Dataset DOI: 10.5061/dryad.j0zpc86t3

Description of the data and file structure

Species distribution modeling was conducted in the program R using the Wallace package (Kass et al. 2018, 2022) and followed the workflow guidelines of Kass et al. (2024), which recommend additional packages to enhance analytical utility. Wallace is built on the shiny web application framework, providing a graphical user interface (GUI) for species distribution modeling (Kass et al. 2022). Many of the supplementary packages outlined by Kass et al. (2024) are already incorporated as GUI options within Wallace, while others can be added as user-specified inputs. For this study, we implemented maximum entropy modeling (Phillips et al. 2006) within the Wallace shiny application.

Files denoted as "Mephits_macroura" are products from use of the Wallace package. 

Files denoted as "Camera" are downloads from our camera trap survey on Wildlife Insights (Below).

Any other file will have additional information. 

Files and variables

File: Mephitis_macroura_fc.LQH_rm.1.5_cloglog_(1).tif.aux.xml

Description: Predicted probability of occurrence.

File: Mephitis_macroura_maxnet_evalTbl.csv

Description: Model evaluation table - Global

File: Mephitis_macroura_maxnet_evalTblBins_(1).csv

Description: Model evaluation table - Across Folds

File: Mephitis_macroura_fc.LQH_rm.1.5_cloglog.tif

Description: Predicted probability of occurrence. 

Camera Trap Survey

Between January 2024 and September 2025, we operated camera traps within Big Bend National Park, Black Gap Wildlife Management Area, and El Carmen Land and Conservation Co., LLC, CEMEX INC., Texas - all adjacent properties within southern Brewster County, TX. At both Black Gap and El Carmen, cameras were placed in part at guzzlers, water catchment devices designed to provide water to wildlife, and at random locations. Within Big Bend National Park, cameras were deployed at random locations only. 

Images from the camera trap survey were uploaded, identified, and catalogued on Wildlife Insights (Perkins et al. 2024). There are 4 csv files downloaded from Wildlife Insights. Additionally, there is a pdf (Camera_Trap_-_Data-Dictionary-for-Data-Download.pdf) that defines every column within these 4 csv files. 

Camera_Trap_Deployments.csv

• See Camera_Trap_-_Data-Dictionary-for-Data-Download.pdf.
• Additional pertinent information provided below. 
• Column Information:
• project_ID: Unique Wildlife Insights Project ID
• deployment_ID: Unique identifier of site, camera #, and deployment date
• placename: Identifier of site and camera #
• longitude, latitude: Camera deployment location
• start_date, end date: Date of camera deployment. Date of deployment end either due to camera removal or camera failure. 
• bait_type: Whether bait was used. None used.
• feature_type: Whether the camera was placed on an anthropogenic or physical feature and what type. 
• feature_type_methodology: If feature type was other, cells are filled with either "Drainage" or "Desert Wash". All empty cells are NA. 
• camera_id: Unique numerical camera identifier - applied by WI. 
• camera_name: Unique camera name that combines site and camera number - internal. Redundant between deployment ID and placename. 
• quiet_period: Time, in seconds, between successive detections. 
• camera_functioning: Binary. Camera functioning indicates the camera was operational the entire period. All others indicate the loss of function during the deployment period and cause.
• sensor_height: Height the sensor was deployed at. All are "knee height".
• sensor_orientation: Angle the camera sensor was deployed at. All are parallel to the ground. 
• detection_distance: Estimate of distance camera can detect wildlife in meters. 
• recorded_remarks: Any additional notes on the camera deployment

