Data from: Seeing in the dark: Using thermal imaging to directly observe nocturnal migration
Data files
Dec 16, 2025 version files 435.28 KB
-
Counts_of_nocturnal_migrants_Cape_May_Sep-Oct2022.csv
426.90 KB
-
README.md
8.38 KB
Abstract
Nocturnal migration has fascinated and puzzled ornithologists for centuries. Today, using technologies from weather surveillance radar to multi-sensor geolocators, we can study the continental magnitude of these flights and the intricate details of individual journeys. Yet, we still lack a way to directly observe migrants, hampering our ability to understand migration at the individual and species level. Combining recent advances in thermal imaging optics and digital photography, we detected, illuminated, and identified nocturnally migrating birds at low altitudes (0–300 m). We describe insights this technique has to offer using ~200 observation hours during three autumns (2020–2022) at Cape May, New Jersey, USA. Our novel approach allowed us to observe migratory behavior at night, identify thousands of nocturnally migrating birds, and begin quantifying the passage of silent individuals and silent species. Aside from a few highly vocal families—primarily thrushes, sparrows, and wood-warblers—the vast majority of nocturnally migrating birds passed silently or inaudibly over Cape May. With acoustic-only monitoring, all of these individuals and nearly two-thirds of families (62 %) would have been missed. In our sample, social behavior was restricted to waterbirds, and no compact flocks of passerines were observed more than an hour after dusk, even for diurnally gregarious species. Because this labor-intensive technique cannot yet be automated, it does not lend itself to broad-scale migration monitoring. However, when integrated with acoustic monitoring, the local, fine-scale resolution of these data can complement and ground-truth both radar and acoustic studies. These portable, easily deployable technologies are particularly well-suited when nocturnal migrants are nearest the ground and may help to visualize how birds interact with wind turbines and other tall structures. By transforming the invisible into the visible, this method provides a new window into the study of nocturnal migration.
Dataset DOI: 10.5061/dryad.tmpg4f5cz
Description of the data and file structure
This dataset contains count data of nocturnal migrants from 32 well-annotated observations (eBird checklists) collected over 20 nights (65.3 hours) between 20 September and 28 October 2022. These observations were recorded from two primary vantage points in Cape May, New Jersey, USA: Cape May Point State Park (38°55'58'' N, 74°57'28'' W) and the western dunes of Cape May city (38°55'48'' N, 74°55'29'' W). Nocturnally migrating birds were identified using a combination of criteria: photographs of illuminated birds, expert visual identifications (with binoculars) of illuminated birds, and expert auditory identifications of distinctive call notes. Within this Sep-Oct 2022 dataset, 105 unique species or species complexes were identified from 26 avian families. Please see Table S2 and Table S3 in the Supplementary Material of the associated publication for those condensed summaries, respectively. In all, 5355 free-flying individuals were identified to at least the family level.
Files and variables
File: Counts_of_nocturnal_migrants_Cape_May_Sep-Oct2022.csv
Description: This long-form spreadsheet has been simplified, cleaned, and curated from the original dataset that was downloaded from eBird (username: capemaync). Most notably, we have manually removed all records of stationary individuals (e.g., birds singing at dusk or roosting on a nearby pond), focusing exclusively on individuals that appeared to be in active nocturnal migration. We also eliminated any eBird entries that lacked details about how a particular species was detected (see "Detection.Method" below). With the exception of "Detection.Method", all remaining columns are standard in downloads of individual eBird accounts.
Variables
- Submission.ID: The unique eBird checklist ID generated for a specific observer x observation combination. Note that, even if all checklists are identical, this can mean that six different observers would have six different checklist IDs for the same observation period. In our case here, these unique checklist identifiers belong to a joint eBird account (username: capemaync). By appending these alphanumerics to the following base url https://ebird.org/checklist/, you can then access each checklist online (e.g., https://ebird.org/checklist/S119476315).
- Common.Name: The common name of a species or taxon according to the 2023 eBird/Clements taxonomy: https://www.birds.cornell.edu/clementschecklist/introduction/updateindex/october-2023/download Note that this list includes other taxonomic entities, such as birds not identified to species (e.g., Yellow-billed/Black-billed Cuckoo) as well as other individuals identified to higher taxonomic units (e.g., Catharus sp., wren sp.).
