Dataset DOI: 10.5061/dryad.j9kd51cr5

Description of the data and file structure

---

Principal Investigator Contact Information

• Name: Jessie Lilly
• Email: jessiemlilly@gmail.com
• Institution: Fisheries and Oceans Canada

Alternate Contact Information

• Name: Martha Robertson
• Email: martha.robertson@dfo-mpo.gc.ca
• Institution: Fisheries and Oceans Canada

Dataset Overview

This dataset contains the data and code required to replicate the analyses in Lilly et al. (2026) all analyses were conducted in R .

The results presented in this paper were part of a larger project co-developed by a collaborative network of researchers from 14 organizations (Indigenous, Government, NGOs, and academia) with the goal of improving knowledge of Atlantic salmon migration routes at sea.

The current three-year (2021 to 2023) study focused on the early marine migration routes of smolts from 32 river populations throughout eastern Canada to:

1. Determine their migration route from their natal river to the Labrador Sea;
2. Describe the timing of migration in relation to modelled sea surface temperature (SST); and
3. As the data allowed, estimate smolt survival from rivers within the Gulf of St. Lawrence to the Labrador Sea.

Summary of Key Results

• The route taken from natal rivers to feeding and overwintering areas was mainly dependent on whether the population was from within or outside the Gulf of St. Lawrence.
• In general, smolts from within the Gulf migrated through the Strait of Belle Isle, whereas other populations migrated north along the continental shelf and east of Newfoundland.
• Despite significant variation in spring smolt migration timing within the studied latitudinal range, populations showed temporal overlap in marine regions along their route at preferred water temperatures between 4–10ºC.
• Smolts avoided very cold water (< 3ºC) and delayed their northward migration to await the slower seasonal warming of the Newfoundland and Labrador Shelf.

Data Spatial Scope

Tagging metadata

Atlantic salmon smolts were captured in rivers during their spring (April to June) seaward migration throughout eastern Canada (Fig. 1).From 2021 to 2023, 3900 smolts (2021: n = 1239, 2022: n = 1334, 2023: n = 1327) were captured and tagged from 32 rivers (2021: n = 19, 2022: n = 24, 2023: n = 21; Fig 1).

For summary purposes, rivers were categorized by their respective Species at Risk Act (SARA) designatable unit (DU), which groups river populations based on geographic or genetic distinctness: Labrador (DU2), Northeast Newfoundland (DU3), South Newfoundland (DU4), Northwest Newfoundland (DU6), Quebec Western North Shore (DU8), Inner St. Lawrence (DU10), Gaspé-Southern Gulf of St. Lawrence (DU12), Eastern Cape Breton (DU13), Nova Scotia Southern Upland (DU14), Inner Bay of Fundy (DU15), Outer Bay of Fundy (DU16).

Receiver metadata

Acoustic transmitter data were collected using both stationary and mobile acoustic receivers (InnovaSea Ltd., Nova Scotia, Canada) deployed as part of the current study and 45 other unrelated projects (Fig.2). The majority of the data were managed and shared through the Ocean Tracking Network (OTN; Dalhousie University, Halifax, Canada). Project details are provided in Table S2.

Funding

This research was supported by the Environmental Studies Research Fund (ESRF; esrfunds.org) to Fisheries and Oceans Canada, Newfoundland and Labrador Region.

Project: 2019-01S: Atlantic salmon in the Eastern Canadian offshore regions (ESRF Regions 8 to 15): timing, duration and the effects of environmental variability and climate change.

Animal Care Approval

The capture and tagging methods used in the current study followed the Canadian Council on Animal Care guidelines and were approved by Fisheries and Oceans Canada Regional Animal Care Committees:

• Northwest Atlantic Fisheries Centre
  • Animal Use Protocols: NAFC-2021-02, NAFC-2022-02 (renewal), NAFC-2023-02
• Maritimes Region
  • AUP-MAR-21-29, AUP-MAR-22-25, AUP-MAR-23-08
• Gulf Region
  • AUP-GULF-21-30, AUP-GULF-22-40

Data Sources

Acoustic telemetry data

Acoustic telemetry data was obtained from matched detection files supplied by the Ocean Tracking Network based on the Atlantic salmon smolts tagged as part of this project (see: OTN Detection Extract Documentation).

