Rise of the machines: Integrating technology with playback experiments to study cetacean social cognition in the wild

Cetaceans represent an evolutionary peak in terms of their cognitive capacities, complex communication systems and their structured, multilevel societies. However, the difficulty of observing their behaviour underwater means that studying whale and dolphin sociality in the wild poses some significant methodological challenges. Traditionally, playback experiments have been used to explore aspects of communication and cognition in whales and dolphins, particularly with trained animals under human care. However, while these studies have provided major breakthroughs in our understanding of cetacean social cognition, it is difficult to know whether these findings generalize to wild animals. In recent years, new state‐of‐the‐art technology (drones and non‐invasive sound and movement tags) have revolutionized the field of marine mammal behaviour, providing unparalleled information on the fine‐scale behaviour of individuals in the wild. Here, we review the state of the field, combining published studies with our own extensive experience, to demonstrate how these new technologies fundamentally change the behavioural metrics that we are able to measure; allowing us to move from categorical observations to quantifying fine‐scale changes in movement, activity and vocal behaviour. We discuss how conducting playback experiments alongside these new technologies combines rigorous experimental design with strong ecological validity and increased reproducibility and can be adapted for many social species, setting the standard for high‐calibre, field‐based experiments that explore animal social cognition in the wild.


| INTRODUC TI ON TO THE HIS TORY OF INVE S TI G ATING CE TACE AN SO CIAL COG NITI ON
Cetaceans have long fascinated researchers due to their multi-level societies (Cantor et al., 2015;Connor et al., 2011), their culturally transmitted dialects and foraging traditions (Allen et al., 2013;Ford, 1991;Rendell & Whitehead, 2003;Wild et al., 2019;Wild et al., 2020;Yurk et al., 2002), their cooperative relationships (Connor & Krützen, 2015;Gazda et al., 2005) and their vocal flexibility (Janik, 2014). The complex social structures of cetacean species, and their highly differentiated relationships therein, has meant that they have evolved sophisticated ways of keeping track of conspecifics; a component of social living that is believed to be the driving force of social cognition. However, their aquatic lifestyle, and the difficulty of observing cetaceans underwater means that studying their social behaviour in the wild remains challenging.
Cetaceans live in a world governed by sound rather than light.
As a consequence, they rely on sound for communication, navigation and for localizing and capturing prey, and have in turn evolved some of the most specialized adaptations for sound production and perception. Groundbreaking research on dolphin social cognition has, over the years, been primarily conducted with animals under human care. Controlled playback experiments have revealed much about the perceptive and cognitive abilities of cetaceans, mainly bottlenose dolphins, that could not be easily obtained in the wild.
For example, a seminal playback study from the early 1980s, showed that a trained common bottlenose dolphin could mimic computergenerated sounds with high fidelity and reliability, and this work is still widely cited as experimental evidence that the species has an impressive talent for vocal flexibility and imitation (Richards et al., 1984). Sound playback experiments have also been used with common bottlenose dolphins under human care to measure cardiac responses to conspecific vocalizations (Miksis et al., 2001), to determine how individuals discriminate between different whistle types (Harley, 2008), to show that individuals are capable of cross-modal perception of identity by sound and taste (Bruck et al., 2022), and to identify critical time intervals for vocal turn-taking (King et al., 2014;Nakahara & Miyazaki, 2011).
In more recent years, two-way playback experiments have allowed researchers to uncover the function of cetacean vocalizations within an interactive setting (King, 2015). This flexible playback design allows the researcher to manipulate the sounds played back as a function of the focal subject's current vocal behaviour and therefore present vocal signals interactively (King, 2015). One study using interactive playbacks with common bottlenose dolphins under human care elicited calling in focal animals and then played back either the same call produced by the focal animal or a different call (control) to simulate call matching by conspecifics (King et al., 2014). Individuals showed a strong vocal response to being matched but no difference in approach distance and swimming behaviour between the two playback types, suggesting that call matching is an affiliative means of addressing specific individuals (King et al., 2014). At a similar time, a large-scale study with 43 common bottlenose dolphins under human care, from six different facilities, conducted a habituationdishabituation sound playback experiment to determine if individuals recognized the signature whistles of former tank mates, with unfamiliar signature whistles used as a control (Bruck, 2013). In this experiment, only the physical response was measured, namely head turns towards the speaker, approach distance and swimming behaviour, which were assigned a qualitative response level.
The author found that bottlenose dolphins showed a much stronger physical response to familiar signature whistles of former tank mates, suggestive of long-term social recognition (up to 20 years) in this species (Bruck, 2013). While these studies have provided major breakthroughs in our understanding of cetacean social cognition, it is difficult to know whether findings would generalize to wild animals. Social experiences are likely very different for dolphins under human care compared to their wild counterparts and it is, therefore, desirable to complement this research with experiments on wild cetaceans within the socioecological settings in which their cognition and communication systems evolved. Yet taking these types of experiments from the laboratory into the field remains challenging.
Playback experiments with wild cetaceans began over 50 years ago, with this early body of work showing that the detection of conspecific sounds could trigger behavioural responses in a variety of cetacean species. Some of the first studies involved playing back conspecific sounds to wild and captive belugas, who modified their vocal behaviour in response to playbacks (Morgan, 1970(Morgan, , 1973. This was followed by a study published in Science in 1980, where researchers played back a variety of conspecific, heterospecific and man-made sound to demonstrate that southern right whales responded to conspecific sounds by increasing their rate of vocal production and approaching the speaker, but did not vocalize or approach when exposed to other sounds (Clark & Clark, 1980). Over 20 years later, this finding was replicated in northern right whales, where the playbacks of calls from social groups elicited approaches in the focal animals, particularly if they were males (Parks, 2003).
