Social behavior among nocturnally migrating birds revealed by automated moonwatching
Data files
Nov 02, 2023 version files 7.79 MB
-
allRotCrds.csv
4.89 MB
-
allRotTrks.csv
1.36 MB
-
group_migration_taxa_notes.csv
3.83 KB
-
GroupsTracks.csv
1.51 MB
-
README.md
7.46 KB
-
VideoTimes.csv
10.45 KB
Abstract
Migrating birds often fly in group formations during the daytime; whereas at night, it is generally presumed that they fly singly. However, it is difficult to quantify group behavior during nocturnal migration as there are few means of directly observing interactions among individuals. We employed an automated form of moonwatching to estimate percentages of birds that appear to migrate in groups during the night within the Central Flyway of North America. We compared percentages of birds in groups across the spring and fall and examined overnight temporal patterns of group behavior. We found groups were rare in both seasons, never exceeding 10% of birds observed, and were almost nonexistent during the fall. We also observed an overnight pattern of group behavior in the spring wherein groups were more commonly detected early in the night and again just before migration activity ceased. This finding may be related to changes in species composition of migrants throughout the night, or alternatively it suggests that group formation may be associated with flocking activity on the ground as groups are most prevalent when birds begin and end a night of migration.
https://doi.org/10.5061/dryad.0k6djhb62
This data set contains observations of migratory birds derived from automated moonwatching. This method entails making a video recording of the moon magnified by a telescope and subsequently applying a machine learning workflow to extract the screen coordinates of bird silhouettes captured by the video to assemble flight paths. Details on the hardware and analysis workflow are published elsewhere (https://doi.org/10.1093/icb/icac053 and https://doi.org/10.1016/j.ohx.2020.e00106).
Description of the data and file structure
The primary data consist of two comma separated text files. The file called "allRotCrds.csv" contains the coordinates of all detected bird silhouettes. This data set has five columns defined below:
- trackID - an alphanumeric identifier that links each silhouette to a bird track (sequence of silhouettes from consecutive frames that belong to the same bird). The trackID consists of the name of the video file from which the silhouette was detected, an underscore symbol, and then the number of the track that the silhouette belongs to.
- id - a number identifier unique within each video that was assigned to the silhouette as part of the sihouette extraction workflow.
- f - the sequential number of the video frame from which the silhouette was extracted
- size - a measure of the number of pixels that constitute the silhouette
- tier - an indicator of how distinct the silhouette was from the background. lower values are assigned to clearer silhouettes.
- xScr - x coordinate for the location of the silhouette on the screen (value is in pixels)
- yScr - y coordinate for the location of the silhouette on the screen (value is in pixels)
- xAdj - a modification of xScr wherein coordinates are rotated relative to the position of the moon such that the y axis is alighned with due North
- yAdj - a modification of yScr wherein coordinates are rotated relative to the position of the moon such that the y axis is alighned with due North
The other data file, called "allRotTrks.csv" contains summary information for each of the tracks represented across several data lines in the "allRotCrds.csv" dataset. Each line in "allRotTrks.csv" pertains to a flight path for an individual bird that was assembled from sequential detections of silhouettes across video frames that conformed to a linear pattern. This data set has the following columns:
- trackID - Same as above
- timeUTC - the date and time corresponding to the recording of the first silhouette detected in the track in coordinated universal time.
- sunSet - date and time of the sunset that immediately preceded the recording of the track in coordinated universal time
- sunRise - date and time of the sunrise that immediately followed the recording of the track in coordinated universal time
- minAfterSet - the. number of minutes between sunset and the recording of the track
- minBefRise - the. number of minutes between the recording of the track and sunrise
- locLon - longitude of the recording location.
- locLat - latitude of the recording location.
- startFrm - the first video frame that contains a silhouette that was included in the track
- endFrm - the first video frame that contains a silhouette that was included in the track
- noFrames - the total duration of the track in frames (i.e. endFrame minus startFrame plus 1)
- frmsDet - the number of frames in which a silhouette was actually detected (it was common for birds to go undetected in some frames, and their presence could only be inferred)
- durSec - the duration of the track in seconds
- moonSpd - the speed at which the silhouette moved across the moon, measured in moon diameters (pixels) per second
- scrnAzimuth - the direction of movement across the video screen in degrees
- avgPixSz - an average of the silhouette size across video frames (in pixels)
- moonElevation - the moon elevation angle in degrees up from the horizon at the time and place of the the recording of the track.
