Ariana Strandburg-Peshkin

Deep learning has been widely adopted in the field of bioacoustics, enabling behavioral and ecological research into many sound producing animals through the use of specially designed and trained models. So-called foundational models can generalize to novel species or tasks, but require larger and more diverse training datasets. But the broad range of frequencies that animals produce poses a problem: bioacoustic datasets are recorded at various sample rates to capture different frequency ranges. Combining these data to train generalist models becomes exceedingly costly the wider the frequency range is expanded with high resolution data, becoming untenable for bioacousticians. To approach this problem, we performed experiments using a state-of-the-art transformer model, animal2vec, and a custom dataset consisting of recordings with various sample rates and species whose vocalizations cover a broad frequency range. We designed a positional encoding that provides inductive information about the sample rate, but found that the model is natively able to compensate for the perceived time stretch in the receptive field induced by varying sample rates, with no effect on performance compared to the standard approach of upsampling. This opens up new possibilities of training powerful models on broad bioacoustic datasets and the creation of more generalizable tools for bioacousticians.

Communication is fundamental to sociality and can shape interactions that are critical for individual fitness. Vocal communication is particularly important for many group-living animals, as it all...

When individual preferences for collective outcomes diverge, cohesive animal groups often coalesce on the majority opinion. However, majority-based decision rules may be counterbalanced by other factors, particularly in heterogeneous groups with differentiated social relationships, and these factors could produce inequality in social influence. We used multi-sensor tracking collars to collect detailed and simultaneous data on the movements and vocalizations of almost all members of three wild white-nosed coati ( Nasua narica ) groups, and analyzed 2,401 individual decisions between conflicting travel directions that were indicated by acoustically signaling groupmates. Decision-making was shared: individuals favored directions that had majority support, and we found evidence that they used acoustic signals and movement cues to infer majority support. Individuals were also more likely to choose directions favored by closer kin and by groupmates in more frontward spatial positions. Although decisions were shared, influence was not equally distributed across individuals. Individuals who consistently formed majorities or occupied frontward spatial positions had higher influence over travel direction. By explicitly linking decisions by individual followers to emergent patterns of influence among potential leaders, our results suggest that influence is a complex product of higher-order interactions that are likely dependent on group demography and socio-spatial structure.

For many group living species, vocal signals are a vital form of communication needed to coordinate social behaviours. Investigating these processes first requires a detailed description of the species' vocal repertoire. White-nosed coatis (Nasua narica), which forage and move together in forest habitats, are thought to rely on vocalisations to coordinate movements and maintain group cohesion. However, quantitative studies of their vocal repertoire are lacking. We examined the vocal repertoire of white-nosed coatis from wild populations in Panama and Arizona, USA, to gain a more comprehensive view of their calling behaviours. By combining traditional acoustic analyses with an unsupervised approach based on spectrogram structure, we characterised the diversity of calls in this species and described the temporal and structural features of their vocalisations. We identified 19 call types, with some of these calls emitted in multi-syllable call sequences or in fast succession. We found variability in call rates among group members, which may be driven by differences in social status within the group. In addition, our results indicate that white-nosed coatis likely have individually-recognisable vocalisations. This study provides a foundational description of the white-nosed coati vocal repertoire, laying the groundwork for future research on vocal communication in this species.

Here, we present HyenaSET, a large (~1640 hours) bioacoustic dataset derived from collar-mounted audio recorders deployed on 19 spotted hyenas in the Maasai Mara National Reserve, Kenya. Within this dataset, 243 hours have been strongly labeled by identifying the onset and offset of all vocalizations as well as their types, and the labels have been validated by experts on hyena vocalizations and behavior. Within the strongly labeled data, the total amount of time hyenas were vocalizing was 9.5 hours (3.9%). Furthermore, each vocalization has been manually identified as "focal" (emitted by the hyena wearing the collar) or "non-focal" (emitted by a nearby conspecific), making use of information from collar-mounted accelerometers that picked up vibrations of the animal's throat when it produced vocalizations. In addition to the labeled data, we also provide a large corpus of unlabeled data from the same recordings, which can be used for un- or self-supervised machine learning tasks. To ensure reproducibility of this dataset as a benchmark in machine learning studies, we present it alongside five stratified cross-validation train/test splits to enable accurate comparisons, and we also provide a train/test split in which specific individuals are left out of the training set to assess generalizability across individuals. Finally, as a performance benchmark, we present baseline results for this dataset using animal2vec, a recently developed transformer-based model optimized for bioacoustic data.

