The fate of 15 N-nitrate in mesocosms from five European peatlands differing in long-term nitrogen deposition rate

Elevated nitrogen (N) deposition changes the retention, transformation, and fluxes of N in ombrotrophic peatlands. To evaluate such effects we applied a N tracer (NH4 NO3) at a rate of 2.3 g N m −2 yr to mesocosms of five European peatlands with differing long-term N deposition rates for a period of 76 days of dry and 90 days of wet conditions. We determined background N content and moss length growth, and recovered the N tracer from the mosses, graminoids, shrubs, the peat, and dissolved N. Background N contents in Sphagnum mosses increased from 5.5 (Degerö Stormyr, deposition <0.2 g N m yr) up to 12.2 mg g (Frölichshaier Sattelmoor, 4.7–6.0 g N m yr). In peat from Degerö, nitrate and ammonium concentrations were below 3 mg L, whereas up to 30 (nitrate) and 11 mg L (ammonium) was found in peat from Frölichshaier Sattelmoor. Sphagnum mosses (down to 5 cm below surface) generally intercepted large amounts of N (0.2–0.35 mg g) and retained the tracer most effectively relative to their biomass. Similar quantities of the N were recovered from the peat, followed by shrubs, graminoids, and the dissolved pool. At the most polluted sites we recovered more N from shrubs (up to 12.4 %) and from nitrate and ammonium (up to 0.7 %). However, no impact of N deposition on N retention by Sphagnum could be identified and their length growth was highest under high N background deposition. Our experiment suggests that the decline in N retention at levels above ca. 1.5 g m yr, as expressed by elevated near-surface peat N content and increased dissolved N concentrations, is likely more modest than previously thought. This conclusion is related to the finding that Sphagnum species can apparently thrive at elevated long-term N deposition rates in European peatlands.

Ombrotrophic peatlands are exclusively fed by rainwater and especially sensitive to atmospheric N inputs.Over the last century many of these ecosystems have received inputs of atmospheric N deposition even 10-fold larger than the pre-industrial levels (van Aardenne et al., 2001).In Europe, where N deposition strongly varies spatially, the mean change in N deposition between 1950-2000 and 1900-1950 was between 0.4 and 1.6 m −2 yr −1 (Granath et al., 2014).Although the future estimates of N emissions (NO x and NH 3 ) vary, with some indicating a substantial further increase (Galloway et al., 2004) and others suggesting stabilization or even slight decrease in emissions by 2030 given an appropriate Published by Copernicus Publications on behalf of the European Geosciences Union.
emission control (Cofala et al., 2007), the importance of elevated N inputs to peatlands remains undisputable.
Elevated N supply has been documented to cause various shifts in peatland N cycling such as a decrease in C : N ratio in near surface peat (Bragazza et al., 2012), stimulation of denitrification (Francez et al., 2011), and an increase in N concentration in interstitial waters (Limpens et al., 2003).It has also been reported that plant community change from Sphagnum-dominated to vascular-dominated and the enhanced N inputs may turn peatlands into C sources (Bubier et al., 2007;Wiedermann et al., 2007).The effect of N deposition on the retention ability of Sphagnum mosses is also dependent on N deposition rate.At low N deposition levels the growth of Sphagnum can be promoted (Aerts et al., 1992), which suggests that also N retention is raised.At high N deposition, however, Sphagnum was found to lose its filtering capacity (Lamers et al., 2000;Limpens and Berendse, 2003).As a result elevated N input to ombrotrophic peatlands and declining uptake by Sphagnum could potentially enhance denitrification due to increased N availability in the pore waters of affected systems (Hayden and Ross, 2005).
A quantification of N transformation rates and fluxes under conditions of varying long-term N deposition is crucial for understanding the mechanisms and robustness of N retention.Several experimental studies have addressed the impact of increased N deposition on the functioning of peatland vegetation (e.g., Aldous, 2002a;Curtis et al., 2005;Williams et al., 1999b).Lower N retention in the upper stem moss tissue (Aldous, 2002b) and higher N 2 O emissions (Regina et al., 1996) have been documented with decreasing water table.However, to our knowledge, the fate of N in peatland ecosystems under varying N inputs and hydrological conditions has not been examined extensively yet (Blodau et al., 2006;Xing et al., 2010).Studies on the fate of N within moss, vascular plants, and upper peat layer, including the pore water, are needed in light of the continued, high levels of N emissions.In this contribution, we investigate how the allocation of N in plants and peat is altered under different N pollution levels using 15 N-labeled nitrate.We conducted a controlled mesocosm experiment with intact peat cores, including the vegetation, from five European peatlands, which have a history of differing long-term N deposition.This approach allowed us to compare the fate of deposited 15 N at sites that have undergone decades of differing N deposition and in this way avoid the shortcomings of short-term fertilization studies, in which the timescale of manipulation is generally too short for an adjustment of ecosystem structure and functioning.Labeled N was used to investigate how the allocation of N in plants and peat was altered under the different long-term N pollution levels the sites were exposed to.We hypothesized that the sites with a legacy of low N deposition would experience low mobility and high retention of N in vegetation and near-surface peat, whereas the sites with a legacy of higher N deposition would experience high mobility and low retention of the element.We further expected nearly complete absorp-tion of the applied 15 N by Sphagnum at low background N deposition, and diminished absorption by mosses and raised uptake by vascular plants at background N deposition exceeding levels of about 1.0-1.5 g N m −2 yr −1 (Bragazza et al., 2006;Lamers et al., 2000).
To address the hypothesis, we specifically examined 1. differences in elemental N content in the peat and vegetation before the experiment and dissolved nitrate, ammonium, dissolved organic, and total nitrogen concentrations during the experiment; 2. differences in the distribution of 15 N stemming from experimentally deposited 15 N-nitrate between plant functional types, peat, and dissolved nitrate and ammonium; 3. differences in the "efficiency" of the vegetation and peat to retain the deposited 15 N during the experiment.

