Vibroacoustic Response of the Tympanic Membrane to Hyoid-Borne Sound Generated during Echolocation in Bats

Abstract

Synopsis
The hyoid apparatus in laryngeally echolocating bats is unique as it forms a mechanical connection between the larynx and auditory bullae, which has been hypothesized to transfer the outgoing echolocation call to the middle ear during call emission. Previous finite element modeling (FEM) found that hyoid-borne sound can reach the bulla at an amplitude likely heard by echolocating bats; however, that study did not model how or if the signal could reach the inner ear (or cochlea). One route that sound could take is via stimulation of the eardrum—similarly to that of air-conducted sound. We used micro computed tomography (μCT) data to build models of the hyoid apparatus and middle ear from six species of bats with variable morphology. Using FEM, we ran harmonic response analyses to measure the vibroacoustic response of the tympanic membrane due to hyoid-borne sound generated during echolocation and found that hyoid-borne sound in all six species stimulated the eardrum within a range likely heard by bats. Although there was variation in the efficiency between models, there are no obvious morphological patterns to account for it. This suggests that hyoid morphology in laryngeal echolocators is likely driven by other associated functions.

Full text

Integrative Organismal Biology
Integrative Organismal Biology , pp. 1–12
https://doi.org/10.1093/iob/obad004 A Journal of the Society for Integrative and Comparative Biology
ARTICLE
Vibroacoustic Response of the Tympanic Membrane to
Hyoid-Borne Sound Generated during Echolocation in Bats
C .C .G. Snipes and R.T. Carter
1
Department of Biological Sciences, East Te n n e s s e e State University, 127 Gilbreath Dr, Johnson City, TN 37614, USA
1
E-mail: carterrt@etsu.edu
Synopsis The hyoid apparatus in laryngeally echolocating bats is unique as it forms a mechanical connection between the
larynx and auditory bullae, which has been hypothesized to transfer the outgoing echolocation call to the middle ear during
call emission. Previous nite element modeling (FEM) found that hyoid-borne sound can reach the bulla at an amplitude likely
heard by echolocating bats; however, that study did not model how or if the signal could reach the inner ear (or cochlea).
One route that sound could take is via stimulation of the eardrum—similarly to that of air-conducted sound. We used micro
computed tomography ( μCT) data to build models of the hyoid apparatus and middle ear from six species of bats with variable
morphology. Using FEM, we ran harmonic response analyses to measure the vibroacoustic response of the tympanic membrane
due to hyoid-borne sound generated during echolocation and found that hyoid-borne sound in all six species stimulated the
eardrum within a range likely heard by bats. Although there was variation in the eciency between models, there are no obvious
morphological patterns to account for it. This suggests that hyoid morphology in laryngeal echolocators is likely driven by other
associated functions.
Introduction
The echolocation calls of bats are generated either via
tongue clicking or by vibration of the vocal cords in
the larynx. Laryngeal echolocation in bats coincides
with adaptations of the skull and neck not found in
tongue clicking bats. These adaptations include en-
larged basal turns of the cochlea and increased stiness
of the basilar membrane, which increases hearing
sensitivity to the higher frequencies associated with
echolocation calls ( Kössl and Va t e r 1995 ; Va t e r and
Kössl 2011 ). Laryngeal echolocators also have enlarged,
reinforced cricoid, thyroid, and arytenoid cartilages,
and hypertrophied intrinsic musculature that allows
for the production of powerful, high-frequency calls
( Carter 2020 ). Perhaps most notable is the attened,
paddle-like cranial end of the stylohyal bones that
articulate with the auditory bullae, which is considered
a characteristic indicative of laryngeal echolocation in
bats ( Ve s e l ka et al. 2010 ). This is unusual among mam-
mals, as the hyoid typically does not articulate with
other bones but instead is suspended in the throat via
ligaments and muscles and serves as a dynamic anchor
for the complex musculature associated with chew-
ing, swallowing, and vocalization. Since the stylohyal
bone is the distal portion of the hyoid apparatus, the
unique stylohyal-auditory bulla articulation in laryn-
geal echolocators completes a bony connection from
the larynx (site of call production) to the auditory bulla
(site of echo reception) via the hyoid apparatus ( Fig. 1 ).
This is particularly interesting because neurological
research on echolocating bats shows that bats must rst
register their outgoing calls to subsequently register
the returning echoes ( O’Neill and Suga 1979 ; Suga et
al. 1979 ). Given the close proximity of the larynx to
the ear, coupling the two would theoretically provide a
more direct means of transferring the call from the site
of production to the inner ear and therefore the brain.
Finite-element (FE) modeling of this connection be-
tween larynx and ear indicates that sound can be ef-
fectively transmitted from the laryngeal surface of the
hyoid to the auditory bullae in Artibeus jamaicensis
(a low duty cycle [LDC]/frequency modulated [FM]
echolocator) and Rhinolophus pusillus (a high duty cy-
cle [HDC]/narrow band [NB] echolocator) ( Snipes and
Carter 2022 ). Here, duty cycle refers to the length of
time in a call sequence in which there is out going
©TheAuthor(s) 2023. Published by Oxford University Press on behalf of the Society for Integrative and Comparative Biology. This is an Open
Access article distributed under the terms of the Creative Commons Attri bution License ( https://creativecommons.org/licenses/by/4.0/ ), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