In this latter study, a qualitative level of playback response was calculated based on approach distance and swim speed, and a single hydrophone was used to record any vocal responses (Parks, 2003).
Early work with humpback whales also revealed that focal animals rapidly approached the sound source after playbacks of feeding or social calls, but not to playbacks of male song, suggesting that song may play a role in male-male competition rather than in mate attraction (Mobley et al., 1988;Tyack, 1983). Playback experiments have been conducted with sperm whales, where researchers played back sperm whale social calls, that is, codas, to social or resting groups at the surface to assess the function of stereotyped coda patterns, but they found few consistent responses (Rendell & Whitehead, 2005).
In this study, each group was exposed to two coda playbacks, one from their vocal clan (members of the same vocal clan use the same coda dialects) and one from a different vocal clan. However, there was large variation in coda production rates in response to playback, with no influence of clan relationship (Rendell & Whitehead, 2005).
Indeed, the most common response was a cessation of coda production, although some groups did commence coda production or switch coda types (Rendell & Whitehead, 2005). As the authors point out, interpretation of these results is challenging, which may be due to the choice of the single vocal response variable, and future studies should measure changes in individual movement as well as orientation in response to playbacks (Rendell & Whitehead, 2005).
These early studies were certainly influential in setting the scene for future experimental work with wild cetaceans but measuring individual or group level responses to these playbacks was limited. Most of these studies, if they measured a movement response, used either theodolite tracking (a surveying instrument that measures angles in the horizontal and vertical planes, which allows animal movement to be tracked visually) or laser range finders to quantify positions and movements of animals at the surface, but this requires either an elevated position for theodolite tracking or fairly close proximity with laser range finders in order to obtain accurate distance measurements ( Figure 1). Furthermore, measuring subtle changes in orientation was nigh on impossible without some aerial view of the focal animal(s).
Even for playback experiments conducted in more recent years, the limitations of quantifying movement responses have remained.
Playback studies comparing the reaction of fish-eating killer whales to calls from the same and different pods (Filatova et al., 2011) or the reaction of bottlenose dolphins to copies of their own signature whistle or different signature whistles (King & Janik, 2013) both relied on estimating directional movement response of the animals (when at the surface) to or from the boat using observer estimates or laser range finders. For both these studies, the strongest measured response was an increase in vocal activity to calls from the same pod (Filatova et al., 2011) or replying with their signature whistle when hearing a copy of said whistle (King & Janik, 2013). Together, these studies highlight the logistical challenges of conducting sound playbacks at sea with free-ranging animals and the difficulty of measuring physical behavioural responses underwater (Deecke, 2007).
One group of researchers have managed to collect fine-scale behavioural data from wild cetaceans by conducting playback experiments with wild, temporarily restrained common bottlenose dolphins. In Sarasota Bay, Florida, the capture and release of wild dolphins are conducted annually for health assessment purposes and life-history studies (Wells et al., 2004). Researchers have used these events to conduct playback experiments to restrained individuals to determine the role signature whistles and non-signature whistles play in individual recognition in bottlenose dolphins (Janik et al., 2006;Sayigh et al., 1999;Sayigh et al., 2017). These studies began in the 1990s where whistle playbacks were conducted to temporarily restrained dolphins to determine whether there was sufficient information in their individually distinctive signature whistles for individual recognition (Sayigh et al., 1999).
A within-subject paired playback design was used, where individuals were either played the signature whistle of a close relative F I G U R E 1 Methods for observing and quantifying cetacean responses to sound playback experiments. (a) Playback experiments with temporarily restrained, wild dolphins during annual health checks, using head-turns as a response measure. Photo courtesy of Jim Schulz, Chicago zoological society under NMFS scientific research permit no. 522-1785 and reproduced with permission from .  (treatment) or the signature whistle of a familiar, similar-aged nonrelative (control) (Sayigh et al., 1999). A measurement tape was held above the animal along the length of its body with the playback speaker positioned at 90° (Figure 1), in order to measure the number of head turns towards the speaker that were >20° in magnitude. The vocal response of the focal animal was also measured during these playback experiments. This set-up allowed, for the first time, more subtle behavioural responses, such as head turns towards the speaker, to be recorded. In this first study, animals were significantly more likely to turn towards the speaker for the treatment playback than the control playback (Sayigh et al., 1999), but there was no difference in vocal response to the playback types (Sayigh, 1992;Sayigh et al., 1999). In order to demonstrate that identity information was encoded in the frequency modulation pattern of the signature whistle, a later study used the same experimental design combined with computer-generated playback whistles where all voice features had been removed (Janik et al., 2006). Again, there was a significant difference in head turns towards the speaker between the treatment and control but no difference in vocal response (Janik et al., 2006). Similarly, a subsequent study showed no difference in head turns towards the speaker in response to non-signature whistles from close relatives or from known unrelated individuals, suggesting that dolphins do not use voice cues to identify individuals .
These experiments have been pivotal in furthering our understanding of individual recognition in dolphins, and have also revealed that subtle movement may be a more salient response variable to measure when investigating recognition.  , these were limited in their applicability due to their large sizes and difficulty of deployment. On the other hand, UAVs are small, easy to deploy from research vessels, commercially available from many different vendors and relatively affordable ( Figure 1; Table 1).

