Introduction

Effects of atmospheric nitrogen deposition on semi-natural ecosystems have been intensively studied in the last three decades. Nutrient-poor and weakly buffered ecosystems are particularly vulnerable. In dry ecosystems in north-western and central Europe, negative impacts of high atmospheric deposition (15–30 kg N ha−1 y−1) have been described for forests (Houdijk and others 1993; van Dijk and others 1989, 1990), heathlands (Bobbink and others 1992; Power and others 1998; Roelofs 1986; van den Berg and others 2005) and some types of coastal dunes (Jones and others 2004; Kooijman and others 1998). However, even low levels of atmospheric deposition may have negative effects. For coastal dunes of the Baltic Sea, which only receive 3–8 kg N ha−1 y−1, critical loads for dry, acid, lichen-rich grasslands as low as 4–6 kg N ha−1 y−1 have been suggested (Remke and others 2009).

Increased atmospheric N-deposition may affect various ecosystem patterns and processes. Elevated N-loads not only stimulate eutrophication and plant growth, but also lead to acidification and loss of buffer capacity, increasing the availability of toxic metals (Al3+ and Fe3+) within the soil (Aerts and Bobbink 1999; Heij and Schneider 1991). Al and Fe concentrations of 50–100 ppm in plant tissue can harm acid intolerant species and disturb the P-metabolism (Fink 2007). Higher soil Al/Ca-ratios are assumed to be toxic for plants growing at intermediate pH levels (de Graaf and others 1997) as Ca2+ inhibits the uptake of Al3+ (Marschner 1995).

High N-deposition may also accelerate succession rates in acid grasslands, heathlands and dune pools (Bobbink and others 1998; Achermann and Bobbink 2003). Vegetation becomes dominated by tall graminoids like Deschampsia flexuosa (L.) Trin., Molinia caerulea (L.) Moench, Carex arenaria L. and Calamagrostis epigejos (L.) Roth (Bobbink and others 1998; Kooijman and others 1998; Remke and others 2009). Processes behind this species change are a switch from competition for nutrients to a competition for light and space, and resistance of plant species to toxicity such as high metal concentrations or a very low pH (Bobbink and others 1998; de Graaf and others 1997; Kleijn and others 2008). In the Baltic region, dry, lichen-rich dune grasslands are reduced as N-affected dunes become dominated by Carex arenaria (Remke and others 2009). Under unpolluted conditions, the succession pathway starts with bare sand, which is then slowly colonized by a few pioneer graminoids like Corynephorus canescens P. Beauv. and Carex arenaria, by lichens like Cetraria muricata (Ach.) Eckfeldt and mosses like Ceratodon purpureus (Hedw.) Brid. With time, the bare soil is totally covered by low, but highly characteristic vegetation. These short grasslands contain about 20–25 plant species (per relevee of 16–25 m2), mostly lichens and mosses (Ellenberg 1996). Grey dunes and wet dune slacks add an essential part to the species diversity of coastal dunes and habitats. Species diversity of coastal habitats comprises 40–70% of the total diversity in Flanderen and The Netherlands (Kooijman 2004; Provoost and Bonte 2004), about 25% on the West Frisian islands (Niedringhaus and others 2008), and forms therefore an essential part of total biodiversity.

Succession from bare sand to fully developed lichen-rich dune grassland may take 50–70 years, and is accompanied by a moderate decrease in soil pH (0.5–1 pH units) and accumulation of organic matter (Ellenberg 1996). Eventually taller graminoids, dwarf shrubs or trees become dominant. The sequence from young short via old to tall grassland is one of the major succession lines for Baltic coastal dunes and has been described for the Vistula Spit (Steffen 1931), the Łeba bar (Hueck 1932) as well as the Curonian spit (Paul 1953). However, with increased levels of atmospheric N-deposition, crucial soil processes may change, and lead to dominances of dense, tall grass stands.

