Effect of nitrogen (N) deposition on soil-N processes: a holistic approach

Nitrogen (N) deposition is a serious environmental issue for soil fertility and human wellbeing. Studies on various terrestrial ecosystems yielded fragmented information on soil-N status (microbial biomass- and mineral-N) and dynamics (N-mineralization and -leaching) whereas the holistic view on this issue is relatively unknown. A complete understanding of soil-N status and dynamics in response to N deposition is essential for sustainable management of ecosystem structure and function as needed for human wellbeing. Therefore, we conducted an experiment in the N-limited tropical grassland to explore the question whether N-deposition weakens the soil-N status and dynamics; if yes, then what could be the optimum amount of deposited N and the related controlling mechanism? We undertook a 3-year (2013–2016) experimental N fertilization (control, 30, 60, 90, 120, and 150 kg N ha−1 year−1) study (using urea as a source of N deposition). The data from a total of 72, 1 × 1 m plots (six treatments with 12 replicates) were collected and properly analysed with statistical software. N deposition caused significant differences in the parameters of soil-N status and dynamics. The responses of microbial biomass-N, N-mineralization, and mineral-N to the N deposition were quadratic (maximum values were in N90) whereas N-leaching showed a linear response. Compared to control, N deposition (30–150 kg N) consistently enhanced (29–96%) leaching of N. As a mechanism, acidification induced aluminium toxicity, carbon to nitrogen ratio and litter decomposition governed the soil-N status and dynamics. N deposition over and above 90 kg ha−1 year−1 resulted in a negative feedback to soil N transformation and availability. Hence, N deposition below 90 kg ha−1 year−1 could be a limit for the sustainable functioning of the tropical or similar grasslands.


Materials and methods
Study location. The study was conducted on 72 plots on the campus of the Banaras Hindu University (24° 18′ N and 83° 03′ E and 76 m m.a.s.l. altitude), during July 2013 to June 2016 in the Indo Gangetic Basin of eastern Uttar Pradesh located in Varanasi district of India (Fig. 1). The climate of the study area is tropical monsoonal with three different seasons; a cold winter (November-February), a hot summer (April-June) and a warm rainy season (July-September). October and March are transitional periods between rainy and winter, and between winter and summer seasons, respectively 59 . During the study period, mean maximum temperature was 31 °C while mean minimum temperature was 20 °C and the mean annual precipitation was 969 mm. The soil is categorised as Banaras Type III 60 , alluvial, well-drained, pale brown, silty loam, inceptisol 61 and moderately Scientific RepoRtS | (2020) 10:10470 | https://doi.org/10.1038/s41598-020-67368-w www.nature.com/scientificreports/ fertile being low in available N and medium in available phosphorus and potassium 62 with neutral to alkaline soil pH 63 . The campus of the Banaras Hindu University is spread in ≈ 520 ha land area having a luxuriant growth of natural flora. Azadiracta indica, Dalbergia sissoo, Madhuca longifolia, Mangifera indica, Sterculia alata, Tamarindus indica, Tectona grandis, Zizyphys glaberrima, etc. are locally top canopy dominating species while Alysicarpus monilifer, Cynodon dactylon, Cyperus compressus, Desmodium gangeticum, Dichanthium annulatum, Evolvulus nummularius, Imperata cylindrica, Malvastrum coromandelianum, Oplismenus burmannii, Sida acuta are the locally dominating species of ground vegetation 21,56 . Experimental design. A 20 × 20 m open area having natural herbaceous vegetation and substantially away from the buildings was selected in the horticulture premise of the University. Within this; 72, 1 × 1 m experimental plots (plot size determined by species-area curve), arranged in 6 parallel rows (12, 1 × 1 m plots in each row) were established. Surface-to-belowground boundary of each 1 × 1 m plot was permanently demarcated by using bricks and cement (10 cm wide and 50 cm depth). A 1 m distance between two 1 × 1 m plots was kept as buffer zone to protect against boundary effects due to the migration of N out of the sampling areas. Six treatments of N, each replicated twelve times, were randomly established on the basis of lottery method: control (without N), N 30 (30 kg N ha −1 year −1 ), N 60 (60 kg N ha −1 year −1 ), N 90 (90 kg N ha −1 year −1 ), N 120 (120 kg N ha −1 year −1 ) and N 150 (150 kg N ha −1 year −1 ) 64 .
