Peatlands store more nitrogen (N) per unit area than any other soil type (Limpens et al. 2003). There are concerns that increasing N inputs will lead to peat degradation, elevated emissions of greenhouse gases, and accelerated warming (Woodwell et al. 1995). Under high N deposition, Sphagnum is no longer able to prevent N leaching into deeper peat layers (Lamers et al. 2000). Consequently, roots of invading vascular plants oxygenate deeper peat and augment organic matter decomposition (Bragazza et al. 2013).

Construction of long-term N mass budgets in peat cores is one approach to assessing the response of peatlands to global change (Wu and Blodau 2013). Recently, two studies have observed “excess N accumulation” in vertical peat profiles in high-latitude wetlands, characterized by very low deposition of reactive nitrogen. Vile et al. (2014) reported excess N accumulation in ombrotrophic (rain-fed) bogs, whereas Larmola et al. (2014) arrived at a similar conclusion for minerotrophic fens. A comparison was made between the cumulative N deposition in recent decades, and the amount of N accumulated in 210Pb-dated peat cores over the same period of time. Such studies assume that N deposition in remote unpolluted areas has never been greater than today. The authors measure present-day N fluxes via rainfall, and extrapolate the unknown contribution of N by horizontal deposition (fog interception and dry deposition). Most studies assume that horizontal deposition can be as high as vertical (Vile et al. 2014). Total deposition is calculated by doubling the vertical deposition to safely account for the unknown, horizontal portion. Clearly, excess N stored in peat can be better quantified after actual measurements of horizontal N deposition into peat bogs are performed. Nitrogen budget discrepancies can then be more rigorously identified and interpreted.

Our objective was to estimate the maximum contribution of horizontal deposition to total N deposition at two polluted ombrotrophic peat bogs in Central Europe. We hypothesized that in Sphagnum-dominated landscapes, N flux via horizontal deposition is smaller than N flux via vertical deposition. We also hypothesized that a more polluted site would exhibit a higher contribution of horizontal deposition to total N input, compared to a less polluted site. Both study sites are located in the Czech Republic (Fig. 1). The peat bogs of Kunstatska kaple (KB; north) and Blatenska slat (BS; south) are situated on mountain plateaux at elevations greater than 1,000 m (Table  1). KB is located close to a large cluster of coal burning power plants (Silesia, Poland), and was historically more polluted with sulfur and nitrogen oxides (SOx and NOx), compared to BS. Between 1970 and 1996, the area of KB (north) was affected by massive spruce die-back related to acid rain. In 1999, the area around SB (south) was affected by spruce defoliation caused by a bark beetle infestation. Nitrogen deposition at both sites has been decreasing since the early 1990s (Novak et al. 2014a, b).

Fig. 1
figure 1

Location of the study sites. High-elevation areas characterized by frequent fogs are marked with dark grey colour. Lowlands (while and light gray) are associated with fewer foggy days per year. Data by Czech Hydrometeorological Institute, Prague

Full size image
Table 1 Study site characteristics
Full size table

Fig. S1 in the Electronic Annex gives monthly data on the water table depth at three previously studied ombrotrophic bogs in the Czech Republic. The average water table depth across these sites during the growing season was 12 cm below the top of Sphagnum capitula. At KB and BS, the water table depth at the time of sampling was 7 cm below bog surface. At each of our study sites, we collected five typical Sphagnum cuspidatum plants. For planimetry, we considered 7-cm long S. cuspidatum plants. The number of leaves was counted under a binocular microscope (Nikon SMZ 800 N). The surface area of leaves and stems was approximated by triangles and cylinders. The surface area of individual leaves was measured using the NIS ELEMENTS AR 3.2 software on a Nikon Eclipse 600 polarizing microscope. Fig. S2 (Electronic Annex) gives an example of the planimetric procedure, along with a generalized mathematical formula for the calculation of Sphagnum surface. The surface of an average Sphagnum plant was multiplied by 175, the average number of Sphagnum plants per research plot (a 10-cm diameter circle; an average for three replicate plots). The calculated S. cuspidatum surface area per a 78.5-cm2 research plot was 5,660 cm2. The same surface area was modelled by 7 cm-high “bristles”, formed by polyethylene (PE) strands. Loose vertical strands were prepared from mesh (0.3 by 0.3 mm openings) by removing horizontal strands. The surface area of the removed horizontal strands was disregarded in surface area calculations. At the bottom of the loose vertical strands, an intact stripe of the mesh was preserved for easier handling. 150-cm long, 8 cm-wide strips were wrinkled and folded at the bottom, using a nylon thread (Fig. S3, Electronic Annex). In all, a 13.5-meter long strip of the mesh was used in one fog collector, separated into nine bristles. The nine bristles completely filled a cylindrical PE funnel, 10-cm in diameter (Fig. S4, Electronic Annex). The funnel was attached to a 2-L PE bottle for solute collection. The top of the fog collector was situated at a similar level as the surrounding Sphagnum capitula. For the measurement of horizontal deposition, three replicate collectors were used at each site. Vertical deposition was simultaneously measured by a cylindrical PE funnel, 10-cm in diameter, at each site.

