Limited geographic variation in the acoustic structure of and responses to adult male alarm barks of African green monkeys
- 1k Downloads
The global diversity of human languages is a remarkable feature of our species, which requires a capacity for rapid vocal learning. Given that primate alarm calling systems have played an important role in the language origin debate, identifying geographic variation in primate alarm calls and understanding the underlying causal mechanisms are important steps to help uncover evolutionary precursors to language. This study investigates geographic variation in the alarm bark of the widely distributed African green monkey (Chlorocebus). To quantify geographic variation in spectral and temporal call structure, acoustic analysis was used to compare the adult male barks of green monkeys (Chlorocebus sabaeus) and two subspecies of vervet (Chlorocebus pygerythrus pygerythrus and Chlorocebus pygerythrus hilgerti). Playback experiments were also carried out to test whether adult male vervets would distinguish between the barks of own-group males, unknown conspecific males and green monkey males. Acoustic analysis showed that, whilst similar in overall structure, the barks of green monkeys could be distinguished from vervet barks with a high degree of accuracy; the barks of vervet subspecies could also be discriminated, although to a lesser degree. Males responded most strongly to unknown conspecific males’ barks, and exhibited responses typical of leopard-avoidance and territorial defence. Taken together, these findings indicate that variation in alarm calls can be best explained by phylogenetic distance, and that intra- and inter-species differences are relevant during social interactions. Moreover, barks may function as an alarm and display call, which could explain the observed sexual dimorphism in barks in this genus.
KeywordsAlarm call Chlorocebus Geographic variation Language evolution Vervet monkey Vocal communication
Human speech displays extensive regional differences in the language spoken, dialect and accent, and it has been proposed that the emergence of vocal learning was a critical step during the evolution of the modern language faculty (Janik and Slater 1997; Oller and Griebel 2008). Vocal communication in nonhuman animals also exhibits geographic variation (Weilgart and Whitehead 1997; Slobodchikoff et al. 1998; Bradbury et al. 2001; Davidson and Wilkinson 2002; Smith and Hunter 2005; Delgado 2007), but whilst speech patterns are strongly influenced by learning, the ability to produce novel vocalisations as a result of experience has been identified in only a few other distantly related taxonomic groups (Janik and Slater 1997). In nonhuman primates (hereafter primates), vocal learning is notably scarce and may be limited to the modification of existing vocalisations as a result of social interaction (reviewed in Egnor and Hauser 2004; Hammerschmidt and Fischer 2008). Such vocal modification is often inferred when population differences in vocal structure cannot be explained by genetic or habitat structure (Fischer et al. 1998; Crockford et al. 2004). Studies investigating the causal mechanisms underlying vocal variation in primates are therefore a first step towards a better understanding of the factors that drive acoustic divergence, and ultimately vocal learning.
Between-species variation in primate loud-call structure is generally attributed to genetic differences (Oates and Trocco 1983; Brockelman and Schilling 1984; Méndez-Cárdenas et al. 2008; Wich et al. 2008; Thinh et al. 2011; Meyer et al. 2012), and species-specific differences in such calls can be used as a non-invasive tool for discriminating between cryptic species (Zimmermann et al. 2000). Despite the important role that primate alarm calls have played in the search for the roots of human language, and having been identified as a flexible, context and audience-dependent behaviour requiring sophisticated cognitive processes (Zuberbühler 2007), far less is known about geographic variation in primate alarm calls. In a recent review, Wilkins and colleagues (2013) introduced a framework for the comparative study of acoustic divergence across a broad range of taxa. Within this review, they identified three scenarios under which acoustic characteristics may change in relation to (neutral) genetic distance. Firstly, acoustic and genetic distance may covary as a result of genetic drift; secondly, acoustic characteristics may diverge more rapidly due to sexual selection in association with ecological differentiation or a genetic mutation related to signal production and species recognition, and finally acoustic signals may diverge slowly at first and then rapidly increase in variation if the costs of hybridisation drive reproductive character displacement. Clearly, broad comparative acoustic analyses including phylogenetic reconstruction are needed to identify the evolutionary processes giving rise to observed acoustic variation. Quantifying the geographic variation present in alarm calls within and between closely related primate species, and investigating how vocal variation affects receiver responses will contribute to a clearer understanding of the dynamics of primate vocal evolution.
