Occurrence and spatial variations
Of the 140 target PPCPs, 42 were detected in the two lakes at one or more sampling sites. Among these compounds, five PPCPs, carbamazepine, diethyltoluamide, lidocaine, cotinine, chlorpheniramine, and triadimefon were found in both lakes. Approximately 11.5% (16 out of 140) of the analyzed PPCPs in Tai Lake were detected at the concentrations above the limit of detection levels, and circa 22.3% (31 out of 140) in Baiyangdian Lake. Concentrations of PPCPs in surface waters from Tai and Baiyangdian Lakes are summarized in Fig. 2. There were differences in the number, types and concentrations of PPCPs detected in waters of the two lakes. This result might be caused by several factors, including differences in use and release, removal efficiency of the WWTPs, degradation rate, temperature and dilution of receiving waters [8, 41]. Compared with Baiyangdian Lake, Tai Lake possess richer water resources and higher temperature that could result in PPCPs being diluted and more quickly being degraded. Consumption of caffeine is associated with aspects of financial capability [9], so it was more frequently detected in Tai Lake, which located in the eastern developed area of China. The bactericide, triadimefon and insect repellent, diethyltoluamide were consumed more in agricultural regions, near Baiyangdian Lake. Maximum concentrations of antibiotics sulfamethoxazole, griseofulvin, and lincomycin reported in Baiyangdian Lake were 34.5, 12.2, and 113.4 ng/L, while the antibiotic trimethoprim was 0.6 ng/L observed in Tai Lake. These results were consistent with amounts of antibiotics used in the two regions [2].
Total concentrations of the 16 PPCPs detected at ten sites in Tai Lake, ranged from 26.02 to 42.72 ng/L, with a median concentration of 31.11 ng/L. Maximum concentrations of individual chemicals ranged from 0.04 ng/L for famotidine to 25.77 ng/L for caffeine, and detection of frequencies of 10%-100%. Nine chemicals, caffeine, valsartan, diethyltoluamide, cotinine, carbamazepine, chlorpheniramine, trimethoprim, and triadimefon, were positively detected in all 10 samples. Concentrations of carbamazepine, ranging from 0.63 to 1.86 ng/L, were similar to concentrations previously reported for Tai Lake, which ranged from 0.24 to 8.74 ng/L [19]. However, the maximum concentration of ibuprofen was found to be 1.48 ng/L, which was less than those mean concentrations reported previously for Tai Lake (65.3 ng/L) [19], Liao River (246 ng/L) [23] and the Peal River (1417 ng/L) [42]. Diclofenac, propranolol and erythromycin were not detectable in all water samples during this study, have been previously reported to occur at relatively great concentrations [19].
In Tai Lake, coefficients of variation (CV; CV=mean concentration/standard deviation) for concentrations of individual PPCPs ranged from 12% to 258%. This result indicated spatial variations and large differences among chemicals. Spatial variation might result from a combination of distances of sampling sites from sources of emissions and variations in volumes of discharges [43]. For example, the predominant pollutant, caffeine, was present at relatively great concentrations (up to 25.77 ng/L) at Meiliang Bay (T5, T8, T9, T10) near a densely populated and scenic spot [44] and a relatively stable, thermal stratification that had been established for some days [45]. Eight PPCPs occurred at the greatest concentrations at T7, because the Yincungang River receives domestic wastewater from a densely populated and urbanized area [45]. These results were in general agreement with previous observations in sediments [20], where caffeine was the dominant pollutant near the Yincungang River estuary, and PPCPs in the west of Tai Lake exhibited greater concentrations than those other locations sites.
