Methane Levels of a River Network in Wuxi City, China and Response to Water Governance

: The majority of rivers are a CH 4 source that accounts for an important proportion of annual global emissions. However, CH 4 evasion from urban river networks has received disproportionately less attention than their contribution. The e ﬀ ect of water governance on water quality and CH 4 emission in urban areas remains unclear. Water quality, CH 4 concentrations, and ﬂuxes from a river network in Binhu District, Wuxi City, and their response to water governance were analyzed in this study. CH 4 concentrations in the investigated rivers ranged from 0.05 µ mol L − 1 to 16.37 µ mol L − 1 (2.47 ± 4.5 µ mol L − 1 , medium 0.23 µ mol L − 1 ), and CH 4 di ﬀ usive ﬂuxes were 75.55 ± 171.78 µ mol m − 2 h − 1 with a medium of 6.50 µ mol m − 2 h − 1 . CH 4 concentration showed a signiﬁcant correlation with water quality parameters, especially for NH 3 –N (r = 0.84, p < 0.001). Signiﬁcant di ﬀ erences in water quality and CH 4 levels were found between sites that had conducted water management and those that continued to exhibit poor water quality. Our analysis showed that rivers under water governance have a positive tendency toward water ecological restoration, and a signiﬁcant decrease in CH 4 e ﬄ ux to the air can be achieved after extensive and intensiﬁed water governance.


Introduction
CH 4 is a potent greenhouse gas, and the global warming potential per CH 4 molecule is approximately 30 times greater than that of CO 2 at a 100-year time scale. The increase in atmospheric CH 4 concentrations has contributed~23% to the additional radiative forcing accumulated in the lower atmosphere since 1750 [1,2]. The recent global estimate of CH 4 from freshwaters was 103 Tg CH 4 year −1 , which can substantially affect the estimate of global land greenhouse gas sink [3]. However, CH 4 dynamic in streams and rivers has received remarkably less attention than those in lakes, reservoirs, and wetlands [4,5]. The global-scale estimate revealed an annual emission from streams and rivers of 26.8 Tg CH 4 , which is equivalent to~15-40% of wetland and lake effluxes, respectively [6]. Among the river CH 4 sources, CH 4 evasion from urban river networks has received disproportionately less attention compared with their contribution. Carbon evasion from the Shanghai river network can offset up to 40% of regional terrestrial net ecosystem production and 10% of net carbon uptake in the river-dominated East China Sea [7]. Both studies of rivers in Shanghai and Chongqing found that urban and suburban areas contribute higher river CH 4 concentration and flux than rural locations due to the worse water quality of the former [7,8]. Therefore, CH 4 evasion from urban river networks requires additional attention as the urbanization process and proportion of urban land use increase globally.
However, despite the inevitable increase of sewage during urbanization, urban water governance has concurrently taken effect not only in Western countries but also in the developing world.

Sampling and Measurements
Eighteen rivers located in the Binhu District of Wuxi City were investigated in this study (Table  1). According to river length, 2 to 4 sampling sites per river were chosen to reach 45 sampling sites. CH4 concentrations in the surface water were investigated in 21 sampling sites. The water quality of all the 45 sampling sites was measured.
Sampling was conducted in September 2019. To measure CH4 concentration in the surface water, triplicate river waters at the surface depth of 0-10 cm were collected in glass vials pre-added with 2 g of sodium chloride, and the vials were immediately sealed with butyl stoppers. Sediment samples were collected using a sediment sampler (Ekman-Birge type). To measure CH4 concentration in the sediment, after retrieval, triplicate surface sediments (5 mL) were collected using a 10 mL cutoff plastic syringe and extruded into 50 mL glass vials containing 5 mL of 4% NaOH [12]. The remaining sediment samples were transferred immediately into a plastic bag and refrigerated for further analysis. Vials with water and sediment were refrigerated at 4 °C, and CH4 concentration was measured within 24 h. Water temperature, dissolved oxygen (DO) concentration, pH, and oxidationreduction potential (ORP) were measured using a multiparameter water quality probe (YSI, Yellow Springs, OH, USA), and the transparency (Secchi depth, SD) was obtained via in situ measurements with a Secchi disk. Surface water (1 L) was collected, stored in the dark, and frozen at −20 °C for further analysis.

