Disinfection tackling the COVID-19 pandemic causes disinfection by-products (DBPs) accumulation and threatens aquatic ecosystems

To fight against the coronavirus infectious disease-2019 (COVID-19), chlorine-based disinfectants are extensively or even over used for water, surface and personal care decontamination. The risks of disinfection by-products (DBPs) have been alerted to cause serious secondary pollution; however, there is still lack of evidence. This study collected 110 water samples from nine lakes and two rivers in Wuhan during the COVID-19 pandemic and comprehensively analyzed the occurrence of eighteen DBPs. Trihalomethanes, halonitromethanes, halogen acetonitriles and nitrosamines had a high detection frequency and were 0.99-14.26, ND-4.62, ND-1.09 and 0.0414-0.0861 μg/L, respectively, all lower than the maximum contamination level (MCL) suggested by China and USA. Haloacetic acids were detected in all lakes and Yangtze River and ranged from 33.8 to 856.1 μg/L, much higher than the MCL. Haloacetic acids and halogen acetonitriles accounted for 74.2-95.1% of the total cytotoxicity (0.38-3.62×10 5 ); halonitromethanes (94.0-98.7%) contributed to the majority of genotoxicity (0.52-5.17×10 4 ). Dichloroacetic acid exhibited significant ecological risks to green algae in two lakes and Yangtze River (risk quotient >10), and all the other DBPs showed negligible risks (risk quotient <0.01) to fish, daphnid or green algae. Correlation and redundancy analysis identified strong correlations between total organic carbon, conductivity, NH 3 -N, turbidity and DBPs. DBP composition and the fluorescence indices of dissolved organic matters together categorized all lakes into two types. Type-I lakes contained all DBP categories, driven by total organic carbon and secondarily formed by residual active chlorine with natural organic matters; Type-II lakes and Yangtze River only had high levels of haloacetic acids and small