Study sites
We selected five Sphagnum-dominated peatlands in northern and western Europe: Degerö Stormyr (DS), Sweden; Lille Vildmose (LV), Denmark; Fenn's, Whixall, and Bettisfield Mosses NNR (WM), United Kingdom; Cors Fochno (CF), Wales, United Kingdom; and Frölichshaier Sattelmoor (FS), Germany.The differences in the soil acidity between the sites were small -pH ranged between 3.6 and 4.0 and was typical for ombrotrophic peatlands (Rydin and Jeglum, 2006).Site specific characteristics as well as the typical vegetation are presented in Table 1.

Experimental design and sampling
In the summer of 2009 we collected three intact peat monoliths ("cores", "mesocosms", 30 cm in diameter, 40 cm long) from lawns of the five sites.This resulted in 15 mesocosms, which were subsequently incubated in a greenhouse chamber at 20 • C, ca.60 % RH, and were exposed to 12 h light-dark cycles.Directly after collection, all mesocosms were drained and then rewetted to a water table of 28 cm ± 3 cm below the moss layer.The experiment commenced in November 2009, after a 2-to 3-month (depending on the collection date) acclimation period.During the first 76 days all mesocosm were subjected to low water table level (−28 cm ± 3 cm), here also called "dry period".It is important to note that the moss layer and peat remained moist even during this period due to frequent irrigation.Between day 77 and 78 we raised the water level up to ca. −8 cm using deionized water and maintained it there (±3 cm) for the next 90 days, until the end of the experiment in the middle of April ("wet period").Mesocosms were watered twice a week using a selfmade sprinkler.During the acclimation period we added Site-specific data -DS (Granberg et al., 2001;Sagerfors, 2007); CF, WM http://www.metoffice.gov.uk/climate/uk/averages/19712000/;LV (Charman et al., 2007); FS (Baumann, 2009;Jensen, 1990).N and S deposition data (wet and dry deposition) -national models: DS (Karlsson et al., 2012), CF and WM for year 2009, Department for Environment Food and Rural Affairs (DEFRA) accessed August 2012 at http://pollutantdeposition.defra.gov.uk/data,LV for year 2004 (Ellermann et al., 2004), FS for 1993 (Gauger, 2009); measured values: DS for 1990-1992 (Granberg, et al., 2001), CF and WM data not available, LV (Ellermann et al., 2004), FS (Böhlmann et al., 2005); IDEM model (Pieterse et al., 2007).Abbreviations: Lat -latitude; Long -longitude; MAP -mean annual precipitation; MAT -mean annual temperature; MASL -meters above sea level; bsl -below sea level; IDEM -Integrated Deposition Model (Bleeker et al., 2004).* The area of Fenn's, Whixall, and Bettisfield Mosses.
a synthetic rainwater solution, whose chemical composition was based on the average European rainwater composition (Appelo and Postma, 2005), with N concentrations adjusted for each site.Specifically, the concentrations were 16.65 mg L −1 NH 15 4 NO 3 , and 2.94 (CaCl 2 × 2H 2 O), 1.63 (MgCl), 3.70 (Na 2 SO 4 ), 0.52 (K 2 SO 4 ), 0.029 (H 2 SO 4 ) and 0.002 (HCl) mg L −1 during the experiment.In the acclimation period N deposition differed between sites and was adjusted to 1.57 (DS), 6.26 (CF), 9.39 (WM), 12.53 (LV) and 16.44 (FS) mg L −1 NH 4 NO 3 .Throughout the experiment (November-April), we substituted NH 4 NO 3 in the irrigate with the NH 15 4 NO 3 tracer at 2.3 g of N m −2 yr −1 (1.1 g of N m −2 added over the experimental period) to each mesocosm.This deposition level was found by Blodau et al. (2006) to be almost fully absorbed by mosses in meso-cosms from unpolluted sites, whereas breakthrough of the tracer was expected in mesocosms from N-polluted sites.Each tracer application (48 in total) was followed by an addition of distilled water to the cores using the sprinkler.This procedure was adopted to rinse off the tracer from the plant surfaces and to adjust the water table.While this procedure cannot guarantee that no tracer remained on plant surfaces, the large number of irrigation and rinsing events should ensure that only little of the applied 15 N tracer remained physically attached to the plant surfaces by the termination of the experiment and sampling of biomass.The 15 N application thus allowed us to trace the fate of N in peat cores exposed to different background N-deposition levels.The remaining rainwater solution stayed unchanged.When referring to the "start" and the "end" of the experiment we refer to the first 15 N tracer application and the sampling following the last 15 N application, respectively.
Throughout the experiment, we recorded the water level in all mesocosms twice a week, and adjusted it in the outlined way.Peat soil moisture (SM) content was logged hourly (ECH 2 O EC-5 FD sensors, Em5b loggers, Decagon Devices Inc.) at four depths in one core from each site.We calibrated the sensors for each site and for the top and the bottom of the peat core (see Supplement).Polynomial equations (R 2 > 0.82) were used to calculate the volumetric water content at 1 h time steps.
Sampling of peat gas and water was performed fortnightly.The extracted gas volume was replaced with nitrogen gas (N 2 ) at each sampling depth.Gas flux from the mesocosm surface was measured during 20 min long sampling period, including five sampling points in 5 min intervals.We used plexiglas chambers (27 cm in height and 29 cm in diameter) with an inlet and outlet tube at the top.These data will be reported in a separate contribution (in preparation).The peat water was sampled using Rhizon ® samplers (polymer, 1 µm pore size, Eijkelkamp Agrisearch Equipment) at four depths: 9, 16, 23, and 30 cm below the Sphagnum surface and through the PE tube from the bottom of each core.The pH of peat water was measured immediately using a glass electrode.The water samples were stored cool or were frozen for later analyses.
The length growth of Sphagnum moss community was measured in all mesocosms using three cranked wires (Clymo, 1970) per mesocosm.Wire lengths were measured shortly after installing, in the middle and at the end of the experiment.Single species of Sphagnum, shrubs, and graminoids designated for isotopic analyses were sampled from each mesocosm at the end of dry and at the end of wet period.Plant samples used for estimation of background 15 N contents were collected directly from the sampling sites.
After termination of the experiment in April 2010 total aboveground biomass of each core was quantified.We clipped the whole plant material from the mesocosms' surface, distinguishing between main plant functional types (shrubs, graminoids, and mosses).Mosses were collected down to 5 cm below the moss canopy.After the biomass collection, we sequentially extracted the pore water from individual depth layers through the gas sampler openings.The extraction continued from the top of the core downwards until no water from the single layer could be obtained.Due to a large water volume required for the 15 N analysis and differences in a field capacity between the cores and specific peat layers, it was not possible to obtain exactly the same layer depths for all mesocosms.Finally, the cores were extracted and one part of each core cut into 2-8 cm (higher resolution at the top and lower at the bottom) segments for 15 N analyses.The remaining parts were used for bulk density (BD) estimations.The peat for background 15 N measurements was collected before the first tracer application using a stainless steel microcorer (3 cm in diameter).
All organic samples were dried at 70 • C for at least 24 h or until no more mass loss could be recorded.The samples used for aboveground biomass and bulk density estimation were then weighed and the samples destined for isotopic analysis were ground with a ball mill to fine powder and stored in an evacuated desiccator prior to analysis.
When reporting sampling depths, we report the depth relevant not to the rim of the mesocosm tube but to the Sphagnum surface, thus obtaining discrepancies of up to 1 cm between the cores.

