Conservation and restoration of Nardus grasslands in the Swiss northern Alps

Aim: Species‐rich Nardus grasslands are high nature‐value habitats. In Switzerland, many of these grasslands are degraded even though they have been under protec‐ tion since the 1980s. Degradation shows two divergent trends: Nardus grasslands are either dominated by Nardus stricta or by eutrophic plants, both trends leading to the disappearance of typical Nardus grassland species. With this study, we aim to identify the factors that could be adjusted to conserve the integrity of this habitat. Location: Bernese Alps, Switzerland. Methods: In 2016, we investigated the underlying causes of this degradation process by assessing vegetation composition in 48 Nardus grasslands located in the Swiss northern Alps of canton Bern and

In the Bernese Alps, many species-rich Nardus grasslands are protected through environmental contracts to ensure extensive management practices such as late mowing or low-intensity grazing, and the prohibition of fertiliser and herbicide applications (Regierungsrat des Kantons Bern, 2001). Despite the management regulations, in 2011-2014 the authorities of the canton of Bern became aware that many species-rich Nardus grasslands were degraded (Table 1). The observed degradation trend was dichotomous (Appendix S1). First, the cover of N. stricta strongly increased in some grasslands, which triggered a decrease in the total number of plant species. One mechanism leading to N. stricta dominance (its cover can reach up to 80-90%) is the selective grazing by livestock. They avoid N. stricta due to the high abrasiveness and the poor digestibility of the plant leaves (Armstrong, Common, & Davies, 1989, 1986Maag, Nösberger, & Lüscher, 2001;Massey, Ennos, & Hartley, 2007). The dense root system of this grass increases its dominance by inhibiting colonisation by other plant species (Fischer & Wipf, 2002). Second, eutrophication leads to the disappearance of N. stricta and other typical Nardus grassland species in favour of more common plant species better adapted to enriched soil conditions.
In the past, research focussed mainly on how to convert Nardus grasslands into more productive meadows and pastures with higher agricultural value (e.g., Dietl, , 1998;Perkins, 1968). More recently, with the increased awareness about the loss of this valuable habitat, some ecological studies have shown that inadequate management such as over-or undergrazing can cause severe species declines (Fischer & Wipf, 2002;Rudmann-Maurer, Weyand, Fischer, & Stocklin, 2008;Stevanovic et al., 2008). Nevertheless, there is only little knowledge about how it is possible to maintain the biological integrity of Nardus grasslands with its share of typical Nardus grassland species. Moreover, knowledge about the response of the invertebrates occurring along the two mentioned Nardus grassland degradation trends is lacking. This is surprising given that Nardus grasslands are inhabited by many rare and endangered invertebrate species (Galvànek & Janàk, 2008).
In this study we investigated the influence of soil conditions, local management practices and environmental variables on the degradation (and subsequent declassification) of species-rich Nardus grasslands. This was done by assessing the vegetation composition of 28 degraded and 20 non-degraded Nardus grasslands located in the Alps of the canton of Bern, Switzerland. Ultimately, the aim was to identify the factors that could be adjusted to restore the integrity of the habitat. This question was posed on the online platform "Marktplatz für Forschungsfragen" of the Swiss Biodiversity Forum by the Bernese cantonal authorities. On this platform stakeholders such as policy-makers or private environmental organisations can post scientific questions that are relevant for their work but not scientifically investigated yet (Suhner, Pauli, & Stapfler, 2015). This encourages collaboration between researchers and practitioners to enhance the uptake of effective evidence-based conservation guidance to preserve and restore biodiversity.

