Sociality predicts orangutan vocal phenotype

Abstract

In humans, individuals' social setting determines which and how language is acquired. Social seclusion experiments show that sociality also guides vocal development in songbirds and marmoset monkeys, but absence of similar great ape data has been interpreted as support to saltational notions for language origin, even if such laboratorial protocols are unethical with great apes. Here we characterize the repertoire entropy of orangutan individuals and show that in the wild, different degrees of sociality across populations are associated with different 'vocal personalities' in the form of distinct regimes of alarm call variants. In high-density populations, individuals are vocally more original and acoustically unpredictable but new call variants are short lived, whereas individuals in low-density populations are more conformative and acoustically consistent but also exhibit more complex call repertoires. Findings provide non-invasive evidence that sociality predicts vocal phenotype in a wild great ape. They prove false hypotheses that discredit great apes as having hardwired vocal development programmes and non-plastic vocal behaviour. Social settings mould vocal output in hominids besides humans.

Fig. 1. Spectrographic representation of orangutan kiss-squeak alarm calls and analytical flow chart.

Spectrographic representation of six orangutan kiss-squeaks, where darker colours denote louder sound frequencies. Dashed lines indicate the manual selection from which kiss-squeak maximum frequency (mxf) and duration (dur) were extracted, and how the two acoustic parameters were them processed to calculcate their corresponding entropy parameters per individual per context, where E is emergence, S is self-organization and C is complexity. P and p are probabilities, K is a constraint that constrains E, S and C, H is normalized entropy and y represents a call variant. (Methods and Supplementary Data 5).

Fig. 2. Effect of orangutan density on repertoire entropy of alarm calls.

Frequency-based measures are shown in magenta and time-based measures in cyan. Shaded areas represent the 95% confidence interval around the mean, and small diamonds represent data points. Populations by order of increasing density (in number of individuals per km²): Sampan Getek, Sikundur, Sebangau, Tuanan, Gunung Palung and Suaq. Emergence and self-organization are inversely proportional and depicted together for ease of interpretation only. *** denotes significant effects as calculated by linear mixed model analysis after P adjustment for false discovery rate (Table 1 and Supplementary Data 4). Graphic representations are based on raw data; differences between density levels are based on model estimates.