2 C . C . G. Snipes and R. T. Car ter
Fig. 1 Volume-rendered lateral and ventral views of the cranium (brown), trachea/larynx (gray), and hyoid apparatus from R. f er rumequinum.
The hyoid apparatus consists of the fused basihyal and thyrohyals (blue), hypohyal (green), ceratohyal (purple), and stylohyal (yellow) bone(s).
The h ypoh yal, ceratoh yal, and styloh yal bones are collecti vel y ref er red to as the anterior cornu, and the bony segments are connected via
cartilaginous joints (gray). In laryngeal echolocators, the stylohyal bones articulate with the auditory bullae (orange), which houses the TM and
middle ear bones that transfer airborne sound to the cochleae (red).
sound, where LDC bats use 10% of the call sequence and
HDC bats use more than 50% of each sequence ( Fenton
et al. 2012 ). While both A. jamaicensis and R. pusillus
exhibit spatulate stylohyals that wrap around the bul-
lae, there is variation in the placement and extent to
which they articulate with the bulla. The LDC echolo-
cator ( A. jamaicensis ) has stylohyals that wrap around
the lateral side of the bullae, while the HDC echoloca-
tor ( R. pusillus ) has stylohyals that wrap around the me-
dial rim of the bullae ( Fig. 2 ). Snipes and Carter (2022)
did not include a tympanic membrane (TM) in their FE
models but instead used the vibration of the bulla as ev-
idence that sound likely moved into the inner ear via
bone conduction or a rocking motion of the bulla/TM
unit in the lateral-medial plane, which presumably sets
the ear ossicles into motion. In that study, we modeled
varying levels of constraint (0, 1, 3, and 5 xed points)
on the basihyal to evaluate the eect of muscle attach-
ments and found dierences in the performance of our
R. pusillus and A. jamaicensis models. As basihyal con-
straint was increased, the displacement of the bulla in
the A. jamaicensis model quickly dropped below the as-
signed threshold of 2.9e-11 m; and conversely, the bulla
of the R. pusillus model exhibited displacement peaks
above the assigned threshold at all levels of constraint
( Snipes and Carter 2022 ). These results lead us to con-
sider whether HDC echolocators could use vibration of
the bulla and bone conduction to transfer sound into
the inner ear. The relatively poor performance of the
constrained A. jamaicensis model ( Snipes and Carter
2022 ), and the placement of the spatulate end of the sty-
lohyal on the bulla ( Fig. 2 ) may mean that excitation
of the TM (similar to what airborne sound would do) is
the most ecient route for hyoid-borne sound to reach
the inner ear in LDC bats.
As the mammalian middle ear is ecient at sound
transfer, the connection between the hyoid and ear ossi-
cles may be the most direct way to register outgoing calls
to the cochlea. Bats may also use the muscles of the mid-
dle ear to attenuate the loud outgoing call so that the soft
returning echo can be perceived by the cochlea ( Henson
1970 ). Therefore, a route through the middle ear would
enable bats to control the amplitude of the outgoing
call arriving at the cochlea. In the present study, we
used FE models to assess whether hyoid-borne sound
could displace the TM within a range that bats can per-
ceive. To do this, we used experimental data from TM
excitation ( Manley et al. 1972 ) and neurophysiological
( Hener et al. 2013 ) studies of hearing in bats to verify
our models and estimate a minimum hearing thresh-
old for TM displacement. We hypothesized that bone-
conducted sound through the hyoid would stimulate

Hyoid conducted hearing in bats 3
Fig. 2 Stylohyal-auditory bulla articulation from LDC/FM and HDC/NB echolocators. The styloh yal (yellow) articulates with the lateral rim of
the auditory bulla (orange) in LDC/FM echolocators, whereas the stylohyal articulates with the medial rim of the auditory bulla in HDC/NB
echolocators.
Ta b l e 1 Catalog ID, species, data identier/link, and scan settings for all modeled specimens.
Catalog # Species Identier Pixel spacing ( μm) Voltage (kV) Filter (mm)
AMNH 245591 R. ferrumequinum http://n2t.net/ark:/87602/m4/491720 16 80 none
AMNH 48028 R. rouxi http://n2t.net/ark:/87602/m4/491709 15 80 Al-0.5
MVZ 122932 R. hildebrandtii http://n2t.net/ark:/87602/m4/491705 20 80 Al-0.5
MVZ 112095 H. diadema http://n2t.net/ark:/87602/m4/491624 21 80 Al-0.5
UMMZ 163615 M. spasma https://doi.org/10.17602/M2/M57216 46.92 95 none
L-RC:colonyadult A. jamaicensis http://n2t.net/ark:/87602/m4/491726 18 110 Al-0.5
the TM within a range likely heard by bats. This hypoth-
esis would be supported if displacement of the TM in
response to hyoid-borne sound exceeds the estimated
hearing threshold. We also hypothesized that the po-
sition of the spatulate end of the stylohyal, relative to
the plane of the TM, would aect the degree to which
the TM is displaced in response to hyoid-borne sound.
Support for this would be found if hyoid-borne sound
in LDC bats displaced the TM more than in HDC bats
( Fig. 2 ).
Methods
Specimens, scanning, and construction of 3D
models
Models of the hyoid apparatus, auditory bullae, and
TM were built from μCT data of R. ferrumequinum
(HD C/NB), R. rouxi (HD C/NB), R. hildebrandtii
(HDC/NB), Hipposideros diadema (HDC/NB),
Megaderma spasma (LDC/FM) and A. jamaicensis
(LDC/FM) specimens ( Ta b l e 1 ; Fig. 3 A–F). Species
were selected to provide variable hyoid morphology
for our models, variable phylogenetic position (Yinte-
rochiroptera and Yangochiroptera), and variable call
structure at the level of the larynx (e.g., LDC/FM vs
HDC/NB). However, we were restricted to species
that have extensively ossied hyoids, as this allowed
easy segmentation with traditional μCT and did not
require staining museum specimens with contrast.
Three Rhinolophus species were selected as we noticed
variation in the morphology of the anterior cornua
among species within this genus during our survey
of available μCT datasets. Specically, the number
of ossied elements proximal to the basihyal varied

