SITE DESCRIPTION

The research was conducted at the Teberda Biosphere Reserve (Northwestern Caucasus, Russia). The study site was located at Mount Malaya Khatipara (43°27′N, 41°42′E) at 2750 m a.s.l. in an alpine lichen heath, which has been the subject of diverse ecological studies during more than 20 years (Onipchenko, 2004). The climate of the area is characterized by low air temperatures (mean annual temperature is -1.2 °C, mean July temperature is 7.9 °C) and high annual precipitation (1400 mm). The lichen heath covers snow-free wind-exposed ridges and upper slopes in the alpine landscape. The plant community is dominated by fruticose lichens (mainly Cetraria islandica), which cover about 30%-60% of the area. Vascular plants are represented by more than 40 species. The most common are Anemone speciosa, Antennaria dioica, Trifolium polyphyllum, Festuca ovina, Carex sempervirens, Carex umbrosa, Campanula tridentata, and Vaccinium vitisidaea. Detailed description of the site plant community, including species' nomenclature, is published elsewhere (Onipchenko, 2002).

STUDIED SPECIES

Twenty-five species from eight functional/mycorrhizal groups (three species per group, except legumes which were represented by three nodulated and one non-nodulated [Trifolium polyphyllum] species) were studied. The different functional groups with respect to nutrient uptake strategy were represented by ericoid mycorrhizal species (ERI), ectomycorrhizal species (ECT), arbuscular mycorrhizal forbs (AM—FORB), arbuscular mycorrhizal grasses (AM—GRA), arbuscular mycorrhizal nodulated legumes (FAB—N), non-mycorrhizal graminoids (sedges and rushes) (NOM—GRA), non-mycorrhizal hemiparasites (NOM—SP), and orchids (ORC) (Table 1). Type of mycorrhiza and mycorrhizal infection rate for arbuscular mycorrhizal species (proportion of thin root mesoderm cells occupied by fungi) were studied earlier (Onipchenko and Zobel, 2000; Onipchenko, 2004).

Foliage from these plant species was sampled during one week at the beginning of August in five replications from a sampling site of 30 × 30 m in size. In general, leaves from five separate plants per species were collected, although in some cases several individuals of a species were pooled to make one replicate sample.

CHEMICAL ANALYSES

Total N was determined by dry combustion on an Elementar Vario EL elemental analyzer. Natural ¹⁵N abundance of plant samples was determined by dry combustion on a Carlo Erba NC 2500 elemental analyzer coupled with a Delta^plus continuous-flow isotope ratio mass spectrometer (Thermo Finnigan, Bremen, Germany) and expressed as δvalues, which were defined as:

STATISTICS

Nested design ANOVAs in general linear models were applied to estimate differences in nitrogen content and δ¹⁵N for plant species and functional types, with three species nested in each functional type (and excluding the non-nodulated legume). In case of significant effects, post hoc Tukey HSD test was used for comparison of means between individual species or functional groups. The calculations were made in Statistica 6.0 for Windows.

To compare nitrogen content and δ¹⁵N between three nodulated legumes and the non-nodulated legume Trifolium polyphyllum, one-way ANOVA was used. Correlation coefficients were calculated to test for a relationship between mycorrhizal infection rate and nitrogen content or δ¹⁵N for AM forbs and grasses.

NITROGEN CONCENTRATION

The N concentration varied significantly and widely among alpine plant species and functional groups of lichen heath (Tables 2 and 3, Fig. 1). Festuca ovina (AM-GRA) had the lowest N concentration (0.97%), whereas Astragalus levieri (FAB-N) and Traunsteinera globosa (ORC) were the most N-rich (3.30% and 3.34%, respectively). The groups of N₂-fixing legumes, orchids, arbuscular mycorrhizal forbs, and ectomycorrhizal species generally had higher N concentrations than other groups. Non-mycorrhizal hemiparasites and ericoid mycorrhizal species tended to have a somewhat higher N concentration among the four groups with relatively low N concentrations. Graminoids, irrespective of their mycorrhizal status, had the lowest N leaf concentration. So, we may represent the following ranking of the leaf N concentration among alpine heath species:

TABLE 1

Mycorrhizal and functional types of the studied plant species.

