Benefits but not the dual functions of submissive signals differ between two Malagasy primates

Abstract

Many animals use formalized signals to communicate dominance relationships. In some primates, such as macaques, the function of such signals varies with dominance style. Despotic species produce unidirectional submission signals that have a dual function: in conflict contexts, they signal a willingness to withdraw, whereas in peaceful contexts, they indicate the agreement to subordination. More despotic species produce these calls to a lesser extent than less despotic species. Here, we investigated whether the use of unidirectional submission signals is also related to dominance style in two lemur species and whether signalling subordination stabilizes social relationships at the group level. Ring-tailed lemurs (Lemur catta) exhibit a more despotic dominance hierarchy than Verreaux’s sifakas (Propithecus verreauxi). We observed social interactions in 75 dyads of Verreaux’s sifakas and 118 dyads of ring-tailed lemurs. Both species used unidirectional submissive calls that have a dual function, potentially suggesting convergent evolution of the function of these signals in independent primate lineages. However, signalling subordination did not stabilize social relationships at the group level in both species. Additionally, subordination occurred more frequently in dyads of the more despotic ring-tailed lemurs than in Verreaux’s sifakas, indicating opposite patterns to macaques in the coevolution of social traits with dominance style.

This article is part of the theme issue ‘The power of sound: unravelling how acoustic communication shapes group dynamics’.

1. Introduction

Communication is an essential tool to mediate social behaviour and should, hence, have co-evolved with a species’ social organization and structure [1–3]. In particular, group-living species characterized by stable long-term memberships evolved behavioural strategies such as formalized dominance relationships, defined as asymmetries in aggression by one animal towards another animal, to limit the costs of conflicts [4,5]. To this end, conventions, i.e. formalized submissive signals such as facial expressions or vocalizations, evolved in many species to establish and maintain dominance relationships [5–9]. In addition, the outcome of conflicts is also under strong selection because an individual’s status in the dominance hierarchy can strongly influence its health, physiological stress response, access to resources and reproductive success [10,11].

The use and function of submissive signals vary across species with different dominance styles. Despotic societies are characterized by asymmetrical social relationships, steep linear dominance hierarchies and the specific use of submissive signals. In Japanese macaques (Macaca fuscata), for example, a facial expression, the silent bared-teeth display (hereafter SBT), is produced unidirectionally by subordinates and serves to signal the submissive status to dominants [12]. In contrast, in more tolerant species where relationships are more equal, resulting in shallower dominance hierarchies, the SBT display is used in several affiliative contexts [6,8,13,14]. For example, in Tonkean macaques (Macaca tonkeana) and crested macaques (Macaca nigra), the SBT display is used flexibly, irrespective of dominance status and appears to serve to signal general appeasement and increase affiliation [6,14,15]. Hence, the same signal can serve to communicate either fear and submission or affiliative intent, depending on the species’ social structure [15–17].

Interestingly, in some species in which subordinate individuals produce a unidirectional submissive signal, it serves both to signal submission and also affiliative intent, depending on the context of usage [9,18,19]. For example, in pig-tailed macaques (Macaca nemestrina), subordinates produce SBT displays unidirectionally in response to an immediate threat from a dominant group member but also in peaceful contexts without an immediate threat [9]. Submissive signals that are emitted during a conflict are considered to signal submission (an immediate behavioural response), whereas the same signals emitted in unprovoked contexts are considered to signal subordination (a long-term pattern of behaviour). Individuals benefit from signalling subordination by fighting less, grooming and reconciling more often than individuals that never signal subordination [9]. Similarly, in rhesus macaques (Macaca mulatta), individuals that signal subordination benefit from increased grooming and experience less severe aggression [18]. In both pig-tailed and rhesus macaques, submission signalling is associated with greater dominance relationship certainty, and individuals receiving subordination signals from many different signallers experience more social power by more frequently policing the fights of others [18,20–22]. Hence, the ability to communicate long-term behaviour has been suggested to reduce uncertainty in social relationships, allowing for higher-quality relationships and leading to more robust dominance relationships [9].