Camera_Trap_Images.csv

• All detections from camera trap project that were catalogued within Wildlife Insights. 
• Blank cells are present in this spreadsheet. They are present in the taxonomic columns when species identification could not be resolved at a lower level and in the bounding boxes column when a detection box was not generated per image. These cells are intentionally left blank. 
• Future users should note that our cameras produced 3 images per detection event and all 3 images may not be categorized as the same species based upon how we reviewed and curated the images on Wildlife Insights. For example, image one may incompletely capture information capable of identifying the species (and in this instance be categorized as Canidae) while the next two images capture this information and are categorized as gray fox. Subsequent assessment of the dataset within an r package such as camtrapR (Niedballa et al. 2016) to identify species categorization errors (e.g., 3 images within ~ 5 seconds) will flag this as a potential error. Future users should use the Wildlife Insights project (Perkins et al. 2024) to resolve these issues. 
• Project personnel have mammal expertise with a carnivore concentration. Any user of this dataset outside of mammalian carnivores should consider using the Wildlife Insights project (Perkins et al. 2024) to confirm species identification. 
• Of note, avian species were primarily categorized as "avian" outside of turkey vulture and scaled quail. Detections included both black vulture and Montezuma quail, but these were less than 1% compared to the former species. 
• Of note, mule and white-tailed deer occur within the study area. White-tailed deer were only definitively identified from El Carmen. Generally, deer on El Carmen were categorized as Odocoileus sp. while deer at Black Gap and Big Bend were categorized as mule deer. 
• Of note, up to 3 cottontail rabbit species are hypothesized to occur within the study region. We are unaware of any reliable way to differentiate among these species from image only. Future users should elevate this consideration above any categorizations within the dataset. 
• Of note, collared peccary were frequently and abundantly detected. These images were reviewed a bit more closely as we have collaborative interest of this species (as of fall 2025). Feral pigs were present, but seemingly extremely rare (e.g., 1 known large boar moving among 2 of the properties). American black bear were present and detected relatively frequent. Efforts were made to resolve the Wildlife Insights AI initial identification across these 3 species. Broadly, temporal use of this dataset for collared peccary with a temporal designation of independence of 1 hour or less, may benefit from review of the Wildlife Insights project. 
• Of note, up to 4 currently extant canine species may occur within the study area (as of fall 2025). Recent survey of the general area failed to detect kit fox (Hewitt et al. 2022) and to our knowledge, red fox have never been located as far south in Brewster County as our survey area (Schmidly and Bradley 2016). While the Wildlife Insights AI flagged some detections as red fox or Vulpes species, we categorized all of these as grey fox. Across all canine images, there were none identified as kit fox or red fox that closely resembled these species.

Camera_Trap_Identifiers.csv

• Camera trap identifiers including Wildlife Insights designated camera trap ID (camera_id) and internal camera trap ID (camera name). 

Camera_Trap_Projects_Metadata.csv

• Project metadata associated with the Wildlife Insights project. 

Mephitis macroura "habitat ranking"

We used information from the 10 m resolution Texas Parks and Wildlife Department Ecological Mapping System (Elliot et al. 2014) to assess and rank potential M. macroura habitat associated with 11 occurrence records. This ranking classification was used solely to inform randomly stratified camera-trap survey locations. To identify the vegetative communities associated with M. macroura, we first applied a 1-kilometer buffer around each location and then conducted a summary analysis of the proportion of vegetation types within each buffer. To check for consistency across the Chisos Mountains, this procedure was repeated on a subset of nine lower-elevation locations and again on all 11 locations, but with vegetation types found only in the Chisos Mountains removed. In each instance, the ranking of the top three vegetation types remained the same: succulent desert scrub, mixed desert shrubland, and hill and foothill grassland vegetation types ranked the highest and together accounted for 52% of the land-cover at 10 m resolution in the global assessment.

Environmental_1-km_Buffers_from_BIBE_MEMA_Observations.xlsx

Workbooks:

• All Data: All data exported from the 1-km buffers surrounding each of the 11 hooded skunk occurrence records. Data is from the Texas Parks and Wildlife Department Ecological Mapping System (Elliot et al. 2014).
• Location: Name of individual camera trap associated with occurrence record. 
• Lat/Long. Latitude or Longitude.
• Veg Community: Vegetation community classifier. 
• Hectares: Size, in hectares, of each polygon.
• Truncated Data. Same dataset as "All Data", but truncated to only the 5 columns previous identified. 
• Individual Locations %:
• Pivot Tables summing the hectares of vegetative community per class. 
• Station: Camera Trap Station.
• Total HA: Total Hectares per Veg Class. 
• %: Decimal value for the percentage each veg class contributes to the overall 1-km buffer. 
• Global %
• Pivot table with all vegetation classes summed across the 11 buffers
• Cumulative: Total Hectares of each vegetation class. 
• Overall %: Percentage each vegetation class contributed to the 11 buffers
• Green: Vegetation Classes representing more than 5% of the global total and found throughout the study area.
• Red: Vegetation Classes representing more than 5% of the global total but primarily found in the Chisos Mountains, with secondary presence in the Northern Rosillos and Sierra del Carmens (e.g., higher elevation vegetation types). 
• High Elevation:
• Subset of locations (~2/3 of total) that were relatively higher in location compared to all occurrence locations or were "in between" higher and lower. 
• The only new column is Rank. Of these cameras and vegetation classes, numerical rankings were applied based upon overall percentage. Vegetation classes less than 5% were not considered. 
• Lower Elevation: 
• Same as the higher elevation workbook except these are the lower elevation occurrence locations and those "in between".

