How zoo-housed chimpanzees (Pan troglodytes) target gestural communication within and between age groups

Resumo A comunicação gestual nos primatas não-humanos evoluiu como uma resposta à complexidade social. Neste âmbito, machos e fêmeas tendem a usar diferentes tipos de gestos, provavelmente consequências das diferentes pressões sociais a que estão sujeitos. Neste estudo, realizou-se uma análise intra e interclasses etárias dos gestos produzidos em contextos distintos. Para tal, durante um período de 3 meses, observou-se uma colónia de chimpanzés (Pan troglodytes) em cativeiro. Inicialmente, os dados foram recolhidos através de uma amostragem ad libitum, para a elaboração do catálogo gestual. Posteriormente, a amostragem focal determinou quem gesticula com quem, e em que contexto. De um modo geral, os resultados mostraram que, em alguns casos, os chimpanzés juvenis tendem a gesticular para outros chimpanzés de Abstract Gestural communication among nonhuman primates evolved as a response to their complex social environment. In this scope, males and females, adults and nonadults employ different gestures, probably due to their distinct social roles. In this study, a within and between age group analysis of the gestures produced in different contexts was carried out. For this purpose, a community of 16 captive chimpanzees (Pan troglodytes) was observed during a 3-month period. Initially, data were collected through ad libitum sampling in order to identify their gestural repertoire. Subsequently, focal sampling was used to identify who gesticulated with whom and in what context. Overall, the results showed that juvenile chimpanzees tend to direct their gestures to different age groups according to the context; more


