Characteristics of Atmospheric Reactive Nitrogen Deposition in Nyingchi City

Atmospheric reactive nitrogen (N) deposition has been proven to be an important nutrient input from external environments to forest ecosystems. However, the magnitude of atmospheric N deposition in the Tibetan region of China is not well known. In this study, multi-year (between 2005 and 2016) measurements of dry and wet N deposition were carried out in Nyingchi (NC) city, southeastern Tibet. Bulk deposition was collected by the rain gauge method; dry deposition was calculated by the inferential method, namely, multiplying ambient N concentrations by dry deposition velocity (Vd) of the N species. During the entire period, annual bulk and dry N deposition fluxes averaged 2.19 and 1.85 kg N ha−1 yr−1, respectively. Total N deposition fluxes (the sum of reduced and oxidized N species in dry and bulk deposition) showed an obvious increasing trend, especially for oxidized N species. Both bulk and dry N deposition showed a consistent seasonal pattern, with the highest fluxes in summer and the lowest in winter. Our findings suggest that N deposition to the urban environment in southeast Tibet has recently shifted from ammonium-dominated to nitrate-dominated conditions.


Results
Monthly N concentrations in rainwater and bulk deposition fluxes in NC City. Monthly volume-weight mean concentrations were 0.14-0.68 mg N L −1 for NH 4 + and 0.09-0.39 mg N L −1 for NO 3 − in rainwater (Fig. 1a). Monthly bulk deposition fluxes of NH 4 + and NO 3 − were in the ranges of 0.01-0.26 kg N ha −1 and 0.01-0.16 kg N ha −1 , respectively (Fig. 1b). Bulk N concentrations showed an obvious seasonal variation, with the lowest values in summer and the highest in winter. By contrast, bulk deposition fluxes showed an opposite seasonal behaviour to concentrations. www.nature.com/scientificreports www.nature.com/scientificreports/ Monthly atmospheric N concentrations and dry N deposition fluxes in NC city. Monthly mean concentrations of atmospheric NH 3 , HNO 3 , pNH 4 + , pNO 3 − , NO 2 were been monitored over three years (2009,2015 and 2016) (Fig. 2a,b). Monthly mean concentrations of NH 3 and pNH 4 + were in the ranges 0.39-2.41 and 0.18-1.01 μg N m −3 , respectively. The highest monthly NH 3 concentration appeared in April, approximately 6 times higher than the lowest value observed in December. The highest monthly pNH 4 + concentration occurred in July, approximately 5 times higher than the minimum value in January. Monthly mean concentrations of NO 2 , pNO 3 − and HNO 3 were in the ranges of 0.65-1.26, 0.16-0.33 and 0.01-0.08 μg N m −3 , respectively, and the highest monthly concentrations of NO 2 , pNO 3 − and HNO 3 appeared in August, August and June, respectively. The lowest monthly concentrations of NO 2 , pNO 3 − and HNO 3 were observed in January, April, and December, respectively.
Monthly deposition fluxes of NH 3 and pNH 4 + were in the ranges 0.02-0.17 and 0.01-0.05 kg N ha −1 , respectively. Monthly deposition fluxes NO 2 , pNO 3 − and HNO 3 were in the ranges 0.01-0.03, 0.01-0.02 and 0.01-0.02 kg N ha −1 . Both reduced N and oxidized N deposition fluxes showed an increasing trend in the front six months and a decreasing trend in the later six months in a year. Furthermore, NH 3 deposition fluxes dominated reduced N deposition in all months, but oxidized N had no unique variations, NO 2 deposition fluxes were the main oxidized N species from April to October and pNO 3 − deposition fluxes was main oxidized N species in January, February, March, November and December (Fig. 2c, Relative N deposition source attribution. Fertilizer use, transportation and livestock were contributed largely to total N deposition, with the percentages of 26.3%, 16.5% and 15.7%, respectively. Other sources, including human waste, residential activities, soil, lighting and biomass burning, altogether contributed 34.5% of the total N deposition. Industry and power had minor contribution to total N deposition, with percentages of 4.2% and 2.9% (Fig. 4).

