Nitrate is an important nitrogen source for Arctic tundra plants

Significance

How terrestrial plants use N and respond to soil N loading is central to evaluating and predicting changing ecosystem structure and function with climate warming and N pollution. Here, evidence from NO₃⁻ in plant tissues has uncovered the uptake and assimilation of soil NO₃⁻ by Arctic tundra plants, which has long been assumed negligible. Soil NO₃⁻ contributed about one-third of the bulk N used by tundra plants of northern Alaska. Accordingly, the importance of soil NO₃⁻ for tundra plants should be considered in future studies on N and C cycling in Arctic ecosystems where C sequestration is strongly determined by N availability.

Abstract

Plant nitrogen (N) use is a key component of the N cycle in terrestrial ecosystems. The supply of N to plants affects community species composition and ecosystem processes such as photosynthesis and carbon (C) accumulation. However, the availabilities and relative importance of different N forms to plants are not well understood. While nitrate (NO₃⁻) is a major N form used by plants worldwide, it is discounted as a N source for Arctic tundra plants because of extremely low NO₃⁻ concentrations in Arctic tundra soils, undetectable soil nitrification, and plant-tissue NO₃⁻ that is typically below detection limits. Here we reexamine NO₃⁻ use by tundra plants using a sensitive denitrifier method to analyze plant-tissue NO₃⁻. Soil-derived NO₃⁻ was detected in tundra plant tissues, and tundra plants took up soil NO₃⁻ at comparable rates to plants from relatively NO₃⁻-rich ecosystems in other biomes. Nitrate assimilation determined by ¹⁵N enrichments of leaf NO₃⁻ relative to soil NO₃⁻ accounted for 4 to 52% (as estimated by a Bayesian isotope-mixing model) of species-specific total leaf N of Alaskan tundra plants. Our finding that in situ soil NO₃⁻ availability for tundra plants is high has important implications for Arctic ecosystems, not only in determining species compositions, but also in determining the loss of N from soils via leaching and denitrification. Plant N uptake and soil N losses can strongly influence C uptake and accumulation in tundra soils. Accordingly, this evidence of NO₃⁻ availability in tundra soils is crucial for predicting C storage in tundra.

Nitrogen (N) is often the nutrient that most limits terrestrial plant growth, making plant N availability a key determinant of primary productivity in terrestrial ecosystems (1). Hence, improved knowledge of in situ plant N availability and consequent plant N use is crucial for better evaluating and predicting responses of vegetation to climate change and N loading (2, 3). However, the availability of N to terrestrial plants is difficult to evaluate using measurements of soil N because of strong plant–microbe and plant–plant competition for N and the resulting rapid turnover of soil N pools (4).

Arctic ecosystems are typically characterized by strong N limitation (1). Because of high carbon (C) stocks in permafrost soil and their sensitivity to environmental change, the Arctic C cycle has important implications for global C balance and C-climate feedbacks (5, 6). Although it remains difficult to budget N inputs in the Arctic, the Arctic biome is a potential sink for anthropogenic N pollutants (7). So far, long-term N addition experiments have revealed that elevated N inputs into Arctic tundra ecosystems change C accumulation and species diversity (5, 8, 9). Field observations and isotope labeling experiments provide evidence of how added N has altered the distribution, fate, biotic use, and losses of N in Arctic tundra ecosystems (10–15). These studies indicate that a better understanding of in situ N availability in Arctic ecosystems is important because C and N cycles are tightly coupled between the vegetation and soils, and elevated N loading can influence the Arctic’s C balance (5, 16).

Nitrate (NO₃⁻) is a common and pivotal plant-available N form in addition to ammonium (NH₄⁺) and some forms of dissolved organic N (DON) (1). Until the 1990s, researchers underestimated the availability of soil NO₃⁻ to microbes because microbial uptake of NO₃⁻ often results in very low NO₃⁻ standing stock and low or negative net NO₃⁻ production (nitrification) rates in soil, even when gross nitrification rates are high (17–19). However, it remains undetermined how important soil NO₃⁻ is for plants because of inadequate understanding of in situ plant NO₃⁻ use. In Arctic tundra, NO₃⁻ availability can be increased by direct release from thawing permafrost, melting snow, and increased nitrification resulting from elevated N loading and warming temperatures (7, 14, 20). Elevated NO₃⁻ availability to tundra plants can change interspecific N competition and N-use strategies of tundra plants (9, 13, 21), potentially resulting in the spread of NO₃⁻-adapted species and altering the partitioning of above-ground vs. below-ground biomass (18, 22–24). These factors could alter CO₂ fixation by vegetation and the quantity and quality of litter inputs to the soil, which would then change microbial breakdown of soil C and the emission and uptake of greenhouse gases (5, 8, 25–27). Accordingly, soil NO₃⁻ availability and plant NO₃⁻ use have important implications for both N and C cycles in Arctic tundra.

