Vocal tract allometry in a mammalian vocal learner

Abstract

Acoustic allometry occurs when features of animal vocalisations can be predicted from body size measurements. Despite this being considered the norm, allometry sometimes breaks, resulting in species sounding smaller or larger than expected for their size. A recent hypothesis suggests that allometry-breaking mammals cluster into two groups: those with anatomical adaptations to their vocal tracts and those capable of learning new sounds (vocal learners). Here, we tested which mechanism is used to escape from acoustic allometry by probing vocal tract allometry in a proven mammalian vocal learner, the harbour seal (Phoca vitulina). We tested whether vocal tract structures and body size scale allometrically in 68 young individuals. We found that both body length and body mass accurately predict vocal tract length and one tracheal dimension. Independently, body length predicts vocal fold length while body mass predicts a second tracheal dimension. All vocal tract measures are larger in weaners than in pups and some structures are sexually dimorphic within age classes. We conclude that harbour seals do comply with anatomical allometric constraints. However, allometry between body size and vocal fold length seems to emerge after puppyhood, suggesting that ontogeny may modulate the anatomy-learning distinction previously hypothesised as clear cut. We suggest that seals, and perhaps other species producing signals that deviate from those expected from their vocal tract dimensions, may break allometry without morphological adaptations. In seals, and potentially other vocal learning mammals, advanced neural control over vocal organs may be the main mechanism for breaking acoustic allometry.

Keywords: Acoustic allometry; Harbour seal; Larynx; Pinniped; Trachea; Vocal anatomy; Vocal tract.

Fig. 1.

Phylogenetic generalised least squares regressions between frequency parameters and body mass across 164 mammalian species. Left, frequency range; right, maximum frequency. All variables are log-transformed and the figure is adapted from Ravignani and Garcia (2022). The dotted lines represent a threshold at 2.5 standard deviations from the main regression lines used to define outliers. Non-outlier species (which show acoustic allometry between frequency parameters and body mass) are represented by small circles, and outlier species (which escape acoustic allometry) are represented by large circles. The two red data points, representing harbour seals, are both outliers.

Fig. 2.

Lack of acoustic allometry relationships in harbour seals. (A) Correlations between median fundamental frequency (f₀) for each noise condition (silence, low and high) and body mass. The respective correlation coefficients (τ) and associated P-values for each correlation are reported above the regression line. At first sight, the characteristic inverse relationship between f₀ and body size may seem present, but there is some overlap in the range of f₀ values (whiskers on the right side of the plot) produced by individuals of differing body size between noise conditions. Non-significant P-values suggest that, at least in this sample, there is a lack of acoustic allometry. In addition, allometry may break if calls are produced in different noise conditions. In other words, do the environmental conditions in which vocalisations are produced strongly affect the f₀ values, as much as or even more than body mass? (B–D) Density distributions produced by computing 10,000 different combinations of randomly selected median f₀ values (1 of the 3 median frequency values per seal) to assess whether allometric relationships hold across noise conditions. The coloured vertical lines in these plots represent the respective median values for each of the noise conditions. The median value of the distribution is represented by black circle on the density curve. B shows the density distribution of the Kendall rank correlation coefficients. The median value lies around −0.18, pointing to a weak negative correlation. C shows the density distribution of the correlation P-values associated with the correlations from B. The median P-value is 0.38 which means that in most of the simulated cases we would not reject the null hypothesis (i.e. the correlation is not significantly different from 0). In fact, in only 2.2% of cases (217 out of 10,000) is the correlation significant; this is indicated by the red vertical line. In other words, in 10,000 simulated samples of 8 seals, we generally found no acoustic allometry. D shows the density distribution of the simulated linear regression coefficients (β), where the median value is −10.8 Hz. Given a 5.1 kg difference in body mass between the smallest and the largest seal, we would expect, on average, a frequency shift of 55.08 Hz. For every individual, we calculated the difference of the median f₀ values between the silent and high noise condition; the median range across all individuals was 73.6 Hz. This suggests that the differences caused by individual variability in f₀ in response to noise conditions are larger than the f₀ differences expected from body mass differences alone. Seals of differing body sizes (e.g. 7 versus 12 kg) could thus potentially produce the same f₀ value. This would mean that, in harbour seal pups, vocal plasticity can outweigh and mask acoustic allometric relationships.

Fig. 3.

Illustration of the source–filter theory of sound production using the vocal anatomy of the harbour seal.

Fig. 4.

Vocal anatomy of the harbour seal. (A) The main anatomical structures composing the vocal tract. (B) The measurements shown on a digital rendering. (C) The measurements shown on a picture of a hemi-larynx from a harbour seal pup. In C, the black square outlined on the piece of white paper serves as a reference and is exactly 1 cm². The vocal tract measurements taken include (1) vocal tract length (VTL), (2) vocal fold length (VFL), (3) vocal fold thickness (VFT), (4) subglottic-tracheal dorsoventral distance 1 (STDV1) and (5) subglottic-tracheal dorsoventral distance 2 (STDV2).

Fig. 5.

Boxplots illustrating the significant differences in anatomical measurements between pups and weaners. (A) VTL, (B) VFL, (C) STDV1 and (D) STDV2. Boxplots show median, upper and lower quartiles and 1.5× the interquartile range. The level of significance is denoted by asterisks (***P=0.001).

Fig. 6.

Boxplots illustrating the significant differences in anatomical measurements according to sex. (A) VTL in pups. (B) STDV1 in weaners. The level of significance is denoted by asterisks (*P=0.05 and **P=0.01).

Fig. 7.

Predicted effects of the body length and sex interaction for VTL (top) and the body mass and age interaction for VFL (bottom). The shading around each line of best fit indicates the 95% confidence interval.