| CHANG ING PER S PEC TIVE S -NE W TECHNOLOG IE S E X TEND OBS ERVATIONAL
UAVs are already widely used in the field of marine mammal research, especially for measuring cetacean body condition using photogrammetry (Azizeh et al., 2021;Christiansen et al., 2020Christiansen et al., , 2021Irschick et al., 2020), for surveying marine megafauna (Hodgson et al., 2017) and for quantifying cetacean response to disturbance (Durban et al., 2022). More recently, however, UAVs have been used to collect fine-scale behavioural data from cetacean groups.
For example, UAVs have been used to observe direct social interactions in a pod of southern resident killer whales in order to gain a better understanding of the role of both age and sex in structuring social relationships (Weiss et al., 2021). UAVs, either launched from a research vessel or from shore, were used to collect video data on synchronous surfacings and affiliative physical contact between members of the same pod. These interaction data were compared to proximity-based association data, which is a traditional measure of cetacean social network structure. While neither age nor sex predicted association strength, both significantly predicted rates of synchronous surfacing and contact behaviour, revealing that age and sex are important determinants of social interactions in killer whales (Weiss et al., 2021). Importantly, this highlights that cetacean social relationships are not being fully characterized by proximity-based association data alone (Weiss et al., 2021). A similar approach using UAVs to measure synchronous surfacing events was used to quantify social relationships in a group of male Risso's dolphins (Hartman et al., 2020). These studies are the first to use behavioural data collected from UAVs to construct cetacean social networks (Hartman et al., 2020;Weiss et al., 2021).
UAVs have also allowed researchers to gain unique insights into cetacean foraging strategies by recording previously undescribed behaviours that are challenging to detect for boat-based observers (Ramos, Santoya, et al., 2021). UAVs allow individuals to be visually tracked when underwater and recent studies have used UAVs to provide evidence of group hunting in harbour porpoises (Ortiz et al., 2021), a cetacean species believed to be relatively solitary; of food sharing in rough-toothed dolphins ; and of a foraging specialization called mud-ring feeding in a population of bottlenose dolphins in the Caribbean, a behaviour that was previously thought to be limited to bottlenose dolphins in Florida (Ramos, Santoya, et al., 2021). These studies all convincingly demonstrate that UAVs are a powerful tool for providing new insight into the complexity of cetacean social organization and behaviour in the wild.