Discussion
Lots of studies have revealed that the lowest concentrations of N species in bulk precipitation are associated with the highest precipitation rates in summer 18,19 , furthermore, atmospheric N bulk deposition fluxes are positively correlated with local precipitation 20 . Monthly precipitation and bulk DIN deposition fluxes were found to be positively correlated (R 2 = 0.976, P < 0.01) (Fig. 1c). Furthermore, bulk DIN deposition fluxes summed 2.06 kg N www.nature.com/scientificreports www.nature.com/scientificreports/ ha −1 from March to October (Fig. 1b), which accounted for 94% of annual bulk DIN deposition, suggesting that bulk deposition mainly occurred in the plant growth season.
Seasonal variations of atmospheric N species concentrations have been investigated in many studies 18,19,21 . High atmospheric NH 3 concentrations in summer could be a result of agricultural activities (e.g., N-fertilization) and seasonal changes in temperature 22,23 . Unlike NH 3 , NO 2 concentrations exhibited a weaker seasonal variation, because they mainly come from industrial and transportation emissions which are non-seasonal emission sources 24 . Atmospheric NH 3 concentrations were relatively high from April to August (Fig. 2a), which corresponds to local agriculture activities. It is interesting that ambient NO 2 concentrations were obviously lower in January and February (Fig. 2b), in which relatively higher NO 2 concentrations were found in other regions of China 20 . This is because, on one hand, there was no winter domestic heating in NC City. On the other hand, the city focused on tourism and there were few people in winter. As a result, transportation emissions as the main NO 2 source were low in these two months.  www.nature.com/scientificreports www.nature.com/scientificreports/ Zhu et al. 16 reported that the average wet N deposition flux was 13.69 kg N ha −1 yr −1 in China. Our measured bulk N deposition flux in Nyingchi city was 2.19 kg N ha −1 yr −1 (Fig. 3a), just one-sixth of the national level. Not surprisingly, wet N deposition fluxes were influenced by precipitation and different kinds of atmospheric N concentrations 20 . It has been commonly accepted that atmospheric N concentration can be largely affected by anthropogenic activities 9 ; the Qinghai-Tibet plateau was considered a region of less human disturbance and a low N deposition area 9 . It could be seen that both NO 3 − and NH 4 + concentrations in rainwater were much lower than in other areas (Table 1), partially resulting in lower N deposition fluxes. At a large spatial scale, Liu et al. 17 reported that wet N deposition fluxes in the Tibetan Plateau were 0.51 kg N ha −1 yr −1 for NO 3 − and 1.06 for kg N ha −1 yr −1 NH 4 The mean concentration and dry deposition fluxes of NH 3 , HNO 3 , pNH 4 + , pNO 3 − and NO 2 were all lower than the mean values reported across China 20 . Regions with high atmospheric N concentrations were usually coupled with intensive agricultural activities, high population density and a rapidly developing economy. However, the differences of N deposition fluxes in different regions are due to large differences in economic function, weather condition and land use type 34 . Compared with concentrations and deposition fluxes in other areas (Table 2), all types of N species were much lower than in other regions. In addition, NH 3 and NO 2 dominated atmospheric N in NC city, the same as in other areas 5 .