Chemical analyses
In the peat pore water obtained by rhizon samplers we measured nitrate (NO − 3 ) concentrations by ion chromatography (Metrohm IC 733 with suppressor module, Metrosep Dual 3 column).Ammonium (NH + 4 ) concentrations were quantified photometrically at 690 nm (ISO 7150/1) on a DR 3800 photometer using the NH + 4 quick test method (ammonium test, range: 0.01-3.00NH 4 -N mg L −1 , Merck).Total dissolved N (TDN) and particulate N (PN) were measured on a TOC/TN analyzer (multi N / C 2100, Analytik Jena, NPOC analysis).Dissolved organic N (DON) concentrations were calculated as the difference between TDN and dissolved inorganic N (DIN) (NO − 3 plus NH + 4 ).In peat pore gases obtained by silicon gas samplers and in chamber measurements, nitrous oxide (N 2 O) was measured on a Varian gas chromatograph with flame ionization (FID) and electron capture (ECD) detectors.We did not detect any N 2 O above background in the pore water and flux from the mesocosms throughout the experiment; data on N 2 O are thus not reported in this contribution.

Isotopic analyses
The ratios of natural abundance and tracer 15 N/ 14 N as well as total C and N were quantified in the dried plant material and the peat using an elemental analyzer (NA 1108 -natural samples, NC 2500-labeled samples, CE instruments, Milan, Italy), connected via ConFlo III interface to a δS IR-MS (natural/labeled, Finnigan/Thermo Fisher Scientific, Bremen, Germany).Dissolved inorganic 15 N content in pore water samples was assessed by a sequential trapping of volatilized NH + 4 and NO − 3 (Brooks et al., 1989) -procedure modified and described in detail in Blodau et al. (2006).
High purity of the added tracer (98 at.% 15 N) and high frequency of watering events resulted in extremely high isotopic signatures of most peat and plant samples, exceeding the calibration range of the instrument.We thus diluted all samples with a natural plant material of known 15 N content and remeasured the samples.Comparison of the two data sets revealed a reasonably close correlation (y = 0.9623x +0.0036, R 2 = 0.71; see Supplement Fig. S1).For interpretation we thus used the original 15 N data set.
The 15 N abundances (at.% 15 N excess) were obtained by subtracting the background 15 N of plants, peat and traps (diffusion procedure) from the labeled samples.

Data handling and statistical analyses
We calculated area-based plant biomass, peat density, and total nitrogen (TN) content of the ecosystem pools as well as volume-based 15 N content in peat and plant pools.Additionally, 15 N retention efficiency and 15 N recovery from the pools was estimated.The relative share of each of the ecosystem pools to the total 15 N recovered (%) was obtained by dividing their tracer mass by the total mass of the tracer added to the system.The sum of the relative shares of each pool gave the total 15 N recovery.
Statistical analyses were performed with the SPSS 20.0 package and SigmaPlot for Windows, version 11.0.The effect induced by differing long-term N deposition (sites) on N pools and 15 N retention and of wet and dry conditions on 15 N uptake by plants was analyzed by one-way analysis of variance (ANOVA) followed by a post hoc test (Tukey or Bonferroni).The effect of differing long-term N deposition (sites) on 15 N abundance in vegetation pools after the dry and wet phase was analyzed with one-way analysis of covariance (ANCOVA) using wet and dry conditions as covariate and log-transforming data when needed.Significance was set at p ≤ 0.05, unless stated otherwise.
Before reporting the results of N concentration measurements, a few words on the uncertainty in data are needed.We encountered an issue when evaluating the pore water concentration data from FS cores -NO − 3 concentrations in the first part of the experiment by far exceeded the measured TDN concentrations.A repeated measurement of the problematic samples did not give a solution to that problem.Thus, the DON data for the first 98 days needed to be omitted, otherwise showing negative values.

Plant biomass and growth
Total plant biomass at the end of the experiment was largest in the Welsh cores, with 1530 ± 313 (CF) and 1397 ± 693 g m −2 (WM), and smallest in German Frölichshaier Sattelmoor (FS) mesocosms, with 1178 ± 67 g m −2 (Fig. 1a).Bryophytes accounted for 62 to 87 % of the total (Fig. 1b), when defining living biomass as reaching down to a depth of 5 cm below the Sphagnum capitulum.Shrub biomass was significantly higher than graminoid biomass with the exception of Degerö Stormyr (DS) (Fig. 1b).The highest percentage of shrubs (30 %) was found in FS cores.Vascular plant biomass increased at all sites during the experiment, particularly after raising the water table.The highest increase -from 140 to 440 g DW m −2 -was recorded in FS cores (Fig. S2); overall biomass gain decreased in the order FS > WM > CF > LV ∼ = DS.Sphagnum growth differed substantially among the sites.In Frölichshaier Sattelmoor (FS) mean growth rate was 2.36 mm month −1 and at Whixall Moss (WM) only 0.27 mm month −1 ; growth declined in the order FS > DS > LV > CF > WM (Fig. S3).During the dry period Sphagnum mosses did not desiccate and actively grew, likely due to the frequent irrigation.We measured an increase in growth at CF, WM, and Lille Vildmose (LV) cores during wet period (Fig. S3), but only for CF was this increase significant (P <0.02).

