The Impacts of Nitrogen Pollution and Urbanization on the Carbon Dioxide Emission from Sewage-Draining River Networks

Carbon dioxide (CO2) emissions from river water have sparked worldwide concerns due to supersaturate CO2 levels in the majority of global rivers, while the knowledge on the associations among nitrogen pollution, urbanization, and CO2 emissions is still limited. In this study, the CO2 partial pressure (pCO2), carbon and nitrogen species, and water parameters in sewage-draining river networks were investigated. Extremely high pCO2 levels were observed in sewage and drainage river waters, such as Longfeng River, Beijing-drainage River, and Beitang-drainage River, which were approximately 4 times higher than the averaged pCO2 in worldwide rivers. Correlations of carbon/nitrogen species and pCO2 indicated that carbon dioxide in rural rivers and sewage waters primarily originated from soil aeration zones and biological processes of organic carbon/nitrogen input from drainage waters, while that in urban rivers and lakes was mainly dominated by organic matter degradation and biological respiration. Enhanced internal primary productivity played critical roles in absorbing CO2 by photosynthesis in some unsaturated pCO2 sampling sites. Additionally, higher pCO2 levels have been observed with higher NH4+-N and lower DO. CO2 fluxes in sewage waters exhibited extremely high levels compared with those of natural rivers. The results could provide implications for assessing CO2 emissions in diverse waters and fulfilling water management polices when considering water contamination under intense anthropogenic activities.


Introduction
River ecosystems play dominant roles in conveying materials between inland water and oceans, which has implications for the circulation of matter and energy in the global carbon pool, as well as the global carbon budget and climate change [1,2]. Generally, the supersaturate carbon dioxide (CO 2 ) levels of most river ecosystems cause CO 2 to be released from rivers to the atmosphere, making river ecosystems a vital source of the total global CO 2 emissions [3,4]. Global CO 2 emissions have been estimated at 1.80 petagrams of carbon per year (Pg C·yr -1 ) from rivers and 0.32 Pg C·yr -1 from lakes or reservoirs, being equivalent to approximately 2.34 times the total carbon transferred from inland water to oceans (0.90 Pg C·yr -1 ) [5,6]. Owing to the enormous challenges of carbon assessment and environmental management in the global ecosystem, CO 2 emissions from river networks have inspired great interest around the world. Therefore, much efforts have been made to assess the contributions of CO 2 emissions from natural river ecosystems, e.g., Amazon and Mekong River systems [7], Guadalete River [8], and the Wujiang River [9]. However, urban river networks are intensively impacted by anthropogenic activities (e.g., sewage draining and urban land coverages) compared with those of natural river networks, which generally exhibit more complex watershed environmental conditions dominating the production and 2 of 15 transportation of CO 2 [10]. Integrated studies have affirmed that watershed urbanization profoundly amplifies the uncertainty of riverine CO 2 emissions, while the fact that few works have concentrated on urban river networks suggests more efforts to fill the gap [11].
Mechanisms of CO 2 emissions from river ecosystems are associated with a combination of internal factors (temperature, pH, dissolved oxygen, and organic matters) and external factors (geomorphology, hydrology, precipitation, and anthropogenic activities) [8,12], which are further impacted by autotrophic and heterotrophic microbial activities in waters, photosynthesis of aquatic vegetation, organic matter degradation, and mineralization [13]. The drastic disturbance in urban river networks commonly expressing low dissolved oxygen (DO) and high nutrient or organic carbon concentration is closely related to the higher partial pressure of carbon dioxide (pCO 2 ) levels than rivers in less impacted areas. It is principally attributed to the fact that nutrients stimulate microbe reproduction and modify biochemical environmental conditions, accelerating CO 2 release from waters [14,15]. Multiple studies have also emphasized that the significantly different nutrient/carbon inputs from draining waters and soil CO 2 influxes between various urban rivers would profoundly enlarge the uncertainty of CO 2 emissions from regional river ecosystems [16][17][18]. Although these works supplied some valid results, the correlations between anthropogenic nutrient loading and CO 2 emissions, and the factors affecting on the distributions of CO 2 emissions from urban rivers, are still limited. The lack of investigation on overall river networks in specific regions, which has primarily concentrated on local rivers, lakes, and estuaries [19][20][21], is also not conducive to the construction of a global carbon budget system.
Enormous regions around the world have been experiencing rapid urbanizations. The uncertainty and complexity of sources of contaminant waters (especially domestic/industrial sewage-draining water and runoff coming from agricultural soil) would destroy the nitrogenous and carbonic balance of aquatic ecosystems [18]. Hence, focusing on CO 2 emissions in urban river networks is beneficial to recognize the response of aquatic ecosystems to anthropogenic activities as well as assessing the global carbon budget. Tianjin, one of the most prosperous coastal cities in China, receives sewage waters from local and surrounding areas (including Beijing and Hebei Province), regarded as important sources of CO 2 emissions. The aims of this study are to (1) disclose the spatial distribution patterns of pCO 2 and CO 2 emission fluxes from different waters; (2) explore the mechanisms producing CO 2 with various nitrogen concentrations; and (3) reveal the primary factors affecting CO 2 emissions in a eutrophic aquatic environment. The results of this study would be helpful for assessing the source and sink effects of CO 2 in urban and rural rivers under the influence of multiple factors, providing a database for future studies on rivers with similar characteristics of nitrogen pollution and urbanization worldwide, and fulfilling water management polices when considering the contribution of global urban rivers to carbon cycling and CO 2 emissions.