Sampling and Measurements
Eighteen rivers located in the Binhu District of Wuxi City were investigated in this study (Table 1). According to river length, 2 to 4 sampling sites per river were chosen to reach 45 sampling sites. CH 4 concentrations in the surface water were investigated in 21 sampling sites. The water quality of all the 45 sampling sites was measured.
Sampling was conducted in September 2019. To measure CH 4 concentration in the surface water, triplicate river waters at the surface depth of 0-10 cm were collected in glass vials pre-added with 2 g of sodium chloride, and the vials were immediately sealed with butyl stoppers. Sediment samples were collected using a sediment sampler (Ekman-Birge type). To measure CH 4 concentration in the sediment, after retrieval, triplicate surface sediments (5 mL) were collected using a 10 mL cutoff plastic syringe and extruded into 50 mL glass vials containing 5 mL of 4% NaOH [12]. The remaining sediment samples were transferred immediately into a plastic bag and refrigerated for further analysis. Vials with water and sediment were refrigerated at 4 • C, and CH 4 concentration was measured within 24 h. Water temperature, dissolved oxygen (DO) concentration, pH, and oxidation-reduction potential (ORP) were measured using a multiparameter water quality probe (YSI, Yellow Springs, OH, USA), and the transparency (Secchi depth, SD) was obtained via in situ measurements with a Secchi disk. Surface water (1 L) was collected, stored in the dark, and frozen at −20 • C for further analysis. CH 4 concentrations in the headspace were determined using gas chromatography (7890B with a Porapak Q column and a flame ionization detector, Agilent Technologies, Lexington, MA, USA) via manual injection of 5 mL headspace gas sucking from the vials with a 5 mL syringe. CH 4 concentrations in the river water and sediments were calculated using the headspace equilibration method according to Fick's law [13]. Please refer to paper [14] for the detailed steps. Note: *, river width; & , maximum water depth of the sampling site during sampling. $ , group A includes sites with TLI ( ) < 60, group B represents 60 < TLI ( ) < 70, and group C is TLI ( ) > 70. # indicates whether water governance was conducted, and sampling sites were separated into "human" (H) and "natural" groups. On the basis of TLI ( ) and vegetation status, the "natural" group was further divided into "natural good" (NG) and "natural bad" (NB) groups.
Collected water samples were measured for water quality parameters, including total nitrogen (TN), total phosphorus (TP), chemical oxygen demand (COD Mn ), nitrate (NO 3 -N), ammonia nitrogen (NH 3 -N), phosphate (PO 4 -P), and chlorophyll a (Chla). TN and TP were photometrically determined using a UV-VIS spectrophotometer (UV3600, Shimadzu, Japan) [15]. Water was filtered through 0.7 µm membranes (GF/F, Whatman, Maidstone, UK), and Chla was extracted from the membrane with 90% acetone and measured via spectrophotometric analysis [16]. Concentrations of NO 3 -N, NH 3 -N, and PO 4 -P in filtered water were measured with a continuous flow analyzer (San Plus, SKALAR, Breda, Netherlands). The dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) contents were measured with a total organic carbon analyzer (TOC-V CPN, Shimadzu, Kyoto, Japan). COD Mn was determined via UV-VIS spectrometry [17]. Sediment water content was measured by drying the wet sediment to a constant weight at 105 • C. Loss on ignition (LOI) was obtained by igniting the dried sediment at 550 • C for 4 h [18].

Air-Water CH 4 Diffusive Flux
The diffusive flux of CH 4 in the air-water interface is expressed as follows: where F is the diffusive flux, k CH4 is the piston velocity (m d −1 ), C w is the CH 4 concentration measured in the water (µmol L −1 ), and C eq is the CH 4 concentration in the water at equilibrium with the atmosphere. The following gas transfer velocity of CH 4 (k CH4 ) was estimated using normalized k that corresponds to k CH4 at 20 • C (k 600 ), in situ water temperature, and Schmidt number (Sc): where n is the proportionality coefficient determined by the wind velocity. Here, n was set to 0.5 to represent a moderately turbulent surface in this study [7,19]. Sc and k 600 were determined using the measured in situ water temperature and mean wind speed of the study month (September 2019) collected from the website of the Chinese Meteorological Bureau [20,21].