Despite these morphological and genetic differences, vocal comparisons amongst African green monkey populations were, until now, limited to qualitative descriptions. These observations suggested that vocalisations are spectrally stable among vervet subspecies and between vervet and tantalus (Chlorocebus tantalus) populations, but that they may exhibit differences in temporal characteristics (Struhsaker 1970). A quantitative analysis of intra- and inter-species vocal differences and how African green monkeys respond to the vocalisations of other populations was, however, lacking. This is somewhat surprising, as the alarm calls of East African vervets constitute one of the most well-known examples of so-called functionally referential signals (Seyfarth et al. 1980), and studies of geographic variation in this genus may offer insights into the flexibility present in their vocal behaviour, and thus shed light on the mechanisms that give rise to context-specific calling.
In this study, we carried out structural analyses of bark calls produced by adult male South African vervets (C. pygerythrus pygerythrus), adult male East African vervets (C. pygerythrus hilgerti) and adult male green monkeys (C. sabaeus) to investigate vocal variation between and within species, including the variation between individuals within the South African population. Following this, we tested how South African adult male vervets perceive the barks of own-group males and unknown males of the same subspecies, and unknown male green monkeys. The adult males of all three populations have been reported to produce barks in response to terrestrial predators, and playback experiments have shown that East African vervets frequently respond to conspecifics’ alarm barks by climbing a tree, an appropriate leopard-avoidance behaviour (Seyfarth et al. 1980). Similar leopard-avoidance behaviour was observed in green monkeys, with subjects always climbing more than 2 m into a tree in responses to a leopard model (Price and Fischer 2013). Thus, if South African vervets recognise intra- and inter-specific bark calls as indicating the presence of a terrestrial predator, they should respond by climbing to a height of more than 2 m into a tree. However, acoustically similar barks are also given during aggressive encounters (Struhsaker 1967; Galat and Galat-Luong 1976; Cheney and Seyfarth 1992), which frequently entail males chasing after another male (TP personal observation).
The aim of this study was thus to identify the degree of variability present with the structure of bark calls, to investigate how this variability is perceived, and to offer insights into the function of adult male barks. Given the limited evidence for vocal learning in nonhuman primates (Hammerschmidt and Fischer 2008), we predicted to find little variation between and no variation within species. Considering the presumed stabilising selection acting on alarm calls (Struhsaker 1970), and the costs of not responding to a putative alarm call, we expected calls to cause males to climb into trees, and furthermore we expected relatively little variation in listeners’ responses to the playback of calls with different origins.
Geographic and individual differences in call structure
Adult male bark vocalisations were recorded from green monkeys, East African vervets and South African vervets at three geographically distant study sites within the range of African green monkeys (Fig. 1). All study subjects were habituated to human presence and were recognised individually from natural markings on the face and body. The barks of East African vervets were recorded by Thomas Struhsaker (June 1963–May 1964), and Robert Seyfarth and Dorothy Cheney (1977–1988) as part of their earlier studies on several free-ranging groups within the semi-arid acacia savannah of Amboseli National Park (2°39′49 S; 37°15′16 E) in Kenya, and these calls were subsequently made available for inclusion within this study. Green monkey barks were recorded by TP over two field seasons (January–June 2010 and 2011) from four free-ranging groups and two solitary males within Niokolo-Koba National Park (13°01′34″ N, 13°17′41″ W), an area in southeastern Senegal consisting mainly of Sudano-Guinnean savannah interspersed with woodland and gallery forest (Frederiksen and Lawesson 1992). South African vervet barks were recorded by TP and ON (January–June 2012) from five free-ranging groups located within the Loskop Dam Nature Reserve (25°25′18S; 29°18′29E) in South Africa, which contains a mixture of open grassland, acacia dominated woodland and low mountains with open tree savanna (Filmalter 2010).