| Study area
In 2011-2014, the canton of Bern performed an inventory of all previously known species-rich Nardus grasslands in the Bernese F I G U R E 1 A Nardus stricta turf early in the season when its nutrient level is still high. Picture by Nora Rieder [Colour figure can be viewed at wileyonlinelibrary.com] KURTOGULLARI eT AL.
Alps, comprising 112 pastures (grazed grasslands) and 69 meadows (mown grasslands). During the inventory, these grasslands were categorised as being either still species-rich or degraded. Categorisation as a species-rich Nardus meadow required the presence of eight Nardus grassland indicator species (NGIS, listed in Table 2) within a circular survey plot (diameter 6 m). Meadows having less species were categorised as degraded; the main cause of this degradation, either N. stricta or eutrophic plants dominance, was assessed qualitatively. Some additional non-biological reasons for degradation (e.g., the size of the grassland being too small to be registered as habitat of national importance) were also recorded. For pastures lying above the utilised agricultural area (which has an upper limit defined by the Swiss Federal Office for Agriculture varying between 800 m and 1,700 m a.s.l. in the study region), there was an additional requirement for species-rich status: the cover of N. stricta plus the cover of eutrophic plants (listed in Table 1) should not exceed 50%.
The degradation status of the 181 Nardus grasslands in the Bernese Alps is summarised in Table 1. The study sites were selected among the grasslands of the Bernese inventory and followed a stratified random design. For pastures, the same number of fields in each category (species-rich, dominated by N. stricta or eutrophic plants) and of both elevation categories (high or low elevation separated by the median elevation of 1,620 m) were selected. Among meadows, the cantonal inventory included only eight degraded Nardus meadows, which were all considered in this study. For each degraded meadow, a species-rich meadow in close proximity and within the same elevation belt (high or low) was randomly chosen for comparison. All the chosen grasslands were grouped into six geographical study regions (Diemtigtal, Kandertal, Tschingel, Lenk, Niesen, Zweisimmen; Figure 2). Over the whole study area, the mean annual rainfall was 1,338 mm between 1981 and 2010 (Bundesamt für Meterologie und Klimatologie, 2016). The average size of the studied Nardus grasslands was 2.2 ha (range: 0.3-17 ha) and elevation ranged from 1,000 m to 2,013 m. The minimal distance between two grasslands was 1.25 km. After visiting the selected grasslands, four meadows were excluded from analyses because their management has been abandoned, or the vegetation composition was drastically different from a typical Nardus grassland. Furthermore, one meadow had been converted into a pasture and thus was included as an additional pasture. As a result, 48 (37 pastures and 11 meadows) of the 52 originally selected grasslands were included in this study (Appendix S2).

| Data collection
Vegetation surveys were performed between May and July 2016 on all 48 Nardus grasslands. Within a representative area of the grassland, a 3-m radius sampling plot (28.26 m 2 ) was randomly located. All vascular plants present in the plot were recorded and their absolute cover was estimated. The GPS location, aspect and slope were recorded for each vegetation plot and soil was collected with 10 samples along a transect crossing the plot. The soil was dried overnight in an oven at 105°C, ground and sieved (mesh size <1 mm). Soil (10 g) was mixed with 25 ml of 0.01 molar calcium chloride (CaCl 2 ) solution, and the pH of the suspension was measured after 2 h. We measured the total of organic carbon and nitrogen in 12 mg of homogenised and dried soil samples with a CNS elemental analyser (Vario EL cube, Elementar). The ratio of C to N was calculated as an indicator of nitrogen in its plant-available form (Hodge, Robinson, & Fitter, 2000). To determine the plant-available phosphorus, the phosphate content in the soil was measured with the Olson method (Pansu & Gautheyrou, 2007).
Orthopteran densities (number of individuals per m 2 ) were assessed in August 2016 using a biocenometer built from a net and two plastic rings with a ground area of 1 m 2 . In each grassland, 12 orthopteran density samples were taken along one or two diagonal transects depending on the shape of the meadow. A minimal distance of 10 m was left between the density samples and from the edges of the grassland, (as performed by Humbert, Ghazoul, Richner, & Walter, 2012). Sampling took place only during warm and sunny days. Adult orthopterans were identified to species level while nymphs (only the last larval stage was considered) were classified into their suborder (Caelifera or Ensifera). In order to obtain a more comprehensive list of orthopteran species, two people additionally scanned (visually and acoustically) the grasslands for at least 20 min.
In every second biocenometer sample, the vegetation height was measured using an A4 clear plastic sleeve. The plastic sleeve was dropped from a height of 1 m and the minimal and maximal vertical distance from the edge of the sleeve to the ground were measured.  (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005), 181 meadows and pastures were registered as species-rich Nardus grasslands. In the years 2011-2014 the grasslands were monitored again and severe degradation was observed, although all grasslands were under environmental contract to ensure extensive management. The percentages and reasons of degradation are given in the table.
All farmers managing the 48 grasslands were interviewed to obtain information on the management practices applied in 2016 and also in the past (see Appendix S3 for the detailed list of questions).
Unfortunately, it was not possible to estimate the exact grazing intensity on the Nardus pastures because they were often a subset of a bigger pasture; thus, it is unknown how much time the livestock actually spent within a particular field.

| Data analysis
Due to the different numbers of replicates of pastures (n = 37) and meadows (n = 11), the causes of Nardus grassland degradation and its impact on orthopterans were analysed separately. The response variables of interest were the number of NGIS, N. stricta cover, cover of eutrophic plants and the orthopteran variables (Ensifera, Caelifera and overall species richness as well as density). Strongly rightskewed responses and explanatory variables were log-transformed.

