J Comp Physiol A (2014) 200:311–316 DOI 10.1007/s00359-014-0885-3

Original Paper

Auditory sexual difference in the large odorous graminea

Wei-Rong Liu · Jun-Xian Shen · Yu-Jiao Zhang · Zhi-Min Xu · Zhi Qi · Mao-Qiang Xue

Received: 8 August 2013 / Revised: 15 January 2014 / Accepted: 22 January 2014 / Published online: 9 February 2014 © Springer-Verlag Berlin Heidelberg 2014

Abstract Acoustic communication is an important Abbreviations behavior in frog courtship. Male and female of most AENFP Auditory evoked near-field potential species, except the concave-eared torrent frog Odorrana BEF Best excitatory frequency tormota, have largely similar audiograms. The large odorous CF Characteristic frequency frogs (Odorrana graminea) are sympatric with O. tormota, RMS root mean square but have no ear canals. The difference in hearing between SPl Sound pressure level two sexes of the frog is unknown. We recorded auditory TS Torus semicircularis evoked near-field potentials and single-unit responses from the auditory midbrain (the torus semicircularis) to deter- mine auditory frequency sensitivity and threshold. The Introduction results show that males have the upper frequency limit at 24 kHz and females have the upper limit at 16 kHz. The Acoustic communication is often associated with the more sensitive frequency range is 3–15 kHz for males and behaviors of frogs, including territorial behavior, mate 1–8 kHz for females. Males have the minimum threshold at finding, courtship and aggression (Zelick et al. 1999). Dur- 11 kHz (58 dB SPL), higher about 5 dB than that at 3 kHz ing the breeding season, an adult male usually produces for females. The best excitatory frequencies of single units advertisement calls to attract receptive female to the calling are mostly between 3 and 5 kHz in females and at 7–8 kHz male, followed by courtship, leading to amplexus. in males. The underlying mechanism of auditory sexual However, a sex difference in the peripheral auditory sen- differences is discussed. sitivity was observed in the American bullfrog (Rana cates- beiana) (Hetherington 1994; Mason et al. 2003) and in the Keywords Auditory evoked near-field potential · tree frog (Eleutherodactylus coqui) (Narins and Capranica Auditory threshold · Sex difference · Frogs · Torus 1976). A recent study through behavioral and physiologi- semicircularis cal experiments and laser Doppler vibrometer measure- ments demonstrated that the concave-eared torrent frogs (Odorrana tormota) have significant sexual differences in auditory frequency sensitivity (Shen et al. 2011a). Males of O. tormota have the ultrasonic communication capacity, W.-R. Liu · Y.-J. Zhang · Z. Qi · M.-Q. Xue but females exhibit no ultrasonic sensitivity. It appears that Department of Basic Medical Sciences, Medical College ultrasonic components in male’s calls might be not essen- of Xiamen University, Xiamen 361102, China e-mail: [email protected] tial to attract female approaching during courtship. The large odorous frog (Odorrana graminea, previously J.-X. Shen (*) · Z.-M. Xu O. livida) is an arboreal and nocturnal species sympatric State Key Laboratory of Brain and Cognitive Science, Institute with O. tormota living in the Huangshan Hot Spring area, of Biophysics, Chinese Academy of Sciences, Beijing 100101, China China. Males of O. graminea produce diverse types of e-mail: [email protected] calls, most of them containing ultrasonic harmonics (Shen