Analyzed occurrence of 31 PPCPs detected in Baiyangdian Lake, total concentrations at ten sites ranged from 622 to 2781 ng/L, with a median of 1421 ng/L. Although most individual compounds occurred at lesser concentrations, total concentrations of PPCPs exceeded 1 µg/L at eight locations. Mean concentrations of individual chemicals ranged from 0.79 to 329.18 ng/L. Concentrations of diethyltoluamide (329.18 ng/L), methylparaben (201.47 ng/L), florfenicol (196.74 ng/L), metformin (183.14 ng/L) were greatest among PPCPs. Frequencies of detection for the 31 PPCPs were 10%-100%, among which 24 chemicals were detected at all the ten locations. Concentrations ranged from 0.94 to 113.40 ng/L. These results are consistent with those of previous studies where 22 antibiotics were observed in waters of Baiyangdian Lake and tributaries [21]. In that previous study, antibiotics occurred widely in waters samples, with sulfamethoxazole occurring at the greatest concentrations, with a maximum concentration of 940 ng/L, which was much greater than that of 34.54 ng/L, observed in this study.
All 31 PPCPs were found at B10, and the total concentrations of combined PPCPs was 2780 ng/L. The most likely reason for this is that location B10 is located near the lakeshore, at a frequently visited, scenic spot, and near the estuary of Fuhe River that receives huge amount of wastewater from Baoding City. Results of previous studies indicated that sewage discharged from Baoding City with over one million residents is likely to be the main source of PPCPs to Baiyangdian Lake [47, 48]. In addition, relatively great concentrations of PPCPs were found at B9 (1856.9 ng/L) and B1 (1751.7 ng/L). As expected, least concentrations of PPCPs (622-1,190 ng/L) in water samples were observed in the middle of the lake (B2, B3, B4, B5), where there was little direct influence by human being activities. These studies demonstrated that human activities played a key role in distributions of PPCPs in Baiyangdian Lake.
Screening-level risk assessment
Chronic toxicity data for detected compounds were collected and PNEC values were calculated by use of a conservative AF (Table S2), excepted for valsartan, miconazole, triamterene, clopidol, nalidixic acid, triclabendazole, and cetirizine, since ecotoxicological data were not available. The 16 PPCPs in Tai Lake and 28 PPCPs in Baiyangdian Lake were ranked, in descending order, by RQ values (Fig. 3). In the two lakes, a total of 7 PPCPs yielded PI values greater than zero, including carbamazepine in both lakes (Table 1).
Table 1 Prioritization of target compounds based on PI values (PI>0).
Rank
|
Chemical
|
CAS
|
RQmax
|
F (%)
|
PI
|
Tai Lake
|
|
|
|
|
|
1
|
caffeine
|
58-08-2
|
10.3
|
100
|
10.3
|
2
|
carbamazepine
|
298-46-4
|
1.9
|
70
|
1.3
|
3
|
ibuprofen
|
15687-27-1
|
1.5
|
20
|
0.3
|
Baiyangdian Lake
|
|
|
|
|
|
1
|
sulfamethoxazole
|
723-46-6
|
1128.6
|
100
|
1128.6
|
2
|
carbamazepine epoxide
|
36507-30-9
|
25.6
|
100
|
25.6
|
3
|
carbamazepine
|
298-46-4
|
20.5
|
100
|
20.5
|
4
|
diethyltoluamide
|
134-62-3
|
10.5
|
100
|
10.5
|
5
|
triclosan
|
3380-34-5
|
1.2
|
10
|
0.12
|
Note: RQmax reference risk quotient based on the maximum measured concentration; F reference frequency of PNEC exceedance; PI reference prioritization index.
In Tai Lake, three pharmaceuticals, caffeine, carbamazepine, and ibuprofen, posed risk to aquatic organisms, with PI values of 10.3, 1.3, and 0.3 respectively. The maximum RQ value for caffeine was 10.3, and the frequency of PNEC exceedance was 100%, which would mean that a great or moderate environmental risk at all 10 locations was probable. RQ values of carbamazepine and ibuprofen ranged from 0.63 to 1.86, 0.25 to 1.48, and frequencies of exceeding the PNECs were 70% and 20%, respectively, which indicated environmental risks in Tai Lake. Frequencies of exceedances for the remaining 12 PPCPs were zero, while that of sildenafil and diethyltoluamide presented least risk (0.1<RQ<1), indicated potential risk to aquatic organisms and should be paid more attention in the future.
