Extreme tadpoles: The morphology of the fossorial megophryid larva, Leptobrachella mjobergi
The bizarre larvae of Leptobrachella mjobergi are fossorial and live in the gravel beds of small streams. These tadpoles are vermiform in body shape. Here we present details on their skeleton and musculature, particularly of the head. The entire cranium and its associated musculature are reconstructed in three dimensions from serial histological sections. The hyobranchial apparatus is highly reduced. The head of the L. mjobergi larva is more mobile than in other anuran species. This mobility can largely be ascribed to the exclusion of the notochord from the cranial base and an articulation of the foramen magnum floor with the atlas of the tadpole. The articulation is unique among anuran species, but design parallels can be drawn to salamanders and the articulation between atlas and axis in mammals. In L. mjobergi, the atlas forms an anterior dens that articulates with the basal plate in an accessory, third occipital articular face. The muscle arrangements deviate from the patterns found in other tadpoles: For instance, epaxial and ventral trunk muscles reach far forward onto the skull. The post-cranial skeleton of L. mjobergi is considerably longer than that of other anurans: it comprises a total of 35 vertebrae, including more than 20 post-sacral perichordal centra. Despite a number of features in cranial and axial morphology of L. mjobergi, which appear to be adaptations to its fossorial mode of life, the species clearly shares other features with its megophryid and pelobatid relatives.
Zoology 109 (2006) 26–42
Extreme tadpoles: The morphology of the fossorial megophryid larva,
, Stefan Hertwig
, Indraneil Das
Biozentrum Grindel und Zoologisches Museum, Martin-Luther-King-Platz 3, D-20146 Hamburg, Germany
¨r Spezielle Zoologie und Evolutionsbiologie, Erbertstr. 1, D-07743 Jena, Germany
Institute of Biodiversity and Environmental Conservation, Universiti Malaysia Sarawak, 94300 Kota Samarahan, Sarawak,
Received 15 June 2005; received in revised form 9 September 2005; accepted 30 September 2005
The bizarre larvae of Leptobrachella mjobergi are fossorial and live in the gravel beds of small streams. These
tadpoles are vermiform in body shape. Here we present details on their skeleton and musculature, particularly of the
head. The entire cranium and its associated musculature are reconstructed in three dimensions from serial histological
sections. The hyobranchial apparatus is highly reduced. The head of the L. mjobergi larva is more mobile than in other
anuran species. This mobility can largely be ascribed to the exclusion of the notochord from the cranial base and an
articulation of the foramen magnum ﬂoor with the atlas of the tadpole. The articulation is unique among anuran
species, but design parallels can be drawn to salamanders and the articulation between atlas and axis in mammals. In
L. mjobergi, the atlas forms an anterior dens that articulates with the basal plate in an accessory, third occipital
articular face. The muscle arrangements deviate from the patterns found in other tadpoles: For instance, epaxial and
ventral trunk muscles reach far forward onto the skull. The post-cranial skeleton of L. mjobergi is considerably longer
than that of other anurans: it comprises a total of 35 vertebrae, including more than 20 post-sacral perichordal centra.
Despite a number of features in cranial and axial morphology of L. mjobergi, which appear to be adaptations to its
fossorial mode of life, the species clearly shares other features with its megophryid and pelobatid relatives.
r2005 Elsevier GmbH. All rights reserved.
Keywords: Pelobatidae; Megophryidae; Anuran larvae; Cranial morphology; Cranial musculature; Fossorial tadpole
With over 5000 extant species, anurans are the most
successful group of lissamphibians (Frost, 2004). The
remarkable diversity of reproductive modes, larval
forms and adaptive strategies exhibited by anurans are
unquestionably major determinants of their evolution-
ary success (McDiarmid and Altig, 1999). Ecomorpho-
logical guilds have been described containing distantly
related species indicating much of convergence in the
evolution of adaptive types and resource use. In
different regions of the world equivalent microhabitats
are used by similar larval ecomorphs belonging to
different frog taxa (Orton, 1953;Altig and Johnston,
1989;Altig and McDiarmid, 1999a).
An elongate and slender larval ecomorph has evolved
in the megophryid genus Leptobrachella (Inger, 1983,
ARTICLE IN PRESS
0944-2006/$ - see front matter r2005 Elsevier GmbH. All rights reserved.
E-mail address: email@example.com (A. Haas).
1985). Species of Leptobrachella are restricted to Borneo
and the Natuna Island where the adults live along river
banks (Dring, 1983;Inger, 1983;Inger and Stuebing,
1991). Currently, seven species are recognized (Frost,
2004). The body can be described as vermiform or eel-
like, and has an almost seamless transition from the
narrow, cylindrical trunk into the strong tail. The tail ﬁn
is very low (Fig. 1). The combination of these unusual
external features is indicative of a fossorial life style. The
external morphology and details of the bucco-pharyn-
geal cavity of Leptobrachella mjobergi larvae have been
described by Inger (1983, 1985). A similarly elongate
body form is also known from the closely related
Leptolalax (Inger, 1985). Apart from megophryids,
slender tadpoles have been described for unrelated taxa,
such as the ranid Staurois (Inger and Wassersug, 1990;
Malkmus et al., 1999), and centrolenids (Mijares-
Urrutia, 1990;Jaramillo et al., 1997;Ibanez et al.,
1999;Noonan and Bonett, 2003;Altig and McDiarmid,
We collected L. mjobergi tadpoles from a small stream
in a secondary forest patch, within a largely cultivated
landscape. Tadpoles were found among the superﬁcial
layer of gravel, particularly in rifﬂes in the middle of the
stream (see Fig. 1 with gravel from the collection site).
Syntopic amphibian fauna included Ichthyophis sp.,
Staurois guttatus,Pelophryne signata,Ansonia minuta,
Megophrys nasuta,Limnonectes malesianus, and Nyctix-
The morphology of the musculo-skeletal apparatus in
anuran tadpoles has been reviewed recently (Cannatella,
ˇek, 2003). The musculo-skeletal architecture
of the cranium is profoundly different in the tadpole and
frog stage (Gaupp, 1893). The one is transformed into
the other in a dramatic metamorphosis (de Jongh, 1968).
Although interspeciﬁc variation in tadpole morphology
has been documented and its usefulness for phylogenetic
reconstruction proven for a variety of species (Pu´ gener
et al., 2003;Haas, 2003), the limits to which the general
tadpole body plan has been modiﬁed in evolution in
various groups have not been explored at depth. The
study of tadpoles with extreme life histories is essential
to understand innovative adaptations and alterations in
An account on cranial skeletal and muscular features
of L. mjobergi tadpoles is not available. In the present
study, we give a detailed description of the cranial
musculoskeletal system of L. mjobergi and address the
following questions: (1) Which features distinguish the
tadpole of L. mjobergi from other anurans? (2) Can
these features be related with a fossorial lifestyle and a
particular feeding mode? (3) Finally, do the special
features of L. mjobergi obscure its phylogenetic related-
ness to other megophryids? Although cranial morphol-
ogy is the main focus, we will give preliminary data on
Materials and methods
Specimens of L. mjobergi tadpoles were collected from
a small stream within a secondary forest at the foothills
of Gunung Penrissen (N 0110902600 , E 11011702900 ) near
Anna Rais village, Sarawak State, Malaysia (Western
Borneo). Adults of L. mjobergi were found calling in
numbers along the banks of the brook. Assignment of
larvae to the species was based on the descriptions
by Inger (1983, 1985), and the abundance of adult
L. mjobergi in the habitat.
Tadpoles were anesthetized and killed in chlorobuta-
nol (Sigma T-5138), ﬁxed and stored in neutral buffered
formalin (4%). Four sources of information were used
for this study: (1) plain preserved specimens (external
characters); (2) serially sectioned specimens (soft tissue,
musculature and skeletal characters); (3) manually
ARTICLE IN PRESS
Fig. 1. Movements of living Leptobrachella mjobergi. Two
individuals (a and b/c) were captured and transferred to an
aquarium. The specimen in (a) shows strong dorsal extension
of the trunk vertebral column. The specimen in (b) and (c)
demonstrates burrowing abilities in (b) and ventral ﬂexion of
head in (c). The gravel in (b) and (c) was taken from the
collection site at Annah Rais.