1 3 312 J Comp Physiol A (2014) 200:311–316 et al. 2011b). A preliminary study indicated that males of reading RMS). Each stimulus at different frequencies O. graminea have the ability to hear ultrasound (Feng et al. was modulated relative to 90 dB SPL using a Program- 2006). It is unclear whether males communicate with ultra- mable Attenuator (PA5, TDT) and the BrainWare software sound and females detect ultrasound as well. (distributed by TDT). The stimulus presentation and data To explore the auditory sexual difference, we recorded acquisition were done in an automated mode with the auditory evoked near-field potentials (AENFP) and single- BrainWare. unit responses from the torus semicircularis (TS) of mid- brain in O. graminea. The findings indicate that males have Neurophysiology ultrasonic communication capacity and females are insensi- tive to ultrasound. Glass microelectrodes filled with 3 M sodium acetate (impedances 1–10 MΩ) were used to record AENFPs or single-unit spikes induced by tone bursts (frequency range Materials and methods 1–30 kHz) from the frog’s TS. An indifferent electrode was placed at the nearby muscles. The first site for microelec- trode was located at the dorsal surface of the torus away from the midline of the brain about 1.0 mm, then orthogo- Thirty-four male and eight female frogs of O. graminea nally inserted into the surface of the TS by a remote-con- (Boulenger) were collected in the Huangshan Hot Spring trolled Pulse Motor Micro-Drive Micromanipulator (SM- area, China in May and June of 2009 and 2011, and carried 21, Narishige, Tokyo, Japan) with an accuracy of 1 μm. into the Institute of Biophysics in Beijing for electrophysi- Neurons were identified along the penetration track by their ological study. The methods were similar to those described responses to sound in various sound pressure levels for all in the previous papers (Feng et al. 2006; Shen et al. 2011a) frequencies. A subsequent penetration site for electrode with slight modifications. Briefly, frogs were lightly anes- was placed caudo-rostrally or medial-laterally about 50 μm thetized by immersed in a 0.2 % solution of tricaine meth- away from the preceding site. A lower band-pass filter anesulfonate (MS222) for 2–5 min. After anesthesia, the (20–200 Hz) was used for AENFP recordings and single- body was wrapped in cotton gauze and the head was unit responses were band-pass filtered between 300 Hz and unwrapped to make the sound transmission uninterrupted. 3 kHz. Neural signals were recorded by RA4PA Preamp To immobilize the frog, the jaw and hind limbs of the frog and RA16 Medusa Base (TDT), monitored visually and were pinned on a platform made from paraffin. The skin on extracted using BrainWare and stored on a hard drive. the dorsal surface of the head was incised, and a small hole AENFP measured from each recording site was averaged was made in the skull above the TS. During the record- over 20 trials at corresponding sound pressure levels across ing session, the frog was placed on a vibration-free plat- frequencies. The amplitude and latency of all AENFPs form inside a sound-proof and anechoic room with periodic from one recording site were read out from digital infor- addition of 0.1 % MS222 to keep the frog in light anesthe- mation of their waveforms by the BrainWare, forming a sia. The temperature was maintained at 19–21 °C by an air dataset, in which the maximum peak-to-peak amplitude of conditioner. AENFPs was as 100 %, amplitudes of other AENFPs were normalized. The threshold for AENFP at a frequency is the Sound signals minimum sound pressure level of a stimulus as the point at which the amplitude of AENFP is equal to or greater than Sound signals were generated by an RP2.1 Enhanced 120 % of the amplitudes of other waves. The threshold for Real-time Processor [Tucker Davis Technologies (TDT) single unit is the minimum sound pressure level of a stimu- System 3, USA] and amplified via a power amplifier (GF- lus as the point at which spike activity of certain latency is 10, China), and broadcasted from the loudspeaker (fre- elicited at least three times per 10 stimuli. A total of 74 sets quency range 1–30 kHz; FE87E, Fostex, Japan) placed of AENFPs were measured from 8 females and 114 sets at a distance of 50 cm from the frog’s eardrum on the of AENFPs measured from 34 males, and the single-unit side contralateral to the recording site in the TS. Acous- spike activities were measured from 3 females (47 units) tic stimuli consisted of tone bursts (50-ms duration, 5-ms and 15 males (81 units). rise and fall times, presented at a rate of 1/s). The sound pressure levels of the stimulation system were measured Data analysis using a condenser microphone (Brüel and Kjaer 4135, Denmark) placed at the position corresponding to the The results were presented as mean SE. The Origin 7.0 ± frog’ eardrum and a precision measuring amplifier (Brüel (Originlab, Northampton, MA, USA) and one-way ANOVA and Kjaer 2610, Denmark) in dB SPL (re. 20 μPa) (fast were used for statistical analyses and plotting graphs.