All considered variables and transformations are listed in Appendix
S4. To improve model convergence, the variables were standardised (mean = 0, standard deviation = 1). If two explanatory variables had a Spearman correlation coefficient >0.7, only the biologically more meaningful variable was retained.
The statistical analysis of the pastures was conducted in two steps. First, a pre-selection of the explanatory variables was done: univariate linear mixed-effects models (LMM) were used to iden-  For meadows, no model was selected due to the low number of replicates (n = 11). All the explanatory variables were tested in univariate LMMs to identify any associations (p < 0.1) with one of the seven response variables. The variables with p < 0.1 were ranked according to their absolute estimated effect size; only the three variables with the highest absolute estimate are shown in Table 4. Lastly, we compared number of NGIS, N. stricta cover, eutrophic plant species cover and orthopteran responses between pastures and meadows by using LMMs with region as random factor. The relative number of pastures and meadows that had become degraded among the 181 original Nardus grasslands of the Bernese Alps (Table 1) was also compared using a two-proportions χ 2 test.
All models were checked for equal variance and normality by using Tukey-Anscombe and Q-Q plots. To visualise the data, a redundancy analysis of the response variables and the explanatory variables TA B L E 2 List of the Nardus grassland indicator species (NGIS) and eutrophic plant species found in our study sites  (Eggenberg, Dalang, Dipner, & Mayer, 2001). In this study, the number of NGIS per grassland (within a sampling plot, diameter 6 m) was used as a proxy for Nardus grassland quality. In contrast, a high coverage of the more common and less specialist eutrophic plant species indicates a low quality of the Nardus grasslands (Eggenberg et al., 2001).
resulting from model averaging was performed. All statistical analyses were run in R Studio version 1.1.463 (R Core Team, 2014).

| RE SULTS
Overall, 252 vascular plant species with an average of 48 species

| Pasture analysis
The management practices of 2016 were applied for on average 20   (Table 3). In north-exposed pastures, we found a low species richness (all three variables) and density. In pastures with few NGIS, we found many Ensifera species. Pastures with a higher mean vegetation height harboured more species (mainly Ensifera) and a higher orthopteran density; a higher vegetation structure also promoted greater species richness but this time dominated by Caelifera species. Further, we found no significant effect of eutrophic plants on the orthopteran assembly, whereas N. stricta had a negative effect on orthopteran density (p = 0.038), but no effect on species richness. The redundancy analysis plot shows an overview of the relations between all response and key explanatory variables (Figure 4).

| Meadow analysis
All meadows were mown once a year and the management practices from 2016 had been applied for 39 (±37) years. The mowing event occurred around 5 August. According to the cantonal assessment, seven meadows were degraded and 60 were species-rich. In the studied meadows (n = 11), we found that highly elevated ones with Note: For each response variable (a-g), the explanatory variables retained after model selection and model averaging are given. Unconditional SE is the standard error, which is not conditional on the model, meaning that it is more precise because does not depend on the number of fitted parameters (Burnham & Anderson, 2002). Relative importance (rel. importance) is calculated by summing up all Akaike weights of the models within ΔAIC < 2 where the predictor variable occurs (Burnham & Anderson, 2002). Definitions of all variables are provided in Appendix S4 and all original best models within ΔAICc < 2 can be found as Appendix S5 in the supporting information. Significant effects are highlighted in bold.