Introduction
Coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has aroused over 160 million cases and 3.4 million deaths until 10 th May, 2021.Fecal-oral transmission route is reported by World Health Organization (WHO) for the spread of SARS-CoV-2 1 , and some evidences show the spillover of SARS-CoV-2 from wastewater 2 or solid wastes 3 .Disinfection is therefore used as the most effective strategy to prevent SARS-CoV-2 from spreading.Viable disinfectants for SARS-CoV-2 outlined by the United States Environmental Protection Agency (USEPA) include chlorine-based disinfectants, hydrogen peroxide, ethanol, octanoic acid, peroxyacetic acid and quaternary ammonium 4 .Among them, chlorinebased disinfectants are most used because of its potency and low price 5,6 .Particularly, Chinese government has expended a huge amount of chlorine-based disinfectants for both indoor and outdoor spaces, and at least 2000 tons of disinfectants are estimated to use in Wuhan alone in the initial stage of the COVID-19 pandemic 7 .These disinfectants may eventually adduct and accumulate in lakes and rivers through runoff and sewage, posing a significant threat to aquatic ecosystem and drinking water safety 8 .However, there is still lack of evidence on the hypothetical consequence from disinfectant overuse during the COVID-19 pandemic.
Chlorine-based disinfectants can directly destroy cell walls or oxidize proteins of organisms 8 .In addition, they can react with various kinds of organic matters, e.g., natural organic matters, effluent suspended solids, microorganisms, algal toxins, and anthropogenic contaminants, to form unintended disinfection by-products (DBPs) 9,10 .Up to now, more than 800 DBPs have been identified, including trihalomethanes (THMs), haloacetic acids (HAAs), halogen acetonitriles (HANs), halonitromethanes (HNMs), haloacetamides (HAcAms), haloketones (HKs) and nitrosamines (NAs).Since many DBPs are reported to possess cytotoxicity, mutagenicity, genotoxicity, and/or teratogenicity, increasing concerns are raised for public health 11 .Microorganisms, plants and animals in aquatic ecosystems are highly vulnerable to environmental insult from residual DBPs, which in turn affects human beings by biomagnification through the food chain 12 .Yadav et al. reported the effects of DBPs on microorganisms that both microbial diversity and community structure of aquatic ecosystem were altered after disinfection 13 .
THM exposure leads to developmental anomalies of zebrafish, and HAAs including tribromoacetic acid (TBA) and dichloroacetic acid (DCAA) are responsible for embryo malformations 14 .DBPs exposure occurs in three ways including inhalation, ingestion and dermal absorption, leading to high levels of skin permeable and volatile DBPs in the blood 15,16 .Several epidemiological studies have suggested that long-term exposure to DBPs is related to adverse human health outcomes including higher incidence of bladder and colorectal cancer 17,18 .Since THMs are soluble in fat and also highly volatile, they are found in various food such as ice cream, juices and soft drinks.In addition, THMs are reported to bioaccumulate in adipose tissue, kidneys, lungs and liver when quantity and route being exposed 19,20 .Previous studies have suggested that many DBPs are intermediate compounds that can transform into other end products by hydrolysis or chlorination 21 .For example, HANs are susceptible to further convert to haloacetic acids, especially trihaloacetonitriles 22 .Accordingly, many DBPs have been classified as potential carcinogenic compounds to humans by International Agency for Research on Cancer (IARC) 15 .
Several regulatory guidelines are established by authorities like WHO, USEPA and Bureau of China Standard for DBPs in drinking water.The maximum contamination level (MCL) is 80 μg/L for THMs and 60 μg/L for HAAs in USA 23 .In China, it is 60 μg/L for trichloromethane (TCM), 60 μg/L for monobromodichloromethane (BDCM), 100 μg/L for dibromochloromethane (DBCM), 100 μg/L for tribromethane (TBM), 50 μg/L for dichloroacetic acid (DCAA) and 100 μg/L for trichloroacetic acid (TCAA) 24 .These guidelines mainly focus on drinking water or swimming pool; however, limited studies address DBPs in natural waters, particularly after the intensive disinfection activities in the pandemics like COVID-19, when excessive DBPs in chlorinated sewage effluents and surrounding soils eventually extravasate into the receiving surface waters and contaminate the drinking water sources.As COVID-19 continues to spread across the world, the increasing consumption of chlorine-based disinfectants and production of DBPs might result in a global secondary disaster in aquatic ecosystems 7 .Therefore, it is of great significance to assess the occurrence, distribution and impacts of DBPs in aquatic ecosystems.Till now, only two studies reported the occurrence and distribution of DBPs in several rivers and lakes during the COVID-19 pandemic in China 25,26 and did not find a significant increase.There is still lack of a comprehensive survey targeting a broad range of water sources in megacities.This work studied the levels and compositions of DBPs in 110 water samples from nine lakes and two rivers in Wuhan in the initial stage of the COVID-2019, and for the first time found a significant accumulation and unneglected risks of HAAs in natural waters.We aimed to evaluate the levels of DBPs in lakes and rivers, assess their toxicities and ecological impacts, and explain the formation mechanisms of DBPs by exploring the organic matter compositions in these aquatic systems.Our findings highlight the secondary environmental risks after the intensive disinfection activities in the pandemics, and offer suggestions for disinfection strategies and DBPs risk management in natural waters.

Levels of key DBPs in lakes and rivers
Mean concentrations and compositions of key DBPs in 110 water samples are illustrated in Figure 2.Among them, HAAs (33.8-856.1 μg/L) were of the highest level in lakes, followed by THMs (0.99-14.3 μg/L), HNMs (ND-4.62 μg/L), HANs (ND-1.09μg/L) and NAs (0.0414-0.0861 μg/L).Only DCAA was detected among three HAAs and the concentration exceeded the MCL of the Standards for Drinking Water Quality of China (GB5749-2006) except for samples from MSH.The highest mean concentration of HAAs was observed in JYH (856.1 μg/L), followed by NH (330.1 μg/L) and HH (123.7 μg/L).
Other DBPs were all below the MCL.For river samples, only waters from Yangtze River (YR) had a wide variety of DBPs, and HAAs had the highest concentration (mean 487.1 μg/L), approximately 8.7 times higher than the MCL.In contrast, only limited DBPs were detected in HR, and none of them exceeded the MCL.