1 3 J Comp Physiol A (2014) 200:311–316 313

Fig. 1 Auditory evoked near- field potentials measured from the torus semicircularis of the midbrain in O. graminea. The AENFP waveforms were recorded and averaged from a female (a) and a male (c) in response to 20 bursts presented at frequency range of 1–24 kHz at 90 dB SPL and rate of 1/s. Lines with different colors represent different tone fre- quency shown in the right color palette. b and d Peak-to-peak amplitude (black) and response latency (blue) of the AENFPs are depicted as a function of tone frequency, respectively, corresponding to a and c

Results value of the peak ranged from 1 to 7 kHz for females, while from 2 to 9 kHz (or even 13 kHz) for males. Statistic analy- Auditory evoked near‑field potentials from TS sis indicates significant difference in the AENFP amplitude versus frequency relationship between the two sexes at all The AENFPs recorded from the TS of a female and a tested frequencies [one-way ANOVA with Bonferroni cor- male of O. graminea are shown in Fig. 1a, c, respectively. rection, 18.1 < F < 88.6, P < 0.0001; sample number (Nf) In females, AENFPs were always recorded from the TS for each frequency ranged from 4 to 72 for 8 females; Nm in response to tone bursts (frequency ranged from 1 to from 15 to 99 for 34 males], except at 9 kHz (ANOVA, 15 kHz, sound pressure level at 90 dB SPL), and seldom F 1.73, P 0.19). The line chart is characterized 1,170 = = evoked by the stimuli at 16 kHz and 102 dB SPL, indi- by both the kurtosis (ku) and skewness (sk) of the two data cating that the upper frequency limit was around 16 kHz sets. The measurements reveal that the relative amplitude for females (Fig. 1a). In males, AENFPs were always curve in females with high kurtosis (ku 1.25) tends to f = recorded under stimulation at a frequency range between 1 have a peak at 3 kHz and the distribution is unimodal. In and 22 kHz at 90 dB SPL (Fig. 1c) and sometimes evoked contrast, the relative amplitude curve in males with a low with stimuli at 24 kHz and 96 dB SPL, indicating that kurtosis (ku 0.98) tends to have a flat top at 4–5 kHz. m = males have frequency sensitivity up to 24 kHz. Peak-to- However, the two relative amplitude curves have similar peak AENFP amplitude and response latency as a func- skewness (sk 0.16 for females; sk 0.22 for males), f = m = tion of stimulus frequency were shown in Fig. 1b, d for which means the similar distributions of two groups of the female and the male, respectively, with the maximum measures. AENFP amplitude at 3 kHz for the female and 5 kHz for Figure 2b illustrates the average latency versus fre- the male. quency curves, which show sexual difference in two sexes To explore sexual differences in frequency sensitivity, of the frog (ANOVA, 1.86 < F < 61.6, P < 0.001), except peak-to-peak amplitudes of AENFPs were normalized to at 8 kHz (ANOVA, F 0.29, P 0.48). The shortest 1,95 = = the maximum amplitude in each AENFP dataset recorded latency is similar of around 10.3 ms in average for two from one recording site across all frequencies in order to sexes of the frog (at 5 and 7 kHz for females; at 11 kHz for construct relative peak-to-peak amplitude of AENFP ver- males). However, the latency curve in females have higher sus frequency curves (Fig. 2a). The maximum AENFP kurtosis (ku 2.41) than that in males (ku 0.97), f = m = response was measured at 3 kHz for females. In contrast, a which means that females have a frequency range in the peak was at 4–5 kHz for males. The frequency width at half latency curve narrower than that males do. The skewness

1 3 314 J Comp Physiol A (2014) 200:311–316

Fig. 3 Audiograms of O. graminea. Averaged threshold versus fre- quency relationship of the AENFPs in females (45 sets, n 8; open circles) and males (114 sets, n 34; solid circles). Dashed= and dot- = ted lines indicate the auditory sensitive frequency bandwidth (W10, 10 dB above threshold at CF) for females and males, respectively. Error bars are 1 SE of the mean ±