Despite its potential importance, NO₃⁻ availability and the contribution of different N forms to plant N use have been unclear in Arctic tundra (21, 28). Four decades of research show that tundra plants rely on soil NH₄⁺ and DON (e.g., direct uptake of free amino acids) to meet growth requirements for N (12, 21, 28–31). In contrast, researchers generally have considered plant NO₃⁻ use to be negligible in the Arctic for several reasons. First, NO₃⁻ concentrations in soils are often low or undetectable, and soil net nitrification rates seldom show positive values (SI Appendix, Figs. S1 and S2), presumably because of low temperature, low soil NH₄⁺ availability, and low soil pH, together with high microbial N demand (32, 33). Second, plant-tissue NO₃⁻, a common marker of plant NO₃⁻ uptake, is rarely detected in tundra plants with conventional analytical methods (11, 12, 34).

We argue that the importance of NO₃⁻ to plants in such seemingly low-NO₃⁻ Arctic tundra ecosystems remains an open question for several reasons. First, although extractable soil NO₃⁻ concentrations are typically low in Arctic tundra soils, NO₃⁻ is sometimes present in measurable amounts and contributes nontrivial fractions of total extractable N (TEN) stocks similar to high-NO₃⁻ ecosystems (SI Appendix, Fig. S2B). Second, rates of in situ NO₃⁻ reductase activity (NRA), which is inducible and reflects the enzymatic NO₃⁻ reduction occurring in plants, are measurable in tundra plants and are not distinct from NRA rates measured in plants at lower latitudes (SI Appendix, Fig. S3). Accordingly, the abilities of Arctic tundra plants to assimilate NO₃⁻ are comparable to those of plants in relatively NO₃⁻-rich ecosystems. Third, controlled experiments revealed that tundra plants took up NH₄⁺ and NO₃⁻ at similar rates (9, 12, 29) or even took up NO₃⁻ at higher rates (33). Field ¹⁵N application (7, 13, 31) and modeling results (35) confirmed that tundra plants can assimilate NO₃⁻, NH₄⁺, and amino acids. All these observations illustrate that NO₃⁻ is an important soil N source in Arctic tundra and that tundra plants can use NO₃⁻. However, the relative importance of soil NO₃⁻ for plants in Arctic tundra ecosystems is unknown because we lack measures of in situ plant NO₃⁻ use and how it compares to that of plants in other NO₃⁻-poor or NO₃⁻-rich ecosystems.

Results and Discussion

Using the highly sensitive denitrifier method (detailed in Materials and Methods), we analyzed concentrations and stable isotope compositions of NO₃⁻ in tissues of dominant plant species in Alaskan tundra ecosystems. We then compared our results with those for plants from relatively high-N or high-NO₃⁻ ecosystems in lower-latitude regions (Figs. 1 and 2). Such comparisons of Arctic sites to non-Arctic sites, using both traditional and new methods, are important for understanding soil N cycling (particularly soil NO₃⁻ availability) and for placing the N uptake abilities of tundra plants into a broader context.

Fig. 1.

Concentrations of NO₃⁻ in plant leaves (A) and roots (B) across different ecosystems. The box encompasses the 25th to 75th percentiles, and whiskers are the SD values. The line and square in each box mark the median and mean values of studied plants at each site, respectively. Unique letters above the boxes mark significant differences at the level of P < 0.05. Detailed site information, including site abbreviation definitions, and species-specific values are given in SI Appendix, Tables S1 and S2. dw, dry weight.

Fig. 2.

