Data from: Using individual-based modelling to anticipate future conflict potential for expanding European bison populations
Data files
Jun 15, 2026 version files 5.60 MB
-
data.zip
5.55 MB
-
R_scripts.zip
41.36 KB
-
README.md
12.88 KB
Abstract
European bison (Bison bonasus) populations are recovering and expanding in parts of their historical range, increasingly using landscapes shaped by agriculture, forestry, and transport infrastructure. While this recovery creates opportunities for ecological restoration, it also makes it important to anticipate where human-bison conflicts may emerge and how such conflicts could affect future population development. This dataset accompanies a study that mapped and modelled potential future conflict between European bison and people across Poland, Germany, Czechia, Slovakia, and Austria. We developed spatial indicators of relative conflict potential for three major conflict types: crop damage, forestry damage, and traffic collisions, based on land cover and land use data, habitat structure, and bison resource preferences. We then linked these indicators with outputs from a spatially explicit individual-based population model simulating population growth, dispersal, and patch occupancy over 100 years. The study also tested two conflict scenarios, representing population regulation and increased traffic mortality, to assess their effects on population trends. Our model projected continued growth and range expansion, while anticipating increasing conflict risk and region-specific hotspots. Both traffic mortality and conflict-driven population regulation substantially reduced future growth and expansion, thereby restricting European bison recovery. The analyses also identified low-conflict coldspots that may be promising candidates for restoration efforts. Together, the data and R scripts document the workflow used to create landscape-based conflict proxies and link these with population projections. They allow reproducibility of the associated analyses and provide a basis for further work on coexistence planning, conflict mitigation, and wildlife restoration in human-dominated landscapes.
Dataset DOI: 10.5061/dryad.r4xgxd2v8
Description of the data and file structure
This dataset accompanies a study that mapped and modelled potential future conflict between European bison and people in Central Europe. We developed spatial indicators of relative conflict potential for three major conflict types: crop damage, forestry damage, and traffic collisions, based on land cover and land use data, habitat structure, and bison resource preferences. We then linked these indicators with outputs from a spatially explicit individual-based population model simulating population growth, dispersal, and patch occupancy over 100 years. The study also tested two conflict scenarios, representing population regulation and increased traffic mortality, to assess their effects on population trends.
Files and variables
File: data.zip
Description:
- Subfolder "ancillary data" contains ancillary data to derive patch-level conflict indicators.
- habitat_patches.shp - shapefile of all potential habitat patches considered in the conflict potential analyses (Source: Bluhm et al. 2023)
- habitat_patches.tif - raster of all potential habitat patches considered in the conflict potential analyses. Patches are numbered 1-176, and the non-patch area has value 0 (Source: Bluhm et al. 2023)
- habitat_patches_na.tif - raster of all potential habitat patches considered in the conflict potential analyses. Patches are numbered 1-176, and the non-patch area is NA (Source: Bluhm et al. 2023)
- habitat_patches_b5.shp - shapefile of all potential habitat patches + a 5-km buffer around the patch area, considered in the conflict analyses, in line with regularly observed movement distances of E. bison from the forest edge
- studyarea_adm0.shp - shapefile of the full study area.
- Subfolder "RangeShiftR" contains input files for the RangeShiftR population models
- ebison_patches.txt - all potential habitat patches. (Source: Bluhm et al. 2023)
- ebison_hsi.txt - habitat suitability index values used to derive patch-specific carrying capacity (Source: Bluhm et al. 2023)
- ebison_cost.txt - movement resistance surface representing the pixel-level cost of movement for individuals across the study area during dispersal (Source: Bluhm et al. 2023)
- b_init_pop_fm1.txt - initial population data, indicating the number of individuals across sex and age classes within the initially occupied habitat patches at the start of the population model simulation in the baseline model.
- b_init_pop_fm_batch1_sim1.txt - initial population data, indicating the number of individuals across sex and age classes within the initially occupied habitat patches at the start of the population model simulation in the conflict-driven population regulation scenario with subsequent 5-year time steps.
