Group cohesion in foraging meerkats: follow the moving ‘vocal hot spot’ — TEI-tagged view

Gabriella E. C. GallpersName, Marta B. ManserpersName (2017date). “Group cohesion in foraging meerkats: follow the moving ‘vocal hot spot’.” Royal Society Open Sciencetitle·j.

Object of study — whether foragingterm·key meerkatsterm·key follow the group's moving vocal hotspotterm·dom of close callsterm·dom to keep cohesion

Research question — how groups keep cohesion through communication other than vision

Contribution — a two-sided playbackterm·key experiment creating artificial vocal hotspotsterm·dom at the front and back of foragingterm·key groups

Full text — tagged

Group cohesionterm·key in foragingterm·key meerkatsterm·key: follow the moving ‘vocal hot spot’

Abstract Group coordinationterm·key, when ‘on the move’ or when visibility is low, is a challenge faced by many social living animals. While some animals manage to maintain cohesion solely through visual contact, the mechanism of group cohesionterm·key through other modes of communication, a necessity when visual contact is reduced, is not yet understood. Meerkatsterm·key ( Suricata suricatta ), a small, social carnivore, forage as a cohesive group while moving continuously. While foragingterm·key, they frequently emit ‘close callsterm·dom’, soft close-range contact calls. Variations in their call rates based on their local environment, coupled with individual movement, produce a dynamic acoustic landscape with a moving ‘vocal hotspotterm·dom’ of the highest calling activity. We investigated whether meerkatsterm·key follow such a vocal hotspotterm·dom by playing back close callsterm·dom of multiple individuals to foragingterm·key meerkatsterm·key from the front and back edge of the group simultaneously. These two artificially induced vocal hotspotsterm·dom caused the group to spatially elongate and split into two subgroupsterm·key. We conclude that meerkatsterm·key use the emergent dynamic call pattern of the group to adjust their movement direction and maintain cohesion. Our study describes a highly flexible mechanism for the maintenance of group cohesionterm·key through vocal communication, for mobile species in habitats with low visibility and where movement decisions need to be adjusted continuously to changing environmental conditions.

Keywords: close callsterm·dom, cohesion, coordinationterm·key, group split, meerkatterm·key, vocal hot spot Introduction Members of cohesively foragingterm·key groups must stay within close range of their group mates [ 1 , 2 ] to achieve the benefits of group living. The mechanisms to reach group cohesionterm·key vary between species and are usually based on simple interaction rules such as attraction, repulsion and alignment between individuals, often achieved through visual cues, such as described for fish schools [ 3 , 4 ]. However, different mechanisms are needed in moving animal groups in structured habitats, when the location of an individual is highly dynamic and visual contact is limited. In these cases, other modalities of communication between individual group members become important to maintain group cohesionterm·key. Many bird and mammal species use vocalizations to reduce the risk of separation, especially during foragingterm·key [ 5 , 6 ]. ‘Close callsterm·dom’, soft close-range contact calls, are frequently produced during group movement, and are thought to function primarily to maintain group cohesionterm·key [ 5 – 8 ]. When visibility is low, individuals often change the structure of their calls [ 9 ] or their call rate with increasing separation risk [ 5 , 7 ]. Most research has focused on changes in the vocal behaviour of individuals based on their relative spatial location to other group members [ 5 – 7 , 9 ], but has not addressed the specific mechanisms by which contact calls lead to group cohesionterm·key. Here, we investigated whether foragingterm·key meerkatsterm·key ( Suricata suricatta ) use the distribution of close callsterm·dom in the group (given by the number of individuals and their call rate at a specific location) and follow in the direction of ‘vocal hotspotsterm·dom’, areas with many closely aggregated individuals calling at high rates [ 10 ], to maintain cohesion during movement.

