warkentin KM (2017) Patterns of parental
care in Neotropical glassfrogs: eldwork alters hypotheses of sex‐
role evolution. Journal of Evolutionary Biology 30(5): 898–914.
Duarte-Marín S, Rada M, Rivera-Correa M, Caorsi V, Barona E,
González-Duran G, Vargas-Salinas F (2022) Tic, Tii and Trii calls:
Advertisement call descriptions for eight glass frogs from Colombia
and analysis of the structure of auditory signals in Centrolenidae.
Bioacoustics 32(2): 143–180.
Frost DR (2023) Amphibian Species of the World: an Online Reference.
Version 6.1 [accessed 4 May 2023] Electronic Database accessible
at American Museum of Natural History, New York. https://amphib-
Glaw F, Vences M (1997) Anuran eye colouration: denitions, variation,
taxonomic implications and possible functions. In: Böhme W, Bis-
cho W, Ziegler T (Eds) Herpetologia Bonnensis. SEH Proceedings,
Bonn, 125–138.
Guayasamin JM, Castroviejo-Fisher S, Trueb L, Ayarzagüena J, Rada
M, Vilà C (2008) Phylogenetic relationships of glassfrogs (Centro-
lenidae) based on mitochondrial and nuclear genes. Molecular Phy-
logenetics and Evolution 48: 574–595.
Guayasamin JM, Castroviejo-Fisher S, Trueb L, Ayarzagüena J, Rada
M, Vilà C (2009) Phylogenetic systematics of glassfrogs (Amphibia:
Centrolenidae) and their sister taxon Allophryne ruthveni. Zootaxa
2100: 1–97.
Guayasamin JM, Cisneros-Heredia DF, McDiarmid RW, Peña O, Hutter
CR (2020) Glassfrogs of Ecuador: diversity, evolution and conser-
vation. Diversity 12(6): e222.
Hrbek T, Larson A (1999) The evolution of diapause in the killish fam-
ily Rivulidae (Atherinomorpha, Cyprinodontiformes): a molecular
phylogenetic and biogeographic perspective. Evolution 53: 1200–
Hutter CR, Esobar-Lasso S, Rojas-Morales JA, Gutiérrez-Cárdenas
PDA, Imba H, Guayasamin JM (2013b) The territoriality, vocal-
izations and aggressive interactions of the red-spotted glassfrog,
Nymphargus grandisonae Cochran and Goin, 1970 (Anura: Centro-
lenidae). Journal of Natural History 47: 3011–3032.
Hutter CR, Guayasamin JM, Wiens JJ (2013a) Explaining Andean
megadiversity: the evolutionary and ecological causes of glassfrog
elevational richness patterns. Ecology Letters 16(9): 1135–1144.
Kalyaanamoorthy S, Minh BQ, Wong TKF, von Haeseler A, Jermiin LS
(2017) ModelFinder: Fast model selection for accurate phylogenet-
ic estimates. Nature Methods 14: 587–589.
Katoh K, Standley DM (2013) MAFFT multiple sequence alignment
software version 7: improvements in performance and usability. Mo-
lecular Biology and Evolution 30: 772–780.
Kocher TD, Thomas WK, Meyer A, Edwards SV, Pääbo S, Villablan-
ca FX, Wilson AC (1989) Dynamics of mitochondrial DNA evo-
lution in animals: Amplication and sequencing with conserved
primers. Proceedings of the National Academy of Sciences of the
United States of America 86: 6196–6200.
Köhler J, Castillo-Urbina E, Aguilar-Puntriano C, Vences M, Glaw F
(2022) Rediscovery, redescription and identity of Pristimantis neb-
ulosus (Henle, 1992), and description of a new terrestrial-breeding
frog from montane rainforests of central Peru (Anura, Straboman-
tidae). Zoosystematics and Evolution 98: 213–232. https://doi.
Köhler J, Jansen M, Rodríguez A, Kok PJR, Toledo LF, Emmrich M,
Glaw F, Haddad CFB, Rödel M-O, Vences M (2017) The use of
bioacoustics in anuran taxonomy: theory, terminology, methods and
recommendations for best practice. Zootaxa 4251: 1–124. https://
Evolutionary Systematics 7 2023, 195–209
Kumar S, Stecher G, Tamura K (2016) MEGA7: Molecular evolutionary
genetics analysis version 7.0 for bigger datasets. Molecular Biology and
Evolution 33: 1870–1874.