| Moving observations below the surface
Over the last two decades, high-resolution sound and movement archival or biologging tags, including the DTAG (Johnson et al., 2009;Johnson & Tyack, 2003), Acousonde (Burgess, 2009)  Tags have changed the way we can investigate behaviour and cognition in the wild. Sound and movement recording tags were initially invented to investigate how cetaceans respond to anthropogenic noise (Tyack, 2009); as such, they have been extensively used for understanding how a variety of cetacean species respond to, and in turn are impacted by, Navy sonar (i.e. Harris et al., 2018;Miller et al., 2012;Southall et al., 2012), large commercial vessels (Aguilar Soto et al., 2006;Holt et al., 2021;Wisniewska et al., 2018) and echosounders (Quick et al., 2017). Along the way, behavioural response studies have enabled a range of investigations on how different cetaceans respond to other sounds in their environment, including potential predators such as killer whales (Cure et al., 2012;Curé et al., 2015). These studies have revealed that long-finned pilot whales are capable of complex acoustic discrimination between fish-eating and mammal-eating killer whales (Cure et al., 2019), and have also found that species-specific differences in anti-predator responses seem to predict how species respond to, and are impacted by, anthropogenic noise (Miller et al., 2022).
In toothed whales, acoustic tags allow us to tap into the echolocation behaviour (Madsen et al., 2005) and even the returning echoic information (Johnson et al., 2004), which echolocating cetaceans use for decision-making . suggesting that calls are important for maintaining or re-establishing contact with social groups (Jensen et al., 2011). In long-finned pilot whales, tags combined with focal follow observations have shown that call rates increase during dispersed group foraging but that tight surface groups can be relatively quiet, supporting the idea that calls are especially important for group cohesion during periods of active foraging (Visser et al., 2017). Further, because tags allow for tracking sound production concurrently with movement even when far from the surface, they have helped us understand the acoustic and social behaviour of a range of cetacean species that communicate out of sight of vessel-based observers, from the function of slow clicks (Madsen et al., 2002;Oliveira et al., 2013) and codas (Madsen, 2012;Oliveira et al., 2016) in sperm whales, to acoustic communication in bottom-foraging humpback whales (Parks et al., 2014) or to the context of fin whale  and blue whale (Lewis et al., 2018) sound production. ing to pre-playback behaviour), the approach/avoidance distance in metres (i.e. the maximum distance that the subject moved towards or away from the sound source), the change in orientation in degrees (i.e. the relative change in heading upon responding to the playback) and whether the subject oriented towards the source when responding to the playback. The authors found that while there was no difference in latency to response; response duration, approach distance and orientation towards the source all differed significantly between the treatment and control playbacks. Notably, a vocal response by subjects (recorded with a towed array system: [Quick et al., 2008]) was only evident in 5%

