Musculoskeletal morphogenesis supports the convergent evolution of bat laryngeal echolocation

Abstract

The order Chiroptera (bats) is the second largest group of mammals. One of the essential adaptations that have allowed bats to dominate the night skies is laryngeal echolocation, where bats emit ultrasonic pulses and listen to the returned echo to produce high-resolution ‘images’ of their surroundings. There are two possible scenarios for the evolutionary origin of laryngeal echolocation in bats: (1) a single origin in a common ancestor followed by the secondary loss in Pteropodidae, or (2) two convergent origins in Rhinolophoidea and Yangochiroptera. Although data from palaeontological, anatomical, developmental and genomic studies of auditory apparatuses exist, they remain inconclusive concerning the evolutionary origin of bat laryngeal echolocation. Here we compared musculoskeletal morphogenesis of the larynx in several chiropteran lineages and found distinct laryngeal modifications in two echolocating lineages, rhinolophoids and yangochiropterans. Our findings support the second scenario that rhinolophoids and yangochiropterans convergently evolved advanced laryngeal echolocation through anatomical modifications of the larynx for ultrasonic sound generation and refinement of the auditory apparatuses for more detailed sound perception.

1. Introduction

The order Chiroptera (bats) is the second largest mammalian order, including more than 1400 species [1,2]. The adaptive radiation of bats and consequently high species richness are largely due to the acquisition of powered flight and echolocation [1]. These unique abilities have enabled bats to fly and feed in the night sky [1,3–6]. Most echolocating bats use laryngeal echolocation, emitting ultrasonic sound (up to 20 kHz) pulses from their mouth or nose and interpreting reflected echoes to determine the position and characteristics of surrounding objects [1,3,5,6]. The laryngeal echolocation is achieved by three consecutive activities [7]: (1) generation of ultrasonic sounds in the larynx, (2) reception of ultrasonic sounds with the auditory apparatus, and (3) reconstruction of spatial image based on the acoustic information processed by appropriate brain areas. The family Pteropodidae are the only bats that do not use laryngeal echolocation, with most pteropodid species not echolocating at all (figure 1a).

Figure 1.

Evolutionary scenarios for the origin of laryngeal echolocation in bats and the functional mechanism for ultrasonic generation in bat larynxes. (a) Molecular phylogeny of bats on which acquisition timing of laryngeal echolocation ability was mapped. The families including the species used in the adult anatomical examinations are highlighted with bold characters. The families including the species used in the embryological analyses are highlighted with underlines. (b) Schematic representations of a miniopterid bat larynx illustrating the general mechanism for high-frequency sound generation in bats: contraction of the anterior and posterior cricothyroid muscles (1), followed by ventroposterior rotation of thyroid cartilage (2), prevention of ventroposterior rotation of arytenoid cartilage by contraction of posterior cricoarytenoid muscle which acts as antagonist muscle to the anterior and posterior cricothyroid muscles (3), and finally, production of ultrasonic sounds by the vibration of the stretched vocal fold and vocal membrane (4). The orientation of the larynx illustrations is indicated to their upper-right: a, anterior; d, dorsal; p, posterior; v, ventral. ac, arytenoid cartilage; cc, cricoid cartilage; CT, cricothyroid muscle; hy, hyoid bone; PCA, posterior cricoarytenoid muscle; tc, thyroid cartilage; tr, trachea; vf, vocal fold; vm, vocal membrane.

Molecular-based current phylogeny of bats divides the order into two suborders: Yinpterochiroptera (Pteropodidae and the superfamily Rhinolophoidea) and Yangochiroptera (all other families) [8–10] (figure 1a). This phylogenetic relationship suggests two possible scenarios for the evolution of laryngeal echolocation [8,11]. The first scenario is that laryngeal echolocation was acquired in the common ancestor of bats and subsequently lost in the family Pteropodidae (figure 1a). The second scenario is that laryngeal echolocation was convergently acquired by the superfamily Rhinolophoidea and suborder Yangochirotpera.

Evidence for the two possible scenarios of laryngeal echolocation is mixed. Fossil bat skull morphology suggests that the common ancestor of Chiroptera used echolocation, supporting the first scenario [12,13] (but see for an opposed view [14]). Furthermore, it has been reported that the relative size of auditory brain regions (e.g. the auditory cortex and inferior colliculus) of the chiropteran ancestor is equivariant to that of extant echolocating bats, supporting the first scenario as well [15]. A study that investigated the growth rate of the cochlea during bat embryogenesis showed that the cochlea in all lineages, including non-echolocating pteropodid bats, was significantly larger than that of non-echolocating mammals at its initial developmental stage, providing further support for the first scenario [16].