In Baiyangdian Lake, five compounds were determined to pose measurable environmental risks, sulfamethoxazole, carbamazepine epoxide, carbamazepine, diethyltoluamide, and triclosan with PI values of 3454, 25.6, 20.5, 10.5, and 1.1 respectively. Sulfamethoxazole exhibited greater risks to non-target organisms in Baiyangdian Lake for 100% of samples, with RQs ranged from 758 to 3,454, because of its toxic potency to Caenorhabditis elegans. Carbamazepine epoxide, carbamazepine, and diethyltoluamide represented great or moderate risks at all 10 sites, with RQs ranging from 3.2 to 25.6, 3.8 to 20.5, and 3.2 to 10.5, respectively. Triclosan exhibits a moderate risk (RQ=1.2) in only 10% of samples, with concentrations greater than PNECs observed only at B10. Amongst target compounds, 22 PPCPs presented RQ < 0.1 for all studied samples, indicating di minimis risks to aquatic ecosystems in Baiyangdian.
The PPCPs priority list partially overlaps with compounds prioritized in earlier studies. Eighteen PPCPs were selected and RQ values of sulfamethoxazole and carbamazepine were less than 0.01 in Baiyangdian Lake [22]. That result can be attributed to the relatively great values for PNECs of those two chemicals. In a case study performed in Europe, 42 compounds were prioritized and 5 of them can also be found in this priority list, i.e. caffeine, ibuprofen, triclosan, sulfamethoxazole, and carbamazepine [8].
Probabilistic analysis of risk
In the present study, based on prioritization indexes, caffeine, carbamazepine, diethyltoluamide, carbamazepine epoxide, and sulfamethoxazole were identified as posing great or moderate risk in Baiyangdian Lake or in Tai Lake. Therefore, they were assessed by a higher-tier assessment based on variability in exposure and ecotoxicity data. Toxicity data used were reported in Supporting Information (Table S3), and data sets were tested for log-normal distribution by use of the Shapiro-Wilk test (p<0.05) prior to application of parametric statistics (Table 2). Joint probability curves for each compound, excluding carbamazepine epoxide, for which too few toxicity data were available to provide sufficient meaningful resolution, were derived by integrating the distribution for surface water concentrations with chronic toxicity effects on varies species to indicate the probability of exceeding effects of differing magnitudes (Fig. 4). Three reference lines were used to categorize risk as de minimis, lesser, intermediate or greater [38, 39]. Each point on the curve represents both the probability that the chosen proportion of species will be affected and the frequency with which that magnitude of effect would be exceeded in surface waters.
Table 2 Parameters of joint probability curves (JPCs) for screened PPCPs
Chemical
|
N
|
Mean
(ng/L)
|
SD
|
CV
|
Shapiro-Wilk test for
log-normal distribution
|
exposure data set
|
|
|
|
|
|
Caffeine
|
10
|
12.51
|
7.08
|
0.57
|
0.140
|
carbamazepine
(Tai Lake)
|
10
|
1.18
|
0.35
|
0.30
|
0.471
|
carbamazepine
(Baiyangdian Lake)
|
10
|
13.26
|
4.93
|
0.37
|
0.056
|
diethyltoluamide
|
10
|
329.18
|
117.48
|
0.36
|
0.062
|
sulfamethoxazole
|
10
|
1.18
|
0.35
|
0.30
|
0.194
|
toxicity data set
|
|
|
|
|
|
Caffeine
|
17
|
6949090
|
23411509
|
3.37
|
0.355
|
carbamazepine
|
24
|
1051828
|
3858902
|
3.67
|
0.879
|
diethyltoluamide
|
7
|
14724975
|
19980915
|
136
|
0.066
|
sulfamethoxazole
|
9
|
68611
|
165164
|
2.41
|
0.272
|
Notes: N refers to Number of data; SD refers to standard deviation; CV refers to coefficient of variation.