Before the N treatment to the plots; soil bulk density, soil-porosity, -water holding capacity, -sand, -silt and -clay contents for the experimental plots were analysed. Across the plots; values of soil bulk density (g cm −3 ), percentage soil-porosity, -water holding capacity, -sand, -silt and -clay contents varied from 1. 23-1.26, 52-54, 48-49, 8-9, 77-79 and 12-13, respectively. Statistically; none of these soil variables varied due to designated N-levels. Thus, before the initiation of the experiment, the selected plots were homogenous in soil characteristics.

N-inputs.
Similar to other studies 65,66 , urea fertilizer was used as a source of N deposition because it has high (46%) N content, zero phosphorus and potassium and is comparatively inexpensive, stable, and easy to handle and it alone contributed more than 50% of the global atmospheric-N deposition 2 . The urea was applied in the evening in each month to avoid the N loss due to volatization 21,64 . For calculating the monthly doses, the total annual N dose of each N level was equally divided into 12 applications. We deliberately used a wide range of N-level with a maximum of 150 kg N ha −1 year −1 to understand the measureable responses of soil-N attributes and threshold tolerance of soil N to the N deposition within the tropical grassland. www.nature.com/scientificreports/ Sampling and analyses. The soils were sampled at two depths (0-10 cm and 90-100 cm depths from the soil surface). For each N-level, three soil samples (0 to 10 cm depth) were collected from each 1 × 1 m plot, for each month of the year starting from 2013 to 2016 by using a 5 cm-diameter corer. For each soil depth, the three soil samples collected from each 1 × 1 m plot were combined to form a composite soil sample for each plot. These composite soil samples of 0-10 cm depth were gently homogenized. Carefully, large roots, woods, litters and all fine roots were removed from the composite soil samples. One part of soil sample was air dried, sieved through 2 mm mesh screen and analyzed soil-pH, total organiccarbon (C), total soil-N (TN), ammonium-N (NH 4 + -N) and nitrate-N (NO 3 − -N). Soils of 0-10 cm depth were used for the analysis of soil N mineralization, microbial biomass carbon; MBC and MBN. To understand the leaching; the soils of 90-100 cm depth were collected by inserting a 100 cm long metallic corer into soil by avoiding the root injury. These soil samples were used for the estimation of NH 4 + -N and NO 3 − -N 67 . Total soil-N (TN) was determined by micro-Kjeldahl digestion method 68 . NH 4 + -N was extracted by 2 M KCl and analyzed by using the phenate method 69 . The concentration of NO 3 − -N was analyzed by the phenol disulphonic acid method after extraction by CaSO 4 68 . For the analysis of litter decomposition, nylon net litter bag (10 × 10 cm) technique 70 was adopted. In this analysis, 100 g of air-dried mixed leaf litters of grasses and forbs were used. The mesh size of the litter bags was 1 mm which easily allows the movement and activity of soil microorganisms. Litter decomposition was determined by computing the decay constant (k). The negative exponential decay (k = − ln (X t /X 0 )/T) model was used to compute the k 71-73 . In the equation; X 0 is the initial dry weight, X t the dry weight remaining at the end of the investigation time T (1 month).
The in situ buried bag technique was adopted for N-mineralization. Before incubation, the NH 4 + -N and NO 3 − -N concentrations were determined for zero-month sampling. Using a large sealed polythene bag, a portion of fresh soil sample (200 g) was incubated in soil at a depth of 0-10 cm on the same microsite from which the samples had been collected for the analysis of NH 4 + -N and NO 3 − -N. After one month of field incubation, the incubated bags were collected for the analysis of NH 4 + -N and NO 3 − -N. Again, a portion of fresh soil sample (200 g) was incubated on the same microsite from which the samples had been collected for the analyses of NH 4 + -N and NO 3 − -N and after one month of incubation the incubated samples were re-collected and NH 4 + -N and NO 3 − -N were re-analyzed. These analyses were repeated for each month of the entire experimental period after an interval of 30 days from 1st July 2013 to 24th June 2016. The increase in the concentrations of NH 4 + -N and NO 3 − -N after field incubation is referred to as ammonification and nitrification, respectively and the increase in the amount of NH 4 + -N plus NO 3 − -N over the course of field incubation is defined as total N mineralization 73 . MBC and MBN were determined by the chloroform fumigation-extraction method using 0.5 M K 2 SO 4 as an extractant 74 . The organic-C of extract was estimated by oxidation with potassium dichromate. The difference in the organic-C content between the fumigated and unfumigated extracts was converted to MBC by dividing with a conversion factor of 0.45 75 . The MBN was estimated by micro-Kjeldahl digestion procedure from the extracts. The difference in N content between the fumigated and unfumigated extracts was converted to MBN by dividing with a conversion factor of 0.54 76 .