In all studies, adult male barks were recorded ad libitum when the context of calling could be confirmed as the presence of a feline terrestrial predator, either by observing the predator or hearing its vocalisations. Whilst the natural occurrence of bark calls was not uncommon, it was often not possible to confirm whether a terrestrial predator was present at these times. Following numerous studies that have successfully used the presentation of predator models to elicit alarm calling (e.g. Coss et al. 2007; Arnold et al. 2008; Wheeler 2010), spontaneous barks from all field sites were supplemented with barks produced in response to leopard models. Altogether, five different leopard models were used. East African vervet calls were recorded at a distance of 0.5 to 7 m using a Nagra III or a Nagra SNN tape recorder at 9.5 cm/s and MKH 804 Sennheiser directional microphones (R. Seyfarth and D. Cheney) or an Uher 4000 Report-S portable tape recorder at 7 ½ ips with a frequency response of about 40–20,000 cps (manufacturer’s specifications), parabolic reflector and a Shure dynamic microphone (T. Struhsaker; no information on distance available for these calls). These calls were later digitised at 22.05 or 44.1 kHz with a 16-bit resolution. South African vervet and green monkey calls were recorded by TP at a distance of 3 to 10 m using a digital Marantz PMD661 solid-state recorder (44.1 kHz sampling rate, 16bits accuracy) connected to a Sennheiser ME66/K6 directional microphone.
Bark vocalisations were frequently produced in long calling bouts. Consecutive bark elements (the basic units represented by a continuous sound) were identified using the pulse-train analysis of Avisoft SASLab Pro (version 5.1.17) and the start and end points of each element were recorded. From these labels, we calculated inter-unit intervals and used a log survivor function (Slater and Lester 1982) to determine a time threshold of 75 ms, below which elements were classified as belonging to the same call. On the basis of this, bark calls can be made up of one or more bark elements, with inter-call intervals exceeding all intra-call intervals and multi-unit barks frequently containing first exhalation (Ex1), within-call exhalation (Ex2) and inhalation (Inhal) call elements (definition of terms modified from Struhsaker 1967). It was not possible to analyse all calls from a calling bout, as males were often too far away from the microphone when they started calling and/or many calls were overlapped by the calls of conspecifics. As such, call samples were taken from as near as possible to the start of each calling bout, but had to be tailored to the number and quality of calls recorded.
To investigate population-level differences in adult male bark vocalisations, we analysed barks from 12 green monkeys, 12 South African vervets and 13 East African vervets. Unless insufficient calls of adequate quality were available, 10 bark calls were selected from the calling bout of each individual, resulting in a total call sample of 352 bark calls (120 green monkey barks, 120 South African vervet barks and 112 East African vervet barks). To investigate individual-level differences, we analysed the bark vocalisations of six adult male South African vervets. We could not extract uninterrupted whole calls from all bouts of each individual, and as such we were not able to test temporal differences at the level of the whole call. To identify differences in element structure, we selected 20 Ex1 bark elements from five calling bouts per male, resulting in a call sample of 100 bark elements per male and a total of 600 bark elements.
