Nutrient limitations inferred from elemental stoichiometry
Nardus and chalk grasslands are both considered oligotrophic environments with low levels of available N and P (Smith 1980; Schelfhout et al. 2017). In the present study, the mean ratios of N to P in both grassland types were within the range of 10–20, within which, according to Güsewell (2004), no clear limitation by N or P occurs but a co–limitation by both nutrients is observed. However, according to the criteria of Koerselman and Meuleman (1996), the N:P ratio of the herbage of the chalk grassland was above the critical value of 16, which indicates a relative limitation by P, whereas the average N:P ratio for the Nardus grassland was slightly below the critical value of 14, indicating a limitation by N. Other studies conducted in other European mountain areas (Kirkham 2001; Klaudisová et al. 2009; Busqué and Bedía 2013) yielded mean N:P ratios of 12–14 for N. stricta and Nardus grasslands that are similar to those of the Nardus grassland in the present study and indicate limitation or co–limitation by N. Conversely, Bobbink et al. (1989) reported N:P values of 14–18 and above for Bromion chalk grasslands, which indicate limitation or co–limitation by P.
Moreover, the soil enzyme C:P ratio was < 1 in both the Nardus and chalk grasslands, indicating that more effort was directed at acquiring and cycling P relative to processing C, which suggests a P deficiency in both grassland types (Liu et al. 2020; Lasota et al. 2022). Phosphorus limitation can occur in high-mountain vegetation even with N:P ratios < 10 (Wang et al. 2017) because of the high demand for P, which is used by plants to protect tissues and membranes against the cold (Marschner 2012). Both the Nardus and chalk grasslands are regarded as adapted to P-limited or colimited habitats because of the efficient P uptake and use strategies of their dominant species (Köhler et al. 2001; Van der Krift and Berendse 2002; Hejcman et al. 2014). Consequently, they have been shown to spread with decreasing P (Klaudisová et al. 2009) or increasing N availability (Bobbink et al. 1989; Wilson et al. 1995; Leith et al. 1999; Stevens et al. 2011). The elemental ratios in the present study indicate stronger P versus N limitation in the chalk grassland than in the Nardus grassland. This is also supported by the finding in the previous study by Badía–Villas et al. (2020) of greater 15N fractionation in chalk grasslands compared to Nardus grasslands, which suggests a relative excess of N over other more limiting nutrients in the chalk grassland (Xu et al. 2014).
The N:S ratios of the herbage were below the critical values established by Stevens and Watson (1986), Mahot (2005) and Ryant and Skládanka (2009), which indicates a sufficiency of S in both vegetation types. In contrast, the N:K and P:K ratios were well above the various critical values proposed in other studies (Dampney 1992; Pegtel et al. 1996; Olde Venterik et al. 2003; Lawniczak et al. 2009), which suggests a strong limitation by K in both grassland types.
Nutrient uptake as affected by nutrient availability in soil
According to customary ranges for soil nutrients (Jones 2001; Horneck et al. 2011), Olsen–P and ammonium–N were low, sulfate–S was high and exchangeable K+ was low to moderate in the soils of both grassland types, whereas exchangeable Ca2+ varied from low–moderate in the Nardus grassland to moderate–high in the chalk grassland. Only ammonium–N and exchangeable Ca2+ showed significant differences between the two grassland types.
Ammonium–N was higher in the Nardus grassland, with values similar to those reported by Badía et al. (2008) for other Nardus grasslands in the same region. Higher ammonium can result from more active N cycling (Zhou et al. 2022), as suggested by the correlation between ammonium–N and URE, and from the more acidic conditions in the Nardus grassland, which mitigate volatilization losses (Woodmanse et al. 1981; Cameron et al. 2013).