Study Site
is a densely populated (13.87 million people) coastal metropolis in China, located northeast of North China Plain and downstream of the Haihe River Basin, which is enclosed to the east by the Bohai Sea. The climate in the area is warm temperate sub-humid monsoon, with higher temperature in the summer (monthly average temperature of 28 • C) and an average annual temperature of 14.8 • C. The mean annual precipitation is 520-660 mm with approximately 75% occurring from June to August. There are five major rivers, the Haihe River, Yongdingxin River, Chaobaixin River, Duliujian River, and Jiyun River, in Tianjin ( Figure 1). Among these, the Haihe River flows through the urban area and eventually enters into the Bohai Sea. Tianjin serves as one of the most important economic centers of China experiencing rapid urbanization processes, which convert the river ecosystem due to anthropogenic activities. Tianjin river networks accept sewage from the neighboring areas as well as load their own water environmental pollution, threatening the water resource safety, to which more attention should be paid [22]. the Haihe River flows through the urban area and eventually enters into the Bohai Sea. Tianjin serves as one of the most important economic centers of China experiencing rapid urbanization processes, which convert the river ecosystem due to anthropogenic activities. Tianjin river networks accept sewage from the neighboring areas as well as load their own water environmental pollution, threatening the water resource safety, to which more attention should be paid [22].

Sampling and Analysis
Previous studies have verified that summer is a sensitive period for CO2 production and emissions because of promoted and active carbon biogeochemical processes [10,23]. Based on the climatic conditions and characteristics of the main rivers, water samples were collected throughout the Tianjin municipality in June 2015, including 11 sampling sites in lakes/reservoirs, 5 sampling sites in sewage-draining waters, and 41 sampling sites in rivers ( Figure 1). Among them, 18 samples in urban rivers (mainly in Haihe River) and 23 sampling sites in rural rivers were deployed to better investigate the CO2 emissions under various anthropogenic activities.
Surface water samples (depth 0.5 m) were collected by a Niskin sampler in triplicate and all the water samples were stored away from light at 4 °C until analysis in a laboratory. The water physical-chemical variables, including water temperature (T), dissolved oxygen (DO), electric conductivity (EC), pH, and total dissolved solids (TDSs), were determined in situ with a multi-parameter monitor (Yellow Springs Instrument, YSI-6600 V2, United States Gimcheon Instruments Inc.). Alkalinity (Alk) of samples was measured using a titration method with 0.01 mol·L −1 HCl in the field. The concentration of dissolved inorganic carbon (DIC) in water was calculated by the carbonate equilibrium system, detailed formulations are shown in Section 2.3. The concentration of dissolved organic carbon (DOC) in water was measured with an elemental analyzer (Vario TOC cube, Elementar, Langenselbold, Germany). The concentrations of NH4 + -N, NO3 − -N, NO2 − -N, and total nitrogen (TN) were determined using automatic flow analysis (AA3 Auto Analyzer, SEAL, Norderstedt, Germany) in a laboratory. However, because high concentrations of