Differences (∆ values) in δ¹⁵N (A) and δ¹⁸O (B) between leaf NO₃⁻ and soil NO₃⁻ across different ecosystems. The box encompasses the 25th to 75th percentiles, whiskers are the SD values, and the red line and red square in each box mark the median and mean values, respectively. Unique letters above the boxes indicate significant differences at the level of P < 0.05. The ∆ values were calculated using replicate values of plant tissues minus mean values of soil in corresponding sites (SI Appendix, Fig. S8 and Table S1).

The Uptake of NO₃⁻ in Plants.

The existence of NO₃⁻ in plant tissues is evidence for NO₃⁻ uptake from the soil or atmosphere because NO₃⁻ production in non-N₂ fixing plants is negligible under normal conditions (36–40). Although NO₃⁻ can be produced from the oxidation of nitric oxide (NO) both enzymatically and nonenzymatically in non-N₂ fixing plants (37–40), the rates are very low in natural environments (41–44), especially compared with the pool sizes of NO₃⁻ detected in plants of this study. Besides, while NO₃⁻ production by nonsymbiotic hemoglobin is possible in anoxic conditions (38, 39) and with high ambient NO concentrations (40), neither anoxic conditions nor high ambient NO applies to the present study.

We detected unexpectedly high NO₃⁻ concentrations in leaves and roots of the tundra plant species studied (Fig. 1 and SI Appendix, Tables S1 and S2). First, of the 153 tundra plant samples analyzed, 143 had measurable NO₃⁻ concentrations (detailed in Materials and Methods). Some species (e.g., Polygonum bistorta) had higher foliar NO₃⁻ than low-latitude forest species, including those in high-NO₃⁻ environments (Fig. 1A and SI Appendix, Table S2). Second, ratios of leaf NO₃⁻ to soil NO₃⁻ and of root NO₃⁻ to soil NO₃⁻ were similar between tundra and lower-latitude ecosystems or even higher in tundra than in some lower-latitude ecosystems (SI Appendix, Fig. S4). These results provide evidence of high NO₃⁻ uptake of tundra plants despite much lower concentrations of NO₃⁻ in tundra soils. Thus, we conclude that tundra plants can take up NO₃⁻ as efficiently as plants from relatively NO₃⁻-rich ecosystems in other biomes. In addition, NO₃⁻ additions to soils enhanced leaf NO₃⁻ concentrations in most tundra plants (SI Appendix, Figs. S5 and S6). This result is evidence that plant NO₃⁻ uptake is responsive to soil NO₃⁻ variations in Arctic tundra ecosystems. Such responses and patterns of NO₃⁻ uptake among studied species are useful for interpreting changes in functional traits and the structure of tundra plant communities in response to projected increases of soil NO₃⁻ with climate warming and elevated N deposition (1, 45).

The Sources of NO₃⁻ in Plants.

We used the ∆¹⁷O signatures of leaf NO₃⁻ (∆¹⁷O_leaf) to verify the mixing of atmospheric-derived NO₃⁻ [∆¹⁷O_atm > 0 per mille (‰) due to an enrichment in ¹⁷O during photochemical oxidization of nitrogen oxides (NOx) by O₃] with soil-derived NO₃⁻ (∆¹⁷O_soil = 0‰ because of no ¹⁷O excess in atmospheric O₂ and soil H₂O molecules) (46–48). Leaf NO₃⁻ of P. bistorta showed no ¹⁷O isotope anomaly (∆¹⁷O values = 0.0‰; SI Appendix, Fig. S7), indicating that the NO₃⁻ detected in this species was purely soil derived. Clearly, soil NO₃⁻ is available to, and taken up by, tundra plants.

In contrast, positive ∆¹⁷O_leaf values in low-latitude forests (SI Appendix, Fig. S7) indicate the direct leaf absorption of atmospheric-derived NO₃⁻ (∆¹⁷O > 0‰) or possibly the root uptake of NO₃⁻ at the surface soil with positive ∆¹⁷O values (49). We used mean ∆¹⁷O values of precipitation NO₃⁻ measured in the Tama-Kyuryo Field Museum forest in temperate Japan (TML) (see SI Appendix, Table S1 for descriptions of the forest sites used in this study) (49); in Guiyang in subtropical China (this study); and in Jianfengling forests in Hainan, tropical China (49) as ∆¹⁷O_atm values in the studied temperate, subtropical, and tropical forests, respectively (SI Appendix, Fig. S7). We then estimated mixing ratios of atmospheric-derived NO₃⁻ (∆¹⁷O_leaf:∆¹⁷O_atm) for plants in lower-latitude ecosystems. The results showed that atmospheric-derived NO₃⁻ accounted for, on average, 35% (6 to 86%) of total leaf NO₃⁻ in measured samples from lower-latitude forests.