By contrast, the internal microstructure and developmental pattern of the cochlea differ between echolocating rhinolophoid and yangochiropteran bats, suggesting the second scenario [17,18]. Furthermore, positive selection of the genes involved in hearing and vocalization has been detected in the two echolocating bat linages and not in non-echolocating pteropodid bats, supporting the second scenario that suggests the independent evolution of higher frequency hearing ability in the two echolocating taxa [19,20]. Consequently, the evolutionary history of bat laryngeal echolocation remains controversial [5,6].

In this paper, we examined whether the anatomy and morphogenesis of bat larynxes support a particular scenario for the evolutionary origin of laryngeal echolocation. Bats generate ultrasonic sounds by vibrating the vocal folds of the larynx (figure 1b and electronic supplementary material, figure S1). Several morphological adaptations have been reported in the larynx of yangochiropteran bats [21–24]. For example, an extremely thin membrane, called the vocal membrane, protrudes from the tip of the vocal folds and acts as a low-mass oscillator that can vibrate faster than the vocal folds, thereby generating ultrasonic sounds [24–26] (figure 1b). In addition, the cricothyroid muscle, whose contraction produces high-frequency sound in mammals (figure 1b and electronic supplementary material, figure S1), is enlarged and specialized into superfast muscle (SFM) in yangochiropteran bats, allowing them to emit several dozens of ultrasonic pulses per second [27–29]. Morphological and molecular comparisons of the larynx between echolocating and non-echolocating linages, and within echolocating linages, may shed light on the origin of chiropteran laryngeal echolocation [30].

Here, we first compared laryngeal musculoskeletal anatomy among echolocating Rhinolophoidea (3 species) and Yangochiroptera (3 species), as well as the non-echolocating Pteropodidae (3 species; electronic supplementary material, table S2). We then compared laryngeal musculoskeletal development among these three lineages. Finally, we investigated the expression patterns of myosin heavy and light chain family genes uniquely downregulated in the SFM to compare the specialization pattern of SFM among the three lineages. Our results suggest that laryngeal echolocation evolved convergently in the Rhinolophoidea and Yangochiroptera.

2. Results

(a) . Laryngeal musculoskeletal anatomy

We compared laryngeal musculoskeletal anatomy among bats with and without laryngeal echolocation to identify any underlying morphological differences (figure 2 and electronic supplementary material, figure S2). In the ventral side of the larynx, the spatial pattern and homology of the muscles were highly conserved among Pteropodidae (Rousettus aegyptiacus, R. leschenaultii, and Pteropus dasymallus), Rhinolophoidea (Rhinolophus ferrumequinum, Hipposideros turpis, Rhinopoma cystops), and the mole (Mogera imaizumii; order Eulipotyphla) analysed as an outgroup (figure 2b–d and electronic supplementary material, figure S2). The muscular architecture of the above bats was comparable even with those of mice and humans reported previously [31] (electronic supplementary material, figure S1).

Figure 2.

Comparison of laryngeal musculoskeletal anatomy in Mogera imaizumii, Rousettus aegyptiacus, Rhinolophus ferrumequinum, and Miniopterus fuliginosus. (a) Anterior views of the head of the animal examined with phylogenetic relationships shown above. (b–e) Ventral view of the larynx. Note that the ventral side of M. fuliginosus larynx is completely covered by muscular tissue. (f–i) Dorsal view of the larynx. (f′–i′) Horizontal sections at the plane shown by dotted lines in (f–i), respectively. The dorsal crest and posterior cricoarytenoid muscle of R. ferrumequinum are substantially enlarged. (j–m) Lateral view of the larynx. Remarkably, three muscles are present only in M. fuliginosus and the most anterior one, named preanterior cricothyroid muscle by the authors, has never been recognized so far. (n) Schematic illustrations of (j–m), respectively. The orientation of models and sections is indicated in the right side of panels: a, anterior; d, dorsal; p, posterior; v, ventral. ac, arytenoid cartilage; ACT, anterior cricothyroid muscle; cc, cricoid cartilage; dc, dorsal crest; PACT, prepanterior cricothyroid muscle; PCA, posterior cricoarytenoid muscle; PCT, posterior cricothyroid muscle; tc, thyroid cartilage. Scale bars are 10 mm.