A. Haas et al. / Zoology 109 (2006) 26–42 27
dissected specimens (soft tissue, musculature); and (4)
cleared and stained whole-mount preparations (mostly
skeletal characters). Whole-mount specimens were
processed according to the protocol of Taylor and van
For serial sectioning, two specimens were decalciﬁed
in 2% ascorbic acid (Dietrich and Fontaine, 1975),
dehydrated, embedded in parafﬁn and sectioned at 7
and 8 mm thickness, respectively. Two other specimens
were embedded in Historesin
(Leica) and sectioned at
3mm. Sectioning was performed with a Microm HM360
rotary microtome equipped with the Microm water
transfer system for parafﬁn, and a glass knife for resin
sectioning, respectively. Parafﬁn sections were stained
with Heidenhain’s Azan (Bo
¨ck, 1989); plastic resin
sections in methylene blue and basic fuchsin solution
(6 parts ethanol 100%, 4 parts methylene blue 0.13%, 3
parts basic fuchsin 0.13%, 8 parts sodium borate 1%).
Specimens for manual dissection were prepared by
applying only the ﬁrst staining step (Alcian blue) of the
clearing and staining protocol with subsequent rinsing
and transfer to 70% ethanol. This procedure contrasts
dark blue stained cartilages against white muscles.
Drawings were made either from digital photographs
(Zeiss SV11 stereomicroscope equipped with a digital
video camera ColorView 12, software analySIS
Soft Imaging System GmbH, Germany) or with a
camera lucida on a Zeiss SV11.
Data on individuals examined are summarized in
Table 1. The serially sectioned specimen BrachellaM1
was chosen for a three-dimensional reconstruction of the
complete cranium and associated musculature. Only the
muscles of the right side of the body were reconstructed.
Every second histological section was photographed
digitally with a Canon Powershot S50 camera mounted
on a Leica MZ 9.5 stereomicroscope and connected to a
Apple Macintosh G5. Similar to an account given in
Haas and Fischer (1997), digital images were imported
as background image planes into Alias-Wavefront
5.01 software. Contour lines of bones, cartilages
and muscles were digitized manually from image planes.
Contour lines were aligned at the proper distance to
each other and moved and rotated manually in the
transverse plane to generate best ﬁt. A cleared specimen
was used as additional reference for alignment. Subse-
quently, surfaces were built starting with shape primi-
tives and forming and reﬁning them with various tools
to closely ﬁt the surfaces to the contour line
stacks. Reliable reconstruction is impossible when a
muscle has widely spaced ﬁbers or ﬁber bundles and
when the artifacts and distortions due to histological
sectioning are much greater than the ﬁbers’ diameters.
This was particularly true for the extremely ﬁne and
spaced ﬁbers of the m. interhyoideus posterior. This
muscle was not reconstructed. In the m. levator arcuus
branchialis (I+II) the muscle’s ﬁbers were approxi-
mated in position, length, and number as closely as
possible. Others, namely the m. mandibulolabialis and
the m. transversus ventralis IV had less ﬁber spacing
than the previous two muscles and were reconstructed as
closed surface for simplicity. Final 3D surface recon-
structions were rendered with Maya
The anatomical terminology largely follows the
summary in Roc
ˇek (2003) for skeletal structures and
Haas (1997, 2001, 2003) for hyobranchial and jaw
musculature. Post-cranial muscles were identiﬁed ac-
cording to Gaupp (1896).
A supplementary movie ﬁle is available on the
journal’s pages (www.elsevier.de/zool) for download.
The movie shows an animation of the computer-
reconstructed skull of a Leptobrachella mjobergi larva
(Quicktime 7 required).
The body is vermiform. Head and trunk are oblong,
and together account for up to one-third of the total
length (Fig. 1). The trunk is approximately cylindrical in
cross section but the head is bluntly conical and slightly
depressed. The trunk is capable of considerable hyper-
extension (Fig. 1a). The pigmentation is uniformly dark
brown above. Small specimens o20 mm are only faintly
brown with overall white-bluish to pinkish appearance.
The venter is light blue in life and the gut coils are only
ARTICLE IN PRESS
Table 1. List of specimens examined for this study
Study identiﬁcation Head–body length Total length Preparation
BrachellaM 1 9.0 27.4 Serial section, parafﬁn, 3D
BrachellaM 2 7.5 22.3 Serial section, parafﬁn
21.0 Serial section, plastic resin
17.0 Serial section, plastic resin
BrachellaM 5 9.0 26.6 Micro-dissection
BrachellaM 6 8.8 25.5 Cleared and stained
¼no measurements available.
A. Haas et al. / Zoology 109 (2006) 26–4228
faintly visible. The oral disk is distinctly protruding
antero-ventrally and funnel-shaped with deep dorsal
and ventral disk emarginations that incompletely sub-
divide the oral funnel into right and left halves. The oral
disk is without pigmentation; keratodonts are absent.
The marginal papillae stand in an uniserial line at the
margin of the oral disk. Some submarginal papillae are
present around the mouth oriﬁce. The naris is closer to
the snout than to the eye. The eyes are positioned
dorsolaterally and are relatively small and sunken in;
they hardly protrude beyond the contour of the body.
The spiraculum is sinistral, long and tubular, clearly
protruding from the body wall. The vent is dextral. The
dorsal ﬁn originates slightly posterior to the trunk–tail
junction. It is low on the proximal half of the tail,
expanding only posteriorly. The tip of the tail is blunt.
The muscular part of the tail is strong and myotomes are
clearly visible. The specimens examined ranged from 17
to 27.5 mm in total length and would fall into stage 25
according to Gosner (1960) (Table 1). Application of
this staging table, however, is not meaningful in
L. mjobergi, because hind limb development appears
to be more delayed relative to growth and the
development of internal structures than in most other
tadpole species. The maximum head–body length is
13.8 mm (Inger, 1983).
Neurocranium and ﬁrst visceral arch
The neurocranium is oblong, narrow, and approxi-
mately parallel sided (Fig. 2). The lateral contours do
not expand but rather narrow at the posterior part of
the palatoquadrate. The widest point of the neurocra-
nium is at the mid-level of the ovoid otic capsules. Arcus
occipitalis and tectum synoticum form the foramen
magnum and connect the otic capsules to the planum
basale and to each other. The arcus occipitalis also
delimits the foramen jugulare posteriorly. The otic
capsule is well chondriﬁed and has four foramina: the
foramen endolypmphaticum inferius posteriorly, the
foramen endolymphaticum superius posteromedially,
the foramen acusticum medially close to the planum
basale, and the foramen endolymphaticum dorsally in
the medial capsular wall. The fenestra ovalis is formed
but no operculum is present. The medial wall of the otic
capsules is conﬂuent with the planum basale.
The posterior margin of the planum basale is deeply
emarginated. The notochord does not enter the planum.
The planum is approximately as wide as long. Ante-
riorly it gives rise to the trabeculae cranii. They form an
elliptical arch and meet anteriorly at the planum
trabeculare anticum (Figs. 2–4). The trabeculae cranii
encircle the fenestra intertrabeculare, which is unchon-
driﬁed. However, the bony ring of the parasphenoid ﬁts
into the intertrabecular opening.
The cornua trabeculae originate from the planum
trabeculare anticum anteriorly (Figs. 2 and 3). They are
fused for most of their lengths. Distally they diverge,
curve laterad and form a pronounced lateral process for
articulation with the cartilago labialis superior. None of
the specimens has chondriﬁed nasal structures.
Anteriorly, the palatoquadrate is connected to the
trabecula cranii by the commissura quadrato-cranialis
anterior. Posteriorly, the upward curving processus
ascendens quadrati connects to the pila antotica of the
cranial sidewall (Figs. 2 and 5). A processus oticus is
absent. The palatoquadrate is widest anteriorly and thin
and slender at the processus ascendens. Anteriorly, the
pars articularis quadrati—mostly its lateral margin—
articulates with the cartilago meckeli. The palatoqua-
drate articulates with the ceratohyale ventrally (Figs. 4
The processus muscularis quadrati is visible most
clearly in lateral view (Fig. 5). It is relatively narrow at
the tip. A commissura quadrato-orbitalis, connecting
the process’s tip to the sidewall of the braincase, is
absent. The tip of the processus muscularis extends
anteriorly. Its posterior edge is ﬂat, almost horizontal in
orientation, whereas the anterior edge is steep and
overhanging. The processus muscularis bears a promi-
nent lateral process close to its anterior margin.
This process borders the m. orbitohyoideus anteriorly
The sidewall of the cavum cranii is only weakly
chondriﬁed. The processus ascendens clearly connects to
the pila antotica above the center level of the foramen
oculomotorii (Fig. 5), i.e., a high processus ascendens is
present in the species. An arch of cartilage originates
from the pila antotica anteriorly and curves ventrad to
connect to the trabecula cranii and encircle the foramen
oculomotorii. The sidewall does not form a foramen
opticum. The nervus opticus simply passes through the
membraneous lateral wall of the cavum cranii anterior
to the foramen oculomotorii. The pila antotica gives rise
to anterior and posterior processes dorsally. Both form
the upper marginal cartilages of the sidewall. Fronto-
parietals have not yet formed.