Our field experiment was carried out between April 19 and November 15, 2013. Fog and rain collectors were exposed to atmospheric deposition for 84 days at KB and 92 days at BS. One precipitation sampling was performed in spring, and two precipitation samplings were performed in autumn. No sampling took place between July and September because of lower fog occurrence. For the July–September period, fog collectors were removed from the research plots. At the end of each sampling, the PE samplers were thoroughly rinsed by DDW, the sample volume and the concentrations of NO3 −, NH4 + and total organic nitrogen (TON), were measured. Nitrogen fluxes were calculated for a 12-week period. Horizontal N deposition was calculated by subtracting vertical deposition from total deposition onto fog collectors. Statistical analysis was performed using the PASW (Version 18) software by SPSS.

The more polluted site KB exhibited 3.4 times higher horizontal N deposition than the less polluted site BS (Fig. 2; S. cuspidatum). Relative to vertical N deposition, horizontal N deposition added 69 % N at KB and 35 % N at BS (Fig. 2). KB exhibited a higher contribution of horizontal deposition to total N input, compared to BS. Ammonium–N input was greater than nitrate–N input at both sites (Fig. 2). This indicates that agricultural N sources were more important at both sites than industrial N sources. Organic N represented the smallest N input, averaging 14 % of total N at both sites.

Fig. 2
figure 2

A comparison of horizontal (grey) and vertical (black) N deposition in Sphagnum bogs (mean ± SE). The sum of horizontal and vertical deposition was measured by PE collectors with a high surface area, identical to that of locally growing Sphagnum. Vertical deposition was measured by rain collectors installed in an unforested area. Total Nr = NO3 −-N + NH4 +-N + organic N. For the statistical evaluation, a mixed model with covariant structure was used, corresponding to repeated measures (PASW 18 software by SPSS). KB and BS differed significantly in the sum of vertical and horizontal N deposition (p = 0.001). Deposition of NO3 −-N, NH4 +-N and organic N also differed significantly (p = 0.150). Different letters mark significantly different values (p < 0.05). Only two replicate measurements were available for organic N at BS (*)

Full size image

Our sampling design focussed on fog-rich, snow-free seasons (spring and autumn, cf. Fig. S5, Electronic Annex). Therefore, our estimated percentage of horizontally deposited N, relative to rainfall N, represents an upper limit. In the following paragraphs, we discuss some of the caveats intrinsic to our approach, and attempt to scale our temperate-zone S. cuspidatum results to boreal regions. Compared to our Central European study sites, peat bogs in the boreal zone are usually characterized by lower elevation, lower precipitation totals, fewer foggy days per year, less pollution, and more densely packed Sphagnum carpets.

Large segments of the boreal regions are located on the Canadian and Siberian Shield at elevations of 300–400 m, i.e., 600–700 m lower than our study sites. As seen in Fig. 1, fog frequency in Central Europe decreases with a decreasing elevation. In the Central part of the Czech Republic (elevation of 350 m), less than 60 foggy days per year were recorded by the Czech Hydrometeorological Institute (Fig. 1). This is about five times less than in the mountainous regions near our study sites (up to 300 foggy days; Fig. S5, Electronic Annex). Rainfall totals at our upland study sites and the lowland center of the country differ by a factor of 2.5 (ca. 1235 mm vs. 500 mm per year). For the KB–BS nitrogen pollution range, a proportional scaling down the fog-frequency gradient gives 14–28 % horizontal N deposition, relative to vertical N deposition. We note that in other parts of the world, low-elevation sites may be associated with a very high fog frequency, depending on local climatic conditions and the amount of condensation nuclei (dust particles) present in the atmosphere.

In unpolluted boreal regions, Vile et al. (2014) reported a vertical deposition of 1 kg N ha−1 year−1. In Fig. 2, a less N-polluted site had a lower horizontal N deposition, relative to vertical N deposition, than a more N-polluted site (35 vs. 69 %). Accordingly, the unpolluted sites studied by Vile et al. (2014) may be characterized by a total Nr deposition of 1.35 kg ha−1 year−1, or less. For a more rigorous extrapolation of our data (Fig. 2) to the pristine sites studied by Vile et al. (2014), horizontal deposition should be measured at more than just two sites along the N-pollution gradient.

