Sound source localization and segregation with internally coupled ears: the treefrog model

Abstract

Acoustic signaling plays key roles in mediating many of the reproductive and social behaviors of anurans (frogs and toads). Moreover, acoustic signaling often occurs at night, in structurally complex habitats, such as densely vegetated ponds, and in dense breeding choruses characterized by high levels of background noise and acoustic clutter. Fundamental to anuran behavior is the ability of the auditory system to determine accurately the location from where sounds originate in space (sound source localization) and to assign specific sounds in the complex acoustic milieu of a chorus to their correct sources (sound source segregation). Here, we review anatomical, biophysical, neurophysiological, and behavioral studies aimed at identifying how the internally coupled ears of frogs contribute to sound source localization and segregation. Our review focuses on treefrogs in the genus Hyla, as they are the most thoroughly studied frogs in terms of sound source localization and segregation. They also represent promising model systems for future work aimed at understanding better how internally coupled ears contribute to sound source localization and segregation. We conclude our review by enumerating directions for future research on these animals that will require the collaborative efforts of biologists, physicists, and roboticists.

Keywords: Auditory grouping; Auditory scene analysis; Auditory stream segregation; Pressure difference receiver; Pressure gradient receiver.

Fig. 1

Anatomy of internally coupled ears in hylid treefrogs. a An adult female of Cope’s gray treefrog, H. chrysoscelis, with a white arrow depicting the tympanum. b Schematic of the middle ear of the Pacific treefrog, Pseudacris (formerly Hyla) regilla, redrawn from Lombard and Straughan (1974). c Magnetic resonance imaging (MRI) scan of a Cope’s gray treefrog made with a 9.4-T magnet with 31-cm bore, redrawn from Bee (2015). Abbreviations: col columella (red), ec extracolumella (yellow), et Eustachian tube, mc mouth cavity, op operculum (green), opm opercularis muscle (brown), ss suprascapula, sc (inner ear) semicircular canal, t tympanum

Fig. 2

Directionality of the tympanum in Hyla versicolor. The plot shows vibration amplitude as a function of source incidence angle in azimuth in 30° steps (relative to the snout at 0°). The center of the plot corresponds to a vibration amplitude of 10 nm; distance between the concentric reference circles is 10 dB. Data are shown for three frequencies: 1080 Hz (blue circles), 1520 Hz (red squares), and 2200 Hz (green triangles). The tympanum’s frequency response is shown at each angle (solid black lines), and the response from 60° is re-plotted as a gray area behind each spectrum. Note that the greatest directionality is generally seen at frequencies intermediate between the two peaks of the bimodal frequency response of the tympanum (e.g., 1520 Hz) and that the two peaks correspond approximately to the lower peak (e.g., 1080 Hz) and upper peak (e.g., 2200 Hz) of conspecific advertisement calls. Redrawn from Jørgensen and Gerhardt (1991)

Fig. 3

Directionality of the tympanum in Hyla chrysoscelis. The plot shows the transfer function of tympanum vibration velocity (color, dB re 1 mm/s/Pa) as a function of direction (x-axis, ipsilateral angles positive, frontal direction is 0) and frequency. a Lungs inflated. b Lungs deflated. c Lungs manually re-inflated. Note the two peaks of tympanum vibration near 1.4 and 2.5 kHz and the pronounced directionality between these two frequencies. Note also how the spectral peak at 1.4 kHz (red arrow) and strong directionality observed in the range of 1.6–1.9 kHz (black arrow) disappear when the lungs are deflated. The data depicted here are from a male frog. Reprinted from Journal of Comparative Physiology A, volume 200, M. S. Caldwell, N. Lee, K. M. Schrode, A. R. Johns, J. Christensen-Dalsgaard, M. A. Bee, “Spatial hearing in Cope’s gray treefrog: II. Frequency-dependent directionality in the amplitude and phase of tympanum vibrations,” pp. 285–304, Copyright (2014), with permission from Springer