The lower jaw is segmented into two functional units, the
cartilago meckeli laterally and the cartilago labialis inferior
(or infrarostral cartilage) medially. Meckel’s cartilage is
robust and conspicuously convex dorsally (Fig. 7). It forms
the processes present in most anuran larvae: dorsomedial
and ventromedial processes as part of the articulation with
the infrarostral cartilage, and the processus retroarticularis
as insertion point for the angularis-muscle group. The
anterior face of the cartilago meckeli is deeply concave. The
cartilago labialis inferior is U-shaped in frontal view, with a
broad and ﬂat middle part. Both infrarostral cartilages are
connected by a medial symphysis.
ARTICLE IN PRESS
A. Haas et al. / Zoology 109 (2006) 26–42 29
The movable upper jaw, cartilagines labiales super-
iores (or suprarostral cartilages) are U-shaped in ventral
view (Figs. 2 and 4). Right and left suprarostrals are
fused in the midline but bear a clearly discernible suture
(Fig. 4). There is no sign of subdivision of the cartilago
labialis superior into a pars corporis and pars alaris as in
many other frog species. Distally, the cartilage forms a
broad and blunt posterodorsal process. It articulates
syndesmotically with the adrostral cartilage (Figs. 2 and
3). The latter is a smoothly L-shaped body of cartilage
positioned dorsolateral to the posterodorsal process and
anterodorsal to the cartilago meckeli.
The hyobranchial skeleton (Fig. 7) is remarkable in
many respects. The axis running from the processus
anterior to the processus posterior of the ceratohyale is
long, whereas the processus lateralis hyalis is short in
comparison. This geometry gives the ceratohyale an
overall oblique orientation (Fig. 7). The processus
lateralis hyalis is broadly expanded dorsoventrally
(Fig. 5), forming an articular condylus for articulation
with the palatoquadrate at its dorsal edge. Anterolateral
processes are not formed at the ceratohyalia.
The elongation of the basibranchiale parallels the
antero-posterior extension of the ceratohyale. The
basibranchiale connects by synchondroses (stained only
faintly by Alcian blue) to the plana hypobranchiales
posteriorly. A processus urobranchialis on the ventral
side of the basibranchiale is absent.
The plana hypobranchiales are fused medially (synch-
ondrosis). The posterior, tapering end of each planum
continues posteriorly in a rod of cartilage that, after a
short distance, expands into spiculum IV that is ﬂat and
bears four horizontal projections (Fig. 7).
Ceratobranchiale I originates from the lateral margin
of the planum hypobranchiale. It forms a broad
processus anterior branchialis that extends into the
ARTICLE IN PRESS
Fig. 2. Drawing of a cleared and double-stained specimen (study ID BrachellaM6): (a) ventral view, and (b) dorsal view of
neurocranium, ﬁrst visceral arch, and ﬁrst three vertebrae.
A. Haas et al. / Zoology 109 (2006) 26–4230
posterior concavity of the ceratohyale. Beyond the
processus anterior branchialis the ceratobranchiale I
continues laterally as a very slender rod of cartilage bent
in an S-shape (Fig. 7). All four ceratobranchialia are
thin and end freely distally; commissurae terminals and
lateral projections (common in other species) are absent,
and there is no commissura proximalis I.
Ceratobranchialia II and III are fully conﬂuent
proximally, i.e., the commissura proximalis II is present.
Both ceratobranchialia form a broad U-shaped arch,
which is open distally. The proximal base of the arch
connects to the posterior prolongation of the planum
hypobranchiale (Figs. 4 and 7). At the same transverse
level, a strong processus branchialis projects ventrally
from the connected ceratobranchials and bends medially
with its tip.
The spicula are unique in several features. The
proximal end of spiculum I is in touch with, but does
not fuse conﬂuently with, ceratobranchiale I. In arching
posterolaterally, spiculum I gets close to ceratobran-
chiale II. Spiculum II is connected to ceratobranchiale II
and ceratobranchiale I only by perichondrial contact.
Spiculum III likely is the most anterolateral, ﬁnger-like
projection from the plate extension of spiculum IV.
Ceratobranchiale IV descends from the ventral side of
the spiculum IV-plate. After a short distance it bends
laterally and soon meets ceratobranchiale III. Both are
connected by the ﬁbers of their perichondrium but they
are not conﬂuent.
Although the condylus of the ceratohyale is located
relatively far posteriorly, the long ceratohyalia ﬁll much
of the space between the partes articulares quadrati
(Fig. 4). The branchial part of the hyobranchial
apparatus is not the basket-shaped structure commonly
found in other species; rather, it is small and ﬂat (Figs. 4,
5 and 10). It hardly goes beyond the otic capsule
contours in ventral view (Fig. 4).
Haas and Richards (1998) proposed 16 landmarks to
estimate the relative contribution of the ceratohyal,
hypobranchial plate, and branchial basket to the bucco-
pharyngeal ﬂoor area. In L. mjobergi, the total area is
composed of 42% ceratobranchial (CH), 18% hypo-
branchial (HB), and 40% branchial basket (BB) area
(specimen BrachellaM6). Thus, BB is relatively reduced
and CH relatively enlarged in L. mjobergi (24–28% and
55–62%, respectively, in generalized Litoria tadpoles;
Haas and Richards, 1998). The lever arm ratio of the
ceratohyal (b/a, Fig. 8;Wassersug and Hoff, 1979) is 0.3
when measured perpendicular to the longitudinal body
axis, but approx. 0.4 when measured along an oblique
ARTICLE IN PRESS
Fig. 3. Three-dimensional computer reconstruction from serial histological sections of specimen BrachellaM1. Cranium, including
hyobranchial apparatus, and atlas in dorsal views. In (a) the complete cranial musculature is shown. In (b) superﬁcial muscles are
hidden to expose some deep muscles.
A. Haas et al. / Zoology 109 (2006) 26–42 31
axis (x,Fig. 8). The geometry of the ceratohyale
(a¼661;Fig. 8) and the oblique orientation of the m.
orbitohyoideus (Fig. 6) suggest an oblique axis of
rotation of the ceratohyale.
The notochord is the major element in the larval axial
skeleton. Unlike typical pond tadpoles from other
ARTICLE IN PRESS
Fig. 4. Three-dimensional computer reconstruction from serial histological sections of specimen BrachellaM1. Cranium, including
hyobranchial apparatus and atlas, in ventral views: (a) all muscles shown, and (b) superﬁcial muscles made invisible.
Fig. 5. Three-dimensional computer reconstruction from serial histological sections of specimen BrachellaM1. Cranium, including
hyobranchial apparatus and atlas, in lateral view.
A. Haas et al. / Zoology 109 (2006) 26–4232
species, the notochord in L. mjobergi does not project
into the planum basale of the neurocranium (Figs. 3 and
4); rather, it ends within the ﬁrst presacral vertebra, the
atlas. The anterior tip of the notochord lies in a
prominent, ossiﬁed anterior process of the atlas. The
process has similarity to the dens axis of amniotes and is
considered a dens analog (Fig. 4). This dens is fastened
to the planum basale by a ligamentum transversum
ARTICLE IN PRESS
Fig. 6. Three-dimensional computer reconstruction from serial histological sections of specimen BrachellaM1. Cranium, including
hyobranchial apparatus and atlas, in lateral views. In (a) all cranial muscles are shown, and in (b) some superﬁcial muscles are
hidden to expose deeper muscles.
A. Haas et al. / Zoology 109 (2006) 26–42 33
occipitalis (new term for tadpoles) that runs over its tip
(not shown in ﬁgures).
The atlas forms two relatively large anterolateral
processes (proc. articulares; atlantal cotyles) to arti-
culate with the arcus occipitalis of the neurocranium
(Figs. 3 and 4). The faces of the processus articulares are
concave anteriorly to match the convex shape of the
arcus’ condyle. The dens of the atlas is the third, medial
articulation of the atlas with the neurocranium.