4 C . C . G. Snipes and R. T. Car ter
Fig. 3 3D models/geometry used for the FE models. Ventral and lateral views of the hyoid apparatus and auditory bullae from ( A ) A. jamaicensis ,
( B ) M. spasma , ( C ) R. f er rumequinum , ( D ) R. rouxi , ( E ) R. hildebrandtii , and ( F ) H. diadema. Bones are color coded as follows: fused basihyal and
th yroh yals (blue), h ypoh yal (g reen), ceratoh yal (purple), styloh yal ( yellow), auditory bulla (orange), and inter vening car tilaginous segments
(gray).
in number ( Fig. 3 C–E), and we wanted to capture
potential performance dierences resulting from these
morphologies. The other families in this study exhib-
ited relatively uniform hyoid morphology within their
respective families. Including species that make use of
LD C/FM and HD C/NB echolocation ( Rhinolophus and
Hipposideros ) was important as these two groups have
been shown to undergo dierent ontogenetic steps
in the formation of their stylohyal—tympanic bone
articulation that leads to dierent adult morphology
( Fig. 2 ) and may represent convergent evolution of this
morphology ( Nojiri et al. 2021 ). Furthermore, the large
size of the cochlea in Rhinolophus and Hipposideros bats
results in an auditory bulla that contacts the cochlea

Hyoid conducted hearing in bats 5
and therefore could transmit sound into the inner ear
via bone conduction rather than through the TM and
ear ossicles.
The R. ferrumequinum and R. rouxi specimens were
provided by the American Museum of Natural His-
tory, and the H. diadema specimen was provided by
the Berkeley Museum of Verte b r a t e Zoology. All were
scanned with a Bruker Skyscan 1273 and reconstructed
with Bruker proprietary software at East Tennessee
State University. The M. spasma specimen was scanned
with a Nikon Metrology XT H 225 ST housed in the
Earth and Environmental Sciences Department at the
University of Michigan and provided by the Museum
of Zoology at the University of Michigan. The A. ja-
maicensis specimen was provided by the East Tennessee
State University and came from a captive colony housed
at the University of Northern Colorado (see Carter et
al. 2014 ). The A. jamaicensis specimen was scanned
using a Scanco μCT 50 at the Va n d erbi l t Center for
Small Animal Imaging at Vande r b i l t University and re-
constructed with the Datos ǀx 2 reconstruction software
(General Electric Company). To capture the shape of
the TM in A. jamaicensis, contrast enhanced μCT us-
ing phosphomolybdic acid (PMA; Gignac et al. 2016 )
was used, which allowed for the visualization of soft tis-
sues. Following staining with PMA, the specimen was
rescanned with a Bruker Skyscan 1273 at East Ten -
nessee State University. All specimens were stored in
70% ethanol.
Segmentations of the hyoid apparatus (basihyal,
hypohyal, ceratohyal, and stylohyal) and the middle ear
(auditory bullae, TM, tympanic annulus, and malleus)
were created in Dragony (Object Research Systems,
Montreal, Quebec, Canada) ( Fig. 3 A–F). The malleus
was included in each model due to its attachment to
the TM and apparent fusion to the auditory bullae in
the scans. For the A. jamaicensis model, the contrast
enhanced μCT data were imported into the same Drag-
ony session as the corresponding traditional μCT scan
and aligned using the image registration tool. This al-
lowed for the TM segmentation to be exported and
assembled in the anatomically correct location in space
relative to the traditional μCT scan. Ultimately, this
workow provided an accurate representation of the
TM and annulus for this species and informed us on
the precise attachment of the tympanic annulus to the
tympanic bone for the remaining models ( Fig. 4 A). To
build the TM for the remaining models, we used the
dierences in pixel intensity of the space on the medial
side of the TM (middle ear) compared to the lateral side
(external meatus) (4B). The dierence in pixel intensity
was likely due to the middle ear of the specimens being
lled with 70% ethanol while the meatus was l le d with
air during scanning. The TM was created by segmenting
Fig. 4 Transverse slices through the auditory bulla and cochlea from
a contrast enhanced μCT scan of A. jamaicensis ( A ) and a μCT scan
of M. spasma ( B ). Due to the contrast between the lateral and medial
sides of the TM, we were able to segment the negative space on the
lateral side (i.e., air) to get the shape of the TM, which was used to
construct the TM with NURBS surfaces ( C and D ). The medial side
of the TM was likely lled with uid as these specimens were stored in
70% ethanol, resulting in the contrast between the outer and middle
ear ca vities. Ar rows indicate the manubrium of the malleus, which
articulates with the medial side of the TM. Note that the slices are
from dierent planes within the middle and inner ears.
the air on the lateral side of the TM (acoustic meatus)
and segmentations were then exported as triangu-
lated surfaces (.stl les) and assembled in SpaceClaim
(Canonsburg, PA, USA). The 3D sketch and skin sur-
face tools were used to wrap a Non-uniform Rational B-
spline (NURBS) on the surface of the air segmentation
that contacted the TM, creating the surface of the TM
for most specimens ( Fig. 4 C–D). We also laid NURBS
on the TM segmented from the contrast enhanced μCT
dataset to build the A. jamaicensis model. To attach
the TM to the bullae, we built the tympanic annulus by
laying a NURBS surface on the area where it attaches
to the bullae and then used the “funnel” tool to blend
that surface to the corresponding edge of the TM. The
cartilaginous segments between each bony segment of
the hyoid were also created by laying NURBS surfaces
on the ends of each corresponding bone, blending them
using the “funnel” tool, and then converting them to

