Responses of Net Anthropogenic N Inputs and Export Fluxes in the Megacity of Chengdu, China

Abstract

Anthropogenic N inputs have become progressively more problematic and have profoundly affected the water quality in megacities throughout China. Thus, to design and implement appropriate megalopolis watershed management, it is important to understand the relationship between N inputs and exports and to identify the N pollution sources. To that end, in this work, the net anthropogenic N inputs (NANI) in Chengdu City were estimated based on statistical data collected between 1970 and 2019. N input fluxes and pollution sources were estimated through sample collection and field measurements that were performed between 2017 and 2019, while nitrate (${\mathrm{NO}}_3^{-}$) was identified using stable isotope and Bayesian model (SIAR) analysis. The NANI was found to be affected primarily by livestock and poultry consumption of N rich feed. Moreover, the N export fluxes and runoff showed a high degree of correlation. Notably, ${\mathrm{NO}}_3^{-}$ fluxes exhibited a significant increase over the course of the study period, such that, by 2019, the total N fluxes (18,883.85 N kg/${\mathrm{km}}^2$) exceeded the NANI (17,093.87 N kg/${\mathrm{km}}^2$). The results indicate that although livestock and poultry farming were the original primary sources of NANI, their contributions declined on an annual basis. Moreover, with the emphasis placed on point source management in Chengdu City, domestic sewage discharge has been significantly reduced. Therefore, N retention in groundwater is thought to be the factor driving the N flux increase. These findings are pivotal to solving the N pollution problem in megacities like Chengdu (China).

1. Introduction

Balancing population growth, industrial development, and environmental quality is a challenge that is experienced on a global scale [1,2,3,4,5]. Among a plethora of issues, rapid urbanization has led to surface runoff pollution problems—such as pollutants flowing from paved roads into rivers with rainwater, industrial and domestic sewage discharge, atmospheric deposition pollution, and excessive fertilizer application [6,7,8,9,10]. The Yangtze River is considered the birthplace of Chinese civilization and is a strategic economic development area in China. As the furthest upstream megacity in the Yangtze River Basin, Chengdu City water pollution directly impacts economic development and water quality in the Yangtze River’s lower reaches.

N inputs from anthropogenic activities have affected the global N cycle worldwide [11,12,13]. As such, nitrate (${\mathrm{NO}}_3^{-}$) pollution in surface water has gradually attracted attention from the scientific community. This study is particularly relevant because decision makers in most Chinese cities disregard nitrate pollution and are, therefore, unable to propose appropriate control measures in the context of global nitrate pollution. Notably, ${\mathrm{NO}}_3^{-}$ is the main form of N in surface water [14,15,16], and excessive ${\mathrm{NO}}_3^{-}$ inputs can lead to eutrophication. Furthermore, long-term ${\mathrm{NO}}_3^{-}$ consumption poses a serious threat to human health (e.g., gastric cancer [16,17,18] and blue baby syndrome [19,20,21]). Moreover, once the water environment is polluted with ${\mathrm{NO}}_3^{-}$, self-purification is unlikely to occur [22,23,24]. To ensure that the water quality remains healthy, it is particularly important to clarify the source(s) of ${\mathrm{NO}}_3^{-}$ in surface water and explore its transformation pathway.${\mathrm{NO}}_3^{-}$ surface water sources are complex and varied, and include: precipitation, soil N mineralization, field fertilization, septic tank discharge, manure, domestic sewage, and industrial wastewater [25,26,27,28]. The N and O isotope compositions in ${\mathrm{NO}}_3^{-}$ that originate from different sources present distinct characteristics, which can be used to identify the ${\mathrm{NO}}_3^{-}$ sources and analyze the N cycling process [15,29]. As compared to source pollution models (e.g., SWAT, AGNPS, etc.), using ${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$ and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ to identify nitrate sources is simpler and faster [30,31], and was therefore the selected method for this investigation.

The goal of this study was to estimate the net anthropogenic N inputs (NANI) and the ${\mathrm{NO}}_3^{-}$-N export fluxes of the Yangtze River Basin in Chengdu. Moreover, the N pollution sources were investigated based on ${\mathrm{NO}}_3^{-}$ dual stable isotopes (${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$ and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$) to determine the response relationship between anthropogenic activities and environmental water quality. The results presented herein can be used to support watershed management in megacities, enrich the global ${\mathrm{NO}}_3^{-}$ isotope database, and promote the development of isotope geochemistry research.