In specimen BrachellaM6, transverse processes are
present on vertebrae II and III. The neural arches of the
atlas arise from the dorsolateral side of the centrum,
posterior to the processus articulares. The atlas’ neural
arches are devoid of transverse processes. The neural
arches’ dorsal ends bend abruptly posteriorly and are
broad and cartilaginous. Their posteroventral corners
form postzygapophysial articulations with the prezyga-
pophysis of the subsequent vertebra. In specimen
BrachellaM6 (Fig. 9), only presacral vertebrae V and
VI form closed neural arch rings around the vertebral
canal. All other neural arches are open dorsomedially at
Vertebral centra vary in size along the antero-poster-
ior axis. The atlas is shortest and vertebral length
increases in the subsequent vertebrae (Fig. 9). As in
most other frogs, vertebra IX is the prospective
postmetamorphic sacral vertebra. Its neural arch has a
prezygapophysis to articulate with vertebra VIII but no
postzygapophysis to vertebra X. From vertebra IX on,
ten large centra follow that are notably increased in
height. The following more posterior vertebrae diminish
in size gradually and the last one is only a faintly stained
ossiﬁcation at the perichordal sheath. In specimen
BrachellaM6, there are 35 vertebral centra ossiﬁcations
(Fig. 9). The notochord is fully encircled by the vertebral
ossiﬁcations (perichordal formation).
The m. mandibulolabialis comprises superior and
inferior parts, which extend into the anterior and
posterior parts of the funnel-shaped oral disk, respec-
tively. Keratodonts are absent, thus the muscle’s ﬁbers
do not insert in keratodont ridges; rather, ﬁbers run
toward the marginal area of the oral disk. Distally, the
ﬁbers are single layered and slightly spaced (recon-
structed as a closed sheath in the 3D model for
simplicity; Figs. 4 and 6). The m. mandibulolabialis
inferior originates with bundles of ﬁbers from two sites:
the posterior side of the processus ventromedialis and
the anterior face concavity of Meckel’s cartilage, lateral
to the cartilago labialis inferior. The m. mandibulola-
bialis superior originates lateral to the inferior part,
somewhat dorsal on the anterior face of cartilago
meckeli (Figs. 6 and 7).
A m. submentalis is absent. The m. intermandibularis
originates relatively far laterally from the anterior face
of cartilago meckeli, immediately ventral to the origin of
the m. mandibulolabialis superior. The bundle of ﬁbers
curves around cartilago meckeli ventrally. The muscle
ﬂattens out distally, approximately ventral to the pars
reuniens (Fig. 4). The m. intermandibularis meets its
counterpart from the other side in a median raphe of
The m. levator mandibulae internus arises from the
anteroventral face of the otic capsule and adjacent
planum basale (parachordal) (Figs. 3, 6 and 10). It runs
anteriorly over the fenestra subocularis. It crosses the
pars articularis quadrati dorsally but is ventral to all
other muscles of the levator series. It inserts on the most
lateral prominence of cartilago meckeli (Fig. 6).
The m. levator mandibulae longus superﬁcialis is
located in an unusual medial position. Three heads
originate posteriorly, unite more anteriorly, and ﬁnally
insert via a short tendon on the processus dorsomedialis
meckeli (Fig. 6). The origins are: (1) lateral side of
trabeculae cranii (posterior to level of pila antotica); (2)
pila antotica, processus ascendens and cupula anterior
of otic capsule; and (3) connective tissue dorsolateral to
otic capsule and fascia of m. levator mandibulae longus
The m. levator mandibulae longus profundus is the
largest of the jaw levators (Fig. 3). Its superﬁcial ﬁbers
are almost conﬂuent posteriorly with the paravertebral
ARTICLE IN PRESS
Fig. 7. Three-dimensional computer reconstruction from serial
histological sections of specimen BrachellaM1. Lower jaw
cartilages and muscles originating from it. Frontal view.
A. Haas et al. / Zoology 109 (2006) 26–4234
system. The longissimus tract and the m. lev. mand.
profundus are separated only by a tendinous inscription.
Deep bundles of the profundus originate from the otic
capsule and the posterior parts of the palatoquadrate,
including the processus ascendens. The internal ﬁber
architecture of the profundus is characterized by loosely
spaced bundles of ﬁbers (Fig. 10). This does not seem to
be a histological artifact as the feature is present in all
specimens sectioned, both parafﬁn and plastic resin
embedded, as well as in different larval sizes. The ﬂeshy
part of the profundus ends approximately at the level of
the jaw joint from where it continues as a long tendon
(Fig. 6) that runs over the cartilago meckeli, bends
ventromediad, posterior to the adrostral, and passes
medial to the m. mandibulolabialis to attach at the
posteroventral margin of the cartilago labialis inferior.
The m. levator mandibulae articularis is a short and
thin muscle that originates from the dorsal side of the
palatoquadrate at the base of the processus muscularis
(Fig. 3). It attaches to the lateral part of cartilago
meckeli, dorsal to the insertion of the m. lev. mand.
internus (Fig. 6) and crosses the latter.
The m. levator mandibulae externus is short and thin.
It originates from the medial side of the processus
muscularis quadrati (Fig. 3). From there it extends
anteriorly for a short distance to merge with the tendon
of the m. levator mandibulae longus profundus. The m.
levator mandibulae lateralis is absent.
The angularis group is part of the depressor
mandibulae group (cranial nerve VII innervation).
Three muscles of different lengths and orientations
belong to this group; all of them insert on the processus
retroarticularis meckeli (Figs. 4 and 6). The m. quad-
ratoangularis is the most ventromedial of the three. It
originates from the ventral side of the pars articularis
quadrati (Fig. 4). It originates immediately anterior to
ARTICLE IN PRESS
Fig. 9. Vertebral column. Drawing of a cleared and double-stained specimen (study ID BrachellaM6). Note the number and sizes of
vertebrae. Arrow points to future sacral vertebra (IX). Bone stippled, cartilage shaded.
Fig. 8. Hyobranchial apparatus in dorsal view. Drawing of a cleared and double-stained specimen (study ID BrachellaM6). The
ceratohyalia and basibranchiale are particularly long in the anterior–posterior axis. The ceratobranchialia are thin curved rods of
cartilage without any lateral projections. The ceratobranchialia do not form commissurae terminales at their ends (see text for
details). Lever arm ratio: b/a.
A. Haas et al. / Zoology 109 (2006) 26–42 35
the quadrato-hyal articulation. The insertion is ﬂeshy,
without a notable tendon, at the ventromedial side of
the processus retroarticularis. The m. hyoangularis is the
strongest muscle of the group as estimated from cross-
sectional area. Several more or less separate tracts of
ﬁbers originating from the ceratohyale belong to this
muscle (Fig. 4). The m. suspensorioangularis has its
ﬂeshy origin along the lateral side of the processus
muscularis quadrati (Fig. 6). In lateral view it is covered
by the mm. orbitohyoideus and suspensoriohyoideus.
The insertion of the suspensorioangularis is mediated by
a long tendon that merges with the tendon of the m.
hyoangularis and inserts ventrally at the processus
retroarticularis of cartilago meckeli.
Despite belonging to the depressor mandibulae group
by innervation (cranial nerve VII) and metamorphic
fate, the mm. orbitohyoideus and suspensoriohyoideus
(Fig. 6) act exclusively on the ceratohyale in larvae. Both
originate laterally from the processus muscularis quad-
rati; the m. orbitohyoideus arises broadly from the tip
area of the process, whereas the suspensoriohyoideus
arises from the posterior margin of the process and the
adjacent arcus subocularis quadrati. The m. suspensor-
iohyoideus is medial to the orbitohyoideus (Fig. 6).
The m. interhyoideus connects the processus laterales
hyalis ventrally in a straight line (Fig. 4). Right and left
muscle parts meet in a medial tendinous inscription. The
m. interhyoideus posterior is present as a veil of
extremely ﬁne, scattered and spaced ﬁbers underlying
the epithelium of the posterolateral and posterior wall of
the cavum peribranchiale. Due to its small ﬁber
diameter and scattered ﬁbers it was not possible to
reconstruct the muscle reliably from histological serial
sections. It is not shown in the ﬁgures.
Muscles associated with the branchial arches are
weakly developed. The branchial levators are small. The
mm. levatores arcuum branchialium I+II are merely
loosely scattered ﬁbers in a connective tissue sheath (it is
hardly possible to reconstruct those ﬁbers precisely from
parafﬁn sections; Figs. 3, 4 and 6 give only an
approximation of size, number, and orientation). The
muscle ﬁbers of branchial levators III+IV are much
more organized in a bundle and attach to the terminal
tips of ceratobranchiale III and IV, respectively (Fig. 6).
ARTICLE IN PRESS
Fig. 10. Transverse section through the anterior otic region. The section illustrates several mentioned features: the spaced ﬁbers of
the m. levator mandibulae superﬁcialis, dermal layer thickness of the skin, ﬂat branchial apparatus, buccal ﬂoor and buccal roof
papillae, lack of ﬁlter plates.