Meerkatsterm·key, cooperatively breeding mammals, live in social groups of up to 50 individuals [ 11 ] which forage cohesively, typically 1–10 m next to each other [ 10 ]. Individuals search for prey in the sand with their heads orientated downwards [ 11 ], thus impairing visual communication. They instead rely on an array of vocal signals to coordinate their activities [ 12 ]. During foragingterm·key, close callsterm·dom are the most frequently emitted vocalization and are thought to maintain group cohesionterm·key [ 10 , 12 ]. Meerkatsterm·key adjust their call rates depending on their social environment [ 10 , 13 , 14 ] with call rates decreasing with increasing distance to their closest neighbour, and increasing towards the front of the group, relative to the direction of movement [ 10 ]. Thus, a spatial pattern of calls, areas with few individuals calling at low rates and areas with many individuals calling at high rates (vocal hotspotterm·dom), emerges with a vocal hotspotterm·dom typically towards the centre-front of the group [ 10 ]. Here, we hypothesize that meerkatsterm·key use the vocal hotspotterm·dom to determine where most group members are located relative to their own spatial location, thus guiding each individual's future movement towards them. To test this hypothesis, we manipulated the distribution of close callsterm·dom in the group. We played back calls of multiple individuals from two sides, the front and the back edge of the group ( figure 1 ), creating two artificial vocal hotspotsterm·dom, and moved them into opposite directions. If meerkatsterm·key are guided by the distribution of close callsterm·dom in the group in their future movement direction, we would expect individuals to follow the closest artificially induced vocal hotspotterm·dom. Thus, we predict the group to elongate and/or split into two subgroupsterm·key, as individuals at the front and back end of the group perceive two different vocal hot spots.

Diagram of the experimental set-up: the length (blue dashed line) and width (green line) of a group (grey points) in the direction of movement (black arrow) as estimated in our data. The black squares indicate the location of the speakers in relation to the group during each playbackterm·key.

Material and methods The study was conducted at the Kalahari Meerkatterm·key Project, Kuruman River Reserve, in the Northern Cape of South Africa. Data were collected from January to March 2015 on eight groups, ranging from 9 to 22 individuals (mean ± s.d. = 16 ± 4) during the morning foragingterm·key session. During the period of data collection, daily temperatures ranged from 10°C to 42°C and daily rainfall from 0.0 to 30.0 mm m −2 (with a median daily rainfall of 3.1 mm m −2 on a total of 16 days during the three months of data collection). A more detailed description of the climate and habitat are provided by Clutton-Brock et al. [ 15 ]. In contrast to winter when meerkatsterm·key mostly forage continuously for the whole day, in summer meerkatsterm·key forage for a couple of hours, starting just after dawn and in the evening before sunset, and rest during the hottest part of the day. All animals were visually identifiable through dye mark combinations [ 16 ] and were habituated to close human observation within less than 1 m.

Recordings and playbacksterm·key of close callsterm·dom To test whether meerkatsterm·key follow areas with a high number of close callsterm·dom, we performed a playbackterm·key experiment, inducing two artificial vocal hotspotsterm·dom, areas with the highest calling rate. Calls, used in the playbackterm·key sound files, were recorded using a directional microphone (Sennheiser ME66 with K6 powering module) connected to a recorder (Marantz PMD660, sampling frequency 44.1 kHz, 16 bit) at a distance of 0.3–1.5 m from the individual. Playbackterm·key sound files were created for each group in Cool Edit Pro 2.0 (Syntrillium Software Corporation) by selecting high-quality calls from the recordings of the same group, where the playbacksterm·key were performed.

We created two 20 min playbackterm·key tracks for each group with close callsterm·dom of four adult subordinate group members. The call rate of each individual on the track reflected its ‘natural’ call rate, observed during previous foragingterm·key sessions and was the same for both playbackterm·key tracks. However, to avoid playing exactly the same sound files back simultaneously from both loud speakers, we randomized the order of calls of recorded individuals within each track, by assigning two random numbers to each call, one for the time within the playbackterm·key and one for the call itself. We played these close callterm·dom sound files as test condition to eight meerkatterm·key groups from the front and back edges of the group simultaneously. As a control, we played background noise, recorded in the group's home range when no meerkatsterm·key were present, with the same set-up. We played both conditions to a group during the same morning session (starting 30 min after the onset of foragingterm·key to midday), to avoid confounding time effects due to possible changes in habitat and climate between sessions. The sound files were played from a Marantz recorder connected to a speaker (X-Mini Uno XAM14) attached to the leg of the observers at about 15 cm above ground. The order in which we played the close callsterm·dom (test condition) and the background noise (control condition) was counterbalanced, with at least 15 min between conditions.

During the playbackterm·key, observers with the speakers stood at the edge of the group, maintaining a constant distance of 3 m to the closest meerkatterm·key. Hence, when meerkatsterm·key moved towards the speaker, the speaker was slowly moved away from the group to maintain a constant distance. Throughout the playbackterm·key, the observers recorded the behaviour of the group on a video camera (Sony Handycam 3.3MPXL). Every 2 min (time steps), we estimated the spatial width of the group as the distance between the furthest two animals at an angle of 90° to the direction of group movement. We also estimated the length of a group as the distance between the first and the last meerkatterm·key in the direction of group movement ( figure 1 ). We further recorded the occurrence of subgroupsterm·key, i.e. subclusters of at least three animals, as well as the distance (gap sizeterm·var) between the edges of these subgroupsterm·key. To account for variation in overall group dispersion, we normalized the gap sizeterm·var by subtracting the average distance between neighbouring meerkatsterm·key.

