Introduction

Anthropogenic N pollution entering the atmosphere is increasing every year1. It is predicted that global anthropogenic N emissions will be 200 Tg y−1 by 20502. Atmospheric N deposition can influence soil chemistry3, lacustrine and estuarine eutrophication4, biological diversity5, greenhouse gas balance6, and even human health7. The levels and sources of atmospheric N deposition urgently need to be assessed.

Due to lack protective cuticles and have large surface areas, moss receives N passively and effectively from atmospheric N deposition8,9,10,11. Mosses have therefore been used widely to easily and cheaply acquire relatively high spatial resolution data on long-term atmospheric N deposition. Strong relationships between atmospheric N deposition and bulk N concentrations in moss tissues have been found in numerous studies12,13,14,15,16,17.

In field studies, FAA has been found to be more sensitivity to atmospheric N deposition than bulk N in moss18,19. Laboratory experiments using 15N labelled compounds have indicated that N-containing compounds taken up by moss are immediately assimilated as glutamine (Gln) and then transformed into other FAAs to avoid toxic NH4+ concentrations accumulating in the cells20,21. Strong links between the concentrations of some FAAs and atmospheric N deposition have been found for vascular plants18,22,23,24,25,26,27. However, different types of FAAs have been found to accumulate in different plant species25,28. It is still unclear whether FAA concentrations in moss can be used to quantitatively indicate N deposition and which specific FAAs respond most to atmospheric N deposition.

Moss depends on atmospheric N deposition as its main N source and the low isotopic fractionation is accompanied with the uptake of N by moss13,29,30. Bulk N isotope compositions (δ15Nbulk) in moss have therefore been used to indicate the dominant sources of N deposition31,32,33,34,35,36. For example, δ15Nbulk for moss has been found to significantly negatively correlate with the wet deposition NH4+/NO3− ratio27,37. However, influences of atmospheric N deposition on N utilization and metabolism of moss were hindered by only analyzing the δ15Nbulk38,39. FAAs are important N-containing biomolecules that play central roles in N metabolism in plants and have been shown to be sensitive to atmospheric N pollution40. FAA δ15N values can be used to help evaluate the responses of N metabolism in plants to environmental N inputs. Bol, et al.41 found that the histidine (His) and phenylalanine (Phe) δ15N values can be used to differentiate functional strategy in relative to acquisition of available N sources. Xu and Xiao42 found that the δ15N values of some FAAs and total FAAs (TFAA) in needles were depleted when the contribution of traffic was lower. However, Yoneyama and Tanaka43 found that a significant isotopic fractionation was connected to the metabolism of FAA in plants. In previous studies differences between δ15NFAA values among individual FAA up to 36‰ have been found38,42. However, no study explore how δ15NFAA pattern reflects atmospheric N source when strong isotope fractionation occurs through FAA metabolism and which specific FAA best reflect the isotope signatures of atmospheric N sources in moss has yet been performed. It is therefore necessary to investigate the relationship between 15N abundances in individual FAAs and atmospheric N sources.

In this study, we determined the FAA N concentrations, δ15NFAA values, bulk N concentrations and δ15Nbulk values in moss samples. The FAA and bulk data were compared to determine whether FAAs can be effectively used to assess N deposition. The aims were (1) to assess the relationship between FAA N concentrations and atmospheric N deposition, (2) to determine how to using highly variable δ15NFAA values indicate atmospheric N sources, and (3) to determine which specific δ15NFAA value best reflects N sources to the atmosphere.

Results

Bulk N concentrations and δ15Nbulk

The bulk N concentrations in the moss samples were 1.1%–3.0%, and the mean was 1.9% ± 0.6%, as shown in Fig. 1a. The mean bulk N concentrations in moss from the seven sites in Nanchang City were decreased in the order urban centre (2.7% ± 0.4%), landfill (2.5% ± 0.3%), airport (1.9% ± 0.3%), zoo (1.4% ± 0.2%) and suburbs (1.2% ± 0.1%). The mean bulk N concentrations were significant higher in Urban than those in Suburbs (p < 0.05).