The aim of this study is to determine,(i) which soil processes promote the shift from lichen-rich, short, dry grasslands to a dominance of Carex arenaria, (ii) which process changes can be linked to atmospheric N-deposition, (iii) to specify characteristics of Carex arenaria stands under different atmospheric N-loads, and (iv) to describe the effects of elevated deposition loads on species composition and richness. Finally, early indicators for the influence of low atmospheric N-deposition levels are listed, and recommendations for restoration of dry, coastal dunes dominated by tall graminoids are given.

Field Sites

The 19 coastal dune sites are spread over a large geographical gradient from the Kattegat in the west of the Southern Baltic to the Baltic proper in the east (Figure 1; Remke and others 2009). The research area therefore stretches from the Atlantic through Central Europe to the Baltic terrestrial ecoregion (Olson and others 2004). The average annual temperature decreases from 8°C in Denmark to 5°C in Estonia (Table 1) as the climate gets more continental. Another gradient exists for sea water salinity and therefore salt-spray. Salinity decreases steeply within in the Kattegat from 24 to approximately 10 PSU and ranges mainly between 7 and 8 PSU in the Southern Baltic and Baltic proper (Table 1). All 19 sites have parent sand material with pHNaCl below 6.5 and receive precipitation of 500–700 mm y−1 (Table 1). All sites are open dunes without any physical barrier like a forest towards the sea, and have less than 10% cover of trees. During the last decade, management impacts such as grazing or burning have not been intensive.

Figure 1
figure 1

Map of field sites (reprinted from Remke and others 2009, with permission from Elsevier).

Full size image
Table 1 Field Sites with Synonym, Country, Geographical UTM Coordinates (from Google Earth), Long-Term Annual Mean Values for Temperature and Precipitation, and Sea Water Salinity
Full size table

The dataset was separated into two main groups, with pH of the parent material below (‘acid’) or above (‘slightly calcareous’) 6.0 (Table 2), that is, below or within the carbonate buffer range. Above pH 6.5–6.2, the system is buffered by dissolution of calcium carbonate below pH 6.0 by cation exchange, and below pH 5.0 by aluminium and iron buffer systems (Scheffer and Schachtschabel 1998). In addition, both pH-groups were separated into two N-deposition classes below and above a total wet N-deposition of 5 kg N ha−1 y−1, estimated by N-content of the biomonitor Cladonia portentosa (Remke and others 2009; Table 2). The calculated total wet N-deposition ranged from 2.6 to 7.8 total N kg ha−1 y−1, but critical loads were estimated as 4–6 kg N ha−1 y−1 (Remke and others 2009). The N-content in C. portentosa proved to be a suitable biomonitor of N-deposition in remote areas as it explained approximately 50% of the variation in wet N-deposition measured at nearby EMEP stations (regression equation: C. portentosa tissue N [%] = 0.0228 × N-deposition [kg ha−1 y−1] + 0.3385, R 2 = 0.5223, P = 0.008; further discussion see Remke and others 2009).

Table 2 Classification of Field Sites
Full size table

Materials and Methods

In each of the 19 dune sites, three 2 × 2 m2 plots were selected in June–July 2005, in a sequence from young short via old short to tall grassland. First, early successional stage is characterized by short, dry grassland, with at least 30% open sand area, second stage by older and lichen-rich, short grassland, with open sand below 5%, and third stage is dominated by tall grasses with at least approximately 50% cover of sand sedge (Carex arenaria). In each plot, species cover was estimated. If necessary, species were dried and identified using a microscope and coloring techniques. For species determination the following literature was used: for vascular plants Rothmaler and others (2002), Oberdorfer (1994) and Hegi and Conert (1998), additionally for Salix spp. Lautenschlager-Fleury and Lautenschlager (1994), for mosses Frahm and Frey (1992) and for lichens Wirth (1995). Within each plot, aboveground biomass was sampled in a circle with a diameter of 29.5 cm. Within each circle, thickness of the humus horizon (Ol, Of, Oh) was measured (Finnern and others 1994). Below this, a soil sample mixed out of three soil cores (each core cutter 100 cm3) was collected from the top 10 cm of the mineral soil layer. In addition, at each site, three soil samples (top 2–3 cm) were taken along the first dune ridge towards the open sea to determine the pH of the parent sand material. All samples were dried at 70°C (plant) and 40°C (soil) for 24 h.

