The Human Health Assessment to Phthalate Acid Esters (PAEs) and Potential Probability Prediction by Chromophoric Dissolved Organic Matter EEM-FRI Fluorescence in Erlong Lake

Phthalate acid esters (PAEs) are suspected to cause wide environmental pollution and have adverse effects on human health. Three priority control phthalates, namely dimethyl phthalate (DMP), diethyl phthalate (DEP), and dibutyl phthalate (DBP), were determined in 45 water samples from the largest drinking water source in Jilin Province. Chromophoric-dissolved organic matter (CDOM), which are composed of complex compounds and are a proxy for water quality, can be monitored using a fluorometer. This study attempted to understand the correlations of the CDOM fluorescence regional integration (FRI) components with PAEs and CDOM characteristics under seasonal and spatial variations in the Erlong Lake. The characteristics of the CDOM absorption parameters in different water samples showed a higher aromatic content and molecular weight in October because of increased terrestrial inputs. The Σ3PAEs concentrations ranged from 0.231 mg L−1 to 0.435 mg L−1 in water, and DEP contributed to more than 90% of the Σ3PAEs. The FRI method identified five fluorescence components: one tyrosine-like (R1), one tryptophan-like (R2), one fulvic-like (R3), one microbial protein-like (R4), and one humic-like (R5) component. However, significant relationships exist between DEP and R3 (R2 = 0.78, p < 0.001), R4 (R2 = 0.77, p < 0.001), and R5 (R2 = 0.58, p < 0.001). Quantifying the relationship between CDOM and PAEs was highly significant, because the results will simplify the componential analysis of pollutants from a spatiotemporal perspective as compared to traditional chemical measurements. The human health risk assessment results revealed no human health risk (HQ < 1) in the Erlong Lake basin.


Introduction
Lakes are critical in the transportation, transformation, and storage of large amounts of carbon from terrestrial sources, and they contribute to regional effects on climate [1,2]. With the development of industry and agriculture, lakes are disturbed by anthropogenic activities, such as agricultural, industrial activities, and living organic pollutants that destroy water quality and ecological balance, carcinogenic risk of DEHP (diethylhexyl phthalate) exceeded the threshold limit value in agricultural soils. The risks that are posed by PAEs from drinking water on human health still exist [16].
The Erlong Lake is the drinking source of Siping city, Northeast plain of China. It has an irrigation area of approximately 6700-hm 2 cultivated land, which is the largest reservoir in the Jilin province. The increase agricultural activities, industrial wastewater, and domestic sewage discharge contributed to the increased pollution of Erlong Lake. The water quality of the Erlong Lake is closely related to the health of residents, growth of crops, fisheries, and ecosystems in the region. To maintain agricultural yields and activities, agricultural plastic films, fertilizers, and pesticides are used, which introduces more PAE contaminants to the lake. Subsequently, various farming activities, the location of the sewage discharge outlets, and residential density cause spatial variations in the PAEs in water, which may lead to differences in exposure risk levels. Nevertheless, the issue of pollution from PAE compounds has not been given enough attention. Ensuring safe drinking water for human health is a global problem [17].
Accordingly, the objectives of this study are to investigate: (i) the temporal and spatial distributions and variations in selected PAE compounds in the Erlong Lake, and assess the risks to human health, (ii) the sources and compositions of FDOM by the EEM-FRI method, and (iii) the relationship between five EEM-FRI-divided FDOM components and PAEs in surface waters, and discuss the potential probability in using CDOM fluorescence as a tool for predicting the concentrations of PAEs. These results provide a basis for the evaluation of health risk based on remote sensing. Similarly, quantitative assessments and the monitoring of PAEs in the Erlong Lake provide an early pollution warning and management measures for water sources conservation.