Nitrogen contents
Weight-based (Fig. 2) and area-based (Fig. 3) N contents differed between the sites at the end of the experiment, albeit often not significantly.The average N content of peat, plant, and dissolved pools, expressed as the ratio of the mass of N present in the pool to the total mass of the pool, generally increased with long-term N deposition (Fig. 2).Surface peat layer (5-15 cm depth) contained between 5.3 ± 0.3 (DS) and 13.4 ± 2.3 (CF) mg N g −1 and this difference was significant.Sphagnum (capitulum and stem) and shrubs had similar N contents varying between 4.9 ± 2.5 and 9.8 ± 3.9 mg g −1 , and 5.7 ± 0.8 and 10.7 ± 1.7 mg g −1 , respectively.The N content in Sphagnum was generally higher in the apical part than in the stem part and ranged between 5.6 ± 0.5 (DS) and 12.2 ± 6.4 (FS) mg N g −1 (Fig. S5).We recorded a very high shrub N content at Degerö Stormyr resulting from high N content of Andromeda polifolia leaves at this site (Fig. 2).Graminoids were the most N-rich plant functional type (8.4 ± 1.5-18.9± 4.0 mg g −1 ) and their N content differed significantly (F (4, 10) = 8.099, p = 0.004) between Frölichshaier Sattelmoor and the rest of the sites.The least N was found in the dissolved pool (Fig. 2).Total dissolved N content in the surface layer of peat (0-15 cm) ranged from 1.95 ± 0.12 to 6.17 ± 1.27 µg g −1 and differed significantly between LV cores and DS and WM cores (F (4, 10) = 5.82, p = 0.01).ANOVA further showed that N content in the ecosystem pools significantly differed from each other; the post hoc test indicated that at the level of sites the differences between graminoids, Sphagnum and shrubs were significant at the DS site (p = 0.002 and p = 0.018) and FS site (p = 0.01 and p = 0.001).Significant differences also occurred between graminoids and peat (DS, FS, CF), as well as shrubs and Sphagnum and peat (CF).
Area-based N content of ecosystem pools, calculated by multiplying biomass (plants), bulk density (peat), or volumetric water content (Supplement Table S1) by N content after termination of the experiment, varied between the sites but the differences were mostly not significant (Fig. 3).Due to its large mass, peat contained by far the largest N pool and held between 22.8 ± 4.6 (DS) and 63.5 ± 11.5 (CF) g N m −2 in the 10 cm subsurface layer.4.0 ± 0.8 (WM) and 9.0 ± 0.3 (LV) g N m −2 .Total N content was similar for shrubs and graminoids varying between 0.6 ± 0.16 and 2.8 ± 0.61 g N m −2 and between 0.4 ± 0.25 and 1.7 ± 2.07 g N m −2 , respectively.The N content in these two plant functional types was highest at Frölichshaier Sattelmoor (Fig. 3), and statistically differed from that of the DS site.Lowest TDN values were detected at low long-term N deposition for shrubs and at intermediate long-term N deposition levels for graminoids (Fig. 3).The dissolved pool contained the smallest quantities of N. Total dissolved N content increased with long-term N deposition and ranged from 0.03 to 0.63 g N m −2 (Fig. 3).The largest amounts of TDN were found in the LV site and differed significantly from both DS and WM (Fig. 3).
In Sphagnum mosses, molar ratios of C / N were significantly related to long-term N deposition at the sites and responded to the experimental treatment with the deposition of the NH 15 4 NO 3 tracer.Ratios of C / N in Sphagnum capitula sampled at the sites before the experiment declined from above 100 (Degerö Stormyr, DS) to 36 in German Frölichshaier Sattelmoor (FS) following a log function (Fig. S4, R 2 = 0.96, p = 0.00022).Experimental deposition of 2.3 g N m −2 yr −1 , i.e., 1.1 g N m −2 during the experiment, led to a decline of C / N in DS and an increase in FS, and little change otherwise; wet conditions partly reversed this effect (Fig. S4).

Nitrogen uptake by plants
Uptake of 15 N during the experiment varied between plant functional types and was influenced by water table position.The highest 15 N enrichment was observed in the Sphagnum's apical part, followed by Sphagnum stem, shrubs, and graminoids (Fig. 4).N deposition had a significant effect on 15 N content in Sphagnum capitula (ANCOVA, p = 0.029) and graminoids (p = 0.042) but not on Sphagnum stems and shrubs.Enrichment with 15 N in Sphagnum (0-5 cm) also visually decreased with long-term N deposition, with Degerö Stormyr having highest and Frölichshaier Sattelmoor smallest uptake during the dry period at 5.9 and 2.9 % in the capitulum, respectively (Fig. 4a).Tracer uptake by the capitulum of Sphagnum was significantly elevated during the dry period (Fig. 4a), as confirmed by one-way ANCOVA (p<0.001).Unlike in Sphagnum, the largest 15 N enrichment in shrubs occurred during the wet period (Fig. 4c and d).Shrubs assimilated on average twice as much of the tracer during this second part of the experiment (Fig. 4c; one-way ANCOVA, p<0.001).Effects were not significant in regard to graminoids and Sphagnum stems.

