Ammonium first: natural mosses prefer atmospheric ammonium but vary utilization of dissolved organic nitrogen depending on habitat and nitrogen deposition


  • Xue-Yan Liu,

    1. State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, China
    2. Institute of Agriculture, Tokyo University of Agriculture and Technology, Fuchu, Japan
    Search for more papers by this author
  • Keisuke Koba,

    Corresponding author
    1. Institute of Agriculture, Tokyo University of Agriculture and Technology, Fuchu, Japan
    • State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, China
    Search for more papers by this author
  • Akiko Makabe,

    1. Institute of Agriculture, Tokyo University of Agriculture and Technology, Fuchu, Japan
    Search for more papers by this author
  • Xiao-Dong Li,

    1. State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, China
    Search for more papers by this author
  • Muneoki Yoh,

    1. Institute of Agriculture, Tokyo University of Agriculture and Technology, Fuchu, Japan
    Search for more papers by this author
  • Cong-Qiang Liu

    1. State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, China
    Search for more papers by this author

Author for correspondence:

Keisuke Koba

Tel: +81 42 367 5951



  • Mosses, among all types of terrestrial vegetation, are excellent scavengers of anthropogenic nitrogen (N), but their utilization of dissolved organic N (DON) and their reliance on atmospheric N remain uncharacterized in natural environments, which obscures their roles in N cycles.
  • Natural 15N abundance of N sources (nitrate (math formula), ammonium (math formula) and DON in deposition and soil) for epilithic and terricolous mosses was analyzed at sites with different N depositions at Guiyang, China. Moss math formula assimilation was inhibited substantially by the high supply of math formula and DON. Therefore, contributions of math formula and DON to moss N were partitioned using isotopic mass-balance methods.
  • The N contributions averaged 56% and 46% from atmospheric math formula, and 44% and 17% from atmospheric DON in epilithic and terricolous mosses, respectively. In terricolous mosses, soil math formula and soil DON accounted for 16% and 21% of bulk N, which are higher than current estimations obtained using 15N-labeling methods. Moreover, anthropogenic math formula deposition suppressed utilization of DON and soil N because of the preference of moss for math formula under elevated math formula deposition.
  • These results underscore the dominance of, and preference for, atmospheric math formula in moss N utilization, and highlight the importance of considering DON and soil N sources when estimating moss N sequestration and the impacts of N deposition on mosses.


Nitrogen (N) is an important plant nutrient. Since the 19th Century, anthropogenic N deposition has been increasing globally, triggering major changes in terrestrial ecosystems, including aspects of N and carbon (C) dynamics (Aber et al., 1998; McLauchlan et al., 2007; Lovett & Goodale, 2011) and floristic diversity (Bobbink et al., 2010; De Vries et al., 2010). Therefore, it is important to gain insights into how the N inputs change the N dynamics in terrestrial ecosystems (Phoenix et al., 2012; Templer et al., 2012). Ascertaining changes of the ecosystem N dynamics with increased N inputs is also necessary to estimate changes in ecosystem functions (Manning et al., 2006; Pardo et al., 2012). The preferences of plants and microbes for different N forms among ammonium (math formula), dissolved organic N (DON), and nitrate (math formula) play an important role in determining the ecosystem N dynamics (Northup et al., 1995; Lovett & Mitchell, 2004) and in determining the fates of N input into natural ecosystems (Durka et al., 1994; Liu et al., 2012c). Nevertheless, it remains extremely difficult to evaluate plant N preferences, partly because these three forms of soil N have different physical, chemical, and biological characteristics, and therefore have different availabilities to plants (Kaye & Hart, 1997; Abaas et al., 2012).

Nitrogen utilization by plants includes N uptake and assimilation. For N assimilation, the incorporation of different N forms into plant biomass differs in assimilation costs (Gutschick, 1981; Clarkson, 1985; Li et al., 2013). The assimilation cost of amino acids, a small but important DON component, is expected to be lower than that of math formula, which must be attached to a C skeleton before use, and much lower than that of math formula, which requires additional reduction steps to math formula (Clarkson, 1985; Bloom et al., 1992; Chapin et al., 1993). These differences partially explain the preferences for math formula or amino acids observed in vascular plants when different N forms are supplied in equal doses (Kronzucker et al., 1997; Houlton et al., 2007; Wang & Macko, 2011). However, the roots of vascular plants actually take up N through different pathways (e.g. mycorrhizal symbionts) and from different soil depths with different availabilities (McKane et al., 2002 ; Kohzu et al., 2003), which obscures the real reasons for N preferences in natural vascular plants. In contrast to vascular plants, mosses lack a rooting system. With leaves only one cell thick and no cuticular barrier in most taxa, moss cells are exposed directly to environmental N sources (Glime, 2007). Moss N uptake might not cause substantial preference because nutrients can enter moss tissues easily through cation exchange and the proton (H+) pump (for math formula and amino acids) and through cotransport (for math formula) with positively charged ions (Raven et al., 1998; Glime, 2007). When the supply rates of different N forms were identical, N preferences observed in mosses were derived mainly from the assimilation (Pearce et al., 2003; Wiedermann et al., 2009). Nevertheless, no suitable method exists to confirm the N preference of mosses in natural environments with variable availabilities of different N forms. It is therefore unknown if moss N preferences will vary along with the N source availability.