Many social animals exhibit fission-fusion dynamics, where group members split into subgroups and come back together. Studying these dynamics can reveal how individuals weigh the costs and benefits of sociality. Most studies infer the drivers of fission-fusion dynamics by examining subgroup association patterns, yet different underlying processes can produce similar patterns. Here we take a fundamentally different approach by focusing on subgrouping events themselves, where decision-making plays out in real time. Using multi-sensor tracking collars, we collected simultaneous movement and vocalisation data from all group members within two wild white-nosed coati groups. We developed analytical tools to extract and characterise fission and fusion events, thus establishing a framework for quantifying the spatiotemporal dynamics of these events and the use of vocalisations before, during, and after them. We found that fissions typically occurred when groups were initially stationary, indicating that splitting does not result from loss of coordination while moving. Subgrouping was associated with reduced aggression, yet aggressive vocalisations did not precede splits, supporting the hypothesis that subgrouping is a pre-emptive mechanism to manage within-group conflict. Contact calls in moving subgroups increased before fissions and fusions, indicating that vocal communication is key to the coordination of these collective movement decisions.

Physical grooming is often used to maintain social bonds in animal groups, but opportunities to groom every group member may be limited. Vocal exchanges may provide an alternative way to sustain social ties without physical proximity. To explore the link between vocal exchanges and social relationships, we examined how dominance and dyadic bond strength influence the intensity of vocal interaction in meerkats (Suricata suricatta). Focusing on “sunning calls”, socially facilitated vocalizations produced when meerkats warm themselves in the sun, we conducted experiments in which we played sunning calls to individuals and recorded their responses. Overall, subordinate individuals increased their call rate in response to dominant calls, whereas dominant receivers showed little or no change. The effect of tie strength was weak overall but varied across dyad types, tending to be negative in dominant–subordinate pairs and slightly positive among subordinates. These results support the notion that sunning calls serve a social regulatory function, possibly acting as bonding or appeasing signals primarily directed toward dominants. The asymmetry between dominant and subordinate responses highlights the strategic nature of social relationships within a despotic society. Engaging in continuous reciprocal interactions may promote tolerance and reduce conflict; thus, investing in such interactions with influential dominants would align with the “vocal grooming” function of these exchanges. Future studies that causally link vocal interactions to long-term relationship outcomes will be key to understanding how such vocal exchanges complement physical affiliation across species.

Social animals often face a trade-off between the costs of foraging competition among group members and the benefits of protection from predators offered by group living. The spatial position of an individual in relation to the other group members during foraging can mediate the effects of this trade-off as individuals at the front or edge may have better access to food resources, but also higher predation risk than individuals near the centre of the group. Using meerkats (Suricata suricatta) as a model species, we investigated the effect of individual spatial position within a group on foraging success. We determined the spatial position of individuals in a meerkat group by fitting the animals with high-resolution GPS loggers. As a proxy of foraging success, we used meerkats’ individual body weight differences between the start and the end of daily data collection over foraging periods (3 hours). We found significant individual differences in meerkats’ spatial positions within the group. In addition, age-dependent differences in spatial position became obvious, with younger individuals spending more time in the centre of the group and less time in the front. However, younger individuals who spent more time on the side of the group relative to older individuals had higher daily weight gain, indicating more successful foraging. In older individuals, we found that the dominant females tended to spend more time towards the front of the group, but gained less weight in this location, contrary to the predicted association between front edge of the group and better access to food resources. Our results suggest that the relationship between weight gain and spatial position is highly nuanced and likely dependent on more than just trade-offs between foraging success and predation risk. Highlights Spatial position relative to other group members during foraging was a highly repeatable trait for individual meerkats, though time spent near the front of the group was more strongly influenced by individual traits (age, sex, and social rank) Younger meerkats spent less time in the front and more time in the centre of the group relative to older meerkats, but had the highest foraging success towards the sides of the group Dominant female meerkats spent more time near the front of the group, but had reduced foraging success in this position Meerkats may trade-off more than just foraging success and predation risk when making decisions about where to position themselves relative to other group members

Collective behaviour, social interactions and leadership in animal groups are often driven by individual differences. However, most studies focus on same-species groups, in which individual variation is relatively low. Multispecies groups, however, entail interactions among highly divergent phenotypes, ranging from simple exploitative actions to complex coordinated networks. Here we studied hunting groups of otherwise-solitary Octopus cyanea and multiple fish species, to unravel hidden mechanisms of leadership and associated dynamics in functional nature and complexity, when divergence is maximized. Using three-dimensional field-based tracking and field experiments, we found that these groups exhibit complex functional dynamics and composition-dependent properties. Social influence is hierarchically distributed over multiscale dimensions representing role specializations: fish (particularly goatfish) drive environmental exploration, deciding where, while the octopus decides if, and when, the group moves. Thus, ‘classical leadership’ can be insufficient to describe complex heterogeneous systems, in which leadership instead can be driven by both stimulating and inhibiting movement. Furthermore, group composition altered individual investment and collective action, triggering partner control mechanisms (that is, punching) and benefits for the de facto leader, the octopus. This seemingly non-social invertebrate flexibly adapts to heterospecific actions, showing hallmarks of social competence and cognition. These findings expand our current understanding of what leadership is and what sociality is.