Librado P, Rozas J (2009) DnaSP. Version 5. A software for compre-
hensive analysis of DNA polymorphism data. Bioinformatics 25:
Martin AP (1999) Substitution rates of organelle and nuclear genes in
sharks: implicating metabolic rate (again). Molecular Biology and
Evolution 16: 996–1002.
McDiarmid RW (1994) Preparing amphibians as scienti c specimens.
In: Heyer WR, Donnelly MA, McDiarmid RW, Hayek L-AC, Foster
MS (Eds) Measuring and Monitoring Biological Diversity. Standard
Methods for Amphibians. Smithsonian Institution Press, Washing-
ton, 289–296.
Nguyen LT, Schmidt HA, von Haeseler A, Minh BQ (2015) IQ-TREE: a
fast and eective stochastic algorithm for estimating maximum-like-
lihood phylogenies. Molecular Biology and Evolution 32: 268–274.
Palumbi SR, Martin A, Romano S, McMillan WO, Stice L, Grabowski
G (1991) The Simple Fool’s Guide to PCR, Version 2.0. Privately
published, University of Hawaii.
Puillandre N, Brouillet S, Achaz G (2021) ASAP: assemble species by
automatic partitioning. Molecular Ecology Resources 21: 609–620.
Rivera N, Folt B (2018) Community assembly of glass frogs (Centro-
lenidae) in a Neotropical wet forest: a test of the river zonation
hypothesis. Journal of Tropical Ecology 34: 108–120. https://doi.
Salzburger W, Ewing GB, von Haeseler A (2011) The performance
of phylogenetic algorithms in estimating haplotype genealogies
with migration. Molecular Ecology 20: 1952–1963. https://doi.
Scherz MD, Glaw F, Hutter CR, Bletz MC, Rakotoarison A, Köhler J,
Vences M (2019) Species complexes and the importance of Data De-
cient classication in Red List assessments: The case of Hyloba-
trachus frogs. PLoS ONE 14(8): e0219437.
Stephens M, Smith NJ, Donnelly P (2001) A new statistical method for
haplotype reconstruction from population data. American Journal of
Human Genetics 68: 978–989.
Taboada C, Delia J, Chen M, Ma C, Peng X, Zhu X, Jiang L, Vu T,
Zhou Q, Yao J, O’Connell L, Johnsen S (2022) Glassfrogs conceal
blood in their liver to maintain transparency. Science 378(6626):
Terán-Valdez A, Guayasamin JM (2014) The tadpole of the glassfrog
Chimerella mariaelenae (Anura: Centrolenidae). CienciAmérica 3:
Twomey E, Delia J, Castroviejo-Fisher S (2014) A review of north-
ern Peruvian glassfrogs (Centrolenidae), with the description of
four new remarkable species. Zootaxa 3851: 1–87. https://doi.
Vences M, Kosuch J, Glaw F, Böhme W, Veith M (2003) Molecular
phylogeny of hyperoliid treefrogs: biogeographic origin of Mala-
gasy and Seychellean taxa and re-analysis of familial paraphyly.
Journal of Zoological Systematics and Evolutionary Research 41:
Vences M, Miralles A, Brouillet S, Ducasse J, Fedosov A, Kharchev V,
Kostadinov I, Kumari S, Patmanidis S, Scherz MD, Puillandre N,
Renner SS (2021) iTaxoTools 0.1: Kickstarting a specimen-based
software toolkit for taxonomists. Megataxa 6: 77–92. https://doi.
Vences M, Patmanidis S, Kharchev V, Renner SS (2022) Concatenator,
a user-friendly program to concatenate DNA sequences, implement-
ing graphical user interfaces for MAFFT and FastTree. Bioinformat-
ics Advances 2: vbac050.
... Fieldwork was conducted in November 2019 in different areas of the departments Huánuco and Ucayali in central Peru (see also Castillo-Urbina et al. 2021;Köhler et al. 2022Köhler et al. , 2023. Specimens were observed and collected during opportunistic searching at night using torches and headlamps, often guided by the sounds emitted by calling males. ...