---

Temperature data

Daily SST at receiver locations was assessed using an ocean circulation model configuration for the Western North Atlantic (WNA). The model is an implementation of the Nucleus for European Modelling of the Ocean (NEMO 3.6 model; Madec et al., 2016). The domain covers the North Atlantic with a 1/4° horizontal resolution and includes a child domain of the WNA with a horizontal resolution of 1/12° with two-way coupling (see DNAODS: DFO North Atlantic Ocean-ice Downscaling System in Han et al., 2022).

The model successfully generated seasonal and interannual variations of major shelf-scale currents, such as the Labrador Current and Nova Scotian Current, and adequately reproduced SSTs (Dufresne et al., 2025).

For this study, the model output from the child domain only was used, which covers most of the region occupied by the North American Atlantic salmon.

Recommended Citation

Lilly, J., Burke, C., Dufresne, C., Lavoie, D., April, J., Carr, J ... Robertson, M.J. Atlantic salmon post-smolt migration from Canadian rivers to the Labrador Sea: routes, timing and influence of sea surface temperature. Canadian Journal of Fisheries and Aquatic Sciences.

Description of the data and file structure

Multiple files in this repository require R and R Markdown to generate. Users are provided with the appropriate scripts and datasets to run the analyses should they choose to do so.

All datasets used to run each script were generated from the master_data.csv file and are described in more detail below.

master_dataset.csv

• Number of variables: 25
• Number of rows: 1,675

Variable list

• serialno_tag (character):
  The unique tag identifier of each post-smolt.

• transmitter (character):
  The manufacturer prefix, code set, and unique tag identifier (e.g., A69-1602-44705).

• Date_Hit (POSIXt):
  The initial timestamp when an Atlantic salmon smolt was detected at each ecoregion quadrant, main receiver line or estuary included in this study. 

• SerialNo_Rec (character):
  The unique serial number of each acoustic receiver.

• Station_Name (character):
  Name of each acoustic receiver station. NA values indicate that the station name was not provided by the contributing project; only the geographic coordinates (latitude and longitude are available). 

• Latitude (numeric):
  Latitude of the deployed acoustic receiver.

• Longitude (numeric):
  Longitude of the deployed acoustic receiver.

• Location (character):
  The location of each acoustic receiver, including ecoregions, main receiver lines, and estuaries. This variable was used to describe post-smolt migration routes by grouping detections into broader spatial units. 

  Location values are not restricted to a fixed set of categories and include a mixture of: 

  • Ecoregions (e.g., Labrador Shelf, Scotian Shelf)

  • Subdivided ecoregion quadrants (e.g., Gulf of St. Lawrence 1-2, Grand Banks 1-4)

  • Specific receiver lines (e.g., Halifax Line, Strait of Belle Isle)

  • Estuaries (e.g., Fortune Bay, Placentia Bay)

    The primary ecoregions used in this study include: 

    • Gulf of St. Lawrence (GS1–2)

    • Labrador Shelf (LBS1–4)

    • Newfoundland Shelf (NLS1–4)

    • Grand Banks (GRB1–4)

    • Scotian Shelf (SCS1–4)

• Location_abr (character):
  Abbreviated label representing each quadrant of an ecoregion, main receiver line, or estuary.

• temp (numeric):
  The daily average sea surface temperature (°C) at the time of detection, matched to nearest receiver grid.

• acoustictagfamily (character):
  Type of acoustic tag used on the detected smolt.

• tag_life_days (numeric): The estimated duration the tag is battery is estimated to last (in days). 

• min_delay (numeric):
  The shortest programmed time between tag transmissions (seconds).

• max_delay (numeric):
  The longest programmed time between tag transmissions (seconds).

• cosewic_name (character):
  Species at Risk Act (SARA) designatable unit name for the smolt's river of origin.

• cosewic_du (character):
  SARA designatable unit number.

• specific_river_name (character):
  Specific river or tributary name where smolts originated.

• river (character):
  Mainstem river name (tributaries grouped under the same main river).

• lat_capture (numeric):
  Latitude where smolts were captured.

• long_capture (numeric):
  Longitude where smolts were captured.

• lat_release (numeric):
  Latitude where smolts were released.

• long_release (numeric):
  Longitude where smolts were released.