2. Materials and Methods

2.1. Study Area

As one of the seven megacities located in China, Chengdu has a population of 20.94 million—68.74% of whom are from 15–59 years old. Thus, Chengdu’s Working Age population is numerically superior to those of Beijing and Shanghai. Furthermore, the urbanization rate has reached a total of 77%, which is lower than that of China’s other megacities. Both of these values emphasize Chengdu’s extensive development potential. In 2020, Chengdu’s GDP was 1.77 trillion CNY, the 8th highest in China.

Chengdu was built in the upper reaches of the Yangtze River Basin (Figure 1) and covers an area of 14,335 ${\mathrm{km}}^2$. It is located within the subtropical humid monsoon climate zone, and experiences an average annual temperature of 15.2–16.6 °C and an annual rainfall of 800–1400 mm. Interestingly, Chengdu straddles two water systems, the Minjiang River and the Tuojiang River. The Minjiang River, which was once considered the Yangtze River’s main stream, divides into the Jinma River Basin (JM) and the Jinjiang River Basin (J) at the Dujiangyan Fish Mouth (i.e., part of a famous ancient water project). Since ancient times, fish mouths have provided a steady flow of water to the J throughout the year, allowing agricultural irrigation and preventing floods. Excess water flows to the JM, which is mainly used for flood discharge. Although the Tuojiang River has its own water system, it actually draws water from the Minjiang River. Notably, the JM, J, and Tuojiang River Basin (T) account for 44.43%, 15.94%, and 39.63% of the total watershed, respectively.

2.2. Calculation of the NANI

The NANI values in three watersheds were quantified based on sampling conducted between 1970 and 2020. Data for 1980, 1990, 2000, and 2010–2019 were obtained from the Chengdu Statistical Yearbook, while data for 1970 were estimated by time-line interpolation. Most of the values were traced to food/feed inputs, agricultural fixation, fertilizer application, and atmospheric deposition. This relationship is mathematically expressed as [32,33]:

$$\mathrm{NANI} = N_{\mathrm{im}} + N_{\mathrm{fer}} + N_{\mathrm{cro}} + N_{\mathrm{dep}}$$

(1)

where $N_{\mathrm{im}}$= food/feed N inputs, $N_{\mathrm{fer}}$= fertilizer N inputs, $N_{\mathrm{cro}}$ = N from crop fixation, and $N_{\mathrm{dep}}$ = N from atmospheric deposition. All the variables are presented in $\mathrm{kg}/\left({\mathrm{km}}^2\mathrm{a}\right)$.

2.2.1. Calculation of $N_{\mathrm{im}}$

The $N_{\mathrm{im}}$ values were calculated by summing the anthropogenic and livestock N consumption (i.e., the positive N fluxes into the study area), and then subtracting the livestock and agricultural crop N production (i.e., the negative N fluxes from the study area). This formula is mathematically expressed as [32,33]:

$$N_{\mathrm{im}} = N_{\mathrm{hc}} + N_{\mathrm{lc}}-N_{\mathrm{lp}}-N_{\mathrm{cp}}$$

(2)

where $N_{\mathrm{hc}}$ and $N_{\mathrm{lc}}$ represent the N consumption from food (anthropogenic) and feed (livestock), respectively, while $N_{\mathrm{lp}}$ and $N_{\mathrm{cp}}$ are the N contents in livestock/poultry products and agricultural crop products, respectively. All the variables are presented in $\mathrm{kg}/\left({\mathrm{km}}^2 \mathrm{a}\right)$.