Toxicity and ecological risks of DBPs in lakes and rivers
The cytotoxicity indices and fractions of each DBPs in lakes and rivers of Wuhan are illustrated in Figure 4A and 4B, respectively.Total DBPs cytotoxicity ranged from 3.80×10 4 (HGH) to 3.62×10 5 (JYH) in lakes.HAAs and HANs accounted for 74.2-95.1% of total DBPs cytotoxicity, which was approximately 1-3 orders of magnitude higher than THMs, HNMs or NAs.Both HAAs and HANs exhibited the major cytotoxicity in the type-I lakes, whereas HAAs dominated the cytotoxicity in the type-II lakes.Only DBPs in the type-I lakes exhibited significant genotoxicities ranging from 5.24×10 3 to 5.17×10 4 (Figure 4C).
RQs represent the potential ecological risks of DBPs in aquatic ecosystem, and our data suggested neglectable risks of DBPs to fish and daphnid in lakes and rivers of Wuhan, as their RQs were less than 0.1 (Figure 5A and 5B).Nevertheless, DCAA exhibited significant ecological risks to green algae, which were adverse in HJH, JYH and YR (RQs>10) and moderate in other seven lakes (1.0<RQs<10) (Figure 5C).All the other DBPs showed insignificant risks (RQs<0.01) to green algae.
Typical EEM spectra of natural DOMs in lakes and rivers are shown in Figure 6A.The spectra exhibited some similarities in peak locations among DOMs from the nine lakes, such as region IV (microbial by-products) and region V (humic acids).The peaks of tryptophan-like proteins in region II were prominent, whereas the peaks of tyrosine-like proteins in region I were less noticeable.Figure 6B illustrated
RDA score plot between water physiochemical variables and DPBs (Figure 7B) illustrates that the first (RDA1) and second (RDA2) components explained 19.76% and 2.58% of the total variance, respectively.Among all the variables, conductivity (10.4%, p=0.002),TOC (7.1%, p=0.004), turbidity (3.3%, p=0.02) and NH3-N (3.0%, p=0.042) contributed to the major variance of DBP composition.Water samples from the type-I lakes, type-ⅠI lakes and rivers were clearly separated.DBPs in the type-I lakes were mainly linked with TOC, whereas conductivity had more critical roles in the type-II lakes and two rivers, segregated by NH3-N.As for fluorescence indices of DOMs, they together explained 20.88% of the total variance of DBP composition (Figure 7C).Peak T/C (4.1%, p=0.02) and Region IV (3.9%, p=0.01) were significant variable.