In addition, the two species of macaques differ in their dominance style, with rhesus macaques exhibiting the most despotic dominance style among macaques, while pig-tailed macaques exhibit a less despotic dominance style. According to Thierry [8], greater symmetry and uncertainty about aggression outcomes create room for negotiation in less despotic societies, suggesting that signalling subordination might be less important in more despotic societies [18]. In line with this assumption, despite an overall similar function of submissive signals, the more despotic rhesus macaques signalled subordination less frequently than pig-tailed macaques [9,18].

Non-human primates use not only facial expression but also vocalizations as submissive signals, such as the spat call of ring-tailed lemurs (Lemur catta) or the chatter call of Verreaux’s sifakas (Propithecus verreauxi) [3,7,19,23]. Both species produce these submissive calls not only during conflicts but also in peaceful contexts [7,19]. In Verreaux’s sifakas, the submissive chatter also seems to have a dual function by signalling both submission and subordination. Individuals signalling subordination by producing submissive calls more often in peaceful contexts benefit from increased affiliation and decreased aggression [19]. Hence, independent of the communicative modality with which submissive signals are produced, they appear to exhibit similar functions.

However, it is unknown whether the dual function of submissive acoustic signals also varies with dominance style, and whether these have evolved independently in all radiations of primates or only in African monkeys, which include macaques. Therefore, we investigated whether the dual function of submissive signals covaries with dominance style in Verreaux’s sifakas and ring-tailed lemurs. These two species belong to an adaptive radiation of primates endemic to Madagascar that evolved group living independently of macaques and of each other [24]. Both species live in multi-male and multi-female groups, in which females are dominant over males, but they vary in their dominance style [7,25–27]. Verreaux’s sifakas live in small groups of about six individuals [28], characterized by a more relaxed nonlinear and transitive dominance hierarchy. Ring-tailed lemurs, in contrast, live in larger groups of about 15 individuals, characterized by a steep linear, non-transitive dominance hierarchy [7,25,26]. Since the ability to communicate long-term behaviour has been suggested to reduce uncertainty in social relationships [9] and signalling subordination, in particular, is associated with greater dominance relationship certainty [22], Verreaux’s sifakas may experience higher levels of uncertainty compared to ring-tailed lemurs because of their nonlinear, transitive dominance hierarchy. Therefore, we predicted that in Verreaux’s sifakas, more dyads signal subordination and that the rate of subordination signalling is higher than in ring-tailed lemurs. In addition, we predicted that in both species, more frequent subordination signallers benefit by receiving more affiliation and less aggression, with this effect being more pronounced in Verreaux’s sifakas than in ring-tailed lemurs. Since kin, sex and age may influence affiliation and aggression, we also considered these effects in our analyses. In macaques, individuals receiving subordination signals more frequently enjoyed greater social power by also more frequently policing the fights of others, thereby stabilizing the social network [18,20–22]. Although neither lemur species polices the fights of others [7,29], signalling subordination may nevertheless have a stabilizing effect at the group level. Therefore, we also investigated whether subordination signalling is associated with more homogenous and, hence, more balanced as well as more stable social relationships.

2. Methods

(a). Study sites and behavioural observations

We studied wild Verreaux’s sifakas in Kirindy Forest, western Madagascar, where the German Primate Center has operated a research station since 1993 [28]. As part of a long-term study, sifakas are individually marked with nylon collars and well habituated to the presence of human observers [28]. We conducted a total of 504 h of behavioural observations on 22 adult individuals belonging to four groups, of which two groups have been observed in two different years. We observed social interactions between 75 dyads, of which 7 dyads were observed in both observation periods.

We studied wild ring-tailed lemurs in Berenty Private Reserve, southern Madagascar. Ring-tailed lemurs were individually identified by unique facial patterns or other natural markings. This population has been regularly studied since 1966 [30] and is also well habituated to the presence of human observers. We conducted a total of 447 h of behavioural observations on 23 individuals living in two different groups and observed social interactions between 118 dyads.

We conducted continuous focal animal observations (Verreaux’s sifakas: 60 min, ring-tailed lemurs: 30 min) and protocolled the duration of time spent in affiliation, i.e. grooming and being in body contact. In addition, we recorded the frequency of submissive calls given in conflict and peaceful contexts. A conflict was defined as any interaction involving aggression (cuff, bite, lunge, displacement). A submissive conflict call was defined as occurring immediately (within 5 s) following aggression. A submissive peaceful call was defined as occurring during peaceful interactions such as affinitive or affiliative contact, and not in response to aggression. Finally, we protocolled the occurrence of aggressions (cuff, bite, lunge and displacement) and calculated the rate of aggressions by excluding aggressive interactions in which a conflict call was given.