Figure 1
figure 1

Concentrations of TN, TFAA, Gln, Asn, Glu and Arg in moss from Suburban, Zoo, Airport, Landfill and Urban in Nanchang city: (a) TN, (b) TFAA, (c) Gln, (d) Asn, (e) Glu and (f) Arg. Bars represent mean values ± standard deviations. Significantly different mean values (HSD Tukey’s, p < 0.05) of TN and FAA from different sampling sites are indicated with superscript letters ‘A’ and ‘B’.

Full size image

As shown in Fig. 2, most moss samples had negative mean δ15Nbulk values. The mean δ15Nbulk was −4.0‰ ± 2.9‰ and the interquartile range was −5.7‰ to −1.3‰.

Figure 2
figure 2

Moss δ15N of individual FAAs in NanChang city. The vertical lines represent standard deviations. Moss δ15NTFAA and δ15NTN were showd in box plot. The box encloses 50% of the data, the whiskers 90% of the data, the solid lines is the median, the dashed line is the mean, solid circles are outliers. The δ15N ranges of the potential N sources are also included in the figure. The date of NHx δ15N values from excretory wastes is cited from Freyer59; Heaton60 and Moore61. NHx δ15N values from agricultural source is referenced from Xiao et al.58. The δ15N value of NOx is cited from Freyer59 and Saurer et al.67.

Full size image

FAA concentrations

The Ala, Arg, Asn, Asp, Gaba, Gln, Glu, Gly, His, Ile, Leu, Lys, Met, Phe, Pro, Ser, Thr, Trp, Tyr, and Val concentrations in the moss samples are shown in Table S1. The concentrations of TFAA (943.9–11100.5 μg g−1; Fig. 1b), Gln (not detected to 303.5 μg g−1; Fig. 1c), Asn (6.18–750.9 μg g−1; Fig. 1d), Glu (49.7–2159.9 μg g−1; Fig. 1e), and Arg (114.0–2117.3 μg g−1; Fig. 1f) in the samples from the different sites varied in similar ways to the bulk N concentrations (Fig. 1a). The FAA concentrations were significantly higher in the samples from the urban centre than from the suburbs (p < 0.05).

The N concentrations of Arg, Asn, Asp, Gln, Glu, Ser, and TFAA strongly positively correlated with atmospheric N deposition (P < 0.05) (Fig. 3). The equations for the relationships between the Arg, Asn, Asp, Gln, Glu, Ser, and TFAA concentrations and atmospheric N deposition are shown in Table S2.

Figure 3
figure 3

Relationships between concentrations of individual FAA (expressed as N concentrations) in moss and estimated total atmospheric N deposition. CFAA-N calculated by: CFAA-N = CFAA∙n∙14. CFAA is the molar concentration of each amino acid; n is the nitrogen atoms contained in each AA; 14 is the relative molecular mass of nitrogen atom. Total atmospheric N deposition (x) at each sampling sites was estimated using the linear correlation equation (y = 0.052x + 0.73, R2 = 0.70, P < 0.001; Xiao et al.58) between atmospheric N deposition (x) values and the corresponding moss TN concentrations (y) from the Yangtze River drainage basin.

Full size image

δ15N values for individual FAAs (δ15NFAA)

The δ15NFAA values for the moss samples varied widely, from −19.3‰ to +16.1‰ (Fig. 4). The FAAs were placed in four groups depending on how the δ15NFAA compared with the δ15Nbulk interquartile range. As shown in Fig. 2, the mean δ15N values for Ala (−4.1‰), Glu (−4.2‰), and Lys (−4.0‰) (group d) were close to the mean δ15Nbulk (−4.0‰). The mean δ15N values for Arg (−2.0‰), Gln (−1.7‰), Ile (−5.1‰), Leu (−5.6‰), Tyr (−2.9‰), and Val (−5.1‰) (group c) were between the δ15Nbulk interquartiles (−5.7‰ and −1.3‰). The δ15N values for Gly (−14.3‰), His (−7.5‰), Met (−7.2‰), Ser (−9.3‰), Thr (−7.1‰), and Trp (−9.0‰) (group b) were below than the lower δ15Nbulk quartile. The δ15N values for Asn (+0.5‰), Asp (−1.2‰), Gaba (−1.1‰), Phe (+2.8‰), and Pro (−1.2‰) (group a) were higher than the upper δ15Nbulk quartile.