The natural abundance of N isotope (δ15N; the 15N : 14N ratio expressed relative to atmospheric N2) in natural plants can integrate available N sources and physiological processes (Högberg, 1997; Robinson, 2001; Craine et al., 2009). Compared with the isotopic labeling method (Wanek & Zotz, 2011), δ15N analysis avoids artificial N addition and therefore presents no risk of changing the soil N pool and plant N-uptake kinetics. In particular, using the δ15N of mosses to evaluate their N sources is likely to yield better results than for vascular plants, because most mosses have no cuticular barrier, stomatal regulation, or root mycorrhizal mediation, with these organs and regulations apparent 15N fractionations can be expressed during N acquisition and reallocation in vascular plants (Handley & Scrimgeour, 1997; Evans, 2001; Gebauer & Meyer, 2003; Hobbie & Högberg, 2012). Consequently, the δ15N analysis of mosses can reveal dominant N sources and or N preferences in N assimilation. In studies in China, for example, the δ15N of mosses on bare rock showed low values in regions where inorganic N deposition was dominated by math formula (Liu et al., 2008; Xiao & Liu, 2011). However, the extent to which mosses rely on math formula and the extent to which they use math formula and DON remain open questions. Such knowledge is of global importance because math formula is dominant in N deposition in most regions of the world (Pearson & Stewart, 1993; Stevens et al., 2011). The preferred utilization of math formula over math formula has also been emphasized for vascular plants because math formula preference interacts closely with ecosystem processes and functioning (Houlton et al., 2007; Kahmen et al., 2008; Boudsocq et al., 2012). Measurements of stable isotopes (δ15N and δ18O) of tissue math formula in mosses (Liu et al., 2012a) revealed that moss math formula assimilation is inducible when math formula is the sole N source (Liu et al., 2012b), but moss math formula assimilation was found to be negligible when the supply rate of reduced dissolved N (RDN; math formula plus DON) was significantly higher than that of math formula in natural environments (e.g. Liu et al., 2012c). This low assimilation of math formula in mosses across different habitats resulted from the inhibition of nitrate reductase activity (NRA) by the high supply rate of RDN (detailed in Liu et al., 2012c; detailed mechanisms are reviewed by Dortch, 1990). Consequently, measuring δ15N of math formula and DON in deposition allows further partitioning of moss math formula and DON assimilation, an exploration of math formula preference in epilithic mosses, and an evaluation of the bioavailability of DON in RDN-dominated deposition.

The exploration of DON assimilation in mosses is important for two main reasons. First, present knowledge of moss N sequestration and N deposition effects on mosses largely relates to inorganic N (Paulissen et al., 2004; Bragazza et al., 2005; Gundale et al., 2011) and N2 fixation (for feather mosses; DeLuca et al., 2008; Ackermann et al., 2012). However, the assimilation of DON and its interaction with inorganic N (math formula and math formula) assimilation remain unclear. This poses the question of whether the importance of inorganic N deposition to terrestrial mosses has been overrated because of neglect or underestimation of the true contribution of DON. Second, exploration of DON assimilation can expand the characterization of moss N preference (Soares & Pearson, 1997; Arróniz-Crespo et al., 2008). Mosses prefer math formula when using inorganic N sources because of the high energy cost of math formula reduction and the potential for avoiding excessive math formula accumulation (Pearson & Stewart, 1993; for vascular plants, Kronzucker et al., 2001). However, it is unclear whether there is a moss N preference for math formula over DON because in natural environments the bioavailability of DON has not been fully characterized. Through 15N labeling, some laboratory and field studies have revealed considerable utilizations of amino acids in mosses and influences of amino acid accumulation on inorganic N metabolism (Baxter et al., 1992; Nordin & Gunnarsson, 2000). Forsum et al. (2006) applied 50 kg N ha−1 yr−1 (math formula: math formula : glycine = 1 : 1 : 1) to mosses and found a clear preference for amino acid N over math formula, although the assimilation of glycine remained lower than that of math formula. Similarly, Wanek & Pörtl (2008) reported that the uptake rates of math formula or glycine were two times higher than those of math formula. By 15N-labeling of math formula, math formula, alanine, and glutamic acids, Wiedermann et al. (2009) found that mosses preferred math formula and DON, with very low uptake of math formula under different levels of N deposition. These studies demonstrated that, in natural conditions, math formula is a negligible N source for mosses, but amino acids may contribute substantially to moss N sequestration. It is noteworthy that amino acids account for only a small proportion of DON, and that the bioavailable fraction of DON is expected to be larger than that of amino acids (Neff et al., 2002). To date, only the δ15N of ‘bulk’ DON can be measured routinely (Knapp et al., 2005; Koba et al., 2010a,b; Lachouani et al., 2010), but this allows exploration of whether plants indeed prefer math formula when isotopes of major dissolved N in natural environments are characterized (Houlton et al., 2007; Kahmen et al., 2008; Takebayashi et al., 2010). Thereby, the importance of DON in moss N assimilation can be estimated.

Four sets of moss–soil systems (mosses on bare rock, mosses growing on the soil of the rock surface, and terricolous mosses in open fields and on forest floors) under high (urban area; 21 kg N ha−1 yr−1) and low (suburban and rural areas; 10–12 kg N ha−1 yr−1) N deposition rates were investigated in the Guiyang area of China. Because of the dominance of math formula or RDN in total dissolved N (TDN) deposition (Table S1), depleted δ15N values have been observed in mosses, especially species on bare rock (Liu et al., 2008). Moss NRA was inhibited substantially by high atmospherically derived (atm-) RDN. Therefore, a negligible contribution of math formula was found in moss N assimilation in this area (Liu et al., 2012c). By measurement of the pool sizes and δ15N signatures of dissolved N (math formula, DON, and math formula) in soil attached with mosses, the present work aimed (1) to verify the importance of atm-math formula for mosses in natural habitats through determination and consideration of the contributions of atm-DON and soil N, and (2) to explore whether mosses prefer atm-math formula over DON or soil N by comparing their contributions in mosses with their availabilities across the gradient of anthropogenic math formula (Liu et al., 2008a). Our main hypothesis is that moss N assimilation is dominated by atm-math formula, not only because of high deposition of atm-math formula, but also because of the preference for atm-math formula over DON and soil N.

