The role of technology and policy in mitigating regional nitrogen pollution

Human activity greatly influences nitrogen (N) pollution in urbanized and adjacent areas. We comprehensively studied the N cycling in an urban–rural complex system, the Greater Hangzhou Area (GHA) in southeastern China. Our results indicated that subsurface N accumulation doubled, riverine N export tripled and atmospheric N pollutants increased 2.5 times within the GHA from 1980–2004. Agriculture was the largest N pollution source to air and water before 2000, whereas industry and human living gradually became the primary N pollution sources with the socioeconomic development. Based on the sensitivity analysis, we developed a scenario analysis to quantify the effects of technology and policy on environmental N dynamics. The fertilization reduction scenario suggested that the groundwater N pollution could decrease by 17% with less than a 5% reduction in crop production; the N effluent standard revision scenario led to a surface water N pollution reduction of 45%; the constructed wetlands implementation scenario could reduce surface water pollution by 43%–64%. Lastly, the technological improvement scenario mitigated atmospheric N pollution by 65%. Technologies play a key role in atmospheric N pollution control, policies mainly contribute to groundwater N pollution control, while technology and policy both work on surface water N mitigation within an urban–rural complex.


Introduction
Human activities contribute more than 50% of the total reactive nitrogen (N R ) inputs to the global budget for the planet's land surfaces [1,2]. Natural factors (geomorphology, climate, meteorology, etc) may drive temporal and spatial heterogeneity of nitrogen (N) cycling between regions [3,4]. However, the magnitude and effectiveness caused by human activity are much larger than those driven by natural processing at a specific location [5,6], especially in an urban-rural complex system (URC), where the city core strongly affects its periphery areas [7]. Differences in environmental N pollution between developed and developing countries indicate that socioeconomic development influences N flux [8]. For example, industrial and domestic point sources of N pollution have been dramatically reduced in developed countries through technology and policy innovation, but remain a serious problem in many developing countries where rapid urbanization is presently occurring [9,10].
This may result from a lack of environmental awareness, appropriate technology, policy, and management to control the spread of pollution in developing countries [5,7].
The role of technology and policy in mitigating N pollution has attracted much attention [5,11], but much of this has focused on cases only in developed countries, such as the Long Island Sound [12] and Chesapeake Bay Program [13], with less attention given to developing countries [5]. Sophisticated environmental technologies and managerial practices in developing countries rely heavily on financial backing [6], making the situation difficult for developing countries. To solve these problems, integrated biogeochemical models based upon real-world cases, such as models applied to some developed countries as they progress [14], are required in developing countries. These models help in systemically analyzing the importance of technological, policy, and economic factors on the temporal and spatial dynamics of N pollution while providing policy makers with a means to achieve optimal regulation.
The southeastern coast, with the fastest economic growth and urbanization, is the most developed as well as polluted region in China [15]. Intensive agriculture and aquaculture in this humid region have a history of several thousand years, and fertilizer over-consumption during the past three decades [16] has led to substantial environmental problems [7]. The Greater Hangzhou Area (GHA) in southeastern China is a typical case of a URC, to illustrate the human mediated N cycling in a highly polluted region (figure 1) [7]. In this study, we: (i) develop a comprehensive N cycling budget (model) for a URC; (ii) quantify the influence of technology and/or policy on environmental N pollution and forecast N pollution dynamics under different scenarios; and (iii) achieve more sustainable N management strategies for developing countries.

System dynamics
A URC contains four functional groups based upon mutual services, with humans as the central node: N-processor, Nconsumer, N-remover, and N-life-supporter groups [7]. The processor group includes six subsystems: cropland, urban lawn, forest-grassland, livestock, aquaculture (freshwater), and industry. These modules can process the fixed N input into the food chain and biomass products that will then be used by consumers. The consumer group includes both the human and pet subsystems since they are the end of the food chain, as well as the end of the industrial chain. The remover group refers to artificial facilities to treat N R waste with processes that convert N R into N g (nitrogen gas) and includes the wastewater treatment subsystem in this study. The lifesupporter group includes the near-surface atmosphere, surface water, and subsurface subsystems that closely relate to all organisms (including humans) within a URC. For the purpose of this study, the simulation timeline dates from 1980 (the point from which rapid economic growth and urbanization occurred in China) to 2050 with a time step of one year (for details see supplementary data available at stacks.iop.org/ERL/6/014011/ mmedia).