Introduction
Over the past decades, research on gestural communication in nonhuman primates has become a rather appealing area due to the phylogenetic proximity between humans and other primates. Evidences on shared communication characteristics in nonhuman and human primates have emerged, regarding flexibility, intention, and learning opportunity (Liebal and Call, 2012), cultural variation and mutual understanding (e.g. chimpanzees and bonobos: Pollick and de Waal, 2007; bonobos: Graham et al., 2017) or even neuronal mechanisms underlying language production in humans and gestural production in nonhuman primates (e.g. southern pig-tailed macaques: Rizzolatti et al., 1996;chim-panzees, bonobos, and western gorilla: Cantalupo and Hopkins, 2001).
Besides the hypothesized association with the human language, gestural communication used by nonhuman primates plays a fundamental role in the transmission of information among individuals. Therefore, its study allows for a deeper understanding of how primates interact with conspecifics, and how they deal with a vast group of challenges, mainly of a social nature (Liebal et al., 2013). In fact, according to some authors (Maestripieri, 1999;Call and Tomasello, 2007), the gestural communication of nonhuman primates is influenced by the complexity of their social environment. That is, the higher the social complexity of a particular species, the greater will be the complexity and variety of its gestural reper-classes etárias semelhantes, tendo em conta o contexto; mais especificamente, estes direcionam gestos em contexto de brincadeira para outros chimpanzés juvenis, e comunicam com chimpanzés adultos em contexto de locomoção e afiliação. Com base nestas evidências, sugerimos a existência de um certo grau de flexibilidade na capacidade de comunicar por gestos dos chimpanzés juvenis, uma vez que estes tendem a direcionar os seus gestos para recetores de classes etárias aparentemente mais adequados, tendo em conta o contexto do sinal gestual.
Palavras-chave: Gestos; chimpanzé; flexibilidade; brincadeira. specifically, juvenile chimpanzees frequently gesticulate within their age group in play contexts, and with older individuals in locomotion and affiliation contexts. Based on this, a certain degree of flexibility in juvenile chimpanzees gestural signalling is suggested, to the extent that they rather direct their gestural signs to chimpanzees of the same age group with the aim of involving themselves in the activities' context in which the gestural sign is produced.
toire. Taking chimpanzees as an example, their gestural repertoire is one of the most varied (i.e., in terms of the number of different gestures produced) and complex (in meaning and morphology), reflecting an adaptation to the innumerable demands of their social systems (Roberts et al., 2012a), which comprises formation of coalitions (Gilby et al., 2012), conflict resolution (Fuentes et al., 2002), and mating strategies (Duffy et al., 2007).
Lately, new data supporting the hypothesis that social complexity affects gestural communication of nonhuman primates arose. Scott (2013) found differences in the use of gestural signs by opposite sex chimpanzees when subjected to social pressures. In intersexual interactions, males resorted more often to agonistic gestural signs, as a control attitude toward females, while females gesticulated more frequently in affiliation contexts, to avoid retaliation and aggressive behaviour of males. On the other hand, Roberts and Roberts (2015) found that, in mating circumstances and also for chimpanzees, there was an adaptation in the sensory modality of subordinate males' gestural communication in the absence or presence of alpha males. In the presence of alpha males, subtle tactile gestural or visual signs were used to address females, whereas in their absence, acoustic gestural signs were preferentially used to get females attention.
Gestural communication also differs according to the nonhuman primate ontogeny. In this sense, Tomasello et al. (1989) stressed that some gestural signs used by juvenile chimpanzees were not adopted by adults and vice-versa; and that other gestural signs were used by individuals of distinct age groups, though with different purposes. For example, the arm raise (an individual raises his arm, as if to hit, and then charges other) gestural sign is made by juvenile chimpanzees only to engage in play activities (Tomasello et al., 1985), and the throwback head is used both by young siamangs (Symphalangus syndactylus) to play and by adults to start copulation (Liebal et al., 2004a).
Additionally, Call and Tomasello (2007) also suggested that, in chimpanzees, gestural repertoire tends to increase during infancy and youth, but it decreases once the individual becomes an adult. Hobaiter and Byrne (2011a) proposed that the reduction of gestural repertoire in adulthood may be explained by a higher capacity of understanding gestural signs, restricting communication to more effective gestures, in a phenomenon called repertoire tunning. On the other hand, younger individuals are less experienced and thus incapable of understanding gestures efficacy; accordingly, they use a greater number of signs that constitute a larger gestural repertoire, although with a smaller efficiency in comparison to adults. Call and Tomasello (2007) and Liebal et al. (2013) added that younger chimpanzees are highly involved in play activities, which require a higher flux of communi-cation. Actually, according to Frohlich et al. (2016a;, playing is a decisive element in the acquisition of gestural signs in an early phase of the individual's ontogeny. They state that playing provides socialization opportunities as well as the necessary experiences to stimulate the development of a complete gestural repertoire -flexible and intentional -and to improve, simultaneously, the capacity of using proper gestural signs for the accomplishment of a specific goal. Although individuals in different ontogeny stages also suffer distinct social pressures (Hamada and Udono, 2006;Watts and Pusey, 2002), very little is known about the pattern of gestural communication in individuals of different age groups. Liebal and Call (2012) underlined this idea when alerting for the scarcity of published data on differences in gestural signs use among independent groups of the same community of nonhuman primates. As mentioned above, some evidences about these differences -e.g., between males and femaleshave been reported (e.g. Scott, 2013;Roberts and Roberts, 2015). Nevertheless, on what concerns age groups, research has been mostly focused on the effect of social complexity in the use of gestural signs, as well as on the development of gestural repertoire throughout the ontogeny of individuals, namely on how signs are acquired in the first years of life.
Endeavouring to overcome this lacuna, the present study aims to analyse how individuals from a chimpanzee (Pan troglodytes) captive colony produce gestural signs between and within distinct age groups in different contexts. For this purpose, we hypothesized that the context of gesture communication would differ between and within age groups interactions, for both juvenile and adult chimpanzees, and match their social priorities. To this end, we predicted that, on the one hand, adult chimpanzees would be more likely to employ sex and grooming within age group interactions, and, on the other hand, juvenile chimpanzees would be more likely to use play gestures within age group interactions, but locomotion, food and affiliation gestures between age group interactions.

