Data from: The dynamics of dominance in a ‘despotic’ society
Data files
Oct 29, 2024 version files 146.57 KB
-
rank_dynamics_data.csv
113.31 KB
-
rank_dynamics.Rmd
32.06 KB
-
README.md
1.20 KB
Apr 01, 2025 version files 129.68 KB
-
rank_dynamics.csv
97.79 KB
-
rank_dynamics.Rmd
30.30 KB
-
README.md
1.60 KB
Abstract
Dominance hierarchies are a key feature in the dynamics of animal social groups, playing a crucial role in fostering group stability. Despite often being viewed as static, persistent linear structures, hierarchies are fundamentally dynamic and can change over time due to ecological conditions, demographic changes and ontogenetic development. There are numerous methods used to construct hierarchies and quantify individual dominance rank, however, methods to capture the dynamics of a hierarchy across time have only recently been developed. As such, relatively little is known about the longitudinal hierarchy dynamics in many social species, including non-human primates and the timescale at which these hierarchy dynamics play out. Here we consider the longitudinal hierarchy dynamics across a four-year period in a large group of rhesus macaques. We investigated group- and individual-level predictors of active rank dynamics, or dynamics that arise from rank reversals. We found that, despite rhesus macaques being considered to have relatively stable hierarchies, there was significant active rank mobility in both males and females, even in the face of limited resource competition. Female rank change was not solely driven by matrilineal structure or demographic processes and females may also opportunistically ascend in rank. Further, we found strong links between rank certainty and hierarchy dynamics with periods of high hierarchy instability associated with low mean dominance certainty. Lastly, we found limited evidence of associations between periods of high active rank dynamics and social global network structure. This suggests more localised dynamics during hierarchy instability are at play rather than widescale network reorganisation. Together, these results stress the importance of considering social context in rank dynamics, illustrate the dynamic nature of macaque dominance rank and further highlight the opportunistic nature of the species.
https://doi.org/10.5061/dryad.1ns1rn935
Description of the data and file structure
All data used are contained in the rank_dynamics_data.csv file. This file contains rank dynamics calculations as well as individual- and group-level predictors.
Variables:
- Period: study period number
- Id: animal ID
- Sex: male/female
- Age: animal age in years
- Group size: number of individuals in the group in that time period
- delta.active: active rank dynamics
- rank.decile
- dc: dominance certainty
- sd: dominance certainty standard deviation
- centralization: network-level centralization
- average_degree: average degree of network
- delta.active.scaled: active rank dynamics scaled by group size (for fig. 2 only)
- delta.passive.scaled: passive rank dynamics scaled by group size (for fig. 2 only)
Missing values (NA) in the delta.active, delta.active.scaled and delta.passive.scaled columns indicate periods where no hierarchy dynamics can be calculated (i.e. an individual’s first study period). These are filtered out during analysis but contained in the dataset.
Code/Software
All analytical code is contained in the “rank dynamics.Rmd” file.
Version changes
April-2025: Rank dynamics recalculated to remove 4 individuals from their final data bin to rectify inaccurate inclusion. Minor variable naming fixes for clarity in “rank dynamics.Rmd” file. “delta.active” column added to “rank_dynamics_data.csv” which was excluded in previous version.
Data were collected on a single social group (formed in 2003) of rhesus macaques (Macaca mulatta) housed in a 0.5-acre outdoor corral from March 2016 to November 2019. Behavioural observations were focused on all animals 3 years and involved the collection of aggressive, submissive, and affiliative behaviour. Dominanace ranks were caluclated in 90-day blocks and longitudinal hierarchy dynamics (Strauss and Holekamp 2019) were calculated across 14 time periods. Agonistic data were used to construct social networks for each time point, and average degree centrality and centralisation were extracted for each time period.