Statistical analysis Statistical tests were carried out using R (v. 3.3.1) [ 17 ]. In order to investigate whether groups more likely split when playing back close callsterm·dom, compared with background noise, we fitted three linear mixed effects modelsterm·mth (LMMsterm·mth) [ 18 ] with the gap sizeterm·var, the length and the width of the group as responses, and the interaction between the playbackterm·key condition and the time steps within each playbackterm·key as explanatory variables. The time steps refer to the 2 min intervals at which we documented the length and width of the group and the gap sizeterm·var between subgroupsterm·key. The intercept of the model represents the predicted values for the control playbackterm·key of background noise at time step zero. The coefficients of the model (‘estimate’ in table 1 ) give the difference in the mean of the levels of a categorical explanatory variable and the slope for continuous variables. It also indicates the direction of the effect of a factor (positive or negative) on the response variable. Gap sizeterm·var was log transformed before fitting the model to ensure the normal distribution of the data. We included the identity of the group as a random factor and accounted for multiple testing using a Bonferroni correctionterm·mth.

Results of the LMMsterm·mth testing how the length, width and the formation of subgroupsterm·key were affected by the playbackterm·key of close callsterm·dom and the playbackterm·key of background noise. Significant effects are highlighted in italics. Owing to the Bonferroni correctionterm·mth for multiple testing, the α level was reduced from 0.05 to 0.017 to account for multiple testing. The time steps are the 2 min intervals within the playbackterm·key at which the length, width and gap sizeterm·var were estimated. The intercept of the model represents the predicted values for the control playbackterm·key of background noise at time step 0. The coefficients of the model (estimate) give the difference in the mean of the levels of a categorical explanatory variable and the slope for continuous variables. It also indicates the direction of the effect of a factor (positive or negative) on the response variable.

Results When artificially inducing two vocal hot spots by playbacksterm·key of close callsterm·dom at the front and back edge in meerkatterm·key groups, individuals followed the nearby speaker, i.e. the closest location with the highest calling rate. Thus, the groups spatially elongated in the direction of the speakers, and the gap between subgroupsterm·key became significantly larger when playing back close callsterm·dom compared with background noise from the two opposite sides of the group ( table 1 and figure 2 a , b ). However, subgroupsterm·key reunited towards the end of the playbackterm·key ( figure 2 a,b ). Groups did not widen based on playbackterm·key condition ( table 1 and figure 2 c ). The time steps within each playbackterm·key caused an elongation of groups, but did not influence the gap sizeterm·var, nor the width of the group. We found no interaction between the time step within each playbackterm·key and the playbackterm·key condition ( table 1 and figure 2 ).

The mean (points) and standard deviation (whiskers) of a group's length ( a ), the gap sizeterm·var between subgroupsterm·key ( b ) and its width ( c ) during the playbackterm·key of background noise (grey) and the playbackterm·key of close callsterm·dom (black), measured at 2 min intervals throughout the experiment.

Discussion Our experiment reveals that meerkatsterm·key use close callsterm·dom, constantly emitted by their group members while foragingterm·key, to make decisions about their own movement direction. Creating two artificial vocal hotspotsterm·dom of highest call rates by playing back close callsterm·dom of several individuals at opposite sides of the group resulted in an increase in group elongationterm·dom and a higher probability of splitting into two subgroupsterm·key. This effect decreased after some time into the playbackterm·key, likely due to individual meerkatsterm·key of most groups realizing that they were getting separated and starting to produce alert calls, causing the two subgroupsterm·key to reunite (G.E.C.G. 2015, personal observation).

