Nitrogen atmospheric deposition driven by seasonal processes in a Brazilian region with agricultural background

Understanding the seasonal patterns and influencing factors of nitrogen atmospheric deposition is essential to evaluate human impacts on the air quality and nitrogen biogeochemical cycle. However, evaluation of the nitrogen deposition flux, especially in South America agricultural regions, has not been fully investigated. In this paper, we quantified the atmospheric wet deposition fluxes of total dissolved nitrogen (TDN), dissolved organic nitrogen (DON), and dissolved inorganic nitrogen (DIN), in a region with agricultural and livestock predominance in the Southern Minas Gerais region, Brazil, from May 2018 to April 2019. Deposition fluxes of nitrogen species in the wet season (October–March) were on average 4.8-fold higher than those in the dry season, which revealed significant seasonal variations driven largely by the seasonality of rainfall and agricultural operations. We also found high NO3−/NH4+ ratios (average = 8.25), with higher values in dry season (NO3−/NH4+  = 12.8) in comparison with wet season (NO3−/NH4+  = 4.48), which revealed a higher relative contribution of NOx emissions from traffic sources in dry season. We also estimated the influence of atmospheric deposition of inorganic nitrogen (N-DIN) on environmental ecosystems, being 2.01 kgNha−1 year−1 with potential risk of acidification and eutrophication of 30%. Therefore, attention should be paid to the role of wet atmospheric deposition of nitrogen as a source of nitrogen environmental pollution in agricultural regions.


Introduction
The nitrogen cycle is essential to living organisms and triggers several natural reservoir processes, and in the last centuries, such processes have been changed and ascribed to anthropogenic source inputs (Jaffe and Weiss-Penzias 2003). From the local to global scale, several activities, such as fossil fuel combustion, mobile exhaust engines, and agricultural activities, including fertilizer use and livestock husbandry, account for specific variations and the increase of atmospheric reactive nitrogen (Nr) emissions (Liu et al. 2013;Xing et al. 2018). Galloway et al. (2004) estimated that NO x and NH 3 global atmospheric emissions will increase from 23 TgNyear −1 in 1860 to 189 TgNyear −1 in 2050. The main forms of Nr in the atmosphere are divided among (i) the inorganic reduced forms of nitrogen (e.g., NH 3 , NH 4 + ); (ii) inorganic oxidized forms (e.g., NO x , HNO 3 , N 2 O, NO 3 − ); and (iii) organic compounds (e.g., urea, amines, proteins, nucleic acids) (Galloway et al. 2004;Xing et al. 2018;Zeng et al. 2020). These species may be considerably modified by (photo)chemical reactions and enter into biogeochemical cycles of terrestrial and aquatic ecosystems (Qiao et al. 2018).
The nitrogen species analysis through wet, dry, or bulk deposition is a key process in better understanding Responsible Editor: Gerhard Lammel * Marcelo Vieira-Filho marcelo.filho@ufla.br the human impacts on the nitrogen biogeochemical cycle (Wang et al. 2013;Song et al. 2017;Souza et al. 2020).
Since net primary production of most terrestrial ecosystems is limited by nitrogen availability, deposition of reactive nitrogen becomes a source of nutrients in these sites and could improve productivity (Dentener et al. 2006;Vet et al. 2014;Xu et al. 2020). In contrast, the excessive nitrogen input may cause acidification of forest soils, eutrophication, unbalance, and decreases in biodiversity, and enhanced greenhouse gas emissions (Bobbink et al. 2010;Stevens et al. 2018;Wang et al. 2018). During the wet atmospheric deposition, nitrogen is supplied in soluble form (total dissolved nitrogen (TDN)), which encompasses dissolved inorganic (DIN: NH 4 + + NO 3 − ) and organic nitrogen (DON) (González Benítez et al. 2009;Violaki et al. 2010;Rasse et al. 2018). TDN atmospheric deposition has been extensively investigated worldwide (Zhang et al. 2008;Violaki et al. 2010;Hofhansl et al. 2011;Zamora et al. 2011;Matsumoto et al. 2020). Cui et al. (2014) and Tu et al. (2013) reported that wet deposition fluxes of TDN in agricultural and forest ecosystems in China ranged from 37.37 to 113.8 kgNha −1 year −1 , respectively, with DON accounting for approximately 26% of TDN. In Brazil, studies have shown that both urban and forest regions receive a considerable TDN deposition (5-18 kgNha −1 year −1 ) by bulk deposition (Schroth et al. 2001;Parron et al. 2011;Araujo et al. 2015;Souza et al. 2015). According to these sites, DON deposition ranged between 30 and 58% of TDN, highlighting its potential importance to N cycling in different ecosystems. DON contribution to TDN varies due to local sources, seasonal patterns, and mixed sources; thus, it is desirable to evaluate the partition from natural to anthropogenic sources and their environmental impacts. Moreover, studies from 2010s highlighted some of these mechanisms and pointed out the ubiquitous role of DON in atmospheric deposition Xing et al. 2017;Rasse et al. 2018).
Despite the fact that nitrogen atmospheric deposition has been adequately estimated in Asia, Europe, and North America through monitoring networks , available data for many Southern hemisphere developing countries are either scarce or non-existent (Vet et al. 2014). As it pertains to Brazil, nitrogen atmospheric deposition data comes from private and isolated initiatives or from research groups limited to specific regions for restricted periods, which impoverishes the global analysis of the nitrogen input impacts in ecosystems of the said country (Parron et al. 2011;Araujo et al. 2015;Souza et al. 2015;Vieira-Filho et al. 2016). In this perspective, we evaluated and quantified the atmospheric deposition fluxes of TDN, DIN, and DON in a region with agricultural influence in the Southern Minas Gerais region, Brazil.