Shen et al. 2011b; the dominant frequency ranged from 1.2 to 14.7 kHz). Males have the minimum threshold at 11 kHz (58.0 0.5 dB SPL; 114 sets of AENFPs) with W ± 10 ranged from 3 to 15 kHz. The upper frequency limit was at 16 kHz for females and 24 kHz for males. Statistic analy- sis indicates significant differences in the AENFP thresh- olds at all frequencies between the two sexes (ANOVA,

7.48 < F < 651.9, P < 0.0001; Nf ranged from 5 to 45 for 8 females; Nm from 14 to 112 for 34 males), except at 8 kHz (ANOVA, F 3.14, P 0.08) and 9 kHz Fig. 2 The averaged relative peak-to-peak amplitude and latency of 1,154 = = (F 0.56, P 0.45). The measurements reveal that the AENFPs versus frequency relationships. Sound pressure levels of 1,154 = = stimuli were 90 dB SPL for all frequencies. a Peak-to-peak ampli- threshold curve in females with high kurtosis (ku 1.22) f = tudes of the AENFPs are normalized to the maximum in each sets and tends to have a trough between 3 and 7 kHz. In contrast, then averaged for each frequency (74 sets of AENFPs, n 8 females, the threshold curve in males with a little low kurtosis open circles; 101 sets, n 15 males, solid circles). Dashed= and dot- = (ku 0.90) tends to have a wider trough ranged from 3 to ted lines indicate the frequency widths at the half maximum of the m = AENFPs for females and males, respectively. b Averaged latency 13 kHz. The skewness of the threshold curve is also differ- versus frequency relationships of AENFPs for females (71 sets) and ent (skf 0.69 for females; skm 1.42 for males), which males (70 sets). Error bars are 1 SE of the mean = = ± means dissimilar distributions of two datasets. of the latency curve is similar (sk 1.54 for females, f = sk 1.34 for males). Auditory responses of single TS units m = A total of 47 single TS units from 3 females and 81 units Audiograms from 15 males were extracellularly recorded to deter- mine the best excitatory frequency (BEF) of each unit. In Figure 3 illustrates the AENFP threshold versus frequency females, about 40.4 % (19/47) units had the BEF at 3 kHz, relationships for female and male O. graminea. The char- 21.3 % (10/47) at 4 kHz, 27.7 % (13/47) at 5 kHz, 10.6 % acteristic frequency (CF) was around 3 kHz for females (5/47) at 7 kHz and no unit found above 7 kHz. In con- (threshold, 52.8 0.8 dB SPL; 45 sets of AENFPs), and trast, in males about 12.3 % (10/81) units had the BEF at ± the frequency-sensitive bandwidth (W10, frequency range 3 kHz, 16 % (13/81) at 4 kHz, 27.2 % (22/81) at 7 kHz, at 10 dB above the minimum threshold at CF) ranged 30.9 % (25/81) at 8 kHz, and 13.6 % (11/81) at 11 kHz. from 1 to 8 kHz, corresponding to the fundamental fre- The average threshold versus frequency curves of single quency (F0) of male vocalization (about 1.5–8.5 kHz, TS units are shown in Fig. 4 for females and males. There

1 3 J Comp Physiol A (2014) 200:311–316 315

cavitympanum (Arch et al. 2008 and 2009), respectively. They possess deeply sunken tympana and ear canals as a resonator (Feng et al. 2006), which have been suggested to be critical for ultrasonic hearing. However, the males of O. graminea have thicker tympanic membranes [at the rim, 27 μm, but 3–4 μm in O. tormota (Feng et al. 2006)] and no ear canals, which may be the reason why the upper fre- quency limit of hearing in males of O. graminea is consid- erably lower than that in males of O. tormota.

