Patch quality affords behavioral flexibility in collared pikas (Ochotona collaris), potentially increasing climate resiliency
Data files
May 08, 2026 version files 6.53 MB
-
COPI_IndivRecords_1min_Wall.csv
15.61 KB
-
COPI2022_TempLogger.csv
6.51 MB
-
README.md
9.29 KB
Abstract
Climate disruption threatens the persistence of temperature-sensitive species in many regions. Some species may be able to moderate hyperthermia risk by using cooler subsurface microclimates, a strategy that could afford some level of population resilience. However, such behavioral responses may interact with environmental factors, making it challenging to determine vulnerability. Here we assessed the extent to which collared pikas (Ochotona collaris), a temperature-sensitive boreal lagomorph, use below-talus microclimates as thermal refuges and whether environmental factors influence the plasticity of these responses, in order to assess the species’ vulnerability to regional warming. We monitored pika activity in Denali National Park and Preserve, Alaska, USA, with camera traps and temperature loggers (above and below ground), finding that pikas adjusted their activity in response to temperature, with peak activity at 13.9oC. Activity levels appeared driven by surface rather than subsurface temperatures, highlighting that regional warming could influence the species in ways that are not buffered by below ground thermal refugia. However, we also detected an interaction between patch quality and surface temperature, such that pikas were more active at cooler temperatures in talus patches with more vegetation. This suggests that pikas in better habitat could be more resilient to warming as they are afforded more flexibility in their diel activity patterns before reaching the upper limits of thermal tolerance. Overall, our results highlight that behavior and environment can interact to influence species’ resilience to climate change.
Dataset DOI: 10.5061/dryad.gtht76j2v
Description of the data and file structure
We conducted research north of the Alaska Range in the northeast portion of Denali during the summer of 2022. Our study area encompassed all land within 8 km of the first 73 km of the road corridor that runs from the park entrance in the east to mid-way through the park in the west. We deployed camera traps (Reconyx HyperFire) and temperature loggers (HOBOware Pro v2) between 1 June and 20 July and collected them between 8 and 12 August, 2022.
To reduce the possible influence of human disturbance while maximizing accessibility and survey efforts, we established sites between 100 m and 4 km from the road and at least 50 m from maintained trails. Deploying camera traps and temperature loggers enabled us to continuously monitor behavioral patterns with minimal disruptions to individuals and exact date and time records.
At each site, we recorded a path along the talus patch perimeter to measure the talus patch size of each occupied site. In areas where the talus patch perimeter was ambiguous, we defined the border conservatively, using cliffs, scree, and tundra as patch edges. We then estimated percent cover of both forbs and graminoids within a patch, denoting scores based on Daubenmire’s (1959) vegetation scale with additional categories as described by Rodhouse et al. (2010). Scores for each site were then totaled to calculate a measure of intra-talus forage availability, which we defined as patch quality. We defined latrines as piles of at least 20 scat pellets that included some fresh fecal matter and fresh scat as those appearing dark or green in color and pliable in texture, not brittle. Starting at the lowest elevation enabled us to scan cracks and crevices up in elevation, reducing surveyors’ potential to miss signs. Surveyors scanned the ground across a patch while communicating with adjacent surveyors to avoid double counting any detections. This process continued until the entire patch was surveyed.
Within each occupied talus patch, we placed a camera trap facing the highest concentration of recent pika activity, including individual sightings, fresh scat, and fresh haypiles, while minimizing interference from direct sunlight. Cameras were attached to steel rebar poles (61 cm long) secured in the talus at heights ranging from 17 to 60 cm above the talus (avg. 48 cm) and set to record 3 images in rapid fire at medium/high sensitivity with a 15 sec quiet period. We then placed two temperature loggers within 5 m of the camera trap: one in the shade and one between 0.5 and 1 m below ground in a location with accessible retrieval options, similar to the methods by Wright and Stewart (2018). The temperature loggers were carefully positioned among the rocks to prevent direct exposure to sunlight and were set to record temperature at 10-min. intervals for the duration of the study, starting at midnight the morning after they were launched. At the end of the survey period, we removed all cameras and temperature loggers and extracted temperature data with HOBOware (Onset Computer Corporation, 2022).
We combined our camera images, temperature logger data, and site-level survey data into one comprehensive activity record for collared pikas at each site. For our camera trap images, we assigned metadata tags to images with pikas present using digiKam (digiKam developers team, 2022) and extracted metadata information on species, date, and time with camtrapR (Niedballa et al., 2016). We combined images into a record table with a minimum delta time of 1 min., then summed total occurrence records at each station over 3-hr intervals for the duration of camera deployment. These records were then matched to their respective 3-hr intervals for each camera throughout the duration of the study period, with all remaining time intervals marked as 0 detections. For our temperature loggers, we averaged all 10-min. measurements within each 3-hr interval and joined them to our detection data.