Figure 4
figure 4

The δ15N values of free amino acids (‰) vs. the concentrations of free amino acids (expressed as N concentrations, μg/g) in mosses. δ15NTFAA = −3.1‰ is concentration-weighted average nitrogen isotope of free amino acids calculated by Rayleigh equilibrium equation.

Full size image

Concentration-weighted mean δ15N values for the TFAA (δ15NTFAA)

The FAA δ15N values varied widely, there being a 35‰ difference between the highest and lowest, as shown in Fig. 4. The δ15NTFAA values were calculated using the isotope mass-balance equation

$${{\rm{\delta }}}^{15}{{\rm{N}}}_{{\rm{TFAA}}}=\frac{\sum {{\rm{\delta }}}^{15}{\rm{Ni}}\cdot {\rm{Ci}}\cdot {{\rm{n}}}_{{\rm{i}}}}{\sum {\rm{Ci}}\cdot {{\rm{n}}}_{{\rm{i}}}},$$
(1)

where δ15Ni is the δ15N value for FAA i, Ci is the molar concentration of FAA i, and ni is the number of N atoms in FAA i. The mean δ15NTFAA was −3.1‰ ± 3.2‰ (Fig. 4) and the interquartile range (−5.2‰ to −1.3‰) was similar to the δ15Nbulk interquartile range (Fig. 2).

Fractionation of individual FAA normalized to δ15Nbulk

The method used to calculate positive and negative fractionation of individual FAA normalized to δ15Nbulk is shown in Fig. S1 The FAAs were divided into three groups depending on the δ15NFAA values relative to the mean δ15Nbulk (−4.0‰). In group δ1, the FAA δ15N values were >0‰. In group δ2, the FAA δ15N values were >−4.0‰ but <0‰. In group δ3, the FAA δ15N values were <−4.0‰. N isotope fractionation relative to δ15Nbulk for all three groups was calculated using equation 2.

$${{\rm{\Delta }}}^{{\rm{15}}}{\rm{N}}=\frac{\sum ({{\rm{\delta }}}^{{\rm{15}}}{\rm{Ni}}+{\rm{4}}){\rm{Ci}}}{\sum {\rm{Ci}}}$$
(2)

The total positive fractionation relative to δ15Nbulk (Δ15Npositive +3.4‰) was equal to the total negative fractionation relative to δ15Nbulk (Δ15Nnegative −3.6‰) (Fig. 5).

Figure 5
figure 5

Relationship between δ15Nsource, δ15NTFAA and δ15Nbulk N. Total 15N-enrichment for individual FAA and total 15N-depletion vs. δ15Nbulk (−4.0‰) are also showed.

Full size image

Spearman correlations between δ15N FAA, δ15Nbulk, and δ15NTFAA

Linear regression analyses indicated that the δ15Nbulk values significantly correlated with the δ15NTFAA values and the δ15N values for Ala, Gaba, His, Ile, Leu, Lys, and Ser (p < 0.05). δ15NTFAA significantly correlated with the δ15N values for most of the FAAs (Ala, Arg, Asn, Asp, Gaba, Glu, Gly, His, Ile, Leu, Lys, Pro, Ser, and Val) (p < 0.05). δ15NGlu correlated with the δ15N values for almost all of the FAAs except for Gaba, Gln, Ile, Met, Phe, Thr, Trp, and Tyr (p < 0.05) (Table S3).

