Spatial autocorrelation and temporal variation of contaminants of emerging concern in a typical urbanizing river

Abstract

The distribution and fate of contaminants of emerging concern (CECs) was studied in relation to hydrological conditions, land use characteristics, and spatial contiguity in Houxi River. Thirty-four CECs were detected in the surface water during a three-year sampling campaign. Caffeine was most prevalent (99% frequency), while bisphenol A had the highest median concentration (78.2 ng/L) among the detected CECs. Caffeine and the other prevalent CECs lincomycin and bisphenol A, with median concentrations of 3.89 ng/L, 0.26 ng/L, and 78.2 ng/L, respectively, were positively correlated with land use types related to anthropogenic activities (grass, barren, built up, and cropland areas and landscape indexes for human activities). The analysis of similarities revealed significant annual variations, with increasing trends in both the concentrations and detection frequencies of CECs. Spatial variations were demonstrated by higher concentrations and detection frequencies downstream compared to upstream. The singular value decomposition analysis revealed that the downstream sites were the major contributors (55.6%–100%) to the spatial variability of most CECs. Moran's I analysis based on downstream contiguity indicated strong spatial autocorrelation among the connected sites for most CECs. This was further supported by longer correlation lengths for 18 CECs than the average distance between the sampling sites. The spatial autocorrelation can be attributed to the physicochemical properties of CECs and local hydrological dynamics, including temperature, wind speed, and sunshine hours. For most CECs, local contribution predominated over neighbor influence with an average value of 75.5%. The results of this study provide new insight to evaluate CEC distributions, which will be beneficial to policymakers for the management and prioritization of CEC contaminants in the Houxi watershed.