Detailed temporal structure of communication networks in groups of songbirds

Collective signalling is shaped by feedbacks between signaller variation, receiver perception and acoustic environment in a simulated communication network

Communication takes place within a network of multiple signallers and receivers. Social network analysis provides tools to quantify how an individual's social positioning affects group dynamics and the subsequent biological consequences. However, network analysis is rarely applied to animal communication, likely due to the logistical difficulties of monitoring natural communication networks. We generated a simulated communication network to investigate how variation in individual communication behaviours generates network effects, and how this communication network's structure feeds back to affect future signalling interactions. We simulated competitive acoustic signalling interactions among chorusing individuals and varied several parameters related to communication and chorus size to examine their effects on calling output and social connections. Larger choruses had higher noise levels, and this reduced network density and altered the relationships between individual traits and communication network position. Hearing sensitivity interacted with chorus size to affect both individuals' positions in the network and the acoustic output of the chorus. Physical proximity to competitors influenced signalling, but a distinctive communication network structure emerged when signal active space was limited. Our model raises novel predictions about communication networks that could be tested experimentally and identifies aspects of information processing in complex environments that remain to be investigated. This article is part of the theme issue 'The power of sound: unravelling how acoustic communication shapes group dynamics'.

Exploring the role of vocalizations in regulating group dynamics

Because of the diverging needs of individuals, group life can lead to disputes and competition, but it also has many advantages, such as reduced predation risk, information sharing and increased hunting success. Social animals have to maintain group cohesion and need to synchronize activities, such as foraging, resting, social interactions and movements, in order to thrive in groups. Acoustic signals are highly relevant for social dynamics, some because they are long-ranging and others because they are short-ranging, which may serve important within-group functions. However, although there has been an increase in studies concentrating on acoustic communication within groups in the past decade, many aspects of how vocalizations relate to group dynamics are still poorly understood. The aim of this review is to present an overview of our current knowledge on the role of vocalizations in regulating social group dynamics, identify knowledge gaps and recommend potential future research directions. We review the role that vocalizations play in (i) collective movement, (ii) separation risk and cohesion maintenance, (iii) fission-fusion dynamics, and (iv) social networks. We recommend that future studies aim to increase the diversity of studied species and strengthen the integration of state-of-the-art tools to study social dynamics and acoustic signals. This article is part of the theme issue 'The power of sound: unravelling how acoustic communication shapes group dynamics'.

The ecology of zebra finch song and its implications for vocal communication in multi-level societies

Acoustic signalling is crucial in affecting movements and in social interactions. In species with dynamic social structures, such as multi-level societies, acoustic signals can provide a key mechanism allowing individuals to identify and find or avoid each other and to exchange information. Yet, if the spacing between individuals regularly exceeds the maximum signalling range, the relation between movements and signals becomes more complex. As the best-studied songbird in captivity, the zebra finch (Taeniopygia castanotis) is a species with individually distinct songs that are audible over just a few metres and a widely ranging dynamic multi-level social organization in the wild, raising questions on the actual role of its song in social cohesion and coordination. Here, we provide an overview of birdsong in social organizations (networks) and use the ecology of the zebra finch and male song to discuss how singing can facilitate social cohesion and coordination in species where the signal range is very short. We raise the question of the extent to which zebra finches are a representative species to understand the function of song in communication, and we broaden current views on the function of birdsong and its individual signature. This article is part of the theme issue 'The power of sound: unravelling how acoustic communication shapes group dynamics'.

A model of marmoset monkey vocal turn-taking

Vocal turn-taking has been described in a diversity of species. Yet, a model that is able to capture the various processes underlying this social behaviour across species has not been developed. To this end, here we recorded a large and diverse dataset of marmoset monkey vocal behaviour in social contexts comprising one, two and three callers and developed a model to determine the keystone factors that affect the dynamics of these natural communicative interactions. Notably, marmoset turn-taking did not abide by coupled-oscillator dynamics, but rather call timing was overwhelmingly stochastic in these exchanges. Our features-based model revealed four key factors that encapsulate the majority of patterns evident in the behaviour, ranging from internal processes, such as particular states of the individual driving increased calling, to social context-driven suppression of calling. These findings indicate that marmoset vocal turn-taking is affected by a broader suite of mechanisms than previously considered and that our model provides a predictive framework with which to further explicate this natural behaviour at both the behavioural and neurobiological levels, and for direct comparisons with the analogous behaviour in other species.