Materials and Methods

Study area and sample collection

The Guiyang area, located in southwestern China, has a typical subtropical monsoon climate and an average altitude of 1250 m (a detailed description is given in Supporting Information Notes S1). The wet deposition was collected from December 2008 to September 2009 at an urban site in Guiyang (detailed in Notes S1). In July 2010, epilithic mosses on bare rock (EB), mosses growing on the soil of the rock surface (ES), and terricolous mosses in open fields (TO) and on the floor of pine forests (TF), as well as soil attached with mosses (< 3 cm), were collected along a northeastern urban (U)–suburban (S)–rural (R) transect in the Guiyang area (sampling sites are shown in Fig. S1). These habitats represent typical habitats of natural mosses. For moss sampling, mature and green tissues at five to 10 subsites were collected and then mixed as composite samples for each site. To avoid possible interspecific differences, moss species with uniform or similar morphological traits were considered. Each sample of mosses on rock (EB and ES) was a mixture of Eurohypnum (mainly Eurohypnum julaceum and Eurohypnum leptothallum (c. mull.) ando), Hypnum (Hypnum plumaeforme Wils.), and Haplocladium (mainly Haplocladium microphyllum Hedw.), whereas terricolous moss samples included species of Hypnum (mainly Hypnum plumaeforme Wils.), Thuidium (mainly Thuidium cymbifolium (Dozy et Molk.) Dozy et Molk.), and Haplocladium microphyllum Hedw. These species, all pleurocarpous, have a widespread distribution all over the world. For soil sampling, only soil to a depth of 3 cm was collected because our observations and experience in the field suggested that soil-anchored rhizoids or moss layers were generally 3 cm deep. Mosses on forest floors were sampled from three unmanaged pine (Pinus massoniana Lamb. var) forests. Three sampling plots were set in each forest. Then mosses and soil (< 3 cm) at three to five subsites were combined to form one sample in each plot. For ES and TO, eligible soil and moss samples were found to be insufficient at each sampling site (marked in Fig. S1) for all experimental analyses. Therefore, samples were mixed for U1 and U2, U3 and U4, and U5 and U6 in urban areas, for S1 and S2, S3 and S4, and S5 and S6 in suburban areas, and for R1 and R2 and R3 and R4 in rural areas (Fig. S1). Consequently, the number of replicates was three for samples of ES, TO, and TF.

Experimental analyses

Within 8 h after sampling, sample pretreatments were conducted in the laboratory at the Institute of Geochemistry (CAS), Guiyang, China. Some of the fresh soil was used to determine water contents. The remainder was passed through a 2-mm mesh sieve to remove impurities and coarse fragments. Some of the sieved soil was used to determine pH (H2O). The remainder was used immediately for extraction of dissolved inorganic N (DIN: math formula plus math formula) and DON, with a ratio of 10 g of fresh soil to 100 ml of 2 M KCl solution. The mixture of soil and KCl solution was shaken for 1.5 h, then centrifuged. The supernatant was filtered using a glass filter (GF/F; Whatman, Maidstone, UK). The KCl salts and filters were heated to 450°C for 4 h to reduce the N blank before use. A subsample of the sieved soil was dried at 60°C to measure water, C, and N contents as well as δ15N values. The preparation of moss samples has been described by Liu et al. (2012c).

The ball-milled soil samples, diffusion samples of soil math formula (see details in Notes S1), soil extraction solutions (frozen at −20°C) and the sieved fresh soil (at 4°C) were shipped to Tokyo University of Agriculture and Technology within 3 d after sampling. The methods described by Takebayashi et al. (2010) were used to incubate soil for determination of net N nitrification and N mineralization rates. The concentrations and stable isotopes of bulk N in soil, math formula, math formula and TDN in soil extracts were analyzed using the methods described by Koba et al. (2010a) and by Takebayashi et al. (2010) (details in Notes S1). The RDN was calculated as the difference between TDN and math formula. DON was calculated as the difference between TDN and DIN. The δ15N values of DON and RDN were calculated, respectively, using the following mass and isotopic balance equations:

display math
display math

The δ15N was expressed as (Coplen, 2011):

display math

where 15N/14N in samples and the standard (atmospheric N2, and Rstandard = 0.0036765).


Dissolved N forms in soil attached with mosses

Table 1 presents characteristics (pH, C : N ratio, net N nitrification and mineralization rates) of moss soil. Details are shown in Notes S2. Dissolved N concentrations in the soil of epilithic mosses were higher than those in soils of terricolous mosses, with the lowest in soils of forest mosses (Table 1). The soils of epilithic mosses showed higher dissolved N concentration in urban than in rural areas (< 0.05), although such spatial differences were not observed in the soils of terricolous mosses. Moreover, the pool of RDN was significantly (< 0.05) larger than that of math formula in moss soil, showing much higher RDN: math formula ratios (6–71; Table 1) than that in wet deposition (4.9; Table S1).

Table 1. Bulk nitrogen (N), net N nitrification and mineralization rates, pool sizes of dissolved N, and the natural abundance of N isotope (δ15N) values in soil under mosses in the Guiyang area
HabitatSitepHBulk N (mg g−1)C : NNet N ratesNit (%)Pool size (mg N kg−1 soil)RDN: math formulaNδ15N/‰
NitMin math formula math formula DONRDNTDNRDNTDNBulk N
  1. Reported are means ± SD (= 3). Net N nitrification and mineralization rates are shown in mg N kg−1 d−1.

  2. Nit, nitrification; Min, mineralization.

  3. math formula, nitrate; math formula, ammonium; DON, dissolved organic N; RDN, reduced dissolved N; TDN, total dissolved N.

  4. ES, mosses growing on the soil of the rock surface; TO, terricolous mosses in open fields; TF, terricolous mosses in pine forests.