KURTOGULLARI eT AL.
low nitrogen deposition and high soil phosphate harboured many NGIS. The cover of eutrophic plant species was high in meadows with a high soil pH, whereas N. stricta was dominant in remote, highly elevated and flat meadows (Table 4). For the orthopterans, the number of species per meadow was on average 6.5 (±2.2) and the density 0.7 (±0.9) individuals per m 2 . Ensifera species were more frequent in meadows with a high soil C:N ratio. The orthopteran density was highest in meadows harbouring few plant species in total but a high number of NGIS and having a low cover of eutrophic plants. Both orthopteran density and Ensifera richness were low if the meadow was freshly cut relative to the sampling day.

| Comparison of pastures and meadows
There were no significant differences in the number of NGIS, eutrophic plants or orthopteran response variables between pastures and meadows. However, there was a slightly higher N. stricta cover in pastures than meadows (Estimate = 11.52, SE = 5.86, p = 0.056).
However, among the 181 initial Nardus grasslands of the Bernese Alps the proportion of pastures that was degraded was significantly higher than those of meadows (χ 2 = 13.75, p < 0.001).

| D ISCUSS I ON
In this study, we investigated the potential drivers of the ongoing degradation of species-rich Nardus grasslands in the Swiss Alps.
Mown Nardus grasslands (meadows) are rarely degraded in compari- subsections, results on pastures and meadows are discussed separately before being compared. Finally, management recommendations are given in the light of our findings.

| Nardus pastures
In this study, the number of Nardus grassland indicator species (NGIS, see Table 2) was used as a proxy for Nardus grassland quality.
These indicator species are specialised on Nardus grasslands and can only be found abundantly if neither of the two degradation trends (namely eutrophication or dominance of N. stricta) is present.
We found that the vegetation composition differs according to the grazing management: on summer pastures, which are grazed once over the whole summer, we found more NGIS than in spring and autumn pastures that are grazed twice a year. Accordingly, eutrophic plants were more frequent in spring and autumn pastures than in summer pastures. One factor explaining this relationship is elevation, meaning that summer pastures with many NGIS are at high elevations whereas spring and autumn pastures with high cover of eutrophic plants are located at lower elevations. This is in accordance with observations by Roth, Kohli, Rihm, Amrhein, and Achermann (2015) who found that in Switzerland the maximum number of oligotrophic species lies at higher elevation than the maximum total species richness. Atmospheric nitrogen deposition was found to be higher at low elevations, which may increase the competition pressure on NGIS by eutrophic plants. Next to elevation, the number of grazing days differs among grazing management types. On summer pastures, the cattle spend on average more days than on pastures grazed in spring and autumn. Although we miss a direct measure of grazing intensity, we suspect grazing is more intensive within the generally smaller spring and autumn pastures than in the larger summer pastures as the number of livestock units per area is higher, though for a shorter time. The relationship between the area of the grasslands and the number of grazing days cannot be derived in this study, as the grazed areas are often (especially in the summer pastures) much larger than the area of the Nardus grassland itself. NGIS occur more often in the large and lowintensity summer pastures, whereas eutrophic plants occur more frequently on small high-intensity pastures. This is in line with literature stating that intensive grazing increases dung deposition and lowers the C:N ratio in the soil, which is the main reason for the shift towards an eutrophic plant community (Lezama & Paruelo, 2016;Parolo, Abeli, Gusmeroli, & Rossi, 2011). In addition, we found that The second Nardus grassland degradation trend towards N. stricta dominance is most frequent in pastures with one grazing period (typical summer pastures). In these low-intensity and longterm-grazed summer pastures, selective grazing by livestock is presumably high because cattle prefer feeding on plants with higher nutrient content and simultaneously avoids the unpalatable grass N. stricta (Armstrong, Common, & Davies, 1989;Grant, Torvell, Sim, Small, & Armstrong, 1996;Parolo et al., 2011). Selective grazing consequently enhances the abundance of unpalatable species on certain parts of the pastures (Adler, Raff, & Lauenroth, 2001;Gusewell, Jewell, & Edwards, 2005). The late start of grazing in the summer pastures impairs forage quality of N. stricta even more (Bovolenta, Spanghero, Dovier, Orlandi, & Clementel, 2008;Meisser et al., 2014). Furthermore, north-exposed pastures are highly dominated by N. stricta because they are more acid than south-exposed ones.
Additionally, N. stricta is able to tolerate the oscillations in humidity on north-exposed pastures (Egli et al., 2006;Lauber, Wagner, & Gygax, 2012) which could be a further advantage in competing with other plants. Interestingly, the C:N ratio was not retained in the best model of N. stricta cover and also does not significantly correlate with it. Weigelt, Bol, and Bardgett (2005) found that N. stricta takes up rather high amounts of nitrogen despite its low productivity.
For this reason, fertilising pastures to get rid of N. stricta as is proposed by many farmers might be more detrimental for NGIS than for N. stricta itself (Hegg, Feller, Dähler, & Scherrer, 1992).
For orthopteran species richness, vegetation structure was found to be more important than vegetation height alone. As a high vegetation structure indicates a high structural heterogeneity, this finding can be explained by the habitat heterogeneity hypothesis F I G U R E 3 Differences in the vegetation according to grazing management. Nardus stricta and NGIS (Nardus grassland indicator species, see Table 2) are more frequent on summer pastures whereas on spring-and autumn-grazed pastures, eutrophic plants are dominant. The number of grazing days also differs between the two categories: summer pastures are grazed long-term, whereas spring and autumn pastures are grazed twice a year for a short period.  (Dennis, Young, & Gordon, 1998), which stipulates that a structurally diverse vegetation provides different habitats for various species (see also Jerrentrup, Wrage-Monnig, Rover, & Isselstein, 2014;Wettstein & Schmid, 1999). The management providing higher structural heterogeneity is low to intermediate grazing intensity (Fabriciusova, Kanuch, & Kristin, 2011). When comparing the species richness of the two orthopteran suborders Ensifera and Caelifera, we observe a stronger dependence on high vegetation for Ensifera than for Caelifera (see also Marini, Fontana, Battisti, & Gaston, 2009b;Willott & Hassall, 1998). A second important variable was the southern aspect, which was crucial for all four orthopteran response variables, since they need a high temperature to perform their physiological activities and egg development (Baur et al., 2006;Pradervand et al., 2013;Sutcliffe, Batary, Becker, Orci, & Leuschner, 2015). Further, we found that the dominance of the tough and nutrient-poor grass N. stricta has a negative impact on orthopteran density. Franzke, Unsicker, Specht, Kohler, and Weisser (2010) observed that Pseudochorthippus parallelus selectively avoids tough grass species, and Isern-Vallerdu and Pedrocchi (1994) as well as Blumer and Diemer (1996) found that certain orthopteran species would feed on N. stricta if they have no alternative. Results also indicate that the number of NGIS has a negative impact on Ensifera species richness.
Since NGIS make up only 4.6% (±6.2%) of the vegetation cover and as the relative importance of that variable in the best model was low (0.19), we do not think that NGIS directly influences orthopterans.
The mechanism is rather indirect via the high elevation and northern aspect of NGIS-rich pastures, both leading to cold and harsh conditions unfavourable for many orthopteran species (Pradervand et al., 2013;Sutcliffe et al., 2015;Wettstein & Schmid, 1999).