(b). Statistical analyses

To examine whether the usage of conflict and peaceful calls influenced affiliative and aggressive behaviour, we first calculated the proportion of peaceful calls for each dyad (the frequency of peaceful calls divided by the sum of peaceful and conflict calls). We used a Kruskal–Wallis test to investigate whether the proportion of peaceful calls differed between the composition of kin dyads, based on maternal relatedness (kin: both partners are kin, non-kin: partners are unrelated, unknown: at least one partner is of an unknown kin relationship) and the composition of sex dyads (female–female, female–male, male–male). We used a Mann–Whitney U-test to investigate differences among age classes (adult–adult, adult–juvenile). We used a proportion test to investigate whether the proportions of dyads producing peaceful calls, only conflict calls or no submissive calls differ between species. A t‐test was used to examine whether the rate of peaceful calls differed between species. For both species, we fitted generalized linear mixed models with a beta error distribution and logit link function (package glmmTMB [31]) to estimate whether the proportion of time spent affiliating per hour was predicted by the proportion of peaceful calls, composition of the sex dyad, age class and maternal relatedness as fixed factors. For Verreaux’s sifakas, we included the identities of the individuals forming each dyad as a random factor and a nested factor of group identity and year to account for the two groups that were observed in two different years. For ring-tailed lemurs, we included individual identities of the dyad and group identity as random factors.

To investigate whether signalling submission results in more homogenous social relationships at the group level, we also calculated the coefficient of variation (CoV) of affiliative relationships. This represents the number of differentiated relationships within a group, with smaller CoVs indicating that individuals interact more evenly with group members, whereas higher CoVs indicate more heterogeneous interactions. We used Spearman rank correlation to investigate whether CoVs of affiliative relationships correlate with the proportion of dyads in a group exchanging peaceful calls and the mean proportion of exchanged peaceful calls per group, respectively. Since we studied only two groups of ring-tailed lemurs, we conducted correlations for both species together and for seven groups of Verreaux’s sifakas because one group consisted of only two individuals, for which we could not calculate a CoV.

In both species, we additionally fitted linear mixed models (package lme) [32] to estimate whether the rate of aggressive interactions per hour was predicted by the proportion of peaceful calls, composition of the sex dyad, age class and maternal relatedness. For Verreaux’s sifakas, we included individual identities of the dyad and a nested factor of group identity and year as random factors. For ring-tailed lemurs, we included individual identities of the dyad and group identity as random factors.

(c). Model implementation

All analyses were conducted using R (v. 4.2.2 [33]). To ease model convergence, we centred the proportion of peaceful calls at a mean of zero and a standard deviation of 1 before including them in the models. We included all theoretically identifiable random slopes to avoid type I errors [34] but these slopes had to be removed in the end because the models would not converge. To test the significance of predictors as a whole, we compared the fit of the full model with that of the null model comprising only random factors [35,36]. Additionally, we checked GLMMs for overdispersion using the package ‘DHARMa’ [37]. We obtained confidence intervals for all models by means of parametric bootstraps using the function ‘bootMer’ of the package ‘lme4’ [32], applying 1000 parametric bootstraps. We checked for collinearity by determining variance inflation factors for a standard linear model without random effects using the package ‘car’ (v. 3.0.11) [38]. To estimate model stability, we proceeded by dropping levels of the random effect one at a time from the dataset and compared the obtained estimates to the estimates obtained for the full dataset. Models estimating the proportion of time spent affiliating revealed good model stability, and models estimating aggression rates revealed moderate model stability.

3. Results

(a). Usage of submissive calls in peaceful and conflict contexts

In Verreaux’s sifakas, we observed social interactions in 75 dyads. Overall, 55% of dyads (n = 41) exhibited peaceful signalling, with 25 dyads producing submissive calls in both peaceful and conflict contexts and 16 dyads producing submissive calls in peaceful contexts only. Nine per cent of dyads (n = 7) uttered submissive calls in conflict contexts only, and 36% of dyads (n = 27) never produced any submissive calls. Verreaux’s sifakas produced peaceful calls at an average rate of 0.41 ± 0.42 calls per hour (mean ± s.d.). The proportion of peaceful calls did not differ between dyads of kin and sex categories or age class (Kruskall–Wallis: kin dyads: χ ² = 3.78, d.f. = 2, p = 0.151; sex dyads: χ ² = 0.50, d.f. = 2, p = 0.779; age class: Mann–Whitney U-test: U = 541.5, p = 0.081).