By contrast, we identified unique morphological features in yangochiropteran species (figure 2e and electronic supplementary material, figure S2). For example, the right and left laminae of the thyroid cartilage (tc) were completely covered ventrally by muscular tissue in Miniopterus fuliginosus, Pipistrellus abramus, and Carollia perspicillata (figure 2e and electronic supplementary material, figure S2). Such expansion of the muscular tissue over the ventral side of the larynx has been reported in other yangochiropteran species, including Emballonura raffrayana (family Emballonuridae) [32], Pteronotus parnellii (family Mormoopidae) [33], and Nycteris javanica (family Nycteridae) [34]. Although the muscular tissue covering the ventral lamina of the thyroid cartilage (tc) in yangochiropteran bats was previously thought to be homologous with the anterior cricothyroid muscle (ACT), we found three distinct muscles in the ventral side of the yangochiropteran larynx, the anterior cricothyroid muscle (ACT) and posterior cricothyroid muscle (PCT) and a previously unreported muscle (figure 2m,n, and electronic supplementary material, figure S2). This novel muscle was located anterior to the anterior cricothyroid muscle (ACT) and inserted into the top edge of the thyroid cartilage (tc; figure 2e,m,n, and electronic supplementary material, figure S2). We named this novel muscle the ‘preanterior cricothyroid muscle’ (PACT). The anterior cricothyroid muscle (ACT) and posterior cricothyroid muscle (PCT) of yangochiropteran species were inserted into the bottom edge of the thyroid cartilage (tc) as in other mammalian larynxes (figure 2m,n). The cricoid cartilage (cc) of yangochiropteran species possessed a pair of unique anterolateral protrusions, providing attachment sites for the preanterior cricothyroid muscles (PACT), anterior cricothyroid muscles (ACT), and posterior cricothyroid muscles (PCT; figure 2n and electronic supplementary material, figure S2).

In the dorsal side of the larynx, the musculoskeletal architecture was highly conserved among Ptreropodidae, Yangochiroptera, the mole M. imaizumii, mice, and humans [31] (figure 2j,k,m,n; electronic supplementary material, figures S1 and S2). By contrast, rhinolophoid species uniquely possessed an extremely projected ‘dorsal crest’ (dc), a cartilaginous peak running anteroposteriorly along the midline of the cricoid cartilage (cc), as reported by [21] (figure 2h,h′, and electronic supplementary material, figure S2). Along with this dorsal projection, the posterior cricoarytenoid muscles (PCA) originating from the dorsal crest were substantially expanded in rhinolophoid species compared to pteropodids, yangochiropterans, and non-bat mammals (figure 2h,h′, and electronic supplementary material, figure S2).

(b) . Laryngeal musculoskeletal tissue development

To confirm the independence and developmental origin of the preanterior cricothyroid muscle uniquely observed in the larynx of yangochiropteran species, we compared laryngeal musculoskeletal morphogenesis among representative species of Pteropodidae (Rousettus aegyptiacus), Rhinolopoidea (Rhinolophus ferrumequinum), and Yangochiroptera (Miniopterus fuliginosus). At embryonic stage (St.) 17, an anterior protrusion of muscular tissue was only observed in the anterior cricothyroid muscle (ACT) of M. fuliginosus (presented by asterisks in figure 3d,d″). At St. 20, the anterior protrusion was split from the anterior cricothyroid muscle (ACT) and formed the independent preanterior cricothyroid muscle (PACT) (figure 3g,g″).

Figure 3.

Larynx morphogenesis in bats with or without laryngeal echolocation ability. (a) Embryo of Rhinolophus ferrumequinum at embryonic stage (St.) 20 (left), a three-dimensional model of the embryonic head prepared from serial sections (centre), and a higher magnification reconstructed image of the laryngeal musculoskeletal tissue (right). (b–d) Three-dimensional model lateral views of the larynxes of Rousettus aegyptiacus, R. ferrumequinum, and Miniopterus filiginosus at St. 17. (e–g) Three-dimensional model lateral views of larynxes at St. 20. (b′–g′,b″–g″) Histological sections of the planes shown by dotted lines in the three-dimensional models (b–g), respectively. Model and section orientations are indicated on the far right: a, anterior; d, dorsal; p, posterior; v, ventral. Three-dimensional models and horizontal sections show cartilage (light blue), the posterior cricothyroid and posterior cricoarytenoid muscles (light red), preanterior cricothyroid muscle (pink), and anterior cricothyroid muscle (red). Asterisks in (d,d″) show the anterior protrusion of muscular tissue from the developing anterior cricothyroid muscle. The muscular protrusion splits from the anterior cricothyroid muscle to become an independent ‘preanterior cricothyroid muscle’ in St. 20 M. fuliginosus embryos. ac, arytenoid cartilage; ACT, anterior cricothyroid muscle; cc, cricoid cartilage; dc, dorsal crest; e, oesophagus; PACT, prepanterior cricothyroid muscle; PCA, posterior cricoarytenoid muscle; PCT, posterior cricothyroid muscle; tc, thyroid cartilage; tr, trachea. Scale bars are 10 µm.