System parameterization and calibration
The system parameters can be divided into two types. Type I (available measured data) represents the basic information on the GHA, that is population, cropland area and production, fossil fuel consumption, and wastewater flows. Timeline data from 1980-2004 were obtained directly from statistical documentation (table S1 available at stacks.iop.org/ERL/6/ 014011/mmedia) while values from 2005-2050 are estimated based upon empirical regression (table S4 available at stacks.iop.org/ERL/6/014011/mmedia) or system assumptions. For example, the value for cropland yield was estimated by the regression equation associated with fertilizer application, while the population growth rate was assumed to be decreased from current rates (0.8%) to 0.3% in 2050 [17]. Two-thirds of the historical data were applied to calibration while the remaining one-third were applied to validation (figure 2). Type II (available coefficient data) represents information developed from coefficients that relate to N, e.g., the excretion per capita livestock, the harvest index of crops, the non-symbiotic N fixation rate of cropland, etc, which were obtained from various published peer-reviewed papers or technical reports (for details see supplementary data available at stacks.iop.org/ ERL/6/014011/mmedia).

Validation and sensitivity analysis
The simulated results were validated against N flux historical data to test their accuracy (figure 2). Historical data of NO x emissions, wastewater discharge, riverine N export, and cropland production were used for validation against the simulation results (table S1 available at stacks.iop.org/ERL/ 6/014011/mmedia). NO x emissions (r 2 = 0.94), wastewater emissions (r 2 = 0.95), riverine N exportation (r 2 = 0.95), and cropland production (r 2 = 0.91) all showed satisfactory agreement between the simulation and historical data (P < 0.001, figure 2). Although a system development for this study was used to simulate N dynamics based upon the best available information, the estimations still contained uncertainties due to the input variables, parameters and structure. Based on the Monte Carlo simulation, we found that all output uncertainties are within the mean values ±5% (figure S4 available at stacks.iop.org/ERL/6/014011/mmedia) except for the cropland production, which was under 10% (for details see supplementary data available at stacks.iop.org/ERL/6/014011/ mmedia).
Before scenario analysis, sensitivity analysis was conducted by means of changing one variable while keeping all other variables constant in order to identify the key variables driving N pollution. This demonstrated that (table S5 available at stacks.iop.org/ERL/6/014011/mmedia): (i) the fertilization rate and cropland area had a relatively large effect on subsurface N; (ii) the wastewater treatment plant (WTP) N effluent standard (WNES) and constructed wetlands (CW) were the factors affecting riverine N; while (iii) population growth, fossil fuel consumption, and technological innovation on O x emission intensity were the main influences on atmospheric N pollution.