Description of the acoustic parameters used to describe the temporal and spectral structure of barks
Duration of call
Ex1 duration (ms)
Duration of single or first call element
The number of elements per call
Mean fundamental frequency across all time segments
F0 start (Hz)
Fundamental frequency at the start of the call unit
F0 end (Hz)
Fundamental frequency at the end of the call unit
F0 linear trend
Factor of linear trend of fundamental frequency
Percentage of tonal time segments for which F0 can be calculated
Median first frequency quartile across all time segments
First quartile_1–4 (Hz)
Mean first frequency quartile at 1st, 2nd, 3rd and 4th temporal quartiles
Median second frequency quartile across all time segments
Second quartile_1–4 (Hz)
Mean second frequency quartile at 1st, 2nd, 3rd and 4th temporal quartiles
Median third frequency quartile across all time segments
Third quartile_1–4 (Hz)
Mean third frequency quartile at 1st, 2nd, 3rd and 4th temporal quartiles
Frequency range (Hz)
Mean frequency range
Peak frequency (Hz)
Median peak frequency across all time segments
Peak frequency_1–4 (Hz)
Mean peak frequency at 1st, 2nd, 3rd and 4th temporal quartiles
PF linear trend
Factor of linear trend of peak frequency
PF deviation (Hz)
Mean deviation between peak frequency and linear trend
To assess population and individual differences in male bark calls, we first applied a stepwise method to identify a subset of optimum variables for each classification. We set population or caller identity as the grouping variable and entered all temporal and spectral parameters into a stepwise variable selection using the stepclass function of the R package “klaR” (Weihs et al. 2005) with leave-one-out cross-validation. To assess the degree to which barks could be correctly assigned and to determine which structural properties contributed most to differentiating between the different populations or callers, we then entered the selected variables into a linear discriminant analysis (LDA) using the lda function of the R package “MASS” (Venables and Ripley 2002) with a jack-knife leave-one-out method. We compared the classification results of the LDA to those of a nested permuted discriminant function analyses (pDFA; Mundry and Sommer 2007) to control for the pseudo-replication introduced by using multiple calls from a single calling bout. All statistical analyses were carried out using R (R Development Core Team 2011), and we ensured that assumptions were met (see S1-S2 in Online Resource 1) before running tests.
Percentage of calls assigned to each population and descriptive statistics (mean ± SD) of acoustic parameters used for classification
Ex1 duration duration
262 ± 110
254 ± 40
−0.14 ± 0.1
2197 ± 708
128 ± 68
99 ± 16
282 ± 57
−0.11 ± 0.2
3116 ± 831
101 ± 53
113 ± 19
320 ± 61
−0.10 ± 0.2
3565 ± 645
156 ± 115
Stepwise variable selection carried out to look at individual differences in the acoustic structure of Ex1 call elements identified five acoustic parameters to enter into a subsequent LDA. These parameters were F0, first quartile_1, first quartile_2, first quartile_4 and Ex1 duration. Entering these parameters into LDA (using a leave-one-out method) resulted in 70 % correct assignment (range = 59–94 %) of South African vervet barks, very similar to the 71 % correctly assigned with pDFA. The first discriminant function described 74 % of the total variance explained and was most influenced by F0. The second discriminant function described 17 % of the total variance explained and was most influenced by measures of the first quartile energy band.
Behavioural responses to bark vocalisations
Playback experiments were conducted with South African vervets between January and June 2012 by TP and ON. Study subjects were 11 habituated and individually recognised adult male South African vervets from four free-ranging groups within Loskop Dam Nature Reserve in South Africa. For each experiment, we played back bark calls elicited by a leopard model. These barks originated from own-group adult males (“South-own”), unknown adult males of the same subspecies (“South-unknown”) and adult male green monkeys (“West”). Barks used as stimuli for the West and South-own playback conditions were selected from recordings of green monkeys and South African vervets that were used in the structural analysis of bark calls. Barks used as stimuli for the South-unknown playback condition were recorded by Daniel van der Post from free-ranging groups in the Okavango Delta (18°25′42 S; 21°53′39 E) and Limpopo (22°54′25 S; 28°2′28 E) both in Botwana, and from Lajuma Research Centre (23°2′21S; 29°26′58E) in South Africa. Playback stimuli were made up of 6–12 barks units with a high signal-to-noise ratio that were produced as single and double unit exhalations sometimes interspersed with inhalation units. All bark units were taken from a single calling bout, with inter-call intervals, call compositions and maximum amplitude held constant between conditions. Reflecting temporal differences identified in the above section of this study, however, mean duration of call elements was longer in West playback stimuli than in South-unknown and South-own stimuli.
Barks were played back to male vervets using a within-subjects design such that, with one exception, each of the eleven subjects experienced one playback of each condition. This exception was due to a male migration which left one study subject as the sole male of his group before a final experiment could be carried out. We therefore conducted a total of 32 playback experiments (11 West, 11 South-unknown and 10 South-own). The order of playback trials was balanced across conditions, and to avoid habituation effects, playback experiments were carried out on one group with a minimum separation of 6 days. To avoid pseudo-replication, a different playback sequence was used for each playback experiment. As much as possible, each call sequence was produced by a different individual, and no individual contributed calls to more than two playback stimuli.