TABLE 2

Results of nested ANOVA (species are nested in functional type) for nitrogen concentration and δ¹⁵N.

TABLE 3

Leaf N concentration and δ¹⁵N for studied alpine species from different functional types (see Table 1 for abbreviations).

FIGURE 1.

Leaf N concentration for different functional types of alpine plants (mean and standard error) (see Table 1 for abbreviations). Significant (p < 0.05) differences are shown by different letters.

FIGURE 2.

Leaf N concentration and δ¹⁵N for 3 nitrogen fixing and one non-fixing (Trifolium polyphyllum) legumes in the alpine lichen heath. Significant (p < 0.05) differences are shown by different letters.

However, within separate functional groups the variation was also high, and in each group there were species differing significantly in N concentration from each other. For example, ectomycorrhizal Polygonum viviparum had more than twice the N concentration (3.12%) of another ectomycorrhizal species-Juniperus communis (1.48%). Within other groups, pronounced differences were also observed (Table 3). A separate one-way ANOVA for legumes indicated that leaf N concentration in non-fixing Trifolium polyphyllum was significantly lower than the N concentrations for other legumes (Fig. 2).

FIGURE 3.

Relationship between N leaf concentration and rate of mycorrhizal infection (%) for 6 arbuscular mycorrhizal (AM) plant species.

There was a highly significant positive correlation (r² = 0.70) for six AM species (forbs and grasses) between N leaf concentration and rate (degree) of AM infection (Fig. 3). The same tendency is apparent within three species of grasses, but not within forbs.

¹⁵N ABUNDANCE

Natural abundance of ¹⁵N in the plant leaves also varied widely among species within the lichen heath community, ranging from -5.5‰ in Bromus variegatus (AM-GRA) to +2.3‰ Traunsteinera globosa (ORC). All orchid species had about zero or positive δ¹⁵N; their mean of +0.8‰ was significantly higher than that in any other group, excluding the marginally ¹⁵N-depleted non-mycorrhizal graminoids (sedges and rushes) and nodulated legumes (Table 3, Fig. 4). In contrast, there was no single most ¹⁵N-depleted group. Pronounced ¹⁵N depletion was typical for arbuscular mycorrhizal forbs and grasses, non-mycorrhizal hemiparasites, and ericoid mycorrhizal species, with δ¹⁵N values ranging from -3.8‰ to -2.9‰ and not significantly differing from each other. Ectomycorrhizal species were intermediate. We can summarize these results in the following ranking of leaf ¹⁵N enrichment among alpine heath species:

Similar to N concentration, δ¹⁵N varied widely within separate functional groups, often by more than 2‰. For example, Helictotrichon versicolor and Bromus variegatus (AM-GRA species) has δ¹⁵N values of -2.2‰ and -5.5‰, respectively (Table 3). Within all legumes, non-nodulated Trifolium polyphyllum had the lowest δ¹⁵N (-2.6‰), which was close to the values for the ERI and ECT mycorrhizal groups (Fig. 4). This value was significantly lower than δ¹⁵N of two other legume species-Oxytropis kubanensis and Astragalus levieri-but there was no significant difference between δ¹⁵N for Trifolium polyphyllum and Anthyllis vulneraria (Fig. 2).

In contrast to N concentration, δ¹⁵N did not show any relationships with mycorrhizal infection rate for six AM species.

Across all samples, foliar δ¹⁵N increased with increasing N concentrations (r² = 0.22, P < 0.001) (Fig. 5, part a). The group of non-mycorrhizal sedges, being strongly N-depleted and relatively ¹⁵N-enriched, was the main functional group that decreased the strength of this correlation. A highly significant correlation (r² = 0.83, P < 0.001) between N concentration and δ¹⁵N was observed within the group of legume species (Fig. 5, part b).

FIGURE 4.

δ¹⁵N for different functional types of alpine plants (mean and standard error) (see Table 1 for abbreviations). Significant (p < 0.05) differences are shown by different letters.

FIGURE 5.

Relationships between N concentration and δ¹⁵N in alpine plants: (a) all species; (b) legume species.