Auditory sexual difference and its underlying mechanism

A sex difference in the peripheral auditory sensitivity was Fig. 4 Auditory threshold versus frequency relationship of single TS observed in the American bullfrog (R. catesbeiana) (Heth- units in O. graminea. The auditory thresholds of single TS units are erington 1994; Mason et al. 2003) and the tree frog (E. averaged from 47 single TS units (3 females; open circles) and 81 coqui) (Narins and Capranica 1976). Recent behavioral and units (15 males; solid circles), respectively. Error bars are 1 SE of the mean ± physiological study combined with laser Doppler vibrom- eter measurements demonstrated that the Chinese concave- eared frog O. tormota has remarkable sexual differences in was a significant sexual difference across BEFs and sex in hearing (Shen et al. 2011a).

O. graminea (ANOVA, 32.4 < F < 130.4, P < 0.0001; Nf To avoid sampling bias, a total of 3–8 electrode pen- ranged from 10 to 47 for 3 females; Nm from 7 to 81 for 15 etration sites per frog were mainly located at the center males), except at 7 kHz (ANOVA, F 0.19, P 0.66). of the dorsal surface of the TS with an inter-site distance 1,124 = = of about 50 μm by a micromanipulator. AENFPs or sin- gle units were successively identified along a penetration Discussion track in depth about 80 μm per step. AENFP and single- unit responses were largely recorded from the principal Different tactics of three sympatric species to cope nucleus of TS (cf. Arch et al. 2011, Fig. 2). The results with noise demonstrated evident differences in hearing between the two sexes of the frog. Females have an upper frequency Recent studies demonstrate that frogs emit diverse broad- limit of hearing at 16 kHz with no ultrasonic sensitivity, band signals as an evolutionary adaption in noisy environ- whereas males can detect ultrasound up to 24 kHz. Males ments (Feng et al. 2006; Shen et al. 2011a, b). O. schmack- and females have similar minimum latency of 10.3 ms at eri, O. graminea and O. tormota are sympatric species the sensitive frequency ranges (Fig. 2b), although only living in the vicinity of a fast-flowing mountain stream. higher frequencies above 8 kHz show longer latencies in Males of O. schmackeri and O. graminea call loudly from females than males. As to audiograms (Fig. 3), males have rocks in the stream at night. O. schmackeri inhabits in an a wider frequency-sensitive range with an upward shift of open and moist environment close to the stream, while about 5 kHz relative to that in females. Males have higher O. graminea and O. tormota move generally around the minimum threshold of about 8 dB than females, whereas brushwood (Fei et al. 2009). Males of O. schmackeri have females have higher thresholds at higher frequencies above a sensitive frequency range between 2 and 4 kHz (Yu et al. 9 kHz than males. These differences may primarily result 2006), just above the main energy distribution of noise in from sexual dimorphism of the auditory organs (especially the same area (Narins et al. 2004). Males of O. tormota tympanic membranes) and their physical property in the have evolved very thin eardrums to help them hear ultra- frog (Feng et al. 2006; Shen et al. 2011a). For example, sound up to 34 kHz (Feng et al. 2006; Shen et al. 2011a), larger and thicker tympanic membranes in females of O. and males of O. graminea have an upper frequency limit of tormota demonstrate lower vibration velocity amplitudes 22 kHz (Feng et al. 2006). Thus, O. graminea may be an in response to sound at higher frequencies (above 10 kHz) in-between species in the upper frequency limit of hearing. (Shen et al. 2011a, Fig. 4; Gridi-Papp et al. 2008, Fig. 2d) and thus longer latencies than those in males. However, Ultrasound sensitivity and ear canals in frogs the middle ear morphology and biomechanics are not well known for O. graminea. Male frogs are able to detect frequencies up to 34 kHz Although at most eight locations of electrode penetra- in O. tormota (Feng et al. 2006) and 38 kHz in Huia tions per frog were used, there was no obvious difference