$N_{\mathrm{hc}}$ was estimated based on the population size and the protein-N content of the food intake. As such, $N_{\mathrm{hc}}$ was calculated separately for urban and rural populations. In contrast, $N_{\mathrm{lc}}$ was calculated according to the amount of livestock and poultry and their respective N intake level. $N_{\mathrm{lp}}$ represented the N production from animals—i.e., in meat, milk, and eggs—while $N_{\mathrm{cp}}$ represented the N content of the crop harvest in the catchment. The values of these variables were estimated by multiplying the production quantity by the N content produced by each animal and crop. The livestock and poultry species considered here included mainly pigs, cattle, sheep, rabbits, and poultry, while the agricultural crop products were mainly paddy, wheat, corn, rapeseed, and potato. Spoilage and inedible components were assumed to cause a 10% loss in the amount of products available for consumption.

2.2.2. Calculation of $N_{\mathrm{fer}}$

$N_{\mathrm{fer}}$ was mainly derived from urea-N and combined fertilizers. It is a scalar number representing the quantity of N fertilizer and the N content of combined fertilizers (~32.2%) [32].

2.2.3. Calculation of $N_{\mathrm{cro}}$

$N_{\mathrm{cro}}$ was estimated by multiplying the area of each N-fixing crop by its N fixation rate and summing all the results. The N-fixing crops within the study area mainly included soybean, rice, and peanut plants—each of which exhibited N fixation rates of 9600, 4480, and 8000 N kg/${\mathrm{km}}^2$, respectively [32].

2.2.4. Calculation of $N_{\mathrm{dep}}$

${\mathrm{NO}}_y$-N, ${\mathrm{NH}}_x$-N, and atmospheric organic nitrogen (AON) were considered when calculating $N_{\mathrm{dep}}$. The inorganic N deposition data were derived from a dataset in the China Science Data Network, which is based on data collected between 1996 and 2015. This dataset, which depicts the spatial arrangement of inorganic N deposition in China, was the first of its type to be openly accessible, and currently is considered the most reliable, given the available options. AON deposition was assumed to be 15% of the total inorganic N deposition [32]. Missing yearly data were calculated by linear difference.

2.3. River N Export Fluxes

Monthly flow, total nitrogen (TN), and ammonia-N (${\mathrm{NH}}_4^{+}$-N) data from 2011 to 2016 were obtained from the Central Monitoring Station in Chengdu. Based on the 2017–2019 data, which were empirically obtained (Section 2.4.), ${\mathrm{NO}}_3^{-}$-N was determined by calculating the difference between TN and ${\mathrm{NH}}_4^{+}$-N (${\mathrm{NO}}_3^{-}$-N = (TN − ${\mathrm{NH}}_4^{+}$-N) × 0.87. The following formula was used to quantify the N export fluxes (3):

$$F = \frac{Q\times C\times {10}^5}{A\times 3.17}$$

(3)

where F represents the riverine N export fluxes ($\mathrm{kg}/\left({\mathrm{km}}^2 \mathrm{a}\right))$, Q = runoff (${\mathrm{m}}^3/\mathrm{s}$), C = annual average N concentration (mg/L), and A = the basin area (${\mathrm{km}}^2$) [34]. Moreover, “${10}^5$” and “3.17” are the conversion factors for converting mg/L to kg/${\mathrm{m}}^3$ and seconds (s) to annual (a).

2.4. Sample Collection and Analysis

From 2017–2019, every month, 75 samples were collected along the full length of the watershed in Chengdu (Figure 1). At each sample site, the water flow was measured using doppler ultrasonography (RiverSurvey M9, SonTek, San Diego, CA, USA). Next, water samples were collected and analyzed for ${\mathrm{NO}}_3^{-}$, ${\mathrm{NH}}_4^{+}$, nitrite (${\mathrm{NO}}_2^{-}$), and TN. ${\mathrm{NO}}_3^{-}$ and ${\mathrm{NO}}_2^{-}$ were measured using ion chromatography (883 Basic IC, Metrohm, HeriSau, Switzerland), while the ${\mathrm{NH}}_4^{+}$ concentrations were determined by spectrophotometry (722N, Shanghai Jingke, Shanghai, China). The TN was digested with alkaline K persulfate and analyzed by spectrophotometry (UV752, Shanghai Jingke, Shanghai, China) after reducing the ${\mathrm{NO}}_3^{-}$-N to ${\mathrm{NO}}_2^{-}$-N.