Discussion
To fight against COVID-19, chlorine-based disinfectants are extensively or even over used for water, surface and personal care decontamination, raising the concerns about the formation, release and accumulation of DBPs in surrounding environment 7 .
Disinfection with chlorine-based disinfectants can directly generate high levels of DBPs in sewage and sludge, and the residual active chlorine might further react with NOMs in natural waters or sediments, causing secondary contamination in lakes, rivers and even surrounding soils 8 .These DBPs hypothetically extravasate into the receiving rivers and lakes, posing a great risk to aquatic ecosystem and public health 7 .Till now, there is still lack of direct evidence, and our study is the first report unraveling the secondary pollution and ecological risks of DBPs in lakes and drinking water sources of Wuhan, resulted from disinfectant overuse during the COVID-19 pandemic.Our findings comprehensively visualized the occurrence and distribution of eighteen DBPs, assessed their toxicities and ecological risks, and discussed the formation mechanisms, showing great significance to evaluate the consequence of intensified disinfection on aquatic ecosystem.DBP levels are reported to vary greatly across water samples, attributing to the difference in disinfectant type, dosage, water physiochemical variables, and the presence of organic and halogenated molecules in water 27 .Comparing to other previously reported DPBs levels in wastewater effluents (13.5 to 234.2 μg/L) 28 , drinking water (86.0-122.2μg/L) 29 and swimming pools (85.2-224.0μg/L) 30 (Table 2), our results suggested a significant increase and accumulation of DBPs in surface waters of Wuhan after using superfluous disinfectants to decontaminate wastewater, surface and personal care in the COVID-19 pandemic.Particularly, HAAs were 33.8-856.1 μg/L in almost all lakes and Yangtze River, much higher than the MCL of the Standards for Drinking Water Quality of China (GB5749-2006).Previous studies found different levels of HAAs in drinking water and wastewater effluents, highlighting the necessity to prevent HAAs formation in drinking water disinfection 23,28 .Nevertheless, HAAs accumulation in natural surface water has not been reported, and the significant increasing HAAs in lakes and rivers of Wuhan in this study prove that intensive disinfection in megacities during the COVID-19 pandemic can cause remarkable DBP contamination in surface water and raises special concerns of disinfection strategy.DBP composition in surface waters of Wuhan was different from other studies.THMs were normally the most prevalent chlorinated DBPs 23 , whereas they were much lower in our study, only ranging from 0.99 to 14.26 μg/L and accounting for 0.52-21.11% of total DBPs.In contrast, the average levels of HAAs (33.8-856.1 μg/L) were much higher and dominated DBPs, consistent with Lee's work that HAAs (35.2-747.1 μg/L) was main DBPs in chlorinated-swimming pool water, accounting for 72.6% of total DBPs 31 .It might be explained by higher volatility and stability of THMs than HAAs in natural environment.
Unlike previous laboratory studies collecting samples promptly after the addition of chlorine-based disinfectants, it normally takes several hours to days for lakes and rivers to receive sewage containing DBPs from pipeline networks, which have many lifting and pumping operations from disinfection points to the final receiving water 16 .Highly volatile chemicals like THMs are therefore easier to evaporate into atmosphere and their concentrations then decline in natural water.Another possible explanation is the secondary formation of DBPs in the receiving water due to the excess active chlorine residual in sewage after intensive disinfection activities.Higher temperature, microbial activities and DOMs are reported to benefit HAA formation 23,32 .In this study, the lakes and rivers of Wuhan had poor water quality with high levels of COD and TOC.More precisely, COD in HH exceeded the standard of Class V according to the Environmental Quality Standards for Surface Water (GB 3838-2002) regulated by Chinese Ministry of Health 33 .Together with the relatively higher temperature ranging from 15 to 30 ℃ during the sampling period, more HAAs might be formed than THMs in natural surface waters.
Last but not the least, higher dosage and residuals of chlorine-based disinfectants can favor the formation of HAAs over THMs 34 .Accordingly, HAAs rather than other DBPs exhibited a striking high level in lakes and rivers of Wuhan, and they should be specifically monitored in other surface waters receiving disinfected sewage in cities with intensive disinfection activities to fight against COVID-19.
Besides total amount, each DBP category exhibited unique compositions comparing to previous studies.TCM and DBCM were the dominant THMs in lakes and rivers of Wuhan, and BDCM and TBM were not detected (Figure 3A).It was different from another study reporting high levels of TCM (20.9 μg/L, approximately 90% of THMs) but neglectable DBCM and TBM (<0.2 μg/L) in chlorinated swimming pool waters 31 .Among HAAs, only DCAA was detected in lakes and YR, while TCAA and MCAA were non-detectable.Other studies have inconsistent results that TCAA was the most common HAAs, followed by DCAA, DBCAA, MCAA and MBAA, in drinking water 23 , chlorinated wastewater effluents 28 , and chlorinated swimming pools 31 .The low levels of HANs in our study (ND-1.09μg/L) were consistent with other studies on drinking water (0.6-24 μg/L) 28 and wastewater effluents (2.9-48.6 μg/L) 31 .However, TCAN was the major HANs (Figure 3B), different from a study on the chlorinated swimming pool waters that DCAN (74.7%),BCAN (15.8%) and DBCN (9.5%) were predominant HANs and TCAN was below the limit of detection 31 .Only TCNM was detected among all HNMs in lakes (Figure 3C) and similar with the concentrations in the chlorinated drinking water (0.16-1.5 μg/L) 28 , meeting well with previous studies that the TCNM (>94%) were the major HNMs in chlorinated water 35,36 .
The dominance of TCAN and TCNM might be explained by slow decomposition of TCAN through hydrolysis and natural transformation from HCNM to TCM 31 .Our results suggested that DBPs in natural surface waters receiving sewage containing DBPs from intensive disinfection during the COVID-19 pandemic had unique signature and compositions.