Trophic Level Index (TLI)
As recommended by the China National Environmental Monitoring Center, a trophic level index (TLI) was used to assess the water quality in this study. TLI is a summation of the TLI (Chla) and the weighted TLI of the other parameters, which are based on correlations with Chla [22]. The TLI can be expressed as follows: where TLI ( ) is the comprehensive TLI, TLI (j) is the TLI of parameter j, m is the number of parameters used, and W j is the weight of the parameter TLI (j), which is determined using the correlation of parameter j with Chla. Wj is calculated as follows: where r 1j represents the correlation of parameter j with Chla. In this study, TN, TP, SD, and COD Mn , along with Chla, were used to assess the water quality, and their correlation coefficients with Chla were set to 0.82, 0.84, −0.83, and 0.83, respectively, in accordance with an investigation on 26 Chinese lakes [23]. The TLI of each parameter can be calculated as follows: TLI (TP) = 10 × (9.436 + 1.624 lnTP), where the unit of Chla is µg L −1 ; the units of TN, TP, and COD Mn are mg L −1 ; and SD represents the Secchi depth (unit: m).

Statistical Analysis
Water quality variables and CH 4 concentrations of the surface water were tested for normality distribution using the Shapiro-Wilk test. The majority of the variables with non-normal distributions were log-transformed, except COD Mn , which conducted reciprocal transformation, to approach normality. Stepwise multiple regressions were applied to identify which variables explained the variance of CH 4 concentration, and p < 0.05 determined the significance probability of each control variable. R 2 and F-tests were applied to evaluate the regression equation. The above analyses were performed using IBM SPSS 20. CH 4 concentrations of the surface water in the sampling sites were calculated according to the regression function. In terms of exposure to management activities, 45 sampling sites were separated into "human" (H) and "natural" groups. Based on the TLI and the vegetation status during sampling, the sites of the natural group were further divided into "natural good" (NG) and "natural bad" (NB) groups (Table 1). CH 4 and water quality variables were analyzed via principal component analysis (PCA) using R packages FactoMineR [24] and factoextra [25,26]. One-way analysis of variance (ANOVA) and a post hoc "Bonferroni" method were adopted to test the significant differences of CH 4 and water quality variables among the three groups (p < 0.05) with IBM SPSS 20.

CH 4 Concentrations and Diffusive Fluxes
CH 4 concentrations in the investigated rivers ranged from 0.05 µmol L −1 to 16.37 µmol L −1 (2.47 ± 4.5 µmol L −1 , medium 0.23 µmol L −1 ) based on 21 measured data, and according to the measured in situ water temperature and mean wind speed from the meteorological website (~2.5 m s −1 ), the calculated CH 4 diffusive fluxes were 75.55 ± 171.78 µmol m −2 h −1 with a medium of 6.50 µmol m −2 h −1 ( Table 2). Great values of CH 4 concentrations were observed at sites 9 and 27, which were the only two sites with CH 4 concentrations higher than 10 µmol L −1 . Four sites, namely, 18, 21, 24, and 26, demonstrated CH 4 concentrations higher than 1 µmol L −1 . These results indicated high CH 4 concentration in Xianjingbang and Lucunhe Rivers and in the downstream of Chendahe. The surface water DOC content of the 21 sampling sites ranged from 1.54 mg L −1 at site 18 to 30.09 mg L −1 at site 25, and no significant difference was found between the sites with water treatment and those under natural status (p > 0.05). The CH 4 content in the surface sediment ranged from 173.37 µmol L −1 to 1028.89 µmol L −1 (430.73 ± 234.12 µmol L −1 , medium 423.49 µmol L −1 ) and had no significant relationship with CH 4 in the water and water quality variables (p > 0.05). Compared with the surface water CH 4 , the sediment CH 4 was significantly higher at site 18 (1028.89 µmol L −1 ) but lower at site 21 (173.37 µmol L −1 ) ( Table 2). The sediment water content and LOI could not adequately explain the CH 4 level in the sediment (p > 0.05). For example, site 8 had a high level of water content and LOI, but its sediment CH 4 was only in the middle level (448.86 µmol L −1 ). However, site 17 showed an extremely low sediment CH 4 (18.01 µmol L −1 ) compared with the other sites, and this result is consistent with its low sediment water content and LOI. Table 2. Mean value ± standard deviations of CH 4 concentrations in the surface water (CH 4 -wat) and sediment (CH 4 -sed); CH 4 diffusive fluxes from water to the atmosphere (CH 4 -flux), as well as the in situ water temperature (WT), dissolved organic carbon (DOC) concentrations, and the sediment water content; and loss on ignition (LOI) of 21 sampling sites. Notes: Sites with shadow represent sites with water governance and nd denotes no data.

