J Comp Physiol A (1998) 182: 737±746
Ó Springer-Verlag 1998
ORIGINAL PAPER
F. Ladich á H. Y. Yan
Correlation between auditory sensitivity and vocalization in anabantoid ®shes
Accepted: 26 November 1997
Abstract Several anabantoid species produce broadband sounds with high-pitched dominant frequencies (0.8±2.5 kHz), which contrast with generally low-frequency hearing abilities in (perciform) ®shes. Utilizing a recently developed auditory brainstem response recording-technique, auditory sensitivities of the gouramis Trichopsis vittata, T. pumila, Colisa lalia, Macropodus opercularis and Trichogaster trichopterus were investigated and compared with the sound characteristics of the respective species. All ®ve species exhibited enhanced sound-detecting abilities and perceived tone bursts up to 5 kHz, which quali®es this group as hearing specialists. All ®shes possessed a high-frequency sensitivity maximum between 800 Hz and 1500 Hz. Lowest hearing thresholds were found in T. trichopterus (76 dB re 1 lPa at 800 Hz). Dominant frequencies of sounds correspond with the best hearing bandwidth in T. vittata (1±2 kHz) and C. lalia (0.8±1 kHz). In the smallest species, T. pumila, dominant frequencies of acoustic signals (1.5± 2.5 kHz) do not match lowest thresholds, which were below 1.5 kHz. However, of all species studied, T. pumila had best hearing sensitivity at frequencies above 2 kHz. The association between high-pitched sounds and hearing may be caused by the suprabranchial airbreathing chamber, which, lying close to the hearing and sonic organs, enhances both sound perception and emission at its resonant frequency. Key words Auditory sensitivity á Sound spectra á ABR á Evolution á Communication
F. Ladich (&) Institute of Zoology, University of Vienna, Althanstraûe 14, A-1090 Vienna, Austria Fax: +43-131 336-778 e-mail:
[email protected] H.Y. Yan University of Kentucky, T.H. Morgan School of Biological Sciences, Lexington, KY 40506-0225, USA
Abbreviations ABR auditory brainstem response á SBC suprabranchial chamber á SPL sound pressure level
Introduction Investigations on sound production and hearing in ®shes have revealed a high diversity of sounds produced as well as of hearing abilities (for reviews see Fine et al. 1977; Myrberg 1981; Fay 1988; Ladich 1997a). These diversities are due to the large variety of sound-generating mechanisms (Tavolga 1971) on the one hand and to accessory hearing structures on the other. Low-frequency sounds (100±300 Hz) are produced by various swimbladder drumming muscles, e.g., in cat®shes (Ladich 1997b), toad®shes (Fine 1978; Brantley and Bass 1994), and cods (Hawkins and Rasmussen 1978) or by pectoral girdle vibrations (Barber and Mowbray 1956; Ladich 1989). Some ®shes also emit a variety of highfrequency pulsed sounds by rubbing bony structures against each other, e.g., pectoral spines against the cleithrum in cat®shes (Schachner and Schaller 1981) or pharyngeal teeth in cichlids (Lanzing 1974). In order to maximize the eectiveness of intraspeci®c communication by sound, natural selection would favor that the main energy content of sounds is generated within the best hearing range of a particular ®sh species. In general all ®sh species can detect the kinetic component of lowfrequency sound (Hawkins 1993; Popper and Fay 1993). On the other hand, enhanced hearing abilities largely depend on coupling between air-®lled chambers and the inner ear, which enables ®sh to perceive the pressure component of underwater sounds in the far-®eld of the sound source (Popper and Fay 1973; Hawkins and Myrberg 1983; Hawkins 1993). Thus, ®shes lacking a gasbladder are restricted to low frequencies and highamplitude sounds, e.g., ¯at®shes (Chapman and Sand 1974) or sculpins (Enger and Anderson 1967), whereas those possessing a connection between the swimbladder and inner ear hear frequencies up to several kilohertz. The best known ``hearing specialists'' are otophysans