Sampling site
The state of Minas Gerais is located in the Southeastern region of Brazil, where the largest metropolitan regions of the country are located, e.g., São Paulo, Rio de Janeiro, and Belo Horizonte (Fig. 1). The major atmospheric pollutant emissions in this region are related to transport, farming, biomass burning, and industrial activities such as mining, metallurgical, agroindustrial, and chemical facilities . This study was conducted in the Southern Minas Gerais state region, which encompasses 21.8% of agricultural commodities (mainly from coffee crop production) and accounts for 12% of the state's gross domestic product (Almeida et al. 2017).
The sampling collection was conducted specifically in Lavras city (21° 13′ 45.3″ S and 44° 58′ 32.4″ W), 241 km from the Atlantic Ocean ( Fig. 1). Lavras has an area of 564.744 km 2 , 919 m of altitude, and a population of 102,728 inhabitants, occupying the fifth place among the most populous cities in the Southern region of Minas Gerais (IBGE 2018). Approximately 19% (107 km 2 ) of the total area of Lavras is associated with agricultural activities, mainly coffee production (IBGE 2019). IBGE (2014) estimated that an average synthetic nitrogen fertilizer application rate in Minas Gerais is approximately 110.4 kgha −1 year −1 . Lavras city vehicular fleet counts approximately 50 thousand light-duty vehicles, comprising 54% automobiles and 26% motorcycles. Moreover, its vehicle fleet is about 15 years old on average, with 62% of its passenger cars produced before 2010 and 14% before 1990 (DENATRAN 2018).
The Köppen-type climate of the region is subtropical Cwa with well-defined seasons, and rainfall concentrated in summer (Junqueira Junior et al. 2019). Long-term average annual precipitation (1981-2010) is 1462 mm, and 85% of the rainfall occurs in the wet period (October to March). The mean annual temperature is 20.3 °C ranging from 16.9 to 22.5 °C (INMET 2019a).

