Scholarlink

<p>Capturing the spatial and temporal dynamics of zooplankton and aquatic environmental elements in the drinking water source watersheds is pivotal for the scientific, holistic, and precise development of a reservoir ecological health assessment system. However, research on the interaction between zooplankton and aquatic environment in a cascade reservoir used as drinking water sources remains limited. From 2022 to 2024, we combined field surveys in combination with artificial intelligence ecological networks and hybrid modeling methods, and systematically revealed multi-scale ecological interactions in subtropical mountain reservoirs. Results reveal that body size mediated the niche differentiation. Rotifera (&lt;650 µm) showed a rapid nutrient response, whereas Copepoda (200–2 800 µm) dominated via environmental adaptability, governed by metabolic scaling principles (<i>R</i><sup>2</sup>=0.78 for total nitrogen (TN)-phytoplankton coupling). Multidimensional drivers including hydraulic gradients (0.12–0.68 m/s), stoichiometric imbalance (TN:total phosphorus (TP)=18.7±3.2), and hypoxia-induced Copepoda dominance (37%–42% abundance at dissolved oxygen (DO)&lt;4.2 mg/L), collectively shaped the community dynamics. Innovatively, recurrent neural network (RNN)-Eco networks (term frequency-inverse document frequency (TF-IDF)-weighted co-occurrence, <i>α</i>=0.82; latent dirichlet allocation (LDA) clustering) decoded hydrodynamic-driven architectures: lentic zones showed long-path networks (path=4.7) contrasting with lotic zones’ clustered topologies (cluster=0.68). The principal component analysis (PCA)-RNN hybrid model outperformed conventional methods by 23.7%, prioritizing cyanobacterial control (cyanobacterial index (PCY) weight=0.378) and flow-induced Cladocera habitat shifts (&gt;0.45 m/s triggers 41% migration). Three operational thresholds were proposed to optimize reservoir management: maintaining TN:TP ratios below 12:1 to mitigate cyanobacteria-zooplankton decoupling, preserving flow velocity gradients (0.2–0.6 m/s) to sustain habitat heterogeneity, and limiting interannual conductivity variation to &lt;15% to prevent adaptive mismatches. This study established a transformative framework linking ecological theory with precision water resource management, and provided quantifiable metrics for anthropogenic impact assessment in climate-vulnerable ecosystems.</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

Hierarchical drivers and adaptive strategies of zooplankton to aquatic environment in anthropogenically modified cascade reservoir ecosystems

  • 期刊论文

  • 发表日期：

    2026年1月26日

查看全文
  • Jingyun Yin,
  • Jihong Xia,
  • Chuanbin Dou,
  • Yue Wang,
  • Qianqian Mao,
  • Zhicheng Liu,
  • Yuanshuo Lu,
  • Tao Hu,
  • Huijie Yang,
  • Yifei Ren,
  • Guofu Lin

摘要

Capturing the spatial and temporal dynamics of zooplankton and aquatic environmental elements in the drinking water source watersheds is pivotal for the scientific, holistic, and precise development of a reservoir ecological health assessment system. However, research on the interaction between zooplankton and aquatic environment in a cascade reservoir used as drinking water sources remains limited. From 2022 to 2024, we combined field surveys in combination with artificial intelligence ecological networks and hybrid modeling methods, and systematically revealed multi-scale ecological interactions in subtropical mountain reservoirs. Results reveal that body size mediated the niche differentiation. Rotifera (<650 µm) showed a rapid nutrient response, whereas Copepoda (200–2 800 µm) dominated via environmental adaptability, governed by metabolic scaling principles (R2=0.78 for total nitrogen (TN)-phytoplankton coupling). Multidimensional drivers including hydraulic gradients (0.12–0.68 m/s), stoichiometric imbalance (TN:total phosphorus (TP)=18.7±3.2), and hypoxia-induced Copepoda dominance (37%–42% abundance at dissolved oxygen (DO)<4.2 mg/L), collectively shaped the community dynamics. Innovatively, recurrent neural network (RNN)-Eco networks (term frequency-inverse document frequency (TF-IDF)-weighted co-occurrence, α=0.82; latent dirichlet allocation (LDA) clustering) decoded hydrodynamic-driven architectures: lentic zones showed long-path networks (path=4.7) contrasting with lotic zones’ clustered topologies (cluster=0.68). The principal component analysis (PCA)-RNN hybrid model outperformed conventional methods by 23.7%, prioritizing cyanobacterial control (cyanobacterial index (PCY) weight=0.378) and flow-induced Cladocera habitat shifts (>0.45 m/s triggers 41% migration). Three operational thresholds were proposed to optimize reservoir management: maintaining TN:TP ratios below 12:1 to mitigate cyanobacteria-zooplankton decoupling, preserving flow velocity gradients (0.2–0.6 m/s) to sustain habitat heterogeneity, and limiting interannual conductivity variation to <15% to prevent adaptive mismatches. This study established a transformative framework linking ecological theory with precision water resource management, and provided quantifiable metrics for anthropogenic impact assessment in climate-vulnerable ecosystems.