In ring-tailed lemurs, we observed social interactions in 118 dyads. Seventy-nine per cent of dyads (n = 93) exchanged peaceful calls, with 80 dyads producing submissive calls in both peaceful and conflict contexts and 13 dyads producing submissive calls in peaceful contexts only. No dyad signalled submissive calls in conflict contexts only, and 21% of dyads (n = 25) never produced a submissive call. Ring-tailed lemurs produced peaceful calls at an average rate of 0.54 ± 0.34 calls per hour. The proportion of peaceful calls did not differ between dyads of kin and sex categories, but did differ between age classes (Kruskall–Wallis: kin dyads: χ ² = 3.78, d.f. = 2, p = 0.151; sex dyads: χ ² = 0.69, d.f. = 2, p = 0.709; age class: Mann–Whitney U-test: U = 984.5, p = 0.003). Peaceful calls were produced at a higher rate between adults (0.62 ± 0.22 calls per hour) than between adults and juveniles (0.33 ± 0.46 calls per hour).

In contrast to our prediction, the proportion of dyads exchanging submissive calls in peaceful contexts was lower in Verreaux’s sifakas than in ring-tailed lemurs (figure 1, proportion test: p = χ ² = 11.84, d.f. = 1, p < 0.001). However, the proportion of dyads producing only conflict calls or never a submissive call was higher in Verreaux’s sifakas than in ring-tailed lemurs (figure 1, proportion test: only conflict calls: p = χ ² = 8.91, d.f. = 1, p = 0.003; no calls: p = χ ² = 4.39, d.f. = 1, p = 0.036). In contrast to our prediction, Verreaux’s sifakas produced peaceful calls at a lower rate (0.41 ± 0.42 calls per hour) than ring-tailed lemurs, with on average 0.54 ± 0.34 calls per hour (t‐test: t = 2.40, p = 0.017).

Figure 1.

Proportion of dyads producing signals in peaceful contexts (peaceful and conflict contexts), only in conflict contexts, or never producing submissive calls in Verreaux’s sifakas (yellow) and ring-tailed lemurs (blue).

(b). Influence of the usage of peaceful calls on time spent affiliating

In Verreaux’s sifakas, the model estimating whether the proportion of time spent in affiliation was predicted by the proportion of peaceful calls, kin, sex composition of the dyad or age class was overall not significant (see electronic supplementary material, table S2a, likelihood ratio test full-null model comparison: χ ² = 6.60, d.f. = 6, p = 0.360). Therefore, we do not discuss these results any further .

In ring-tailed lemurs, the proportion of time spent in affiliation was predicted not only by the proportion of peaceful calls but also by kin and sex composition of the dyad, as well as age class (table 1, figure 2, likelihood ratio test full-null model comparison: χ ² = 66.47, d.f. = 6, p < 0.001). Dyads that exchanged more numerous peaceful calls spent more time in affiliation. In addition, maternally related dyads spent more time in affiliation compared to dyads with at least one non-related partner or one partner whose kinship was not known (figure 3a ). Female–female dyads spent more time in affiliation than female–male, but not male–male dyads (figure 3b ) and adult dyads spent less time in affiliation than adult–juvenile dyads (figure 3c ).

Table 1.

Results of the models estimating the influence of the proportion of peaceful calls, sex composition of the dyad, maternal kin relationship and age class on affiliative behaviour in ring-tailed lemurs.

model	term	estimate	s.e.	lower IC	upper IC	p
Proportion of affiliative behaviour in ring-tailed lemurs	intercept	−2.76	0.17	−3.10	−2.49	a
	prop. peaceful calls	0.17	0.06	0.04	0.30	0.006
	sex dyad (f–m) b	−0.46	0.13	−0.72	−0.18	<0.001
	sex dyad (m–m) b	−0.28	0.20	−0.72	−0.16	0.163
	kin (no-kin) b	−1.73	0.16	−2.02	−1.42	<0.001
	kin (unknown) b	−1.77	0.27	−2.31	−1.26	<0.001
	age class (ad–juv) b	0.45	0.15	0.12	0.78	0.003

^a Not shown because of having a very limited interpretation.