NITROGEN CONCENTRATION IN ALPINE PLANTS

Leaf N concentrations in our study for all but four of the studied species were lower than the mean values for alpine plants in the Alps (2.87%) and the northern Scandes (3.18%) (Körner, 1989). The mean value for the Caucasian species was 2.11%, which is lower than the N concentration in leaves of plant species even from low altitude in the Alps (2.40%) (Körner, 1989). Monson et al. (2001) reported mean foliar N concentration in alpine forbs in Niwot Ridge (Colorado) of about 3.1%, which is higher than our data for AM forbs (2.4%). Relatively low N concentrations in the species of the lichen heath may be explained by a general tendency of decreasing N concentration in alpine plants with decrease of latitude (Körner, 2003). Also, nitrogen availability in lichen heath is the lowest among the plant communities in the same region (Makarov et al., 2003), as evidenced by a doubling of productivity after N fertilization (Soudzilovskaia et al., 2005).

Hemiparasitic plants usually have higher N and other nutrient concentrations than their hosts (Press et al., 1999). Our results for three Scrophulariaceae species (annual Euphrasia ossica and perennial Pedicularis comosa and P. caucasica) showed that N leaf concentration was lower than those values for AM-FORB but higher than for NOM-GRA (both groups can serve as hosts for studied hemiparasites [Popova, 1966]). We can suppose that the results indicate relatively low or facultative root parasitic activity of the studied species.

Nitrogen concentration in leaves of Trifolium polyphyllum was significantly lower (2.05%) than in other studied legumes (2.64%-3.30%). This result confirms our visual observations of the absence of nodules in the root system of Trifolium polyphyllum. The non-nitrogen fixing status of Trifolium polyphyllum was also confirmed by its positive responses to N fertilization (Soudzilovskaia et al., 2005; Soudzilovskaia and Onipchenko, 2005; Soudzilovskaia et al., 2012) and by the acetylene reduction method (Makarov et al., 2011). Trifolium polyphyllum is therefore the first known temperate non-fixing legume besides previously reported tropical examples (Sprent, 2005). The N concentrations of the three other legume species that do have root nodules correspond to those reported from high mountain ecosystems in Colorado and the Alps (Bowman et al., 1996; Jacot et al., 2000), confirming the adaptiveness of this strategy at high altitude.

Arbuscular mycorrhizal fungi are known to provide plant access to organic soil phosphates and improve mainly phosphate acquisition of plants (Cavagnaro et al., 2003; Smith and Read, 2008). Nitrogen capture from organic material and transferring N to plants via AM is a recently discovered phenomenon (Nakano et al., 2001; Jin et al., 2005; Leigh et al., 2009). However, to our knowledge there is no example in the literature of positive relationships between rate of AM infection and leaf N concentration. For example, the alpine species Ranunculus adoneus had no relationship between AM root colonization and leaf N concentration (Monson et al., 2001). In spite of the limited number of AM species in the present research, the strong positive correlation between these parameters (Fig. 3) is consistent with AM fungi influencing plant nitrogen uptake as well.

¹⁵N NATURAL ABUNDANCE IN ALPINE PLANTS

The δ¹⁵N variation among plant species could be due to several factors. Many surveys highlighted the influence of the mycorrhizal status of species (Schulze et al., 1994; Michelsen et al., 1996, 1998; Nadelhoffer et al., 1996; Högberg et al., 1999; Hobbie et al., 2000; Emmerton et al., 2001; Hobbie and Colpaert, 2003), while other factors (species-specific differences in isotopic fractionation, differences in plant preference for N forms and root placement at different soil depths) could also be important (Schulze et al., 1994; Emmerton et al., 2001; Miller and Bowman, 2002).

Relatively high ¹⁵N abundances of plant leaves of ORC, FAB-N, and NOM-GRA species from an alpine lichen heath correspond to previously published results. For instance, the highest δ¹⁵N values of orchids were also observed in forest and some grassland sites in Bavaria and southern France and were explained by utilization of fungi-derived ¹⁵N-enriched nitrogen (Gebauer and Meyer, 2003).

