In the previous chapter, it was shown that a three-component structure of relationship quality exists within this study group and gives evidence to suggest that this theoretical framework is likely to be reflected in other captive chimpanzee groups. Chapter 3 explains the quality and asymmetrical nature of chimpanzee relationships at the dyadic level. However, it is important to understand how a relationship between two individuals contributes to the dynamics of an entire social group as the frequency and nature of chimpanzee interactions are highly susceptible to temporal change. It is, therefore, important to further quantify social relationships at the group level, which can be achieved through the use of statistical tools. Social network analysis (SNA) is an analytical technique which has been employed with growing popularity to the study of animal behaviour. The potential application of SNA, to aid in the study of social relationships among chimpanzees, will be explored in this chapter.

Primate species housed in captivity often live in closed social groups, which are underpinned by individual recognition and a shared history of repeated social interaction with specific group members (Aureli & Schaffner, 2002; Silk, 2007). In most such groups, social interactions are not evenly spread between adult group members, with individuals selectively preferring a few key individuals (McFarland & Majolo, 2011). Such non-random social biases can either reflect constraints imposed by the physical environment and/or living in a captive setting (Barrett & Henzi, 2006), or can indicate real social preferences in the sense that individuals within a group actively seek each other out as social partners. In order to understand the complexity of the social structure of chimpanzees housed in captivity, and to be able to link the behaviour of individuals with the dynamic of the group as a whole, it is important to consider, analyse and quantify all relationships connecting each member of the group. Social Network Analysis (SNA) offers an excellent framework to do this by providing a way to describe the continuum of social complexity that encompasses each social relationship within a group.

Social network analysis examines a social group as a network of individuals, or “nodes”, connected by social ties (Sueur et al. 2011), analysing each individual in the context of their relationships with other group members. SNA comprises a collection of statistical models, rooted in the theoretical framework of graph theory, which generates a visual map of the connections among individuals and provides quantitative measures that substantiate the underlying patterns of each relationship (Whitehead, 2008). These techniques have been successfully applied to aid in the description of complex social interaction patterns between group-living non-human primates (Clark, 2011; Rushmore et al., 2013), whilst providing a deeper understanding of their social complexity by measuring social relationships directly (Whitehead, 2008).

As previously stated, the social structure of a group concerns how relationships are patterned. The focus on patterns of relationships is what distinguishes SNA from other analytical techniques. This approach can analyse the structural properties of a network as a whole or analyse an individual’s role within it. SNA allows for the identification of key individuals and examines how they fragment or integrate the overall structure of a social group (Beisner et al., 2010). Therefore, social network analysis could provide an insight into the quality of social relationships through quantitative measures such as “reciprocity”, “centarality”, and “fragmentation”; important measures to consider when studying the complex and highly asymmetrical nature of social relationships between chimpanzees.

In group-living primate communities, social services are frequently exchanged between group members (Kutsukake & Castles, 2004). However, social relationships are often asymmetric as the relative importance of a relationship to each member of a dyad is unlikely to be equal (Silk, 2002). Among chimpanzees, living in large and complex social groups, there are vast differences in the abilities of group members to effectively obtain a resource or gain competitive power based on their dominance status (Maynard-Smith & Parker, 1976). As a consequence of living in a fluctuating social network, individuals often have to make trade-offs in order to balance the advantages and disadvantages brought about by group living (Conradt & Roper, 2005). The strategies that enable individuals to achieve this typically affect the pattern and frequency of interactions with specific group members resulting in non-random associations (Ramos-Fernandez et al., 2006). Studying the patterns of this assortativity among captive chimpanzees using SNA can allow us to explore how individual behaviour influences what happens at the dyadic and group level. Furthermore, sociometric analysis can provide an objective, systematic way to determine dominance rank and centrality measures can be used to identify important individuals within the social group and their relative contribution to each relationship with a social partner.