Retention and recovery of added 15 N
To examine the retention capabilities of each of the pools during the experiment, retention efficiency per gram of dry weight was calculated from the quotient of 15 N retained at the end of the experiment divided by dry biomass or dry peat mass, respectively.The sites substantially differed in 15 N retention efficiency, when all pools were accounted for, yet differences between sites were statistically significant only with regard to the shrub pool and sites DS and CF as well as DS and WM (Fig. 5).Vegetation was up to four times more effective in retaining N than peat and Sphagnum moss was significantly more efficient than graminoids and peat (p = 0.007 and p = 0.001), as were the shrubs (p = 0.013).Cors Fochno and Whixall Moss vegetation had the smallest retention efficiency varying between 0.05 ± 0.02 in graminoids (CF) and 0.2 ± 0.06 in Sphagnum (0-5 cm) (WM) (Fig. 5).Interestingly, Sphagnum from both the sites with low (DS) and the sites with high (LV, FS) level of background N input was equally effective in absorbing N, but peat in LV and FS cores retained substantially less N than the in WM cores (Fig. 5).
During the course of the experiment we applied 38 mg (572 mg m −2 ) of 15 N to each mesocosm.After 24 weeks, the recovery, i.e., the mass of 15 N divided by amount of 15 N added, of the tracer from all N pools ranged from 87.5 to 131 % (Table 2).Recoveries above 100 % were most likely caused by the uncertainty in the peat bulk density.The depth resolution used in 15 N sampling (6-8 sections per core) was much higher than the resolution used in the bulk density measurements, which encompassed only 2-3 depth intervals per core.Especially the 15 N content in the most upper peat layers, where BD was possibly lower than the average BD of a thicker peat layer, may have been overestimated.Sphagnum and peat contained the largest fraction of the tracer and accounted for 21.9 to 68.9 and 23.6 to 65.6 % of the total, respectively.Vascular plants contained only between 5.6 and 13.6 % of the recovered 15 N.The shrubs had an up to 13 times higher share than the graminoids (Table 2).The shrubs in cores from Frölichshaier Sattelmoor were particularly efficient in retaining the tracer and accounted for 91 % of the retention in vascular plants (Table 2).Dissolved inorganic 15 N contributed little to the retention of 15 N at levels below 1 %, but its share increased with background N deposition rate (Table 2).This pool accounted for 0.01 % (DS), followed by 0.2 % (CF and WM), 0.4 (FS), and 0.7 % (LV).

Dissolved nitrogen
The dissolved N concentration increased with long-term N deposition at the sites during the experiment (Fig. 6, Table 1).Nitrate was detectable predominantly above the water level but remained low (< 3 mg L −1 ) in the majority of the cores (Fig. 6).Only in the cores from Frölichshaier Sattelmoor, NO − 3 concentrations reached 30 mg L −1 and were especially high during the first 35 days of the dry period.Ammonium concentrations reached 11 mg L −1 at the site characterized by high levels of N input (FS) and 2 mg L −1 at the site with low level of N input (DS) (Fig. 6).The smallest detectable NH + 4 concentrations were found at DS and were as low as 0.3 mg L −1 at a depth of 30 cm halfway through the experiment (Fig. 6).The highest NH + 4 concentrations were found at FS, 15-20 cm below the Sphagnum surface, i.e., ca.10-15 cm above water level, during the dry period (Fig. 6).Between days 77 and 78 ammonium concentrations declined, which was probably caused by a dilution from wetting, but returned to previous levels within 5 days (Fig. 6).Total dissolved N remained low at the low long-term N deposition sites (DS) and increased substantially in the high long-term N deposition site (FS) (Fig. S6a).Higher concentrations were found mainly above the water table (Fig. 6).The concentrations varied from 1.3 to 5.1 mg L −1 at Degerö Stormyr and from 2.4 to 17.3 mg L −1 at Frölichshaier Sattelmoor.Dissolved organic N also increased with long-term N deposition during the experiment (Fig. 6, Table 1).Sites with a legacy of high long-term N deposition (LV, FS) had elevated DON concentrations reaching 5.3 mg L −1 (FS), whereas the other sites contained less than 1 mg DON L −1 .Particulate N content at the bottom outlet of the cores was between 0.2 and 5 mg L −1 (Fig. S6b).The highest PN concentrations, recorded for sites with intermediate and high long-term N deposition levels (WM, LV, FS), were measured during first 15 days of the experiment and declined substantially thereafter (Fig. S6b).At Degerö Stormyr, PN and TDN contents in the pore water were similar and generally did not exceed 3 mg L −1 (Fig. S6).At other sites, PN was on average 20 to 90 % lower than TDN.

15 N distribution in peat and pore water
The distribution of the experimentally applied 15 N in peat was influenced by background N deposition rates and decreased with depth.A sharp decrease in the 15 N signal, expressed as the difference between the 15 N signal recorded at the end of the experiment and the natural 15 N background, was observed between 10 and 15 cm depth and 78-98 % of the total 15 N peat pool was retained in the top 20 cm of the peat profile (Fig. 7).The isotopic signal was elevated most at sites with the low and intermediate long-term N deposition levels and reached 4 at.% 15 N excess in the near-surface peat (Fig. 7a) at the end of the experiment.At sites with high long-term N deposition rates, however, the amount of 15 N in  excess of the background was much lower -in a range of 0.2-0.6 at.% in the top layers (Fig. 7, DS).The mean values oscillated between 0.28 (LV) and 1.95 (DS) at.% 15 N excess.
Volume-based 15 N content in peat was calculated by multiplying the experimental isotopic signal of the 15 N content in peat layers by their bulk density, and the pattern was similar to the distribution of the isotopic signal (Fig. 7b).The isotopic signal in pore water was coupled to the 15 N distribution in the peat, as contents decreased both with depth and background N deposition (Fig. 8a and b).The 15 N-NO − 3 and 15 N-NH + 4 concentrations were only lower than expected in FS cores, which was most probably a result of dilution with abundant, unlabeled dissolved N at this site.The 15 N excess in nitrate was similar among sites (Fig. 8a), but it substantially differed regarding ammonium (Fig. 8b).Volumebased dissolved inorganic 15 N content -the product of DIN concentration, volumetric water content, and at.% 15 N-DIN excess -differed significantly (Kruskal-Wallis, p<0.001) between nitrate and ammonium but not between sites (Fig. 8c  and d).The differences mostly resulted from the discrepancies in concentrations between these two N species (Fig. 6).The ammonium concentration was up to 100 times higher than the corresponding nitrate concentration.Concentration of 15 N-NH + 4 generally decreased with depth, whereas 15 N-NO − 3 did not show any specific pattern and remained at low levels at Degerö Stormyr, Cors Fochno, and Whixall Moss, with exception of the bottom layer (Fig. 8c and d).
A more detailed examination of the individual cores from Degerö Stormyr (low background N deposition) and Frölichshaier Sattelmoor (high background N deposition) revealed two patterns of the dissolved 15 N-NH + 4 distribution in the peat (Fig. 9).Frölichshaier Sattelmoor had up to 10fold larger at.% 15 N excess than Degerö Stormyr (0.036-0.349 at.% 15 N-NH + 4 excess and 0.002-0.045at.% 15 N-NH + 4 excess, respectively).The 15 N-NH + 4 concentration decreased gradually with depth in the FS cores, but peaked between 16 and 21 cm below the surface in the DS replicates (Fig. 9).