Study Area and Water Sampling
The Erlong Lake (43 • 7 45 -43 • 20 34 N, 124 • 46 1 -124 • 58 29 E) is located in the east Siping City of the Jilin Province, China. The east Liaohe River is the primary supply source and pollution source of the Erlong Lake. The lake area above the dam site is 3799 km 2 and it is located in the middle and upper reaches of the east Liaohe River basin. The water surface of the Erlong Lake spans two provinces, three cities and five counties, and is the water source of the residents of Siping city and the irrigation source for approximately 6700 hm 2 of cultivated land. The total storage capacity is 1.762 billion m 3 , and the average flow rate of the designed flow is 10,100 m 3 /s. The Erlong Lake is located in the transitional zone of humid climate and semi-humid climate with an average air temperature of 5.8 • C and annual precipitation of 650 mm [18].
A total of 45 water samples that were collected at 25 points on the lake surface were recorded by a global positioning system ( Figure 1). The detailed sampling location information regarding each field survey from the various sampling points is shown in Figure 1. The sampling courses were performed in June (normal season) and October (dry season) in 2017. The water samples were collected using acid-cleaned Niskin bottles, and subsequently the water sample was placed in a brown bottle. At each point, 4 L water samples were collected from a depth of 0-0.2 m from the surface. The collected samples were stored at 4 • C in coolers and were subsequently transported to the laboratory within 3-5 h. The physical and chemical parameters were determined within 24 h.

Water Quality Determination
The water quality parameters, electrical conductivity (EC) and pH, of each water sample were measured onsite using a portable multiparameter water quality analyzer (Hach, Loveland, CO, USA). By referring to the Environmental Quality Standards for Surface Water (GB3838-2002, China) (http://kjs.mep.gov.cn/), chemical oxygen demand (COD) was determined using dichromate, ammonia nitrogen (NH 3 -N) via Nessler's reagent colorimetry, and total phosphorous (TP) using the molybdenum blue method after the samples were digested with potassium peroxydisulfate [19]. To determine the dissolved organic carbon (DOC) concentrations, all of the water samples were filtered through a pre-combusted Whatman GF/F (1825-047) filter (0.45 µm) under low vacuum and were measured by the Shimadzu TOC-VCPN analyzer. Chlorophyll a (Chl-a) was extracted from the filtered (0.45-µm Whatman GF/F) sample using a 90% acetone solution; subsequently, the concentrations were determined using a UV spectrophotometer (UV-2006 PC, Shimadzu, Kyoto, Japan); the detailed process can be found in Song et al. (2012) [20]. The total suspended matter (TSM), inorganic suspended matter (ISM), and organic suspended matter (OSM) were determined by gravimetrical analysis. Detailed descriptions about this analysis are available in Song et al. (2012) [19].

Sample Extraction and Chemical Analysis
The water samples were extracted using a classical soil-phase extraction method [15]. Water samples of 2 L had previously been acidified to pH 7 by hydrochloric acid, using OASIS HLB to enrich the PAEs in water. Subsequently, 5% methyl alcohol was used to clean the column, 9-mL of a mixed solution of 95% diethyl ether, and 5% methyl alcohol was used to elute the sample with the flow velocity of 1 mL min −1 . Finally, the remains were reduced to 1 mL under gentle nitrogen flow. Prior to the instrumental analysis, the 0.05 ng internal standard was added to the sample.
Quantification of the PAEs was performed on the GCMS-QP 2010 Plus gas chromatography-mass spectrometry system (Shimadzu Corporation, Kyoto, Japan). A DB-5 MS (30 m × 0.25 mm i.d.; 0.25-µm film thickness, Agilent, Santa Clara, CA, USA) fused-silica capillary column was used to separate the target compounds. The column temperature program was initiated at 60 • C for 1 min and increased to 220 • C at a rate of 20 • C/min, held for 1 min, and finally to 280 • C for 5 min and held for 8 min. The carrier gas helium (99.999% purity) was flowed at a constant flow rate of 1.0 mL min −1 . A 1-µL sample was injected in the splitless mode. The temperatures of the injector, each extract, the GC-MS transfer line, and the post run were 250 • C, 250 • C, 280 • C, and 285 • C, respectively. These were operated in the electron impact mode (70 eV).