ESUrban6.5 ± 0.27.0 ± 0.517.2 ± 1.31.1 ± 0.44.9 ± 0.622 ± 510.2 ± 1.241.0 ± 10.555.0 ± 19.896.0 ± 24.5106.2 ± 25.79 ± 14.6 ± 1.63.6 ± 1.20.7 ± 0.1
Suburban6.7 ± 0.15.6 ± 0.619.4 ± 1.11.2 ± 0.23.8 ± 0.133 ± 37.6 ± 1.639.8 ± 4.548.7 ± 8.388.5 ± 12.896.1 ± 14.312 ± 15.8 ± 0.35.3 ± 0.33.8 ± 0.1
Rural6.3 ± 0.25.0 ± 0.921.7 ± 1.51.0 ± 0.12.4 ± 0.026 ± 27.0 ± 1.437.6 ± 8.844.1 ± 11.181.7 ± 13.888.7 ± 12.612 ± 45.1 ± 0.94.8 ± 0.91.3 ± 0.3
TOUrban6.1 ± 0.26.0 ± 1.615.6 ± 0.60.9 ± 0.12.9 ± 0.533 ± 210.1 ± 3.035.1 ± 2.220.6 ± 6.455.7 ± 8.165.8 ± 11.26 ± 15.4 ± 1.04.1 ± 1.51.1 ± 0.7
Suburban6.2 ± 0.24.1 ± 0.820.2 ± 2.41.3 ± 0.61.9 ± 0.664 ± 135.3 ± 2.540.5 ± 18.622.4 ± 3.862.9 ± 22.268.2 ± 24.612 ± 26.2 ± 0.65.7 ± 0.75.1 ± 0.2
Rural6.0 ± 0.35.2 ± 1.115.5 ± 1.40.4 ± 0.11.4 ± 0.029 ± 73.7 ± 0.510.5 ± 1.344.2 ± 6.354.7 ± 7.558.3 ± 7.015 ± 45.5 ± 1.25.1 ± 1.02.4 ± 0.4
TFUrban5.3 ± 0.24.0 ± 1.112.2 ± 3.30.3 ± 0.12.5 ± 0.410 ± 31.4 ± 0.77.9 ± 0.714.3 ± 1.422.1 ± 1.823.5 ± 1.618 ± 85.3 ± 1.64.9 ± 1.73.4 ± 0.6
Suburban5.9 ± 0.23.7 ± 0.712.1 ± 1.70.1 ± 0.12.4 ± 0.45 ± 20.4 ± 0.27.3 ± 3.116.9 ± 5.124.2 ± 2.124.6 ± 2.271 ± 334.3 ± 0.94.2 ± 0.93.4 ± 1.1
Rural5.6 ± 0.33.7 ± 0.816.1 ± 4.70.1 ± 0.12.7 ± 0.37 ± 30.7 ± 0.41.0 ± 0.122.2 ± 2.023.2 ± 2.123.9 ± 2.343 ± 313.5 ± 0.73.4 ± 0.62.8 ± 0.6

Variations of δ15N in bulk N (0.6–5.2‰), in TDN (2.2–6.4‰), and in RDN (2.8–6.9‰) showed no clear pattern in urban or rural areas (Table 1). The δ15N values of TDN (4.6 ± 1.2‰) and DON (3.8 ± 2.0‰) were generally more positive than that of bulk N (2.7 ± 1.4‰). Soil math formula showed the most positive δ15N values (2.5–11.2‰), in contrast to the lowest δ15N signatures of math formula (−8.9 to 2.1‰) (Fig. 1). In general, soil δ15N-math formula was lower at urban sites than at rural sites, and lower in soil in open fields than in soil in pine forests (Fig. 1).

Figure 1.

The natural abundance of nitrogen (N) isotope (δ15N) signatures of moss N, dissolved N (math formula, dissolved organic N (DON) and math formula) in wet deposition and moss soil in the Guiyang area. EB, epilithic mosses on bare rock; ES, mosses growing on the soil of the rock surface; TO, terricolous mosses in open fields; TF, terricolous mosses in pine forests. Solid and dashed lines within the boxes show the median and the mean, respectively. The box boundary shows the 25th and 75th percentiles. Dots aside from the caps indicate each outlier of data points. The δ15N of math formula in deposition is after Xiao et al. (2012). The δ15N of mosses, total dissolved N (TDN) and math formula in deposition follow Liu et al. (2012c). The δ15N of DON in deposition was calculated using the isotopic mass -balance equation (δ15NDON = {δ15NTDN × [TDN] – δ15NNO3- × [math formula] – δ15NNH4+ × [math formula]}/[DON]) based on monthly mean concentrations and δ15N values. Details of sampling and analyses of deposited N are provided in Liu et al. (2012c).

Moss bulk N and δ15N signatures

Moss N concentration varied from 14.1 to 26.5 mg g−1, differing among habitats (mean values from Liu et al., 2012c are listed in Table S2) and generally increased under elevated N deposition in urban areas. Epilithic mosses with soil had higher N than mosses on bare rock, although mosses in pine forests had the lowest N. Moss δ15N ranged between –12.9‰ and –1.3‰ (Fig. 1 and Table S2; Liu et al., 2012c). Overall, the δ15N was lower in urban mosses than in rural mosses, and lower in mosses on bare rock than in mosses growing on soil.

