Biological relevance and methodological implications of unexpected hearing thresholds in a diving bird

Abstract

Many animals alternate between different media, such as air and water, thanks to specific adaptations. Among birds, penguins (Sphenisciformes) have the most extreme morphological, physiological, and behavioural adaptations to their amphibious lifestyle. Their auditory perception of sound, potentially matching different impedances in air and under water, is largely unknown particularly in terms of whether their underwater adaptations may have affected their in-air hearing capacity. In this context, we investigated the hearing ability of four captive Humboldt penguins (Spheniscus humboldti) in air using psychophysical hearing tests. The 50% hit rate was found to be below 76 dB rms re 20 µPa between 0.250 kHz and 10 kHz, with most sensitive hearing at 2 kHz (mean threshold of 15.3 dB rms re 20 µPa). The four penguins showed large inter-individual variation in sensitivity at a given frequency but within a common audiogram shape. Despite the variability, penguins detected 0.250 kHz at comparably low sound levels (mean = 36.8 dB rms re 20 µPa) after a rapid decline of sensitivity at 0.500 kHz (mean = 64.1 dB rms re 20 µPa). This finding was unexpected, and it is therefore difficult to interpret whether it is an artefact of the methods or a biologically relevant finding. An extensive discussion is presented and suggests that this finding may be biologically relevant but would need further investigation to reveal the underlying mechanisms.

Keywords: Behaviour; In-air hearing; Penguins; Psychophysics; Threshold; Training.

Fig. 1

Top: diagram of equipment schematic for the signal production and receiving chains; Bottom: diagram of the behavioural audiogram set up during a hearing test session inside the anechoic box.

Fig. 2

Sound stimulus recording at 2 kHz with background noise (sampling rate 48 kHz, 24 bits signal at -36 dB re 1 V output at the maximum tested level; top in time domain and middle in frequency domain (FFT size 4096). Bottom: Self-noise of Norsonic sound level instrument (sampling rate 48 kHz, 16 bits, FFT size 4096).

Fig. 3

In-air psychoacoustics audiogram for the four Humboldt penguins. The hearing thresholds (individual symbols) were derived from signal detection theory at A) 50% hit rate, B) 50% probability of cumulative distribution function of hit rate. Yellow Dot = Frieda (female), Red Square = Jakob (male), Green Triangle = Gustel (female), Blue Star = Lemmy (male). The same-coloured straight lines connect the data points of the individuals. Not all frequencies were usable for all individuals and therefore missing in the figure. See supplementary figures for d’ audiograms of individuals.

Fig. 4

Audiogram of mean thresholds of all four animals from different cut off limits (mean hr0.5 = mean values at threshold cut level of 50% hit rate, mean overall = mean data from all cut off levels including all animals also if data had outliers, robust values mean overall = mean data of only the most robust data points were included, amount of data points changed for animal and frequency; for specific data used see Table 1).

Fig. 5

Hit rate as a function of sound pressure level (SPL) shown per frequency for one to four animals after cleaning the data. Psychometric functions generated by probit fitting for each tested frequency and individual. See supplementary material Fig. S5 for other frequencies tested in the study.

Fig. 6

Sound pressure level as a function of d-prime plotted with a linear regression for each session type in an example of two different animals at two different frequencies. Each coloured dot represents pooled sessions per sound pressure level with constant stimulus (c) session type and each grey triangle representing pooled sessions per sound pressure level of the methods of constants (sc) session type. Grey horizontal line is at a commonly used d-prime value of 1. Session types were not combined for threshold determination due to different slope steepness. Adjusted R² and p values are seen in the plot for each session type.

Fig. 7

Audiogram of psychophysical hearing thresholds of currently available water birds. (Data acquired from Hill, 2017; Maxwell et al., 2017; Therrien, 2014; Trainer, 1946 (in Hill, 2017)) now also including the Humboldt penguin (mean 50% hit rate ± SD, n = 1 – 4, see Table 1) from this study.