• date_capture (POSIXt):
  Date when smolts were captured.

• fl_mm (numeric):
  Fork length (mm) of the smolt.

• ww_g (numeric):
  Weight (g) of the smolt.

figure_3_example.zip

Figure_3_data.csv

This dataset was produced using the R package VTrack(v2.4.0; Campbell et al. 2012). 

To assess general migration patterns of Atlantic salmon post-smolts during their early marine phase, a migration route was defined as a sequence of consecutive detections of an individual across two or more distinct spatial groupings, including ecoregions and/or quadrants, estuaries, or main receiver lines. Accordingly, routes could only be identified for post-smolts detected in multiple distinct spatial groupings. 

The full dataset and associated VTrack code are not included, as they are currently being used in ongoing analyses and manuscript preparation. The dataset provided here contains all information required to reproduce Figure 3. The full dataset can be made available by the authors upon reasonable request. 

Migration routes were extracted using the RunResidence function from the VTrack package in R. For this analysis, duplicate routes among post-smolts originating from the same river were removed to avoid redundancy.

The resulting non-resident event file was converted back into a standard data frame format to facilitate plotting, such as for generating Figure_3_data.csv.

Variable list

• serialno_tag (character):
  The unique tag identifier of each post-smolt.
• date_hit (POSIXt):
  The timestamp when an Atlantic salmon smolt was detected at an acoustic receiver.
• Location (character):
  The ecoregion, main receiver line, or estuary in which each receiver was located. Ecoregions in this study included the Gulf of St. Lawrence, Davis Strait, and the four continental shelves surrounding the east coast of Canada: Scotian Shelf, Grand Banks, Northeast Newfoundland Shelf, and Labrador Shelf.
• Latitude (numeric):
  Latitude of each acoustic receiver. 
• Longitude (numeric):
  Longitude of each acoustic receiver. 
• river (character):
  Mainstem river name (tributaries grouped under the same main river).
• lat_release (numeric):
  Latitude where smolts were released.
• long_release (numeric):
  Longitude where smolts were released.
• date_capture (POSIXt):
  Date when smolts were captured.
• fl_mm (numeric):
  Fork length (mm) of the smolt.
• ww_g (numeric):
  Weight (g) of the smolt.

Folder: figure_4.zip

figure_4_data.csv

This dataset was subset from the "master_dataset.csv" file and contains the daily detections of Atlantic salmon smolt at acoustic receivers. To create this dataset, duplicate daily detections of the same smolt on an acoustic receiver were removed.

Figure 4, created by C. Dufresne and D. Lavoie, depicts the monthly detection locations of acoustically tagged Atlantic salmon smolt (black circles), along with the corresponding mean sea surface temperatures extracted from the Nucleus for European Modelling of the Ocean (NEMO) 3.6 model (Madec et al., 2016), from May to August over the three years 2021 to 2023.

The NEMO oceanographic model and the code used to produce this figure are currently being used by C. Dufresne and D. Lavoie to generate results for another manuscript and cannot be shared at this time.

• Number of variables: 8
• Number of rows: 10911
• serialno_tag (character):
  The unique tag identifier of each post-smolt.
• transmitter (character):
  The manufacturer prefix, code set, and unique tag identifier (e.g., A69-1602-44705).
• date (POSIXt):
  The date (yyyy-mm-dd) when an Atlantic salmon smolt was detected at an acoustic receiver.
• serialNo_rec (character):
  The unique serial number of each acoustic receiver.
• deploy_lat (numeric):
  Latitude of the deployed acoustic receiver.
• deploy_lon (numeric):
  Longitude of the deployed acoustic receiver.
• location (character):
  The quadrant of each acoustic receiver within an ecoregion.
  Used to describe post-smolt migration routes by subdividing ecoregions:
  • Gulf of St. Lawrence (GS1–2)
  • Labrador Shelf (LBS1–4)
  • Newfoundland Shelf (NLS1–4)
  • Grand Banks (GRB1–4)
  • Scotian Shelf (SCS1–4)
  • Davis Strait (DAST)
• location_abr (character): Abbreviated ecoregion quadrant code.