6 C . C . G. Snipes and R. T. Car ter
Ta b l e 2 Material proper ties f or bone and car tilage ( Cur rey, 2006 )
and tympanic membrane (TM) and tympanic annulus (TA) ( Caminos
et al., 2018 ) assigned to all FE models.
Material Density (g/cm
3
) Young’s modulus (Pa) Poisson’s ratio
Bone 2000 2.00E + 10 0.3
Cartilage 1100 1.20E + 07 0.3
TM 1200 3.20E + 07 0.3
TA 1200 3.20E + 07 0.3
triangulated surfaces. All triangulated surfaces were
then regularized to ensure a uniform mesh, converted
to solid bodies (except the TM, which was modeled as
a surface body), and saved as Spaceclaim les for FE
analysis within ANSYS (Canonsburg, PA, USA).
FE setup, TM validation, hearing threshold, and
harmonic response analyses
We ran a series of harmonic response analyses using
modal superposition within ANSYS. Material proper-
ties for bone, cartilage, the tympanic annulus, and the
TM were assumed to be isotropically elastic ( Dumont
et al. 2005 ; Snipes and Carter 2022 ). Bone, cartilage,
the tympanic annulus, and the TM were assigned mate-
rial properties taken from the literature ( Currey 2006 ;
Caminos et al. 2018 ) ( Tabl e 2 ). Although the TM in
bats can range from 20–100 μm thick ( Henson 1970 ),
the TM of A. jamaicensis was 40 μm thick (contrast-
enhanced μCT) and thus used in all TM models for
uniformity. This was necessary to ensure all TMs be-
haved similarly to airborne sound so that dierences in
its response to hyoid-borne sound could be attributed
to hyoid morphology alone. All connections were as-
signed as bonded (no sliding or separation between
faces or edges), and a contact tool was used to ensure
contacts had been assigned accurately by ANSYS. Sim-
ilar to Snipes and Carter (2022) , a series of xed sup-
ports were added to the tympanic bullae, basihyals, and
thyrohyals to hold the model in space along surfaces
that closely articulate with the surrounding anatomy
( Fig. 5 A). Due to the large number of muscle at-
tachments on the basihyal, we added ve xed sup-
ports on its ventral surface of the basihyal which rep-
resent attachments of the geniohyoideus, hyoglossus,
mandibulo-hyoid, and sterno-hyoideus ( Griths 1982 ,
1994 ; Griths et al. 1992 ). One xed point was added to
the ends of each of the thyrohyals to simulate the articu-
lation with the thyroid cartilage of the larynx. Four xed
points were assigned along the surface of the auditory
bullae that closely articulates with the skull. A damp-
ing coecient of 0.02 was applied to the entire model to
account for the loss of kinetic/oscillatory energy to the
surrounding tissue via friction ( Dodge et al. 2012 ). All
solid bodies (bones, cartilages, and the tympanic annu-
Fig. 5 Ventral ( A ) and dorso-lateral ( B ) views of the geometry from
the A. jamaicensis FE model. All xed points are indicated with red tri-
angles ( A ) with: four xed supports on the ventral surface of the basi-
hyal to model muscle attachments, two xed supports on the ends
of the th yroh yals to model their attachment to the larynx, and ve
xed supports on the surface of the auditory bullae that closely ar-
ticulates with the skull. TM displacement data were generated in the
axis orthogonal to the plane of the TM, indicated by the blue axis
on the triad ( B ). Bones are color coded as follows: fused basihyal
and th yroh yals (blue), h ypoh yal (g reen), ceratoh yal (purple), stylo-
hyal (yellow), auditory bulla (orange), and intervening cartilaginous
segments (gray).
lus) were meshed with a ne mesh using quadratic ele-
ments, and the TM surface body was meshed with shell
elements. The resultant meshes of all six models ranged
from 200,000 to 360,000 10-noded tetrahedral elements
with 400,000–700,000 nodes. The resultant meshes of
the TMs ranged from 4177 to 10,690 shell elements with
8538–19,959 nodes.
The rst analysis was set up to ensure the TM models
functioned realistically in response to airborne sound
and that the membrane from each model had similar
displacement values across frequencies. This was done
to ensure that any dierence in the TM’s response to
hyoid-borne sound was due to variation in hyoid mor-
phology and not something intrinsically dierent be-
tween the TM models themselves. To verify the models
behaved like that of a real TM, we compared the dis-
placement behavior of the TM models to experimental
data collected from the TM of a live bat. Manley et al.
(1972) excited the TM of a live Eptesicus pumilis with
a 100 dB airborne sound and, along with velocity data,
recorded an average maximum displacement value of
6.5e–8 m at 2.5 kHz and 2.9e–10 m at 100 kHz with an
overall decrease in the average maximum displacement
values as frequency increased. To recreate this experi-
ment with the digital models, we excited the lateral side
of each TM model with a sinusoidal (harmonic) pres-
sure of 2 Pa (equivalent to 100 dB, SPL ref 20 μPa; all
reports of sound pressure level hence forth are refer-
enced to 20 μPa) and recorded the average maximum