SMART-BARN: Scalable multimodal arena for real-time tracking behavior of animals in large numbers

The SMART-BARN (scalable multimodal arena for real-time tracking behavior of animals in large numbers) achieves fast, robust acquisition of movement, behavior, communication, and interactions of animals in groups, within a large (14.7 meters by 6.6 meters by 3.8 meters), three-dimensional environment using multiple information channels. Behavior is measured from a wide range of taxa (insects, birds, mammals, etc.) and body size (from moths to humans) simultaneously. This system integrates multiple, concurrent measurement techniques including submillimeter precision and high-speed (300 hertz) motion capture, acoustic recording and localization, automated behavioral recognition (computer vision), and remote computer-controlled interactive units (e.g., automated feeders and animal-borne devices). The data streams are available in real time allowing highly controlled and behavior-dependent closed-loop experiments, while producing comprehensive datasets for offline analysis. The diverse capabilities of SMART-BARN are demonstrated through three challenging avian case studies, while highlighting its broad applicability to the fine-scale analysis of collective animal behavior across species.

Computational bioacoustics with deep learning: a review and roadmap

Animal vocalisations and natural soundscapes are fascinating objects of study, and contain valuable evidence about animal behaviours, populations and ecosystems. They are studied in bioacoustics and ecoacoustics, with signal processing and analysis an important component. Computational bioacoustics has accelerated in recent decades due to the growth of affordable digital sound recording devices, and to huge progress in informatics such as big data, signal processing and machine learning. Methods are inherited from the wider field of deep learning, including speech and image processing. However, the tasks, demands and data characteristics are often different from those addressed in speech or music analysis. There remain unsolved problems, and tasks for which evidence is surely present in many acoustic signals, but not yet realised. In this paper I perform a review of the state of the art in deep learning for computational bioacoustics, aiming to clarify key concepts and identify and analyse knowledge gaps. Based on this, I offer a subjective but principled roadmap for computational bioacoustics with deep learning: topics that the community should aim to address, in order to make the most of future developments in AI and informatics, and to use audio data in answering zoological and ecological questions.

Deep perceptual embeddings for unlabelled animal sound events

Evaluating sound similarity is a fundamental building block in acoustic perception and computational analysis. Traditional data-driven analyses of perceptual similarity are based on heuristics or simplified linear models, and are thus limited. Deep learning embeddings, often using triplet networks, have been useful in many fields. However, such networks are usually trained using large class-labelled datasets. Such labels are not always feasible to acquire. We explore data-driven neural embeddings for sound event representation when class labels are absent, instead utilising proxies of perceptual similarity judgements. Ultimately, our target is to create a perceptual embedding space that reflects animals' perception of sound. We create deep perceptual embeddings for bird sounds using triplet models. In order to deal with the challenging nature of triplet loss training with the lack of class-labelled data, we utilise multidimensional scaling (MDS) pretraining, attention pooling, and a triplet mining scheme. We also evaluate the advantage of triplet learning compared to learning a neural embedding from a model trained on MDS alone. Using computational proxies of similarity judgements, we demonstrate the feasibility of the method to develop perceptual models for a wide range of data based on behavioural judgements, helping us understand how animals perceive sounds.

Familiarity enhances moment-to-moment behavioral coordination in zebra finch (Taeniopygia guttata) dyads

An individual's ability to respond to and align with the behavior of others is a fundamental component of social behavior. Zebra finches form lifelong monogamous pair bonds; however, zebra finches are also gregarious and can form strong social bonds with same-sex conspecifics. Here, we quantified behavior during brief 10-min reunions for males and females in five types of social conditions: monogamously bonded opposite-sex partners, familiar same-sex, familiar opposite-sex, novel same-sex, and novel opposite-sex dyads. We analyzed these interactions in three ways. First, we quantified overall activity levels (call and movement rates) for each individual. Second, we measured how coordinated calls and movements were by calculating (a) the percent difference in activity rates as an estimate of how similar calling and movement activity were between individuals within a dyad, and (b) the sliding correlation coefficients for time-stamped calls and movements for each dyad. Finally, we described multimodal behavioral profiles of coordination using principal component analyses. Overall, females were more active than males. For both females and males, activity levels as well as the coordination of calls and movements were significantly affected by social condition. In general, monogamous partners, female familiar same-sex dyads and familiar opposite-sex dyads were the most coordinated. This effect of familiarity shows that moment-to-moment behavioral coordination can be influenced by prior social experiences. Quantifying patterns of coordination or social synchrony may prove valuable for understanding the effects of social experience on brain and behavior. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Call and be counted! Can we reliably estimate the number of callers in the indri's (Indri indri) song?