Sampling and Analysis
Previous studies have verified that summer is a sensitive period for CO 2 production and emissions because of promoted and active carbon biogeochemical processes [10,23]. Based on the climatic conditions and characteristics of the main rivers, water samples were collected throughout the Tianjin municipality in June 2015, including 11 sampling sites in lakes/reservoirs, 5 sampling sites in sewage-draining waters, and 41 sampling sites in rivers ( Figure 1). Among them, 18 samples in urban rivers (mainly in Haihe River) and 23 sampling sites in rural rivers were deployed to better investigate the CO 2 emissions under various anthropogenic activities.
Surface water samples (depth 0.5 m) were collected by a Niskin sampler in triplicate and all the water samples were stored away from light at 4 • C until analysis in a laboratory. The water physical-chemical variables, including water temperature (T), dissolved oxygen (DO), electric conductivity (EC), pH, and total dissolved solids (TDSs), were determined in situ with a multi-parameter monitor (Yellow Springs Instrument, YSI-6600 V2, United States Gimcheon Instruments Inc.). Alkalinity (Alk) of samples was measured using a titration method with 0.01 mol·L −1 HCl in the field. The concentration of dissolved inorganic carbon (DIC) in water was calculated by the carbonate equilibrium system, detailed formulations are shown in Section 2.3. The concentration of dissolved organic carbon (DOC) in water was measured with an elemental analyzer (Vario TOC cube, Elementar, Langenselbold, Germany). The concentrations of NH 4 + -N, NO 3 − -N, NO 2 − -N, and total nitrogen (TN) were determined using automatic flow analysis (AA3 Auto Analyzer, SEAL, Norderstedt, Germany) in a laboratory. However, because high concentrations of organic matter and heavy metals in sewage water and effluents would affect the accuracies of the measurements of TN, the AA3 Auto Analyzer was equipped with Cd columns and effective UV lamps to ensure the digestion rate. By following the protocols of the People's Republic of China State Environmental Protection Standards: Determination of Total Nitrogen by Continuous Flow Analysis (CFA) and N-(1-naphthyl) Ethylene Diamine Dihydrochloride Spectrophotometry (Chinese Edition) (HJ 667-2013), the concentrations of total nitrogen were determined. Dissolved organic nitrogen (DON) was calculated by subtracting the dissolved inorganic nitrogen (DIN, including NH 4 + -N, NO 3 − -N, and NO 2 − -N) from TN. The authoritative protocol manual handbook in China (The water and wastewater monitoring and analysis methods (version 4)) was applied to determine the sewage-draining waters. Laboratory standards, blanks, and replicates were applied to guarantee that the accuracies of all the analyses were better than ±5%.

The Calculation of Partial Pressure of Carbon Dioxide in River Waters
The principal buffer mechanism in a freshwater system is carbonate equilibrium. H 2 CO 3 , HCO 3 − , CO 3 2− , and aqueous CO 2 make up the total dissolved inorganic carbon (DIC) in water, which is affected by pH, water temperature, and ionic strength [24,25]. Previous studies have reported that the alkalinity-pH-temperature method might overestimate the values of pCO 2 at low pH, while they could be accurately described with pH > 7.2 and total alkalinity > 1 mmol·L −1 [25,26]. Based on the pH (ranging from 7.29 to 10.41) and alkalinity (exceeding 1.43 mmol·L −1 ) in water samples in this study area, the alkalinity-pH-temperature method is employed to calculate pCO 2 . The pCO 2 values are calculated according to the carbonate equilibrium system which is described by the following equations: where K i are DIC dissociation constants related to temperature. Based on these, pCO 2 (µatm), H 2 CO 3 *, and CO 3 2− are evaluated with HCO 3 − , water temperature (T), and pH, and the relative calculating equations are as follows.

CO 2 Emission Flux Calculation
CO 2 concentration gradient and gas exchange coefficient commonly govern the CO 2 exchange rate between water and atmosphere [2]. According to the theoretical diffusion model of CO 2 , which has been efficiently employed at the water-air interface in previous studies [5,14], the CO 2 emission fluxes in this work are evaluated as follows: where F CO 2 (mmol·m −2 ·d −1 ) is the CO 2 emission flux, K CO 2 (µatm) is the Henry's law constant of CO 2 calibrated by measured T ( • C) in situ, k (cm·h −1 ) is the gas exchange coefficient at the air-water interface, pCO 2.water and pCO 2.air are the partial pressure of CO 2 in water and air, respectively. Among these, k is uncertainty that should be cautiously confirmed by specific environmental conditions. Researchers have made great efforts on combining multiple factors (such as the wind speed, temperature, stream slope, discharge, and depth) with estimation models that have been widely employed in previous studies to optimize the calculation model of gas transfer velocity [25,27,28]. Wind speed is a primary cause of turbulence in low-gradient river systems that strongly relate to k 600 [7,29]. In addition, it was proposed that averaged k could be used to estimate gas fluxes from water [30]. Considering the low terrain and steady river flow velocity in this work, k is evaluated by a typical model primarily correlating it with the wind speed and temperature. We also adopted the averaged value (k = 4 cm·h −1 ) to calculate CO 2 emission fluxes and inconspicuous differences between the two estimation methods were found. Relevant calculation equations are listed as follows [31]: in which k 600 is the gas exchange coefficient of CO 2 calibrated with wind speed U 10 (m·s −1 ) which is the wind speed 10 m above the water surface. The value of x is determined by U 10 , x = −0.67 or x = −0.5 when U 10 < 3.7 m·s −1 and U 10 > 3.7 m·s −1 , respectively. S C (cm·h −1 ) is the Schmidt number represented by T ( • C). The monthly averaged wind speed of Tianjin in June (from 2014 to 2021) was obtained from the China Meteorological Data Network.