- patches_survival_scaling_traffic.tif - raster with patch-specific scaling indices for the baseline mortality values based on the patch-specific values of traffic conflict potential, used in the traffic mortality conflict scenario of the population model.
- Subfolder "intermediate_results" contains intermediate results required to derive patch-level conflict indicators and to link these indicators with the population projections.
- _habitat_patches_proxies.rds: data frame with derived values of various land cover, land use, and habitat structure input data for each potential habitat patch, used for calculating the indices of conflict potential. The data frame contains the following variables:
- patch_id: unique ID for each potential habitat patch
- patch_area_km2: patch area in km2
- patch_area_km2_b: area of the buffered patch (patch + 5km buffer) in km2
- forest_share_potapov: share of forest cover within the patch, derived from the Potapov et al. 2022 land cover data
- forest_share_potapov_b: share of forest cover in the buffered patch area, derived from the Potapov et al. 2022 land cover data
- grassland_share_patch: share of grassland within the patch, derived from the Potapov et al. 2022 land cover data
- grassland_share_patch_b: share of grassland in the buffered patch area, derived from the Potapov et al. 2022 land cover data
- cropland_share_potapov: share of croplands within the patch, derived from the Potapov et al. 2022 land cover data
- cropland_share_potapov_b: share of croplands in the buffered patch area, derived from the Potapov et al. 2022 land cover data
- forage_availability_pre: mean forage availability value of the patch
- edge_density_f_c_potapov: density of forest-cropland edges in the buffered patch area in m/ha
- edge_density_f_c_potapov_INSIDE: density of forest-cropland edges within the patch in m/ha
- forestry_conflict_pre: mean value of forest management scores within patch, weighted by forest openness
- road_density_primary: density of roads with OSM type “primary” within the patch, in m/m2
- road_density_b_primary: density of roads with OSM type “primary” in the buffered patch area, in m/m2
- road_density_secondary: density of roads with OSM type “secondary” within the patch, in m/m2
- road_density_b_secondary: density of roads with OSM type “secondary” in the buffered patch area, in m/m2
- railway_density: density of railway lines within the patch, in m/m2
- railway_density_b: density of railway lines within the buffered patch area, in m/m2
- conflict_index_results.rds: data frame with the results of the conflict potential estimate calculations for all 176 potential habitat patches as well as for the nine extant population home ranges. The data frame contains the following variables:
- patch_id: unique ID for each potential habitat patch
- Population: name of extant population (for rows representing extant population home ranges or respective patches containing these home ranges); all unoccupied potential habitat patches are named “Other”
- pop_id: unique ID for each row that represents on of the extant populations; other rows are NA
- crop_conflict: crop conflict index, scaled 0-1
- forestry_conflict: forestry conflict index, scaled 0-1
- traffic_conflict: traffic conflict index, scaled 0-1
- pop_df_baseline_raw.rds - data frame with the results of the baseline population model containing information on the sex- and age-specific number of individuals for each habitat patch and model replicate over time (in 5-year steps). The data frame contains the following variables:
- Rep: Number of the model replicate (0-99 for 100 replicate model runs)
- Year: Year of the simulation
- PatchID: unique ID of the habitat patch
- Ncells: number of cells of the habitat patch (1 km2 resolution of the patch raster)
- NInd: Total number of individuals
- Nfemales_stage1: Number of female yearlings
- Nmales_stage1: Number of male yearlings
- Nfemales_stage2: Number of female subadults
- Nmales_stage2: Number of male subadults
- Nfemales_stage3: Number of female reproductive adults
- Nmales_stage3: Number of male reproductive adults
- Nfemales_stage4: Number of female senescent adults
- Nmales_stage4: Number of male senescent adults
- NJuvFemales: Number of female juveniles
- NJuvMales: Number of male juveniles
- model: number assigned to the respective model version
- scenario: name of the associated model scenario
- batch: model batch number
- conflict_pop_size_densities_patches.rds - data frame where low-conflict population densities are estimated for each habitat patch based on the size and habitat suitability of patches, their equilibrium densities in the baseline model, and the patch-specific crop and forestry conflict potential indices. These assumed low-conflict densities are then used as input for the conflict-driven population regulation scenario. The data frame contains the following variables:
- patch_id: unique ID for each patch
- patch_area_km2: size of patch in km2
- mean_forestry_crop_inverted: inverted average of the crop conflict index and forestry conflict index
- mean_hsi: average habitat suitability value of the patch
- patch_size: size of patch in km2
- carrying_capacity_K012: theoretical carrying capacity of the patch as calculated in RangeShiftR (mean_hsi * patch_size * K_parameter)
- theoretical_density_K012: population density of the patch at theoretical carrying capacity
- assumed_equilibrium_density_baseline: theoretical equilibrium density of the patch inferred from the relation of carrying capacity to simulated equilibrium densities of occupied patches in the baseline RangeShiftR model
- assumed_equilibrium_abundance_baseline: European bison abundance in the patch resulting from the equilibrium density
- conflict_adjusted_equilibrium_density: equilibrium density scaled by mean_forestry_crop_inverted, used as target population density in the population regulation scenario
- conflict_adjusted_equilibrium_abundance: equilibrium European bison abundance scaled by mean_forestry_crop_inverted, used as target population density in the population regulation scenario
- _habitat_patches_proxies.rds: data frame with derived values of various land cover, land use, and habitat structure input data for each potential habitat patch, used for calculating the indices of conflict potential. The data frame contains the following variables:
File: R_scripts.zip
Description: R scripts used for data analyses, modeling, and creating result figures (see Code/software for details). Replace “PATH” with your working directory in the file paths of the R scripts to load the uploaded data files and run the scripts.
The input data layers related to forest cover, forest type, tree cover density, grassland, cropland, forest management, soil type, roads and railways, which were used for calculating the spatial conflict indicators (script 01_patches_conflict_proxies_pre.R) cannot be published under a CC0-license here, but can be accessed from other sources as follows:
- Forest variables: Copernicus High Resolution Layers https://land.copernicus.eu/en/products/high-resolution-layer-forests-and-tree-cover
- Grassland and cropland: Potapov et al. 2022 land cover data https://glad.umd.edu/dataset/GLCLUC2020
- Soil type: JRC European Soil Data Center (ESDAC) https://esdac.jrc.ec.europa.eu/resource-type/datasets.
- Forest management: dataverse website of the Vrije Universiteit Amsterdam: https://doi.org/10.34894/HQIJN5
- Road and railway data: OpenStreetMap website or from https://download.geofabrik.de/
Code/software
We ran all analyses in R version 4.2.2, using the packages: sf, terra, tidyverse, landscapemetrics, readr, patchwork, RangeShiftR (v1.0.5 and its spatial demography extension). The following scripts contain the data analyses and models:
- 01_patches_conflict_proxies_pre.R - This script contains the extraction and calculation of the input variables for the conflict index calculations for each individual habitat patch. In a separate script, the final conflict index values are calculated from these input values.
- 02_conflict_indicators.R - Application of the formulas described in the main article to derive conflict potential estimates across habitat patches for crop, forestry, and traffic conflict, based on landscape characteristics and habitat structure.
- 03_conflict_indicators_results.R - Create result figures and maps of the estimated conflict potential.
- 04_population_model_baseline.R - Run the RangeShiftR spatial population model for European bison, i.e., the baseline model without explicit conflict impacts.
- 05_conflict_hotspots_coldspots.R - Linking the population model results with the conflict indicators to identify future conflict hotspots and coldspots over time.
- 06_population_model_baseline_5year_steps.R - Population model version split into subsequent 5-year steps in order to ensure consistency with the population regulation scenario
- 07_scenario_population_regulation.R - Run the conflict-driven population regulation scenario where emerging population densities within each patch are evaluated against the previously estimated low-conflict densities.
- 08_scenario_traffic_mortality.R - Run the traffic mortality conflict scenario where patch-specific mortality of European bison individuals is increased according to the estimated traffic conflict potential.
- 09_population_model_results.R - Pulling and summarizing the population results of all RangeShiftR population models.
- 10_population_model_result_figures.R - Creating result figures of population trajectories for the baseline and conflict scenario models.