Quality Analysis and Quality Control
An appropriate concentration of the standard solution was spiked into each sample to estimate the recovery and the performance of the methods. During the analysis, a procedural blank, a matrix-spiked sample, and duplicates were processed for each batch of the water samples. The recoveries from the standard mixture solution of three PAEs, including dimethyl phthalate (DMP), diethyl phthalate (DEP), and dibutyl phthalate (DBP) (50 ng L −1 each) ranged from 76.3% to 108.6% in the spiked water. Each water sample was analyzed in triplicate with relative standard deviations of less than 23%. The method determination limit of the PAEs ranged from 0.08 ng L −1 to 0.51 ng L −1 .

Chromophoric-Dissolved Organic Matter Absorbance and Three-Dimensional Fluorescence
The water samples were first filtered under low vacuum through a pre-combusted Whatman GF/F (1825-047) filter (0.7-µm pore size), and subsequently, through a pre-rinsed 25-mm diameter Millipore membrane cellulose filter (0.22-µm pore size) into brown glass bottles [6]. The Shimadzu UV-2600 spectrophotometer with 1-cm quartz cells between 200 nm and 800 nm at 1 nm increments with Milli-Q water as a reference was used to measure the twice-filtered water samples [20]. The CDOM absorption coefficient α CDOM (λ ) was calculated from the measured optical density (OD), as follows: where γ is the cuvette path length (0.01 m) and 2.303 is the conversion factor. OD(λ) is the optical density at the same wavelength [6]. The CDOM EEMs were obtained using the Hitachi F-7000 fluorescence spectrometer (Hitachi High-Technologies, Tokyo, Japan) with a 700-V xenon lamp as the light source [20]. The wavelength ranges of excitation were 220-450 nm at 5-nm sampling interval and 250-600 nm for the emission at a 1-nm sampling interval, with the scanning speed being maintained at 1200 nm/min. The spectrum of the Milli-Q water recorded as the blank was subtracted from all of the EEMs to eliminate the water Raman scatter peaks [21]. To eliminate the inner-filter effect and the Raleigh scattering, the EEM fluorescence spectra need to be corrected for absorbance; more details can be found in Li et al., 2017 [6].
The fluorescence indices FI 370 and FI 310 were used to characterize the CDOM source.  [22] as the ratio of the excitation fluorescence intensity to that of the emission intensity, as follows: E x /E m = (370/450 nm)/(370/500 nm) [2], to distinguish between terrestrially (FI 370 < 1.4) and microbially (FI 370 > 1.9) derived fulvic acids. To determine the contribution of the autochthonous, FI 310 is defined as the ratio of excitation and emission intensities, as follows: E x /E m = (310/380 nm)/(310/430 nm) [2]. When the fluorescence index FI 310 is lower than 0.7, it indicates the low levels of autochthonous CDOM components in water bodies. When the fluorescence index FI 310 is higher than 0.8, it indicates the large amounts of autochthonous CDOM components in the water bodies due to the biological activity. When the fluorescence index FI 310 is between 0.7 and 0.8, it implies intermediate autochthonous CDOM components [2,23].

Health Risk Assessment
By referring to the widely applied risk assessment guidelines that are recommended by the United States Environmental Protection Agency (USEPA) (2013) [14,15], this study assesses the potential health risk to inhabitants, as well as the non-cancer and carcinogenic risks of PAEs. The non-carcinogenic risk is considered by the hazard quotient (HQ) determined by the average daily dosage (ADD, mg kg −1 day −1 ) and the reference dosage (RfD, mg kg −1 day −1 ) to represent the exposure pathway of the intake, ingestion, dermal absorption, and inhalation. The formulas are the following: where C is the contaminant concentration (mg kg −1 d −1 ), DR is the daily consumption rate (L d −1 ), EF is the exposure frequency (d year −1 ), ED is the exposure duration (year), BW is the body weight (kg), and AT is the average lifetime exposure (d). RfD is defined as the daily maximum permissible level of pollutants. RfD and ADD were collected from the integrated risk information system (IRIS) database that was developed by the EPA. The hazard index (HI) was calculated to assess the non-cancer risk by the level comparison of the PAE compounds. It is considered that HI > 1 indicates a high non-cancer risk, while HI < 1 indicates a low non-cancer risk. The carcinogenic risk is considered by the carcinogenic risk (CR) (unitless), as follows: where CFS is the slope factor of the carcinogen ((mg kg −1 d −1 ) −1 ). The risk index (RI) < 1 × 10 −6 indicates a very low carcinogenic risk, whereas RI > 1 × 10 −4 means an unacceptable risk level.