Subjects
The colony studied was composed by 7 male and 9 female chimpanzees, from which 5 were juveniles (aged between 4 and 7 years old) and 11 were adults (aged between 13 and 30 years old), all living at the Lisbon Zoo (Portugal). Neither kinship, nor the dominance rank in the colony were known. During the day, chimpanzees had unrestricted access to the outdoor captivity (1220 m 2 , approximately), composed by a grass field delimited by a stream and made up of tree trunks, a wooden climbing facility, ropes, blankets and stairs leading to the indoor captivity (470 m 2 , approximately) that also contained a wooden climbing facility and where chimpanzees stayed overnight and during heavy rain days, and cleaning and feeding operations (every day between 9 am and 10 am).

Procedures
Data was collected by the first author (MO) of this study between 10 am and 4 pm, during a 3-month period, from December 2 nd , 2015 to February 23 rd , 2016.

Data Collection
With the aim of recording interactions during which chimpanzees displayed gestural signs to communicate with each other, ad libitum sampling was adopted in a first phase, followed by a focal sampling (Altmann, 1974;Martin and Bateson, 1993), each one serving a different purpose. Ad libitum sampling allowed the elaboration of the gestural repertoire. In order to guarantee the accuracy of the data collected, the observation in this phase was based on the protocol described in table 1.
At the end of the first phase, the gestural repertoire (table 2) was produced according to the following procedures: (1) morphological description of gestural signs [analogous to Hobaiter and Byrne (2011b), or Roberts et al. (2014)] always Table 1. Ad libitum observation sampling protocol.

Item
Operational Definition Criteria

Gesture
The communication signal produced by the expressive movements of the head, limbs, or other body parts and directed to a recipient with the purpose of influencing their behaviour in a specific way (Liebal et al., 2004b;Roberts et al., 2014).
(2) Gesture production in a way that ensures the recipient can visualize it. Otherwise, the signaller manipulates the recipient' s attention via auditory and/or tactile signs or by moving forward to the recipient before gesture production (Liebal et al., 2013;Roberts et al., 2012b).
(3) Persistence or production of additional gestures in response to an unsuccessful communication attempt (Liebal et al., 2013;Roberts et al., 2012b).

Apparently Satisfactory Outcome
An outcome that ends the signaller' s intention to communicate, if the purpose of the gesture was satisfied (Hobaiter and Byrne, 2014).
Based on the observation of the impact of the recipient' s behavioural response towards the signaller, according to the functional context of the interaction (Roberts et al., 2012b;Moore, 2014;Cartmill and Byrne, 2010).
beginning the designation attributed to each gestural sign by a verb in the infinitive form, as proposed by Nishida et al. (2010); (2) allocation of the functional context to each gesture; and (3) correspondence between each gestural sign and the respective and apparently satisfactory outcome. The functional context of each gesture was also determined based on its apparently satisfactory outcome (Tomasello et al., 1997) and in line with the definitions proposed by Pollick and de Wall (2007), namely for play, affiliative, agonistic, feed, locomotion, sexual and groom contexts.
As an example of these procedures, the direct upper limb gestural sign is morphologically described by the stretch of one of the upper limbs towards the receiver or another location, and it is performed by a chimpanzee who looks to move to another location, in the context of locomotion (see table 2). It should be noted that intra-or inter-observer tests were not carried out. Not only the geomorphological characteristics of the chimpanzees' enclosure were not suitable for video equipment installation, but also the viewing angle did not allow for an observation spectrum capable of covering the entire space. Hence, the observer was forced to change position on several occasions, during date collection. Plus, no gestural repertoires on the population under study were available. Alternatively, in an attempt at minimizing this handicap, the gestural repertoire obtained through ad libitum sampling was compared with the ones gathered by Roberts et al. (2014), Scott (2013), McCarthy et al. (2013), and Hobaiter and Byrne (2011b). From the 30 gestural signs identified, only one did not have a partial or total correspondence with the morphological descriptions presented by the above mentioned authors.
Subsequently, focal samplings were collected. During each 15 minutes session, a focal chimpanzee was followed and observed as a signaller (the chimpanzee who produces an intentional gesture, Roberts et al. 2012a) or a recipient (the chimpanzee to whom the gesture was directed, as determined through the orientation of the head and the body of the signaller, during or immediately after the gesture production, Roberts et al. 2012a) of a gestural sign. In each interaction, along with the gestural signs, the identification of the recipient (if the focal chimpanzee was the signaller) or the signaller (if the focal chimpanzee was the recipient) was registered. The behavioural response of the recipient was also registered (present or absent). The focal sampling sessions were not randomized. The selection of the focal chimpanzee was based on the criterion of the number of focal sessions already done to ensure that all subjects had the same number of focal sessions and identical observation times, as suggested by Martin and Bateson (1993).
The gestures previously identified and described at the gestural repertoire Table 2. Gestural repertoire with all gestures observed during ad libitum sampling for the chimpanzee colony, living at the Lisbon Zoo (Jardim Zoológico de Lisboa). This repertoire encompasses the gesture designation, its morphological description, functional context and apparently satisfied outcome, and the age group who produced the gesture.