N₂-fixing legumes that partly acquire atmospheric N₂ have higher δ¹⁵N (close to naught) relative to non-fixing species growing in N-limited conditions of tundra and alpine ecosystems (Bowman et al., 1996; Michelsen et al., 1996; Nadelhoffer et al., 1996; Körner, 2003; Hobbie et al., 2005). This difference in ¹⁵N signature between N₂-fixing and non-fixing species is often used to assess the degree of N₂ fixation by symbiotic bacteria using the dual-sources model (Shearer and Kohl, 1986). Although this method requires care in selecting non-fixing reference plants, which ideally should be characterized by the same preference of soil N and the same temporal and spatial N uptake pattern as N-fixing species, it was successfully applied for alpine ecosystems (Bowman et al., 1996). The studied alpine lichen heath provides a unique test case for revealing N₂-fixation with the δ¹⁵N method. The non-nodulated legume Trifolium polyphyllum provides a taxonomically close reference species for N₂-fixing legumes. The lowest N concentration and δ¹⁵N in Trifolium polyphyllum leaves (Fig. 2) and highly significant correlation between N concentration and δ¹⁵N for all legume species (Fig. 5, part b) allow us to conclude that the ¹⁵N natural abundance method can provide an acceptable estimation of N₂-fixation. Applying the dual-source model and the result of Bowman et al. (1996) that ¹⁵N enrichment of the legume grown solely with atmospheric N₂ is close to 0‰, we calculated that the percentage of plant N that is fixed from the atmosphere varied from 27% in Anthyllis vulneraria to 69% and 73% for Oxytropis kubanensis and Astragalus levieri, respectively (Makarov et al., 2011). The latter results are in close correspondence with the large proportions of atmospherically fixed N₂ (from 70% up to 100%) to satisfy N requirements of legumes in other mountains (Bowman et al., 1996; Jacot et al., 2000).

The ¹⁵N enrichment of non-mycorrhizal graminoids (sedges and rushes) of alpine lichen heath corresponds to the high ¹⁵N content in plant species from this functional group within two plant communities in northern Swedish Lapland where Carex spp. and Luzula arcuata had δ¹⁵N values between -0.5‰ and +2.5‰ (Michelsen et al., 1996). Non-mycorrhizal graminoids in Alaskan tundra (Eriophorum vaginatum and Carex bigelowii) and in Niwot Ridge alpine tundra (Kobresia myosuroides, Carex rupestris, Luzula spicata) were also among the species with the highest δ¹⁵N (up to +2.3‰) (Nadelhoffer et al., 1996; Miller and Bowman, 2002; Körner, 2003). According to Körner (2003), a probable reason for high ¹⁵N enrichment of Cyperaceae species (up to +3.5‰) could be their access to ¹⁵N-enriched stable forms of organic N. However, the more recent finding of considerable ¹⁵N enrichment of DON (Pörtl et al., 2007) indicated that the reason could also be access to labile forms of organic N. In particular, Raab et al. (1996, 1999) demonstrated that non-mycorrhizal alpine Cyperaceae species had uptake rates for glycine similar to or substantially greater than those for inorganic N.

Isotopic nitrogen fractionation during ammonium and nitrate plant assimilation should essentially complicate the measure of the δ¹⁵N of available N in the soil (Emmerton et al., 2001). However, ¹⁵N discrimination during nitrogen uptake will only be observed when plant nitrogen demand is relatively low, compared with the nitrogen available in the solution, and will probably decrease to negligible under conditions of low N concentration and highly efficient N uptake typical for most natural ecosystems. Hence, in natural N-limited ecosystems the δ¹⁵N of non-mycorrhizal plants is a good approximation of δ¹⁵N of the available N (Högberg et al., 1999; Hobbie et al., 2005). This assumption was confirmed by very similar δ¹⁵N values of non-mycorrhizal plants (between -2‰ and +2‰) (Michelsen et al., 1996) and of NH₄⁺-N (between -1.5‰ and +1.4‰) (Makarov et al., 2008) from the surface soil horizon in tundra ecosystems in northern Sweden. Similar δ¹⁵N were also found for non-mycorrhizal plants, total soil N, ammonium, and nitrate N in the arctic tundra ecosystem in Alaska (Hobbie and Hobbie, 2006).

