Scholarlink

<p>Aggressive behavior in insects plays a crucial role in various ecological and evolutionary processes, influencing resource acquisition, mate competition, and social organization. Stag beetles (Lucanidae) represent a compelling model system for studying aggression, particularly male-male combat. However, studying their interactions in a contest test with low light—a common condition for nocturnal species—and a complex background to mimic their natural environment remains challenging due to poor visibility and tracking limitations. Here, we adapted Detect Any Mouse Model (DAMM), a machine learning-driven tracking framework, to detect and track the position of each male stag beetle of two species (<i>C. mniszechi</i> and <i>C. speciosus</i>), followed by the calculation of several essential behavior endpoints to quantitatively analyze their behaviors with a conspecific during the test. The calculated object detection metrics indicated that the DAMM performed well in detecting and classifying objects. Meanwhile, based on the behavior endpoints, <i>C. mniszechi</i> displayed more robust behaviors during the test compared to <i>C. speciosus</i>, which was indicated by the tendency of this species to secure the food tube while fighting another contestant to defend their position. In addition, the calculated fighting-related-behavior endpoints also showed consistent results with the manual observation, highlighting the validity of the calculated endpoints. To sum up, the adaptation of DAMM to study insect behavior enables researchers to quantify their behaviors and automation that would undoubtedly lead to unprecedentedly detailed and objective insights into the complex behaviors of insects, especially stag beetles, contributing to a deeper understanding of their ecology, evolution, and the intricate workings of the natural world.</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

Quantitative measurement of stag beetle behaviors in low light and complex background conditions using DAMM (detect any mouse model)

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月18日

查看全文
  • Gilbert Audira,
  • Chung-Hsin Huang,
  • Mahmood Yousaf,
  • I-Min Tso,
  • Chung-Ping Lin,
  • Chung-Der Hsiao

摘要

Aggressive behavior in insects plays a crucial role in various ecological and evolutionary processes, influencing resource acquisition, mate competition, and social organization. Stag beetles (Lucanidae) represent a compelling model system for studying aggression, particularly male-male combat. However, studying their interactions in a contest test with low light—a common condition for nocturnal species—and a complex background to mimic their natural environment remains challenging due to poor visibility and tracking limitations. Here, we adapted Detect Any Mouse Model (DAMM), a machine learning-driven tracking framework, to detect and track the position of each male stag beetle of two species (C. mniszechi and C. speciosus), followed by the calculation of several essential behavior endpoints to quantitatively analyze their behaviors with a conspecific during the test. The calculated object detection metrics indicated that the DAMM performed well in detecting and classifying objects. Meanwhile, based on the behavior endpoints, C. mniszechi displayed more robust behaviors during the test compared to C. speciosus, which was indicated by the tendency of this species to secure the food tube while fighting another contestant to defend their position. In addition, the calculated fighting-related-behavior endpoints also showed consistent results with the manual observation, highlighting the validity of the calculated endpoints. To sum up, the adaptation of DAMM to study insect behavior enables researchers to quantify their behaviors and automation that would undoubtedly lead to unprecedentedly detailed and objective insights into the complex behaviors of insects, especially stag beetles, contributing to a deeper understanding of their ecology, evolution, and the intricate workings of the natural world.