738
(cyprinids, cat®shes, characids) in which a chain of bony ossicles transmits volume changes of the swimbladder in the sound ®eld to the inner ear (von Frisch 1936; Hawkins 1981; for a comparison of audiograms see Hawkins and Myrberg 1983; Fay 1988). In general it is assumed that frequencies used in hearing and sound production are matched in ®shes, but investigations focusing on both are rare. A correspondence between hearing thresholds and main frequencies of sounds was found in the midshipman, damsel®sh, and piranha (Cohen and Winn 1967; Myrberg and Spires 1980; Stabentheiner 1988; McKibben and Bass 1996) and a slight mismatch was observed in the toad®sh (Fine 1981). Schellart and Popper (1992) analyzed correlations between best frequencies of hearing and dominant frequencies of sounds and found a correlation coecient of 0.56 in 15 species. Nevertheless, all these studies focus on frequencies below 800 Hz. This correlation has not been tested for ®sh that produce higher-pitched sounds; in particular, no species is known to have best hearing sensitivities above 1 kHz. Anabantoids (labyrinth ®shes) frequently emit stridulation sounds that are variably associated with social behavior. These sounds are probably produced by their pharyngeal teeth (Kratochvil 1985). Additionally, croaking gouramis (genus Trichopsis) have developed a unique pectoral ®n sound-producing mechanism and regularly utilize sounds during agonistic behavior (Marshall 1966; Ladich et al. 1992a,b). The sonic organ consists of enhanced pectoral ®n tendons; these snap over bony elevations of ®n rays, like guitar strings, to produce a pulsed sound during rapid pectoral ®n beating (Kratochvil 1978). Ladich et al. (1992a) demonstrated that the main frequencies of these sounds are concentrated between 1000 Hz and 1300 Hz in T. vittata, and between 1800 Hz and 2600 Hz in the smallest species, T. pumila. Previous ®ndings on hearing abilities in gouramis are contradictory. Schneider (1941) found in behavioral experiments that the paradise ®sh Macropodus opercularis could detect sounds up to 4500 Hz, due to their suprabranchial chamber. Anabantoid ®shes possess an air®lled chamber located dorsal to the gills (suprabranchial chamber or labyrinth), which is utilized for air-breathing. In contrast, audiograms of the blue gourami (Trichogaster trichopterus) and the kissing gourami (Helostoma temmincki) based on saccular microphonics (Saidel and Popper 1987) revealed best auditory sensitivity between 200 Hz and 300 Hz, and showed a steep decrease in sensitivity above 1000 Hz. The aim of the present study is two-fold. The ®rst is to analyze hearing sensitivities in labyrinth ®shes by focusing on vocalizing species in order to determine whether there is a consistency in enhanced hearing abilities within this ®sh group. The second is to investigate whether auditory sensitivity matches the high-pitched sounds produced, especially in the genus Trichopsis. The auditory brainstem response (ABR) recording technique, an electrophysiological, non-invasive, far-®eld recording method recently adapted to
®shes (Kenyon et al., 1998) was utilized to analyze auditory thresholds of croaking, pygmy, blue, and dwarf gouramis as well as paradise ®sh. The audiograms obtained were then correlated and compared to spectra of speci®c sounds produced by tested species.