Big Bend National Park (BIBE) Camera Trap Survey

Between 2021 and 2021, National Park Service Personnel operated a camera trap survey comprised of 12 cameras arrayed in a grid across the Chisos Mountain range within the boundaries of Big Bend National Park. Under permit, images previously categorized as "skunk" were provided by the NPS, categorized, and results presented. Image metadata was tagged with a camera station identifier and species. Next, the camtrapR package (Niedballa et al. 2016) was used to define temporal independence as one visitation per species per camera trap per 12 hours. 

BIBE_Camera_Trap_Survey.csv

• Images are of 4 skunk species from camera deployed within Big Bend National Park. 
• Note some images contain multiple species including Urocyon cinereoargenteus and Canis latrans. 
• Station: Station identifier based on station names provided by BIBE personnel. 
• Species: Scientific name of species identified in image.
• DateTimeOriginal: Month/Day/Year; Hour:Minute
• Date: Month/Day/Year
• Time: Hour:Minute:Second
• delta.time.secs; delta.time.minute.; delta.time.hour:
• Per species, the first visit per camera trap is designated as 0, 0, 0 in these columns. After this visit, all successive visits that occurred 12 hours after the previous, are categorized as temporally independent and given a unique row with these columns representing the time (hours, minutes, and seconds) since the previous unique visit. 
• metadata_Site: All are BIBE - Big Bend National Park. 
• metadata_Species: Scientific name. 
• metadata_Camera: Camera station identifier. 
• n_images: Total number of images associated with each unique visit (e.g., all images of the species per camera per 12 hours). Cameras were set to record 3 images per trigger. 
• HierarchicalSubject: Hierarchical column that contains site, species, and camera using both commas and magritte pipes. 

Code/software

Wallace (Kass et al. 2018, 2022) via program r. 

References

Kass, J. M., Vilela B., Aiello-Lammens M. E., Muscarella R., Merow C., Anderson R. P. 2018. Wallace: A flexible platform for reproducible modeling of species niches and distributions built for community expansion. Methods in Ecology and Evolution, 9:1151-1156.

Kass, J.M., Pinilla-Buitrago, G.E, Paz, A., Johnson, B.A., Grisales-Betancur, V., Meenan, S.I., Attali, D., Broennimann, O., Galante, P.J., Maitner, B.S., Owens, H.L., Varela, S., Aiello-Lammens, M.E., Merow, C., Blair, M.E., Anderson R.P. 2022. wallace 2: a shiny app for modeling species niches and distributions redesigned to facilitate expansion via module contributions. Ecography, 2023: e06547. DOI: 10.1111/ecog.06547.

Niedballa, J., R. Sollmann, A. Courtiol, and A. Wilting. 2016. camtrapR: an R package for efficient camera trap data management. Methods in Ecology and Evolution 7:1457-1462. doi: 10.1111/2041-210X.12600.

Phillips, S. J., R. P. Anderson, and R. E. Schapire. 2006. Maximum entropy modeling of species geographic distributions. Ecological Modeling 190:231-259.

Access information

Other publicly accessible locations of the data:

• Perkins, J. C., R. C. Dowler, and R. D. Stevens. 2024. Last updated August 2025. Survey of the carnivore community of southern Brewster County. http://n2t.net/ark:/63614/w12008605. Accessed via wildlifeinsights.org on 2025-08-25.

Data was derived from the following sources:

• GBIF: Global Biodiversity Information Facility (GBIF). 2025. Mephitis macroura Occurrence Download. Accessed 25 April 2025. DOI: 10.15468/dl.rw3qjb.
• WorldClim: Fick, S. E., and R. J. Hijmans. 2017. WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. International Journal of Climatology 37(12): 4302–4315. https://doi.org/10.1002/joc.5086.
• iNaturalist: iNaturalist. 2025. Hooded Skunk Project. https://www.inaturalist.org/projects/hooded-skunk. Project created December 2024.
• Perkins, J. C., R. C. Dowler, and R. D. Stevens. 2024. Last updated October 2025. Survey of the carnivore community of southern Brewster County. http://n2t.net/ark:/63614/w12008605. Accessed via wildlifeinsights.org on 2025-10-17. 
• Elliott, L. F., A. Treuer-Kuehn, C. F. Blodgett, C. D. True, D. German, and D. D. Diamond. 2009-2014. Ecological Systems of Texas: 391 Mapped Types. Phase 1 – 6, 10-meter resolution Geodatabase, Interpretive Guides, and Technical Type Descriptions. Texas Parks & Wildlife Department and Texas Water Development Board, Austin, Texas. Documents and Data Available at: https://tpwd.texas.gov/gis/programs/landscape-ecology/by-ecoregion-vector.