白洋淀浅水湖泊地处我国华北平原地区腹地，在调节京津冀地区气候、维护华北地区湿地生态系统稳定性、保护湿地生物多样性等方面都发挥着关键作用。由于近年来的气候变化和高强度人类活动影响，浅水湖泊水环境发生变化，引发水质恶化、生物多样性降低等生态问题。如何基于白洋淀大型底栖动物对多重环境因子的响应规律，探索保障全淀最适多样性目标所对应的生态补水方案，成为白洋淀生态补水与水资源优化配置需要重点关注的问题。

本文基于2017-2023年的实测水文水质、沉积物及水生态数据，探究了白洋淀大型底栖动物群落结构及多样性的时空变异规律，采用典范对应及冗余分析方法分析了大型底栖动物群落结构对环境因子变化的响应。随后，基于粒子群智能算法优化后的随机森林算法构建大型底栖动物多样性评估模型。同时，根据白洋淀水文资料、水文边界情况构建MIKE21水动力水质耦合模型，模拟不同生态补水工况下关键因子分布特征，进而通过优化后的随机森林模型（PSO-RF）对不同情景下的多样性等级分布进行预测，最终依据最适多样性目标比选出最优补水方案。主要研究结论如下：

（1）2017-2023年白洋淀大型底栖动物结构及多样性变异规律

研究期间共调查物种72种，隶属于4门7纲，其中寡毛纲、腹足纲和昆虫纲分别占据统计总数的18.38%、27.47%和52.67%，其余纲则分布较少。白洋淀自2019年及之后在大型底栖动物群落结构及多样性指数上呈现出明显的分异特征，总体而言2019年之后大型底栖动物物种丰度及多样性不断提升，且优势种逐渐由昆虫纲变迁至寡毛纲和腹足纲。大型底栖动物群落结构在季节上呈现春秋无显著差异，夏季与春秋呈明显差异；在夏季群落分布多受水质本底值、污染源排放特征影响，而秋季群落分布主导因素为补水水源与方向，在空间上呈现明显的群落结构差异。与其他湖泊相比而言，白洋淀当前群落结构与多样性尚处于结构不稳定、多样性偏低的阶段。

（2）白洋淀大型底栖动物群落结构对环境因子的响应规律

通过CCA/RDA分析手段探究大型底栖动物群落结构在春、夏、秋三季分别对于生境变化的响应。春季对大型底栖动物群落结构产生显著影响的关键环境因子为TN、pH、DO，且春季优势种铜锈环棱螺、梨形环棱螺易在水体TN含量较高的环境中生长，红裸须摇蚊、霍普水丝蚓偏好pH较高的水域，而赤豆螺则喜好有氧环境；夏季大型底栖动物群落结构的关键因子依次为水温、C、Cd、pH、盐度和有机质N含量，且寡毛纲类物种丰度与pH、盐度和Cd含量呈正相关关系，腹足纲类物种丰度更偏好水温较高的环境；秋季大型底栖动结构的关键因子主要为沉积物中S、Cu、P含量，沉积物中P、Cu含量较高时更适合霍普水丝蚓、苏式尾鳃蚓、羽摇蚊、梨形环棱螺及铜锈环棱螺的生长，而含S量较高时则更适宜多足摇蚊、槲豆螺及赤豆螺的生长。

（3）白洋淀大型底栖动物多样性评估模型构建

引入粒子群智能算法优化机器学习模型，选用随机森林、支持向量回归和模糊神经网络进行评估性能比选。比选后判定采用粒子群优化后的RF模型拟合性能良好（R²=0.81），优选出用于构建大型底栖动物多样性评估模型的PSO-RF算法。随后进一步在PSO-RF模型中对影响多样性指数的关键因子进行排序筛选，重要系数大于0.5的环境因子依次为：TN、水深、溶解氧、P、pH、Zn、Cu。筛选和构建大型底栖动物多样性评估模型的过程可以作为理论方法迁移至其他研究区域。

（4）基于大型底栖动物多样性等级预测的生态补水优选方案研究

选取引黄济淀工程、南水北调中线工程及水库调水等多种工况组合而成的五种情景，经MIKE21模型模拟不同情景下的关键因子分布后，输入PSO-RF模型中对各情景多样性进行等级预测，综合评价发现，情景3（引黄济淀工程水量3亿m³，水库调水0.5亿m³）在短期内对水域大型底栖动物多样性的局部提升有明显的效果当补水情景发生变化时，其高等级（high）对应面积占比达到最大，占全淀面积百分比达23.8%，较其他情景下的高等级（high）面积占比提升2%~6.5%；而从全域平均大型底栖动物多样性提升而言，五种情景中平均多样性指数为情景5（引黄济淀工程水量2.0亿m³，水库调水0.5亿m³，南水北调工程引水 0.5亿m³）最高。在情景5中高等级（high）占比为34.34%，中等级（medium）占比为39.86%，占据全域面积超74%。

本文在分析各环境因子相互作用的基础上，构建白洋淀大型底栖动物多样性评估模型，量化大型底栖动物多样性与关键环境因子之间的耦联关系，揭示大型底栖动物多样性对关键环境因子变化的响应规律，旨在促进湿地生态健康维持和生物多样性保护，为白洋淀生态修复与生态服务功能提升提供理论支撑。

