CmaxVideoAnalysis_2019_dutycycled_MW.xlsx

Towed cameras were programmed to record video data at 1080p resolution at 50 frames per second until either the memory or power were exhausted. Video recording was continuous in 2018, but a duty cycle of 30 seconds of recording every 5 minutes was used in 2019 to extend the recording duration. The 2018 continuous video data were therefore sub-sampled to 30-second video sequences every 5 minutes to match the duty cycling used in 2019. We also compared habitat use and behaviours revealed using the duty-cycled video with the continuous video data (collected in 2018), to understand what may have been missed in the duty-cycled data. All video data were observed in VLC Media Player (Version 3.0.8). Initial assessment of all data was undertaken by viewing each 30-second video file at twice the native speed to ascertain information on the following: (i) ambient lighting conditions (more than 50% of video data gathered in dark conditions, termed ‘blackout’, separated into day and night) and (ii) seabed presence and habitat type (whether seabed was visible in more than 50% of data and dominant habitat type). Seabed habitats were classified following the European Nature Information System (EUNIS; https://eunis.eea.europa.eu) habitat classification system, and also included surface and mid-water swimming where neither the seabed nor surface were visible. The frequency and extent of other behaviours, including feeding behaviour (when the shark’s mouths was open for more than 50% of the video duration, i.e. >15 seconds), presence of conspecifics and other species were recorded, and species were identified to the finest possible taxonomic level. Where necessary, video data were watched at native or slowed to half speed to ascertain behaviour changes or other events. To investigate the initial responses of basking sharks to tagging, Tail Beat Frequency (TBF) was calculated as the number of lateral undulatory movements visible in the video data. Using data collected in 2018, where video recording was continuous, TBF was estimated every minute (using 30 second of data), and every five minutes for the 2019 duty cycled data for the first 30 minutes following tagging. TBF was also estimated for each 30-second video sequence of surface feeding behaviour

拖曳式摄像头（Towed cameras）被预设为以1080p分辨率、每秒50帧的规格录制视频数据，直至内存耗尽或电力耗尽。2018年的视频录制为连续模式，但2019年为延长整体录制时长，采用了每5分钟录制30秒的占空比方案。因此研究人员将2018年采集的连续视频数据按每5分钟截取一段30秒视频序列的方式进行二次采样，以匹配2019年采用的占空比录制规则。
我们还将采用占空比模式录制的视频所揭示的栖息地利用与行为数据，与2018年获取的连续视频数据进行对比分析，以明确占空比模式下可能遗漏的观测信息。所有视频数据均通过VLC媒体播放器（VLC Media Player，版本3.0.8）完成观测。
对所有数据的初步评估采用以2倍原生速度播放每段30秒视频文件的方式，以获取两类核心信息：(i) 环境光照条件（超过50%的视频数据采集于黑暗环境，该环境被定义为“黑障（blackout）”，并按昼夜时段进行划分）；(ii) 海底存在情况与栖息地类型（判断是否有超过50%的视频片段可观测到海底，以及该区域的优势栖息地类型）。
海底栖息地按照欧洲自然信息系统（European Nature Information System, EUNIS；https://eunis.eea.europa.eu）的栖息地分类体系进行划分，同时也涵盖了既无法观测到海底也无法观测到海面的表层及中层水域游动场景。
研究人员同时记录了其他行为的发生频次与覆盖范围，包括取食行为（当鲨鱼嘴部张开时长超过视频总时长的50%，即超过15秒时）、同种个体及其他物种的出现情况，并尽可能将物种鉴定至最精细的分类学层级。必要时，会以原生速度或半速播放视频片段，以确认行为变化或其他关键事件。
为探究姥鲨（basking sharks）对标记行为的初始响应，研究人员计算了尾拍频率（Tail Beat Frequency, TBF），即视频数据中可见的侧向摆动动作的总数量。基于2018年的连续录制视频数据，TBF的估算频率为每分钟1次（采用30秒的数据片段进行计算）；而针对2019年占空比模式采集的数据，则在标记后的前30分钟内每5分钟估算1次TBF。此外，研究人员还针对每一段包含表层取食行为的30秒视频序列估算了TBF。