Playbacks were initiated when the study subject was sitting resting on the ground or low down (<2 m) in a tree, and when the caller (in south-own condition) or another adult male (south-unknown and west conditions) was out of sight. This was to control for the subject responding to observed male presence rather than to identity cues present in the broadcast call. Playbacks were not carried out within 60 min of the natural production of bark calls or any other alarm calling within earshot of the experimenter. In addition, since we were interested in the propensity of the subject to climb up into a tree, it was necessary that at the time of the experiment, there was a tree of >2 m in close vicinity to the subject. Prior to an experiment, the loudspeaker was hung using a net bag behind a natural obstacle at 1–2 m from the ground and 42–59 m from the study subject. Playback stimuli were broadcast using a Marantz PMD-661 solid-state recorder connected to a loudspeaker (David Active, VISONIK, Berlin), with maximum amplitude set within the range of natural calling behaviour (66–79 dB at 10 m from source, measured using a Voltcraft 322 sound level meter). Experiments were discarded if the subject moved prior to stimulus presentation (n = 1) if there were technical problems with the equipment (n = 2), if the individual was lost before the end of the experiment (n = 1), or if the subject’s behaviour was altered by the presence of human food (n = 1).
Following each playback, a single subject was filmed for at least 30 s using a Sony Handycam (DCR-HC90E). At the end of these 30 s, we recorded whether the subject had climbed more than 2 m up into a tree, and, using a tape measure, the maximum horizontal distance travelled. Thirty seconds was selected because it was the maximum time period that all subjects were visible on film, importantly all subjects that responded by climbing up into a tree did so well within this time frame (maximum 8 s after the playback stimulus). To assess behavioural effects over a longer time period, subjects were followed for 30 min (from the point at which the playback was broadcast), and at each 5-min interval, we recorded their height (as being more or less than 2 m from the ground), and their position using a handheld GPS (Garmin GPSMAP 60CSx). We additionally took GPS points of the position of the loudspeaker, the subject’s initial position at time of playback, and of the subject’s position 3 min after the playback experiment.
Description of the behavioural measures used to describe subjects’ responses to playback experiments
Strength of response:
First orientation (s)
Duration of first orientation towards loud speaker
Initial displacement (m)
Maximum distance travelled within 30 s of experiment
Does subject climb to >2 m within 30 s of experiment
Is subject >2 m high within the 30 min following experiment
Male-Male competitive response
Initial approach (m)
Distance approached towards loudspeaker 3 min post-experiment
Minimum approach (m)
Minimum distance to loudspeaker within 30 min of experiment
To test whether bark origin would have an effect on the strength of response, we used a general linear mixed model (GLMM) with Gaussian error structure to assess differences in the duration of subjects’ first orientation towards the playback speaker, and a GLMM with Poisson error structure to assess differences in initial displacement. We ran GLMMs with binomial error structure to test the effect of bark origin on leopard-typical response behaviours, more specifically looking at differences in whether the male climbed immediately up into a tree (arboreal escape) and whether the male spent more time up in a tree over the next 30 min (time arboreal). Lastly, to test whether bark origin would have an effect on the male-male aggressive behaviours, we used GLMM with Gaussian error structure to analyse differences in the distance approached towards the speaker after 3 min (initial approach) and the minimum distance to loudspeaker over the next 30 min (minimum approach).
For all GLMMs, playback condition (west, south-own, south-unknown) and sequence order were entered as fixed effects and subject was entered as a random effect; to test for differences in subjects’ height over the 30-min period, experiment was added as an additional random effect to account for interval data being included as separate data points. Maximum likelihood was used to achieve more reliable P values for models run using a Gaussian error structure. Likelihood ratio tests were used to compare full models with null models (comprising only of sequence order and the random effect), and Markov Chain Monte Carlo sampling using the functions pvals.fnc and aovlmer.fnc of the R package languageR (Baayen 2011) was applied to calculate P values for the different levels for behaviours that differed between playback conditions. All models were fitted in R using the function lmer of the R package lme4. For details of test assumptions, see S3 in Online Resource 1.