We only found ng/L level of NAs in lakes and rivers of Wuhan (Figure 3D), consistent with previous studies on drinking waters and wastewater effluents 28,37 .Particularly, NDMA and NDEA were predominant NAs (Figure 3D), similar as a study on river waters that NDMA (0.0015-0.017 μg/L) and NDEA (0.0014-0.0095 μg/L) were major NAs 38 .Wang et al. reported that wastewater effluents from upstream cities might be a crucial source of NAs 35 , and the tap water in megacities like Shanghai are heavily polluted by NAs compared to other areas along Yangtze River.As NAs are formed by the reactions between monochloramine and organic amine precursors 37 or nitrite chlorination in the presence of nitrosamine precursor 17 , they are not correlated with the intensively used chlorine-based disinfectants during the COVID-19 pandemic in Wuhan.Accordingly, all NAs had similar concentrations and compositions across all lakes and rivers in this study, documenting a DBP background from other disinfection activities.
According to the occurrence and composition of DBPs, two types of lakes were classified in Wuhan (Figure 3).Type-I lakes (HJH, NH, LZH, MSH and HGH) had high levels of TCM, DBCM, TCAN and TCNM (Figure 3), and they are mainly located in urban areas with higher population and disinfection density.The intensive disinfection activities in these regions left excessive disinfectants on roads and surrounding soils, eventually extravasating into the receiving lakes through rain pipes after rain off 25 .These residual disinfectants might react with NOMs in lakes and formed secondary DBPs.It was supported by the higher explanation of TOC in DBPs composition in RDA score plots (Figure 7B) and positive correlations between three kinds of DBPs (HANs, HNMs, THMs) and TOC in PCA score plots (Figure 7A).Further evidence comes from the structural characteristics of DOMs across lakes and rivers using EEM spectra, which provide information of chromophores and fluorophores 39 .DOMs in type-I lakes exhibited low HIXs (<4.0), high BIXs (0.90 to 1.12), moderate FIs (1.72-2.06),and high Peak T/C (1.53-2.71),suggesting weak humification, autochthonous and microbial-driven features 40 .
Previous studies revealed that THMs were strongly correlated with humic-like components (region V) 41 .Despite the similar distribution of region V between type-I (11.8-20.9%)and type-II (11.7-22.3%)lakes, TOC in type-I lakes (5.10-16.0mg/L) were much higher than type-II lakes (4.95-8.00mg/L), indicating higher levels of precursors for THMs.
Similarly, the precursors for HANs strongly correlated with protein-like component (region I and II) 42 , which were also remarkably higher in type-I lakes than in type-II lakes.
In contrast, type-II lakes (DH, HH, JYH and TXH) only had a low level of TCM and DBCM (Figure 3), and they are in suburban regions with lower population density, large area and mainly rain-sewage mixed flow pipelines.There were relatively weaker disinfection activities, and the low levels of residual disinfectants in runoff might directly react with DOMs in sewage in the pipelines and only primary DBPs were formed.Therefore, type-II lakes exhibited different DBPs composition, predominantly explained by conductivity, turbidity and NH3-N (main contaminants in sewage) in RDA results (Figure 7B).Similarly, there are many discharge points of rain-sewage mixed or diversion pipelines along Yangtze River (YR) 43 , which received effluents from both type-I and type-II lakes and had high levels of HAAs and small amounts of THMs (Figure 2A).Instead, Han River (HR) is the main drinking water sources for Wuhan and nearly all the outfalls have been banned since 2002 44 .Therefore, Han River did not receive sewage containing high levels of DBPs from intensive disinfection activities during the pandemic and there were only NAs irrelevant to the chlorine-based disinfectants in HR (Figure 2A).Our findings proved that the high levels of DBPs in Yangtze River were mainly derived from discharge from pipelines and surrounding lakes, and the residual DBPs might spread and threaten the downstream aquatic ecosystems of Yangtze River.
DBPs in lakes and rivers of Wuhan exhibited considerable cytotoxicities, significant genotoxicities and non-neglectable ecological risks (Figure 4).Cytotoxicities were mainly derived from HAAs and HANs, which were approximately 1-3 orders of magnitude higher than THMs, HNMs or NAs (Figure 4A).More precisely, HAAs exhibited the highest concentration (Figure 2A) and strong cytotoxicity 45 , contributing to the majority of cytotoxicities (15.7-98.6%) in lakes.As for HR containing neglectable HAAs and HANs, NAs had the highest cytotoxicity (Figure 4B).Although THMs, HAAs and NAs were main DBPs, neither of them exhibited genotoxicity or lack information (Table S2) 46,47 .Instead, HNMs contributed to the major genotoxicities (94.0-98.7%),approximately 16.0-49.4times higher than HANs.All DBPs exhibited limited or nonnegligible ecological risks to fish, daphnid and green algae in lakes and rivers of Wuhan (Figure 5A and 5B), except for DCAA which imposed significant risks to green algae.All these results were consistent with some previous reports about the cytotoxicities of HANs in wastewater effluents 28,45 , major cytotoxicities of HANs and NAs in chlorinated effluents 28 , significant cytotoxicities of HANs, haloacetaldehydes and HAAs in chlorinated saline groundwater 48 , stronger genotoxicities of nitrogen-containing HANs and HNMs 49 in disinfected recreational pools 50 , and remarkable ecological risks of HAAs for green algae in chlorinated wastewater effluents 28 .Comparing to other studies on wastewater effluents (cytotoxicities of 10 5 -10 7 ) 28 and drinking water (cytotoxicities of 10 2 -10 5 and genotoxicities of 10 2 -10 3 ) 51 , the risks of DBPs in lakes and rivers of Wuhan were nonnegligible.Regarding the facts that some lakes and HR are water sources of Wuhan and residual DBPs can transport to downstream of Yangtze River, DBPs derived from intensive disinfection activities in Wuhan had significant ecological risks and could threaten drinking water safety.It is therefore of urgency to reconsider the disinfection strategies for the prevention and control of emerging infectious diseases during the pandemics, and appropriate disinfection management including disinfectant type and dosage is critical to mitigate secondary environmental pollution in the post-COVID-19 era.