At St. 17, the relative height of the dorsal crest (dc) was similar among the three bat species (figure 3b′–d′). However, mesenchymal cells were highly condensed in dorsal crest's (dc) dorsal region in R. ferrumequinum, foreshadowing the substantial elongation of the dorsal crest (dc) in later stage embryos (figure 3c′). At St. 20, the dorsal crest's (dc) relative height in R. ferrumequinum was already comparable to that of adults (figure 3f′).

(c) . Superfast muscle specific gene expression during laryngeal ontogeny

Previous studies reported that the anterior cricothyroid muscle (ACT) was specialized into SFM in Yangochiroptera, and that genes encoding slow-twitch muscle fibre proteins, including Myosin heavy chain 6 (Myh6) and Myosin light chain 3 (Myl3), were substantially downregulated in the muscles of adult yangochiropterans [28,29]. To assess when the anterior cricothyroid muscle (ACT) specializes into SFM during yangochiropteran ontogeny, as well as whether the laryngeal muscles of pteropodid and rhinolophoid bats also specialize into SFM, we compared the expression patterns of Myh6 and Myl3 using in situ hybridization on St. 20 R. aegyptiacus, R. ferrumequnum, and M. fuliginosus embryos (figure 4). Myosin heavy chain 4 (Myh4), which is expressed in all muscles, was used as a control. All three genes were expressed in all laryngeal muscular tissues and there were no differences in expression patterns (figure 4).

Figure 4.

Identical expression patterns of myosin genes in the laryngeal muscles of stage 20 bat embryos. (a–f) Myh4 in situ hybridization at the sections containing the posterior cricoarytenoid muscle (a–c) and cricothyroid muscles (d–f). (g–l) Myh6 in situ hybridization at the sections containing the posterior cricoarytenoid muscle (g–i) and cricothyroid muscles (j–l). (m–r) Myl3 in situ hybridization at the sections containing the posterior cricoarytenoid muscle (m–o) and cricothyroid muscles (p–r). These myosin genes were expressed in all laryngeal muscular tissues of the St. 20 bat embryos. Section orientations are indicated on the far right: d, dorsal; v, ventral. ACT, anterior cricothyroid muscle; cc, cricoid cartilage; dc, dorsal crest; e, oesophagus; PACT, prepanterior cricothyroid muscle; PCA, posterior cricoarytenoid muscle; PCT, posterior cricothyroid muscle; tc, thyroid cartilage; tr, trachea. Scale bars are 100 µm.

3. Discussion

This is the first study to explore the origin of laryngeal echolocation by comparing laryngeal musculoskeletal morphogenesis among bat orders. Our results revealed that the larynxes of the two echolocating linages, Rhinolophoidea and Yangochiroptera, have been anatomically modified in different ways to generate ultrasonic sounds (figure 5), while the only non-echolocating linage, Pteropodidae, has retained laryngeal morphology similar to other mammals (figure 5).

Figure 5.

Evolutionary scenario of ultrasonic sound generation suggested by the comparison of laryngeal morphogenesis among three bat linages. Our study demonstrated that the larynx of Rhinolophoidea has been modified on the dorsal side, with a dorsally elongated dorsal crest and expansion of the posterior cricoarytenoid. On the other hand, the larynx of Yangochiroptera has been modified on the ventral side, through acquisition of the novel preanterior cricothyroid muscle. Non-echolocating pteropodids preserve the ancestral laryngeal condition. Distinct modifications of the larynx suggest that ultrasonic sound generation has evolved convergently in two linages of echolocating bats.