A. Haas et al. / Zoology 109 (2006) 26–4236
Branchial levator III originates from the otic capsule
dorsal and posterior to the foramen ovale. Levator IV
originates from connective tissue ventral to the cupula
posterior of the otic capsule. The m. transversus
ventralis IV originates from the same strand of
connective tissue more medially and expands mediad
and anteriad as a horizontal fan of loose ﬁbers
(reconstructed as closed surface in the 3D reconstruc-
tion). The short m. dilatator laryngis is embedded in the
latter muscle (Fig. 4).
Three mm. constrictores branchiales are present
(II–IV) in the branchial septa. Constrictor I (sensu
Haas, 1997) is absent. All three constrictors insert on the
arch formed by ceratobranchialia II and III in reversed
order, i.e., constrictor II inserts most posteriorly,
constrictor IV most anteriorly (Fig. 4). Constrictor II
originates from the distal end of ceratobranchiale II.
Constrictores III and IV both take origin from the tip of
The processus branchialis is the attachment site for
the mm. subarcualis rectus I, subarc. rectus II–IV,
subarc. rectus accessorius, and subarcualis obliquus II
(Fig. 4). The m. subarcualis rectus I has, unlike in most
other known tadpoles, only one head. The site of origin
is at the lateral side of the processus posterior hyalis.
Conversely, the m. subarcualis rectus II–IV is short. It
connects the processus branchialis to the distal cerato-
branchiale IV. The m. subarcualis rectus accessorius
originates from the ventral side of the spiculum IV plate.
It inserts ventral to the m. subarc. rectus IV on the
processus branchialis. Finally, the m. subarcualis
obliquus II originates from the processus branchialis
and reaches anteromedially to the mid-sagittal plane
where it forms a weak tendon. The tendon runs rostrad
in the mediosagittal plane and attaches to the ﬂat
ventral side of the posterior end of the basibranchiale
(a processus urobranchialis is not formed).
In most tadpoles, the m. diaphragmatobranchialis
attaches to ceratobranchiale III. In L. mjobergi, the
muscle is detached from the branchial apparatus and
instead inserts on the lateral wall of the otic capsule
(canalis semi-circularis lateralis). The muscle takes
origin from the connective tissue of the abdominal
septum. The muscle consists of two columns tightly
bundled along most of their length, but diverging at
both ends (Figs. 4 and 6). The m. rectus cervicis
originates from the same abdominal septum tissue and
runs between the columns of m. diaphragmatobran-
chialis (Fig. 6). The m. rectus cervicis is a thick muscle,
however, internally composed of loosely spaced bundles
of ﬁbers. The m. rectus cervicis inserts far anteriorly on
the ventral side of the ceratohyale (processus anterior).
Another longitudinal muscle runs along the internal
wall of the operculum (i.e., ventral lining of cavum
peribranchiale). The muscle likely is an anterior exten-
sion of the m. rectus abdominis system. It is a ﬂat muscle
band composed of few ﬁber tracts. Anteriorly it attaches
to a protuberance, the ventromedial face of the pars
articularis quadrati (Fig. 4).
The moderately thick m. geniohyoideus connects the
posterior margin of the cartilago labialis inferior (near
its symphysis) to the ventral side of the planum
hypobranchiale (Fig. 4).
Here we give only a preliminary account of those
postcranial muscles that relate to the cranium. Most
evidently the paraxial longissimus tract extends far
anteriorly on the otic capsule. The muscle’s almost
seamless transition into the m. levator mandibulae
longus profundus (Figs. 3 and 6) has already been
noted above and is a unique feature in L. mjobergi.
The m. intertransversarius capitis inferior (Fig. 4)
connects the processus transversum of vertebra II (not
shown) to the ventral parts of the posterior otic capsule.
One bundle of ﬁbers attaches to the otic capsule just
ventrolaterally to the foramen perilymphaticum inferius;
the second bundle attaches nearby but more medially at
the cartilage connecting the otic capsule and the planum
basale. The m. intertransversarius capitis superior also
originates from processus transversum II but inserts at
the arcus occipitalis and adjacent otic capsule cartilage
(sulcus occipitalis, Fig. 3).
Immediately posterior to the m. diaphragmatobran-
chialis are two muscles, both part of the body wall
musculature. Pending further veriﬁcation, based on
more extensive developmental series, these muscles are
tentatively assigned to the m. obliquus. The anterior
one, the pars scapularis (Gaupp, 1896) of the m.
obliquus originates from the fascia of the paravertebral
longissimus system and inserts on connective tissue
ventrally where m. rectus cervicis, m. rectus abdominis,
and m. diaphragmatobranchialis meet. The posterior
portion, the m. obliquus proper, has dorsal and ventral
connections to the skin. Ventromedially, the ﬁbers of
the pars scapularis could hardly be distinguished in
histological sections from the lateral ﬁbers of the m.
rectus abdominis; both were reconstructed as one body
(Fig. 4). An alternative interpretation would be that
what we tentatively call pars scapularis here could be a
lateral expansion of the rectus abdominis group.
The vermiform tadpole of L. mjobergi lives in the
gravel beds of small forest streams. We never collected
these larvae from other microhabitats in the same
stream, such as leaf litter accumulations (where the
ARTICLE IN PRESS
A. Haas et al. / Zoology 109 (2006) 26–42 37
related M. nasuta is common) or sandy stretches of the
stream. In an aquarium ﬁlled with gravel from the
collecting site (Fig. 1), we observed that large specimens
of L. mjobergi are not only able to use existing crevices,
but will also push their way into spaces smaller than the
body diameter. Thus, L. mjobergi is an active burrowing
tadpole. Locomotion is eel-like and characterized by
unusual mobility of the head and trunk (Fig. 1). We
assume that interstitial spaces are used opportunisti-
cally, particularly by small larvae lacking the force to
push aside gravel. Larger individuals are able to push
aside superﬁcial gravel, and might just burrow deep
enough to conceal themselves. Small tadpoles of early
larval stages may be able to get deeper into the sediment
just by using interstices. The burrow use and burrowing
behaviors of L. mjobergi are impossible to observe in the
ﬁeld and need further investigation under laboratory
conditions. In preliminary aquarium experiments,
L. mjobergi tadpoles did not burrow when offered sand
Active burrowing by tadpoles is rare but has been
reported in the South American microhylid Otophryne
pyburni (Wassersug and Pyburn, 1987) and in various
neotropical centrolenid tadpoles. Similar to L. mjobergi,
the larvae of O. pyburni live shallowly buried at the
bottom of clear streams; however, the two species prefer
different sediments: sand in the case of O. pyburni and
gravel in the case of L. mjobergi.
Neotropical centrolenid tadpoles live in accumulated
leaf litter and stream sediments, and are sometimes
found in moist plant debris at some distance from the
stream (Villa and Valerio, 1982;Lynch et al., 1983;
Mijares-Urrutia, 1990;Hoff et al., 1999). Some were
reported to burrow as deep as 200 mm into benthic
debris (Savage, 2002). Mijares-Urrutia (1990) found
larvae of Centrolenella andina in quiet pools with slow
currents. Tadpoles hide superﬁcially in the sediment and
among rocks and fallen leaves and escape with bursts of
free swimming when disturbed. Mijares-Urrutia did not
give information about the grain size of the sediment. In
contrast, we found L. mjobergi only in fast ﬂowing, rifﬂe
sections of the streams in gravel sediments. We never
saw any L. mjobergi tadpole resting exposed or
swimming in escape.
With the exception of Wassersug and Pyburn’s (1987)
description of O. pyburni, anatomical accounts of
burrowing or fossorial tadpoles are scarce in the
literature; none of the centrolenid larvae has been
analyzed in great detail (except some character states
for Cochranella granulosa in Haas, 2003). Superﬁcially,
centrolenid tadpoles are quite similar to the larvae of
Leptobrachella in body shape (Starrett, 1960;Mijares-
Urrutia, 1990). Centrolenid tadpoles have long tails with
low tailﬁns and thick muscular parts. Their eyes are
rudimentary and pigmentation may be reduced (Villa
and Valerio, 1982;Altig and McDiarmid, 1999b). Long
and slender tadpoles living in leaf litter have evolved in
the southeast Asian Staurois (Inger and Wassersug,
1990;Malkmus et al., 1999, pers. obs.).