Occurrence of orientation, initial displacement and initial approach behaviours
Orientation towards speaker
Description of response behaviours measured as mean ± SD or occurrence, and results from general linear mixed models for the six behavioural variables measured during playback experiments
First orientation (s)
2.1 ± 2.0
5.3 ± 2.9
1.8 ± 1.6
Initial displacement (m)
1.0 ± 1.6
3.8 ± 7.1
1.9 ± 4.2
Time points arboreal
Initial approach (m)*
5.1 ± 10.4
-2.5 ± 15.0
0.5 ± 11.9
Minimum approach (m)
53 ± 18
29 ± 15
42 ± 22
Population differences in call structure
Despite their overall acoustic similarity, the bark calls of West African green monkeys, East African and South African vervets could be distinguished on the basis of spectral and temporal differences in call structure. Of the three populations, correct classification at the species level was high (96 % of calls). The calls of vervet subspecies were also distinguishable, although to a lesser degree, being correctly classified in 77 % of cases, indicating that call differences were larger between than within species. This finding is not suggestive of vocal learning, which leads to rapid changes in vocal structure independent of genetic differences between populations. In addition, vocal differences are unlikely to have arisen as a result of adaptation to the local acoustic niche, because all three populations inhabit mixed savannah woodland, with larger variation in habitat type within than between sites. Instead, results are in accordance with phylogenetic data which identifies green monkeys as especially distinct from other Chlorocebus taxa (Haus et al. 2013) with an estimated divergence time of 2.76–2.81 MYA, compared to 1.59–1.48 MYA divergence of vervet subspecies (Wertheim and Worobey 2007). Thus this study indicates that, unlike geographic variation in patterns of human speech, acoustic divergence in the structure of the African green monkey bark is likely the result of phylogenetic distance rather than learning processes. A similar correspondence between acoustic divergence and phylogenetic distance was found for the long calls of leaf monkeys (Presbytis spp., Meyer et al. 2012) as well as the song of crested gibbons (Nomascus spp.,Thinh et al. 2011).
In contrast, the alarm calls of western grey and eastern rufous mouse lemurs (Microcebus murinus and M. rufus), despite diverging 6.7–7.3 MYA (Weisrock et al. 2012), do not differ (Zimmermann et al. 2000). One explanation for these differences could be that African green monkey barks, similar to the loud calls of leaf monkeys and crested gibbons, also play a role outside of alarm contexts (Struhsaker 1967), and they may therefore be less subjected to stabilising selection than the more stimulus-specific mouse lemur alarms.
With regards to how African green monkey barks differ, this study identified the duration of Ex1 bark elements as the most influential parameter distinguishing between green monkey and vervet barks, with green monkeys producing longer, but also lower frequency, call elements than vervets. East African vervets on the other hand produced barks with a lower fundamental frequency and a larger frequency range than their South African sister taxon. The production of calls with lower frequency bands and lower F0 could well be the result of a larger body size (and corresponding larger vocal anatomy), a correlation found to hold across a wide range of primate species (Hauser 1993). Support for this explanation comes from a comparative study of cranial measurements across the genus which identifies green monkey samples as being larger than those of vervets (Elton et al. 2010). Interestingly, patterns of skull size within the genus map to clinal variation in rainfall, and it has been hypothesised that larger body size is influenced by differences in habitat productivity (Cardini et al. 2007). Size differences between South and East African vervets are, however, unclear (Pasternak et al. 2013). The prolonged bark produced by male green monkeys could also be the result of a larger body size in relation to lung capacity (Fitch and Hauser 2002), but it seems unlikely that this could account for such large differences in call duration. If, as suggested above, male barks function as a sexual display as well as an alarm call, the longer bark duration could be a sexually selected trait.