Discussion

Strong relationships between individual FAA (mainly Arg, Asn, Asp, Glu, Gln, Ser, and TFAA) and atmospheric N deposition have been found in various moss species (Table S2)19,44,45,46. In our study, the concentrations of TFAA and some FAAs also varied spatially in similar ways to the bulk N concentration (Fig. 1) and positively correlated with atmospheric N deposition (Fig. 3). The abilities of FAAs in moss to respond to N inputs are related to the chemical and physiological characteristics of the FAAs. When high N deposition occurs, the Gln, Arg, and Asn concentrations increased because these FAAs have low C:N ratios40. Moreover, larger changes in free amino acid concentrations responded to increased atmospheric N additions than total N has been observed in various studies. Baxter, et al.47 found dramatic transient increases in the concentrations of Arg (by a factor of 19), Asn (by a factor of 4), and Gln (by a factor of 3) in moss exposed to 0.1 mM NH4+ for 20 d. Huhn and Schulz18 found that Arg accumulated much more strongly in Rӧsa (high N concentrations) than in Neuglobsow (low N deposition), the Arg concentration being 150 times higher in moss from Rӧsa than in moss from Neuglobsow. Similarly, in this study, we found that Glu (7-fold), Arg (9-fold), Gln (12-fold) and Asn (4-fold) increased large proportion than total nitrogen (2-fold). The synthesis of N-rich FAAs minimizes carbon use for storing N and avoids toxic concentrations of NH4+ accumulating in plant tissues48,49,50,51. Additional metabolic features have been found to be responsible for increases in the concentrations of these FAAs in high nitrogen deposition. For example, Arg is more soluble than other FAAs19,25,52, Glu plays a central role in N uptake18, Gln increases the photosynthetic capacities of plants44,53, and Ser is involved in the photorespiratory N cycle47. It is therefore possible that the concentrations, expressed as N concentrations, of some FAAs (Arg, Asn, Asp, Gln, Glu, Ser, and TFAA) in moss could indicate current atmospheric N deposition.

The development of isotopic analysis methods has led to δ15N values for amino acids being regarded as important tracers of the sources of and transformation processes affecting N-containing compounds in plant tissues41,42,54. However, the δ15N values for the 20 FAAs mentioned above were very different, suggesting marked N isotope fractionation occurred during the uptake, translocation, biosynthetic, and metabolic pathways38,55,56. Gauthier, et al.38 found that isotope fractionation between nitrate and Glu gave a δ15N value of 15.8‰ and that isotope fractionation associated with Asn synthesis from Asp gave δ15N values up to 36‰. In our study, the FAA δ15N values for moss covered a wide range, from −19.3‰ to +16.1‰ (Fig. 4). Little or no N isotope fractionation has been assumed to occur during N uptake and translocation in mosses34, so the large variations in the FAA δ15N values could mainly have been caused by FAA metabolism pathways.

Numerous atmospheric N compounds are directly taken up by moss, and little isotopic fractionation is associated with N assimilation. It has previously been found that δ15Nbulk values for mosses are good indicators of atmospheric N sources30,36,57,58. The mean δ15Nbulk value for the moss samples from Nanchang City was −4.0‰ ± 2.9‰ (range −9.3‰ to +0.9‰). According to δ15N inventories for potential N sources59,60,61, atmospheric N may be deposited in Nanchang mainly as NHy (negative δ15N values, group f in Fig. 2) originally emitted in animal excreta (−15.0‰ to −5.0‰)59,60,61 and during agricultural processes (−5‰ to 0‰)58. This conclusion was drawn because of the negative δ15N values. This result agreed with the results of a previous study of urban, rural, and forested sites in South China11. However, most of the δ15NFAA values were very different from the δ15Nbulk values and the δ15NFAA range was much wider (35‰) than the δ15Nbulk range (10‰). It would therefore have been somewhat difficult for the δ15NFAA values to indicate the N sources because of isotopic fractionation caused by metabolism, as discussed above.