Nitrogen transfer within the mesocosm system
The aim of this mesocosm study was to characterize changes in the transport and transformation of N within ombrotrophic peatlands that are induced by long-term N deposition.We are aware of a number of experimental limitations.First of all we did not address all aspects of N pools and transformations -uptake of N by microorganisms and transfer of 15 N into DON were, for instance, not investigated.It should be kept in mind as well that the results were obtained with experi- mental deposition of nitrate, which is generally the dominant inorganic N source in non-agricultural areas (Galloway et al., 2004), yet may not be the preferred form of N assimilated by mosses (Wiedermann et al., 2009).We also have no means to quantify how much of the applied 15 N nitrate remained physically attached on plant surfaces rather than being absorbed, although we attempted to minimize this problem by frequent irrigation and rinsing with distilled water.Despite these limitations, the results at hand provide some insight into the fate of N entering ombrotrophic bogs in the form of nitrate subsequent to varying long-term N deposition rate.While 15 N retention in Sphagnum and peat dominated in all cores (Table 2), the distribution of the 15 N tracer between the mesocosm pools differed depending on the background N deposition.We found a decline of N recovery from the peat pool and an increase in N recovery from shrubs and the dissolved pool with higher long-term N deposition (Tables 1, 2).A similar change in the N recovery of added 15 N-nitrate was observed by Xing et al. (2010) after 7 years in a nutrient fertilization study at the Mer Bleue bog in Ontario, where controls and treatments were fertilized with 1.6, 3.2, and 6.4 g N m −2 yr −1 applied in 7 doses during the vegetation period.The retention efficiency of vegetation pools was lowest in the mesocosms from the British sites characterized by the intermediate longterm N deposition levels (Fig. 5), which is in conflict with expectations.Curtis et al. (2005), for example, argued for a retention efficiency peaking at intermediate deposition levels and decreasing towards polluted sites.The data obtained by us do not allow clarifying the reason for this phenomenon.
The differences, which were not significant, may just be a result of differences in the biomass pool sizes and differing plant species intercepting the tracer.The peat contained not only the largest N pool (Fig. 3) but also similar or even more 15 N than the moss layer (Table 2), which here was defined to reach down to a depth of 5 cm beneath the moss canopy.This finding is in agreement with a previous mesocosm study performed by Blodau et al. (2006) using peat cores and 15 N-nitrate as a tracer from the Mer Bleue Bog exposed to approximately 1.6 g N m −2 yr −1 and the Experimental Lake Area (Ontario) exposed to approximately 0.3 g N m −2 yr −1 , and the previously mentioned field study by Xing et al. (2010).A substantial mass transfer of 15 N into the peat occurred, although only very little dissolved 15 N was recovered (Table 2).Blodau et al. (2006) could not explain the N transfer into the peat by solute movement and speculated that 15 N was transported in the form of the particulate organic N (PON), which was not analyzed in their study.The movement of PON could, to some extent, explain high recoveries of 15 N from the peat pool also in this case, although no experimental evidence can be provided.One can speculate that the PN concentrations measured at the bottom outlet of the cores were probably only a fraction of those in the upper layers considering that the highest TDN concentrations were measured between 15 and 25 cm below the Sphagnum surface (Fig. 6).
Mosses not only intercepted large amounts of 15 N in the mesocosms but also retained the tracer most effectively relative to their biomass.Since Sphagnum serves as an effective filter for inorganic N under unpolluted conditions (Lee et al., 1990), we expected 15 N-nitrate to accumulate most strongly in the mosses of Degerö Stormyr (Table 1, Fig. 2), and inorganic N concentration to be very low beneath.The Swedish site was indeed most efficient in retaining N, storing on average 0.25 mg N g −1 , mostly in Sphagnum (0-5 cm) and shrubs (Fig. 5).We also recovered 65 % of tracer from the Degerö Stormyr Sphagnum pool only (Table 2) and measured the highest 15 N content in the near-surface peat (Fig. 7) at that site.In line with our hypothesis that N mobility would increase with a legacy of long-term N deposition, particularly above previously found N threshold values of 1.0-1.5 g N m −2 yr −1 (Bragazza et al., 2006;Lamers et al., 2000), we found polluted sites (LV and FS) to contain the largest dissolved N pools (Fig. 3) and the highest nitrate concentrations (Fig. 6).Nitrate not assimilated by N-rich Sphagnum (Fig. 2) leached through the Sphagnum layer and entered the rooting zone of the vascular plants.This finding was also reflected in the higher 15 N recovery from shrubs with increasing background N deposition (Table 2).
Remarkably, however, large amounts of 15 N were still retained in the N-rich Sphagnum layers.Recoveries of 58 and 45.5 % for Lille Vildmose and Frölichshaier Sattelmoor, respectively, were in the range of the recovery recorded for the Sphagnum layer (0-5 cm) from Degerö Stormyr.The finding suggests that under N-rich conditions Sphagnum mosses can still filter inorganic N, most likely by the ongoing formation of new biomass, which continuously requires N. Recovery of 15 N from the Sphagnum layer is further in agreement with the Sphagnum biomass and growth data obtained in this study.In mesocosms from Lille Vildmose and Frölichshaier Sattelmoor, Sphagnum mosses had a similar biomass than in the other mesocosms at the end of the experiment (Fig. 1).Particularly in mesocosms from N-rich Frölichshaier Sattelmoor, moss length growth was strong during the dry and wet period as well (Fig. S3).The ongoing 15 N uptake by mosses in the experiment thus appears plausible.Our results corroborate previous studies where high N deposition rates did not seem to have a detrimental effect on Sphagnum growth (e.g., Granath et al., 2009;Heijmans et al., 2001;Limpens and Berendse, 2003) and other, indirect changes, such as in the abundance of vascular plants, seemed to be more important for Sphagnum production (Juutinen et al., 2010;Limpens et al., 2011).When the plant-peat soil system remains intact and dominated by mosses and shrubs, as was the case across the selected sites, a threshold at which the filter function of Sphagnum is entirely lost does not seem plausible.
The robustness of N uptake by mosses may further be assisted by physiological acclimation and species replacement.We allude that an adjustment to higher N inputs in Lille Vildmose and Frölichshaier Sattelmoor may have contributed to the greater resilience of Sphagnum species against enhanced long-term N deposition.The question arises whether (a) high N retention in the N-rich Sphagnum layers was a consequence of the ability to better utilize N by fast-growing species (S. papillosum (FS), Table 1) or whether (b) a shift to more tolerant species had occurred.Heijmans et al. (2001) found that Sphagnum growth during three growing seasons was not limited by high (5 g N m −2 yr −1 ) levels of N deposition, and mosses previously exposed to a background deposition of ca.1.5 g N m −2 yr −1 were still able to assimilate a large portion of experimentally deposited N. The potential ecophysiological adjustment of mosses to an elevated N deposition could evolve, for example, by a gradual increase in the Sphagnum tissue N content (Wiedermann et al., 2009).Given that the species in the Lille Vildmose and Frölichshaier Sattelmoor mesocosms differed, it cannot be ruled out a priori either that a shift to different species assisted in maintaining the filtration capacity of Sphagnum at these sites.Historical information of Loeske (1903) does not support this idea, however, because in the beginning of the 20th century S. papillosum already dominated in the area of Torfhaus, where the Frölichshaier Sattelmoor mesocosms were sampled.Thus, the scenario of species' shift does not seem credible, at least with regard to the Frölichshaier Sattelmoor peatland.
Nitrogen uptake by Sphagnum and higher plants was influenced by peat water content.Since there is no mechanism for controlling N uptake in Sphagnum (Jauhiainen et al., 1998), a prolonged exposure to high ambient N concentrations may threaten their vitality.Sphagnum mosses can acquire N in three ways: (1) from airborne drift via wet and dry deposition (e.g., Li and Vitt, 1997), (2) by recycling from older parts to the youngest, most actively growing ones (Aldous, 2002a;Malmer, 1988;Urban and Eisenreich, 1988), and (3) by N 2 fixation (Damman, 1978;Hemond, 1983).During the first part of the experiment, under dry conditions but with moss growth ongoing, tracer input from the top was the main N source for mosses.Non-symbiotic N 2 fixation might have also played a role but was not assessed.Evidence for the significance of this process is limited to date.Waughman and Bellamy (1980) found bogs to generally have a low activity of nitrogenase and demonstrated an annual N 2 fixation of 0.07 g N m −2 , which is low in comparison to input by precipitation.However, recently, Larmola et al. (2013) estimated the annual N 2 fixation at the Finnish peatland to range from 0.1 to 2.9 g N m −2 -values up to 10 times greater than the inorganic N deposition rate recorded at the studied region.As for the differing effect of raised water table on 15 N uptake by mosses and vascular plants we can only speculate.More N, particularly ammonium, from deeper, now inundated layers likely reached the living tissue of Sphagnum plants.This additional, 15 N-poor influx probably diluted the 15 N input from irrigation, resulting in a lower Sphagnum 15 N content during the wet period (Fig. 4).Vascular plants, however, assimilated substantially larger amounts of the tracer during the second part of the experiment (Fig. 4).As the direct absorption of atmospheric N through leaf tissue is minimal, root uptake from pore water is the most common N supply for shrubs (Li and Vitt, 1997).Shrubs can experience oxygen deficiency when faced with continuously high water levels since they lack aerenchyma.It seems plausible that inundation led to a relocation of root activity to the upper, aerated peat layers, where more 15 N was present according to our measurements (Figs. 7 and 8).Murphy et al. (2009) suggested that shallow water tables in bog systems promote a shallower root distribution.It is also possible that more of the added 15 N surpassed the moss layer and entered the rooting zone of the vascular plants due to a generally raised N supply to the mosses when water table was high.

