Alexander J. Pritchard a, c

Brianne A. Beisner b

Amy Nathman a

Brenda McCowan a, c

a: California National Primate Research Center, University of California Davis, Davis, CA, USA;

b: Emory National Primate Research Center Field Station, Division of Animal Resources, Emory University, Lawrenceville, GA, USA;

c: Department of Population Health & Reproduction, School of Veterinary Medicine, University of California Davis, Davis, CA, USA

Article

https://doi.org/10.1080/10888705.2024.2303679

Data Dryad repository

https://doi.org/10.5061/dryad.g79cnp5wk

Description of the data and file structure

Data are contained in the NMKO_Data_List.Rdata file. This can be read into R and preprocessed using Run-This-First_NMKO_Extract-Dataframes.R and will assemble 26 R objects. These objects are as follows:

Alf_NM_Codes = a dataframe with columns, in order, identifying the groups with a code, the alpha male ID (alfID) and the removed natal males' ID (nmID) for each of the groups.
NMKO_Conflict = a dataframe of all of the conflict reversals that go against the hierarchy during baseline and PKO. With columns, in order, showing: group code, subject ID, project period (baseline or perturbation), count of initiations of reversal conflicts, days of observation, rates of conflict reversals, ordinal rank as a proportion (RankProp), 'YOB' for year of birth, and 'Age' in years.

The remaining 24 R objects are lists of five weighted edgelists with each embedded dataframe representing each of the five groups with Recipient and Initiator, as well as count of each behavior during Baseline and PKO (Treatment). Lists are named with the group identifier (A-E). Lists are also labelled by a code that represents what each list contains, with BEHAVIOR_Period_Sex. Possible combinations include one of four behavior types ('SBT' for silent-bared-teeth, 'Grm' for Grooming, 'Disp' for displacements, 'DAG' for directed aggression), across the two periods ('Base' for Baseline; PKO for Post-knockout Perturbation), and three different sex combinations ('ALL' for all adult males and females, "F" for females-only, "M" for males-only).

NMKO_Comparing_Metrics.csv is a result file comparing our group-level hierarchical metrics. Columns include, in order, group identifier (A-E), project Period ('Baseline' for Baseline; PKO for Post-knockout Perturbation), three different sex combinations under Net ('ALL' for all adult males and females, "F" for females-only, "M" for males-only), Metrics of hierarchical structure ('Prop' for proportion of nodes in cyclic interactions, 'ttri' a metric of triangle transitivity, 'h' is the h' index for linearity of the hierarchical structure, 'GRC' is the Global Reach Centrality metric).

Code/Software

You will need R for most of this code, and the following packages: igraph, sna, EloRating, tidyr, ggpubr, brms, bayesplot, ggplot2
In Python, you will need: Networkx, pandas, and base functions (csv, and random).

After importing the data into R using the Run-This-First... file, users can run several analyses:
Analysis_NMKO_NM-vs-Alf-Jaccard.R executes code to extract the Jaccard similarity between grooming, displacement, and directed aggression ego-networks for the alpha and natal males.
Analysis_NMKO_Authority-Prop-ttri-h-getGRC.R executes code to extract authority scores for the alpha and natal males based on displacement networks for male-only and mixed-sex groupings. Group-level metrics are also calculated for displacement networks, including: proportion of nodes in cycles, h' index, triangle transitivity, and then data are exported for use in Python to calculate global reach centrality (GRC - see below).
Analysis_NMKO_Reversals-Model.R includes code to execute a Bayesian Regression Model using Stan for the assessing whether the initiation and count of reversals increased during treatment, relative to baseline.
NMKO_GRC_and_Permutation.txt includes Python code to extract and permute GRC values from the displacement and pSBT networks. Executing this will create multiple csv files that will be read into the following file.
Plots_NMKO_Comparing-Hierarchy-Metrics.R will make plots based on the above output. For user convenience, we have included our output in the file: NMKO_Comparing_Metrics.csv. We have included Permutation .csv output files as a zip (Permutation_NMKO_Results.zip). Users can replicate these output by executing all the above code, but will need to manually add the GRC values to the output csv from Analysis_NMKO_Authority-Prop-ttri-h-getGRC.R