^b z-transformed to mean of zero and a standard deviation of 1; mean and standard deviation of the original predictors: proportion peaceful calls: mean = 0.54 ± 0.34, sex dyad: f–f: mean = 0.02 ± 0.03, f–m: mean = 0.08 ± 0.02, m–m: mean = 0.007 ± 0.01, kin relationship: mean = 0.05 ± 0.03, no-kin: mean = 0.005 ± 0.01, unknown: mean = 0.008 ± 0.01, age class: adult–adult: mean = 0.008 ± 0.02, adult–juvenile: mean = 0.02 ± 0.03.

Figure 2.

Proportion of time spent affiliating as a function of the proportion of peaceful calls in ring-tailed lemurs. The dashed line is the regression line, and the grey polygon the 95% confidence intervals.

Figure 3.

Proportion of time spent affiliating as a function of (a) kin and (b) sex composition of the dyad and (c) age classes in ring-tailed lemurs.

The coefficient of variance (CoV) of affiliative behaviour (Verreaux’s sifakas: 127 ± 9.46; ring-tailed lemurs: 202 ± 24.89) did not correlate with the proportion of dyads exchanging peaceful calls or the mean proportion of peaceful calls in the dataset comprising both species as well as in the dataset comprising only Verreaux’s sifakas (table 2). Hence, signalling submission does not appear to result in more homogenous social relationships at the group level.

Table 2.

Spearman rank correlations between the coefficient of variance of affiliative behaviour and the proportion of dyads producing peaceful calls as well as the mean proportion of peaceful calls per group.

group	correlation between CoV of affiliative behaviour and the proportion of dyads producing peaceful calls	correlation between CoV of affiliative behaviour and the mean proportion of peaceful calls
both species (n = 9)	R = 0.24, p = 0.527	R = 0.02, , p = 0.966
Verreaux’s sifakas (n = 7)	R = −0.09, p = 0.846	R = 0.13, p = 0.788

(c). Influence of the usage of peaceful calls on aggression rates

In Verreaux’s sifakas, the model estimating whether the rate of aggression was predicted by the proportion of exchanged peaceful calls, kin, composition of the sex dyad or age class was overall not significant (electronic supplementary material, table S2b; likelihood ratio test full-null model comparison: χ ² = 9.81, d.f. = 6, p = 0.133).

In ring-tailed lemurs, aggression rates were predicted by the proportion of exchanged peaceful calls and by the composition of the sex dyad (table 3, likelihood ratio test full-null model comparison: χ ² = 31.64, d.f. = 6, p < 0.001). Dyads that exchanged peaceful calls at a lower rate exhibited higher rates of aggression (figure 4). In addition, female–female dyads exhibited higher rates of aggression in comparison to female–male or male–male dyads. Kin or age class did not predict rates of aggression (table 3).

Table 3.

Results of the models estimating the influence of the proportion of peaceful calls, sex composition of the dyad, maternal– kin relationship and age class on aggression rates in ring-tailed lemurs.

model	term	estimate	s.e.	lower IC	upper IC	p
Aggression rates of ring-taled lemurs	intercept	0.27	0.07	0.13	0.40	a
	prop.peaceful calls	−0.03	0.01	−0.06	−0.00	0.043
	sex dyad (f–m) b	−0.16	0.04	−0.23	−0.09	<0.001
	sex dyad (m–m) b	−0.26	0.05	−0.36	−0.15	<0.001
	kin (no-kin) b	0.03	0.04	−0.04	0.12	0.413
	kin (unknown) b	−0.04	0.06	−1.16	0.08	0.474
	age class (ad–juv) b	0.10	0.07	−0.06	0.25	0.192

^a Not shown because of the very limited interpretation.

^b z-transformed to a mean of zero and a standard deviation of 1; mean and standard deviation of the original predictors: proportion of peaceful calls: mean = 0.54 ± 0.34, sex dyad: f–f: mean = 0.42 ± 0.42, f–m: mean = 0.18 ± 0.20, m–m: mean = 0.06 ± 0.08, kin relationship: mean = 0.22±0.21, no-kin: mean = 0.23 ± 0.30, unknown: mean = 0.03 ± 0.03, age class: adult–adult: mean = 0.22 ± 0.30, adult–juvenile: mean = 0.17 ± 0.20.

