Data from: Come from away: Reconstructing the long-range migratory flight of spruce budworm moths to Newfoundland, Canada
Data files
Aug 28, 2025 version files 1.04 MB
-
About_the_NF_data_file.pdf
935.61 KB
-
Backward_trajectories.zip
72.35 KB
-
Meteorological_profiles_along_SBW_trajectory.zip
25.27 KB
-
README.md
5.64 KB
-
SBW_reconstructed_migratory_flight.zip
1.68 KB
Abstract
In this study, we combined Lagrangian atmospheric models with ecophysiological factors to reconstruct a migratory flight of spruce budworm (SBW, Choristoneura fumiferana) moths. This insect is a major defoliator in boreal forests in Eastern Canada and was captured in large numbers in a light trap on the night of 26-27 July 2017 at Sally's Cove, on the west coast of Newfoundland, Canada.
First, we used a simple trajectory model in backward mode in time to identify the most likely source region of the SBW moths south of Anticosti Island, in the Gulf of St. Lawrence, where moth take-off was likely facilitated by orographic lift.
Second, using a thermo-circadian model of take-off flight and the dispersion model MLDP (Modèle lagrangien de dispersion des particules), we determined that, during their 375 km journey, moths travelled a few hundred meters above surface level, under windy and low-humidity conditions.
Although the moths followed a trajectory too low to be tracked by a nearby weather radar, potential bioscatter reflectivities that could potentially be SBW bioscatter were detected at higher altitudes, suggesting that the moths that landed in Sally’s Cove could have been part of a larger dispersal event.
Dataset DOI: 10.5061/dryad.t1g1jwtfs
Description of the data and file structure
The present study focuses on a dispersal event detected by Rhainds et al. (2022) on the night of 26-27 July 2017, close to the town of Sally’s Cove, located on the west coast of Newfoundland (49.73 ° N, 57.93 ° W). We hypothesized that, because SBW populations were at low density levels everywhere on the island of Newfoundland at the time, the immigrants observed in Sally’s Cove must have come from high population density areas on the mainland. To test this hypothesis, we developed and applied a methodology to reconstruct SBW dispersal events from observed arrival location and time.
Since the per capita rate of dispersal is density-dependent, the potential sources of migrants are generally assumed to be areas of high population density (Anderson and Sturtevant, 2011; Sturtevant et al., 2013; Garcia et al., 2024). We used annual forest defoliation levels as an indicator of population density (Rhainds & Therrien, 2025).
To simulate the atmospheric dispersal of SBW, we used the Lagrangian modelling approach, which consists of tracking the motion of air parcels as they move through the atmosphere. A comprehensive explanation of the conceptual approach and applications of Lagrangian modelling, including air parcels, can be found in Lin et al. (2013). We used two Lagrangian models: a trajectory model executed in backward mode in time to identify the most likely source area and a dispersion model run in forward mode in time to reconstruct the moths' flight. Both models assume that the moths travel passively with the wind. The trajectory model, described in D’Amours et al. (2015), is a simple tool designed to quickly calculate a few air parcel trajectories (typically of the order of about ten) moving in the 3-D wind field of the atmosphere. In this model, only the advection by the mean wind (that is, atmospheric transport) is considered. The dispersion model used was the Lagrangian stochastic particle dispersion model MLDP (Modèle lagrangien de dispersion de particules). In MLDP, dispersion is modelled by calculating the trajectories of a large number of air parcels, typically several million, in two parts: 1) 3-D displacements due to the transport by the synoptic (non-turbulent) component of the mean wind provided by the meteorological model, and 2) 3-D displacements due to turbulent stochastic motions that are unresolved (at subgrid scale) by the meteorological model. Therefore, unlike the trajectory model, MLDP simulates the dispersion of a tracer, which allows the computation of concentrations of this tracer in the atmosphere, typically expressed as mass per unit volume of air.
We analyzed single-polarization C-band radar data from the region of interest to detect the presence of reflectivity signals resembling those of insects. Anticosti Island and the Jacques Cartier Strait lie outside the coverage of Canadian mainland radars, while the western region of Newfoundland is covered by the Marble Mountain radar (CXME), located at 48.930 ° N and 57.835 ° W, 90 km south of Sally’s Cove. Consequently, we extracted raw data from the Marble Mountain radar station (559 m above sea level, ASL; wavelength: 5.33 cm, beam width: 0.62°) for 26-27 July 2017, thus covering the night preceding the insect captures at Sally’s Cove as well as the night of capture. We processed the radar scans as described by Boulanger et al. (2017) using the BUGTRACKER package (Fabry and Hogg, 2020). In short, the four lowest elevation radar volume scans (0°, 0.2°, 0.4°, 0.5°) were read into Python, and signals from ground clutter (calculated using scans between 00 to 09 UTC on Jul 26), precipitation, and potential vertebrates (reflectivity > 25 dBZ) were censored. Scans were projected onto a 500 m x 500 m Cartesian grid, and a composite image of potential SBW activity within each grid cell was calculated using the maximum reflectivity across elevation scans.