- moonAzimuth - the moon azimuth angle in degrees East of North at the time and place of the the recording of the track.
- moonDistance - the moon's distance from the earth at the time and place of the the recording.
- camRotCorr - a camera rotation correction value in degrees for aligning the x axis with the horizon.
- moonDiamPix - the maximum visible moon diameter in pixels
- vidDim - the height and width of the analyzed video frame in pixels. (As part of the analysis workflow, the video is cropped to a square that barely encompasses the moon. This value is the height and width of that square).
- bearing - the flight direction in degrees East of North.
- maxPix - the maximum number of pixels attributed to the silhouette across all frames in a track.
In analyzing our data it was necessary to quantify the total amount of video footage used. To do so we complied a simple data table called "VideoTimes.csv" with the following columns:
- vidName - the name of the video file.
- durMinutes - the duration of the video in minutes.
There is s a third dataset that is derived from "allRotTrks.csv. " This data set, called "GroupsTracks.csv" classifies all of the bird tracks as singletons or members of a group. The grouping is done using Ward hierarchical clustering to group individuals that are detected at about the same time and that demonstrate the same general flight direction and velocity. The code to generate this file is included in this repository ("groups_exploration_Final.R"). The columns for this file include all of those listed above in "allRotTrks.csv" plus the following:
- site - A four letter code for the location where the relevant footage was recorded (Same as the first four characters of each movie name.
- tempGrpID - A temporary group ID assigned during the analysis.
- GrpID - A unique group ID number shared by all birds deemed to be in a group together
- GrpSize - the group size (number of birds) that corresponds to the group ID
- yfSlope - Rate of silhouetted position change in the y axis, or the change in the y coordinate over the change in frame number.
- xfSlope - Rate of silhouetted position change in the x axis, or the change in the x coordinate over the change in frame number
- minDist - the minimum distance in pixels to the nearest bird in the group
- maxDist - the maximum distance in pixels to the nearest bird in the group
A final data set includes taxonomic notes from an effort to determine whether each group of birds observed was composed of passerines or non-passerines. These data were generated manually by viewing the relevant video frames and make a determination based on body shape and flap rate. The data set has only three columns:
- Group - a group identifier that includes the relevant video file and the group number.
- species - an imprecise taxonomic determination, either passerine, nonpasserine, or waterfowl.
- GrpNo - the group number extracted from Group.
Sharing/Access information
Data and analysis files are also available on OSF:
Code/Software
This dataset was analyzed in R. The relevant code are available on the open science framework: https://osf.io/bv5hj
We used the LunAero system (Honeycutt et al. 2020;m https://doi.org/10.1016/j.ohx.2020.e00106) to record video footage of the moon at sites in Central Oklahoma including the Kessler Atmospheric and Ecological Field Station (34.98°N, 97.52°W); private residences in Norman, Oklahoma (35.22°N, 97.44°W); and the U.S. Department of Energy Atmospheric Radiation Measurement Southern Great Plains Lab (36.61°N, 97.48°W). These sites aligned north to south and were all within the North American Central Flyway (Buhnerkempe et al. 2016). Much of the video footage we collected was unusable due to cloudy skies and/or technical issues with the hardware. Hence, we screened the footage to identify recordings where the video was suitable for further analysis. All footage used in our analyses were collected within 5 days of a full moon or when the moon was at least 66% illuminated. The steps used to extract, filter, and validate flight paths from our video footage are described in Honeycutt and Bridge (2022; https://doi.org/10.1093/icb/icac053). This workflow yielded a dataset wherein all detected bird silhouettes were incorporated into individual flight paths along with their screen coordinates (x and y) and timestamps.
- Bridge, Eli et al. (2023), Social behavior among nocturnally migrating birds revealed by automated moonwatching, , Article, https://doi.org/10.5281/zenodo.10070098
- Bridge, Eli S; Honeycutt, Wesley T; Chen, Angela J et al. (2024). Social behavior among nocturnally migrating birds revealed by automated moonwatching. Ornithology. https://doi.org/10.1093/ornithology/ukad055