Calcium was the only element among the nutrients studied that showed a significant correlation between its available concentration in the soil and its levels in the herbage. The concentrations of exchangeable Ca2+ in the soils of both grasslands were equal to or above the range of 1–3 cmolc kg–1 typical for ‘calcicole’ plant species and well above the value of 0.5 cmolc kg–1 below which ‘calcifuge’ species usually occur (Cross and Lambers 2021). Calcareous soils usually exhibit Ca levels that exceed the demand for this element by any plant species but can lead to stress due to a deficiency of other nutrients (Arnesen et al. 2007; Körner, 2021). According to our findings, the high availability of Ca adversely affected P intake, as shown by the decrease in plant P with increasing plant Ca and soil Ca. In the chalk grassland, high Ca resulted in greater limitation by P than by N, as can be inferred from the values of the N:P ratio of the herbage, and led to higher enzyme demand for P relative to N compared to the Nardus grassland (Fig. 6). Conversely, in the Nardus grassland, decalcification somewhat alleviated P limitation, which, according to the N:P ratios, shifted to N and P colimitation or even N limitation at some sites and increased the enzyme demand for N relative to P (Fig. 6). Contrary to other studies on Nardus grasslands (Kirkham, 2001; Klaudisová et al. 2009), the herbage P concentrations were not related to the soil available P, which may occur in calcareous P-limited grasslands because P acquisition depends more on P mobilization by root exudates than on uptake of soluble inorganic P (Klaus et al. 2016).
The results of the RDA (Fig. 7) also suggest that the enzyme demand for S relative to C, N and P may increase as the enzyme demand for P versus N decreases. Enhanced productivity resulting from sufficient P availability might explain a stronger demand for S, which is consistent with the observations that the levels of plant S increase with increasing plant P and decreasing Ca (Fig. 2). The plant dynamics from chalk to Nardus grasslands were also related to higher enzyme ratios of C to N, P and S. This is likely to be the consequence of the higher concentration of C-rich structural carbohydrates in the Nardus grassland, as suggested by the positive correlation between GLU activity and the concentrations of CEL and HEM, which act as substrates (Wang et al. 2020).
The composition of the herbage indicated a severe K deficiency that cannot be solely explained by soil K availability, which was within the optimal range according to Jones (2001), but appeared instead related to the K+:Ca2+ ratios. Calcium and K are mutually antagonistic in their uptake by plants, so high Ca in the presence of low–moderate K can decrease K intake enough to induce K deficiency (Cahoon and Grummet 1954). The K+:Ca2+ ratio has long been reported to be decisive for the calcicole–calcifuge behavior of plants (De Bilde, 1978; Korcak, 1987) and a major factor influencing the composition of chalk grassland communities (Austin, 1968).
Effect of plant traits on plant composition
The chemical and elemental composition of plants is genetically and physiologically controlled and varies between species and, to a more limited extent, among individuals (Kay et al. 2005; Elser et al. 2010). The results of the present study reveal that the shift between Nardus grassland and chalk grassland involved a shift in plant functional groups with contrasting compositional traits. The two grasslands differed mainly in the relative proportions of graminoids, which were more abundant in the Nardus grassland mainly due to the dominance of N. stricta, relative to forbs and legumes, which were predominant in the chalk grassland. The greater proportion of graminoids affected the organic composition of herbage, which became richer in cellulose and hemicellulose and poorer in lignin, as is characteristic of graminoids compared to nongraminoid herbs (Gordon, 1989; Marinas and García–González 2006; Poca et al. 2014). A larger amount of structural carbohydrates reflects a greater investment in grass stems (lignin was a minor constituent in both grasslands), indicating higher density and stronger competition for light in the Nardus grassland than in the chalk grassland, which would be consistent with the less severe nutrient limitation (Irving, 2015; Postman et al. 2020; Rehling et al. 2021). In fact, Nardus grasslands tend to be denser, and the plants tend to be taller than those in chalk grasslands, which could effectively mean more competition for light.
The greater proportion of graminoids in the Nardus grassland was also related to higher P and lower Ca concentrations, which agrees with the finding by Marinas and García–González (2006) of higher P and lower Ca levels in graminoids compared to non–graminoids in subalpine Pyrenean grasslands. A higher investment in stem structures can affect the plant elemental composition since it means less investment in leaf mesophyll and epidermis, which contain most of the N, P and K in plants (Meerts, 1997). However, in our study, the greater proportion of graminoids and higher concentration of structural carbohydrates were related to higher P and unrelated to N or K. Furthermore, the higher levels of P result in lower N:P ratios in the Nardus grassland compared to chalk grassland, which contradicts the usual finding that a higher proportion of graminoids is associated with higher N:P ratios and, consequently, with an increasing limitation of P relative to N (Güsewell, 2004; He et al. 2006). On the other hand, forbs and legumes are generally reported to show higher Ca concentrations than graminoids (Meerts, 1997; Mládková et al. 2018; Kajzrová et al. 2022), which relates to the large amount of Ca in the form of Ca–pectates in the cell walls of dicotyledons (including legumes and many forbs) (White and Broadley 2003; Mládková et al. 2018).