Estimating the number of animals participating in a choral display may contribute reliable information on animal population estimates, particularly when environmental or behavioral factors restrict the possibility of visual surveys. Difficulties in providing a reliable estimate of the number of singers in a chorus are many (e.g., background noise masking, overlap). In this work, we contributed data on the vocal chorusing of the indri lemurs (Indri indri), which emit howling cries, known as songs, uttered by two to five individuals. We examined whether we could estimate the number of emitters in a chorus by screening the fundamental frequency in the spectrograms and the total duration of the songs, and the reliability of those methods when compared to the real chorus size. The spectrographic investigation appears to provide reliable information on the number of animals participating in the chorusing only when this number is limited to two or three singers. We also found that the Acoustic Complexity Index positively correlated with the real chorus size, showing that an automated analysis of the chorus may provide information about the number of singers. We can state that song duration shows a correlation with the number of emitters but also shows a remarkable variation that remains unexplained. The accuracy of the estimates can reflect the high variability in chorus size, which could be affected by group composition, season and context. In future research, a greater focus on analyzing frequency change occurring during these collective vocal displays should improve our ability to detect individuals and allow a finer tuning of the acoustic methods that may serve for monitoring chorusing mammals.

Dawn chorus interpretation differs when using songs or calls: the Dupont's Lark Chersophilus duponti case

Background

Territorial songbirds vocalise intensively before sunrise and then decrease their vocal activity. This creates a communication network that disseminates essential information for both males and females. The function of dawn chorus in birds has been frequently interpreted according to seasonal variation of singing as the breeding season advances, but potential differences in seasonal variation of song and calls for the same species have not been taken into account.

Methods

We chose Dupont’s Lark as a model species to study whether the seasonal pattern of dawn chorus differs between singing and calling activity, because in this species most daily songs and calls are uttered at dawn. We registered vocal activity of Dupont’s Lark before and around dawn in three different populations, through repeated sampling over the entire breeding season of two consecutive years.

Results

We found that dawn singing parameters remained constant or presented an increasing trend while dawn calling activity decreased as breeding season advanced. We also found different daily patterns for singing and calling, with birds calling mostly during the first 30 minutes of dawn choruses and singing peaking afterwards.

Discussion

The different time patterns of songs and calls may indicate that they serve diverse functions at dawn in the Dupont’s Lark. Relaxation of dawn calling activity after the first month of the breeding season would suggest that dawn calling may be mainly related to mate attraction, while constant dawn singing throughout the breeding period would support a relationship of dawn singing to territorial defence. Our study highlights that the type of vocalisation used is an important factor to consider in further research on dawn choruses, since results may differ depending on whether calls or songs are analysed.

A spatiotemporal analysis of acoustic interactions between great reed warblers (Acrocephalus arundinaceus) using microphone arrays and robot audition software HARK

Acoustic interactions are important for understanding intra‐ and interspecific communication in songbird communities from the viewpoint of soundscape ecology. It has been suggested that birds may divide up sound space to increase communication efficiency in such a manner that they tend to avoid overlap with other birds when they sing. We are interested in clarifying the dynamics underlying the process as an example of complex systems based on short‐term behavioral plasticity. However, it is very problematic to manually collect spatiotemporal patterns of acoustic events in natural habitats using data derived from a standard single‐channel recording of several species singing simultaneously. Our purpose here was to investigate fine‐scale spatiotemporal acoustic interactions of the great reed warbler. We surveyed spatial and temporal patterns of several vocalizing color‐banded great reed warblers (Acrocephalus arundinaceus) using an open‐source software for robot audition HARK (Honda Research Institute Japan Audition for Robots with Kyoto University) and three new 16‐channel, stand‐alone, and water‐resistant microphone arrays, named DACHO spread out in the bird's habitat. We first show that our system estimated the location of two color‐banded individuals’ song posts with mean error distance of 5.5 ± 4.5 m from the location of observed song posts. We then evaluated the temporal localization accuracy of the songs by comparing the duration of localized songs around the song posts with those annotated by human observers, with an accuracy score of average 0.89 for one bird that stayed at one song post. We further found significant temporal overlap avoidance and an asymmetric relationship between songs of the two singing individuals, using transfer entropy. We believe that our system and analytical approach contribute to a better understanding of fine‐scale acoustic interactions in time and space in bird communities.

Power-law scaling of calling dynamics in zebra finches

Social mammals and birds have a rich repertoire of communication calls. Some call types are uttered rarely but in specific contexts while others are produced in large numbers but are not linked to a specific context. An example for the latter is the "stack" call that zebra finches (Taeniopygia guttata) utter thousands of times per day in a seemingly erratic manner. We quantified this calling activity of captive zebra finches by using on-bird telemetric microphones that permitted a precise temporal resolution. We separated the calling interactions into the reactive and the self-contained calls. Despite a large dynamic range in the succession of calling events, the temporal distribution of the reactive and the self-contained callings was characterized by a power-law with exponents ranging between 2 and 3, which implies that all calls in that scale have similar dynamic patterns. As birds underwent physiological (water availability) and social (separation from the reproductive partner) changes, their calling dynamics changed. Power-law scaling provided an accurate description of these changes, such that the calling dynamics may inform about an individual's physiological and/or social situations state, even though a single "stack" call has no predetermined meaning.