Files and variables
File: About_the_NF_data_file.pdf
Description: Description and interpretation of the Mable Mountain (Newfoundland, Canada) radar data
File: SBW_reconstructed_migratory_flight.zip
Description: 3D position and environmental data of the reconstructed migratory flight trajectory of spruce budworm moths from the island of Anticosti to Sally's Cove (Newfoundland)
File: Meteorological_profiles_along_SBW_trajectory.zip
Description: Meteorological vertical profiles along the migratory flight of the spruce budworm moths. A text file is included that describes the data.
File: Backward_trajectories.zip
Description: Raw results from the backward trajectories starting at Sally's Cove. A text file is included that describes the data.
Access information
Other publicly accessible locations of the data:
Data was derived from the following sources:
Data file: Using the R package ‘terra’, the time series of raster maps in Azimuthal Equidistant Projection describing maximum reflectivity (calculated from the lowest 5 elevation scans) for each 10-minute interval from Jul 26 2000 UTC to Jul 27 0900 UTC were collated as a SpatRaster. This data object was then saved as a netCDF file. Ground clutter (but not precipitation) had been filtered from each raster using a radar-specific clutter mask.
Background: In 27 July 2017, a large increase in spruce budworm moths was detected in traps located along the east coast of Newfoundland (REF) and near the radar station located at Marble Mountain, Newfoundland (48.9303 N, -57.83462 E). Computer simulations indicate that these moths may have been derived from a mass dispersal event with a source population on Anticosti Island.
Ground clutter mask: Ground clutter was identified using scans from July 26, 0000-0900 UTC. Ground clutter on the two lowest elevation scans (0°, 0.2°) was identified on the east coast of Quebec (NNW of the radar station) and around the radar station itself in Newfoundland. On the three higher elevation scans (0.4°, 0.6°, 0.8°), there were small, scattered areas of ground clutter around the radar station (Figure 1).
Results: Sufficient data were available to search for migrating SBW on two nights: (a) Jul 26 0000 to 0900 UTC and (b) Jul 26 2000 to 0900 UTC:
(a) Jul 26 0000 to 0900 UTC: Two distinct patches of targets were identified (Figure 2):
1. At 0010hrs UTC there was a small area of moderate reflectivity (> 2dBZH) north of the radar station (west of Hawkes Bay) on the three lowest elevation scans (0°,0.2°, 0.4°). This patch moved towards the west and disappeared from view at 0220hrs over the Atlantic Ocean .
2. At 0530hrs, scattered low reflectivity (< -0.1 dBZH) over the Gulf of St Lawrence to the NNW of the radar station was present on the lowest elevation scan (0°). At 0620hrs this area of low reflectivity substantially increased in size and began to move SSE towards the radar station. At first these targets were only detected using the lowest elevation scan, but as the targets moved closer to the radar station they were also detected by higher elevation scans. This patch was still present on the final scan at 0900hrs UTC.
(b) Jul 26 2000 to Jul 27 0900 UTC: Four distinct patches of targets were identified (Figure 3):
1. Ground clutter was very poorly filtered from the scans, with large areas of high reflectivity remaining around the radar station, and on the Quebec mainland.
2. A patch of high reflectivity was detected on all elevation scans approximate 220km to the north of the radar station at 0010 UTC. This patch moved slowly to the east over time and disappeared from scans by 0810 hrs UTC. This patch describes precipitation clutter.
3. Patches of high reflectivity were detected on the two lowest elevation scans (0°, 0.2°) approximately 200km to the SSW of the radar station between 0010-0220 hrs, 0250-0510 hrs, and 0640-0820 hrs UTC.
4. A patch of high reflectivity over the Gulf of St Lawrence approximately 150km WNW of the radar station, was detected by the lowest elevation scan (0°) between 0010-0810 hrs UTC. This patch was also detected within the 0.2° elevation scans from 0100-0810 hrs UTC and on the 0.4° elevation scan between 0200-0.250 and between 0620-0810. On the composite scan, the relative density of high reflectivity cells, and the width of the high reflectivity patch, were both highest between 0120-0240 hrs.
Additional small areas of moderate reflectivity were detected within the 0.6° elevation scan (but not lower scans), close to the Quebec mainland between 0440-0600 hrs.
Discussion: The characteristics of this radar are not conducive to detecting SBW. At low elevations, most sectors are blocked by terrain (Figure 4), leaving only small segments to the south, west and north visible. Within these available sectors, the radar poorly samples the volume below 1000m where migrating SBW are typically located. The lowest radar beam (0°) does not sample air space below approx 500m a.s.l. (Figure 5), and at a distance of 150km from the radar station, does not sample airspace below 1000m a.s.l. The potential SBW targets in the fourth group are located in the 1000 to 3000m elevation range at 150m from the radar station, but because of blocked sectors it is not possible to determine where they departed from or landed. Radial velocity of these potential SBW on the 1.5° elevation Dopplar scan are between -2 < v ≤ 0 m/s, indicating stationary objects or objects with very slow radial movement toward the radar station. Wind direction during this period was WSW (257°) at an average wind speed of 10 m/s (NCEP reanalysis data).