Full-text available
We studied the taxonomic status of a population of Pristimantis from the southern Cordillera Azul, Departamento Huánuco, central Peru. A phylogenetic analysis based on the mitochondrial 16S rRNA gene revealed that it represents a lineage within the Pristimantis lacrimosus species group, being the closest relative of a species of uncertain taxonomic status from a lowland rainforest in central Peru (Panguana), and P. pulchridormientes from the Tingo Maria National Park. However, the focal lineage is divergent from all nominal species in the P. lacrimosus group for which respective data are available by >7.9% uncorrected pairwise distance in the 16S rRNA gene fragment. An integrative taxonomic approach, including morphological and bioacoustic analyses, provided multiple lines of evidence for the focal specimens belonging to an unnamed evolutionary lineage at the species level that we describe and name herein. The systematics of Peruvian populations associated with the P. lacrimosus group are discussed, particularly highlighting problematic taxa with uncertain taxonomic status and unknown relationships. We point to scientific challenges and actions needed to achieve a better taxonomic resolution of this species-rich clade of frogs.
Full-text available
Transparency in animals is a complex form of camouflage involving mechanisms that reduce light scattering and absorption throughout the organism. In vertebrates, attaining transparency is difficult because their circulatory system is full of red blood cells (RBCs) that strongly attenuate light. Here, we document how glassfrogs overcome this challenge by concealing these cells from view. Using photoacoustic imaging to track RBCs in vivo, we show that resting glassfrogs increase transparency two- to threefold by removing ~89% of their RBCs from circulation and packing them within their liver. Vertebrate transparency thus requires both see-through tissues and active mechanisms that "clear" respiratory pigments from these tissues. Furthermore, glassfrogs' ability to regulate the location, density, and packing of RBCs without clotting offers insight in metabolic, hemodynamic, and blood-clot research.
Full-text available
Motivation Phylogenetic and phylogenomic analyses require multi-gene input files in different formats, but there are few user-friendly programs facilitating the workflow of combining, concatenating or separating, aligning and exploring multi-gene data sets. Results We present Concatenator, a user-friendly GUI-driven program that accepts single-marker and multi-marker DNA sequences in different input formats, including Fasta, Phylip, and Nexus, and that outputs concatenated sequences as single-marker or multi-marker Fasta, interleaved nexus, or Phylip files, including command files for downstream model selection in IQ-TREE. It includes the option to (re)align markers with MAFFT and produces exploratory trees with FastTree. Although tailored for medium-sized phylogenetic projects, Concatenator is able to process phylogenomic data sets of up to 30,000 markers. Availability and implementation Concatenator is written in Python, with C extensions for MAFFT and FastTree. Compiled stand-alone executables of Concatenator for MS Windows and Mac OS along with a detailed manual can be downloaded from; the source code is openly available on GitHub ( Supplementary information Supplementary data are available at Bioinformatics Advances online.
Full-text available
The taxonomic status of the strabomantid frog species Pristimantis nebulosus (Henle, 1992), originating from the southern Cordillera Azul in central Peru, is investigated based on examination of the holotype and its comparison with freshly collected topotypic material. Following current standards, we provide a redescription of the holotype. It is in a rather poor state and exhibits certain damages and preservation artifacts, conditions that have hampered an allocation of this nominal taxon to any known living population of Pristimantis in the past. Our detailed specimen-to-specimen comparison provided morphological evidence for the conspecifity of the holotype and freshly collected topotypes. Molecular phylogenetic analysis, based on the mitochondrial 16S gene fragment places P. nebulosus in the P. conspicillatus species group, being closely related to P. bipunctatus and an undescribed candidate species from the Cordillera de Carpish. From both, P. nebulosus differs by considerable divergence in the 16S gene (p-distance 4.1–6.2%). Based on the specimens available, we provide an updated diagnosis for P. nebulosus , compare it to other species in the P. conspicillatus group and describe its advertisement call. In addition, we describe and name the closely related candidate species from the Cordillera de Carpish. It is sister to P. bipunctatus and P. nebulosus , but differs from both mainly by its tuberculate dorsal skin (versus shagreen) and divergence in the 16S gene (3.3–4.1%). We briefly discuss cryptic species diversity in the P. conspicillatus and P. danae species groups and provide justification for the description of a singleton species.