Conclusion
Our study first reported the increasing levels of DBPs, particularly HAAs, in lakes and rivers of Wuhan during the initial stage of the COVID-19 pandemic.They were mainly derived from the intensive disinfection activities in urban areas, and imposed nonnegligible toxicities and significant ecological risks to surrounding receptors or even downstream of Yangtze River.Additionally, the composition of DBPs varied across surface waters and were remarkably different from previous studies on chlorinated drinking water or wastewater, explained by the long-distance transport in sewage pipelines and secondary formation with NOMs in surface waters.Our findings provide the first evidence that the excessive disinfection activities in megacities during the pandemics do dispense DBPs in the surrounding receiving waters and threaten aquatic ecosystems, highlighting the importance for appropriate disinfection strategy and management for the prevention and control of emerging infectious diseases during the pandemics.

Study area and sampling
The study area is in Wuhan, Hubei Province (China), which is about 8569 km 2 with estimated population of 11.2 million.As illustrated in Figure 1 and Table S1, two types of surface water including nine lakes (Huangjiahu Lake, HJH; Nanhu Lake, NH; Liangzihu Lake, LZH; Moshuihu Lake, MSH; Tangxun Lake, TXH; Houhu Lake, HH; Donghu Lake, DH; Houguanhu Lake, HGH; Jinyinhu Lake, JYH) and two rivers (Yangtze River, YR; Hanjiang River, HR) were collected in May and June 2020, which was the initial stage of the COVID-2019 pandemic.The distances between sampling sites were about 0.3-4.0km and 2.0-9.0 km in lakes and rivers, respectively.All sampling sites in two rivers are drinking water sources of Wuhan.Water samples were transported to the laboratory within 4 h, stored at 4 ℃ and filtered through a 0.2 μm membrane filter prior to instrumental analysis.