Water Chemistry
The spatial-temporal variations in water quality are related to land use and land cover [24]. When all of the water samples were pooled together in the Erlong Lake, a significant seasonal variability of water quality parameters occurred, i.e., pH, EC, DOC, Chl-a, TP, NH 3 -N, COD, ISM, and OSM for the 45 water samples are displayed in Table 1. The pH in June (average ± SD, 8.0 ± 0.26) is higher than that in October (average ± SD, 7.8 ± 0.04), which was associated with a high EC (average ± SD, 463.616.76 µS cm −1 ) in June and a lower EC (average ± SD, 349.814.75 µS cm −1 ) in October. DOC represents an essential link between terrestrial and aquatic ecosystems [25,26]. The average DOC concentration in October (23.63 mg/L) is lower than that in June (26.38 mg/L). Likewise, a higher Chl-a, COD, ISM, and OSM were also found in water samples in June than in October. The average Chl-a concentration in June (39.01 µg L −1 ) is higher than that in October (14.51 µg L −1 ); further, during the sampling period, many floating algae appeared on the lake surface. This result indicated the higher contribution of algal and microbial activity in June, which is related to the increased temperature and sunlight in the summer [25]. However, low nutrients, i.e., TP and NH 3 -N concentration, were exhibited in June (Table 1). This may be associated with the higher precipitation in the summer owing to the dilution effect. COD is related to the organic pollution levels; the average COD concentration in June (28.28 mg L −1 ) is higher than that in October (22.57 mg L −1 ), signifying more human and farming activities in June. The differences were assessed statistically (ANOVA), and the p value for each water quality parameter in June and October was lower than 0.05.

Temporal and Spatial Variations of Phthalate Acid Esters
The distributions of the three US EPA priority PAEs (Σ3PAE, including dimethyl phthalate (DMP), diethyl phthalate (DEP), dibutyl phthalate (DBP)) in the Erlong Lake from the sampling sites were investigated in June and October. The results of the relative contributions of the three PAE congeners are presented in Table 2 and Figure 3. The mean concentrations of DBP detected in the present study were well above the reference dose (RfD: 3 µg L −1 ), and is regarded as unsafe in China for surface water (Environmental Quality Standard for Surface Water of China, GB3838-2002) ( Table 2). The Σ3PAEs concentrations ranged from 0.231 mg L −1 to 0.435 mg L −1 , and the arithmetic mean is 0.318 mg L −1 , from 0.447 mg L −1 to 0.654 mg L −1 , and the arithmetic mean is 0.546 mg L −1 for June and October, respectively. Among the three PAEs that were detected in the Erlong Lake in June, DMP, DEP, and DBP were detected in all of the samples at average concentrations of 0.013, 0.299, and 0.006 mg L −1 , respectively; in October, the average concentrations of DMP, DEP, and DBP were 0.006, 0.533, 0.007 mg L −1 , respectively. The value of DEP was significantly higher in October than in June (ANOVA, p < 0.05); this phenomenon was caused by the seasonal hydrology and temperature [16]. The seasonal average of DMP, DEP, and DBP concentrations were in the order of 0.01, 0.403, and 0.007 mg L −1 . When considering the individual PAE congeners in this area, the results show that DEP is the most abundant in the water samples, which is the major component of three PAEs, contributing to 95.9% (Table 2 and Figure 3). The next most dominant PAE was DMP, contributing to 2.4%, followed by DBP, contributing 1.7% of the ∑3PAEs concentrations in all 45 water samples. The differences were assessed statistically (ANOVA), and the p value for DEP and DBP in June and October were lower than 0.05.  As shown in Figure 3, significant variations in DMP, DEP, and DBP occurred at different sampling points in June and October. It is noteworthy that the PAE levels may be related to the types of local waste discharge, such as sewage water, food packaging, and scrap material during the sampling period. In addition, obvious differences were found in the proportions of DMP and DBP to the total PAEs during the two periods, indicating various possible sources of PAEs. A previous study has shown the positive correlation between the concentrations of PAEs with the low-molecular-weight DEP and DMP; further, the agricultural runoff and the solubility of DMP and DEP indicate their potential abundance in water (dissolution phase) [27]. DBPs are used in epoxy resins and special adhesive formulations, indicating that the PAEs at the sampling locations are from industrial pollution [16]. The concentration of DBP and DEP were 0.012 mg L −1 and 0.033 mg L −1 at the 13th point in October and June, respectively. The 13th sampling point that was near the sewage discharge outlet and the values of DBP and DEP at point of 13 were higher than in other sampling sites, indicating that the source of DBP and DEP were from the allochthonous. In general, the results show that the spatial distributions of PAEs in the Erlong Lake were site specific.