Materials and methods Gouramis were obtained from local pet suppliers and maintained in ®ltered aquaria at 25 1 °C. They were fed frozen Artemia and commercially prepared ¯ake foods. Eorts were made to provide quiet environments for the animals (e.g., no submerged pumps or airstones). Eleven specimens of T. vittata (0.70±1.56 g) and nine specimens of T. pumila (0.28±0.41 g), Colisa lalia (1.45±2.74 g), M. opercularis (1.24±5.40 g), and Trichogaster trichopterus (3.22± 6.60 g) were used during this study. Only ®sh which were sexually mature or close to sexual maturity were used during this study. Experimental setup The ABR recording protocol followed that of Kenyon et al. (1998). Therefore, only a brief description of the method is given here. Test subjects were secured in a 15-l plastic tub (38 cm ´ 24.5 cm ´ 14.5 cm) ®lled with water and adjusted so that the nape of the head was just above the surface of the water, and a respiration pipette was inserted into the subject's mouth. Respiration was achieved through a simple temperature-controlled (25 1 °C) gravity-feed system. Preliminary trials showed that immobilizing the animals with a curariform agent increased the auditory brainstem response and lowered the auditory threshold values. Therefore, all animals were injected with gallamine triethiodide (Flaxedil) in order to reduce myogenic noise levels. The dosage required was 0.3±0.5 lg g)1 for Trichopsis, 1.0±2.0 lg g)1 for Trichogaster, and 5.0±10 lg g)1 for Colisa and Macropodus. The plastic tub rested on an air table (Kinetic Systems model 1201), and the entire setup was enclosed in a walk-in sound-proof room (2 m ´ 3 m ´ 2 m, Industrial Acoustics Company). Electrodes were pressed ®rmly against the skin, which was covered by a small piece of Kimwipes tissue paper to keep it moist, in order to ensure proper contact during experiments. Loss of contact immediately resulted in an increase of the noise level. The contacting point of both electrodes was positioned about 2 mm above the water surface. The recording electrode was placed on the midline of the skull, over the medulla region. The reference electrode was placed approximately 1 cm in front of the recording electrode. The relative position of the two electrodes could be displaced by a few millimeters with no discernible changes in the response waveforms. Recording electrodes consisted of Te¯on-insulated silver wire (0.25 mm diameter) with ca. 1 mm of exposed tip. Shielded electrode leads were attached to the dierential inputs of an a.c. preampli®er (Grass P-15, 40 dB gain, high-pass @ 30 Hz, low-pass @ 3000 Hz). A hydrophone (Celesco LC-10) placed close to the right side of the otic region was used to monitor stimulus sound pressure. A second Grass P-15 (40 dB gain, high-pass @ 10 Hz, low-pass @ 10 kHz) was used to amplify the hydrophone input. Speakers, suspended in air, were mounted 1 m above the test subject. For frequencies below 3000 Hz, a 30-cm-diameter ``woofer'' (Pioneer, frequency response 19±5 kHz ) was used, while for higher frequencies a 12-cm midrange speaker (Pyle MR 516, frequency response 500±11 kHz) was used. ABR recording apparatus and stimulus presentation Both sound stimuli presentation and ABR waveform recording were accomplished via a Tucker-Davis Technologies (Gainesville, Fla., USA) modular rack-mount system controlled by an opticallylinked 66 MHz 486 PC containing a TDT digital processing board
739 and running TDT ``Bio-Sig'' software. Sound stimuli waveforms were constructed using TDT ``Sig-Gen'' software, and fed through a DA1 digital-analog converter, a PA4 programmable attenuator, and a power ampli®er (QSC Audio Products, Model USA 370) which drove the speaker. The hydrophone preamp output cable was fed to one channel of an AD1 analog-digital converter, while the electrode preamp output was ®rst passed through a PC1 spike conditioner (which provided an additional 60 dB gain and 3000-Hz low-pass ®lter) before reaching the AD1. Both tone bursts and clicks were presented to test subjects. Clicks were 0.1 ms in duration, and presented at a rate of 38.2 s)1 (to prevent phase locking with any 60-Hz noise). The number of cycles in a tone burst was adjusted according to frequency in order to obtain the best combination of stimulus rise time (shorter rise time greater ecacy at generating ABRs) and peak frequency bandwidth (longer duration sharper spectral peak) (Silman and Silverman 1991). One thousand stimuli at each polarity (90° and 270°) were averaged by the Bio-Sig software. ABR traces gained at opposite polarities were averaged together forming a 2000-stimulus trace, in order to eliminate any stimulus artifact. At each tested frequency and sound pressure level (SPL) this was done twice and overlaid to examine if traces were reproducible. SPL was reduced in 5-dB or 3-dB steps until the ABR waveform was no longer apparent. The lowest SPL for which a repeatable ABR trace could be obtained, as determined by overlaying replicate traces, was considered the threshold. This method of visual inspection/correlation is the traditional means of determining threshold in ABR audiometry (Kileny and Shea 1986; Gorga et al. 1988; Hall 1992; Song and Schacht 1996). Once the threshold level was determined, the hydrophone recording was analyzed to determine the root mean square (RMS) SPL, based on the method of Burkhard (1984). Using the capabilities of the Bio-Sig software, cursors were placed 1 cycle apart on either side of the largest (i.e., center) sinusoid of a particular tone burst recording. The software then calculated the RMS of the waveform between the cursors, and calibration factors were applied to determine actual SPL in decibels re 1 lPa. Animals were tested at frequencies of 100, 200, 300, 400, 500, 600, 800, 1000, 1500, 2000, 2500, 3000, 4000, and 5000 Hz. Because we intended to investigate the correlation between high-pitched sounds and audition in anabantoids, only soundpressure measurements were conducted. According to our present knowledge it is very unlikely that ®shes perceive particle motion at frequencies above 500 Hz (Hawkins 1993; Popper and Fay 1993).