Full-text available
In anurans, vocalisations are the main behavioural modality of communication. Hence, the description of acoustic signals in anur-ans is important for understanding many aspects of their biology. We describe for the first time the advertisement calls for eight glass frog species (Centrolene antioquiensis, "Centrolene" robledoi, Nymphargus caucanus, N. chami, N. ignotus, N. rosada, N. spilotus, Sachatamia orejuela) and provide additional data on the recently described advertisement calls of Espadarana audax. In addition, we review the current knowledge of advertisement calls for all glass frog species (Centrolenidae). Based on the predominant temporal and the spectral structure, we identified three major types of calls in the family: 1) calls consisting of unpulsed short notes with amplitude modulation, similar to a 'Tic', 2) calls consisting of one long note (whistled) without amplitude modulation, similar to a 'Tii' and 3) calls consisting of pulsed or pulsatile notes, similar to a 'Trii'. We mapped these acoustic characters in the context of the evolutionary history of Centrolenidae. Descriptions presented here offer evidence to recognise most centrolenid calls using measurable characters in the field or laboratory. As such, we hope to stimulate future studies based on bioacoustical analysis in this widespread and highly diverse Neotropical clade.
Full-text available
While powerful and user-friendly software suites exist for phylogenetics, and an impressive cybertaxomic infrastructure of online species databases has been set up in the past two decades, software targeted explicitly at facilitating alpha-taxonomic work, i.e., delimiting and diagnosing species, is still in its infancy. Here we present a project to develop a bioinformatic toolkit for taxonomy, based on open-source Python code, including tools focusing on species delimitation and diagnosis and centered around specimen identifiers. At the core of iTaxoTools is user-friendliness, with numerous autocorrect options for data files and with intuitive graphical user interfaces. Assembled standalone executables for all tools or a suite of tools with a launcher window will be distributed for Windows, Linux, and Mac OS systems, and in the future also implemented on a web server. The initial version (iTaxoTools 0.1) distributed with this paper ( contains graphical user interface (GUI) versions of six species delimitation programs (ABGD, ASAP, DELINEATE, GMYC, PTP, tr2) and a simple threshold-clustering delimitation tool. There are also new Python implementations of existing algorithms, including tools to compute pairwise DNA distances, ultrametric time trees based on non-parametric rate smoothing, species-diagnostic nucleotide positions, and standard morphometric analyses. Other utilities convert among different formats of molecular sequences, geographical coordinates, and units; merge, split and prune sequence files, tables and species partition files; and perform simple statistical tests. As a future perspective, we envisage iTaxoTools to become part of a bioinformatic pipeline for next-generation taxonomy that accelerates the inventory of life while maintaining high-quality species hypotheses. The open source code and binaries of all tools are available from Github ( and further information from the website (
Full-text available
We studied the status of toads of the genus Rhinella collected in the southern Cordillera Azul, central Peru. Molecular analysis of the mitochondrial 16S rRNA gene revealed them to be members of the recently proposed Rhinella festae species group, and sister to R. lilyrodriguezae, a species known from northern areas of the Cordillera Azul. The new specimens are differentiated from R. lilyrodriguezae and other species of Rhinella by substantial genetic divergence in the studied gene fragment (> 5% uncorrected pairwise distance) and several qualitative morphological characters, providing combined evidence for a divergent evolutionary lineage. We consequently describe the specimens from the southern part of the Cordillera Azul in Departamento Huánuco as a new species, Rhinella chullachaki sp. n. We briefly discuss the definition and content of species groups in Rhinella as well as the difficulties hampering taxonomic resolution within this species-rich genus.
Full-text available
Glassfrogs (family: Centrolenidae) represent a fantastic radiation (~150 described species) of Neotropical anurans that originated in South America and dispersed into Central America. In this study, we review the systematics of Ecuadorian glassfrogs, providing species accounts of all 60 species, including three new species described herein. For all Ecuadorian species, we provide new information on the evolution, morphology, biology, conservation, and distribution. We present a new molecular phylogeny for Centrolenidae and address cryptic diversity within the family. We employ a candidate species system and designate 24 putative new species that require further study to determine their species status. We find that, in some cases, currently recognized species lack justification; specifically, we place Centrolene gemmata and Centrolene scirtetes under the synonymy of Centrolene lynchi; C. guanacarum and C. bacata under the synonymy of Centrolene sanchezi; Cochranella phryxa under the synonymy of Cochranella resplendens; and Hyalinobatrachium ruedai under the synonymy of Hyalinobatrachium munozorum. We also find that diversification patterns are mostly congruent with allopatric speciation, facilitated by barriers to gene flow (e.g., valleys, mountains, linearity of the Andes), and that niche conservatism is a dominant feature in the family. Conservation threats are diverse, but habitat destruction and climate change are of particular concern. The most imperiled glassfrogs in Ecuador are Centrolene buckleyi, C. charapita, C. geckoidea, C. medemi, C. pipilata, Cochranella mache, Nymphargus balionotus, N. manduriacu, N. megacheirus, and N. sucre, all of which are considered Critically Endangered. Lastly, we identify priority areas for glassfrog conservation in Ecuador.