Although SNA has gained rapid popularity within the field of primatology, many previous studies have largely focussed on interactions within primate groups housed in laboratory settings or in the wild (reviewed by Sueur et al., 2011), and have sought to address theoretical rather than applied questions regarding the importance and frequency of social interactions between group members. In contrast, the application of SNA to primate groups housed in a zoo environment has been largely understudied and the advantages of using this analytical technique when studied zoo-housed primate groups has been seemingly overlooked. In zoos, spatial limitations and husbandry practises often mean that group sizes are small (Hosey, 2005). However, these small social groups can yield much smaller social networks in comparison to many wild primate groups and the validity of SNA is dependent on the collection of valid samples of relationship data rather than the number of individuals that comprise a social network (Whitehead, 2008). Additionally, some features of the zoo environment can actually help to bypass frequently occurring sampling issues that are encountered when studying wild populations (Clarke et al., 1982). For example, close-range observations carried out over a sustained period of time can usually be achieved when documenting the interactions of zoo primates, whereas a continuous sampling method recording detailed interaction data can be difficult to collect in the wild. Furthermore, although some spatial limitations exist in a zoo setting, populations are “closed” so that all individuals live within a clearly defined boundary within an exhibit. In contrast, defining “social group” among wild animals has been shown to be difficult (e.g. Franz and Nunn, 2009; Ramos-Fernandez et al., 2009), with differing definitions existing within the literature often involving the assumption that all individuals seen together are interacting or have some degree of association with each other.

Previous studies on social networks in primates have primarily focussed on analysing subgroup interactions, sub-grouping patterns during fission or grooming reciprocity across the entire network. Furthermore, recent network studies have utilised a purely simulated “experimental knockout” procedure whereby key individuals are removed from a network theoretically as a means of understanding how network stability could change over time (Flack et al., 2006; James et al. 2009). The use of “experimental knockout” has generated much debate within the literature as natural primate groups have an instinctive tendency to maintain group cohesion due to internal and external environmental factors not taken into account in a simulated experiment (Lehmann et al., 2007). Additionally, simulated data may not be reflected or supported in the observational data collected. This experimental technique has been frequently employed when studying primate groups over a relatively short amount of time and in situations where the temporal changes to group dynamics cannot be conclusively examined.

It is also clear that a large proportion of the literature regarding social networks in primates is devoted to social interactions among laboratory housed or wild macaque groups (e.g. Flack et al., 2006; Voelkl and Noe, 2008; Franz and Nunn, 2009; Sueur et al., 2010). However, the species-specific social style of macaques has meant that authors using SNA to quantify a social network have yielded limited agreement and differing housing conditions have produced varying results in regard to the frequency of interaction between social partners. In contrast, chimpanzees are excellent study subjects for SNA as they are universally associated with having highly complex social lives. However, as yet there have been relatively few studies to apply modern SNA techniques to quantify measures of relationship quality or the asymmetrical nature of primate social relationships.

It is important to note that in June 2014 (one month before data collection had commenced) an infant chimpanzee, “Velu”, was born into the study group. Velu and his mother, Heleen, were kept segregated from the rest of the study group for a period of 8 weeks whilst the zookeepers at the ‘Budongo Trail’ monitored the health and progress of both individuals. Prior to their full integration, both mother and infant chimpanzee were held in an off-show area of the ‘Budongo Trail’, which allowed for visual but non-physical contact with other group members. The birth of a chimpanzee directly before the study period resulted in a unique situation whereby social interaction data was able to be collected for 12 months following the birth and integration of an infant chimpanzee and mother in captivity, which could then be compared with a previous study, outlined in Chapter 3 of this thesis, prior to the birth.

The purpose of this study was multi-faceted, focusing primarily on four keys areas of research:

1) To identify if social partners with more valuable, compatible and secure relationships are more connected within the entire network.

2) To test whether distinct subgroups existed within the chimpanzee’s social networks, in line with the naturally occurring fission-fusion dynamics of chimpanzee social structure.

3) To observe temporal changes to chimpanzee social group dynamics and the causes of these behavioural group shifts.

4) To investigate the influence of relationship asymmetry by identifying if more symmetrical relationships equate to stronger associations within the network.