Hyoid conducted hearing in bats 7
Fig. 6 Geometry, including excitation surfaces and surfaces from which results data were generated, for the Harmonic response/modal su-
perposition analyses on R. f er rumequinum . ( A ) To verify the TM geometry, the TM was excited with 100 dB sound/pressure on the lateral
surface, and response data were generated from the same lateral surface of the TM. ( B ) To establish the displacement hearing threshold, the
validated TM was then excited with 0 dB sound/pressure, and response data were generated from the same lateral surface of the TM. ( C )
To mimic an outgoing echolocation call, the laryngeal surface of the basihyal was excited with a 120 dB sound/pressure, and response data
were generated from the lateral surface of the TM. Bones are color coded as follows: fused basihyal and th yroh yals (blue), h ypoh yal (g reen),
ceratohyal (purple), stylohyal (yellow), and auditory bulla (orange). The intervening cartilage segments are gray.
displacement values of the TM in the axis orthogonal to
the plane of the TM ( Fig. 5 B) from 0–150 kHz ( Fig. 6 A).
Additionally, when exposed to sound, a real TM does
not displace as a rigid, piston-like unit but instead with
spatial patterns ( Khanna and Ton n d o r f 1972 , Cheng
et al. 2019 ). Contour plots were generated at 50, 100,
and 150 kHz to conrm the modeled TM displacements
in response to airborne sound were realistic.
Once each TM model was validated, we estimated
a hearing threshold by measuring the displacement of
each TM model when excited with the lowest intensity
of airborne sound that is audible in bats. As these are
linear models, we divided the TM displacement at
100 dB SPL by a factor of 1e5, which is equivalent to
exciting the TM with 20 μPa (0 dB) and measuring the
average maximum displacement of the TM from 0–150
kHz ( Fig. 6 B). We chose 20 μPa (0 dB) because species
such as Desmodus rotundus and R. ferrumequinum have
been reported to hear sounds as low as –5 dB and other
species have thresholds slightly above 0 dB ( Hener
et al. 2013 ). Although hearing thresholds do vary across
species, we feel that 0 dB eectively approximates
the lowest hearing threshold in most echolocating
bats.
Lastly, to measure the TM response to an outgoing
echolocation call, the laryngeal surface of the basihyal
was excited with a sinusoidal pressure of 20 Pa (120 dB),
and the average maximum displacement values in the
axis orthogonal to the plane of the TM were measured
from 0–150 kHz ( Fig. 6 C). The intensity of the outgoing
call was chosen based on evidence that Rhinolophus
bats emit calls as loud as 28.2 Pa (123 dB) mea-
sured around 10 cm from the face ( Wat e rs and Jones
1995 ), and phyllostomid fruit bats, like A. jamaicensis ,