Nitrogen transformations and transport in the dissolved phase
The 15 N tracer penetrated deeper into the peat and more 15 N was recovered from shrubs and graminoids in the cores from the sites with a legacy of strongly elevated N deposition (LV, FS).This finding agrees with the hypothesis of increased N mobility and decreased retention in mosses and near-surface peat with long-term N deposition (Table 1).We did not detect any N 2 O fluxes but we observed dissolved inorganic N in the pore water and recorded higher concentrations of organic N at polluted sites.Concentrations of N 2 O fluxes are difficult to quantify due to the episodic nature of N 2 O production.Moreover, acidic conditions typical for ombrotrophic peatlands can slow down or even inhibit denitrification (Williams et al., 1999a).The increased concentrations of DIN and TDN detected below the Sphagnum layer might be evidence of reduced retention of the Sphagnum layer caused by long-term elevated atmospheric N inputs at the study sites (FS).Such an observation was, for example, reported in the studies by Lamers et al. (2000) and Limpens et al. (2003).The concentrations of DON also increased with long-term N deposition (Table 1) and were highest at the Frölichshaier Sattelmoor site (Fig. 6).A similar pattern was recorded by Bragazza and Limpens (2004) during a field study of six European bogs spanning a range of chronic atmospheric N inputs from 0.2 to 2.0 g m −2 yr −1 .In their study, both DIN and DON concentrations, measured in water samples collected from plots characterized by similar moss composition, increased with background N deposition but the latter increased more sharply.Baxter et al. (1992) reported an accumulation of amino acids in mosses receiving NH 4 NO 3 , which could be a potential source of DON released into the water surrounding mosses.
The tracer application as NH 15 4 NO 3 indicates that all 15 Nammonium found in pore water (Figs.8 and 9) had to result from transformation of nitrate to ammonium.Such a process can take place in either the soil through microbial reduction (Hemond, 1983) or in the plant tissue through an enzymatic reduction of assimilated nitrate (Guerrero et al., 1981).At Degerö Stormyr, where only very little nitrate was present (Fig. 6), there was no potential for nitrate reduction.In Cors Fochno and Whixall Moss cores the 15 N application increased the amount of nitrate over the duration of the dry period (Fig. 6) and more 15 N-NH + 4 was found at these sites as well (Fig. 8).The highest 15 N-NH + 4 excess was recorded at Lille Vildmose, where nitrate concentrations were also high.In Frölichshaier Sattelmoor cores, where nitrate concentrations reached 38 mg L −1 , most probably nitrate ammonification took place since ammonium accumulated towards the end of the dry period (Fig. 6).The lower isotopic signal recorded at the Frölichshaier Sattelmoor mesocosms could imply a dilution effect caused by lower concentration of N in the applied solution in comparison to the background N values.
The mechanisms of ammonium release may have changed with long-term N deposition, as suggested by the distribution of the 15 N tracer (Fig. 9) in the mesocosms: at the sites with low N inputs (DS), nitrate was apparently first assimilated by the plant layer and later partly released as ammonium from the vascular plant roots.In these cores a direct microbial transformation from nitrate to ammonium seems implausible, because nitrate concentration levels were very low (Fig. 6).At the sites with elevated N inputs (FS) some nitrate apparently bypassed the living moss layer (Fig. 8c) and was converted to ammonium in the pore water.However, this pattern does not necessarily indicate a loss of the Sphagnum filter.Even though the N content of Sphagnum tissue at Frölichshaier Sattelmoor was at 10 ± 3.8 mg g −1 (Fig. 2)the range encompassing the threshold N concentration suggested by Lamers et al. (2000) to inhibit further N content increase -the mosses did not suffer a substantial decline in biomass in the field and grew most of all in the experiment (Fig. S3, Table S1).