Based on these results, the four PPCPs in the two lakes posed lesser to greater risks to aquatic organisms. Risk, based on chronic toxicity data, for sulfamethoxazole in Baiyangdian Lake was categorized as greater, with a maximum risk product of 16.62%. For caffeine in Tai Lake, diethyltoluamide and carbamazepine in Baiyangdian Lake, concentrations represent intermediate risk of chronic effects with maximum risk products of 7.25%, 6.68%, and 2.76%, respectively. Lesser risk of chronic effects on aquatic organisms was identified for carbamazepine in Tai Lake, with maximum risk products of 0.82%. Results from the estimated risk curves can also be used to describe the probability of exceeding various percentages of effects. The probability of exceeding 5% adverse effect depended on the most sensitive species, while the shape of the risk curve was related to ranges and variability of datasets (Fig. 5). For example, JPCs for sulfamethoxazole were classified as greater risk to more than 20% of species, but slightly above the reference line for lesser risk to 30% of species. This is because concentrations of sulfamethoxazole were only slightly greater than thresholds for adverse effects on the most sensitive species, and CVs for estimates of exposure were much less than those for relative potencies among species. In other words, ecological risk would not occur if the most sensitive species was not native species or not important for the local aquatic ecosystem. Therefore, PNEC derived by the most sensitive species and risk assessment according to RQ is likely to be over protective of aquatic ecosystems, and ecological risk assessment based on multiple species is necessary.
Uncertainty Analysis
Due to the limited measured surface water concentrations and the lack of data, for toxic potencies of some of the PPCPs to aquatic organisms, some uncertainty in conclusions reached was unavoidable. To more accurately describe exposure and ecological risks, measured concentrations of PPCPs at various temporal scales in waters are required. For seven chemicals, no conclusion can be drawn because ecotoxicological data were not available. However, environmental risks of some drugs are of concern, and due to their great frequency of detection, especially for valsartan in Tai Lake, clopidol and triclabendazole in Baiyangdian Lake. Furthermore, toxicity arising from complex mixtures of PPCPs, each of which occurred at small concentrations that would result in di minimis risks, could lead to additive or synergistic interactions, as demonstrated for similar acting compounds such as antibiotics [49]. This means that even though individual PPCPs are present in relatively small concentrations that do not elicit significant toxic effects, PPCPs mixtures can still exert considerable ecotoxicity. Further research on the risk of these detected compounds should be considered based on combined toxic. At another level, the risk that PPCPs might pose to aquatic species is not only directly related to toxicity of dissolved substances but also to possible bioaccumulation through the food web [50, 51]. For example, bioconcentration factors measured for ibuprofen in rainbow trout (Oncorhynchus mykiss) bile were 14,000~49,000 [52], also Coogan et al. [53] revealed accumulation of triclosan in filamentous algae species with the bioaccumulation factor ranged from 900~2,100, suggesting a high bioconcentration in aquatic organisms [54].
There were also limitations imposed by chiral chemicals that might exhibit significant differences in biodegradation and toxic potency among enantiomers [55]. The enantioselective biodegradation and ecotoxicity of chiral PPCPs tend to complicate their potential risk [56]. For example, when waters from several lakes and rivers in Switzerland were investigated, enantiomeric ratios (ER) of ibuprofen ranged from 0.7 to 4.2 [57]. There appears to be a trend toward lesser ERs (closer to racemic) during the warmer season, and greater ERs in winter, with concentration of S-ibuprofen higher than R-ibuprofen. Results of previous studies have shown that inhibition of prostaglandins by R-ibuprofen was 100 times than that of S-ibuprofen [58]. On the contrary, inhibition of cyclooxygenase by R-ibuprofen was 1.4 times less than that of S-ibuprofen [59]. In this study, enantioselectivity of the three chiral pharmaceuticals, caffeine, carbamazepine and ibuprofen that have potential risk in surface waters were not analyzed. Therefore, the risks of such chemicals might have been underestimated or overestimated, and this is likely to change drastically as new information becomes available. Further considerations on ecotoxicity effect of chiral pharmaceuticals in the aquatic environment are possibly needed, which could provide scientific basis and technical support to improve the accuracy of ecological risk assessment.