The ${\mathrm{NO}}_3^{-}$ ${\delta }^{15}\mathrm{N}$ and ${\delta }^{18}\mathrm{O}$ analysis was performed after the water samples were filtered in situ with disposable filter devices (0.45 μm pore size, 25 mm diameter, Whatman, GD/X, Maidstone, UK), frozen, and stored at <4 °C in PET centrifuge tubes (15 mL, Sterile, Corning, NY, USA). The ${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$ and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ were determined via the bacterial denitrification method using an isotope mass spectrometer (MAT 253, ThermoFisher, Waltham, MA, USA) [27,35,36].

2.5. Bayesian Model in R (SIAR) Mixing Model Estimation

The ${\mathrm{NO}}_3^{-}$ source contributions to the river water samples were quantified using the SIAR mixing model, which can be mathematically expressed as follows [37,38]:

$$\left\{\begin{array}{c}{X}_{ij} = {\displaystyle {\displaystyle \sum }_{k = 1}^k}{P}_k\left(S_{jk} + C_{jk}\right) + {\epsilon }_{jk}\\ S_{jk}~N({\mu }_{jk},{\omega }_{jk}^2)\\ C_{jk}~N({\lambda }_{jk},{\tau }_{jk}^2)\\ {\epsilon }_{ij}~N(0,{\sigma }_j^2)\end{array}\right.$$

(4)

where X_ij is the isotope value j of the mixture i (i = 1, 2, 3, …, i and j = 1, 2, 3, …, j); P_k is the proportion of source k (which needs to be estimated by the SIAR model); S_jk is the source value k on isotope j (k = 1, 2, 3, …, k) and is normally distributed, with a mean value = μ_jk and standard deviation = ω_jk [39]; and C_jk is the fractionation factor for isotope j on source k and is normally distributed, with a mean value = λ_jk and standard deviation = τ_jk. In addition, ε_ij is the residual error, which represents the additional, unquantified variation between individual mixtures and is normally distributed, with a mean value = 0 and standard deviation = σ_j [40].

The contributions from each ${\mathrm{NO}}_3^{-}$ source in the Chengdu City watershed were estimated based on the measured ${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$ (‰) and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ (‰) values. Specific isotope values were considered to determine the ${\mathrm{NO}}_3^{-}$ contribution of four potential sources: precipitation (NP), fertilizer nitrification (NF), soil N (SN), and manure and sewage (M & S). These specific isotope values were obtained through monitoring and from published literature (

Table 1. ${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$ and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ values of various ${\mathrm{NO}}_3^{-}$ sources.

Sources		n	${\mathbf{\delta }}^{\mathbf{15}}\mathbf{N}$-${\mathbf{NO}}_{\mathbf{3}}^{\mathbf{-}}$ (‰)		${\mathbf{\delta }}^{\mathbf{18}}\mathbf{O}$-${\mathbf{NO}}_{\mathbf{3}}^{\mathbf{-}}$ (‰)
			Mean	SD	Mean	SD
${\mathrm{NO}}_3^{-}$ in precipitation (NP)		26	2.3	5.4	30.9	9.4
Fertilizer N (NF)		12	−3.5	9.1	−1.1	2.1
Soil N (SN) a		-	4.6	4.4	−0.3	6.8
Manure and sewage (M & S)	Manure	8	22.9	8.4	3.1	5.0
	Sewage	50	14.4	7.8	−1.9	4.7
	Total	58	15.58	8.4	−1.3	4.9

^a Data obtained from [14,41,42,43,44,45,46].

). The fractionation factor value (C_jk) in Equation (2) was assumed to be 0, because denitrification could not to occur in the stream water (DO > 2 mg/L).