Folder: figure_5.zip

figure_5_temp_data.csv

This dataset is used in figure_5.Rmd and contains the average daily SSTs that were extracted from the Nucleus for European Modelling of the Ocean (NEMO) 3.6 model (Madec et al., 2016) grid cell located closest to the acoustic receiver with the highest daily number of post-smolt detections in each location.

• Number of variables: 5
• Number of rows: 6470
• temp (numeric):
  The daily average sea surface temperature (°C) at the time of detection, matched to nearest receiver grid
• date (POSIXt):
  The date (yyyy-mm-dd) when an Atlantic salmon smolt was detected at an acoustic receiver.
• Latitude (numeric):
  Latitude of the deployed acoustic receiver.
• Longitude (numeric):
  Longitude of the deployed acoustic receiver.
• Location (character):
  The subdivision of an ecoregion or the main receiver line where detections occurred. Ecoregions in this study included the Gulf of St. Lawrence, Davis Strait, and the four continental shelves surrounding the east coast of Canada: Scotian Shelf, Grand Banks, Northeast Newfoundland Shelf, and Labrador Shelf.

Folder: CMR_Analysis

cmr_data.csv

This dataset contains the capture histories of post-smolts at the Strait of Belle Isle line and is used in cmr_analysis.R.

• Number of variables: 13
• Number of rows: 731
• River (character):
  Mainstem river name (tributaries grouped under the same main river) where smolts originated from.
• Specific_River_Name (character):
  Mainstem river name or tributary name of each River. 
• Release (numeric):
  A column indicating the smolt was released.
• Estuary (numeric):
  A column indicating whether a smolt was detected in their natal estuary (1 = detected, 0 = not detected).
• SOBI_1 (numeric):
  A column indicating whether a smolt was detected at the first SOBI receiver line (1 = detected, 0 = not detected).
• SOBI_2 (numeric):
  A column indicating whether a smolt was detected at the second SOBI receiver line (1 = detected, 0 = not detected).
• tag_group (character):
  The acoustic tag model type.
• tag_power (character):
  The transmit power level of the tag (i.e., how strongly the tag emits its acoustic signal (ping)).
• tag_mindelay (numeric):
  The shortest programmed time between tag transmissions (seconds).
• fl_mm (numeric):
  Fork length (mm) of the smolt.
• ww_g (numeric):
  Weight (g) of the smolt.

Code/software

Language and Environment

R environment for Statistical Computing

Version

R 4.4.3

Folder: figure_3_example_zip

Figure_3_example_du14.RMD

This code produces a map of the east coast of Canada using the basemap function from the ggOceanMaps package (v.2.2.0; Vihtakari, 2024). Routes between ecoregion/quadrants, estuaries and main receiver lines were added with the geom_segment function in the ggplot2 package (v4.0.1; Kahle and Wickham, 2023). The proportion of post-smolts from each river detected at ecoregions/quadrants, estuary and main receiver line—expressed as a percentage of the total number tagged in each river/DU—was visualized on the same map using scale_fill_gradientn in ggplot2. Yellow to red, pink to purple, and light to dark green circle gradients represent the percentage of post-smolts detected in each ecoregion/quadrant, estuary, or main receiver line during 2021, 2022, and 2023, respectively. When post-smolts were detected at the same location across multiple years, the circle produced by scale_fill_gradientn was subdivided manually in Microsoft PowerPoint to depict detections across years. To adjust for routes drawn by geom_segment that overlapped land areas, route maps were imported into PowerPoint, and the routes were re-drawn using the “curve” tool.

Version

R 4.4.3

Dependencies

• tidyverse
•lubridate •sp •sf •maps •ggspatial •ggOceanMaps •ggnewscale •grid

Folder: figure_5.zip

Figure_5.RMD

This R Markdown file contains the code used to produce the output shown in Figure 5 of the manuscript. Figure 5. is a Violin plot depicting the dates (day-month) and daily average sea surface temperatures (SST) for tagged Atlantic salmon post-smolts from each designatable unit, across ecoregions, main receiver lines, and Southern Newfoundland Bays. 

Version

R 4.4.3

Dependencies

• readr
• dplyr
• doBy
• janitor
• viridis
• ggplot2
• tidyverse •gridExtra

Folder: CMR_analysis.zip

cmr_analysis.R

This R file contains the code used to produce the capture-mark-recapture (CMR) survival analysis results.