The lichen, Cladonia portentosa (Dufour) Coem., was used as biomonitor for atmospheric N deposition (Remke and others 2009). Within each plot, ten 10 × 10 cm2 samples of C. portentosa were cut with a pair of scissors out of the center of a lichen patch of approximately 0.4–0.5 m diameter (Søchting 1995). Only the top 2 cm of the Cladonia were collected to exclude nutrient uptake from the soil. The samples were cleaned roughly and stored in a paper bag. In the laboratory, all samples were cleaned thoroughly, washed briefly in distilled water and dried at 70°C for 24 h.

Fresh biomass was separated into cryptogams (lichens and mosses), vascular plants and litter, and weighed after drying at 70°C for 24 h. A subsample of sand sedge was taken for further analysis. Sand sedge and Cladonia portentosa samples were ground in a centrifugal mill (rotational speed 18,000 for 1–2 × min, FRITSCH pulverisette 14, Idar-Oberstein, Germany). Total nitrogen and CN-ratio of plant and soil material (finely ground in a centrifugal ball mill, Fritsch, Idar-Oberstein, Germany) were determined with a C/N-analyzer (CHNOS element analyzer vario EL III, elementar Analysensysteme, Hanau, Germany). Total P was analyzed for sand sedge only, with 200 mg ground material digested in sealed Teflon vessels in a Milestone microwave oven (type Ethos D, Milestone Inc., Sorisole, Italy) after addition of 4 ml HNO3 (65%) and 1 ml H2O2 (30%) (Kingston and Haswell 1997).

Soil samples were sieved with a 2 mm mesh-size before the following analyses. Soil organic matter content was determined as loss on ignition (LOI) at 550°C for 8 h, and pH was measured in 0.2 M NaCl. Total extractable amounts of Al, Ca, Fe, Mg, Mn, P, S, and Zn were measured in these 0.2 M NaCl extracts, and Na in a double deionized water extract with inductively coupled plasma emission spectrophotometry (ICP-OES: IRIS Intrepid II XDL, Thermo Fisher Electric, Breda, The Netherlands). Nitrogen mineralization was measured in a laboratory incubation experiment of 26 days, with 60 g of sieved soil in glass beakers sealed with parafilm kept at 40% water holding capacity and 25°C in total darkness. NO3 − and NH4 + were measured at the start and end of the incubation period, using extractions with double deionized water and 0.2 M NaCl, respectively. Ortho-P was measured in double deionized water extracts at the start of the incubation period. Nitrate, ammonium and phosphate were measured colorimetrically with an Auto Analyzer 3 system (Bran + Luebbe, Norderstedt, Germany), using ammonium molybdate (Henriksen 1965), hydrazine sulphate (Technicon 1969) and salicylate (Grasshoff and Johannsen 1977), respectively.

Linear models and linear mixed effect models were fitted using R (R Development Core Team 2008), followed by model justification procedures. Mixed effect models were applied to overcome spatial pseudo-replication within the dataset (field site as random factor). If fitted models were not justified, generalized linear models (glm) with gamma error distributions were fitted or data were transformed by log or (double) square root before regression analysis. Multiple regressions were carried out starting with all environmental factors and subsequently simplified using stepwise backward deletion until the minimal adequate model was reached. To avoid over-dispersion no more variables than the replicate number divided by three were fitted (Crawley 2005, 2007). Tests between two classes were performed by Student’s t test, if the data were normally distributed, otherwise the non-parametric Kruskal–Wallis test was performed.

