Introduction

Acoustic signals play an important role in social interactions among mammals [1–4] and birds [5–6]. In mammals, vocalisations seem to contribute to social interactions such as partner and offspring recognition [7–11], to promote group cohesion [1,12–14], and to facilitate social interactions [15–18]. In some species, calls may play more than one functional purpose and encode information on sex, age and social status. For example, the hyena’s (Crocuta crocuta) giggle, often referred as the hyena’s laugh, encodes information about the animal’s identity, social status and age [17]. After hearing this call from one individual, hyenas’ clan-mates can decide whether to join the giggler or not depending on caller identity [17]. In white rhinoceros (Ceratotherium cottoni), the panting call is used as a greeting and for sex and age recognition [18–20]. During territorial encounters, rhino males are more likely to react more strongly to panting from a subordinate male, possibly to confirm its social status [21].

Grunt-like calls are present in the vocal repertoire of many group-living mammals and seem to facilitate social interaction, especially between lower and higher-ranking group members [15], [16], [22], [23]. In chimpanzees (Pan troglodytes), which live in a fission-fusion society with linear hierarchy, for instance, the pant-grunt vocalization is typically a greeting signal [24]. This kind of call is emitted by lower-ranking individuals towards higher-ranking males or females [25]. In addition, pant-grunts were reported to play a role in reconciliation after aggressive encounters [24], [26] or to express fear [26], [27], and as a submissive-signal [22], [27]. In rhesus macaque (Macaca mulatta) and baboons (Papio cynocephalus ursinus), dominant females usually emit quiet grunts to communicate their intent to behave peacefully [15], [23], [28]. In the white-lipped peccary (Tayassu pecari), a group-living mammal, grunts are also present [29] and may be used to facilitate social interactions.

The white-lipped peccary is an ungulate that lives in dense Neotropical forests [29] in mixed-sex and stable herds numbering in the hundreds [30]. These herds are characterised by high cohesiveness, even during seasons of food-shortage [30], [31]. In captivity, the species shows a strict dominance hierarchy in which both males and females can reach the highest dominance rank [32], [33]. Although little is known about that peccary’s vocal repertoire [29], the white-lipped grunt is their most frequent vocalisation; it is a loud, conspicuous, fast-rising and noisy sound emission used for short-distance communication [29], [32], [33], [34] (Fig 1). The grunts are usually emitted following aggressive interactions during feeding time in both captive and free-ranging animals [29], [32], [33], [34]. Because of the context in which this vocalization is used, it has the potential to mediate social conflicts by preventing or mitigating aggression, contributing to the stability of that peccary group.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Grunt call emitted by a white-lipped peccary (Tayassu pecari).

https://doi.org/10.1371/journal.pone.0158665.g001

Our aim, therefore, was to investigate the possible functions of peccary’s grunts and their relationship to an individual’s social rank, identity and sexual dimorphism. If the grunt-like calls are used to mitigate conflicts by signaling the caller’s social status, as observed in non-human primates [22], we expect that lower-ranking individuals will emit grunts more often than dominant ones to indicate their submissive status. Furthermore, the acoustic structure of calls used to mediate agonistic interactions may provide cues about the sender’s motivational state [35], [36]. For instance, aggressive individuals usually produce low-frequency sounds with a broad frequency band, while frightened submissive individuals produce sounds with higher frequency and narrow frequency band [35]. Together with low-frequency sounds, spectral noisy vocalizations, which reflect patterns of deterministic chaos in laryngeal vibration [37], can also provide cues about the individuals’ dominance rank [38] and about the severity of excitement or aggression [36], [39]. Thus, in our study we also expect higher-ranking individuals to demonstrate willingness to counter-attack in a stronger manner than lower-ranking individuals, using longer, noisier and lower-frequency calls [35], [36]. In addition, we do not expect to find individual differences in grunts, because individual recognition is not necessary for conflict avoidance [16], [22]. Finally, as females seem to be valuable in peccary society, as would be suggested by the fact that they usually occupy higher ranks after farrowing (giving birth to a litter) [32], we would expect to find acoustic differences in grunts between sexes to allow female recognition, given that there are no obvious visual cues that permit sexual identification.