Fig. 4

Directionality of auditory nerve responses in Hyla versicolor. The central figure in each panel is a polar plot showing spike rate at 10 dB above threshold as a function of azimuthal sound incidence angle for a a low-frequency fiber with a characteristic frequency of 395 Hz or b a high-frequency fiber with a characteristic frequency of 1705 Hz (Christensen-Dalsgaard, unpublished data). Recordings were made from the auditory nerve on the animal’s right side. In a, circular grid spacing is 30 spikes/s; in b it is 10 spikes/s. Surrounding each polar plot are peristimulus time histograms showing the relative magnitudes of responses as a function of azimuthal sound incidence angle (all on the scale indicated); below each histogram is a depiction of the stimulus waveform

Fig. 5

Simulated EI neuron response to free-field sound. In this model, contralateral is excitatory and ipsilateral is inhibitory when leading by up to 1 ms. The response was constructed by comparing each ipsilateral and contralateral spike in an auditory nerve recording (as in Fig. 9). The peristimulus time histograms and colored curves in the polar plot show the responses of right (red) and left (blue) EI neurons to simultaneous calls from both sides (Note the color coding of the y-axes for the two overlaid histograms; histograms at all angles are on the scale indicated). The black curve shows the actual nerve spike rate data (i.e., before EI processing) at a stimulus level of 70 dB SPL (Christensen-Dalsgaard, unpublished data)

Fig. 6

Measures of behavioral performance in closed-loop phonotaxis tests of source localization in azimuth in green treefrogs (Hyla cinerea). Histogram showing distributions of head orientation angles (α) and jump error angles (γ) when females engaged in head scanning behavior. Insets show how head orientation angle and jump error angle were computed. Data are from Rheinlaender et al. (1979)

Fig. 7

Measures of behavioral performance in open-loop phonotaxis tests of source localization in azimuth in a barking treefrogs (Hyla gratiosa) and b Cope’s gray treefrog (H. chrysoscelis). Shown here are the mean orientation angles of subjects after making a translational or rotational movement relative to the position of a source of advertisement calls at sound incident angles in the frontal hemifield between −45° (left) and +45° (right). Redrawn from data in Klump and Gerhardt (1989) and Caldwell and Bee (2014)

Fig. 8

Acoustic triangulation hypothesis for source distance estimation in frogs. Depicted here is the change in the angles (θ₁ and θ₂) between a female and two calling males as the female moves through the chorus environment. The hypothesis holds that females estimate source distance by attending to the rate at which these angles change as they move (Murphy 2008). Adapted from Murphy (2008), assessment of distance to potential mates by female barking treefrogs (Hyla gratiosa). Journal of Comparative Psychology 122, 264–273, with permission from the American Psychological Association

Fig. 9

Spatial release from masking in Cope’s gray treefrog (Hyla chrysoscelis). Depicted here are masked signal recognition thresholds measured in the presence of “chorus-shaped noise” (i.e., noise having the long-term spectrum of natural breeding choruses) that was either co-located with the signal or separated from it by 90° around a circular test arena. Subjects exhibited spatial release from masking on approximately 70 % of trials (left), whereas differences in threshold were negligible on the remaining 30 % of trials (right). Redrawn from data in Nityananda and Bee (2012)

Fig. 10

Spatial coherence as a cue for sequential auditory grouping in eastern gray treefrogs (Hyla versicolor) and Cope’s gray treefrogs (H. chrysoscelis). a Schematic illustration of how pairs of pulse trains, each with a pulse rate in the range of H. versicolor, were used as stimuli in two-alternative choice tests or single-stimulus, no-choice tests. The two versicolor-like pulse trains (e.g., 20 pulses/s) were temporally interleaved to create a single chrysoscelis-like pulse train (e.g., 40 pulses/s), but presented from spatially separated speakers. b–g Results from phonotaxis tests examining the effect of spatial separation on sequential auditory grouping in H. versicolor (b–d, after Schwartz and Gerhardt 1995) and H. chrysoscelis (e–g, after Bee and Riemersma 2008). b–d Depictions of speaker configurations used in the two-alternative choice tests of Schwartz and Gerhardt (1995) and their results plotted in the form of preference functions for various choice tests. e–g Depictions of speaker configurations used in the single-stimulus tests of Bee and Riemersma (2008) and their results showing response latencies and the proportions of subjects responding as a function of spatial separation. Reprinted from International Journal of Psychophysiology 95(2), M. A. Bee, “Treefrogs as animal models for research on auditory scene analysis and the cocktail party problem,” pp. 216–237, Copyright (2015), with permission from Elsevier