Files and variables
File: COPI_IndivRecords_1min_Wall.csv
Description: Summary of collared pika (Ochotonas collaris) detections in 1 minute intervals at each site with accompanying site location covariates.
Variables
- Station: Camera site location unique identifier.
- n: Number of total collared pika (Ochotona collaris) detection events (in 1 min. intervals) at this site across the study period.
- latrines: Total number of latrines counted during the plot survey. We defined latrines as piles of at least 20 scat pellets that included some fresh fecal matter and defined fresh scat as those appearing dark or green in color and pliable in texture, not brittle.
- Setup_Date: Date camera site was established in the field. (MM/DD/YY)
- Retrieval_Date: Date camera was removed from the field. (MM/DD/YY)
- Time.Retrieval: Time camera was removed from the field.
- trap.ngt: Total number of trap nights across the study period.
- n.trap: Number of total collared pika (Ochotona collaris) detections at this site per trap night.
- gram: Graminoid site score (0 to 8). This is an estimated percent cover of graminoids within a patch, denoted as a numeric score based on Daubenmire’s (1959) vegetation scale with additional categories as described by Rodhouse et al. (2010). Categories include 0 (0%), 1 (trace, <1%), 2 (1-5%), 3 (5-25%), 4 (25-50%), 5 (50-75%), 6 (75-95%), 7 (95-99%), 8 (100%).
- forb: Forb site score (0 to 8). This is an estimated percent cover of forb within a patch, denoted as a numeric score based on Daubenmire’s (1959) vegetation scale with additional categories as described by Rodhouse et al. (2010). Categories include 0 (0%), 1 (trace, <1%), 2 (1-5%), 3 (5-25%), 4 (25-50%), 5 (50-75%), 6 (75-95%), 7 (95-99%), 8 (100%). For this measurement, we also included low-cover shrubs (namely, Dryas spp.) as forbs.
- length: Visual estimate of the approximate length (m) of the surveyed talus patch. In areas where the talus patch perimeter was ambiguous, we defined the border conservatively, using cliffs, scree, and tundra as patch edges.
- width: Visual estimate of the approximate width (m) of the surveyed talus patch. In areas where the talus patch perimeter was ambiguous, we defined the border conservatively, using cliffs, scree, and tundra as patch edges.
- quality: Patch quality score (0 to 16), defined as intra-talus forage availability. This is a combined total of the scores for graminoids and forbs (see above).
- area: Talus patch area (m2) based on GPS tracks of the defined talus perimeter.
- elev: Elevation (m) at camera site from the ArcticDEM (U.S. Geological Survey 2020) digital elevation model (5 m resolution).
- perm: Continuous probability of near-surface permafrost (Pastick et al. 2015) in percentages (30 m resolution).
- Species: Unique code identifer for species surveyed. In this case, COPI = collared pika.
- Det_DateTime: Date and time of detection event (camera trap photo record).
- Det: Detection (binary). All 1's.
File: COPI2022_TempLogger.csv
Description: Temperature logger data for ground level and sub-surface temperature loggerss at each station (camera site location).
Variables
- Station: Camera site location unique identifier.
- Ground Logger: Surface level temperature logger (HOBOware Pro v2) unique identifier.
- Date Time, GMT-08:00 (GL): Date and timestamp for each temperature reading for the ground logger.
- GL.Temp (°F): Recorded temperature for the ground logger in Fahrenheit.
- GL.Temp (°C): Recorded temperature for the ground logger in Celcius.
- GL.RH (%): Recorded relative humidity in percentages for the ground logger.
- Subsurface Logger: Sub-surface level temperature logger (HOBOware Pro v2) unique identifier. Loggers were placed between 0.5 and 1 m below ground in a location with accessible retrieval options.
- Date Time, GMT-08:00 (SS): Date and timestamp for each temperature reading for the subsurface logger.
- SS.Temp (°F): Recorded temperature for the subsurface logger in Fahrenheit.
- SS.Temp (°C): Recorded temperature for the subsurface logger in Celcius.
- SS.RH (%): Recorded relative humidity in percentages for the subsurface logger.
Code/software
Camera photo data was assigned metadata tags with digiKam (digiKam developers team, 2022) and processed with camtrapR (Niedballa et al., 2016).
All data was analyzed in R (R Core Team 2022), and included the following packages:
- tidyverse: workflow and data organization
- sf: extracting spatiotemporal data from the camera site locations.
- camtrapR: importing and organizing camera trap data to create a record table of occurrences.
- glmmTMB (Brooks et al., 2017): For Poisson generalized linear mixed-effects model, with camera site as a random effect on the zero-inflated portion of the model
Code for this research is provided on Github: Wall, JL. (2026). Patch-quality-collared-pikas. [R]. GitHub. https://github.com/wall-jennifer/patch-quality-collared-pikas