Behavioural responses to bark vocalisations
The duration of the subject’s first orientation towards the speaker was significantly longer in response to unknown conspecific calls than to known conspecific calls, strongly suggesting that adult male vervets differentiate between the calls of known and unknown males and are more attentive to the presence of “strangers.” That males differentiate between these call categories is supported by the earlier finding of this study that male barks exhibit individual differences in call structure. This result does not, however, require that males recognise group members individually (i.e. “true” individual recognition; Tibbetts and Dale 2007). That males are more attentive to unknown calls is not surprising, as vervet monkeys live in relatively stable multi-male multi-female groups (Fedigan and Fedigan 1988) and actively defend their territory against intruding males (Struhsaker 1967). Additionally, subjects’ longer orientation to unknown conspecific barks than to unknown inter-specific barks suggests that inter-specific differences in this call are relevant in inter-group relationships. We predicted that subjects should respond to all barks with the leopard-typical response of climbing up into a tree. We found that subjects climbed into a tree on hearing the playback stimulus in 8/32 experiments only, with no effect of playback condition. Thus, our results show that conspecific and inter-specific barks elicit leopard-typical escape responses, demonstrating some degree of the predicted constraint that receivers should be under to respond appropriately to alarm calls. Nevertheless, the finding that arboreal escape responses were elicited in only 25 % of trials suggests that, at least for adult males, contextual cues such as conspecifics’ behaviour at the time of the experiment or the subject’s prior experience may also play a role in the attribution of meaning to perceived barks. Interestingly, on hearing the playback stimulus, males also approached the playback speaker in 25 % of trials, suggesting that barks may also function as an intra-sexual display and elicit territorial defence behaviours. In addition, males were more likely to move nearer towards the playback speaker in the 30 min following experiments after hearing the calls of an unknown conspecific than after hearing the calls of a known conspecific or hetero-specific male, corroborating the assumption that males are able to differentiate between the calls of own-group and unknown males, and conspecific and hetero-specific males, and are more attentive to the calls of unknown males of the same species.
Subtle variation in the structure of the African green monkey bark can be used to differentiate between species and to a lesser extent subspecies. This strengthens the general consensus that phylogenetic differences account for a large degree of the vocal variability found within and between primate species, not only in loud calls but also in calls produced in alarm contexts. Future studies comparing geographic variation in call types that are either subject to sexual selection or not, will provide insights into the evolutionary dynamics of acoustic divergence in primates. From the perspective of bark perception, this study suggests that adult males differentiate between the barks of own-group and unknown males, and respond more strongly to unknown conspecific than unknown heterospecific barks. Thus within and between-species variation in bark structure appears to be important in regulating inter-group relationships, or relationships among males. That males respond to barks both by climbing into the trees and by approaching the speaker suggests that adult male barks may have a dual function; a terrestrial alarm call and a male display. More generally, this study supports the hypothesis that, in contrast to human speech, flexibility in primate vocalisations is shaped by call function.
We thank Mpumalanga Parks Board for permission to work at Loskop Dam Nature Reserve, and the Diréction des Parcs Nationaux and Ministère de l'Environnement et de la Protéction de la Nature de la République du Sénégal for permission to work in Niokolo-Koba National Park. We are also grateful to Robert Seyfarth, Dorothy Cheney, Tom Struhsaker and Daniel van der Post for kindly providing vervet calls for inclusion within this study; to Jannie and Elizma Coetzee, Tersia de Beer and Alan Barrett for their logistical support while working in South Africa; and to Jacky Bassene, Cheikh Sane, Samba Ciss, Nele Fröse and Matthias Klapproth for their assistance with data collection in Senegal. Thanks also go to Mathieu Cantat for his assistance with video analysis, to Christof Neumann for statistical advice and to Tanja Haus and Gisela Kopp for their comments on the manuscript. The study was supported by funding from the Deutscher Akademischer Austauschdienst, The Leakey Foundation and The Wenner-Gren Foundation.
The experiments described in this paper comply with the current laws of Senegal, South Africa and Germany and were conducted with the agreement of the Mpumalanga Parks Board, the Diréction des Parcs Nationaux and the Ministère de l'Environnement et de la Protéction de la Nature de la République du Sénégal (permit numbers: 0383/24/03/2009; 0373/10/3/2012).
Conflict of interest
The authors declare that they have no conflict of interest.