For the analysis of soil base cation and non-base cations, dried soil samples in triplicate were homogenized by grinding to fine powder followed by digestion in di-acid (HNO 3 and HClO 4 in 9:4 ratio) solution 77 . The contents of soil Na + , K + , Al 3+ and Fe 3+ were determined with Atomic Absorption Spectrophotometer; AAS (Analyst-800, PerkinElmer Inc., Norwalk, CT, USA). For all the metals; blank and standards (Sisco Research Laboratories Pvt. Ltd., India) were run after every five samples to check the accuracy and precision of the results (within 2% of the certified value).
The analyses of litter decay constant, C/N ratio, pH and MBC/MBN ratio were used for explaining the patterns of soil-N pools and dynamics to the N depositions. Similarly, base and non-base cations were used. We used these variables as explanatory variables because under the scenario of N deposition these variables are supposed to be interlinked with the soil-N pools and dynamics.
For understanding the effects of N deposition on the response variables, the effect sizes (response ratio; RR) were computed following the equation of Hedges et al., (1999) 78 . For a given variable; the RR was estimated as the ratio of its value in the N treatment group (X t ) to that in the control group (X c ). It was transformed in log scale to improve its statistical behaviour, hence the ln RR = ln (X t /X c ) or ln X t − ln X c equation was used. ln RR was assumed to follow a normal distribution 79,80 . The optimum amount of N deposition (before the negative feedback) required for the maximum beneficial responses of the selected variables of soil-N status and dynamics were computed based on the best-fitted regression equations between the N-levels and corresponding parameters of soil-N status and dynamics.
Statistical analyses. Repeated measures analysis of variance (RANOVA) procedure selecting general linear model (GLM) option in SPSS package 81 was used to notice the effects of year, month, and season on the parameters of soil-N status and dynamics. In these analyses, year and season were used as within-subject variable and N-levels as between-subject factor. Tukey's HSD test was used to determine the significance of differences in the values of these variables between different treatment pairs. Pearson correlation coefficient was established between the different response variables with the help of the SPSS package 81 . The MBN and N mineralization parameters were linearly regressed with the soil-pH, TN, C/N ratio and decay constant opting linear regression option in SPSS software 81 . A path analysis was constructed by using AMOS 16.0 software 82 , which executes the "structural equation modelling/analysis of covariance structures/causal modelling". It was based on the linear correlation analysis and represented graphically to visualize the direct and indirect interactions among the predictors and the dependent variables 83

Results
Soil pH. Season, year, N-levels and their interactions caused significant variations in the soil-pH ( Table 1).
The soil-pH decreased with the progress of the experimental duration and N-levels. Effect size analysis also indicated a consistent decline of soil-pH due to increasing amount of N deposition (Fig. 2).

Concentrations of soil base and non-base cations.
Across the N-levels, the concentrations (g Kg −1 of soil) of Na + and K + soil base cations ranged from 1.03 to 1.34 and 6.37-7.52, respectively. The concentrations (g Kg −1 of soil) of Al 3+ and Fe 3+ (non-base cations) varied from 36.63 to 48.79 and 8.95-9.59, respectively ( Table 2). The values were greater in N treated plots compared to control plots. The K + (R = 0.97, P ≤ 0.002) and Al 3+ (R = 0.98, P ≤ 0.001) positively responded to the N-levels ( Fig. 3). Interestingly, the Al 3+ was linearly and negatively related with the soil-pH (Fig. 3).