Calculations of the deposited N and soil N contributions to moss bulk N

The assimilation of math formula was negligible as a result of the inhibition of moss NRA by RDN associated with much higher deposition of RDN than that of math formula (described in Liu et al., 2012c). Therefore, mosses in our study area mainly assimilated math formula and DON. Consequently, the proportional contributions (f, expressed as percentages hereafter) of atm-DON (fatm-DON) and atm-math formula (fatm-NH4+ = 1 − fatm-DON) to bulk N of mosses on bare rock (Nmoss-rock) were calculated using a two-source mixing model:

display math(Eqn 1)

Because the mean δ15Natm-NH4+ was −14.8‰ and the mean δ15Natm-DON was 6.8‰ (Table S1; Liu et al., 2012c), then

display math(Eqn 2)

In contrast to mosses on bare rock, N in other mosses (Nmoss-soil) was derived from both deposition (mainly atm-RDN: atm-math formula and atm-DON) and soil (mainly soil-derived (soil-) RDN: soil-math formula and soil-DON). The total contribution of soil-RDN and deposited RDN can be estimated using a two-source mixing model (detailed in Notes S3; shown in Fig. S2). The proportional contributions of explicit N species are calculable using the ‘IsoSource’ model (Phillips & Gregg, 2003) (Eqn 3).

display math(Eqn 3)

The entry of N into moss tissues has no substantial 15N fractionation (Liu et al., 2012b). Therefore, this model iteratively generates source isotopic mixtures of which the proportions (f) sum to 1 (fatm-NH4+ + fatm-DON + fsoil-NH4+ + fsoil-DON = 1). It compares each calculation against a known mixture (δ15Nmoss-soil) and retains only those mixtures that satisfy the known value within some mass balance tolerance, as defined using a data set of feasible solutions. Although this model can only generate feasible solutions, it nevertheless provides a systematic means of constraining the attribution of N sources in an underdetermined system. In our case, the calculated mixtures reflecting combinations of the δ15N values of atm-math formula, atm-DON, soil-math formula, soil-DON, and moss N, we applied a mass balance tolerance of 0.2%. The original data output from the model for all replicate samples is presented in Fig. S3.

For mosses on bare rock, the fatm-NH4+ (56 ± 13%; 44–90%) was higher than fatm-DON (44 ± 13%; 10–55%) (Table 2) and significant differences (< 0.05) were found between urban and rural areas (Fig. 2a). In urban areas, the fatm-NH4+ was 68 ± 15%, which decreased to 49 ± 1% and 56 ± 3% in suburban and rural areas, respectively, with a corresponding increase in the fatm-DON (Fig. 2a).

Table 2. Average proportional contributions (f; %) of different nitrogen (N) sources in mosses growing on bare rock and soil in the Guiyang area
Moss habitatExplicit N speciesatm versus soilmath formula vs DON
f atm-NH4+ f atm-DON f soil-NH4+ f soil-DON f atm-N f soil-N f total-NH4+ f total-DON
  1. Reported data are means ± SD. fatm-N = fatm-NH4+ + fatm-DON; fsoil-N = fsoil-NH4+ + fsoil-DON; ftotal-NH4+ = fatm-NH4+ + fsoil-NH4+; ftotal-DON = fatm-DON + fsoil-DON. Mosses on soil include ES, TO and TF. math formula, nitrate; math formula, ammonium; DON, dissolved organic N; atm-N, N from atmospheric deposition; soil-N, N from soil; ES, mosses growing on the soil of the rock surface; TO, terricolous mosses in open fields; TF, terricolous mosses in pine forests.

Bare rock (= 17)56 ± 1344 ± 13
Soil (= 27)46 ± 1017 ± 516 ± 421 ± 763 ± 837 ± 862 ± 838 ± 8
Figure 2.

Comparisons of the proportional contributions of atmospherically derived (atm-) math formula with those of (a) atmospherically derived total dissolved nitrogen (atm-DON) in mosses on bare rock (= 6 for urban and suburban sites; = 5 for rural sites), (b) atm-DON in mosses on soil (= 9), (c) total DON (atm-DON plus soil-DON) in mosses on soil (= 9), and (d) total soil N (soil math formula plus soil-DON) in mosses on soil (= 9). Mosses on soil include mosses growing on the soil of the rock surface (ES), terricolous mosses in open fields (TO), and terricolous mosses on floors of pine forests (TF). Reported are means ± SD. Mean values of percentage ranges (for each replicate sample) output from the two-source mixing model and the ‘IsoSource’ model were used.

On average, mosses on soil derived more N from deposition (63 ± 8%; fatm-N = fatm-NH4+ + fatm-DON) than from soil (37 ± 8%; fsoil-N = fsoil-NH4+ + fsoil-DON), and more N from math formula-N (62 ± 8%; ftotal-NH4+ = fatm-NH4+ + fsoil-NH4+) than from DON (38 ± 8%; ftotal-DON fatm-DON + fsoil-DON) (Table 2). The atm-math formula showed the highest contribution of all N sources (Fig. S3), with mean fatm-NH4+ of 46 ± 10% (36–54%) in mosses on soil (Table 2). For atm-DON, soil-math formula, and soil-DON, the mean contributions were, respectively, 17 ± 5%, 16 ± 4%, and 21 ± 7% (Table 2). Similar to the pattern for mosses on bare rock (Fig. 2a), the fatm-NH4+ in all mosses on soil showed a clear decrease from the urban (56 ± 11%) to suburban (42 ± 4%) and rural areas (40 ± 4%) (Fig. 2b). By contrast, the contributions of atm-DON (fatm-DON; Fig. 2b), total DON (Fig. 2c) and soil N (Fig. 2d) showed an opposite pattern to that of atm-math formula from urban to rural areas. Moreover, neither proportional nor real contributions of soil N (soil-math formula and soil-DON) in Nmoss (real contribution = proportional contribution × Nmoss; in mg N g−1 DW) responded clearly to soil N availability (Fig. S4).