Statistical Analysis
IBM SPSS Statistics 19 was used for the descriptive statistical analysis, variance analysis, and linear correlation analysis. One-way analysis of variance (ANOVA) was performed to examine the differences in pCO 2 /F CO 2 from different sampling sites. Correlation coefficients between pCO 2 /F CO 2 and environmental parameters were determined by linear regression analysis. Statistical significance was found when the p value < 0.05 for all the variance analyses and linear correlation analyses. All the figures were made by Origin 2018, R Studio, and Grapher 15.

Variations in Water Quality Parameters
Statistics of water quality parameters including water temperature (T), dissolved oxygen (DO), total dissolved solids (TDSs), electric conductivity (EC), pH, and alkalinity (Alk) are listed in Table 1. In June, surface water temperatures ranged between 25.60 • C and 34.80 • C. The pH ranged from 7.29 to 10.41 and the maximum value was detected in rural rivers (R23, Ziya River) while the minimum value was found in sewage-draining waters (R9, Longfeng-drainage River) (Figure 1). Large ranges of TDSs in river waters in Tianjin have been observed, with the highest value in the estuary of the Duliujian River (R27, 22,900 mg·L −1 ) and with lowest value in the Yuqiao Reservoir (L1, 250 mg·L −1 ). DO generally demonstrated oversaturated conditions in these waters, with an average value of saturation of 102%, and ranged from 2.15 to 14.24 mg·L −1 . As shown in Figure 2, from upstream to downstream, water parameters presented obvious longitudinal variations in all the studied rivers, with different spatial patterns. Since significant correlations (R = 0.79, p < 0.01) have been observed between DO and pH, the spatial distributions of DO and pH exhibited a similar tendency in all major rivers. In Haihe River, DO and pH increased remarkably in the urban area, while the EC and Alk decreased significantly in the same reaches. Except DO and pH in Jiyun River and Chaobaixin River, water parameters in most of the rivers increased from upstream to downstream and estuaries or coastal area.  Note: The data shown above are minimum-maximum, and average values ± mean deviation.
As shown in Figure 2, from upstream to downstream, water parameters presented obvious longitudinal variations in all the studied rivers, with different spatial patterns. Since significant correlations (R = 0.79, p < 0.01) have been observed between DO and pH, the spatial distributions of DO and pH exhibited a similar tendency in all major rivers. In Haihe River, DO and pH increased remarkably in the urban area, while the EC and Alk decreased significantly in the same reaches. Except DO and pH in Jiyun River and Chaobaixin River, water parameters in most of the rivers increased from upstream to downstream and estuaries or coastal area.

Variations in Nitrogen and Carbon Species
A box plot of nitrogen species in different waters is presented in Figure 3. The concentrations of TN ranged from 0.83 to 10.20 mg·L −1 in lakes, 2.11 to 16.97 mg·L −1 in rural rivers, 3.51 to 16.51 mg·L −1 in urban rivers, and 3.51 to 15.82 mg·L −1 in sewage waters, and mean values of NO 3 − -N were higher than other nitrogen species (NH 4 + -N, NO 2 − -N, and DON) except those in sewage-draining waters ( Table 2). Rural rivers and sewage-draining waters presented higher nitrogen concentrations in diverse waters, and urban rivers exhibited considerably higher nitrogen concentration than lakes (p < 0.05) (Figure 3). Higher NH 4 + -N concentrations were observed in anoxic surface waters, such as R24 in urban rivers, R9 and R20 in sewage-draining waters, where the DO concentrations were all below 3.67 mg·L −1 . Similar findings have been reported from studies in a coastal lagoon [21] and anoxic water layers in the hypolimnion of a lake/reservoir [27] as well. Except the lake waters, the NH 4 + -N in surface waters had a negative relationship with concentrations of DO (Figure 4). NH4 -N concentrations were observed in anoxic surface waters, such as R24 in urban ers, R9 and R20 in sewage-draining waters, where the DO concentrations were all be 3.67 mg·L −1 . Similar findings have been reported from studies in a coastal lagoon [21] anoxic water layers in the hypolimnion of a lake/reservoir [27] as well. Except the waters, the NH4 + -N in surface waters had a negative relationship with concentration DO (Figure 4). Note: The data shown above are average values ± mean deviation. PON is particle organic nitro

