Dominance_scripts_repository

This repository contains basic scripts to calculate Elo rank using the EloRating package by Neumann & Kulik (10.32614/CRAN.package.EloRating), using the agonistic interactions recorded at IVP.

Several conditions have to be met:

1. Interactions are between two individuals, and no support was given to any of the two individuals.
2. The interaction has to have a clear winner. This is generally decided by whichever individual retreats, flees or shows other clear subordinate behaviours. Otherwise, the individual that is clearly most aggressive is chosen as the winner. For an overview of the behaviours taken into consideration, see line 79-83 of Create_WinnerLoser.R.
3. The interactions should generally be filtered by age and sex class, i.e., only adult males vs adult males, or only adult females vs adult females. A male is considered an adult after his first dispersal. A female (in this code) is considered an adult when she reaches 3 years of age.

To calculate dominance you need the following data:

1. Ad lib agonistic interactions.
2. Focal agonistic interactions (optional).
3. The life history file.

Before calculating rank, you need to create:

1. Presence/absence matrices for all applicable groups (using life history)
2. WinnerLoser file showing the winner and loser of each clear agonistic interactions (using FinalAgonistic, which uses the raw agonistic data)

There are several ways you can calculate rank. For example, you might be interested in a male's rank on each day he alarm calls. In that case, it might be best to calculate his rank on that specific day, ensuring you use a set amount of data (for example, the three previous months). If this is what you're after, please see the Elo_rating.R script. In this script, we calculate the elo rating on each day an individual occurs in a dataset. All you need for this is a dataset with a column ID, Group and Date. When you apply this method, make sure you correct for potential autocorrelation in your models (since we use similar data to calculate the rank of day 2010-06-24 as on day 2010-07-24 -- two months of overlap). Another method might be that you're interested in the average rank of females over the years. In that case, please refer to Monthly_elo.R. In this script, we calculate the elo rating consistently each month, again using three months of previous data. All you need for this script is the winnerloser file, presence files and a window of time that you're interested in. Again, depending on how you use this data, keep in mind that there might be overlap in data in how rank is calculated, in which case you should again correct for autocorrelation.