Methods

Study site and animals

We conducted an observational study with 19 adult white-lipped peccaries (9 males, 10 females) at the Applied Ethology Laboratory of the Universidade Estadual de Santa Cruz, Bahia, Brazil (14°47’39.8”S, 39°10’27.7”W). The animals (Table 1) were all adults (M_weight = 39.3 SD = 6.0) from the same family herd (N = 19) strongly related, born and raised in captivity. Because the animals came from another captive breeding center, we could not determine their exact age (range: 3–8 years old). The animals were identified with plastic ear tags of different shapes and housed in a 940m² paddock surrounded by a 1.5 m high wire fence. The paddock had a dirt floor with the presence of medium-sized and tall trees and bushes, which provided natural shade. Food was provided twice a day (at 8:30 am and 4:00 pm) and water was available ad libitum in a water trough.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Mean values (± standard deviation) of the acoustic parameters for grunts emitted by white-lipped peccaries according to the ranking position.

Lower numbers correspond to higher positions in the dominance hierarchy. All measurements were taken from 50 samples from each individual. The percentage of correct cross-classified grunt calls for each individual based on the conventional (DFA) and permuted discriminant analysis (pDFA) is also presented, based on 37 calls for each individual.

https://doi.org/10.1371/journal.pone.0158665.t001

Results

From the 1,880 agonistic interactions observed among all dyads of the group, we obtained a linear dominance rank order (linearity index h’ = 0.9; P < 0.0001) Table 1), which included both males and females without circular relationships. Moreover, the directional consistency index (DCI = 0.9) showed an almost complete unidirectionality of submissive gestures.

The dominant male was the only individual that did not emit grunts (Table 1). The lower the animal rank, the higher the number of grunts emitted (r_s = 0.54, N = 19, P < 0.01). Moreover, while the mean minimum frequency of grunts decreased as the peccary’s rank increased (r_s = 0.52, N = 18, P < 0.05), the other two acoustic parameters were not correlated with animal rank (mean HNR: r_s = -0.24, N = 18, P = 0.34; mean bandwidth: r_s = 0.24, N = 18, P = 0.33). We verified that grunt production often resulted in the end of conflict (N = 400) instead of aggression continuation (N = 13) (Chi square = 362.64, df = 1, P < 0.0001). Moreover, the end of conflict in response to a grunt occurred on average (±SD) of 1.3±0.6s after the grunt (N = 400), while subsequent aggression occurred on average (±SD) of 6.8±5.1s following the grunt (N = 13).

Grunts were emitted as solo units (elements) or sequences of 2 to 20 elements (mode = 3, N of sequences = 90), with a mean duration of 1.63 ± 1.34 sec (mean ± SD, N of sequences = 90). The conventional DFA differentiated the grunts among individuals (Wilks Lambda = 0.52, P < 0.0001, N = 666). The first two discriminant functions (DF) explained 61% of the variation in the acoustic structure of individuals’ grunts (Table 2). The pulse duration (s) showed the highest loading for DF1, while the peak frequency centre (Hz) showed the highest loading for DF2 (Table 2). The greatest source of signal discrimination came from the peak frequency, which showed the highest load of all variables. Cross-validation analysis based on conventional DFA correctly assigned the grunts to each sender with a mean accuracy of 18.5%, which was higher than that expected by chance (5.6%; Binomial test: P < 0.001). The permuted DFA (pDFA) confirmed that individuals vary in the acoustic structure of their grunts (P < 0.001). Cross-validation analysis correctly assigned the grunts to each sender with a mean accuracy of 16% (based on 100 random selections of the original data). Although the percentage of correct classification was very low, it was greater than the 6% expected by chance (based on 1000 permutations of the original data set). The cross-validation revealed a great variation in the percentage of correct classification across callers, ranging from 0% to 49% (Table 1; Fig 2). The pDFA revealed that there was no difference between the acoustic features of male and female grunts. Although the cross-validation correctly assigned the grunts to males and females with a mean accuracy of 53.72% (based on 100 random selections of the original data), the original data set does not classify better than the permuted data (average percentage of correct classifications for permuted data set = 53.36; P = 0.50).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Distribution of discriminant scores for each individual along the two canonical discriminant functions.