The plant functional type also influenced the N levels, which increased with increasing amounts of legumes. Legumes are important N fixers, which results in the luxury consumption of N (Freschet et al. 2017) and thus in higher N concentrations and higher ratios of N relative to other elements compared to nonlegumes (He et al. 2007; Di Palo and Fornara 2017). However, the N concentrations were similar in the two grasslands (and the N:P ratios were higher in the Nardus grassland) presumably because legumes were minor populations in both plant communities (albeit more abundant in the chalk grassland). The soil C:N ratio was higher in the Nardus grassland than in the chalk grassland, but this can likely be explained by the fact that a lower concentration of structural components leads to faster decomposition of organic matter, resulting in lower C:N (Badea et al. 2020). Faster decomposition can also result in lighter δ13C values in the chalk grassland (Badea et al. 2020), a finding observed in a previous study by Badía–Villas et al. (2020).
Topographical controls on elemental stoichiometry
The small-scale coexistence of Nardus grasslands and chalk grasslands was related to contrasting elemental composition and demand, particularly of Ca and P. Heterogeneous elemental composition is a key factor in the coexistence of plant species (He et al. 2008; Hong et al. 2015) and communities (Yan et al. 2019; Lin et al. 2022) in high mountain areas. Differences in elemental composition mirror different nutrient requirements, resulting in niche separation (Sardans and Peñuelas 2014), and the high spatial variability in mountain soils provides a diversity of niche spaces for plants with different requirements (Antonelli et al. 2018). The heterogeneity of mountain soils is largely contributed by microtopography (Hiller and Müterthies 2005; Holtmeier and Broll 2018), which has been shown to have a considerable effect on the nutrient distribution in calcareous ranges (Sebastiá 2004; Michalet et al. 2002; Giaccone et al. 2019) and is even more influential than the calcareous versus siliceous nature of bedrock (Michalet et al. 2002). Distinguishing topographic from geochemical effects is critical for understanding how bedrock geology influences the biodiversity of high mountain ecosystems (Rahbek et al. 2019).
In the present study, the differences in elemental composition between Nardus and chalk grasslands were derived from the calcareous substrate and limitations resulting from excess Ca but were controlled by the local topography. Erosion in mountain areas is usually uneven due to the complexity of the topography, with local areas of both soil thinning and thickening occurring (Amundson et al. 2015). In the study area, the two grassland types settle on different sediment levels separated by a centimeter–scale unevenness produced by gully erosion, which was likely triggered by the removal of the original forest for pasture in the past (Badía-Villas et al. 2021). Soil thinning brings the bedrock closer to the soil surface, which in calcareous ranges leads to increased levels of lime and pH (Tovar et al., 2012; Liu et al. 2018), whereas thickened soils are more prone to leaching and debasification (Amundson et al. 2005). In the study area, higher Ca results in increased P limitation in the eroded soils, causing them to be colonized by the chalk grasslands, which can be regarded as a typical calcicole adapted to (but likely not demanding) extreme Ca contents. In contrast, Nardus grassland, which is not typically calcifuge but is still intolerant to excess Ca, grows in noneroded soils where P limitation is gradually alleviated by debasification. The lower biomass of the chalk grasslands allows for further soil erosion (Liu et al. 2018) as opposed to the greater stability of the soils of the Nardus grassland, which contributes to stabilizing the mosaic of communities by positive feedback mechanisms (Armas–Herrera et al. 2020). To our knowledge, reports on Ca–induced P limitation being driven by local topography are limited for upland areas, but it is a well-known phenomenon for calcareous fens (e.g., Boyer and Wheeler 1989; Boeye et al. 1997), where P limitation and plant communities are patchily distributed due to fluctuations in the base-rich water table due to microtopography.