8 C . C . G. Snipes and R. T. Car ter
Fig. 7 TM displacements (m) in response to a 100 dB excitation on the lateral surface of the TM from H. diadema (solid blue line), R. f er rume-
quinum (dashed green line), R. rouxi (dotted green line), R. hildebrandtii (solid green line), M. spasma (solid yellow line), and A. jamaicensis (solid
red line) . Data were generated in the axis orthogonal to the plane of the TM and therefore in the direction that would set the ear ossicles
into motion during airborne hearing. The experimental data from Manley et al. (1972) used to verify our TM models are indicated by the black
horizontal lines at 6.5e–8 m (2.5 kHz) and 2.9e–10 m (100 kHz).
can emit calls as loud as 6.3 Pa (110 dB) ( Brinkløv
et al. 2009 ). Average maximum displacement values
were compared to the estimated hearing threshold
to determine if a bat could hear hyoid conducted
sound via the TM during call emission. For these
models, data points falling above the average hear-
ing threshold (established in the previous analy-
sis) were considered audible, and conversely, data
points falling below that threshold were considered
inaudible.
Results
TM validation through airborne sound and
estimated hearing threshold
Although our TM displacement data did not exactly
match those reported by Manley et al. (1972) , we con-
sidered them realistic enough to test our hypothe-
ses ( Fig. 7 ). Additionally, examination of the contour
plots at 50, 100, and 150 kHz indicates the TM re-
sponds with spatial patterns that are qualitatively sim-
ilar to those of other species ( Fig. 8 ) ( Khanna and
Tonndor f 1972 ; Cheng et al. 2019 ). Given the sim-
ilar displacement values of each TM in response to
a stimulus at 100 and 0 dB (determined in the pre-
vious analysis) across all frequencies (0–150 kHz),
these data were averaged and used to represent the
estimated hearing threshold across all models in this
study.
Vibroacoustic response of the TM to hyoid-borne
sound
For all species, TM displacements were orders of magni-
tude greater than the estimated hearing threshold dur-
ing hyoid-borne propagation of sound ( Fig. 9 ). There
were no discernible patterns in the performance of
LD C/FM vs HD C/NB models in displacing the TM, nor
were there obvious eects due to the variation in mor-
phology of the proximal elements of the anterior cor-
nua in the dierent Rhinolophus species. The contour
plots of the TM at 50, 100, and 150 kHz depict spatial
displacement patterns similar to that of airborne sound
and thus indicate that the TM responds to hyoid-borne
sound ( Fig. 8 ).
Discussion
Our results support the hypothesis that hyoid-borne
sound generated during echolocation call emission
would stimulate the TM within a range likely heard by
bats. Moreover, the TM displacements were orders of
magnitude greater than the estimated hearing threshold
in most of our models. While our previous work indi-
cated that sound could arrive at the bulla with an inten-
sity that bats could likely hear, we did not test whether
it was transferred to the inner ear (or cochlea) through
the middle ear or by direct stimulation via bone/soft
tissue conduction ( Snipes and Carter 2022 ). The
data from the present study show that TM vibration,

Hyoid conducted hearing in bats 9
Fig. 8 Displacement contour plots of the TM at 50, 100, and 150 kHz in response to airborne and hyoid-borne sound excitations. Warmer
colors indicate areas of greater displacement (peaks), while cooler colors indicate areas with less displacement. TM displacements across a
range of frequencies are characterized by f e wer peaks at lower frequencies and an increase in the number of peaks as frequency increases.
Our results show that this is the case in both air- and hyoid-borne excitation.
Fig. 9 Average maximum TM displacements (m) from 0 to 150 kHz in response to 120 dB excitation on the laryngeal surface of the basihyal
from H. diadema (solid blue line), R. f er rumequinum (dashed green line), R. rouxi (dotted green line), R. hildebrandtii (solid green line), M. spasma
(solid yellow line), and A. jamaicensis (solid red line) . Data were generated in the plane orthogonal to the plane of the TM and therefore in
the direction that would set the ear ossicles into motion during airborne hearing. The average maximum TM displacement in response to a
0 dB excitation on the lateral side of the TM was measured from 0 to 150 kHz, and the average from each model was used to represent the
average lowest hearing displacement threshold across species (solid black line).
like that during airborne hearing, can be used to
transfer the hyoid-borne call through the middle ear.
Our second hypothesis, that gross dierences in sty-
lohyal morphology and its articulation with the bulla
would aect the performance of the system, was not
supported. The TMs from LDC and HDC bats re-
sponded similarly at frequencies below 20 kHz, and al-
though there was some variation at frequencies above