Sampling campaign
From May 2018 until April 2019, wet deposition samples were collected after each rainfall event (daily at 9:00 AM local time) using 3 Ville de Paris-type rain gauges. Moreover, 61 different rainfall events were collected with at least 5 mm (critical volume necessary for laboratory analysis). In addition, monthly samples were analyzed, adding an aliquot of each rain gauge in the same flask. Regarding preservation procedures and analytical methodologies, we followed the criteria adopted by Standard Methods for the Examination of Water and Wastewater (APHA/AWWA/WEF 2014).
In the same period (May 2018-April 2019), we also collected 36 bulk deposition samples through a high-density polyethylene bucket (NALGON) of 10 L with a collecting area of 439 cm 2 . To prevent sunlight effects and reduce litterfall in samples, the collector was placed inside a sunprotective PVC structure and covered with a nylon mesh. In this case, the sampling period was around 7 days. In the absence of precipitation, 50 mL of deionized water was added in order to analyze soluble species. It is important to note that the sampling collector was installed 1.5 m above ground level and rinsed several times with ultrapure water Milli-Q (Millipore, electrical resistivity 18 MΩ) in order to follow GAW's sampling procedures (WMO 2004). In addition, blank sample analyses were carried out throughout the experimental campaign.

Analytical procedures
Total Kjeldahl nitrogen (TKN) was determined according to the macro-Kjeldahl method (ABNT 1997) for monthly ammonium and organic nitrogen quantification in the wet deposition samples. The process for TKN analysis consists in converting organic nitrogen to ammoniacal nitrogen by acid digestion, and thereafter, the sample pH is raised for the ammoniacal nitrogen distillation; lastly, the nitrogen was quantified by titrimetric method. For TKN analysis, the detection limit was calculated as 0.36 mgL −1 .
We measured pH and DIN concentrations from bulk deposition samples. The pH measurements were obtained by using a pH meter (AKSO AK model 151), calibrated with buffer solutions (pH 4.0 and 7.0). In order to quantify nitrogen inorganic species (NH 4 + and NO 3 − ), one sample aliquot was filtered with a 0.22-μm-diameter membrane (Millex), stored in conditioned polyethylene bottles, and kept at − 18 °C until chemical analysis. The ion chromatograph (IC) (Metrohm model 851) was equipped with anionic column (Metrosep A Supp 5-250 mm × 4 mm) and cationic column (Metrosep C2 150-150 mm × 4 mm). Analytical quantification was performed using an external calibration curve from the standard concentrations for each ion. We calculated detection limit (DL) values from the parameters obtained from the analysis, by the method of the least squares, from the calibration curve (y = a + bx) and corresponded to the white sign (or linear coefficient) plus 3 times the standard deviation (sd) of the angular coefficient (sd y/x ), that is, DL = a + 3 sd y/x . Both NH 4 + and NO 3 − species presented DL less than 0.01 mgL −1 . Blank sample concentrations were quantified and subtracted by 0.36, 0.05, and 0.15 mgL −1 for TKN, NH 4 + , and NO 3 − , respectively.

Flux calculations and statistical analyses
We estimated nitrogen inputs as the product between concentrations of TKN, NH 4 + , and NO 3 − species and collected rainfall amounts. The monthly and annual nitrogen deposition flux was expressed using Eq. 1 Zhang et al. 2020).
where I represents the input (kgha −1 month −1 or kgha −1 year −1 ), C represents the nitrogen specie concentration (mgL −1 ), V represents the volume sample (L), A represents the collector area (m 2 ), i refers to the ith sample, and n is the total number of samples at the corresponding monthly or annual scale. For bulk deposition samples in which we added 50 mL of deionized water due to rainfall absence, we considered this volume for calculations.
From wet deposition inputs, we calculated DON by determining the difference between TKN and NH 4 + ( ). It is valuable to report that, for bulk deposition samples, we estimated wet deposition inputs according to Filoso et al. (2003), in which N inputs from wet deposition were considered 50% of bulk deposition. Although this assumption may overestimate or underestimate the nitrogen wet deposition fluxes, there is no previous study accounting for this estimate in the Southern Minas Gerais region. In addition, the Filoso et al. (2003) study was conducted in the countryside of Brazil's Southeastern region, where ~ 31% of the area is associated with agricultural activities.
We carried out parametric and non-parametric statistical tests accordingly, following the Shapiro-Wilk test (Shapiro and Wilk 1965). Thus, a one-way analysis of variance (ANOVA) and Kruskal-Wallis test were performed to detect rainfall and deposition flux differences of the nitrogen species among the wet and dry seasons (Romero Orué et al. 2019;Liu et al. 2021). In addition, a correlation analysis was applied to identify associations between the different nitrogen forms Rasse et al. 2018;Xing et al. 2018). We emphasize that all statistical analysis and data processing were performed in the R programming environment, through which we applied several functions of the stats and ggplot2 packages (Wickham 2016; R Core Team 2019). (1)

