The gestural repertoire of Bwindi mountain gorillas (Gorilla beringei beringei): gesture form and frequency of use

Abstract

Over recent decades comprehensive catalogues of vocal, facial, and gestural signals have been established for most great ape species; however, a systematic description of wild gorilla gestural behaviour, particularly of the Eastern gorilla species, remains missing. We address this absence by cataloguing the physical form of gestural units used by 49 habituated wild mountain gorillas (Gorilla beringei beringei) from four social units in the Bwindi Impenetrable National Park, Uganda (n = 157 observation days over 8 months). We obtained a dataset of n = 3220 instances of intentional gesture, coded with a systematic ELAN-based framework (GesturalOrigins). Mountain gorillas employed a repertoire of 63 gesture actions, including potentially species-specific units, across 10 behavioural contexts. A latent class analysis on variants of gesture action expression split units further into 126 finer-grained forms ('morphs'). We observed ~ 6 gestures per hour of observation time and species-level repertoire size was similar to those reported in both Pan species. Our study constitutes the first systematic description of the mountain gorilla gestural repertoire, providing a new understanding of their communication, filling current gaps in great ape gestural phylogeny, and complementing previous studies on their vocal signals. Living in cohesive, small-sized female-male bonded social units, gorillas show striking differences in social organisation as compared to Pan species and provide crucial context for theories on the potential ancestral states of human communicative behaviour.

Keywords: Gesture phylogeny; Gorilla communication; Great ape gesture; Intentionality; Language evolution; Signal.

Fig. 1

Potentially mountain gorilla specific gesture action examples. There are 15 gesture actions in the repertoire that have not been previously described in the great ape gesture literature and have the potential to be mountain gorilla specific. Four are illustrated here: Body cross (a), Lay on (b), Lean in (c), and Jab (d). The pictures capture the point in time where the gesture action is fully in place (MAU end, see Table S10 in ESM)

Fig. 2

Detection asymptote for gesture actions. When plotting the identified repertoire units (here gesture actions) against the number of observations (gesture tokens), visual assessment of the asymptote allows us to estimate how likely it is that we have been able to fully describe all (habitual) gestural behaviour. We included all gesture actions (GAs) with at least 3 observations (n = 63 gesture actions; n = 3203 gesture tokens)

Fig. 3

Token distribution of mountain gorilla gesture morphs in polymorphic gesture actions. The mountain gorilla morph analysis resulted in 126 gesture morphs. Only the morphs of polymorphic gesture actions are represented here (n = 89 morphs, n = 26 gesture actions). Morphs belonging to the same gesture action are grouped next to each other and are indicated with the same colour. The suffix ‘1_3’ indicates the first morph of a gesture action that has three morphs in total, and so forth. For clarity we removed the label of the first morph of Object move (ObjectMove.1_3) and replaced it with +

Fig. 4

Individual repertoire sizes (IRS_1) across sex and maturation classes. Box plot showing individual repertoire sizes (n = 33 data points) across four maturation classes (cat-1: infants; cat-2: juveniles, cat-3: subadults, blackbacks, nulliparous females; cat-4: adults) differentiating males and females within each class. IRS values were calculated by considering every gesture action that a given signaller produced as part of their repertoire (even where n = 1). Note that 6 out of 27 signallers contributed to more than one maturation class in this comparison

Fig. 5

Distribution of Signaller’s context (n = 10) following n = 1462 gestural communications. Definitions of contexts are listed in Table S8. The group ‘Other’ (in grey) comprises all contexts with a percentage contribution < 1%, the group ‘Unknown’ (in grey) comprises cases where we were unable to estimate the signaller’s behavioural context following a gestural interaction