Chemical analysis of water physiochemical variables
Total organic carbon (TOC) was analyzed on a Shimadzu TOC-Analyzer (TOC-L CPN, Japan).Chemical oxygen demand (COD), total phosphorus (TP), total nitrogen (TN) and ammonia nitrogen (NH3-N) were analyzed according to the standard methods 26 .The pH and conductivity were measured with a WTW Multi 3410 equipped with a SenTix 940 pH electrode (WTW, Germany) and an HI 9835 conductivity meter (Hanna, Italy), respectively.

Excitation-emission matrix fluorescence spectroscopic analysis
Besides anthropogenic sources, dissolved organic matters (DOMs) in natural waters are well-recognized as the crucial precursors of DBP.As a complex mixture of organic matters with diverse chemical structures and reactivities, DOMs were characterized by excitationemission matrix (EEM) fluorescence spectroscopy with high sensitivity and selectivity 52 , using a Hitachi spectrofluorometer (F-7000 FL, Japan) at room temperature (20-23 ℃)    with a 1-cm quartz cuvette.The test conditions were as follows: excitation wavelength range: 200-400 nm (step 5 nm), emission wavelength range: 250-500 nm (step 5 nm), scan speed: 2400 nm/min.The spectrofluorometer was auto-zeroed prior to analysis and the fluorescence intensities were normalized by the Raman peak of Milli-Q water at λex=348 nm.
Fluorescence regional integration (FRI) and several fluorescence indices were used to delineate EEM data.EEM peaks were first divided into five regions including aromatic proteins such as tyrosine and tryptophan (regions I and II), fulvic acids (region III), soluble microbial by-products (region IV) and humic acids (region V).The distribution of five regions can be calculated according to a previous study 53

DBPs analysis
Five categories of DBPs including four THMs, three HAAs, one HAN, one HNM and nine NAs were quantitatively analyzed.THMs measurement was conducted on a gas chromatograph-mass spectrometer (GC-MS) (Agilent 7890B-5977C, USA) equipped with a purge and trap autosampler and a DB-624 column (60 m×0.25 mm×1.4 μm).For HAAs, samples were filtered with a Ba/Ag/H column and 0.2 μm microporous membrane, and then analyzed on an ion chromatograph (IC) (Thermo Integrion HPIC, USA).HANs and HNMs analyses were performed on a GC equipped with an electron capture detector (GC-ECD) (Agilent 7890A) and a DB-17MS capillary column (30 m×0.25 mm×0.25 μm).
Solid phase extraction (SPE) and liquid chromatograph-mass spectrometer (LC-MS) (Shim-pack LCMS-8) were used for the measurement of NAs.

Toxicity and ecological risk assessment of DBPs
Both cytotoxicity and genotoxicity are widely used to assess the overall toxicity of DBPs in drinking water 28 .The unit cytotoxicity and genotoxicity index values were defined as the reciprocal of the median LC50 and median CHO SCGE genotoxic potency for each DBP, respectively 46 .Cytotoxicity and genotoxicity indices of each DBPs were calculated by multiplying unit toxicity indices and DBPs concentrations, as listed in Table S2 46,47 .
In addition, risk quotient (RQ) was used to assess the potential ecological risks of DBPs in aquatic ecosystems.Three different model organisms including fish, daphnid and green algae were evaluated at the acute level of DBPs using LC50 or EC50.RQs of each DBP for a certain taxonomic group were calculated by the following equation (4) 28,54 .S3). RQ<0.1 is regarded as no adverse effect; 0.1<RQ<1 represents low risk; 1<RQ<10 indicates moderate risk; RQ>10 is considered adverse risk to aquatic organisms 54,55 .