Fluorescence Regional Integration-Divided Fluorescent-Dissolved Organic Matter Components
FRI has been widely used to quantitatively analyze and study the EEM fluorescence characteristics of CDOM [11,28]. The total fluorescence intensities, the fluorescence intensities of five components, and the relative contributions of different components to the total fluorescence intensities P i showed seasonal variations. As shown in Figure 4a,b, EEM was divided into five regions by FRI, each representing a CDOM fluorescence fraction: the tyrosine-like (R1), tryptophan-like (R2), microbial protein-like (R4), fulvic-like (R3), and humic-like (R5) fluorescent components [11,29]. The excitation-emission area volumes Φ i and P i (i = 1, 2, 3, 4, 5) were the proportion of the total fluorescence intensities and the relative contributions of five different components to the total fluorescence intensities, respectively. As shown in Figure 3a,b, the total fluorescence intensities F SUM decreased from 5.1 × 10 10 to 2.8 × 10 10 nm in June and October. The fluorescence intensities Φ 5 account for 50% in June and 54% in October of the total fluorescence intensities. The fluorescence intensities Φ 3 and Φ 5 were dominant in the two months and they were from the allochthonous. The value of P (3+5) was increased from 82.4% in June to 86.4% in October, indicating more allochthonous substances in October.

Chromophoric-Dissolved Organic Matter Absorption and Fluorescence Characteristic
The composition and distributions of CDOMs were impacted by the environmental factors (water quality, soil, vegetation, land use, and rainfall) and could generally affect the absorption and fluorescence of CDOMs at certain wavelengths [20]. Generally, the absorption coefficient a CDOM (350) is used to characterize the concentration of CDOMs [7,29] and a CDOM (254) can be used as a proxy for characterizing the aromaticity of CDOMs [7,30]. The average a CDOM (350) and a CDOM (254) in October have exhibited significantly higher values than in June ( Table 3). The spectral slopes S 275-295 and E 250:365 [a CDOM (250)/a CDOM (365)] could be used as an indicator for terrigenous DOC percentage, which were used to track the changes in CDOM molecule size [31]. The S 275-295 values (0.24 ± 0.12 nm −1 ) in October were lower than those in June (0.26 ± 0.10 nm −1 ), and they exhibited a consistent tendency of E 250:365 and S 275-295 (ANOVA, p < 0.05). This indicated that the increase in aromatic compounds and the percentage of high-molecular-weight fulvic acid of CDOMs in October was greater in October. The high E 250:365 values are associated with the low content of aromatic hydrocarbon and the molecule weights were remarkable in June. These results that are associated with the higher contribution from algal and microbial activity in June are consistent with the results of water quality presented in Table 1. Table 3. Chromophoric-dissolved organic matter (CDOM) absorption parameters from water samples collected in the Erlong Lake (Avg. ± SD). As an effective index to characterize the DOC concentration and DOM aromaticity, we used SUVA 254 values with the ratio of a CDOM (254)/DOC [30][31][32]. Higher SUVA 254 values indicate aquatic systems with abundant vascular plant inputs, and the allochthonous sources dominated the organic matter content [31][32][33]. Meanwhile, the lower values indicate more autochthonous sources (algal and microbial). As shown in Table 3, the average value of SUVA 254 was higher in October than in June (ANOVA, p < 0.05). Relatively lower SUVA 254 measurements were found from point 18 to point 25, indicating that the aromatic moieties of CDOM in this environment were lower when compared with the other points in June and October (Figure 5c [25], point 5 had a lower color DOC concentration that was resulting from the algal-derived DOC in the waters. As shown in Table 3 and Figure 5a, the average values of the fluorescence index FI 370 is 1.39 in June and 1.14 in October (ANOVA, p < 0.05); the average values of the fluorescence index FI 310 is 1.04 in June and 0.81 in October (ANOVA, p < 0.05). In June, the average value of FI 310 was above 0.8 and FI 370 was approximately equal to 1.4, indicating that CDOM sources were derived from the autochthonous; in October, the average value of FI 310 is higher than 0.8 and FI 370 is less than 1.4; the primary origin of CDOM is the terrestrial humic-like substances. A positive linear relationship was also found between