Scenario preparation
Based on sensitivity analysis (table S5 available at stacks.iop.org/ERL/6/014011/mmedia) and the potential changing ranges of different parameters, we developed seven scenarios to quantify the effects of technology and policy on environmental N dynamics (table 1). In scenario 1 (S1), we introduced CW, a novel technology for environmental N pollution control [18,19], to our system. The effect of CW on N pollution control from agriculture and WTP in a URC (figure S1 available at stacks.iop.org/ERL/6/014011/mmedia) was quantified with a N removal rate in the range ∼50-80% under different engineering design and management [18,19] (S1.2-1.3). Scenario 2 is technological innovation on NO x emission intensity, varying across different countries [20], from fossil fuel consumption. The NO x emission per GDP in China is currently 14.5 kg per 1000 USD (S2.1), much higher than the average level seen in mid-level developed countries (1.9 kg per 1000 USD, S2.2) and developed countries (1.0 kg per 1000 USD, S2.3) [20]. Scenario 3 is fertilization reduction (S3.2-3.3), since the N fertilization rate is currently ∼50% higher than the optimum value recommended within the GHA [7,16]. Scenario 4 is cropland area decrease, due to population growth and economic development (S4.2-4.3). Scenario 5 is WNES revision (S5.2-5.3), since the effluents from WTP have been one of the main sources of N pollution to the regional environment [7,21]. Scenario 6 is the population growth rate changing. It may slow down with a 'one child policy' and natural landscape protection or increase with rapid industrial development that could possibly attract immigration (S6.3). Scenario 7 is fossil fuel consumption rate reduction (S7. 2-7.3) according to the Gothenburg Protocol adopted in 1999 to abate acidification and eutrophication [22].

Subsurface N
The subsurface N accumulation rapidly increased from 4. Under the fertilization reduction scenario (S3.3), with the synthetic fertilizer application rate reduced from 400 to 200 kg N ha −1 yr −1 , subsurface N is predicted to decrease by 17% compared to the BAU scenario, while food production would decrease by approximately 4%.
If the cropland area is decreased by ∼50% due to urban expansion (S4.3), subsurface N would decrease by ∼21% ( figure 4). This scenario, however, reduces production by 36%, and weakens the self-sufficiency of the GHA in terms of food production while increasing food importation by 125%.

Riverine N
Riverine N export rapidly increased from 18.9±0.53 to 69.  Under the WNES revision scenarios (S5), effluent N concentration reduced from the current level of 18 to 8 mg l −1 or even as low as 3 mg l −1 , riverine N export would decrease by 30% or 45% compared to the BAU scenario, respectively (figure 4). If a 3 mg l −1 WNES was put into practice (S5.3), the average riverine N concentration would be less than 0.8 mg l −1 .
Under the CW application scenarios (S1), CWs intercept N transported from agricultural non-point pollution (cropland and aquaculture), livestock factories, and effluent from WTPs (S1) and reduce riverine N export by ∼43-64% (figure 4). The average riverine N concentration would be less than 0.5 mg l −1 if the CW application scenario (S1.3) were enacted. Under population scenarios, population growth rate changes (S6) contribute less than 10% variation to atmospheric N pollution. Fossil fuel reduction (S7) by 30-50% compared to the BAU scenario would decrease atmospheric N pollution by 21-35%, respectively (figure 5).

Atmospheric
Under technology innovation scenarios (S2), the average atmospheric N pollution would decrease by ∼60% and ∼65%, respectively, when the NO x emission intensity within the GHA would reach the mid-and high-level strata seen in developed countries. In addition, combining the consumption reduction scenario (S7) and technology scenario (S2), average atmospheric N pollution would decrease by ∼65-68% (figure 5).

Coupling WTP and CW. The vast investment in WTP
has dramatically reduced the industrial and domestic pointsource N pollution in many parts of developed countries [6,8]. With BAU, WTP effluent within the GHA will contribute more than 50% of the surface water N pollution by 2050; combining the discharge of wastewater that is not being treated, wastewater N will dominate 65% of all surface water N pollution. High levels of precipitation force water cycling to couple with the point of N pollution, ultimately leading to widespread N pollution [7]. By 2005, N pollutants from the GHA emitted to the East China Sea via the Qiantang River, coupled with N pollutants from nearby URCs, have created a polluted coastal area greater than 65 000 km 2 within the East China Sea ( [23], figure S3 available at stacks.iop.org/ ERL/6/014011/mmedia). The total polluted coastal area in China reached 139 000 km 2 in 2005 [24]. If no territorial surface water pollution controls were in existence, the East China Sea would be the next 'dead zone' (a condition caused by N pollution transferred from land to ocean that has led to the excess algae bloom seen in the Gulf of Mexico costal ecosystems [25]).
Fortunately, an emerging low cost and high efficiency ecological technique, CW, may help to solve this problem. The CW technique works well for wastewater N removal by means of microbial nitrification/denitrification, plant uptake, and substrate absorption [19]. CWs have been used for wastewater treatment for several decades and still work well [18]. The average N removal rate of CWs is ∼50% in China currently [19], and ∼80% N removal rate is observed in other parts of the world due to better design and management [18]. By coupling WTPs and CWs (e.g., CWs are placed the downstream from WTPs), surface water N pollution would decrease by ∼25-40% by 2050 within the GHA (figure 4, S1). Although uncertainties exist (figure S4 available at stacks.iop.org/ERL/6/014011/mmedia), with the technology improvement [18], CWs would play a better role in developing countries where more than half of the wastewater is not being treated [6,19].

