Age Differences in the responses of vervet monkeys to terrestrial alarm calls
Description
Here, we explore the development of alarm call responses in wild vervet monkeys (Chlorocebus pygerythrus) living at the Samara Game Reserve, South Africa. Vervet monkeys at this site live in relatively large social groups, and adult responses to alarms have previously shown less uniformity than in other sites where group sizes are smaller. We presented monkeys played back recordings of terrestrial alarm calls produced by individuals of different age/sex classes. We then video-recorded and scored the responses of receivers along an ordinal maturity scale. We used a mixed-effects ordinal logistic regression model within a Bayesian framework to explore how response intensity is affected by the age/sex of the caller, and the age of receivers. Our analysis showed that younger monkeys (<2yrs) exhibit strong evasive responses to call stimuli, regardless of the age/sex class of the caller. The intensity of these responses decreases with age, with responses to non-alarm calls decreasing earlier in development compared to alarm calls. Adult responses to alarms in this population are equally as likely to be characterized by a general increase in vigilance as they are to consist of an evasive response. We suggest that responses in younger individuals at Samara are mediated by a generalized startle responses to loud, plosive noises. Development of more specific responses throughout ontogeny is likely to be dependent on learning from adult models, whose milder responses reflect specific to local socioecological conditions at Samara. Raw data and R code (R Markdown) included in upload
Files
Steps to reproduce
We collected data from three habituated groups of vervet monkeys occupying adjacent, overlapping territories in semi-arid riverine Acacia woodland at the Samara Game Reserve, Eastern Cape Province, South Africa, from May 2016 to May 2017. We conducted 36 playback trials to investigate the effect of caller and receiver age/sex class on responses to alarm calls. We played back the alarm calls given to terrestrial predators by either an adult male, an adult female, or a juvenile. We also ran five control trials by broadcasting male “waa” calls, which, while produced by subordinate adult males during agonistic encounters. We used a Marantz PMD661 digital recorder connected to a MiniVox Lite public address 160 30-watt speaker to present playback stimuli. We placed the speaker under natural cover ahead of a moving group of individuals. During each trial, trained field assistants with video cameras followed and filmed monkeys of known age/sex and identity. An assistant trained in focal data collection but unaware of the study’s goals and sampling protocols transcribed all behaviours and behavioural changes over the duration of each video. We used the video transcriptions to identify one of three response categories, ordered from weakest to most intense: i. No response, ii. Vigilance, iii. Overt. To determine whether response intensity was dependent on the age of the receiver and the age of the caller, we first constructed a mixed effects ordinal logistic regression model, with a cumulative logit link function. We ran the model within a Bayesian framework, using the “brms” package in R 4.0.3, specifying four chains and 2000 iterations, and setting weakly informative priors centred 10 207 on zero (i.e., normal (0,1)). The response variable was the ordinal, three-category strength of response. We included the 4-level categorical variable stimulus type (Alarm call produced by i. juvenile, ii. adult female, iii. adult male, and iv. Control call), and receiver age class as fixed effects. To assess whether the effects of stimulus type were dependant on the age class of the receiver , we specified an interaction between stimulus type and receiver age. We also included terrain type (on the ground, on the ground under cover, in a tree or shrub), and the relative distance between the playback speaker and receiver (3 to 10m, 11 to 20m, 21 to 30m) as fixed effects. To control for the possibility that a receiver's response was in some way affected by the number of animals in the vicinity, we included the number of neighbours as a fixed effect. Receiver and caller identity were specified as crossed random effects.