Fluorescence Regional Integration Fluorescence Component Versus Chromophoric-Dissolved Organic Matter
As shown in Table 4, significant positive correlations exist between F R for fulvic-like components R3, microbial protein-like R4, and humic-like components R5 (p < 0.01), indicating that they may originate from similar sources. However, no strong correlations were found for the tyrosine-like R1, the tryptophan-like R2, and the microbial protein-like R4 for the water samples in the Erlong Lake. The tryptophan-like components F R2 showed a medium positive correlation with the tyrosine-like F R1 (R 2 = 0.68, N = 45), fulvic-like components F R3 (R 2 = 0.67, N = 45), and humic-like components F R5 (R 2 = 0.69, N = 45), indicating that parts of these F R1 fluorescent components were likely from some common sources. Significant positive correlations were found between F R3 , F R4 , and F R5 (R 2 > 0.88, p < 0.01).

Fluorescence Regional Integration Fluorescent Components Versus Water Quality Parameters
As shown in Table 5, significant correlations were exhibited between DOC and F R for the fulvic-like R3 (R 2 = 0.68, p < 0.01; N = 45), the microbial protein-like R4 (R 2 = 0.68, p < 0.01; N = 45), and the humic-like R5 (R 2 = 0.58, p < 0.01; N = 45), respectively. Likewise, Chl-a also showed medium relationships with F R for the fulvic-like R3 (R 2 = 0.57, p < 0.01; N = 45), and the microbial protein-like component R 4 (R 2 = 0.50, p < 0.01; N = 45). It demonstrated that some DOCs, Chl-a, and CDOMs from the water samples in the Erlong Lake stemmed from the fulvic-like and microbial protein-like component sources. The presence of microbial protein-like component R4 in the Erlong Lake samples was similar to the degradation of phytoplankton releases and humus components that were caused by microbial activities [9] and microbial oxidized components [34]. The fulvic-like R3 were derived from terrestrial substances from the soil, wetlands, or agricultural sites. Table 5. Pearson correlation analysis between water quality parameters and fluorescence regional integration (FRI) fluorescent components. Notably, medium correlations exist between a(350) and F R for the fulvic-like R3 (R 2 = 0.57, p < 0.01; N = 45), the microbial protein-like component R4 (R 2 = 0.52, p < 0.01; N = 45), and humic-like R5 (R 2 = 0.55, N = 45) ( Table 5). These results imply that the CDOM with humic-like and microbial protein-like components were from a common source in the lake. However, a relatively weak relationship was revealed between a(350) and F R for R1 and R2, respectively. In summary, based on such discrepancies, we speculate that the terrigenous pollutants and autochthonous tryptophan-like substances that were caused by microbial activities were important factors regulating the migration and transformation of CDOMs in the Erlong Lake.

