README: Beneath the Antarctic sea-ice: Fine-scale analysis of Weddell seal (Leptonychotes weddellii) behaviour and predator-prey interactions, using micro-sonar data in Terre Adélie.

Abstract

In this study, we tried to assess
i) whether and how female Weddell seals feed (frequency, depth, duration) during lactation,

ii) what is their utilization of a limited foraging area (benthic or pelagic dives) as they are spatially constrained by the presence of their pup,

iii) how can we characterize their foraging dives and the approach/catching phases using new tools providing a more detailed description of their behaviour.

Sonar tags were deployed on three breeding female Weddell seals in Terre Adélie (East Antarctica) in November 2019, to study animals’ movements and dives at high resolution (3D acceleration, magnetometry, time and depth and GPS location), as well as information on prey and predator-prey interactions using acoustic data.

Description of the data and file structure

Usage note

Raw data were calibrated and pre-processed following a routine that has been developed by Mark P. Johnson, with the help of Pauline Goulet. Please refer to Mark Johnson (markjohnson@bio.au.dk) for code distribution.

Codes for acceleration data processing, isolation of prey capture attempts from acceleration data, and visual analysis of the echograms for fine scale predator-prey interaction description have been written by Pauline Goulet. Please refer to Pauline Goulet (paulinegouletpro@gmail.com) for code distribution.

The tags were deployed for seven days in Terre Adélie (East Antarctica). The three tags are hereafter referred to as 314a, 314b and 315a respectively.

Datasets included

"NameOfTag" is either “314a”, “314b” or “315a”.

1) lw19_NameOfTag_sensor.nc: NetCDF file containing calibrated data recorded by the tag (acceleration in each of the three axis, depth, magnetometry) and general information on tag deployment, with:

• in the “A” structure: acceleration data in each of the three axis, with metadata,

• in the “info” structure: metadata on the tag deployment date, location, type, and other information,

• in the “J” structure: RMS norm-jerk data (magnitude of the quick acceleration of the head), obtained from tri-axial acceleration data, with metadata,

• in the “P” structure: calibrated depth data in meters, with metadata,

• in the “T” structure: temperature data in degrees Celsius, with metadata.

2) lw19_NameOfTag_dives.mat: MATLAB structure containing information on dives, dives shallower than 5m were removed during pre-processing of the data, with:

• dive start and end time, expressed as the number of seconds elapsed since the tag was turned on,

• other time information such as the duration of the dive, or the start time in datenum format,

• maximum depth (dives.max),

• coordinates, latitude and longitude in degrees.

3) lw19_NameOfTag_summary_dives.csv: CSV file summarising available dive data, with the depth (meters), time (datenum format) and dive index of each recorded point of the dive.

4) lw19_NameOfTag_cluster.csv: CSV file containing the cluster analysis result, with the cluster number and the index of the corresponding dive (obtained after running code “lw1_functional_pca_on_dive_shapes”).

5) lw19_NameOfTag_jerks_fullbandwidth.mat: jerk data, i.e., rapid head movement detected from tri-axial acceleration data, at full bandwidth (initial frequency of recording, 200 to 250Hz), and metadata on tag deployment, with:

- In the “info” structure: metadata on tag deployment

- In the “jerks” structure:

• jerk start and end time, expressed as the number of seconds elapsed since the tag was turned on,

• other time information such as the duration of the jerk, or the start time in datenum format,

• depth of the jerk, in meters,

• coordinates, latitude and longitude in degrees.

6) lw19_NameOfTag_argos_track.mat: 3D reconstruction of the animal’s trajectory, in local coordinates (dead-reckoned track), from Argos transmitted localisation data, with:

• serial date numbers (datenum MATLAB format) elapsed from the reference date 1-Jan-0000,

• coordinates, latitude and longitude in degrees.

7) lw19_NameOfTag_sunangle.mat: sun elevation during dives, with:

• the sun angle at the moment of the dive, expressed in degrees,

• dive start and end time, expressed as the number of seconds elapsed since the tag was turned on,

• other time information such as the duration of the dive, or the start time in datenum format,

• maximum depth (dives.max),

• coordinates, latitude and longitude in degrees.

8) lw19_sonar_visual_analysis_results.mat: MATLAB structure containing results from the visual analysis of the echograms (sonar data) corresponding to the moment when jerks (rapid acceleration of the head) where identified, in random order, with:

• information on the corresponding jerk: individual (“prefix”), jerk number for the given individual (“index_relative_indiv”), start and end time of the jerk, depth in meters, and date (in datetime and datenum format),

• results of the visual analysis: whether the bottom or a prey were visible (“presence_bottom” and “presence_prey”, 1:yes, 0: no), whether there was a single prey or a school of prey (“indiv_prey”, 1: single, 0: school), whether the prey reacted or not (“prey_reaction”, 1: yes, 0: not), the type of reaction (“prey_reaction_type”, 1: the prey leaves the bottom, 2: the prey escapes from the predator), the time and distance of the reaction (“prey_reaction_time” expressed as the number of seconds after the start of the corresponding jerk, and “prey_reaction_distance” in meters), and the prey acoustic size estimated from the number of pixels (“prey_size”).

Missing values are denoted by NaN.

Code/Software

To run the analysis, we used Matlab version R2021b (The MathWorks, Natick, 242 MA, USA) and R Studio software, version 1.3.1093 (R Core Team, 2020).

The code for the functional principal component analysis (“lw1_functional_pca_on_dive_shapes”) has been written with the help of David Nerini (Mediterranean Institute of Oceanography, Pytheas Institute, UMR 7294, 13288 Marseille, France). The code is shared here in Markdown script and HTML knitted formats.

Codes are numbered in the order that they should be run.

Acknowledgements

We are greatly indebted to Mark P. Johnson who designed the sonar tags, supervised this work, and developed the algorithm necessary to analyse the sonar tag data.

This work was supported by the French Polar Institute (IPEV) program 1182 ASSET, the “Sentinel of the Sea-Ice Environment” SENSEI BNP-Paribas Fondation project, and the CNES-TOSCA program “Weddell seals bio-oceanographers of the Antarctic sea-ice” (P.I.: J.B. Charrassin and S. Labrousse).

We also thank Prof. P. T. Madsen and the MSCA EU grant ‘’FEAST’’. Karine Heerah received fundings from a Marie Sklodowska Curie fellowship for this project (project 792042 - FEAST - H2020-MSCA-IF-2017).

Adelie Antoine received a PhD fellowship for the GDR OMER (CNRS).

All experiments were conducted under the ethical regulation approval of the French sub-Antarctic and Antarctic territories (TAAF) ethic committee. All animals in this study were treated in accordance with the “Comité de l’Environnement Polaire” Ethic Committee guidelines and the work was conducted under permit #2019-107 of Terre Australes et Antarctiques Française.

We warmly thank all the fieldworkers of the 69th (2019) and 70th (2020) expeditions to Dumont D’Urville, and the IPEV program 109 who assisted in the field.