Data from: Movement-integrated habitat selection reveals wolves balance ease of travel with human avoidance in a risk-reward trade-off
Data files
Feb 19, 2025 version files 1.04 GB
-
JAPPL-2024-00408_Wolf-Crossing-Dataset.csv
532.15 MB
-
JAPPL-2024-00408_Wolf-Proximity-Dataset.csv
506.25 MB
-
README.md
8.42 KB
Abstract
Anthropogenic linear features often alter wildlife behaviour and movement. Landscape features, such as habitat, can have important mediating effects on wildlife response to disturbance and yet are rarely explicitly considered in how habitat and disturbance interact. We tested the movement and space-use responses of GPS-collared wolves to linear features with respect to adjacent habitat variation. We simultaneously modeled wolf movement and selection within a conditional logistic regression framework (integrated Step Selection Analysis). We explicitly considered how adjacent habitat alters these responses through putative effects such as movement friction. Classifying linear features based on the selection and movement response of wolves revealed that pairing transmission lines and primary roads increased the avoidance response to be greater than either feature on their own and provided evidence of a semi-permeable barrier to movement. In contrast, features with reduced human activity, including secondary and tertiary roads were highly selected for and function as movement corridors.
Synthesis and applications: Explicitly parameterizing adjacent habitats provides evidence that where a linear feature is routed and which habitats it interacts with will have the greatest implications for wolf behavioural responses. Reduced avoidance behaviour in highly risky environments signifies the importance of habitat for maintaining landscape connectivity, particularly when routing multiple different features parallel and near each other. Increased vegetation density along linear features also reduced movement advantages putatively by increasing friction, indicating that actively decommissioning other features such as secondary roads could be an effective mitigation strategy for reducing wolf encounters with prey. Knowing the influence of adjacent habitats on the likelihood of wolves selecting for a given linear feature creates a context to minimize the impact of new anthropogenic features on behaviour.
https://doi.org/10.5061/dryad.ksn02v7g1
Description of the data and file structure
We captured and collared 46 wolves across 12 packs in eastern Manitoba between 2014-2019. Each wolf was fit with a GPS telemetry collar programmed to collect a GPS relocation every two hours across all seasons.
To produce the dataset, GPS locations were used to generate steps (linear connection between consecutive locations) using the integrated Step Selection Function (iSSA) framework. For a given used step, we randomly generated ten steps based on observed distributions of individual-level movement behaviour. We then extracted habitat covariates for each start and end point of a step, including proportion of forest, distance to linear features, and time of day.
GPS locations have been removed for conservation purposes. Please contact authors directly if required.
Files and variables
File: JAPPL-2024-00408_Wolf-Crossing-Dataset.csv
Description: Dataset used to run wolf iSSA model for crossing linear features
Variables
- uniqueID: unique row ID
- Date: date of GPS location in YYYY-MM-DD format
- Animal_ID: unique ID for each tagged animal
- burst_: unique identifier for each step burst
- sl_: step length in meters
- ta_: turn angle in radians
- t1_: timestamp (date and time in CST) at start of step
- t2_: timestamp (date and time in CST) at end of step
- dt_: time difference between t1 and t2
- step_id_: unique identifier for a given step
- case_: logical; indicates used (TRUE) or available (FALSE) step
- cos_ta: cosine transformed turn angle
- log_sl: log transformed step length
- lnDistTo_PR_start: log transformed distance to nearest primary road at the start of the step
- lnDistTo_SR_start: log transformed distance to nearest secondary road at the start of the step
- lnDistTo_TL_start: log transformed distance to line 77 at the start of the step
- lnDistTo_TLAll_start: log transformed distance to nearest transmission line at the start of the step
- lnDistTo_TLN_start: log transformed distance to line PQ95 at the start of the step
- lnDistTo_TLP_start: log transformed distance to line L5/L47 at the start of the step
- lnDistTo_TR_start: log transformed distance to nearest tertiary road at the start of the step
- lnDistTo_W_Temp_start: log transformed distance to nearest waterway at the start of the step
- lnDistTo_PR_end: log transformed distance to nearest primary road at the end of the step
- lnDistTo_SR_end: log transformed distance to nearest secondary road at the end of the step
- lnDistTo_TL_end: log transformed distance to line 77 at the end of the step
- lnDistTo_TLAll_end: log transformed distance to nearest transmission line at the end of the step
- lnDistTo_TLN_end: log transformed distance to line PQ95 at the end of the step
- lnDistTo_TLP_end: log transformed distance to line L5/L47 at the end of the step
- lnDistTo_TR_end: log transformed distance to nearest tertiary road at the start of the step
- lnDistTo_W_Temp_end: log transformed distance to nearest waterway at the end of the step
- OpenBuff_St: proportion of open habitat within a 100 m buffer of the start of the step
- ConiferBuff_St: proportion of coniferous habitat within a 100 m buffer of the start of the step
- DecidBuff_St: proportion of deciduous habitat within a 100 m buffer of the start of the step
- MixedBuff_St: proportion of mixed habitat within a 100 m buffer of the start of the step
- OpenBuff_End: proportion of open habitat within a 100 m buffer of the end of the step
- ConiferBuff_End: proportion of coniferous habitat within a 100 m buffer of the end of the step
- DecidBuff_End: proportion of deciduous habitat within a 100 m buffer of the end of the step
- MixedBuff_End: proportion of mixed habitat within a 100 m buffer of the end of the step
- Time: time in CST of location fix
- Day: binary; 1 if location was taken during daylight, 0 otherwise
- Night: binary; 1 if location was taken during night, 0 otherwise
- Twilight: binary; 1 if location was taken during twilight, 0 otherwise
- PRcross: binary; 1 if step crossed a primary road, 0 otherwise
- SRcross: binary; 1 if step crossed a secondary road, 0 otherwise
- TRcross: binary; 1 if step crossed a tertiary road, 0 otherwise
- TLcross: binary; 1 if step crossed Line 77, 0 otherwise
- TLPcross: binary; 1 if step crossed Line L5/L47, 0 otherwise
- TLNcross: binary; 1 if step crossed Line PQ95, 0 otherwise
- TLAllcross: binary; 1 if step crossed a transmission line, 0 otherwise
- Wcross: binary; 1 if step crossed a waterway, 0 otherwise
- ToD: factor with 3 levels; Day, Night, or Twilight
- wolf_step_id: unique identifier for a combination of animal ID and step ID
File: JAPPL-2024-00408_Wolf-Proximity-Dataset.csv
Description: Dataset used to run wolf iSSA model for linear feature proximity.