It has been shown in numerous studies that FAA δ15N values are related to fractionation in the FAA metabolic pathways38,39,43,62,63,64. In this study the AA-δ15N pattern for free amino acid contrasted to the average value of δ15NTFAA, to discuss the fractionation with free amino acids metabolic pathways. Compared to the average value of δ15NTFAA, Gln, Phe, Tyr, Asn and Asp have higher δ15N value vs. δ15NTFAA (Fig. 2). Relative enrichment of 15N in Phe has been found to be related to kinetic isotope effects associated with the Phenylalanine ammonia-lyase catalyses Phe deamination, leaving the residual Phe relatively enriched in 15N39,62,65. Tyr is catalyzed by tyrosine ammonia-lyase to 4-hydroxycinnamate, which associated with marked 15N enrichment in Tyr.The δ15N value of Pro is positive than the value of δ15NTFAA, it could be explained by the kinetic isotope effect involved in the catabolism of Pro is greater than that its biosynthesis procedure or the biosynthesis of Pro is an thermodynamic procedure54. Relative 15N-enrichment in Asp is caused by the transfer of the amino group from Glu to oxaloacetate to form Asp, involving the formation of a protonated Schiff base, favouring 15N for Asp production56. Styring, et al.54 attributed 15N enrichment in Asn in cereal grains to Asn acting as a transport metabolite. The amino group of Asn is incorporated into other amino acids through transamination with a-keto acids, involving kinetic isotope fractionation discriminating against 15N. On the other hand, Gly and Ser have depleted δ15N values vs. δ15NTFAA (Fig. 2). Gly and Ser involve the photorespiratory cycle in the plants. 15N-depletion in Gly and Ser was possibly caused by 15N-depletion reaction during photorespiration related to Gly and Ser formation, e.g., isotope effect associated with transamination from Glu to produce Gly and discrimination against 15N associated with the reaction that converts Gly to Ser38,63,66.

Obviously using δ15NFAA values to indicate atmospheric N sources could therefore be affected by isotopic fractionation during FAA metabolic reactions in moss, as discussed above. The δ15N values for some FAAs may not reliably reflect atmospheric N sources. For example, using 15N-enriched FAAs (e.g., Phe, δ15NPhe 2.8‰ ± 2.7‰) to identify the main sources of atmospheric N deposition would incorrectly identify the source of N deposition in Nanchang City as being traffic-derived NO2 (δ15N +1.3‰ to +6.4‰)67, whereas using FAAs with more negative δ15N values (e.g., Gly, δ15NGly −14.3‰ ± 2.7‰) would indicate the sources being animal excreta (δ15N −15.2‰ to −8.9‰) and sewage (δ15N −15‰ to −4‰)59,60. We attempted to use δ15NTFAA as an indicator to solve this. As shown in Fig. 5, the sum of the positive differences between individual δ15NFAA values and δ15Nbulk (Δ15Npositive +3.4‰) was equal to the sum of the negative differences between individual δ15NFAA values and δ15Nbulk (Δ15Nnegative −3.6‰), implying that the TFAAs were isotopically equilibrated during FAA metabolism in the moss. The mean δ15NTFAA (−3.1‰ ± 3.2‰) was close to δ15Nbulk (−4.0‰ ± 2.9‰), and the δ15NTFAA interquartile range (−5.2‰ to −1.3‰) was similar to the δ15Nbulk interquartile range (−5.7‰ to −1.3‰) (group e in Fig. 2), that is, δ15NTFAA ≈ δ15Nbulk ≈ δ15NSource. The Pearson correlations indicated that δ15NTFAA significantly correlated with δ15Nbulk (Table S3). We therefore concluded that little isotopic fractionation occurs between TFAA and bulk N, meaning δ15NTFAA for moss can be used to indicate atmospheric N sources.