| Advantages and pitfalls
In combination with playback experiments, these new technologies extend our capacity to observe responses of wild, unrestrained cetaceans ( Table 1). While typical vessel-based observations are limited to observing cetaceans at the surface, aerial observations conducted in relatively calm weather can track movement of animals that are submerged in several metres of water (King et al., 2021;Ortiz et al., 2021;Ramos, Santoya, et al., 2021;Weiss et al., 2021). As a result, much more subtle responses such as orienting behaviours like head-turns or short-term changes in orientation or speed can be accurately captured. Tags  and 30 s (T + 30) after we played the whistle of their core alliance member (treatment) and another known adult male (control). Location of playback source is located with the arrow. Animals showed a strong physical response by turning towards the playback speaker for the treatment playback but not for the control playback (note: slight head turn for the control playback is visible at T − 0). No vocal response was given to either playback. It is possible to differentiate between individuals from the drone footage using body markings and pigmentation, but supplemental visual information from the vessel (e.g. identifying individuals through dorsal fin characteristics) is also recorded to ensure individual animals can be tracked continuously from the drone. Data taken from (King et al., 2021).
visually in several delphinid species in response to playbacks of killer whale calls (Bowers et al., 2018;Visser et al., 2016). However, tracking conspecifics can also be critical for investigating signals used for coordination between individuals, or to mediate interactions or mitigate aggression during reproductive conflicts.
Tags have an innate advantage for capturing vocal responses to playbacks. While mobile localization arrays (Quick et al., 2008) can be combined with drone observations to capture acoustic responses from the exposed individuals (King et al., 2021), they can be limited by background or flow noise. In contrast, acoustic tags record sound signals from focal individuals and nearby conspecifics with a high signal-to-noise ratio (SNR) and are unlikely to miss signals from the tagged animal, meaning that lack of acoustic responses or quieting behaviours (in response to potential perceived threats) can be accurately assessed (Figure 3). However, it can be challenging to unequivocally differentiate communication signals from a tagged animal from those of nearby conspecifics (Johnson et al., 2009;Saddler et al., 2017;Stimpert et al., 2020) and this potential ambiguity needs to be factored into the planning and design of experiments.
Another advantage of tags is that they allow for measuring the received sound level and in-situ SNR of the playback signal that animals were exposed to (DeRuiter et al., 2013). Proper experimental design should standardize the received sound exposure levels during trials by calibrating emitted source levels and by controlling transmission range, but sound propagation conditions can vary greatly in complex shallow-water habitats (i.e. Quintana-Rizzo et al., 2006) and it is not always easy to control exact transmission distance when conducting experiments to freely moving animal groups. By quantifying received sound level for transmissions that do not overlap with other noise such as from the tag breaking the sea surface, this variation can be incorporated directly into response models and thus improve statistical power. While this can obviously influence response intensity, temporal variables such as response latency may also be slightly affected by differences in exposure level since the auditory system tends to respond faster

F I G U R E 3
Measuring behavioural responses to sound playback experiments with tags. Data from a tagged pilot whale exposed to killer whale playback. Top: Time-frequency spectrogram of a 30-s playback period showing five playback stimuli (red) and an acoustic call (blue) from within the social group right after the first playback stimulus. Two additional playback stimuli were transmitted but masked by noise from surfacing at t = 5 s and t = 20s. Middle: Diving behaviour of animal during playback period and during an equivalent period before and after playback. Bottom: Heading of animal relative to magnetic north, showing a 60-degree short-term orienting response after start of playback. Right: Dead-reckoned track reconstruction showing the deviation in trajectory of the tagged whale during onset of playback. Note that concurrent vessel observations were needed to establish that this was an orienting response towards the playback vessel (approximate direction where playback signal is coming from indicated with the black arrow).
to signals of higher intensity for both humans (Wolfe et al., 1978) and dolphins (Ridgway et al., 1981).
Finally, both tags and drones help understand the context under which playbacks take place. With vessel-based observations, it can be difficult to identify many subsurface behaviours.
In particular, foraging behaviours can be easy to miss unless they include a surface component such as a tail slap or other visible element. Yet, the behavioural context is likely to mediate how focal animals respond to playback experiments. For example, prey mediates behavioural responses to sonar in blue whales, where animals tend to respond more during deep foraging dives than during shallow foraging dives Southall et al., 2019). Similar differences in response likelihood as a function of foraging context have also been found for several delphinids . These contextual effects are likely to be important for mediating more subtle responses to conspecific signals used to investigate cognition. From an aerial perspective, however, it is much easier to recognize animals looking for or even chasing prey near the surface, and thus the risk of missing foraging activity is much lower (King and Jensen, pers. obs.; also see Torres et al., 2018;Ortiz et al., 2021;. Acoustic tags likewise help identify foraging behaviour because echolocation clicks and foraging buzzes are recorded directly by the tag irrespective of how far away from the surface an animal moves (Johnson et al., 2004).   (Figures 2 and 3). Ideally, a within-subject experimental design with individuals exposed to both treatment and control playbacks should be utilized, as this design offers significantly more power for the smaller sample sizes that are common when conducting playbacks to wild cetaceans (e.g. Barluet de Beauchesne et al., 2021;Cure et al., 2012;Curé et al., 2015;Deecke, 2007;King & Janik, 2013;Miller et al., 2012). Researchers should also strive to make their playback signals as realistic as possible, as any deviation from realism would likely introduce uninterpretable noise into any behavioural responses. As an example, we suggest that playback sequences incorporate small variations in amplitude from call to call, or small amounts of jitter in the inter-call interval to make call sequences seem less synthetic, as well as ensuring that the frequency response of the playback system is suitable for the signals of interest. Continually recording animal heading and drone position in relation to the boat using a time stamped voice recorder can also help with post-processing the drone video.