Partial Pressure of CO2 in Different River Waters
In total, the pCO2 averaged 2663.21 μatm in surface waters in the river networks in Tianjin, fluctuating within a wide range (standard deviation: 4124.99 μatm). Compared with the average values of global rivers (averaging 3100 μatm) [5,14], the pCO2 in most of the rivers in Tianjin presented a high level. Specifically, the mean values of pCO2 in sewage waters (11,350.59 ± 7243.23 μatm) were significantly higher than those of urban rivers (3035.41 ± 3510.66 μatm), rural rivers (1516.58 ± 1598.54μatm), and lakes (628.16 ± 807.27 μatm). The results of Pearson's correlation coefficient showed that pCO2 consistently exhibited negative correlations and was statistically significant with pH and DO, while it showed positive correlations with DON ( Figure 4). Additionally, pCO2 was positively related to DIC in river waters in Tianjin, especially in urban rivers, with the highest correlation coefficient (R = 0.94, p < 0.01). The significant correlations between nitrogen species (TN, NH4 + -N, NO3 − -N, and DON) and pCO2 have been observed in lakes and rural rivers.

Fluxes of CO2 in River Networks
Distributions of CO2 fluxes were consistent to those of pCO2, while there were obvious discrepancies when evaluating the various surface waters in Tianjin ( Figure 5). Calculated CO2 fluxes ranged from −10.47 mmol·m −2 ·d −1 (in Ziya River, R23) to 538.93 mmol·m −2 ·d −1 (in Beijing-drainage River, R20). The highest CO2 fluxes were found in sew-

Partial Pressure of CO 2 in Different River Waters
In total, the pCO 2 averaged 2663.21 µatm in surface waters in the river networks in Tianjin, fluctuating within a wide range (standard deviation: 4124.99 µatm). Compared with the average values of global rivers (averaging 3100 µatm) [5,14], the pCO 2 in most of the rivers in Tianjin presented a high level. Specifically, the mean values of pCO 2 in sewage waters (11,350.59 ± 7243.23 µatm) were significantly higher than those of urban rivers (3035.41 ± 3510.66 µatm), rural rivers (1516.58 ± 1598.54 µatm), and lakes (628.16 ± 807.27 µatm). The results of Pearson's correlation coefficient showed that pCO 2 consistently exhibited negative correlations and was statistically significant with pH and DO, while it showed positive correlations with DON ( Figure 4). Additionally, pCO 2 was positively related to DIC in river waters in Tianjin, especially in urban rivers, with the highest correlation coefficient (R = 0.94, p < 0.01). The significant correlations between nitrogen species (TN, NH 4 + -N, NO 3 − -N, and DON) and pCO 2 have been observed in lakes and rural rivers.

Fluxes of CO 2 in River Networks
Distributions of CO 2 fluxes were consistent to those of pCO 2 , while there were obvious discrepancies when evaluating the various surface waters in Tianjin ( Figure 5). Calculated CO 2 fluxes ranged from −10.47 mmol·m −2 ·d −1 (in Ziya River, R23) to 538.93 mmol·m −2 ·d −1 (in Beijing-drainage River, R20). The highest CO 2 fluxes were found in sewage waters (319.54 ± 210.54 mmol·m −2 ·d −1 ) compared to those in urban rivers (77.95 ± 102.30 mmol·m −2 ·d −1 ), lakes (7.75 ± 23.51 mmol·m −2 ·d −1 ), and rural rivers (33.42 ± 46.55 mmol·m −2 ·d −1 ) (Figure 6). Statistically significant differences in CO 2 fluxes in lakes from those in other river waters (p < 0.05) have been observed, while there were no obvious variations between sewage waters and urban rivers (p > 0.05). In hypoxic surface waters with higher CO 2 supersaturated conditions, increased CO 2 fluxes were also detected, such as the sampling sites L5 in Qilihai Wetland, R35 in Yongdingxin River, and R20 in Beijing-drainage River. Health 2022, 19, x FOR PEER REVIEW 9 mmol·m −2 ·d −1 ) ( Figure 6). Statistically significant differences in CO2 fluxes in lakes f those in other river waters (p < 0.05) have been observed, while there were no obv variations between sewage waters and urban rivers (p > 0.05). In hypoxic surface wa with higher CO2 supersaturated conditions, increased CO2 fluxes were also detected, s as the sampling sites L5 in Qilihai Wetland, R35 in Yongdingxin River, and R20 in Beij drainage River.