Most δ15NFAA values have not been compared with δ15Nbulk values, so it is not clear which δ15NFAA values in moss best indicate N source signatures. Only similar trends in δ15NFAA and δ15Nsources have been reported in previous publications. Chikaraishi, et al.68 found more 15N-depleted FAAs in moss from more industrial areas than in moss from more agricultural areas. Xu and Xiao42 found that Ala, Arg, Asp, Glu, His, Ile, Lys, Pro, Ser, and TFAA were more 15N-depleted in needles from sites far from highways than in needles from sites near highways, suggesting that atmospheric NHx-N from soil emissions affect δ15NFAA values more for needles far from highways than for needles near highways. However, δ15N values for most FAAs used as indicators were quite different from δ15N values for environmental N sources in a study by Xu and Xiao42. They found δ15NGln values < −8‰ for new needles at 800 from the highway. These δ15NGln values may possibly indicating a more 15N-depleted N source such as animal excreta (δ15N −15‰ to −5‰) rather than NHx-N from soil (δ15N −5.8‰ to −3.3‰)59,60. If using δ15N of specific free amino acid with large fractionation in their metabolism to indicate atmospheric N sources, a misleading conclusion would be obtained. It was unexpected that only a portion of the free amino acid δ15N values can hold N source signatures. We compared the δ15NFAA to δ15Nbulk values for the 20 FAAs to identify which δ15N values best indicated atmospheric N sources. The mean δ15NGlu, δ15NAla, and δ15NLys values were very similar to the δ15Nbulk values (Fig. 2). This may have been because no or little isotope fractionation was associated with the metabolic pathways of these FAAs. The main roles of Glu in FAA metabolism in plant tissues are to provide an amino group for the biosynthesis of other amino acids and to receive amino groups from the catabolism of other FAAs, which were confirmed in needles, cereal, pulse, algae and wheat tissues41,54,69,70,71. Our results confirmed this from the N isotope viewpoint in that δ15NGlu significantly correlated with the δ15N values for most of the FAAs and with δ15NTFAA (p < 0.05) (Table S3) and in that δ15NGlu (−4.0‰) was similar to δ15NTFAA (−3.1‰) (Fig. 2). We also found that the measured δ15NAla value was similar to the measured δ15NGlu, which would have been because kinetic isotope effects on the biosynthesis of Ala from pyruvate and Glu are weak39,43,72. Numerous previously also found that biosynthesizing branched-chain from pyruvate and Glu associated by low kinetic isotope effect39,72. Lys displayed no significant offset to the average value of δ15NTFAA. Gauthier, et al.38 found that, in plants, Lys acquires N derived from Glu, so δ15NLys will reflect δ15NGlu. This could help explain why δ15NLys was equal to the mean δ15Nbulk in our study. The δ15Nbulk value for moss reliably indicates atmospheric N sources, so we concluded that free Ala, Glu, and Lys (which are little affected by kinetic isotope effects during metabolism) may preserve information on atmospheric N sources.

Conclusions

The concentrations (expressed as N) of some FAAs (e.g., Arg, Asn, Asp, Gln, Glu, Ser, and TFAA) in moss were positively correlated with total atmospheric N deposition, indicating that the concentrations of those FAAs in moss could indicate atmospheric N deposition with a good degree of sensitivity.

We first used the FAA N isotope compositions to determine whether FAA metabolism in moss could reflect atmospheric N sources. The FAA δ15N values for the moss varied widely, probably mainly caused by the FAA metabolic pathways in the moss. However, total FAAs are at isotopic equilibrium during FAA metabolism and that the moss δ15NTFAA value could reliably indicate atmospheric N sources. We also found that the δ15N values of some FAAs (such as Ala, Glu, and Lys) preserve information on atmospheric N sources as well as δ15Nbulk preserves this information because little isotope fractionation occurs in the metabolic pathways of these FAAs.

Future work should include an investigation of FAA δ15N variability in vascular plants under different N deposition conditions to allow the kinetic isotope effects of N transport in different plant organs to be investigated.