Estimation of moss preference for NH+4 versus DON

Plant N preference for a given N source ‘A’ over the other N source ‘B’ (βA) can result from both uptake and assimilation processes (Boudsocq et al., 2012; Li et al., 2013). The N preference during uptake is related to external abundance and mobility of specific N species, and root properties, whereas the N preference during assimilation is associated with the energy cost of specific N incorporation (Gutschick, 1981; Clarkson, 1985; Bloom et al., 1992). Previous to this work, the uptake preference was often evaluated using the ratio of N uptake rates between ‘A’ and ‘B’ when they were equally available (e.g. preference for math formula over math formula; Chapin et al., 1986; Kronzucker et al., 1997; Näsholm et al., 1998; preference for glycine over math formula; Chapin et al., 1993; Kielland, 1997; Raab et al., 1999). Moss N uptake might not result in preferences because mosses have very simple tissue structure. Moreover, cation exchanges (for math formula and amino acids) occur simultaneously with anion cotransport (for math formula). Consequently, to ascertain the differences between the assimilation and relative availability of specific N forms, we can explore N preference although in natural conditions N forms are supplied with variable ratios. A preferential assimilation will result in a higher contribution of ‘A’ in overall assimilation of ‘A’ and ‘B’ than the proportional contribution of ‘A’ in the total availability of ‘A’ and ‘B’. Accordingly, the preference for atm-math formula over atm-DON (βatm-NH4+) can be inferred for all mosses as the difference between (the proportional contribution of atm-math formula assimilation in atm-RDN assimilation) and (the proportional contribution of atm-math formula in atm-RDN; 62%; Table S1).

display math(Eqn 4)

The preference for soil-math formula over soil-DON (βsoil-NH4+) can be described for mosses on soil as the difference between the proportional contribution of soil-math formula assimilation in soil-RDN assimilation and the proportional contribution of soil-math formula in soil-RDN, as

display math(Eqn 5)

where fatm-RDN fatm-NH4+ + fatm-DON and fsoil-RDN fsoil-NH4+ + fsoil-DON. The fatm-NH4+, fatm-DON, fsoil-NH4+ and fsoil-DON were calculated using Eqns 1-3. The averaged values of each replicate sample were used for the βNH4+ calculation. Positive values, 0, and negative values of β, respectively, show a preference for math formula, no preference, and a preference for DON.

For atmospheric N sources, mosses in all habitats had the highest βatm-NH4+ values in the urban area (Fig. 3a). In suburban and rural areas, the βatm-NH4+ value generally decreased; in particular, the βatm-NH4+ values became negative for mosses on bare rock (Fig. 3a). By contrast, for soil N sources, the βsoil-NH4+ values were negative for terricolous mosses in open fields in the urban area, whereas mosses in pine forests had positive βsoil-NH4+ values (Fig. 3b). A negative correlation was found between the βatm-NH4+ and βsoil-NH4+ for mosses on soil (Fig. 4).

Figure 3.

Preference (β) for (a) atmospherically derived (atm-) math formula over atmospherically derived total dissolved nitrogen (atm-DON) (βatm-NH4+; (Eqn 4)), and (b) soil-NH4+ over soil-DON (βsoil-NH4+; (Eqn 5)) in mosses of different habitats. Positive β values denote a math formula preference. Negative values show a DON preference; β = 0 shows no preference (dashed lines). Solid and dotted lines within the boxes denote the median and the mean, respectively.

Figure 4.

Correlation between the preference for atmospherically derived math formula over atmospherically derived dissolved organic N (DON) (βatm-NH4+; (Eqn 4)) and the preference for soil-NH4+ over soil-DON (βsoil-NH4+; (Eqn 5)) in mosses on soils (including epilithic mosses with soil, and terricolous mosses in open fields and on floors of pine forests) in the Guiyang area.


Contributions of atm-NH+4 and atm-DON in mosses on bare rock

With no N supply from substrates, %N and δ15N in mosses on bare rock are good indicators of anthropogenic N deposition (Pearson et al., 2000). Low moss δ15N observed in the Guiyang area reflected the dominance of RDN (low δ15N; Table S1) in N deposition. Lower moss δ15N in the urban than in the rural area (Fig. 1) was attributed mainly to high math formula from urban sewage/waste NH3 emission, which created an urban–rural gradient of anthropogenic N deposition and which was more 15N-depleted than math formula from soil/fertilizer NH3 in rural areas (Liu et al., 2012c).

Calculations based on the two-source mixing model revealed higher contributions from atm-math formula (56%) than from atm-DON (44%) to N in mosses on bare rock, particularly in urban areas (fatm-NH4+ = 68%) (Table 2; Fig. 2a). However, in suburban and rural areas where anthropogenic math formula deposition was low, the contribution from atm-DON became higher than in the urban area (Fig. 2a). Attention therefore must be devoted to the contribution of atmospheric DON to the moss N economy. Otherwise, the ecophysiological impacts of anthropogenic atm-math formula deposition on mosses can be overestimated substantially in our study area. Moreover, mosses can adjust the N-assimilating regime in response to anthropogenic N deposition (Wiedermann et al., 2009). The evaluation of math formula preference using βatm-NH4+ revealed that urban mosses on bare rock preferred atm-math formula over atm-DON (0 < βatm-NH4+ < 1; Fig. 3a). Because both math formula and DON can be absorbed into moss cells through cation exchange and the proton (H+) pump, the uptake might not result in a substantial preference. The observed atm-math formula preference was generated mainly from assimilation associated with inherent N demand, and potentially with the purpose of reducing impacts of excessive math formula accumulation on moss growth (Limpens & Berendse, 2003), as evidenced by higher efficiencies of intra-plant math formula assimilation than assimilation of other N forms (Pearson & Stewart, 1993; Kronzucker et al., 2001). This mechanism was supported in particular by higher βatm-NH4+ for mosses under high math formula pollution in the urban area (Fig. 3a). In rural areas, the low anthropogenic math formula (Liu et al., 2012c) and the supply of atm-math formula might not satisfy moss N demand. For these reasons, epilithic mosses showed greater reliance on atm-DON and showed little or no preference for atm-math formula (Fig. 3a). Accordingly, the N assimilation in epilithic mosses was dominated by atm-math formula because the preference for math formula was associated with a higher rate of math formula deposition. The atm-DON contributed a substantial fraction of moss N assimilation, although lower fatm-DON in the rural area than fatm-math formula in the urban area showed a compromise of atm-DON assimilation to atm-math formula assimilation (Fig. 3a). Similarly a down regulation of elevated anthropogenic DIN deposition on moss N2 fixation has been observed in feather mosses of boreal regions (DeLuca et al., 2008; Ackermann et al., 2012).