Water Quality
The TLI ranged from 52 to 84, with high values representing high eutrophication levels ( Figure 1). The study sites were divided into the following groups: Groups A, B, and C included sites with TLI < 60, 60 < TLI < 70, and TLI > 70, respectively (Table 1, Group 1). Groups A, B, and C had 19, 20, and 6 sampling sites, respectively. Several sites in group A were macrophyte-dominated, and macrophytes were planted artificially in sites 13 and 18 or grew naturally in site 43 ( Figure 1). The sites in group A obtained TN and TP ranges of 1.2-4.33 and 0.066-0.32 mg L −1 , respectively. Floating algae were observed on the water surface during sampling in the majority of the sites in group B and in all the sites in group C (Figure 1). TN and TP ranges in the group B sites were 1.91-7.54 and 0.15-0.44 mg L −1 , respectively. Six sites (4, 8, 9, 24, 26, and 27) in group C showed a significantly high nutrition level, with TN and TP reaching 26 mg L −1 at site 26 and 2.4 mg L −1 at site 4, respectively. These findings indicate extremely poor water quality in Dongxinhe, Xianjingbang, and Lucunhe Rivers. Moreover, these six sites had significantly higher NH 3 -N and NO 3 -N contents and lower DO than the other sites ( Figure 2) but with no water quality control, except site 8. By contrast, rivers located in the south of Lake Lihu were relatively clear, although under natural state, especially for Miaoqiaobang, Renzigang, and Shanxihe Rivers (Figure 1). Several sites with good water quality were located in the lakeshore and received water from Lake Lihu, such as 39, 40, 41, 13, and 20. This result indirectly reflects the good water quality in Lake Lihu.

Water Governance, CH 4 , and Water Quality
CH 4 concentration in the surface water of the 21 sampling sites had a significant correlation with DO, ORP, N and P contents, and COD Mn (p < 0.05), with the most significant correlation with NH 3 -N among other variables (Figure 4). The significant correlation of CH 4 concentration with the TLI (r = 0.63, p < 0.005) was not as strong as that of CH 4 with NH 3 -N (r = 0.84, p < 0.001). Apart from water quality, CH 4 concentration was also correlated with water depth (r = −0.49, p < 0.05). However, according to the sampling record, extremely low water depths at sites 26 and 27 were caused by pumping; if these two sites were excluded, then CH 4 concentration was not correlated with water depth (p = 0.30). The stepwise regression analysis determined the most robust result of CH 4 using the variables NH 3 -N and ORP (Function 4). CH 4 concentration in the surface water of the sampling sites was reconstructed using Function 10 as follows: Water 2020, 12, x FOR PEER REVIEW 9 of 15