4.1.2.
Reducing fuel-N. Atmospheric N pollution in developed countries has been somewhat reduced via technological improvements despite large increases in burning fossil fuels [11,26]. For instance, the N deposition rate has been reduced from over 10 kg N ha −1 yr −1 in the 1970s to 3.0 kg N ha −1 yr −1 in the United States of America, and 6.8 kg N ha −1 yr −1 in Europe currently [27]. However, it remains a serious issue in China [15]. The total N deposition rate was as high as 22.6 kg N ha −1 yr −1 in 2004 within the GHA [7], about twice the average in China (12.4 kg N ha −1 yr −1 [28]). Moreover, atmospheric annual average NO 2 concentrations within the GHA amounted to 0.055 mg m −3 [7], which is beyond the clean air quality standard (0.04 mg m −3 ) of China [29]. Fossil fuel combustion was the largest source of atmospheric N pollution after 2000 within the GHA (figure 3), and NO x emissions per GDP were greater by a factor of ∼7-15 compared to those seen in developed countries [20]. Reasons for this include: (i) energy consumption per GDP is high [30]; (ii) coal makes up 60-70% of total energy consumption and it emits relatively more NO x during combustion [15]; and (iii) techniques of N removal from coal are still less developed [7]. Technology improvement in N removal from coal, coal-burning economy transition, and enhancing the efficiency of energy use [11,30], should be attained.

Policy drivers in mitigating N pollution
4.2.1. Precision agriculture. The development of precision agriculture and new crop varieties have resulted in much larger grain harvests with only slight increases, or in some cases a decrease, in the use of synthetic N fertilizer in developed countries since the 1980s [26,31]. In the case of China with its enormous population and limited farmland, high yields are the first priority. This, in combination with the extra 'insurance' of fertilizer application, leads to over-fertilization [16]. In 2004, for example, the intensive double-cropping system within the GHA consumed 400 kg N ha −1 yr −1 synthetic fertilizer and 150-200 kg N ha −1 yr −1 manure (mainly human and livestock excretion). The total applied N reached as high as 275-300 kg N ha −1 per crop, about twice that in Europe and the United States of America (where 150-180 kg N ha −1 per crop is typical [5,31]). Fertilizer over-consumption was as high as 70% according to the optimal fertilization rate set by Ju et al [16], and led to serious non-point source pollution [7].
Although there are still no policies that control overfertilization in China, as is also true in most of the world, whether developed or developing [5], the national 'Prescription Filled According to Soil Test Result in Fertilization' (PSTF) program was implemented from 2005 on [32], supplying basic data and a foundation for policy establishment. Increasing R&D funding for technology exploitation, reducing or abolishing fertilizer subsides, and introducing N fertilizer overconsumption taxes [16] based upon the PSTF would all help to achieve the fertilizer reduction scenarios within the GHA (S3). In addition, re-coupling animal and crop production systems can increase the agricultural N use efficiency [11]. Improving local extension services, providing knowledge and training for farmers [16], and providing environmental awareness education for the public [14] would ultimately help to avoid serious environmental degradation.