Full-text available
We present the latest version of the Molecular Evolutionary Genetics Analysis (MEGA) software, which contains many sophisticated methods and tools for phylogenomics and phylomedicine. In this major upgrade, MEGA has been optimized for use on 64-bit computing systems for analyzing bigger datasets. Researchers can now explore and analyze tens of thousands of sequences in MEGA. The new version also provides an advanced wizard for building timetrees and includes a new functionality to automatically predict gene duplication events in gene family trees. The 64-bit MEGA is made available in two interfaces: graphical and command line. The graphical user interface (GUI) is a native Microsoft Windows application that can also be used on Mac OSX. The command line MEGA is available as native applications for Windows, Linux, and Mac OSX. They are intended for use in high-throughput and scripted analysis. Both versions are available from free of charge.
Full-text available
Taxonomy is the cornerstone of extinction risk assessments. Currently, the IUCN Red List treats species complexes either under a single overarching species name—resulting in an unhelpfully broad circumscription and underestimated threat assessment that does not apply to any one species lineage—or omits them altogether—resulting in the omission of species that should be assessed. We argue that taxonomic uncertainty alone, as in species complexes, should be grounds for assessment as Data Deficient (DD). Yet, use of the DD category is currently discouraged, resulting in assessments based on poor data quality and dismissal of the importance of taxonomic confidence in conservation. This policy may be leading to volatile and unwarranted assessments of hundreds of species across the world, and needs to be revised. To illustrate this point, we here present a partial taxonomic revision of torrent frogs from eastern Madagascar in the Mantidactylus subgenus Hylobatrachus. Two named species, Mantidactylus (Hylobatrachus) lugubris and M. (H.) cowanii, and several undescribed candidate species are recognised, but the application of the available names has been somewhat ambiguous. In a recent re-assessment of its conservation status, M. (H.) lugubris was assessed including all complex members except M. (H.) cowanii within its distribution, giving it a status of Least Concern and distribution over most of eastern Madagascar. After describing two of the unnamed lineages as Mantidactylus (Hylobatrachus) atsimo sp. nov. (from southeastern Madagascar) and Mantidactylus (Hylobatrachus) petakorona sp. nov. (from the Marojejy Massif in northeastern Madagascar), we show that Mantidactylus (Hylobatrachus) lugubris is restricted to the central east of Madagascar, highlighting the inaccuracy of its current Red List assessment. We propose to re-assess its status under a more restrictive definition that omits well-defined candidate species, thus representing the actual species to which its assessment refers, to the best of current knowledge. We recommend that for species complexes in general, (1) nominal lineages that can be confidently restricted should be assessed under the strict definition, (2) non-nominal species-level lineages and ambiguous names should be prioritised for taxonomic research, and (3) ambiguous names should be assessed as DD to highlight the deficiency in data on their taxonomic status, which is an impediment to their conservation. This would reduce ambiguity and underestimation of threats involved in assessing species complexes, and place the appropriate emphasis on the importance of taxonomy in anchoring conservation.
We describe ASAP (Assemble Species by Automatic Partitioning), a new method to build species partitions from single locus sequence alignments (i.e. barcode datasets). ASAP is efficient enough to split datasets as large 104 sequences into putative species in several minutes. Although grounded in evolutionary theory, ASAP is the implementation of a hierarchical clustering algorithm that only uses pairwise genetic distances, avoiding the computational burden of phylogenetic reconstruction. Importantly, ASAP proposes species partitions ranked by a new scoring system that uses no biological prior insight of intra‐specific diversity. ASAP is a stand‐alone program that can be used either through a graphical web‐interface or that can be downloaded and compiled for local usage. We have assessed its power along with three others programs (ABGD, PTP and GMYC) on 10 real COI barcode datasets representing various degrees of challenge (from small and easy cases to large and complicated datasets). We also used Monte‐Carlo simulations of a multi‐species coalescent framework to assess the strengths and weaknesses of ASAP and the other programs. Through these analyses, we demonstrate that ASAP has the potential to become a major tool for taxonomists as it proposes rapidly in a full graphical exploratory interface relevant species hypothesis as a first step of the integrative taxonomy process.