1 3 316 J Comp Physiol A (2014) 200:311–316 found among adjacent penetration sites in relative ampli- Capranica RR, Moffat AJM (1983) Neuroethological correlates of tudes, frequency-sensitive range and latency of AENFPs. sound communication in anurans. In: Ewert JP, Capranica RR, Ingle DJ (eds) Advances in vertebrate neuroethology. Plenum, However, the BEF of single TS units identified within the New York, pp 701–730 same penetration track showed a dorsoventrally decreasing Fei L, Hu SQ, Ye CY, Huang YZ et al (2009) Fauna Sinica. Amphibia tendency, which needs to be further explored. vol. 3, Anura Ranidae. Science Press, Beijing, pp 1219–1224 When making comparisons between near-field poten- Feng AS, Narins PM, Xu CH, Lin WY, Yu ZL, Qiu Q, Xu ZM, Shen JX (2006) Ultrasonic communication in frogs. Nature tial and single-unit threshold data, the minimum thresholds 440:333–336 determined for near-field potentials were generally lower Gridi-Papp M, Feng AS, Shen JX, Yu ZL, Rosowski JJ, Narins PM about 5–8 dB than those for single TS units (Figs. 3, 4). (2008) Active control of ultrasonic hearing in frogs. Proc Natl The disparity results from different threshold definition Acad Sci USA 105:11013–11018 Hetherington TE (1994) Sexual differences in the tympanic frequency as above-mentioned in “Methods”. Near-field potentials responses of the American bullfrog (Rana catesbeiana). J Acoust reflect the activity summation of multiple neurons at a Soc Am 96:1186–1188 recording site in response to sound. In fact, when the sound Mason MJ, Lin CC, Narins PM (2003) Sex differences in the mid- pressure level of a stimulus at a given frequency is equal dle ear of the bullfrog (Rana catesbeiana). Brain Behav Evol 61:91–101 to the threshold for AENFP, it could elicit spike activity of McClelland BE, Wilczynski W, Rand AS (1997) Sexual dimorphism certain latency only one or two times per 10 stimuli from and species differences in the neurophysiology and morphology single unit at the same recording site. In contrast, AENFP of the acoustic communication system of two neotropical hylids. can be evoked of considerable amplitude when the sound J Comp Physiol A 180:451–462 Narins PM, Capranica RR (1976) Sexual differences in the audi- pressure level of a stimulus is equal to the threshold for sin- tory system of the tree frog Eleutherodactylus coqui. Science gle unit at the same recording site. 192:378–380 Narins PM, Feng AS, Lin WY, Schnitzler HU, Denzinger A, Suthers Acknowledgments This work was supported by the grants from RA, Xu CH (2004) Old World frog and bird vocalizations contain the National Natural Science Foundation of China (No. 30730029 to prominent ultrasonic harmonics. J Acoust Soc Am 115:910–913 J.X.S.; Nos. 31070741 and 31270891 to Z.Q.). All experiments were Shen JX, Xu ZM, Yu ZL, Wang S, Zheng DZ, Fan SC (2011a) Ultra- conducted following the Animal Care and Use Guidelines approved sonic frogs show extraordinary sex differences in auditory fre- by the Institute of Biophysics, Chinese Academy of Sciences. quency sensitivity. Nat Commun 2:342 Shen JX, Xu ZM, Feng AS, Narins PM (2011b) Large odorous frogs (Odorrana graminea) produce ultrasonic calls. J Comp Physiol A 197:1027–1030 References Yu ZL, Qiu Q, Xu ZM, Shen JX (2006) Auditory response character- istics of the piebald odorous frog and their implications. J Comp Arch VS, Grafe TU, Narins PM (2008) Ultrasonic signalling by a Physiol A 192:801–806 Bornean frog. Biol Lett 4:19–22 Zelick R, Mann DA, Popper AN (1999) Acoustic communication in Arch VS, Grafe TU, Gridi-Papp M, Narins PM (2009) Pure ultrasonic fishes and frogs. In: Fay RR, Popper AN (eds) Comparative hear- communication in an endemic Bornean frog. PLoS One 4:e5413 ing: fish and . Springer, New York, pp 363–411 Arch VS, Burmeister SS, Feng AS, Shen J-X, Narins PM (2011) Ultrasound-evoked immediate early gene expression in the brain- stem of the Chinese torrent frog, Odorrana tormota. J Comp Physiol A 197:667–675

1 3