3. Results

3.1. Variation of the NANI in Chengdu between 1970 and 2019

The 2010 NANI values in Chengdu reflected a change in N inputs derived from anthropogenic activity. The NANI increased between 1970 and 2010, and its cascade declined after 2010 (Figure 2). Over the first four decades, the NANI multi-year average value was 18,429.06 N kg/(${\mathrm{km}}^2 \mathrm{a}$), significantly higher than the average for China as a whole (11,109 N kg/(${\mathrm{km}}^2 \mathrm{a}$)). Furthermore, the NANI value more than doubled—increasing from 10,261.27 N kg/(${\mathrm{km}}^2 \mathrm{a}$) in 1970 to 26,047.28 N kg/(${\mathrm{km}}^2 \mathrm{a}$) in 2010. Additionally, during this period, the $N_{\mathrm{im}}$ increased significantly due to rapid population growth in Chengdu. In contrast, since 2010, the $N_{\mathrm{im}}$ has decreased significantly—from 15,819.39 N kg/(${\mathrm{km}}^2 \mathrm{a}$) to 6837.71 N kg/(${\mathrm{km}}^2 \mathrm{a}$)—while the $N_{\mathrm{fer}}$ has been declining yearly due to an accelerated rate of urbanization and a continuous decrease in cultivated land use. Notably, prior to 2010, the $N_{\mathrm{fer}}$ was the most important NANI component, yet since 2010, the $N_{\mathrm{im}}$ has maintained the position of prominence. The $N_{\mathrm{dep}}$ contribution, which is related to the interpolation calculation, increased steadily each year. Private car ownership is very common in Chengdu (4.39 million private cars, 2019), and in China, it is second only to Beijing (4.97 million private cars, 2019). Given that private car ownership has increased three-fold over the past decade, the $N_{\mathrm{dep}}$ is expected to continue increasing.

Between 1970 and 2019, the NANI and the $N_{\mathrm{im}}$ followed similar trends (Figure 2), and the components of $N_{\mathrm{im}}$ followed similar trends to those of the $N_{\mathrm{im}}$ and $N_{\mathrm{lc}}$ (Figure 3). These results indicate that N consumption in livestock and poultry feed was a leading contributor to both the $N_{\mathrm{im}}$ and the NANI. This finding is consistent with the extensive pig market in Chengdu (1.58 million pigs sold, 2019), which is generally substantially larger than in other Chinese megacities. For example, in 2019, the total number of pigs sold in Peking = 0.13 million, Shanghai = 0.57 million, and Wuhan = 1.05 million (Chinese (2020), Chengdu (2020) and Wuhan (2020) Statistical Yearbooks). Prior to 2010, the $N_{\mathrm{hc}}$ slowly increased, eventually stabilizing at ~4582.01 N kg/(${\mathrm{km}}^2 \mathrm{a}$). Since 2010, despite the considerable population increase, no significant changes in the $N_{\mathrm{hc}}$ were observed, as the land area also expanded. Furthermore, between 1970 and 2019, the $N_{\mathrm{cp}}$ decreased yearly in response to the decrease in cultivated area and crop yield. Before 2000, agriculture and animal husbandry were well developed in Chengdu. However, as the degree of urbanization increased, the percentage of people participating in these activities significantly decreased. Finally, from 1970–2019, the dominant $N_{\mathrm{im}}$ components $N_{\mathrm{im}}$ shifted from crop N production, which decreased from 72–25%, to livestock and poultry feed N consumption, which increased from 2–42%.

3.2. Variations in the Riverine TN Export Fluxes between 2011 and 2019

Figure 4 shows that N export fluxes fluctuated between 2011 and 2014 (average value = 15,692.07 N $\mathrm{kg}/({\mathrm{km}}^2\mathrm{a})$), decreased significantly in 2015, and then gradually increased through 2019. ${\mathrm{NO}}_3^{-}$-N export fluxes were the main contributors (49–76%) to the TN and showed a similar trend. The ${\mathrm{NH}}_4^{+}$-N export fluxes’ variation trend was similar to that of surface runoff, suggesting that the former was controlled by the latter. Notably, the TN export fluxes showed an overall opposite trend to that of the NANI, and in 2019, their values exceeded those of the NANI. Given these results, it can be inferred that N retention is occurring in the study area.

