Scholarlink

<p>Breathing and diving behaviours of air-breathing marine animals are intricately linked to their physiological needs and environmental conditions. Sea turtles exhibit specialized respiratory adaptations that support prolonged dives, yet surfacing to breathe exposes them to significant anthropogenic risks. The study of the intricate relationship between the plurality of diving and surfacing factors and the breathing requirements is frequently overlooked, often relying on indirect methodologies with limited sensitivity. The integration of advanced tools, such as high-resolution biologgers and cameras, and machine learning (ML) algorithms holds the potential to address these gaps. This study investigated how dive parameters, surfacing dynamics, and underwater behaviours are associated with breathing patterns in loggerhead turtles (<i>Caretta caretta</i>) at foraging grounds, and evaluated the detection of breathing, diving, and surfacing events performed using pressure sensors and video recordings. Carapace-mounted multisensor loggers integrated with cameras were deployed on 23 turtles obtaining high-resolution data on diving dynamics, surface intervals, and single breathing events. Breath number increased with surfacing duration, despite plateauing for surfacing events exceeding ~ 2.5&#xa0;min. Higher breath counts were associated with increased rotational movements and a greater proportion of benthic crawling in the subsequent dive, suggesting enhanced respiratory preparation before undertaking these dives that may be more energetically/physiologically demanding. Camera validation revealed that pressure sensors underestimated the occurrence of brief surfacing events and overestimated their duration. ML models showed moderate/low efficacy (recall: 0.62–0.63) in detecting individual breaths, underscoring the challenges of identifying breathing events solely through carapace-mounted sensors. However, pitch angle variance emerged as a promising proxy of breathing events.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Sea turtle breathing patterns and involved factors investigated through animal-borne camera and sensors

  • 期刊论文

  • 发表日期：

    2026年1月8日

查看全文
  • Chiara Agabiti,
  • Federica Restelli,
  • Daniela Freggi,
  • Pasquale Salvemini,
  • Paolo Luschi,
  • Paolo Casale

摘要

Breathing and diving behaviours of air-breathing marine animals are intricately linked to their physiological needs and environmental conditions. Sea turtles exhibit specialized respiratory adaptations that support prolonged dives, yet surfacing to breathe exposes them to significant anthropogenic risks. The study of the intricate relationship between the plurality of diving and surfacing factors and the breathing requirements is frequently overlooked, often relying on indirect methodologies with limited sensitivity. The integration of advanced tools, such as high-resolution biologgers and cameras, and machine learning (ML) algorithms holds the potential to address these gaps. This study investigated how dive parameters, surfacing dynamics, and underwater behaviours are associated with breathing patterns in loggerhead turtles (Caretta caretta) at foraging grounds, and evaluated the detection of breathing, diving, and surfacing events performed using pressure sensors and video recordings. Carapace-mounted multisensor loggers integrated with cameras were deployed on 23 turtles obtaining high-resolution data on diving dynamics, surface intervals, and single breathing events. Breath number increased with surfacing duration, despite plateauing for surfacing events exceeding ~ 2.5 min. Higher breath counts were associated with increased rotational movements and a greater proportion of benthic crawling in the subsequent dive, suggesting enhanced respiratory preparation before undertaking these dives that may be more energetically/physiologically demanding. Camera validation revealed that pressure sensors underestimated the occurrence of brief surfacing events and overestimated their duration. ML models showed moderate/low efficacy (recall: 0.62–0.63) in detecting individual breaths, underscoring the challenges of identifying breathing events solely through carapace-mounted sensors. However, pitch angle variance emerged as a promising proxy of breathing events.