Figure 4.

Aggression rates as a function of the proportion of peaceful calls in ring-tailed lemurs. The dashed line represents the regression lines, and the grey polygon the 95% confidence intervals.

4. Discussion

In this study, we show that unidirectional submissive signals have a dual function in lemurs. In both Verreaux's sifakas and ring-tailed lemurs, unidirectional submissive signals were produced in peaceful and conflict contexts. In contrast to our prediction, fewer dyads signalled subordination at a lower rate in Verreaux’s sifakas than in ring-tailed lemurs. In Verreaux’s sifakas individuals signalling subordination did not appear to enjoy higher rates of affiliation than individuals signalling subordination less often. However, in ring-tailed lemurs, individuals signalling subordination enjoyed higher rates of affiliation than individuals signalling subordination less often. Moreover, individuals signalling subordination more often experienced lower rates of aggression in ring-tailed lemurs but not in Verreaux’s sifakas. These results suggest that signalling subordination may reduce uncertainty in social relationships in the more despotic ring-tailed lemurs but not in Verreaux’s sifakas. Moreover, signalling subordination did not covary with the coefficient of variation of affiliative relationships per group. Hence, patterns and benefits of signalling subordination differed between the two lemur species, but in the opposite pattern than predicted by their dominance style.

(a). Usage of submissive calls in peaceful and conflict contexts

In line with earlier studies [9,18,19], both lemur species used unidirectional signals in conflict and peaceful contexts. By using the same signal, they communicate both the willingness to withdraw in the present fight and the agreement to a subordinate pattern [9]. It has been suggested that signalling subordination to reduce uncertainty might be less important in more despotic societies, such as, shown by despotic rhesus macaques, who signal subordination at a lower rate than the less despotic pig-tailed macaques [18]. In contrast, in Verreaux’s sifakas, which are characterized by a nonlinear and transitive dominance hierarchy [25,26], fewer dyads signalled subordination at a lower rate than in ring-tailed lemurs. In addition, in Verreaux’s sifakas, about 10% of dyads produced submissive calls in conflict contexts only, suggesting that individuals in these dyads only responded to an immediate threat but never signalled the behavioural state of subordination, whereas in ring-tailed lemurs, not a single dyad produced submissive calls in agonistic contexts only. In a similar vein, the proportion of dyads never signalling submission in any context was higher in Verreaux’s sifakas than in ring-tailed lemurs, suggesting that the more despotic ring-tailed lemurs appear to signal submission more often to reduce uncertainty in social relationships than the less despotic Verreaux’s sifakas to reduce uncertainty in social relationships. Hence, patterns of signalling subordination differed between the two species, but in the opposite pattern than predicted by their dominance style.

(b). Benefits of signalling subordination

Ring-tailed lemurs signalling subordination at a higher rate benefitted from an increase in affiliation and a decrease in aggression in comparison to individuals signalling subordination at a lower rate. Thus, individuals signalling subordination enjoyed better social relationships. Affiliation was additionally moderated by kin—a pattern often observed in despotic societies [39]—and the sex composition of the dyad (with female–female dyads exhibiting higher rates than male–male or female–male dyads, reflecting female dominance and strong competition among females [7,40]), as well as by age classes (with dyads of adults and juveniles exhibiting higher rates of affiliation than dyads of adults only).

In Verreaux’s sifakas, signalling subordination, however, appeared to have no effect on affiliation and aggression. Therefore, potential benefits of signalling subordination in this population of Verreaux’s sifakas could not be demonstrated in this study. The dual function of submissive calls has also been investigated in another Verreaux’s sifaka population at Kirindy Mitea National Park [19]. In this population, more dyads (94%) signalled subordination than in Kirindy Forest (55%), whereas the proportion of dyads producing submissive calls in conflict contexts (6% versus 10%) did not differ strongly. Moreover, in the Kirindy Mitea population, Verreaux’s sifakas signalling subordination benefitted more often from an increase in grooming and a decrease in aggression, which is in contrast to the results of the present study. In principle, both populations exhibit similar patterns in female dominance style [27,41], suggesting that potential variation in social structure may not explain the variation in signalling subordination between these two populations. In the Kirindy Mitea population, however, only 18 dyads were observed, necessitating a different statistical approach to estimate the effect of single predictors in separate models, whereas we estimated the effect of several predictors in a single model. Hence, it is difficult to conclude whether differences in signalling subordination might present real population differences or whether they result from different statistical approaches.