Our anatomical and embryological analyses showed that the three yangochiropteran species representing the families Miniopteridae, Vespertilionidae, and Phyllostomidae possess the previously undescribed preanterior cricothyroid muscle (PACT), located ventral to the anterior cricothyroid muscle (ACT) (figures 2–4). The preanterior cricothyroid muscle (PACT) originates from the cricoid cartilage (cc), runs in the rostral direction covering the ventral surface of the thyroid cartilage (tc) completely, and inserts into the upper edge of the thyroid cartilage (tc) (figures 2 and 3, and electronic supplementary material, figure S1). This novel muscle is developmentally derived from the anterior cricothyroid muscle (ACT; figure 3).

Our results and those of previous studies suggest different morphological synapomorphies in the larynxes of Yangochiroptera and Rhinolophoidea. Schematic figures by Griffiths showed that the ventral plane of the thyroid cartilage (tc) was completely covered by muscular tissue in the yangochiropteran families, Emballonuridae, Nycteridae, and Mormoopidae [32–34]. This muscular tissue may correspond to the preanterior cricothyroid muscle (PACT) that we describe in this study, suggesting that the preanterior cricothyroid muscle (PACT) is common among Yangochiroptera (figure 5). The rhinolophoid species we studied (families Rhinolophidae, Hipposideridae and Rhinopomatidae) possessed dorsally elongated dorsal crest (dc) and highly developed posterior cricoarytenoid muscles (PCA) (figure 2; electronic supplementary material, figures S1 and S2). A previous study found that a species of the rhinolophoid family Megadermatidae also had an elongated dorsal crest (dc) with the posterior cricoarytenoid muscles (PCA), suggesting that these modifications may be Rhinolophoidea synapomorphies [22] (figure 5).

Yangochiropterans appear to produce the ultrasonic frequencies necessary for laryngeal echolocation through ventral modifications to the muscles between the thyroid cartilage (tc) and the cricoid cartilage (cc). These muscles play a major role in producing high frequency vocalizations in mammals [24] (figure 1 and electronic supplementary material, figure S1). Contraction of these muscles causes ventroposterior rotation of the thyroid cartilage (tc) and puts high tension on the vocal folds (vf), which can then be vibrated through exhalation to generate high frequency sounds (figure 1). In yangochiropteran species, because the preanterior cricothyroid muscle (PACT) is inserted into the upper edge of the thyroid cartilage, its contraction should enable the thyroid cartilage (tc) to tilt deeper ventroposteriorly, providing higher tension to the vocal folds (vf) and thus generating extremely high frequencies.

On the other hand, our results suggest that rhinolophoid species generate ultrasonic sounds using morphological modifications of the dorsal side of the larynx. The posterior cricoarytenoid muscles (PCA) play an antagonist role to the anterior cricothyroid muscles (ACT) and posterior cricothyroid muscles (PCT) during the generation of high-frequency sounds in mammals (figure 1). When the anterior cricothyroid muscles (ACT) and posterior cricothyroid muscles (PCT) are contracted, the posterior cricoarytenoid muscles (PCA) pull the arytenoid cartilages (ac) dorsally to provide tension to the vocal folds (vf) (figure 1b). In rhinolophoid species, the dorsally elongated dorsal crest (dc) should allow the posterior cricoarytenoid muscles (PCA) to pull the arytenoid cartilages (ac) further dorsally compared to that in other mammals, providing higher tension to the vocal folds (vf) for the production of ultrasonic frequencies.

Previous studies showed that the anterior cricothyroid muscle (ACT) of yangochiropterans is specialized into SFM, where the genes of slow-twitch muscle fibre proteins (e.g. Myh6 and Myl3) are substantially downregulated [28,29]. Since laryngeal morphology is almost identical between St. 20 bat embryos and adults, we speculated that expressions of Myh6 and Myl3 were substantially downregulated in the cricothyroid muscle (ACT) of M. fuliginosus at St. 20. However, we detected intense expression of Myh6 and Myl3 in all laryngeal muscles of St. 20 bat embryos, suggesting that the laryngeal muscles are not specialized into SFM during embryogenesis. Mead et al. [35] suggested that specialization of SFM in songbirds occurs after they hatch, facilitated by learning vocalizations from the parents [35]. Bats have also been reported to learn vocalizations, including echolocation frequencies, from their mothers [36]. Furthermore, it has been reported that ultrasonic sound generation at the larynx required at least two weeks after birth in bats [37]. Considering these findings, bat larynx muscles may specialize into SFM postnatally facilitated by vocal learning similar to songbirds, although laryngeal anatomical architecture is established prenatally as reported in this study.

