我国流域铜的长期水质基准预测模型研究——MLR vs. BLM
Prediction Model for Setting Long-term Water Quality Criteria of Copper in Chinese River Basins: MLR vs. BLM
-
摘要: 实际水环境中金属的毒性效应受多种环境要素的复合影响,利用模型定量表征金属形态和生物有效性是制定“原位”水质基准和风险评估的前提和基础。国际上推荐使用的生物配位体模型(Biotic Ligand Model,BLM)因物种局限性和建模过程不透明性,限制了在水质管理中的推广应用。本研究拟在已有生物配位体理论基础上,基于我国本土物种的慢性毒理学数据,通过水环境参数的精简优化,建立与BLM模型功能相当、且预测准确度更高的水质基准校验方法。以BLM模型最早开展研究的铜为例,选择溶解性有机碳(Dissolved Organic Carbon,DOC)、硬度(Hardness,H)和pH三个环境参数,采用含有交互项的多元线性回归(Multiple Linear Regression,MLR)方法,发现我国4门22种水生生物的慢性毒性终点与H、pH和DOC的对数呈显著正相关(r2=0.5531,F=61.23,p<0.0001,AIC=0.9457,BIC=23.43)。与BLM模型相比,基于MLR获得的校验模型预测准确率(RFx,2.0)提升了20%,残差分析的等级评分为0.739(MLR)和0.466(BLM)。通过对环境参数进行梯度化赋值,计算22种水生生物的毒性预测值,发现Sigmoidal-Weibull函数对物种敏感度分布(Species Sensitivity Distributions,SSDs)曲线的拟合效果最佳,并在此基础上获得三参数耦合的铜长期基准校验方程(R2 = 0.9902,RMSE = 0.04106,F = 7238,p < 0.001)。研究成果不仅为“因地制宜”制定流域铜的保护水生生物基准提供可能,而且为其他金属的水生态风险防控提供科学依据和技术支撑,助力流域水质管理的差异化和精准化。Abstract: In actual water environment, the toxic effects of metals are affected by multiple environmental factors. It is the premise and basis for setting "in situ" water quality criteria (WQC) and performing risk assessments by using the quantitative characterization of metal morphology and bioavailability. The Biotic Ligand Model (BLM) is limited in water quality management due to difference of species and opacity of the modeling process. Based on existing biological ligand theory and chronic toxicological data for native species in China, the present study aims to establish an approach to predict WQC within the framework of the BLM model and acquire higher prediction accuracy through simplification and optimization of water environmental parameters. Taking copper as an example, the chronic toxicity endpoints of 4 phyla and 22 species in China, by Multiple Linear Regression (MLR) analysis, were positively correlated with the logarithm of Dissolved Organic Carbon (DOC), Hardness (H), and pH(r2=0.5531,F=61.23,p<0.0001,AIC=0.9457,BIC=23.43. Compared with the BLM model, the prediction accuracy of our MLR prediction model was increased by 20% (RFx, 2.0), and the rating scores of residual analyses were 0.739 (MLR) and 0.466 (BLM). Through gradient assignment of environmental parameters, predictive values for 22 aquatic organisms were calculated, indicating that the Sigmoidal-Weibull function had the best fitting by the Species Sensitivity distribution (SSDs) (R2 = 0.9902,RMSE = 0.04106,F = 7238,p < 0.001). These findings not only provide the possibility to derive the site-specific WQC of copper in Chinese river basins, but also introduce scientific basis and technical support for the control of aquatic ecological risks of other metals and enhance water quality management distinctively and precisely.
点击查看大图
计量
- 文章访问数: 124
- HTML全文浏览量: 11
- PDF下载量: 11
- 被引次数: 0