Two different mechanisms might lead to groups splitting: individuals might orientate and follow towards the closest speaker, as the closest vocal hotspotterm·dom, or individuals at the back of the group might move more slowly due to hearing many calls around them, but follow a general movement direction given by individuals in the front or being determined as commonly used foragingterm·key route. Given that not only the observer at the front of the group, but both observers slowly moved apart from each other and away from the closest meerkatsterm·key, the first option is more likely. Meerkatsterm·key followed the speakers carried by the observers, who decided on the movement direction (G.E.C.G. 2015, personal observation). Whether the movement direction of animals at the front of the group might have coincided with the direction chosen by the group is impossible to tell. However, the individuals in the back turned around to follow the speaker in the opposite direction. Thus, if we had played the calls from both sides instead of the front and back of a group, we would expect the same results. Following a moving vocal hot spot allows each individual to constantly direct towards the part of the moving group where likely several members aggregate, and ensures group cohesionterm·key even if visual contact is restricted due to vegetation and topographical barriers, and no general foragingterm·key direction or route governs a group's movement.

Group coordinationterm·key has mainly been studied in the context of movement initiation; however, how social animals coordinate movement while moving with restricted visual contact has been more challenging to investigate. Meerkatsterm·key adjust their close callterm·dom rate based on their own relative spatial location within the group as well as their social environment, calling less with increasing distance to the closest neighbour, and at higher rates towards the front of the group as well as when in close proximity to dominant individuals [ 10 , 13 ]. By following the emergent vocal hotspotterm·dom, meerkatsterm·key thus adjust their movement direction presumably to avoid losing contact with their group. Our results support the findings for wild sooty mangabeys ( Cerocebus atys ) [ 19 ] that an individual's movement is influenced by its social environment and the call rates of other group members. Sooty mangabeys change their movement speed and potentially their direction depending on the call rate of their own subgroupterm·key and call rates of other primate species frequently associating with them. Nevertheless, it is unclear how individual mangabeys adjust their own call rate to their surroundings and contribute to the movement of others through their own vocal behaviour.

Based on prior knowledge and the results of this experimental study, we can integrate our understanding on both the production [ 10 ] and the receiver side of a moving vocal hotspotterm·dom, defined as the location of the highest call rate within a group, and suggest it as a highly flexible coordinationterm·key mechanism enabling adjustment to fast-changing movement decisions. This mechanism shows similarities to cliff swallow ‘squeak calls’ used to attract conspecifics to moving insect swarms [ 20 ]. In a more general context, the locations with highest densities of vocal signals function similar to ant pheromone trails to recruit and direct other group members to a specific location, though in the case of ants with stationary destinations [ 21 ]. This suggests that signal hotspots, consisting of vocalizations or signals of other modalities, provide a robust and flexible way for individuals to track the core of the group and maintain cohesion during foragingterm·key. Previous studies show that close callterm·dom rate increases towards the front of the groups progression [ 10 ], suggesting that individuals located at the front might actively increase their call rate. Furthermore, meerkatsterm·key are able to distinguish between close callsterm·dom of different individuals [ 14 ] and might therefore take the identity of the callers into account when following a signal hotspot. Whether these potentially more informed individuals can ‘lead’ the group [ 19 , 20 ] by affecting the location of the signal hotspot with an increase in their own call rate, thereby influencing the movement of others, is yet to be explored.

Acknowledgements We thank the Kalahari Research Trust for permission to work at the Kuruman River Reserve, and the neighbouring farmers to use their land. We also thank Tim Clutton-Brock and Dave Gaynor for the organization of the field site and their input on the fieldwork, and the managers and volunteers of the KMP for maintaining the habituation and basic data collection on the meerkatsterm·key. Many thanks to Bruce Boatman for his assistance with the playbackterm·key experiments and Ariana Strandburg-Peshkin for comments on the manuscript.

Ethics All data collection adhered to ASAB guidelines. All research was conducted under the permission of the ethical committee of Pretoria University (permit number: EC031-13) and the Northern Cape Conservation Service (FAUNA 192/2014), South Africa.

Data accessibility Data are available via Dryad ( http://dx.doi.org/10.5061/dryad.qb562 ) [ 22 ].

Authors' contributions G.E.C.G. and M.B.M. designed the experiment and data collection; G.E.C.G. carried out the experiment, collected and analysed the data. Both authors wrote the article and gave final approval for publication.

Competing interests We declare we have no competing interests.

Funding This study was funded by the Swiss National Science Foundation (grant no. PDFMP3_141768 to M.B.M.). This paper has relied on records of individual identities and/or life histories maintained by the Kalahari Meerkatterm·key Project (KMP), which has been supported by the European Research Council (grant no. 294494 to T.H. Clutton-Brock since 1 July 2012), the University of Zurich and the Mammal Research Institute at the University of Pretoria.

References Associated Data Data Citations Data Availability Statement Data are available via Dryad ( http://dx.doi.org/10.5061/dryad.qb562 ) [ 22 ].