Conditions of N-limitation are quite applicable to the alpine lichen heath where inorganic N concentrations in the soil solution are typically low at about 0.05 mg L^-1 both N-NH₄⁺ and N-NO₃^- (Makarov et al., 2003). However, in our case, alpine non-mycorrhizal sedges were much more ¹⁵N-enriched (between -0.2‰ and -1.5‰) than the dominant form of inorganic soil N-exchangeable NH₄⁺-N (between -2.6‰ and -5.1‰) (Makarov et al., 2008). It seems likely that the ¹⁵N abundance of alpine non-mycorrhizal plants is the result of uptake of a mixture of N sources.

¹⁵N AND TYPE OF MYCORRHIZA

In spite of the prevalent opinion that there is the great difference between ericoid mycorrhizal and ectomycorrhizal species on the one hand (more negative δ¹⁵N values), and non-mycorrhizal or arbuscular mycorrhizal species on the other hand (less negative δ¹⁵N values) (Schulze et al., 1994; Nadelhoffer et al., 1996; Michelsen et al., 1996, 1998; Körner, 2003; Craine et al., 2009), it was not confirmed in our study in an alpine heath community, where ¹⁵N abundance of ericoid mycorrhizal and especially ectomycorrhizal species tended to be rather higher than that of arbuscular mycorrhizal forbs and grasses (Fig. 4). Some other previous results also did not correspond to this prevalent pattern. For example, the isotopic similarity within ecto- and arbuscular mycorrhizal species in forest sites in Bavaria was explained by utilization of isotopically similar N compounds (Gebauer and Meyer, 2003). Also, Hobbie et al. (2005) demonstrated that some arbuscular mycorrhizal species (Luetkea pectinata) had similar δ¹⁵N values with ecto- and ericoid mycorrhizal plants in the Cascade Mountains of Washington, U.S.A. Last, our study of AM species in subarctic meadow communities in the Khibiny Mountains (NW Russia) and Abisko region (N Sweden) indicated that these plants were characterized by low δ¹⁵N values close to ERI and ECT species (not published).

These findings could be connected with increased N isotope fractionation by arbuscular mycorrhizal fungi and preferential transfer of ¹⁵N-depleted compounds from fungi to host plants in a similar fashion to the well-established transfer in ectomycorrhizal symbioses (Hobbie and Hobbie, 2008; Craine et al., 2009). However, it remains unclear why it occurs in meadow communities where nitrogen is probably more available to plants in comparison with heaths for which much higher δ¹⁵N values of AM species are typical (Schulze et al., 1994; Nadelhoffer et al., 1996; Michelsen et al., 1996, 1998). In addition, culture studies investigating the effects of arbuscular mycorrhizal colonization on plant δ¹⁵N indicated only modest changes and sometimes resulted in ¹⁵N enrichment rather than depletion (Handley et al., 1993, 1999; Azcon-Aguilar et al., 1998).

On the other hand, the rate of ¹⁵N fractionation by ecto- and ericoid mycorrhizal fungi probably depends on N availability. In conditions of strongly pronounced N limitation, plants take up the most part of N through mycorrhiza and thus become ¹⁵N-depleted, while for ecosystems with higher N availability and lower importance of mycorrhiza in plant N nutrition, similarity between ecto-, ericoid, and arbuscular mycorrhizal species is more typical (Hobbie et al., 2000, 2005). The alpine lichen heath, while being the most N limited among alpine ecosystems in the Teberda Reserve, is probably not as limited as arctic and subarctic tundra ecosystems, because annual net N mineralization rate in the heath soil (0.6 g m^-2) (Makarov et al., 2003) is higher than in soils of Arctic ecosystems (0.05–0.5 g m^-2) (Giblin et al., 1991; Schmidt et al., 2002). Probably this is the reason of relatively high δ¹⁵N values of ERI and ECT plant species, which isn't comparable with values close to -8‰ or -9‰, as it was observed when studying subarctic communities (Schulze et al., 1994; Nadelhoffer et al., 1996; Michelsen et al., 1996, 1998).