Results

Acidification

Acidification rates differed between early and later successional stages. Initial acidification, that is, differences in pH between parent material and the first succession stage, was negatively correlated with acidification during subsequent succession (Figure 2). At sites with small pH differences between parent material and the first stage of succession, the pH decreased by nearly two units during the succession from the first to the third stage. Sites following this pattern were Akmensrags, Nagliu, Pajuris, Syrodde, Pape and Gellen (Figure 2). Sites with high pH differences (~2 units) between parent material and initial succession stage, showed minimal pH decrease with further succession. Dünenheide, Harilaid, Pavilosta, Holtemmen, Keibu and Korshage were examples for this pattern. The early acidification, that is, the high pH-decrease between the parent sand material and the first stage, shows no link to N-deposition levels. The early decrease in pH (parent sand—first stage) cannot be linked to N-deposition (regression results: acid P = 0.595; slightly calcareous P = 0.845; all P = 0.660). Only the pH itself differed with N-deposition. Especially at acid sites and later successional stages the pH was significantly lower (0.3–0.5 units) at N-affected sites (Table 3).

Figure 2
figure 2

Relationship between the pH difference of the parent sand material and the first vegetation succession stage and the pH difference within succession, that is, first and the third succession stage. Every point is a site and extreme sites are labelled. For example, Pajuris or Akmensrags have a low pH decrease between the parent sand material and the first succession stage, and Pavilosta and Harilaid have a high pH difference within later succession (regression equation: y = −0.74x + 1.45, R 2 = 0.5166, P = 0.001).

Full size image
Table 3 Soil and Plant Data for All Sites, Separated for pH Classes (Acid, Slightly Calcareous), Deposition Classes (N-Unaffected and N-Affected) and Vegetation Succession Stages (First, Second and Third stage)
Full size table

Eutrophication

Soil pH and organic matter (LOI) were closely related and changed during succession, but were also affected by N-deposition (Figure 3). Organic matter increased exponentially with decreasing pH, but only in areas with high N-deposition. Plots with more than 1.0–1.5% soil organic matter and a pH below 4.0 belonged mainly to the third succession stage at N-affected sites, which are densely covered by Carex arenaria. A significant increase in organic matter with N-deposition occurred only at sites with acid parent material (Table 3). At these sites, organic matter concentration was two to four times higher at N-affected than at N-unaffected sites. At sites with slightly calcareous parent material, however, organic matter increased during succession, but did not differ between low and high N-deposition areas (Table 3).

Figure 3
figure 3

Organic matter content (LOI) versus pH of all vegetation units.

Full size image

During succession, net N-mineralization generally increased from the first to the third stage (Table 3). The amount of N-mineralization per unit soil organic matter was positively correlated with pH (Figure 4) and furthermore related to N-deposition. The increase of N-mineralization/LOI with increasing pH was three times steeper at N-affected than at N-unaffected sites (Figure 4), and N-mineralization in total was elevated two to three times at sites with high N deposition (Table 3).

Figure 4
figure 4

N-mineralization per day per LOI (g kg−1) against the pH (NaCl) for acid and slightly calcareous sites separately. (Regression results: acid sites: N-affected y = 0.238x − 0.712 (P = 0.000), N-unaffected y = 0.087x − 0.218 (P = 0.001); slightly calcareous sites: N-affected y = 0.092x − 0.194 (P = 0.005), N-unaffected not regression equation as P = 0.104).

Full size image

Toxicity

The Al/Ca-ratio increased with increasing N-deposition only at sites with acid parent material (second stage, Table 4), and was twice as high at N-affected than at N-unaffected sites (second stage, Table 3). At acid sites, the Al/Ca-ratio exceeded the values of one in all succession stages. At sites with slightly calcareous parent material, Al/Ca-ratios were 5–10 times lower. Salt-extractable Fe soil concentrations were at least two times higher at acid than at slightly calcareous sites. At acid sites, the Fe content was 1.5–3 times higher under high compared to low N-deposition.