Sewage-Draining Dominated the Spatial Distributions of pCO2 in River Networks
Sewage-draining was linked to the aquatic environmental conditions, carbon o trogen accumulation, and organic matter decomposition rate, which obviously modi the intensity of riverine contamination levels and CO2 emissions [28,32]. Diverse river ters in Tianjin showed different contamination levels, mainly due to the complic  (Figure 6). Statistically significant differences in CO2 fluxes in lakes from those in other river waters (p < 0.05) have been observed, while there were no obvious variations between sewage waters and urban rivers (p > 0.05). In hypoxic surface waters with higher CO2 supersaturated conditions, increased CO2 fluxes were also detected, such as the sampling sites L5 in Qilihai Wetland, R35 in Yongdingxin River, and R20 in Beijing drainage River.

Sewage-Draining Dominated the Spatial Distributions of pCO2 in River Networks
Sewage-draining was linked to the aquatic environmental conditions, carbon or ni trogen accumulation, and organic matter decomposition rate, which obviously modified the intensity of riverine contamination levels and CO2 emissions [28,32]. Diverse river wa ters in Tianjin showed different contamination levels, mainly due to the complicated source of sewage-draining waters under intense anthropogenic activities [22,33] Lakes/reservoirs generally served as the water sources and nature reserves in Tianjin were less affected by sewage, and expressed the lowest pCO2 values in Tianjin River net

Sewage-Draining Dominated the Spatial Distributions of pCO 2 in River Networks
Sewage-draining was linked to the aquatic environmental conditions, carbon or nitrogen accumulation, and organic matter decomposition rate, which obviously modified the intensity of riverine contamination levels and CO 2 emissions [28,32]. Diverse river waters in Tianjin showed different contamination levels, mainly due to the complicated source of sewage-draining waters under intense anthropogenic activities [22,33]. Lakes/reservoirs generally served as the water sources and nature reserves in Tianjin, were less affected by sewage, and expressed the lowest pCO 2 values in Tianjin River networks ( Figure 5). Rural rivers generally accepting a large amount of domestic/industrial sewage have overwhelmingly altered aquatic ecosystems, while urban rivers could be considered as landscape waters with drastic disturbance by human activities. Higher pCO 2 values were generally accompanied with the low pH and high DIC in direct sewage-draining waters [11,25]. In this study, sampling sites in sewage-draining waters in Longfeng River (R8 and R9), Beijing-drainage River (R20), and Beitang-drainage River (R45) all presented higher pCO 2 (approximately 4 times that of averaged pCO 2 in worldwide rivers, i.e., 3100 µatm) [5] with lower pH (ranged from 7.29 to 7.85) and higher DIC (ranged from 4.05 to 6.85 mol·L −1 ) when compared with other waters in this study.
Traditionally, increased pCO 2 values were consistent with population density [34], and it was applicable in this study (with lower pCO 2 levels in rural rivers and lakes than in urban rivers and sewage waters). Here, we selected the Haihe River to disclose the effects of different pollution sources on pCO 2 in river waters, with the most obvious variations in environmental conditions (Figure 2). The higher pCO 2 values exhibited in R8 and R20 along the Haihe River were mainly attributed to the direct reception of large amounts of sewage in the upper reaches [11]. A similar observation was also made in urban rivers with direct sewage-draining waters in Beitang-drainage River (R45). However, compared with those in rural rivers, lower pCO 2 values in urban areas (especially in R35, R36, and R37) might result from the artificial dredging project which reduced biological respiration and organic matter degradation in river waters and sediments [35]. Additionally, the positive correlations between pCO 2 and DOC/DON in urban rivers (Figure 4) also indicated that increased organic matter played important roles in the CO 2 production in urban rivers. These results highlighted that sewage-draining would explain the differences in environmental conditions and spatial variations in pCO 2 in river waters, and sewage inputs increased pCO 2 levels.