| Logistical and methodological considerations
Tags impose some additional constraints on playback experiments due to the need for instrumenting animals before experiments. Behavioural response studies conducted to tagged marine mammals have used periods of minimum 45 min (baleen whales) or 2 h (toothed whales) to ensure animals return to baseline behaviour before experiments (Southall et al., 2012). Interpreting responses to playbacks from tag recordings also necessitate additional planning. Finally, it is becoming increasingly important to consider whether the study animals are representative of the broader population, for example using the STRANGE framework (Webster & Rutz, 2020).
Performing experiments to wild cetaceans alleviate many of the concerns of generalizability that we may expect from studies of animals under human care. However, we still need to think about whether our observational tool might favour specific subsets of a population.
Tagging of animals can be challenging, and it is likely that some group members are easier to instrument with tags than other animals, and that the differences that shape this might have some influence on how animals in turn respond to playbacks. In delphinids, it is often easier to tag animals that are larger (typically older individuals, or male individuals in sexually dimorphic species such as killer whales or pilot whales). But individual differences in some personality traits, like boldness, may also influence how easy it is to instrument animals. These differences are important to consider during planning of experiments, and to try to mitigate as much as possible when instrumenting animals. For example, researchers could randomly select which animals to tag when first encountering a group, to avoid tagging just the boldest individuals. Drones, on the other hand, are extremely unlikely to have any inbuilt biases and thus naturally make it easy to generalize results out to the broader population.

| E THIC AL CONS IDER ATIONS
As with any animal research, there are important ethical considerations when it comes to designing and conducting sound playback experiments and/or using animal-borne tags or UAV technology.
Critically, researchers must have good understanding of the behavioural or vocal repertoire, and natural sound levels, of the focal species, the context in which these calls or signals are used, and the animal's motivation for communicating. This knowledge will ensure that any playback experiments are well designed and that any likely effects of stimuli on the focal subject are known prior to the playback. Generally, we have no reason to expect any mediumor long-term negative impacts from well-designed sound playback experiments. These approaches have been successfully used for decades and have provided remarkable insight into the natural behaviour of marine mammals. Several studies have shown that UAVs are associated with minimal underwater noise disturbance and generally have little impact on cetaceans if flown at appropriate altitudes (Christiansen et al., 2016b;Fettermann et al., 2019;Raoult et al., 2020). Responses to tagging vary from species to species and quantitative studies of how long it takes for animals to return to baseline behaviours are scarce. However, suction-cup tags are considered relatively non-invasive and unlikely to result in more than minor discomfort (Andrews et al., 2019). It is unclear how much short-term changes in behaviour are due to the direct physical disturbance during tagging rather than a response to the manoeuvring of the research vessel and associated underwater noise. Ongoing developments in cetacean tagging techniques, such as drone-based tagging, may in the future facilitate faster tag deployment without the need for close approaches with a research vessel and thus help decrease such secondary impacts on behaviour of tagged animals.

| CON CLUS IONS
To summarize, new technologies such as non-invasive sound and movement tags and drone technology fundamentally change the behavioural metrics that we are able to measure; allowing us to move from categorical observations to the quantification of finescale changes in movement, activity and vocal behaviour. Thus, many of the long-standing challenges of studying whale and dolphin behaviour in the wild have been circumvented, facilitating a more detailed study of social cognition in wild, free-ranging cetaceans.
However, while cetaceans have been the focus of this review, these technologies can easily be adapted for use with other taxa. Indeed, the use of drone imagery has already been used to quantify social structure in feral horses (Maeda et al., 2021), sound and movement tags have been adapted for a variety of terrestrial social mammals (see Demartsev et al., 2022, this issue), and deep learning methods offer an automated tool kit for measuring animal behaviour from videos (Graving et al., 2019). While the socioecology of a species may dictate which of these technologies is more suitable, the wider application of drones and sound and movement tags opens the door for more high-calibre, field-based experiments that explore animal sociality in the wild.

AUTHOR S' CONTRIBUTIONS
S.L.K. and F.H.J. conceived the idea, wrote the manuscript and gave final approval for publication.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/2041-210X.13935.

DATA AVA I L A B I L I T Y S TAT E M E N T
Not applicable.