Atmospheric deposition seasonal patterns
From May 2018 until April 2019, the total rainfall collected by wet deposition (location depicted in Fig. 1) was 1524.6 mm, which represents 99% of the total rainfall reported by Lavras' weather station for the same period (INMET 2019b). In comparison with climatological values, the sampling period showed a surplus of 63 mm (rainfall positive anomaly) in comparison with climatological values; such differences represent almost 5% above long-term annual average rainfall . Approximately 85% of the rainfall occurred in the wet season (October to March) (Fig. 2a), which is in accordance with the expected pattern of the region. Regarding the atmospheric bulk deposition, the total rainfall collected in the sampling period was 1050.4 mm, which in comparison with Lavras' weather station represents 68%. In addition, such differences were expected due to wet and bulk deposition collection methods; moreover, sample loss by evaporation is inherent to atmospheric deposition sampling. Conversely, we can assume that most of the atmospheric processes were represented by both atmospheric deposition samples collected.

Total dissolved nitrogen (TDN)
Deposition flux of TDN ranged from 0.328 to 2.869 kgha −1 month −1 in June and December, respectively (Fig. 2b). According to the climatological normal for Lavras, December regularly presents the most rainfall and, as expected, the highest rainfall amount collected. Alternatively, rainfall in June was 22 times lower than December 2018. Moreover, TDN fluxes showed significant correlation (p value < 0.05) with rainfall reported by Lavras weather station in the study period (r = 0.95). Thus, this pattern suggests that monthly variability of TDN deposition was influenced by the rainfall distribution pattern. Although several meteorological variables could interfere in the samples, we chose to report only the precipitation due to its direct impact in the atmospheric deposition fluxes. Similar results were observed in the Northern Indo Gangetic Plain, Nepal (Bhattarai et al. 2021); in the Sichuan province, southwest China (Deng et al. 2018); and in Lagos Lagoon basin, Nigeria (Oladosu et al. 2017), where wet deposition flux of nitrogen increased with the increase in precipitation volume. We calculated an annual TDN atmospheric deposition of 16.73 kgha −1 year −1 , which was comparable with the global estimates higher than 8 kgha −1 year −1 at sites like eastern North America, Southern Brazil, Europe, and Asia (Vet et al. 2014). In this perspective, Souza et al. (2015) (n = 40) reported a similar TDN bulk atmospheric deposition ranging from 12.1 to 17.2 kgha −1 year −1 in a forest area and a coastal urban area in the Southeast region of Brazil, where the increasing in the TDN deposition was not related to increase in the annual rainfall, but due to the proximity of the Rio de Janeiro metropolitan region. In contrast, Araujo et al. (2015) (n = 20) monitored an urban area in the Northeast region of Brazil, with similar annual average precipitation (1500 mm) in comparison with our study region; however, the bulk TDN deposition was four times lower (4.32 kgha −1 year −1 ). The results reported in the cited studies suggest that other factors control TDN fluxes besides rainfall, suggesting that agricultural sources are main drivers for nitrogen fluxes in the studied site.

