Revealing heavy metal correlations with water quality and tracking its latent factors by canonical correlation analysis and structural equation modeling in Dongjianghu Lake

Abstract

Decreasing levels of water quality and elevated concentrations of heavy metals in freshwaters can pose global challenges for drinking water sources. Multivariate statistical techniques have been applied on data matrices of water quality and heavy metals for keen characterization of their spatio-temporal variations, exploration of latent factors, and identification of pollution sources. Non-metric multidimensional scaling (nMDS), canonical correlation analysis (CCA), and structural equation modeling (SEM) were employed to process data matrices of the water quality and heavy metals with 14 parameters measured at 13 sampling sites in Dongjianghu Lake in March, June, August, and December 2016. The sampling sites were grouped into three clusters using the nMDS, suggesting that the increasing order of the water quality levels was approximately midstream < downstream < upstream and lake. The CCA of 14 parameters proved that the Escherichia coli, COD_Mn, TP, TN, TEMP, DO, and pH were the latent factors to distinguish the sampling sites, suggesting that the natural disturbances further influenced the lake and upstream, while the anthropogenic activities further influenced the midstream and downstream. The CCA of the heavy metals exhibited that the COD_Mn, F^-, and E. coli were the latent factors of the Cu, Zn, and As, while the DO and TEMP were the latent factors of the Cd. This indicated that the Cu, As, and Zn were mainly associated with the anthropogenic activities, while the Cd was predominantly relative to the natural conditions. The SEM of the water quality and heavy metals showed that the weights of COD_Mn (28.64%), NH₃-N (14.96%), BOD₅ (14.32%), TN (12.88%), and TP (10.18%) were higher than those of the pH (8.37%), DO (7.73%), TEMP (2.58%), and E. coli (0.34%). This indicated that the former exhibited strong influences on the heavy metals than the latter. Moreover, the COD_Mn and BOD₅ were the key factors of the heavy metals, which should be attributed to the no-point sources, especially the exploitation mining and mill tailings. The water quality assessment by the nMDS, CCA, and SEM can determine the status, trend corresponding to its standards, and trace latent factors and identify possible pollution sources. The study could provide a guide for water quality evaluation and pollution control.

Keywords: Canonical correlation analysis; Heavy metals; Latent factor; Structural equation modeling; Water quality.