Partitioning of atm-N and soil-N contributions in mosses growing on soil

The utilization of soil N sources, which were generally more 15N-enriched than those from atmospheric deposition, caused higher δ15N in mosses on N-available substrates than in those on bare rock (Fig. 1). The low net N rates of acidic soil in pine forests resulted in low soil N availability to forest mosses. Consequently, bulk N and tissue math formula (Liu et al., 2012c) were lower in forest mosses than in other mosses (Tables 1, S4). This raised the question of whether atmospheric N is still important for mosses that have opportunities to use soil N sources, especially under high soil N availability. Moreover, lower C : N, higher soil math formula and nitrification rates, and lower δ15N of soil math formula were observed in urban than in rural areas (Fig. 1, Table 1). These results demonstrated that N availability and cycling processes in underlying soils showed a response to elevated N deposition, although direct atmospheric N inputs were largely retained by moss layers. It is therefore necessary to ascertain whether the utilization of soil N in mosses will increase when N supply from both soil and deposition is elevated by anthropogenic N pollution.

Showing consistency with mosses on bare rock, the math formula assimilation in mosses on soil was inhibited because RDN was much higher than math formula in both deposition and soil (Tables 1, S1; detailed in Liu et al., 2012c). Provided that mosses had the capability to assimilate math formula, the fNO3– (mean > 38%) would show higher percentages than the fNH4+ (data not shown; calculated using IsoSource). Such a greater reliance on math formula than on math formula and preference for math formula over math formula is unlikely in our study area. First, the inhibitory effect of math formula on math formula utilization is well known to prevail among microbes (Rice & Tiedje, 1989), phytoplankton (reviewed by Dortch, 1990), and vascular plants (Kronzucker et al., 1999; Aslam et al., 2001; Wang & Macko, 2011). Evidence from phytoplankton showed that the inhibitory effect occurs at math formula concentrations higher than c. 1 μM (a concentration much lower than math formula concentrations in precipitation and soil solutions) (Dortch, 1990). Second, many 15N-tracer experiments have reported extremely low math formula utilization in mosses, even at the same concentrations as other N sources (Soares & Pearson, 1997; Forsum et al., 2006; Wanek & Pörtl, 2008; Wiedermann et al., 2009). In our study area, we inferred that mosses trapped math formula and that the storage of math formula depended on external availability, but math formula assimilation did not occur to a substantial degree under higher deposition of RDN (Liu et al., 2012b,c). Therefore, the partitioning of soil and deposited N sources was conducted on math formula and DON for mosses on soil (Figs 2, S2).

There are two main implications of our results. First, terricolous mosses did rely more on N from deposition (63%) than from soil (Table 2), but the mean soil N contribution of 37% (Table 2) was much higher than the current estimation based on short-term (7 d) and solely inorganic 15N additions (math formula : math formula = 1 : 1) in mat-forming mosses (2–9%; Ayres et al., 2006). Evaluation of moss N sequestration and N deposition effects should therefore consider the assimilation of soil-derived N. Neglecting this source might result in considerable overstatement of the importance and effects of N deposition. Moreover, the contribution of soil N did not change with soil N availability (Fig. S4). Forest mosses, which showed little difference in N% and δ15N from tericolous mosses in open fields (suggesting little canopy effect at least on deposition δ15N in our study; Table S2), had substantially lower soil N availability, but the corresponding contributions of soil N in forest mosses were comparable to those of mosses in open fields (Fig. S3). By contrast, the proportional contribution of soil N was even lower in mosses of urban areas (Fig. 2d), where both soil N availability and N deposition were elevated by anthropogenic math formula deposition (Liu et al., 2008, 2012c). The pattern is also apparent in calculations based on δ15N of RDN (Fig. S2 and Notes S3), confirming that atmospheric N is still more important for mosses than soil N sources, even when soil N availability is elevated. The contribution of soil N to moss N was not responsive to soil N availability (Fig. S4), but decreased with anthropogenic math formula deposition (Figs 2d, S3).

Second, math formula was the dominant N form among N species assimilated by mosses (Table 2). This dominance is attributable to the dominance of math formula among available N sources (Tables 1 and S1) and/or to the preference for math formula over DON. The greatest contribution was from atm-math formula (46%; Table 2, Fig. 2). This result added quantitative and field-based evidence that math formula deposition plays a major role in altering moss N metabolism and species composition (Baxter et al., 1992; Aerts & Bobbink, 1999). However, an average 38% of moss N was contributed from DON (ftotal-DON; Table 2), suggesting that, in natural habitats, DON was a notable N contributor to mosses, although the availability of coexisting DIN was high. Before this study, the utilization of DON (mainly as amino acid N) was estimated mainly in vascular plants and was based on isotopic labeling methods. Using 14C-labeling, the uptake of free amino acids by nonmycorrhizal Eriophorum vaginatum L. was estimated as accounting for > 60% of the N absorbed by this species in arctic tundra (Chapin et al., 1993). By injecting 13C-labeled and 15N-labeled glycine into the organic layer in an old boreal forest, it was estimated that at least 91%, 64%, and 42% of N was taken up, respectively, in intact glycine in dwarf shrubs (Vaccinium myrtillus L.), grass (Deschampsia flexuosa (L.) Trin.), and coniferous trees (Pinus sylvestris L. and Picea abies (L.) Karst.) (Näsholm et al., 1998). Additional evidence obtained from arctic moss (Sphagnum rubellum Wils.; Kielland, 1997), subalpine/alpine and temperate species (Raab et al., 1999), and also agricultural plants (Näsholm et al., 2000) revealed a widespread capability to utilize DON among plants, irrespective of the type of mycorrhizal association and the availability of DIN (Näsholm et al., 2009). Our results, based on the natural δ15N partitioning method, emphasize the importance of DON for moss N nutrition and add to the knowledge of mosses' utilization of organic N sources (Forsum et al., 2006).