https://doi.org/10.1371/journal.pone.0158665.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Coefficients of the two main discriminant functions (DF) indicating the relative contribution of each acoustic parameter (variable) for discriminating between individuals’ grunt calls.

The values in bold indicate the parameters with higher loading for the discriminant functions (DF1 and DF2).

https://doi.org/10.1371/journal.pone.0158665.t002

Discussion

Our study supported our prediction that grunt-like calls play a role in white-lipped peccary social interactions by preventing or mitigating aggression. Low-ranking callers emitted grunts more often than higher-ranking ones, whereas the alpha male never emitted this vocalization. We verified that after hearing grunt emissions, dominant opponents ceased the conflict or at least delayed the continuation of aggression for some seconds, possibly reducing the severity of agonistic interaction. The DFA indicated that grunts were individually distinctive; however, a low accuracy of assigning grunts to individuals was detected. These results seems to point to at least two scenarios: (i) individual recognition is not the main function of grunts, perhaps because it is not necessary for peccary social organization, or (ii) the acoustic parameters used in the present study did not allow acoustic cues to the callers’ identity to be detected. Moreover, contrary to our expectations, there was no sexual dimorphism in the acoustic parameters of grunts in this species.

Top-ranking white-lipped peccaries emitted grunts with a lower mean minimum frequency. The higher pitched grunts of subordinate peccaries may be influenced by social experience, and may reflect affective state and levels of excitement or aggression [35], [36]. The observed correlation between individuals’ rank and mean minimum frequency among animals could also be explained by individuals’ age and consequent weight, instead of social rank. The animals' body size is likely to affect the spectral amplitude distribution, such as peak frequency, which may have lower values in larger animals [63]. In southern white rhinos, for instance, pants showed differences between sex, depending on the age class and social status of the signaller [21]. As all peccaries in this study were adults and their weight was similar (Table 1), we did not expect to find any influence of age on individuals’ rank and minimum frequency. Here the correlation between animals’ rank and mean minimum frequency may be more reasonably explained by the affective state of the individuals [35]. Mathevon et al. [17], studying Crocuta crocuta, in accordance with Morton rule [35], report that aggressive individuals usually produce lower frequency sounds with a broad frequency band, while frightened submissive individuals produce sounds of higher frequency and a narrow frequency band.

Once the hierarchy is established, the alpha male is not threatened by other group members [32], [33, 64], and this could explain the fact that the alpha male did not emit grunts during our study. In contrast, lower-ranking peccaries needed to signal their rank in a more conspicuous way, which allowed dominants to modulate their response, avoiding unnecessary levels of aggression. By signalling its submission status, the subordinate peccary can provide a ritualized way to avoid or prevent continuation of aggression. Grunts in peccaries show some similarities with non-human primates. The pant-grunt, for instance, which is primarily considered as greeting signal in chimpanzees [24], is also related to dominance-submission relationships [65], because subordinate individuals usually address pant-grunts to dominant males or females [22], [24], [25]. Sakamaki [22] reported that pant-grunt vocalization in chimpanzees was present in 80% of aggression encounters; similarly, peccaries usually emit frequent grunts in social conflicts during feeding [29], [32], [33]. In chimpanzees, it was also observed that pant-grunts together with agonistic interactions participate to form a single hierarchy of social status allowing males to be ranked in distinct social levels [66]. In savannah baboons, grunts are also related to the social status of individuals involved [16]. This vocalization, however, occurs in a different situation from the peccary’s grunts. While grunts in peccaries are emitted by subordinates, the dominant female in baboons uses a grunt call to the subordinate that then allows the former dominant to approach less aggressively, as a reconciliatory interaction [16]. The authors suggest that the grunt can reduce victims’ anxiety and can also facilitate interaction between adversaries, despite not changing their social relationships [67]. Therefore, our findings suggest that grunts play a role in the social dominance of peccaries, similarly to what occurs in chimpanzees [22], [66]. Moreover, rather than reducing aggression per se, as happens among baboons [16], peccary’s grunts may help to maintain (at least in captivity) the strict dominance hierarchy among group individuals.