Variables
- uniqueID: unique row ID
- Date: date of GPS location in YYYY-MM-DD format
- Animal_ID: unique ID for each tagged animal
- burst_: unique identifier for each step burst
- sl_: step length in meters
- ta_: turn angle in radians
- t1_: timestamp (date and time in CST) at start of step
- t2_: timestamp (date and time in CST) at end of step
- dt_: time difference between t1 and t2
- step_id_: unique identifier for a given step
- case_: logical; indicates used (TRUE) or available (FALSE) step
- cos_ta: cosine transformed turn angle
- log_sl: log transformed step length
- lnDistTo_PR_start: log transformed distance to nearest primary road at the start of the step
- lnDistTo_SR_start: log transformed distance to nearest secondary road at the start of the step
- lnDistTo_TL_start: log transformed distance to line 77 at the start of the step
- lnDistTo_TLAll_start: log transformed distance to nearest transmission line at the start of the step
- lnDistTo_TLN_start: log transformed distance to line PQ95 at the start of the step
- lnDistTo_TLP_start: log transformed distance to line L5/L47 at the start of the step
- lnDistTo_TR_start: log transformed distance to nearest tertiary road at the start of the step
- lnDistTo_W_Temp_start: log transformed distance to nearest waterway at the start of the step
- lnDistTo_PR_end: log transformed distance to nearest primary road at the end of the step
- lnDistTo_SR_end: log transformed distance to nearest secondary road at the end of the step
- lnDistTo_TL_end: log transformed distance to line 77 at the end of the step
- lnDistTo_TLAll_end: log transformed distance to nearest transmission line at the end of the step
- lnDistTo_TLN_end: log transformed distance to line PQ95 at the end of the step
- lnDistTo_TLP_end: log transformed distance to line L5/L47 at the end of the step
- lnDistTo_TR_end: log transformed distance to nearest tertiary road at the start of the step
- lnDistTo_W_Temp_end: log transformed distance to nearest waterway at the end of the step
- OpenBuff_St: proportion of open habitat within a 100 m buffer of the start of the step
- ConiferBuff_St: proportion of coniferous habitat within a 100 m buffer of the start of the step
- DecidBuff_St: proportion of deciduous habitat within a 100 m buffer of the start of the step
- MixedBuff_St: proportion of mixed habitat within a 100 m buffer of the start of the step
- OpenBuff_End: proportion of open habitat within a 100 m buffer of the end of the step
- ConiferBuff_End: proportion of coniferous habitat within a 100 m buffer of the end of the step
- DecidBuff_End: proportion of deciduous habitat within a 100 m buffer of the end of the step
- MixedBuff_End: proportion of mixed habitat within a 100 m buffer of the end of the step
- Time: time in CST of location fix
- Day: binary; 1 if location was taken during daylight, 0 otherwise
- Night: binary; 1 if location was taken during night, 0 otherwise
- Twilight: binary; 1 if location was taken during twilight, 0 otherwise
- ToD: factor with 3 levels; Day, Night, or Twilight
- wolf_step_id: unique identifier for a combination of animal ID and step ID
Code/software
R programming language is required to run the files. Code and scripts can be found on Zenodo: https://doi.org/10.5281/zenodo.14201743
We captured and collared 46 wolves across 12 packs in eastern Manitoba between 2014-2019. Each wolf was fit with a GPS telemetry collar programmed to collect a GPS relocation every two hours across all seasons.
To produce the dataset, GPS locations were used to generate steps (linear connection between consecutive locations) using the integrated Step Selection Function (iSSA) framework. For a given used step, we randomly generated ten steps based on observed distributions of individual-level movement behaviour. We then extracted habitat covariates for each start and end point of a step, including proportion of forest, distance to linear features, and time of day.