NO₃⁻ Assimilation in Plants.

Higher δ¹⁵N and δ¹⁸O values in plant-tissue NO₃⁻ relative to source NO₃⁻ could provide new evidence for in situ plant NO₃⁻ assimilation because NO₃⁻ reduction via NO₃⁻ reductase would cause ¹⁵N and ¹⁸O enrichments in the unassimilated NO₃⁻ (2, 50–52). Accordingly, we calculated differences (∆ values) between isotopic values of tissue NO₃⁻ (δ¹⁵N and δ¹⁸O) in each plant sample and mean values of soil NO₃⁻ in corresponding ecosystems (Fig. 2 and SI Appendix, Fig. S8).

In northern Alaska, δ¹⁵N values of soil NO₃⁻ were 1.0‰ at Toolik Field Station (TFS) (see SI Appendix, Table S1) (21, 53) and 0.5 ± 4.7‰ at Barrow (54). Atmospheric-derived NO₃⁻ in snowmelt had lower δ¹⁵N values of −4.8 ± 1.0‰ at Barrow (54) and much lower values of −8.6 ± 0.7‰ at a high Arctic site at Midtre Lovénbreen, Svalbard (55). Compared with δ¹⁵N values of soil- or atmospheric-derived NO₃⁻ (SI Appendix, Fig. S8A), the higher δ¹⁵N values of leaf NO₃⁻ in tundra of northern Alaska (positive Δδ¹⁵N values; Fig. 2A) are evidence for in situ NO₃⁻ assimilation in tundra plants (Fig. 2A). The δ¹⁸O values of NO₃⁻ produced in high-centered soil polygons averaged −4.4 ± 2.7‰ at Barrow (54). By comparison, distinctly higher δ¹⁸O values of leaf NO₃⁻ than those of soil NO₃⁻ (positive ∆δ¹⁸O values; Fig. 2B) also provide evidence for in situ NO₃⁻ assimilation in tundra plants.

In non-Arctic sites, higher δ¹⁸O values of leaf NO₃⁻ than those of a soil- and atmospheric-derived NO₃⁻ mixture (distributed on the mixing line; Fig. 3) indicated assimilation of the mixed NO₃⁻ pool in the studied plants. However, higher ¹⁸O enrichments (SI Appendix, Fig. S8) might be due, in part, to contributions from high δ¹⁸O values of atmospheric-derived NO₃⁻ (57). Major uncertainties existed in fractional contributions of atmospheric-derived NO₃⁻ in leaf NO₃⁻ because of limited ∆¹⁷O data of leaf NO₃⁻ and lack of explicit ∆¹⁷O values of atmospheric NO₃⁻. Precipitation NO₃⁻ might not fully represent all atmospheric NO₃⁻ contributions to plant leaves; in addition, it is even more difficult to determine reasonable δ¹⁵N and δ¹⁸O end-member values of atmospheric-derived NO₃⁻ in plant leaves. Despite these problems, NO₃⁻ isotopes in plant tissues did provide information on plant NO₃⁻ sources and uptake in disturbed ecosystems.

Fig. 3.

Δ¹⁷O vs. δ¹⁸O plots of NO₃⁻ in soil, leaves, and atmospheric (Atmos, as precipitation or snow) deposition across different ecosystems. The mixing lines of Arctic and tropical sites (y = 2.52x − 4.42 and y = 2.97x + 0.58, respectively) were based on isotopic values of soil NO₃⁻ (n = 18) (54) and snowpack NO₃⁻ (n = 12) (56) at Barrow, and of soil NO₃⁻ (n = 18) and precipitation NO₃⁻ (n = 3) at Jianfengling in tropical China (49), respectively. The mixing line of temperate sites (y = 2.64x + 3.82) was based on isotopic values of soil NO₃⁻ at Japanese temperate sites (n = 22) and precipitation at TML (n = 12) in this study. The mixing line of subtropical sites (y = 2.87x + 0.91) was based on isotopic values of soil NO₃⁻ (n = 29) at subtropical sites and precipitation NO₃⁻ at Guiyang, China (n = 3) in this study. The Δ¹⁷O of soil NO₃⁻ was assumed to be zero.