Gesture
Morphological   were converted into 3-letter codes [analogously to Liebal et al. (2004b) and Roberts et al. (2014)] in order to avoid redundancies and to improve data collection efficiency (Lehner, 1996). For example, the direct upper limb was coded into dul.

Data Analysis Gesture Rate
Only data collected during the focal sampling were considered for calculation of the gesture rate. This rate (g) was calculated for each individual in the sample, by dividing the total number of gestural signs produced under focal sampling (n) by the total number of hours of observation (t). Regarding the total number of hours of observation, each chimpanzee was observed during 12 focal sessions. In most chimpanzees, 12 focal sessions are equivalent of three hours of observation. However, in some cases, during the observation session, the focal chimpanzee temporarily was no longer in sight. This out of sight amount of time was not accounted. That is, the observation time, in hours, ranges from 2.75 hours to 3 hours -i.e., all chimpanzees had the same number of focal sessions, but not all had the same observation time.
The gesture rate was measure for the situational contexts described in table 3.
Besides gesture rate analysis, the behavioural response of the recipients was also analysed. For this responsiveness rate (r), the total number of interactions with recipient's behavioural response (n 1 ) was divided by the total number of interactions (n 2 ) in three situational contexts: overall interactions, overall intra-age group interactions (juvenile-juvenile, adult-adult), and overall inter-age group interactions (juvenile-adult, adult-juvenile).

Statistical Analysis
Collected data was statistically processed and analysed using IBM SPSS Statistics 23. The gesture rate for each chimpanzee and for each situational context was first measured, as described above. Then, and given the small sample size (N=16), the Shapiro-Wilk test was run and all variables in study were not normally distributed. Subsequently, the Mann-Whitney test was used to compare the gesture rate between juvenile and adult chimpanzees in all interactions (overall and for each functional context), in same age group interactions (overall and for each functional context) and in different age group interactions (overall and for each functional context). The same procedure was used to compare the responsiveness rate in overall, intraage group and inter-age group interactions. Finally, the Wilcoxon test was used to compare the gesture rate between same age groups and different age groups interactions (overall and for each functional context). Table 3. Situational contexts in which gesture rate was measured and its requirements.

Situational context Functional context Recipient
All interactions (overall) All All All intra-age group interactions (overall) All Only same age group recipients

Results
First of all, no gestures were found in either feeding or agonistic contexts. Overall, almost 75% of the gestures observed were produced by juvenile chimpanzees. Table 4 briefly summarizes the percentage of gestures produced in each functional context. On the one hand, from an age group view, play and locomotion were the contexts in which most gestures were displayed by juvenile and adult chimpanzees, respectively. On the other hand, from a context view, play and locomotion, and sex and groom were the contexts in which there was a greater predominance of gestures employed by juvenile and adult chimpanzees, respectively.

