Data and supporting information from: Habitat heterogeneity and anthropogenic disturbance jointly affect species spatial associations and persistence
Data files
Feb 09, 2026 version files 71.19 KB
-
Asian_badger.csv
6.06 KB
-
Cattle.csv
6.05 KB
-
Hog_badger.csv
6.06 KB
-
Horse.csv
6.05 KB
-
Parameters.csv
2.84 KB
-
README.md
7.82 KB
-
Red_fox.csv
6.05 KB
-
Sika_deer.csv
6.05 KB
-
Stray_dog.csv
6.05 KB
-
Tufted_deer.csv
6.05 KB
-
Wild_boar.csv
6.05 KB
-
Wolf.csv
6.05 KB
Abstract
Interspecific spatial associations are a key component of mammalian community assembly, yet their persistence in heterogeneous, human-dominated landscapes remains poorly understood. This dataset was developed to investigate the mechanisms underlying these associations. It comprises two integrated components: 1) systematically collected occurrence records for ten sympatric large- and medium-sized mammal species from extensive camera-trap surveys, and 2) a suite of environmental covariates characterizing anthropogenic pressure and 3D habitat heterogeneity at each survey site. The dataset also includes complete model selection results and summarized model outputs as Supplemental information. Analyses using this dataset indicate that various forms of human disturbance cumulatively influence interspecific spatial associations, while three-dimensional habitat heterogeneity mediates these anthropogenic effects. This curated dataset provides a foundation for understanding the persistence mechanism of mammal communities, thereby strengthening our understanding of biodiversity conservation in a human-dominated landscape with 3D heterogeneity.
We have submitted occurrence records for ten species (Red_fox.csv, Sika_deer.csv, Stray_dog.csv, Tufted_deer.csv, Wild_boar.csv, Wolf.csv, Asian_badger.csv, Cattle.csv, Hog_badger.csv, Horse.csv) , environmental variables (Parameters.csv), along with model selection results (Data_S1_Five_model_selection_outputs_applied_five_distinct_communities.xlsx, Data_S2_Five_model_selection_outputs_applied_ten_distinct_communitie.xlsx) and summarized model outputs (Data_S3_The_influence_of_environmental_characteristics_on_occurrence_of_mammals.xlsx, Data_S4_The_influence_of_environmental_characteristics_on_species_co-occurrence_of_five_communities.xlsx, Data_S5_The_influence_of_environmental_characteristics_on_species_co-occurrence_of_ten_communities.xlsx ) as Supplemental information.
Description of the data and file structure
Red_fox.csv:
Detection / non-detection data for Red_fox. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Sika_deer.csv:
Detection / non-detection data for Sika_deer. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Stray_dog.csv:
Detection / non-detection data for Stray_dog. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Tufted_deer.csv:
Detection / non-detection data for Tufted_deer. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Wild_boar.csv:
Detection / non-detection data for Wild_boar. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Wolf.csv:
Detection / non-detection data for Wolf. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Asian_badger.csv:
Detection / non-detection data for Asian_badger. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Cattle.csv:
Detection / non-detection data for Cattle. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Hog_badger.csv:
Detection / non-detection data for Hog_badger. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Horse.csv:
Detection / non-detection data for Horse. ID is camera trap site identifier, Ri represents the i-th replicate. 1 indicates detection, 0 indicates non-detection, NA indicates no sampling.
Parameters.csv:
Environmental variables. PA is protected area zoning-specific, slope is topographic slope, W_NEAR_DIST is proximity to key water resources, C_NEAR_DIST is proximity to the countryside, PAD_d is habitat vertical complexity, CHM_sd is habitat horizontal complexity.
Files available on Zenodo (https://doi.org/10.5281/zenodo.15108262):
Data_S1_Five_model_selection_outputs_applied_five_distinct_communities.xlsx:
Model selection outputs of the five candidate models applied to five distinct communities in different season. M1 is analogous to a conventional multi-species occurrence model, positing that the presence or absence of one species is independent of another. M2 is an intercept-only model that assumes species influence each other's occurrence patterns regardless of local environmental conditions. M3, M4, and M5 also acknowledge interdependence among species occurrence patterns, with spatial associations varying in relation to PAD_d, CHM_sd, and C_dis, respectively. The models are ranked in descending order of fit based on the Akaike Information Criterion (AIC). ΔAIC denotes the difference in AIC values relative to the top-performing model. Models with ΔAIC values less than 2 are considered to have statistically comparable support and are highlighted in bold. M2, M3, M4, and M5, was purely a function of species-specific and pairwise interaction parameters, assuming the conditional probability 3 or more species occurred together rarely.
Data_S2_Five_model_selection_outputs_applied_ten_distinct_communitie.xlsx:
Model selection outputs of the five candidate models applied to ten distinct communities in different season. M1 is analogous to a conventional multi-species occurrence model, positing that the presence or absence of one species is independent of another. M6 is an intercept-only model that assumes species influence each other's occurrence patterns regardless of local environmental conditions. M7, M8, and M9 also acknowledge interdependence among species occurrence patterns, with spatial associations varying in relation to PAD_d, CHM_sd, and C_dis, respectively. The models are ranked in descending order of fit based on the Akaike Information Criterion (AIC). ΔAIC denotes the difference in AIC values relative to the top-performing model. Models with ΔAIC values less than 2 are considered to have statistically comparable support and are highlighted in bold. M6, M7, M8, and M9 account for the conditional probability of three species co-occurring, in addition to species-specific and pairwise interaction parameters.
Data_S4_The_influence_of_environmental_characteristics_on_occurrence_of_mammals.xlsx:
This quantitative summary is derived from multi-species, multi-state occupancy models, which evaluate the effects of environmental characteristics (PAD_d, CHM_sd, C_dis，W_dis) on ten species occupancy. We present the mean of the distribution, along with uncertainty conveyed through 75 % and 95 % confidence intervals (CI). Parameters exhibiting moderate support (where the 75 % CI does not include zero) are bolded, while those with substantial support (where the 95 % CI does not include zero) are bolded and underlined.
Data_S4_The_influence_of_environmental_characteristics_on_species_occurrence_of_five_communities.xlsx:
Summary of model outputs demonstrating the influence of local environmental characteristics on species occurrence. This quantitative summary is derived from multi-species, multi-state occupancy models, which evaluate the effects of environmental characteristics (PAD_d, CHM_sd, C_dis) on species occupancy. We present the mean of the distribution, along with uncertainty conveyed through 75 % and 95 % confidence intervals (CI). Parameters exhibiting moderate support (where the 75 % CI does not include zero) are bolded, while those with substantial support (where the 95 % CI does not include zero) are bolded and underlined.
Data_S5_The_influence_of_environmental_characteristics_on_species_occurrence_of_24_communities.xlsx:
Summary of model outputs demonstrating the influence of livestock and local environmental characteristics on wild species occurrence. This quantitative summary is derived from multi-species, multi-state occupancy models, which evaluate the effects of environmental characteristics (PAD_d, CHM_sd, C_dis) on species occupancy. We present the mean of the distribution, along with uncertainty conveyed through 75 % and 95 % confidence intervals (CI). Parameters exhibiting moderate support (where the 75 % CI does not include zero) are bolded, while those with substantial support (where the 95 % CI does not include zero) are bolded and underlined.