Materials and Methods

Sample collection and treatment

Haplocladium microphyllum (Hedw.) moss samples were collected from urban, suburban, landfill, and airport sites in Nanchang City (South China) in July 2017. The sampling locations are shown in Fig. 6. Only green, healthy moss was sampled. The sampling sites were chosen based on the results of previous studies73,74. Each moss sample was collected from natural rocks in an open field away from overhanging vegetation or tree canopy. Sites were excluded if they could have been affected by point sources of N, such as soil, surface water, or domestic animals. Each sample was collected at least 500 m from any main road and at least 100 m from any other road or a house. Two–four sampling sites were selected in each plot, and 5–10 subsamples were collected at each site, then the subsamples were mixed (to ensure each sample was representative).

Figure 6
figure 6

The locations of moss sampling sites in Nanchang city. The locational map was modified from Google Earth 7.1.5.1557 (http://earth.google.com).

Full size image

Each moss sample was immediately placed in a chilled insulated box. Adsorbed pollutants were removed by gently rinsing each sample with deionized water several times. Half of each washed sample was dried at 80 °C for 1 d, and the other half was freeze-dried. Each dry sample was ground to a fine powder and stored at −80 °C until analysis68.

Bulk N and δ15Nbulk analyses

The bulk N concentration (expressed in % on a dry weight basis) and δ15Nbulk were determined simultaneously using a Flash EA 2000 elemental analyser (Thermo Scientific, Bremen, Germany) connected to a Thermo MAT253 plus isotope ratio mass spectrometer (Thermo Scientific, Bremen, Germany). The N concentration analytical precision was better than 0.1%. The δ15N method was calibrated by analysing caffeine (IAEA-600, δ15N = +1.0‰), ammonium sulfate (USGS25, δ15N = −30.4‰), and L-glutamic acid (USGS 41a, δ15N = +47.6‰) standards with each set of samples. The δ15N analytical precision (standard deviation; n = 3) was better than ±0.05‰. The isotope ratios were expressed in per mil (‰) relative to atmospheric N2. Each N concentration and δ15N value reported is the mean of at least three measurements.

FAA extraction, purification, and derivatization

The FAAs were extracted using a method described by Gauthier, et al.38. Briefly, 0.2–1 g of moss powder was suspended in distilled water, centrifuged for 5 min at 10000 g and 5 °C, then the supernatant was transferred to another centrifuge tube. The sample was extracted again, and the supernatants were mixed and heated to 100 °C for 5 min to precipitate proteins. The extract was then centrifuged for 5 min at 10000 g and 5 °C, then 100 μL of 1 nmol μL−1 α-aminobutyric acid was added to act as an internal reference (δ15N −8.17‰ ± 0.03‰). The extract was then freeze-dried and resuspended in 1 mL of 0.1 mol L−1 HCl. The extract was then passed through a cation exchange column (Dowex 50WX8 H+, 200–400 mesh; Sigma-Aldrich, St Louis, MO, USA), and the amino-acid-enriched fraction was stored at −80 °C until analysis.

tert-Butyldimethylsilyl (tBDMS)derivatives of the amino acids were prepared following methods described by Molero, et al.55 and Zhang, et al.75. Approximately 150 μg anhydrous Na2SO4, 50 μL pyridine, and 50 μL N-methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide were added in sequence to freeze dried amino acids, then the mixture was incubated at 70 °C for 1 h.

Determining amino acid concentrations and δ15N values

Amino acid concentrations and compound-specific structural and δ15N values were determined by analysing the tert-butyldimethylsilyl derivatives by gas chromatography (GC)/MS/IRMS. The GC/MS/IRMS instrument had a Trace GC instrument (Thermo Fisher Scientific), from which ~10% of the outflow entered a ISQ QD single quadrupole MS instrument (Thermo Fisher Scientific) to allow concentration and structural information to be acquired for each eluting peak. The remaining ~90% of the outflow entered a Thermo GC-isolink, in which the eluted compounds were oxidized and reduced to form CO2 and N2. The gases then entered a ConFlo IV interface (Thermo Fisher Scientific) and then a Delta V IRMS instrument (Thermo Fisher Scientific) to allow δ15N isotope data to be acquired.