With rapid economic development, NH 3 and NO x emissions increased continuously and N deposition fluxes were enhanced on a national scale 6,9 . Compared with dry N deposition fluxes in 2009, they increased significantly in 2015 and 2016 ( Fig. 3b) in NC city; furthermore, deposition fluxes of oxidized N increased more obviously than that of reduced N from 2009 to 2016. We used statistics from National Bureau of Statistics of Tibet to estimate the possible N source in this study ( Table 3). The amount of agricultural fertilizer application and agricultural N fertilizer application, as the main NH 3 source, increased 1.34 × 10 4 t and 0.38 × 10 4 t from 2009 to 2015, an approximately 28% and 22% increase, respectively; urban population increased from 0.66 million to 0.9 million, an approximately 36% increase; vehicles were considered to be the largest source of NO x , private cars and operated cars (bus or truck) increased 17.38 × 10 4 and 2.75 × 10 4 , an approximately 186% and 123% increase. The GAINS model (http://www.iiasa.ac.at/) showed that NH 3 pollution was significantly higher than NO x pollution in Tibet as well, but the increased magnitude of NO x was higher, NH 3  Xu et al. 25 reported that fertilizer use and transport were the main N sources in China and Yangtze River basin based on GEOS-Chem (http://geos-chem.org) model; similar results were founded in this study, fertilizer use (26.3%) and transport (16.5%) were the main N deposition source at our monitoring site. In addition, livestock was the main N deposition source in this study. This could be explained by the Nr emission source, as shown in Fig. 4, fertilizer, transportation and livestock were all dominated Nr emission sources at our monitoring site (Table 4). Interestingly, others (including human waste, residential activities, soil, lighting and biomass burning) contribution to total N deposition were reached 34.5%, One explanation was that there was no obvious N pollution source was found around NC City; as a result, natural N sources which are unchangeable to some extent in different areas, tend to be more important. The same result was also found in a previous study 25 .
Dry N deposition fluxes were estimated by different N species concentrations multiplied by their deposition velocities in our research. As in previous studies 18,19 , different N species velocities have monthly variations, which were high in summer and low in winner. The same variations were found in this study; furthermore, the V d of NH 3  www.nature.com/scientificreports www.nature.com/scientificreports/ and 0.01-0.10 cm s −1 , respectively. In conclusion, monthly dry N deposition flux variations were determined by monthly variations in both N concentrations and deposition velocity.
Since bulk N deposition was measured directly and gaseous dry N deposition was estimated based on modeled V d , bulk N deposition flux was more accurate than dry deposition fluxes in this work. However, the rainfall amounts were so low in some precipitation events that that could not finish a complete chemical analysis could not be achieved (<20 ml), due to the high N concentration in low precipitation events, and bulk N deposition flux might be underestimated in this work. Furthermore, rain gauges were continuously opened during the experiment period, and some dry N deposition fluxes were included in bulk N deposition estimations. Clearly, if missing rainfall samples were calculated and additional dry N deposition fluxes were excluded, the assessment of N deposition flux could be more precise.   www.nature.com/scientificreports www.nature.com/scientificreports/ For dry N deposition flux measurements, both the collection bucket method and the DELTA system were widely used. Dry N deposition fluxes collected by the former were lower than the latter because of the particle NH 4 NO 3 volatilization 39,40 . However, the particle cut-off size for the DELTA system was 4-5 μm, and large particulate matter of diameter >10 mm is deposited by gravity, defined as sedimentary deposition 41 . In seasons with less rainfall, large atmospheric particles were suffused and dry N deposition collected by a bucket could be more effective. In addition, NH 3 flux is bi-directional, often both emitted from and deposited to land or plant canopy, and net deposition flux is hard to accurately determine in this experiment. Moreover, modeled dry deposition fluxes are very uncertain and can seldom be compared to measurements 40 . Deposition velocities of different N species were influenced by local meteorological conditions and the underlying surfaces; the meteorological conditions in NC City frequently changed and the microclimate was complex, which bring trapped for model V d estimation in our research area.
Total N deposition fluxes reached 4.75 kg N ha −1 yr −1 in NC city in 2016, it is no doubt that atmospheric N pollution in NC city was not serious. However, it great attention should be paid to the fact that N had the highest efficiency when atmospheric N deposition was between 4 and 10 kg N ha −1 yr −1 for plant growth and carbon sequestration 15 . Numerous simulated N deposition experiments have been conducted in forest ecosystems and have shown that additional N input could promote plant growth and carbon sequestration 14,[42][43][44] ; when total N deposition ranged between 3-11 kg N ha −1 yr −1 , the annual above-ground growth of surviving trees and net annual above-ground carbon increment increased approximately 0.6 t C ha −1 yr −1 , a nearly 5.5% increase per kg N ha −1 yr −1 45 . Magnani et al. 46 also suggested that net ecosystem production in forests showed small changes when N deposition flux was lower than 3 kg N ha −1 yr −1 ; however, that value could be increased from less than 1 t C ha −1 yr −1 to more than 4 t C ha −1 yr −1 when N deposition fluxes ranged from 3-9 kg N ha −1 yr −1 . As total N deposition flux reached 4.75 kg N ha −1 in 2016 and might continue increasing in coming years, our results suggested that more carbon could be fixed in southeast Tibet.
