Data from: Seasonal dynamics of resource availability and human presence shape habitat use of large carnivores in Chad

Understanding how seasonal changes in resources and human activity influence large carnivore space use is critical for effective conservation, yet such information is scarce for Central African savanna ecosystems. This study used occupancy modelling with data from camera traps deployed during the dry and wet seasons in Zakouma National Park (NP), Chad, to assess seasonal patterns of large carnivore occurrence and their environmental drivers. Five large carnivore species were detected: lion, leopard, cheetah, spotted hyaena, and striped hyaena. Prey availability emerged as the primary driver of carnivore space use in the dry season. Detections declined markedly in the wet season; occupancy remained similar across seasons for lions and leopards, whereas other species exhibited reduced occupancy in the wet season. Our results indicate that Zakouma NP functions as a critical refuge during the dry season, when anthropogenic pressure intensifies in community lands due to the movement of transhumant pastoralists. During the wet season, reduced human presence in the broader landscape, combined with decreasing prey availability and flooding that renders parts of Zakouma NP unsuitable, allows several species to expand their range beyond the park. Within the Greater Zakouma Ecosystem, these seasonal dynamics highlight the importance of landscape-level conservation approaches that account for seasonal shifts in resource availability and human land use to support large carnivore persistence. ## Description of the data and file structure Data were collected through a systematic camera-trap survey designed to quantify seasonal variation in large carnivore occurrence and space use in Zakouma National Park, Chad. Camera traps were deployed at a network of fixed sampling sites during both the dry and wet seasons of 2020, using a standardized placement protocol to ensure comparability across seasons. Each camera station operated continuously throughout its respective sampling period and recorded date- and time-stamped detections of terrestrial mammals. The survey targeted large carnivores, including lions, leopards, cheetahs, spotted hyaenas, and striped hyaenas, and was structured to generate detection histories suitable for occupancy modeling. Site-level environmental and anthropogenic covariates were compiled for each camera location to characterize seasonal variation in prey availability, flooding, and human activity. These data were used to evaluate how seasonal changes in resources and human presence influence carnivore space use within and beyond the protected area. Download all files in the same folder before running the R scripts. ### Files and variables #### Excel files: ##### File: ZNP2020_carnivore_dataset.xlsx **Description:** dataset including all detections of large carnivores during the survey period. Variables: station name, Date and Time, Species and Season.  ##### File: ZNP2020_covariatesRAI_use.xlsx **Description:** Site-level covariates that are included in the analysis. * Wet2020_RAIprey_log, Wet2020_RAIlarge_log, Wet2020_RAImed_log, Wet2020_RAIsmall_log, Dry2020_RAIprey_log, Dry2020_RAIlarge_log, Dry2020_RAImed_log, Dry2020_RAIsmall_log: log-transformed Relative Abundance Index of wild prey for the wet and dry season. Prey species were separated into large prey of > 250kg: giraffe, buffalo and roan antelope; medium prey between 70 kg and 250kg: waterbuck, greater kudu, hartebeest and tiang; and small prey <70kg: kob, warthog, bushbuck, reedbuck, red fronted gazelle, oribi and common duiker. * Dry2020_RAIlivestock_cap_log: log-transformed Relative Abundance Index of livestock capped livestock at the 99th percentile due to the presence of an extremely high outlier. * MakeModel: made and model of the camera traps Cuddeback Xchange (Cuddeback, Green Bay, Wisconsin, USA), Bushnell CORE, Bushnell Trophy Cam HD Aggressor (Bushnell Outdoor Products, Overland Park, Kansas, USA). * Flash: white flash (W) and dark flash (D) including covert flash (CV) and infra-red flash (IR). * hab_road: True or False, if the camera was deployed along a park road. * hab_water: True or False, if the camera was deployed close to available surface water. * Dist_border_m: distance to park boundary calculated from up-to-date spatial files provided by the park management (African Parks). * Dist_Villa_m: distance to villages calculated from up-to-date spatial files provided by the park management (African Parks). * X5x5_parkroad_m_log: density of park roads calculated from up-to-date spatial files provided by the park management (African Parks). * NDVI_Dry20, NDVI_Wet20: Normalized Difference Vegetation Index calculated using Google Earth Engine with Sentinel-2 satellite images with a 10m resolution, calculating the mean values for each grid square over each survey period. * DryWater_GEEdist_m_log: distance from permanent water in the dry season, identified using Google Earth Engine with Sentinel-2 satellite images with a 10m resolution. * Flood_area: surface area covered in water in the wet season per grid cell, in sqkm, extracted from ground-truthed spatial layer of flooded areas provided by the park management (African Parks). The continuous variables were z-standardized (x − x/σ) to a mean 0 and unit standard deviation. ##### ##### File: ZNP2020_table5d_effort_Dry2020_wide.xlsx ##### File: ZNP2020_table5d_effort_Wet2020_wide.xlsx **Description:** Effort (number of trap nights active) per sampling occasion (5 days) per station for the dry and wet season, also including total days the camera was active over the survey period. ##### ##### File: ZNP2020_table5d_Dry2020_leopard_wide.xlsx ##### File: ZNP2020_table5d_Dry2020_lion_wide.xlsx ##### File: ZNP2020_table5d_Dry2020_spottedhy_wide.xlsx ##### File: ZNP2020_table5d_Dry2020_cheetah_wide.xlsx ##### File: ZNP2020_table5d_Dry2020_stripedhy_wide.xlsx ##### File: ZNP2020_table5d_Wet2020_leopard_wide.xlsx ##### File: ZNP2020_table5d_Wet2020_spottedhy_wide.xlsx ##### File: ZNP2020_table5d_Wet2020_lion_wide.xlsx **Description:** Detection tables for the different carnivore species and different seasons derived from the ZNP2020_carnivore_dataset with sampling occasions of 5 days. 1 = animal detected; 0 = animal not detected; blank = camera not active during the sampling occasion. #### R scripts: ##### File: compare_det_occ_models.R ##### File: plot_occupancy_effect.R ##### File: run_ppc_diagnostics.R **Description:** source scripts that run in the background of the analysis scripts to compare multiple models (compare_det_occ_models.R), run diagnostics (run_ppc_diagnostics.R) and plot occupancy effects (run_ppc_diagnostics.R) ##### File: Occupancy_analysis_spOcc_final_-_after_PR1_share.R ##### **File:** Occupancy_analysis_spOcc_wet_season_final_-_after_PR1_share.R **Description:** R scripts to run the analysis with spOccupancy for the dry (Occupancy_analysis_spOcc_final_-after_PR1*.*R) and wet season (Occupancy_analysis_spOcc_wet_season_finale_-_after_PR1.R). Download all files in the same folder before running the scripts. ## Code/software All analyses were run in R version 4.4.2 (2024-10-31 ucrt) -- "Pile of Leaves" with RStudio 2024.12.0+467 "Kousa Dogwood" Release (cf37a3e5488c937207f992226d255be71f5e3f41, 2024-12-11) for windows.