for distances of 12 cm. Correlation coecients were calculated for total curves (audiogram versus spectrum) and for the frequency range including the lowest hearing threshold 1 octave. In order to compare the relationship between audiograms and sounds produced, previously published data on sounds of Colisa (Schuster 1986) were added to the graphs. Furthermore, the audiograms of T. trichopterus obtained by microphonics (Saidel and Popper 1987) and ABR measurements (present study) were compared using Student's t-test.
Results ABR waveforms and threshold determinations All gouramis examined showed ABRs to stimulation with click and sinusoid tone bursts from 100 Hz to 5 kHz. A typical suprathreshold ABR consisted of a series of four to nine rapid downward peaks superimposed over a slow negative de¯ection lasting approximately 4±9 ms at low frequencies to around 2 ms in response to clicks and high-frequency tone bursts (Fig. 1). ABR waveforms showed typical characteristics for auditory-evoked potentials at suprathreshold levels
Comparison between audiograms One threshold point was measured at each frequency for each ®sh. Threshold values from all individuals as measured at 14 dierent frequencies were averaged to produce audiograms for each species. Audiograms were compared between species by two-way ANOVA followed by Bonferroni's pairwise multiple comparison procedure. In order to calculate interspeci®c dierences at particular frequencies, one-way ANOVA and Bonferroni's test were used. Signi®cance level was adjusted to the number of frequencies tested. Increase in hearing thresholds in decibels from 1.5 kHz to 3.0 kHz and 2.5 kHz to 5.0 kHz were calculated for T. vittata and T. pumila and compared interspeci®cally. Sound spectra determination and comparison to audiograms Sounds of ten T. vittata (0.82±1.80 g) and ten T. pumila (0.26± 0.45 g) were analyzed. Sound spectra of ®ve to six sounds per individual were averaged by S±Tools, the Integrated Workstation for Acoustics, Speech and Signal Processing developed by the Research Laboratory of Acoustics at the Austrian Academy of Sciences. Relative amplitudes were measured for frequencies used in audiogram determination and means were calculated. Absolute SPL values were determined by comparing relative RMS data measured by S_Tools and absolute SPL values published previously for Trichopsis spp. (Ladich et al. 1992a). SPL values were calculated
Fig. 1 Auditory brainstem response (ABR) waveforms of the dwarf gourami (Colisa lalia) in response to tone bursts of dierent frequencies presented at sound pressure levels 20 dB above hearing threshold (upper ®ve traces). The last trace represents an ABR waveform of a croaking gourami (Trichopsis vittata) in response to a click sound
740 Table 1 Threshold values for investigated anabantoids (dB re 1 lPa). n = 11 for T. vittata and n = 9 for all other species. Freq. frequency (Hz) Freq.