- Baayen RH (2011) languageR: data sets and functions with “Analyzing Linguistic Data: A practical introduction to statistics”, http://CRAN.R-project.org/package=languageR
- Bradbury JW, Cortopassi KA, Clemmons JR (2001) Geographical variation in the contact calls of orange fronted parakeets. Auk 118:958Google Scholar
- Cheney DL, Seyfarth RM (1992) How monkeys see the world: inside the mind of another species. University of Chicago Press, ChicagoGoogle Scholar
- Development Core Team R (2011) R: A language and environment for statistical computing. R Foundation for Statistical Computing, ViennaGoogle Scholar
- Fedigan L, Fedigan M (1988) Cercopithecus aethiops: a review of field studies. In: Gautier-Hion A, Bourlière F, Gautier J-P, Kingdon J (eds) A primate radiation: evolutionary biology of the African Guenons. Cambridge University Press, pp 389–411Google Scholar
- Filmalter N (2010) A vegetation classification and management plan for the Hondekraal section of the Loskop Dam Nature Reserve. University of South Africa, DissertationGoogle Scholar
- Fitch WT, Hauser MD (2002) Unpacking “honesty”: vertebrate vocal production and the evolution of acoustic signals. Spr Hdb Aud 16:65–137Google Scholar
- Galat G, Galat-Luong A (1976) La colonisation de la mangrove par Cercopithecus aethiops sabaeus au Senegal. Terre Vie 30:3–30Google Scholar
- Groves C (2001) Primate taxonomy. Smithsonian Institution Press, Washington, DCGoogle Scholar
- Groves C (2005) Order primates. In: Wilson DE, Reeder DM (eds) Mammal species of the world. A taxonomic and geographic reference. JHU Press, Baltimore, pp 111–184Google Scholar
- Hammerschmidt K, Fischer J (2008) Constraints in primate vocal production. In: Oller DK, Griebel U (eds) Evolution of communicative flexibility: complexity, creativity, and adaptability in human and animal communication. MIT Press, Cambridge, MA, pp 93–121Google Scholar
- Hill WCO (1966) Primates comparative anatomy and taxonomy VI Catarrhini Cercopithecoidea Cercopithecinae. Edinburgh University Press, Edinburgh, No 3fGoogle Scholar
- Kingdon J (1997) The Kingdon field guide to African mammals. Academic Press, LondonGoogle Scholar
- Lernould J (1988) Classification and geographical distribution of guenons: a review. In: Bourlière F, Gautier JP, Kingdon J (eds) Gautier-Hion A. Cambridge University Press, Cambridge, pp 54–78Google Scholar
- Napier PH (1981) Catalogue of primates in the British Museum (Natural History) and elsewhere in the British Isles. Part II: Family Cercopithecidae, subfamily Cercopithecinae. British Museum (Natural History), LondonGoogle Scholar
- Oller DK, Griebel U (2008) Signal and functional flexibility in the emergence of communication systems: the editors introduction. In: Oller DK, Griebel U (eds) Evolution of communicative flexibility: complexity, creativity, and adaptability in human and animal communication. MIT Press, Cambridge, MA, pp 3–8Google Scholar
- Price T, Fischer J (2013) Meaning attribution in the West African green monkey: influence of call type and context. Anim Cogn 17:277–286Google Scholar
- Slobodchikoff CN, Ackers SH, Van Ert M (1998) Geographic variation in alarm calls of Gunnison’s prairie dogs. J Mammal 79:1265–1272Google Scholar
- Struhsaker TT (1967) Auditory communication among vervet monkeys (Cercopithecus aethiops). In: Altmann S (ed) Social communication among primates. University of Chicago Press, Chicago, pp 281–325Google Scholar
- Struhsaker TT (1970) Phylogenetic implications of some vocalizations of Cercopithecus monkeys. In: Napier JR, Napier PH (eds) Old World monkeys: evolution, systematics, and behavior. Academic Press, New York, pp 365–444Google Scholar
- Wich SA, Schel AM, de Vries H (2008) Geographic variation in Thomas langur (Presbytis thomasi) loud calls. Am J Primatol 70:566–574Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.