Biogeochemical Processes Served as the Primary Mechanisms of CO 2 Production
It has been noticed that there are two production mechanisms of CO 2 in waters: the chemical weathering and the biological processes. Chemical weathering dissolves silicate or carbonate rocks by consuming CO 2 and generating DIC, which obviously occurred in the Karst watershed in China [9,24,36]. Biologically, the photosynthesis and respiration of aquatic plants, microbial respiration and reproduction, and organic matter degradation commonly dominate CO 2 production and release [13]. Additionally, eutrophic waters in Tianjin provided favorable conditions for phytoplankton reproduction, and high temperatures in the rainy season also stimulated microbial activities, which traditionally promote the primary productivity of phytoplankton fixing a large amount of CO 2 as important carbon sinks in the water [22,37].
Generally, water-rock reaction, CO 2 input from the soil aeration zone, and the dissolution of atmospheric CO 2 are commonly regarded as the primary sources of CO 2 in rivers [17,24,28]. However, because of low cover rates of exposed rocks or outcrops in the studied river watersheds, and the geological type was mainly quaternary sediment, and the water-rock reaction in the studied river waters could be neglected. Generally, the correlations of pH and DIC have been used to characterize the influencing factors of CO 2 production and transformations in river waters [21]. In this study, pH presented negative correlations with DIC in all the river waters (including lake waters and river waters) (R = −0.38, p < 0.01) including Haihe River (R = −0.78, p < 0.01) (Figure 6), revealing the potential contributions of photosynthetic processes to DIC uptake and CO 2 consumption in water, but they have obviously been affected by other processes and influencing factors [13]. Specifically, most of the urban rivers and lakes have very low water flow and long residence time, providing ideal static conditions for biogeochemical carbon transformations, including photosynthesis, respiration, and organic matter degradation [25].
On the one hand, in rural rivers and sewage rivers, frequent runoff processes in the rainy season during the sampling period transferred CO 2 from the soil aeration zone to river water, increasing the pCO 2 levels in the rural river water; on the other hand, a large amount of organic carbon/nitrogen input from drainage waters would promote CO 2 production by stimulating the organic matter degradation and biological respiration in sewage waters. Additionally, in the lakes and urban rivers, good positive relationships between DOC and pCO 2 (Figure 4) indicated that CO 2 production was mainly dominated by internal organic matter degradation and biological respiration.
Even though most of the river waters in the study area were sources of atmospheric CO 2 in the rainy season, unsaturated pCO 2 has been found in some of the sites in the lakes (L2, L4, L9, and L10) and urban areas (R33, R34, R37, R38, R39, R42, and R43), which may be attributed to the photosynthesis enhanced by algal bloom and well-growing aquatic plants in the rainy season with proper temperature and nutrient supply [20,33,38].

Effects of Carbon/Nitrogen Increases on pCO 2
Dramatic anthropogenic activities with increasing pollutant discharge have contributed to complicated environmental conditions in river ecosystems, and potentially provided huge challenges to biochemical processes [15,18]. In lakes, positive correlations of DON, inorganic nitrogen (NO 3 − -N and NH 4 + -N), and DOC have been observed, especially the significantly positive correlations between NO 3 − -N and DON (R = 0.84, p < 0.01) ( Figure 4). Generally, higher temperatures and carbon/nitrogen inputs relatively encouraged microbial activities and influenced the contributions of pCO 2 in river ecosystems [8,12,25]. Nutrient supplementation was also verified as an important factor influencing pCO 2 by enhancing the primary production and mineralization of organic matter [39][40][41]. Therefore, CO 2 emissions of lakes were mainly controlled by photosynthesis and mineralization of organic matter, which could be disclosed by the correlations between environmental parameters and nitrogen species.
Higher concentrations of NH 4 + -N have been observed in the Longfeng River (R8 and R9), Beijing-drainage River (R20), and Jiyun River (R17), which correspond to the amount of sewage waters produced by anthropogenic activities. The concentrations of NH 4 + -N are normally considered as the crucial indicator of sewage discharges from anthropogenic activities [42], which have been found to be the highest in sewage waters ( Table 2). Consistently with previous studies, increased pCO 2 with higher nitrogen concentrations was also found [43]. On the other hand, the positive correlations between NH 4 + -N and pCO 2 (R = 0.54, p < 0.01) also verified the effect of nitrogen inputs from sewage-draining waters on pCO 2 levels in rural rivers and sewage waters.
The distributions of nitrogen species (NH 4 + -N, NO 3 − -N, and DON), environmental parameters, and pCO 2 along the Haihe River, with distinct longitudinal variations (Figures 2 and 7), would be effective evidence on the influence of carbon/nitrogen concentrations. Urban rivers impacted more by anthropogenic activities undergoing intense biochemical processes exhibited higher pCO 2 levels [22,36]. The concentrations of NH 4 + -N were roughly decreased while DO and pH levels increased, in contrast to NO 3 − -N (especially in urban areas), demonstrating that the CO 2 emissions from river waters were produced by the microbial degradation under aerobic and hypoxic conditions [44]. The highest NO 3 − -N (10.35 mg·L −1 ) and DOC (9.07 mg·L −1 ) but lower NH 4 + -N (0.15 mg·L −1 ) and DON (0.10 mg·L −1 ) in R37 have been observed along the Haihe River, which could be explained by the principial biogeochemical processes in this river water being aerobic mineralization (DO saturation was 131%). Obviously positive correlations along the Haihe River between pCO 2 and DON (R = 0.84, p < 0.01) and DIC (R = 0.78, p < 0.01) illustrated that higher nutrient concentrations stimulated microbial activities, and the reception of wastewaters enhanced pCO 2 values [33,45]. Thus, CO 2 emissions should be investigated in detail because of the complexity of sources of carbon/nitrogen concentrations.