In macaques, individuals that received subordination signals more frequently enjoyed greater social power, allowing them to police others and, thus, to stabilize the social network [18,20–22]. Although lemurs do not police the fights of others [29], we investigated whether signalling subordination might nevertheless be associated with more balanced and, thus, more stable social relationships at the group level. However, we found that neither the proportion of dyads signalling subordination nor the proportion of peaceful calls covaried with the coefficient of variation of affiliative relationships per group, suggesting that signalling subordination in lemurs is not associated with more balanced and more stable social relationships at the group level. The gain in social power through receiving more frequent subordination signals might, therefore, be a mechanism for stabilizing social networks only in species that police the fights of others.

(c). Evolution of submissive signals

In the genus Macaca, it has been suggested that more egalitarian species evolved more communicative signals than despotic species as a means to negotiate social relationships because more egalitarian societies exhibit greater symmetry in dominance relationships and, hence, greater uncertainty about aggression outcomes [8,42,43]. In line with this notion, the less despotic pig-tailed macaques signalled subordination more frequently than the more despotic rhesus macaques [18]. However, we found the opposite pattern in the two lemur species studied here. Moreover, the more despotic ring-tailed lemurs evolved more signals in the visual, olfactory and vocal communicative domains than Verreaux’s sifakas [3], indicating that the relationship between dominance style and number of signals also reflects a pattern that is opposite to what we see in macaque societies. Lemurs evolved group living independently from anthropoid primates [24] and deviate from them in several morphological and social traits, such as the lack of sexual dimorphism, sex bias in adult sex ratios, strict female dominance and mild genital masculinization, which have been referred to as the lemur syndrome [27,44]. Nevertheless, these two lineages exhibit a convergent pattern in aspects of social evolution as, for example, the coevolution of communicative complexity and sociality [3,44].

In the two investigated lemur species, the dual function of subordination signalling resembled, in principle, that of macaques, apparently suggesting convergent evolution in the function of these signals. However, the frequency of signalling varied with dominance style in lemurs in the opposite way compared to macaques. In macaques, communication and dominance style co-evolved in species belonging to the same genus, and it has been suggested that, in particular, the use of the submissive SBT might have been present already in their common ancestor [22]. In this study, however, we compared lemur species belonging to two different families that evolved group living independently from each other [24]. Despite the independent evolution of group living, both lemur species have evolved vocal signals to mediate dominance relationships, signalling the willingness to withdraw in agonistic contexts and agreeing to a subordinate pattern in peaceful contexts. Since the frequency of signalling subordination varied in lemurs with dominance style in the opposite way to what is seen in macaques, communication and dominance style across families did not co-evolve in a similar way as within the genus of macaques. Hence, superimposing patterns of the coevolution of social traits with dominance style, i.e. variation in social structure, on other taxa outside the genus of macaques should be done with caution, and more studies in other group-living primate lineages are recommended.

Ethics

We adhered to the 'Guidelines for the ethical treatment of non-human animals in behavioural research and teaching' as published in Animal Behavior (2023, 195, I-XI) and the laws of the country where the research was conducted. This study was approved by the Département de Biologie Animale of the University of Antananarivo and the CAFF/CORE of the Direction des Eaux and Forêts de Madagascar. Research on Verreaux’s sifakas was authorized by the CNFEREF Morondava, Madagascar. Research on ring-tailed lemurs was authorized by the family de Heaulme, the owners of the Berenty Private Reserve. Additionally, the Animal Welfare Body of the German Primate Center confirmed that the observations are in accordance with the legal and ethical requirements of appropriate animal procedures using non-human primates in Germany (No. 14-E-2).

Data accessibility

Supplementary material is available online [45].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

C.F.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, visualization, writing—original draft; K.D.: data curation, methodology, writing—review and editing; F.R.: resources, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

No funding has been received for this article.