3.3. Identifying the $N{O}_3^{-}$-N Pollution Sources Using Data from 2017–2019

Previous studies have highlighted the prominence of ${\mathrm{NO}}_3^{-}$-N pollution in recent years and its close relationship to urban development. Dual stable isotopes (e.g., ${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$ and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$) are known as good tracers of ${\mathrm{NO}}_3^{-}$ pollution in rivers. In this work, the dual stable isotope approach was used in combination with the SIAR model to estimate the contribution rates of different ${\mathrm{NO}}_3^{-}$-N sources ${\mathrm{NO}}_3^{-}$ (Figure 5 and Figure 6). The results showed that the ${\mathrm{NO}}_3^{-}$-N export fluxes increased significantly between 2017 and 2019, and that the main contributing sources changed during the same period. Specifically, the contribution from fertilizers decreased significantly, while those of M & S became more prominent. In addition, soil erosion also provided a steady contribution (Figure 5 and

Table 2. Estimation of different N source contributions.

Year	Category	NP	NF	SN	M & S
2017	Low 95% hdr	0.00	0.27	0.01	0.27
	High 95% hdr	0.05	0.52	0.43	0.45
	mode	0.01	0.33	0.31	0.35
	mean	0.02	0.38	0.23	0.36
2018	Low 95% hdr	0.04	0.00	0.01	0.42
	High 95% hdr	0.12	0.25	0.45	0.70
	mode	0.08	0.03	0.22	0.57
	mean	0.08	0.12	0.24	0.56
2019	Low 95% hdr	0.08	0.00	0.04	0.33
	High 95% hdr	0.14	0.32	0.57	0.50
	mode	0.09	0.05	0.50	0.37
	mean	0.11	0.15	0.33	0.41
2017–2019	Low 95% hdr	0.05	0.01	0.02	0.34
	High 95% hdr	0.11	0.35	0.57	0.53
	mode	0.08	0.16	0.35	0.44
	mean	0.08	0.19	0.29	0.43

). These results are consistent with those from previous investigations.

4. Discussion

4.1. Relationships between N Budgets in the Chengdu City River Systems

To further confirm the relationship between the N budget and anthropogenic activities (e.g., population and GDP), several correlation analyses were carried out. Anthropogenic activity indicators and N inputs data were used to analyze the 1970–2019 period, and surface runoff and N export data were used to analyze the 2011–2019 period. As shown in Figure 7, a significant correlation was observed between the TN and the ${\mathrm{NO}}_3^{-}$-N export fluxes (${\mathrm{R}}^2$ = 0.785, p < 0.05). Moreover, there were simultaneous significant correlations between the ${\mathrm{NO}}_3^{-}$-N export fluxes and the crop N products (${\mathrm{R}}^2$= 0.786, p < 0.05), the crop N fixation (${\mathrm{R}}^2$= −0.676, p < 0.05), the atmospheric deposition (${\mathrm{R}}^2$= 0.759, p < 0.05), the human population (${\mathrm{R}}^2$ = 0.827, p < 0.01), and the GDP (${\mathrm{R}}^2$= 0.763, p < 0.05). In addition, the ${\mathrm{NH}}_4^{+}$-N export fluxes were correlated with the NANI (${\mathrm{R}}^2$ = 0.677, p < 0.05), the $N_{\mathrm{im}}$ (${\mathrm{R}}^2$ = 0.727, p < 0.05), the $N_{\mathrm{lc}}$ (${\mathrm{R}}^2$= 0.716, p < 0.05), the $N_{\mathrm{dep}}$ (${\mathrm{R}}^2$= −0.892, p < 0.01), the human population (${\mathrm{R}}^2$= −0.921, p < 0.01), the GDP (${\mathrm{R}}^2$= −0.868, p < 0.01), and the flow (${\mathrm{R}}^2$= 0.734, p < 0.05).

In summary, both the ${\mathrm{NO}}_3^{-}$-N and ${\mathrm{NH}}_4^{+}$-N export fluxes appeared to reflect changes in the human population and in the economic development. Notably, the ${\mathrm{NO}}_3^{-}$-N export fluxes were significantly affected by agriculture and atmospheric deposition, while those of ${\mathrm{NH}}_4^{+}$-N were significantly affected by animal husbandry. Furthermore, the NANI and $N_{\mathrm{im}}$ were significantly correlated with the ${\mathrm{NH}}_4^{+}$-N export fluxes, indicating that human activities heavily contributed to ${\mathrm{NH}}_4^{+}$ content.