Therefore we assume that the reason of absence of a significant difference between AM, ERI, and ECT species in the Caucasian alpine heath community isn't clear and may result from both processes: the increased N isotope fractionation by arbuscular mycorrhizal fungi and the decreased role of ecto- and ericoid mycorrhizal fungi in the flux of N.

VARIATION IN δ¹⁵N VALUES WITHIN FUNCTIONAL GROUPS

The variability of δ¹⁵N within separate plant functional groups, as discussed above for legumes, can be due to species-specificity in isotopic fractionation (Emmerton et al., 2001), differences in plant preference for N forms (Schulze et al., 1994; Miller and Bowman, 2002), and their root placement at different soil depths (Schulze et al., 1994; Nadelhoffer et al., 1996).

We have no experimental data to evaluate the first two determinants, while the effect of rooting depth was not confirmed in our study, because δ¹⁵N varied widely among species of the same functional group with similar rooting depths. For example, arbuscular mycorrhizal grasses Helictotrichon versicolor and Bromus variegatus, both have rooting depths mainly within the upper 5–10 cm of soil but very different δ¹⁵N values of -2.2‰ and -5.5‰, respectively. Similarly, Anemone speciosa and Campanula tridentata (AM-FORB), both with deeper (15–20 cm) root systems (Onipchenko, 1987), have δ¹⁵N values of -2.2‰ and -4.4‰, respectively.

The dependence of δ¹⁵N of plant leaves on rooting depth is one of the least confirmed in field observations. We do not know of any studies that experimentally demonstrate any relationships between δ¹⁵N of plant leaves and rooting depth, although Schulze et al. (1994) and Nadelhoffer et al. (1996) hypothesized that deep rooting system can be responsible for the δ¹⁵N increase of forest and Arctic tundra plant species (see Introduction). Similar to our study, considerable foliar δ¹⁵N differences within functional groups were found for forbs, grasses, and sedges (rushes) in two dry meadow alpine tundra sites in the Rocky Mountains, Colorado (Miller and Bowman, 2002), while there were no differences in the δ¹⁵N of bulk soils within the upper 15 cm where roots of dry meadow species are concentrated. Therefore, the authors concluded that the observed variation in the foliar δ¹⁵N was due to factors other than rooting depth and was attributed to the acquisition of different N forms. At the same time, δ¹⁵N of non-mycorrhizal and ectomycorrhizal fine roots in European forest soils usually increased with soil depth (Högberg et al., 1996), and the idea of Taylor et al. (1997) that δ¹⁵N values of ectomycorrhizal fungi may be a function of the soil depth at which they utilize the N compounds was subsequently confirmed by Wallander et al. (2004) for the ectomycorrhizal mycelia, the δ¹⁵N of which increased with soil depth in two forest types.

Though we assumed that low δ¹⁵N values of the studied AM species may result from the increased N isotope fractionation by arbuscular mycorrhizal fungi, it also remains unclear whether arbuscular mycorrhizal fungi contribute to variation in foliar δ¹⁵N, because we did not find any relationships with mycorrhizal colonization rate for six AM species.

RELATION BETWEEN FOLIAR N AND δ¹⁵N

Positive relationships between foliar δ¹⁵N and N concentration were reported at the global (Craine et al., 2009) and landscape (regional) scale (Hobbie et al., 2000), indicating the effect of N availability. Hobbie et al. (2000) determined δ¹⁵N in plant materials across six sites representing different postdeglaciation ages at Glacier Bay, Alaska. When compared across all sites and species, the foliar δ¹⁵N values and N concentrations were generally lower at sites of low N availability, suggesting either an increased fraction of N obtained from mycorrhizal uptake, or a reduced proportion of mycorrhizal N transferred to vegetation. However, within our alpine lichen heath, higher foliar N concentrations and δ¹⁵N values are mostly connected with atmospheric N₂ fixation by legumes (see also Hobbie et al., 2000, for Glacier Bay) or utilization of fungi-derived ¹⁵N-enriched N by orchids and accompanying species.