Water Governance, CH4, and Water Quality
CH4 concentration in the surface water of the 21 sampling sites had a significant correlation with DO, ORP, N and P contents, and CODMn (p < 0.05), with the most significant correlation with NH3-N among other variables (Figure 4). The significant correlation of CH4 concentration with the TLI (r = 0.63, p < 0.005) was not as strong as that of CH4 with NH3-N (r = 0.84, p < 0.001). Apart from water quality, CH4 concentration was also correlated with water depth (r = −0.49, p < 0.05). However, according to the sampling record, extremely low water depths at sites 26 and 27 were caused by pumping; if these two sites were excluded, then CH4 concentration was not correlated with water depth (p = 0.30). The stepwise regression analysis determined the most robust result of CH4 using the variables NH3-N and ORP (Function 4). CH4 concentration in the surface water of the sampling sites was reconstructed using Function 10 as follows: Log CH4 = 0.97 × log NH3-N − 0.005 × ORP + 0.445 (adj R 2 = 0.67, p < 0.001) (10) Among the 21 sites for CH4 sampling, 9 sites conducted water governance (shaded sites in Table  2). The management practices were mainly for contamination control, such as the development of a sewage water treatment system and the separation of rainwater and sewage systems, and aquatic ecological rehabilitation, such as dredging, aeration, and ecological floating bed. We confirmed via observation and investigation that sites 5, 14, 17, and 21 had conducted dredging; sites 17 and 18 were performing aeration when sampling; and sites 19 and 21 were equipped with aeration devices but were not functioning. Finally, 20 sampling sites that conducted water governance were selected and separated into group H (Table 1). In this study, the sampling sites with water governance, regardless of how many measures were used, were all placed in group H. The 25 remaining sites were separated into groups NG and NB according to the TLI ( Table 1). The sites in the NG group obtained a maximum TLI of 60 and were typically macrophyte-dominated, and the TLI of the NB group was higher than 60, with floating algae present in most sites. Among the 21 sites for CH 4 sampling, 9 sites conducted water governance (shaded sites in Table 2). The management practices were mainly for contamination control, such as the development of a sewage water treatment system and the separation of rainwater and sewage systems, and aquatic ecological rehabilitation, such as dredging, aeration, and ecological floating bed. We confirmed via observation and investigation that sites 5, 14, 17, and 21 had conducted dredging; sites 17 and 18 were performing aeration when sampling; and sites 19 and 21 were equipped with aeration devices but were not functioning. Finally, 20 sampling sites that conducted water governance were selected and separated into group H (Table 1). In this study, the sampling sites with water governance, regardless of how many measures were used, were all placed in group H. The 25 remaining sites were separated into groups NG and NB according to the TLI ( Table 1). The sites in the NG group obtained a maximum TLI of 60 and were typically macrophyte-dominated, and the TLI of the NB group was higher than 60, with floating algae present in most sites.
According to the reconstructed CH 4 data, the mean values of CH 4 concentrations and diffusive fluxes from groups H, NB, and NG were 0.72 µmol L −1 (23.33 mmol m −2 d −1 ), 8.91 µmol L −1 (277.92 mmol m −2 d −1 ), and 0.24 µmol L −1 (7.50 mmol m −2 d −1 ), respectively. ANOVA revealed a significant difference in CH 4 concentration of the surface water among groups H, NB, and NG. Group H showed significantly different CH 4 from group NB (p < 0.001) but no significant difference with group NG according to the post hoc test (Table 3). DO, TN, TP, COD, NH 3 -N, and PO 4 -P were significantly different between groups H and NB (p < 0.01 for DO and p < 0.001 for others), while CH 4 and water quality variables had no significant difference between groups H and NG, except COD Mn , which demonstrated significance at p < 0.05 ( Table 3). The PCA results also showed the long distance between groups H and NB on the PC1 axis, while groups H and NG partially coincided on the PC1 axis ( Figure 5). Table 3. p-Values of one-way ANOVA and post hoc test to determine the significant difference between CH 4 and water quality among groups H, NB, and NG. Only significant p-values were presented in post hoc tests.