The low accuracy of grunt identification suggests that caller recognition in peccaries may be less important than its function as a strong sign of individual social status. In baboons, for instance, the authors suggest that reconciliatory grunts may indicate the low probability of aggression occurrence, and victims possibly learn by experience or observation [16]. Cheney and Seyfarth [16], suggest that in baboons, grunts can alter the recipient’s behaviour and increase the probability that dominant females will be able to interact with victims (subordinate). Thus, our results indicate that the most important function of grunts in peccary society is to indicate submission and refusal to be supplanted by dominants, as observed in baboons. Furthermore, white-lipped peccaries live in herds which number up to 300 individuals [30] and this makes it more difficult to assess any kind of acoustic parameters associated with identity. Bergman [68], comparing individual recognition among males in geladas (Theropithecus gelada) and baboons (Papio ursinus), hypothesized that to track individual social information of many individuals (baboons: 2–15 and geladas: over 30 in band and 100 in herd) is cognitively challenging for geladas, which we suggest is also true for peccaries.

Our prediction of sexual differences in grunt vocalization has to be rejected, in line with previous studies, which did not find any dimorphic differences between males and females of white-lipped peccary and suggest that peccary groups compete equally for limited resources and establish a unique social dominance hierarchy [32], [33]. In spotted hyenas (Crocuta crocuta), for instance, the giggle calls do not show reliable differences between male and females [17]. Therefore, other acoustic parameters such as formant-like spectral features (vocal tract resonances) may provide information on sex differences and animals’ identity. However, in the present study we could not measure the peccary’s formants due to the lack of appropriate knowledge about peccaries’ vocal tract morphology. In addition, chemical communication could also support some sexual recognition in white-lipped peccary [29], but further studies need to be carried out to answer the dimorphism question as reported for other species [69], [70].

Our findings are particularly important to improve the current understanding of the role of grunt-like calls in herd-living mammals with linear dominant hierarchies, which strongly suggests that they are involved in the maintenance of herd social stability and cohesion. Although further studies are required to improve our understanding of the acoustic function and structure of grunts and other white-lipped peccary vocalizations, this work is the first step in improving our knowledge of their acoustic repertoire.

Supporting Information

Acknowledgments

We are grateful to Plos One’s Editor (Elisabetta Palagi) and anonymous reviewers for all contributions to the manuscript. We thank Roger Mundry for kindly providing his pDFA R script and instructions to conduct this analysis. We thank Michele D. Singh (University of West Indies), Dr Fátima Felisberti (Kingston University) and Dr. Michael Mendl (University of Bristol) for their kind review of the English manuscript. We would also like to thank Dr. Tokumaru, R.S. from the Department of Developmental Psychology of the Federal University of Espirito Santo for suggestions concerning the manuscript. We also thank all the staff of the Laboratory of Applied Ethology at the State University of Santa Cruz. This work was supported by the Coordination for the Improvement of Higher Education Personnel (CAPES), UESC and the National Council for Scientific and Technological Development (CNPq) under process no. 300587/2009-0 and 306154/2010-2, respectively.