Fluorescence Regional Integration Fluorescent Components Versus Phthalate Acid Esters
When all the samples were pooled together (N = 45), a correlation also existed between the FRI fluorescent components and PAEs congeners, as shown in Table 6. As shown in Figure 6, in the Erlong Lake, R3 and DEP showed a significant correlation with the correlation coefficient of 0.78 (two-tailed, p < 0.001). Likewise, R4 (R 2 = 0.77, p < 0.001), and R5 (R 2 = 0.58, p < 0.001) also showed a significant correlation with DEP, respectively. ∑3PAEs and R3, R4, and R5 showed significant correlations with correlation coefficients of 0.76 (two-tailed, p < 0.001), 0.75 (two-tailed, p < 0.001), 0.58 (two-tailed, p < 0.001), respectively. We speculated that the CDOMs and PAEs underwent chemical adsorption and contained fluorophore; consequently, they had similar origins to the waters that were affected strongly by terrestrial sewage inputs and agricultural insecticides. The result shows a positive correlation between the concentrations of PAEs with DEP and the agricultural soil runoff. It shows the contribution of terrestrial fulvic-like fluorophores, humic-like fluorophores, and microbial protein-like to R3, R4, and R5. The solubility of DEP indicates the reason for their potential enrichment in water and with extremely low production and consumption capacity [26]. PAEs contributed part of the CDOM absorption in natural surface waters because a small amount of organic matter was present. Although the results were limited by the conditions of the single research area, they may provide an overall situation of PAEs that are associated with FRI fluorescent components. Based on the correlation between DEP, R3, and R4 of water samples that were collected in the Erlong Lake, the CDOM fluorescence using a fluorometer has the potential probability to monitor PAEs in natural surfaces. However, it requires large amounts of samples for verifications and adjustments in different regions. In the future, we will continue studying the chemical transformation between humic-like, fulvic-like, and PAEs.

Health Risk Assessment
According to the results above, the concentration of DBP in Erlong Lake were higher than the recommended drinking water limit (i.e., the level present in public water supplies must not exceed the drinking water standard: 3 µg L −1 for China, GB3838-2002). The values of the parameters for non-cancer risks can be referred from the US EPA (2013), Kong et al. (2017) [16], and Wang et al. (2015) [35]. Assuming a daily water consumption rate of 2 L and an average body weight of 60 kg for adults, the values of EF, ED, and AT were 350 d year −1 , 30 years, and 26,280 days, respectively. The RfD values of DBP and DEP via drinking water from the Erlong Lake was estimated to be 100 µg kg −1 day −1 and 800 µg kg −1 day −1 , respectively by the US EPA. The RfD value of DMP was 1000 µg kg −1 day −1 according to Wang et al. 2015 [35]. Among the individual PAE congeners that were studied, DBP, DEP, and DMP were recognized as non-cancer compounds. The non-carcinogenic risk assessment is based on a nonlinear model. In this study, the non-carcinogenic risk of PAE compounds to local inhabitants through the dietary route was evaluated. The HQ was employed to assess the non-carcinogenic risk in this study.
As shown in Figure 7, higher values were observed in October and a higher HQ value was observed at site 13. All of the non-cancer risks from the PAE values that were estimated in this study were far below the recommended limits (HQ < 1) in the two months. Therefore, no human health risks by PAE compounds occurred in the Erlong Lake. However, PAEs are partially metabolized by organisms, and future experiments will be focused on the factors affecting the concentration of PAEs in water and the relationship between the fluorescence component of CDOMs and PAE congeners.

Conclusions
The Erlong Lake is the largest drinking source of the residents of Siping City in the Jilin Province, and the irrigation area of cultivated land is approximately 6700 hm 2 . The complex optical properties of the Erlong Lake were caused by the large amounts of pollutants and dissolved constituents. This study exemplified the spatial and the temporal characteristics of three PAEs congeners and the CDOM characteristics, based on the FRI fluorescent components of 45 water samples that were collected in 2017. The Σ3PAEs concentrations ranged from 0.231 mg L −1 to 0.435 mg L −1 , and DEP accounted more than 90% of ∑3PAEs in this lake. In October, lower CDOM optical parameters of a CDOM (254), E250:365, S 275-295 , and S R signified the increased aromatic content and the molecular weight of CDOM. A positive correlation existed between the CDOM FRI fluorescent components and DEP (two-tailed, p < 0.001), and they were affected by the individual samples with high PAEs. Based on the HQ, PAE compounds in the Erlong Lake did not pose a risk to human health. CDOM fluorescence might be a potential tool for monitoring PAE concentration and transport, as well as for rapid health risk assessments based on many experiments in the future.