Statistical data analysis
a similar distribution of FRI among all lakes and rivers.Region II (22.7-29.2%)was the highest for all samples, followed by region III (19.3-27.1%),region IV (12.9-23.1%),region V (11.7-22.3%)and region I (10.7-20.4%).These indicated the presence of many tryptophan-like proteins and fulvic acids in water resources of Wuhan.More precisely, HIX values (1.83-3.31)were less than 4.0 (Figure 6B), suggesting a weak humification degree of DOMs.BIX values ranging from 0.90 to 1.65 suggested the autochthonous production of most natural DOMs.In most lakes, FI values ranged from 1.92 to 2.15 and NOMs were freshly produced from biological activities, whereas the lower FI values in LZH (1.73), YR (1.83) and HR (1.72) documented their formation from both microbial and terrestrial activities.Peak T/C values in YR (1.21) and HR (1.41) were much lower than those in lakes (1.54-3.39),indicating poorer biodegradability of natural DOMs in rivers.
. The fluorescence indices of humification index (HIX), autochthonous biological index (BIX), fluorescence index (FI) and peak T/C (the intensity ratio of peak T to peak C) were then calculated.HIX and BIX reflects the humification degree and microbial aspect of DOM properties, respectively.FI indicates the possibility of microbial production, and peak T/C indicates the biodegradability of DOMs 40 .They were calculated by the following Equations (1) to (3).
is the concentration of DBPs in natural waters, and LC50 and EC50 of each DBPs were derived from ECOSAR Program developed by USEPA (Table Differences in DBP concentrations between sites, lakes and rivers were examined using one-way analysis of variance (ANOVA) at p=0.05 level.Non-parametric Spearman's correlation analysis between DBPs, water physiochemical variables, and fluorescence indices were performed by SPSS 24.0.Redundancy analysis (RDA) was used to explore the relationships between DBPs, water physiochemical variables, and fluorescence indices using Canoco 5.0.Contributions Y.L., J.Q. and D.Z. designed the whole work.X.Z., L.W. H.L., W.L., C.Y. and X.W. collected water samples.S.C. did the chemical analysis.X.Z.coordinated data analyses and wrote the first draft.Y.X., X.H. and D.Z. edited the manuscript.

Figure 1 .
Figure 1.Geographic location of 110 sampling sites in nine lakes and two rivers of Wuhan, China.

Figure 2 .
Figure 2. Total amount (A) and composition (B) of key DBPs in lakes and rivers of Wuhan in the initial stage of COVID-19.

Figure 3 .
Figure 3.The composition of individual THMs (A), HANs (B), HNMs (C), and NAs (D) in lakes and rivers of Wuhan in the initial stage of COVID-19.

Figure 4 .
Figure 4. Cytotoxicity indices (A) and fraction (B) of DPBs in in nine lakes and two rivers of Wuhan.Genotoxicity indices (C) and fraction (D) of DPBs in in nine lakes and two rivers of Wuhan.

Figure 5 .
Figure 5. Ecological risk assessment by RQ values of DBPs in rivers and lakes of Wuhan.(A) Fish; (B) daphnid; (C) green algae.

Figure 6 .
Figure 6.(A) Typical excitation-emission matrix (EEM) of dissolved organic matters (DOMs) in lakes and rivers of Wuhan.(B) Composition of fluorescence regional integration (FRI) in different lakes and rivers.(C) Fluorescence indices (FI) of DOMs in different lakes and rivers.

Figure 3 The
Figure 3

Tables 594 Table 1 .
Water physiochemical variables in lakes and rivers of Wuhan.