ANOVA
Post Hoc "Bonferroni" Post Hoc "Bonferroni" According to the reconstructed CH4 data, the mean values of CH4 concentrations and diffusive fluxes from groups H, NB, and NG were 0.72 μmol L −1 (23.33 mmol m −2 d −1 ), 8.91 μmol L −1 (277.92 mmol m −2 d −1 ), and 0.24 μmol L −1 (7.50 mmol m −2 d −1 ), respectively. ANOVA revealed a significant difference in CH4 concentration of the surface water among groups H, NB, and NG. Group H showed significantly different CH4 from group NB (p < 0.001) but no significant difference with group NG according to the post hoc test (Table 3). DO, TN, TP, COD, NH3-N, and PO4-P were significantly different between groups H and NB (p < 0.01 for DO and p < 0.001 for others), while CH4 and water quality variables had no significant difference between groups H and NG, except CODMn, which demonstrated significance at p < 0.05 ( Table 3). The PCA results also showed the long distance between groups H and NB on the PC1 axis, while groups H and NG partially coincided on the PC1 axis ( Figure 5). Table 3. p-Values of one-way ANOVA and post hoc test to determine the significant difference between CH4 and water quality among groups H, NB, and NG. Only significant p-values were presented in post hoc tests.  5. PCA of water quality variables and CH4. Colored arrows denote contributions of parameters to principal components. Labels for the scatter represent sampling sites, which were divided into three groups. Group H includes sites with water governance. Group NB includes sites that had naturally poor water quality. Group NG represents sites with naturally good water quality.

ANOVA Post Hoc "Bonferroni" Post Hoc "Bonferroni" p-Values p-Values between Groups H and NB p-Values between Groups H and NG
According to the individual PCA shown in Figure 5, the dispersed site distribution in group H indicated the different effects of water governance. For example, sites 18 and 13 demonstrated high water quality after water governance; however, although various measures were applied, the water Figure 5. PCA of water quality variables and CH 4 . Colored arrows denote contributions of parameters to principal components. Labels for the scatter represent sampling sites, which were divided into three groups. Group H includes sites with water governance. Group NB includes sites that had naturally poor water quality. Group NG represents sites with naturally good water quality. According to the individual PCA shown in Figure 5, the dispersed site distribution in group H indicated the different effects of water governance. For example, sites 18 and 13 demonstrated high water quality after water governance; however, although various measures were applied, the water at site 9 continued to have extremely poor quality. Among the 20 sites in group H, only 8 sites (1, 6, 7, 13, 18, 23, 37, and 39) had TLI less than 60. Further analysis determined that although NH 3 -N decreased significantly after water governance (Table 3), NO 3 -N still exhibited a high level in a few sites, including sites 30, 45, 42, and 36 ( Figure 5).