100 200 300 400 500 600 800 1000 1500 2000 2500 3000 4000 5000
T. vittata
T. pumila
C. lalia
M. opercularis
T. trichopterus
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Mean
SD
96.8 97.1 98.4 99.1 100.4 101 95.3 91.5 88.5 95.1 100.6 111.8 122.1 130.3
3.5 3.9 3 5.5 5.3 6.9 6.2 6.4 5.5 4.9 3.5 3.4 3.7 2.9
93.1 93.4 95.8 93.6 99.6 102.0 99.9 95.3 100.2 101.3 103.8 107.7 112.2 121.4
6.5 4.5 5.0 3.5 4.0 7.0 5.3 5.7 4.2 3.0 5.1 4.9 4.2 3.9
93.9 96.3 97.7 95.4 96.0 94.0 93.7 89.9 93.3 95.9 103.4 116.7 127.2 134.9
8.2 4.4 5.1 6.7 7.4 6.5 6.9 7.0 6.7 9.2 8.7 6.6 5.5 4.9
88.9 88.9 92.5 92.9 97.2 99.3 96.7 92.7 96.3 100.8 109.0 119.6 128.3 135.4
3.5 3.8 1.7 4.4 4.8 3.1 6.8 6.2 5.0 5.1 4.1 4.5 4.9 3.5
91.1 90.8 85.2 82.7 80.0 77.0 76.2 77.4 85.1 93.6 102.2 115.0 124.8 132.8
4.1 5.0 4.8 4.4 6.3 4.6 6.3 6.3 4.3 2.4 3.6 7.2 3.5 3.3
in all ®ve species. The ABR traces obtained of acoustic stimuli presented at opposite polarities (90° versus 270°) did not cancel each other out when averaged. In contrast, averaged sound-pressure waveforms presented at two polarities (90° versus 270°) always cancel each other out. The onset latency of the ABR varied with stimulus frequency, ranging from 7.5 ms after stimulus onset at 100 Hz to as little as 0.3 ms with clicks and 5000-Hz tone bursts. The onset latency of the ABR increased with decreasing stimulus amplitude at a rate of approximately 0.05 ms per 5 dB in click sound level. We were never able to record ABR from dead ®sh (n 8). All anabantoids examined were sensitive to high-frequency sounds and can therefore be regarded as nonotophysan hearing specialists (Fig. 2). Sensitivity maxima between about 800 Hz and 1500 Hz were present in all species (Table 1). Two-way ANOVA revealed signi®cant dierences between audiograms (F 63.4, df 4, 13, P < 0.001). The blue gourami T. trichopterus
diered signi®cantly from all other species (Table 2). It had the lowest absolute auditory thresholds between 400 Hz and 1000 Hz of all anabantoids (Fig. 2, Table 3). The lowest part of its audiogram was between 600 Hz and 800 Hz, with mean thresholds of 77 dB and 76.2 dB (re 1 lPa), respectively. The best high-frequency hearing sensitivity (3±5 kHz), on the other hand, was found in the smallest species, the pygmy gourami T. pumila (Fig. 2, Table 3). T. vittata had the most pronounced high-frequency hearing maximum between 1000 Hz and about 1600 Hz. Sounds produced and correlation to auditory sensitivity Croaking sounds uttered by Trichopsis during agonistic interactions are built up of series of double pulses averaging 180 ms in duration in T. vittata and 300 ms in T. pumila (Fig. 3A, B). Pulses were broad band with dominant frequencies concentrated above 1 kHz in both species and with frequency maxima of up to several kilohertz. The audiogram of T. vittata (Fig. 4A) clearly shows a match between sensitivity maximum and the
Table 2 Interspeci®c dierences between audiograms calculated by two-way ANOVA and pairwise multiple comparison procedure (Bonferroni's method). Cl C. lalia, Mo M. opercularis, Tp T. pumila, Tt T. trichopterus, Tv T. vittata
Fig. 2 Audiograms of the ®ve anabantoids obtained by the ABR recording method
Tv ± Tp Tv ± Cl Tv ± Mo Tv ± Tt Tp ± Cl Tp ± Mo Tp ± Tt Cl ± Mo Cl ± Tt Mo ± Tt
Di. of means
t
P
0.62 0.03 0.76 8.15 0.64 1.37 7.54 0.73 8.18 8.91
0.98 0.04 1.20 12.95 0.97 2.08 11.82 1.11 12.39 13.50
NS NS NS