Dissolved inorganic and organic nitrogen (DIN and DON)
In the sampling campaign, NO 3 − fluxes ranged from 0.011 (August) to 0.831 kgha −1 month −1 (January), while NH 4 + deposition varied from 0.003 kgha −1 month −1 in May to 0.343 kgha −1 month −1 in March. The combined deposition flux of DIN (NO 3 − + NH 4 + ) ranged between 0.016 (August) and 1.074 kgha −1 month −1 (March), with NO 3 − contributing more than 56% of DIN throughout the study period. Regarding organic nitrogen, DON deposition fluxes ranged from 0.203 to 2.103 kgha −1 month −1 in June and December, respectively. In this sense, DON was the predominant nitrogen species throughout the sampling campaign, with a monthly contribution relative to TDN ranging from 50 to 98%. In addition, the same as that of TDN, the monthly deposition fluxes of inorganic and organic nitrogen species were controlled by the precipitation, since NO 3 − , NH 4 + , and DON presented strong and significant correlations (p value < 0.05) with rainfall (r = 0.68, r = 0.79, and 0.87, respectively).
In order to characterize the seasonal variability, monthly deposition fluxes of NO 3 − , NH 4 + , and DON were combined according to seasons: the wet season (October-March) and the dry season (April-September). Nitrogen deposition fluxes in the wet season reached values of 3.38, 1.23, and 8.11 kgha −1 for NO 3 − , NH 4 + , and DON respectively, which were 5.15, 6.74, and 2.56 times higher than deposition fluxes of these species in the dry season, revealing a significant distinct seasonal pattern (ANOVA test; p value < 0.05). It is important to note that the highest deposition fluxes for both inorganic and organic nitrogen species occurred in the months that concentrate the fertilizer application period in coffee crop plantations (October-March) (Freitas 2017), suggesting that agricultural emissions induce strong seasonal variations in the deposition fluxes, driven largely by the seasonality of agricultural operations.
Based on the monthly deposition f lux data, we found a strong positive correlation between NH 4 + and NO 3 − (ρ = 0.78; p < 0.05), indicating likely sources, mainly agricultural production, suggesting NH 4 NO 3 formation. Among the inorganic species, NH 4 + had a higher rate of increase when comparing the dry and wet seasons (574%) and also a higher correlation with rainfall (r = 0.79) than NO 3 − (r = 0.68). This behavior emphasizes the influence of fertilizer application and rainfall on NH 4 + deposition and reveals a slight presence of nonagricultural factors in NO 3 − deposition, likely fossil fuel combustion sources.
It should be noted that DON is constitutive of a variety of organic compounds, of which many originate from diverse natural and anthropogenic sources (Cape et al. 2011). In this sense, previous studies carried out in agricultural regions indicated that DON in the atmosphere may be related with the application of urea and organic fertilizers as livestock manure (Ham and Tamiya 2006;Zhang et al. 2012). Figure 2b shows DON inputs with the lowest rate of increase across both seasons (156%) and the highest correlation with rainfall (r = 0.87), suggesting that DON is more influenced by rainfall and other mixed sources, such as litterfall and biomass burning. Moreover, the moderate correlations between DON and NH 4 + (ρ = 0.67; p < 0.05) and between DON and NO 3 − (ρ = 0.55; p < 0.10) suggest that DON also may be partially derived from other sources, such as soil dust, biological emissions, biomass burning, and industrial emissions. Although the Southern Minas Gerais region could not be described as an industrialized area, Pereira et al. (2021) showed that cement manufacturing and forest fires influenced atmospheric deposition chemistry in the region.