The ‘single origin’ scenario for the evolution of bat laryngeal echolocation has been dominant thus far, supported by the fact that fossil bats seem to possess the auditory apparatus necessary for echolocation [12,13,15,16] (figure 1). However, recent studies have shown that the echolocating rhinolophoid and yangochiropteran lineages have distinct anatomical structures and spatio-temporal morphogenetic patterns in their auditory apparatuses, while pteropodid auditory apparatuses are similar to those of non-bat mammals, supporting the ‘dual origin’ scenario [17,18] (figure 1). We discovered distinct modifications in the laryngeal musculoskeletal morphogenesis of rhinolophoids and yangochiropterans, and no laryngeal modification in pteropodids, suggesting that laryngeal echolocation evolved convergently in the two echolocating linages (figure 5). Taken together, our findings provide further support for the ‘dual origin’ hypothesis, in which rhinolophoids and yangochiropterans evolved distinct methods for producing and processing ultrasonic laryngeal echolocation.

Our work established foundation for further studies of bat larynx to reveal not only bat laryngeal echolocation origin but also their unique evolutionary history. In the future, functional morphological analyses should be done to understand the relationship between laryngeal anatomy and acoustic properties of the ultrasonic sounds produced in bats [30]. Moreover, new insights into evolutionary history of the bat laryngeal echolocation should be obtained from observation of innervation pattern of the laryngeal muscles [30,37] and studies of molecular mechanisms that regulate the laryngeal morphogenesis and SFM formation in the echolocating bat lineages.

4. Material and methods

(a) . Collection of adult bat specimens

Adult specimens of nine bat species were used for anatomical observation (electronic supplementary material, table S2). Adults of Rousettus aegyptiacus (n = 3) were provided by Noichi Zoological Park of Kochi Prefecture and Osaka Tennoji Zoo, and collected by the authors at El Beheira, Egypt. Adults of R. leschenaulti (n = 2) were provided by Ueno Zoological Gardens. Adults of Pteropus dasymallus (n = 2) were provided by Okinawa Zoo & Museum. Adults of Rhinopoma cystops (n = 2) were collected by the authors from Bahr Algeer, Giza, Egypt. Adults of Hipposideros turpis (n = 2) and Pipistrellus abramus (n = 2), and frozen adults of Carollia perspicillata (n = 3) were supplied from the National Museum of Nature and Science, Tokyo. Adults of Rhinolophus ferrumequinum (n = 3) and Miniopterus fuliginosus (n = 3) were collected by the authors in Niigata and Wakayama prefectures, Japan, respectively, during May and June 2016–2022 (permits 2-2-1-2-2-3, 3-3-1–3-3-4 and 4-1-1-4-1-3 for Niigata prefecture and 03190002-1-03190002-2, 03190002-1-03190002-2, 02220002-1v02220002-2, 03070003-1-03070003-3 and 02240001-1-02240001-2 for Wakayama prefecture). For adult mole specimens, Mogera imaizumii (n = 3) were collected by the authors in Chiba prefectures, Japan during March and April 2021 (permits 2062-1, 218-1).

(b) . Collection and staging of embryos

To obtain embryonic materials for analyses, wild adult female Rhinolophus ferrumequinum and Miniopterus fuliginosus were captured in Niigata and Wakayama prefectures, Japan, respectively, during May and June 2016–2022 (permits 2-2-1-2-2-3, 3-3-1-3-3-4 and 4-1-1-4-1-3 for Niigata prefecture and 03190002-1-03190002-2, 03190002-1-03190002-2, 02220002-1-02220002-2, 03070003-1-03070003-3 and 02240001-1-02240001-2 for Wakayama prefecture). Wild adult female Rousettus aegyptiacus were captured in El Beheira and Giza, Egypt, during April and May 2017–2020. These animals were sacrificed after anaesthetization by exposure to isoflurane. Each bat's embryos were dissected out from the uterus and washed with phosphate-buffered saline (PBS, pH 7.4). Embryos were fixed with appropriate solutions for subsequent staining methods. After fixation, embryos of R. ferrumequinum, M. fuliginosus, and R. aegyptiacus were staged following [38,39], and [40], respectively. Live animal procedures were approved by the Committee on the Ethics of Animal Experiments of the Faculty of Science, Toho University, for R. ferrumequinum and M. fuliginosus (permits 17-53-301, 18-54-301, 19-55-301, 20-51-449, and 21-51-449), and by the Committee on the Ethics of Animal Experiments of the Zoology Department, Faculty of Science, Fayoum University, for R. aegyptiacus.