Within Age Group Interactions
When considering only interactions within the same age group (juvenile-juvenile, adult-adult), significant differences were obtained between the individuals' global gesture rates (U=0.000; z=-3.136; p≤0.001). However, when each context was analysed separately, significant differences were only obtained for the play context (U=0.000; z=3.580; p≤0.001). In this particular case, as shown in figure 2, the gesture rate was considerably higher

Between Age Group Interactions
As for the interactions between individuals of different age groups (juvenile-adult; adult-juvenile), significant differences were also obtained in the global gesture rates (U=0.000; z=-3.127; p≤0.001). Regarding each context, differences were found in affiliation (U=1.000; z=-3.233; p=0.001) and locomotion (U=2.000; z=-2.906; p=0.002), where the gesture rate of juvenile chimpanzees was higher, as shown in figure 3.

Within vs Between Age Group Interactions
When each age group was analysed separately (Figure 4) -and once more with the aim of comparing the gesture rate intra-and inter-age group in different contexts -, statistically significant differences were only obtained for juvenile individuals in play (z=-2.023; p=0.031), locomotion (z=-2.086; p=0.036) and affiliation (z=-2.023; p=0.031) contexts. In the first case, intra-age group gestural communication was significantly higher than the recorded for the inter-age group. The opposite was found for the locomotion and affiliation contexts. No significant differences were found in the grooming context for juvenile individuals (z=-0.816; p=0.375). Regarding adult individuals, no differences were found between the rate of gestures intra-and inter-age group in play (z=-0.447; p=0.500), locomotion (z=-1.137; p=0.155), affiliation (z=-1.725; p=0.078), grooming (z=-1.580; p=0.070) or sexual (z=-0.184; p=0.500) contexts.