Contributions of Soil NO₃⁻ to Total N in Tundra Plants.

Compared with plants in relatively N-rich ecosystems, tundra plants showed a similar distribution of leaf total N concentrations but a much wider distribution of leaf total (bulk) δ¹⁵N values (SI Appendix, Fig. S9). The wider distribution of leaf total δ¹⁵N values arises because of the strong niche differentiation of N-use regimes among tundra plants (13, 58). However, δ¹⁵N values of total N in tundra plants (−11.2 to 5.8‰ in Alaska) are generally lower than those of soil NH₄⁺ [around 12.3 ± 3.6‰ (this study); 4.4 ± 0.9‰ (53); and 1.4 ± 0.5‰ (21)], although some DON components are ¹⁵N depleted [around −5.7‰ for hydrolyzable amino acids (HAA) at Imnavait Creek (IMT) in northern Alaska; see SI Appendix, Table S1] (Fig. 4). This disparity between the δ¹⁵N signatures of plant total N vs. soil N sources exists even when isotopic fractionations for NH₄⁺ and HAA assimilation by mycorrhizal plants are considered. Given plant NO₃⁻ uptake and assimilation as indicated by NO₃⁻ in plant tissues, soil NO₃⁻ should be considered when using δ¹⁵N methods to evaluate in situ contributions of soil N sources to total N of tundra plants.

Fig. 4.

δ¹⁵N values of leaf total N and soil N sources of tundra plants in Alaska. AM, arbuscular mycorrhiza; ECM, ectomycorrhiza; ERM, ericoid mycorrhiza; NM, nonmycorrhiza. The box encompasses the 25th to 75th percentiles, and whiskers are the SD values. The line in each box marks the mean value. Plant δ¹⁵N data were summarized from ref. 58 and those of SI Appendix, Fig. S9B. The empty squares show soil δ¹⁵N data reported at IMT (53) and the blue-filled circle shows data at TFS (21). Soil δ¹⁵N-NO₃⁻ values of other sites are summarized from available data of non-Arctic sites in this study; soil δ¹⁵N-NO₃⁻ values at Barrow are cited from ref. 54.

Proportional contributions (f, expressed as a percentage) of soil NO₃⁻ to total N in tundra plants were estimated using δ¹⁵N values of soil N (NO₃⁻, NH₄⁺, and HAA) and δ¹⁵N values of leaf total N in a Bayesian isotope-mixing model [Stable Isotope Analysis in R (SIAR) (cran.r-project.org/web/packages/siar/index.html) (59)] (Fig. 5). The SIAR model uses a Bayesian framework to establish a logical prior distribution (60) for estimating f values, and then determines the probability distribution for the f values of each source (soil NO₃⁻, NH₄⁺, and HAA, in this study) to the mixture (total N of plant leaves, in this study). We contend that this approach provides reliable estimations of fractional contributions of different N sources to plant total N because the mixing model considers isotope effects during plant N uptake (¹⁵ε values hereafter) and variability in both source δ¹⁵N values and plant δ¹⁵N values (61).

Fig. 5.

Proportional contributions (mean ± SD) of soil NO₃⁻, NH₄⁺, and HAA to leaf total N of tundra plants in Alaska. The ¹⁵ε values [0‰ for NM plants, −5.0‰ for AM plants, −6.9‰ for ECM plants, and −7.7‰ for ERM plants (21, 62)] were considered for NO₃⁻, NH₄⁺, and HAA (scenario 1); for NH₄⁺ and HAA only (scenario 2); for HAA only (scenario 3); for none of NO₃⁻, NH₄⁺, and HAA (scenario 4).

