لخّصلي

Materials and methods Study area description To test the integrated approach, a typical agricultural watershed in Eastern China was selected as the case study area.For a specific sampling site, the MLR-APCS model can be expressed as follows: Ci 1/4 X0i th?pj 1/4 1Xji APCSj d3TH where Ci is the excessive concentration of trace metal i in sediment; APCSj is the absolute principal component score of anthropogenic source j; Xji is the regression coefficient; X0i is the contribution of undefined anthropogenic sources by PCA; and the combined term Xji x APCSj represents the contribution of anthropogenic source j. Lead isotopic mixing model Lead isotopic composition was characterized to discriminate the different Pb sources, and their respective contributions were calculated by using mixing model.The calculation equations in this software are as follows: f s 1/4 p1 f 1 th p2 f 2 th?th pn f n d4TH 1 1/4 p1 th p th pn d5TH where fs is the Pb isotopic composition of sediment sample; f1, f2, and fn are the Pb isotopic compositions of different sources; p1, p2, and pn are the contributions of different sources (%).In this study, there are 2 Pb isotopic compositions (206Pb/207Pb and 208 207 Pb/ Pb) and 4 sources (background soil, chemical fertilizer, livestock manure, and atmospheric deposition).If there are n isotopic compositions and n + 1 sources, we can find certain contribution values, but if there are > n + 1 sources, feasible solutions can be obtained according to isotopic mass balance (Phillips and Gregg 2003).The total concentrations of Cd, Cu, Ni, Pb, Cr, Zn, Fe, and Al in the digestion solutions were then analyzed by an inductively coupled plasma-atomic emission spectroscopy (ICPAES; IRIS Intrepid II XSP, Thermo Electron, USA).With the EF value for each trace metal, its excessive concentration in sediment can be determined as follows (Huang et al. 2018): X excessive 1/4 X sediment-Ysediment dX=YTHbaseline 1/4 X sediment d1-1=EFTH d2TH Multiple linear regression of absolute principal component scores The MLR-APCS model was used to estimate respective contributions of major anthropogenic trace metal sources in the river sediments.Because the source discrimination power of Pb stable isotopes is mainly due to 206Pb, 207Pb, and 208Pb (Sangster et al. 2000), they were measured for sediment, soil, and anthropogenic source samples by an inductively coupled plasma-mass spectrometry (ICP-MS; Agilent 7500, Agilent Technologies, USA).1 Location of the study area showing land uses and sampling sites For sediment and soil samples, they were digested in a closed poly-tetrafluoroethylene system with an acid mixture of HNO3 (5 mL)-HF (1 mL)-HClO4 (1 mL) at 180 ?C for 10 h (Huang et al. 2015).The method detection limits, defined as 3 times the standard deviation of blank measurements, were 0.1, 0.4, 0.4, 1.4, 0.5, 1.2, 8.9, and 8.9 mg/L for Cd, Cu, Ni, Pb, Cr, Zn, Fe, and Al, respectively.Quantitative source identification methods Enrichment factor (EF) The EFs were calculated to determine the excessive concentrations of trace metals in the river sediments.In addition, 5 atmospheric deposition samples were collected from long-term monitoring sites, which evenly distributed in the watershed for collecting total suspended particulates (Fig.Currently, the dry farmland has occupied the largest watershed area, where wheat is the main crop being cultivated and rainfall, surface, and groundwater supply all irrigation water.APCS were obtained from the principal component analysis (PCA), which is a data dimensionality reduction analysis and can identify the major trace metal sources (Mico et al. 2006; Cai et al. 2019b).A multi-element standard solution (GSB04-1767-2004, Guobiao Testing & Certification Co., Ltd., China) was used to establish calibration curves.For quality control, each digestion batch included 2 reagent blanks, 1 standard reference material, and 1 duplicate sample to assess the accuracy and precision of the measurements.Replicate analysis of the standard reference material GBW-07304 (National Research Center for Geo analysis, China) showed good accuracy, with overall recovery rates ranging from 97 to 106%.Metal concentration and Pb isotope analyses Fig.1). 1).1). 1).

