Arctic migrating barnacle geese utilise accommodation fields in a new agricultural staging area
Data files
Nov 25, 2024 version files 3.79 MB
-
Code.R
5.48 KB
-
Data_after_HMM.RData
891.86 KB
-
Data_for_HMM.RData
830.04 KB
-
Data_for_iSSA.RData
1.03 MB
-
Field_selection_algorithm.R
2.98 KB
-
README.md
6.33 KB
-
Sampled_data_model1.RData
288.45 KB
-
Sampled_data_model2.RData
701.90 KB
-
Sampled_data_model3.RData
36.08 KB
Abstract
The recovery of threatened species after conservation measures can lead to human-wildlife conflicts. One example of such is the recent population growth of the Barnacle Goose Branta leucopsis, a large herbivorous bird. During migration, geese stage in large numbers on agricultural fields in range countries and cause substantial damage to farms. A combination of repelling fields, where geese were chased off by humans, and accommodation fields, which provide refuge sites for foraging geese, has been suggested as an effective management tool to mitigate conflicts.
Using high-resolution satellite tracking data, we investigated habitat selection of 41 barnacle geese staging in Northern Karelia, Finland, during spring 2021. We estimated relative habitat use by these geese and conducted a fine-scale analysis of their use of different fields by employing Hidden Markov Models and integrated step-selection analysis. Fields included normal crop (no goose management), project and other (private and Nature 2000 area) accommodation fields, and repelling fields. Project accommodation and repelling fields were established on areas known to have a long history of high grazing pressure by barnacle geese.
We found that behavioural data of geese can be categorized into three different states (static, slow, and fast movement). Static and slow states were used for local field selection, fast state for field selection in the regional area, and all states for field selection after leaving a repelling field.
Overall, relative habitat use indicated that geese utilise accommodation fields more than expected by their availability. Integrated step-selection analyses revealed that geese avoided normal and repelling versus project accommodation fields at the regional scale. At the local scale, they preferred project accommodation fields over all other fields. After leaving a repelling field, geese did not show preference for any accommodation over repelling fields.
Synthesis and application: Geese show individual selection for accommodation fields compared to normal or repelling fields across several scales. Our results suggest that the accommodation field concept – consisting of refuge areas and no-go areas where geese are repelled from – can help to mitigate the human-wildlife conflict using local stakeholders' knowledge.
https://doi.org/10.5061/dryad.tb2rbp0bt
Description of the data and file structure
This data was used to address the functionality of accommodation fields in mitigating the damage at a new staging site (Northern Karelia, Finland) of barnacle geese belonging to the population breeding in the Russian Arctic. Birds were trapped during early May 2021 with a cannon net while they were grazing on fields. We equipped 50 adult barnacle geese (30 males and 20 females) with solar-powered GPS-GSM/GPRS transmitter neckbands (OrniTrack OT-NL40-3GC, Ornitela, UAB, Lithuania) and individually numbered metal leg rings. GPS resolution was set to record positions every 10 min and the transmission interval was set to 1 h. Data on location and delimitation of individual fields as well as information on field use (crop) were obtained from the Finnish Food Authority. Fields were classified into normal fields (two categories: edible and non-edible according to geese’ foraging preferences), accommodation fields (two categories: project vs. private and Natura 2000 fields), and repelling fields. The lake information was obtained from the National Land Survey of Finland Topographic Database 12/2022.
To open the .RData files, please open the script (Code.R) with R in RStudio first and use the given readRDS commands for reading/opening the data. The data can then be viewed in RStudio or saved in another format if needed.
Files and variables
File: Code.R
Description: This R file contains the code that we used to generate the HMM and iSSA results presented in the paper.
File: Field_selection_algorithm.R
Description: This R file contains the algorithm for our modified iSSA approach to reflect our interest in specific field selection by foraging geese at different scales. In general, instead of sampling available locations based on step length and turning angle distributions, we sample from available fields. Hence, for each step that ends in a field, we chose a circular area with a radius of 5 km from the starting point of the step (or GPS-location). For each field within that radius (also including fields that are only partly within this area), we calculated the distance to the starting point. When sampling these fields randomly, the size of the field and the distance to the starting point were used as sampling weights. This means that larger fields are more likely to be chosen, reflecting geese’ general preference for larger fields. For the distance to the starting point, we used a gamma distribution fitted to the empirical distribution of step lengths as the probability distribution. Hence, closer fields are favoured over the more distant ones.
File: Data_for_HMM.RData
Description: The cleaned GPS data to be used for running the hidden Markov model.
Variables
- “ID”: burst ID (a composite number of “individual” and “burst”)
- “step”: step length in metres
- “angle”: turning angle of two consecutive steps in radians
- “individual”: ID number of the GPS transmitter
- “time”: date and the time when the GPS location was recorded (UTC)
- “burst”: burst number
- “dist_gf”: distance to the nearest accommodation field (gf = goosefield)
- “x”: longitude of the recorded location
- “y”: latitude of the recorded location
File: Data_after_HMM.RData
Description: Same as “Data_for_HMM.RData” but with the attached states and habitat locations.
Variables
- “ID”: burst ID (a composite number of “individual” and “burst”)
- “step”: step length in metres
- “angle”: turning angle of two consecutive steps in radians
- “individual”: ID number of the GPS transmitter
- “time”: date and the time when the GPS location was recorded (UTC)
- “burst”: burst number
- “dist_gf”: distance to the nearest accommodation field (gf = goosefield)
- “x”: longitude of the recorded location
- “y”: lattitude of the recorded location
- “state3”: state associated to the step (1 = static movement; 2 = slow movement; 3 = fast movement)
- “habitat”: habitat in which the GPS observation was located
File: Data_for_iSSA.RData
Description: The step data used for running the integrated step (or here rather field) selection analyses.
Variables
- “x1_”: starting longitude of the step
- “x2_”: ending longitude of the step
- “y1_”: starting latitude of the step
- “y2_”: ending latitude of the step
- “sl_”: step length in metres
- “ta_”: turning angle between the previous step and this one in radians
- “t1_”: starting date and time of the step
- “t2_”: ending date and time of the step
- “state3_start”: HMM state when the step started
- “state3_end”: HMM state when the step ended
- “habitat_start”: habitat in which the step started
- “habitat_end”: habitat in which the step ended
- “hour_start”: hour of the day when the step started (24 hours, UTC)
- “tod_start””: time of the day when the step started (day vs night)
File: Sampled_data_model1.RData
Description: Ready sampled data to be imported to R to reproduce the results for model 1 (field selection at a regional scale).
Variables
- “x1_”: starting longitude of the step
- “x2_”: ending longitude of the step
- “y1_”: starting latitude of the step
- “y2_”: ending latitude of the step
- “state3_start”: HMM state when the step started
- “habitat_start”: habitat in which the step started
- “sl_”: step length in metres
- “habitat_end”: habitat in which the step ended
- “lsize_end”: size of the field in which the step ended (in square meters)
- “case”: indicator for whether the step is observed (TRUE) or sampled (FALSE)
- “step_id_”: grouping number for observed step and it’s sampled replicates
File: Sampled_data_model2.RData
Description: Ready sampled data to be imported to R to reproduce the results for model 2 (field selection at a local level).
Variables
Same as for Sampled_data_model1.RData
File: Sampled_data_model3.RData
Description: Ready sampled data to be imported to R to reproduce the results for model 3 (field selection after geese leave from a repelling field).
Variables
Same as for Sampled_data_model1.RData and Sampled_data_model2.RData