Data from: Culling-induced perturbation of social networks of wild geese reinforces rather than disrupts associations among survivors
Cite this dataset
Royle, Nick et al. (2023). Data from: Culling-induced perturbation of social networks of wild geese reinforces rather than disrupts associations among survivors [Dataset]. Dryad. https://doi.org/10.5061/dryad.h70rxwdq8
Abstract
Wildlife populations may be the subject of management interventions for disease control that can have unintended, counterproductive effects. Social structure exerts a strong influence over infectious disease transmission in addition to other characteristics of populations such as size and density that are the primary target for disease control. Social network approaches have been widely used to understand disease transmission in wildlife but rarely in the context of perturbations, such as culling, despite the likely impacts of such disturbance on social structure and disease dynamics. Here we present a ‘removal’ study of a free-living population of resident Canada geese Branta canadensis, a highly social species that is frequently managed by culling and can carry pathogens relevant to human and domestic animal health. We quantified social network structure and spatial behaviour before and after controlled culling of individuals during the summer moult. Culling did not substantially increase individual social connectivity. Individuals that moulted at cull sites or were formerly strongly associated with removed birds were more likely to strengthen and maintain any surviving existing associations while also forming new associations. However, the establishment of new associations was largely compensatory (with only small increases in the number and strength of connections) and occurred locally. Synthesis & applications: geese that survived the cull responded by strengthening existing social relationships and forming new, compensatory relationships with birds local to them in the network. In the short-term such compensatory adjustments to patterns of association in response to culling could facilitate pathogen transmission. But in the longer term, controlled culling of geese is unlikely to strongly influence pathogen spread and may even slow transmission into new social clusters by reducing wider mixing. When managing wildlife for disease control, in addition to changes in social network structure the prevalence of infection at the time of the cull and the mode of transmission (e.g., direct versus environmental) will also be critical determinants of disease transmission risk in perturbed populations of geese and other wild animals.
README: Culling-induced perturbation of social networks of wild geese reinforces rather than disrupts associations among survivors
https://doi.org/10.5061/dryad.h70rxwdq8
Contains the datasets and R code necessary to replicate the analyses in the paper
Description of the data and file structure
Four files:
1. 'CanadaGoosePerturbationScript5.csv' - The R code required to replicate the analysis and data visualisation. It is fully annotated code.
2. 'RS_CWP.csv' - raw resightings data used in social network construction
3. 'Raw RS.csv'- adapted version of the resightings data to facilitate social network construction
4. 'range model df.csv' - derived ranging data used for the spatial analyses in the paper
README for 'RS_CWP.csv'
Eight columns:
$Event - unique identifier for each group/grouping event observed (numeric)
$Site - Location of Observation
$E or W - Whether in the east, central or west part of the Cotswold Water Park. Contains NAs for locations outside the Cotswold Water Park.
$Date - Date of observation
$Collar - unique individual ID (2 letter code)
$BTO - unique individual ID (BTO ringing code)
$Sex - Male/Female when known (otherwise NA). A '?' mark represents a bird that is probably but not confirmed to be the assigned sex.
$Obs - unique identifier for each observation (numeric)
README for 'Raw RS.csv'
Four columns:
$event - unique identifier for each group/grouping event observed (numeric)
$collar - unique individual ID (2 letter code)
$precull - binary indicator of whether group was observed before cull (0/1)
$postcull - binary indicator of whether group was observed after cull (0/1)
README for 'range model df.csv'
Nine columns:
$Collar - unique individual ID (2 letter code)
$Sex - Male/Female when known (otherwise NA)
$mean_dist_cull - mean moult site distance from cull sites (in kilometres) (NA indicates moult site unknown)
$min_dist_cull - minimum moult site distance from cull sites (in kilometres) (NA indicates moult site unknown)
$cullsite2013 - did the individual moult at a cull site in 2013? (Y/N)
$HR50yr2 - home range area in year 2 calculated from 50% kernel density (NA indicates insufficient information to calculate home range size)
$HR95yr2 - home range area in year 2 calculated from 95% kernel density (NA indicates insufficient information to calculate home range size)
$HR50yr1 - home range area in year 1 calculated from 50% kernel density (NA indicates insufficient information to calculate home range size)
$HR95yr1 - home range area in year 1 calculated from 95% kernel density (NA indicates insufficient information to calculate home range size)
Sharing/Access information
These files are not shared elsewhere. No additional files are required to replicate the analysis.
README for 'RS_CWP.csv'
Raw observational data.
README for 'Raw RS.csv'
Resightings data are raw observations. The $precull and $postcull columns are derived from observation dates.
README for 'range model df.csv'
Home range areas were calculated from raw resightings data (not provided with spatial locations here).
Code/Software
Datasets are used in the R code provided in this repository
Funding
University of Exeter
Animal and Plant Health Agency