Table 4 Regression Results of Various Factors (Dependent Factor) with Atmospheric N-Deposition (Measured as Lichen-N; Independent Factor)
Full size table

Characteristics of Carex arenaria Plants and Vegetation Stands

Increased atmospheric N-deposition (measured as lichen-N) was positively correlated to the N-content of Carex tissue, but only in early succession stages at acid sites and all sites pooled (Table 4). N/P-ratios (Carex) were positively correlated to N-deposition in all three succession stages at acid sites (Table 4). Tissue P-concentrations of Carex arenaria did not differ among sites or succession stages, and ranged between 51.2 and 56.6 μmol P g−1. There was no significant relationship between tissue-N of Carex and total N, NH4 + or NO3 − in the soil.

At acid sites, cover of Carex arenaria significantly increased under N-affected conditions in later stages of succession (Tables 3 and 4, Figure 5). Carex cover was 1.5–2 times higher under N-affected conditions. In early stages, or slightly calcareous sites, Carex cover was not affected by N-deposition. In the third succession stage, Carex cover was negatively correlated to species numbers of lichens and mosses (y = −0.09x + 8.48, P = 0.000) and positively to total biomass of vascular plants (y = 0.47x + 33.70, P = 0.027). Carex cover was not significantly correlated to soil organic matter. Furthermore, longitude, latitude, climate data (temperature, precipitation) and salinity were not significantly correlated to Carex cover in multiple regression.

Figure 5
figure 5

Box and whisker plots for the cover of Carex arenaria per plot at acid and slightly calcareous and N-unaffected versus N-affected sites for the three different vegetation succession stages. Significant differences (P < 0.05) between N-unaffected and N-affected sites are shown with an asterisk (*).

Full size image

Species Richness

At acid sites, the species number of foliose lichens, as well as number of all lichen species together (second and third stage; Figure 6), and grasses and forbs (third stage) were significantly lower (two to three species) under high N-deposition (Table 5). Foliose lichens such as Cladonia glauca Flörke, Cl. macilenta (Leighton) Arnold, Cl. coccifera (L.) Willd. and Cetraria aculeata (Schreber) Fr./muricata (Ach.) Eckfeldt, grasses such as Festuca polesica Zapał. and Ammophila arenaria (L.). Link, and forbs such as Hypochaeris radicata L., Thymus serpyllum L. and Hieracium umbellatum L. were absent or had lower cover at N-affected sites. In contrast, Carex arenaria and Corynephorus canescens were more abundant at acid sites with high N-deposition, and Rumex acetosella L. was exclusively found at these sites.

Figure 6
figure 6

Species number of foliose, shrub and total lichen (per 4 m2 plot) at acid and slightly calcareous and N-unaffected versus N-affected sites for the three different vegetation succession stages. Significant differences (P < 0.05) between N-unaffected and N-affected sites are shown with an asterisk (*) beside each bar for different lichen groups and on the top of the bar for total lichen species number (foliose and shrub lichen data pooled).

Full size image
Table 5 Species Richness for Different Life Forms Separated for Deposition Classes, pH Classes and Succession Stages
Full size table

The most important predictors for the decrease in species numbers at acid sites were selected with multiple regression. For the second succession stage, the model explained 20–50% of the variance, with Al/Ca-ratio and its (statistical) interaction with pH and N-deposition as most important factors. For the third stage, only N-deposition was a significant predictor for species richness and explained 30–60% of the variance (Table 6). Species with preference for low Al/Ca-ratios were Hieracium umbellatum (higher coverage), Cladonia glauca (higher coverage and occurrence), and Jasione montana L., Cladonia coccifera and Cladonia floerkeana (Fr.) V.Wirth (higher occurrence). Species excluded from sites with high Al/Ca-ratios were Cladonia glauca, Viola ssp. and Luzula ssp. However, Rumex acetosella, Anthoxanthum odoratum L. and Corynephorus canescens were more abundant at high Al/Ca-ratios.

