Data from: Spatial strategies in non-territorial societies: How feral horses maintain boundaries with other groups
Data files
Nov 13, 2025 version files 793.22 KB
-
overlap_info.xlsx
275.70 KB
-
rawdata_position2017.xlsx
506.94 KB
-
rcode.Rmd
6.31 KB
-
README.md
4.27 KB
Abstract
Encounters between competitive individuals or groups are common in social animals and can involve costly aggression; thus, animals often employ strategies to minimise direct conflict. However, research on whether and how animals adjust their group spatial structure when they encounter or spatially co-occur with a different group remains limited. We investigated how non-territorial units in feral horse multilevel societies manage spatial encounters with neighbouring units. We observed 25 reproductive units in northern Portugal, using drones, and employed statistical analyses to quantify spatial dynamics. We found that horses actively adjusted their spatial formations depending on the proximity of other units. Units became more circular and cohesive when other units approached, but when extremely close, they elongated their shapes to avoid boundary crossings. These adjustments indicate that horses maintain flexible unit boundaries to prevent mixing with other units. A notable exception was a particular pair of units that frequently crossed boundaries and intermixed, representing a unique social level within the horse multilevel society. Overall, our findings indicate that feral horses respond adaptively to inter-unit encounters by balancing the benefits of aggregation with the need to reduce direct overlap, providing new insights into the spatial organisation of non-territorial societies.
Dataset DOI: 10.5061/dryad.m905qfvft
Description of the data and file structure
The positional data of feral horses was acquired in Serra D'Arga, Portugal, in 2017. We used drones to obtain the images of the horses and identified the individuals from the aerial photographs.
Files and variables
File: rawdata_position2017.xlsx
Description: the positional data of feral horses
Variables
- ID: individual ID
- group: group ID
- age: age class. 'foal' as 0 year old, 'young' as 1-4 years old, adult as 5 years old or older.
- SEX: sex. 'NA' means sex was not identified.
- code: the observation code. 60102 represents 6 = June, 01 = 1st, 02 = 2nd observation of the day.
- zone: whether the individual was observed at zone 1 or zone 2 in the observation site (see Figure 1 in the main manuscript).
- xcoord, ycoord: the geographical coordinates of the individual. The coordinate system is WGS 84 UTM zone 29N
File: overlap_info.xlsx
Description: the information of the possible overlap events
Variables
'overlap' sheet : the code and group ID when possible overlap events occurred
- uzukobe: "1" = the pair is Kobe-Uzumasa, "0" = the pair is not Kobe-Uzumasa
- code2: it consist of 5 digit to specify the observation time + unit name. For example, 60201Asaka means Asaka unit observed at the 1st observation on June 2nd.
'radius_nofoal' sheet : the centre and the radius of units
- code: same as code2 in the 'overlap' sheet
- radius: the radius of the unit (m)
- xcenter, ycenter: the coordinates of the unit center. The coordinate system is WGS 84 UTM zone 29N
- code2: the observation code. 60102 represents 6 = June, 01 = 1st, 02 = 2nd observation of the day.
- date: the date of observation. 601 represents 6 = June, 01 = 1st.
- group: unit name
File: rcode.Rmd
Description: R code for randomisation analysis. This R script runs spatial randomization tests to see how often two animal groups’ ranges (especially “Uzumasa” and “Kobe”) overlap by chance. It does this by shuffling group labels, randomizing positions, or rotating their locations, then checking polygon intersections 1000 times.
Condition 1: Shuffling positions among members of the two units; we randomly shuffled the positions of individuals within two units. As males and females tend to occupy different positions within a group, the positions of the males were fixed, and only the positions of the other members were interchanged. In the case of a single-male unit, males were located at the periphery of the unit, and in the case of a multiple male unit, subordinate males tended to be located at the periphery, with dominant males inside. If the observed data under this condition did not significantly deviate from the randomised data, this would indicate no intra- or inter-unit differentiation in inter-individual distances.
Condition 2: Temporal permutation without changing unit centres; the internal arrangement of individuals in each unit was replaced with that from a different, randomly selected observation time when the units were not involved in a possible overlap event. The central position of the unit is preserved. This condition tested whether horses altered their positions and intra-unit distance in response to the proximity or arrangement of other units. If the units maintained their positions regardless of the distance from other units, the observed data would match the randomised data.
Condition 3: Rotation around the centre; each unit was rotated around its centre at a random angle (0–2π radians) without changing its centre-to-centre distance to other units. As inter-individual distances within a unit and distance between two-unit centres were preserved, this condition facilitated assessment of whether the overall shape, not intra-unit distance, of a unit changed depending on its proximity to the other unit.
Code/software
MS Excel, R studio used for the analysis
Access information
Other publicly accessible locations of the data:
- NA
Data was derived from the following sources:
- NA