NO 3 − /NH 4 + ratio and pH
Several studies have been using the NO 3 − /NH 4 + ratio as a reliable proxy for assessing the relative contributions of oxidized and reduced nitrogen species in the atmospheric deposition (Hunová et al. 2017;Xing et al. 2017;Zheng et al. 2020). In our study, the monthly mean values of NO 3 − /NH 4 + ratios varied between 1.36 in April and 33.7 in May, with an annual average of 8.25 (Fig. 3). This pattern reveals monthly NO 3 − /NH 4 + ratios > 1 for the entire study period, thereby indicating that oxidized nitrogen played a more important role for the nitrogen deposition flux. It is noteworthy that pH values were lower in May (pH = 5.75) than April (pH = 6.03), which was expected, since NO 3 − is well recognized as an acidity proxy, whereas NH 4 + is associated with NH 3 gas (Seinfeld and Pandis 1998). In addition, despite NO 3 − /NH 4 + ratios > 1, none of the samples presented pH values below the limit of 5.60, a critical value for determining acid rain samples. Thus, all samples were alkaline with an average pH of 5. 99 (5.73-6.19).
Aiming to identify seasonal patterns, we calculated an average NO 3 − /NH 4 + ratio of 12.8 and 4.48 in the dry and wet periods, respectively. In general, NH 4 + can be directly attributed to ammonia (NH 3 ) emissions, mainly due to agricultural activities, such as fertilizer application and livestock production (Zheng et al. 2020), and to a lesser extent due to the combustion of fossil fuels (Liu et al. 2016). Regarding the driving factors of NO 3 − , although it is linked to high NO x emissions mainly from vehicle exhaust systems in large urban centers (Cui et al. 2014;Wu et al. 2018), recent findings identified soil emissions as a significant origin of high emissions of NO x in agricultural regions (Molina-Herrera et al. 2017;Almaraz et al. 2018). In this sense, the decrease of the NO 3 − /NH 4 + ratio in the wet season implies an increase of NH 3 emission in the wet season. In another perspective, the NO 3 − /NH 4 + ratio closer to the unit also shows the importance of NO x emissions from the fertilized soil. This behavior suggests a strong seasonality of agricultural activities in the studied region, due to nitrogen fertilizer application.
Moreover, fertilization coupled with heavy rain and high temperature promotes strong volatility of NH 3 and NO x , which could lead to volatilization rates of 8% and 11%, respectively (Kurvits and Matta 1998;Calvo-Fernández et al. 2017;Freitas 2017). By contrast, the higher NO 3 − /NH 4 + ratio in the dry season reflects the NO x emissions from anthropogenic sources, as well as the lower temperatures and higher relative humidity contribute to higher formation of nitrate due to gas-particle reactions (Dong et al. 2020). It is valuable to note that NO 3 − /NH 4 + ratios displayed statistically non-significant differences between the dry and wet seasons, which may be associated with high ratio values in January (10.1) and low values in August (2.10). This nonseasonal behavior of nitrogen species could be associated with the negative rainfall anomaly in January (− 129 mm) and positive in August (+ 53 mm).

Worldwide deposition comparisons
On an annual basis, NO 3 − and NH 4 + deposition fluxes were 4.04 and 1.41 kgha −1 year −1 , accounting for 24% and 8% of the TDN, respectively. The NO 3 − deposition flux was similar to that reported by Souza et al. (2015) in a coastal urban area in Brazil and 2.6-fold lower than reported in urban and agricultural areas in China (Cui et al. 2014;Deng et al. 2018) ( Table 1). The same as NO 3 − , the NH 4 + deposition flux was lower than reported in agricultural and urban areas in China (Cui et al. 2014;Deng et al. 2018). In addition, it agrees with Ounissi et al. (2021), Cao et al. (2019), and by Izquieta-Rojano et al. (2016) in Algeria coastal, Japanese forest, and Iberian Peninsula urban areas, respectively. Regarding DON, the annual deposition flux was 11.3 kgha −1 year −1 , which is similar to areas with agricultural background in the Iberian Peninsula and China (Cui et al. 2014;Izquieta-Rojano et al. 2016;Deng et al. 2018), as well as urban areas in Brazil and China Deng et al. 2018). Moreover, DON deposition flux was about fivefold higher than reported by Schroth et al. (2001) in forest areas. DON represented the major fraction of the TDN (68%), similar to Cao et al. (2019) in a forest area in Japan. Such high relative contributions also were reported in Chinese inland regions (Liu et al. 2016;Wu et al. 2018) and are reasonable due to fertilizer application in croplands.
In general, NH 4 + showed the lowest values when comparing with all locations, whereas NO 3 − levels in the studied region were similar to urban zones and DON has values comparable to urban and agricultural areas. In addition, when comparing all agricultural background locations (Table 1), one can assume that nitrogen deposition in this study is within the levels estimated by other studies, thereby indicating that agricultural regions, especially Lavras, are hotspots of nitrogen deposition.