A more detailed comparison can be made between the
tadpoles of L. mjobergi and O. pyburni. Various features
of both taxa, particularly of the head, can be hypothe-
sized as adaptations to the respective sediment used. In
O. pyburni, the skull is dorsoventrally ﬂattened and
broad cartilagines labiales superiores and cornua
trabeculae support the snout; both have been inter-
preted as adaptations to sand penetration (Wassersug
and Pyburn, 1987). In contrast, the head of L. mjobergi
is narrow and rounded, followed by an almost
cylindrical trunk. This confers an eel-like body shape
that facilitates entering crevices of unpredictable shape
The neurocranial and jaw cartilages of L. mjobergi are
robust but lack the hypertrophy of snout cartilages and
snout skin of O. pyburni. The epidermis of the entire
body is supported by a ﬁrm and thick dermal tissue layer
of collagen ﬁber in L. mjobergi (Fig. 10), which likely
protects the tadpole among the rough-edged gravel
(Fig. 1). Various soft tissue structures of O. pyburni that
apparently serve to block or trap sand grains and
prevent sand from entering the pharynx and gills (e.g.,
expanded ventral velum; see Wassersug and Pyburn,
1987) are absent in L. mjobergi.
Differences prevail, but L. mjobergi and O. pyburni
also have some common features. The muscular part of
the tail is high at the body–tail junction and the tail ﬁns
are low. The tails are used for propulsion in the
substrate as well as swimming outside the substrate.
The eyes are small and non-protruding. These features
also apply to the fossorial tadpoles of centrolenids and
Staurois.InL. mjobergi and O. pyburni, the dorsal and
ventral trunk muscles extend further rostrally than in
other tadpoles and allow for ﬂexion and extension of the
cranium. These muscle characters are unknown for
centrolenids and Staurois.
From the anatomical evidence as well as observations
of living animals (Fig. 1), the maneuverability of the
head and the anterior vertebral column in L. mjobergi is
unusual for anuran larvae. Flexion of the body axis in
the sagittal plane has been reported before in other
species, but without indication that bending took place
between the head and ﬁrst vertebra (Wassersug, 1992).
We present anatomical and photographic (Fig. 1)
evidence that bending can occur between head and
vertebral column. In six lateral view images we
measured the angles between a line connecting the
nostril to the center of the eye and a line in parallel with
the anterior trunk dorsal contour, including extension
ARTICLE IN PRESS
A. Haas et al. / Zoology 109 (2006) 26–4238
(Fig. 1a), ﬂexion (Fig. 1c), and intermediate resting
positions. The angles ranged from 1381to 1701giving an
approximate range of dorso-ventral head mobility of
321.InFig. 1a hyperextension seems to stem from both
head extension plus a smooth upward arching along all
of the presacral vertebral column. In contrast, the
ventral ﬂexion of the head in Fig. 1c appears to be
restricted to the occipital joint. Lateral bending could
not be determined from the existing photographic
We consider two primary features to account for head
maneuverability in L. mjobergi: the exclusion of the
notochord from the planum basale and the design of the
atlanto-occipital joint. Similar joints were reported in
salamanders and a fossil caecilian with a median process
called tuberculum interglenoideum or odontoid process
(Wake and Lawson, 1973;Duellman and Trueb, 1992;
Jenkins and Walsh, 1993). We use the term dens atlantis
in L. mjobergi to stress the analogy, not homology, of
these structures in amphibians.
Apart from some amphibians, the similarities of the
L. mjobergi occipital joint to the design of the atlanto-
axial articulation in mammals, including humans
(Leonhardt et al., 1987), are obvious. These involve
two inclined lateral articular faces and a medial
dens that establishes a third articulation. Both in the
L. mjobergi occipital joint and in the mammalian
atlanto-axial joint the dens is retained by a transverse
ligament. Based on structural analogies we assume that
similar axes of rotation are realized in the atlanto-
occipital articulation of L. mjobergi and the atlanto-
axial articulation in mammals (e.g., for humans
Leonhardt et al., 1987). We are unaware of any data
on rotation in the occipital joint of salamanders. Similar
to the dens axis in humans (Leonhardt et al., 1987), in
L. mjobergi rotation in the occipital joint around the
body axis could happen in the longitudinal axis of the
dens atlantis, and lateral and dorsoventral movements
by translocation in the atlantal cotyles. In L. mjobergi
we assume that the curvatures of the atlantal cotyles and
occipital arches deﬁne the movement of the atlas on a
circular radius in sagittal and frontal planes, respec-
tively. The centers of such dorso-ventral and lateral
movements must be more anteriorly than the articula-
tion itself in a virtual center of rotation approximately in
the hypophyseal region, depending on the radius of the
Our functional understanding of tadpole musculature
relies largely on anatomical relationships. Few studies
have tested functional hypotheses for anuran tadpoles
by electromyography (Gradwell, 1972b;Larson and
Reilly, 2003). Activity of most of the thin branchial
muscles has never been recorded with EMG in tadpoles.
Although EMG measurements have not been made in
L. mjobergi, several muscles likely drive cranial move-
ments: the paravertebral longissimus system reaches far
rostrally and presumably serves for dorsal extension of
the head (assisted by the m. levator mandibulae longus
profundus; Fig. 6). The mm. rectus cervicis et rectus
abdominis also reach far rostrally, ventral to the
rotational axis, and therefore could cause ventral ﬂexion
of the head relative to the trunk. Asymmetric contrac-
tions of the paravertebral longissimus system, the m.
rectus cervicis and/or the mm. intertransversarii capitis
could generate lateral bending of the head. The m.
obliquus (pars scapularis) and m. diaphragmatobran-
chialis might rotate the head around the longitudinal
axis of the dens atlantis. The fact that the m.
diaphragmatobranchialis has no connection to the
ceratobranchiale III, where it inserts in most species
(Gradwell, 1972a;Haas, 1997;Haas and Richards,
1998), corroborates that this muscle has assumed a new
function in head movements of L. mjobergi larvae. In
other tadpoles it appears to lower the ceratobranchials
Kenny (1969a) and Gradwell (1972b) speculated
about the function of the branchial muscles in tadpoles.
As judged by anatomy alone, several muscles likely are
involved in depressing the branchial apparatus (m.
diaphragmatobranchialis, m. rectus cervicis), others in
elevating it (mm. levatores arcuus branchialium I–IV). It
is unclear whether this branchial pump is recruited
in normal cyclic irrigation (Larson and Reilly, 2003). In
L. mjobergi, the branchial basket is reduced and
branchial muscles are feeble. Furthermore, the m.
diaphragmatobranchialis is not available for branchial
lowering. The overall weak structural condition suggests
that branchial arch movements do not contribute
signiﬁcantly to normal irrigation cycles, in terms of a
forceful branchial pump supporting the buccal pumping
mechanism. Rather, the branchial apparatus probably
follows the movements of the ceratohyale passively. The
feeble branchial muscles may still play a role in
occasional hyper-inspiration or hyper-expiration.
Another unusual feature of L. mjobergi is the shape of
the parasphenoid. In the stages examined it is a long oval
ring with a posterior rectangular plate. The parasphe-
noid abuts the medial edges of the trabeculae cranii. To
our knowledge such a shape of the parasphenoid has not
been reported in anuran larvae before.
Currently we can only speculate about the feeding
habits of L. mjobergi. Buccal volume, lever–arm ratio,
and buccal area can be indicative of macrophagous,
generalized, or microphagous feeding preferences in a
given species (Wassersug and Hoff, 1979). In L. mjobergi
the branchial area (40% of the bucco-pharyngeal ﬂoor
area) is relatively reduced and the ceratohyal area (42%)
relatively enlarged (55–62% and 24–28%, respectively,
ARTICLE IN PRESS
A. Haas et al. / Zoology 109 (2006) 26–42 39
in generalized Litoria tadpoles; Haas and Richards,
1998). The lever–arm ratios of the ceratohyale (0.29–0.5)
fall into the groups of benthic generalized tadpoles on
the one end and macrophagous tadpoles on the other
end of the range (Wassersug and Hoff, 1979), depending
on the axis of measurement. The measurements do not
allow a clear assignment of L. mjobergi to the
classiﬁcation of feeding modes based on hyobranchial
geometry as in Wassersug and Hoff (1979).
In L. mjobergi the branchial basket barely extends
beyond the outlines of the upper skull parts, the
ceratobranchials are very thin and terminally not fused
(commissurae terminales absent). The branchial basket
is very ﬂat and ﬁlter plates are absent (Inger, 1983).
Enlarged ceratobranchials and relatively reduced bran-
chial baskets occur in suctorial rheophilous tadpoles
(Wassersug and Hoff, 1979;Haas and Richards, 1998).