Soil microbial biomass nitrogen (MBN).
Across the N-levels, MBN (µg g −1 ) varied from 415; N 150 level to 545; N 90 level (Table 2). RANOVA showed significant effects of year, season, N-levels, season × year, season × N-levels, year × N-levels and season × year × N-levels on the MBN (Table 1). Yearly, the MBN was maximum in second-year of the experiment and minimum in the third-year of the experiment (Table 2). Tukey's HSD test showed significant differences in the values of the MBN among the different year-pairs (Table 2). Season-wise, it was minimum in rainy and maximum in the dry season ( Table 2). The Tukey's test yielded a significant difference in the MBN between rainy and winter, between winter and summer and between summer and rainy seasons ( Table 2). Response size of the MBN to the N-levels was humped-shape ( Fig. 2). Compared to control, the percent change in the MBN for each N-level varied between -11; N 150 level and 21; N 90 level (Fig. 4).  Table 2). The values of these parameters were low in N 0 and high in N 90 (Table 2). Year wise, the ammonification (µg g −1 month −1 ), nitrification (µg g −1 month −1 ) and total N-mineralization (µg g −1 month −1 ) varied between 11.36 and 13.79, 7.59 and 9.52, and 18.95 and 23.31, respectively. The values were minimum for the third-year of the experiment and maximum for the first-year of the experiment. Minimum values were observed in the dry season and maximum in the rainy season (Table 2). Results showed differences in the values of total N-mineralization as well as its components due to differences in the year, season, N-levels and their possible interactions ( Table 1). The Tukey's test suggested major variations in the values of the third-year N-mineralization with those of the first and second years. Similarly, these parameters varied between different season-pairs ( Table 2). The response effect sizes of these parameters to the N-levels were quadratic (Fig. 2). Compared to the N 0 , the percent changes in ammonification, nitrification and total N-mineralization for each N-level varied from 31 to 93, 19-97 and 27-94. Also, these parameters showed quadratic responses to the N-levels (Fig. 4).  www.nature.com/scientificreports/ The ammonification, nitrification and total N-mineralization promptly increased from the end of the dry season to the onset of the rainy season (Fig. 5). Such pattern indicated a quick response of N mineralization to the first-rain event after a period of dry season.  (Table 2). RANOVA showed considerable variations in the NH 4 + -N, NO 3 − -N and inorganic-N due to year, season, N-levels and their interactions (Table 1). Moreover, the mineralized N substantially varied between the years and between the seasons ( Table 2). The effect size responses of the mineralized-N to the N-levels were positive and humped-shape (Fig. 2). The results showed synchronization of NH 4 + -N, NO 3 − -N and inorganic-N with those of the ammonification, nitrification and total N-mineralization. Thus, it seems that the available-N was controlled by the rates of N-mineralization as also suggested by path analysis (Fig. 6) (Table 1). Individually, these attributes were one and half times greater in the third-year of the experiment than the first-year of the experiment, 11-17 times greater in the rainy season than the dry season and consistently increased with the N-levels (Table 2). Interestingly, the values were two-fold greater in the N 150 than the N 0 (Table 2). Also, the effect size analysis indicated increasing pattern of N leaching along the increasing rate of N depositions (Fig. 2). Compared to control, N deposition favoured the leaching of NH 4 + -N by 20-107%, NO 3 − -N by 33-95% and inorganic-N by 29-97%. Overall, on average N deposition enhanced leaching of NH 4 + -N by 56% and NO 3 − -N by 65% (Fig. 4).

Rate of N-mineralization.
Pearson correlation analysis indicated that the leaching of NH 4 + -N, NO 3 − -N and inorganic-N were positively related with TN, decay constant and rates of N-mineralization (Table 3). In contrast to these relations, the N leaching parameters were negatively related with those of soil-pH, MBN, C/N ratio, MBC/MBN ratio, NH 4 + -N, NO 3 − -N and inorganic-N of the 0-10 cm soil depth (Table 3). Finally; path analysis was performed to identify the direct and indirect effects of soil variables on the parameters of inorganic-N leaching. Results showed that the N-levels, soil-pH, decay constant, N-mineralization and C/N ratio caused direct effects on the leaching of inorganic-N. As a main result; the rate of N deposition and soil-pH mediated by it firmly determined the leaching of inorganic-N (Fig. 6).