Table 6 Multiple Regression Results of Species Groups with the Factors Lichen-N, Al/Ca-ratio and pH
Full size table

At sites with slightly calcareous parent material, the effect of N-deposition was less pronounced than at acid sites, and even slightly positive for overall species richness (Table 5). In multiple regression, Al/Ca-ratio and pH mainly determined species richness (Table 6). N-affected sites had one species of foliose lichen (third stage), grass (second stage) and forb (first stage) more, but one moss species less (second stage). Cladonia coniocrea auct. (foliose lichen), Hypochaeris radicata (forb), Anthoxanthum odoratum and Festuca rubra L. (grasses) were more abundant at high N-deposition. Dicranum scoparium Hedw. (second stage) had exceptionally high coverages (30–40%) at high N-deposition. However, Pleurozium schreberi (Brid.) Mitt. (moss), Cetraria aculeata/muricata and C. islandica (L.) Ach. (first and second stage) occurred only at N-unaffected sites.

Discussion

Which Processes Change During Grass Encroachment?

At these 19 coastal dune sites within the Baltic Sea region even medium loads of wet N-deposition (5–8 kg N ha−1 y−1) have an impact on soil and vegetation factors. At these medium N-loads factors are changing significantly only at acid sites (parent sand pH < 6.0) not at slightly calcareous sites (parent sand pH > 6.0). The pH of the parent sand material is a main factor responsible for the reaction of the system to these low to medium N-loads. Although pH-differences are small between acid and slightly calcareous sites, they determine which buffer range is acting and how quickly the buffering capacity of cations is depleted, if toxic levels of soluble metals can occur or if mineralization processes are hampered.

The acidity of the soil has a major influence on soil processes. Some field sites show the expected pattern of acidification during vegetation succession in coastal dunes. The pH remains almost constant during the transition from parent sand material to first stage, but decreases with increasing successional age (Ellenberg 1996). In contrast to these are sites where the pH decreases already between the parent sand material and the first succession stage, after which it remains stable. This pH reduction during the early successional stages was not correlated to N deposition, but may instead be caused by anthropogenic sulphur deposition, for example, ship traffic. Sulphur generally generates two times more protons than reduced or oxidized nitrogen (NEGTAP 2001). Pure chemical reactions might prevail in the pure sands at initial successional stages, whereas biological transformations of N such as nitrification may dominate the acidifying processes in later succession stages with higher soil organic matter contents. This is supported by the lower pH values of later successional stages at higher N-deposition in this study.

As soil pH decreases, organic matter (LOI) accumulates exponentially (this paper; Stützer 1998; Scheffer and Schachtschabel 1998). In these dry, oligotrophic systems, organic matter retains moisture and stabilizes the soil micro-climate. Most typical dry dune plant species are adapted to fluctuating, extreme temperatures, drought and frequent soil moisture changes. The more stabilized conditions, which prevail during organic matter accumulation, support the growth of ruderal, nitrophilous species at the cost of typical dry dune species like Jasione montana. Graminoids such as Carex arenaria, Calamagrostis epigejos and Festuca rubra become dominant, where a sufficient layer of organic matter is available or the top sand layer is nutrient rich (Boorman and Van der Maarel 1997). This observation is supported by the current dataset. At acid sites, the total standing vascular plant biomass and the cover of C. arenaria are 1.5–3 times higher at N-affected than at N-unaffected sites.

At older successional stages, the higher N-content and N/P-ratio in Carex tissue at N-affected sites is in contrast with the higher C/N-ratio in the soil. Different explanations are possible. One is that surplus mineralized N is taken up directly by Carex. In addition, in acid soils more N might be available for vascular plants in competition with microbes as microbial communities seem to have a lower demand for N (Kooijman and Besse 2002; Kooijman and others 2008). Another possible explanation is, that NO3 −, which is not taken up, is readily leached. Furthermore, the immobilized N can be bound in structures with higher C/N-ratios like fungi in these acid soils, which might then result in a higher C/N-ratio for the total soil.