We note that our estimates of S. cuspidatum surfaces may be too high for typical boreal Sphagnum species, such as S. fuscum and S. capillifolium. We were not able to identify S. fuscum at our relatively warm sites. We scaled our data to two forms of S. capillifolium, a red short form, typical of thick carpets (Fig. S6), and a green tall, “fluffy” form. Planimetry was carried out on both S. capillifolium forms, as described above (n = 3). The surface of the red dense-carpet form of S. capillifolium was measured only to the depth of 1.5 cm from the top of the capitula (cf., Fig. S6, Electronic Annex), whereas the surface of the green S. capillifolium form was measured to the depth of 7 cm, similar to S. cuspidatum. It was unlikely that fog droplets (mean diameter of 10 μm; Pruppacher and Klett 1997) could freely penetrate deeper than 1.5 cm below the tops of the capitula of the red S. capillifolium form. The calculated Sphagnum surface area per 78.5-cm2 research plot was 3,666 cm2 for the red S. capillifolium, and 1,867 cm2 for the green S. capillifolium. In Fig. 3, we scaled horizontal N deposition in S. cuspidatum to lower surface areas of S. capillifolium. Because plastic collector surfaces eliminated inner cycling of deposited N within aboveground Sphagnum biomass, we suggest that the amount of horizontally captured N may be directly proportional to Sphagnum surface area, i.e., land surface roughness. Horizontal N deposition in Fig. 3 (grey circles, y-axis) was then recalculated relative to the vertical N deposition. Horizontal N deposition may contribute between 12 % (minimum; green S. capillifolium at BS) and 45 % (maximum; red S. capillifolium at KB) to vertical N deposition. Generally, our scaling trials resulted in a horizontal N deposition not exceeding 45 % of the vertical N input.

Fig. 3
figure 3

Extrapolation of horizontal deposition from Central European peat bogs to Sphagnum species prevailing in boreal regions. Scaling from S. cuspidatum to S. capillifolium along a Sphagnum surface area gradient. See text for details

Full size image

Table  2 gives data on atmospheric N deposition in two spruce-forested catchments, situated in the vicinity of KB and BS, respectively (Fig. 1). Nitrogen input fluxes in the catchments have been measured monthly for a period of 20 years (1994–2013). In a spruce-Sphagnum analogy, open-area deposition exactly corresponds to vertical deposition, while forest canopy throughfall only vaguely corresponds to total deposition (PE fog collectors). Similar to our fog collectors, canopy throughfall includes fog interception and dry deposition, in addition to rainfall. In contrast to our fog collectors, canopy throughfall contains N newly leached from the foliage, while a portion of atmospherically deposited N is immobilized in the aboveground biomass (Gebauer et al. 1994). In the more polluted north, canopy throughfall added 120 % N relative to vertical N deposition (Table  2). In the medium-polluted south, canopy throughfall added 21 % N relative to vertical N deposition, i.e., 10 times less. Spruce canopy throughfall exhibited a larger between-site variability in captured N, compared to PE collectors in peatlands (ten-fold vs. two-fold, cf., Table 2; Fig. 2). We propose that forest canopy throughfall is not a suitable proxy for estimating horizontal N deposition into open Sphagnum-dominated peatlands in the Czech Republic. The amount of N immobilized in, and leached from, forest canopy may not be at a steady state. The N sink/source balance in aboveground forest biomass depends on a number of site-specific parameters, such as tree species, age and health status, N pollution level, and chemical speciation of incoming N. In peat-bog N mass balances, plastic constructs, which eliminate cycling of atmospherically derived N in aboveground biomass (cf., Clymo 1963), may be a more promising proxy for estimating horizontal wet and dry deposition than actual Sphagnum throughfall collectors.

Table 2 Atmospheric deposition of reactive nitrogen
Full size table

We note that our plastic constructs do not precisely mimic the forms of Sphagnum plants. In future studies, the effect of Sphagnum forms may be tested by a set of fog collectors with an identical surface area but contrasting surface geometry. It also remains to be seen whether interception of alkaline fog droplets is more efficient on an acidic Sphagnum surface than on a neutral surface of a PE construct. If so, this mechanism may affect the cumulative N deposition only at the begining of a foggy period. Over time, pH of the solutes on Sphagnum surface is likely to increase, converging to the pH of the atmospheric deposition.

We conclude that in reconstructions of long-term N deposition into open Sphagnum bogs, horizontal deposition should be considered as a significant additional N input. However, even in high-N deposition areas, this unmeasured horizontal N input is likely smaller than the measured vertical N input via rainfall. Preliminary scaling of Central European S. cuspidatum data to conditions typical of the boreal regions confirmed this conclusion. For biogeochemical modelling in Sphagnum-dominated landscapes, a factor of 1.7, or less, can be used to convert vertical to total N deposition. Our data support the conclusions by Vile et al. (2014) who argued that horizontal N deposition was too low to explain excess N storage in peat and invoked high N2 fixation rates in pristine wetlands.