In conclusion, a long-term N deposition monitoring site was established in NC City to measure dry deposition for 3 years (2009,, and bulk deposition for 6 years (2005-07, 2009, 2015-16). Averaged bulk N deposition flux was 2.19 kg N ha −1 yr −1 and averaged dry deposition flux was 1.85 kg N ha −1 yr −1 in NC City. Atmospheric N concentration and deposition flux were largely enhanced. Both atmospheric reduced N and oxidized N concentration increased with local economic development. Although N deposition was still dominated by reduced N, oxidized N deposition fluxes increased more obviously. In general, atmospheric N deposition flux was 4.75 kg N ha −1 yr −1 in NC City in 2016, and the influence of N deposition on the local forest ecosystem should be given attention, as much as plant growth and carbon sequestration.

Materials and Methods
Site description. The monitoring site was located at Xizang Agriculture and Animal Husbandry College (29.66°N 94.34°E 2990 m a.s.l.), NC city (Fig. 5). NC City is located beside the Niyang River, which is one of the main tributaries of the Brahmaputra. The climate is mainly dominated by the Indian Ocean and the Pacific warm current. The regional climate is the cold temperate humid subalpine zone, with weather data presented in Table 5. Tourism is the local economic pillar industry. There was no heavy industry nearby, and potential emission sources were small villages and agricultural fields.
Collection of N compounds in air. Ambient concentrations of gaseous (NH 3 , HNO 3 , NO 2 ) and particulate (pNH 4 + and pNO 3 − ) N compounds were measured for three years (2009, 2015 and 2016) using an active sampling system (DEnuder for Long Term Atmospheric sampling-DELTA) 41 . The system was based on a set of bore glass denuder traps through which a laminar flow of air is driven by a pump (pumping rates were 0.3-0.4 L min −1 ) set at the end of the system. A first set of denuders was coated with citric acid to capture gaseous NH 3 , and a second set was coated with potassium hydroxide to trap gaseous HNO 3 . Aerosols pass through the denuders without reacting and then are collected by a couple of filters placed downstream of the denuders. These filters were treated with the same alkaline (for pNH 4 + sampling) and acid (for pNO 3 − sampling) solutions as the denuders. Ambient  Table 4. Annual total NH 3 and NO x emissions over Qinghai-Tibet plateau and monitoring site (kg N ha −1 yr −1 ). a Fertilizer NH 3 emissions include both chemical fertilizer and manure fertilizer. b Fuel combustion of NH 3 emissions include emission from power plant, industry, transportation and residential. c Natural NO x emissions include emissions from soil, lighting and biomass burning.
www.nature.com/scientificreports www.nature.com/scientificreports/ NO 2 concentrations were measured using Gradko passive diffusion tub with three replicates. The DELTA system and Gradko NO 2 tubes were placed at a height of 1.5 m above the ground, and these samplers were exposed to ambient air for one month (one sample per month). After sampling, exposed DELTA sampling train and NO 2 diffusion tubes were returned to the laboratory for sample analysis. precipitation collection. A rain gauge was installed beside the DELTA to measure monthly bulk deposition, at a height of 1.5 m above ground level. Rainfall was collected immediately after each precipitation event, and then transferred into 100 ml polyethylene bottles. The precipitation samples were thoroughly mixed and transferred to clean polyethylene bottles (50 ml), and then stored at −20 °C until chemical analysis within one month. After collection, the rain gauge was rinsed with deionized water. The measurements were performed for the 6 years (2005-2007, 2009, 2015, and 2016).   www.nature.com/scientificreports www.nature.com/scientificreports/ Analytical procedures. The exposed DELTA sampling trains and NO 2 diffusion tubes were sent to China Agriculture University after each monthly collection was finished, then stored at 4 °C and analysed within one month. Different N species were extracted by different solutions, gaseous NH 3 was captured and pNH 4 + in alkaline filters in the DELTA system were extracted by 0.05% H 2 O 2 solution; gaseous HNO 3 was captured and pNO 3 − in acid filters in the DELTA system were extracted by high-purity water. Nitrate (NO 3 − ) and ammonium (NH 4 + ) in the extracted and filtered solutions were measured by an AA3 continuous-flow analyser (Bran + Luebbe GmbH, Norderstedt, Germany), the detection limits of NH 4 + and NO 2 − were 0.01 mg N L −1 . The blank samples were also analyzed to eliminate the errors in the experimental process.