Feasibility of effluent standards revision and CW applications.
As mentioned above, the WTP tertiary treatment is currently unfeasible in China due to its high cost [30]. For instance, the GHA would need to invest 1.7 billion USD, ∼15% of the total GDP, to update its WTP system and reach 3 mg l −1 WNES (S5.3) according to the cost budget carried out by the CBF [21]. This is the reason for the current low percentage of wastewater being treated and the low WENS within the GHA. Prompted by the 'Taihu Lake water pollution incident' in 2006 and 2007 [33], governments revised the WNES of newly established WTPs from 20 to 15 mg N l −1 in this region. However, even if this standard were followed, the WTP effluent N to surface water would continue to grow and become the primary pollution source in the next two decades ( figure 3). Therefore, governments should strictly revise the WNES to a level they can afford, e.g., 8 mg N l −1 (S5.2), which can lighten the burden on surface water.
Constructed wetlands have been highlighted as more economical and feasible than WTPs in China [19]. The cost of building a CW is only one-third to one-half the cost of a WTP while achieving the same wastewater treatment capacity [19]. Moreover, the costs of operation and maintenance are less than one-tenth of those of WTPs [34]. Although CWs require three to five times larger area than WTPs to remove an equal quantity of N R (table S4 available at stacks.iop.org/ERL/ 6/014011/mmedia), they can provide additional ecosystem benefits such as bioenergy resources and habitat conservation zones for birds while providing recreational and amenity usage [19,34]. Meanwhile, CWs can work well on pollution control through midcourse N interception (i.e., intercept N before it is discharged to small watersheds from agricultural and domestic activities) [18,19]. The CW application scenario (figure 4; S1) requires 2.8-5.5% of cropland area (0.28-0.55% of the total land area) within the GHA to intercept N runoff from agricultural sources and WTP by 2050. Considering the unused marginal land, it is possible for the GHA to use less than 5% of its cropland area for CW applications.

Clean air act. Atmospheric pollution control in
China has shown some initial success, especially concerning SO 2 emission control. However, the recent change in acid rain chemistry from a sulfate-type to nitrate-type indicates an accelerating atmospheric N problem [15]. In 2009, China's central government made atmospheric N control the primary environmental protection goal of the 12th '5-Year Plan' (2011)(2012)(2013)(2014)(2015). New atmospheric quality and N emission standards are ongoing as well. These 'top-down' policies will drive the technological improvement (S2) by means of attracting more financial support to upgrade techniques and equipment. Additionally, the management of multi-source N emission is essential [2,22] since fossil fuel combustion is predicted to contribute only 10% of N air pollution while other sources will contribute more than 80% within the GHA by 2050, after achieving the mid-level technology seen today in developed countries (figure 5).

Policy and technology collaboration
Our results provide a case study on how technology and policy can work toward pollution control in a URC, where pollution would extend beyond the URC borders and affect nearby regions, e.g. water pollution of an upstream URC affects a downstream URC located in the same watershed. Research results from Chesapeake Bay [12] and other areas in the US [4,13] supply good examples of policy collaboration between multiple administration units on pollution control at the watershed level. As opposed to regional policy collaboration, technology cooperation usually happens at national and global scales [11]. The poor pollution control technology and deficient financial support in developing countries compared to the advanced technology and sufficient financial support in developed countries reveals that there is enormous potential for international collaboration to protect our environment [22]. A successful example of such a collaboration is the CW technology transfer. The CW system was introduced to China from Europe beginning in 1996, and has been applied in many areas in China [19]. In addition, the environmental impacts would extend beyond national borders and affect the entire world. For example, GHA's riverine N export greatly contributes to the pollution of the East China Sea [23], air N pollutants are propelled eastwards by prevailing winds, reaching North America [30], contributing to the global change, and regional N 2 O emissions lead to global warming [22]. International collaboration, therefore, must be strengthened to combat environmental pollution globally.