- Scientific.Name: The scientific name of a species or taxon according to the 2023 eBird/Clements taxonomy: https://www.birds.cornell.edu/clementschecklist/introduction/updateindex/october-2023/download
- Taxonomic.Order: The taxonomic position of a species or taxon according to the 2023 eBird/Clements taxonomy: https://www.birds.cornell.edu/clementschecklist/introduction/updateindex/october-2023/download
- Count: The number of individuals that were registered during a particular observation period (eBird checklist). Note that, in some cases, these will not match the associated eBird checklist if, for instance, some individuals of that species were detected roosting on a pond, in which case those individuals were subtracted from the overall count.
- Location: Location names for one of three eBird hotspots around Cape May, New Jersey: Cape Island--Cape May (city), Cape Island--Cape May Point SP (CMPSP), or the Cape May–Lewes Ferry Terminal.
- Date: One of 20 dates between 20 September and 28 October 2022. Note that some of these nights contain multiple observation periods (checklists).
- Time: The start time (12-hour clock) for each observation period.
- Protocol: In this case, all eBird checklists were conducted using the "Stationary Count" protocol: https://support.ebird.org/en/support/solutions/articles/48000950859-guide-to-ebird-protocols#anchorStationary
- Duration..Min.: The duration (in minutes) of each observation period. On the eBird website, these times are presented in hours and minutes instead; thus, 172 minutes equals 2 hours, 52 minutes.
- All.Obs.Reported: In eBird parlance, this prompt asks whether or not the observation is a "complete checklist": "A Complete Checklist is any eBird list where birding was your primary purpose, and every species you could identify to the best of your ability, by sight and/or sound, is reported." In this spreadsheet, all checklists are complete ('1'). See the following help page for more information: https://support.ebird.org/en/support/solutions/articles/48000967748-birding-as-your-primary-purpose-and-complete-checklists
- Number.of.Observers: The number of observers during an observation period, including joint eBird accounts such as "capemaync", which do not pertain to an individual person.
- Detection.Method: This is the only column added to a standard eBird export and the most important variable for this study. For all taxa included here, one of four different categories will be listed:
-
VAM: Visual of Active Migrant. In this case, all individuals for a given taxon (row) were sight-only identifications.
-
NFC: Nocturnal Flight Calls. In this case, all individuals for a given row were only identified using auditory cues (e.g., species-specific call notes).
-
Both: In this case, all individuals in a given row were identified using both sight and sound. An example of this is two American Bitterns (Botaurus lentiginosus) on 5 October 2022, both of which were seen and heard calling. Within this dataset, only 10 individual birds fit this criteria.
-
Mixture: In this case, at least one individual for a taxon was seen and another heard, but the annotations (Observation.Details) were insufficient to distinguish how many individuals fall into each category. For example, on 23 September 2022, 27 Common Yellowthroats (Geothlypis trichas) were recorded, but the details for that specific observation simply say "mix of VAM/ NFC". Because we are unable to determine how many birds were VAM versus NFC, we simply lumped these observations together as "Mixture". Two large flights of Yellow-rumped Warblers (Setophaga coronata) on 14 and 18 October 2022 (300 on each night) account for just over half of all individuals in this category.
Finally, note that in this spreadsheet, the same taxon might be listed twice during a single observation period if individuals were registered using more than one detection mode. One such example is Common Yellowthroat on 20 September 2022: one row lists the 23 individuals that were only identified using visual cues (VAM) and another row lists the 17 individuals that were only identified using auditory cues (NFC).
-
- Checklist.Comments: A text narrative at the start of each checklist that often details the weather conditions (e.g., wind, temperature, cloud cover), flight direction, and the various observers present.
- Observation.Details: This column contains the critical details that were used to populate the "Detection.Method" column, as well as other interesting notes.
Access information
Other publicly accessible locations of the data:
- For more information, please see the online Supplementary Material connected to the associated publication.
Data was derived from the following sources: