Gyrfalcon prey abundance and their habitat associations in a changing Arctic
Data files
Dec 19, 2024 version files 372.54 KB
-
README.md
13.20 KB
-
ropt.csv
112.80 KB
-
squir.csv
112.80 KB
-
widedf.csv
20.97 KB
-
wipt.csv
112.76 KB
Abstract
Arctic habitats are changing rapidly and altering trophic webs and ecosystem functioning. Understanding how species’ abundances and distributions differ among Arctic habitats is important in predicting future species shifts and trophic-web consequences. We aimed to determine the habitat-abundance relationships for three small herbivores on the Seward Peninsula of Alaska, USA by fitting data from 983 point counts (collected during 2019, 2021, 2022) with N-mixture models that account for imperfect detection. These herbivore species, Willow Ptarmigan [Lagopus lagopus], Rock Ptarmigan [L. muta], and Arctic ground squirrels [Urocitellus parryii]), are fundamental to tundra food webs, and primary prey for Arctic raptors including Gyrfalcons (Falco rusticolus). Second, we aimed to map herbivore densities within Gyrfalcon breeding territories. Third, we aimed to evaluate whether Gyrfalcons were more likely to occupy territories with higher prey densities using a multi-season occupancy model coupled with occupancy observations from helicopter surveys conducted during 2016-2022 at 97 Gyrfalcon territories. We found that male Willow Ptarmigan were more abundant in areas with greater cover of tundra, tall shrubs, and tussock tundra. Conversely, male Rock Ptarmigan were more abundant in areas with greater cover of sparse vegetation and tundra. Arctic ground squirrels were more abundant at higher elevations with greater cover of sparse vegetation and low shrubs. Willow Ptarmigan were widespread within Gyrfalcon breeding territories, whereas Rock Ptarmigan and Arctic ground squirrels had patchier distributions with few areas of high abundance. Lastly, Gyrfalcons were more likely to occupy territories with higher densities of Willow Ptarmigan and Arctic ground squirrels. As the Artic continues to warm, Rock Ptarmigan and Arctic ground squirrels may be vulnerable to ongoing shrub encroachment, whereas Willow Ptarmigan may benefit. By tying abundances of three prey to Gyrfalcon occupancy, our results contribute to understanding potential impacts on higher levels of this Arctic trophic web.
README: Gyrfalcon prey abundance and their habitat associations in a changing Arctic
https://doi.org/10.5061/dryad.djh9w0w94
Author: Michaela Gustafson
Contact: michagustaf@gmail.com
Manuscript DOI: 10.1002/ECE3.70763
Study area: Seward Peninsula, Alaska USA
Dates: 2016, 2019-2022
Species: Willow Ptarmigan (Lagopus lagopus), Rock Ptarmigan (Lagopus muta), Arctic Ground Squirrels (Urocitellus parryii), Gyrfalcon (Falco rusticolus)
Last updated: 11 December 2024
Description of the data and file structure
This study investigates the habitat-abundance relationships of three herbivore species—Willow Ptarmigan, Rock Ptarmigan, and Arctic Ground Squirrels to understand how their distributions may change in response to habitat shifts due to climate change. Data were collected across 983 point count locations during the summers of 2019, 2021, and 2022. These observations were analyzed using N-mixture models to account for imperfect detection, providing more accurate estimates of species abundance in different habitat types.
The study also involved mapping the herbivore densities within 97 Gyrfalcon breeding territories using the same point count data. Additionally, a multi-season occupancy model was employed to assess whether Gyrfalcons were more likely to occupy territories with higher prey densities, using occupancy data obtained from helicopter surveys conducted between 2016 and 2022. These combined efforts allowed for the evaluation of how herbivore distributions and abundances relate to Gyrfalcon territory occupancy and the broader trophic web dynamics in the Arctic.
Files and variables
File: ropt.csv
Description: Male Rock Ptarmigan observations and standardized covariates used in analysis. Rock Ptarmigan observations collected May-July 2019, 2021, and 2022.