Materials and methods
Study area description
To test the integrated approach, a typical agricultural watershed in Eastern China was selected as the case study area. This watershed has a total drainage area of approximately 122 km2, which is characterized by asymmetrical hills with a maximum elevation close to 1000 m (Fig. 1). In history, this watershed was covered mainly with natural forestland, but it has experienced an extensive agricultural reclamation since the 1980s. Currently, the dry farmland has occupied the largest watershed area, where wheat is the main crop being cultivated and rainfall, surface, and groundwater supply all irrigation water. As it is located in a hilly region, there are no obvious industrial point sources in this watershed. The watershed climate is continental monsoon, with the average annual temperature and rainfall of 13.2 °C and 790 mm, respectively. Local rivers are influenced by a seasonal hydrological regime, which generally flow from northeast to southwest.
Samples collection and pretreatment
In April 2017, a total of 19 river sediment sites were selected from the typical agricultural watershed (Fig. 1). At each sediment sampling site, 3 sub-samples (0–10cm depth) were collected, pooled and mixed into a composite sample. In order to obtain watershed background values, 9 soil samples were also collected from selected reference sites (Fig. 1). These reference sites were located in natural forestland area, where the soils have almost not been impacted by agricultural activities. At each reference site, 3 sub-samples were taken at a minimum of 40-cm depth to avoid surface contamination by atmospheric deposition, and they were then fully mixed to get a composite sample. For Pb isotopic analysis, 10 chemical fertilizer and livestock manure samples were obtained from local farmers. In addition, 5 atmospheric deposition samples were collected from long-term monitoring sites, which evenly distributed in the watershed for collecting total suspended particulates (Fig. 1). The atmospheric deposition monitoring was performed between 8 April 2016 and 8 April 2017 using clean bottles. After collection, all samples were sealed in polyethylene bags and transported to the laboratory immediately. There, they were air dried at room temperature, ground with an agate mortar, and passed through a 63-μm mesh sieve.
Metal concentration and Pb isotope analyses
Fig. 1 Location of the study area showing land uses and sampling
sites
For sediment and soil samples, they were digested in a closed poly-tetrafluoroethylene system with an acid mixture of HNO3 (5 mL)–HF (1 mL)–HClO4 (1 mL) at 180 °C for 10 h (Huang et al. 2015). The total concentrations of Cd, Cu, Ni, Pb, Cr, Zn, Fe, and Al in the digestion solutions were then analyzed by an inductively coupled plasma-atomic emission spectroscopy (ICPAES; IRIS Intrepid II XSP, Thermo Electron, USA). A multi-element standard solution (GSB04-1767-2004, Guobiao Testing & Certification Co., Ltd., China) was used to establish calibration curves. For quality control, each digestion batch included 2 reagent blanks, 1 standard reference material, and 1 duplicate sample to assess the accuracy and precision of the measurements. Replicate analysis of the standard reference material GBW-07304 (National Research Center for Geo analysis, China) showed good accuracy, with overall recovery rates ranging from 97 to 106%. The relative standard deviations for duplicate samples were less than 5%. The method detection limits, defined as 3 times the standard deviation of blank measurements, were 0.1, 0.4, 0.4, 1.4, 0.5, 1.2, 8.9, and 8.9 mg/L for Cd, Cu, Ni, Pb, Cr, Zn, Fe, and Al, respectively.