4.2. Changes in N Inputs, Outputs, and Sources in Response to Urban Development and Management

Prior to 2010, Chengdu was a third-tier Chinese city that lacked subways and economic development zones. As such, it was not included among China’s top 10 GDP cities. On 12 May 2008, the city experienced a massive earthquake, which triggered a large scale rebuilding effort soon after the disaster. Since 2010, 13 subway lines have been constructed and Chengdu’s GDP has tripled. Moreover, the city now holds a top spot among China’s new first-tier cities. Overall, since 2010, Chengdu’s population, economy, and urban area have rapidly developed, and variations in the NANI and its components reflect these changes. Specifically, food N consumption increased with population growth, and as arable land acreage decreased, the amount of crops and fertilizer use decreased in parallel. Independent from the food N variations, NANI values were always related to N consumption in livestock and poultry feed (an aspect that has been neglected in previous studies).

In June 2017, Chengdu’s municipality issued the “Ten Water Control Measures” work plan and set a goal of remediating 296 black and smelly river sections by the end of 2020. Following publication of this work plan, agricultural non-point source sewage discharge, which would normally enter the river, has been regulated through ditches, and fertilizer pollution has been effectively controlled. In fact, since 2017, Chengdu’s sewage treatment plant has been upgraded, resulting in lower N emissions, and the amount of N input from anthropogenic activity has been decreasing yearly. Livestock/poultry breeding and domestic sewage are unlikely to be the main cause of the observed increase in N outputs. Alternatively, this trend might be explained by the gradual release of N trapped in groundwater into the river. This explanation is supported by the fact that an analysis of 13 groundwater samples in Chengdu exhibited ${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$ and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ values of 15.9 ± 8.0 ‰ and 4.6 ± 8.1 ‰, respectively. Using the same tracer method in this study showed that, as with surface water, the main source of ${\mathrm{NO}}_3^{-}$ in groundwater is M & S. These results strongly indicate that groundwater significantly contributes to the ${\mathrm{NO}}_3^{-}$ content in surface runoff.

4.3. Key Conversion of River N

The results described in Section 3.2 and Section 3.3 suggest, to a certain extent, the occurrence of N retention in the study area. Notably, nitrification and denitrification are key processes involved in N retention in rivers. In particular, nitrification is the process by which autotrophic microorganisms convert ${\mathrm{NH}}_4^{+}$ to ${\mathrm{NO}}_3^{-}$ via N oxidation, resulting in little N isotope fractionation. Theoretically, one third of the O in the ${\mathrm{NO}}_3^{-}$ produced by nitrification is derived from dissolved oxygen (DO) [48], whereas the remaining two thirds are derived from water and are produced during the oxidation of ${\mathrm{NH}}_4^{+}$ to ${\mathrm{NO}}_3^{-}$ [49,50]. This relationship can be expressed as follows: ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ = 2/3 ${\delta }^{18}\mathrm{O}$-H₂O + 1/3 ${\delta }^{18}\mathrm{O}$-air [51]. In this study, the ${\delta }^{18}\mathrm{O}$ values of the newly formed ${\mathrm{NO}}_3^{-}$ indicated that nitrification was the main factor determining ${\mathrm{NO}}_3^{-}$ formation. Considering the ${\delta }^{18}\mathrm{O}$ value for atmospheric O (23.5‰) and the typical ${\delta }^{18}\mathrm{O}$ values of precipitation in Chengdu (ranging from −16.7 to 1.9‰) [52], the calculated theoretical ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ values derived from nitrification in Chengdu City’s river water ranged from −3.3 to 9.1‰. These data support the presence of nitrification reactions in surface water. Figure 6 shows that the measured ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ values for the samples collected between 2018 and 2019 were consistent with the theoretical nitrification values. However, most of the samples collected in 2017 depict values outside this range. Implementation of the “Ten Water Control Measures” activities in 2017 is a possible cause of this phenomenon. In that year, a large input of ${\mathrm{NH}}_4^{+}$-N led to increased nitrification and, hence, to the formation of higher amounts of ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ from water O. The water was low in ${\delta }^{18}\mathrm{O}$-${\mathrm{H}}_2\mathrm{O}$, resulting in low ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ levels.