The instrument conditions are described below. The injection volume was 0.2–1.0 μL, and splitless mode was used. The autosampler injector temperature was 270 °C. Separation was achieved using a DB-5 column (30 m long, 0.25 mm i.d., 0.25 μm film thickness; Agilent Technologies, Santa Clara, CA, USA). The carrier gas was helium, and the flow rate was 1.0 mL/min. The system was back-flushed with helium for 900 s during each analysis. The GC oven temperature started at 90 °C (held for 1 min), then increased at 12 °C min−1 to 150 °C (held for 5 min), increased at 3 °C min−1 to 220 °C, then increased at 12 °C min−1 to 285 °C (held for 7.5 min). The combustion reactor was held at 1,000 °C.

The linearity of the GC/MS method was assessed by evaporating, derivatizing, and analysing a series of standards containing 20 amino acids at concentrations of 0.04–1 mM. Each standard contained alanine (Ala), γ-aminobutyric acid (Gaba), arginine (Arg), asparagine (Asn), aspartate (Asp), glutamine (Gln), glutamate (Glu), glycine (Gly), histidine (His), isoleucine (Ile), leucine (Leu), lysine (Lys), methionine (Met), phenylalanine (Phe), prolineb (Pro), serine (Ser), threonine (Thr), tryptophan (Trp), tyrosine (Tyr), and valine (Val). The concentration of each amino acid was determined from the GC-MS signal using the relevant calibration curve produced from the standard amino acid mix data and corrected for the α-aminobutyric acid recovery. The R2 values for the calibration curves were 0.9909–0.9969, indicating that the method was accurate.

A derivatized mixture of 20 amino acid standards and several single amino acid standards (Ala Gly3, Gly4, Phe, USGS40, USGS41a, and Val) with known δ15N values (−26.35 to +47.55‰) was prepared to allow instrumental performance to be monitored and drift to be corrected. The amino acids were successfully converted into TBDMS derivatives and could be completely resolved by GC-C-IRMS (Fig. S2). The results are shown in Table S4. The α-aminobutyric acid (internal standard) δ15N value for each sample was used to confirm that the isotope measurements were reproducible. The 20 amino acid standard mixture was analysed after every three samples during a GC/MS/IRMS run to assess the isotope measurement reproducibility and normalize the δ15N values of the amino acids in the samples42. The amount of sample analysed by GC/MS/IRMS needed to be considered. Standard mixture containing the 20 amino acids each at an equivalent of 0.8 nmol (equivalent to FAA concentration of 9–20 μg g−1 in moss) was analysed to allow the δ15N values of low concentrations of amino acids to be determined. The FAA concentrations expressed as N concentrations in our samples were higher than these concentrations. The δ15N measurement precisions (n = 9) for the derivatized amino acid standard mixtures were 0.5‰–1.4‰ (Table S4). The δ15N values for the underivatized amino acids measured by elemental analysis/IRMS correlated with the δ15N values for the derivatized amino acids measured by GC/MS/IRMS (R2 = 0.997, P < 0.001). The differences between the empirically corrected δ15N values measured by elemental analysis/IRMS and GC/MS/IRMS were 0.1‰–1.3‰ (Table S4). Each value reported here is the mean of at least three δ15N determinations.

Statistical analysis

Statistical analyses were performed using SPSS 16.0 software (IBM, Armonk, NY, USA). The statistical significances of differences in the FAA contents of samples from different sites were tested using the one-way analysis of variance method and Tukey-HSD tests, and differences were considered significant at P < 0.05. Correlations between δ15NFAA, δ15NTFAA, and δ15Nbulk were assessed using Pearson correlation coefficients (r). Linear regressions were used to identify correlations between the FAA concentrations and estimated atmospheric N deposition. Most graphs were drawn using SigmaPlot 10.0 software (Systat Software, San Jose, CA, USA).