| Nardus meadows
Like in Nardus pastures, NGIS were more numerous at higher elevations. Further, the number of NGIS was positively correlated with phosphate concentration in the soil. Although counterintuitive at first sight, because usually plant species richness decreases with increasing phosphate availability (Gilbert, Gowing, & Wallace, 2009), in very acid soils phosphate can hardly be taken up by plants, leading to phosphate accumulation (Kooijman, Dopheide, Sevink, Takken, & Verstraten, 1998). These large phosphate reserves make the meadows rich in NGIS vulnerable to nitrogen deposition (Stevens et al., 2010). As a corollary and similar to pastures, eutrophic plant cover increased with increasing soil pH at the cost of NGIS.
Nardus stricta dominance was most pronounced within flat and remote meadows located at high elevation, which cannot be explained by the management because all meadows were mown once a year between 19 July and 31 August. It is possible that wild ungulates are responsible for this pattern because they are known to have a strong impact on alpine grassland composition, as they feed preferably on nutrient-rich plants (Marchiori, Sturaro, & Ramanzin, 2012).
When looking at the orthopterans, mowing led to direct mortality of many individuals and the disappearance of Ensifera species (Humbert, Ghazoul, Richner, & Walter, 2010). Interestingly, orthopteran density was higher in meadows with a high number of NGIS and with increasing plant species richness, probably because the density was decreased as a result of invading eutrophic plants . This is in line with the result that a high cover of eutrophic plants had a negative effect on the orthopteran density in meadows. Ensifera species richness was also lower in soils with a low C:N ratio, corresponding probably to high nitrogen availability. The mechanism behind the negative impact of eutrophication on the orthopterans might be the colder microclimate and the decreased vegetation structure in the higher and denser vegetation (Marini, Fontana, Battisti, & Gaston, 2009a;Willott & Hassall, 1998).