10 C . C . G. Snipes and R. T. Car ter
20 kHz, it did not fall along any obvious morpho-
logical or phylogenetic lines ( Fig. 8 ). The R. ferrme-
quinum model diered from those of R. rouxi and
R. hildebrandtii , particularly at frequencies above 70
kHz, where there was a reduction in the amplitude
of displacements. Compared to the R. rouxi and R.
hildebrandtii models, R. ferrumequinum had the largest
number of bony elements (four) that make up the an-
terior cornua which could be responsible for the lower
displacements in the higher frequency range. However,
if element number of the anterior cornua alone ex-
plained variation in the TMs response to hyoid-borne
sound, then the R. rouxi model (two elements) would
have resulted in the largest displacements among Rhi-
nolophus models. This was not the case, as the R. hilde-
brandtii model (three elements) had the highest TM
displacements across most frequencies. This indicates
that the vibroacoustic response of the hyoid, bulla, and
TM during sound transfer is complex and hard to esti-
mate using gross morphology alone.
It also means that detailed performance is hard to ac-
curately model without more data on the exact nature
of materials, connections, and boundary conditions for
each species used. As we assumed many variables to be
equal across our models, we included values that do not
exactly match reality. For example, we modeled all our
TMs using the same material properties and thickness
when there is likely variation across species. This is not
a problem when testing hypotheses on the overall abil-
ity of the system to transmit sound into the ear but can
become problematic when testing hypotheses on ner
scale performance dierences between dierent groups
of bats.
The emitted echolocation pulse can be as high as
ve orders of magnitude louder than the returning
echo, and raises the question of how bats can hear an
echo after such a loud initial pulse. Some research sug-
gests that the middle ear muscles contract during vocal-
ization and might attenuate the initial outgoing pulse
as it passes through the middle ear ( Henson 1965 ),
but experiments on the role of these muscles in at-
tenuating the bats own vocalization have yielded con-
tradictory results and remain unresolved ( Neuweiler
2000 ). If it is the case that middle ear muscles (e.g.,
stapedius) help attenuate these calls during produc-
tion, then a route from the hyoid into the TM and
through the middle ear would provide a route where
attenuation of the call can be modulated as needed. A
bone conducted route, where the call passes into the
bulla from the hyoid and then stimulates the cochlea
directly through vibration, would bypass the ear os-
sicles and thus not provide an opportunity for atten-
uation via middle ear muscles. Of course, some bats
could use a combination of the two routes (middle ear
and bone conducted) to stimulate the cochlea and pro-
vide the neurologic registration of the outgoing call
in the brain. This scenario is more likely in Rhinolo-
phus and Hipposideros , as they possess large cochlea that
contact the bulla and could therefore transfer a vibra-
tion from the bulla directly into the cochlea. Further-
more, our previous models found that R. pusillus was
capable of eectively displacing the bulla during hyoid-
borne sound transfer, whereas A. jamaicensis was not
( Snipes and Carter 2022 ).
As previously mentioned, HDC/NB echolocators
would likely experience an outgoing call reaching the
TM through the hyoid and a returning echo reaching
the TM through the air at the same time. This scenario
would result in mixing of the two signals at the TM or
ear ossicles. If these two signals are at slightly dierent
frequencies due to Doppler shift, they will presumably
create a beat-note through periods of constructive and
destructive interference when mixed ( Wittrock 2010 ). If
the two arriving calls are at the same frequency and in
phase, then TM or ear ossicle displacement is expected
to be maximized through constructive interference.
Conversely, if the arriving calls are the same frequency
but out of phase, then TM or ear ossicle displacement
is expected to be attenuated through destructive inter-
ference. All of these call interactions require explicit dy-
namic FE modeling where each time step is dened, and
therefore where not modeled in the present study.
While we did include variable hyoid morphology
from various bat taxa in this study, the hyoid models
all contained a series of bones and cartilaginous seg-
ments, which is not the case in all echolocators ( Sprague
1943 ). In some genera of laryngeal echolocators ( Eptesi-
cus , Minipterus , and Kerivoula ) the proximal hypohyal
is fascial, which may aect the ability of the hyoid
to transmit the outgoing call from the larynx to the
ear. Interestingly, tongue clicking echolocators in the
genus Rousettus lack a stylohyal—tympanic bone ar-
ticulation but do possess a facial attachment between
the two bones. If fascial connections within the hyoid
pose no problem in transmitting the outgoing call to
the ear, then a tongue—generated call could also pass
through the hyoid to the ear as the basihyal provides at-
tachment for tongue musculature. Of course, all these
hypotheses remain to be tested and warrant further
modeling.
In summary, the hyoid of laryngeally echolocating
bats can transfer laryngeally produced sound to the
ear, but we found no gross morphological patterns
associated with sound transfer eciency. This suggests
that hyoid morphology and complexity within and be-
tween HDC/NB and LDC/FM echolocators have been
minimally aected by selection for sound conduction.
Given the range of functions associated with the hyoid,