Jointly estimated survival and detection rates were obtained using traditional CMR analytical approaches (Lebreton et al., 1992) implemented via the RMark package (v3.0.6; Laake, 2013) in R version 4.3.1 (R Core Team, 2023).

The CMR approach follows individuals over time, sampled at successive capture or detection occasions. In this study, time and location were treated as exchangeable, with detections possible on the SOBI-1 and SOBI-2 acoustic receiver lines. Three-character encounter histories of 1s (“detected”) and 0s (“not detected”) were created for each tagged post-smolt.

Detection was modeled with a tag type effect (three levels) and an additive location effect (two levels: SOBI-1 and SOBI-2). A combined detection probability for both SOBI receiver lines was calculated as the complement of the product of the probabilities of non-detection on either line, for each tag type.

Survival was modeled with a river/tributary of tagging effect (14 levels). To examine the potential influence of migration distance on survival, the river/tributary effect was replaced with a continuous covariate representing distance (km) from the river or tributary outlet to SOBI-1. Additionally, to assess whether post-smolt size influenced survival to SOBI within individual rivers or tributaries, a model including an interaction between tagging river and fork length (as an individual covariate) was constructed.

Version

R 4.4.3

Dependencies

• readxl
• here
• RMark

References (in this ReadME)

Campbell, H.A., Watts, M.E., Dwyer, R.G., Franklin, C.E. 2012. V-Track: software for analysing and visualising animal movement from acoustic telemetry detections, Mar. Freshw.      Res.  63:815-820 https://doi.org/10.1071/MF12194.             

Dufresne, C., Lavoie, D., Robertson, M., April, J., Burke, C., Carr, J., et al. (2025). The Labrador Current cold front shaping the Atlantic salmon homing migration routes from the waters off Southern Greenland to eastern North America. Progress in Oceanography, 233, 103439. https://doi.org/10.1016/j.pocean.2025.103439

Han, G., Bianucci, L., Chassé, J., Foreman, M., Holden, J., Horwitz, R., et al. (2022). Status of DFO ocean modelling: A national review. Canadian Technical Report of Hydrography and Ocean Sciences, 338, viii + 58 p.
Laake, J.L. 2013. RMark: an R interface for analysis of capture-recapture data with MARK. National Marine Fisheries Service. Alaska Fisheries Science Center, Processed Report 2013-01. Seattle.

Kahle, D., Wickham, H. 2023. ggmap: Spatial visualization with ggplot2. R J., 5(1): 144–161. https://doi.org/10.32614/rj-2013-014.

Laake, J.L. 2013. RMark: an R interface for analysis of capture-recapture data with MARK. National Marine Fisheries Service. Alaska Fisheries Science Center, Processed Report 2013-01. Seattle.

Lebreton, J.D., Burnham, K. P., Clobert, J., Anderson, D. R. 1992. Modeling survival and testing biological hypotheses using marked animals: A unified approach with case studies. Ecol. Monogr. 62(1): 67–118. https://doi.org/10.2307/2937171.

Madec, G., Bourdallé-Badie, R., Bouttier, P., Bricaud, C., Bruciaferri, D., Calvert, D. (2016). NEMO ocean engine. Notes du Pôle de modélisation de l'Institut Pierre-Simon Laplace, v3.6-patch, Number 27. https://doi.org/10.5281/zenodo.3248739

Vihtakari, M. 2024. ggOceanMaps: Plot data on Oceanographic maps using ‘ggplot2’. R package version 2.2.0. https://CRAN.R-project.org/package=ggOceanMaps.

Access information

Other publicly accessible locations of the data:

• N/A

Data was derived from the following sources:

• The master dataset master_dataset.csv was the main dataset used for all analyses. It contains the initial detection timestamps of Atlantic salmon smolts at acoustic receivers deployed within each location used in this study (ecoregions/quadrants, main receiver lines, and estuaries) along the eastern coast of Canada. These data were compiled from multiple projects affiliated with the Ocean Tracking Network (see Table S2 for references to the specific projects).
• Daily SST values at receiver locations was assessed using an ocean circulation model configuration for the Western North Atlantic (WNA). The model is an implementation of the Nucleus for European Modelling of the Ocean (NEMO) 3.6 model (Madec et al., 2016).