本文基于2023年8月在浙江省枸杞岛贻贝浮筏式养殖场利用网采镜检以及水下数字全息粒子成像系统（Lisst holo2）进行调查获得的浮游植物实测数据，对网采数据和粒子影像数据进行综合比较研究，并分析其群落结构晨昏动态变化及影响因子。调查期间共发现5门38属75种浮游植物，其中夜光藻（Noctiluca scintillans）和笔尖形根管藻（Rhizosolenia styliformis）等为主要优势种。浮游植物种类数量和丰度晨昏变化明显，正午浮游植物大多都朝底层移动，底层浮游植物种类数和丰度远大于表层，而到傍晚表底层种类数和丰度差异减小，主要存在以夜光藻为主的甲藻昼降夜升的垂直移动现象。浮游植物丰度与潮流之间存在一定联系，养殖水层与潮汐之间的相关性相对较强，浮游植物丰度随涨潮有所上升。养殖场海域表现为硅藻爆发向夜光藻爆发转变的生长模式。通过构建浮游植物共现网络，发现养殖水层的浮游植物种类数量和丰度相对全水层来说较少，但边数量为186，多于全水层，因此养殖层浮游植物网络更为复杂，群落结构稳定性更强，浮游植物之间的互作关系、共生关系强于全水层。Lisst holo2观测与传统采样相比，解决了其样品保存和处理繁琐，数据获取滞后以及难以测量浮游植物粒径大小等问题，本研究验证其在大粒径藻类（如夜光藻）监测中的可靠性，有助于未来海洋生态系统监测与保护的研究。

This study is based on phytoplankton field data collected in August 2023 from the suspended mussel farms of Gouqi Island, Zhejiang Province, using the underwater digital holographic particle imaging system (Lisst holo2) and traditional net sampling combined with microscopy. A comprehensive comparative analysis of particle imaging data and net sampling data was conducted, with a focus on diurnal variations in phytoplankton community structure and influencing factors. During the survey, 75 phytoplankton species belonging to 38 genera and 5 phyla were identified, with dominant species including Noctiluca scintillans and Rhizosolenia styliformis.Significant diurnal variations in phytoplankton species richness and abundance were observed. At midday, most phytoplankton migrated toward deeper layers, resulting in significantly higher species numbers and abundance in the bottom layer compared to the surface. By evening, differences between surface and bottom layers diminished, with a vertical migration pattern dominated by dinoflagellates (primarily Noctiluca scintillans), characterized by daytime descent and nighttime ascent. Phytoplankton abundance exhibited a moderate correlation with tidal cycles, showing a notable increase during flood tide within the aquaculture water layer. The study area displayed a transition from diatom-dominated blooms to Noctiluca scintillans-dominated blooms.Phytoplankton co-occurrence network analysis revealed that the aquaculture layer had fewer species and lower abundance compared to the whole water column. However, its network complexity (186 edges) was higher, indicating stronger community stability and more intensive species interactions (e.g., mutualism or competition) within the aquaculture layer. Compared to traditional sampling methods, Lisst holo2 addresses limitations such as cumbersome sample preservation and processing, delayed data acquisition, and challenges in measuring phytoplankton particle size. This study validates its reliability in monitoring large-sized algae (e.g., Noctiluca scintillans), highlighting its potential for future marine ecosystem monitoring and conservation research.