Discussion
Significantly high CH 4 concentrations in the surface water and CH 4 diffusive fluxes occurred in hyper-eutrophicated river sites (TLI > 70). This result is consistent with the findings of previous studies that found a positive correlation between CH 4 emission and water quality [27,28]. Oxygen depletion and increased carbon loading are considered the primary drivers of high CH 4 production and emissions in eutrophic lakes [29,30]. In this study, DOC showed no correlation with TN and TP, and was even negatively correlated with Chla, which was probably because DOC in rivers is primarily allochthonous rather than autochthonous. The DOC content exhibited no relationship with CH 4 , and the LOI was also inconsistent with the CH 4 content in sediments (Table 2, Figure 3). These results indicate that CH 4 production and emission in the investigated rivers were not limited by the carbon content.
Net CH 4 flux is determined by the balance in its production, oxidation, and transport. In this study, surface water CH 4 concentration was not correlated with sediment CH 4 content but was significantly correlated with water DO and ORP. These results suggest the importance of CH 4 oxidation in the water column in the regulation of CH 4 flux. Carbon and nitrogen circulations are interactive processes in the river ecosystem, and nitrogen can be a key contributing factor to long-term carbon balance [31]. We found a significantly high correlation between CH 4 in the water and NH 3 -N content (Figure 4). NH 3 -N was much higher than NO 3 -N in the hyper-eutrophicated sites in group C ( Figure 2) and may serve as the main N source for photosynthesis of vegetation, which applied abundant substrate for CH 4 production. Moreover, NH 3 -N is a factor that influences CH 4 oxidation in aquatic sediments based on the competitive and noncompetitive inhibition mechanisms [32,33]. In summary, eutrophication due to the increase in N loading and decrease in DO in the investigated rivers likely benefits CH 4 production and inhibits CH 4 oxidation and evasion in air.
The CH 4 concentration data were comparable with the global-scale estimate of fluvial CH 4 concentration (1.35 ± 5.16 µmol L −1 , medium 0.25 µmol L −1 ), but the CH 4 diffusive flux was much lower than the global estimated level (342.50 ± 1062.50 µmol m −2 h −1 , medium 35.83 µmol m −2 h −1 ) [6]. This finding may be caused by the low wind speed and weak turbulence of water in urban rivers based on the stagnant flow velocity, which had a direct influence on the diffusion of CH 4 from the water to the air [6]. The CH 4 data from urban river networks or typical rivers in cities were collected for comparison with the CH 4 results in this study (Table 4). An average estimate of the CH 4 concentration in these urban rivers was equal to 10.8 µmol L −1 , with high variability and a medium of 2.0 µmol L −1 . CH 4 concentration is generally high in urban rivers, and extremely high CH 4 levels are demonstrated in rivers that have been highly polluted by sewage. Moreover, rivers in rural areas with water influx from fertilized cropland or industrial areas can also be significant CH 4 sources. A comparison of CH 4 levels from different cities, such as Shanghai, Chongqing, Chennai, and Glasgow, showed that metropolises contributed significant CH 4 emissions, and rivers in Wuxi City showed the same level of CH 4 content as those in Tianjin and Hefei (Table 4). Similar to this study, the water quality had a significant influence on CH 4 production and emission in other urban river systems, especially for DO and NH 3 -N (Table 4). Rivers are an important component of urban ecosystems and are highly threatened by humans because of reduced connectivity by dams, river channelization, and disposal of urban and industrial sewage water. Such human activities have induced to the degradation of river ecosystems [34,35] and additional CH 4 emission. Given the rapid urbanization in global areas, previous studies have emphasized the notification of carbon evasion from urban rivers [7,8]. However, previous studies ignored the effect of water governance on water quality and CH 4 emission. Water treatment practices, such as dredging and aeration, should contribute to the removal of nutrient elements and influence the biogeochemical process [41]. For example, a study in Lake Taihu found that oxygen penetration depth increased with a thickening oxidation layer in newborn sediments after dredging [42]. In this study, the increase of oxygen and the reducing organic matter content due to water governance could benefit CH 4 oxidation but could result in less CH 4 production, for example, the low sediment CH 4 concentration and LOI at site 17 after dredging. However, more studies are needed for in-depth exploration of the influential mechanisms of various water quality management measures on CH 4 dynamics.
According to our statistics, 58% of the sampling sites in the study area conducted water treatment, and this proportion increased to 84% if sites with naturally good water quality were excluded. Notably, under these control measures, the treated sites (group H) showed significantly better water quality than the sites in the NB group and had no significant difference with that in the NG group (Table 3, Figure 5), thereby indicating a positive tendency toward water ecological restoration. The CH 4 diffusive flux of the sites in group H (23.33 µmol m −2 h −1 ) was 12 times lower than that of the sites in group NB (277.92 µmol m −2 h −1 ) and 3 times higher than that of those in group NG (7.50 µmol m −2 h −1 ). These findings suggest that CH 4 evasion to the air decreases significantly after extensive and intensified water governance. Water management on a worldwide scale started at the end of the twentieth century, and old rivers, canals, and lakes were first restored in Western cities. The high environmental demand and standard in developing countries or recently developed countries contributes to a gradual increase of water quality control [43]. Hence, along with the improvement of the water environment, CH 4 emission in urban rivers would also significantly decrease.
Our study has the following limitations: (1) The sampling in this study was conducted at the end of the wet and hot season when high CH 4 production and emission occurred according to our seasonal study of CH 4 dynamics in Lake Taihu [44]. Hence, our results only represent the water quality and CH 4 emission characteristics in the wet and hot season. Multiple samplings at different times are necessary to obtain comprehensive and year-round conclusions on the effect of water governance. (2) Ebullition may be an important way of CH 4 emission considering the oversaturation of CH 4 concentration in sediments, but it was excluded in our study (Table 2). Moreover, CH 4 ebullition is considered to be regulated by sediment structure, such as grain size [45]. Hence, water treatment activities, especially for dredging, should have a strong effect on CH 4 bubble formation and emission. In this respect, the reduced effect of CH 4 emission due to water governance may have been underestimated in our study.

Conflicts of Interest:
The authors declare no conflict of interest.