CO2 Fluxes
Similar to most river ecosystems under intense anthropogenic activities receiving a substantial amount of nutrients, CO2 fluxes demonstrated obvious positive correlations with the pCO2 values in Tianjin (n = 57, p < 0.01). In general, CO2 emissions from river ecosystems were mainly dominated by pCO2 levels at the air-water interface and gas transfer velocity [4,5,14]. As shown above, variations in averaged CO2 fluxes in river waters were higher in sewage waters and urban rivers while lower in lakes and rural rivers, illustrating the stronger effects of carbon/nitrogen concentrations on CO2 emissions in this study. CO2 fluxes from other river waters (rural rivers, urban rivers, and lakes) around the world are exhibited in Table 3. CO2 fluxes in urban rivers in Tianjin were estimated to be marginally lower than those in highly polluted eutrophic rivers [25,46], while significantly higher than those in other rural rivers [9,25,47]. CO2 fluxes in sewage waters exhibited extremely high levels compared with those of natural rivers. The annual CO2 fluxes were estimated to be 1.08 × 10 13 mmol (i.e., 1.08 × 10 -4 Pg C) according to the water area of the Tianjin River network (with 884 km 2 ) [22], which was a major source of atmospheric CO2. Despite the fact that projected values of CO2 emissions in this study were lower than global CO2 emissions from river ecosystems (1.80 Pg C·yr −1 from rivers and 0.32 Pg C·yr −1 from lakes) [5,6], the vast water area around the world should be given more consideration when considering the effects of CO2 emissions from water.

CO 2 Fluxes
Similar to most river ecosystems under intense anthropogenic activities receiving a substantial amount of nutrients, CO 2 fluxes demonstrated obvious positive correlations with the pCO 2 values in Tianjin (n = 57, p < 0.01). In general, CO 2 emissions from river ecosystems were mainly dominated by pCO 2 levels at the air-water interface and gas transfer velocity [4,5,14]. As shown above, variations in averaged CO 2 fluxes in river waters were higher in sewage waters and urban rivers while lower in lakes and rural rivers, illustrating the stronger effects of carbon/nitrogen concentrations on CO 2 emissions in this study. CO 2 fluxes from other river waters (rural rivers, urban rivers, and lakes) around the world are exhibited in Table 3. CO 2 fluxes in urban rivers in Tianjin were estimated to be marginally lower than those in highly polluted eutrophic rivers [25,46], while significantly higher than those in other rural rivers [9,25,47]. CO 2 fluxes in sewage waters exhibited extremely high levels compared with those of natural rivers. The annual CO 2 fluxes were estimated to be 1.08 × 10 13 mmol (i.e., 1.08 × 10 -4 Pg C) according to the water area of the Tianjin River network (with 884 km 2 ) [22], which was a major source of atmospheric CO 2 . Despite the fact that projected values of CO 2 emissions in this study were lower than global CO 2 emissions from river ecosystems (1.80 Pg C·yr −1 from rivers and 0.32 Pg C·yr −1 from lakes) [5,6], the vast water area around the world should be given more consideration when considering the effects of CO 2 emissions from water.

Conclusions
The study revealed the impacts of urbanization and nitrogen pollution on CO 2 emissions and production in river networks in Tianjin. Increasing sewage effluent and pollutant discharge in river ecosystems dramatically altered the environmental conditions, potentially triggering biochemical processes and CO 2 production. The spatial distributions of CO 2 emissions presented obvious differences mainly attributed to the complexity of sources of sewage-draining waters altering biological processes in river waters. Urban rivers and sewage waters exhibited higher pCO 2 and CO 2 fluxes than rural rivers and lakes, which might be dominated by the enhanced biochemical processes of respiration with higher carbon/nitrogen concentrations. Sampling sites with unsaturated pCO 2 levels in lakes and urban rivers were mainly controlled by photosynthesis and organic degradation. The correlations between pCO 2 and organic carbon/nitrogen demonstrated that nutrient accumulation stimulated microbial activities, and the reception of wastewaters could enhance pCO 2 levels. We highlighted the correlations between pCO 2 and carbon/nitrogen species in a regional river network and provided a theoretical basis for water management when considering CO 2 emissions under dramatic anthropogenic activities.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The datasets used or analyzed during the current study are available from the corresponding author on reasonable request.