In this study, the δ¹⁵N values (mean ± SD) of soil NO₃⁻ at Barrow [0.5 ± 4.7‰ (54)], soil NH₄⁺ at IMT and TFS (11.5 ± 8.4‰, this study and ref. 21), and soil HAA at IMT [−5.7 ± 2.2‰; (53)] were used as source δ¹⁵N values. For nonmycorrhizal (NM) plants, leaf δ¹⁵N values were mainly controlled by the δ¹⁵N values and f values of source N (NO₃⁻, NH₄⁺, and HAA), assuming negligible isotope effects during the acquisition processes of source N from soil into NM plants (i.e., ¹⁵ε = 0‰). For mycorrhizal plants, the ¹⁵ε values during the acquisition processes of soil N sources were calculated as the net differences of leaf δ¹⁵N values between mycorrhizal and NM plants. The same ¹⁵ε value was assumed for plant species associated with the same type of mycorrhiza and for N forms absorbed through the same type of mycorrhiza. In Alaskan tundra, the ¹⁵ε values for plant species associated with arbuscular mycorrhizae (AM), ectomycorrhizae (ECM), and ericoid mycorrhizae (ERM) were estimated as net δ¹⁵N differences from NM plants—that is, −5.0‰, −6.9‰, and −7.7‰, respectively (21, 62), which differed from the ¹⁵ε values normalized for worldwide plants [−2.0‰, −3.2‰, and −5.9‰, respectively (63)]. Our ¹⁵ε values (0‰ for NM plants, −5.0‰ for AM plants, −6.9‰ for ECM plants, and −7.7‰ for ERM plants) were considered under four scenarios (scenario 1: for NO₃⁻, NH₄⁺, and HAA; scenario 2: for NH₄⁺ and HAA only; scenario 3: for HAA only; and scenario 4: for none of NO₃⁻, NH₄⁺, and HAA) (Fig. 5). Estimates from natural ¹⁵N evidence were that NO₃⁻ assimilation accounted for 4 to 52% of species-specific leaf total N (around one-third, on average) of Alaskan tundra plants (Fig. 5), thereby demonstrating the importance of soil NO₃⁻ relative to soil NH₄⁺ and HAA for N use by many tundra plants. These findings also enhance understanding of N competition among plant species and between plants and microbes in Arctic tundra ecosystems, and how that may affect changes in species community composition and productivity with climate change and N pollution.

Materials and Methods

Study Sites and Sampling.

To evaluate in situ NO₃⁻ uptake and assimilation in terrestrial plants in relation to NO₃⁻ availability, we selected 18 sites (see descriptions in SI Appendix, Table S1) across a distinct gradient of soil NO₃⁻ (SI Appendix, Fig. S2), including one tropical and four subtropical sites in southwestern China; nine temperate sites in central, southern, and western Japan; and four Arctic tundra sites in northern Alaska. Among them, Tsukuba Forest Experimental Watershed (TKB) and Tama-Kyuryo Field Museum upper slope (TMU) and lower slope (TML) (SI Appendix, Table S1) are characterized by high soil NO₃⁻ or N saturation (49, 64, 65), while the Arctic sites TFS, Sagavanirktok River Valley (SAG), and IMT (SI Appendix, Table S1) are characterized by unmeasurable nitrification rates and negligible soil NO₃⁻ and, thus, are assumed to be typically low-NO₃⁻ ecosystems (SI Appendix, Fig. S2). In total, 28 plant species in the above study sites were sampled for fine roots (roughly <5 mm in diameter and <20 cm in spatial distribution of soil depth) or mature sunlit leaves. The studied plants in each ecosystem include dominant indigenous species (SI Appendix, Table S1). The design of this study allows us to evaluate plant NO₃⁻ use at the species and ecosystem levels.

Soil N Analyses.

Soil N concentrations and net N transformation rates (mineralization and nitrification) were measured as indices of potentially available NO₃⁻ for both plants and soil microbes. Information on soil types and samplings, N variables, and corresponding methods used for each ecosystem are summarized in SI Appendix, Table S1. Concentrations of NO₃⁻ and NH₄⁺ in soil solutions, extracts of fresh soils, and extracts of incubated soils (for net N mineralization and net nitrification rates) were determined colorimetrically. TEN was digested to NO₃⁻ using alkaline persulfate digestion and its concentration measured as NO₃⁻ on the autoanalyzer (specified in SI Appendix, Table S1). In-house standards (alanine, glycine, and histidine) dissolved in corresponding extracts were used for calibrating the concentrations of TEN and estimating the effect of the N blank from reagents (the same as that described in ref. 65). The soil extractable organic N was calculated as the difference between soil TEN and extractable inorganic N.

