The importance of partner inclusion criteria for understanding drivers of social variation among individuals: Data from blue monkeys

Individuals in social species vary in their sociality, allowing inferences about how social ties influence fitness, but also raising questions about the drivers of observed variation in sociality. We examined how sociodemographic, social, and individual factors were associated with inter-individual variation in the sociality of wild blue monkeys (Cercopithecus mitis stuhlmanni), and how these associations are affected by the set of partners included in the analysis. Using data from focal follows of adult females in 12 groups over 13 years, we measured individual variation in five measures of sociality (number of ties, number of strong ties, number of weak ties, total tie strength, and evenness of tie strength), binned in annual periods. We used linear mixed models to assess the effects of sociodemographic, social, and individual factors on these five measures. We repeated this analysis while limiting observations to peer partners (other adult females) or to non-peer partners (juveniles and adult males) to evaluate how including different sets of partners in the analysis affected the results. Three data files were used in these analyses, each the result of applying different criteria for who was included as a potential partner for the adult female subjects. 

Methods
These data were collected as part of a long-term observation program, which began in 1979, focusing on individually identified arboreal blue monkeys (Cercopithecus mitis stuhlmanni) inhabiting the Kakamega Forest, a rain forest in western Kenya (0°19 N, 34°52 E, 1,580 m., ca. 2000 mm rainfall annually. The data represent a 13-year period (September 2006-August 2019 inclusive), during which the number of study groups increased from 4 to 8 via natural group fissions. We considered parent and daughter groups as distinct units in our analyses, which included 4-8 groups at any one time (N=12 total), ranging in size from 6-76 individuals. Members of all study groups were identifiable by sight using natural physical variation, and their life histories (date of birth, age at each offspring’s birth for females) were known from long-term near-daily monitoring.

Data on social behavior were collected during focal animal samples of adult (parous) females in all study groups. We first trained the observer team to correctly identify all group members (they had to do this without error five times in a row) and to achieve >95% agreement in focal sample scoring. Observers chose focal subjects to prioritize an equal accumulation of observation time across groups, individuals, and time of day (morning, midday, and afternoon hours). Focal follows were scheduled to last 30 min, but subjects occasionally went out of sight. Observers searched for them for up to 15 minutes and continued the follow if they found the focal subject, counting only in-sight time. We aborted and discarded follows that lasted less than 20 minutes.

During each follow, the observer made instantaneous records at 1-minute intervals of the focal subject’s activity, identifying any social partners. We examined affiliative social contact in this report, which included both giving or receiving grooming and sitting in body contact. We defined social partners as those individuals who were grooming, being groomed by, or sitting in contact with the focal subject. A focal subject could have multiple social partners at once.

Observers also monitored agonistic behavior, both during focal follows and whenever it was witnessed at other times. Agonistic interactions included aggressive and submissive behavioral elements, as well as approach-avoid interactions. These interactions served as input for calculating dominance ranks, using the I&SI method as implemented in Domicalc (Schmidt and DeVries 2013).