However, even more reduced branchial baskets and
larger ceratohyalia were reported for macrophagous
tadpoles (Hyla nana,Lavilla, 1990;Vera Candioti et al.,
2004;Lepidobatrachus laevis,Ruibal and Thomas, 1988;
Anotheca spinosa,Wassersug and Hoff, 1979; pers.
obs.). L. mjobergi lacks the enormous development of
the m. orbitohyoideus found in macrophagous Hyla
nana (Vera Candioti et al., 2004) or ovophagous
Anotheca spinosa (pers. obs.). In these species, this
muscle generates the forces for suction. However, the
special insertion pattern of the m. rectus cervicis in
L. mjobergi suggests that this muscle might act
synergistically with the moderately developed m. orbi-
tohyoideus to generate larger forces.
Histological sections of the gut, so far examined, do
not support our initial hypothesis that L. mjobergi was a
macrophagous tadpole (see, for example, worm feeding
reported in Hyla nana;Vera Candioti et al., 2004). Gut
contents did not contain large objects such as nematodes
or oligochaete worms. Rather, we found predominantly
heterogeneous organic debris without identiﬁable parts
of animals or plants. However, L. mjobergi lacks those
structures identiﬁed as indicative of efﬁcient micropha-
gous suspension feeding (and found in the fossorial
O. pyburni): high ﬁlter plates and ﬁlter rufﬂes.
As striking as the lack of elaborate ﬁlter plates (Fig. 10)
are the numerous papillae (some long, but most of
them pustules) on the buccal ﬂoor and buccal roof
arenas (Inger, 1983). The buccal ﬂoor and roof pustules
contain secretory cells. Strings of mucus could entrap
particles in the center of the buccal cavity and transport
particles directly to the esophagus; rather than mucous
entrapment above the ﬁlter plates in most tadpoles
(Kenny, 1969a, b).
The oral disk has a characteristic cup- or funnel-
shaped structure and is well equipped with muscle ﬁbers
of the m. mandibulolabialis, though its function remains
obscure. However, the shape of this derived oral disk
and the lack of keratodonts means that a substrate-
scraping mode of feeding, typical of most stream-
associated tadpoles (Taylor et al., 1996), is unlikely.
The skeletal structures of the jaws give no further clues
with respect to feeding. The jaw cartilages are reminis-
cent of those of other pelobatids (e.g., Roc
ˇek, 1981), and
differ only in proportion.
In sum, some features of the feeding apparatus of
L. mjobergi are unique and do not compare readily to
other known feeding types in tadpoles. Larger specimen
samples and evaluation of their gut contents will be
necessary to reveal resource use in L. mjobergi.
L. mjobergi belongs to the Megophryidae (Frost,
2004), a southeast Asian group of ground dwelling
frogs, formerly included in the Pelobatidae. Megophryi-
dae form a monophyletic group with other Old World
pelobatoids (Haas, 2003;Garcı
´s et al., 2003;
Pu´ gener et al., 2003; Hoegg et al., 2004;Roelants and
Bossuyt, 2005;San Mauro et al., 2005). Lathrop (1997)
reviewed megophyid taxonomy. Generic relationships
were addressed by Zheng et al. (2004), however, without
considering Leptobrachella.Ramaswami (1943) de-
scribed the rostral cartilages in Megophrys. Despite the
vermiform body shape and the derived anatomical
features in the musculo-skeletal apparatus of L.
mjobergi, a number of larval morphological characters
show its phylogenetic relationships: the excessive num-
ber of larval tail vertebrae has been reported before in
Megophrys major (Grifﬁths, 1963) and is present in M.
nasuta as well (pers. obs.). The large and elongate
adrostral cartilage, presence of a m. mandibulolabialis
superior, and insertion of m. subarcualis rectus II–IV on
ceratobranchiale III are derived character states of the
Eurasian megophryid and pelobatid clade in Haas
(2003) and are shared by L. mjobergi indicating its
megophryid relatedness. The only species known to
possess the m. subarcualis rectus accessorius are the
megophryids L. mjobergi,Leptobrachium hasselti and
Megophrys montana, and the non-pelobatid South
African Heleophryne natalensis (Haas, 2003). L. mjo-
bergi differs from M. montana and L. hasseltii (Haas,
2003) in having a high suspensorium (low in the two
We wish to thank the Sarawak Forest Department, in
particular Datuk Cheong Ek Choon, Director, and
Bolhan Budeng, for issuing collecting permits
(NPW.907.4-36); the Economic Planning Unit, The
Prime Minister’s Department, Malaysia, and especially
Mrs. Munirah Abd. Manan, for issuing research permit
ARTICLE IN PRESS
A. Haas et al. / Zoology 109 (2006) 26–4240
No. 1168 to A. Haas. K. Felbel and E. Gretscher
skillfully prepared serial sections of the specimens. We
thank Greg Handrigan and Richard Wassersug for
thoughtful criticism and valuable comments on an early
draft of this work. Finally, we are grateful to Volkswa-
gen-Stiftung, Germany, who supported the study with
Grant I/79 405.
Appendix A. Supplementary data
The online version of this article contains additional
supplementary data. Please search for doi:10.1016/
j.zool.2005.09.008 at http://dx.doi.org.
Altig, R., Johnston, F., 1989. Guilds of anuran larvae:
relationships among developmental modes, morphologies,
and habitat. Herpetol. Monogr. 3, 81–109.
Altig, R., McDiarmid, R.W., 1999a. Diversity: familial and
generic characterizations. In: McDiarmid, R.W., Altig, R.
(Eds.), Tadpoles. University of Chicago Press, Chicago,
Altig, R., McDiarmid, R.W., 1999b. Body plan: development
and morphology. In: McDiarmid, R.W., Altig, R. (Eds.),
Tadpoles. University of Chicago Press, Chicago, pp. 24–51.
¨ck, P., 1989. Romeis Mikroskopische Technik. Urban &
¨nchen, Wien, Baltimore.
Cannatella, D., 1999. Architecture: cranial and axial muscu-
loskeleton. In: McDiarmid, R.W., Altig, R. (Eds.), Tad-
poles. University of Chicago Press, Chicago, pp. 52–91.
De Jongh, H.J., 1968. Functional morphology of the jaw
apparatus of larval and metamorphosing Rana temporaria.
Neth. J. Zool. 18, 1–103.
Dietrich, H.F., Fontaine, A.R., 1975. A decalciﬁcation method
for ultrastructure of echinoderm tissues. Stain Technol. 50,
Dring, J.C.M., 1983. Frogs of the genus Leptobrachella
(Pelobatidae). Amphibia-Reptilia 4, 89–102.
Duellman, W.E., Trueb, L., 1992. Biology of the Amphibians.
McGraw-Hill, New York.
Frost, D.R., 2004. Amphibian Species of the World. Version
3.0 (22 August 2004). American Museum of Natural
History, New York http://research.amnh.org/herpetology/
´s, M., Buchholz, D.R., Parra-Olea, G., 2003.
Phylogenetic relationships of Pelobatoidea re-examined
using mtDNA. Mol. Phylogenet. Evol. 28, 12–23.
Gaupp, E., 1893. Beitraege zur Morphologie des Schaedels. I.
Primordial-Cranium und Kieferbogen von Rana fusca.
Morphol. Arbeit. 2, 275–481.
Gaupp, E., 1896. Ecker’s und Wiedersheim’s Anatomie des
Frosches. Erste Abteilung. Lehre vom Skelett und vom
Muskelsystem, third ed. Vieweg, Braunschweig.
Gosner, K.L., 1960. A simpliﬁed table for staging anuran
embryos and larvae with notes on identiﬁcation. Herpeto-
logica 16, 183–190.
Gradwell, N., 1972a. Gill irrigation in Rana catesbeiana. Part
I. On the anatomical basis. Canad. J. Zool. 50, 481–499.
Gradwell, N., 1972b. Gill irrigation in Rana catesbeiana. Part
II. On the musculoskeletal mechanism. Canad. J. Zool. 50,
Grifﬁths, I.G., 1963. The phylogeny of the Salientia. Biol. Rev.
Haas, A., 1997. The larval hyobranchial apparatus of
discoglossoid frogs: its structure and bearing on the
systematics of the Anura (Amphibia: Anura). J. Zool. Syst.
Evol. Res. 53, 179–197.
Haas, A., 2001. Mandibular arch musculature of anuran
tadpoles, with comments on homologies of amphibian jaw
muscles. J. Morphol. 247, 1–33.
Haas, A., 2003. The phylogeny of frogs as inferred from
primarily larval characters (Amphibia, Anura). Cladistics
Haas, A., Fischer, M.S., 1997. Three-dimensional reconstruc-
tion of histological sections using modern product-design
software. Anat. Rec. 249, 510–516.