Located in the hinterland of the North China Plain, Baiyangdian shallow water lake plays a key role in regulating the climate of Beijing-Tianjin-Hebei region, maintaining the stability of the wetland ecosystem in North China, and protecting the wetland biodiversity. In recent years, due to the impact of climate change and high-intensity human activities, the water environment of shallow lakes has changed, causing ecological problems such as water quality deterioration and biodiversity reduction. Based on the ecological and environmental effects of the diversity of large benthic animals, how to explore the response rules of Baiyangdian benthic animals to multiple environmental factors, so as to determine the corresponding ecological water replenishment plan to ensure the optimal diversity of the lake has become a key issue that needs to be paid attention to in Baiyangdian ecological water replenishment and optimal allocation of water resources.

Based on the measured data from 2017 to 2023, the distribution variation of macroinvertebrate structure and diversity in Baiyangdian Lake was investigated. The relationship of macrobenthic structure and its crucial stressors was revealed by CCA/RDA method. Optimizing random forest algorithm using particle swarm intelligence algorithm, a benthic diversity assessment model was constructed. At the same time, the MIKE21 hydrodynamic and water quality coupling model was built, and the distribution of key factors under different scenarios was simulated, and the diversity level distribution under different scenarios was predicted by optimized RF model (PSO-RF) model. Finally, the optimal water replenishment scheme was selected according to the optimal diversity target ratio. The main conclusions are as follows:

(1) Macroinvertebrates community structure and biodiversity characteristics from 2017 to 2023 in Baiyangdian Lake

A total of 72 species were investigated, belonging to 4 phyla and 7 classes, among which Oligochaeta, Gastropoda and Insecta accounted for 18.38%, 27.47% and 52.67%, respectively, while the other classes were less distributed. Baiyangdian showed obvious differentiation in macrobenthic community structure and diversity index from 2019 and thereafter. In general, the abundance and diversity of macrobenthic species increased continuously after 2019, and the dominant species gradually changed from insecta to oligochaeta and Gastropod. The community structure of macrobenthos showed no significant difference in spring and autumn, but was significantly different in summer and spring. The community distribution in summer was mostly affected by the background value of water quality and emission characteristics of pollution sources, while the dominant factor of the community distribution in autumn was the source and direction of replenishment water, showing obvious differences in community structure in space. Compared with other lakes, Baiyangdian community structure and diversity are still in the stage of structural instability and low diversity.

(2) Response of macrobenthos community structure to environmental factors in Baiyangdian Lake

The responses of macrobenthos community structure to habitat changes in spring, summer and autumn were investigated by CCA/RDA analysis. In spring, the key environmental factors that significantly affected the community structure of macrobenthos were TN, pH and DO, and the dominant species in spring were prone to grow in the environment with higher TN content in water. The key factors of macrobenthic community structure in summer were water temperature, C, Cd, pH, salinity and organic matter N content, and the abundance of oligochaeta species was positively correlated with pH, salinity and Cd content. The contents of S, Cu and P in the sediments were the key factors of the macrobenthic dynamic structure in autumn, and the higher S content was more suitable for the growth of Chirus multipes, Mistlebean snail and red bean snail.

(3) Construction of evaluation model for macrobenthic fauna diversity in Baiyangdian Lake

Particle swarm intelligence algorithm was introduced to optimize the machine learning model, and random forest, support vector regression and fuzzy neural network were selected to evaluate the performance comparison. After selection, it was determined that the RF model after particle swarm optimization had good fitting performance (R² = 0.81), and the PSO-RF algorithm was selected to construct the evaluation model of large benthic animal diversity. Then, the key factors affecting the diversity index were further sorted and screened in the PSO-RF model. The environmental factors with important coefficients greater than 0.5 were successively TN, water depth, dissolved oxygen, P, pH, Zn and Cu. The process of screening and building a large benthic diversity assessment model can be transferred to other study areas as a theoretical approach.

(4) Research on the optimal scheme of ecological water replenishment based on the prediction of the diversity level of large benthic animals

After simulating the distribution of key factors under different scenarios with the MIKE21 model, it was input into the PSO-RF model to predict the level of diversity. The comprehensive evaluation found that scenario 3 (Yellow River Diversion Project 300 million m³, water transfer of reservoirs 50 million m³) had an obvious effect on the local improvement of water diversity in the short term. When the water replenishment scenario changed, the corresponding area of the high level reached the largest proportion. It accounted for 23.8% of the total lake area, which was 2%~6.5% higher than that of other scenarios. In terms of global average diversity improvement, the global average diversity index was the highest in scenario 5 and the lowest in scenario 1. In scenario 5, high grade accounted for 34.34%, medium grade accounted for 39.86%, occupying more than 74% of the whole area.

Based on the quantitative consideration of the interaction of various environmental factors, this study aims to build an assessment model of macrobenthic animal diversity in Baiyangdian, quantify the coupling relationship between macrobenthic animal diversity and key environmental factors, and reveal the response of macrobenthic animal diversity to key environmental factors, so as to promote the maintenance of wetland ecological health and the protection of biodiversity. It can provide a theoretical support for ecological restoration and ecological service function enhancement in Baiyangdian.