However, the strategies of moss DON utilization differed between atm-DON and soil-DON. For atmospheric N sources, DON was deposited onto moss layers with atm-math formula. Thereby, the assimilation occurred when atm-math formula could not meet moss N demand. However, the absorption of soil-DON may be more complex because the bioavailability of N deposition and the competition for soil N with soil microbes must be considered. Using the 15N-labeling method, Harrison et al. (2007) found that coexisting plants of a temperate grassland showed a consistent preference for soil DIN over soil DON and for simple (glycine and serine) over more complex amino acids (phenylalanine), but competition between plants and soil microbes can complicate the picture and cause differences over time. For mosses, the motivation of using soil-DON might be complex: the supply of deposited N sources should at least be considered. In addition, more studies are needed to assess the explicit δ15N compositions of amino acids, and bioavailable and residual fractions of DON to examine the relative availability of N deposition and soil N sources to mosses. By estimating the preference for math formula over DON, a difference was found between atm-math formula and soil-math formula. Fig. 3(a) shows that mosses preferred math formula to DON when assimilating atmospheric N sources (0 < βatm-NH4+), no matter how high or low the availability of soil N sources was. As explained for mosses on bare rock, this fact reflected the dominance of math formula in N deposition and a strategy to avoid tissue math formula accumulation from continuous atm-math formula inputs. Consequently, a greater preference for math formula occurred for mosses under higher math formula deposition in urban areas than in rural areas (Fig. 3a). However, mosses showed no consistent preference for math formula in using soil N sources (Fig. 3b), but exhibited an opposite pattern of βsoil-NH4+ to the βatm-NH4+ (Fig. 4) associated primarily with moss N demand and overall N availability. In urban areas, higher math formula supply from deposition and higher soil-DON availability caused greater preferences for atm-math formula and soil-DON (Fig. 3). In rural areas, lower atm-math formula supply and lower soil-DON availability caused a lower preference for atm-math formula but a greater preference for soil-math formula.

Uncertainties and future work

Isotopic mass balance calculations in this work were based on negligible isotopic fractionation and preference during moss N uptake, and negligible math formula assimilation in mosses under high supply of RDN. However, the isotopic variations of plant N pool and N sources are factors that might influence isotopic partitioning of plant N sources. For mosses, it is difficult to define the time-scale (or temporal stability) of recorded moss δ15N and the influence of short-term N assimilation on moss δ15N. For N sources, although particulate N in deposition might not enter moss tissues without the help of rainwater, differences might exist between the δ15N of wet N deposition and the δ15N of usable N in total deposition (dry and wet deposition). Moreover, the measured DON was actually a salt-extractable fraction passed through a GF/F filter, although it might represent an important fraction of the whole DON pool. The DON includes complex compounds with different molecular weights, and presumably different bioavailabilities (Knapp et al., 2005). Yet the explicit fraction and exact δ15N values of moss-assimilated DON in bulk DON have not been characterized to date. Furthermore, explicit δ15N signatures in N deposition have not been measured across all moss habitats in our study. The deposition data collected in the urban area suggest that the δ15N values of each N source are temporarily variable to a great degree. Therefore, their mean values were taken as end members of source N in the whole study area. All above-mentioned assumptions and possible d15N variations in mosses and N sources need further verification in future studies using stable isotopes to constrain moss N utilization.


This study has advanced the application of the nondisturbing 15N natural abundance method for interpreting plant N utilization in natural environments. Ammonium was confirmed as the main form of N species assimilated by mosses, showing higher dependence on and greater preference for atm-math formula. However, DON was revealed as an important N source, with a mean contribution of 38%, especially in mosses with no N available in substrates and in mosses receiving lower math formula deposition. On average, 37% of N in mosses was derived from soil, which is greater than estimations based on 15N-labeling of inorganic N. These results do not support the idea of exclusive reliance of mosses on atmospheric DIN deposition, and underscore the importance of considering DON and soil N utilization in estimating N sequestration and N deposition impacts in moss ecosystems. Moreover, the contributions of DON and soil N sources decreased with that of atm-math formula, which was related to math formula preference associated with high anthropogenic math formula deposition. Furthermore, terricolous mosses were found to prefer atm-math formula over atm-DON, and to prefer soil-DON over soil-math formula when math formula availability was elevated in deposits and soil. The elucidation of these mechanisms provides new insights into moss N strategies pursued in response to anthropogenic N pollution, especially in regions with math formula-dominated N deposition.


We thank Prof. Janice Glime for helpful comments on the initial draft of this manuscript. We also thank Dr Hongwei Xiao for collecting rainwater, and Drs Haifeng Fan, Takebayashi Yu, Fujun Yue and Liran Bao for help with experiments. This work was supported by a Grant for Projects for the Protection, Preservation and Restoration of Cultural Properties in Japan by the Sumitomo Foundation, Grants-in-Aid for Creative Scientific Research (No. 21310008), the Program to Create an Independent Research Environment for Young Researchers from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and the NEXT Program (GS008) from the Japan Society for the Promotion of Science (JSPS). X-Y.L. was also supported by the National Natural Science Foundation of China (nos. 40903012, 41021062 and 41273026) and the JSPS postdoctoral program for foreign researchers (no. 09F09316).