Hyoid conducted hearing in bats 11
along with specialized vocalizations and call emission
in echolocating bats (i.e., nasal vs oral emission), we
suggest that a comparative functional study of its me-
chanical properties alongside a broad morphological
evaluation could provide insight into the evolution of
echolocation in bats and a novel view of the integration
and evolvability of the hyoid apparatus across taxa.
Acknowledgements
The authors thank Peter Newman for helping build the
FE models, without his expertise this paper would not
have been possible. We also thank Dr Nancy Simmons
and Eleanor Hoeger from the American Museum of
Natural History and Dr Chris Conroy at the Museum of
Vert e b r a t e Zoology, Berkley, for loaning us specimens.
The authors also want to thank Dr Cody Thompson
at the University of Michigan Museum of Zoology for
sharing the M. spasma scan . We would also like to thank
Dr Scott Pedersen for reviewing earlier drafts of this
manuscript.
Funding
This research was supported with startup funds from
East Tennessee State University and a Major Re-
search Instrumentation Grant from the National Sci-
ence Foundation (NSF award # 2018559).
Declaration of competing Interests
The authors declare no competing interests.
Data Availability
All triangulated surface models and the μCT datasets
from which they were derived are available via a unique
identier on Morphosource ( w w w.mor p hosour ce.org )
( Table 1 ).
References
Brinkløv S, Kalko EK, Surlykke A. 2009. Intense echolocation
calls from two “whispering” bats, Artibeus jamaicensis and
Macrophyllum macrophyllum (Phyllostomidae). J Exp Biol
212:11–20.
Caminos L, Garcia-Manrique J, Lima-Rodriguez A, Gonzalez-
Herrera A. 2018. Analysis of the mechanical properties of
the human tympanic membrane and its inuence on the dy-
namic behaviour of the human hearing system. Appl Bionics
Biomech 2018:1736957.
Carter RT. 2020. Reinforcement of the larynx and trachea in
echolocating and non-echolocating bats. J Anat 237: 495–503.
Carter RT, Shaw JB, Adams RA. 2014. Ontogeny of vocalization
in Jamaican fruit bats with implications for the evolution of
echolocation. J Zool 293:25–32.
Cheng JT, Maftoon N, Guignard J, Ravicz ME, Rosowski J. 2019.
Tympanic membrane surface motions in forward and reverse
middle ear transmissions. J Acoust Soc Am 145:272.
Currey JD. 2006 Bones: structure and mechanics. Vol . 269.
Princeton (NJ): Princeton University Press. p. 124–9.
Dumont ER, Piccirillo J, Grosse IR. 2005. Finite-element analysis
of biting behavior and bone stress in the facial skeletons of bats.
Anat Rec A Discov Mol Cell Evol Biol 283:319–30.
Dodge T, Wanis M, Ayoub R, Zhao L, Watts NB, Bhattacharya
A, Akkus O, Robling A, Yok o t a H. 2012. Mechanical load-
ing, damping, and load-driven bone formation in mouse tib-
iae. Bone 51:810–8.
Fenton MB, Faure PA, Ratclie JM. 2012. Evolution of high duty
cycle echolocation in bats. J Exp Biol 215:2935–44.
Gignac PM, Kley NJ, Clarke JA, Colbert MW, Morhardt AC, Ce-
rio D, Cost IN, Cox PG, Daza JD, Early CM et al. 2016. Dif-
fusible iodine-based contrast-enhanced computed tomogra-
phy (diceCT): an emerging tool for rapid, high-resolution, 3-D
imaging of metazoan soft tissues. J Anat 228:889–909.
Griths TA. 1982. Systematics of the new world nectar-feeding
bats (Mammalia, Phyllostomidae), based on the morphology
of the hyoid and lingual regions. American Museum Novitates
No. 2742. New Yo r k (NY): American Museum of Natural His-
tory, p. 1–45.
Griths TA. 1994. Phylogenetic systematics of slit- faced bats
(Chiroptera, Nycteridae) based on hyoid and other morphol-
ogy. American Museum Novitates No. 3090. New Yo r k (NY):
American Museum of Natural History, p. 1–17.
Griths TA, Truckenbrod A, Sponholtz PJ. 1992. Systematics of
megadermatid bats (Chiroptera, Megadermatidae) based on
hyoid morphology. American Museum Novitates No. 3041.
New Yor k (NY): American Museum of Natural History, p. 1–
21.
Hener RS, Koay G, Hener HE. 2013. Hearing in American leaf-
nosed bats. IV: the common vampire bat, Desmodus rotundus .
Hear Res 296:42–50.
Henson OW. 1965. The activity and function of the middle-ear
muscles in echo-locating bats. J Physiol 180:871–87.
Henson OW. 1970. The ear and audition. In:WA, Wimsatt (ed.).
Biology of bats. New Yo r k (NY): Academic Press.
Khanna SM, Tonndorf J. 1972. Tympanic membrane vibrations
in cats studied by time-averaged holography. J Acoust Soc Am
51:1904–20.
Kössl M, Va t e r M. 1995. Cochlear structure and function in bats.
In:AN Popper, RR, Fay (eds.). Hearing by bats. New Yo r k (NY):
Springer, p. 191–234.
Manley GA, Irvine DRF, Johnstone BM. 1972. Frequency re-
sponse of bat tympanic membrane. Nature 237:112–3.
Neuweiler G. 2000. The biology of bats. Oxford: Oxford Univer-
sity Press, p. 140–206.
Nojiri T, Wilson LA, López-Aguirre C, Tu VT, Kuratani S, Ito
K, Higashiyama H, Son NT, Fukui D, Sadier A et al. 2021.
Embryonic evidence uncovers convergent origins of laryngeal
echolocation in bats. Curr Biol 31:1353–65.
O’Neill WE, Suga N. 1979. Ta r g et range-sensitive neurons in the
auditory cortex of the mustache bat. Science 203: 69–73.
Snipes CCG, Carter RT. 2022. The hyoid as a sound conducting
apparatus in laryngeally echolocating bats. J Anat 240:1020–
33.
Sprague JM. 1943. The hyoid region of placental mammals with
especial reference to the bats. Am J Anat 72:385–472.
Suga N, O’Neill WE, Manabe T. 1979. Harmonic-sensitive neu-
rons in the auditory cortex of the mustache bat. Science
203:270–4.
Vat e r M, Kössl M. 2011. Comparative aspects of cochlear func-
tional organization in mammals. Hear Res 273:89–99.

12 C . C . G. Snipes and R. T. Car ter
Veselka N, McErlain DD, Holdsworth DW, Eger JL, Chhem
RK, Mason MJ, Brain KL, Faure PA, Fenton MB. 2010. A
bony connection signals laryngeal echolocation in bats. Na-
ture 463:939–42.
Wat e r s DA, Jones G. 1995. Echolocation call structure and in-
tensity in ve species of insectivorous bats. J Exp Biol 198:
475–89.
Wittrock U. 2010. Laryngeally echolocating bats. Nature 466: E6.