Conclusions
Our results confirm that a legacy of elevated N deposition in ombrotrophic peatlands, especially above levels of about 1.0-1.5 g N m −2 yr −1 , leads to lower retention of N, increased inorganic N concentrations in pore waters, and enhanced transfer of the deposited N to shrubs.Higher inorganic and organic dissolved N concentrations and more N in near-surface peat were found, and more of the 15 N label was recovered from the aboveground shrub biomass at the polluted sites compared to the nearly pristine Degerö Stormyr mire in Northern Sweden.As a consequence, a higher mobility of dissolved N in peat is probable under more realistic hydrologic conditions than present in our experiment.The shift to higher N mobility in the plant-peat soil system was, however, more gradual and modest than we expected.It was not visibly linked to the disappearance of Sphagnum mosses at the high long-term N deposition or to shifts in species composition, if the somewhat limited information from our most polluted site at the beginning of the 20th century is to be believed.Raised uptake of N by vascular plants at sites exposed to higher long-term N deposition was accompanied by a Sphagnum N absorption that remained substantial.Some previous studies (Bubier et al., 2007;Gunnarsson and Rydin, 2000;Juutinen et al., 2010) alluded that the vitality of Sphagnum suffers with increasing N saturation, which was undoubtedly the case in the mentioned fertilization studies.Under conditions of our short-term experiments, however, the filter function of mosses appeared to remain mostly intact, even when sites had been exposed to long-term deposition levels considerably higher than 1.0-1.5 g N m −2 yr −1 .
The Supplement related to this article is available online at doi:10.5194/bg-13-707-2016-supplement.

Figure 3 .
Figure3.Total nitrogen content (mean ±SD, n = 3) in the ecosystem pools investigated standardized to the peatland surface area.The Sphagnum pool contained mass from 0 to 5 cm below surface.Letters a and b -significant differences in the mean values of the sites at each pool (one-way ANOVA, Tukey test).A bar without a letter -no significant difference.Site abbreviations as in Fig.1.

Figure 4 .
Figure 4.15 N abundance (mean ±SD, n = 3) expressed as at.% 15 N excess in vegetation during dry and during wet period.(a) Sphagnum capitulum, (b) Sphagnum stem, (c) shrubs, and (d) graminoids.One-way ANCOVA indicated a significant effect of N deposition for Sphagnum capitulum and graminoids, as well as a significant effect of the dry and wet period for Sphagnum capitulum and shrubs; however, a Tukey post hoc test failed to identify significant differences for individual treatments.Site abbreviations as in Fig.2.

Figure 5 .
Figure5.15 N retention efficiency (mean ±SD, n = 3) of ecosystem pools, calculated by dividing retained 15 N in each pool at the end of the experiment by dry mass of the corresponding pool."Average" represents the mean 15 N retention efficiency of all pools (mean ±SD, n = 4).Letters a and b -significant differences in the mean values of the sites at each pool (one-way ANOVA, Tukey test).A bar without a letter -no significant difference.Site abbreviations as in Fig.1.

Figure 9 .
Figure 9. Atomic percent of 15 N-NH +4 excess in pore water of three replicates from Degerö Stormyr (left) and Frölichshaier Sattelmoor (right).Each symbol in the profile represents the middle depth of the section from which the water was extracted.The scale of the x axes varies across panels.

Table 1 .
Characteristics of the sampled peatlands.

Table 2 .
Recovery of 15 N (%) for all sites and percent contribution of N ecosystem pools to the total recovery at each site after 24 weeks of tracer application.NO − 3 and NH + 4 were analyzed in peat pore water.Standard deviations are given in parentheses.