KOSMOS 2020 Peru mesocosm on ecosystem responses to different light and upwelling intensities: zooplankton dynamics via Plankton Imaging with Scanning Optics (PIScO)@en

Zooplankton dynamics during the mesocosm experiment in the Peruvian upwelling system near Lima from February-April 2020. Light (ambient = high vs. reduced = low) was manipulated from the beginning (day 0) and upwelling (0 to 60% deep water addition) two weeks into the experiment (day 13). Copepods >200 µm were sampled every two days using Plankton Imaging with Scanning Optics (PIScO) (Taucher et al. 2022). This camera system was lowered from the surface to the sediment trap, scanning at three frames per second. Each image sampled 0.577 L of seawater, summing to 99 L per camera cast. Images were analyzed for copepods with an object detection algorithm trained by a human observer on 11,269 images (73% for both precision and recall). For each image, abundance and individual size (equivalent spherical diameter, ESD) was calculated. The data shown here represent averages across the entire water column, ranging from 1.15 to 10.5 m depth. Methodological details in Goldenberg et al. and Thielecke et al. (manuscripts in progress).2020年2月至4月，秘鲁利马附近海域上升流系统中型实验生态系（mesocosm experiment）实验中的浮游动物动态数据集。实验初始当日（第0天）即对光照条件进行调控，设置高光照（自然光照组）与低光照（弱光组）两种处理；实验开展至第13天（即实验启动后两周）时，启动上升流模拟操作，向系统中添加0至60%体积占比的深层海水。 每两日采用扫描光学浮游生物成像系统（Plankton Imaging with Scanning Optics, PIScO）对粒径大于200微米的桡足类进行采样（Taucher等，2022）。该成像系统从水面下放至沉积物捕集器位置，以每秒3帧的速率完成扫描成像。单次成像可采集0.577升海水，每轮相机投放采样总容积达99升。 研究人员借助经人类观察员在11269张图像上训练得到的目标检测算法，对采集图像开展桡足类识别分析，该算法的精确率与召回率均为73%。针对每张图像，计算得到桡足类的丰度以及个体尺寸参数——等效球形直径（equivalent spherical diameter, ESD）。 本次展示的数据为整个水柱的平均结果，采样水深范围为1.15米至10.5米。相关方法学细节可参考Goldenberg等与Thielecke等（待发表手稿）。