Direct observation of foraging behavior is not always possible, especially for marine species that hunt underwater. However, biologging and tracking devices have provided detailed information about how various species use their habitat. From these indirect observations, researchers have inferred behaviors to address a variety of research questions, including the definition of ecological niches. In this study, we deployed video cameras with GPS and time-depth recorders on 16 chinstrap penguins (Pygoscelis antarcticus) during the brood phase of the 2018–2019 breeding season on Signy (South Orkney Islands). More than 57 h of footage covering 770 dives were scrutinized by two observers. The outcome of each dive was classified as either no krill encounter, individual krill or krill swarm encounter and the number of prey items caught per dive was estimated. Other variables derived from the logging devices or from the environment were used to train a machine-learning algorithm to predict the outcome of each dive. Our results show that despite some limitations, the data collected from the footage was reliable. We also demonstrate that it was possible to accurately predict the outcome of each dive from dive and horizontal movement variables in a manner that has not been used for penguins previously. For example, our models show that a fast dive ascent rate and a high density of dives are good indicators of krill and especially of swarm encounter. Finally, we discuss how video footage can help build accurate habitat models to provide wider knowledge about predator behavior or prey distribution.

中文翻译：

---

使用生物记录仪和动物传播相机预测帽带企鹅的觅食潜水结果

直接观察觅食行为并不总是可能的，尤其是对于在水下捕猎的海洋物种。然而，生物记录和跟踪设备提供了有关各种物种如何使用其栖息地的详细信息。从这些间接观察中，研究人员推断出行为来解决各种研究问题，包括生态位的定义。在这项研究中，我们在 Signy（南奥克尼群岛）的 2018-2019 年繁殖季节的育雏阶段为 16 只帽带企鹅（Pygoscelis antarcticus）部署了带 GPS 和时间深度记录器的摄像机。两名观察员仔细检查了超过 57 小时、涵盖 770 次潜水的镜头。每次潜水的结果都被归类为没有遇到磷虾，估计了个体磷虾或磷虾群遭遇以及每次潜水捕获的猎物数量。来自测井设备或环境的其他变量被用来训练机器学习算法来预测每次潜水的结果。我们的结果表明，尽管存在一些限制，但从镜头中收集的数据是可靠的。我们还证明，可以以以前未用于企鹅的方式从潜水和水平运动变量中准确预测每次潜水的结果。例如，我们的模型表明，快速的潜水上升率和高密度的潜水是磷虾，尤其是群体遭遇的良好指标。最后，我们讨论了视频片段如何帮助建立准确的栖息地模型，以提供有关捕食者行为或猎物分布的更广泛知识。来自测井设备或环境的其他变量被用来训练机器学习算法来预测每次潜水的结果。我们的结果表明，尽管存在一些限制，但从镜头中收集的数据是可靠的。我们还证明，可以以以前未用于企鹅的方式从潜水和水平运动变量中准确预测每次潜水的结果。例如，我们的模型表明，快速的潜水上升率和高密度的潜水是磷虾，尤其是群体遭遇的良好指标。最后，我们讨论了视频片段如何帮助建立准确的栖息地模型，以提供有关捕食者行为或猎物分布的更广泛知识。来自测井设备或环境的其他变量被用来训练机器学习算法来预测每次潜水的结果。我们的结果表明，尽管存在一些限制，但从镜头中收集的数据是可靠的。我们还证明，可以以以前未用于企鹅的方式从潜水和水平运动变量中准确预测每次潜水的结果。例如，我们的模型表明，快速的潜水上升率和高密度的潜水是磷虾，尤其是群体遭遇的良好指标。最后，我们讨论了视频片段如何帮助建立准确的栖息地模型，以提供有关捕食者行为或猎物分布的更广泛知识。