(c) . Anatomy of the adult larynx

The larynx was dissected out from the throat of adult bat specimens and the associated tissues were removed for clearer observation of larynx morphology. The larynxes were fixed in 4% paraformaldehyde in PBS and then washed with PBS to remove fixative. The larynxes were observed and imaged with a digital camera (Leica IC90E) mounted on a dissecting microscope (Leica M125).

(d) . Histological analysis for describing laryngeal morphogenesis

Bat embryos at the developmental stage 17 and 20 (n = 2 for the three species respectively; electronic supplementary material, table S2) were fixed with 4% PFA or Serra's solution (100% ethanol : 37% formaldehyde solution : glacial acetic acid, 6 : 3 : 1) for 24 h, then dehydrated using an ethanol series, embedded in paraffin, and sectioned at a thickness of 10 µm. Sections were deparaffinized and stained with hematoxylin, eosin, and alcian blue solutions following standard protocols. Stained sections were photographed with a digital camera (Nikon DS-Fi3) mounted on an optical microscope (Nikon Eclipse Ni). Amira 5.3.0 software (Visage Imaging GmbH, Berlin) was used to reconstruct laryngeal musculoskeletal tissues in three dimensions based on the serial sections.

(e) . Cloning of the genes and RNA probe synthesis

Total RNA was extracted from embryos using NucleoSpin RNA (Macherey-Nagel) and cDNA was synthesized with a PrimeScript TM II 1st strand cDNA Synthesis Kit (TaKaRa). The polymerase chain reaction was performed with PrimeStar GXL and ExTaq DNA polymerases (TaKaRa) to amplify DNA fragments. The primer sequences used for isolation of the fragments are listed in electronic supplementary material, table S1. These fragments were subcloned using the pGEM-T Easy Vector (Promega) and DH5α competent cells (TOYOBO) and then sequenced using Sanger sequencing. Sequence data were registered for BLAST search (https://blast.ncbi.nlm.nih.gov/Blast.cgi) to identify orthologous genes. Phylogenetic trees were constructed with the maximum-likelihood method in PhyML 3.0 under the JTT + G model (www.atgc-montpellier.fr/phyml/). DNA templates with Sp6 and T7 promoters were generated by PCR before synthesis of digoxigenin (DIG)-labelled antisense RNA riboprobes. The templates were then transcribed with an appropriate RNA polymerase. All new DNA sequence information was deposited in the DNA Databank of Japan (DDBJ) database (https://www.ddbj.nig.ac.jp).

(f) . Gene expression analysis

Section in situ hybridization was conducted for St. 20 R. ferrumequinum, M. fuliginosus, and R. aegyptiacus embryos (n = 2; electronic supplementary material, table S2). Embryos were embedded in paraffin and sliced with a microtome (Leica RM2125 RTS) at a thickness of 10 µm. Deparaffinized sections were hybridized with DIG-labelled riboprobes. Anti-digoxigenin-AP antibody, NBT (nitro blue tetrazolium), and BCIP (5-bromo-4-chloro-3-indolylphosphate) (11093274910, 11383213001 and 11383221001, Roche) were used to detect signals [41,42]. Stained materials were imaged with a digital camera (Nikon DS-Fi3) mounted on an optical microscope (Nikon Eclipse Ni). Larynxes of two to three embryos were examined to confirm each gene's expression pattern.

Ethics

This work did not require ethical approval from a human subject or animal welfare committee.

Data accessibility

DNA sequences reported in this paper are available at public databases under NCBI accession nos. PP066094–PP066102.

Supplementary material is available online [43].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors' contributions

K.U.: conceptualization, data curation, formal analysis, funding acquisition, investigation, resources, visualization, writing—original draft, writing—review and editing; T.Y.: formal analysis, investigation; E.R.K.: resources; M.T.: conceptualization, funding acquisition, project administration, resources, supervision, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

K.U. is funded by the Sasakawa Scientific Research Grant (grant no. 2018-5005) and JSPS KAKENHI Grant (grant no. JP20J15442). This study was partly supported by MEXT/JSPS KAKENHI grant number JP21KK0133 to M.T.