Denitrification plays an important role in the N cycle, as it can lead to a significant fractional distillation of N isotopes. A significant negative relationship is known to occur between ${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$ and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$. As such, ${\delta }^{15}\mathrm{N}$-${\mathrm{NO}}_3^{-}$/${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ ratios, which typically range from 2:1 to 1:1, have traditionally been used to diagnose denitrification [53]. However, we observed a significant positive relationship between ${\delta }^{15}\mathrm{N}$-$—$ and ${\delta }^{18}\mathrm{O}$-${\mathrm{NO}}_3^{-}$ (${\mathrm{R}}^2$ = 0.384, p < 0.01), with a linear regression slope of 0.31 (which deviated from the denitrification value range). Furthermore, the DO concentration in surface water was much higher than that ideal for denitrification (<1 mg O₂/L) [54]. Therefore, the denitrification process can reasonably be ignored in this study, as a considerable amount of ${\mathrm{NO}}_3^{-}$-N w—retained in the water system due to the dominance of nitrification. In this study, no significant inverse relationship was observed between ${\mathrm{NH}}_4^{+}$ and ${\mathrm{NO}}_3^{-}$. This indicates that ${\mathrm{NH}}_4^{+}$ nitrification did not contribute much of the ${\mathrm{NO}}_3^{-}$ detected in surface runoff. As such, most of the ${\mathrm{NO}}_3^{-}$ was derived from direct inputs. These findings support the groundwater contribution hypothesis presented in Section 4.2.

4.4. Prediction of N Retention

According to Section 2.3, the 2011–2019 measured runoff data from the Yangtze River Basin in Chengdu City and the corresponding NANI data were applied to iterate to convergence using the multiple regression model. This process is mathematically expressed as:

$$\mathrm{Y} = 103.9230 \times {\mathrm{Q}}^{0.7873} \times \exp \left(1.8982 \times {10}^{-6} \times \mathrm{NANI}\right)$$

(5)

Figure 8 shows that the TN export fluxes predicted using Equation (5) were basically consistent with the values measured between 2011 and 2019; hence, they reliably reflect the river N output. Based on Figure 4 and Figure 6, it can be inferred that over the next ten years the NANI will continue to decrease, while the TN will fluctuate with surface runoff. Furthermore, these predictions indicate that during the next decade, the average NANI value will reach 17,412.74 N kg/(${\mathrm{km}}^2 \mathrm{a}$), while the TN export fluxes will equal 17,709.37 N kg/(${\mathrm{km}}^2 \mathrm{a}$). Thus, these two variables will approximately counterbalance each other.

5. Conclusions

This study focused on the NANI characteristics for the period 1970–2019, the N export fluxes for the period 2011–2019, and the ${\mathrm{NO}}_3^{-}$ sources for the period 2017–2019. Given the four decades analyzed during this investigation, evolution of the NANI can be divided into two main stages: a first stage marked by a significant increase (1970–2010) and a second stage marked by a steep decline (2010–2019). Since 2010, $N_{\mathrm{im}}$ has been the leading source of NANI in the Chengdu City river basin (52.3%), while $N_{\mathrm{lc}}$ has been the leading source of $N_{\mathrm{im}}$ (53.0%). According to the data for the 2011–2019 period, the ${\mathrm{NO}}_3^{-}$-N export fluxes increased significantly toward the end of this period, and consequently, the N output exceeded the NANI value. Based on the isotopic tracing results, the sources of the TN export fluxes changed at the end of the period. Fertilizer contributions decreased significantly, and that of M & S became dominant. Considering this information in conjunction with the actual environmental management and the N budget characteristics, we suggest that a reverse excess of N output was more likely caused by N retention. Given these results, groundwater monitoring should be included in future environmental management strategies. This study also showed that the ${\mathrm{NO}}_3^{-}$-N export fluxes were significantly and positively correlated with both population and economic indicators. As such, future urbanization development should consider ${\mathrm{NO}}_3^{-}$-N pollution. The significant correlation observed between ${\mathrm{NO}}_3^{-}$ pollution and atmospheric deposition also highlights the need of monitoring atmospheric N deposition and N pollution. Overall, the results of this study deepen our understanding of N sources, fluxes, long-term changes, and future trends in the context of megacities and underscore the need to promote development of environmental management within these locales.