Because the source discrimination power of Pb stable isotopes is mainly due to 206Pb, 207Pb, and 208Pb (Sangster et al. 2000), they were measured for sediment, soil, and anthropogenic source samples by an inductively coupled plasma-mass spectrometry (ICP-MS; Agilent 7500, Agilent Technologies, USA). During ICP-MS measurement, the peak pattern of maximum (20 points) was chosen, and the integration time was set as 0.3 s (Luo et al. 2015). To optimize the analytical performance, the digestion solutions above were diluted with 5% HNO3 to approximately 20 μg/L according to total Pb concentration (Margui et al. 2006). The method detection limits were 0.06, 0.07, and 0.07 μg/L for 206Pb,
207 208
Pb, and Pb, respectively. The relative standard deviations for duplicate samples were generally lower than 1%. In addition, solutions of the standard reference material NIST 981 (National Institute of Standards and Technology, USA) were measured after every 5 samples for calibration and quality control. The average measured ratios of 206Pb/207Pb and 208Pb/207Pb for the standard reference material were 1.091 and 2.371, which were generally in good agreement with the certified values of 1.093 and 2.370, respectively.
Quantitative source identification methods
Enrichment factor (EF)
The EFs were calculated to determine the excessive concentrations of trace metals in the river sediments. It was defined as the concentration ratio of a metal of interest to a reference metal in sediment, divided by their baseline ratio (N’guessan et al. 2009).
¼ ððX=YÞsediment ð Þ
EF1
X=YÞbaseline
where X is the total concentration of the metal of interest, and Y is the total concentration of the reference metal. Because Al is more abundant than Fe in concentration and has no obvious anthropogenic sources in the agricultural watershed, it was selected for reference purpose. In addition, previous studies have suggested that the local soil background values can give a more reliable basis than the average crust values when calculating EF (Blaser et al. 2000). Therefore, the total concentrations of X and Y in the watershed background soils were used as the baseline values. With the EF value for each trace metal, its excessive concentration in sediment can be determined as follows (Huang et al. 2018):
X excessive ¼ X sediment−Ysediment ðX=YÞbaseline ¼ X sediment ð1−1=EFÞ ð2Þ
Multiple linear regression of absolute principal component scores
The MLR-APCS model was used to estimate respective contributions of major anthropogenic trace metal sources in the river sediments. This model is based on the assumption that all potential sources contribute linearly to the total trace metal concentration in a receptor site (Haji Gholizadeh et al. 2016). APCS were obtained from the principal component analysis (PCA), which is a data dimensionality reduction analysis and can identify the major trace metal sources (Mico et al. 2006; Cai et al. 2019b). MLR was subsequently run with the trace metal concentration as dependent variable and the APCS as independent variables. In this study, the excessive concentrations of trace metals in sediments rather than total concentrations of them were used in PCA to identify major anthropogenic trace metal sources. For a specific sampling site, the MLR-APCS model can be expressed as follows:
Ci ¼ X0i þ∑pj¼1Xji APCSj ð3Þ
where Ci is the excessive concentration of trace metal i in sediment; APCSj is the absolute principal component score of anthropogenic source j; Xji is the regression coefficient; X0i is the contribution of undefined anthropogenic sources by PCA; and the combined term Xji × APCSj represents the contribution of anthropogenic source j.
Lead isotopic mixing model
Lead isotopic composition was characterized to discriminate the different Pb sources, and their respective contributions were calculated by using mixing model. Previousstudieshave used a three end-member mixing model to calculate source contributions, but it can only cope with one natural and two anthropogenic sources (Luo et al. 2015). IsoSource is a mixing model calculation software, which can cope with more than three sources (Phillips and Gregg 2003; Huang et al. 2015). The calculation equations in this software are as follows:
f s ¼ p1 f 1 þ p2 f 2 þ⋯þ pn f n ð4Þ
1 ¼ p1 þ p þ pn ð5Þ
where fs is the Pb isotopic composition of sediment sample; f1, f2, and fn are the Pb isotopic compositions of different sources; p1, p2, and pn are the contributions of different sources (%). If there are n isotopic compositions and n + 1 sources, we can find certain contribution values, but if there are > n + 1 sources, feasible solutions can be obtained according to isotopic mass balance (Phillips and Gregg 2003). In this study, there are 2 Pb isotopic compositions (206Pb/207Pb and
208 207 Pb/ Pb) and 4 sources (background soil, chemical fertilizer, livestock manure, and atmospheric deposition).