Although nitrification is hampered at a pH below 4.2 (Aerts and Bobbink 1999; Roelofs and others 1985), the higher amount of organic matter at acid, N-affected sites nevertheless increases N-mineralization to the same level as at slightly calcareous sites (Table 3). At acid sites, microbes have a lower demand for N, and therefore the net availability of N for vascular plants, for example, Carex, seems to be higher.

Al/Ca-ratios as well as available Fe concentrations are higher at N-affected, acid sites compared to N-unaffected sites. Carex which is more abundant at these sites seems not to be hampered by these high metal concentrations, though metal toxicity might inhibit the growth of many typical plants of dry, lichen-rich dune grasslands. In addition, Carex can grow taller and has greater growth plasticity (Tietema 1981) than typical dry dune grassland species, and can thus dominate the system.

A N/P-ratio of 7–12 indicates that Carex is far from being P-limited and still N-limited (Koerselman and Meuleman 1996). This is supported by the total tissue concentration of P in Carex. P-limitation and critical values depend on plant species and plant tissue part, but a tissue concentration of 50–60 μmol P g−1 dw, which is the range for Carex at these 19 sites, is critical but not severely limiting (Troelstra and others 2001).

Carex arenaria is therefore assumed to find optimal growth conditions in Baltic dune sites with high N deposition as it is not hampered by low pH conditions, but is favored by increased availability of its limiting nutrient, nitrogen. With its larger structure, C. arenaria can probably comb out more atmospheric N via its leaves as Ammophila arenaria does (Heil and others 1988). The N-content and N/P-ratio of Carex increase with rising N-deposition mainly at older successional stages, but no correlation was apparent between different soil N-forms and Carex tissue N. At older successional stages, atmospheric N may therefore contribute more to the nutrition of Carex than soil-derived N. With its extensive rhizome network, C. arenaria can exploit and transport temporally and spatially widespread resources nearly all year round (Noble and Marshall 1983; D’Hertenfeld and Falkengren-Grerup 2002; D’Hertenfeld and Jonsdottir 1999) and thereby efficiently use the surplus nitrogen supply. C. arenaria finally wins the competition under higher N-loads. Organic matter content above 1.0–1.5% and pH below 4.0 might be the threshold conditions for a system shift to a vegetation dominated by Carex. Similar shifts in competitive relationships at increasing N-levels have been reported from heathlands and coniferous forests (Aerts and others 1990; Berendse and Aerts 1984; Heil and Bruggink 1987; Keller and Redbo-Torstensson 1995). Another possible reason for grass encroachment is lack of (rabbit) grazing (Veer and Kooijman 1997), but rabbits have not been important grazers according to local site managers and nature conservationists within the 19 dune sites around the Baltic Sea during the last one to two decades, and no other herbivores like elk, red deer, roe deer or hare have been reported to have a major impact on these dry grassland dune systems. Thus, grazing can be excluded as an important factor for grass encroachment by Carex arenaria in Baltic dunes.

Character of Carex-Dominated Vegetation Units

Dominance by C. arenaria can vary considerably. Such vegetation units sometimes are species-rich and open (Harilaid or Pavilosta), but sometimes a species-poor, thick grass sward (Korshage or Gellen). The species-rich, more open Carex-dominated vegetation is a community known from Baltic dunes since the 1920s (Steffen 1931; Hueck 1932; Paul 1953). At the Curonian Spit, a slightly calcareous and N-unaffected site today, Paul (1953) recorded during the late 1930s an average of 12 species in the optimal phase of a C. arenaria community (Cladonia spp. excluded). This number is still the same today, and, quite remarkably, species assemblages have not changed during the last seven decades. A decreased number of species (9) at slightly calcareous, N-affected Baltic dunes is comparable to Dutch dunes (Veer and Kooijman 1997).

At higher N loadings, Carex dominance is characterized by a dense grass sward. With increasing biomass the light availability decreases, and a dense root network is established (Veer and Kooijman 1997). These types of Carex units are probably quite persistent. Once a thick grass sward is set up, it is difficult for other plants and even trees to germinate and establish. Most of the few species that are still occurring in this vegetation unit (three on average) are relics of former, more open stages.