Discussion
Soil acidification, soil base and non-base cations. In the present study, N induced reduction of soil-pH along increasing rate of N deposition could be due to increase in Al 3+ concentration. It is evident because of negative relationship of Al 3+ concentration with soil-pH. Release of carbonic acids during litter decomposition 84 and H + into the soil solutions during nitrification 85 could be probable explanation for reduction of soil-pH owing to N deposition. Also, it is known that breakdown of urea fertilizer (similar to other studies 65,66 , we also    21,31,84 . In other studies differential buffering and mobilization capacities of base and non-base cations are suggested for soil acidification because of N input 57 . Reports indicated that reduction of base cations (Na + and K + ) present in soil system usually delays acid-buffering capacity of the soil 57,86 . Once, these base cations are exhausted, the non-base cations (Al 3+ and Fe 3+ ) mobilize and buffer during the N-induced soil acidification 57,86 . Similar to present study, other studies also reported reduction of soil-pH due to N deposition 21,23,55,66 . Microbial biomass-N. Low MBN in the rainy season and high in the dry season suggested seasonality in the MBN of the tropical grassland 29,87,88. Significantly positive effect sizes of MBN from low to moderate N-levels (N 30 -N 90 ), and negative from moderate to high levels of applied N (N 120 -N 150 ), decline in the third-year of the experiment and humped-shape response across the applied N-levels (by pooling entire data) favoured the assumption that moderate level of N deposition favours maximum MBN while continuous and sufficiently high amount of N depositions retard the conservation of soil-N in microbial biomass 89,90 . Noticeably, regression analysis revealed 63 kg N ha −1 year −1 as an optimum rate of N deposition for maximum accumulation of N in the microbes of tropical grassland. Using 82 published field studies (considering only highest N application rates), Treseder 91 suggested reduction of microbial biomass due to N additions. However, he excluded the data of microcosm or greenhouse-based experiments as well as organic N or urea, or N with phosphorus added data. A recent analysis of Camenzind et al. 52  The labile-N substrate quickly mineralized through the activities of the remaining microbes; hence, a pulse of N-mineralization was likely 95 . It might be expected that the microbes have stored a higher amount of N during the dry season and as they receive a rain event their activities get accelerated, consequently, the start of the rainy season yielded a greater mineralised-N 29,96 . The study yielded humped-shape curves for the effect sizes of N-mineralization parameters to the N-levels. These curves suggested that the N-mineralization was low at low levels of N depositions (N 30 -N 90 ), whereas increased to a maximum at a moderate level (N 90 ) then decreased towards the higher rate of N deposition (N 90 -N 150 ). It could be explained by changes in composition and activities of soil microbes, ' soil-pH and rate of organic matter decomposition in response to N depositions. These patterns are evident due to positive relations of decay constant, negative relations of C/N and MBC/MBN ratios and quadratic relations of soil-pH with those of ammonification, nitrification and total N-mineralization. The path analysis also revealed such mechanisms for the controlling the N-mineralization in the present experiment.
Since the N-mineralization parameters quadratically responded to the N deposition and soil-pH and 90 kg N ha −1 year −1 deposition rate yielded 6.98 soil-pH for maximum N-mineralization, therefore, 90 kg N ha −1 year −1 deposition is thought to be an optimum amount of N for favourable soil-pH that had supported maximum soil N-transformation from organic residues to the mineral-N in the tropical grassland. Similar to the present findings, other temperate studies also suggested moderate level of N deposition for greater soil N-mineralization mediated by microbial communities and their performances [36][37][38][97][98][99] . The poor quality of decomposing materials; high C/N ratio 43,84,100 , poor growth and activities of oligotrophic decomposers; high MBC/MBN ratio 43,[101][102][103] and conditions during the decomposition; low soil-pH and high Al 3+ toxicity 104 , could be major constrains for the transformation of organic residues into the mineralized-N.  www.nature.com/scientificreports/ higher N deposition levels. The study inferred that at the low level of N deposition; probably the soil-N was not sufficient for the activities of ammonifiers and nitrifiers to release the NH 4 + -N and NO 3 − -N. As soon as amount of N deposition was increased, more ammonifiers and nitrifiers get activated for ammonification and nitrification 23 , at adequately high level of N deposition, probably there was loss of additional N via volatilization of ammonium 106 , denitrification of nitrate [106][107][108] and also, possibly consumed by the nitrophilc plants for increasing their biomass 23,[109][110][111] .
