EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Facial Recognition Analyses Reveal Social Networks of Co‐Occurrence at Harbor Seal Haul‐Out Sites

Wyatt Hall, Abigail Hanson, Sydney Dunn, Ava Benton, C Filipowicz, Ahmad Khazaee, Tolga Dinçer, Krista K Ingram iD

Ecology and Evolution · 2026-06 · 16 (6)

DOI: 10.1002/ece3.73856 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Ecology and Evolution, online erschienen am 12.06.2026. Verfasst von Wyatt Hall und 7 weiteren Autor:innen. Beteiligte Einrichtung: Department of Biology Colgate University Hamilton New York USA. Die Arbeit nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Directorate for Computer and Information Science and Engineering.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Understanding the behavior and population dynamics of harbor seals ( Phoca vitulina ), an ecologically significant and widespread coastal pinniped, is vital to effectively manage vulnerable coastal marine ecosystems. A novel, non‐invasive facial recognition technology was used to identify harbor seals at haul‐out sites across two consecutive molting seasons in Middle Bay, Maine. Using a gallery of 672 individual seals found in the bay over a 6‐year period, we recorded the presence of seals at target sites over 30 days across 2 years and constructed social network analyses based on repeated co‐occurrences at these sites within the seasons and across years. Our results revealed strong site fidelity among a subset of seals across years and persistent co‐occurrences between individual seals within seasons and across years, suggesting non‐random haul‐out behaviors within the region. A core group of seals central to the networks were regularly observed at the sites across the span of the molting season, indicating strong site co‐fidelity of seals during this season. Module membership of co‐occurring seals varied across years, but the overall structure of the network (graph density, average degree, average clustering and average path length) remained consistent across years. These findings underscore the robust site fidelity of harbor seals to specific haul‐out locations in Middle Bay and demonstrate that harbor seals are not transient visitors to this region. This study also demonstrates the utility of facial recognition as an effective, non‐invasive method for long‐term monitoring of wild seal movements, distribution, habitat use, and social interactions. Using this approach can provide novel insights into how individuals and populations of wild seals respond to environmental or anthropogenic changes and can inform conservation efforts to mitigate threats to coastal ecosystems.

Schlagwörter

Stichwörter der Autor:innen

association networksfacial recognitionharbor sealmarine mammalphoto IDsite fidelity

Autor:innen und Institutionen

  • Department of Biology Colgate University Hamilton New York USA.
  • Information Technology Colgate University Hamilton New York USA.

ORCID-Profile: Krista K Ingram

Förderung

  • Directorate for Computer and Information Science and Engineering

Zitieren

APA

Hall, W., Hanson, A., Dunn, S., Benton, A., Filipowicz, C., Khazaee, A., Dinçer, T., & Ingram, K. K. (2026). Facial Recognition Analyses Reveal Social Networks of Co‐Occurrence at Harbor Seal Haul‐Out Sites. Ecology and Evolution, 16(6). https://doi.org/10.1002/ece3.73856

Vancouver

Hall W, Hanson A, Dunn S, Benton A, Filipowicz C, Khazaee A, et al. Facial Recognition Analyses Reveal Social Networks of Co‐Occurrence at Harbor Seal Haul‐Out Sites. Ecology and Evolution. 2026;16(6). doi:10.1002/ece3.73856

BibTeX
@article{Hall2026,
  author = {Hall, Wyatt and Hanson, Abigail and Dunn, Sydney and Benton, Ava and Filipowicz, C and Khazaee, Ahmad and Dinçer, Tolga and Ingram, Krista K},
  title = {Facial Recognition Analyses Reveal Social Networks of Co‐Occurrence at Harbor Seal Haul‐Out Sites},
  journal = {Ecology and Evolution},
  year = {2026},
  volume = {16},
  number = {6},
  doi = {10.1002/ece3.73856}
}
Kostenlos starten →