Haas, A., Richards, S.J., 1998. Correlations of cranial
morphology, ecology, and evolution in Australian suctorial
tadpoles of the genera Litoria and Nyctimystes (Amphibia:
Anura: Hylidae: Pelodryadinae). J. Morphol. 238, 109–141.
Hoegg, S., Vences, M., Brinkmann, H., Meyer, A., 2004.
Phylogeny and comparative substitution rates of frogs
inferred from sequences of three nuclear genes. Mol. Biol.
Evol. 21, 1188–1200.
Hoff, K.S., Blaustein, A.R., McDiarmid, R.W., Altig, R.,
1999. Behavior, interactions and their consequences. In:
McDiarmid, R.W., Altig, R. (Eds.), Tadpoles. University of
Chicago Press, Chicago, London, pp. 215–239.
Ibanez, R., Jaramillo, F.E., Jaramillo, A.C.A., 1999. Amplia-
cion del ambito de distribucion y descripcion del renacuajo
de la rana de cristal Hyalinobatrachium aureoguttatum
(Anura: Centrolenidae). Rev. Acad. Colomb. Cienc. Exact.
Fisic. Nat. 23 (Supl. Especial), 293–298.
Inger, R.F., 1983. Larvae of southeast Asian species of
Leptobrachium and Leptobrachella (Anura: Pelobatidae).
In: Rhodin, A., Miyata, K. (Eds.), Advances in Herpetol-
ogy and Evolutionary Biology. Museum of Comparative
Zoology, Massachusetts, pp. 13–32.
Inger, R.F., 1985. Tadpoles of the forested regions of Borneo.
Fieldiana 26, 1–89.
Inger, R.F., Stuebing, R.B., 1991. A new species of frog of the
genus Leptobrachella Smith (Anura: Pelobatidae), with a
key to the species from Borneo. Rafﬂes Bull. Zool. 39,
Inger, R.F., Wassersug, R.J., 1990. A centrolenid-like anuran
larva from southeast Asia. Zool. Sci. 7, 557–561.
Jaramillo, F.E., Jaramillo, C.A., Ibanez, D.R., 1997. Rena-
cuajo de la rana de cristal Hyalinobatrachium colymbiphyl-
lum (Anura: Centrolenidae). Rev. Biol. Trop. 45, 867–870.
Jenkins, F.A., Walsh, D.M., 1993. An early Jurassic caecilian
with limbs. Nature 365, 246–250.
Kenny, J.S., 1969a. Feeding mechanisms in anuran larvae.
J. Zool. London 157, 225–246.
Kenny, J.S., 1969b. Pharyngeal mucous secreting epithelia of
anuran larvae. Acta Zool. 50, 143–153.
ARTICLE IN PRESS
A. Haas et al. / Zoology 109 (2006) 26–42 41
Larson, P.M., Reilly, S.M., 2003. Functional morphology of
feeding and gill irrigation in the anuran tadpole; electro-
myography and muscle function in larval Rana catesbeiana.
J. Morphol. 255, 202–214.
Lathrop, A., 1997. Taxonomic review of the megophryid frogs
(Anura: Pelobatoidea). Asiatic Herpetol. Res. 7, 68–79.
Lavilla, E.O., 1990. The tadpole of Hyla nana. J. Herpetol. 24,
Leonhardt, H., Tillmann, B., To
¨ndury, G., Zilles, K. (Eds.),
1987. Rauber/Kopsch Anatomie des Menschen. Band I:
Bewegungsapparat. Georg Thieme Verlag, Stuttgart.
Lynch, J.D., Ruiz, P.M., Rueda, J.V., 1983. Notes on the
distribution and the reproductive biology of Centrolene
geckoideum Jimenez De la Espada in Colombia and
Ecuador (Amphibia: Centrolenidae). Stud. Neotrop. Fauna
Environ. 18, 239–244.
Malkmus, R., Kosuch, J., Kreutz, J., 1999. Die Larve von
Staurois tuberlinguis Boulenger, 1918. Eine neue centrole-
¨hnliche Kaulquappe aus Borneo (Anura: Ranidae).
Herpetozoa 12, 17–22.
McDiarmid, R.W., Altig, R. (Eds.), 1999. Tadpoles. Uni-
versity of Chicago Press, Chicago.
Mijares-Urrutia, A., 1990. The tadpole of Centrolenella andina
(Anura: Centrolenidae). J. Herpetol. 24, 410–412.
Noonan, B.P., Bonett, R.M., 2003. A new species of
Hyalinobatrachium (Anura: Centrolenidae) from the High-
lands of Guyana. J. Herpetol. 37, 91–97.
Orton, G.L., 1953. The systematics of vertebrate larvae. Syst.
Zool. 2, 63–75.
Pu´ gener, A., Maglia, A.M., Trueb, L., 2003. Revisiting the
contribution of larval characters to an analysis of
phylogenetic relationships of basal anurans. Zool. J. Linn.
Soc. 139, 129–155.
Ramaswami, L.S., 1943. An account of the chondrocranium of
Rana afghana and Megophrys, with a description of the
masticatory musculature of some tadpoles. Proc. Natl. Inst.
Sci. India 9, 43–58.
ˇek, Z., 1981. Cranial anatomy of frogs of the family
Pelobatidae Stannius, 1856, with outlines of their phylo-
geny and systematics. Acta Univ. Carol. Biol. 1980, 1–164.
ˇek, Z., 2003. Larval development and evolutionary origin
of the anuran skull. In: Heatwole, H., Davies, M. (Eds.),
Amphibian Biology. Surrey Beatty & Sons, Chipping
Norton, pp. 1877–1995.
Roelants, K., Bossuyt, F., 2005. Archaeobatrachian paraphyly
and pangaean diversiﬁcation of crown-group frogs. Syst.
Biol. 54, 111–126.
Ruibal, R., Thomas, E., 1988. The obligate carnivorous larvae
of the frog, Lepidobatrachus laevis (Leptodactylidae).
Copeia 1988, 591–604.
San Mauro, D., Vences, M., Alcobendas, M., Zardoya, R.,
Meyer, A., 2005. Initial diversiﬁcation of living amphibians
predated the breakup of Pangaea. Am. Nat. 165, 590–599.
Savage, J.M., 2002. The Amphibians and Reptiles of Costa
Rica. A Herpetofauna Between Two Continents, Between
Two Seas. The University of Chicago Press, Chicago.
Starrett, P., 1960. Description of the tadpoles of Middle
American frogs. Misc. Publ. Mus. Zool. Univ. Michigan
Taylor, W.R., van Dyke, G.C., 1985. Revised procedures for
staining and clearing small ﬁshes and other vertebrates for
bone and cartilage study. Cybium 9, 107–119.
Taylor, C.L., Altig, R., Boyle, C.R., 1996. Oral disc kinematics
of four anuran tadpoles. Herpetol. Nat. Hist. 4, 49–56.
Vera Candioti, M.F., Lavilla, E.O., Echeverria, D.D., 2004.
Feeding mechanism in two treefrogs, Hyla nana and Scinax
nasicus (Anura: Hylidae). J. Morphol. 261, 206–224.
Villa, J., Valerio, C.E., 1982. Red, white and brown!
Preliminary observations on the color of the centrolenid
tadpole (Amphibia: Anura: Centrolenidae). Brenesia 19–20,
Wake, D.B., Lawson, R., 1973. Development and adult
morphology of the vertebral column in the plethodontid
salamander Eurycea bislineata, with comments on vertebral
evolution in the Amphibia. J. Morphol. 139, 251–300.
Wassersug, R., 1992. The basic mechanics of ascent and
descent by anuran larvae (Xenopus laevis). Copeia 3,
Wassersug, R.J., Hoff, K., 1979. A comparative study of the
buccal pumping mechanism of tadpoles. Biol. J. Linn. Soc.
Wassersug, R.J., Pyburn, W.F., 1987. The biology of the Pe-
ret’ toad Otophryne robusta (Microhylidae), with special
consideration of its fossorial larva and systematic relation-
ships. Zool. J. Linn. Soc. 91, 137–169.
Zheng, Y.-C., Mo, B.-H., Liu, Z.-J., Zeng, X.-M., 2004.
Phylogenetic relationships of megophryid genera (Anura:
Megophryidae) based on partial sequences of mitochon-
drial 16S rRNA gene. Zool. Res. 25, 205–213.
ARTICLE IN PRESS
A. Haas et al. / Zoology 109 (2006) 26–4242