| Comparison between Nardus pastures and meadows
Although we could not find differences in NGIS and eutrophic plant cover between pastures and meadows, probably due to the  (Fischer & Wipf, 2002;Gustavsson, Lennartsson, & Emanuelsson, 2007). This would be in line with the slightly higher N. stricta covers in pastures than in meadows found in the current study. A potential reason for the higher quality of Nardus meadows compared to pastures is the homogenous vegetation removal, which does not favour the unpalatable N. stricta like selective grazing does (Fischer & Wipf, 2002) and reduces the dominance of eutrophic plants by lowering the soil nutrient content (Kitchen, Blair, & Callaham, 2009).
Regarding orthopterans, grazing might be the best management option because it provides a more heterogenous vegetation structure, which we found to be the most important factor for species richness and density. Furthermore, mowing leads to direct mortality of the orthopterans, which might affect long-term population growth (Buri et al., 2013;Humbert et al., 2010).

| CON CLUS I ON AND MANAG EMENT RECOMMENDATIONS
In this study, Nardus meadows are less often degraded than pastures, and the high cover of N. stricta in pastures may drive this degradation. To maintain species-rich Nardus meadows in the Alps, the current measures applied in the canton of Bern (i.e., no fertilisation and earliest cut after 15 July) should be continued and are recommended for other similar alpine regions. Contrastingly, for Nardus pastures, some management adaptations are needed. Since the quality of the Nardus pastures is better at high elevation on nutrient-poor soil, restoration actions should focus on low-elevated pastures as proposed by Korzeniak (2016). Both the study by Korzeniak (2016) and our study show that at low elevations, degradation through eutrophic plants is predominant and linked to high nutrient availability. In pastures, further range expansion of N. stricta could probably be limited by a cut after the grazing season since this would reduce the ungrazed N. stricta tussocks. However, due to topographic constraints such as steepness or cow stairs, this is often impossible.
Grazing management should thus imitate mowing as far as possible by homogenising grazing pressure and keeping the nutrient input through dung deposition as low as possible. Therefore, we recommend dividing the pastures into small fenced paddocks to better equalise grazing pressure and dung deposition, as was also proposed by Parolo et al. (2011). Specifically, it is important to choose the right grazer type for the reduction of N. stricta, as sheep were shown to be much worse in reducing N. stricta than cattle (Armstrong, Grant, Common, & Beattie, 1997;Sebastia, de Bello, Puig, & Taull, 2008).
Furthermore, grazing intensity should be reduced on low-elevation pastures, by, for example, moving the cattle to the summer pastures (higher elevation) a bit earlier. On these higher-elevated pastures, before grazing the whole area, cattle should first be enclosed in areas dominated by N. stricta when the plant is still palatable. For grasslands where N. stricta is very dominant, creating gaps in the turf for other plant seedlings by rotavation could be a restoration option (Mitchell, Rose, & Palmer, 2009).
By reducing the amount of N. stricta and eutrophic plants with the proposed measures, the orthopteran density would also be promoted. Moreover, future restorations of degraded Nardus grasslands to enhance orthopteran diversity should prioritise south-exposed pastures because the dominance of N. stricta is less pronounced there and these grasslands have the highest potential for a rich orthopteran assemblage. In meadows, we propose to keep an uncut refuge area when haying, as suggested by Buri et al. (2013), to reduce the direct detrimental effect of the mowing machine. An associated management practice is already applied in certain regions of the Swiss Alps: traditionally known as "Eger Mähder," the meadows are divided in two parts and both parts are mown alternately every two years. However, it should be tested whether or not a supra-annual mowing cycle has negative effects on the vegetation of Nardus grasslands. It should be noted that all results of this study are based on observations and further experimental studies should test the applicability of these recommended measures and their effectiveness in conserving and restoring species-rich Nardus grassland.
The study was carried out in the Bernese Alps, which has a similar climate and land-use patterns as the French, Italian, German and Austrian Alps (e.g., Parolo et al., 2011). In addition, the Nardus grassland assemblage is widespread in Europe, where it faces similar threats and has a poor conservation status (European Environment Agency, 2012; Korzeniak, 2016). Hence, these findings have clear implications beyond Switzerland.

ACK N OWLED G EM ENTS
We would like to thank Maiann Suhner, Brigitte Holzer and Bernhard

DATA AVA I L A B I L I T Y S TAT E M E N T
The raw data of the study can be found online within the supporting information of this paper (Appendix S7 to Appendix S10).