Variables
- id: unique site id
- 1: time interval (1st 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 2: time interval (2nd 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 3: time interval (3rd 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 4: time interval (4th 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 5: time interval (5th 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- observer: unique ID for primary observer of the survey. DS is Observer 1, KW is Observer 2, and MG is Observer 3.
- wind: standardized value of wind(mph) for that site during time of survey. Units: mph
- julian: day of year (not true julian date, but the count starting from January 1st each year) for that site during time of survey. Values standardized. Units: day
- min_after_sun: standardized minutes after end of civil twilight or when the sun reaches 0 degrees on the horizon that the survey at that site took place. Negative indicates survey was conducted before 'sunrise'. Units: minutes
- lowshrub: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- tundra: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- sparseveg: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- elev: standardized elevation for each point. unstandardized elevation value is the mean elevation for land within 800m2 of survey point. Units: meters
File: widedf.csv
Description: Gyrfalcon occupancy data
Variables
- terrXyear: Unique territory and year of survey ID
- territory: Territory name
- year: year survey occurred
- pres_1: Gyrfalcon occupancy status during the 'early' survey that occurred in May
- pres_2: Gyrfalcon occupancy status during the 'late' survey that occurred in June
- jday_1: day of year (from January 1st) the early survey took place. Units: day
- jday_2: day of year (from January 1st) the late survey took place. Units: day
*Cells that contain NA means a survey was not conducted for that site during that year and during that survey round (either early or late).
File: squir.csv
Description: All Arctic ground squirrel observations and standardized covariates used in analysis. Arctic ground squirrel observations collected May-July 2019, 2021, and 2022.
Variables
- id: unique site id
- 1: time interval (1st 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 2: time interval (2nd 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 3: time interval (3rd 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 4: time interval (4th 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 5: time interval (5th 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- wind: standardized value of wind (mph) for that site during time of survey. Units: mph
- julian: day of year (not true julian date, but the count starting from January 1st) for that site during time of survey. Values standardized. Units: day
- min_after_sun: standardized minutes after end of civil twilight or when the sun reaches 0 degrees on the horizon that the survey at that site took place. Negative indicates survey was conducted before 'sunrise'. Units: minute
- observer: unique ID for primary observer of the survey. DS is Observer 1, KW is Observer 2, and MG is Observer 3.
- lowshrub: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- tundra: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- sparseveg: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- elev: standardized elevation for each point. unstandardized elevation value is the mean elevation for land within 800m2 of survey point. Units: meter
File: wipt.csv
Description: Male Willow Ptarmigan observations and standardized covariates used in analysis. Willow Ptarmigan observations collected May-July 2019, 2021, and 2022.
Variables
- id: unique site id
- 1: time interval (1st 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 2: time interval (2nd 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 3: time interval (3rd 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 4: time interval (4th 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- 5: time interval (5th 2-minute period). The number in that row x column is the number of Rock Ptarmigan observed at that site during that specific time interval *** Data was filtered to only include male Rock Ptarmigan that were observed at or less than 400 m from the observer. Distances were measured using a rangefinder.
- observer: unique ID for primary observer of the survey. DS is Observer 1, KW is Observer 2, and MG is Observer 3.
- wind: standardized value of wind(mph) for that site during time of survey. Units: mph
- julian: day of year (not true julian date, but the count starting from January 1st) for that site during time of survey. Values standardized. Units: day
- min_after_sun: standardized minutes after end of civil twilight or when the sun reaches 0 degrees on the horizon that the survey at that site took place. Negative indicates survey was conducted before 'sunrise'. Units: minute
- tallshrub: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- tundra: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- tussock: standardized value of the percent of that land cover type within an 800m2 area of the survey point. Units: %
- elev: standardized elevation for each point. unstandardized elevation value is the mean elevation for land within 800 m x 800 m of survey point. Units: meter
Code/software
Software necessary: Program R, RStudio, and RTools
https://posit.co/download/rstudio-desktop/
https://cran.r-project.org/bin/windows/Rtools/
Packages necessary are listed in the code scripts.
