Data for: Intraguild competition mediates human avoidance in an endangered African large carnivore
Data files
Apr 30, 2025 version files 135.01 MB
-
datasets_archive.RData
134.95 MB
-
random_sample.csv
52.06 KB
-
README.md
3.67 KB
Abstract
The anthropogenic refuge (i.e., human shield) effect is a well-documented consequence of anthropogenic disturbance, whereby prey or subordinate competitors use human-dominated areas to avoid dominant predators. While this effect has been widely demonstrated for prey and, to a lesser extent, mesopredator species, it remains unknown whether it occurs among sympatric species within a top predator guild such as large carnivores. Moreover, there is limited understanding of how intraguild competition may mediate carnivore responses to human disturbance. Here, we examined the spatial behaviour of two sympatric large carnivore species exhibiting intense intraguild competition – lions and African wild dogs – to investigate whether intraguild competition mediates large carnivore responses to human disturbance in support of the anthropogenic refuge hypothesis. We found that lions consistently avoided human-dominated areas. Similarly, we found that wild dogs generally avoided human-dominated areas, but this pattern reversed when in areas of high lion-encounter risk. Such results reveal that the ecology of fear, and in particular the anthropogenic refuge hypothesis, extends beyond traditional predator-prey relationships to competing top predators within a large carnivore guild. Further, our study suggests that intraguild variation in responses to human disturbance may be an emerging niche axis that facilitates coexistence between competing species within the same guild.
https://doi.org/10.5061/dryad.8w9ghx3xx
Description of the data and file structure
This study employs resource selection and step selection analyses on lions (Panthera leo) and African wild dogs (Lycaon pictus) to investigate whether intraguild competition mediates large carnivore responses to human disturbance. Data represent used and simulated locations and their environmental covariates for each species collected from GPS collar data.
Files and variables
File: random_sample.csv
Description: covariates of a random sample of 1,000 locations within the study area for performing a linear regression between lion encounter risk and distance to village
Variables
- point_id: unique ID for each random point
- lion_risk_camtrap: lion encounter risk derived from camera traps
- lion_risk_z: z-scored (standardized) value of lion_risk_camtrap
- village_dist: distance in meters to nearest village
- village_dist_z: z-scored (standardized) value of village_dist
File: datasets_archive.RData
Description: RDATA file type that contains 3 dataset objects when loaded in R
Dataset 1: den_data_archive
Variables
- case_: logical presence (TRUE) or pseudoabsence (FALSE)
- pack: wild dog social group ID
- lion_risk_camtrap: lion encounter risk derived from camera traps
- lion_risk_z: z-scored (standardized) value of lion_risk_camtrap
- village_dist: distance in meters to nearest village
- village_dist_log: natural logarithm of village_dist
- village_dist_log_z: z-scored (standardized) value of village_dist_log
Dataset 2: lion_data_archive
Description: covariates of lion step selection and pseudoabsence locations for performing a step selection analysis
Variables
- id: individual animal ID
- case_: logical presence (TRUE) or pseudoabsence (FALSE)
- step_id_: unique ID for the used step and associated pseudoabsences
- sl_: step length in meters
- log_sl: log of sl_
- ta_: turning angle in radians
- cos_ta: cosine of ta_
- village_dist: distance in meters to nearest village
- village_dist_log: natural logarithm of village_dist
- village_dist_log_z: z-scored (standardized) value of village_dist_log
Dataset 3: dog_data_archive
Description: covariates of African wild dog step selection and pseudoabsence locations for performing a step selection analysis
Variables
- id: individual animal ID
- case_: logical presence (TRUE) or pseudoabsence (FALSE)
- step_id_: unique ID for the used step and associated pseudoabsences
- sl_: step length in meters
- log_sl: log of sl_
- ta_: turning angle in radians
- cos_ta: cosine of ta_
- village_dist: distance in meters to nearest village
- village_dist_log: natural logarithm of village_dist
- village_dist_log_z: z-scored (standardized) value of village_dist_log
- lion_risk_camtrap: lion encounter risk derived from camera traps
- lion_risk_z: z-scored (standardized) value of lion_risk_camtrap
Code/software
Code files:
-
rsf_code_archive.R: code to run resource selection analysis on African wild dog den sites. Takes as input den_rsf_archive.csv.
-
issa_code_archive.R: code to run step selection analysis on African wild dog and lion GPS data. Takes as input lion_issa_archive.csv and dog_issa_archive.csv.
-
linear_regression_archive.R: code to run linear regression analysis between lion encounter risk and distance to village for 1,000 random locations within the study area. Takes as input random_sample.csv.