The stainless steel mesh in the Gradko passive diffusion tube was leached with a mixed solution of sulfonamide, phosphoric acid and N-1-naphthylethylen-diamina (NEDA), and the concentration of NO 2 − was determined by colorimetry at a wavelength of 542 nm. The blank sample was analysed in the same way to eliminate errors during the experiment.
Precipitation samples were filtered with a 0.45 µm syringe filter before determination each month. NH 4 + and NO 3 − concentrations in the filtrate were measured using the AA3 continuous-flow analyser as described above. where D w refers to the mean monthly bulk NO 3 − -N or NH 4 + -N deposition (kg N ha −1 month −1 ); C w is the monthly volume-weighted mean concentration of NO 3 − -N or NH 4 + -N in bulk precipitation (mg N L −1 ); P t is the mean monthly precipitation (mm); 100 is the unit conversion factor.

Estimation of bulk
Estimation of dry N deposition flux. Dry deposition fluxes were estimated by different kinds of atmospheric N concentrations multiplied by their deposition velocities (V d ), as has been widely done in previous studies 20,21 . Monthly averaged atmospheric N concentrations species measurements were as described above. Monthly averaged V d was calculated by the GEOS-Chem chemical transport model (http://geos-chem.org) 47 . The GEOS-Chem CTM is driven by GEOS-5 assimilated meteorological data from the NASA Global Modeling and Assimilation Office (GMAO) with a horizontal resolution of 1/2° latitude × 2/3° longitude and 6 h temporal resolution (3 h for surface variables and mixing depths) 20 .
Source attribution of N deposition. The GEOS-Chem (http://geos-chem.org) model was used to estimate the emission sources of N deposition in NC City, which has been widely used to assess the contribution of different N emission source across China 26,48 . The Nest GEOS-Chem atmospheric chemistry model was used to test the model sensitivity driven by the GEOS-5 assimilated meteorological fields at a horizontal resolution of 1/2° × 2/3°. Based on the Multi-Resolution Emission Inventory of China in 2010 (MEIC, http://meicmodel.org), anthropogenic emissions except NH 3 were used in this research; NH 3 emissions were based on the Regional Emission in Asia (REAS-v2) inventory and improved by Zhao et al. 48 . Details of the model emissions and mechanisms have been provided in Zhao et al. 48 . The NH 3 and NO x emissions over the Qinghai-Tibet pleateau are 1.88 kg N ha −1 yr −1 and 0.69 kg N ha −1 yr −1 in 2010. Fertilizer and livestock were dominant NH 3 sources, accounting for 34% and 53% of the NH 3 emissions, respectively. Natural and transportation were main NO x sources, accounting for 54% and 29% of total NO x emissions, respectively. The NH 3 and NO x emissions were 2.45 and 0.45 kg N ha −1 yr −1 at our monitoring site, respectively. Fertilizer (71%) was the donimated NH 3 emission source and transportation (58%) was the dominant NO x emission source ( Table 4).
The data on possible N emission sources (urban population; agricultural fertilizer application; emissions of nitrogen oxides; waste water ammonia emissions and total nitrogen emissions from wastewater) were obtained from the National Bureau of Statistics of Tibet (http://data.stats.gov.cn/easyquery.htm?cn=C01). statistical analysis. All statistical analyses were performed using SPSS 20.0 (SPSS Inc., Chicago, IL, USA), with significance level set at P < 0.05. All the data were shown as the mean ± standard error (SE) of corresponding months during the observation period. The periods of March-May, June-August, September-November, and January-December were defined as spring, summer, autumn and winter.