Program and package versions used are listed in the manuscript.
File List:
06_WillowTimeModel.qmd: code containing model analysis and model fit for Willow Ptarmigan observations
07_RockTimeModel.qmd: code containing model analysis and model fit for Rock Ptarmigan observations
08_SquirrelTimeModel: code containing model analysis and model fit for Arctic ground squirrel observations
12_GyrOccupancyAnalysis.qmd: code containing occupancy model analysis for Gyrfalcon territories
13_GyrOccEval.qmd: code containing model fit and evaluation for the Gyrfalcon occupancy model
Access information
Other publicly accessible locations of the data:
Data was derived from the following sources:
- Land cover data derived from: https://daac.ornl.gov/ABOVE/guides/Annual_Landcover_ABoVE.html
Wang, J.A., D. Sulla-Menashe, C.E. Woodcock, O. Sonnentag, R.F. Keeling, and M.A. Friedl. 2019. ABoVE: Landsat-derived Annual Dominant Land Cover Across ABoVE Core Domain, 1984-2014. ORNL DAAC, Oak Ridge, Tennessee, USA. https://doi.org/10.3334/ORNLDAAC/1691
Methods
The study conducted multi-species point count surveys to detect bird species and Arctic ground squirrels over three years (2019-2022). Surveys began 30 minutes before sunrise and ended when at least ten points were surveyed or by noon. Each point was surveyed once. The surveys followed a time-removal approach, with specific time intervals acting as replicate survey periods. Observations were made for ten minutes, with a two-minute quiet period beforehand. The first detection of each individual within 400 meters was recorded, excluding subsequent detections of the same individual. Data collected included species, time, distance, bearing, group size, and sex. Individuals not actively using the area (e.g., flyovers) were excluded. Observers also measured temperature, wind speed, and direction at each point. Surveys were not conducted in winds over 24 km/h or in heavy rain or fog.
From 2016 to 2022 (excluding 2020 due to Covid-19), the study monitored 97 Gyrfalcon territories on nesting cliffs, conducting two surveys per territory each year between 1 May and 2 July. Surveys were performed using an R44 helicopter, following established protocols. A territory was considered potentially occupied if a Gyrfalcon adult, egg, or nestling was observed during the survey. Sixty surveys were not completed due to weather, and those were treated as missing data. The study also used prior surveys (2011-2015) to confirm whether territories were historically Gyrfalcon territories, as some were unoccupied or occupied by other raptors during the 2016-2022 period.
The study used vegetation data from the Arctic Boreal Vulnerability Experiment (ABoVE) to describe land cover on the Seward Peninsula. The ABoVE project classified vegetation into 15 types based on dominant vegetation derived from satellite imagery at a 30-meter grid resolution. The dataset from 2014 was used, and relevant habitat types for each species (Willow Ptarmigan, Rock Ptarmigan, and Arctic ground squirrels) were selected based on their preferred habitats. Vegetation types such as tundra, tussock, low shrub, and tall shrub, along with elevation, were chosen for the models. Data for each survey point included vegetation cover within a 0.64-km² area and mean elevation using a digital elevation model (DEM). Areas above 500 m elevation were excluded.
To estimate prey density inside each Gyrfalcon territory, we excluded cells within each territory with vegetation percentage values that fell outside the ranges used in our prey abundance models, and with elevations exceeding 500 m. We calculated the area size of each territory correcting for these removed cells. We summed expected density of each prey from the grid cells inside each Gyrfalcon territory and divided by the corrected territory area. For Willow Ptarmigan and Rock Ptarmigan, densities were expressed as males per square kilometer, whereas for Arctic Ground Squirrels, densities were reported as squirrels per square kilometer.