Further, study believes that whatever N is present in the control plot maybe because of biological N 2 -fixation as well as from the release of N through the microbial decomposition of litters. The N deposition at the rate of 90 kg ha −1 year −1 reduces the N 2 -fixation for a short time and later on; N 2 -fixation is increased because of an increased microbial population 23 . Beyond this limit of N deposition; denitrification (if any); appears to be insignificant; therefore, the loss of extra N may be through volatilization 106 and denitrification 112 , however, at a slow rate. The N deposition at the rate of 150 kg ha −1 year −1 entirely reduces the growth of N 2 -fixing microbes and N 2 -fixation as well as activities and the growth of denitrifying microbes 106,112 . Additionally, a substantial amount of NH 4 + -N and NO 3 − -N may be taken by the nitrophilic species 23,109,110 . In this situation; the quantity of ammonia formation appeared to be reasonably high because of the increased quantity of substrate 106,113,114 and most of this ammonia is being transformed into the NO 3 − -N by nitrifies, whereas remaining ammonia is being volatilized 106 . www.nature.com/scientificreports/ mum in the dry season. The emergence of such patterns may be due to the maximum uptake of mineral-N by the plants for their vigorous growth in the rainy season. At the same time, the increased precipitation and slightly warmer condition during rainy season probably increased the microbial activities and the rates of decomposition which in turn could have increased the rate of N-mineralization (because of positive relationship between decay constant and N-mineralization). Thus, the remaining inorganic-N beyond the demands of plants and microbes possibly resulted into the leaching of inorganic-N through the water 2 . Probably, it could be a reason for high leaching in rainy and low in the dry season. The percentage change of NO 3 − -N leaching is approximately one and a half times greater than the NH 4 + -N. It may be because of greater nitrification and accumulation of NO 3 − -N (as evident by positive relationship between nitrification and NO 3 − -N leaching) which were assumed to be over and above the requirements of the plants and nitrifying microbes and less competition between them for NO 3 − -N 118,119 . Further, higher NO 3 − -N leaching rates than that of NH 4 + -N may be expected because the latter is a preferred form of inorganic N for the  www.nature.com/scientificreports/ biota due to low energetic cost during biological assimilation. Thus, it appears that the high NO 3 − -N leaching could be a dominant form of N-leaching because of biological assimilation controlled NO 3 − -N saturation 120,121 . Also, the possible mechanisms behind it may be explained by the differences in the charges of NH 4 + and NO 3 − . For example, the NH 4 + is positively charged and it binds with the negatively charged clay particles while the negatively charged NO 3 − freely moves with the water molecule until the exchange of anions within the soil is completed 122 . Thus, the study supported the view that the excess N deposition increases the NO 3 − -N leaching 2,58 in the tropical grassland. Its proper management is warranted; otherwise the excess N deposition may cause soil acidification, leaching of N 123 and ultimately participate in the warming of the globe through production of nitrous oxide from soil via nitrification and denitrification by different aerobic and anaerobic microbes 124 . Overall, the current holistic study revealed that the N deposition below 90 kg ha −1 year −1 could be a substantial limit for the healthy soil-N fertility and its transformation in the tropical grassland.

Conclusions
The continuous, as well as an incremental amount of N-levels decreased the soil-pH and increased the Al 3+ concentration within the soil system and changes in these soil variables governed the decomposition of organic materials and N-transformation. Also, the N deposition dependent soil-pH, decay constant and N-mineralization guided the leaching pattern of mineral-N. The N deposition below 90 kg ha −1 year −1 seems to be an optimum limit for the maximum soil-N status and dynamics. The N deposition beyond this limit caused negative feedback to the soil-N fertility and its dynamics. Hence, this holistic approach suggested that the N deposition should not go beyond 90 kg ha −1 year −1 and it should be managed by implementing into a policy for sustainable functioning of the tropical or similar grasslands.