The Effects of N-Deposition on Species Richness

In acid dune systems, even N-loads below 8 kg N ha−1 y−1 can have a negative impact, while slightly calcareous sites are still well enough buffered by the carbonate system (Scheffer and Schachtschabel 1998; Blum 2007), and thus no acidifying impact of this moderate atmospheric N-deposition is occurring. As the three factors soil Al/Ca-ratio, N-deposition and soil pH are strongly intercorrelated, their impact cannot be separated totally, but their importance can be ranked. Kleijn and others (2008) could not observe any systematic difference between Al/Ca-ratio at growth sites of common and rare species, but rare species occurred at a restricted pH-range. In this study, Al/Ca-ratio and N-deposition (measured as lichen N), showed the largest impact on dry coastal dunes.

Foliose lichens, which have their thalli directly flat on the ground, disappear along with two to three forb and grass species. Cladonia glauca and Festuca polesica do not grow at all at N-affected sites. Also in other habitats, coverage and species richness of lichens and mosses decrease at higher atmospheric deposition (Boorman and Fuller 1982; Heil and Diemont 1983; Ketner-Oostra and Sykora 2004; Lee and Caporn 1998; Van Tooren and others 1990). At ambient or artificial total N-deposition of 10 kg N ha−1 y−1 total plant species diversity was reduced in American prairie grasslands (Clark and Tilman 2008), Swedish deciduous forests (Falkengren-Grerup and Diekmann 2003) and British grasslands (Stevens and others 2004, 2006). Particular to Baltic dunes, there seems to be a decrease in foliose lichens even at moderate deposition loads of less than 8 kg N ha−1 y−1 wet deposition, whereas shrub-thalli lichens do not change their coverage under these conditions. At N-affected field sites, soil pH is lower and Al and Fe are more available than under pristine conditions, and hence, lichen species not adapted to these high concentrations might take up more metals, which may affect their vitality (Hauk and others 2002, 2007).

Recommendations for Coastal Dune Management

Ecological thresholds and early indicators for ecosystem changes are well used tools in nature conservation and management, though research of their practical application to various ecosystems is still needed (Groffman and others 2006). This study elucidates the impact of low to medium loads of N deposition on coastal dunes. An early indicator for a dune system change due to N-loads of 5–8 kg N ha−1 y−1 may be an increased drop in soil pH between freshly deposited sand (parent material) and the first successional stage. If the total organic matter content (LOI) in the upper mineral soil horizon of lichen-rich, short grasslands (second stage) is above 1–1.5% and the pH is below 4.0–4.5, the system is about to change. Heavy metals become freely available, for example, Al/Ca-ratios above 1 occur and short grasslands are no longer suitable for slow-growing lichen species. In particular Cladonia species growing with their thalli flat on the ground decrease their cover. Cladonia glauca and Cl. macilenta agg. Hoffm. might be good indicator species of an intact system; higher coverages of Rumex acetosella or Dicranum scoparium in short grasslands indicate the opposite. Slightly calcareous sites are only marginally affected by these relatively low N-loads as the calcium buffer capacity is not completely depleted yet.

Stands of C. arenaria are a natural part of the Baltic coastal ecosystem (Steffen 1931; Hueck 1932; Paul 1953). If the coverage of C. arenaria per field site does not exceed 30–40% (Remke and others 2009), these stands can be species rich (12–16 species per 4 m2 plot). However, C. arenaria stands have become more widespread and more dense. At many sites, a dense, species poor (4–6 species per 4 m2 plot) grass sward has established over about 60–70% of the area (Remke and others 2009). Because soil conditions have been irrevocably altered, restoration of dune grasslands on such sites should include turf stripping, removal of soil enriched with organic matter (Ah-layer) and addition of low doses of lime (Dorland and others 2004; Symes and Day 2003).