Discussion and conclusions
The main aim of this study was to investigate the differences in gestural communication within and between age groups of the chimpanzee community living at the Lisbon Zoo (Lisbon, Portugal). In order to ascertain these differences and to analyse in which contexts there was a higher frequency of intraand inter-age group gestural communication, the observation focused on three parameters: age group of the signaller, the gesture produced (and respective functional context) and the age group of the receiver. In general, the obtained results revealed that, in some cases, juvenile chimpanzees tended to direct their gestures to chimpanzees of apparently more adequate age group to particular contexts.  The within and between age groups analyses have shown that juvenile chimpanzees strongly prefer to communicate by gestures in a play context, but also to direct the same gestures to conspecifics of the same age group. These evidences corroborate the results obtained by Tomasello et al. (1985) and Frohlich et al. (2016a). As such, two plausible and complementary motives emerge to explain why young chimpanzees gesticulate substantially more to others of the same age group in play context.
First of all, several authors (Hobaiter and Byrne, 2011b;Tomasello et al., 1997;King et al., 1980;Palagi et al., 2004;Shimada and Sueur, 2014) state that playing is the main context in which younger individuals of several primate species get involved. Generally, the evolutionary function of playing fosters the development of social, physical and cognitive capacities (Davila-Ross et al., 2011;Cordoni and Palagi, 2011), and promotes the necessary behavioural flexibility to deal with social and ecological needs (Palagi and Paoli, 2007). Therefore, play activities between juvenile individuals bring several benefits, namely socialization, sensori-motor stimulation and physical and cognitive exercising (Bekoff and Byers, 1981). They might also influence future dominance hierarchy (Byers and Walker, 1996;Palagi and Cordoni, 2012), stimulate the learning of behaviour and communication elements from other chimpanzees (Fagen, 1981), reduces social conflicts (Palagi, 2007), and potentiate the practice and testing of communication signs as well as the establishment of social relationships (Goodall, 1968).
Secondly, adult chimpanzees do not show much physical and cognitive availability to get involved in non-priority activities such as playing. That is why younger chimpanzees tend to choose other young individuals to play (Mendonza-Granados and Sommer, 1995;Flack et al., 2004).
Moreover, the results obtained in this study apparently support the social Figure 5. Responsiveness rate comparison of juvenile and adult chimpanzees in overall, intraage group and inter-age group interactions. negotiation hypothesis, which, according to Frolich et al. (2016a), suggests that "gestures are the output of social shaping, shared understanding, and mutual construction in real time by both interactants". The results here presented clearly provide evidence of a high prevalence of gestural signalling in play interactions among young chimpanzees. This suggests that they devote a great part of their time to playing, which is, according to Frohlich et al. (2016a;, an essential component in the early stages of chimpanzees' ontogeny, not only for the development of a functional and complete gestural repertoire, but also for the effective learning of the appropriate gestural signs. The latter should be based on mutual and shared understanding and exhibited to produce different meanings, specific for different contexts.
In play context, the gestural communication pattern of juvenile individuals is compatible with what is considered, in the literature aforementioned, as a typical behaviour of young chimpanzees. In other words, the involvement among young individuals in play activities is rather common and the results obtained corroborate that tendency, attested by a significant rate of gestural signs within the group of juvenile chimpanzees in play context.
On the other hand, on what concerns inter-age groups interactions, the results revealed differences in locomotion and affiliation contexts, in which the gesture rate of juvenile chimpanzees has surpassed the one of adult individuals. Apparently, the fact that locomotion and affiliation include diverse and regular parenting activities may explain why, in these contexts, juvenile chimpanzees tend to direct their gestures to adults. For instances, in locomotion context, younger chimpanzees frequently request 'joint-travel' to adults, with the intent of exploring the surrounding environment, even when they are already physically independent (Tomasello et al., 1989;Tomasello et al., 1985;Flack et al., 2004). This pattern of gestural communication is similar to that found by Halina et al. (2013) in bonobos (Pan paniscus) living in captivity. However, Frohlich et al. (2016b) verified the opposite in wild chimpanzees, i.e., that the initiative of gesticulating in order to require 'joint-travel' was performed by the progenitors. In sum, data here presented do not sustain potential interspecific differences on what concerns maternal styles related to locomotion, as stated by Frohlich et al. (2016b). For this motive, it is suggested that possible socioecological discrepancies between captivity and wild environment might be at the base of the differences found.
The same gestural communication pattern was verified in affiliation context. Younger chimpanzees usually seek their parents or other adult chimpanzees for affiliative activities (Tomasello et al., 1989;Goodall, 1968;Pusey, 1990;Langergraber et al., 2007). This occurs not only due to the affective relationship between them, but because the latter are the most suitable to provide affiliative care. Hayashi and Matsuzawa (2017) also argue that, even after becoming independent and beginning to explore the social and physical environment around them, younger chimpanzees continue to request affiliation activities to their mothers quite often, taking them as a 'secure base' .
As emphasized by Roberts et al. (2012b), gestural communication plays an important role in the way nonhuman primates deal with social complexity. In fact, the results here presented suggest a certain degree of flexibility in gestural signalization of juvenile chimpanzees, showing their tendency to direct gestural signs to get involved with individuals of a similar age-group depending on the context in which the gestural sign is produced.
Finally, the following constraints should be taken into account when interpreting the obtained results: the relatively short period of data collection; the possible kinship and/or dominance rank between the individuals (according to some authors, this may influence several interactions in diverse contexts -e.g. Goodall, 1968;1986;Boeschet al., 2006;Lonsdorf et al., 2014;Foerster et al., 2016;Murray et al., 2006), and, more important, the fact that the data collected was not validated through analyses intra-and inter-observer. To attain a holistic understanding of gestural communication in nonhuman primates, similar analyses, i.e., between/with-in age-groups, should be carried out both in the wild, where primates are exposed to selective active pressures (Frohlich et al., 2016b), and in captive populations.