δ¹⁵N and δ¹⁸O ratios of soil NO₃⁻ were determined using the denitrifier (Pseudomonas aureofaciens) method (described in refs. 65 and 66) that converts NO₃⁻ to nitrous oxide (N₂O) (67, 68). The calibration curve between measured isotope ratios of N₂O and those of NO₃⁻ was prepared using US Geological Survey (USGS)-32, USGS-34, USGS-35, and International Atomic Energy Agency (IAEA) NO₃ standards. Soil NH₄⁺ in 100-mL extracts of IMT soil was separated onto glass filter papers (GF/D; Whatman) using the diffusion method (69), and then the NH₄⁺ diffused on the filter papers was measured for δ¹⁵N values on an elemental analyzer coupled with an isotope ratio mass spectrometer (EA-IRMS) (70) at The Ecosystems Center, Marine Biological Laboratory (Woods Hole, MA). IAEA-N₂ was run with the samples to check the accuracy of δ¹⁵N-NH₄⁺ data. The analytical precision was better than 0.2‰ for δ¹⁵N-NO₃⁻, 0.5‰ for δ¹⁸O-NO₃⁻, and 0.5‰ for δ¹⁵N-NH₄⁺. The respective natural abundances of ¹⁵N and ¹⁸O were reported as δ¹⁵N and δ¹⁸O values expressed in per mille units, as δ¹⁵N or δ¹⁸O = [(R_sample/R_standard) – 1] × 1,000, where R = ¹⁵N/¹⁴N or ¹⁸O/¹⁶O and standards are atmospheric N₂ and standard mean ocean water for N and O, respectively.

Plant N Analyses.

Leaf total N concentrations and total δ¹⁵N values of plant samples were analyzed using an EA-IRMS (detailed in SI Appendix, Table S1). The analytical precision for δ¹⁵N was better than 0.2‰. The leaf NRA assay, which has been used to evaluate the NO₃⁻-reduction potential of tundra plants [expressed per either fresh or dry weight (58, 71)], was conducted for plants at pristine and control sites of IMT, SAG, TFS-MAT (moist acidic tundra), TFS-MNT (moist non-acidic tundra), and at fertilized plots of TFS-MAT (SI Appendix, Table S1 and Fig. S3 A and B). The method of leaf NRA determination was the same as that described in refs. 58, 72, and 73. The NRA data (only those uniformly reported in dry weight) of natural terrestrial plants in low-latitude ecosystems were compiled (SI Appendix, Fig. S3C) for comparing NRA levels between tundra and low-latitude ecosystems.

The concentrations and δ¹⁵N and δ¹⁸O of NO₃⁻ in plants were measured using the sensitive denitrifier method (67, 68) at the Tokyo University of Agriculture and Technology (TUAT; method details are described in refs. 74 and 75). In the present study, 1 of 7 root samples of Eriophorum vaginatum and 7 of 94 leaf samples of tundra plants showed measurable NO₃⁻ concentration as zero, including 5 of 15 Sphagnum samples, 1 of 8 Cassiope tetragona leaf samples, and 1 of 1 Juniperus communis leaf sample.

The ∆¹⁷O values of NO₃⁻ in plant leaves were determined by combining bacterial reduction [i.e., denitrifier method (67, 68)] and the thermal decomposition method (76). First, NO₃⁻ in plant extracts was converted to N₂O using the denitrifier method (67, 68) at TUAT (method details are described in refs. 74 and 75). Next, the gold-tube conversion of bacteria-produced N₂O into N₂ and O₂ was conducted, and Δ¹⁷O values (defined as Δ¹⁷O = [(1 + δ¹⁷O)/(1 + δ¹⁸O)^β] − 1, where the constant β is 0.5247; see refs. 76 and 77) were measured on a Finnigan Delta Plus Advantage IRMS (Thermo Fischer Scientific) at the University of Washington (method details are described in ref. 78). A laboratory standard courtesy of Greg Michalski, Purdue University, West Lafayette, IN [NaNO₃ with Δ¹⁷O = 19.9‰ (79)] and several standards that mimic the 5% and 10% of atmospheric NO₃⁻ (i.e., Δ¹⁷O = 1‰ and 2‰, respectively) were used to check the precision of low Δ¹⁷O samples. The average SDs for replicate analyses of an individual sample were ±0.2‰ for Δ¹⁷O.