Ecological effects of wet nitrogen deposition
Due to TDN values observed in our study, we suggest that nitrogen supply via wet deposition should be considered when calculating optimum nitrogen fertilizer application rates since the study region has a rural background, mostly agricultural. Moreover, nitrogen species deposition by wet deposition provided 16.73 kgha −1 year −1 , which is about 15% of the annual nitrogen fertilizer application in the region, suggesting a major input of nitrogen species, specifically in the wet season.
Regarding N-DIN species, the annual value observed (2.011 kgNha −1 yearr −1 ) via wet deposition was lower than the empirical critical nitrogen load proposed by Pardo et al. (2011) for tropical and subtropical humid forests (5-10 kgNha −1 year −1 ). However, such value does not account for N-DON nitrogen inputs that would surpass such critical range. The excesses of nitrogen deposition over critical loads could lead to environmental impacts in the region, such as soil acidification, which has implications for the chemical availability of metals and other nutrients, influencing plant nutritional status (Stevens et al. 2018).
Excess nitrogen may also cause damage to freshwater aquatic ecosystems (Chen et al. 2018;Clark et al. 2018). Baron et al. (2011) estimated an acidification critical load of 8.0 kgNha −1 year −1 and an even lower nutrient enrichment critical load, ranging from 3.5 to 6.0 kgNha −1 year −1 , for lakes of the Northeastern United States. Thus, our results showed that nitrogen atmospheric deposition of N-DIN (2.011 kgNha −1 year −1 ) has the potential to increase acidification and eutrophication in water bodies in the studied region by atmospheric input by 25% and 33%, respectively.

Conclusion
We assessed the wet atmospheric deposition fluxes of TDN, DON, and DIN (NO 3 − + NH 4 + ) for 1 year (March 2018-April 2019), as well as the factors controlling its seasonal variations in the Southern Minas Gerais region, Brazil, an area with agricultural background. TDN annual flux was 16.73 kgha −1 year −1 , with a dominant contribution of DON (68%), followed by NO 3 − (24%) and NH 4 + (8%). These results showed that TDN deposition should be included in the calculation of optimum nitrogen fertilizer application rates since they provided 15% of the annual nitrogen fertilizer application in Lavras. Moreover, the N-DIN fluxes (2.011 kgNha −1 year −1 ) presented potential to increase acidification and eutrophication in Lavras water bodies by about 30%. Because such value does not account for N-DON fluxes, the TDN inputs would increase even more the nitrogen pollution in the environment.
We observed considerable seasonal variations (ANOVA test; p value < 0.05), since all nitrogen species presented higher fluxes in wet season (NO 3 − = 3.38, NH 4 + = 1.23, and DON = 8.11 kgha −1 ), induced by rainfall and agricultural emissions. On the other hand, moderate correlations between DON and NH 4 + and between DON and NO 3 − (ρ = 0.67, ρ = 0.55, respectively) indicated that DON was partially derived from other sources, such as soil dust, biomass burning, and industrial emissions. We also found high NO 3 − /NH 4 + ratio (average = 8.25), with lower values in the wet season (NO 3 − /NH 4 + = 4.48) and higher values in dry season (NO 3 − /NH 4 + = 12.8). This behavior revealed a higher relative contribution of NO x emissions from traffic sources in the dry season and corroborated with the increase of NH 3 emissions from agricultural sources during wet season. Therefore, our study sheds light on the importance of agricultural activity as a driver of nitrogen wet deposition coupled with its potential to change biogeochemical cycles. Atmospheric emissions resulting from activities in the agricultural sector are not regulated in Brazil and have the potential to degrade environmental quality. Accordingly, the use of decentralized monitoring policy strategies can contribute to the assessment and possible control of these emissions, avoiding their impacts on natural ecosystems. Data availability All meteorological data used in this research was available at INMET (Instituto Nacional de Meteorologia) websites (https:// portal. inmet. gov. br/); all coding and R programing language packages are properly cited across the manuscript.

Declarations
Competing interests The authors declare no competing interests.