Discussion
Our analyses revealed a highly dynamic social network between core units in the C. a. ruwenzorii multi-level society. Contrary to our first prediction, the clan tier of organization was not entirely stable over time. Given the importance of kinship in structuring primate groups (Silk, 2001; 2002), we predicted that this may extend to the clan tier of organization, with core units containing related individuals preferentially clustering (e.g., Papio hamadryas, Theropithecus gelada , Colmenares, 2004; Loxodonta africana , Wittemyer et al., 2005), leading to stable clans within the band. However, we observed two major changes in our study band. First, an all-male unit formed when seven males left the largest core unit and began to range in loose association with the two clans. Second, one core unit moved between clans after dispersal events involving five males. These changes show that clans do shift in core unit composition over time, though not frequently, and that male dispersals can cause this variation.
We found support for our overall hypothesis as both ecological and social variables were important in determining the amount of association among core units. Association patterns fluctuated at both the node and network level, with the largest changes correlating to seasonal shifts in fruit availability. As predicted, core units were more likely to associate, and did so with a larger number of other core units, when fruits were abundant, suggesting that food competition limits operational group size when fruits are scarce. This increase in association appeared to facilitate male dispersals between core units in the band, thus creating a seasonal dispersal pattern. Our analysis of association indices following each male dispersal event within the band revealed that male transfers promote higher than expected dyadic associations between interacting core units in the short-term (1-2 months post-dispersal).
Many species alter their behaviour in response to changing climatic and resource conditions (Candolin & Wong, 2012). Our results show thatC. a. ruwenzorii is no exception. Similar to studies done on other primates (Cercocebus torquatus, Dolado, Cooke, & Beltran, 2016; Rhinopithecus bieti, Ren, Li, Garber, & Li, 2012;Papio hamadryas , Schreier & Swedell, 2012b; Pongo pygmaeus, Sugardjito, Te Boekhorst, & van Hooff, 1987) and non-primates (Orcinus orca, Foster et al., 2012; Loxodonta africana , Wittemyer et al., 2005), we found that C. a. ruwenzorii units increase their association levels during times of peak food availability. Food competition decreases when resources are abundant, allowing animals to aggregate if they choose, which provides benefits for predator avoidance (Hamilton, 1971; Sueur et al., 2011). Species living in a multi-level society benefit from this ability to alter overall group size in response to external pressures (Grueter & van Schaik, 2009). For C. a. ruwenzorii , enlarged group size may even mean an expansion of the microhabitats they are willing to take advantage of. Adams and Teichroeb (2020) found that at Nabugabo, where predation risk is greatest near the ground, C. a. ruwenzorii were willing to come lower in the canopy to find food when more core units were clustered together and predation risk was lessened. The analyses presented here suggest that this niche expansion may occur more often in resource rich seasons when core units are able aggregate.
Although we find correlations between seasonal fruit availability, association patterns and male dispersal, it is important to acknowledge that we cannot determine cause and effect between these phenomena. While we posit that higher fruit availability leads to more clustering among core units, which facilitates male dispersal, it is possible that males prospect more during seasons of food abundance and that male prospecting behaviour drives the observed changes in association patterns. Seasonal dispersal patterns are found in many species (Likicker & Stenseth, 1992) but in most cases, this pattern emerges due to seasonal breeding (e.g., Presbytis entellus, Borries, 2000; Suricata suricatta , Mares et al., 2014; Chlorocebus pygerythrus, Young et al., 2019; Rhinopithecus roxellana, Yao et al., 2011). Breeding is not typically seasonal in black-and-white colobus monkeys (Fashing, 2011) and we do not have data showing seasonal breeding at Nabugabo. Alternatively, it is sometimes advantageous for animals to time dispersal to coincide with high food availability because it allows them to compensate for increased travel, potentially in unfamiliar areas (Pusey & Packer, 1987; Isbell & Van Vuren, 1996). This explanation is unlikely to apply in a multi-level society like that seen in C. a. ruwenzorii as all the core units in our band share a home range (Stead & Teichroeb, 2019). Consequently, male dispersal between units does not require extra travel or moving into a new, unfamiliar area. We suggest that the best explanation for the seasonal pattern of male dispersal that we observe in C. a. ruwenzorii is the opportunity for prospecting provided by greater core unit clustering due to high resource availability. The proximity of so many other core units allows males to assess their composition (i.e., sex ratio) as well as the competitive ability of the males there (Teichroeb et al., 2020), potentially influencing their decision to disperse. In primates, it is common for dispersal to occur during intergroup encounters (e.g.,Macaca mulatta , Boelkins & Wilson, 1972; Erythrocebus patas , Rogers & Chism, 2009; Gorilla beringei , Sicotte, 1993;Rhinopithecus roxellana , Yao et al., 2011) or to groups where prospecting has previously been directed (e.g., Colobus vellerosus , Teichroeb, Wikberg, & Sicotte, 2011).
The persistence of high association indices post-dispersal for core units that have males transfer between them may be a result of the continued bonds between individuals that persist even after the dispersal has taken place. The dispersing individual(s) likely still have ties in their former (sometimes natal) unit, which may contain many kin. However, over time, we see a slow decrease of association between the units individual(s) dispersed to and from, back to the baseline association levels that they had prior to the dispersal event. This decrease in association may be explained by the further integration of the dispersing individual(s) into their new unit, and/or the seasonal decrease in fruit availability, and subsequent increase in food competition. Future research examining how male-male genetic and social relationships impact association patterns over short and long time periods will provide insights into the ways that kinship structures core unit association in tandem with ecological and social factors (e.g., Snyder-Mackler et al., 2014).
To conclude, our results show that in the dynamic social network of Rwenzori Angolan colobus monkeys, core units behaviourally adapt to changing ecological conditions by altering their association patterns. Doing so has cascading effects on the composition of core units, and structure of both the clan and band tiers in this multi-level society. This type of behavioural flexibility allows animals to thrive in dynamic environments (Candolin & Wong, 2012). Our study provides a deeper understanding of the mechanisms underlying the formation of complex multi-level social organizations and some insight into the intertwined temporal effects of ecological and social variables.